Animal: Global

ANIMAL PRODUCTION SYSTEMS
Global Workshop
VER
0
-f I
IDRC
'14C AUI 1.
RISPAL
9
IARC - Lib.
ANIMAL PRODUCTION SYSTEMS

Global Workshop
Edited by: Manuel E. Ruiz
Carlos Sere
Hugo Li Pun
Proceedings of the Global Workshop

on Animal Production Systems, held in
San Jose, Costa Rica, September 15-20, 1991,
sponsored by the International Development
Research Centre (Canada) and the Latin American
Research Network for Animal Production Systems (RISPAL)
Inter-American Institute for Cooperation on Agriculture

San Jose, Costa Rica
1995
Ya
4
C Inter-American Institute for Cooperation on Agriculture (IICA) / Latin

American Research Network for Animal Production Systems (RISPAL).
March, 1995.
All rights reserved. Reproduction of this book, in whole or in part, is prohibited

without the express authorization of the Inter-American Institute for Cooperation
on Agriculture and RISPAL.
The views expressed in signed articles are those of the authors and do not
necessarily reflect those of IICA and RISPAL.
IICA's Editorial Service and Print Shop were responsible for the stylistic revision
and printing of this publication, and RISPAL for the typesetting and layout.
Cover photographs: above right, Dr. Mario Tapia, Consortium for Andean
Sustainable Development (CONDESAN), National Potato Center, Lima, Peru;
above left and below, Dr. Manuel E. Ruiz, RISPAL, IICA, San Jose, Costa Rica.
Global Workshop on Animal Production Systems (1991 : San Jos6,

C.R)
Proceedings / ed. by Manuel E. Ruiz, Carlos Sere, Hugo Li
Pun. - San Jose, C.R. : IDRC : IICA. RISPAL, 1995.
506 p. ; 23 cm. - (Research and Development
Collection / IICA, no. 26)
ISBN 92-9039-267 3
1. Produccion animal. 2. Sistemas de produccion. I. Ruiz,

Manuel E. II. Sere, Carlos. III. Li Pun, Hugo. IV. IICA.
V. IDRC. VI. Titulo. VII. Serie.
AGRIS DEWEY
L01 338.6362
Research and Development Collection No. 26
The Research and Development Collection is designed to disseminate

significant research strategies and results to the agricultural and rural
sectors of the Americas.
San Jose, Costa Rica, 1995

TABLE OF CONTENTS
PREFACE ............................................ 7
INTRODUCTORY REMARKS
Hugo Li Pun .......................................... 9
CASE STUDIES ................................... 23
FODDER BANK: IMPROVING THE NUTRITION OF

CATTLE IN THE SUBHUMID ZONE OF WEST AFRICA
...............
M.A. Mohamed-Saleem and R.R. von Kaufmann 25
A CASE STUDY ON DUAL-PURPOSE GOAT

TECHNOLOGY DEVELOPMENT IN WESTERN KENYA
Fanny B. Nyaribo, J.F.M. Onim,
.........................
P.P. Semenye and B. Mwandotto 55
INTEGRATION OF CROSSBRED DAIRY GOATS INTO

BURUNDI'S FARMING SYSTEMS
Bernard Rey ......................................... 89
THE THREE-STRATA FORAGE SYSTEM FOR

DRYLAND FARMING IN INDONESIA
1. Made Nitis, K. Lana, M. Suarna,
...............................
W. Sukanten and S. Putra 119
TECHNOLOGY ADOPTION AND COMPETITIVENESS IN

SMALL DAIRY FARMS IN COSTA RICA:
A CASE STUDY
Federico Holmann, Ruben D. Estrada,
.....................
Francisco Romero and Luis A. Villegas 141
DUAL-PURPOSE CATTLE PRODUCTION IN GUATEMALA:

A CASE STUDY
Hugo E. Vargas, Ruben Dario Estrada
..............................
and Edgar Fernando Navas 169
4 Animal Production Systems: Global Workshop
EXPERIENCES OF THE INTERNATIONAL CENTERS 181
ANIMAL PRODUCTION SYSTEMS RESEARCH IN NEPAL:

A CASE STUDY
Jim A. Yazman and Enrique Ospina ........................ 183
EXPERIENCES WITH LIVESTOCK SYSTEMS RESEARCH AT ICARDA

Euan F. Thomson ..................................... 199
RESEARCH AT CIAT ON RUMINANT PRODUCTION SYSTEMS

Raul R. Vera ......................................... 219
ASIAN RICE FARMING SYSTEMS NETWORK: COLLABORATIVE

RESEARCH ON CROP-ANIMAL FARMING SYSTEMS
.................................... 229
Virgilio R. Carangal
RISPAL: THE LATIN AMERICAN RESEARCH NETWORK

FOR ANIMAL PRODUCTION SYSTEMS
Manuel E. Ruiz ....................................... 245
DONOR EXPERIENCES AND PERSPECTIVES ....... 267
LIVESTOCK PRODUCTION SYSTEMS RESEARCH:

A DONOR'S PERSPECTIVE
Joyce M. Turk ....................................... 269
IDRC EXPERIENCES IN THE SUPPORT OF ANIMAL PRODUCTION

SYSTEMS RESEARCH IN DEVELOPING COUNTRIES
Hugo Li Pun, Carlos Sere and C. Devendra ..................
279
LIVESTOCK SYSTEMS RESEARCH AT CIRAD

Bernard Rey, Philippe Lhoste
..................................... 299
and Vincent Dolle
ANIMAL PRODUCTION SYSTEMS RESEARCH: GTZ's

EXPERIENCES AND PROSPECTS
GunterKleemann ..................................... 309
ANIMAL PRODUCTION SYSTEMS RESEARCH IN THE TROPICS:

A REVIEW OF DUTCH EXPERIENCES
Johannes B. Schiere .................................. 327
Table of contents 5
SELECTED ISSUES ON ADAPTIVE ANIMAL

PRODUCTION RESEARCH IN AFRICA
A. Roeleveld ......................................... 355
NEW ISSUES ..................................... 371
THE ROLE OF LIVESTOCK IN THE DESIGN OF

SUSTAINABLE LAND-USE SYSTEMS
Robert D. Hart and Michael W. Sands ...................... 373
POLICY IMPACT ASSESSMENT IN SMALL-RUMINANT

PRODUCTION SYSTEMS: THE PERUVIAN CASE
Corinne Valdivia ...................................... 385
GENDER ISSUES IN ANIMAL PRODUCTION

SYSTEMS RESEARCH
July Leesberg ....................................... 417
WORK GROUP SESSIONS ......................... 437

INSTITUTIONAL ISSUES
A report by Working Group 1 ............................ 439
GENDER ISSUES
THE ROLE OF NARIS, IARCS, NGOS, AND NETWORKS
IDENTIFICATION OF ALTERNATIVES
POLICY/COMMODITY SYSTEM PERSPECTIVE
SUSTAINABILITY PERSPECTIVES
METHODOLOGICAL ISSUES
A report of Working Group 7 ............................. 469
6 Animal Production Systems: Global Workshop .
IMPACT ASSESSMENT
CONCLUSIONS AND RECOMMENDATIONS ........ 485
CONCLUSIONS AND RECOMMENDATIONS ................. 487

APPENDICES ..................................... 489
LIST OF PARTICIPANTS ................................ 491
LIST OF ACRONYMS AND ABBREVIATIONS ................. 497

PREFACE
The world as a whole still faces deep-rooted problems that affect not
only the lives of the present population, but those of the future generations
as well. Most pressing among these problems are food supply and secur-
ity, whose intensity will increase with the growing world population,
expected to reach over 10 billion people in the year 2060. Food supply
and security are old, but familiar problems. The urgency is not only the
terrible loss of lives and suffering this generation is witnessing, but also the
effects that hunger has on the intellectual and emotional development of
the surviving young population. This problem is closely associated with
poverty, which in turn is closely linked to rural people. The irony is that
these people are the ones called on to help solve the food needs of the
world.
The rural poor live in areas that are environmentally vulnerable and
often use natural resources over which they have little legal control; they
are often impelled to use lands that are not suitable for agriculture, such
as steep hillsides and tropical forests, and for this reason many associate
the rural poor with natural resource degradation.
One form of land use is agriculture, and one of the major agricultural
activities is animal production. In Latin America, for example, over 75% of
agricultural land is being used for livestock production. In addition,
livestock plays other recognized roles in rural development, as a source of
income, employment, and traction. It contributes benefits to the
environment through nutrient cycling. It also provides security to the rural
poor, as it serves as a buffer against climatic and economic instability.
Livestock in extensive systems in the tropics, especially beef ranching, has
been associated with an increased rate of deforestation. This has high
economic and environmental costs for society; fortunately, more
sustainable production systems are also practiced in tropical areas of
developing countries.
Finding solutions to these problems must break away from long-

established paradigms. Animal agriculture can no longer be production-
driven, but must be environmentally-conscious, natural-resource sparing,
and internationally competitive. For this, new technology will be needed,
but technology alone cannot be expected to solve all problems. Socio-
economic reforms will be needed and recognition of farmers' basic
objectives in life is essential if the new endeavors are to have any chance
of success. Gender roles in agriculture must be acknowledged if equitable
development is to be achieved. Human resource training, at all levels, will
have to be a cornerstone of any development program, and institutional
(local, regional and international) reforms will have to be included in order
to make these institutions more responsive to social and environmental

concerns and more effective in their responsibility to ensure adequate food
supply and distribution.
Important knowledge that will contribute to finding appropriate solutions

has come from holistic systems-oriented research and development
initiatives in animal agriculture. This book contains contributions from well-
recognized leaders in research, development, training and international
cooperation. Experiences from Africa, Asia and Latin America illustrate the
application of the systems approach to animal agriculture and the technical
and socioeconomic results obtained. The views from the international
community (international agriculture centers and donor institutions) provide
an overall, in-depth analysis of the effectiveness of supporting animal
production work, as well as of the future orientations that this activity
should take, and of the new strategies and reforms the institutions should
adopt. It is hoped that lessons derived from past research and
development projects will be incorporated in future endeavors. It is also
expected that an expansion of farming systems research will be used to
address the more complex problems of natural resource utilization in
developing countries.
producing this book, several people must be thanked. First, all the
In
contributors, for their willingness to share their knowledge and wisdom in
the workshop that produced this publication, and for their help in the final
text corrections. Deep appreciation is also due to Dr. Carlos Sere for his
help in planning the workshop and for the technical editing of some of the
papers. Ms. Susan E. Ruiz, with the participation of Ms. Stella Feferbaum,
was responsible for the arduous formal task of editing every paper.
Special thanks are due to the RISPAL personnel Dr. Arnoldo Ruiz and Ms.
Olga Rojas, Roxana Araya and Maria Ileana Mora for their assistance
during the workshop. Ms. Helen Raig, of the IDRC Regional Office in
Montevideo must be recognized for her prompt and effective collaboration
during the organization stage. The computer assembly of the document
was possible thanks to the expertise of Ms. Rita Herrera and Sonia Castro.
Finally, recognition is due to IICA, and particularly its RISPAL network, for
hosting the meeting, and to IDRC for supporting the meeting and the
publication of its proceedings.
Manuel E. Ruiz Hugo Li Pun

RISPAL, Inter-American Institute International Development
for Cooperation on Agriculture Research Centre
Coronado, Costa Rica Ottawa, Canada
INTRODUCTORY REMARKS
Hugo Li Pun' 2
For the past two decades, different agricultural research and

development agencies have been conducting activities with a systems
perspective. This approach has been shown to be of a holistic,
multidisciplinary, and participatory nature. Within these characteristics, a
wide variety of experiences has been developed, depending on the nature
of the problems to be solved and the scope of the projects.
Why systems research?
Research with a systems approach was developed in response to

mounting concern for the problems of small farmers and the realization
that they have contributed substantially to the economies of developing
countries. It was recognized that the technologies generated through the
Green Revolution did not cause the expected impact under conditions of
limited resources, whether of a physical, biological or economic nature
(Simmonds 1985). Therefore, more comprehensive, multidisciplinary
approaches were proposed.
The main concerns in systems-oriented projects included the perceived

needs for food in developing countries and the need to promote more
equitable development. Thus, small farmers were selected as the target
population.
' Senior Program Specialist, Environment and Natural Resources Division, IDRC, Ottawa,
Canada.
2
The author wishes to acknowledge the valuable comments and suggestions of Dr.
Carlos Sere, who reviewed the present document. Thanks are also expressed to Mr.
Gabriel Quijandria, who helped in the preparation of the table and figures and the
gathering of data.
Evolution of systems research
Systems research in general is not a new field. It has been conducted

in different disciplines, including the study of biochemical processes at the
cellular level, the study of digestive and metabolic processes, and the
simulation of animal performance, among others.
The application of systems theory to the problems of small farms,

however, started with cropping systems projects. The main steps of the
methodology have been outlined by different authors (Norman 1976;
Zandstra et al. 1981; Shaner et al. 1982). Basically, it involved the
following steps: characterization, design, on-farm testing, and technology
transfer. Several different variations on this general scheme have been
proposed and put into practice. Experiences varied in different parts of the
world. It is not the purpose of this document to review these in detail, but
rather to point out their main thrusts: to work with different crop
arrangements in time and space, to introduce higher yielding and/or early
maturing varieties, and to study different management practices, to name
a few. Some of these were highly successful; others, not so successful.
These kinds of experiences led to the support of systems research in
animal production on small farms. Research efforts started in the mid-
1970s. Some of those carried out in Central America were pioneering.
After the first studies were started, it was realized that the various steps of
the general systems methodology. could not be used in the same way as
with cropping systems. Differences associated with the long reproductive
cycle of large ruminants, the small number and high variability of
experimental units, the mobility of animals, and the different roles that
animals play on small farms (which could result in conflicting managerial
decisions), among other reasons, made animal production research on
small farms difficult. To overcome some of these problems, a series of
activities was initiated in different parts of the world. In the case of Latin
America, a network, the Latin American Research Network for Animal
Production Systems (RISPAL), was organized; it had been in operation
informally since 1981. In 1986, IICA formally implemented it, assigning a
full-time coordinator.
As some of the projects worked with farmers who had very limited
resources, strong interactions between the different components of the
farming system were recognized. Therefore, projects in mixed systems
were started in Asia, Africa, and Latin America. The success of some of
these projects was limited, due to the complexity of the target system. As
more was learned about the complexity of the systems, the importance of
Introductory Remarks 11
other disciplines was realized. Therefore, sociological studies were

included to better identify farmers' needs and to identify opportunities for
intervention. In many circumstances, technological solutions were not
deemed possible unless changes were promoted in the socioeconomic
environment, which in many cases had been considered a fixed part of the
exogenous factors.
The added dimensions of the problem have led some to claim that the
farming systems research (FSR) approach is limited in scope, impact, and
scientific method; in addition, it has been criticized because of its high
cost, the long descriptive phases, and the large amount of farm data
collected. Conflicts also appeared between FSR and discipline-oriented
researchers, as well as with some research managers. Although there are
indications that some of these criticisms may be valid in certain projects,
it is also true that overexpectations were created in the enthusiasm to
promote FSR-type approaches. This has led to some cases of donor
fatigue, as well as to unfair blame on FSR and a sense of lack of credibility.
In many instances, the complexity of the problems is of such magnitude

that unless structural changes take place, the problems will not be solved.
Therefore, it is unfair to blame the lack of impact on approaches and
methodologies that are still evolving. Furthermore, if the new concerns
(environment, sustainability, gender, and income and employment
generation) are to be tackled, the systems approach need not be changed;
rather, the scope may need to be widened in order to encompass a more
comprehensive view and inferences need to be built from the
agroecosystem hierarchy to the regional and country hierarchies (Hart
1979). In that respect, policy, markets, and consumer aspects may need
to be included in a systematic way. The dynamism of the exogenous
factors and their impact on the target systems need to be analyzed more
carefully.
This whole discussion may lead to the idea that the problem is
tremendously complex and that finding solutions will require the
participation of a wide range of disciplines and very complex studies. In
reality, the magnitude of the problem depends on the type of target system
and on the circumstances surrounding it. The extreme is the situation of
the minifundio in marginal areas, where the systems are complex and the
possibilities for improvement are limited. However, there are also favorable
cases, where systems are more specialized, and the conditions in terms of
resources (land, capital, type of soils) and climate are better.
Comparing the issues to be resolved during the past decade with the
situation in the 1990s, it can be concluded that the situation has not
improved substantially. In most developing countries, both food
consumption (especially animal protein consumption) and per capita gross
domestic product are still very low (Tables 1 a, 1b, 1c and Figs. 1, 2, and
3).
The consumption of animal protein in such countries is associated with

increases in the income of the population (Mellor 1989). If the economies
of these developing countries improve in the years to come, the potential
for increased demand of animal products is high, in view of current low
levels of consumption and slow growth rates of animal protein
consumption (Figs. 4, 5, and 6). Therefore, the need to increase animal
production in developing countries still exists. There is a great challenge
for systems researchers, not only to resolve the issues of the past, but also
to incorporate the new ones (environment, sustainability, income and
employment generation, gender), as well as the development of
appropriate agricultural policies. Lastly, but equally important, the
constrained external financial resources are generating pressure for a more
efficient use of resources to solve critical problems.
Objectives of the present meeting
To analyze case studies of animal production systems research in

Africa, Asia and Latin America in order to assess the methodologies
followed, the results, the constraints, and also to derive lessons for
future activities.
To discuss new development concerns and how they can be

incorporated in the research agenda.
To identify constraints on animal production systems research and to

propose suggestions to overcome them.
To promote the exchange of information across regions, as well as the

more efficient use of resources for research, through horizontal
collaboration among institutions in developing countries.
It is hoped that the conclusions to be derived from this meeting will be

useful in guiding future work in animal production systems research.
Table la. Protein consumption levels, Africa, 1979-81 and 1986-89
Total Animal Total AnimalAnimal protein

protein protein protein protein consumption
g g g g ann. growth
ountry 1979/1981 1979/1981 1986/1989 1986/1989 rate %
Algeria 64.1 11.4 70.4 14.6 2.8

Benin 45.4 5.5 46.6 5.0 -1.1
Botswana 66.4 20.0 68.3 16.6 -2.0
Burkina Faso 56.5 5.1 61.5 5.2 0.2
Burundi 77.4 3.5 73.5 1.9 -6.6
Cameroon 51.7 8.3 51.0 10.3 2.4
Cape Verde 64.3 15.7 61.7 12.6 -2.4
Central Afr. Rep. 40.1 8.1 44.4 10.1 2.5
Comoros 38.3 8.8 41.6 8.7 -0.1
Congo 40.2 12.8 46.8 17.0 3.2
Cbte d'Ivoire 52.3 11.7 49.2 10.4 -1.3
Gambia 46.9 9.3 55.6 11.9 2.8
Ghana 39.8 9.4 42.3 9.2 -0.2
Guinea 44.3 4.3 44.0 4.4 0.3
Kenya 56.4 12.3 55.5 14.1 1.5
Lesotho 66.2 9.5 63.4 8.9 -0.7
Liberia 41.1 6.2 38.9 5.4 -1.5
Madagascar 53.6 9.9 48.6 8.9 -1.2
Malawi 64.0 4.8 58.5 4.6 -0.5
Mali 49.9 11.5 59.0 9.5 -2.1
Mauritania 63.2 25.2 71.0 25.2 0.0
Mauritius 59.2 19.4 60.4 17.8 -1.0
Morocco 68.7 9.5 76.3 10.4 1.0
Mozambique 30.2 3.4 27.0 3.0 -1.4
Niger 64.0 9.6 65.0 8.2 -1.7
Nigeria 49.0 8.2 45.8 3.7 -8.5
Reunion 60.1 18.8 59.8 19.8 0.6
Rwanda 48.6 2.7 48.7 2.8 0.4
Sao Tom
and Principe 46.0 6.8 46.1 12.0 6.5
Senegal 66.1 14.8 67.9 17.6 1.9
Sierra Leone 43.2 9.1 37.8 6.6 -3.5
Somalia 52.3 24.0 52.5 23.1 -0.4
Swaziland 62.8 21.1 58.9 17.3 -2.2
Seychelles 60.3 29.7 56.8 25.0 -1.9
Tanzania 52.2 9.1 52.8 9.9 0.9
Togo 45.9 4.7 48.3 5.8 2.4
Tunisia 73.0 12.4 79.1 14.9 2.1
Uganda 49.2 10.3 45.4 7.8 -3.0
Zaire 30.2 3.1 30.9 4.2 3.4
Zambia 54.6 6.2 53.0 6.2 0.0
Zimbabwe 54.8 6.5 53.2 6.2 -0.5
Table 1b. Protein consumption levels, Near and Far East, 1979-81 and 1986-89
Total Animal Total Animal Animal protein

protein protein protein protein consumption
g g g g ann. growth
Region Country 1979/81 1979/1981 1986/1989 1986/1989 rate %
NE (1) Egypt 72.9 9.4 78.6 10.7 1.4

Kuwait 92.3 49.7 92.0 50.1 0.1
Libya 81.0 27.5 78.7 26.2 -0.5
Saudi Arabia 76.9 31.4 84.3 34.1 0.9
Sudan 61.4 20.2 56.3 18.7 -0.9
Syria 80.0 21.2 84.2 20.6 -0.3
Turkey 86.2 19.9 84.1 17.9 -1.2
U. Arab 100.1 53.9 99.3 47.5 -1.4
Emirates
Yemen AR 68.0 14.6 64.8 14.8 0.2
Yemen DEM 61.2 20.9 69.7 22.2 0.7
FE (2) Bangladesh 41.6 4.7 41.1 4.7 0.0

Brunei Darus 71.9 36.5 78.5 38.8 0.7
Hong Kong 85.3 50.3 88.3 53.9 0.8
India 50.0 5.6 51.3 7.2 2.8
Indonesia 50.2 6.1 57.8 7.8 2.8
South Korea 75.5 18.6 78.1 26.0 3.8
Macau 67.2 36.2 61.0 31.2 -1.6
Malaysia 58.8 25.6 57.6 26.1 0.2
Nepal 51.0 7.6 52.3 7.6 0.0
Pakistan 59.9 12.0 57.0 12.9 0.8
Philippines 53.1 21.0 51.5 20.1 -0.5
Singapore 72.1 39.0 81.4 46.8 2.0
Sri Lanka 44.5 8.8 47.1 8.9 0.1
Thailand 48.2 14.3 49.1 14.8 0.4
CPE (3) China 55.1 7.6 63.6 12.1 5.3

North Korea 83.0 15.2 88.9 18.1 2.0
Mongolia 86.7 55.2 84.0 49.2 -1.3
(1) Near East
(2) Far East
(3) Centrally Planned Economies.

Table 1c. Protein consumption levels, Latin America, 1979-81 and 1986-89
Total Animal Total Animal protein

Animal
protein protein protein proteinconsumption
g g g g ann. growth
ountry 1979/1981 1979/1981 1986/1989 1986/1989 rate %
Antigua 59.9 38.4 67.4 42.0 1.0

Barbuda
Argentina 107.9 71.8 105.7 68.2 -0.6
Bahamas 67.0 42.2 76.6 48.1 1.5
Barbados 86.5 51.7 99.0 61.6 2.0
Belize 68.7 33.0 70.3 36.3 1.1
Bolivia 54.4 17.8 57.5 18.7 0.5
Brazil 61.1 22.7 62.5 23.9 0.6
Chile 73.1 27.5 68.8 25.5 -0.8
Colombia 54.3 23.4 55.8 23.9 0.2
Costa Rica 63.8 31.1 73.4 30.0 -0.4
Cuba 73.0 35.1 77.4 37.3 0.7
Dominican 49.2 19.2 48.7 17.3 -1.2
Republic
Ecuador 48.2 22.1 53.6 24.7 1.2
French Guiana 79.4 46.3 91.5 52.2 1.3
Grenada 70.2 35.1 82.2 46.3 3.1
Guadeloupe 81.4 47.4 89.2 52.7 1.2
Guatemala 57.0 12.3 60.2 11.8 -0.5
Guyana 61.2 24.7 59.2 23.5 -0.6
Haiti 47.7 7.9 49.4 9.7 2.3
Honduras 54.0 15.6 51.2 13.5 -1.6
Jamaica 61.4 26.1 63.6 26.1 0.0
Martinique 81.0 44.0 84.8 45.3 0.3
Mexico 78.0 28.9 81.0 30.1 0.5
New Antilles 93.1 61.1 88.8 58.1 -0.6
Panama 58.5 29.3 63.3 32.3 1.1
Paraguay 77.7 33.1 74.2 30.2 -1.0
Peru 57.5 20.0 60.1 21.8 1.0
Saint Lucia 60.4 32.6 76.5 42.9 3.1
Saint Vincent 56.6 22.5 66.8 31.5 3.8
Suriname 64.0 26.9 65.1 24.4 -1.1
Trinidad & 75.5 34.5 76.2 35.0 0.2
Tobago
Uruguay 84.2 53.2 84.6 51.4 -0.4
Venezuela 71.4 36.3 67.1 32.8 -1.1
26 1
13
241
13
c.
13
13
0 4
0
2-I
0
0.4 0.8 1.2 1.6 2.0 2.4
Per capita gross national product, US $1000
r
Fig. 1. Animal protein consumption versus income level, Africa, 1988
Calculated from FAO (1989).

60 1
2o. 50
o
U
rn 40-i
o
0
nE 30-
N
C
O
U
11
1
13
1 1
0
1
2 3 4 5 6 7 8 9 10
Per capita gross notional product, US $1000
Fig. 2. Animal protein, consumption versus income level, Asia, 1988.

70 I
60-
o
0
0 0
0
30-0 0
0 13
O p 13
0
O
a 0
=0 20.1
E
c
0 O
T 0
= 10-1 11
V
0.6 1.0 1.4 1.8 2.2 2.6 3.0 3.4
0.2
Per capita gross national product, US$ 1000
1988.
Fig. 3. Animal protein, consumption versus income level, Latin America,

o
0 4 8 12 16 20 24
Daily APC. 9 Per copita
Fig. 4. Growth rate of animal protein consumption (APC) versus protein

consumption in Africa.
Calculated from FAO (IM).
C
6
3 m
2 11
ai
I
o
i
a 6 r1 u
L)
aa -3
-4
-5
-6
if
0 10 20 30 40 50 60
Doily APC, g per capita

consumption in Asia.
4
O
3 13
I 0 0 13
11
13
0- e O
0 0 a
0 0
0
0 O 13
0
-2
-3
-4
0 10 20 30 40 50 60 70
Daily APC, g per capita
consumption in Latin America.
LITERATURE CITED
HART, R.D. 1979. Agroecosistemas: Conceptos basicos. Turrialba, C.R.,

Tropical Agricultural Research and Training Center (CATIE). 211 p.
FAO (FOOD AND AGRICULTURE ORGANIZATION). 1989. Production

yearbook, Vol. 42. Rome, Italy. Food and Agriculture Organization.
MELLOR, J.W. 1989. The political and economic context for development
of animal agriculture in developing countries. In AID Summary report
of the Animal Agricultural Symposium: Development priorities toward the
year 2000. Washington, D.C. p. 7-20.
NORMAN, D.W. 1976. The social scientist in farming systems research.

Invited paper in the Workshop on Farming Systems Research in Mali,
sponsored by Institut d'Economie and the Ford Foundation. Institut
d'Economie Rurale, Mali. Nov. 15-20, 1976. 11 p.
SHANER, W.W.; PHILLIP, P.E.; SHMEL, W.R. 1982. Farming systems

research and development. Boulder, Colorado, Westview Press. 414 p.
SIMMONDS, N.W. 1985. Farming systems research: A review. World

Bank Technical Paper No. 43. Washington, D.C. 109 p.
ZANDSTRA, H.G.; PRICE, E.G.; LITSINGER, J.A.; MORRIS, R.A. 1981., A

methodology for on-farm cropping systems research. Manila,
Philippines. The International Rice Research Institute. 149 p.
CASE STUDIES
FODDER BANK: IMPROVING THE NUTRITION
OF CATTLE IN THE SUBHUMID
ZONE OF WEST AFRICA
M.A. Mohamed-Saleem and R.R. von Kaufmann'
BACKGROUND
ILCA acknowledged the potential for increased animal agriculture in the

subhumid zone of West Africa by establishing a research base at Kaduna
in northern Nigeria. FuIBe agropastoralists were selected as the target
group because they own most of the cattle in the zone and cattle are the
predominant livestock, at least in terms of biomass and traded products.
After consultation with experts, farm surveys, and a simulation analysis,
ILCA chose to focus the research on overcoming the dry-season nutrition
constraint. Although nomads used to be the principal cattle owners in
West Africa, agropastoralism and mixed farming are becoming the
dominant modes of livestock production. Agropastoralists are settling in
the wetter zones where the trypanosomiasis challenge is declining as
insecticides against the carrier tsetse fly (Glossina spp.) are used and land
is cleared by cultivators, reducing the tsetse habitat. Previously cattle
could only use the zone in the dry season.
The West African region is faced with exponential population growth

and a widening gap between the demand for food and food production.
In many areas, productivity is suffering as land is being exploited beyond
its biophysical limits. This makes the relatively underutilized and lightly
populated subhumid zone an attractive proposition for agricultural
development.
The subhumid zone is defined by the isolines of 180 to 270 days of

growing period and a 900 to 1,500 mm rainfall. In West Africa, this is a
continuous belt extending roughly parallel to the equator between latitudes
1
International Livestock Center for Africa (ILCA), Addis Ababa, Ethiopia.
6° and 12° North and covering approximately 1.3 million km2 (23% of the
continent). Included in this zone are portions of Senegal (5%),
Guinea-Bissau (90%), Guinea (90%), Sierra Leone (45%), Mali (5%), We
d'lvoire (30%), Ghana (60%), Togo (80%), Benin (75%), Nigeria (50%), and
Cameroon (10%) (Fig. 1).
Fig. 1. Subhumid zone of Nigeria and Africa
One third (455 500 km2) of the West African subhumid zone is in
Nigeria. Migration of settlers into this zone has been rapid in the past
decade. Transhumant livestock owners are taking up permanent
Case Studies 27
residence, usually on the peripheries of communities of cultivators, where

they find more social amenities and better markets. However, the majority
of the people in the zone are cultivators who depend on cereal grains and
some root crops for subsistence. They own some livestock, mainly poultry
and goats, but increasing numbers are also investing in cattle. The FulBe
are specialized cattle and sheep rearers; even when settled, they continue
to derive most of their livelihood from cattle. However, since they are
relying increasingly on cropping for subsistence and no longer follow
transhumance traditions, they are classified as agropastoralists.
LAND USE IN THE SUBHUMID ZONE
Recent events in the subhumid zone of Nigeria exemplify trends

anticipated for other West African countries. Although only 20% of the
subhumid zone was cultivated in 1983 (Bourn and Milligan 1983), shifting
cultivation was nevertheless fast giving way to more intensive land-use
systems. At the estimated rate of 4.8% expansion in area per year, 50%
of the zone will be under cultivation by the year 2000. Traditionally, land
is seldom bought or sold. Arrangements for its use are negotiated
between the appropriate individuals, groups, or clans. Individual land
holdings can vary widely in area, but the extent of cultivated land is
determined by the amount of available family labor.
The natural fertility of the soil is fast declining because of shorter fallow
periods. The soils of the subhumid zone are typically extremely weathered,
weakly structured, laterized, and deficient in most of the essential nutrients.
The dominant soil types are the tropical ferruginous soils, characterized by
downward movement of clays which solidify with lower layers of iron oxide
to form hardpans.
Cattle are very important in the agricultural economies of the

sub-Saharan African countries. Besides contributing milk and meat, they
also have important traditional economic and social values. Zebu (Bos
indicus) breeds such as the Bunaji (White Fulani) and Sokoto Gudali make
up most of the cattle in the zone. However, trypanotolerant Bos taurus
breeds such as Muturu and N'Dama, as well as fixed indicus x taurus
crosses (such as the Keteku) that are also believed to be trypanotolerant,
are found in the areas that are wetter and more heavily infested with tsetse.
It was estimated by aerial survey that there are 4.5 million head of cattle in
the subhumid zone of Nigeria in the wet season and 4.9 million in the dry
season (Bourn and Milligan 1983). This means that there is now no major
influx of cattle in the dry season as in the past; the zone used to be visited
by nomads in large numbers when the shade-seeking tsetse fly had to
retreat further south.
Cattle productivity in the subhumid zone of Nigeria is poor, even in

comparison with productivity levels found in other stressed environments.
The productivity of nonsupplemented, traditionally managed Bunaji cows
was found to be 51.5 kg of milk per cow per year (Mani et al. 1993),
compared to 181 kg attained under ranch conditions. The dry season
varies from five to eight months, and the quality and quantity of the forage
fluctuates seasonally (Fig. 2). The natural flora is dominated by grass, but
even if this is available in adequate quantities, the intake by ruminants
decreases as the crude protein and digestibility decline with maturity
(Mohamed-Saleem 1984).
Cereal crop residues are available in substantial quantities after harvest,

but they are also of low quality except for a limited period immediately after
harvest, when immature leaves and unharvested panicles are available.
These are rapidly and selectively grazed (Powell 1986), but even this short
boost in the nutritional status of the cows is sufficient to induce estrous
activity (Otchere 1986). Providing cows with cotton seed cake as a
supplement increased calving rates from 33% to 77% (Synge 1980). This,
as well as the response to supplementary feeding of wheat, shown in Table
1, indicate the value of supplementing natural grazing to increase
productivity. However, the practice of feed supplementation with
agroindustrial by-products can not be sustained because of the erratic
supply and constant inflation. These facts drew attention to the possibility
of introducing forage legumes into agropastoral production systems.
GOVERNMENT DEVELOPMENT PRIORITIES AND POLICIES
The government of Nigeria has employed various strategies to improve

livestock productivity. For many years, most of the available resources
were allocated to combating the major epidemic diseases, particularly
rinderpest and contagious bovine pleuropneumonia. By 1950 Nigeria had
these very much under control, but there were still difficulties in
administering health care to mobile herds. This latter problem was the
greatest constraint to the introduction of production innovations.
Settlement of pastoralists thus became a priority. State governments tried
Case Studies 29
to set aside vast areas of land as "grazing reserves" in the belief that the
nomads would settle if they were provided with secure pasture and water.
A lot of effort went into developing infrastructures such as roads, fire
traces, dams, and milk collection centers; however, there was little
response from the nomads.
- Grass CP
Forb C P
Gross DM
16
14
2800
Forb DM / \ \ 12
10 co"
2400
2000
r0
rn 1600
4
1200
d J 2
T
CL
f 800 Mr. 55 to
0
r
m
to
400
M
r 45 ?
I
0 3 I LJ 35
J F M A M J J A S 0 N D
Months
Fig. 2. Generalized productivity and utilization pattern of natural

herbage at Kachia Grazing Reserve, Nigeria
Table 1. Fulani (Bunaji) cattle under traditional management:

Production responses due to supplementary feeding of
wheat
Period Control herds Supplemented herds
- Mean milk yield, kg/head/day) -
Dry season 0.505 0.802

Rainy season 0.911 0.935
Whole year 0.708 0.869
- Mean milk yield per cow over 3 years

(kg/head/day)' -
Dry season 0.183 0.413

Rainy season 0.300 0.450
Whole year 0.245 0.432
1 All possible breeding females, not just those in milk.
Source: Synge (1980).
Meanwhile, the agricultural research stations were introducing and

screening new varieties of forages in attempts to make up for deficiencies
in natural pastures. There was no tradition of pasture seed multiplication;
therefore, seeds of promising species continued to be imported until the
government set up seed multiplication units in the late 1970s. Large tracts
of land were oversown with improved forages with the intention of
improving the composition of the sward, but this did not succeed because
establishment of the sown species was poor; even those that germinated
were grazed off before seed set or were burnt during the dry season.
Milk collection centers associated with the grazing reserves did not
attract the nomads to settle in the reserves. The prices they offered were
too low and did not rise to preserve the household income in times of
reduced supply as the traditional markets do (Waters-Bayer 1985).
Moreover, the sale of milk by itself would make it impossible to sell cooked
Case Studies 31
millet and a sorghum which is traditionally sold in conjunction with it, thus
adding value to the cereal. The intended settlers are agropastoralists, but
there was no provision for cultivation in the early legislation that gave rise
to the grazing reserves. Away from the grazing reserves, land is typically
owned collectively by communities of croppers and is governed by
complex traditions. Pastoralists, who stem from different ethnic groups,
frequently have difficulty obtaining land, especially when the area is
intensively cultivated. Yet the agropastoralists preferred to settle close to
the farming communities where, as noted above, they have access to
amenities and markets for sales of livestock products and purchases of
household food and goods, and where there are more crop residues.
Even if the grazing reserves are individually successful, they can never
serve more than a minority of the cattle owners. The national herd is
estimated at from 10 to 13 million head, owned by about 2% of the people
of Nigeria. That number of cattle requires from 30 to 40 million ha of
grazing land, over a third of the whole country. There is no way that such
a large proportion of the country will be allocated to grazing reserves. The
earlier livestock improvement strategies thus fell short of responding to the
needs and aspirations of the livestock producers.
FODDER BANKS: A LOW-INPUTSUPPLEMENTARY FEED INNOVATION
The following objectives, listed by Kaufmann and Blench (1990), must

be met by any forage development effort if it is to succeed in improving
ruminant livestock productivity:
The innovation must be a low-cost, low-input technology that can be

readily replicated.
Its benefits must be apparent to both agropastoralists (livestock owners)

and croppers (land owners) so that it can be integrated into the
prevailing land-use systems.
Itmust be accepted by, and appropriate to the means of, the extension
services.
ILCA took a farming systems approach and established the necessary

rapport with the principal participants so as to understand their livestock
production systems and aspirations, and then set out to derive
technologies that suited the producers' resources. Many of these aspects

had been overlooked in previous developmental efforts. Three research
sites were selected (Kurmin Biri, Abet, and Ganawuri) because they
represented three different production situations. They were, respectively,
a government-assisted settlement of agropastoralists in a grazing reserve,
a spontaneous settlement of agropastoralists among croppers, and
cattle-owning, crop-farming communities.
In comparison to nomadism and transhumance systems, settled

agropastoralists have limited grazing resources. Their animals lose
condition in the dry season and some that become too weak to last until
the next rainy season have to be sold at very low prices. In spite of this,
settled agropastoralists do not have any tradition of growing forages for
their livestock. The following constraints to forage production were
identified by Mohamed-Saleem and Suleiman (1986):
Acute shortage of labor; all available labor is destined to subsistence

cropping.
Lack of agricultural mechanization, including even animal-drawn

implements.
Little security over land rights, which means farmers can make no
long-term commitments to land development.
Communal grazing by nomadic and transhumant pastoralists, who

compete for the available grazing and do not respect ownership of
.
sown pastures unless they are fenced.
Indiscriminate yearly bush burning which impairs the establishment of

sown forages.
In response to these constraints, ILCA formulated a fodder bank

package using forage legumes because they retain better feed-quality
characteristics than grasses at comparable stages of growth. The legume
fodder banks, which are defined as "enclosed areas of concentrated forage
legumes reserved for dry-season supplementary grazing," in point of fact
originated from participant observation and feedback from one livestock
owner who was settled in Kachia grazing reserve, Kurmin Biri. This
pastoralist started to cooperate with ILCA and national scientists in sowing
Caribbean Stylo (Stylosanthes hamata cv. Verano) as a relay crop with his
sorghum. Since intercropping is a common practice in the area, ILCA
wished to test: (1) the compatibility of a forage with a food crop, (2) the
Case Studies 33
nutritive value of sorghum residue mixed with Stylo as compared to residue

from only sorghum, at the farm level, and (3) the reaction of the
pastoralists to mixing a cereal crop with a forage.
It appeared that the pastoralist simply enjoyed watching ILCA and

government staff working his plot, thereby raising his status in his
community, but at the onset of the next growing season, he fenced off the
previous year's cropped area to allow the Caribbean Stylo to regenerate
and shifted his cropping to another site. This action broke an old and
firmly held tradition that cattle owners were not permitted to erect or
maintain fences to protect pasture. Once the possibility of fencing was
established, the following guidelines were formulated for establishing and
managing fodder banks:
Fence an area of about 4 ha.
Prepare the seedbed by confining cattle overnight in the area or by

grazing down for one to two weeks after seed is broadcast, or by
harrowing, if that is possible.
Broadcast the scarified Stylo seed.
Control fast-growing grasses and other weeds by grazing early in the

growing season.
Allow the forage to bulk-up by deferring further grazing until the dry
season.
Graze the pregnant and lactating animals at a stocking rate of about five
animals per hectare, depending on the forage available, for 2.5 h per
day, normally after the cattle return from grazing.
Ensure that sufficient seeds drop and enough stubble remains for
regeneration in the next season.
Two trial fodder banks were established in 1980 for agropastoralists: one
in the vicinity of a crop-farming community at Abet and the other in the
Kachia grazing reserve. The fodder banks were sited near the cattle
owner's homestead for better supervision and protection. The grazing-
reserve management had a tractor which was used to prepare the seed
bed, while the pastoralist, settled outside the grazing reserve, was
persuaded to kraal (confine) his herd on the selected site during the dry
season to allow for the preparation of the seed bed. Bush poles and
branches cut from the vicinity were used to make the boundary fences.
Because the major objective of the fodder bank was to provide protein
supplementation, the agronomic management practices aimed at
concentrating the legume component of the pasture. This called for very
close monitoring of the growth of seedlings by frequent visits of ILCA staff
to the site for interaction with the owners. Daily milk recordings were
taken, and ILCA enumerators also monitored demographic events such as
deaths, sales, purchases, and births of animals. During the cropping
season, all farming operations, as well as inputs and crop yields, were
recorded. In a later trial, the labor inputs to cropping outside and inside
the fodder bank were studied by recording the time taken per unit area for
each operation from tillage, through planting and weeding, to harvest
(Tarawali et al. 1987). FulBe women are not involved in cropping activities
other than' harvest; this fact was taken into account through careful
analysis of intensive case studies (Waters-Bayer 1984).
In the initial stages, ILCA ensured strict compliance with the guidelines
ithad set out for managing fodder banks. At the end of the growing
season, biomass productivity and forage quality were determined. ILCA
enumerators also ensured that when the herd returned from daily grazing,
lactating and pregnant cows were separated out and put on the fodder
bank. They were rejoined to the rest of the herd after 2.5 hours on the
fodder bank. It was estimated that a 4-ha fodder bank could support 15
to 20 cows for up to 180 days in the dry season. This would provide the
animals with the equivalent of feeding 1 kg of cottonseed cake daily
(Otchere 1986). The productivity of the researcher-managed fodder banks
met the technical expectations (Table 2), and the cattle responded to the
feed supplementation. More milk was taken off from the supplemented
cows than from the controls, and their calves were heavier at one year of
age. However, the most significant result was improved viability of calves.
Calves from supplemented cows had a viability of 88% at 365 days,
compared to 67% for calves of nonsupplemented dams (Otchere 1986).
The first two years of implementing the fodder bank concept was a
learning period for ILCA staff, extension officers, and the agropastoralists.
The cost of every input was met by ILCA, and the management was
supervised by ILCA staff; however, the beneficiaries, as well as other
agropastoralists living in the neighborhood , were consulted frequently
(Taylor-Powell and Suleiman 1986). In 1982 eight more fodder banks were
established, but ILCA supplied only seeds to them; all other costs and the
management were the responsibility of the owners.
Case Studies 35
Table 2. Stylo productivity and quality in researcher-managed and researcher-

implemented fodder banks during the 1981-1982 dry season in the subhumid
zone of Nigeria
Months
Location Observations Oct. 1981 Dec. 1981 Feb. 1982 Apr. 1882
Kurmin Biri Total DM, kg/ha 4,191

Proportion as Stylo, % 68.0
Weight of Stylo, kg/ha 2,850
Stylo CP, % 13.0 10.4 9.8 7.9
Stylo CP, kg/ha 370 90
Abet Total DM, kg/ha 4,900

Proportion as Stylo, % 63.0
Weight of Stylo, kg/ha 3,087
Stylo CP, % 12.6 11.3 8.9 7.2
Stylo CP, kg/ha 389 88
VARYING PRODUCER CIRCUMSTANCES AND

TECHNOLOGY CHANGES
As the beneficiaries undertook to pay more of the fodder bank costs,

they began to express their own ideas and preferences. Some of their
decisions were at variance with the researcher's prescriptions:
Grazing of fodder banks early in the growing season was needed to

selectively control the fast-growing grasses which would otherwise
smother the slow growing legumes. This was unacceptable to the
owners, who feared that their animals would be infested by worms from
the dung that was deposited during kraaling for preparing the seed
beds.
Some animals die or have to be sold prematurely at reduced prices

during the dry season when feed quality is very poor. The fodder bank
owners chose to counter this by grazing all their animals, rather than
just a select few, on the fodder banks, even though it meant much less
feed per animal. This also happened to be less demanding on labor
since it did not require tedious daily separation. The grazing frequency
and duration was also changed to suit the owner's convenience.
Some preferred utilizing the fodder banks early in the dry season when
the feed quality was good, while others reserved it till the latter part of
the dry season as a last resort. Some owners believed that by grazing
fodder banks early, towards the end of the wet season, when the feed
quality was best, the animals attained a better condition to withstand the
stress of the dry season. Ex-post financial analysis confirmed that this
was a wise policy, especially in view of the owners' short-time horizons
and high risk aversion. The ex-ante analysis had not included the
forced sales, which had not been reported in the literature and were
unknown to the experts, who were for the most part extrapolating their
advice from experience with subhumid conditions in the semiarid zones.
Some agropastoralists who wanted fodder banks were discouraged

because of the prevailing land- tenure systems which made it difficult for
them to acquire large enough land areas for meaningful fodder banks.
The insecurity of land-use rights also made them reluctant to invest in
land except for short-term benefits such as cropping.
The inability to sustain legume dominance after a number of years

frustrated some of the owners as the benefit to their animals began to
decline accordingly. The falling legume composition resulted either
from overgrazing or from invasion of nitrophilous grasses in response
to the accrual of nitrogen, fixed by legume-associated rhizobia in the
soil.
Feedback from the participants stimulated the researchers to initiate

remedial research. Alternative land preparation methods were substituted
for kraaling so that the owners could agree to graze the fodder banks early
in the wet season without fear of worm infestations. The owners were also
happy to keep all the manure for their crop fields rather than to devote it
to forage production. Searches for other productive, disease-tolerant and
more competitive legume species were undertaken to suit different
ecological niches within the subhumid zone. Nitrogen-demanding cereal
crops were introduced periodically to alternate with years of legume growth
to take advantage of and reduce the soil nitrogen levels. This resulted in
compatible pasture-based farming possibilities.
There were significant increases of grain and fodder yield for maize
planted after one to three years of Stylo growth, as compared to
continuously cropped fields and even after bush fallow (Table 3). The
increase in maize yield is attributed to the increased soil nitrogen and the
better physical properties acquired by the soil due to the Stylo
(Mohamed-Saleem and Otsyina 1986). Similar benefits were demonstrated
Case Studies 37
with millet (ILCA 1990). These results were particularly interesting to crop
farmers who were depending on purchased fertilizers to sustain yields.
Table 3. Estimated level of N utilization in maize production from soil with different
management histories at Kurmin Biri
Maize Amount of N (kg/ha)

History of land use yield with required by cropped soil
no N applied for equivalent yields
Cropped for 3 years 461
Uncropped for many years 1,275 30
Stvlosanthes hamata for 2 years 1,329 32
S. hamata for 3 years 2,507 90
S. guianensis for 1 year 1,643 44
S. guianensis for 2 years 2,696 110
In the intensively farmed areas, it was difficult to secure 4-ha blocks for
fodder banks for cattle. However, goats are the major livestock owned by
smallholders. With these farmers, mini-fodder banks of less than 0.25 ha
were established to restore cropped soils and cater to the goats.
There is no doubt that the users are in the best position to expose
limitations in new technology. But premature exposure of a technology to
potential beneficiaries raises the risks that the target group will lose
confidence in the technology and the researcher. The timing of on-farm
testing of new technology needs careful consideration. Sensitivity and a
readiness to return to the drawing board with the beneficiaries' ideas are
essential.
CHALLENGES IN IMPLEMENTING INNOVATIONS
Credit from state lending associations (registered credit cooperatives),

as agents of the National Agricultural and Cooperative Bank, became
available in 1984. This was intended to make it easy for those who wanted
to establish fodder banks. The extension responsibilities were gradually

taken up by the National Livestock Projects Division (NLPD) of the Federal
Livestock Department of the Federal Ministry of Agriculture. The Division
undertook the management of a World Bank-assisted livestock
development project, and this project incorporated a target of establishing
2,000 fodder banks by 1990. This would have meant establishing 400-500
fodder banks per year beginning in 1986.
Although producer demand was high, the extension agencies were

unable to meet the targets. Unlike the crop sector, very few extension
messages had been available to the livestock sector and, consequently,
the fodder bank concept was promptly given country-wide coverage. Staff
with varied backgrounds in biological and social sciences (ranging from
ordinary diplomas to post-graduate degrees) were recruited and involved
in the task of popularizing fodder banks. Crash training programs were
arranged to familiarize the staff with the principles of fodder banks, but in
the field there were large differences in staff performance, commitment,
and understanding of the technology.
The extension message was also complicated by the combination of the

originally simple concepts of fodder banks with other innovations and
issues (such as smallholder cattle fattening, model farms and settlement
rights), which were being tackled simultaneously by the same extension
agents.
Despite a firm commitment, the genuine interest of all concerned and the
availability of external financing, there were difficulties in securing inputs at
the appropriate times. Fencing materials, seeds and fertilizers were not
available in the required quantities. This was not surprising, given the
magnitude of the task and the attempt to by-pass the private sector in
supplying the inputs to rural areas. At the recommended rate of 10 kg/ha
of Caribbean Stylo, 20,000 kg of seed were required annually to establish
500 fodder banks of 4 ha each. In trying to reach this target, seeds were
bought by the government from all sources without adequate quality
control, and this compounded the problem by requiring a higher seeding
rate. Many of the fodder-bank owners insisted on having more secure
metal fence posts, and this increased procurement problems because they
had to be imported. Because of governmental regulations, procurement
had to follow elaborate procedures not conducive to timely deliveries at
remote field sites. Procurement and delivery were also complicated by
differences in priorities among the various functional units in the
organization because fodder-bank development was only one of many
activities of the division.
Case Studies 39
IMPACT ANALYSIS
Field situation
Eleven years after the first trial fodder bank was established in Nigeria,
was time to start asking how far the innovation had spread and what
it
effect it had had on the livestock of the target beneficiaries.
At least two major phases can be distinguished in the development and

extension of the fodder bank innovation. The spread of fodder banks in
the first, or development, phase should be credited to ILCA and
cooperating national agencies which facilitated their establishment, directly
as well as indirectly. During this phase, ILCA also helped establish fodder
banks in Cameroon and Mali with the support of responsible ministry
officials in those countries. The spread of fodder banks after 1986 can be
taken as the second, or extension, phase, which was entirely the
responsibility of the NLPD. Ministries and agencies in other countries were
also involved, such as in Benin, northern Cameroon, and as far east as the
Central Republic.
In Nigeria, Caribbean Stylo, and to a much lesser extent Cook Stylo

(Stylosanthes guianensis Cook), were the primary species used for fodder
banks. Land suitability assessments were carried out for Caribbean Stylo-
based fodder banks in- 10 West countries using the FA02 Agroecological
Zones Model (Kassam et al. 1988).
This revealed considerable potential for the spread of fodder banks in

those countries (Fig 3). Even a small proportion of the suitable land under
Stylo (e.g., 10%) could contribute significant amounts of good-quality
fodder to the national feed budgets of these countries (Figs. 4a, 4b).
Legume forages such as Stylosanthes could increase crop yields through
their positive effects on soil. As pressure increases for arable land, forage
legumes are expected to play a more significant role in low external input
cropping systems in conserving soil and sustaining grain production. Yet
the number of fodder banks in Nigeria has not increased as much as had
been hoped, and the prospects for their adoption in other countries remain
largely unexplored.
2 United Nations Food and Agriculture Organization.

Fig. 3. Suitable land area for Caribbean Stylo in the subhumid zone
of West Africa
Fig. 4a. Potential Caribbean Stylo productivity in the subhumid zone

of some West countries (assuming that 25% of the area
would be cultivated using a low level of inputs).
Case Studies 41
Fig. 4b. Potential Verano Stylosanthes productivity in the subhumid

zone of some West countries (assuming that 25% of the
area would be cultivated using high a level of inputs)
As explained earlier, ILCA worked mainly with FulBe agropastoralists and

confined its activities to a 50-km radius, in addition to conducting
on-station research at Kachia grazing reserve. ILCA's involvement meant
that guidelines were closely adhered to in the first phase, even by the
mixed farmers that adopted fodder banks. Because ILCA had paid for the
initial inputs to encourage adoption of the trial fodder banks and because
some of the participants (particularly those in the grazing reserves) had

been exposed to the use of tractors in previous government projects, the
new participants expected similar support. Older participants expected the
subsidies to continue, and the new ones expected to be given their share
of the costs of establishing fodder banks, including preparation of their
crop land by tractors. When extension was taken over by the NLPD,
during the second phase, there was a dramatic change in location, target
group, and management of fodder banks. They were established outside
the subhumid zone using the same legume species recommended for the
subhumid zone. More mixed farmers from different ethnic groups and
clans were recruited. The sizes of the fodder banks varied from the
recommended 4 ha because securing an area of that size in one block
became difficult, especially where cropping intensity was high. The
extension officers did not have the required experience and were less
committed to the tasks expected of them.
Fodder banks were originally designed for settled FulBe agropastoralists,

but mixed farmers from other ethnic groups became the largest
beneficiaries after the credit scheme was introduced (Fig. 5). Non-FulBe
adopters had easier access to extension services not only because there
were more extension staff from their own communities but also because
they had better land rights to present as credit collateral than the FulBe.
The total number of fodder banks in 1990 was 530, as against the 2,000
planned. As indicated above, there are several reasons for this. Fodder-
bank owners who have been extensively interviewed were of the opinion
that lack of extension was the most important reason. Another important
reason was inappropriate land tenure. FulBe agropastoralists, having
larger herds, found the standard fodder banks too small and, therefore,
thought that the cost of the fodder bank was too high for grazing and
inadequate to feed their whole herd. The extension staff rated
inappropriate land tenure and inadequate procurement and delivery of
inputs as the primary reasons for slow adoption. According to the fodder
bank owners, the main source of information about the fodder bank was
the Cattle Breeders Association, and not the government extension system.
This left them short of technical advice.
Nonadopters of fodder banks reported that secure land rights were a

prerequisite for any long-term investment, especially in fast-expanding
cropping areas where it was most difficult to secure land. Farmers were
increasingly likely to claim back or encroach on land that was formerly
avoided as being too marginal for cropping.
Case Studies 43
Fig. 5. Fodder bank adoption in Nigeria (by ethnic group)
Source: 1989 survey by the National Livestock Division, Nigeria.
For some, the motivation to adopt lay in the hope that by being
cooperative with government agents they would gain access to veterinary
services, future government loans, and preferential treatment for inputs
such as fertilizer and supplementary feeds. In fact, fodder banks that
belonged to these farmers were often neglected, but the owner was still
regarded as a participant by the extension agency, which was reluctant to
reduce its apparent impact. This seemed particularly true with the
adopters in the government-assisted grazing reserves where FulBe
pastoralists have been unwilling to shoulder many of the establishment and
management tasks required by the fodder bank. Those farmers who have
seen the value of forage legumes hoped for high proportions of the sown
legume in the sward at the end of the rainy season. However, maintaining
a high proportion of legumes in the fodder banks becomes more difficult
over time, particularly if the sown legume is a good fixer of nitrogen. As
nitrogen accrues in the soil, the fodder bank is invaded by nitrophilous
grasses. This can be counteracted by growing crops, but it requires skilful
management to establish a spatial and temporal crop-forage balance.
Researchers' view
The success or failure of fodder banks will ultimately depend on the

degree to which the innovation coincides with the prospective owners'
objectives. Some of the objectives are: (1) increased cow productivity and
calf survival, (2) increased total herd welfare and avoidance of distress
sales, (3) increased crop yields, and (4) increased security of tenure.
The priorities of individuals will determine how the fodder banks are
used. Recent developments in the economy of Nigeria have had a
significant impact on both the costs and availability of inputs as well as on
producer prices, and thus on the profitability of fodder banks. Until 1985,
the exchange rate was fixed by the government at an approximately US
dollar/Naira parity, which made imported technical inputs readily available
on the market. After the introduction of the structural adjustment program
in 1986, the Naira fell to roughly 20% of its former value; thus, imported
inputs such as steel posts and galvanized sheep-proof fencing became
scarce.
In 1985, a fodder bank cost 2,502 Naira to establish and 498 Naira
annually to maintain (Table 4). By 1989 this had risen to 6,044 Naira, with
1,466 Naira in recurrent costs if inputs were purchased on the open
market, as compared with 3,828 Naira and 708 Naira for fodder banks
using inputs supplied by the NLPD.
A first test of the financial merit of the investment in fodder banks is

whether it is likely to provide a protein feed supplement cheaper than the
most popular alternative feed (in this case, cottonseed cake). Table 5
compares costs based on 1984-1985 and 1989 prices with the average dry
matter production of existing fodder banks, assuming a constant 9% crude
protein content for the Stylo.
Although a fodder bank requires considerable capital investment, the

saving in recurrent costs, as compared to purchasing feed, more than
compensates. Assuming no inflation, the fodder bank was estimated to
Case Studies 45
have a positive present value at a 15% rate of return even if the price of
cottonseed cake dropped to 0.25 Naira/kg. Moreover, a major attraction
of fodder banks is the relative freedom from the effects of cost-inflation
because the recurrent costs are low. With only 10% inflation (actual
inflation was around 30% during 1984-1985 and over 50% in 1988-1989),
cottonseed cake would have to sell at less than 0.20 Naira/kg before the
internal rate of return in favor of fodder banks dropped below 15 percent.
If other factors associated with cottonseed cake (such as the unreliability
of supplies and storage losses) are considered, fodder banks become even
more attractive.
Table 4. Costs of fodder banks compared with purchased cottonseed cake
Cottonseed cake
Fodder bank, 4 ha supplying equal
amount of protein
Total dry matter, at 4,000 kg/ha 16,000 kg
Grazable dry matter, 50% 8,000 kg
Protein (9% with fodder bank,

30% with cottonseed cake) 720 kg 720 kg
Required dry matter 2,400 kg
Capital cost, Nairas
1985 2,502 0
19891 6,044 0
19892 3,828
Recurrent cost, Nairas
1985 498 840
19891 1,466 2,064
19892 708
I Open market price.

2 Inputs supplied by the National Livestock Projects Division.
Source: ILCA records.

To establish just how viable fodder banks really are as an investment,

and to look at the internal rate of return over a longer-time scale, these
projected costs were integrated with recorded data on animal productivity.
A herd model for bio-economic analysis (Kaufmann et al. 1990) was used
to simulate pastoral herd dynamics, with and without supplementation.
The model accepted data on factors such as herd structure, animal
productivity, offtake, forage production, and feed supplementation, with all
relevant costs. The model then calculated herd structures and offtake over
ten years and provided financial analysis. The model, written in Lotus 123,
provides for simultaneous runs, with and without the intervention being
appraised. It is easy to run sensitivity analysis for the effect of changes in
technical inputs (such as feed) on output parameters (such as kg of milk
produced) and financial criteria (such as return on capital invested). Fig.
6 shows potential build-up of the herd over ten years with fodder bank
supplementation, using a constant offtake rate of the traditional 9% per
year.
Table 5. Principal economic returns to agropastoral fodder bank owners over a period of
10 years in 1989, all figures in Naira (except for internal rate of return)
Net Internal 10th year 10th year 10th year

Factors present rate of herd value herd value incremental
analyzed value return without with net revenue
@ 20% % intervention intervention
Improved herd
productivity
alone
1985 140.8 20.6 15,234 26,862 2,509

19891 13,868 49 54,293 98,692 12,141
19892 1,331 22.3 49,907 90,833 4,950
Plus reduced
forced sales
19891 19,987 65.1 54,293 98,692 14,329

19892 7,455 33.9 49,932 90,833 7,138
Plus increased
crop yields
19891 21,843 64.9 54,293 98,692 15,735

19892 9,312 36.0 49,932 90,833 8,544
I Open market price.

2
Inputs supplied by the extension agency (NLPD).
Case Studies 47
120
100
1 2 4 a e 7 to
Simulation years
Fig. 6. Model herd structure with fodder bank intervention
A comparison of exits from supplemented and nonsupplemented herds

(Fig. 7) shows approximately similar levels of sales and deaths, fewer
forced sales of sick animals, and more transfers out. This suggests that
pastoralists will be inclined to use the increased productivity, brought about
by investment in fodder banks, to enable adult sons to establish herds of
their own, rather than to increase sales and cash revenue. Under
prevailing management conditions, with free access to grazing, this
appears to be a sound financial policy for the agropastoralists. The
opportunity cost of retaining extra animals is minimal, and there are few
alternative investments available to the agropastoralists that would be as
resistant to the present high rate of inflation in Nigeria.
Animal Production Systems: Global Workshop
Fig. 7. Herd exits with and without fodder banks
The results shown in Table 5 are based on 1989 market prices, with a
strong element of conservatism in the price of milk so as to ensure that the
model applies to rural producers, who are too distant from towns to sell
milk at the best prices. The model suggests that, even with these
assumptions, the internal rate of return would be above 22 percent. The
federal government's Structural Adjustment Policy has worked in favor of
cattle producers because imports of animals and dairy products have risen
as the exchange rate of the Naira has fallen. This suggests that fodder
banks are reasonable investments, even when considering only the primary
objective of improving cattle productivity. Moreover, there are significant
Case Studies 49
add-on benefits (Table 5), such as avoidance of distress sales and

increased crop yields. Owners without fodder banks usually have up to
two salvage sales per year; the incidence of this among fodder bank
owners was almost negligible. That this is an effect of feed quality and
occurs even when plenty of grazing is available has been borne out by
on-station experience.
FUTURE PROSPECTS
To a region facing livestock feed deficits, the fodder bank concept offers
a self-help option for producing good-quality supplementary feed on a
sustainable basis. Because previous feed improvement strategies were
ineffective, great care was taken in the research reported above to tailor
fodder banks appropriately to producer circumstances. Inputs and
management requirements for fodder banks are simple and within the
reach of the targeted beneficiaries. The potential benefits to livestock, and
thus to household income, can be substantial.
There is ample scope for the adoption of fodder banks in the subhumid
zone. Assuming that 4 ha is sufficient for a typical 50-head herd, this
would involve a total of 360,000 hectares to cater to the 4.5 million head
of cattle estimated to be in the subhumid zone of Nigeria. If this were to
happen, simulation models have predicted an increase of 51 million kg of
beef and 22 million liters of milk at a 100% adoption level. Although this
suggests that there can be significant benefits, there are still many doubts
about the future prospects for fodder banks. Do they require too much
extension support or too many inputs? Is the scope of this concept being
over-extended?
Some observers suggest that since the zone is relatively underutilized

for crops, there is plenty of range, fallow, and idle lands for grazing; they
argue that, even with the threat of seasonal feed deficits, livestock owners
do not have great incentive to grow forages. It is further argued that in
situations of intensive cultivation there are plenty of crop residues available
in the dry season.
Scientists, who have longer-term expectations than businessmen, see

the prospects for fodder banks in the subhumid zone differently. At the
rate that cultivation is increasing, livestock mobility will be severely
restricted and agropastoralists will have to become sedentary. Farmers in
the zone are also investing in cattle, and they are not in the habit of
moving in search of feed for livestock. These trends will create greater
incentives for finding alternative renewable feed sources. Equally important
as cultivation intensity are the issues of land degradation and crop-free rest
periods. These have been found to be critical even at the highest input
levels involving the use of chemical fertilizers (Young and Wright 1984).
Tropical soils, particularly in the subhumid zones, are fragile, and rest
periods seem to be essential for recouping organic matter to sustain
productivity levels. It is in this context that fast-growing and strong-rooting
forage legumes will be important in the near future.
There will be more and more farms in the subhumid zone where crops
are planted on one side of the farm while the sown legumes are
regenerating on the other after the previous year(s) of cropping. This trend
needs to, be encouraged to improve farmer-agropastoral relations.
Potential land improvement may be a good incentive for the farmers to
release land to agropastoralists for fodder banks. Cultivation of part of the
fodder bank by the farmer each year will raise crop yields to compensate
for leasing the land. There are opportunities for beneficial compromises
that could meet the aspirations of both parties.
Today, there is a greater appreciation of the benefits of integrated land-

use systems. Whatever the operational merits, separate developmental
policies for the crop and livestock sectors seem inappropriate in West
Africa. The National Livestock Projects Division is now seriously seeking
the support of the Agricultural (crop) Development Project extension staff.
However, inappropriate manipulations of the technology may lead to
disastrous results. ILCA had some experience of this in the process of
promoting fodder banks. For example, the subhumid zone is not entirely
uniform. The growing season varies from six to eight months. There are
very few forage legume species tested in Nigeria, and Caribbean Stylo is
the only one with seed available in commercial quantities and with a
tolerance of the disease anthracnose. The extension agencies therefore
used Caribbean Stylo everywhere in popularizing fodder banks throughout
the country, but this brought major difficulties because it is not appropriate
to all locations.
In the wetter areas, the anthracnose challenge is so high that it

overcomes the Stylo tolerance. In drier areas, the length of the growing
period can be too short to support the generation cycle of this variety,
creating problems with seed-set and low biomass production. These
considerations were disregarded in the urgency to extend the fodder- bank
concept widely. The criteria for targeting the fodder-bank technology was
Case Studies 51
simply the recognition of a feed-constraint problem. Therefore, wherever

there were cattle, the agency recommended the same fodder bank
package.
The diagnostic techniques of farming systems analysis tend to focus on

remedies for present land-use systems and recognized farm-level
constraints, and pay too little attention to biophysical limitations. This
encourages extrapolation of technology, verified in a particular location,
without giving sufficient consideration to production potentials in relation
to land quality at new sites, even within the same recommendation domain.
ILCA believes that geographic information and land evaluation techniques
should be called upon to assist in site selection and in delineating
appropriate areas for technology transfer.
Development agencies are rightly impatient, but in view of the normal

research station-to-farm cycle of 15 to 20 years, the fodder bank concept
is still relatively new in West Africa. In fact, growing forages for livestock
feeding is in itself a completely new proposition, and the West farming
community still has to come to terms with it. The different responses by
the Nigerian farming community to the fodder bank concept is, therefore,
not only understandable, but also encouraging, because it demonstrates
serious interest in overcoming the weakness and making the innovation
more adaptable and adoptable. Good loan repayment records, willingness
of the successful owners to extend their fodder banks, and the growing
use of forages for smallholder dairy development indicate that fodder
banks will playa significant role in the longer term.
The problems and opportunities for fodder banks now fall in the realms
of adaptation and adoption. ILCA, as an international center, has no
comparative advantage in this. Therefore, it is encouraging that, since the
research started, the National Animal Production Research Institute has set
up a livestock systems research program which is interested in fodder
banks, and the Institute of Agricultural Research is involved in research on
the forage-crop interface with fodder bank owners. Both institutes are part
of Ahmadu Bello University and the research is largely funded by the
National Livestock Projects Division.
There will be a continuing need for scientists: (1) to search for better
germplasm for the different agroclimatic niches within and outside the
subhumid zone, (2) to find alternative feed improvement techniques suited
to the different production circumstances, and (3) to develop increasing
capability to generate, analyze, and model production processes for
situations where a number of conflicting goals have to be reconciled and
the choice of certain development paths becomes dependent on the

relative value attached to each of these goals. This is where interactive
contact among the interest groups (government, development agencies
and beneficiaries) will have to be strong in order to promote fodder banks
as an innovation requiring low external inputs for ruminant feed
supplementation in West Africa.
LITERATURE CITED
BOURN, D.; MILLIGAN, K. 1983. The dynamics of cattle distribution in the

Nigerian subhumid zone. Addis Ababa, Ethiopia. International
Livestock Centre for Africa (ILCA), Subhumid Program, Kaduna. 23 p.
ILCA (INTERNATIONAL LIVESTOCK CENTRE FOR AFRICA). 1990. Annual

Program Report. Addis Ababa, Ethiopia, ILCA. 207 p.
KASSAM, A.H.; VAN VELTHUZIEN, H.T.; MOHAMED-SALEEM, M.A. 1988.

Fodder productivity assessment for West Africa. Report 1: Land
suitability assessment for rainfed production of Verano Stylo. Addis
Ababa, Ethiopia, ILCA/FAO, February 1988. 59 p.
KAUFMANN, R. VON; BLENCH, R.M. 1990. Increasing the use of draught

animal power in the subhumid zone of Nigeria. In Workshop on draft
animal power research and development (1., 1990, Zara, Nigeria).
Proceedings. J.O. Cefu, E.O. Otchere (Eds.). Addis Ababa, Ethiopia,
ILCA. p. 17-31.
KAUFMANN, R. VON; McINTIRE, J.; ITTY, P. 1990. ILCA bio-economic

model for microcomputer (IBIEHM). Addis Ababa, Ethiopia. ILCA. 65
p.
MANI, R.I.; KAUFMANN, R. VON; OTCHERE, E.O. 1993. The productivity

of Bunaji (White Fulani) cattle and the effect of forage legume
supplementation in improving their productivity in agropastoral herds in
subhumid zone of Nigeria. Addis Ababa, Ethiopia. ILCA Monograph
9301309. 93 p.
Case Studies 53
MOHAMED-SALEEM, M.A. 1984. The establishment and management of

fodder banks. In ILCA/NAPRI Symposium on livestock production in
the subhumid zone of Nigeria. Proceedings. R. von Kaufmann, S.
Chater, R. Blench (Eds.). Addis Ababa, Ethiopia, ILCA. p. 226-250.
MOHAMED-SALEEM, M.A.; OTSYINA, R.M. 1986. Grain yield of maize

and nitrogen contribution following Stylosanthes pasture in the Nigeria
subhumid zone. Experimental Agriculture 22:207-214.
MOHAMED-SALEEM, M.A.; SULEIMAN, H. 1986. Fodder banks: Dry-

season feed supplementation for traditionally managed cattle in the
subhumid zone. World Animal Review 59:11-17.
OTCHERE, E.O. 1986. The effects of supplementing feeding of

traditionally managed Bunaji cows. In Livestock systems research in
Nigeria's subhumid zone. R. von Kaufmann, S. Chater, R. Blench
(Eds.). ILCA/NAPRI Symposium (2., 29 Oct.- 2 Nov., 1984, Kaduna,
Nigeria). Proceedings. Addis Ababa, Ethiopia, ILCA. p. 226-250.
POWELL, J.M. 1986. Contribution of fractioned crop residues to dry

season livestock feed resources in the Abet plains, Central Nigeria.
ILCA Newsletter 4, 4:5-7.
SYNGE, B. 1980. Factors limiting cattle productivity in highland areas of

Nigeria. Easter Bush, Midlothian, Scotland, Center for Tropical
Veterinary Medicine. Report. 50 p.
TARAWALI, G.; MOHAMED-SALEEM, M.A.; KAUFMANN, R. VON. 1987.

Comparison of labour requirements for cropping maize on land
previously used for Stylosanthes fodder banks and on adjacent fallows
in the subhumid zone of Nigeria. ILCA Bulletin No. 26. p. 36-39.
TAYLOR-POWELL, E.; SULEIMAN, H. 1986. Extension implementation of

ILCA interventions: Dry-season cow season cow supplementation with
fodder banks. In Livestock systems research in Nigeria's subhumid
zone. R. von Kaufmann, S. Chater, R. Blench (Eds.). ILCA/NAPRI
Symposium (2., 29 Oct.- 2 Nov., 1984, Kaduna, Nigeria). Proceedings.
Addis Ababa, Ethiopia, ILCA. p. 382-390.
WATERS-BAYER, A.; BAYER, W. 1984. The Fulani of Abet: Research for

development of agropastoralism in the West African savannah. Revised
Version. Entwicklung and landlicher Raum 18(5):16-20.
WATERS-BAYER, A. 1985. Dairying by settled Fulani women in Central

Nigeria and some implications for dairy development. ODI Pastoral
Development Network (UK), Paper No. 20b. 24 p.
YOUNG, A.; WRIGHT, A.C.S. 1984. Rest period requirements of tropical

and sub-tropical soils under annual crops. In Report on the second
FAO/UNFPA expert consultation on land resources for populations of
the future. Rome, Food and Agriculture Organization of the United
Nations. p. 197-221.
A CASE STUDY ON DUAL-PURPOSE GOAT
TECHNOLOGY DEVELOPMENT
IN WESTERN KENYA'
Fanny B. Nyaribo2, J.F.M. Onim2,

P.P. Semenye and B.
Mwandotto3.4
INTRODUCTION
The Small-Ruminant Collaborative Research Support Program (SR-CRSP)

is a research and training program currently operating in Indonesia, Kenya,
and Morocco. Major funding for the SR-CRSP comes from the United
States Agency for International Development (USAID), with matching
contributions from participating U.S. universities, research institutions, and
host-country government agencies. Institutions collaborating in Kenya are
Washington State and Colorado State Universities in charge of animal
health and vaccine development research; Texas A&M University, breeding
and systems analysis; University of Missouri, sociology; Winrock
International Institute for Agricultural Development, economics and
production systems, which includes feed resources, and animal nutrition
and management research; and the Kenya Agricultural Research Institute
(KAKI) which has taken over from the Ministry of Livestock Development
as the host-country institution.
y Funding to conduct this case study was provided by the Agriculture, Food and Nutrition
Sciences Division of the International Development Research Centre.
2 Winrock International Institute for Agricultural Development, SR-CRSP, P.O. Box 252,
Maseno, Kenya. F.B. Nyaribo is presently at Washington State University, Pullman, WA,
U.S.A.
3 Texas A & M University. SR-CRSP, P. O. Box 58137, Nairobi, Kenya.
4 The authors are indebted to F. Rurangirwa (Washington State University) and P. Rwambo
(Colorado State University) for helpful comments and suggestions.
Since 1980, the Kenya SR-CRSP has been conducting on-station and on-
farm dual-purpose goat (DPG) research on 147 smallholder farms in
Kakamega, Siaya and Kisumu districts. In addition, on-farm research in
Kimuso was started in 1988. However, for logistical reasons some of the
research is conducted at sites away from western Kenya. This includes
animal health and vaccine development which are carried out at the
Veterinary Research Laboratory at Kabete in the outskirts of Nairobi and
breeding and systems analysis which are carried out in Naivasha, about
100 km northwest of Nairobi. The on-farm testing of DPGs and the
component research activities involving nutrition and management, feed
resources, economics, and sociology projects are conducted at Maseno
Veterinary Farm, about 425 km northwest of the city of Nairobi.
The prime objectives of the program are to develop dual-purpose (milk

and meat) goat production systems suitable for the densely populated
medium- to high-potential zones of western Kenya as a means of improv-
ing the welfare of resource-poor farmers in the region; to increase local
research capacity through on-the-job training as well as formal training at
the postgraduate level; and to strengthen local research institutions.
THE KENYA DUAL-PURPOSE GOAT
Development of the Kenya dual-purpose goat (here referred to as the

KDPG) is based on the concept of a composite breed with equal
proportions of the East African, Galla, Toggenburg, and Anglo Nubian
breeds. This composite breed is expected to weigh 35 kg, produce 2 kg
of milk at peak lactation with one kid, and have at least 120 d of lactation
(Mwandotto and Taylor 1991). Since only a few composites have been
produced, animals used so far for on-farm testing are first crosses (F1
generation) between indigenous (East African and Galla) and exotic
(Toggenburg and Anglo Nubian) breeds. The F1s are the intermediates in
the process of breeding a 4-way composite. The development of a breed
is a long-term process and the scientists could not wait for the final KDPG
inorder to start component technology development to support the new
breed. In view of this, a decision was made to carry out on-farm testing
with the few available F1 crosses. DPG production data presented in this
paper relate, therefore, to the F, crosses used in on-farm trials in western
Kenya.
Case Studies 57
The west Kenya region was deemed suitable for DPG research as, over
the years, cattle numbers have declined and the human population has
continued to increase at a high rate (3.9% per year for the Kakamega
district) according to Lihanda (1978). With diminishing land/labor ratios,
it was hypothesized that the DPG might be a good alternative to a dairy
cow; some of the advantages that DPGs have over cattle, particularly in
small farms, are the following:
DPGs are prolific, with high twinning rates and kidding intervals of less
than a year; thus they have a high offtake rate. A farmer owning three
to five does can expect eight or more offspring for sale or slaughter.
Conversely, Zebu cows reared in the same environment have calving
intervals of 14 to 16 months, which translates to .5 calves a year
(Semenye et al. 1989). .
DPGs provide quality food protein via consumption of milk by family

members and augment the household's cash income through the sale
of surplus milk, offspring, and other goat products.
DPGs have low initial and maintenance costs; thus, animal losses may
.be a more acceptable risk to the farmer.
The cost of feed is low because goats graze on marginal land areas and
on available crop residues.
Products come in small amounts that are readily consumed or sold by

the family.
DESCRIPTION OF PROJECT STUDY SITES
On-farm testing of the DPG is being carried out in six villages in Western
and Nyanza provinces in western Kenya. The villages are: Hamisi,
Kaimosi, Masumbi, Muhanda, Rabuor, and Lela. Masumbi, Lela and
Rabuor lie in the lowlands of the Lake Victoria basin at an altitude range
of 1,296-1,469 m above sea level (Table 1). Average annual rainfall is 1230,
1400, and 1832 mm, respectively. Kaimosi, Hamisi, and Muhanda are
located in the highlands west of the Rift Valley at an altitude range of 1,623
to 1,693 m with a mean annual rainfall of 1,815 mm for Kaimosi and 1,875
mm for both Hamisi and Muhanda. The rainfall has a bimodal distribution,
with long rains lasting from late February to late July. The short rains fall
between September and November. About two-thirds of the rain falls in
the long-rain period.
The soils in the region are of low fertility with some variations, except for
Maseno station, which has moderate to high fertility (Table 1). Onim et al.
(1990) sampled soils on 38 farms in the region and found 92% to be
deficient in phosphorus. Other limiting nutrients were sodium, potassium,
magnesium, nitrogen, and organic carbon. The dominant soil type is deep
red loam with good drainage and pH levels ranging from 4.0 to 6.5. Other
soil limitations include steep slopes with shallow soils not suitable for
cultivation, particularly in Hamisi and Muhanda.
RESEARCH METHODOLOGY
The dual-purpose goat research program follows the farming systems

research (FSR) framework. This approach is appropriate due to the
complex, intensive crop-livestock systems found in the area, the limited
farm resources, the semisubsistence nature of households, and the
interdependence between household production and consumption. The
suitability of FSR in this type of farming environment is well-documented
(Gilbert et al. 1980; Collinson 1982; Harwood 1982).
Table 1. Altitude, rainfall, and soils in project study sites
Mean Annual Soil

Site Province altitude rainfall fertility soil
m mm group limitations
Kaimosi Western 1,693 1,815 Low Steep slopes

Hamisi Western 1,682 1,875 Low Steep slopes
Muhanda Western 1,623 1,875 Low Steep slopes
Rabuor West/Nyanza 1,469 1,832 Low Steep slopes
Lela Nyanza 1,440 1,400 Variable Shallow
Masumbi Nyanza 1,296 1,230 Variable Shallow
Maseno station Nyanza 1,500 1,678 Moder. high Shallow
Adapted from Jaetzold and Schmidt (1982) and Onim et al. (1985b).
Case Studies 59
Following FSR methodology, between 1980 and 1982 the project carried
out baseline surveys to characterize existing farming systems and to
identify technical (biological) and socioeconomic constraints or
opportunities likely to inhibit or enhance the introduction and adoption of
the DPG (Sands 1983; Noble and Nolan 1983). Results from the baseline
studies formed the basis for subsequent design and evaluation of the new
DPG technologies. The following stages have been complied within the
project's research activities.
Stage 1: Selection of collaborating farmers and baseline surveys
Between 1979 and 1980, a research domain was identified and a random
sample of collaborating farmers was selected to participate in research
activities. An initial sample of 80 farmers drawn from Hamisi, Kaimosi,
Masumbi, and Barding villages was randomly selected using a stratified
random sampling technique. The sample was drawn from a nationwide
sample frame of smallholder farmers used by the Central Bureau of
Statistics of the Ministry of Finance and Planning. A detailed description
of sampling procedures is presented in Sands (1983).
Barding village was later dropped from the sample due to problems with
the project's hired enumerator, who developed a bad relationship with
farmers. From November 1980 to October 1981, a team of project
scientists were involved in the small-farm systems baseline survey. The
team included an agricultural economist, an agronomist, a veterinarian, an
animal scientist, and a sociologist. In addition, a cyclical monitoring survey
was administered every 28 days to collect data on seasonal changes in
farm activities including livestock inventories, household demography,
status of field crops, household consumption patterns, and so on. Very
early in the program, it was determined that health and nutrition would be
major influencing factors in DPG adoption in the region. From initial
baseline surveys (Sands 1983; Noble and Nolan 1983) the following were
also identified as constraints to DPG adoption:
Seasonal fluctuations in the quantity and quality of feed available.
Limited land size and lack of capital, including cash for investment.
Prejudices against goats and goat products.
Lack of a suitable goat genotype able to produce sufficient milk for both
kid requirements and household consumption.
The first on-station experiment involving dual-purpose goats was initiated

in 1981 (Brown and Nderitu 1982). In 1982 a limited number of goats were
placed on farms for performance evaluation (Brown et al. 1983).
Stage 2: Design and screening of potential technologies
Stage 2 continued with constraint analyses, diagnosis of farmer

circumstances, and studies on social acceptability of goat milk (Boor et al.
1984; Nyaribo et al. 1984; Reynolds et al. 1984), as well as the start of the
evaluation of DPG economic competitiveness under farmer conditions
(Mukhebi et al. 1984). Researcher- managed biological trials were also
expanded both on-station and on-farm, as more animals became available
for research (Shavulimo et al. 1984; Onim et al. 1985). This phase lasted
from 1983 to 1985. This was a period for narrowing down the range of
technological alternatives with increased on-farm testing of forage
production and goat nutrition and management strategies, as well as
increased distribution of research goats to participant farmers.
By March 1984, a total of 112 research DPG does and bucks had been
allocated to collaborating farmers in the three original villages (Khainga et
al. 1984). Before animals were placed on the farms, farmers were brought
to the station for a one-day course in dairy goat management. Farmers
were encouraged to build low-cost sheds, to milk lactating does twice a
day (in the morning and evening), to have controlled suckling by kids, and
to cut and carry locally available feeds, as well as cultivated forages
introduced by the feed resources component. In some instances, if
animals placed on farms showed poor performance or were critically ill
they were withdrawn, and replacement animals were given.
Stage 3: Technology testing and verification
The third phase of the project's research (1986-1990) coincided with the
emergence of a distinct technology package. Project scientists felt that the
technology package, although subject to further refinement, was ready for
testing beyond the original sample of 80 farmers. Three new villages,
Muhanda, Rabuor, and Lela, contained a random sample of 75 farmers,
who were added to on-farm trials, technology monitoring, and evaluation.
The three villages are located in the same agroecological zones as the
original ones.
Case Studies 61
The new sample of farmers was treated differently in a number of ways.

Experience from the baseline SFSS indicated that it was not necessary to
administer a long-term survey to gather data on basic features of existing
farming systems, such as land size, cropping patterns, livestock
enterprises, resource endowments, and level and sources of off-farm
income. It was decided that these data should be collected using a rapid-
rural-appraisal survey technique.
Data collection was carried out by the entire on-farm research team.
With the assistance of project enumerators, each farmer was interviewed
by. a survey subteam with at least two scientists. A simple 12-point
questionnaire, which included drawing a sketch map of the farmers'
landholding(s), was used. The survey took about one-half hour, while map
sketches took an additional half hour. Return visits were made in cases
where a farmer had multiple landholdings or large and/or complex parcels
of land. .
Goat distribution not only entailed the willingness of the farmer to

cooperate with project scientists, but also required an explicit contractual
agreement between the two parties. The contract expiration date was
determined by the anticipated termination of donor funds. It was
anticipated that, by this time, project research activities would be
incorporated into KARI's mainline research activities through a proposed
Small-Ruminant Research Program.
Collaboration with farmers has entailed both verbal and cash contract
agreements (Sidahmed et al. 1985). For example, the feed resources team
leased trial plots on farmers' land for cash, and, after sampling, all forages
and grain were given to the farm family. Maintenance operations such as
fencing, cultivation, and weeding were usually contracted with the farmers'
family and cash payments were made. This type of agreement was
necessary for researcher-managed trials. In the third phase, when the
technology package was tested, goat distribution to participant farmers
rested on a contractual agreement which outlined the responsibilities of
both the farmer and the SR-CRSP. The farmers (who were ideally
supposed to have only two cows) were required to do the following:
To plant a fodder bank, following guidelines from the feed resources

team, prior to placement of DPG does. Seedlings for forage
establishment were given by the project free of charge.
To be willing to participate in farmer-researcher meetings organized by

the project.
To follow project recommendations on health, nutrition, breeding, and

feed resource management.
To sell the first female kid born on-farm back to the project at a price
agreed upon in advance in order for the project to accelerate its
breeding program and to have a continued supply of animals for
multiplication.
To allow project scientists to collect data and any information needed

for evaluation and monitoring of the technological package.
The project, on the other hand, was to meet the following conditions:
To provide all technical advice and assistance with regard to rearing the
DPG, including veterinary care; however, farmers would pay for any
veterinary medicines required.
To provide a sufficient number of bucks for on-farm mating with the

does. A mating fee was to be charged by farmers who agreed to keep
the bucks. Alternatively, farmers could make other payment
arrangements based on cultural practices for breeding livestock in their
area.
To provide each household with a minimum of two does, at least one of

them in-kid.
To provide the farmers with DPG bucks of at least 50% dairy genotype
(either Toggenburg or Anglo Nubian).
To treat and vaccinate the animals before distribution to the farmers to

ensure that they were free of any parasites and/or diseases.
To train farmers on milking and milk hygiene techniques and to issue

certificates of attendance at the end of the one-day course.
Finally, not to provide the farmers with free inputs, particularly veterinary
medicines.
Case Studies 63
RESULTS
Farmers' resource endowments
The overall mean farm size for participant farmers is 1.68 ha with a mean
range of 1.30 to 2.56 ha in Hamisi and Masumbi (Table 2). Average
household size ranges between seven and eight persons, yielding
land/labor ratios of between .37 and .16. Mean total labor per household
is 35.3 labor months, which is available most of the year; however, high
labor demand periods occur in April, June, and August when field crop
operations such as land preparation, weeding, and harvesting are at their
peak. The project's labor supply and utilization studies (Nyaribo et al.
1984; Conelly et al. 1987) have determined that labor does not appear to
pose a problem at current livestock numbers, including DPGs.
Table 2. Mean farm resources per farm in project study sites
Mean farm Labor Cash inflowa Cash outflowa

Village n size supply Ksh/yrb Ksh/yrb
ha man-months
Hamisi 23 1.30 38.30 5,465 10,164

Kaimosi 23 1.81 47.20 3,375 10,309
Masumbi 24 2.56 24.00 3,756 5,503
Lela 11 1.85 34.77 na na
Rabuor 35 1.08 29.50 na na
Muhanda 28 1.50 37.50 na na
Total 144
Simple average 1.68 35.21 4198 8659
' Mukhebi et al. (1986).

° 1 U.S. dollar = 16 Ksh, in 1986.
In the Hamisi, Kaimosi, and Masumbi villages, household revenue and

expenditure records were kept for one year to determine the level and
seasonality of cash flow, sources of off-farm income, and degree of farmer
participation in the market (Mukhebi et al. 1986). While cash inflow does
not reflect total farm income, some inferences can be drawn regarding the
availability of cash reserves for household investment; the study revealed
that 66% of farm cash receipts come from external sources such as off-
farm employment, remittances, own businesses, and loans. On an
average, 59% of the households' expenditures were on food items,
indicating that western Kenya smallholders are not pure subsistence

farmers. There may have been some underreporting of cash receipts and,
in some cases, exaggerated expenditures. Assuming that household
expenditures were underreported by at least 50%, it can be inferred that
there is little or no cash saving for any significant investment by the
household. Furthermore, farmers in the sample have reported no use of
institutional credit such as that offered by the Agricultural Finance
Cooperation or the Cooperative Bank.
Crop and livestock production
Maize, the staple food crop, is grown by virtually all farmers and is
normally intercropped with beans (Table 3). Other important food crops
are beans, cassava, sweet potatoes, sorghum, and finger millet. Cash
crops grown are tea, coffee, sugarcane, cotton, eucalyptus trees, and
french beans. Growing french beans is a relatively new enterprise in parts
of Kakamega district and is rapidly being picked up by farmers; the beans
are primarily grown for export, and their rapid adoption is due to a strong
input supply and marketing support system from a private company
around the Muhanda area.
Livestock play an important role in the household economy, and Zebu

cattle are the most valued and predominant species. A project survey
conducted from 1986 to 1988 (before the distribution of DPGs to the latest
set of participant farmers), revealed that there were an average of 2.90,.77,
.74 head of cattle, sheep, and East African goats, respectively (Table 4).
Role of women in livestock production
Due to the high male migration out of the area, the responsibility of the
day-to-day running of the farm is carried out by female household
members. In Hamisi, Kaimosi, Rabuor and Muhanda, 54, 58, 71, and 86%,
respectively, of the farms are managed by women (Table 5). Other studies
in the area have noted this trend (Lihanda 1978; Noble and Nolan 1983).
Male family members and educated youth have moved away to urban
centers and may visit their rural homes only once a year; this in part
accounts for the high off-farm income in the form of remittances reported
by farmers. Intensification of livestock production (particularly in the
highland areas) involves cutting, carrying, feeding, and watering livestock;
labor input in livestock rearing by female members is as high as 40% of
total required time (Conelly et al. 1987).
Case Studies 65
Table 3. Proportion of farmers growing each food or cash crop
Hamisi, Kaimosi Rabuor, Muhanda

Crop type Masumbia Lelab
(n = 70) (n = 74)
Maize/beans 99 99
Bananas 83 86
Cassava 58 78
Sweet potatoes 50 73
Eucalyptus trees 43 55
Cowpeas 40 30
Sorghum 38 41
Coffee 33 19
Kale 28 -
Tea 19 7
Gum trees 18 -
Finger millet 17 -
Sugarcane 10 50
Cotton 1 -
French beans - 5
a According to the project's 1986 rapid rural appraisal in the villages indicated.
b According to the project's 1988 rapid rural appraisal in the villages indicated.
Table 4. Mean livestock population per farm, heads
Local Total
Village n Cattle goats Sheep TLU a
Hamisi 24 2.70 0.30 0.30 2.28

Kaimosi 24 2.30 0.10 0.80 2.05
Masumbi 24 4.30 1.40 0.90 4.18
Lela 11 4.09 2.27 1.46 3.81
Rabuor 35 1.46 0.34 0.71 1.41
Muhanda 28 2.57 0.21 0.29 2.57
a
Tropical Livestock Units as defined by ILCA, where 1 TLU = 250 kg; mature Zebu
cattle = 1 TLU; immature Zebu = 0.30; mature sheep and goats = 0.12; and immature
sheep and goats = 0.04 TLU.
Source: Rapid Rural Appraisal Survey data, 1986 and 1988.

Table 5. Percentage of farms managed by women or men among

participant farmers
Management
Village Male Female
Hamisi 46 54
Kaimosi 42 58
Masumbi 79 21
Lela 82 18
Rabuor 29 71
Muhanda 14 86
Average 49 51
Source: Rapid rural appraisal survey data 1986 and 1988.
Experience gained from conducting interviews with respondents revealed

that the head of the household was not necessarily synonymous with the
farm manager; thus, the designation of farm manager was adopted in 1984.
This has assisted project scientists in targeting appropriate members of the
family for training workshops and field days, as well as in knowing which
members to interview when different kinds of data are needed.
Impact assessment
The area of impact assessment in agricultural and rural development

projects has continued to gain increased attention in the face of declining
resources available for research and technology development. Indeed, it
has always been expected that accountability of research funds must
include an assessment of introduced innovations. While impact
assessment is not a new phenomenon, rarely has this type of analysis
been done, particularly in Africa (Norman 1991). Some of the unresolved
issues include (1) what indicators to use, (2) whether to give equal
attention to both intended and unintended effects (including spin-off
Case Studies 67
technology beyond the project's mandate), (3) cost effectiveness of data

collection, (4) mode of analysis, (5) level and rigor of analysis, and (6)
carrying out partial analyses (which only look at the benefits) versus
complete analyses (which take into account the costs and benefits of the
research process) (Kumar 1989a, 1989b; Norman 1991). The present
document focuses on observed trends of key variables which serve as
indicators of potential impact of the technologies introduced by the project.
In addition, linear programming was used to simulate anticipated
technology impacts, as well as the effects of policy and institutional
support systems that play a pivotal role in the uptake of new agricultural
technologies.
Results from the 1991 livestock inventory survey. A livestock inventory

taken in July 1991 among collaborating farmers showed that overall mean
livestock numbers per farm have increased from 4.42 to 6.28 heads per
farm. However, on a species by species basis, the number of cattle per
farm has declined from 2.90 (from 1986 to 1988) to an average of 2.70
heads per farm in 1991 (Table 6). Local goats increased marginally from
.77 to 1.04 heads while sheep declined from .74 to .58 heads per farm.
The higher average in the most recent survey was due to the DPGs that
were given to participant farmers; this raised the overall mean figure by at
least two animals, the number originally given to each farm household.
It is interesting to note that no local goats were observed in Hamisi and

Kaimosi. The 3% proportion of local goats observed in the baseline survey
had disappeared by 1991; while this is a small percentage, farmers in these
two villages have had the longest exposure and experience with DPGs, and
it may be hypothesized that they now prefer the DPG over the local East
African goat.
Table 6. Mean livestock population in July 1991, heads/household
Village n DPGs Local goats Sheep Cattle
Hamisi 27 1.85 none 0.48 2.59

Kaimosi 20 2.95 none 0.40 2.35
Masumbi 19 1.30 0.45 1.40 4.15
Lela 9 4.00 3.50 0.67 2.89
Rabuor 22 2.05 0.09' 0.23 2.18
Muhanda 34 1.71 0.12 0.29 2.06
Fig. 1 shows a comparison of mean tropical livestock units (TLU) per

farm calculated from baseline data collected in 1986 (for Hamisi, Kaimosi,
and Masumbi) and 1988 (Muhanda, Rabuor, and Lela), before the DPG
was distributed. This is compared to TLUs calculated from a livestock
inventory covering all six villages in July 1991. In the second survey
period, every village except Rabuor registered a decline in overall livestock
numbers. In the first survey period the ratio of cattle to goats and sheep
was about two to one, in the second period it declined to 1.46:.1; and if
DPGs are included, the ratio declines further to .57:1. Declining livestock
numbers may be an indication of an increasing population which needs
more land for food crop production at the expense of livestock. Further,
the fact that cattle numbers are declining can be viewed as an indication
of increased impoverishment since small ruminants in the area have
generally been viewed as the poor man's cow." More rigorous statistical
analyses and surveys of nonparticipant farmers (as controls) are needed
to verify this trend. The trend also needs scrutiny both at the regional and
national levels, particularly in densely populated smallholder areas of the
country.
® 1986/88 1991
Hamisi Kaimosi Masumbi Lela Rabour Muhanda
Livestock = sheep, local goats, and cattle
Fig. 1. Mean total livestock per farm (excluding DPGs) in two survey
periods.
Case Studies 69
Fig. 2 shows the mean total number of livestock per farm in July 1991.
A comparison of livestock numbers in the two survey periods indicates that
the presence of DPGs has increased overall in terms of mean numbers per
household. As previously noted, it appears that DPGs have replaced the
local goat in two of the villages with longest exposure to the project. The
question is whether current and anticipated higher DPG stocking rates are
sustainable, given the observed declining trend of livestock and the
winding-down of the project in western Kenya before the final KDPG is
ready for distribution.
® 1986/88 1991
J
Hamisi Kaimosi Masumbi Lela Rabour Muhanda
Livestock = sheep, local goats, and cattle
Fig. 2. Mean total livestock per farm (including DPGs) in two survey
periods.
Changes in DPG inventory over the last two years. Between. 1989 and
1990 goat numbers increased from 335 to 359 (Table 7). The number of
DPGs in Masumbi, Lela, and Muhanda declined by 21, 4, and 7%,
respectively; while Hamisi, Kaimosi, and Rabuor showed corresponding
increases of 6, 8, and 43%. This yields a net increase of 7% over 1989-
1990 figures. In the 1990-1991 period, all villages (except Masumbi, with
no change) showed a decline in total goat numbers ranging between 12%
(in Lela) and 53% (in Rabuor).
Table 7. DPG inventory over three survey periods, heads
Village 1989 1991a

1990
Hamisi 65 (80%)b 69 (69%) 49 (71%)

Kaimosi 63 (70%) 68 (62%) 59 (90%)
Masumbi 33 (55%) 26 (73%) 26 (85%)
Lela .26 (65%) 25 (64%) 22 (57%)
Rabuor 67 (84%) 96 (68%) 45 (76%)
Muhanda 81 (88%) 75 (77%) 55 (75%)
Total 335 (77%) 359 (69%) 256 (76%)
a To July 1991.
b In parenthesis, percentage of all DPG goats as females in each village.
Out of 109 animals sold over the three-year period in Hamisi and
Masumbi, 69 were DPGs, compared with 11 sheep and local goats and 29
cattle (Table 8). It is noteworthy that the number of DPGs sold each year
(26, 26, and 17) in 1989 1990, and 1991, respectively, was more or less
maintained in these two villages. Further analysis is needed to determine
whether these sales reflect the true offtake rate. Revenue from DPGs
contributed a significant 30, 24, and 27% to the total value of livestock
sales (Table 9) in 1989, 1990, and the first half of 1991.
Table 8. Number of livestock sold by species and year in five villages, heads
Hamisi and Masumbi Muhanda, Rabuor and Lela
Sheep and Sheep and

Year DPGsa local goats Cattle DPGsa local goats Cattle
1989 26 8 8 60 7 22
1990 26 2 13 38 6 34
1991b 17 1 8 51 7 10
Total 69 11 29 149 20 66
a Sales are for January to June 1991.

b Up to July 1991.
Source: July 1991 survey data.

Case Studies 71
Table 9. Value of livestock sold by species in western Kenya villages, Ksha
Hamisi and Masumbi Muhanda, Rabuor, and Lela
Sheep and % revenue Sheep and % re venue

Year DPGs local goats Cattle from DPG DPGs local goats Cattle from DPG
1989 6,265 2,040 21,025 30 12,000 1,360 33,800 25
1990 6,325 750 26,105 24 8,560 1,520 51,700 14
1991b 4,045 160 15,005 27 10,790 1,690 14,450 40
Total 16,635 2,950 42,550 27 31,350 4,570 99,950 31
a1 US dollar - 16 Ksh, in 1986.

b Up to July 1991.
Source: July 1991 survey data.
In the villages of Rabuor, Muhanda, and Lela, DPG sales showed a more
erratic pattern. A total of 235 animals were sold in 1989, 1990, and first
half of 1991; of these there were 149 DPGs, 14 sheep and local goats, and
66 cattle. The proportion of DPGs to total livestock sold each year under
study was 67, 49, and 75%, which contributed 25, 14, and 40%,
respectively, to revenues from all livestock sales for the three years (Table
9). Fluctuations in the number of DPGs during this period may be
attributed to a number of factors. First is the agreement that the project
had with farmers regarding disposal of animals: as previously reported,
goats were distributed with the understanding that the first crop of female
kids born on-farm would be purchased back by the project; farmers
provided with bucks for breeding purposes could sell their bucks after 15
successful services.
The second possible explanation may be the expiration of the formal

contract between the SR-CRSP and farmers; in September of 1990 farmers
were reminded that they now had full ownership of the goats5. There is
an inherent danger in assessing adoption when contractual agreements
between farmers and researchers exist, because farmers may interpret
5 With the consultation and consent of farmers, project scientists have continued with
technology monitoring and evaluation and collection of other necessary data. This is
expected to continue until the project terminates in 1993.
expiration of the contract as waning support for the introduced technology.

This fact needs to be taken into account during impact and adoption
assessment; since there are several interacting variables in play, it is
difficult to isolate individual effects in order to determine the net impact of
the technology itself.
Third, the problems involved in issuing physical inputs and experimental

animals gratis must also be noted. With the chronic lack of cash, the near-
money nature of small stock in traditional farm environments is likely to
result in their sale, particularly if the farmers view the experimental animals
as "free of charge." Under such circumstances, their disposal (for example,
through sales to meet a family emergency) may be easier than if a farmer
had acquired the animal through some form of cash investment.
Due to the above-mentioned problems, particularly with respect to

acquisition of goats, a different adoption testing ground may be needed,
such as a new set of farmers in the same area with an entirely different
relationship with the project. Testing could also be done in other regions
of the country. In view of these factors, the terms "impact" and "adoption"
take on distinct meanings; of course, it goes without saying that adoption
cannot take place without demonstrated positive impact. A more
appropriate situation for adoption assessment may be provided by the
latest set of farmers from a village near Maseno station. Having heard
about the DPG, the farmers approached the project to inquire how they
could obtain the goats. After several such inquiries, the project bought
local East African goats and bred them to DPG bucks on the station. In-
kid does were then sold to 30 farmers in the nearby Vigina village.
Finally, when DPGs on farm are excluded, the data gathered for this
study suggest a general decline in overall livestock numbers. Farmers may
be selling off their DPGs as a stabilization measure to maintain a carrying
capacity consistent with a desired livestock mix and available, but
dwindling land resources.
Impact of the DPG on family nutrition. In order to have a continuous

supply of milk for the farm household, the project set a target of six
lactating does as being an adequate number based on household
composition in the villages under study. Simulations by the systems
analysis component of the project (Blackburn et al. 1986) indicated that six
mature does with staggered breeding could provide a minimum of 200 ml
of milk/person/d, based on a family of eight. In another study, linear
programming results suggested that about 243 ml/person/d would be
required by a family of eight members based on recommended daily
Case Studies 73
allowances of calories, protein, and seven other leading nutrients (Nyaribo

1989). Actual milk consumption figures of 93 and 145 ml/person/d have
been reported for two of the study sites (Conelly et al. 1986).
Anticipated numbers of DPGs per farm and the resultant higher milk
consumption by households have not been achieved so far for several
reasons. First, there have not been sufficient numbers of research animals
to distribute to collaborating farmers because of the long-term nature of
breeding, selecting, and stabilizing the new genotype. The high variability
in milk production from the F, crosses being used for on-farm testing has
discouraged many farmers from milking since they fear that milk extraction
will adversely affect:kid growth. Second, it was anticipated that farmers
would multiply their original herd to the desired numbers, but this has not
taken place; for diversification and liquidity, farmers still simply rear a mix
of small and large stock. Small stock are easily saleable when small
amounts of cash are needed and are a hedge against market price risk;
large stock are a store of wealth and are sold only when a large sum of
cash is needed. Third, the farmers' traditional practice has been to keep
a mix of livestock species as shown in the survey results6.
Impact assessment via linear programming. Among the analytical
techniques used by the economics component of the project is linear
programming (LP), a variant of mathematical programming. The main
advantage of LP in the context of animal systems research is its capability
to handle a wide range of production and economic relationships; this has
enabled the project to determine the combined effects of component
technologies. In addition to predicting optimal enterprise mixes, LP
analyses provide information on constraining farm resources and give
imputed values (shadow prices) on constraining resources whose market
prices, If they exist, may not reflect their true economic values. This is
especially true of SR-CRSP project sites where land and capital markets
are not well defined and are, therefore, "incomplete" in economic parlance.
Moreover, these two important factors of production have been identified
in baseline studies as likely constraints to adoption of the DPG and
supporting technologies.
Shadow prices obtained via LP can help rank and determine the
magnitude, and therefore the severity, of these resource limitations. This
technique has been used to determine semisubsistence farmer response
6 An inventory of poultry was not taken; however, it should be noted that almost every
farmer keeps some chickens.
to new technologies (Low 1975; Calkins 1981; Verinumbe et al. 1984;

Ngambeki et al. 1990) and to determine farm decision making in
developing countries (Norman 1977; Kamojou and Baker 1980).
Description of the LP model. The formulation of the model was typical

of LP models where the objective is to maximize total gross margin from
the farm's production activities subject to available, but limited resources.
In recognizing the semisubsistence nature of farm households in the
project study sites, family subsistence food requirements were included in
model specifications as a food security measure. Thus the net farm
income predicted by the model is the sum of the variable costs of
production plus the value of household food consumption.
The model was based on one year of production with seasonal

specifications for arable and grazing land, family and hired labor, capital
availability, livestock, and human nutrient requirements based on average
farm family size and composition. Crop production activities relate to the
long- and short-rain growing seasons. Two DPG rearing activities (the first
doe kidding in April and the second, in November) were included. A Zebu
cattle production activity is also specified. A detailed description of the
model is contained in Nyaribo (1989).
The major purpose of the model was to simulate conditions under which
the DPG would be integrated into the smallholder farm setting of western
Kenya using bioeconomic data from the project. The model was used to
test various technology alternatives introduced in a sequential fashion.
Several scenarios were explored; however, only four technology
alternatives are referred to:
Alternative I = The base model reflecting farm production conditions

before the introduction of project technologies.
Alternative Il = Includes the DPG without the new forages introduced

by the project.
Alternative Ill = Includes forages intercropped with food crops and

grown in fence rows, but DPGs are excluded.
Alternative IV = The full technology package (introduced DPGs and

forages); it includes storage of surplus forages and
locally available mixed grasses using a simple hay-
baling technique developed by the project (Onim et al.
1985a).
Case Studies 75
Other considerations taken in modeling were improved goat

management and availability of working capital through seasonal credit;
these are discussed in the Results section.
Production coefficient used in the model. Goat production data for two
levels of management in Hamisi village are displayed in Table 10. Under
the.first level of management, milk offtake is 101 kg (mean daily offtake of
561 ml per lactation day), the doe weighs 32 kg, and has one kidding per
year (with a 66% weaning rate), and the kid weighs 8 kg at weaning.
Under a higher management regime, milk offtake was assumed to increase
to 152 kg, with an annual doe mortality of 10% (rather than 15%), a doe
weight of 35 kg and.a weaning rate of 90% (the kid being weaned at 10
kg). For comparative purposes, Table 11 shows production parameters of
dairy goats in temperate countries as well as in the tropics. Zebu cattle
production parameters specified in the model are displayed in Table 12.
Forage production data (Table 13) were compiled from various sources
(Sands 1983; Hart et al. 1984; Kenya 1987). In cases where inconsistent
estimates were given, the lower estimate was used.
Table 10. Harms! dual-purpose goat production coefficients
Management level
Item Low' Highb
Weaning rate, % 66 90
Doe mortality/year, % 15 10
Cull rate/year, % 10 10
Doe replacement/year, % 25 20
Doe mature live weight, kg 32 35
Weaning weight, kg 8 10
Age at weaning, months 5 5
Lactation length, days 180 180
Milk yield/doe/lactation, kg 151 227
Kid milk requirements, kg 50 75
Milk offtake, kg 101 152
a
Based upon the Nutrition and Management Data Bank, 1987.
b Assuming that annual kid and doe mortality are 10%/yr and annual milk yield/doe
increased by 50 percent.
Table 11. Production coefficients of European dairy goats in temperate regions, in

developing countries in the tropics, and in a subtropical area in Australia, as
compared to the DPG in western Kenyaa
Europe Western
and U.S.A. Tropics Queensland Kenya, DPG
Milk yield/year, kg 808 290 829 151

Fat, % 3.56 - 3.60 4.9
Lactation length, days 280 254 272 180
Kidding interval, days 336 354 360 365
' Adapted from Roy-Smith (1981).
Table 12. Cattle production coefficients in Hamisi village
Milk yield, kg 419

Lactation length, days 305
Calving interval, days 543
Mature weight, kg
males 171
females 213
Source: Sands (1983).
Table 13. Forage production by quarter, kg DM/ha
Quartera
Item Q1 02 03 04 Total
Maize stover - - 3,469 1 995 5,464

Grazing 841 1,593 973 1 017 4,424
Pigeon peab 400 1,700 1,500 700 4,300
Pigeon pea` 2,000 - - - 2,000
Sudan grass 1,000 3,000 1,800 2 000 7,800
Pigeon pea hay - - 750 350 1,100
Mixed grass hay - - 487 509 996
Sudan grass hay - - 900 1 000 1,900
Total kg DM 4,241 6,293 9,879 7,571 27,984
a Q1-Q4 denote the first to the last quarter of the year.

b Pigeon pea from maize-pigeon pea intercrop.
Pigeon pea planted in fence rows.
Case Studies 77
Resource constraints for a 1.30-ha farm. Farmers in Hamisi cultivate an

average of .98 ha of land, leaving .32 ha fallow (Table 14). The fallow area
used for on-farm grazing includes the compound around the homestead,
fence rows, and patches of land with outcrops of rock that are not
cultivable. Off-farm grazing is composed of unimproved pastures along
road sides and school yards and is estimated to be about .40 ha (Hart et
al. 1984). The bulk of labor used in farm production is supplied by the
farm families with an allowance for casual hired labor amounting to 14
person-days per year. The quarterly differences in family labor supply are
due to the fact that in the second and fourth quarter children are at home
on school holidays. Family capital was estimated based on results from
the cash flow study mentioned previously.
Table 14. Resource constraints for a 1.30-ha representative farm in

Hamisi village
Quarters
Item al Q2 03 Q4
Cultivated landb, ha 0.98 0.98 0.98 0.98

On-farm fallow land, ha 0.32 0.32 0.32 0.32
Off-farm grazing land, ha 0.40 0.40 0.40 0.40
Family labor, person-days 354 377 354 377
Hired labor, person-days - 7 7 -
Family capital/yearc, Ksh 821 910 1029 391
a
01-04 denote the first to the last quarter of the year.
b Cultivated land is used in two growing periods determined by the long (March-July) and
short (September-November) rains.
Main source of family capital is remittances.
Comparison of technology altematives by linearprogramming. The base

model (Alternative 1) was run with only traditional sources of available feed,
namely maize stover and grazing. The livestock production activity
included in the model was Zebu cattle rearing. The results indicate that.44
cow-calf units were reared and all available land (.98 ha) was cultivated in
both growing seasons (Table 15). Net farm income (considering variable
costs of production and household food consumption) was Ksh 7,442'.
7
1 US dollar = 16 Ksh, in 1986.
When Alternative II was simulated (that is when the DPG was added to the
base model without any of the new grown forages), the model excluded
the DPGs as part of the solution and the predicted farm plan remained the
same. Introduction of a systematic growing of forages is represented by
Alternative III. The cultivated forages made a dramatic impact on stocking
rates with 2.41 cow-calf units in the LP solution, up from .44 units in the
base model. During the long rains, all cultivable land was used, while
during the short rains only 47% of the land was cultivated due to a lack of
capital. Predicted net farm income increased to Ksh 11,489. Alternative
IV included DPGs, grown forages, and hay storage; under this scenario,
2.87 cow-calf units were included in the LP solution. During the short
rains, 40% of the cultivable land was left fallow and net farm income
increased by 59% over base model conditions, to Ksh 11,847. It is
noteworthy that the DPG was not economically competitive under any of
these technology alternatives.
Table 15. Results of linear program modeling using DPG and

supporting technologies on stocking rates, cultivated land,
and net farm income
Stocking rate Cultivated land, %
Technology DPG Cow-calf Long income

alternatives' units units
I na 0.44
II none 0.44
III none 2.41
IV none 2.87
a
I = Existing condition without DPG.
II = Introduction of DPG without supporting technologies.
Ill = Same as II with introduction of cultivated forages fed fresh.
IV = Full technology package including hay storage.
Improved goat management and capital availability. Project studies have

indicated that the level of livestock management by farmers is generally
low, with minimal external inputs. On a typical day, goats are fed over an
8-hour period (Semenye et al. 1989). About 56% of this time is spent
grazing, while 25% is spent on cut-and-carry forage feeding. Based on this
feeding strategy, dry matter intake (DMI) per goat is 250 g DM/d from
grazing and 207 g DM/d from cut-and-carry to give a total of 456 g DM/d
Case Studies 79
(Otieno and Onim 1991). DMI under pure grazing and full confinement
systems would be 448 g DM and 840 g DM per animal per day,
respectively. This implies that on-farm DMI is a major limitation to goat
performance. In order to minimize this constraint the nutrition and
management team advises farmers to wilt fresh forages so as to enhance
dry matter intake (Siamba et al. 1990). This modification in the animal
feeding strategy also has the potential to reduce the incidence of internal
parasites.
Linear programming was used to determine what changes in farm

income would result from rearing higher-producing animals under more
conducive management conditions than those commonly observed. The
production levels serve as indicators of offtake levels necessary for the
DPGs' biological as well economic competitiveness in the western Kenya
small-scale farm environment. This scenario can be viewed as a proxy for
the final KDPG, whose production capacity will be higher than that
currently observed from F, crosses. As shown in Table 16, both improved
goat management and credit availability dramatically changed the livestock
mix predicted by the model. With improved goat management, but no
credit, the model predicted that 12.84 doe-kid units could be potentially
reared; also, 56% and 94% of the land was left to fallow during the long
and short rains, respectively, all resulting in a farm income of Ksh 13,870.
Table 16. Impact of improved goat management and credit availability with full technology
package on stocking rate, cultivated land, and net farm income
Stocking rate units Cultivated land %

Amount
Technology DPG Cow-calf Long Short borrowed Net income
alternatives rains rains Ksh Ksh
Improved
management 12.84 none 56 94 na 13,870
Credit la 9.26 none 100 100 2,053 24,350
Credit llb 9.26 none 100 100 2,053 24,350
b Credit limit is Ksh 3,359 per ha, set by the Agricultural Finance Corporation for seasonal loans.
No credit limit was set for this particular run of the model.
The two credit levels considered were a land-size based credit limit, and
no credit limit, where it was assumed that farmers could borrow as much
cash as they wanted. In either case, arable land was used in both
seasons, 9.26 DPGs were reared, and net farm income increased to Ksh
24 350. The latter means a 76% increase over income from improved goat
management alone. The magnitude of this figure stresses the extent of the
effects of having capital limitations. However, given the national budgetary

constraints, a massive extension effort and credit for smallholders is
unlikely at the moment. With the Small-Ruminant Research Program
planned by KAKI, a well-focused extension effort along with
encouragement of local funding may be an alternative solution, particularly
through farm level cooperatives and other self-help groups.
Discussion of LP Results. The results clearly indicate that DPG

production levels will need to be raised beyond those currently observed
on farm in order for the enterprise to be economically viable. Note that the
results were based on the use of the proxy F, goats which are an
intermediate genotype; it is anticipated that the final product, the KDPG,
will have higher production levels to ensure economic competitiveness.
One may then wonder why incomes associated with technology
alternatives III and IV increased over base model results without the DPG.
The answer lies in the fact that the supporting forage technologies were
also available to the Zebu cattle, which is an illustration of the effects of the
interaction of new technologies with other enterprises. It is clear from the
results that these interaction effects must be taken into account in impact
and adoption assessment.
The model scenarios of improved goat management and access to credit

simulate the effects of a focused extension effort through farmer training
and the necessary veterinary support system; access to institutional or
other forms of credit markets; and an efficient marketing infrastructure
(since ultimately surplus milk must find its way to local and regional
markets). These results suggest that policy and institutional support
systems will be crucial to the widespread adoption of the DPG, particularly
to the target group of resource-poor farmers.
Factors limiting economic viability of the DPG
Constraints fall into three broad categories: technical (biological),

sociocultural and economic, and institutional and policy support systems.
The first limitation has been discussed previously; however, it should be
reiterated that the final economic viability of the KDPG will depend on an
anticipated on-farm performance superior to that of the current F, crosses.
The sociocultural factors include the high value placed on Zebu cattle as
well as the use of goats for cultural practices which may not necessarily
be consistent with profit maximization. Additionally, avoidance of risk plays
a large role where farmers wish to rear a mix of species for diversification
purposes. The low level of enterprise management is due not only to lack
of know-how, but also to lack of cash to purchase inputs such as
veterinary medicines and services. Farmers have had difficulty in paying
Case Studies 81
for the highly subsidized veterinary supplies provided by the project;

technologies requiring even a minimal capital investment have not been
picked up due to capital shortages. Indeed, the credit analysis suggests
that overall farm productivity has been stunted for the same reasons.
Another limitation to DPG production is the severe land constraint facing

west Kenya smallholders (Nyaribo and Young 1992). The politically and
socially controversial issue of land tenure precludes the opening up of new
farm land either by relocation and/or subdivision of the larger farm estates
in the country. In view of this, it appears that the most viable option for
increasing labor productivity and providing gainful employment for the
rapidly expanding farm labor force is the intensification of livestock
production and the -promotion of private dairy processing and marketing
cooperatives.
CONCLUSIONS, RECOMMENDATIONS, AND LESSONS LEARNED
In spite of farmers' labor investment in rearing the DPGs, they still had
the attitude that the animals were "free of charge;" thus, it was difficult to
assess the true level of adoption. Perhaps some nominal cash investment
should have been applied to eliminate this attitude. There is also a need
to standardize the collection and organization of data to enable
comparison of variables and indicators across regions and over time.
Future surveys, including the proposed ex-post characterization survey,
should be organized and data analyzed along the lines of previous surveys.
The level of disaggregation of data by village and farm size class should
continue to allow for temporal comparisons.
Adoption assessment was hindered by the many interacting variables

which made it difficult to isolate the net impact of the technology. In this
regard, a different testing ground and/or conditions are needed, such as
the new set of farmers from Vigina village who are raising DPGs purchased
from Maseno station without any inputs from the project. Also, a sufficient
time lag is required after project research activities have ended in order to
assess levels of technology adoption, particularly the sustainability of DPGs
and supporting technologies and their contribution to farm families'
economic and nutritional welfare.
Fund allocation to animal systems research, particularly that with a

breeding component, will have to be committed for a minimum of 10 to 12
years and perhaps more. The breeding component of the project needs
to document problems and setbacks encountered in the development of
the KDPG to provide information which can be drawn upon in the future.
For example, it is not quite clear why F, crosses, rather than a generation
closer to the final product, are still being used for on-farm testing. Due to
the nature of FSR, project personnel are involved in extension activities
along with technology generation; as the project comes to the end, there
has been pressure on scientists to extend the SR-CRSP technology
package. It is difficult to do an effective job when scientists have to double
as extensionists and innovators. Perhaps what is needed is the inclusion
of a research-extension liaison component along the Zambian (Kean and
Sigongo 1990) and Zimbabwean models (Avila et al. 1990). While the
systems research approach has major strengths in attempting to meet
farmer research needs in a holistic fashion, most FSR projects have
mandate restrictions to work with a single commodity; for example, the
DPG in this case. Thus, mandate restrictions have resulted in spin-off work
either going unfunded or seeking outside funding or being reprimanded for
overstepping boundaries.
Social scientists have not only played an important role as evaluators but
have also had an input into the generation of technologies and have
enhanced interdisciplinary collaboration. Thus, the belief that social
scientists are needed only at the beginning and at the end of projects is
unjustified.
The project has also had impact through generating information and
creating awareness about goat research in the country and the region.
While this visibility has been positive. for the project, it has also increased
the demand for dual-purpose goats which cannot be currently met due to,
their limited numbers. Through the strong training, component of the
project, at least 25 students have been trained at the M.S. and Ph.D. levels
in the last 10 years. Students have been indispensable in doing research
requiring long gestation periods that otherwise might not have been
possible.
LITERATURE CITED
AVILA, M.; WHINGWHIRI, E.E.; MOMBESHORA, B.G. 1990. Zimbabwe:

Organization and management of on-farm research in the Department of
Research and Specialist Services, Ministry of Lands, Agriculture, and
Rural Resettlement. OFCOR country case'study no. 5. The Hague,
Netherlands, International Service for Agricultural Research (ISNAR). 206
p
Case Studies 83
BLACKBURN, H.D.; CARTWRIGHT, T.C.; HOWARD, P.J.; RUVUNA, F.

1986. Development of dual-purpose goat for smallholders in western
Kenya using computer simulation and systems analysis. College Station,
Texas. Texas Agricultural Experiment Station, Small-Ruminant
Collaborative Research Support Program Technical Report No. 89. 136
p.
BOOR, K.J.; OCHIENG, E.; FITZHUGH, H.A. 1984. Comparison of goat

and cow milk for human consumption in western Kenya. In Small-
Ruminant CRSP Kenya Workshop (3., 1984, Kabete, Kenya).
Proceedings. 57 p.
BROWN, D.L.; FITZHUGH, H.A.; CHAVULIMU, E.; KIGUHI, H.; OTIENO, E.;
AMUHINDA, P. 1983. Evaluation of lactating Toggenburg x East African
does on small farms in western Kenya: Year 1. In Small-Ruminant CRSP
Kenya Workshop (2., 1983, Naivasha, Kenya). Proceedings. Davis,
University of California. p. 11-16.
BROWN, D.L.; NDERITU, M. 1982. Nutritional strategies for dual-purpose

goats. In Small-Ruminant CRSP Kenya Workshop (2., 1982, Kabete,
Kenya). Proceedings. p. 17-19.
CALKINGS, P.H. 1981. Nutritional adaptations of linear programming for

planning rural development. American Journal of Agricultural Economics
63(2):247-254.
COLLINSON, M.P. 1982. Farming systems research in Eastern Africa: The

experience of CIMMYT and some National Agricultural Research Services
1976-81. East Lansing, Michigan State University, International
Development. Paper No. 3. 61 p.
CONELLY, W.T.; MUKHEBI, A.W.; NOLAN, M. 1986. Preliminary

assessment of household dietary patterns and utilization of milk products
in western Kenya. In Small-Ruminant SR-CRSP Workshop. (5., 1986,
Kabete, Kenya). Proceedings. ILRAD. p. 203.
CONELLY, W.T.; MUKHEBI, A.W.; OYUGI, L.; KNIPSCHEER, H.C. 1987.

Household labor allocation in an intensive crop/livestock farming system
in western Kenya. In Farming Systems Symposium. Proceedings.
Fayetteville, University of Arkansas. 15 p.
GILBERT, E.H.; NORMAN, D.W.; WINCH, F.E. 1980. Farming systems

research: appraisal. East Lansing, Michigan, Michigan State
A critical
University, Rural Development Paper No. 6. 135 p.
HART, A.; ONIM, J.F.M.; RUSSO, S.; MATHUVA, M.; FITZHUGH, H.A.
1984. An analytical framework for feed resources research on mixed
farms in western Kenya. Morrilton, Arkansas, Winrock International
Institute for Agricultural Development; Kenya Ministry of Agriculture and
Livestock Development. 43 p.
HARWOOD, R.R. 1982. Farming systems development in a resource-

limiting environment. In Readings in farm systems research and
development and research. W.W. Shaner, P.F. Phillip, W.R. Schmehl
(Eds). Boulder, Colorado, Westview Press. 260 p.
JAETZOLD, R.; SCHMIDT, H. 1982. Farm management handbook of

Kenya. Vol. II: Natural conditions and farm management information,
part A: West Kenya, Nyanza and Western Provinces. Nairobi, Kenya,
Kenya Ministry of Agriculture and the German Agricultural Team of GTZ.
397 p.
KAMAJOU, F.; BAKER, C.B. 1980. Reforming Cameroon's government

credit program: Effects on liquidity management by small farm
borrowers. American Journal of Agricultural Economics 62:709-718.
KEAN, S.; SIGONGO, L. 1990. Research extension liaison officers in

Zambia: Bridging the gap between research and extension. OFCOR
discussion paper no. 1. The Hague, Netherlands. International Service
for Agricultural Research (ISNAR). 21 p.
KENYA, GOVERNMENT OF. 1987. Per hectare gross margins for Nyanza
and Western Provinces. Nairobi, Kenya. Ministry of Agriculture.
Government Printers. 175 p.
KHAINGA, M.S.; FITZHUGH, H.A.; OCHIENG, W.O.; SIDAHMED, A.E.;

BROWN, D.; SANDS, M. 1984. SR-CRSP dual-purpose goats in western
Kenya: History and record. Small-Ruminant CRSP Kenya Workshop (3.,
1984, Kabete, Kenya). Proceedings. p. 20-29.
KUMAR, K. 1989a. Methodologies for assessing the impact of agricultural

and rural development projects: A dialogue. Washington, D.C., United
States Agency for International Development, Center for Development
Information and Evaluation, Program design and evaluation methodology
report No. 30. 60 p.
KUMAR, K. 1989b. Impact indicators: General issues and concerns.

Washington, D.C., United States Agency for International Development,
Center for Development Information and Evaluation, Occasional Paper
No. 30. 44 p.
Case Studies 85
LIHANDA, F.C.L.M. 1978. Adaptive research planning: A CIMMYT case

study of Vihiga division in Kakamega district. Nairobi, Kenya. CIMMYT
Eastern African Economics Programme. 50 p. (Mimeograph).
LOW, A.R.C. 1975. Small farm improvement strategies - the implications

of a computer simulation study of indigenous farming in Southeast
Ghana. Oxford Agricultural Station Bulletin 4:3-19.
MUKHEBI, A.W.; NYARIBO, F.B.; BERNSTEN, R.H.; REYNOLDS, E.;

MBABU, P. 1984. Preliminary socio-economic evaluation of the dual-
purpose goat enterprisein the small-scale farming systems in western
Kenya. In Small-Ruminant SR-CRSP Kenya Workshop (3., 1984, Kabete,
Kenya). Proceedings. Davis, University of California.
MUKHEBI, A.W.; NYARIBO, F.; OYUGI, L.; KNIPSCHEER, H. 1986. Cash

flow patterns of small scale farm households in western Kenya. In Small-
Ruminant CRSP Kenya Workshop (5., 1986, Nairobi, Kenya).
Proceedings. ILRAD. 11 p.
MWANDOTTO, B.A.J.; TAYLOR, J.F. 1991. The Kenya dual-purpose goat.

Brief on progress of the SR-CRSP breeding project. Kabete, Kenya.
Kenya Agricultural Research Institute/SR-CRSP Breeding Project, Kabete
Veterinary Research Laboratory. Annual Report. 16 p.
NGAMBEKI, D.S.; DEUSON, R.R.; LOWENBERG-DeBOER, J. 1990. The

impact of farming systems research on agricultural productivity: The case
of north Cameroon. Journal of Farming Systems Research-Extension
1(1):10-36.
NOBLE, A.L.; NOLAN, M.F. 1983. Sociological constraints and social

possibilities for production of goats in western Kenya. Columbia,
Missouri. University of Missouri, Department of Rural Sociology. Small-
Ruminant CRSP Technical Report Series No. 18. 51 p.
NORMAN, D.W. 1977. Economic rationality of traditional Hausa dryland

farmers in the north of Nigeria. In Tradition and dynamics in small-farm
agriculture: Economic studies in Asia, Africa, and Latin America. R.D.
Steven (Ed.). Ames, Iowa State University Press. p. 63-91.
NORMAN, D.W. 1991. Farming systems research in a declining donor

environment. Invited Opening Address for Cl MMYT-sponsored Workshop
for Research-Extension Administrators from East and Southern Africa, 8-
12 April 1991. Nairobi, Kenya. 23 p.
NYARIBO, F.B.; MUKHEBI, A.W.; BERNSTEN, R.H. 1984. Implications of

family labor use for the adoption of the dual-purpose goat enterprise on
small farms in western Kenya. In Small-Ruminant CRSP Kenya
Workshop (3.,1984, Kabete, Kenya). Proceedings. Davis, University of
California. p. 81-87.
NYARIBO, F. 1989. Integration of dual-purpose goats in small farms of

western Kenya: A linear programming analysis. Pullman, WA,
Washington State University, Ph.D. Dissertation.
NYARIBO, F.; YOUNG, D.L. 1992. Effects of land and capital constraints
on new livestock technology in western Kenya. Journal of International
Agricultural Economics 6(4):353-364.
ONIM, J.F.M.; MATHUVA, N.M.; OTIENO, K.; FITZHUGH, H.A. 1985a.

Simplified hay making for small scale farmers in western Kenya. In
Small-Ruminant CRSP Workshop (4. 1985, Kabete, Kenya). Proceedings.
p. 25-28.
ONIM, J.F.M.; SEMENYE, P.P.; FITZHUGH, H.A. 1985b. Research on feed

resources for small ruminants on smallholder farms in western Kenya.
In PANESA Workshop (2., 1985, Nairobi, Kenya). Proceedings. p. 149-
158.
ONIM, J.F.M.; MATHUVA, N.M.; OTIENO, K.; FITZHUGH, H.A. 1990. Soil
fertility changes and response of maize and beans to green manures of
leucaena, sesbania and pigeonpea. Agroforestry Systems 12:197-215.
OTIENO, K.; ONIM, J.F.M. 1991. The available feed resources on small-
holder farms in western Kenya and their potentials for the dual-purpose
goats. In Small-Ruminant CRSP Workshop (9., 1987, Muguga, Kenya).
Proceedings. p. 117-130.
REYNOLDS, J.E.; MBABU, AX; NOLAN; M. 1984. Extension services and

the smallholder: A report of preliminary findings. In Small-Ruminant
CRSP Kenya Workshop (3., 1984, Kabete, Kenya). Proceedings. p.
125-132.
ROY-SMITH, F. 1981. Intensive milk production from goats in developing

countries. In Intensive Animal Production in Developing Countries. A.J.
Smith, R.G. Gunn (Eds.). Harrogate, U.K. British Society of Animal
Production, Occasional Publication No. 4. 481 p.
Case Studies 87
SANDS, M.W. 1983. The role of livestock on smallholder farms in western

Kenya: Prospects for a dual-purpose goat. Ithaca, New York, Cornell
University. Ph.D. Dissertation. 218 p.
SEMENYE, P.P.; ONIM, J.F.M.; CONELLY, W.T.; FITZHUGH, H.A. 1989.

On-farm evaluation of dual-purpose goat production systems in Kenya.
Journal of Animal Science 67(11):3096-3102.
SHAVULIMO, R.S.; SIDAHMED, A.E.; GISEMBA, F.; ODERO, 0. 1984.

Health problems in goats in western Kenya. In Small-Ruminant SR-CRSP
Kenya Workshop (3., 1984, Kabete, Kenya). Proceedings. p. 153-164.
SIAMBA, D.N.; SEMENYE, P.P.; SHISYA M.; OYALO, S.N. 1990. Effect of
wilting herbage on body weight, egg per gram of faeces and packed cell
volume in dual-purpose goats at Maseno Station, Western Kenya. In
Small-Ruminant CRSP Scientific Workshop (8., 1990, Nairobi, Kenya).
Proceedings. ILRAD. p. 139-144.
SIDAHMED, A.E.; ONIM, J.F.M.; MUKHEBI, A.W.; SHAVULIMO, R.S.;

DeBOER, A.J.; FITZHUGH, H.A. 1985. On-farm trials with dual-purpose
goats on small farms in Western Kenya. In Research methodology for
livestock on-farm trials. Proceedings. T.L. Nardblom, A.K.H. Ahmed,
G.R. Potts (Eds.). Aleppo, Syria, International Center for Agriculture in
the Dry Areas. p. 101-132.
VERINUMBE, I.; KNIPSCHEER, H.C.; ENABOR, E.E. 1984. The economic

potential of leguminous tree crops in zero-tillage cropping Nigeria: A
linear programming model. Agroforestry Systems (1984):129-138.
INTEGRATION OF CROSSBRED DAIRY
GOATS INTO BURUNDI'S
FARMING SYSTEMS
Bernard Rey'
INTRODUCTION
The improvement of milk production from goats and sheep is one

objective of the core research program on small ruminants of the
International Livestock Center for Africa (ILCA 1988). The work on dairy
goats presented here is the result of the collaboration of ILCA with the
Ngozi goat project in Burundi. Two points are noteworthy. Firstly, dairy
goat production has been promoted in mixed crop-livestock systems.
Secondly, crossbreeding has often been used to boost goat productivity.
The Ngozi goat project, started in 1981, promoted a new livestock

system by introducing a new genotype and improving housing, feeding,
health, and marketing. It was recently felt necessary to assess and monitor
its impact, and collaboration with ILCA was initiated in 1989 to achieve this
objective. The aim of this collaborative ILCA-Ngozi project is to gather
results from an analysis of a running trend, using case studies in areas
where development projects have introduced crossbred animals in the
farming system over a relatively long period and assessing the productivity
and role of dairy goats in systems where they are the object of a deve-
lopment effort. A systems approach was adopted to study the key issue
of the genetic component. This approach was developed on two levels:
the animal production system and, more generally the farming system,
because of the strong interactions between agriculture and livestock.
1
International Livestock Center for Africa (ILCA), Addis Ababa, Ethiopia.
BACKGROUND
The Burundi high plateau region lies to the East of the Zaire-Nil Divide
at an average altitude of 1800 meters above sea level. These plateaus are
part of the Great Lakes Highlands, which comprises Burundi and Rwanda,
the Kivu region in Zaire, the Kigezi region of Uganda and the "Buta" region
of Tanzania. A major characteristic of this area is the variable but generally
high population density (up to 400 people/km2), which results in very small
farm sizes.
Historical background
The historical process of agricultural development in this region led to

the current farming system, based on banana and coffee cropping with
cattle raising linked to the farmers' cultural background(s).
The Bantu people reached the region early in the first millennium of our
era. They brought with them an agricultural system based on sorghum,
millet, and legumes, which they implemented in the wooded savannah area
where annual rainfall is between 1,000 and 1,500 mm and altitude ranges
from 1,500 to 1,800 meters. The corresponding cultural organization is still
based on (1) a segmented and highly decentralized political system and
(2) a dispersed settlement pattern with families living outside villages
(Jones and Egli 1984). Social cohesion is ensured by traditional practices
of giving, taking, and drinking beer together. This was initially sorghum
beer, but has been substituted by banana beer, thus giving this crop a
primary place in the cropping system today. The settlement pattern has
been modified towards creating villages in the few areas exposed to slave
trade.
From the middle of this millennium, Nilotic pastoralists migrated south

with their herds into areas where there was plenty of land for grazing.
They progressively settled and adopted the agriculturalists' language and
customs though keeping a different culture and mode of production; they
were, in particular, at the heart of the creation of the Great Lakes
Kingdoms, where cattle ownership and custodianship formed a basis for
political organization. Livestock and agriculture tend, consequently, to
respond to distinct farmer rationales.
Case Studies 91
Growing population density since 19002, principally linked to an

improved health service which reduced mortality, induced modifications of
the farming pattern. The agricultural frontier was expanded to lower (Imbo
plain in Burundi, for instance) and higher altitudes (tea and pyrethrum cash
cropping). Within the frontier, modifications were noticeable: shorter
grazed fallows, cropping of steeper land and swamps, first in dry seasons
and later in wet seasons. This led to an intensive land use pattern in the
current farming system.
Pattern of livestock ownership
The above-mentioned evolution has progressively pushed cattle out of

the system (Figs. 1 a and 1 b) in the high plateaus, as a result of shrinking
pastoral areas. Despite the prestige attached to cattle ownership, their
substitution by small species (goats, rodents, poultry) is taking place
spontaneously among farmers. Jones and Egli (1984) noted that there is
no apparent need to resist or abet the process of replacing cattle with
goats, but suggested the need for improvements in goat raising (health
status, breeding, feeding and marketing).
Fig. ia. Evolution of goat population in the high plateaus of Rwanda

and Burundi (1961-1988).
2 Average density in the region is now 150 people/km2, but reaches 400 in certain areas,
as in Ngozi.
Fig. 1b. Evolution of sheep population in the high plateaus of Burundi

and Rwanda (1961-1988).
The situation described is relatively homogeneous and characteristic of

the Great Lakes highlands, and similar conditions occur in other parts of
Africa (e.g., Ethiopia and Cameroon). Several research and development
projects focusing on goats and having particular emphasis on "dual-
purpose" (meat and milk) animals developed through crossbreeding with
exotic breeds are found in Eastern Africa (Table 1).
Table 1. Goat development projects in Eastern Africa
Country Location Type1 Breed crossbred Year begun

with SEAG2
Kenya Kisumu R Toggenburg 1980

Tanzania Morogoro R Norwegian Landrace 1983
Malawi Bunda R/D Boer and Alpine; in addition,
selection of SEAG 1983
Rwanda D Alpine 1987
Burundi Ngozi D Alpine 1980
Ethiopia Harar D Anglo Nubian 1989
R=research project; D= development project.

2
Small East African goat.
Case Studies 93
The goat development project in Ngozi
Among the East African projects, the Ngozi goat project is the oldest
(more than 10 years) with the largest number of animals distributed on
farms: over 8,000 crossbred animals are born each year in the flocks
owned by farmers participating in the project. This project started in 1980
with funding from the German Agency for Technical Cooperation (GTZ)
and a permanent GTZ staff. The aims of the project are to improve the
farmers' diet by providing them with goat milk and meat in larger quan-
tities, to generate a source of income other than coffee, and to contribute
to maintaining soil fertility by producing manure (Schultze and Ngen-
dahimana 1986). Schmidt (1990) states that goats seem to offer an
appropriate way to compensate for the decreasing availability of meat and
milk from cattle; in fact, half of the farmers in the region already keep the
local small East African goats (SEAG) for meat production.
The main technical innovation in the Ngozi project was the

crossbreeding of the local goat with Alpine bucks to improve potential milk
yield and body size. The project set up 40 mating centers to which
farmers bring their does for mating. Participation in crossbreeding is
entirely voluntary. To encourage farmers to participate, the project
developed its own extension service under the authority of the Livestock
Board and promoted a package including housing, fodder crops, and
health services. Proper milking of goats was also demonstrated. The
project later built a factory to process milk into cheese and organized milk
and cheese marketing channels between farmers and the factory. Thus,
a second key modification of the system was making goat milk a
commercial product.
The project covers 14 communities in an area totaling 2,000 km2. Some

3,500 farmers were registered as taking their does to mating centers where
more than 8,000 matings were recorded each year in 1988, 1989 and 1990.
It is believed that, at the present time, one-third to one-half of the goats in
the three communities closest to the station are crossbred. The milk
factory is now receiving 42,000 liters of fresh milk annually, benefiting some
350 farmers (Schmidt 1990).
With crossbred goats having been adopted by a large number of

farmers over several years, the Ngozi area offers a good opportunity to
carry out a case study to examine the social and technical conditions
pertaining to the development of goat production in crop-based systems
in the East African highlands.
OBJECTIVES AND THE USE OF A SYSTEMS APPROACH
The promotion of crossbred goats and, particularly, of dairy crossbreds

appears to be a major trend in livestock development in the East African
highlands where population density and pressure on land are increasing.
Development agencies have recently focused on this fact and have taken
a variety of approaches, giving priority either to research (Kenya) or to
development (Burundi).
Among the aspects of common interest to these projects are the

integration of goats in the farming system, and the performance of
crossbreds both on station and under farmer management. The ILCA
research objective was therefore twofold: (1) to determine how the
modified animal production system (through crossbreeding) fits into the
farming system, and (2) to determine how the productivity of the livestock
system can be increased in the Ngozi area.
The Ngozi project seeks to affect the farming system through changes
in the goat component. Thus, the analysis of the animal production level
was the first one to consider. Small farmers (such as those in Ngozi) have
limited resources allocated to optimizing their land and/or labor
productivity.
Possibilities for improvements in agricultural productivity appear cons-

trained by decreasing soil fertility and the related importance of tubers and
by an intensive cropping system based on intercropping. As livestock
compete for the same resources, it is important to determine the
socioeconomic objectives pursued by farmers in adopting the crossbred
goats, especially as regards the sustainability of the intervention.
Socioeconomic objectives can be best determined by analyzing the use
farmers make of dual-purpose animals. However, analysis must
concentrate on the animal production system, considered as a subsystem
of the farming system.
A key task for development isto determine which level of crossbreeding

is advisable. This question can be answered from the biological
standpoint. Collecting production information on farmers' flocks and
analyzing it (in this case comparing breeds) has been done for a long time.
The techniques for on-farm data collection are not specific to animal
production systems research.
Case Studies 95
Animal production systems research (APSR) becomes specific when

observed performances are to be explained. Assessment of the genotype
with the best on-farm potential must take into account the trade-off made
by the farmer between his objectives and the animals' biological
performance.
The APSR approach integrates both biological and socioeconomic

elements (farmers' strategies) to understand the current productivity level
and identify entry points for improvement. However, the research reported
here is aimed at understanding the biological performance of the crossbred
goats under farmer management. The current situation allows this
comparison to be made using a large number of animals, the potentials
and limitations of which are known by the farmers since the animals are
already part of the systems they practice.
Finally, considering that an improved genotype has been propagated,

what are the constraints to optimal utilization of its potential productivity?
How do farmers' practices affect goat productivity? Answers to these
questions may give some feedback to disciplinary research for optimizing
this genetic resource while considering the available inputs and their costs
in the farming system as well as the farmers' objectives. Are there
conceptual models to run such an analysis?
Lhoste (1984) proposes a model for the livestock system. The system
is viewed as a triangle; the farmer, the flock and the land are the three
sides which can be characterized by measurements (farmers' objectives,
flock production, fodder, etc.). The corners of the triangle represent the
interfaces, which give consistency to the system and limit the actions on
each element. The analysis of the interfaces comprises, for example,
range management (action of farmers on land), animal product
consumption or sales (farmer decisions) and grazing performance (data,
species consumed).
The Uoste model can help researchers to analyze and conceptualize the
system at the diagnostic stage. For instance, an equilibrium in the
productivity of the three faces of the triangle may be sought; it may be
found that land productivity in Burundi is reaching a ceiling, flock
productivity appears improvable, and that labor productivity is the key
issue.
ORGANIZATION OF THE PROJECT
A central relationship between a national development project and an

international research institute
This research is primarily conducted with collaboration between ILCA

and the Ngozi goat project, which focuses on extension and development
with a minor research activity at its breeding station. The goat project
operates under the Livestock Board, and works in its region of intervention
with no formal linkages with the Buyenzi SRD (Regional Society for
Development), which focuses on coffee but is also in charge of food-crop
extension (Fig. 2).
HOUSEHOLD CONSUMPTION
COFFEE
SWAMP CROPS
(RICE, BEANS)
OTHER
STOCK
INTERCROPS
FOOD CROPS MEAT

(KIDS)
PASTURES
MILK
FODDER CROPS
MATING CENTERS
REGIONAL SOCIETY GOAT PROJECT
FOR DEVELOPMENT
Fig. 2. The Ngozi farming system and extension services.
One of the recent concerns of the Ngozi project has been to set up
monitoring procedures to assess its impact and reorient its efforts as
necessary. Research objectives and plans of action have been decided
upon jointly between the project and ILCA. The project staff are closely
Case Studies 97
involved in the day-to-day implementation of the data collection (staffing,

local expenses, data management), while ILCA staff members regularly visit
the project to set up and monitor the data collection mechanism (sampling,
data measurement techniques, data validation). Data analysis is carried
out as much as possible in Burundi. Although the Ngozi goat project gives
low priority to research, its involvement is essential to ensure confidence
in the data recorded (linkage with the data recording system: computer,
software, checking) and for the adequate interpretation of the data.
Collaboration with the project staff also facilitates the establishment of

research set up at the farm level and brings confidence in the information
collected; contacts with farmers are easier, there is more willingness to
collaborate, and there is a certain continuity in the farmers' involvement.
This involvement of farmers in the diagnostic phase has facilitated the
adoption of some innovations.
Relationships with national research and training institutions
L'lnstitut des Sciences Agronomiques du Burundi (ISABU) and the

Faculty of Agronomy of Burundi University have agreed that the latter will
deal primarily with small ruminant research. The Faculty is consequently
ILCA's principal contact although the Department of Socioeconomics at
ISABU collaborates. The contact with the University also allows for the
involvement of graduate students in the project.
METHODOLOGY
Two levels of observation were considered: the farming system (or farm)
level and the animal production system (or flock) level. In addition, by
closely monitoring animal performance at the individual level, the accuracy
of the flock level information was increased. However, it is not necessary
to collect all information on an individual level; for example, grazing
management is similar for all animals.
Farm choice
The performance of the different crossbred genotypes was monitored

in participating farmers' farms while the performance of small East African
goats was monitored on farms not directly exposed to the project's

extension activities.
The small average flock size (two to five does) forced the selection of a
large number of farms in order to make proper comparisons between
breeds. One hundred participating farmers were randomly selected from
the list of farmers who had brought does for mating, as well as 100 neigh-
boring (nonparticipating) farmers who had never brought their does for
mating. In addition, 53 farmers known as "breeders" (which the project
had chosen to multiply crossbred animals) were included in the survey and
flock monitoring.
Data collection
Data collection was organized in several phases, as shown in Fig. 3.
INDIVIDUAL
SYSTEM
PERFORMANCES
FLOCK SURVEY FARM SURVEY

INITIALIZATION FARM CHARACTERISTICS
ANIMAL IDENTIFICATION
GOAT MANAGEMENT
INDIVIDUAL MEASUREMENTS
GOAT MANAGEMENT
' REPRODUCTION PRODUCTION
DEMOGRAPHY
' PRODUCTIVITY
MONITORING
100-HEAD SUBSAMPLE 25-FARM SUBSAMPLE

PARASITOLOGY
SAMPLE AND TOPIC

EXPERIMENTATION TO BE DETERMINED
Fig. 3. Methodology for data collection, Ngozi Project
Initial visit This visit focused on both the farm and flock levels. All
animals were ear-tagged and identified according to their genotype, sex,
and age. Farm resources were surveyed (land, labor, capital assets) and
Case Studies 99
the products characterizing the farming system were identified and later
used to compare participating and nonparticipating farmers. Elements of
goat management which remain constant throughout the year, but might
influence performance (e.g., housing) were also recorded.
Fortnightly visits. Major management decisions (e.,g., feeding, health

care), were recorded at the flock level. However, most information, such
as all demographic events (birth, death, and purchases or sales of animals)
and elements of productivity (milk offtake per doe, kid growth, adult
weight), was collected at the individual animal level.
Complementary information. Data was collected on two subsamples:

one at the farm/flock levels and the other at the individual level. On 25
farms data was collected to estimate the farm budget and to determine the
contribution of goats to farm income. This subsample was chosen on the
basis of a typology of farms in order to ensure as much variability as
possible. Feces were collected fortnightly from a 100-head subsample to
determine parasite burden. These animals were chosen at random from
all flocks.
Informal survey techniques. At an initial stage, information collected by

informal survey techniques was used to establish the questionnaires and
data collection formats. Nevertheless, the knowledge gained was minimal
compared with the current knowledge of the project staff. At later stages
informal visits allowed for the testing of hypotheses resulting from data
analysis (e.g., farm labor allocation).
Enumerators
Each enumerator surveyed both participating and nonparticipating

farmers in one of the administrative areas in the region where the project
operates. They were specifically hired for this purpose and were given a
bicycle on loan for use in conducting the survey. They reported to the
project once a month with the documents filled in. Data are stored on the
computer in Ngozi.
ANALYTICAL TECHNIQUES USED
Farming system level
The data collected during the initial survey allowed the comparison of
farms having crossbred goats with those without crossbred goats. A least-
square means analysis (SAS procedure) was applied to structural factors:
land resources and use, labor, and capital. Cluster and discriminant
analyses (SPSS procedures) were used to define a typology of goat
farmers and to identify different rationales for keeping crossbred goats.
The farm types identified were used to select a subsample of farms for
which farm budget data were collected, including monthly cash flows for
both crop and livestock enterprises. Farm budgets of the different farm
types were compared; comparisons were made between participating and
nonparticipating farmers, and among distinct farm clusters.
Animal production system level
Animal performance recorded at the individual level allowed for the study
of the effect that crossbreeding had under farmer management as well as
for the evaluation of the impact of current constraints and practices.
Assessment of the farmers' use of the crossbred animals' dual-purpose
potential provided an insight into their reasons for keeping them.
The performance of the different genotypes was compared using the

data collected on farm; these data were also compared with the data
recorded at the project's breeding station. Standard indicators and indices
calculated similarly for on-station and on-farm data were compared using
least-squares means analyses. The history of the Ngozi project facilitated
this approach since it had kept and monitored a flock of crossbred goats
since its inception. Data from this flock provided a unique reference, within
a single environment, against which on-farm performance could be
compared. Several questions are still raised: Which index (or indices) can
be used satisfactorily to summarize flock productivity? How can the spread
over time of animal production be integrated? In the Results section, some
comments are given regarding these questions.
Constraints to improved productivity were identified, and their impact

was determined as far as possible in terms of potential improvement. For
example, it appeared important to define the periods during the year when
Case Studies 101
productivity is most affected3 by factors such as endoparasitism. Some

experiments in a later stage will be designed on the basis of this constraint.
An effort was made to determine those practices that affect, positively

or negatively, animal and flock performance. It was assumed that some
farm practices are rather constant and thus may not be considered as a
source of variability in performance. Others do vary from one farm to
another and may, taken alone or combined, explain differences in
performance and allow for recommendations and/or experimentation. This
approach has been used in crop research but has not been widely used
in livestock research.
Individual performances can be aggregated to estimate flock

performance. The flock is the level of decision making for farmers; the
farmers' strategy should be examined in terms of annual productivity at the
flock level. Biological and simulated economic indicators were used. Each
year may be considered as a replicate, or several years may be
aggregated for analytical purposes.
RESULTS
The main survey was analyzed and its results tested with informal farmer
interviews. This led to a characterization of participatory farmers vis-a-vis
nonparticipatory ones based on the comparison of farm structures and ac-
tivities.
Biological results have not yet been examined thoroughly. The data
collected over 18 months did not allow for a solid assessment of small
ruminant performance, particularly in terms of reproductive indicators and
survival. The study of the female reproductive cycle requires a longer
term. If the data are collected over a two-year period, most females
monitored will have given birth twice or more, and most kids will have
reached maturity. Variation between years might be considered for
examination of flock productivity, although only two years' data would be
available.
3 Milk productionprovides an earlier indication of variations in flock productivity than does

any other product.
Land resources
"Breeders" (farmers participating in the crossbreeding program of the

Ngozi project) possess a farm structure representative of the average goat
farm in the region, with the exception of having a lesser number of fields.
Landholdings of participating farmers were not found to be larger than the
average farm keeping local-type goats (Fig. 4). These two types of farms
are, nevertheless, larger than the overall average farm size (both with and
without goats) in Ngozi, which seems to have stabilized at 0.63 hectares
(SNES 1984, 1987).
Fig. 4. Average farm size in Ngozi, Burundi.
Flock structure and genetic composition
While participating farms do not differ from nonparticipating farms with

respect to flock size, the same cannot be said with respect to the sex ratio
and genetic structure of the flocks (Figs. 5 and 6).
The availability of bucks at the project's mating centers and the

recommendation of culling young males probably led participating farmers
to keep fewer bucks in their flocks than farmers who were not crossbreed-
ing their goats. Breeders had more bucks and young animals (Fig. 5).
Case Studies 103
Fig. 5. Average flock size.
Participating farmers' flocks are made up of approximately one-third 50%

Alpine goats, one-third 75% Alpine crossbreds and one-third small East
African goats. Nonparticipating farmers tend to keep only local-type goats
and a few crossbred animals purchased outside the project mating centers.
Breeders were chosen by the project on the basis of the importance they
attached to 75% Alpine goats and on the quality of the care given to their
animals. These farmers had dropped the small East African goats almost
entirely and had some animals with a very high proportion of Alpine blood
in a flock, which was generally larger than the average found on
participating farms (Fig. 6).
0 Small East African ® 50% Alpine crossbred

® 75% Alpine crossbred O Other crosses
Nonparticipating Participating 'Breeders'
Types of farmers
Fig. 6. Goat genotypes present in Ngozi farms.
Differences in the cropping system
The adoption of crossbred goats is correlated with differences in the

cropping system (where the use of manure probably plays a key role) and,
in a certain number of cases, with the establishment of fodder crops (a
quarter of participating farmers and more than a third of the breeders have
established fodder crops).
Intercropping is the dominant cropping system. Cassava, sorghum, rice,

cocoyam, and beans are present with the same degree of frequency
among all farmer groups (Fig. 7). The same is true for the main cash
crops, which average 353 coffee and 82 banana trees per farm.
Case Studies 105
Fig. 7. Cropping frequency for each type of farmer.
The participating farmers and the breeders grow potatoes, sweet

potatoes, and maize more frequently than do nonparticipating farmers.
Potato, a cash crop grown by progressive farmers, brings a large income
but needs fertile soil and involves the use of chemicals or manure. Only
4.7% of nonparticipating farmers indicated sweet potato as their primary
food crop, while 8.5% of participating farmers and 9.5% of the breeders do
so. Conversely, 11.3% of nonparticipating farmers indicated cocoyam as
their primary food crop, compared with only 6.1 % and 5.3% of participating
farmers and breeders, respectively. Finally, maize was indicated as an
important food crop only by participating farmers.
The relative frequency of potato and maize cropping might be indirectly

influenced by improved goat management. Informal interviews indicated
that the above two crops as well as beans are fertilized with manure in
preference to other crops. Keeping goats in a paddock and using cut-and-
carry fodder seem to increase the quantity of manure collected by farmers,
thus benefiting these three crops. While sweet potato vines constitute a
good fodder, this was not identified as a reason for growing this crop more
frequently when keeping improved animals.
Labor force employment
There was no difference in the family size of participating and nonpartic-

ipating farmers, but the introduction of crossbred goats appeared to modify
the characteristics of labor force employment.
The adoption of crossbred goats by farmers was linked with other

recommended practices: fodder production, herding instead of tethering,
the use of a pen, feeding cut-and-carry fodder to the goats at night. These
management techniques are employed more frequently by participating
farmers. Zero-grazing (total confinement), although practiced by a small
number of these farmers, indicates a significant modification of the
livestock system, going beyond the project recommendations (Fig. 8).
Existence of housing 0Total oonlinsmant F Fodder mops
-150
50 -140
40 -130
s0 -120
20 -110
10
Nonparticipating Participating 'Breeders'

Types of farmers
Fig. 8. Goat management: housing and feeding.
With practices such as these, farmers have accepted a labor-intensive

animal production system. Goat raising appears to be an alternative to off-
farm employment for the head of the household. Participating farmers
Case Studies 107
work less off-farm than the nonparticipating ones (11.4% and 27.6%,
respectively), although their farm size and available labor force are similar.
On the other hand, breeders have a larger family labor force, which allows
them to engage in off-farm activities more actively.
Thus, the improved goat system is at least partially responsible for

higher income and may be a more suitable alternative for farmers than
nonagricultural employment. This assumption is supported by the farm
budget analysis (calculated by monitoring cash flows over one year on 25
farms). For at least six months participating farmers have a higher income
derived from livestock and a higher agricultural income in general (Figs. 9
and 10)4. Milk was an essential element of the income; Schmidt (1990)
estimated that for farmers delivering milk to the factory milk sales
represented 20% of their annual net farm income.
Agr. expenditures MAgr. Income *Cumul. gross margen

2
.........................................:+.::",.........................
-8
July Aug. Sept. Oct. Nov Dec.
1o0
(A .np. raft In 155Q 175 Bunxndl hrnas per US$
1
Fig. 9. Farm budget: nonparticipating farmers.
4 It is noted that coffee harvest occurs in the six months missing in Figs. 9 and 10; this is
the main cash crop for Ngozi farmers.
\
®Agr. Expenditures ®Agr. Income Cumul. gross margen
0
a
10 M
ac M R
3 -2
W
0
.g -
July Aug. Sept. Oct. Nov. Dec.
1990
Exchange rate In 1990: 175 Burundi francs per USS
Fig. 10. Farm budget: participating farms.
Goat farmer strategies
A cluster analysis of 14 quantitative variables identified five types of goat

farmers. Two major factors determined this grouping: one was a set of
variables linked to the farmers' ages, and the other was the structure of the
goat flock. The different types of farmers are characterized as follows:
Type 1: young traditional farmers. These are young farmers with the
least land and labor. The flock size is four head. These farmers did not
respond to the development project: all goats are small East African and
no improved practices were found.
Type 2: goat crossbreeding farmers. These young farmers have only

crossbred goats in their flocks, which are not particularly large. They have
the fewest plots and are not particularly engaged in banana production.
Type 3: young goat farmers. These farmers have large flocks, in which
small East African goats are still a majority, but where crossbreds are
present (2 to 3 head out of 9).
Case Studies 109
Type 4: old goat farmers. This category comprises older farmers with
large flocks similar in size to Type 3). In comparison to Type 3, more
crossbreeding is practiced, as only a third of the animals are SEAG.
Type 5: older farmers. This last type groups older farmers with small
flocks (4 heads) but with more crossbred animals in relative terms (only
one-fourth SEAG). These farms have the largest family size and are also
more involved in off-farm activities than other types.
All five types were represented in the participating farmers group; the
breeder group did not include any traditional farmers while none of the
nonparticipating farmers were involved in crossbreeding (Table 2).
Table 2. Cross-tabulation of farmer types and participation in the

crossbreeding program
Younger Older
Type of farmer 1 2 3 4 5 Total
Nonparticipating 60 - 28 10 7 105
Participating 6 20 19 18 20 83
"Breeders" - 13 3 14 8 38
Total 66 33 50 42 35 226
The factors related to the age of the head of the household led to the
formation of two groups: "younger farmers" (Types 1, 2 and 3) and "older
farmers" (Types 4 and 5). Hubert (1989) has described the process of
wealth accumulation by farmers in Burundi: age is generally correlated with
larger landholdings and more coffee and banana trees. Differences
between younger and older farmers were significant (P<0.01) for the
variables marked with an asterisk (*) in Table 3.
Table 3. Average values of labor, land, and flock composition for five goat farmer types.
Younger farmers Older farmers
Goat farmer types 1 2 3 4 5
Age of household head, yrs. 42.00 44.00 46.00 54.00 510)

Family labor units 3.60 4.90 4.70 4.90 570
Off-farm family labor, months 3.00 2.00 2.00 3.00 70)
Number of plots 5.10 3.60 5.20 4.80 430
Farm area, acres 48.00 66.00 61.00 153.00 1350)
Coffee trees 268.00 269.00 323.00 4,157.00 515m
Banana trees R 122.00 102.00 170.00 398.00 2650)
Total flock 4.00 5.30 9.30 9.00 40)

Small East African goats 3.92 0.30 6.38 3.19 083
50% Alpine crossbreds 0.16 1.58 1.44 2.38 1®
75% Alpine crossbreds 0.08 2.85 1.32 2.81 19
Other crosses 0.00 0.85 0.12 0.81 071
Does 2.30 3.55 5.30 5.60 2.43

Bucks 0.85 0.82 1.58 1.21 0(9
Kids 1.02 0.94 2.38 2.38 1.45
Type 1: Traditional farmer; Type 2: Goat crossbreeding farmer; Type 3: Young goat farmer; Type 4: Old goat
farmer;
Type 5: Older farmers.
Indicates significant differences (P s 0.01) between "younger" and "older" farmers.
The existence of different farmer types suggests that different strategies

should be followed with respect to their flocks and the crossbreeding
scheme. Type 2 farmers appear to represent a strategic choice for
promoting goat farming with crossbreds at the expense of other forms of
capitalization (coffee, banana). Their landholdings facilitate goat farming
because of the few but large plots. These farmers are dynamic and willing
to participate in the program. Older farmers seem to be more involved in
crossbreeding than younger farmers of Types 1 and 3. In both age
categories (not considering Type 1, which does not involve crossbreeding),
small flocks (Types 2 and 5) have proportionately more crossbred animals
than larger flocks (Types 3 and 4), in which capitalization might be a key
motivation. In small flocks, an increase in animal productivity
(crossbreeding) may be a sound strategy.
Case Studies 111
ASSESSMENT OF IMPACT
Assessment of research impact is client-dependent. From this

standpoint, two perspectives may be considered when assessing the
impact of the ILCA livestock system project in Burundi. The traditional
wisdom of FSR defines the farmer as the client; in Ngozi, farm
development itself (through the project's activities) is also a client.
Impact at the farmer level
The dairy goat technological package, promoted by the extension

system and focused on by this study, is not a direct result of the farming
systems research conducted in Burundi. No transferable technology has
yet resulted from the ILCA research, although ILCA has been working with
farmers for over two years. Nonetheless, this research has created it
elements for ex-post analysis, such as some mechanisms of adoption of
the package, where labor employment and productivity criteria are key
determinants. These criteria for adoption, determined at the farmer level
using a systems perspective, should be considered when new techniques
are being designed and tested in a future phase. Better focused, they will
have an earlier impact on the welfare of farmers.
Completion of the goat project in accordance with farmers' objectives

and the sustained presence of improved genotypes will directly affect
farmers' welfare in the short and longer terms. The expected impact will
be centered on technical aspects (modifications of traditional management
practices), but some impact may also be obtained from marketing goat
products.
Impact at the development level
If regional development is considered as the client, this research will also

have an institutional impact. Although the aim of the ILCA research was
not to assess the impact of the development project, the results will
provide the dairy goat project staff with some elements for an assessment
at the farmer level; for example, What target group would correspond to
the participating farmer? Sociologically speaking, who are the "breeders"?
Any modification and/or adaptation of the technological package

resulting from observations of farmers practices or from future experiments
will have an impact on the development project as well as on farmers. If
a "best" level of crossbreeding is defined or if it is decided that farmers

should pay for some extension services or if some extension message
elements are declared irrelevant, then the institutional setup of the project
would be affected. The fact that this research is linked to a development
project increases the probability of achieving a good impact and
assimilation of knowledge.
The Ngozi goat project clearly benefited from the formal analysis of
information that had been collected on the station over a 10-year period,
particularly of information concerning the value of the different crossbreds.
Information collected in Ngozi may help the institutions in charge of

development in Burundi to make policy decisions. Those decisions may
affect the status of the milk factory (private vs. public, subsidized vs. at-
cost) in Ngozi or might result in the setting up of other similar projects in
Burundi.
The research results should also give some feedback to the project staff
and the policy decision makers. To what degree does milking depend on
a market organized by the project through use of incentive prices? Do
farmers really prefer the improved stock over the local one? Will this stock
be conditioned by the permanence of mating centers? How is this stock
able to cope with decreasing farm sizes?5
Regional impact
Farming systems research has been described as highly site-specific.

This specificity is based on the socioeconomic and physical characteristics
of the area. Where existing development efforts play an essential role, the
local trend in farming system dynamics is also an element of this
specificity. The work conducted in Ngozi is a case in point because goat
development has been promoted by the government over the last 10 years.
Some of the conclusions might not even be pertinent for other regions in
Burundi, because they have no access to the extension services for goat
development or to the goat-milk market.
5 Landholdings are mostly inherited, split in equal shares between heirs. Although some
land is progressively bought by older farmers, a son never reaches the farm size of his
father at the same age.
Case Studies 113
In spite of the above-mentioned site specificity, the existence of several

goat crossbreeding projects in different countries (Table 1) will result in
some degree of impact at the regional level. Even though lessons from a
site-specific study might not be applicable beyond the project, it is
important to make the results available to those considering similar
development projects elsewhere. GTZ, which funded the project in Ngozi,
is considering starting a similar project in Kenya. ILCA has a comparative
advantage in that respect, which can be enhanced if modeling and
simulation are developed on the basis of research in other comparable
locations.
CONSTRAINTS AND DIFFICULTIES
The specific characteristics of small ruminants in small farms brought a

set of constraints to the study of this livestock system. There are, in
particular, important and frequent flows of goats in and out of farmers'
flocks, linked to the moderate economic value of the animals; in fact, some
of the participating farmers sold all their crossbred animals. It has also
been argued that the farmer's memory for small stock is much worse than
for cattle, whose individual value is higher and productive cycle longer.
This gives less confidence in the use of reconstructed individual female
productive histories (a tool developed for cattle) to estimate flock produc-
tivity. Finally, the fact that goats exist in small sized flocks increases the
cost of research by making it necessary to include a large number of farms
in any survey; this, in turn, affects tools used for data analysis.
Although the role of livestock in the Ngozi farming system is traditionally

not central, it is apparently gaining in importance. The interactions with
cropping are complex, highlighting the idea of integration in a mixed
farming system. Most techniques used to address APSR in Africa have
been developed in pastoral and agropastoral systems where cropping can
be considered quite independently. A whole-farm model should address
crop-livestock interaction and, particularly, the soil fertility issue; there are
very few such- models which have been effective in developing countries.
Modeling the farming system at a regional level with the interactions
between farm types and the goat project as a variable factor might even
be foreseen as a final aim; however, at the present time ILCA has little
expertise for conducting this type of modeling and simulation.
The livestock system was used as the entry point to the study of the
farming system. This has some limitations. In a comprehensive analysis
of the farming system, other farm commodities may have a higher priority
than goats; thus, the recommendations given on goat production may be
biased by the restriction imposed on the ILCA-Ngozi project.
Consequently, animal production systems research might be seen as
commodity-biased, just as the development project is biased in Ngozi.
This limitation of the approach might be marginal in pastoral livestock
systems but is more important in integrated crop-livestock systems. To
resolve this question on goat development, more contacts with other
development agencies working in the region on nonlivestock issues are
advisable. This is, for instance, the case for coffee plots, where
intercropping fodder is discouraged by the Regional Society for
Development.
RESEARCH PROSPECTS
The data-recording mechanism which has been set up provides

information useful for the design of recommendations to the development
project and for the prioritization of technological alternative testings.
Although permanent data recording systems exist in developed countries,
sometimes under the pressure of development agencies or farmers'
associations, few examples of data recorded for more than a five-year
period can be found in developing countries. One can question how much
can be gained in the accuracy of the data, or what new information would
result from pursuing data collection for one or two more years during the
diagnostic phase.
Designs for on-farm experiments will result from the data analysis
(diagnostic phase). These experiments will require the monitoring of the
same type of information about productivity and flock census as collected
during the diagnostic phase, together with some more specific studies.
For instance, indicators of labor productivity, the importance of which has
been shown, might be added. The degree of pre-existing knowledge about
the system will certainly determine the type of information that will have to
be recorded.
The establishment of a structure to collect information in a large number

of small farms is a costly process while its operation carries only a
marginal cost. Should a new monitoring system be established for any
Case Studies 115
experiment, research costs would be quite high. The current setup seems
appropriate for monitoring on-farm experiments on a characterized farm
subsample and should be maintained.
The ILCA Bioeconomic Herd Model (Kaufmann et al. 1990) will be

adjusted to include the livestock system submodel and to address the
sustainability of the "genetic innovation" with its economic implications. A
comparison between similar projects in the region might be conducted in
a second instance to refine the use of the model.
CONCLUSIONS AND RECOMMENDATIONS
At the current stage, the ILCA animal production systems research

program has not led to technology generation or to technology transfer.
However, it allowed for the analysis of a livestock system which has been
exposed to two major changes: a different genotype (Alpine goats) and a
new commodity (milk). It has one more important feature, which is
collaborative work with a national development project. What lessons can
be drawn?
Animal production systems research should be considered as a proper

means to study the effect of livestock development on the farming system.
The systems approach used by the ILCA-Ngozi project allowed for the ex-
post evaluation of a development activity. The study of the livestock
system has been seen as an entry point to the farming system. Although
this approach has some bias, it allows consideration of improvements for
livestock from the perspective of the farming system, which is the level of
the farmers' rationale and the framework for development. Consequently,
the lessons gained have a local impact as well as a potential impact for
similar development projects and/or situations.
More importance should be given to farmers' diversity and the

dynamics of the farming system in animal production systems research.
Systems research is a good approach to consideration of two important
factors for development: the diversity of farmers and the evolution of
farming and livestock systems. Although these factors can be
endogenous, they can also be the result of development in which case
they usually occur at a faster rate.
Ngozi showed the importance of adding a historical perspective to a

polaroid" diagnosis in APSR. Constraint identification concerns not only
today's problems but also foreseeable ones. Both the trend of farming
evolution and the effect of past and continuing actions on the system must
be considered in order to understand the current situation and design tech-
nologies for the future. A frequent shortcoming of farming systems
research is the pressure to answer a "one-cropping-season" problem;
social sciences must make contributions beyond the simple calculation of
farm budgets if the above aspects are to be dealt with appropriately.
The results presented in this paper show that small farmers in the same
situation respond differently to a proposed innovation. The diversity of
farmers' responses should be given more importance than is usually the
case. Researchers as well as development specialists must be aware of
that diversity when drawing up and proposing alternative improved
technologies.
Livestock systems research should be closely associated with

development, rather than pretending to play its role. The research
program currently being conducted in Burundi has one important
characteristic, which is to work in collaboration with a development project.
This has advantages over setting up such research on farm, since it
creates an audience for the results and provides a tool to implement and
extend recommended practices. This is quite different from what is
described as farming systems research and extension, where research is
the primary goal and extension is taken up at the end without the
necessary means to conduct it properly.
An institutional setup such as that used in the Ngozi goat project permits
the implementation of an effective triangular relationship between research,
development and farmers (Tourte and Billaz 1982). Systems research in
such a context can form the basis for advising on development efforts,
drawing on a good understanding of the system. The feedback is given
to research (research needs) in the perspective of development. Thus, the
body in charge of the extension of research results has been associated
with the design of the research.
The ILCA-Ngozi experience highlights another element of the triangular

relationship: development cannot always wait for specific research results
before being implemented. Ngozi is an example of a successful project
conceived on a sound technical and social knowledge base. Systems
research can now be brought to bear on problems that appear during the
development process, thus sustaining the development effort.
Case Studies 117
Livestock systems research needs some conceptual models. The

characterization of livestock performance under farmer management is a
core result in APSR. This allows the determination of constraints and
opportunities for improvement. The analysis of the data is, nevertheless,
too often restricted to the biological side; socioeconomic factors, mostly
based on the understanding of current practices, should be given greater
weight in the design of alternatives.
Conceptual models for the analysis of a livestock system are needed in

order to integrate the biological indicators and the socioeconomic factors.
The conceptual model proposed by Lhoste (1984) for livestock systems
has been mentioned. This model gives a framework at the diagnostic
stage, but does not facilitate further analysis of livestock productivity.
Other models appear insufficient as well. It has also been argued in this
paper that data collected, and the way it is collected, must use such
models as a framework. Flock monitoring can provide high-quality
information and is cost effective.
LITERATURE CITED
FAO (FOOD AND AGRICULTURE ORGANIZATION). 1988. Production

yearbook, vol. 42. Rome, Italy, FAO Statistics Series No. 88. 350 p.
HUBERT, J.P. 1989. Reflexion sur ('evolution des systemes d'exploitation

agricole au Burundi A partir d'une typologie dans la region du Kirimoro.
Note de discussion. Bujumbura, Burundi. L'Institut des Sciences
Agronomiques du Burundi. 24 p.
ILCA (INTERNATIONAL LIVESTOCK CENTER FOR AFRICA). 1988.

Sustainable production from livestock in sub-Saharan Africa: ILCA's
programme plans and future requirements, 1989-1993. Addis Ababa,
Ethiopia, ILCA. 93 p.
JONES, W.I.; EGLI, R. 1984. Farming systems in Africa. Washington, D.

C., World Bank Technical Paper No. 27. 117 p.
KAUFMANN, R. VON; MCINTIRE, J.; ITTY, P. 1990. ILCA bio-economic

herd model for microcomputer: User's manual. Addis Ababa, Ethiopia.
ILCA. 114 p.
LHOSTE, P. 1984: Le diagnostic sur les systemes d'elevage. Cahiers de

la Recherche Developpement (France) 4: 84-88.
SCHMIDT, U. 1990. Evolution et resultats de neuf annees de projet

germano-burundais dans le cadre de la cooperation technique.
Capricorne (France) 3(1):15-20.
SCHULTZE, A.; NGENDAHIMANA, S. 1986. Experience du projet caprins

Ngozi au Burundi. In Atelier de travail sur les petits ruminants de la
CEPGL. S6minaire de Bukavu, 8-10 janvier, 1985. Proceedings.
Gitega, Burundi, Institute de Recherche Agronomique et Zootechnique.
p. 26-37.
SNES (SERVICE NATIONAL DES ETUDES ET STATISTIQUES). 1984.

Enquete agricole dans la province de Ngozi, 1981. Bujumbura, Burundi,
Ministry of Planning. 252 p + annexes.
SNES. 1987. Systemes de production et analyze de la situation

alimentaire et nutritionnelle dans la regime du Buyenzi, Bujumbura,
Burundi. v. 1: rapport de synthese, 57 p; v. 2: enquetes et annexes.
TOURTE, R.; BILLAZ, R. 1982. Approches des systemes agraires et

fonction recherche developpement: contribution a la mise au point d'une
demarche. Agronomie Tropicale (France) 37(3): 223-231.
THE THREE-STRATA FORAGE SYSTEM FOR
DRYLAND FARMING IN INDONESIA'
1. Made Nifis, K. Lana, M. Suarna,

W. Sukanten and S. Putra2
ABSTRACT
The Three-Strata Forage System (TSFS) experiment was carried out

during a six-year period (1984-1990) to study the role of grasses and
ground legumes (as first stratum), shrub legumes (as second stratum), and
fodder trees (as third stratum), all integrated with food crops and livestock,
in increasing the productivity of dryland farming areas in Bali.
The design was a completely randomized factorial arrangement

consisting of two farming systems: TSFS and NTSFS (Non-Three-Strata
Forage System) and two stocking rates: two and four steers/ha, with 32
plots as replications per treatment. The 64 plots were randomly selected
from the flat (F), sloping (S), and intermediate (FS) topographical
conditions. Each 0.25-ha TSFS was divided into a 0.16-ha core area for
food crops, a 0.09-ha peripheral area for grasses and ground legumes, and
a 200-m circumference for 2,000 shrub legumes and 42 fodder trees. In
contrast, each 0.25-ha NTSFS plot was for food crops only.
TSFS (as compared with NTSFS) produced 43% less food crops but
91 % more livestock feed and 46% more firewood; moreover, the livestock
feed contained 13% more crude protein. TSFS cattle gained 13% more
weight and were 24% more efficient in utilizing the feed while their
1
Data presented in this paper have been obtained from the Three-Strata Forage
Indonesia Project, entitled Three-Strata Systems for Cattle Feeding in Dryland Farming
Area in Bali, financed by IDRC, Canada.
2
Department of Nutrition and Tropical Forage Science, Udayana University, Denpasar,
Bali, Indonesia.
carcasses contained 10% more fat and 9% more rump portion than those
of the NTSFS. Stocking rate of TSFS during the wet and dry seasons was,
respectively, 45% and 30% higher and the carrying capacity was 52%
greater than those of NTSFS. Soil erosion in the TSFS was 51 % less while
its soil organic matter and soil moisture were, respectively, 11 % and 10%
higher than those of NTSFS. Less time was spent in managing livestock
and gathering firewood, allowing TSFS farmers more time to attend social
activities. The overall result was that TSFS generated 30% more farm
income and created a better ecological balance than the NTSFS.
INTRODUCTION
Of the 47 million ha of agricultural land in Indonesia, 85% is dryland

farming area (Anonymous 1986). Of this area only 7% is in the form of
natural grassland while 64% is used for food crops and 29% for plantation
crops. In the smallholder dryland farming area, food crop production is
the main objective, while raising livestock is a sideline activity. No land is
specifically allocated to grow livestock feed; instead, livestock are fed crop
residues, natural grasses, and shrub and tree fodders grown both on the
perimeter of the field and on land not used for food crop production.
It is not uncommon, therefore, that livestock feed is often in short

supply, particularly during the dry season. After the food crop is
harvested, the land, which is usually too dry for another cultivation, is used
for tethered grazing. Overgrazing and uncontrolled lopping of the shrubs
and fodder trees induces soil erosion which, in turn, reduces the
productivity of the dryland farming area.
CONCEPT OF THE THREE-STRATA FORAGE SYSTEM
The Three-Strata Forage System (TSFS) consisted of planting and

harvesting grass, ground legumes, and fodder shrubs and trees so that
ruminant feeds are available all year around (Nitis et al. 1989). The
rationale is that ruminant feed would be supplied by the first stratum (grass
and ground legumes) during the wet season, the second stratum (shrub
legumes) during the dry season, and the third stratum (fodder trees) during
the late dry season. Ground and shrub legumes are included in the TSFS
Case Studies 121
to increase soil fertility through the root nodule contribution of the legume
species and to increase the nutritive value of ruminant diets through the
inclusion of protein-rich legume foliage.
DESCRIPTION OF TSFS
Objectives
The overall objective of the TSFS was to increase the farmer's income
through improved land management involving crop-animal systems. The
specific objective was to define a three-strata production model for crop
and livestock production for the semiarid rainfall areas of Bali. This was
done by (1) evaluating growth and yield of grasses, legumes, shrubs, trees,
and crop components; (2) measuring the nutritive value of the feed
sources produced in the system; (3) measuring the performance (growth,
feed intake, and carcass quality) of Bali cattle; (4) evaluating carrying
capacity; and (5) comparing the economic and ecological characteristics
of TSFS with those of the existing traditional Non-Three-Strata Forage
System (NTSFS).
Design
Both laboratory and on-farm experiments were conducted during a six-

year period (1984-1990). The on-farm experiments consisted of
agronomic, livestock, soil, socioeconomic, extension, and communication
studies. A completely randomized design was used with a factorial
arrangement of treatments consisting of two forage systems (TSFS and
NTSFS), two stocking rates (two and four steers/ha), and 11 and 10
replications for the low and high stocking rates, respectively. The
parametric data were subjected to an analysis of variance (Steel and Torrie
1960); nonparametric data were analyzed with the X2 test (Gomez and
Gomez 1976). Where appropriate, data were subjected to regression
analysis (Sokal and Rolf 1969).
Site
The site chosen was a dryland farming area in the semiarid climatic
zone. The average annual rainfall is 1,681 mm distributed throughout the
4-month wet season, and the relative humidity varies from 69% to 87
percent. The temperature ranges from 25° to 28°C. The topography is

undulating, with 30% flat land. The limestone-based soil is red-brown
Mediterranean type (Winaya et al. 1980), with a pH ranging from 6.7 to 8.7;
organic matter varies from 2.21 % to 2.97% and total nitrogen from 0.045%
to 0.159 percent.
Forage and food crops
Each 0.25-ha NTSFS plot was planted entirely with cash crops; four
fodder trees were planted at the corners of each NTSFS plot as boundary
markers; an additional 0.25-ha area was allocated for tethered grazing, for
each of the 32 units of NTSFS plots.
Each of the 32 units of TSFS covered 0.25 ha and consisted of a 0.16-

ha core area, a 0.09-ha peripheral area, and a 200-m circumference (Fig.
1); the core area was located in the center of the plot and was planted with
cash crops (e.g., corn, soybean, cassava) commonly grown by farmers.
The peripheral area of TSFS was a 5-m wide buffer zone located between
the core and the circumference. It was subdivided into 45 m2 lots (5 m x
9 m). Each lot was planted with buffel grass (Cenchrus ciliaris cv.
Gayndah), common stylo (Slylosanthes guianensis cv. Graham), green
panic (Panicum maximum cv. Trichoglume), centro (Centrosema
pubescens), and Caribbean stylo (Stylosanthes hamata cv. Verano), in that
order. This grass-legume pasture was designated as the first stratum.
The circumference consisted of fodder trees such as Ficus poacellie,

Lannea corromandilica, and Hibiscus tilliaceus planted at 5-m intervals.
Stands of 50 Gliricidia sepium shrubs and 50 Leucaena leucocephala
shrubs were planted alternately between the fodder trees. The Gliricidia
and Leucaena were designated as the second stratum while the Ficus,
Lannea, and Hibiscus trees constituted the third stratum; together, they
formed a hedgerow fence around the TSFS unit. Therefore, one TSFS unit
consisted of 0.16 ha of cash crops for human use, 0.09 ha of pasture, 2000
shrub legumes, and 42 fodder trees.
After the establishment of the TSFS (which took 12 months), cutting of

the first stratum and lopping of the second and third strata were done
twice a year (at the end of the 4-month wet season and at the end of the
8-month dry season). The cash crops were planted at the onset of the wet
season; then the corn was harvested in February, soybean in March and
cassava in July, as commonly done by farmers in Bali. The excess forage
during the wet season was conserved in the form of hay and silage. Even
Case Studies 123
though the silage was rated as good, farmers were reluctant to make it
because of the time required for cutting the forage and preparing the
silage.
1Ma: ipadng: GM an
moo
Hdscus =
Struba: apadn0: 10 an
+
GNrkldla
Leucaene =
- ++
-
0tasasa: row: 10 an
° Cench us - b
Panicum p-
. LOP MW row 10 an
+ Greham atyb =0
+
+
Cenhasetne = c
vmano 'Yo V.
Plot aba-b0mx5om
-
Lot aiaa 5 m x 9 m
o
_ row direction
4--Ym--*
Fig. 1. Core, peripheral and circumference areas of the TSFS.
Source: Nitis (1984).
The association among the three strata and the food crops was studied
by vegetation sampling, using a 0.5 x 0.5 m iron rod quadrate placed at 1
m from the shrubs and trees and on the border of the adjacent grass,
legume, and cash crops (Fig. 2). This operation was done twice a year:
at the end of the 4-month wet season and at the end of the dry season;
botanical composition, yield of plant components and total biomass were
measured.
'
A
0++++* S
++++++---- * --- -o+++++++
u s
° a
b
0
B ....................
O w+o- To:
b
L .
a
Oh orop
Ky: Aa - ON.rgOm.nt of In« S

loeaUon of bwe In the pa 1
} So* Fla. 1 for dWI and
Fig. 2. Location of the samples (numbered squares) in relation to plot

structure
Source: Nitis at al (1989)
Chemical composition and nutritive value studies (in vitro digestibility,

allelopathic factors and antiqualitative substances) were made on 100- to
500-g forage samples obtained from the TSFS and NTSFS plots. The
botanical composition of the feed offered depended on the season and
availability of the plant species. Generally, during the wet season the TSFS
diet consisted of 75% grasses and ground legumes and 25% shrub legume
and tree fodders; during the dry season the diet consisted of 25% grasses
and ground legumes and 75%.shrub legume and tree fodders.
Case Studies 125
Bali cattle
The stocking rate was arranged according to the 0.25-ha plot size. The
two steer/ha stocking rate was equivalent to one steer for two contiguous
plots (0.5 ha total). The four steer/ha stocking rate was equivalent to one
steer per 0.25-ha plot.
In one experiment, the initial weight of the animals averaged 122 kg,
while in another the initial weight was 275 kg. Each steer was weighed
every 28 days. When the steers reached 375 kg live weight, they were
slaughtered for carcass appraisal. For TSFS, forage was cut from the
allocated plot and the cattle were stall-fed continuously; NTSFS cattle were
tethered and grazed during the day and stall-fed at night with the forage
cut from the allocated NTSFS plot.
Soil status
Eighteen months after the TSFS was established, a 150-cm long

galvanized iron pipe was stuck in the middle of the grass and legume lots
at an angle of 8° to 10°. A mark made at ground level represented the
zero reading of erosion. Soil samples for each of the three topographical
conditions were taken at the beginning and end of the experiment in each
of the grass, legume, and food crop lots. Three replicates of each
combination were used.
Socioeconomic and communication surveys
A socioeconomic survey was carried out every month starting three

years after the TSFS experiment was begun in South Bali. Three TSFS
demonstration plots were established in North Bali for the purpose of
studying the socioeconomic, extension (Arga et al. 1990) and
communication (Nuraini et al. 1991 ) aspects of the TSFS.
Farmer selection and participation
In South Bali, the farmers selected owned more than 0.25-ha of land,
tilled their own land, raised cattle, owned water catchments, were members
of farmers' groups, were willing to surrender 0.25 ha of land to forage
production, and showed commitment to active participation in the project.
Of the 26 farmers, eight were women. Each farmer looked after at least
one plot and, at the most, four plots. In North Bali, 84 farmers participated
in the socioeconomic studies. Of the 240 farmers participating in the
extension and communications courses, only eight were women.
Adjustments to the TSFS
The following adjustments were made in the TSFS without affecting its
basic concept: 1) Graham stylo, a volunteer cultivar, was replaced with
Urochloa mosambisensis, a more persistent legume; 2) Stylosanthes
scabra cv. Seca was planted in the Centrosema plots to act as climber for
this creeping legume; 3) because the Leucaena yield was affected by an
infestation Heteripsylla cubana, Acacia villosa was sown along the
Leucaena row; and 4) because terraces could eventually deteriorate and,
therefore, reduce the sustainable use of the hillside, the lower rim of the
terrace was planted with shrub legumes at a spacing of 1 m as well as with
a 1-m wide strip of mixed grass and legume.
IMPACT OF TSFS
As may be seen in Table 1, the six-year experience with TSFS in Bali

permitted an assessment of the positive and negative impacts of this
system on the productivity of dryland
farming area.
Bioeconomic impact
Food. The creation of a 0.09-ha forage boundary in the TSFS resulted

in a reduction of the area destined to food crop production from 0.25 ha
to 0.16 ha, causing a reduction in food crop yield by 40 percent. However,
traditionally, farmers who continuously planted food crops for three years
on sloping land experienced a marked decrease in the food crop yield,
sometimes leading to total failure to produce grains. Thus, they were
forced to fallow their land at least for three years. Compared to this
situation, the decrease in food crop yield in the TSFS plots was only 19%
after three years, which implied the possibility of planting food crops for an
additional year. After four years, however, the decrease in yield was 54
percent. Notwithstanding, it is obvious that there is a need to design
improved food cropping techniques (such as land-use rotation systems) to
achieve sustainable crop yields.
Case Studies 127
Table 1. Average productivity of the TSFS and NTSFS plots'
Attribute TSFS NTSFS

(0.25 ha) (0.50 ha)2
Food, kg dry weight/year 696 1,216

Decrease in food crop yield after four years, % 54 67
Feed, kg dry weight/year 1,096 573
Cattle
o Live weight gain in 3 years, kg/plot 186 166
t. Stocking rate, kg live weight
o Wet season 600 400
t. Dry season 130 100
P. Carrying capacity, kg live weight/year 300 200
c. Carcass
o Loin eye muscle area, cm2 67.57 67.02
P. Backfat thickness, cm 1.94 1.83
P. Time to reach market weight, years 3.80 4.30
P. Time spent in caring, min/day 115 137
P. Animals affected by endoparasite infestation 2 6
Soil status
c. Erosion in 3 years, mm 10.13 23.80
P. Organic matter in 5 years, % 5.29 4.76
P. Total N in 5 years, % 0.80 0.76
Firewood, kg dry weight/year 1,468 261
Kampung hens
o Egg production per brooding hen, units/year 385 247
c Hatchability, % 93 76
Farmer income, Rp/year4 251,228 400,597
Nitis et al. (1989) present a full statistical analysis of these data.

2
Including the 0.25 ha for tethered grazing.
a
Not Inciuaing straw aerivea from the tool crops.
4
1 US$ = 950 Rupiahs.
Feed. The 0.09-ha improved grass and ground legumes in the TSFS
produced more forage than the 0.25 ha of natural pasture (an area used
for tethered grazing in conjunction with the 0.25-ha NTSFS plot). With the
additional 2,000 shrubs and 42 trees, the total wet- and dry-season forage
production of the TSFS was 91 % more than that of NTSFS. Additionally,
Stylosanthes, Centrosema, Acacia villosa, Gliricidia and Leucaena are
forage legumes containing 25% to 30% crude protein. Therefore, diets
derived from the TSFS contained 13% more crude protein than was
possible with the feed obtained from the NTSFS.
Cattle. Due to the high availability of forage and crop residues in the
TSFS plots, the actual stocking rate during the 4-month wet season and
the 8-month dry season was 45% and 30% higher, respectively, than that
of NTSFS. The TSFS carrying capacity for the whole year was calculated
to be 52% higher than that of NTSFS.
Since the TSFS produced not only more but also better quality forages,
the TSFS cattle gained 12% more weight and took 12% less time to reach
market weight. As TSFS steers were always kept in confinement, they
moved around less to get the feed; this may have caused the 6% increase
in fat deposited in the carcass. However, TSFS steers had a slightly larger
loin-eye muscle area, perhaps due to the fact that they consumed better
quality feeds. The higher concentration of shrub and tree fodders in the
cattle's diet resulted in darker meat; however, when the meat was cooked,
this effect disappeared. Due to the management associated with TSFS,
these animals were less infested by endoparasites than NTSFS cattle.
Farm income. The cost of running TSFS was 3% more than NTSFS.
However, TSFS per capita income was 30% higher than that produced by
NTSFS.
Impact on resources
Firewood. Lopping (twice a year) and pollarding (every three years) of

shrubs and trees resulted in the production of 1.5 t/year of firewood per
0.25-ha plot. Since the firewood requirement of a household is 4.2 t
annually, a single 0.25-ha TSFS plot could supply 36% of that requirement.
Soil fertility. Soil erosion in TSFS was 57% less than that of NTSFS.
The slower rate of erosion was presumably due to the density of the grass
and legume cover, which buffered the impact of rainfall while the root
systems held the soil and prevented run off. Even though the effects of the
individual grasses and legumes varied, the average clay, total salt, and
organic matter contents of the TSFS soil were higher than those of the
NTSFS soil. When TSFS is used, the higher yields of corn, soybeans, and
cassava grown in association with some of the grasses and legumes may
be indicators of increased soil fertility.
Planting materials. When all the materials had to be bought, the

establishment of the TSFS was relatively expensive for the smallholders.
However, once established, the TSFS plots became sources of seeds and
cuttings for the establishment of new TSFS plots, at no cost. Three years
Case Studies 129
after establishment, enough cuttings from Gliricidia and vegetative material

from grasses, as well as sufficient seeds from the ground legumes,
Leucaena, and Acacia were obtained to establish three new TSFS plots.
Only the tree cuttings had to be obtained from outside sources since those
in the TSFS were not ready to be used as planting materials before the
third year.
Time. Since the TSFS cattle were always kept in stalls, while the
NTSFS cattle had to be moved from the stall to the grazing plot every day,
the TSFS farmer saved 22 minutes every day. Once every week the
NTSFS farmer had to spend 50 to 80 minutes gathering firewood. The
TSFS farmer, on the other hand, spent no extra time to get firewood since
this was constantly available due to lopping of shrubs and trees, which
provided branches for firewood as well as leaves for the cattle. The TSFS
farmer can save up to 15 hours a month which could be used for other
productive on-farm and off-farm activities.
Secondary benefits
The 0.25-ha TSFS contained 3.3 million plants representing 14 species,

all of which flowered regularly. Thus, they produced at least 330 millions
flowers (Nitis 1989) which could supply pollen and nectar for one or two
colonies of honeybees. Grass seeds, ground legume seeds, and white
ants hiding under the leaf debris were sources of feed for the native
chickens roaming in the TSFS. The native hens in the TSFS produced 56%
more eggs and the hatchability of the eggs was 22% higher than in the
NTSFS. The TSFS strata served as an appropriate breeding ground for
snails. TSFS, therefore, has the potential of becoming a source of snails for
the Back Yard Snail-Keeping Project, currently promoted by the local
government.
Institutional development
Each state university in Indonesia has to carry out its Tri Dharma
mission, which is a program consisting of teaching, research, and
extension. TSFS may be used for this purpose, particularly for the Faculty
of Animal Husbandry.
Teaching. TSFS has been used as practical work for the Agrostology,
Ruminant Nutrition and Beef Cattle Nutrition courses. At the
undergraduate level, TSFS has been used by students working on their
sarjana degree and by staff members pursuing their Master of Science or

Ph.D. degrees. In total, there were 42 students and staff members, (as well
as one postgraduate student from Japan and one from British Columbia,
Canada) using TSFS for their degree requirement.
Research. TSFS research was started in the area of animal nutrition

(disciplinary approach), then developed into an interactive mode involving
two departments in the faculty and later two faculties in the university
(multidisciplinary approach). The TSFS research activity involved both the
academic and development aspects. Academic research was usually
associated with theses leading to professional degrees (sarjana3, Master
of Science or Ph.D.), while development research was oriented mainly
towards solving the farmers' problems. The academic researchers always
discussed problems with their supervisors while the development
researchers did so with the farmers.
In addition to the research carried out by the TSFS team, 38

experiments were conducted as part of degree requirements and 31 for
nondegree requirements. Such work covered a wide range of topics
including grasses, ground legumes, shrubs, trees, cash crops, cattle, soil,
diseases, feed conservation, feed chemical composition and nutritive value,
and socioeconomic aspects. Forty-seven students, 104 lecturers, 24
technicians and 401 farmers participated in this activity.
Extension. Dissemination of research results was carried out through

the education of (passing on knowledge) and service to (passing on
technical skills) those interested in the TSFS.
Forty-one papers directly concerning TSFS and eight papers citing

TSFS have been presented in symposia, seminars, congresses and
workshops within the country and overseas. Seven papers have been
published in journals and one paper appears in a book. One booklet
entitled Petunjuk Praktis Tata Laksana Sistem Tiga Strata (A Practical
Guide for the Establishment and Management of TSFS) was published in
1988 and circulated to government and nongovernmental organizations
directly and indirectly involved in forage and livestock development in
Indonesia.
3 Sarjana is an academic program consisting of a minimum of four and a maximum of

seven years.
Case Studies 131
The local newspapers Bali Post and Nusa Tenggara and the national
newspapers Kompas and Suara Pembaharuan have published articles on
TSFS. The Bali TV station, in its development news program, has
broadcasted a 15-minute program on TSFS activity. The same station has
used TSFS material to produce a video film entitled Menantang Lahan
Kering (Challenges of the Dryland Area). This 25-minute film ranked fifth
in a field of 19 entries in a government-sponsored competition celebrating
the 44th anniversary of radio and television. The Animal Husbandry
Undergraduate Student Association has a fortnightly radio program which
has included a 25-minute special session on TSFS. The Buleleng Regency
radio broadcasting station has transmitted the TSFS dissemination course.
A 3-day short course has been conducted for neighboring farmers. The
40 participants included 34 farmers from 17 farmer groups and six field
extension specialists. The objective of the short course was to provide
technical knowledge on TSFS to farmers interested in applying TSFS
technology. TSFS teams have been sent to Chiangmai, Thailand, to attend
an agricultural communications workshop; to Sri Lanka to attend a
socioeconomic seminar; and to Bangkok, Thailand, to attend a workshop
on audio-visual aid techniques for extension. IDRC4, in association with
Udayana University, has held an international workshop on The Utilization
of Shrub and Tree Fodders by Farm Animals. Fifty delegates from 14
countries discussed 26 papers in this scientific gathering. As will be
detailed in the section "Extent of TSFS Adoption," a number of
organizations have been involved in the establishment of TSFS units, some
of which have served as demonstration plots.
Apart from IQRC officers, in the past eight years 25 visits have been
made to the TSFS units by individuals and government and
nongovernment officers from Indonesia and from overseas. Moreover, in
the past six years 21 persons from within the country and overseas have
requested copies of TSFS publications. This is in addition to the 20
institutions that regularly receive TSFS progress reports and publications.
Conservation of the environment
Soil erosion in TSFS was less than that in NTSFS because of the
buffering effect of the forage and fodder in the first, second, and third
4 See List of Acronyms and Abbreviations.

strata. TSFS created a more pleasing environment, as plants and forages

in all three strata remained green almost all year round.
Mode of TSFS application
The application of TSFS centered on its integration with cash crop and
livestock production (Fig. 3). This integration has the following
advantages: 1) management will be more intensive because farmers will go
to their fields every day to look after cash crops; 2) livestock will not
disturb the cash crops because of the protection provided by a hedgerow
fence; 3) tethered grazing of the livestock will be minimal as TSFS will
supply green feeds year round; 4) soil fertility will be improved by regularly
and evenly spreading barnyard manure on the field; and 5) socioeconomic
well-being will improve as cash crops and TSFS provide for the farmer's
daily needs and generate extra income while livestock will serve as
insurance to meet the farmer's unexpected expenses.
Food F_+ Live-

crops Land stock
(0.25 ha)
Seeds Cash
l Fire- Meat
Cuttings crops Forage wood Others
e
H on y
i
Fig. 3. Mode of TSFS technology application
Source: Nitis et al. (1988a).

Case Studies 133
TSFS ADOPTION
Mechanisms for TSFS adoption at the farm level
The general scheme describing the technology transfer and adoption

process is presented in Fig. 4. Details of specific activities involved are
given subsequently.
Fig. 4. Mechanism of TSFS technology transfer
Source: Nitis et al (1988a)
Consultation with the local government. Before selecting the TSFS site,
consultations were made to (1) the Provincial Bureau of Planning regarding
the general master plan for agricultural development, (2) the Regency
Bureau of Planning regarding priorities in agricultural and livestock
development, and (3) the head of the village regarding the specific location
for forestry, agriculture, and livestock development. Such consultation was
deemed necessary to insure that the TSFS was in line with existing
government programs and did not interfere with those already in progress;
it also served to inform officials of the new development program that was
about to take place.
Participating farmers. Farmers chosen for participation in the TSFS

were those who (1) tilled their own land rather than leased land from other
farmers; (2) owned more than 0.25 ha of land, so that TSFS did not
interfere with farming practices already established; (3) were willing to
convert 0.09 ha of land from food crops into forage production; and (4)
kept livestock for savings and for other unexpected expenses which could
not be covered by the cash crop income.
Demonstration plots. In selecting the site, the following should be

taken into account: (1) the steeper the slope of the land (within certain
limits), the less chance for food crops to succeed and the greater the need
for forage development; (2) preference should be given to land with an
existing live fence boundary since the shrubs and trees can be used as
cuttings for the TSFS; and (3) preference should be given to land currently
tilled for food crop production rather than fallow land which is being used
for tethered grazing.
Training course. The training course for farmers consisted of three

phases. Phase 1 was carried out at the end of the dry season; it described
how to select the land, plow it, dig holes for the second- and third-strata
trees and shrubs, and prepare the plots for the first stratum; planting
techniques for grass, ground legumes, shrub legumes, and fodder trees
were also explained. Phase 2, carried out in the middle of the dry season
of the following year, described how to prevent the neighbor's livestock
from entering the TSFS area, how to manage seeds and vegetative material
in order to start a new TSFS plot the following year, and how to refrain
from overcutting the TSFS when forage availability dwindled. Phase 3,
carried out early in the wet season, described how to cut each stratum so
as to ensure a constant supply of feed throughout the year; in addition, the
farmers were instructed on how to feed the livestock so that the diet
contained a balanced nutrient composition for growth, fattening, or milk
production; the possibility of conserving excess forage in the form of silage
and/or hay was also covered; farmers were also taught the importance of
livestock care and disease prevention.
Do-it-yourself approach. The course was divided into three phases was
to ensure that the farmers had the correct technique fresh in their
memories when working in the field. Depending on the availability of the
TSFS team members, farmers were attended to as a group or on an
individual basis. The team showed the farmer what to do, then asked
him/her to follow the example. When it was apparent that the instruction
was understood, the farmer was left alone to complete the work. A flexible
deadline for the completion of the work was arranged. The team did not
Case Studies 135
necessarily stay the entire time, as some farmers felt uneasy in its
presence. When revisiting, the team did not directly point at the mistakes
that had been made, but rather put more emphasis on what good things
had been done.
Rural family participation. Even though only the head of the family
participated in the training course, when it came to the work in the field the
team encouraged all family members to participate. Such family
participation develops a sense of belonging so that when the head of the
family attends other functions, other family members will take responsibility
for the TSFS activity. The role of the children should not be
underestimated since they will talk with other children at school and make
them aware that the grasses, legumes, shrubs, and trees should not be cut
indiscriminately as is usually done in the public field. The role of the wife
also should not be overlooked.
Comparative group control. Some farmers may have had an interest

in the TSFS, but did not adopt it because of hand labor constraints. These
farmers practiced the traditional system (NTSFS), but were usually good
observers and may have picked up a few components of the TSFS
technology which they judged suitable to their specific location. Such
farmers may be used as a comparative control to monitor the production
performance of the NTSFS. Farmers who did not fall into this category or
were not involved in the TSFS experience could be monitored to estimate
the extent of TSFS technology transfer.
Contact person. To attend the day-to-day activities of the farmer and

to bridge the communication gap between farmers and team members, a
contact person staying at the TSFS location was appointed. The contact
person was trained to be an all-around expert on technical matters, to
make temporary decisions, to act quickly on his own judgement, and to
report immediately to the team, which would then take whatever action
deemed necessary.
Farmer group formation. If a farmer group already existed, the entire

group or members of it formed a TSFS farmer group. If no farmer group
existed, the TSFS group had to be organized from scratch. Members of
the group preferably included both those actively participating in the
demonstration plot (TSFS farmers) and those only interested on the
demonstration plot (NTSFS farmers). The TSFS activity was seen as only
part of the farmer's many group activities such as family planning, child
care, health programs, and home industry. The farmer group had the
option of setting rules and regulations and enforcing a penalty for those
who did not obey these regulations.
Guidance and consultation. Meetings between the team and the

farmers were carried out regularly to discuss various matters and to seek
solutions beneficial to the farmers. Problem solving through personal
contact proved to be effective. Problem solving by voting was avoided at
all times. The team members were instructed to listen, consider, and
acknowledge farmers' suggestions for problem solving whenever
considered appropriate.
Observing the local customs and traditions. Whenever a new

component technology was to be developed, consultations with the
farmers were made so as to avoid any inconvenience to the farmers or
damage to the surroundings. For example, some farmers preferred Ficus
to be planted as far from the house as possible since its shallow root
system could interfere with the foundation of the house. Farmers would
not castrate cattle during the wet season due to the risk of fly infestation
of the wound. Before starting the castration, an offering was given to
ensure a good outcome.
The team attitude. Each team member had to bear in mind the
following points: (1) their presence was to help solve the farmer's problem
rather than to solve a team problem, (2) the farmer had the knowledge
while the team had the science, and (3) the farmer had to feel that the
team belonged to the farmer group and not the reverse.
GO and NGO participation. Apart from consulting the governmental

organization (GO) regarding the TSFS plans, the team also asked the GO
and nongovernmental organizations (NGOs) to visit the site and give talks
to the farmers. In this manner, the interested GOs and NGOs would help
spread the TSFS concept. Whenever appropriate, newspaper reporters
and TV newscasters were also invited to prepare reports in order to help
other farmers become aware of TSFS technology.
Case Studies 137
Extent of TSFS adoption
Major constraints and possible solutions. According to the experience

gained in the project, when transferring TSFS technology to farmers the
following constraints merit special attention:
Farmers must wait one year for full establishment of the first and
second strata and three years for the establishment of the third stratum;
however, the foregone production of the 0.09 ha cash crop would be
more than compensated for by future TSFS cattle sales;
Drought-resistant grasses and ground legumes are difficult to obtain

when TSFS is being newly introduced to a given area; however, once
these plants are established, they become a source of clones and
seeds for other farmers interested in adopting TSFS technology;
The initial TSFS establishment cost of US$150 to US$200 per plot of

0.25 ha is expensive for the smallholders; however, special bank loans
and cooperative support could be arranged without upsetting the
farmer's daily budget.
Expansion of the TSFS. In the past eight years the number of TSFS
units has increased from 32 to 286; it is worth noting that 56% of these
were due to initiatives by private organizations. When the program started
in 1984, 32 TSFS units were established in South Bali. In 1986, the
provincial Bali government established 30 TSFS units in East Bali (Nitis et
al. 1987). The Daya Pertiwi Foundation established 67 units in Nusa
Penida in 1987-1988 (Nitis et al. 1988b) and 20 additional units in South
Malang, East Java, in 1989-1990. In 1988-1989, the Foster Parent Plan
International Bali established 40 TSFS units in East Bali. In 1989, 52
members of the Selonding farmer group adopted TSFS to ensure forage
supply for their livestock. To study the dissemination of the TSFS results,
IDRC established three units in North Bali in 1988. For the 1991-1992
period, the Board of Animal Husbandry at Jakarta is planning to establish
nine TSFS demonstration plots in nine provinces in Indonesia. IDRC has
approved a study involving cattle and goats associated with food crops,
within TSFS, for the 1991-1993 period. FAO5 has approved a TSFS
demonstration plot for breeding cattle and goats integrated with plantation
crops for the 1991-1992 period.
5 United Nations Food and Agriculture Organization.

Flexibility of the adopted TSFS. Partitioning of the TSFS unit into core,
peripheral, and circumference areas did not create any difficulty for the
farmers since (1) the farmers usually set up live fences around their fields
and (2) they usually cannot plow the outer rim of their fields close to the
fence; therefore, a belt of native grasses is always found between the cash
crop and the fence.
The use of three grass species and three legumes for the first stratum,
three shrub legume species for the second stratum, and three fodder tree
species for the third stratum provided the farmer with ample opportunities
to select the species according to availability and need.
Some farmers made the following modifications: (1) variation of strata

dimensions to suit farm size and shape; (2) wider spacing between trees
and shrubs and narrower grass/legume strips when the land size is more
than 0.25 ha and the circumference more than 200 m; (3) when the TSFS
unit bordered other farmer's property, some farmers preferred to plant
Gliricidia, which makes a solid fence; when the unit was within the farmer's
own land, some farmers preferred to plant Leucaena or Acacia villosa,
which are easily managed due to their finer stems; (4) some farmers
preferred to grow all of the 14 Ficus, 14 Lannea or 14 Hibiscus trees in
one continuous row, so as to use them as shading or for storing crop
residues; other farmers preferred not to grow Ficus close to the house to
prevent its roots from damaging the foundation of the house; and (5) on
sloping land some farmers preferred to grow Cenchrus, while on flat land
some farmers preferred to grow Panicum since Panicum was more
resistant to waterlogging during the wet season. These modifications did
not alter the concept and objective of TSFS.
Sustainability of the 32 TSFS and 32 NTSFS plots. In December 1989

the TSFS and NTSFS plots were returned to full management by the
farmers. A study conducted in June 1991 showed the following: 1) all of
the 32 NTSFS plots had been improved by planting grasses at different
widths and by planting a different number of shrubs and trees along the
fence line; 2) eight of the 20 TSFS plots on sloping land had been modified
by planting Panicum, Urochloa, and/or Cenchrus in the core area, which
had originally been planted with food crops; three of the 12 TSFS plots on
flat land had been modified by pulling up all of the first stratum to within
0.5 m from the fence line, leaving only the second and third strata.
Case Studies 139
LITERATURE CITED
ANONYMOUS. 1986. Buku statistic peternakan. Jakarta, Indonesia,

Direktorat Jenderal Peternakan. 143 p.
ARGA, W.; NURAINI, K., SUTRISNA, I.B.; UTSRIYANA, I.N.G. 1990.

Socioeconomic and communication aspects of Three-Strata Forage
System application in North Bali. Third year supplementary progress
report to IDRC, Canada. Denpasar, Bali, Indonesia, Udayana
University. 118 p.
GOMEZ, K.A.; GOMEZ, A.A. 1976. Statistical procedures for agricultural

research with emphasis on rice. Los Banos, Philippines. International
Rice Research Institute. p. 171-181.
NITIS, I.M. 1984. Three-Strata System for cattle feeds and feeding in
dryland farming area in Bali. First year progress report. Denpasar,
Bali, Indonesia. Udayana University, International Development
Research Centre, Canada. 36 p.
NITIS, I.M.; LANA, K.; SUKANTEN, W.; PUTRA, S. 1987. Demoplot Sistem
Tiga Strata di Seraya Karangasem. Pemda Tk. Propinsi Bali.
I
Denpasar, Indonesia. Udayana University. 28 p.
NITIS, I.M.; LANA, K.; SUARNA, M.; SUKANTEN, W.; PUTRA, S.; ARGA, W.
1988a. Increasing the efficiency of dryland farming area in Bali by
Three-Strata Forage System. Seminar on the Role of ASAIHL
Universities in Transfer of Technology, Jakarta, Indonesia, 6-8
December. 21 p.
NITIS, I.M.; SUKANTEN, W.; PUTRA, S.; NGENTEN, W. 1988b. Demoplot

Sitem Tiga Strata di Nusa Penida Bali. Laporan Evaluasi. Yayasan
Daya Pertiwi, Malang, Indonesia. 32 p.
NITIS, I.M. 1989. Pengembangan lebah dalam konsep Sistem Tiga Strata.
DIKLAT Pembinaan dan Pendirian Pusat Percontohan lebah Asia,
Karangasem. 12 p.
NITIS, I.M.; LANA, K.; SUARNA, M.; SUKANTEN, W.; PUTRA, S.; ARGA, W.
1989. Three-Strata Forage System for cattle feeds and feeding in
dryland farming area in Bali. Final Report. Ottawa, Can., International
Development Research Centre. 252 p.
NURAINI, K.; ARGA, W.; SUTRISNA, I.B.; UTSRIYANA, I.G.N. 1991.

Three-Strata Forage System communication through Humlet courses
in North Bali. Supplementary Progress Report to IDRC, Canada.
Denpasar, Indonesia. Udayana University. 89 p.
SOKAL, R.R.M.; ROLF, F.J. 1969. Biometry. San Francisco, Freeman. p.

617-683.
STEEL, R.G.D.; TORRIE, J.R. 1960. Principles and procedures of

statistics. New York, McGraw-Hill. p. 99-128.
WINAYA, P.D.; NUGARI, K.; OKA, K.M.; SUBANDIYASA, N.; MERIT, M.

1980. Reconnaissance soil map for Bali irrigation project. Denpasar,
Indonesia. Udayana University, Water Resources Reports 1 and 2.
TECHNOLOGY ADOPTION AND
COMPETITIVENESS IN SMALL DAIRY
FARMS IN COSTA RICA: A CASE STUDY
Federico Holmann', Ruben D. Estrada2,

Francisco Romero3 and Luis A. Villegas4,5
ABSTRACT
The objectives of this study were to understand the rationale of

technology adoption and the competitiveness of dairy production systems
in open markets, using as a case study the replications of an intensive
milk-production system prototype established in the tropical humid
lowlands of Costa Rica from 1979 to 1984, as part of an agrarian reform
plan. Data used were from (1) a survey of 44 replications in April and June
of 1991 to analyze biological and economic evolution, and (2) preliminary
evaluations conducted in 1985, along with macroeconomic data of the last
20 years that shed light on trends in policies and prices.
Improved pastures were degraded by soil fertility loss, causing a

reduction in milk yield per hectare, even when the amount of feed
concentrate offered was increased. Net income deteriorated with time,
covering only labor costs calculated at the minimum wage. Additionally,
labor costs systematically increased by 4% a year in real terms for the last
1 Tropical Agricultural Research and Training Center (CATIE), Turrialba, Costa Rica.
2 Consultant, Instituto Colombiano Agropecuario, Bogota, Colombia.
3 CATIE. Presently at the Escuela Centroamericana de Ganaderia, Atenas, Costa Rica.
4 Ministry of Agriculture and Livestock. El Barreal, Heredia, Costa Rica.
5 The authors are grateful to the Regional Office for Latin America and the Caribbean of
the International Development Research Centre for financing this study. We also thank
the Instituto de Desarrollo Agrario, for its cooperation in facilitating technical personnel
to conduct the 1991 survey.
20 years. Thus, with labor contributing 62% to 76% of production costs,

these farming systems based their production on family labor, which was
also underutilized by 24% to 37%.
Costa Rica provided incentives for milk production through subsidized

credits, high milk prices, and import restrictions. These effects generated
a small surplus with few possibilities of competing because milk price at
the farmgate was US$0.29/kg, while in the international market it was
US$0.20/kg. Assuming a price reduction at the international level in an
open market economy, these systems would survive less than four years
if they paid the minimum wage for labor. Intensifying production is not
possible due to soil degradation; thus, the only alternative for sustaining
this system is increasing farm size. Research is needed to facilitate the
transition of these farms to alternative uses and to develop mechanisms to
promote dairy expansion in other ecozones offering comparative
advantages in open markets.
INTRODUCTION
By Latin American standards, Costa Rica historically has consumed high

quantities of dairy products. During the 1970s, annual per capita
consumption of dairy products was 141 kg (8.8% imported). During the
same period, domestic milk production increased 4.1% annually while
human population grew 3.2% per year. In the 1980s, annual per capita
consumption increased to 149 kg (3.9% imported), while domestic milk
production and human population increased 3.5% and 2.6%, respectively
(BCCR 1991a). However, since 1988 Costa Rica has generated a small
milk surplus currently exported to the Central American region at
US$0.65/kg as ultra high temperature (UHT) milk, which does not require
refrigeration (Camara de Productores de Leche 1991). If this tendency
toward surplus continues, Costa Rica may have, in this decade, a
significant surplus that will be difficult to dispose of due to a limited market.
Exporting milk in the traditional powder form is not feasible because price
at the farmgate is US$0.29/kg, while in the international market it is about
US$0.20/kg.
Even with domestic milk production and the existence of adequate farm
infrastructure and a road network to support it, there has been discussion
on the country's potential to compete at the regional level in an economy
without subsidies. In the same way, concern about competitiveness
Case Studies 143
prevails in other countries within the region because the comparative

advantages of different technologies have not been identified, and it is
uncertain whether these technologies can compete in open markets.
The objectives of this study were to understand (1) the reasons why
producers adopt or do not adopt certain technologies; (2) the
competitiveness of milk-production systems in open markets; and (3) the
changes that need to be proposed to make better decisions when
allocating research and production resources to achieve international
competitiveness.
To meet these objectives, the intensive milk-production prototype

develop in the Centro Agronomico Tropical de Investigacion y Ensenanza
(CATIE) (Duarte 1991) was adjusted to the target areas, and the resulting
dairy system was established in the settlements of Sonafluca and Rio Frio
from 1979 to 1981 as part of an agrarian reform plan. Also included were
replications established from 1982 to 1984 by the initiative of other
producers. The performance of the farms where the recommended system
was installed was evaluated through a survey conducted in 1985 and
another (executed by the authors) in 1991.
THE INTENSIVE MILK-PRODUCTION PROTOTYPE
Description
In 1976, CATIE established a prototype of a specialized dairy unit for

farmers with limited resources. Research results generated by a
multidisciplinary group were included in this unit as follows (Pezo et al.
1990):
Intensive management of fertilized pasture. The species used was

African Star grass (Cynodon nlemfuensis), fertilized with 250 kg N/ha/year.
The rest period was 21 days and the stocking rate was 6.5 cows/ha.
Adapted animal genotype. Initially, dairy cows were chosen based on

their production levels, regardless of genotype. Later, efforts were made
to include only Criollo x Jersey crossbreds.
Specialized dairying. Machine milking was practiced twice a day

without calf at foot. Only female calves were kept and were artificially
reared with minimum milk (180 kg during the first 60 days) and feed
concentrate (maximum 1 kg/head/day until five months old).
Minimum use of concentrate feeds. Milking cows were offered 2 kg

molasses/cow/day. Additionally, all animals had free access to a mineral
supplement.
Minimum investment in facilities and equipment. The total unit area was
4.5 ha, of which 4.2 ha was under pasture, 180 m2 for the milking shed,
with the rest including alleys and a reserve paddock.
Intensive labor use. The prototype was managed by one operator, who
did all labor for the unit.
Dissemination of the recommended dairy production system

in the countryside
During 1977, the Instituto de Desarrollo Agrario (IDA), an autonomous

institution involved in agrarian reform programs, signed an agreement with
CATIE for the development of a dairy project. As institutional policy, IDA
allocates land to settlers to be paid off in a 15-year period. However, in
practice, settlers end up acquiring the land for free (thus, land allocation
is subsidized by society).
The initial objective was to validate technology demonstrating the

feasibility of producing milk in family units located in the humid lowland
tropics of Costa Rica. Later, the project was requested to promote milk
production in the area (Villegas 1982); thus, it was redesigned as a dairy
development effort.
The approach used was to introduce an improved alternative (an

adaptation of the CATIE prototype). After three years (in 1982), it was left
to the discretion of producers to continue the technological elements they
considered convenient, without intervention of researchers and with limited
participation of technicians providing technical assistance (Pezo et al.
1990).
Characteristics of the target areas
The project was developed in Rio Frio (100-150 meters above sea level)
and Sonafluca (200-250 meters above sea level); average farm size was 10
ha (Murillo and Navarro 1986).
Case Studies 145
Climatic characteristics in both areas are similar. In Rio Frio, the annual
rainfall is higher than Sonafluca (4,120 mm vs. 3,710 mm), but distribution
is uniform throughout the year. Mean annual temperature is 25°C and 26°C
for Rio Frio and Sonafluca, respectively. The predominant landscape in
both ecozones is flat or flat-convex with concave depressions that produce
floods in some areas (Murillo and Navarro 1986).
Changes in the recommended dairy system
Several changes in the CATIE prototype were necessary in order to

adjust the system to the target areas. These changes (Villegas 1982) were
implemented at the beginning of the project and consisted of:
Pasture species. Since Brachiaria ruziziensis was already established

in both ecozones with a high degree of adaptation and availability of
vegetative material, this species was used instead of C. nlemfuensis.
Farm area. Due to marketing and disease problems affecting the crops
common in the zones, it was decided to allocate the total farm area (10 ha)
to the dairy enterprise.
Fertilization level. Credit was available for acquisition of animals for a

herd smaller than CATIE's prototype (i.e., 15 vs. 23 cows), but pasture area
was twice as large. This factor reduced the need for N fertilization. Also,
soils had been in forest and were high in organic matter. Thus, it was
decided to recommend a decrease in fertilization level to 50 kg N/ha/year
after the fourth year of operation.
Other changes. Due to available family labor, machine milking was

replaced by hand milking. Also, to reduce the risk of mastitis, restricted
suckling of calves was implemented instead of artificial rearing.
DATA COLLECTION
Data were obtained through (1) the review of biological and economic
behavior of the CATIE prototype from 1977 to 1990 (Duarte 1991); (2)
reviews of evaluations of the dissemination and behavior of the
recommended dairy systems during the first five years (CATIE 1981; Murillo
and Navarro 1986); (3) examination of macroeconomic information on
Costa Rica during the last 20 years in order to understand the evolution of
prices and policies in the livestock sector; and (4) a survey, conducted in
April and May of 1991 with 44 participating farmers in both ecozones,
aimed at understanding the performance of their units in the last six years.
The 1991 survey was classified by ecozone (Sonafluca and Rio Frio) and
by start-up period in the dairy business [1979-1981 (first); 1982-1984 (last)].
Sample size included 20 farmers in Sonafluca (7 who initiated activities first
and 13 last) and 24 producers in Rio Frio (12 first and 12 last).
Farm owners were personally interviewed by five technicians trained to

execute the survey. Two technicians from IDA interviewed in Sonafluca
and three from CATIE interviewed in Rio Frio. A second interview was
conducted in both zones to collect additional information.
The survey was designed to determine the performance and changes of

the replications .(farms) in terms of milk and beef yield and input use, in
three time periods (initial 1985, and 1990). The data were used
subsequently to calculate costs of feeding (concentrates, molasses, salt,
and minerals), fertilization, and permanent (family and outside) and
occasional labor; income from the sale of milk and beef was also
calculated. Other costs (e.g., animal health) were not included because
the farm survey was time-exhaustive. Since, the survey focused on only
the evolution of the major costs of production, the actual costs may be
higher than those estimated in this study.
All economic data used (e.g., input and output prices, credit) were
adjusted to 1990 U.S. dollars (US$) using the consumer price index and
exchange rates (BCCR 1991a).
ON-FARM PERFORMANCE OF THE RECOMMENDED DAIRY SYSTEM
The main changes between the initially proposed dairy system and how
they were found to be functioning in the 1985 and the 1991 surveys are
described below:
Operation size. Table 1 shows the changes in the operational size of the
surveyed farms. The greatest increase in total area was by those who
entered dairying earlier (1979-1981) in Rio Frio.
Case Studies 147
Stocking rate. Stocking rate did not change. When there was growth
in herd inventory, it was compensated by an increase in pasture area
(Table 1). In Rio Frio, this process was more dynamic and the behavior of
producers was different, depending on when they had begun dairying. The
first producers increased pasture area through buying of land, and the later
ones by renting land.
Pasture species. Initially, milk production in the replication farms was

based on areas of B. ruziziensis although small areas of other improved
pastures existed, especially C. nlemfuensis (Table 2). After 7-10 years of
use, farms were invaded by ratana pasture (Ischaemum indicum). This
invasion process was especially serious in the first farms in Rio Frio, where
the area in ratana went from 10% to 94 percent. In Sonafluca and the last
farms in Rio Frio, the degree of invasion was about 50 percent.
Fertilization level. The strategy for using fertilizers differed in the two
ecozones. In Sonafluca, the quantity applied was more than the 50 kg of
N/ha/year recommended by CATIE (Villegas 1982), but the fertilized area
was limited to that free of ratana. Therefore, the fertilized area decreased
with time, especially from 1985 to 1990 (Table 2). In Rio Frio, fertilization
levels were generally lower.
Feed supplementation. The only supplementation considered by the

CATIE prototype was 2 kg of molasses per/cow/day. However, in the
replication farms, molasses was offered at an average level of 0.7
kg/cow/day. In addition, the replications have used concentrate feeds
since the beginning and their use has increased with time, reaching levels
of 1.4 kg/cow/day (Table 2).
Labor use. Outside labor represented less than 5% of total labor use.
Family labor assigned to dairy chores increased with time in farms where
ratana represented less than 60% of pasture area by 1990. However, in
the first replications of Rio Frio, where ratana had invaded 94% of the
pasture area by 1990, family labor was lower by 15% with respect to the
start-up period (Table 3), which reflected the low labor level required for
pasture maintenance (weed control, etc.) due to the aggressive growth of
the ratana grass.
Milk yield. Although total milk yield for all groups of farms within the
target areas was relatively constant through time, milk yield per hectare
and per milking cow decreased (Table 4). The proportion of milking cows
stayed at high levels (over 70%), although those who began first in Rio Frio
saw this proportion drop from 78% to 71 percent.
148 Animal Production Systems: Global Workshop,
Income level. Table 5 shows net income with and without considering
family labor costs. In general, net income in real terms, especially per
hectare, deteriorated with time, covering only labor costs (the legal
minimum wage and an attached interest).
Table 1. Land use, stocking rate, and herd inventory of milk producers in the ecozones of
Rio Frio and Sonafluca who began dairying between 1979 and 1981 ('First') versus
those who began after 1981 ('Last)
First Last
Variable Initial 1985 1990 Initial 1985 1990
RIO FRIO
Land use, ha
Pasture 9.5 12.5 14.8 7.1 8.5 11.5
Agriculture 0.6 0.3 1.5 0.6 0.5 0.3
Forest 0.5 0.2 0 2.3 1.1 0
Total farm size 10.6 13.0 16.3 10.0 10.1 11.8
Rented pasture 0 0 1.3 0 0 5.6
Stocking rate, AU1 /ha 1.9 2.0 1.9 2.6 2.3 1.80
Herd inventory, heads

Milking cows 12.2 15.8 15.4 10.5 11.4 17.3
Dry cows 3.4 5.3 6.2 5.0 5.1 6.4
Heifers 0.3 1.4 4.1 2.2 1.5 3.8
Calves 2.4 4.7 9.9 0.6 1.8 4.8
Bulls 1.0 0.8 1.1 0.9 0.6 0.9
Total AU 18.3 25.6 31.3 18.7 19.4 30.3
SONAR.UCA
Land use, ha
Pasture 10.0 10.0 9.6 9.2 9.6 10.0
Agriculture 0.1 1.7 2.2 0.9 2.3 2.3
Forest 0 0 0 0.1 0 0
Total farm size 10.1 11.7 11.8 10.2 11.9 12.3
Rented pastures 0 0 1.0 0 0 1.2
Stocking rate, AU/ha 2.3 2.8 2.9 2.5 2.7 2.6
Herd inventory, heads

Milking cows 15.0 16.4 15.3 15.0 16.5 16.3
Dry cows 4.6 4.7 4.4 3.6 4.3 5.5
Heifers 0.9 2.9 3.6 1.5 1.8 4.2
Calves 3.7 7.3 14.4 2.1 3.5 6.6
Bulls 1.0 1.0 1.1 1.0 0.8 0.8
Total AU 23.4 28.4 31.2 22.2 25.0 29.5
Cows=1, heifers=0.8, calves=0.5, and bulls=1.3 AU (animal units).

Case Studies 149
Table 2. Area, species, pasture fertilization, and supplementation utilized by milk producers
in the ecozones of Rio Frio and Sonafluca who began dairying between 1979 and
1981 ('First) versus those who began after 1981 ('Last-).
First Last
PoO FRO
Pasture area, ha
Ratana 1.0 9.2 14.1 1.7 3.9 7.4
Cut-and-carry 0.1 0.2 0.1 0.1 0.2 0.4
Brachiaria 6.8 2.5 0.4 4.0 4.1 3.6
Other improved 1.7 0.8 0.3 1.5 0.8 0.7
Area in ratana, % 10.4 72.4 94.6 23.3 43.3 61.2
Nitrogen
fertilization
Area, ha 2.0 1.4 1.0 1.1 1.0 3.1
Quantity, kg/ha/year 15.3 21.7 15.3 37.3 15.9 82.2
Supplementation,
kg/cow/day
Concentrate 1.2 0.9 1.3 0.8 1.1 1.5
Molasses 0.8 0.7 0.3 0.3 0.5 0.8
SONAFLUCA
Pasture area, ha
Ratana 0 3.9 4.5 0.2 1.8 5.4
Cut-and-carry 0 0.6 0.7 0.2 0.7 1.2
Brachiaria 9.3 5.2 0.9 5.6 4.3 1.1
Other improved 0.7 0 3.6 3.0 2.7 2.1
Area in ratana, % 0 40.2 46.4 2.2 18.9 55.1
Nitrogen
fertilization
Area, ha 6.5 4.5 1.9 4.8 4.8 2.2
Quantity, kg/ha/year 59.7 77.0 85.0 67.3 64.3 125.7
Supplementation,
kg/cow/day
Concentrate 0.2 0.8 1.1 0.8 0.9 1.6
Molasses 0.6 0.8 0.6 0.9 0.9 0.7
Table 3. Labor utilization by milk producers in the ecozones of No Frio and Sonafluca
who began dairying between 1979 and 1981 ('First) versus those who began
after 1981 ('Last'). Standard deviations are in parenthesis
First Last
RIO FRIO
Labor use, person

days/year
Family 45 54 93 03 96 28
(456) (325) (208) (376) (358) (361)
Outside 34 76 67 98 89 97
(44) (74) (72) (83) (93) (116)
Total 779 730 660 901 885 1025
SONAFL UCA
Labor use, person

days/year
Family 08 13 069 99 83 123

(467) (434) (391) (562) (558) (478)
Outside 52 52 66 74 74 83
(68) (39) (69) (81) (72) (89)
Total 860 965 1135 973 1057 1206

Case Studies 151
Table 4. Daily milk production per farm, per hectare, and per milking cow, and percentage
of milking cows of producers in the ecozones of No Frio and Sonafluca who
began dairying between 1979 and 1981 ('First') versus those who began after 1981
('Last'). Standard deviations are in parenthesis
First Last
NO FFOO
Daily milk
production, kg
Total 9.6 07.1 7.9 4.9 0.0 21.7

(56.3) (41.8) (55.7) (27.8) (42.9) (67.7)
Milking cow 7.3 6.8 6.4 6.2 7.9 7.0

(4.6) (2.7) (3.6) (2.7) (3.8) (3.9)
Hectare 9.4 8.6 6.6 9.1 10.6 10.6

(5.9) (3.4) (3.8) (3.9) (5.1) (5.9)
Milking cows, % of
total cows 78.2 74.9 71.3 67.8 69.1 73.0
SONAFLUCA
Daily milk
production, kg
Total 28.6 18.6 15.7 35.4 42.3 20.0

(41.4) (61.8) (33.6) (66.4) (56.6) (40.6)
Milking cow 8.6 7.2 7.6 9.0 8.6 7.4

(2.8) (3.8) (2.2) (4.4) (3.4) (2.5)
Hectare 12.9 11.9 12.0 14.7 14.8 12.0

(4.2) (6.2) (3.5) (7.2) (5.9) (4.8)
Milking cows, % of
total cows 76.5 77.7 77.7 80.6 79.3 74.8
Table 5. Annual net income (including and excluding family labor valued as minimum
wage) of producers in the ecozones of No Frio and Sonafluca who began
dairying between 1979 and 1981 ('First) versus those who began after 1981
(`Last'), in 1990 dollars. Standard deviations are in parenthesis
First Last
RIO FRIO
Annual net income1
US$/farm 8,712 9,607 7,078 6,191 7,667 9,027

(7,213) (4,719) (6,343) (3,079) (4,312) (8,050)
US$/hectare 1,044 814 440 872 902 528
(990) (420) (561) (395) (768) (376)
Annual net income2
US$/farm 3,950 4,793 2,454 1,800

(5,865) (4,080) (3,397) (4,408) (4,825) (5,161)
US$/hectare 481 438. 152 157 212 105
(657) (335) (205) (832) (591) (291)
SONAFLUCA
Annual net income
US$/farm 14,721 10,668 9,952 14,051 12,923 9,182

(5,281) (7,062) (6,173) (7,651) (6,172) (7,983)
US$/hectare 1,472 1,067 939 1,527 1,346 820
(481) (620) (457) (862) (749) (989)
Annual net income2
US$/farm 9,993 3,952 1,622 8,415 5,690 433

(4,922) (8,521) (4,278) (8,222) (6,606) (4,540)
US$/hectare 999 395 153 915 593 39
(447) (792) (429) (989) (801) (502)
Excludes family labor.
z
Includes all available labor calculated with the minimum wage (in 1990, US$5.41/day plus 44% of social
benefits, equivalent to a total of US$7.79/day).
Case Studies 153
EXPLANATION OF PRODUCERS' BEHAVIOR
As previously shown, there were several differences between the

proposed dairy system and the alternatives farmers adopted. These
differences are related to the integration of biological and microeconomic
factors at the farm level as well as to macroeconomic influences. The
factors that have most influenced the producers' decisions are:
Biological factors: Soil type and pasture species
The main results of the soil analysis performed in 1991 are shown in
Table 6. Even though ecozones with similar characteristics are mentioned
in the IDA base document (CATIE 1981), soil conditions differ substantially.
Initially, P content differed greatly between farms in Rio Frio and Sonafluca
(Murillo and Navarro 1986), giving an idea of the original soil material when
producers acquired the land. With time, P content has reached similar
levels, but soil fixation levels in the last analysis indicate two soils with
different development possibilities (Table 6): on one hand, a manageable
level (67%) in Sonafluca, and on the other hand, almost impossible to
correct (92%) in Rio Frio. This is due mainly to the content of other
nutrients in the soils, especially calcium availability, which is higher in
Sonafluca.
Further, the low P content has stimulated the invasion of ratana.

Arosemena (1990) reported an internal P requirement in ratana of 0.15%
(based on the percentage found in tissue) vs. 0.19% in Brachiaria
brizantha, a species known to require less P than B. ruziziensis (CIAT
1987).
Moreover, ratana competes favorably with other species in terms of

speed of ground cover. Arosemena (1990) reported a 28% reduction of
the relative weight of B. brizantha biomass, with respect to ratana when
these competed for light, space, and nutrients. Therefore, ratana not only
requires less P but also outcompetes B. brizantha, suggesting ratana may
also be more aggressive under these soil conditions than B. ruziziensis,
which has an even higher P requirement.
Table 6. Soil analysis at 20 cm depth for farms in No Frio, Sonafluca, and the CATIE
intensive milk production system prototype. Standard deviations are in
parenthesis
Characteristic Rio Frio Sonafluca CATIE'

(n=22) (n=20) (n=3)
pH 5.5 6.0 5.5

(0.3) (0.3) (0.1)
Calcium2 3.8 10.8 9.9

(3.2) (2.7) (2.0)
Magnesium2 1.2 3.0 4.5

(0.8) (0.5) (0.5)
Potassium2 0.3 0.4 0.5

(0.1) (0.1) (0.2)
Extracted acid2 0.4 0.2 0.4

(0.1) (0.1) (0.1)
Phosphorus3 2.8 2.9 7.8

(0.9) (1.5) (2.0)
Phosphorus fixation, % 92.1 67.4 65.0

(7.5) (4.1) (1.3)
1 CATIE (1991b).
2 meq/100 ml soil.
3 mg/l soil. Phosphorus content in same farms in 1983 was 3.5 in Rio Frio and 4.2 in
Sonafluca (Murillo and Navarro 1986).
In Rio Frio, the problem of ratana invasion was exacerbated by the

producers themselves, who initially saw an adequate vegetative behavior
in ratana (Murillo and Navarro 1986). However, by the time they realized
ratana's limited productivity, it was almost impossible to reverse the
invasion. Ratana's aggressiveness can be observed with the producers
that last began dairying in 1982; even though they were more careful in the
management of pasture (Table 2), there was still a trend toward greater
proportions of ratana. In Sonafluca, where the soil is more fertile, this
process was slower, given the lower P fixation levels in the soils, it is a
problem that can be prevented economically. Thus, at the farm level, soil
characteristics and their effects on pastures have influenced the adoption
and non-adoption of the following technologies that had been
recommended by CATIE:
Case Studies 155
Fertilization levels. In less fertile soils and with ratana invasion, the
probability that producers would obtain a response to N fertilization was
low. For this reason, N was used strategically only in areas with cut-and-
carry forage located next to the milking shed, where manure is also
dumped.
When producers began dairying between 1979 and 1984, the average
milk price was at its highest, considering prices from 1970 to 1990, as seen
in Table 7 (Camara Nacional de Productores de Leche 1991). This allowed
a good price relationship with respect to nitrogen, that is, 2 kg of milk
would allow the purchase of 1 kg of N (FERTICA 1991). With the
petroleum crisis in 1981-1982, N prices increased, while the milk price
stayed relatively stable during five years; thus, the milk/nitrogen
relationship deteriorated to 2.8 kg milk:1 kg N, making milk production
based on unfertilized pastures more attractive. This short-term effect and
an increase in labor costs (Table 8; MTSS 1991) could have induced
producers to use pastures with low N (and labor) requirements (e.g.,
ratana). More recently, this relationship has improved to 1.7 kg milk:1 kg
N, and farmers have increased the amount of N applied per hectare,
although its use has been strategically limited to small areas free of ratana.
Table 7. Producer prices per kilogram of milk in 1990 dollars for the
period 1970-1990
Price Price
Year US$/kg Year US$/kg
1970 0.27 1980 0.33

1971 0.26 1981 0.34
1972 0.25 1982 0.33
1973 0.28 1983 0.35
1974 0.27 1984 0.32
1975 0.33 1985 0.30
1976 0.33 1986 0.28
1977 0.31 1987 0.28
1978 0.31 1988 0.27
1979 0.31 1989 0.29
1990 0.29
Table 8. Daily wages for the livestock sector in 1990 US$ (including
social benefits) for the period 1970-1990
Wage Wage
Year US$/day Year US$/day
1970 4.75 1980 6.48
1971 5.00 1981 5.85
1972 5.00 1982 5.65
1973 4.82 1983 6.54
1974 4.87 1984 6.88
1975 4.97 1985 7.36
1976 5.24 1986 7.80
1977 5.69 1987 7.79
1978 6.32 1988 7.56
1979 6.48 1989 7.80

1990 7.79
Supplementation levels. CATIE's prototype was designed to use

molasses (i.e., digestible energy) in a strategic manner to complement the
protein furnished by C. nlemfuensis. With ratana in pastures having low
dry matter and protein contents, dairy cows required concentrate feeds to
ingest adequate amounts of energy and protein to achieve production
levels comparable to those obtained with other species such as B.
ruziziensis. With time, the increasing percentage of ratana in pastures
induced producers to use greater quantities of concentrate feeds. Their
decision to do so was reinforced by the price relationship between
concentrate feeds and molasses, especially in the last five years (1 kg
concentrate = 4.5 kg molasses) relative to the period 1979-1984 (1 kg
concentrate = 6.5 kg molasses) as demonstrated by data published by
CATIE (1991a).
Case Studies 157
Macroeconomic factors: Government incentives
The period from 1970 to 1980 was favorable for beef development. High
international prices stimulated beef exports to the United States and
increased export earnings. In the middle of the 1970s, understanding that
the generation of export earnings from beef and the substitution of dairy
imports required preferential conditions, the government began to stimulate
beef and milk production. Costa Rica went from exporting US$92.4 million
worth of beef in 1973 (US$4.55/kg, in 1990 US$) to US$148.1 million in
1979 (US$4.69/kg) according to the BCCR (1989). High export earnings
from beef stimulated high milk prices, while increased milk production
satisfied domestic consumption requirements without the need to import
milk. Additionally, in 1979 the government restricted dairy imports (PNUD
1979), stimulating the sector by protecting the market from dumping.
This scenario has substantially changed in the beef sector since 1981.
Due to changes in consumption patterns in the United States, beef prices
began to decrease in real terms and the livestock sector showed atypical
behavior with respect to traditional production and price cycles. Thus,
Costa Rica went from exporting beef at US$4.69/kg in 1979 to exporting
it at US$2.51 /kg in 1989 (in 1990 US$). The main consequences of these
changes have been the following:
Subsidized interest rates. For livestock credit allocated between 1970

and 1983, average real interest rates were negative, and in some years
annual rates exceeded -10 percent. This was especially true for those
credits given between 1973-1975 and 1979-1983 (BCCR 1991 b). Since
then, Costa Rica has adjusted its nominal interest rates based on inflation,
generating annual real rates close to 10 percent.
Increase in amount of credit. From 1970 to 1980, the amount of credit

allocated to the livestock sector grew at a rate of 10% a year (from US$1 00
to 280 million/year); however, since 1981 credit has decreased to levels
that in 1989 were similar to those allocated in 1970, that is, US$102
million/year (BCCR 1991b).
Investment in infrastructure. Significant investments were made with

public funds. New roads were built and electricity provided. The road
network tripled from 2,557 km in 1974 (67% paved) to 7,227 km in 1990
(50% paved), representing an annual increase of 10.4% (MOPT 1991).
These investments allowed for a rapid change from beef to dairy and
dual-purpose activities. Between 1973 and 1984, the inventory of females
two years old or older decreased 2.4% per year in beef herds, but
increased 9% per year in specialized dairy herds and 19.2% a year in dual-
purpose herds (Censo Agropecuario 1973, 1984). These investments also
allowed producers to capture benefits created with public funds through
land appreciation and reduced transport costs.
Microeconomic factors: Effect of government incentives

at the farm level
Government incentives have had the following impacts at the farm level:
Subsidized credit. Table 9 shows the amount of credit allocated to and

subsidies received by producers due to negative real interest rates
according to the year they began dairying. As shown, average net subsidy
per producer varied from US$149 to US$11,133 depending on the start-up
year. If the public credit institutions had charged a 10% real interest rate,
these subsidies would have been in the order of US$6,200 to US$18,300.
These benefits contributed positively to the government policy of
maintaining milk prices while labor wages (Table 8) continued to increase.
Table 9. Credit and subsidies received by each producers in FPoo Frio and Sonafluca due
to negative real interest rates (as occurred) and the amounts that credit institutions
could have recovered if the interest rate had been 10% (as could have occurred)
As occurred As could have occurred
Beginning Credit Credit % credit Credit % credit

year given paid Subsidy paid to pay Subsidy to pay
................... US$1 ................... ........... uss$'...........
1979 16,745 8,418 8,327 50.3 23,092 14,674 137.9

1980 18,882 7,749 11,133 41.0 26,039 18,290 137.9
1981 7,586 3,387 4,199 44.6 10,462 7,075 137.9
1982 11,094 8,649 2,245 78.0 16,495 7,846 148.7
1983 12,430 12,281 149 98.8 18,481 6,200 148.7
1984 10,287 8,995 1,292 87.4 15,295 6,300 148.7
1 Unadjusted dollars.
The increase in farm size (Table 1) is explained by the benefits received

as well as by the early biological and economic performance of the
replications. Table 10 contains the regression coeff icients that explain this
Case Studies 159
increase in Rio Frio, which expanded the most. The factors with greatest
importance were years in business (expressed in quadratic form) and areas
allocated to activities other than dairy or agriculture (i.e., forest and/or idle
land). Producers who began dairying earliest with the entire farm area
allocated to dairying had higher credits for acquisition of animals and
benefited most from subsidized interest rates and high milk prices. In Rio
Frio, increases in area occurred before and during the opening of the
highway and electrification in 1986, improvements which also favored those
who began dairy activities first (1979-1981) as, based on the subsidies
received, they acquired land before it appreciated. Those producers with
the most productive cows, and therefore higher use of feed concentrates,
accumulated less capital and limited their increase in area.
Table 10. Estimated value, standard error, and significance level of analysis of variance of
factors responsible for the expansion of farm size in Rio Frio
Estimated Standard Significance

Factor value (Bid error level
Beginning year in dairying (X1) 0.0762 0.0359 0.0488
Idle land (X2) -1.3718 0.5453 0.0222
Consumption of concentrate/cow (X3) -0.4625 0.1519 0.0073
Milk production per cow (X4) -0.0349 0.0268 0.2105
' The mathematical model was Y= B,X, + B3X3 + 84X, with regression constant of 6.0239 with R2=
0.57 and mean square error= 123 with four degrees of freedom for regression and 17 degrees for the
residual. Y = Expansion of farm size, ha.
Capture of benefits. The benefits captured by producers can be seen in Table 11. As
shown, Rio Frio has captured greater benefits through land appreciation, given the initial
level of infrastructure. Also presented in Table 11 is the net present value of savings by
producers from milk transport costs due to the highway construction and installation of
cooling tanks (made possible by the availability of electricity). In Rio Frio, the increase in
land value was 84%, reflecting investments of public funds not recaptured by the
government through tax appreciation. In the case of Sonafluca, the benefits due to public
fund investments were 29% lower because the ecozone already had electricity and paved
roads by the time producers began dairy activities, and also because dairies were located
closer to a milking plant. Considering the credit subsidy and the benefit captured by land
appreciation, each producer directly received between US$31,051 for those who initiated last
in Rio Frio and US$41,887 for those who initiated first in Sonafluca (Table 11).
Table 11. Estimated value of land in 1990 US$, and net present value of benefits captured
by producers in No Frio and Sonafluca through milk transport costs
Rio Frio Sonafluca
Commercial value of land, US$/ha

1980 350 1,100
1990 3,000 3,500
Year of construction
Paved road 1,986 1,979
Electricity 1,986 1,979
Benefit from milk transport costs

Due to paved road' 1,340 0
Due to electricity2 893 688
Total 2,233 688
Total value of subsidy3

First 36,887 41,887
Last 31,051 36,051
' Calculations based on the formula: (KML, * CT * Cm * PL) - (KMCh * CT * PL) / IR / ha

where:
KMLj = Distance in kilometers of dirt road between farm and processing plant, round
trip (200 km for Rio Frio)
CT = Factor for transport cost per kg of milk per km (i.e., 0.000168)
Cm = Overcharging factor for road under bad condition (i.e., 20%)
PL = Annual milk production per farm
KMCh = Distance in kilometers on paved road between farm and processing plant,
round trip (120 km for Rio Frio and 70 km for Sonafluca)
IR = Real interest rate (5%)
ha = Total farm size in hectares
2 Calculations similar to above formula, but dividing final product by a factor of 3, since
milk truck now collects milk every three days and not daily as before, when milk tanks
were not available.
3 Calculated from the summation of credit subsidy, as indicated in Table 9, and land
appreciation (US$30,000/farm for Rio Frio and US$35,000/farm for Sonafluca).
Case Studies 161
FUTURE CHALLENGES
In the two ecozones considered, society invested scarce public

resources in land and infrastructure, expecting a return through higher
production and lower consumer prices. Even though the invested amounts
were adequate to stimulate production, the response of the system was
poor and decreased with time. The analysis of this case study indicates
that the main issues to face are:
Restructuring of the sector
Traditionally, milk production systems depend largely on the use of

family labor, which in 1990 accounted for 62% to 76% of the production
costs considered in this study. Wage rates increased in real terms from
1970 to 1990, averaging 4% per year (Table 8); including social benefits
that producers have to pay, the average minimum wage for the livestock
sector in 1990 was US$7.79/day, based on US$5.41/day plus 44% of
social benefits (MTSS 1991). Thus, if labor wages continue to increase, the
production systems with small areas (10 ha) may not be competitive if they
are to (1) have incomes to meet the minimum wage; (2) reduce production
costs to equal those resulting from international protection against imports,
currently at about US$0.20/kg milk; and (3) compete in an economy
without subsidies. Restructuring of the sector to combat these problems
will require an analysis of the following factors:
Producers. These producers are on average 47 years old. Because of

the dynamics of their systems in the last decade, they have:
Received greater benefits through subsidies (land appreciation and

interest rates) than through their efficiency of milk production. That is,
social investment in the dairies is not in proportion to their production
efficiency. As shown in Table 12, the current systems would not be
viable if society would ask them to return the subsidies given to them in
a 10-year period with a real annual interest rate of 10 percent. Under
these circumstances, their net income, based on real labor efficiency
and excluding family labor, would fluctuate between 33% and 60% of the
1990 minimum wage.
Given their small operational size and constant milk prices, these
systems have survived only because they have received subsidies. This
would not be the case in an open economy, without subsidies and with
mechanisms to recapture investments of public funds.
Table 12. Net income per farm (excluding family labor); subsidy
payment if society would ask each producer to return it;
resulting annual net income (excluding family labor); and
family labor cost valued as minimum wage
Variable First' Last'
Rio Frio
a. Annual net income2 7,078 9,027

b. Subsidy payment3 5,533 4,658
c. Net income after subsidy
payment2 (a-b) 1,545 4,369
d. Family labor cost valued
as minimum wage 4,620 7,227
Sonafluca
a. Annual net income2 9,952 9,182

b. Subsidy payment3 6,283 5,408
c. Net income after subsidy
payment2 (a-b) 3,717 3,774
d. Family labor cost valued
as minimum wage 8,327 8,749
' "First' denoted producers in No Frio and Sonafluca who began dairying between 1979
and 1981. "Last" denoted those that began after 1981.
2 Excluding family labor.
3 Calculated from total value of subsidy received (Table 11), payable in a 10-year period
at a 10% real annual interest rate.
Case Studies 163
Low efficiency in the use of family labor. Table 13 shows the

percentage of underutilized family labor based on producers' criteria.
Family labor efficiency can be easily increased by 24% to 37% without
the incorporation of additional labor and by 67% to 99% with
incorporation of small equipment.
In an economic system where a reduction of production costs is

necessary to compete, this could be a limiting factor, given that family
labor is a fixed cost (without possibility of sporadic use) and,
additionally, is the most important part of production costs (62% to
76%), the only one that has been systematically increasing in the last 20
years and will continue to increase in a development process.
Table 13. Percentage of underutilized labor, with either hand or

machine milking, based on the criteria' of producers in Rio
Frio and Sonafluca who began dairying between 1979-81
("First") versus those who began after 1981 ("Last")
Hand milking Machine milking
First
Sonafluca 29 74
Rio Frio 34 99
Last
Sonafluca 37 67
Rio Frio 24 80
Based on additional milking cows that the family could manage with current available
labor allocated to dairying
Operation size. With an open market economy, a situation could exist

where the milk price is reduced to levels of international protection from
imports (i.e., US$0.20/kg). This would imply a 30% reduction in the
current milk price in Costa Rica (Table 7). An analysis was done with an
electronic spreadsheet to determine the number of years producers could
subsist under such conditions while receiving minimum wage for their
family labor. The parameters used were conservative with respect to what
could happen (a 5% opportunity cost on capital in animals and equipment,
4% annual increase in real terms in labor wages, and a 10% annual
reduction of milk price in real terms until it is stabilized at US$0.20/kg).
Even under these conditions, the current production systems would survive
less than four years, demonstrating the need to increase family labor
efficiency.
Labor efficiency can be improved by increasing land productivity or

operation size. The first alternative has not demonstrated attractive results,
since the replacement of ratana by more efficient forage species would
only increase the viability of the system by one to three additional years.
In this analysis, the area in ratana was replaced with B. ruziziensis, the
species used at the beginning, assuming this would last four years before
degrading again into ratana. Also, based on producers' opinions, it was
assumed that one hectare of B. ruziziensis without fertilization increased
stocking rate by 33% and produced an additional 1.5 kg of milk/cow/day
in relation to ratana.
With a small operation size (10 ha), the cost of labor (US$7.79/day) and
the cost of establishing new pastures (US$350/ha), it is impossible to
produce milk at US$0.20/kg. This suggests that the only alternative for
reducing production costs under the .current scenario is to increase farm
size to above 20 hectares. Reducing family labor is not viable because, in
the majority of cases, family labor represents labor out of the normal job
market (senior citizens or young people of school age), whose efficiency
in many activities does not differ from that of a working adult.
Contribution to development. The objective of IDA has been to provide

employment in rural areas by allocating land to settlers, thus avoiding
migration to main urban centers. In this case study, producers were
receiving a government subsidy of about US$796/family/year at the time
the project was initiated. These public fund investments in land and
subsidized credit have drastically increased their standard of living. Thus,
these subsidies represent the cost that society has paid to generate
employment opportunities through milk production by small producers in
the rural sector.
The dilemma now is that the current production systems are not
economically viable with current labor efficiency and cost. Thus,
alternatives need to be developed to facilitate the transition of land under
these soil conditions to other uses.
Case Studies 165
Research needs
High milk prices, subsidized credit, and public fund investments in

infrastructure have increased the market value of land. These factors have,
in part, contributed to the deforestation process, at a rate that in Costa
Rica during the 1980s was about 500 km2 per year. Thus, in the last 10
years, about 10% of the total area of the country was deforested (CCAD
1991). The solution to this problem appears to include decreasing the
price of milk, reducing public fund investments for infrastructure in fragile
soils, and developing alternative uses for farms located in areas with the
soil characteristics of Sonafluca and Rio Frio.
CONCLUSIONS
Dairy development with small producers in the ecozones considered in

this study would not have been possible without the investment of
subsidized public funds. With realistic interest rates, and with the use of
family labor, these producers would have been receiving the equivalent of
33% to 60% of the minimum wage in 1990. Thus, this dairy development
model can no longer be justified due to increasing labor wages, market
restrictions (milk surplus) at current prices, and conflict with natural
resource conservation goals.
The current capitalization that producers have is superior to that initially

foreseen, even though their productivity was inferior to the level planned
in the initial proposal (CATIE 1981). About 84% and 29% of the
capitalization in Rio Frio and Sonafluca, respectively, was due to public
fund investments captured by producers.
Dairy farmers responded rationally to biological and economic

conditions. Differences with respect to the proposed model were caused
by the rigid character of recommendations in relation to a dynamic
economy (i.e., relative changes in input and output prices, interest rates,
credit availability) and changing scenarios of biological conditions (soil
degradation, nutrient loss).
Without subsidies, given the small size of operations, and with the
existing labor efficiency, protection from dairy imports does not seem
feasible if labor pay similar to the minimum wage is to be maintained, even
with increases of 4% per year in real terms.
Contrary to traditional situations elsewhere, in this case study credit was

not a limiting factor in technology adoption. Producers who accumulated
capital used it to expand farm size instead of increasing productivity. For
the current capitalization levels, this was a logical decision.
The subsidy per hectare fluctuated between US$3,100 and US$4,200

(Table 11). With these investment levels, the government should be more
careful in selecting land for agricultural development plans. The system
should generate US$310/ha to US$420/ha (e.g., 3-4 MT of corn at
international prices) just to pay the interest on capital investment. This
would only be feasible on fertile soils and with lower labor rates, not the
case in Costa Rica.
The objective of farming systems research is to propose small changes

in the system, assuming stable macroeconomic conditions. The results of
this study indicated that these conditions existed at the time the project
began. During the period of the study, great adjustments were
implemented, permitting the same production system to capture from
US$149 to US$11,133 (Table 9), based only on changing interest rates
within a three-year period. Thus, this experience suggests the need for a
permanent feedback at all hierarchical levels to evaluate changes in the
production systems. With an open market economy, research systems will
need more effective multidisciplinary teams to continuously analyze the
influence of macroeconomic aspects.
LITERATURE CITED
AROSEMENA, E. 1990. Determinaci6n de mecanismos de interferencia

por alelopatia y requerimientos externos e internos de f6sforo en pasto
ratana. Turrialba, C.R., CATIE. M.S. Thesis. 87 p.
BCCR (BANCO CENTRAL DE COSTA RICA). 1989. Principales

estadfsticas sobre las transacciones internacionales de Costa Rica. San
Jose, C.R. BCCR, Division Monetaria, Secci6n Balanza de Pagos. 43 p.
Case Studies 167
BCCR. 1991 a. Datos estadisticos. San Jose, C.R. BCCR, Departamento

de Contabilidad Social, Seccion Indices y Estadisticas. Unpublished
data.
BCCR. 1991b. Credito y cuentas bancarias 1970-1990. San Jose, C.R.

BCCR, Division Monetaria. Unplublished data.
CAMARA NACIONAL DE PRODUCTORES DE LECHE. 1991. Datos

estadisticos. San Jose, C.R. Unpublished data.
CATIE (TROPICAL AGRICULTURAL RESEARCH AND TRAINING CENTER).

1981. Fomento a la produccion de leche en las colonias del ITCO:
Informe final 1977-1980. Turrialba, C.R. CATIE. Serie Institucional,
Informe de Progreso No. 14. 39 p.
CATIE. 1991 a. Livestock farm records. Turrialba, C.R., CATIE,

Department of Animal Production. Unpublished data.
CATIE. 1991b. Soil laboratory records. Turrialba, C.R., CATIE,

Sustainable Agriculture Program. Unpublished data.
CCAD (COMISION CENTROAMERICANA DE AMBIENTE Y DESARROLLO).

1991. Plan de accion forestal tropical para Centroamerica. San Jose,
C.R., CCAD. 37 p.
CIAT (INTERNATIONAL CENTER FOR TROPICAL AGRICULTURE). 1987.

Informe anual 1987. Programa de Pastos Tropicales. Cali, Col., CIAT.
irr. p.
DIRECCION GENERAL DE ESTADISTICAS Y CENSOS. 1975. Censo

Agropecuario 1973. San Jose, C.R., Direccion General de Estadisticas
y Censos. 373 p.
DIRECCION GENERAL DE ESTADISTICAS Y CENSOS. 1987. Censo

Agropecuario 1984. San Jose, C.R., Direccion General de Estadisticas
y Censos. 67 p.
DUARTE, O. 1991. Evaluacion biologica y econbmica de un modulo

lechero. Turrialba, C.R., CATIE, M.S. Thesis. 114 p.
FERTICA (FERTILIZANTES DE CENTRO AMERICA). 1991. Datos

estadisticos. San Jose, C.R., FERTICA, Departamento de Ventas.
Unpublished data.
MOPT (MINISTERIO DE OBRAS PUBLICAS Y TRANSPORTES). 1991.

Datos estadisticos. San Jose, C.R., MOPT, Departamento de Estudios
Basicos, Direccion de Planificacion. Unpublished data.
MTSS (MINISTERIO DEL TRABAJO Y SEGURIDAD SOCIAL). 1991.

Consejo nacional de salarios 1969-1990. San Jose, C.R. MTSS.
Unpublished data.
MURILLO, O; NAVARRO, L.A. 1986. Validacion de prototipos de

produccign de leche en la zona atlantica de Costa Rica. Turrialba, C.R.,
CATIE, Serie Tecnica, Informe Tecnico No. 90. 97 p.
PEZO, D.A.; VILLEGAS, L.A.; ROMERO, F. 1990. Modulos lecheros

ITCO/CATIE: Una experiencia de adaptacion tecnologica a nivel de
finca. In Investigacion con pasturas en fincas. Cali, Col., CIAT-RIEPT.
Documento de Trabajo No. 124. p. 203-214.
PNUD (PROGRAMA DE LAS NACIONES UNIDAS PARA EL

DESARROLLO). 1979. Situacion y perspectivas del sector lechero en
Costa Rica. San Jose, C.R., PNUD/OIT/ITCO. Serie Estudios No. 43,
Proyecto COS/72/018 "Empleo y Desarrollo Rural". 88 p.
VILLEGAS, L.A. 1982. Implementacion del sistema CATIE para la

produccion de leche. In Sistemas de produccion con bovinos en el
tropico americano. L. Pearson de Vaccaro (Ed.). Maracay, Venezuela,
Universidad Central de Venezuela. p. 75-88.
DUAL-PURPOSE CATTLE PRODUCTION IN
GUATEMALA: A CASE STUDY
Hugo E. Vargas2,Ruben Dario Estrada3

and Edgar Fernando Navas4
INTRODUCTION
The purpose of this paper is to present the results of a study on the

development of dual-purpose cattle production systems in Guatemala, with
particular reference to the impact of policies on the adoption of livestock
production technologies. This study was undertaken by the Latin American
Research Network for Animal Production Systems (RISPAL) with the
support of the International Development Research Centre (IDRC).
This study analyzes the influence of the macroeconomic framework on

dual-purpose cattle production and on the farm system. It also analyzes
the interactions between policies reflecting national and institutional
objectives and existing production systems, in order to optimize the use
of available resources in research, validation and dissemination of animal
production technology.
Until now, researchers considered microeconomic analysis at the farm

level sufficient to direct their research priorities. In a setting where policies
were mainly determined domestically and were relatively stable, the implicit
1
Abridged version of the report: Desarrollo de la ganaderia bovina de doble prop6sito
en parcelamientos de la costa sur de Guatemala, presented at the Animal Production
Systems Global Workshop, San Jose, Costa Rica, 1991
2 Animal Production Specialist, IICA, Box 1815, 01901 Guatemala, Guatemala.
3 Consultant, Instituto Colombiano Agropecuario, Bogota, Colombia.
4 Agricultural Economist, IICA, Box 1815, 01901 Guatemala, Guatemala.

assumption was that the policy framework was a constant factor not
requiring explicit analysis. However, the globalization and reduction of
trade barriers that is taking place at regional and international levels is
causing such great changes in relative prices that it becomes essential to
take into account the influence of policies, regulations and the increasing
environmental restrictions in order to understand the evolution of farming
systems and the role of technical change.
Therefore, the objectives of this study were to describe the recent

evolution of the production systems and-to understand the adoption and
impact of the technology introduced in the farms of Nueva Concepcion,
Guatemala. The main macroeconomic variables affecting the supply and
demand of milk in the country were analyzed in order to understand their
impact on the development of Guatemala's cattle production.
The activities undertaken included the review of research projects

related to the improvement of dairy and beef production in the region,
analysis of macroeconomic information of the last 20 years, and the
implementation and analysis of a survey of farmers in the Nueva
Concepcion settlement.
BACKGROUND
In Guatemala,the animal production sector contributes 7.5% of the

GNP and employs 15% of the economically active population. Cattle
production is one of the main activities of the sector. It is concentrated in
the southwest and southeast regions (beef and dual-purpose production
systems), and in the central highlands (specialized dairy production). From
1985 to 1989, cattle stocks increased at an annual rate of 2.2%, increasing
from 2.3 to 2.5 million head. This growth has been accompanied by a
change in structure. In recent years, the stocks of dairy cattle have
decreased while those of dual-purpose cattle have increased.
From 1970 to 1988, the annual per capita consumption of milk averaged
38 liters. This consumption level has been maintained due to the
importation of milk, which has quadrupled during this period (from 27 to
87 million liters). This level of per capita consumption is below the
internationally recommended intake levels; worse, it masks substantial
inequality in consumption within the population, a fact related to the high
Case Studies 171
income elasticity of milk and the unequal income distribution. Therefore,

98% of the population consumes less than 29 liters per year (CIPREDA
1991).
This situation has encouraged the implementation of projects in the

southern coast and southeast region of the country by national and
international organizations, with the aim of increasing animal production
and productivity in order to increase the availability of milk and beef. Since
1979, institutions of the public sector have undertaken actions in the areas
of technology generation and technical and credit support for dual-purpose
cattle farmers.
From 1979 to 1983, the Institute for Agricultural Science and

Technology (ICTA), in collaboration with the Tropical Agricultural Research
and Training Center (CATIE), developed a physical production model that
allowed the efficient use of farm resources, and the improvement of the
nutritional level and sanitary conditions of the herd (Vargas 1986). The
results obtained in this project induced the implementation of a dairy
development project in Nueva Concepcion (1983-1985).
In 1986, a new phase of the project was launched to extend the

evaluation and validation of the model to other areas, supported by a loan
from the Inter-American Development Bank (IDB). In 1985, ICTA, the
Direccion General de Servicios Pecuarios (DIGESEPE) and the Faculty of
Veterinary Medicine and Animal Production, of the University of San Carlos,
Guatemala, initiated a research project on Dairy-Beef Production Systems
with financial support from IDRC.
RESULTS
The production system and the technology
The results of the farm survey showed that farming systems have
evolved from crop-based systems (mainly corn) to mixed systems (crops
and dual-purpose cattle). Specifically, in 1991, 9.8% of the farms were in
pure crop systems, 6.3% were pure cattle systems and 83.9% were in
mixed crop-livestock systems. This is further documented by the change
in land use (Table 1).
Table 1. Changes in land use in Nueva Concepcion (% of total area)
Production Initial Present

System (Purchase year) (1991)
Crops 50.1 37.2
Pastures 28.9 60.1
Forest 21.0 2.7
Source: Technology adoption survey (1991).
Regarding the adoption and impact of proposed technologies for the

improvement of farming systems, the survey showed that farms that
received technical assistance from DIGESEPE and ICTA presented higher
adoption levels than those that did not receive technical assistance. As
shown in Table 2, the following technologies depict this pattern of
adoption: rotational grazing (100% vs. 66%), pasture fertilization (22% vs.
6%), sugar cane as nutritional supplement (44% vs. 0%), Napier elephant
grass as nutritional supplement (33% vs 0%), leucaena as protein
supplement (11 % vs. 0%), sanitary control (100% vs. 75%), parasite control
(100% vs. 70%), vitamin supplementation (33% vs. 23%) and mineral
supplementation (44% vs. 0%). The increase in dairy productivity is based
on both increased reproductive performance of the herd and the increase
in production per area unit. Milk productivity per hectare in the farms
using extension services was 41 % higher than among farms without
technical assistance (1,819 vs. 1,289 1/ha/year).
Household income and its relationship with milk consumption
The real income of rural and urban sectors showed a stable

performance until 1980. During the period from 1980 to 1990, household
income declined to less than 50% of 1980 levels, decreasing household
purchasing power level and lowering consumption levels of animal
products.
Milk consumption is directly linked with household income, and

increases in per capita demand can be achieved through real wage
increases (income and price elasticities of 1 and -0.22 respectively),
Case Studies 173
according to USAID (1987). This implies that, in Guatemala, if the trend of

declining household incomes continues, there is no potential for an
increase in milk demand beyond population growth rates.
Table 2. Technology adoption by farmers and milk production (Nueva

Concepcion, 1991)
Technical support
Technologies None Project

Institutions
No. of farms 59.0 9.0
Pastures (% of farms)
. Rotational management 66.7 100.0
o Fertilization 8.8 22.2
Dry season supplementation (% of farms)

o Sugar cane 30.3 55.6
u Silage 1.8 44.4
o Napier 7.0 33.3
o Leucaena 5.3 11.1
Animal health (% of farms)

.External parasite control 70.4 100.0
.Internal parasite control 100.0 100.0
c Vaccination 74.5 88.9
Vitamin supplementation 23.3 33.3
o Mineral supplementation 0.0 44.4
Milk production
u l/farm/year 15,454.0 17,358.0
u l/ha/year 1,518.0 1,819.0
u I/cow/day (dry season) 2.7 3.0
D I/cow/day (rainy season) 3.3 3.9
Source: Technology adoption survey (1991).

In the case of milk supply, technical change has been efficient in

reducing production costs and in increasing milk availability in the country.
At present, assuming that 100% of the small farmers adopt the validated
technologies, it is possible to increase the production of the southern coast
by 61% to 126%, depending on the level of intensification adopted.
Alternatively, if the production is kept constant, then it would be possible
to release 46% to 54% of the land presently used for dual-purpose cattle
production.
The effects of some macroeconomic variables on the farm system
In Nueva Concepcion, the income surplus generated by crops has been

invested in cattle, allowing capital accumulation by the farmers and a
gradual change of the production system, based initially on the return to
hand labor (in crops), and later on a return on the capital (mainly cattle).
This process of capital accumulation in cattle has been stimulated by
financial market distortions. Negative real interest rates from 1970 to 1990
(Fig. 1), caused by inflation,5 make physical assets more attractive than
financial ones. Consequently, interest rates have caused, on one hand, the
maintenance of a relatively extensive livestock production system using
dual-purpose cattle rather than more specialized dairy cows, and on the
other, have limited the expansion of crops, even though crops require a
lower investment and have a shorter pay-back period.
The magnitude of the variation of the real price of corn over the years
constitutes another factor affecting the land-use systems in Nueva
Concepcion. From 1980 to 1990, mean annual corn prices experienced
variations of up to 30% (in real terms), which discouraged corn production
(Morales 1990). In Guatemala, the local currency was overvalued and this
stimulated the consumption of imported goods that could have been
produced locally. In this study, egg production and poultry production are
based on the use of imported corn. This fact is related to the price of
home-grown corn versus imported corn, given an overvalued local
currency.
5 A negative real interest rate means that a bank deposit (with negative interest rates,
Fig. 1) does not maintain its value in real terms.
Case Studies 175
20
-101 .....................................
ac
-201 .......................... ........ .......................... I.
-30
1970 1975 1980 1985 1990
Year
Fig. 1. Guatemala: Changes in real interest rates, 1970-1990
Source: compiled by the authors.
As an additional negative effect, from 1973 to 1990 the government

regulated the consumer price of milk, resulting in a reduction of the
specialized dairy production sector. The intervention in the milk market
affected the relative price of meat with respect to that of milk at the farm
gate (Fig. 2). This relationship was 5 to 1 during the last five years. Taking
into account nutritional aspects and production costs along with this
relative price, farmers found it attractive to shift production from milk to
beef. Nevertheless, in Nueva Concepcion, production systems still include
dual-purpose dairy production, which allows the efficient use of family labor
and provides a readily available cash flow.
YEAR
1971
1981
1991
0 1 2 3 4 5
PRICE RATIO BEEF: MILK
Fig. 2. Guatemala: Relative prices of beef to milk as paid to the

producer (1 kg beef on the hoof/kg milk).
Source: compiled by the authors.
Outlook for Nueva Concepcion production systems
From a national perspective of efficient natural resource use in the

national context and with a favorable macroeconomic situation (local
currency with real value and positive interest rates), fertile lands should be
mainly used for the intensive production of crops. Systems will evolve in
this direction if macro-policies promote a more open economy, with
exchange rates making domestic production of tradable goods like corn
competitive and with positive real interest rates, thus reducing the
incentives to store wealth in cattle investments. Taking into account the
present macroeconomic situation, it can be concluded that the resources
related to the Nueva Concepcion settlement are being used in a socially
suboptimal way. It must be acknowledged that farmers have adapted to
Case Studies 177
the policy framework by developing a relatively stable production system,

involving less work per hectare of land utilized and less risk than that
required to operate an intensive crop system.
CONCLUSIONS
a. The present study shows the complementarity of the analysis of higher-

level systems (macroeconomic situation, regional development) and
traditional farming systems analysis. This complementarity explicitly
addresses the direct and indirect effects of policies and legislation on
the farm system.
b. In Guatemala, the issues related to animal production systems are an

example of situations where policies can have greater impact than
technological change. Nowadays, the technology to increase
production exists, but it is not widely used due to external conditions
which do not encourage its utilization.
c. In Nueva Concepcion, the production system has evolved from a crop-

based system to a mixed system (crops with dual-purpose cattle
production). Given the resource base, particularly the soils involved,
crops could be the main income-generating activity of the farms.
Taking into account the macroeconomic conditions, mixed production
system is more stable.
d. The survey results show that an important difference exists between

farmers that receive technical assistance from DIGESEPE and ICTA and
those that do not, and that the increase in productivity is due to the
reproductive improvement of the herd and the increase of output per
unit area. The improvement in production was obtained through an
intensification of the system by increasing the carrying capacity of the
farms (increased production of forage and its more efficient utilization).
e. The variability observed in the relative price of beef to milk is mainly

due to the pricing policy for milk. Regulation of milk prices has led to
their decline in real terms. This has caused a decrease in specialized
dairy herds. Nevertheless, the national production has not decreased
and it is now concentrated in dual-purpose cattle herds, due to the low
opportunity cost of available family labor on small- and medium-sized
farms, and the need for the continuous cash flow generated by the sale
of milk.
f. Milk consumption is positively related to household income level, which

implies that an increase in national consumption beyond the rate of
population growth can only be achieved by an increase in the earnings
of lower-income families, which constitute a majority of the population.
It is possible to increase the national milk supply with available
technology, but domestic consumption and lack of international
competitiveness, under the present macroeconomic frame, limit growth
prospects for Guatemala's dairy production.
g. The main macroeconomic variables which have influenced the evolution

of cropping farms to crop-livestock systems are interest rates and
exchange rates. The distortions that have existed in these
macroeconomic prices (negative interest rates and overvalued local
currency) have caused alterations in different directions in the milk, beef
and corn markets.
h. Nueva Concepcion will have a larger role if the macroeconomic

conditions change. The current liberalization and globalization
tendencies of the economy, with a reduction of trade barriers, could
change capital market behavior and the trade patterns of goods and
services, where market distortions still exist. These presently restrict
the contribution of the Nueva Concepcion scheme to socioeconomic
development in Guatemala.
LITERATURE CITED
CIPREDA (CENTRO DE COOPERACION INTERNACIONAL PARA LA

PREINVERSION AGRICOLA. 1991. Estudio sobre comercializacion de
leche y sus derivados. Unpublished documento. Guatemala,
CIPREDA.
CONTRERAS, B. 1988. Diagnostico de produccion y consumo del rubro

maiz en Guatemala. Guatemala, MAGA-CADESCA-CEE.
MORALES, C. 1990. Analisis de la intervencion directa en precios del

maiz por INDECA en Guatemala. Santiago, Chile, Pontificia
Universidad Catolica de Chile, Programa de Posgrado en Economia
Agricola.
Case Studies 179
USAID (UNITED STATES AGENCY FOR INTERNATIONAL

DEVELOPMENT). 1987. Estudio del sector lechero, Vol. and II.
I
Guatemala, Oficina de Desarrollo Rural.
VARGAS, H. 1986. Modelo de produccibn bovina de doble proposito para

Nueva Concepcion. Guatemala, Instituto de Ciencia y Tecnologia
Agricolas. Folleto Tecnico No. 36. 36p.
EXPERIENCES OF THE
INTERNATIONAL CENTERS
ANIMAL PRODUCTION SYSTEMS RESEARCH
IN NEPAL: A CASE STUDY
Jim A. Yazman and Enrique Ospina'
INTRODUCTION
Livestock as a component of agricultural systems has long been a

central program theme at Winrock International. From its establishment in
1975, the trustees and staff of the Winrock Livestock Research and Training
Center directed efforts and attention to the improvement of animal
agriculture in the USA and less developed countries. Winrock's mission
was to improve animal agricultural production systems to meet the needs
of a hungry world. In 1985, Winrock merged with two institutions which
had also grown out of the philanthropic tradition of the Rockefeller family:
the International Agricultural Development Service and the Agricultural
Development Council. While the scope of the new Winrock International
Institute for Agricultural Development was broadened to encompass the
many interrelated components of sustainable agricultural development,
animal agriculture remains a substantial program activity.
During the past 15 years, Winrock International has conducted over 60

research, development, and training projects related to animal production
in the countries of Asia, Latin America and the Caribbean, and Africa, as
well as in the USA, primarily in the South and in Arkansas. Many of these
projects were analytical studies with worldwide, regional, or country
focuses. Other projects focused on improved livestock production through
1
The authors are, respectively, Program Officer, Animal Sciences, and Acting Director,
Development Studies Center, Winrock International Institute for Agricultural
Development. Seniority of authorship is shared. Dr. Yazman is presently located at the
Instituto Boliviano de Tecnologia Agropecuaria, SR-CRSP Program, La Paz, Bolivia.
training and applied research, with particular attention given to small

ruminants on resource-limited farms. Policy issues, as well as
technological constraints, have been addressed. The experience acquired
through these projects is recorded in a collection of over 500 books,
papers, and reports.
The purpose of this paper is to review Winrock's experience in

developing capacity in animal production systems research within the
National Agricultural Research Center (NARC) in Nepal. Between 1985 and
1990, Winrock International and Virginia State University collaborated with
the Ministry of Agriculture (MOA) of His Majesty's Government of Nepal
(HMG) in the early development of NARC under funding from the USAID
Nepal Agricultural Research and Production Project (ARPP). NARC is an
appropriate case study useful in focusing discussion on constraints and
opportunities typical of small countries with limited resources to commit to
animal production systems research.
THE NEPAL AGRICULTURAL RESEARCH AND PRODUCTION

PROJECT (ARPP)
USAID has supported the development of capacity in agricultural

research in Nepal since the early 1970s. In 1985, the ARPP was initiated
to provide support for the development of the NARC as the overall
manager of agricultural research under the MOA. The overall goal of the
project was to develop a national agricultural research system (NABS)
which could provide for the technology needs of Nepalese small farmers.
The NARC was officially established in 1987 and based at Khumaltar,

in the Kathmandu valley, near research facilities of the Department of
Agriculture (DOA) and the Department of Livestock Services (DLS).
Resources (personnel, farms and stations, laboratories, and budget) were
transferred from the DOA and DLS to NARC in order to support a nation-
wide research program. Livestock and animal health resources transferred
from the DLS to NARC included five seed and livestock production farms,
a forage analysis laboratory, a semen production facility (bull stud, liquid
nitrogen generator, semen processing lab), and a veterinary diagnostic
laboratory. Thirty-two officer-level staff posts were transferred from the
DLS and assigned to NARC.
International Centers Experiences 185
LIVESTOCK PRODUCTION IN NEPAL
Nepal is a mountainous country with over 16 million people, 85% to

90% of whom earn their living from agriculture and herding. Livestock play
an important role in the Nepalese economy. In the High Hills (>3,500
meters above sea level), herding of sheep, goats, yaks, and yak x cattle
crossbreds provides the major source of income from trading and sale of
meat, cheese, woven cloth, and handicrafts. Overgrazing has caused a
substantial decline in the productivity of herds and flocks in the High Hills.
The Midhills (500 to 3,500 meters above sea level) cover more than 60%
of Nepal's land area. Agriculture is based primarily on rain-fed maize and
rice-wheat cropping systems with intercropping of legumes (lentils, beans,
and black gram). Despite limited land area, the Midhills support some of
the most dense livestock populations in the world (estimated at five animal
units per arable hectare), with cattle, water buffalo, and goats
predominating. Cattle are maintained for draft and to produce manure.
Buffalo are a primary source of milk and an increasingly important source
of meat. Goat is the preferred meat, and sale of goats is an important
source of income for Midhill farmers and High Hills herders.
The Nepalese Terai is an extension of the vast plains of northern India.

Intensively managed, irrigated agriculture, with a predomination of double-
and triple-cropped production systems based on wheat and rice, provides
much of the food supply for the nation's urban areas. Milk production
from buffalo and cattle and the sale of buffalo and goats for meat are
important enterprises for Terai farmers. Murrah buffalo and crosses of
Indian dairy-cattle breeds with native Nepalese cattle from the Midhills
provide much of the milk marketed in Terai cities and Kathmandu. Buffalo
and cattle still provide much of the draft power in the Terai, as well as
manure, which is important as fertilizer and fuel.
In both the Terai and the Midhills, crop residues, tree fodder, and
grazing of communal pastures provide the majority of the diet consumed
by large and small ruminants. Nepalese farmers harvest foliage from over
100 species of native fodder trees grown on private land and in local
forests. Increasing pressure on forests to provide firewood and timber,
particularly in the Midhills, has resulted in a decline in available tree fodder.
Declining forest reserves and reduced farm size have resulted in a
decrease in the number of buffalo and cattle and an increase in the
numbers of goats.
Poultry and swine production has increased rapidly in recent years,

particularly in the Kathmandu valley and major cities of the Terai. Despite
limited feed resources, periodic gluts of poultry meat and eggs are
increasingly common in Kathmandu. Pork consumption is growing rapidly
among ethnic groups in the Terai and Midhills.
Several contagious diseases are endemic to Nepal, including foot-and-

mouth, hemorrhagic septicemia, and rinderpest. Parasitic diseases,
particularly fascioliasis, take a heavy toll of cattle, buffalo, and goats.
Newcastle disease and fowl pox periodically devastate village poultry
flocks.
NARC LIVESTOCK AND ANIMAL HEALTH RESEARCH:

PROBLEMS AND CONSTRAINTS
Crop disciplinary divisions and commodity programs transferred from

the DOA to NARC have a long history of research activity. Commodity
research and development programs for wheat, rice, and maize patterned
after successful All-India programs were organized in Nepal in the 1970s.
While the strength of linkages has varied over the years, Nepalese crop
commodity programs have had the support of the international agricultural
research centers (IARCs) for genetic materials, training, information
services, and new management technology. International and bilateral
donors have long supported MOA crop and horticulture research efforts.
Crop commodity research programs and disciplinary divisions were

closely linked with the extension service under the DOA. Through the
Cropping Systems Program within the USAID/Winrock International
Integrated Cereals Project (ICP, 1978-1985), DOA extension agents, crop
research specialists, and research farm staff received training and
experience in implementing on-farm research. While the impact of DOA
crop research programs was questionable, cropping systems research
techniques were already institutionalized within government crop research
and development agencies at the time NARC was established in 1987.
Early on in the ICP, research surveys by the Socioeconomics Program

indicated the need for a holistic farming systems approach to increasing
total farm productivity, particularly in the Midhills and Terai. Recognition
of the need to broaden agricultural research to encompass other

components of the farming system, particularly livestock and horticulture,
was inherent to the design of the ARPP (the follow-on project to the ICP).
Prior to 1987, very little livestock and animal health research was
carried out by the MOA. The DLS was charged with controlling infectious
diseases and training farmers, as well as with producing seed and
seedlings of improved grasses, legumes, and fodder trees and improved
breeding stock for distribution to farmers. Most new genetic materials
were imported through donor-supported development programs from
Australia, New Zealand, and India.
As was traditional in South Asia, most DLS staff were veterinarians, and
very few officers had graduate degrees in animal production disciplines.
Through donor-funded projects and bilateral aid programs, a number of
DLS officers were trained during the 1970s and 1980s to the M.Sc. level in
India, Australia, New Zealand, and Europe. Limited budgets and service-
oriented mandates precluded participation in research when they returned.
The ARPP committed a substantial amount of resources to support the

development of animal production systems research within NARC. Several
problems and constraints, many of which were common to both crop and
livestock research, precluded the development of an effective animal
production systems research program. The constraints and problems,
discussed below, tend to have a greater impact on animal agriculture
production systems research than on crop production research. Although
derived from the Nepal experience, these problems and constraints are
typical of research programs during the initial stages of the development
of NARS, as seen in other similar Winrock experiences in Bangladesh,
Pakistan, Indonesia, Egypt, Kenya, Peru, and Brazil.
Lack of incentives for researchers
NARC livestock and animal health divisions have a number of well-

trained, skilled research officers capable of leading productive research
programs. However, low salaries and per diem and lack of a performance-
based promotion system severely affect research officer productivity.
Especially affected are younger officers who have recently completed
graduate degrees and are at the bottom of the seniority scale. On-farm
animal production systems research is particularly impacted by low per
diem rates, as officers have to trek long distances, often for days, to reach
livestock at remote outreach sites and in migratory herds and flocks. Daily
costs for food and lodging in remote regions exceed current per diem
rates.
Insufficient operations budget, inadequate infrastructure,

and production mandates
Livestock research division and farm budgets provide little support for
research operations. Personnel costs for divisions and programs typically
represent 60% to 70% of annual budgets leaving little funding available to
cover supplies, equipment, and maintenance and repairs. Laboratories and
livestock research farms suffer from a lack of vehicles, equipment, and
consistent supplies of water and electricity. Limited budgets preclude
equipment purchase and capital improvements. Importation of specialized
biologicals and reagents for animal disease research is difficult without
access to foreign exchange.
Prior to 1987, management of MOA livestock research farms was the

responsibility of the DLS. The principal activities on the farms were
production of breeding stock and seed, and training of farmers. Seed and
livestock produced by the farms are distributed by DLS district extension
offices to promote livestock development and as a demonstration of
government support for farmers and herders.
With limited resources, NARC farms are hard-pressed to meet both

research and production mandates. Reduction of herd and flock sizes to
more efficient levels required for research is precluded by the need to meet
set targets for production. Large herd and flock sizes not only reduce
funds available for research operations, but severely tax farm management
and land resources. Free or subsidized inputs and services provided by
the government retard the development of private suppliers (e.g., purebred
breeders, forage seed producers) who could become collaborators in
research projects and in extension of newly generated technology.
Short-term planning horizons
His Majesty's Government's planning, monitoring, and evaluation

procedures promote short-term planning horizons inappropriate for
livestock research. Projects are considered as "targets" for completion
within one year. Preparation of a report of research completed as
indication of target achievement is more important than adoption rates for
new forage varieties and management systems, or performance indices of

new breeds in farmers' herds and flocks. Programs with built-in
technology generation goals based on long-term planning and appropriate
monitoring and evaluation procedures are needed.
Overlapping research mandates and lack of inter-agency cooperation
Limited budgets, facilities, and manpower make collaboration between

NARC research units imperative. However, cooperation is made difficult
by budget management procedures where expenditures by one unit for
activities reported by another are prohibited. Overlapping research
subjects which can benefit from close collaboration between units include
forage seed production (Forage Research Division and Seed Technology
Improvement Program); crop by-product utilization (Animal Nutrition,
Agricultural Botany, and Agronomy Divisions); and forage legume inoculum
research (Forage Research and Soils Division). Particularly important to
improvement of intensive Midhill and Terai crop-livestock production
systems is research in multipurpose trees. Close collaboration between
the Forage Research and Agronomy Divisions of NARC and the Forestry
Research Program of the Ministry of Forests and Soil Conservation would
be particularly useful, but is made difficult by budget management
procedures.
Lack of integration of farming systems research methodologies
The ARPP provided support for the establishment of the Farming

Systems Research and Development Division (FSRDD). The mandate of
the FSRDD was to develop farming systems research (FSR) methodology
appropriate to Nepalese production systems and to train NARC staff. The
FSRDD was to be staffed by research officers on deputation from the
various NARC crop and livestock research divisions and commodity
programs. Unfortunately, with limited staff resources and budget, the
NARC research divisions, both crop and livestock, delegated all
responsibility for "doing" farming systems research to the FSRDD.
Participation of livestock and animal health researchers in multidisciplinary,
on-farm research was precluded by lack of transport and the need to
commit the limited budget and manpower to the maintenance of
deteriorating station equipment and infrastructure and to complete short-
term research trials on-station. Establishment and management of FSR
sites in the various agroecozones became the primary activity of the
FSRDD, requiring substantial inputs of manpower and budget. The end
result has been that the establishment of a separate FSR division retarded
the integration of a multidisciplinary FSR approach into the NARC research
program.
Lack of close collaboration with the DLS
DOA extension personnel participating in the cropping systems

research programs that were established by the ICP in the 1970s and early
1980s worked alongside DOA researchers in on-farm trials. This close
researcher-extensionist linkage has persisted with the establishment of
NARC. DOA district extension offices carry out field trials of new varieties
and management technology in collaboration with NARC crop research
farms.
A similar close working relationship does not exist between NARC

livestock and animal health researchers and DLS extension staff. DLS
district offices are burdened with responsibilities for controlling disease and
providing services to farmers and herders (artificial insemination, training,
etc). Adding responsibilities for on-farm testing of new breeds and forage
varieties is not feasible. At the same time, if the DLS is left out of the
technology generation "loop," NARC loses access to the input of
experienced extensionists, and technology transfer is slowed.
Lack of linkages with NARS and IARCs
There is no linkage between NARC livestock and animal health

researchers and colleagues in neighboring NABS. Asia has no
international agricultural research center dedicated to livestock research.
Participation by Nepalese livestock researchers in the Crop-Animal
Systems Research Network sponsored by the International Rice Research
Institute has been minimal. Funding for staff to travel to visit NARS in
neighboring countries, to attend international meetings, and to purchase
books and journals has been dependent on donor support.
Limited collaboration with agroentrepreneurs, PVOs/NGOs,

and farmer organizations
The NARC livestock and animal health research program receives little
input from traders, input suppliers, and processors. There is little or no
commercialization of inputs for livestock production in Nepal.
Importation or production of forage seed, semen, and most vaccines,

isdone by HMG. Forage seed, and even cuttings of napier (Pennisetum
purpureum) are produced on HMG farms. Some private suppliers of
leucaena have recently begun marketing seed and seedlings. Artificial
insemination and animal health services are provided by the DLS at highly
subsidized prices.
Collaboration with traders, processors, and merchants would provide

a better focus for NARC research and would facilitate transfer of research
results to farmers and herders. Lack of collaboration with
agroentrepreneurs has meant that NARC researchers have missed
opportunities to do research which could have direct impact in the
marketplace.
At times, the wrong technology has been promoted by NARC and the
DLS due to a lack of understanding of local market conditions. An
example is the promotion of crossbred dairy cattle in remote Midhill
regions. Dairy cattle have been accepted by farmers due to their higher
milk yield and fertility, but Nepalese consumers prefer the taste of buffalo
milk. Buffalo milk is also preferred by middlemen and owners of hotels
and coffee shops because its higher fat and solids content facilitates
dilution with water. Farmers who were encouraged to purchase dairy
cattle have experienced reduced prices for their milk or have lost their
market completely.
Several international private voluntary organizations (PVOs) and local

nongovernmental organizations (NGOs) operate development projects in
Nepal which have livestock components. There is little or no collaboration
between NARC research units and PVOs or NGOs. Project staff often visit
NARC farms to purchase forage seed or breeding stock but rarely become
involved in research programs which could benefit their constituents.
While PVO and NGO projects tend to be localized, they represent an
opportunity for on-farm testing, validation, and rapid transfer of NARC
technology.
Farmers organizations represent another potential source of

collaborative support for NARC animal production systems research
programs. The Agricultural Development Bank of Nepal requires
membership in a farmer organization in order to receive loans to purchase
breeding stock. These farmers organizations are well served by DLS
district offices, but have little or no contact with NARC researchers.
Lack of donor support for animal production systems research
The combination of the above constraints results in a very low level of

productive activity in animal productions systems research and the
inefficient use of available manpower, budget, and infrastructure. As a
direct result, NARC livestock and animal health researchers have been
unable to attract donor support for animal production systems research.
Crop researchers are able to show more immediate impact from their
programs and receive substantial support from international and bilateral
donors. Increased food grain production is accorded high priority in long-
term MOA planning.
Donor-funded livestock projects in Nepal have tended to support

production and credit activities, assuming that existing or imported
production technology was sufficient to meet the needs of small farmers.
Despite growing demand for milk, meat, and eggs and persistent problems
of livestock disease and feed shortages, NARC researchers have not been
able to make a good case for donor support for livestock and animal
health research programs.
PROMOTING EFFECTIVE ANIMAL PRODUCTION SYSTEMS

RESEARCH PROGRAMS IN NEPAL
Winrock°s experiences in the Nepal ARPP suggest that the institutional

environment is as great a factor in the success or failure of animal
production systems research as are the methodologies employed in on-
farm research. Recommendations growing out of the experience in Nepal
can be extrapolated to countries with an animal production systems
research program at a similar stage of development. Recommended steps
to improve the effectiveness of NARC animal production systems research
efforts are discussed below.
Compensation for NARC livestock and animal health researchers

should be increased
Managing on-farm animal production systems research requires

additional effort relative to on-station research, especially in a country like
Nepal. To collect milk yield data, researchers or enumerators must be at
the farm early in the morning or late at night. Collection of slaughter data
often requires presence at the butcher shop in the middle of the night.
Animal health researchers must respond quickly if an animal sickens or
dies and have to handle biological samples in the field which can pose a
personal health risk.
NARC livestock and animal health researchers doing on-farm research

should receive differential compensation relative to their colleagues doing
research on station. Animal health researchers should be compensated to
the level of their colleagues in private practice. Without adequate
compensation, there is little hope of attracting and retaining the skilled and
dedicated professionals required for effective animal production systems
research.
Research planning and budgeting should be long-term

and program-based
A multidisciplinary, production systems approach is required for

technology generation to support increased production on complex,
intensive crop-livestock production systems. In a NARS with a severely
constrained budget, a portion of research support should be committed on
a program basis. Within programs, close collaboration of staff from
relevant disciplinary divisions and agencies, including the extension service
and crop research divisions, should be encouraged through flexible budget
management guidelines and rules. For Nepal, suggested animal
production systems research programs include the following:
milk production on small farms in the Midhills
small farm poultry meat and egg production in remote regions
improved wool and hair production
livestock feed resources (including fodder trees, crop residues, and

improved forages)
high altitude livestock production
improvement of draft and transport
Planning horizons for animal production systems research should be

longer than for crop production systems research: 15 years to 20 years or
more where cattle and buffalo breed development are an important
component of the research program. Progress should be monitored and

impact periodically assessed by setting, and meeting, shorter-term goals
within individual research projects.
An example of where program budgeting can promote greater efficiency

is the testing of new cereal crop varieties. Nepalese farmers, particularly
in the Midhills where buffalo are an important source of family income,
often reject new rice and wheat varieties because of the forage
characteristics of the straw. Straw palatability is reported to be as
important a factor in rejection of varieties as is straw yield. NARC animal
nutrition and crop breeding specialists should collaborate closely under a
common program-based project. Where crop residues play an important
role in feeding systems for livestock, screening cereal crop varieties for
residue feeding value and palatability will save time and increase adoption
rates.
NARC animal production systems research should be market-oriented

and demand-driven
Improved livestock production presents significant opportunities for

increasing Nepalese farm family income. Nepalese farmers and herders
will adopt new production technologies if they are cost effective and
affordable. Animal production systems research in Nepal will be effective
in increasing livestock production only if the new technologies produced
are adopted on a wide scale. Private sector "buy-ins" into systems
research, through collaboration of agroentrepreneurs, PVO/NGO projects,
and farmer groups in on-farm research trials (if not actual financial
support), will be increasingly important in justifying donor and government
support for animal production systems research.
Animal production systems researchers in Nepal have an adequate

understanding of the livestock production problems of farmers and
herders. They have not been effective in identifying new marketing
opportunities and in designing research programs to assist producers to
take advantage of those opportunities. Donor support for traditional
institution-building projects, similar to the ARPP, is declining in Nepal, as
in most countries. As Nepal's economy expands and population grows,
demand for livestock products will increase. Sale of livestock products is
an important source of income for small farmers, particularly in mixed
crop-livestock farms. It is expected that the new government (elected
democratically for the first time in 1990) will support increased exports of
livestock products.
Ties with new "agribusiness" projects may lead to strengthening of

production and processing capacity in addition to providing the producers
with opportunities for increased domestic and export marketing. Following
this new thrust, the follow-on project to the ARPP, the USAID
Agroenterprise and Technology Systems Project (ATSP), will promote
stronger linkages between Nepalese agroentrepreneurs, producer groups,
and NARC researchers. The challenge for NARC animal production
systems researchers will be to design research programs that assist
farmers and herders in taking advantage of new marketing opportunities.
In order to participate in projects such as the ATSP, livestock and animal
health researchers are going to have to demonstrate more clearly the
relevance of their research to the increased production and marketing of
livestock products.
NARC livestock and animal health researchers should establish

linkages with agroentrepreneurs, PVOs/NGOs and farmer
organizations and improve linkages with the DLS
NARC researchers should seek opportunities for collaboration with

agroentrepreneurs, PVOs/NGOs and farmer organizations. Collaboration
will result in the design of research programs that have more market
orientation and are more closely attuned to the real production problems
of producers.
Agroentrepreneurs are the link between producers (farmers and

herders) and the consumer. They have the means to transport, process,
and commercialize livestock products and have important knowledge of
production constraints and consumer preferences.
PVO/NGO field projects present opportunities for supervised testing of

new technology. Field project managers are generally well trained and
dedicated. Farmer organization leaders speak for the primary consumers
of NARC technology. Farmer leaders should be included with PVO/NGO
project managers in research planning, monitoring, and evaluation and
should be encouraged to assist with on-farm trials.
An effort should also be made to bring the DLS into NARC animal
production systems research programs as a partner in on-farm trials using
FSR methodologies. Program funding should provide for the extra
manpower required for DLS district offices to participate in on-farm
research trials. District livestock extension personnel should assist in
planning, monitoring, and evaluation of research programs in their districts.

NARC animal production systems research programs should provide for
training of extensionists in FSR techniques.
FSR approaches should be integral to all NARC research programs
The FSRD Division should take leadership in the development of

appropriate FSR methodologies and in the training of NARC researchers.
Management of FSR sites should be the responsibility of NARC research
farms and stations. FSRDD should not have responsibility for "doing"
farming systems research. Farming systems research should be an
integral part of all NARC research, with collaboration between NARC units
in on-farm research as appropriate.
Available resources should be used more efficiently: Eliminate

production mandates
Greater emphasis should be given to on-farm research in the herds and

flocks of farmers and herders. NARC herds and flocks are expensive to
maintain and inefficient (in terms of yield per unit of labor, feed, etc).
Station-based herds and flocks should only be maintained where the
degree of risk to livestock and humans from technologies being tested is
high, for reproduction of new breeds, and where isolation and control are
required which cannot be achieved in on-farm trials.
NARC's limited resources should be fully committed to research. Most

production activities can be performed more efficiently by farmers and
agroentrepreneurs. Private suppliers of wheat and maize seed have been
developed in Nepal through training, technical assistance from DOA
extension agents, and provision of simple metal storage bins.
Farmer-to-farmer sale of breeding stock (buffalo, cattle, goats) can be

increased if farmers are encouraged with needed technology and technical
assistance, and do not have to compete with HMG farms selling at
subsidized prices or distributing inputs free of charge. Competition
between suppliers in a marketplace free from subsidy and import
restrictions will result in needed production inputs being available at
reasonable prices.
NARC livestock and animal health researchers should establish

linkages with NARS in other countries and with IARCs
NARC animal production systems research program should be

adaptative in nature. This implies that technologies will be "borrowed" from
countries with similar production environments and then tested under and
adapted to local conditions. Particularly important for livestock producers
in Nepal and similar countries is access to new forage germplasm and
veterinary diagnostic procedures, treatments, and vaccines.
The ARPP invested a substantial portion of available funding in facilities

and support for the establishment of linkages between NARC and Asian
NARS and IARCs. IARCs (CIMMYT, IRRI, ICRISAT)2 provided technical
assistance, program support, and outreach services (germplasm, training,
etc.) to NARC research programs. The result is a substantial amount of
resources supporting linkages between NARC crop research programs and
NARS in neighboring countries and IARCs. A similar level of support was
not available to the NARC animal production systems research program.
A number of networks for animal production systems research are

operating in Asia. These are supported by international donors (IDRC,
UNDP, USAID, etc.)2 and provide useful opportunities for collaborative
research, exchange of experience, and sharing of technology. Networks,
however, do not provide the technical backstopping from experienced
researchers and laboratories that is required on a consistent basis to
support research programs in developing countries. To provide for the
long-term technology needs for Asian livestock producers, particularly in
the lesser developed countries of south and southeast Asia, a more
permanent organization should be considered. Given the tremendous
importance of livestock in Asian production systems, the large proportion
of the world's livestock populations found in Asia, and the potential
contribution of livestock to the economies of the less developed countries
in Asia, consideration should be given by the Consultative Group on
International Agricultural Research (CGIAR) to the establishment of an
International Livestock Center for Asia.
Short of establishing a new IARC, the CGIAR should provide funding to

existing centers to establish livestock outreach programs in Nepal and
other Asian countries. The regional outreach program of CIMMYT in

Mexico should be considered as a model for CGIAR centers with livestock

and animal health research components. Regional representatives
collaborate in research projects, provide technical backstopping,
germplasm, and strategic operational support for research and research
outreach. They work to link programs of the various NARS in the region.
Existing IARCs in Asia or the more advanced NARS (Indian Council of
Agriculture Research, the Philippines Council for Agriculture and Resources
Research and Development, etc.) might be provided funds to manage
similar outreach research programs in livestock and animal health in
collaboration with livestock-based IARCs (CIAT, ILCA, ILRAD)2.
CONCLUDING REMARKS
Livestock are an important component of Midhill and Terai farm

production systems providing food, income, power, and manure. They are
the only means Nepalese herders have for exploiting some 18,000 km2 of
rangeland in the Himalayan foothills. Improvement in the productivity of
Nepal's livestock populations will have a significant impact on family
income and national food supplies and current accounts. A major part of
the improvement can be achieved through importation of production
technologies from neighboring countries. Limitations on available
production resources (particularly feed and forage), the intensive nature of
Midhill and Terai production systems, and the unique characteristics of
Nepalese markets dictate that new technologies be tested and evaluated
before being promoted on a wide scale.
NARC has the human resources to operate an effective animal

production systems research program for developing and testing new
production technology. Institutional constraints, particularly those related
to management of human resources and relationships between NARC
research units, have to be removed before an effective program can be
developed. Research will have to be closely oriented to the solution of
existing production problems and to emerging market opportunities. The
needs of a growing population for meat, milk, and eggs provide the
justification for animal production systems research. Better management
of NARC personnel and resources and better design and management of
research programs will allow NARC animal production systems research
programs to achieve impact and generate support from both the
government and donors.
EXPERIENCES WITH LIVESTOCK SYSTEMS
RESEARCH AT ICARDA
Euan F. Thomson' 2
INTRODUCTION
The International Center for Agricultural Research in the Dry Areas

(ICARDA), with headquarters in Aleppo, Syria, was established in 1977 and
is one of the younger centers of the Consultative Group for International
Agricultural Research (CGIAR). ICARDA's global mandate is the improve-
ment of durum wheat, barley, and lentils, and its regional mandate is the
improvement of bread wheat and research on farm resource management
and pastures, forages and livestock. Of the three international agricultural
research centers (IARCs) with research on livestock, ICARDA, with only
one livestock scientist during the 1980s, has a substantially smaller
program than the International Livestock Center for Africa (ILCA) and the
International Center for Tropical Agriculture (CIAT).
This paper is about livestock systems research at ICARDA during the

1980s. It outlines the problems to be solved and the approach used; it
also presents the main projects and results, and indicates some of the
lessons learned, current activities and the outlook. Although it is
undesirable to ignore the non-livestock aspects and results of multi-
disciplinary projects, this has been necessary for the sake of brevity.
Details of such research are found in the literature cited throughout this
document.
1
International Center for Agricultural Research in the Dry Areas (ICARDA), Aleppo, Syria.
2
Sincere thanks are extended to the numerous colleagues who collaborated in the
research reported in this paper, particularly Faik Bahhady, Ronald Jaubert and Tom
Nordblom. The technical assistance of Asnan Termanini, Shahbah Morelli, Hishman
Hreitani, A. K. Ferdawi, Mahmoud Oglah and Safouh Rihawi and his staff is much
appreciated. The views expressed in the paper are those of the author and do not
necessarily represent those of ICARDA.
PROBLEMS TO BE SOLVED
ICARDA's research efforts are directed to solving problems of agri-

cultural sustainability and production in the countries of West Asia and
North Africa (WANA), from Mauritania in the west to Pakistan in the east,
and from Turkey in the north to Ethiopia in the south. The climate of the
WANA countries is harsh. Annual rainfall is between 150 and 600 mm,
received in winter and spring, and is associated with large variations
between and within seasons (Cooper et al. 1987). Winters are cool to
cold, depending on altitude, and summers are dry and hot, with mean
August temperatures often exceeding 35°C. Seventy percent of the area
of WANA is desert and mountains and 22% is ranges and steppes; of the
remaining 8%, about 130 million ha is cultivated for rainfed crops, and
another 35 million ha is used for irrigated crops. This cultivated land is
unable to support a human population which has one of the fastest growth
rates in the world and is estimated to reach 1.5 billion by 2030 (ICARDA
1989). Thus, the WANA region, which includes some of the world's richest
and poorest countries is the largest global food importer, and many of the
imports are high-priced animal products which are demanded by the oil-
rich countries in particular. ICARDA is therefore faced with the difficult task
of assisting the WANA countries in improving food production in a region
where the pressures to do so are enormous but the harsh environment
makes this particularly difficult to achieve in a sustainable manner.
Research aiming to increase the output of animal products faces many

other constraints related to those indicated above. Particularly critical is
the increasing pressure on cultivable land to produce more food, as well
as feed. To reduce competition between these two land uses, additional
feed is expected to come from the better use of crop residues, barley
crops growing in areas too dry for food crops, the use of fallow land for
growing forage crops and pastures, uncultivable communal grazing areas
within the cultivated zone, and from the vast but overgrazed ranges and
steppes. The challenge to scientists at ICARDA and in national agricultural
research systems (NABS) in WANA is to produce this extra feed using
sustainable production systems in countries where the small-ruminant
population continues to expand and the natural pastures are still
succumbing to the plough. Furthermore, this feed must be used efficiently
by fertile, healthy, and productive animals.
THE EVOLUTION OF LIVESTOCK SYSTEMS RESEARCH AT ICARDA
ICARDA has had four major research programs since its establishment:
two are crop improvement programs, and two conduct research on
pastures, forages, and livestock, and on farm resource management.
Surveys, the "unit-farm" project and the Breda on-farm project, which
started in 1977 1979, and 1981, respectively, and included livestock
components, were among the initial activities of the Farming Systems
Program (FSP). In 1983 the Pasture, Forage and Livestock Program
(PFLP) became a systems-oriented program when the first ICARDA
External Review recommended that the livestock research be transferred
from FSP to PFLP. The transfer, which was undertaken to integrate the
livestock with the pasture and forage research, arguably removed an
essential activity from FSP. This illustrates one of the dilemmas facing
both IARCs and NABS; how best to fit the different research disciplines and
commodities into an institutional structure. The IARCs are still grappling
with this dilemma.
In 1982-1983, the PFLP had the following principal scientists:

germplasm specialist, forage agronomist, forage-Medicago pathologist,
and in 1984-1985, after the transfer of livestock research, the Program's
staff was expended to include scientists in the following areas: pasture
ecology, range ecology, microbiology, forage breeding, grazing
management and animal husbandry.
In 1984, the expanded PFLP started new long-term multidisciplinary

projectsin addition to the unit- farm and Breda on-farm projects inherited
from FSP. The comprehensive EI-Tah on-farm project focused on annual
Medicago ("medic") pastures and involved the pasture ecologist, livestock
scientist, microbiologist and an economist from FSP. In 1985, two more
projects were started. One, project ML2, studies the improvement of
marginal land and involves the range ecologist, pasture ecologist, livestock
scientist and the same economist. The other, project L13 which replaced
the unit-farm project, examines the productivity of a number of two-year
wheat-based rotations. It is led by the grazing management specialist and
has input from the pasture ecologist, livestock scientist, and economist.
The Maragha steppe improvement project started in 1989; it is a
collaborative project between the Syrian NABS, the Arab Center for Studies
of Arid Zones and Dry Lands (ACSAD) and ICARDA. Scientists from both
PFLP and the Farm Resource Management Program (FRMP), the new
name of FSP since 1985, also collaborate in other projects involving sheep.
These examples illustrate the systems nature of much of the research in
PFLP, which has an annual budget that increased from US$1.6 million in
the early 1980s to over US$2.0 million during the late 1980s. About one
quarter of this budget has been allocated to animal production systems
research (APSR).
Now that the livestock disciplines in PFLP are being strengthened, this
structure is considered suitable for APSR. However, the desperately low
funding of NABS may make systems research, however desirable, a luxury
well beyond their reach.
THE APPROACH USED
APSR at ICARDA has used the low-technology approach during the past
decade since these technologies have the best chance of being adopted.
The following examples illustrate how existing low- intensity systems are
being fine-tuned to increase output. First, the local Awassi sheep breed
has been used on-station even though it is frequently assumed that
indigenous breeds are inferior. Second, emphasis has been given to
appropriate feeding during the breeding season, late pregnancy, and early
lactation to achieve genetically determined levels of fertility and milk
production. Third, the on-farm research uses current farmer practices
whenever possible but includes small changes in the current system which
have the best chance of being accepted by farmers. ICARDA was applying
the farming systems research (FSR) approach to APSR (conducting
surveys, on-station research and on-farm research) at a time when the FSR
approach was gaining popularity (CGIAR 1978; Fitzhugh et al. 1982).
However, ICARDA has not been involved in the extension or development
stages of FSR but rather in studies on the impact of new technologies on
farm income and the constraints to technology adoption.
AN OUTLINE OF APSR PROJECTS AT ICARDA
Surveys
In 1977, the former FSP initiated a three-year diagnostic survey of crop-

livestock based farming systems in Aleppo Province (ICARDA 1980), and
in 1978 this activity was extended to a sheep-dominated system at the
cultivated margin of the Syrian steppe (Thomson et al. 1989). This put into
practice the belief that prevailing farming systems must be understood
before they can be improved. These surveys and later ones (Mazid and
Hallajian 1983; Goubelat Survey in Tunisia, as described in the ICARDA
Annual Reports) suffered one or more weaknesses, such as overly broad
objectives, small sample size, and long delays between collecting the data
and analyzing and reporting the results. These weaknesses were avoided
by later surveys, such as the reconnaissance surveys of Jaubert (1983),
which had an immediate impact on contemporary on-farm research, as
discussed later in this paper. Other single-visit surveys on specific topics
(Thomson and Bahhady 1983) were conducted in order to identify farmer
practices that could be applied in the unit-farm project.
On-station research
FSR on-station research started in 1979 using unit-farms, a concept that

has been used on several occasions (Menz 1979). Unit-farms are used to
assemble and test, in a whole-farm system context, the different elements
of an improved package of technology, simulating farmer practices as far
as possible on a research station. At one extreme, farmers ran the farms,
as was the case of the participating farmers at ILCA (Gryseels and
Anderson 1983); at the other extreme, scientists were the managers, as
was the case at ICARDA.
The importance of straw in sheep feeding resulted in a research

program on the genetic variability of wheat and barley straw quality, in
collaboration with ICARDA's cereal breeders (Thomson and Ceccarelli
1991). At the same time, studies were conducted into crop-livestock
interactions, such as the green-stage grazing of barley which would later
regrow and be harvested for grain and straw (Nordblom 1983a), and the
decision of when to harvest or graze mature barley crops (Nordblom
1983b).
From 1979 to 1985, two unit-farms served to compare the physical and
economic productivity of a traditional Syrian crop-and-sheep farm system
with one incorporating improved technologies recommended by
component scientists (Thomson et al. 1993). Each unit-farm consisted of
arable land, an experimental sheep flock and native pasture. A third
experimental flock allocated only native pasture and subjected to "poor"
management, simulated the case of a landless sheep-owning farmer. The
three levels of sheep management made it possible to apply input/output
response analysis.
The project incorporated improved sheep-management practices that

farmers were likely to adopt rather than complex imported technologies
that were inappropriate in the prevailing farming and institutional climate.
For example, diets were based on locally available feedstuffs, but
technologies such as synchronized oestrus and use of prolific breeds and
three-lamb-crops-in-two-years, were rejected. A sheep breeding program
was not included in the initial APSR research agenda because ICARDA did
not have the necessary human, technical, or financial resources for such
research. Nor was such a program seen as appropriate, as it would have
duplicated an ongoing project run by the Syrian NABS.
Project L13 is a replicated, large-scale and long-term on-station trial

comparing the crop productivity of wheat/medic, wheat/vetch,
wheat/lentil, wheat/melon and wheat/fallow rotations. The wheat/medic
rotation is repeated three times to allow three stocking rates. The project
is generating unique information about the different rotations and about
sheep productivity from medic pastures stocked at different rates.
Project ML2 studies native grass-legume pastures at ICARDA's research

station. These pastures are dispersed within the cultivated areas of West
Asia and, because they are communal land, suffer from severe overgrazing.
The project has three phosphate fertilizer treatments, each stocked at two
levels, and three replicates (Osman et al. 1991). Eighteen sheep flocks,
each containing five or six sheep, graze the 18 plots. The trial has already
completed seven years (1984-1991).
The Maragha project is studying the use of two edible shrubs, and
Salsola vermiculata and saltbush (Atriplex halimus), to increase the sheep
carrying capacity of the steppe. It has two steppe treatments, unimproved
and improved with shrubs, each stocked at two rates in a replicated
factorial arrangement.
On-farm research
The seven-year Breda on-farm project started in 1981. Its aims were (1)
to use land left fallow after a barley crop to grow legume forage crops, (2)
to study the system of crop use preferred by farmers, and (3) to identify
constraints to the introduction of the crops (Thomson et al. 1992). In
addition to traditional farmer practices, best-bet practices were applied
since recommended agronomic practices, such as levels of phosphate
application for the crops, had not been defined by component scientists
at that time. In retrospect, this was the correct decision since defining
appropriate agronomic practices can take many years, if not decades.

Starting on-farm research (OFR) soon after the on-station research (OSR)
allowed experiences from the OFR to be fed back into OSR. One example
of this was the observation that sheep found forage pea herbage
unpalatable (Thomson 1984; ICARDA 1985).
The EI-Tah project focuses on the use of annual medics in the wheat-
based farming systems of Syria and ecogeographically similar areas of
West Asia (Cocks 1988). The project uses the low-tech approach in
contrast to many previous projects in WANA involving medics in which the
technology was applied as practiced in Australia. This use of an imported
technology is one reason why the "ley farming system" has had low
adoption rates on the small farms that predominate in WANA. The EI-Tah
project, like the other on-farm projects mentioned here, is serving to
develop a methodology which could be used by NABS.
The six-year EI-Bab on-farm project started in 1985 as a continuation of

the Breda on-farm project. It differs from the Breda and EI-Tah projects in
that it includes both forage and annual pasture legumes, and many of the
farmers own larger farms with more fertile land than at Breda. This
inclusion of both forage and annual pasture legumes in the same project
is of importance because they have complementary roles in improved feed
producing systems (Cocks and Thomson 1988).
Another recently completed, four-year, on-farm project addressed the

effects of helminth parasites on sheep productivity (ICARDA 1990). It used
eight flocks and 400 ewes selected from the EI-Bab and El-Tah sample of
farmers and from one other village. This use of farmers with whom
scientists had already developed a rapport made it possible to conduct
more detailed and frequent monitoring of their animals for parasite load,
live weight changes and general management practices.
A BRIEF DESCRIPTION OF MAJOR RESULTS
The following section gives a brief description of the major results from
projects involving small ruminants. It illustrates how information generated
in one component of APSR is used to formulate objectives or modify
directions of other components.
Surveys
The initial diagnostic surveys provided estimates of actual livestock

productivity under farm conditions, and these were compared with
potential levels established under appropriate management conditions in
OSR. The difference between actual and potential levels of productivity,
commonly referred to as the yield gap, is a measure of the effect of the
environment, that is, management factors (such as nutrition, health, or
housing), on the animal. Once the yield gap is known, decisions can be
taken about the need to improve the local genotype or to import exotic
breeds. A wide gap indicates that considerable improvements in flock
output would arise from better nutrition and health care of the local breed.
The surveys showed that in farm flocks, lambing rate was about 85%, lamb
and ewe mortality about 15 and 8%, respectively, lamb daily growth rates
about 170-200 g, and ewe milk yield after weaning about 55 kilograms.
Another major finding of the surveys was the use of large amounts of
concentrates, straws and by-products. This is hardly surprising since
subsidized feed has been available in Syria since the mid-1 970s as a policy
to stop the devastating losses of small ruminants that can occur during a
series of drought years. As a consequence, farmers were able to keep
more animals, even in years of drought, and the population started
expanding and has long been above the carrying capacity of the steppe.
Overgrazing and the encroachment of barley cultivation into the steppe
have destabilized the fragile ecosystem and today the symptoms of
desertification are obvious. Jaubert (1985) likened the nonsustainability of
barley cultivation practices to a mining operation.
The surveys of Jaubert and Oglah (1985) in the Breda area showed that
farmers were growing common chickling (Lathyrus sativus), a forage
legume, for seed and straw. As a result, chickling was immediately added
as a treatment in the Breda project.
On-station research
Unitfarms. Five significant results from the unit-farm project are

presented; two of these illustrate how results from one project can affect
the objectives of another.
First, the potential productivity of the local strain of the Awassi breed
was established and thus the yield gap could be defined. For example,
under good management, lambing rate could be increased by 20%, lamb
mortality reduced by 66%, lamb growth rate nearly doubled, and ewe milk
yield after weaning increased by at least 50 percent. As will be seen
below, these increases in output were too small to be profitable at the
levels of supplementary feeding being used.
Second, common vetch (Vicia sativa), with its erect growth habit, is
morphologically suited for conservation as hay rather than for grazing.
However, from the outset of the unit-farm project its use for grazing by
growing lambs was assessed, particularly on shallow stony soils where
haymaking machinery is easily damaged. Although modest daily lamb
growth rates of 180-220 g were recorded, up to 300 kg live weight gain/ha
was attained, which makes this practice very profitable at prevailing lamb
prices. This form of crop use was immediately incorporated in the Breda
project.
Third, the early research on forage legumes at ICARDA focused on

conserving them as hay for use during periods of feed deficit. However,
the farm-scale operations on the unit-farms, even using modern equipment
far beyond the reach of local farmers, indicated how difficult it is to make
good-quality hay. Temperatures in April and May are still low, the risk of
rain damage is high, and the thick stems of forage legumes lengthen the
drying time. In addition, harvest losses can reach 50% (Thomson et al.
1990). These experiences on the unit-farms resulted in haymaking being
excluded as a treatment in OFR, although it was continued for some time
in on-station rotation trials.
Fourth, the unit-farms generated physical input/output data over a six-

year period for contrasting crop rotations and levels of sheep management
that could be used to run linear programming models investigating different
farm enterprise combinations (Nordblom and Thomson 1987). The models
show how farms with low levels of sheep management would have the
higher sheep numbers and outputs of sheep products, and higher farm
profits, than farms with high levels of sheep management. The results
indicate that the additional output of animal products with better
management, principally nutrition, was too small to cover the extra feed
costs and suggest that the benefits of improved management will be
realized only when it is applied to improved genotypes. The long-term
nature of such projects generates time-series data which can be used in
dynamic models to study the risk of different enterprises.
Fifth, close day-to-day supervision allowed the three experimental flocks

to be used to determine the effect that ewe live weight at mating had on
fertility (Thomson and Bahhady 1988) and to define the seasonal
prevalence of helminth parasites (Thomson and Orita 1988). Results from

these flocks provided information for the design of subsequent on-farm
studies.
Cereal straw quality. Research has shown that genetic variation in the
quality of cereal straws is largely accounted for by the proportion of leaf
in the straw since variation within the leaf and stem fractions is small.
Cereal breeders could use stem or plant height as a simple predictor of
straw quality since it is reasonably well correlated with leaf proportion.
Thus, short plants would tend to have the best straw quality. However,
cereal breeders point out that tall genotypes are preferred for semiarid
regions because in drought years they are still sufficiently tall to be
mechanically harvested (Thomson and Ceccarelli 1991). The sometimes
opposing goals of the nutritionist and the cereal breeder illustrate the
importance of the collaboration between these two disciplines (Reed et al.
1988).
Research on cereal straw improvement using, for example, caustic soda

treatment or ammoniation, was deemed unnecessary because the results
from applying such technologies were well known. Indeed, there was
much doubt about their applicability in local farming systems. NABS were
considered the appropriate authorities for conducting such research.
On-farm research
The Breda project indicated the practical feasibility of livestock on-farm

trials, the animal production potential of the crops (Thomson and Bahhady
1992) and the increases in profits when forage legumes are grown on
fallow land (Thomson et al. 1992). The eight farmers involved chose
grazing of common vetch by growing lambs and harvesting of crops for
seed and straw as the most appropriate practices. After initial reservations,
they were satisfied that growing forage legumes on fallow land had no
detrimental effects on the next barley crop, provided phosphate fertilizer
was applied to the legumes.
Chickling showed itself to be most suited for seed and straw production
but less suited for grazing than the common vetch being tested by
scientists. It also responded well to phosphate fertilizer when harvested for
seed and straw. This response had not been expressed so clearly when
it was harvested at the hay stage in earlier OSR. The focus of OFR on
either grazing or harvesting forage legumes at the mature stage resulted
in these practices replacing hay production in OSR. Forage breeders also

started to measure straw yields, in addition to hay and seed yields, in their
germplasm research.
Adoption rates, the toughest test of the appropriateness of a new

technology, were low, if not zero, in the Breda project. This was partly
because two successive drought years occurred after the project ended;
as a result, farmers exhausted their seed stocks and were unwilling to buy
more in the market since the legume seeds cost twice as much as barley
seed. The availability and price of seed appeared to be the most important
factor constraining the adoption of forage legumes, as appears to be the
case with annual medics.
The recently completed on-farm parasite study showed how use of a

broad-spectrum antihelminthic could have an economically beneficial effect
on ewe fertility and mortality, even if the treatment differences did not
reach statistical significance. This economic benefit is what interests the
farmers, not the statistics! This study also demonstrated again the
feasibility and the large resource needs of OFR.
LESSONS LEARNED
Lessons were continuously being learned and resulted in the initiation

of new projects and modifications of those already established. Some
examples are given in the following sections.
Unit-farms
The main weakness of the unit-farm concept is that it is impossible to

simulate the commercial and decision-making environment of a real farm
because the managers, whether they are farmers or scientists, are not, in
the final analysis, ultimately responsible for the financial risks of the farm
operations. They also usually have access to resources (such as funds to
buy feed or use of modern field equipment) that are not available to local
farmers. This weakness, realized at the start of the project, must be
weighed against the close supervision needed in component research on
livestock.
Linkages between systems and component research
The unit-farm project, with fields of 1.5-2.7 ha needed for grazing flocks
of 40-50 sheep, provided a useful framework for comparing different farm
enterprise combinations. However, involving scientists from other
disciplines was unsuccessful because of the absence of replication, the
small number of rotations and confounding of treatments. In contrast, the
small plot sizes (e.g., 50 m2) used in other on-station rotation trials
precluded the meaningful inclusion of sheep. In the mid-1980s the new
projects L13 and ML2 were established using large plots ranging from 0.4
ha up to over 3 ha and sheep are now successfully part of these projects
which integrate both systems and component research.
The usefulness of on-farm research
As already indicated, OFR provides valuable feedback for making

modifications to OSR. It has also demonstrated to NABS that such
research is indeed feasible, essential, and publishable. However, OFR is
costly, particularly if it involves livestock and fencing, and this will often put
such research beyond the capabilities of NABS unless they are assisted by
IARCs (Tully et al. 1985).
Upstream versus downstream research
There is already a good understanding of the principles of the livestock

and related sciences and their application in the semiarid and arid regions,
and NABS have often verified these principles in the local environment.
IARCs, therefore, must take care not to "reinvent the wheel," but rather to
conduct some upstream research that is usually too costly for NABS to
undertake, although it leads to systems that are more sustainable than
current ones. An example here is research on the genetics of host
resistance to helminth parasites and the immunology of host resistance to
parasitic infection. Vaccination or introduction of genotypes with
resistance parasites is a more sustainable solution to the problem than
to
overzealous use of antihelminthics which risks the development of
resistance to the drugs. However, overemphasis on upstream research will
further widen the gap between what is possible under carefully controlled
conditions and what farmers are actually doing.
The need for critical mass
An IARC needs a critical mass of scientists in a particular discipline to

be an effective partner with NABS. Until recently, ICARDA lacked the
critical mass of livestock scientists, but the situation is now changing.
Statistical principles
A compromise has to be found between statistical principles and

practicality in OFR. At ICARDA, locations for OFR are usually close to all-
weather roads and often within one hour's drive of headquarters. This is
an important requirement in OFR involving livestock that at times requires
weekly, or even daily, visits.
Farmers are often not chosen at random, but for their willingness to
collaborate, number of sheep owned and location of land relative to their
houses. However, unrepresentative farmers are avoided if at all possible.
In the Breda project, the farmers appeared to be representative of the area,
but their poverty was one reason for the low adoption of the technology.
The farmers in the El-Bab project are better endowed, and this should
result in higher adoption rates when the project ends in 1992.
Replication is still possible so that valid statistical tests can be made

between treatments. Sites or flocks become replicates and treatments are
applied within sites or flocks. Other confounding factors, stage of lactation
being one of the most important, can be partially removed by blocking
animals according to these factors and then allocating from within blocks
to treatments. However, the small numbers of animals involved means this
balancing of treatments is often only partly successful.
Involvement of the NARS in OFR
Involvement of the Syrian NABS in OFR was poor in the Breda project,
better in the El-Bab project and excellent in the El-Tah project. This
involvement of NABS is largely due to the location of the on-farm sites.
The Breda and El-Bab locations are conveniently placed for ICARDA
scientists and technical staff but somewhat remote for scientists from the
Syrian NABS. However, the El-Tah site is well located on the main Aleppo-
Damascus road, even though it is still 250 km from Damascus.
Do NARS have the resources to do APSR?
APSR is long-term, costly, and needs a stable institutional environment.

These facts help explain why so little APSR is seen in the NARS of WANA.
OSR may be possible using financially self-sustaining flocks on government
research stations, as with the unit-farm approach described previously.
NARS often have stations with sufficient land, animals and staff to use this
approach. All that is needed is for the land and animals to be managed
in a systems context while, if possible, comparing different treatments on
a long-term basis. This is the case at the Kamishli Station in northeast
Syria where over 100 sheep are used in replicated grazing trials based on
wheat/medic rotations (ICARDA 1990). ICARDA trained the staff, makes
supervisory visits at regular intervals, and has provided only limited
equipment and fencing. In general, though, the project is run entirely by
Syrian scientists.
Sustainability issues
The need to develop sustainable farming systems has always been an

implicit goal of the research activities of PFLP and FRMP. Thus, low-tech
approaches have been used to increase livestock productivity, feed
producing systems have been promoted based on annual forage or
pasture legumes which need phosphate but no nitrogen fertilizer, and use
of anthelminthics was kept to a minimum so that the risk of parasites
developing resistance was reduced. Research on better ways to use
cereal crop residues as feed also recognizes that there is a conflict
between this use and the need to return the residues to the soils, many of
which been depleted of organic matter in WANA. This conflict may be
resolved by feeding the high-quality leaves to livestock and returning the
low-quality stems to the soil.
PRESENT ACTIVITIES AND OUTLOOK
The increased emphasis given to livestock in the recent CGIAR review

of priorities (CGIAR 1989), and the recommendation of the 1988 ICARDA
External Review that livestock research be strengthened, resulted in a plan
to increase, from one to at least four, the number of livestock scientists at
ICARDA by 1994. This increase has already started with the appointment
of a ruminant nutritionist and a livestock network scientist in 1991.
Considerable efforts will be made to maintain the close integration of
livestock with the other research activities of PFLP in particular, and also
with those in the other research programs at ICARDA. This has been the
case in the past but as the number of livestock scientists increases care
must be taken to maintain this integration. Indeed, this integration was,
and should continue to be, a major strength of the Program, a situation
which even today is seldom, if ever, seen in the NARS of WANA, where
livestock production, livestock health, pasture or forage, and range
research is usually carried out by separate departments.
One of the first results of the increased emphasis on small-ruminant

research at ICARDA was the recent purchase of Awassi ewes selected for
improved milk yield. This purchase illustrates three important points.
Firstly, an indigenous breed may have one or more shortcomings which
limit the productivity of a newly designed crop-livestock system. However,
this weakness can be overcome by using superior animals selected from
the local population rather than the use of exotic breeds in crossing
programs.
Secondly, the NARS in most countries usually have an ongoing

selection and crossing program based on their indigenous breeds which
makes it unnecessary to initiate a long and costly breeding program at an
IARC. Thirdly, improved genotypes are often selected under favorable
management conditions which allows maximum expression of genetic
merit. It is the task of an IARC to confirm that superior performance of
improved genotypes is still expressed when the management level is closer
to that found in current farm conditions. Such research will probably lead
to OR which evaluates the impact of improved genotypes on flock
productivity. IARCs have an essential leadership role in such research,
which will be conducted with NABS.
The need to develop sustainable farming systems, which has become

an explicit goal at IARCs such as on ICARDA, will have an impact on the
type of APSR that is conducted in the future. This is particularly true in
arid regions where livestock are the main output of the fragile
agroecosystems. It will be important for ICARDA to help NARS establish
long-term rotation trials based on cereal/legume rotations into which
livestock are integrated. These trials will provide the framework for NARS
scientists from different disciplines to work together. Furthermore, ICARDA
will need to provide guidance to NARS on the appropriateness of imported
technologies. It should continue to show that low-tech practices have a
better chance of adoption by farmers than high-tech practices, although
some of the latter will need to be tested.
CONCLUSIONS
The first decade of APSR at ICARDA used the low-tech approach to

fine-tune existing low-intensity systems. During the coming decade, efforts
to achieve adoption of new technologies will be strengthened but, as
additional livestock scientists are appointed, more attention will be given
to working with NARS in WANA and to upstream research, such as
evaluation of improved genotypes, which should make semi-intensive
systems profitable. This research, aimed at generating more productive
livestock producing systems, will be conducted within the framework of
sustainable agriculture, as has largely been the case in the past. ICARDA
will maintain a balance between upstream and downstream research,
thereby setting an example for NABS.
LITERATURE CITED
CGIAR (CONSULTATIVE GROUP FOR INTERNATIONAL AGRICULTURAL

RESEARCH). 1978. Farming systems research at the international
agricultural research centers. Rome, FAO, Technical Advisory
Committee of the CGIAR. 66 p.
CGIAR. 1989. Priorities review. Washington, D.C., CGIAR Secretariat.
COCKS, P.S. 1988. The role of pasture and forage legumes in livestock
based farming systems. In Nitrogen fixation by legumes in
Mediterranean agriculture. D.P. Beck, L.A. Materon (Eds.). Dordrecht,
The Netherlands, Martinus Nijhoff Publishers. p. 3-10.
COCKS, P.S.; THOMSON, E.F. 1988. Increasing feed resources for small
ruminants in the Mediterranean basin. In Increasing small-ruminant
productivity in semiarid areas. E.F. Thomson, F.S. Thomson (Eds.).
Dordrecht, The Netherlands, Kluwer Academic Publishers. p. 51-66.
COOPER, P.J.M.; GREGORY, D.J.; TULLY, D.; HARRIS, H.C. 1987.

Improving water use efficiency of annual crops in the rainfed farming
systems of West Asia and North Africa. Experimental Agriculture
23:113-148.
FITZHUGH, H.A.; HART, R.D.; MORENO, R.A.; OSUJI, P.O.; RUIZ, M.E.;
SINGH, L. (Eds.)-. 1982. Workshop on research on crop-livestock
systems. Proceedings. Turrialba, C.R., CATIE/CARDI/ Winrock
International. 65 p.
GRYSEELS, G.; ANDERSON, F.M. 1983. Research on farm and livestock

productivity in the central Ethiopian highlands: initial results 1977-1980.
Addis Ababa, Ethiopia. International Livestock Centre for Africa. ILCA
Research Report No. 4. 52 p.
ICARDA (INTERNATIONAL CENTER FOR AGRICULTURAL RESEARCH IN

THE DRY AREAS). 1980. Studies of farming systems in Syria. Aleppo,
Syria. ICARDA. Farming Systems Program Report No. 2, Sections 3
and 5.
ICARDA. 1985. ICARDA Annual Report 1984. Aleppo, Syria, ICARDA.

378 p.
ICARDA. 1989. Medium-Term Plan 1990-94. Aleppo, Syria, ICARDA.

ICARDA-154 En. 149 p.
ICARDA. 1990. Pasture, Forage and Livestock Program: Annual Report

1989. Aleppo, Syria, ICARDA. ICARDA-166 En. 178 p.
JAUBERT, R. Sedentary agriculture in the dry areas of Syria:

1983.
Development problems and implications for ICARDA. Farming Systems
Program. Aleppo, Syria, ICARDA. Manuscript.
JAUBERT, R. 1985. The semi-arid areas of Syria: Farming systems in

decline. In Farming Systems Symposium (4., 1985, Manhattan,
Kansas). Proceedings. C.B. Butler Flora, M. Tomecek (Eds.). Kansas
State University. p. 232-257.
JAUBERT, R.; OGLAH, M. 1985. Farming systems management in the

Bueda/Breda subarea: 1983/84. Aleppo, Syria, ICARDA, Farming
Systems Program. Research Report No. 13. 63 p.
MAZID, A.; HALLAJIAN, M. 1983. Crop-livestock interactions: Information

from a barley survey in Syria. Aleppo, Syria, ICARDA, Farming Systems
Program. 28 p.
MENZ, K.M. 1979. Unit-farms and farming systems research: The IITA
experience. Ibadan, Nigeria, International Institute for Tropical
Agriculture. Discussion Paper No. 3/79. 15 p.
NORDBLOM, T.L. 1983a. Crop-livestock interactions: The case of green-

stage barley grazing. Aleppo, Syria, ICARDA, Farming Systems
Program. Discussion Paper No. 9. 37 p.
NORDBLOM, T.L. 1983b. Livestock-crop interactions: The decision to

harvest or graze mature grain crops. Aleppo, Syria, ICARDA, Farming
Systems Program. Discussion Paper No. 10. 21 p.
NORDBLOM, T.L.; THOMSON, E.F. 1987. A whole-farm model based on

experimental flocks and crop rotations in northwest Syria. Aleppo,
Syria, ICARDA. Publication ICARDA-102 En. 78 p.
OSMAN, A.E.; COCKS, P.S.; RUSSI, L.; PAGNOTTA, M.A. 1991.

Response of Mediterranean grassland to phosphate and stocking rates:
Biomass production and botanical composition. Journal of Agricultural
Science, Cambridge 116:37-46.
REED, J.D.; CAPPER, B.S.; NEATE, P.J.H. (Eds.). 1988. Plant breeding
and the nutritive value of crop residues. Addis Ababa, Ethiopia.
International Livestock Centre for Africa. Proceedings of a workshop
held at ILCA, 7-10 December 1987. 331 p.
THOMSON, E.F.; BAHHADY, F.A. 1983. Aspects of sheep husbandry

systems in Aleppo province of northwest Syria. Aleppo, Syria, ICARDA.
Farming Systems Program. Research Report. 76 p.
THOMSON, E.F. 1984. First experiences with joint managed forage and
grazing trials. In Farming systems research symposium: Animals in the
farming System (3., 1984, Manhattan, Kansas). Proceedings. F.C.
Butler (Ed.). Kansas State University. Farming Systems Research
Paper No. 6. p. 234-250.
THOMSON, E.F.; BAHHADY, F.A. 1988. A note on the effect of live weight
at mating on fertility of Awassi ewes in semiarid northwest Syria. Animal
Production 47:505-508.
THOMSON, E.F.; ORITA, G. 1988. Seasonal prevalence of lungworms of

Awassi sheep in northwest Syria. Tropical Animal Production and
Health 20:187-189.
THOMSON, E.F.; BAHHDAY, F.A.; MARTIN, A. 1989. Sheep husbandry

at the cultivated margin of the northwest Syrian steppe. Aleppo, Syria.
ICARDA. ICARDA-148 En. 92 p.
THOMSON, E.F.; CECCARELLI, S. 1991. Progress and future directions

of applied research on cereal straw quality at ICARDA. In Production
and utilization of lignocellulosics: Plant refinery and breeding, analysis,
feeding to herbivores and economic aspects. G.C. Galletti (Ed.). New
York, Elsevier Applied Science. p. 249-264.
THOMSON, E.F.; BAHHADY, F.A. 1992. On-farm evaluation of pasture

and feed legume crops for increasing sheep production in cereal-based
farming systems in west Asia. In Livestock in Mediterranean cereal
production systems. Wageningen, The Netherlands, PUDOC. p. 219-
224.
THOMSON, E.F.; JAUBERT, R.; OGLAH, M. 1992. Using on-farm trials to

study the benefits of feed legumes in barley-based rotations of
northwest Syria. Experimental Agriculture 28:143-154.
THOMSON, E.F.; BAHHADY, F.A.; NORDBLOM, T.L. 1993. Sheep and

feed outputs, and productivity from model farms in northwest Syria
managed at different levels of intensity. In Animal production in
developing countries: Resolving technical and socioeconomic
constraints. M. Gill, E. Owen, G.E. Pollott, T.L.J. Lawrence (Eds.)
Edinburgh, U.K., British Society of Animal Production, Occasional
Publication No. 16. p. 176-178.
TULLY, D.; THOMSON, E.F.; JAUBERT, R.; NORDBLOM, T.L. 1985. On-
farm trials in northwestern Syria: Testing the feasibility of annual forage
legumes as grazing and as conserved feed. In IDRC-ICARDA
Workshop on livestock on-farm trials. T.L. Nordblom, A.K. Ahmed, G.R.
Potts (Eds.). Ottawa, Canada, International Development Research
Centre. Publication IDRC-242e. p. 209-236.
RESEARCH AT CIAT ON RUMINANT
PRODUCTION SYSTEMS
Raul R. Vera'
INTRODUCTION AND BACKGROUND
Since its beginnings in the late 1970s, farming systems research (FSR)
on ruminant production systems at CIAT has tried to avoid the trap of
believing that diagnosis of problems can substitute for technology
generation. In practice, the opposite may well have been the predominant
case. Based on the assumption, firmly grounded in the empirical
knowledge of highly experienced people, that undernutrition of grazing
ruminants is at the root of poor animal productivity in most of the tropical
American lowlands, technology generation paralleled, and in many
instances preceded, detailed problem identification. In turn, on-farm
research (OFR) attempted to find or define niches for the newly generated
technologies, most of them based on new germplasm, and also
documented unanticipated uses and adaptations of those technologies by
grazers and other users.
The above approach has been conditioned by the fact that in CIAT
research on animal production systems is part of a germplasm program.
Although this characteristic may have constrained and/or biased some of
the FSR activities, the fact that attention was initially concentrated on
systems where ruminant production was the dominant enterprise lessened
the risks of ignoring other significant products as well as interactions
among components of the farming system. On the plus side, it should be
mentioned that this particular organization was probably a determining
factor in keeping FSR activities highly focused. By the same token, both
system diagnosis and on-farm testing of emerging technologies became
institutionalized as an integral part of germplasm evaluation. The
approaches used can best be illustrated by considering two different
cases.
' Tropical Pastures Program, International Center for Tropical Agriculture (CIAT),
Cali,
Colombia.
PROBLEM IDENTIFICATION
In the late 1970s, a major diagnosis of the predominant ranching

systems was conducted at three representative sites in Latin American
tropical savannas, located in Brazil, Colombia, and Venezuela (Vera and
Sere 1985b). Despite the massive nature and cost of this undertaking, it
began with a straightforward hypothesis, which was relatively easy to test.
The prevailing view at the time affirming the existence of large between-
ranch, within-region differences in technical and economic performance
associated with differences in management practices for beef breeding
herds, which had originated from a limited number of on-farm observations
over a short period of time (Stonaker et al. 1976), was rejected. In
addition, the fallacies of the methodologies used to arrive at that initial
hypothesis were revealed. Much work has since been conducted to
elucidate appropriate methodologies for studying reproductive and other
events in extensive beef breeding herds (Amezquita and Vera 1988;
Amezquita et al. n.d.). Even before the study terminated and its results
were analyzed, the project had an important influence in helping reorganize
CIAT's Beef Production Program, which evolved into the present Tropical
Pastures Program, thus recognizing the overwhelming influence of
undernutrition in determining system output.
Similar highly detailed, long-term, costly studies have not been

repeated since then for several reasons. The first one is the high cost in
terms of both financial resources and time. Perhaps equally important is
that, with time, the accumulation of empirical experience and knowledge
led to the use of various other approaches which, although less precise,
were probably far more cost-effective. Two examples in contrasting
socioeconomic settings follow.
THE CASE OF CAQUETA, COLOMBIA
Methodology
The Caqueta Department of Colombia is located in the Amazon basin.

The region was subject to an intensive process of colonization by
dispossessed small farmers in the 1940s and 1950s (Michelsen 1990).
Land tenure has since been legalized, and most settlers own about 30-50
ha each, a size that is considered small for the region as a whole.
Identification of problems and opportunities and the generation of

hypotheses were approached initially through an informal and unstructured
visit to the region, and the visit was similar to a rapid rural appraisal (RRA).
In effect, an interdisciplinary group of scientists, with the participation of
local technical assistants, jointly visited a number of farms in the region,
which is characterized by the almost total predominance of dual-purpose
cattle production systems (Michelsen 1990). This led to a rapid
socioeconomic survey of a restricted universe of farms selling milk to a
multinational enterprise active in the region. A limited number of questions
addressing technical issues was included, and a survey of pastures sown
to Brachiaria decumbens was undertaken (Ramirez and Sere 1989), since
it was hypothesized that pasture degradation in the region was a common
phenomenon, responsible for decreases in carrying capacity and animal
production. The above sequence of events lasted approximately 18
months. Even before the results (Ramirez and Sere 1989) had been
completely analyzed, work preparatory for OFR began, consisting mostly
of the establishment of several seed multiplication areas in cooperation
with both private and government institutions active in the region. These
areas would eventually provide some of the seed (both true seed and
vegetative material) for on-farm testing.
The process of selecting pasture species and mixtures for OFR was
based to a limited extent on locally run small-plot experiments, and to a
much larger extent on information originating in similar agroecological
zones in other countries, available in an extensive data base at CIAT. This
approach clearly illustrates the unplanned, but efficient, interaction between
a highly structured network of agronomic, small-plot experiments promoted
by CIAT throughout the tropical American lowlands and the eventual
selection of germplasm for on-farm testing.
CIAT implemented a very limited program (six farms) of on-farm testing

of grass-legume associations in comparison with grass-alone pastures.
Although other institutions effectively contributed to the identification of the
cooperating farms, the actual monitoring of the results was carried out
solely by CIAT. Nevertheless, this activity encouraged other institutions to
pursue parallel activities addressing the transfer of the technologies being
tested by CIAT. In effect, an increasing number of institutions have joined
the effort without formal agreements, have established additional seed
multiplication lots, and are promoting the planting of the species thus
multiplied. In addition, a local university is actively conducting supporting
research with the help of undergraduate students.
The OFR project is testing the hypothesis that milk production, based
on crossbred cows (Zebu x European breeds), is higher on grass-legume
pastures than on the corresponding grass alone. Due to restrictions on
seed supply and other resources, it was decided initially to establish
relatively small areas (3-6 ha per farm for each type of pasture). This
implies that both the control paddock and the mixture become part of the
normal rotation used by the farmer, generally involving between five and
ten different paddocks. Therefore, the main methodological problem is to
quantify the effect of pasture type on milk yields, since it is confounded
with stage of lactation.
Critical analysis of the project
Within the FSR/OFR experience of the Tropical Pastures Program, this

case has been a resounding success, to the extent that it is highly
probable that if CIAT were to discontinue its on-farm activities in the region,
many if not all of the activities, especially those more closely linked to
technology transfer, would most likely continue unaffected. Let us analyze
this successful outcome.
The author believes that the foremost reason is, on the technical side,
the success of some (although not all) of the pasture species introduced.
The successful ones have proved to be persistent despite varying levels
and intensities of management applied by farmers and have generally
shown their resilience in a variety of different microenvironments. Although
the technical results regarding animal production with these pasture
species (as opposed to the controls) have not yet been completely
analyzed, the majority of the admittedly small number of farmers that were
initially exposed to them are convinced that the new species lead to
improved animal productivity, and several have expanded the areas sown
with them. The strong commitment from several of the institutions involved
is undoubtedly the reason for the continuing existence of joint activities.
Perhaps even more important is the existence of committed individuals in
the institutions. The contribution from private industry has been a key
impetus by providing an assured outlet for any foreseeable increase in milk
production, through the provision of some important inputs (veterinary
medicine, mineral supplements, seeds, etc.), and by making technical
assistance available.
retrospect, it is possible that this one-sided effort in diagnosis and

In
on-farm testing may have ignored some problems and opportunities. For
example, a more comprehensive diagnosis of farming systems might have
led to the incorporation of some forestry components in a region well

known for the extent of deforestation. Similarly, it could be argued that
little or no effort was made to explore the possibilities of reclaiming
degraded pastures via a crop, although the region's topography and likely
marketing constraints make this a doubtful proposition. Even within the
livestock sector of farms, only one constraint has been consistently dealt
with: animal nutrition. Other possible constraints such as animal genotype
have been mostly ignored. On balance, it is argued that when financial
and human resources are extremely limited, a sharply focused effort is
required to rapidly achieve enough success to facilitate continued
progress.
From a methodological point of view, the main difficulty concerns

quantifying the effect of pasture type on milk yield. The design clearly
does not allow for orthodox analyses. Furthermore, other equally
significant effects cannot be measured. For example, as shown in a
separate and controlled experiment (CIAT 1991), a given pasture type may
lead to higher weight gains by cows, which in turn is associated with
improved reproduction. These effects would be totally ignored the animals
rotated through numerous and different pastures. It is a potentially serious
error, as modeling has shown that improved reproduction means an
increased lactation, a fact that leads to a larger difference over the lifetime
of the cow than that observable in one or two consecutive lactations, in
which reproductive differences are ignored or confounded (unpublished
results). Given the complexity of these relationships hypothesis that the
cow's milk production, by virtue of its sensitivity to changes in nutritional
status, is and the ease of its measurement, a relatively simple index of
potential differences between pastures would not be valid. On the other
hand, measuring both milk yields and reproductive performance would
involve not only a much longer observation period, but would also imply
having separate herds for different feeding regimes; this type of
management appears unrealistic in on-farm research, at least in CIAT's
view. Thus, it is suggested that if milk is such a relevant output and cannot
be replaced by another index of animal productivity, then a combination
of on-farm records and on-station experiments, adequately "discounted,"
will have to be used as inputs for mathematical models of animal
production.
As suggested above, even the analysis of milk yield records can cause
difficult methodological problems. Although not the case for Caqueta, milk
yields recorded by both the farmer (voluntarily) and the researcher can be
extremely variable from day to day. An extreme case, from a similar
project in Pucallpa, Peru (also within the Amazon basin), is illustrated for
a single farm and cow inFig. 1. The variability in this case is extreme and
is probably due to the lack of tradition in the region for milking, the
absence of reliable outlets for the product, and the general lack of
expertise among farmers in managing cattle, even though raising livestock
is an important activity there.
A A
A
A
IIA AA
A A
O
0
1
N
h
1
O
h
h
Q
I I
O
Q
1
O
M
I
O
M
i
h
N
i
O
N
' 1
1 O
m
r IM IL
Mm
MOO/ADP/6> 'PIaiR ij iyy LL
THE CASE OF THE COLOMBIAN SAVANNAS
Methodology
In contrast to the previous case, the well-drained Colombian savannas

are characterized by extensive ranching systems. Cow-calf operations
predominate, and over the last 10 years there has been an increasing trend
toward incorporating fattening and, to a much lesser extent, milk
production, based on dual-purpose herds located close to main roads and
towns. An important but decreasing percentage of ranchers are absentee
landowners.
Following the diagnostic study referred to earlier (Vera and Sere

1985b), which identified nutrition as the main limiting factor, experimental
grass-legume pastures were established on a large number of cooperating
ranches (for details, see Vera and Sere 1985a, 1989). Given that the
majority of the paddocks were of a size such that a well-identified group
of animals could be maintained on them year-round, it was relatively easy
to measure animal output in several types of pastures. Over a period of
four to six consecutive years, the technical and economic viability of a
grass-legume pasture was well documented, in terms of weight gain by
steers and replacement heifers and improved reproductive performance in
heifers and cows (Vera and Sere 1989). These results were used
extensively in ex-post economic analyses (Vera and Sere 1989) and in
modeling (Thornton and Vera 1988).
As a consequence of the success of some of the pastures involved, the

Colombian national research institute, ICA2, gradually developed a
program of technology transfer for the region, centered on pasture
development.
Critical analysis of the case
Undoubtedly, the project was successful in first demonstrating the

validity of the hypothesis that grass-legume pastures would contribute to
increased animal production and pasture persistence. Furthermore, it led
to the development of a successful program of technology transfer where
none existed before. The analysis of the results was straightforward and
2 See List of Abbreviations and Acronyms.

relatively orthodox, and there was a massive amount of results available,

regarding both animal production and pasture parameters.
The main limitations on the project were essentially two. In the first
place, it was relatively expensive in terms of time and physical resources.
It is doubtful that this exercise would be repeated today, but it made sense
at a time when there was a need to establish the credibility of the
technology proposed, both at the institutional level and outside it. The
project was instrumental in promoting the release of a commercial cultivar,
a legume whose performance on-station was far less impressive than on-
farm (Ferguson et al. 1989), and also focused attention on a grass which
on-station research had rejected due to low nutritional value.
Simultaneously, the project served to direct the attention of on-station
researchers to soil types poorly studied until then.
Technically, an important limitation on the project was the absence of

proper control pastures (in this case, grass-alone pastures). Although this
comparison was not required to test the initial hypothesis regarding
persistence and productivity of grass-legume pastures, it would have
added credibility to the results.
OUTLOOK
CIAT has elaborated, and the CGIAR3 system has approved, a new
Strategic Plan for the 1990s, which will have a substantial impact on
FSR/OFR activities. The new plan indicates that the existing germplasm
programs will consolidate around germplasm development and
characterization and will gradually move upstream in these and related
activities. A new division will be created to deal with issues of natural
resource management in three selected agroecosystems of tropical
America, namely, the savannas, humid tropics, and hillsides, with overall
support from a Land Use Program that will deal with larger issues. The
three agroecosystem programs will incorporate systems research that up
to now has been the responsibility of the germplasm programs. This move
should facilitate simultaneous consideration and integration of various and
diverse system components. Nevertheless, it should not be assumed that
the new programs will focus exclusively on highly applied or downstream
3 See List of Abbreviations and Acronyms.

activity, since an important hypothesis underlying the new Strategic Plan

is that, at the level of agroecosystem resolution, there are important issues
of strategic research with wide applicability and relevance.
LITERATURE CITED
MEZQUITA, M.C.; VERA, R.R.; LEMA, G. n.d. Pasture evaluation with

breeding herds: A methodology for data analysis. Cali, Col. CIAT.
(Manuscript).
CIAT (INTERNATIONAL CENTER FOR TROPICAL AGRICULTURE). 1991.

Informe Anual 1990. Programa de Pastos Tropicales. Cali, Col. CIAT.
Documento de Trabajo No. 89. irr. p.
FERGUSON, J.E.; VERA, R.R.; TOLEDO, J.M. 1989. Andropogon gayanus

and Stylosanthes capitata in the Colombian Llanos: The path from the
wild towards adoption. In International Grassland Congress 16., 1989,
Nice, France). Proceedings. p. 1343-1344.
RAMIREZ, A.; SERE, C. 1990. Brachiaria decumbens en el Caquet&

Adopcibn y use en ganaderias de doble proposito. Proyecto
Colaborativo Nestle de Colombia, Fondo Ganadero del Valle del
Cauca, INCORA, SENA, Universidad de la Amazonia, ICA, CIAT. Cali,
Col. CIAT. Documento de Trabajo No. 67. 118 p.
MICHELSEN, H. 1990. Analisis del desarrollo de la produccibn de leche

en la zona tropical humeda. El caso del Caqueta, Colombia. Call, Col.
CIAT. Documento de Trabajo No. 60. 68 p.
STONAKER, H.H.; VILLAR, J.; OSORIO, G.; SALAZAR, J. 1976.

Differences among cattle and farms as related to beef cow
reproduction in the eastern plains of Colombia. Tropical Animal Health
and Production 8:147.
THORNTON, P.K.; VERA, J.; VERA. R.R. 1988. Modelo de simulacibn para
los sistemas de produccion de carne en los Llanos Orientales de
Colombia. Pasturas Tropicales 10(1):8-13.
VERA, R.R.; SERE, C. 1985a. Evaluation of tropical pastures species with

a farming system perspective. In International Grassland Congress
(15., 1985, Kyoto, Japan). Proceedings. p. 1187-1188.
VERA, R.R.; SERE, C. (Eds). 1985b. Sistemas de produccion pecuaria

extensiva: Brasil, Colombia, Venezuela. Informe final Proyecto ETES
(Estudio Tecnico y Economico de Sistemas de Produccion Pecuaria
1978-1982). Cali, Col. CIAT. 530 p.
VERA, R.R.; SERE, C. 1990. On-farm results with Andropogon gayanus.

In Andropogon gayanus Kunth: A grass for tropical acid soils. J.M.
Toledo, R. Vera, C. Lascano, J.M. Lenne (Eds.). Cali, Col. CIAT. p.
303-333.
ASIAN RICE FARMING SYSTEMS NETWORK:
COLLABORATIVE RESEARCH ON
CROP-ANIMAL FARMING SYSTEMS
Virgilio R. Carangal'
BACKGROUND
Many rice farmers in developing countries practice mixed farming,

diversifying the use of their resources to produce food for consumption
and income. The production systems practiced are very complex, and
research that uses a commodity approach with high discipline orientation
is not sufficient to develop appropriate technologies for small-scale
farmers. A systems approach to research was started in 1974 at IRRI2
with emphasis on-farm research with farmers participating in the design
and testing. In 1975, the Asian Rice Cropping Systems Network was
established in collaboration with national programs to identify the more
productive rice-based cropping systems acceptable to small-scale farmers.
A conceptual framework for cropping systems research focusing on
farmers' fields with farmers participating in the research process was
developed. The number of participating countries increased from three in
1975 to 12 in 1982 (Bangladesh, Bhutan, Cambodia, Laos, India,
Madagascar, Malaysia, Myanmar, Pakistan, Sri Lanka, South Korea and
Vietnam).
In 1983, the Asian Rice Cropping Systems Working Group, consisting

of the national program leaders and two scientists from IRRI, decided to
expand activities to include animal production because most rice farmers
are engaged in animal production, and, in many cases; receive greater
income from animals than from crop production. The name of the network
i Agronomist and Coordinator, Asian Rice Farming Systems Network, International Rice
Research Institute, P.O. Box 933, Manila, Philippines.
2 International Rice Research Institute.

was changed to Asian Rice Farming Systems Network (ARFSN) in order

to include not only cropping but also animal production, aquaculture and
agroforestry. In 1985, collaborative research on crop-animal farming
systems research was initiated in the Philippines, Indonesia, Thailand and
Nepal; in 1982 China was added to this list, bringing to 17 the number of
participating countries.
provides the coordination of the ARFSN. The coordinating unit is

IRRI
composed of a network coordinator, a senior research assistant, three
research assistants and 10 other supporting staff. Since IRRI's
comparative advantage is on rice, the ARFSN taps national research
institutions to conduct research on other commodities involved in the
production systems and to provide technical backstopping to countries
collaborating with the ARFSN. The lead research institutions are the
University of the Philippines at Los Banos Institute of Animal Science (IAS)
for animal nutrition research and Khon Kaen University for forage crops in
rice farming. A senior animal scientist from IAS and a forage crop
agronomist from Khon Kaen University help the ARFSN Coordinator
provide technical support not only to the key crop-animal farming systems
sites, but also to other countries collaborating with the ARFSN.
Funding for the ARFSN activities comes from different sources. The
coordination unit is funded by the IRRI core budget and IDRC.
Collaborative activities in different countries are mostly funded by the
national agricultural research institutions (NARIs), with some support from
IDRC, USAID, Ford Foundation, the International Fund for Agricultural
Development, the Asian Development Bank and others. Most funding from
the donors is given directly to the country.
OBJECTIVES
The main objective of the ARFSN is to facilitate collaboration between

IRRI and national scientists in order to increase food production in Asia by
identifying more productive rice-based farming systems acceptable to
small-scale farmers. The specific objectives for the crop-animal farming
systems collaboration are given below:
To establish collaborative research on crop-animal production systems

between IRRI and national programs in Asia.
To develop and refine crop-animal research methodologies to be used

in the ARFSN.
To further develop relevant component technologies for both crops and

animal systems.
To facilitate exchange of research information and ideas among

scientists in Asia on crops and animal production systems.
COLLABORATIVE RESEARCH
On-station research highlights
On-station research on forage crop production in rice-based farming

systems is conducted at IRRI, whereas animal nutrition work is conducted
at IAS. The major constraint to animal production is availability of feeds
during the dry season or after the rice cropping season. Small-scale
farmers will not grow forage crops in place of food crops since animal
production is a secondary enterprise. The on-station research at IRRI
focuses on forage crops in irrigated lands, rainfed lowlands and upland
rice-based farming systems, with the objective of producing food grains,
animal feed and green manure.
Several food-forage crop production systems were evaluated, and the

most promising combinations in rainfed lowlands when grown after rice
were food legumes intercropped with Siratro, Crotolaria, Desmanthus and
Verano Stylo (Stylosanthes guyanensis) or cowpea intercropped with
Setaria sphacelata and Setaria splendida; and before rice, mungbean
intercropped with Sesbania.
Under irrigated conditions, the most promising combinations were

sorghum and corn intercropped with Desmanthus and Crotolaria after rice
and mungbean intercropped with Sesbania before rice. In the upland
ecosystems, the most promising after upland rice were Verano Stylo
intercropped with corn and cowpea and peanut intercropped with
Desmanthus. These promising food-forage crop production systems
showed higher forage and grain yields and did not affect the yield of the
food crops. Beneficial effects on the yield of the following rice crop were
obtained after the last forage cutting was incorporated in the soil as green
manure.
Desmanthus is the most promising legume when grown either after rice
both in lowland and upland, as monocrop in the upland, or as a hedgerow
intercropped with food crops. It is now being tested in different sites in the
Philippines, Thailand, Myanmar, Vietnam and Bangladesh.
Seven grasses and four legumes were established on lowland paddy

bunds. Andropogon gayanus, Setaria splendida, Pennisetum purpureum
and Setaria sphacelata were the most promising grasses. Among the
forage legumes the best were Desmanthus virgatus and Clitoria ternatea.
Cropping-pattern trials for rice with forage grass were conducted under
rainfed lowland conditions to evaluate forage production after rice using
four forage species. Brachiaria mutica showed the highest forage yield;
rice following the Setaria species gave significantly higher yields, as
compared to the fallow plot.
Dual-purpose food crops were screened for high stable yield and
biomass production. The most promising crops were cowpea, peanut,
mungbean, pigeon pea and sorghum. There were significant differences
between varieties in terms of biomass production. Protein content was
higher in the legume crop residues than in the cereal residues. There was
a positive correlation between biomass and bean yield in peanut and
pigeon pea.
Several varieties of rice were evaluated for forage and grain yield. They
were cut for forage 40 days after transplanting and then allowed to grow
to maturity for grain production. The grain yield of the uncut treatment
was higher, and the difference between the cut and uncut treatments
ranged from 0.7 MT/ha (IR48) to 1.51 MT/ha (fR74). Yields of fresh forage
obtained by cutting the rice plants were from 3.27 to 5.33 MT/ha.
Six forage grasses and five legume species were tested under upland
conditions. Andropogon gayanus gave the highest yield of forage dry
matter (21.6 MT/ha) and Setaria splendida the highest fresh forage yield
(80.6 MT/ha). Among the forage legumes, Desmanthus had the highest
dry matter yield (18.4 MT/ha).
On-station and on-farm animal nutrition research in the Philippines

focuses mainly on the establishment of feeding values of potential crops
for rice-based farming systems. Five grasses, five legumes, three grain
stovers and 10 grass-legume mixtures were evaluated in terms of chemical
composition and digestibility by cattle and water buffalo. In terms of
energy, Napier (Pennisetum purpureum) and Leucaena gave higher values
among grasses and legumes, respectively. Lowest digestible energy

values were obtained with rice straw, Acacia leaves and pigeon pea leaves.
A marked improvement in the nutritive value resulted from feeding a
combination of grass and legume at a 1:1 ratio when compared to a diet
of only grasses or legumes. Generally, the water buffalo digested dry
matter better than cattle, especially when the ration had a high lignin
content.
Treatment of straw with a lye solution from ash of Amaranthus

spinosus, and rice hull significantly improved dry matter intake and
digestibility for both cattle and water buffaloes.
Two digestibility trials were conducted to compare Sesbania rostrata

with (1) a concentrate as a supplement to grass and (2) mungbean hay as
a supplement to rice straw. In both trials, the supplements were given at
1 % of the body weight. The Sesbania-grass mixture had higher digestibility
values for dry matter, crude protein, cellulose and energy than the grass-
concentrate mixture. However, animal performance was better when
concentrate was fed because intake of Sesbania was limited by its pungent
odor. The live weight gain and feed efficiency were similar in cattle given
mungbean hay as a supplement to straw and in those supplemented with
concentrate.
Two feeding trials with sheep were conducted to evaluate the feeding
value of Gliricidia in different forms and levels in rations based on rice
straw. In the first trial, fresh and dried Gliricidia leaves given at 0.3 and
0.6% of live weight were compared. Results showed that fresh Gliricidia
supplementation significantly increased the voluntary intake of rice straw;
however, there were no significant differences in average daily gain and
feed efficiency among treatments. In the second trial, an increasing level
of Gliricidia (20, 40, 60 and 80% of the ration), as a replacement for rice
straw, showed a corresponding improvement in dry matter and protein
intake of animals; sheep fed 80% Gliricidia had the highest live weight gain
(54.4 g/day).
Crop-animal farming systems key sites
Seven cropping systems sites representing different rice ecosystems

were identified for inclusion of the animal component. These were
Batumarta (Indonesia) and Trece Martires (Philippines) representing the
upland rice ecosystem; Beijing and Zhenjiang (China) representing the
irrigated rice ecosystem; Naldung (Nepal), Khon Kaen (Thailand) and Santa
Barbara (Philippines) representing the rainfed lowland rice ecosystem. The

experiments differed from one location to another depending on the
characteristics of each site
Collaboration in the Philippines is with the Farming Systems and Soil

Resources Institute, Department of Agriculture and Philippine Rice
Research Institute; in Thailand with the Farming Systems Research
Institute, Khon Kaen University and the Department of Livestock
Production; in China with the Chinese Academy of Agricultural Sciences
and Jiangshu Academy of Agricultural Sciences; in Indonesia with the
Central Research Institute for Food Crops and Central Research Institute
for Animal Production; and in Nepal with the National Agriculture Research
Center.
The basic research approach used is similar to the cropping systems

approach except that in the diagnosis or site description a more detailed
information was obtained on the animal component, with emphasis on
production practices, types and uses of animals, feed sources and
practices, problems in crop-animal systems and their interaction with other
on-farm or off-farm enterprises. The steps followed included selection of
target environment, diagnosis of the research site, component or
production systems research, and testing of designed component or
production system. There are three kinds of interventions in the key crop-
animal farming systems sites: intervention in crop and animal systems,
crop-animal integrated trials, and whole-farm testing. In all the steps
mentioned above, farmers are involved and form part of the research team.
On-farm research highlights
Philippines. The two key sites are located in Santa Barbara,

Pangasinan and Trece Martinez, Cavite. During the first three years in
Santa Barbara, cropping pattern testing was conducted to compare rice-
mungbean and rice-cowpea with traditional monocrop rice. Because of the
profitability of the rice-mungbean alternative, the proportion of farmers who
adopted this pattern increased from less than 20% in year 1 to 90% in year
4. Cowpea was introduced in anticipation of an integrated study of
cowpea-pig fattening, with cowpea as a substitute for soybean oil meal in
the pig diet. At present, the rice-mungbean pattern has replaced that of
rice-fallow, while cowpea as a follow-up crop was neglected by the
farmers.
Superimposed component technology studies of the rice-mungbean

pattern were conducted. A fertilizer rate of 30 and 40 kg N/ha for early-
and medium-maturing rice varieties resulted in greater economic benefits
than the farmers' practice. The use of green manure from Sesbania
rostrata in combination with inorganic fertilizers indicated that this can
substitute for 36 kg N/ha from urea. Among the introduced rice varieties,
IR65, a glutinous rice variety, has outperformed the traditional Diket variety
and has since been adopted by the farmers. Directly seeded rice showed
some advantages in terms of higher yield from the main crop and an
additional yield from ratoon crops and lesser tillage operations.
Experiments on post-rice establishment of mungbean showed that
optimum tillage (consisting of plowing, broadcasting of seed and
harrowing) is needed to increase grain yield.
Intercropping Siratro with cowpea and mungbean increased both the

food grain and fodder yields. The soil incorporation of herbage from the
last cutting of the forage legume resulted in higher yields of succeeding
crops. Several cattle feeding studies were conducted using legumes and
in some cases urea, as supplemental feed, to increase protein intake. The
legumes used were Leucaena and Sesbania rostrata. Results indicated
that the average daily gain of experimental animals was increased, as
compared to the control group.
In Santa Barbara, Pangasinan, the total family net income increased

from US$414 in 1984 to US$975 in 1990. There was also a notable
Increase in the contribution of livestock to net farm income, from 26% in
1984 to 60% from 1987 to 1990. .
A study of draft animal power was conducted in order to develop an

in-field draft measurement methodology and to assess the draft animal
performance of cattle and carabaos as affected by rice bran
supplementation. Results showed that the power generated is directly
proportional to the draft force exerted by the animals. Both types of
animals could pull draft equivalent to 10 to 15% of their live weight at a
sustained speed. Pulse rate and respiration rate increased progressively
with heavier draft, although environment had a greater influence on these
physiological parameters, the values being lower during the cooler part of
the day. Despite rice bran supplementation, loss in live weight occurred
during October, November and December in both cattle and carabaos.
However, cattle were able to make moderate live weight gains during the
rest of the year, compared with carabaos which lost weight during the
year.
fattening trial comparing the utilization of cassava plus cowpea

A pig
with the traditional feeding practice used by farmers were conducted at
Pangasinan and Cavite. Results showed that pigs from experimental and
control groups had similar gain in weight but pigs fed the cassava-cowpea
diet required a lesser amount of feed (245 kg) than the control pigs (318
kg) and had a better feed conversion efficiency (4.22 vs. 5.39 kg/kg). Cost
and return analysis showed a reduction in cost of feeding when the
experimental ration was used.
Indonesia. The key site in Indonesia, located in Batumarta, South

Sumatra, is representative of upland rice ecology and is a transmigration
area. In 1989, preproduction trials were conducted in six village units with
30 cooperators. The average farmer's income in 1989 ranged from
US$567 to US$1,670/year. About 65% of the cash income came from
rubber, 11.5% from animal production and only 4.8% from food crops.
Cattle provided a higher income when compared to goats and chickens.
After two years, the number of chicken increased 100% and the
recommended cropping patterns were modified based on farmers' needs
and available resources. An experiment on grazing goats was conducted
on a block of land planted with rubber: five goats were allowed to graze
day and night on about 2,075 m2 of land; there was more than enough
herbage available and only 37% of the potential production was consumed.
No damage of the trees was observed. Weight gain of the bucks after five
months of trial was only 3 kilograms.
In addition to the crop-animal activities, there are many component

research activities at the Batumarta site, such as variety trials of upland
rice, mungbean, cowpea and irrigated rice; alley cropping using Flemingia
and Gliricidia with food crops (upland rice and cowpea); long-term trials
consisting of year-round cropping (intercropping and relay cropping) with
NPK, lime and manure treatments; and nutrient toxicity experiments.
Nepal. The research activities at crop-animal farming systems sites in

Nepal started in Pumdi Bhumdi, a rainfed lowland rice ecosystem in the
midhills. Farmers introduced Napier grass at the edge of the rice terraces
and planted oats after rice in areas where only one rice crop was grown;
with the introduction of oats, milk production increased. In 1988, the crop-
animal key site was transferred to Naldung, near Kathmandu.
The major intervention at the new site consisted of planting forage

grass in bunds for use during the dry season and the planting of fodder
trees not only for feed but also for fuel and timber. Other activities
included rice-wheat cropping pattern trials, variety trials with rice, wheat,
corn and other less important crops. Several forage crops were
introduced, but farmers preferred Napier. In 1989, 13 species of fodder
trees were planted in 63 sets of terraces to study the effects of trees on
food crops and fodder yield of trees. A community silvipastured forest was
established in a 5-ha area.
China. There are two key sites in China: Zhenjiang, Jiangsu and
Changping, Beijing. In Changping, 7 ha were planted with triticale, after
rice, to produce silage for dairy cows; the average yield was 30 MT/ha in
1989. A yield trial involving nine triticale varieties showed that the varieties
8-Forage-6, WOH 813-F4, 8-Forage-5 and WOH 59 were better than the
control variety H-1990 in terms of yield and maturity. Economic benefits
from different cropping patterns in the site were monitored; triticale-rice
showed the highest net returns (US$1,025/ha), followed by wheat-maize
for silage (US$553). Farmers within the site increased the area of triticale
to 30 ha in 1990.
In Zhenjiang, corn was introduced in the rice-based cropping pattern

to increase the feed supply for swine. The upper paddy fields were grown
to wheat-corn, rapeseed-corn and barley/corn-mungbean or radish or
Chinese cabbage instead of rice-wheat or rice-barley. Corn is now being
used in expanding areas. Feeding trials on improved and traditional feeds
given to different swine breeds showed that the improved Yorkshire Taihu
breed, when fed with mixed feed, showed the highest feed efficiency,
shortest feeding duration and highest net return; the second best
performance was obtained when Yorkshire Taihy was fed the traditional
feed; a local breed with traditional feed showed the worst performance.
The cooperative dairy farm in the Zhenjiang site grew a dual-purpose corn
hybrid on 17.5 ha; the average yield was 4.54 MT/ha of grain and 18.8
MT/ha silage. Rye cultivar Dongmu 70 and triticale Oh 1194 were
evaluated with wheat as silage; Dongmu 70 showed the highest fodder
yield (71 MT/ha) and crude protein (1.41 MT/ha).
Thailand. In Thailand, the first phase of the crop-animal farming

systems key site was conducted in Ban Phai, Khon Kaen, which represents
the drought-prone rainfed lowland. Farmers were divided into three
groups: animal-based, crop and animal-based; and crop-based. The
traditional cropping patterns were monocrop rice in the lowlands and
cassava in the uplands. The project introduced peanut-green corn, green
corn-mungbean or cowpea (instead of rice in the upper paddy) and
peanut-green corn, green corn-cowpea, cassava intercropped with green
corn and peanut in the upland. Backyard forage production utilizing Stylo,
Napier and Ruzi (Brachiaria ruziziensis) grass was also introduced for cut-
and-carry feeding in the evenings. However, it was observed that the more
advanced farmers preferred to grow grass in the upper paddy. Artificial
insemination was also introduced to produce American Brahman grade
cows and Holstein Freisian (heifers for dairy and males for beef
production). Calves were also produced for sale to the Central Plain rice
farmers. The net cash farm income of the animal-based farmer was 100%
higher than crop-animal based farmers and 207% higher than crop-based
farmers. In all groups, the income from animal production was higher than
that from crop production.
The methodology and technology from the activities in Ban Phai were
tested in other locations within the same agroecological zone. Three sites
were selected: a rice-based site in Huaybong, Chaiyaphum; a cassava-
based site in Phonpek, Khon Kaen; and a sugarcane-based site in Dong
Muangam, Khon Kaen.
Site description has already been completed and problems have been
identified. The research activities in the rice-based site included: (1)
introduction of additional feed resources such as fodder trees in upper
paddy land and around the homestead; (2) upland crops after rice for grain
production and crop residue for animal feed (peanut, corn, cowpea and
mungbean); (3) backyard pasture (Ruzi, Guinea and Verano Stylo); (4)
forage crops in rice bunds; (5) intercropping forage crops with fruit trees;
(6) increasing the rice production by growing rice in the fallow land (upper
paddy) using direct seeding techniques; and (7) intercropping rice with
Stylo and prevention of animal diseases.
Expansion crop-animal farming systems research
As a result of network activities, the number of crop-animal farming

systems sites in Asia increased from five in 1985 in four countries to 72 in
nine countries in 1990. In the Philippines, approximately 28 sites formerly
studying only cropping systems now include the animal component. The
number of sites also increased in Indonesia, Nepal and Thailand. In
addition, other countries involved in the network, such as Myanmar,
Bangladesh, Pakistan, Korea, and Bhutan, are conducting crop-animal
farming systems research using an approach similar to that of the key
sites. In Bangladesh, animal scientists joined the FSR teams from various
government agencies and institutions; however, in many sites, the
intervention consisted only of forage crops introduced in the rice system
to increase feed availability especially during the dry season.
Forage crop research in rice farming has also expanded in different

countries. In Korea, long-term rice-forage crop trials including different
forage-rice sequences, are being conducted. The most promising forage
crops are Italian rye grass in the southern and central regions and rye
grass in the northern region. In Bangladesh, food-forage crop
intercropping experiments similar to the activities at IRRI are being
conducted. Pakistan is working mainly on rice-berseem with emphasis on
identifying better cultivars that can be grown after rice. Myanmar, Bhutan,
Madagascar, Indonesia, and Thailand are also conducting forage crop
research in rice farming.
Sharing of research information
One of the major roles of the ARFSN is to facilitate the exchange of

research information and methodologies used in farming systems among
scientists in Asia and elsewhere. The research output of the key sites is
disseminated to national programs through workshops, monitoring tours
and meetings. Countries not involved in the collaborative research are
invited to the workshops where methodologies and technologies are
discussed.
Two monitoring tours were organized: one to Nepal and Indonesia in

1985 and the other to China and the Philippines in 1987. The objectives
were to visit the crop-animal farming systems sites, review the plans and
programs for crop-animal farming systems research, and review the
methodologies used. This was an effective mechanism at the beginning
of the network collaboration.
The network also organized four workshops. These workshops

focused on the progress of the collaborative research activities, research
on the animal component for the smallholder farmer, utilization of food
crops for livestock production, forage crops in rice-based cropping
systems, case studies on crop-animal farming systems and research and
extension linkages. In the last two workshops, the crop-animal farming
systems research methodology for rice-based farming was discussed.
Women in rice farming
A major thrust of the collaboration is to include women's concerns in

the rice farming systems research methodology, from site description to
technology development and dissemination. In particular, women's
concerns were included inthe crop-animal farming systems key sites in

Ban Phai (Thailand), Santa Barbara (Philippines), and Batumarta
(Indonesia). In Ban Phai, labor and income contributions of men, women,
and children were analyzed. Three consecutive years of drought resulted
in increased emigration of men, leaving women with the increased burden
of household and farm activities. The results also indicated that sericulture,
a task normally managed by women, had good potential for improving
yields and increasing household income, particularly during the dry season.
Technology for increasing and sustaining the yields of mulberry bushes has
been combined with improved silkworms to raise the income-earning
capacity of organized women in this village.
Women participate in rice and livestock production, glutinous rice

processing, and farm marketing and financial activities in Santa Barbara,
Pangasinan. To increase the productivity of female labor and household
income, several technologies have been introduced and adopted. These
include: (1) IR65, a high-yielding glutinous rice variety; (2) mushroom
culture using crop residues as a substrate, e.g., rice and mungbean straw,
Leucaena leaves, banana leaves, and others; (3) a low-cost, foot-powered
tapak-tapak water pump; and (4) swine rations formulated from farm-grown
sweet potato and cowpea. The project also conducted complementary
analyses of intra- and inter-household time allocation, resource use, and
food consumption for farming systems located in rainfed and partially
irrigated environments. Results from the analyses are contained in a series
of reports and papers presented in conferences and workshops.
The results of the study conducted at the Batumarta site showed that
women are involved not only in household activities but also in crop and
animal production, and off-farm and non-farm activities. Besides being
dominantly engaged in household activities, women are also involved in
farm decision making, particularly in poultry, goat husbandry, and
postharvest processing of cassava, soybean, and mungbean. Men, women
and children contribute 58.5, 38.7 and 2.5%, respectively, of the total labor
in food crop production.
Technologies that can be introduced to reduce drudgery and improve

women's productive efficiency in crop and animal production included
implements such as seeders, weeders, corn shellers, groundnut pod
shelters, soybean pod shellers, pedal rice threshers, and cassava slicers.
Training programs in storage and pest control, seed technology,
composting, pest management, vaccination, poultry health management,
and forage production are needed to enhance women's technical
knowledge.
LESSONS LEARNED
Crop-animal farming systems research needs interdisciplinary and

intercommodity research teams whose composition will depend on the
type of production system in different rice ecosystems. However,
difficulties in organizing the teams arise because the needed specialists
belong to different institutions. A key factor facilitating the integration of
these teams is donor funding.
Collaboration among scientists in crop-animal farming systems

research often depends on the personalities of the scientists concerned.
Constant visits of the Network Coordinator are essential in promoting the
collaboration of different disciplines; during the visits, the different scientists
should be invited to participate and to discuss the progress of the
collaboration and the proposed activities. This is a very important
component of networking. Participation in international and national
workshops and meetings also promotes better collaboration among
scientists from the NARIs.
The best way to disseminate the methodologies and technologies

identified in the key sites and other sites in the network is through
meetings, workshops and field visits. The monitoring tours were effective
in the first three years of the program; but now, with more experienced
scientists in different countries, workshops and field visits are more
effective. The workshops can also be used to identify research issues that
animal and crop researchers should focus on to make their research
relevant. A meeting or workshop at least every two years is needed to
bring scientists together and create and maintain a team spirit.
The practice of centralizing data management in the international

coordination unit is not acceptable to strong national programs, and for
that reason ARFSN decentralized the network data management. Also, the
network has encouraged the NARIs to take the responsibility of writing up
the publications arising from the workshops and meetings.
Animal production is an integral part of many rice-based farming

systems. It is a major source of cash income, food for home consumption,
power for farm operations and manure for organic fertilizer, but overall
productivity is very low. Available recommended technologies from
research institutions have not been effective when tried at the farm level;
thus, there is a need to reorient station research, especially animal

component research, to develop more appropriate technology for small-
scale animal production.
Crop residues (rice straw, corn fodder, wheat straw) are the most
abundant animal feed, but their feed quality is very poor. There are several
production systems that can overcome this constraint, based on the
production of forage legumes and grasses. Promising forage crops
include berseem, Stylo, Desmanthus, Siratro, Crotolaria, Italian rye grass,
triticale. Feeding trials have indicated increased weight gains when these
legumes and grasses are used.
There are varietal differences in residue production of different crops

grown in rice farming systems. Network studies have indicated several
promising dual- purpose (food grain and animal feed) crops such as
cowpea, peanut, pigeonpea, sorghum, and mungbean. Significant
differences in residue yield have been observed. Plant breeders generally
look at only grain yield and should modify their breeding programs so as
to include in their objectives improvement of both plant biomass and grain
yield.
Forage legumes can also be used as green manure. In several studies,

utilization of the last cut of forage legumes as green manure has led to an
increase in the yield of the following rice crop. The most effective have
been Desmanthus, Crotolaria and Stylo.
Even though very intensive systems using forage grass and legumes
have been found promising in different countries, it is necessary to study
long-term effects of continuous cropping on soil fertility and crop
productivity in various rice ecosystems and soil types.
Several forage crops have shown promise either as monocrops or in

combination with other crops. However, availability of seed for research
purposes has been a major problem. It will even be worse when farmers
start adopting the technology.
OUTLOOK
Ecosystem-based research programs are a new strategy for the year

2000 and beyond. AFRSN is already conducting cropping and farming
systems research in three ecosystems: irrigated lands, rainfed lowlands,
and uplands. Since the IRRI research farm is not a representative

environment, future research will be decentralized and will involve the more
advanced NARIs in the development of technologies in different
ecosystems and subecosystems. In order to accomplish this task, a new
approach to research collaboration was established in 1991 through a
multinational ecosystem-based consortium for the rainfed lowland and
upland rice systems. This will enable IRRI and the stronger national
programs to work together in basic, strategic and applied research.
The research output of the research program at IRRI and consortium

sites will be evaluated and disseminated within each country and across
countries through the IRRI-coordinated networks. The activities in each
consortium site can be linked with network trials in different subecosystems
resulting in a type of research program that will be continuous from basic-
strategic to applied and finally adaptive research.
Sustainability of rice farming will be the major concern in rice and rice-
based farming systems research both by IRRI and the consortium.
Consequently, the IRRI-coordinated networks will work together to promote
the development and evaluation of technologies contributing to the
sustainability of rice farming. One of the networks will focus on identifying
high-yielding rice varieties that will reduce the use of chemical inputs and
increase tolerance to stress; another network will develop nutrient
management systems for sustained productivity using reduced levels of
inorganic fertilizer and increased levels of organic fertilizer for rice and rice-
based farming; a third network will integrate pest management with
minimum chemical inputs and more biological control and other crop
management techniques in rice-based farming systems. ARFSN will
develop more productive and sustainable farming systems by bringing
together technologies identified by the IRRI research programs, consortia,
NARIs, and the three other IRRI networks. The final output should be
small-scale production systems that are technically feasible, biologically
and economically viable, socially and culturally acceptable, and
sustainable.
Crop-animal collaboration between ARFSN and the NARIs will continue

in a few farming systems sites representative of different rice
ecosystems/subecosystems and in animal production systems in three
irrigated sites, one deepwater site, two acid upland sites, and three rainfed
lowland sites (one favorable, one drought-prone and one drought- and
submergence-prone).
RISPAL:
THE LATIN AMERICAN RESEARCH NETWORK
FOR ANIMAL PRODUCTION SYSTEMS
Manuel E. Ruiz'
BACKGROUND
In Latin America, as in many other areas of the world, during the 1960s
and early 1970s there was a growing disenchantment with traditional,
discipline-based research as it became more and more evident that much
of the existing technology failed to solve the production problems that
farmers, particularly the smallholder, faced. The mismatch between
technology offer and demand was made even more evident by the so-
called integrated rural development projects such as the Puebla Project in
Mexico (Myren 1971) and the Caqueza Project in Colombia (Zandstra et
al. 1978). The technology generated by the discipline-oriented approach
was seen as being: (1) high-input demanding, (2) labor minimizing, (3) net
income maximizing, (4) unstable, as compared to the technology used by
the farmer, (5) divorced from the social and economic conditions surround-
ing the farmer, (6) inconsistent with the farmer's objectives, (7) hard to
use
in extension programs, and (8) of little use to policy-makers.
In addition,
no attention had been given to the needs and unique characteristics of the
small farmer by research and/or development programs.
In the early 1970s, cropping specialists in Latin America

began a new
approach in research whereby, in close cooperation with the small farmer,
real problems and constraints were identified (and thus
subject to
experimentation), as well as interactions among the farming system's
components and between these and the environment. This systems
approach caught the attention of animal scientists and in 1976 a research
project2 presented by CATIE3 designed to characterize cattle
production
1
Ph.D. in Animal Nutrition, Coordinator of the Latin American
Research Network for
Animal Production Systems (RISPAL), IICA, 2200 Coronado,
Costa Rica.
2
CATIE/IDRC Project "Dual-purpose Cattle Production Systems
Using Crop Residues".
systems in four regions of Costa Rica and to develop feeding systems

based on pastures, crops, and crop residues received the support of
IDRC4. For this, an attempt was made to adapt the research methodology
applied by cropping systems scientists (Zandstra et al. 1981); however, it
was soon realized that this methodology could not be followed in its
entirety (Bore) et al. 1985). The major difficulties arose because (1) for on-
farm experimentation, animal experimental units are very few, (2) animals
within a production system are highly variable due to differences in
physiological function, breed and/or undefined crossbreeding, sex,
production function, and management, and (3) animal response to
treatment is multiparametric (weight changes, milk production and
reproduction, for example).
Thus, methodological difficulties were encountered in the application of

the systems approach to animal production research, especially when
coupled with the novelty of working with the small farmer. However, the
project initiated in 1976 succeeded in drawing the interest of other
research groups; this led to the training in the systems approach (formally,
leading to a Masters degree, and informally through short training
programs) of professionals from various Latin American countries. As a
result, a similar project in Panama was established in 1978 and another
one in the Peruvian Amazon region in 1979.
In view of the methodological limitations and the scarcity of technical

people involved in each of these three projects, it was agreed that perhaps
more progress could be achieved if they pooled knowledge and discussed
topics of common interest. Thus they met in Panama in 1981. This
meeting was followed by others involving an increasing number of projects
and institutions and, as a result, the group began to take shape as a
network sharing a common research methodology and similar objectives
and problems. In 1986, RISPAL was formally created with the financial
support of IDRC and technical coordination responsibilities shared by
IICA5, CATIE and INIAA6. Briefly, the main results obtained from these
meetings were as follows:
3 See List of Acronyms and Abbreviations at the end of the book.
4
See List of Acronyms and Abbreviations at the and of the book.

First meeting: Gualaca, Panama, 198 1.
It was agreed that farming systems research (FSR) methodology

suffered gaps insofar as animal production research was concerned. The
concepts were rather vague and the projects were ill-prepared to fill the
methodological gaps quickly, due to lack of discipline, diversity, and
expertise. However, the exchange of information was useful and it was
deemed necessary to continue this activity, for which IDRC was willing to
lend its support.
Second meeting: Pucallpa, Peru, 1982.
A general methodological scheme for animal production research was

proposed. The statistical problem of on-farm experimentation was
examined, as well as the need to minimize the information obtained
through farm surveys, as it was evident that a backlog of unanalyzed
information was becoming the main result of the projects.
Third meeting: Turrialba, Costa Rica, 1983.
Different procedures for ex-ante analysis were compared, a consensus

was reached on a general methodological scheme for animal production
systems research, and the projects were urged to document experiences
on confrontation of technological alternatives with the farmers.
Fourth meeting: Chiclayo, Peru, 1984.
Various farmer consultation experiences were reviewed. It was

considered convenient to include social scientists in the research teams.
Statistical procedures for planning and testing of on-farm experiments and
comparisons were defined, including the evaluation of technological
alternatives. Guidelines for the economic analysis of farm data were
established.
Fifth meeting: Turrialba, Costa Rica, 1985.
Case studies of technology transfer and monitoring were analyzed,

including the Rio Frio (Costa Rica) and the Nueva Concepcion (Guatemala)
projects. Specific procedures in each phase of the general research
methodology were examined. A committee was formed to propose a
network that would strengthen and lend vitality to the participating projects.
Sixth meeting: Bambito, Panama, 1985.
On-farm experimentation was discussed. The role of modeling in

animal production systems research was broached for the first time. The
usefulness of farm data, not only to characterize the production system but
also to test cause-effect hypotheses, was discussed. The projects
approved both the by-laws of a formal network and the steps to be taken
to find the necessary funding and institutional support; thus RISPAL
became a reality in 1986, with IDRC funding and institutional backing by
IICA, CATIE and INIAA.
Seventh meeting: Lima, Peru, 1987.
The RISPAL projects were exposed to ILCA's7 application of the

systems approach to research. A cooperative agreement was reached
with CIAT's8 Tropical Pastures Network. The need to consider
macroeconomic analysis of livestock research problems was presented to
the network. The link between research projects and agricultural policy
and planning organizations was discussed. A new attempt was made to
define the procedures to follow for the evaluation of technological
alternatives at the farm level.
Eighth meeting: Guatemala City, Guatemala, 1988.
An integrated set of analytical tools was presented concerning the

organization and analysis of data, macroeconomic considerations in the
design of technological alternatives, modeling, ex-ante analysis, statistical
designs and analysis for on-farm trials, and farmer confrontation.
Ninth meeting: Zacatecas, Mexico, 1990.
Transfer of technology was discussed with a view to reaching a

broader, in-depth discussion in the next meeting. The use of secondary
information to define the scope of research projects was applied to the
Mexican case. A procedure for on-farm reproduction studies on dual-
purpose cattle was proposed, thus allowing the RISPAL projects dealing
with this system to make methodological adjustments. The development
and use of simulation models was presented and illustrated with models
a See List of Acronyms and Abbreviations at the end of the book.

for beef cattle, alpacas and sheep. The description of the results obtained
in the Network's nine general meetings, although brief, provides an
indication of the methodological progress achieved, the permeation of the
systems approach in animal production researchers, and the increasingly
interdisciplinary, farmer-linked work being carried out by the member
projects.
OBJECTIVES
RISPAL is a mechanism for regional cooperation in research and

transfer of technology as they pertain to animal production systems, as
such or in conjunction with other agricultural systems. Specifically, RISPAL
has the following objectives:
1. To implement cooperative FSR activities
2. To stimulate the development of a research methodology that uses a

systems approach
3. To validate and recommend technology
4. To strengthen local programs, projects, and institutions through

technical support and training
RISPAL PARTICIPANTS
The network is formed by institutions and scientists actively involved in

animal or mixed production systems research directed to small and
medium landholders. A wide range of institutions and projects covering
different ecosystems, animal species, and target systems are included
(Table 1 and Fig. 1).
Table 1. Participating projects in RISPAL
INSTITUTION PROJECT TITLE ECOLOGICAL AREAS
IDIAP (Panama) Dairy-Beef Feeding Systems Lowland humid tropics and savannahs
ICTA/DIGESEPE/USC/IICA (Guatemala) Dairy-Beef Production Systems Lowland humid tropics
(VITA (Peru) Amazonian Production Systems (Dairy-Beef) Tropical forest
CATIE (Costa Rica) Silvopastoral Systems Tropical forest/lowland humid tropics
CARDI (Guyana) Milk and Dairy-Beef Production Systems Lowland humid tropics
UCCH (Chile) Milk Production Systems Temperate
INIFAP (Mexico) Goat Production Systems Subtropical desert
IVITA (Peru) South American Camelids Tropical highlands
INIAA (Peru) Guinea Pig Production Tropical highlands
ICA (Colombia) Rural Systems (Mixed) Tropical highlands/tropical lowlands
INIAA (Peru) Andean Farming Systems (Mixed) Tropical highlands
FONAIAP (Venezuela) Dairy-Beef Production Systems Dry-humid tropics
UCV (Venezuela) Genetic Improvement of Dairy-Beef Cattle Dry-humid tropics
CATIE (Costa Rica) Agrosilvopastoral Systems Dry tropical hillsides
PARTICIPATING INSTITUTIONS: IICA, IDRC, CE&DAP,

WINROCK INTERNATIONAL
ISAPLAC
CE&DAP ucv
AROA CE&DAP CATIE-3
IDRC IDRC AROA CE&DAP CE&DAP
CIID CIID IDRC AROA AROA
RISPAL RISPAL CIID IDRC IDRC
WINROCK WINROCK RISPAL CIID CIID
ICA-1 ICA-1 WINROCK RISPAL RISPAL
INIAA-C INIAA-C ICA-1 WINROCK WINROCK
CATIE-2 PISA PISA INIAA-C ICA-1 ICA-1
GUAT CATIE-2 CATIE-2 PISA INIAA-C INIAA-C
IVITA-C IVITA-C GUAT GUAT CATIE-2 PISA PISA
UCCH UCCH IVITA-C IVITA-C GUAT CATIE-2 CATIE-2
MAG MAG UCCH UCCH IVITA-C GUAT GUAT
INIFAP INIFAP MAG MAG UCCH IVITA-C IVITA-C
INIAA-L INIAA-L INIAA-L INIFAP INIFAP MAG UCCH UCCH
CARDI CARDI CARDI CARDI INIAA-L INIAA-L INIFAP INIFAP INIFAP
IVITA-P IVITA-P IVITA-P IVITA-P IVITA-P CARDI CARDI CARDI CARDI CARDI
IDIAP IDIAP IDIAP IDIAP IDIAP IDIAP IVITA-P IVITA-P IVITA-P IVITA-P IVITA-P
CATIE-1 CATIE-1 CATIE-1 CATIE-1 CATIE-1 CATIE-1 CATIE-1 IDIAP IDIAP IDIAP IDIAP' IDIAP
78- 7 778-79 80 81 82 83-84 85 86-87 88 89 90 91
Fig. 1. RISPAL's rapid acceptance in the region.
N
N
Explanation of projects and acronyms in Table 1 and Fig. 1
AROA Aroa-Bajo Tocuyo Project "Sistemas de Producci6n de Bovinos de Doble

Prop6sito." Barquisimeto. Fondo Nacional de Investigaciones
Agropecuarias (FONAIAP), Venezuela.
CARDI CARDI/IDRC Project "Milk Production Systems," Caribbean Agricultural

Research and Development Institute (CARDI), Guyana.
CATIE-1 CATIE/IDRC Project "Sistemas de Producci6n de Leche y Carne para

Pequenos Productores Usando Residuos de Cosecha." Tropical Agricultural
Research and Training Center (CATIE), Costa Rica (ended in 1985).
CATIE-2 CATIE/IDRC Project "Sistemas Silvopastoriles." Costa Rica.
CATIE-3 CATIE/IDRC Project "Sistemas Agrosilvopastoriles Sostenidos para

Pequenos Productores del Tr6pico Seco de Centroam6rica." Guatemala,
El Salvador, Honduras, Nicaragua.
CE&DAP Centro de Estudios y Desarrollo Agrario del Peru (Peruvian Center for
Agrarian Studies and Development), Peru.
CENIP CENIP/CIID Project "Sistemas de Producci6n de Bovinos de Doble

Prop6sito." Centro Nacional de Investigaciones Pecuarios (CENIP),
Republica Dominicana (ended in 1989).
GUAT IICA/ICTA/DIGESEPE/USC/IDRC Project "Mejoramiento de Sistemas de

Producti6n de Bovinos de Doble Prop6sito." Inter-American Institute for
Cooperation on Agriculture (IICA), Instituto de Clencia y Tecnologia Agricola
(ICTA), Direcci6n General de Servicios Pecuarios (DIGESEPE), Universidad
de San Carlos (USC), Guatemala.
ICA ICA/IDRC Project "Generaci6n y Transferencia de Tecnologia en Sistemas

de Production." Instituto Colombiano Agropecuario (ICA), Colombia.
IDIAP IDIAP/IDRC Project "Sistemas de Producti6n de Bovinos de Doble

Prop6sito." Instituto de Investigaci6n Agropecuaria de Panama (IDIAP),
Panama.
IDRC International Development Research Centre, Regional Office in Uruguay
IICA Inter-American Institute for Cooperation on Agriculture, Costa Rica.
INIAA-C INIAA/IDRC Project "Sistemas de Produccion de Cuyes." Instituto Nacional

de Investigaci6n Agraria y Agroindustrial (INIAA). Peru.
F
INIAA-L INIAA/IDRC Project "Sistemas de Producci6n de Caprinos." Peru. (ended
in 1988).
INIFAP INIFAP/IDRC Project "Sistemas de Producci6n de Caprinos." Instituto de

Investigaci6n Forestal, Agricola y Pecuaria (INIFAP), Mexico.
ISAPLAC IICA/IDRC Project "Information System in Animal Production for Latin

America and the Caribbean," Costa Rica.
IVITA-C (VITA/IDRC Project "Sistemas de Producci6n de Cam6lidos

Sudamericanos." Instituto Veterinario de Investigaciones Tropicales y de
Altura (IVITA), Peru.
IVITA-P IVITA/IDRC Project "Sistemas de Producci6n Amaz6nicos." Peru.
MAG MAG/IDRC Project "Native Swine." Ministry of Agriculture, El Salvador

(ended in 1988).
PISA INIAA/IDRC Project "Sistemas de Producci6n Andinos." Canadian

International Development Agency (CIDA), IDRC, Peru.
RISPAL IICA/IDRC Project "Latin American Research Network for Animal Production
Systems." Costa Rica.
UCCH Universidad Cat6lica de Chile/IDRC Project "Sistemas de Producci6n de

Leche," Chile.
UCV Universidad Central de Venezuela/IDRC Project "Mejoramiento Gen6tico de

Bovinos de Doble Prop6sito." Venezuela.
WINROCK Winrock International, U.S.A.
METHODOLOGY
All projects in RISPAL share a common methodology based on farming

systems research principles (Fig. 2). An explanation of the network's
research methodology, which has been described by Ruiz (1989), Is not
the purpose of this paper. However, it must be pointed out that the
application of the methodological scheme presented in Fig. 2 is flexible, in
the sense that each project adapts it to suit its particular set of
agroecological characteristics, recommendation domain, target system,
and institutional function and structure.
Selecting the
target system
T
the
mod
Compiling Information Oftersturs.

form survey., extension)
Chr.ot.dz.tlon of
the system
Partial Problem.
I
, Rx..roh
prlorm" gfaoMn
on-(rm .nd in
rk"Ont
Designing aftematives
Eli!fTlnj
Partial
extato
proved
models) solutions
T
& -ant. an.ly.b and
wIth
farms. and extension
agents
On-farm evaluation
i
F Transfer
71
Traditional
system
+ * Improved
system
Fig. 2. Simplified diagram (without feedbacks) for agricultural

production systems research methodology.
Source: Ruiz (1989).
The methodological scheme does not imply a sequential progress

through all of the phases indicated in it. Although, in fact, a step-by-step
approach tended to characterize the pioneering animal production systems
projects, the network encourages leaving out any steps deemed
unnecessary (for example, if key information already exists due to recent
farm characterizations), in order to gain time in the research process.
Finally, it must be noted that the scheme includes field activities having
to do, on one hand, with the collection of farm data (through rapid rural
appraisals and surveys) and, on the other hand, with transfer of technology
methods and techniques, thus providing a template for easy and proper
linkage with development programs, such as has already occurred in
Guatemala and is being considered in Panama.
ORGANIZATION
To pursue its objectives and promote networking, RISPAL is organized

as follows:
Board. Made up of six technical representatives: three elected from

among the participating projects, two from IDRC and one from IICA. Its
main function is to provide general policy guidelines and approve plans.
Advisory group. Composed of leading scientists chosen to provide

technical support and advice to the network Coordinator.
Coordinator. Person responsible for planning and coordinating all

network activities, in close collaboration with the project leaders and
members of the board.
Plenary. Made up of all participating scientists. Voting rights are

bestowed on the project leaders and a representative of each organization
enjoying membership status.
ACTIVITIES
The following activities characterize the network:
General meetings. In recent years, these meetings have been held

every 18 months. Their main purpose is to allow for exchange of
information, discussion of methodology, analysis of experiences, and
planning of follow-up activities. The First General Meeting was held in
1981; the Tenth Meeting took place in October 1991 in Chile.
Workshops. These are designed for the advancement of methodology

in specific areas. The participants include not only RISPAL members but
outside experts as well, the proportion depending on the level of
knowledge already present in the network in the area to be covered.

Workshops have been held in the following areas:
Farm data analysis
Microeconomic analysis (2)
Computerized farm data base organization
Ruminant nutrition research methodology
Social scientists' role and methodology in systems research (2)
Construction and application of simulation models in animal production

systems research
On-farm animal reproduction research methodology
Consultancies. Short-term cross-project consultancies are offered. If

the expertise is not present in the network, outside consultants are
recruited. To ensure follow-up activities, the consultants agree to lend
continuous support after the mission is completed.
Training. RISPAL used to offer short courses. These have now been
discontinued due to limited multiplying effect and relatively low retention
and application of knowledge; however, RISPAL lends its support to
projects that organize short courses and seminars. More emphasis is now
being place on hands-on training, based on a planned program that
considers the projects' needs, the host institutions' capabilities and the
network coordinator's insight.
Publications. The proceedings of the first nine General Meetings have

been published. Also, methodological guidelines have been published on
the topics of Rapid Rural Appraisal and Ruminant Nutrition Research
Methodology. The proceedings of the two workshops on the participation
of social scientists in systems research have also been published. The
following publications are underway: Building and Use of Simulation
Models as a Tool for Animal Production Systems Research, A
Methodological Guideline for On-farm Animal Reproduction Research
Methodology and Glossary of Animal Production System Research
Terminology. Additionally, seventeen selected technical documents are
being translated into English to attend demands from the Caribbean region,
Africa and Asia.
Information system. An offshoot project, coordinated by RISPAL's

headquarters, seeks to enhance communication and exchange of
information. This is the Information System in Animal Production for Latin
America and the Caribbean (ISAPLAC). Data bases are being created in
the various target systems of interest to the network's members. Two
workshops on the operational aspects of this project have been held. Two
bibliographic bulletins have been published, and a vocabulary on animal
production is being prepared with the participation of RISPAL's members.
Promotion of activities. In addition to the activities described above

(which help to promote the network's, philosophy and methodology),
network scientists are encouraged to participate in scientific and technical
meetings. Also, the coordinator of RISPAL actively seeks to establish
working relationships with other networks and organizations sharing
common technical interests. For example, agreements have been reached
with CIAT's Tropical Pasture Evaluation Network, the Southern Cone
Collaborative Program, and the Small Ruminant Network, which operates
in the Andean countries.
Coordination and information exchange. Firm collaborative

arrangements have been made with other networks and cooperative
organizations participating in methodology research workshops, data
sharing and training. ISAPLAC has been successfully developed and
implemented, one of its achievements being the creation of a data base
covering 17 topics related to animal agricultural systems.
METHODOLOGICAL HIGHLIGHTS
Training and consultancies. As animal production systems research is

not part of the training programs given by the region's universities,
RISPAL's role in this area has been very much appreciated. This has been
accomplished through short courses, workshops, consultancies, and
exchange of scientists. Training (over 300 person-days) has been
accomplished in various areas such as farm diagnosis, design and
evaluation of technological alternatives, animal nutrition, pastures,
economics, animal reproduction, project planning and evaluation, and
systems research methodology, while consultancy activities have been
varied and numerous (over 600 person-days) in the same areas.
Institutionalization. With respect to institutionalization of the systems

approach, some institutions in the region have adopted the systems
approach in their research programs. This has been the result of various
factors including: promising results obtained in many of the network's
projects, institutional restructuring, the participation of RISPAL members in
support of the institutions' technical activities, and the promotion of
systems practitioners to decision-making levels. Nevertheless,
institutionalization remains a weak area.
Methodology development. This area constitutes the main technical

thrust. Practically all of RISPAL's publications contain methodological
procedures and analytical tools developed for animal systems researchers.
The methodological research tools and procedures that the network has
been incorporating into its knowledge base can be appreciated in Fig. 3.
This diagram is not intended to offer a complete and detailed account of
the actual set of techniques and methods; it only shows the ones that have
marked a turning point in the course of RISPAL's relatively short life. The
open-ended bars in Fig. 3 indicate that once a methodological tool is taken
up by the network's projects, it is tested, improved and used with
increasing intensity until, hopefully, all or most of the projects use it.
MCM.
aANSFER
MM"
MODELS
MM DATA 0%4V 1TION
MAa40ECONOMIC ANALYSIS
USE OF MODELS
RAPID RURAL APPRAISAL
FSR ECONOMIC ANALYSIS
ER CONFRONTATION TECHNIQUES
EXM/TEANALYSIS
FSR TA AND ANALYSIS
MMMAL FARM DATA
ATIC AND UYWM FARMING

GENERAL ANIMAL PRODUCTION SYSTEMS RESEARCH METODOLOGY
0
1981 82 83 84 85 86 87 88 89 90 91
Fig. 3. Methodological progress in RISPAL.

Methodology and technology dissemination. The RISPAL Coordination

Office has encouraged its projects to release their information on
appropriate technologies for small landholders in a systematic fashion, and
has been very active in the organization of workshops leading to the
preparation, editing and publication of methodological guides.
In summary, RISPAL has facilitated technical cooperation among

researchers and the strengthening of the national institutions which they
represent; to this end, a variety of activities have been conducted so far:
organization of 25 workshops on research methodology, system

modeling and research proposal preparation
training of nearly 400 researchers from 11 Latin American and

Caribbean countries
cross consultancies among the 16 research projects that make up

RISPAL
creation of a data base covering 17 topics related to sustainable animal

production systems
publication of five books on research methodology, two bibliographies

on animal production systems, 21 Newsletters, and nine General
Meeting Proceedings
initiation of translation into English of 17 technical documents on

technology and methodology developed in the network
CONSTRAINTS AND LIMITATIONS
Accepting the systems approach was not difficult for researchers, who
saw its principles as logical; in any case, it was necessary to break away
from the traditional scheme with its obvious limitations (Ruiz 1989).
However, the actual implementation of research with a systems approach
proved difficult because of technical and institutional factors.
Unfamiliarity. Contrary to research by disciplines, the systems research

methodology was not readily recognizable, and this led to a slow rate of
adoption. The very idea of having to work with researchers from other
fields made each individual feel a certain degree of distrust, a fear of losing
leadership within the institution.
Academic background. It has only been in recent years that

universities began to offer a major in agricultural systems analysis or
related fields. At present, nearly all researchers applying the systems
approach are specialists in some discipline and hence have some
difficulties in comprehending the problems from the perspective of other
disciplines.
For the same reasons, difficulties are sometimes encountered when

trying to identify and prioritize problems and when technological options
need to be designed. These difficulties are greater when natural science
and social science researchers are brought together. Thus, considerable
time and effort is spent in converting the researcher to a systems
approach.
Methodology and analytical capacity. The methodology originally

followed by cropping systems researchers was not adequate for animal
systems research. Even though now there is general agreement on the
basic methodological scheme (Fig. 1), the researcher must be prepared to
face limitations specific to animal production systems experimentation;
some of these are given below:
Scarcity of animals, particularly when conducting on-farm research with

small landholders.
High heterogeneity among animals due to differences in physiological

status, production function, genetic make-up, and management level.
Animal response is usually multiparametric. For example, when working

with milk cows, not only milk production is measured; it is also
important to measure changes in weight, reproductive performance,
and other parameters.
Animals do not remain in a given place; they constantly move or are

moved, sometimes even from one farm to another. Under these
conditions, experimental "noise" is exacerbated.
Large animal species, such as bovines and South American camelids,

have long productive and reproductive cycles which increase the risk
of a trial being terminated by unexpected circumstances, thus having
a bearing on the cost and duration of the research.
Due to the newness of systems research, few researchers have

sufficient experience and training to help them discern the type and
amount of information that needs to be collected. On too many occasions
this situation has led to the collection of an excessive amount of data
("better too much than too little"), which may accumulate for lack of
organization and analytical tools. Also, there have been cases where
experiments have been established on farms, treating these as if they were
experiment stations, with no farmer participation and little regard for farm
conditions. When there is no clear knowledge of what information is to be
collected and what analysis should be done, research with a systems
approach may become too time-consuming, giving some grounds to the
charge that this type of research is "just to conduct endless farm surveys."
Overexpectations. Sometimes excessive enthusiasm or overprojections

have aroused unrealistic expectations as to cost-and time-effectiveness of
farming systems research. It is not uncommon to witness a prediction
made by a project leader about what will happen when all farms in the
country or in a region adopt the technological options found most
successful.
Institutionalization. Most research institutions in the region still conduct

discipline-oriented or commodity-oriented research with no ties to farming
systems research. Adoption of'a holistic systems approach usually means
deep changes in the organization and function of the institution, which may
not be desirable. Also, the formation of multidisciplinary teams may be
resisted from within. Fear of the unknown has prevented the adoption of
systems research in some institutions.
According to Sands et al. (1989), many times institutions assign the

least experienced researchers to farming systems projects, while the
experienced ones either resist being involved or prefer to assume
administrative roles. This situation has been observed in some RISPAL
projects and, in most of these cases, it was necessary to make
suggestions as to how to assemble the research teams.
A serious phenomenon that needs to be addressed is the

ineffectiveness of many systems projects in influencing, and becoming
integrated with, their own organizations. This detachment is often due to
a fear of having the organization, or a group within it, take control over the
resources provided by the donor agency; other reasons may be found in
the high turnover of personnel and the pervading politicking which have
instilled insecurity and lack of long-term commitments. Directly related
to this phenomenon is jealousy among nonparticipating colleagues who
see a systems project as a "rich" one both financially as well as technically

(the latter due to participation in RISPAL's activities).
Human resources. Farming systems research requires professionals

with a holistic attitude, sensitive to the environment's limitations and to the
rural family's basic objectives. Few have reached this sensitivity through
self-training. The training of new researchers and technology transferrers
should not ignore proper academic preparation in disciplines, but the main
thrust should be specialization in systems (producing not nutritionists or
plant physiologists, but rather soil-plant-animal specialists, production
systems specialists, agricultural policy systems specialists, or farming
systems-oriented social scientists).
PERSPECTIVES
Animal production in Latin America and the Caribbean must face the
new challenges concerning the fragility of natural resources and the
deterioration of environmental quality. The agricultural sector is directly
affected by these challenges and, therefore, agricultural research,
development, and education will have to undergo profound changes,
making these activities more ecologically-oriented.
With respect to the systems approach, if it has already been confusing

to some, in the future it will become even more complex, as ecological
variables will be added to the present social, economic, physical and
biological variables. Several events must occur if livestock research with
a systems approach is to contribute effectively to the new technological
order that is approaching.
Institutional policy aspects. There is no doubt that Latin America and

Caribbean countries will continue to promote food production with
urgency. To this policy must be added mechanisms that will put a stop to
the destruction of the natural resources, but which, at the same time, will
not limit the people's right to achieve a minimum level of well-being. The
implementation of this double-edged policy (productivity plus
environmental protection) will require a iong-term commitment to maintain
and strengthen national agricultural research systems. If this commitment
is not forthcoming, an alternative would be for the governments to facilitate
the private sector to assume the responsibility of conducting agricultural
research and technology transfer.
Institutions (private and public) need to redefine their objectives in

accordance with the new national policies, to firmly adopt the holistic
systems approach as the basis for their research and technology transfer
programs, and to form interdisciplinary teams with long-term commitments.
International cooperation agencies and donors could change their

strategies so as to facilitate and strengthen the consolidation process that
the national institutions may undergo. For example, instead of providing
direct support to specific projects, they should do so by way of the
institution the project belongs to, while ensuring the institution's
commitment to support the project.
Research methodology. The systems approach is uniquely suited to

help in the conceptualization and solution-seeking required now that
ecological concerns need to be added to everyone's agenda. As new
technological policies which satisfy environmental and production demands
are drafted, the systems approach may become the instrument through
which the new policies can be put into effect. However, certain changes
may need to be inserted in the present systems-oriented research. These
are as follows:
The need to create a data base on land use capacity (or production
potential), determined not just on the basis of ecologically justified
arguments but also on evidence or opportunities for rational use of the
land for production purposes as well as the corresponding
technological requirement.
Information on land-use capacity (both from the social and biological

aspects) should be of primary importance to researchers, as should be
a set of clearly defined research guidelines. The latter, in turn, would
have to be based on macroeconomic studies, market trends,
competitiveness, and sustainability.
Instead of taking the farming system as the focal point of research (as
it is at the present time), the researcher would do well to focus directly
on an agroecological zone or ecosystem. This change in focus would
not ignore the farm, since it is the next level in the hierarchical order of
systems (Hart 1979).
Much has been written and said about agricultural sustainability, but
little has been done to translate this concept into concrete research
actions. For this reason, one of the first commitments in the future
research agenda should be to define which indicators would aid in

determining what is sustainable and what is not.
In view of the international economic contraction, funding of research

programs and projects will become more and more difficult. This is in
conflict with the manifest interest of donor agencies in environmental
and sustainability issues. In any case, horizontal cooperation and
communication through networking is justified more than ever.
Networking has been proven effective in complementing technical
needs and expertise and in promoting exchange of information, but
most of all, in facilitating the effective participation of rural men and
women in the overall process of generating and transferring technology
suitable to their social, economic and ecological environments.
Insofar as RISPAL is concerned, there should be a move towards a

"think tank" mode, taking advantage of its expertise and wealth of
information. The network should stress its analysis and design of
technology, particularly under the new ecological and social constraints
and opportunities.
Microanalyses that will take into consideration international markets,

agricultural policies, and trends can be helpful in identifying research and
development endeavors. Finally, modern communications technology
could be adopted and adapted to strengthen the network's activities.
LITERATURE CITED
BOREL, R.; RUIZ, M.; PEZO, D.; RUIZ, A. 1985. A methodological

approach for the development and evaluation of alternatives for animal
production systems for the small farmer. In Workshop on Tropical
Animal Production Systems (II, Pucallpa, Peru, 1992). Report. H. Li
Pun, H. Zandstra (Eds.). Ottawa, Can., IDRC Manuscript Report IDRC-
MR 62e. p. 36-77.
HART, R. D. 1979. Agroecosistemas: Conceptos b6sicos. Turrialba, C.R.,

MYREN, D. 1971. The Puebla Project: A development strategy for low-

income farmers. Paper presented at a Seminar on Small Farmer's
Development Strategies. Columbus, Ohio, Ohio State University,
September 13-15, 1971.
RUIZ, M. E. 1989. El enfoquede sistemas en la investigacion pecuaria y

su metodologia en America Latina. In Ciencias sociales y enfoque de
sistemas agropecuarios. E. Nolte, M. E. Ruiz (Eds.). Lima, Peru.
RISPAL. p. 9-35.
SANDS, D. M.; EWELL, P.; BIGGS, S.; McALLISTER, J. 1989. Issues

in
institutionalizing on-farm client-oriented research: A review of
experiences from nine national research systems. Quarterly Journal of
International Agriculture 28 (3/4): 279-300.
ZANDSTRA, H.; SWANBERG, K.; ZULBERTI, C.; NESTEL, B. 1978.

Caqueza: Experiencias en desarrollo rural. Ottawa, Can., International
Development Research Centre, IDRC-107s. 386 p.
1981.
ZANDSTRA, H. G.; PRICE, E. C.; LITSINGER, J. A.; MORRIS, R. A.
A methodology for on-farm cropping systems research. Los Banos,
Philippines, The International Rice Research Institute. 147 p.
DONOR EXPERIENCES AND
PERSPECTIVES
LIVESTOCK PRODUCTION SYSTEMS
RESEARCH: A DONOR'S PERSPECTIVE
Joyce M. Turk'
INTRODUCTION
The earlier, more traditional livestock projects developed by the Agency

for International Development (USAID) were designed to export and
impose US improved technologies with little knowledge of indigenous
systems. Projects were designed from the top down with little or no input
from sociologists prior to activities.
A "sunset review" of the early animal agriculture projects reveals that

cattle have pervaded the US livestock industry for nearly a century. Taking
into account that the technical specialists designing, implementing, and
advising animal agriculture projects were drawn from this sector, no
wonder the bias was so widespread! Often livestock projects had minor
impact on improving and sustaining the livestock sector because AID's
efforts were based on hastily-conceived designs that were discrepant with
traditional livestock practices. In addition, host country governments
lacked the necessary management expertise and financial capability to
implement and absorb costs recurring from livestock project activities.
LOOKING BACK: THE PAST TWENTY YEARS
In the past twenty years, AID funded at least 18 animal agriculture

projects, as listed below:
Ms. J. Turk isa Livestock Specialist in the Bureau for Research and Development, Office
of Agriculture (AID), Washington, D.C. 20523, U.S.A. The views expressed herein are the
author's and not necessarily those of the AID.
Small Ruminant Collaborative Research Program (SR-CRSP): Kenya,

Morocco, Peru, Indonesia, and Brazil
Mali Livestock Sector Project
Western Sudan Agriculture Research Project
Niger Range and Livestock Project
Togo Animal Traction Development Project
Sheep Production and Marketing: Bolivia
Belize Livestock and Development Project
North Cameroon Livestock and Agricultural Development Project
Dairy Production Assistance: Tanzania
Small Farmer Swine Repopulation: Dominican Republic
Central African Livestock Production and Marketing Project
Senegal Range and Livestock Development Project
Somalia Central Rangelands Project
Haiti Goat Development
Antigua Sheep Development
Tobago Sheep Station
Kibboko Range Research: Kenya
Management of Agricultural Research and Technology: Pakistan
Notably, the majority of them focused on increasing cattle production:

swine and small ruminants accounted for less than a third of the total
number. Several of these projects did not meet the goals and objectives
set for them, due to inflexible project designs, rigid assumptions, and poor
Donor Experiences and Perspectives 271
management. Many of the 18 projects were terminated in the mid-1 980s;

however, three will continue until the mid-1990s, namely, the Pakistan
project, the Mali Livestock Sector Project, and the SR-CRSP.
LOOKING AROUND: MALI LIVESTOCK SECTOR

PROJECT (1982-PRESENT)
Mali society is agropastoral, and livestock are an important component

of all agricultural production systems. Notwithstanding the devastating
impact of the Sahelian droughts in 1968-1973 and the early 1980s, the
contribution of the livestock sector to Mali's GNP continues to be
substantial. In fact, livestock is probably Mali's primary export, depending
on the level of cotton exports. Importantly, ox traction has been a major
factor in the significant increases in cash crop production, most notably
groundnuts and cotton.
During the period 1963-1982, AID funded a variety of activities in animal

health and livestock production in Mali. In 1963, it loaned US$2 million for
the construction of a small veterinary laboratory and for the training of
Malian specialists to produce vaccines. AID considers the training of
Malians to diagnose diseases and conduct research on trypanosomiasis
important enough to continue its funding through the Mali Livestock Sector
Project. A recent estimate revealed that, over a 20-year period, a 1 %
reduction in overall mortality among cattle in the project area would bring
US$6.5 million in benefits to cattle owners, in net present value terms.
In previous projects, AID funded on-farm and commercial cattle feeding

in an effort to ultimately reduce the wide seasonal swings in livestock
market supply and prices. However, severe problems in operating a
commercial feedlot under Malian management prohibited continued
funding. AID shifted its focus to increasing production and marketing
through improved range management and expansion into new lands. For
largely administrative and formidable logistic reasons, little progress was
made in the remote project site. Hence, the project was terminated.
Nevertheless, recent reassessment of the stall-fattening program revealed
a dramatic, widespread adoption supported on a sustainable basis by
nongovernment organizations, extension organizations, local banks, and,
most significantly, independently by farmers with no outside support!
In a further attempt to support livestock production, AID funded a

technical and economic feasibility study to determine the efficacy of a
chemical tsetse control program in lightly infected or isolated areas of
Zone One. However, preliminary results appeared to be against such
efforts, so future vector-borne disease control research was planned to
focus increasingly on tick-borne diseases.
Finally, a tsetse-trypanosomiasis training project in Mali provided the

basis for a developing national research capability in trypanosomiasis and
other vector-borne diseases of cattle. However, little has been
documented quantifying the impact of this training as opposed to other
activities, and future AID support for animal health in Mali does not focus
on this area.
Having determined that export marketing of Mali's large and small

ruminants contributes significantly to the GNP, AID's continued interest in
and support for that country's livestock sector is based on the following
facts: that animal agriculture is vital to economic development in
developing countries; that long-term donor support critically impacts on a
program's sustainability; and that a systems approach is required to create
viable packages for specific production needs. Such an approach
incorporates policies, technologies and institutional capabilities.
Future plans for this project call for improving livestock nutrition by
integrating crop/livestock production systems, by supporting the Malian
government's efforts to privatize veterinary services, and by strengthening
private livestock organizations, initiating policy dialogues, and establishing
a market information system.
LOOKING FORWARD: THE OTHER SPECIES
For many developing countries, small ruminants fill small niches. As

viable links in the agricultural production chain, these ubiquitous creatures
generate income and employment for women, children, and the elderly.
Outnumbering cattle in many regions, they also fit a niche inhabited by the
urban poor. Their role in sustaining agricultural production systems in
most ecozones clearly shows that small ruminants are vital to economic
development. As sources of meat or quick cash, these drought-resistant
animals cannot be ignored any longer.
Is it no wonder, then, that distant but audible consensus is being heard

among donors? Future projects which integrate small ruminants have
better than average chances for success. AID's Small Ruminant CRSP
(1978 to present) is a classic example.
To improve production of meat, milk, and fiber from sheep, goats and
camelids, AID initiated the SR-CRSP in 1978. A Title XII grant of
US$41,540,000 projected over a 13-year period has been matched by
participating US institutions providing US$15,559,000 (a matching
equivalent estimated at 37% of the funds provided by AID). Host country
contributions for the same period are estimated to be US$23,857,000.
SR-CRSP was developed as a joint effort between AID, 12 US

universities, Winrock International Institute for Agricultural Development,
and governmental agencies in five host countries. Through the years,
changes in research activities have led to a reduction of programs and of
the number of universities and overseas sites. Today, ten US institutions
(nine universities and Winrock International) continue to conduct
multidisciplinary research in four countries.
Initially, the program developed research and training agendas and

selected overseas worksites and collaborators in Brazil, Indonesia, Kenya,
Morocco- and Peru. With small ruminants widely dispersed among the
ecozones of the world, these sites were selected to maintain the global
mandate of the CRSP. Brazil reverted to linkage status in 1985, and the
program left Peru in 1989 due to political violence. SR-CRSP activities will
begin in Bolivia in late 1991 to replace Peru, while Morocco will be phased
down in 1992.
US principal investigators collaborate with host country scientists to

conduct research in six priority areas:
Dual-purpose goat production and management in farming systems
Genetics of prolific sheep
Hair sheep production systems, including grazing sheep under rubber

or other tree canopies
Sustainable agropastoral systems for small ruminant production,

particularly on marginal and fragile lands
Animal health management systems through biotechnology
The Andean Small Ruminant Science Network and other ongoing

networks in Africa and Asia for information exchange
Because of the SR-CRSP's proven track record in human resource and

institutional development, AID has extended project funding until 1995.
The fact that both Kenya and Indonesia are adapting parts of the research
and integrating the project into national research plans points to the quality
and success of SR-CRSP research. Although the national agricultural
research institutes of these countries are struggling to adopt less than
traditional modes of production systems research, such integration
highlights how animal agriculture programs can succeed when they are
well planned, developed, and implemented.
USAID's PERSPECTIVE: NEW INITIATIVES
Within USAID, current shifts in program planning have led regional

bureaus to express less interest in agricultural research and more interest
in agribusiness. Struggling to balance high priority goals with more
traditional objectives, USAID adopted several new initiatives; its current
portfolio is now encompassed by these four new initiatives: family,
democracy, environment and business/development partnership.
Traditionally, many development programs were designed to benefit

individuals or national groups. Now a focus on the family, including the
dynamics within the family, is considered critical to the success or failure
of USAID's development objectives. Given that food production is a family
enterprise, projects will focus principally on the family in order to
understand household decisions which affect development.
Complementary to and supportive of the transition to market-oriented
economies and sustained, broadly-based economic development, the
democracy initiative includes long- and short-term training and scholarship
programs as well as long- and short-term technical assistance. While these
activities are not new to USAID's programs, the focus on democracy in
USAID-assisted countries will stimulate their economic development.
As environmental problems are closely linked to developmental

concerns, the purpose of the environment initiative focuses USAID's
environmental and natural resource interventions on cross-cutting issues.
For example, unsound agricultural practices lead to loss of biodiversity,

loss of tropical forests, and to urban and industrial pollution.
Using three distinct approaches, USAID proposes to develop sound

economic and environmental policies, to strengthen institutions in the host
countries, and to involve the private sector in creative solutions.
International trade, foreign investment, and economic development all

intertwine in the global economy of the 1990s. In partnership with
universities, private voluntary organizations, donor nations, and
international organizations, USAID will develop stronger partnerships with
the private sector to support free markets and broad-based economic
growth.
HOW SMALL RUMINANTS FIT THE NEW INITIATIVES
Sheep production in Indonesia, goat production in Kenya, and camelid

production in Peru are all family enterprises. Whether in densely populated
regions like Indonesia or western Kenya or within the complicated
framework of peasant communities in the Peruvian highlands, small
ruminants sustain familial economies by providing opportunities for labor,
income storage and the sale of milk, manure, or offspring.
Gender roles in small ruminant production and marketing emphasize

family dynamics. Due to the size and docility of the species, women and
children become primary caretakers. and the primary beneficiaries of the
supplemental income brought in by small ruminants. Who cares for the
animals? who makes marketing decisions? who uses by-products? and
who determines on what to spend income generated by sale of animals
and by-products? are important decisions that impact on the family
enterprise.
Transition to market-oriented economies requires sustainable production

of commodities which are high in demand, short in supply. Small
ruminants provide a diversity of products (meat, milk, fiber, hides, fertilizer)
and are integral to sustainable agricultural production systems.
In most countries, meat supply cannot meet demand, and as family

income rises, the demand for meat also increases. In his study on alpaca
meat production and exchange in southern Peru, Primov (1983) documents
how alpacas contribute to sustaining human nutrition. Households located

in the high puna2 rely almost exclusively on animal production, mostly
alpacas and llamas. In the more isolated regions, the barter of alpaca
meat for crops persists as a traditional practice of exchange between the
livestock-producing highlands and crop-producing temperate zones. Small
ruminant production increases the independence of the family unit.
SR-CRSP research assists local producers to meet the growing

consumer demand for meat by improving methods of production through
long- and short-term technical assistance and the training of scientists and
extension workers. Due to their body size, small ruminants can be raised
economically in small numbers (i.e., there isn't much economy of scale).
Risk can be spread over a larger number of animals. Losing a few small
ruminants is less catastrophic than losing a total herd of one cow.
Policy changes in market-oriented countries, coupled with the growing

consumer demand for meat, create opportunities for entrepreneurs in
production and marketing of livestock and by-products. Marketing small
ruminants creates opportunities for both domestic and international
business partnerships.
SR-CRSP research has benefited small ruminant producers in both

developed and developing countries. Some examples include a vaccine
developed for contagious caprine pleuropneumonia; breeding and genetic
research on prolificacy in sheep; systems analysis and modeling which
provide simulations for economic analysis; and research on alpacas.
Private and public sectors benefit mutually through research and
collaboration.
Small ruminants constitute a sustainable system component, and they

unique niches in Africa, Asia, and Latin America, as well as in developed
fit
countries. A sustainable systems approach to small ruminant production
highlights the role of small ruminants in natural resource conservation. Not
only are these animals environmentally friendly (they decrease the need for
and use of inorganic fertilizers and herbicides), but they also convert crop
by-products and residues into higher value products such as meat and
milk.
2 Puna: desolate Andean plateau.

CONCLUSION
Because of the close relationship between household production and

consumption, family enterprises that integrate crops and livestock spread
risk more diversely than do monocrop farmers. Similarly, as a donor, AID
spreads its risk by funding long-term multidisciplinary projects rather than
unidisciplinary ones. Reviewing and comparing traditional animal
agriculture projects with more recent ones demonstrates how a systems
approach to project design and implementation leads to greater probability
of positive impacts.
Looking forward, it appears that AID and other donors must learn a
critical lesson: animal agriculture cannot be dissected from a farming
system without amputating a vital link to economic development at both
the household and national levels.
LITERATURE CITED
PRIMOV, G. 1983. Alpaca meat production and exchange in Southern

Peru. Lima, Peru. University of California, Davis/Instituto Nacional de
Investigacion Agraria y Agroindustrial, SR-CRSP Reporte Tecnico No.
31. 83 p.
IDRC EXPERIENCES IN THE SUPPORT OF
IN DEVELOPING COUNTRIES
Hugo Li Pun', Carlos Sere2 and C. Devendra3A
INTRODUCTION
The International Development Research Centre (IDRC) is an institution

created by the Canadian Parliament to promote and support applied
research in developing countries. Since its creation, IDRC has chosen to
support research for the benefit of small farmers, taking into consideration
their important contribution to national economies, their needs and
aspirations, and the fact that they have been neglected by many research
and development efforts. In pursuing this, IDRC has supported many
projects in which holistic and participatory approaches have been followed.
The reasons for this have been mentioned in the introductory remarks to
the present workshop (Li Pun 1991).
The first of these projects (begun in the mid-1970s) were in cropping

systems in Asia and were linked to the activities of the International Rice
Research Institute (IRRI). These activities were later extended to other
national programs in Asia, Latin America, and Africa. About the same time,
IDRC supported a project in integrated rural development in Caqueza,
Colombia. This also served to expand client-oriented activities in other
parts of the world.
1 Associate Director, Environment and Natural Resources Division, IDRC. Ottawa, Canada.
2 Consultant, International Development Research Center. Montevideo, Uruguay.
3 Senior Program Officer, IDRC. Singapore.
4 The authors wish to express their gratitude to Drs. O.B. Smith and B. Kiflewahid for the
data provided to prepare Tables 5 and 6, and Mr. G. Quijandria for the analysis and
preparation of tables and figures; also to Ing. Helen Raij for the review of the manuscript.
In 1976 IDRC supported the first animal production systems project in

Central America at the Tropical Agriculture Research and Training Center
(CATIE). This was followed by a large number of projects in other Latin
American countries, eventually leading to the formation of the Latin
American Research Network for Animal Production Systems (RISPAL). The
RISPAL network has operated informally with IDRC's support since 1981
and has been formally coordinated by the Inter-American Institute for
Cooperation on Agriculture (IICA) since 1986. Animal production systems
projects were also supported by IDRC in Africa starting in 1983; however,
the networks supported by IDRC in that region have been commodity-
oriented (pastures: PANESA; by-products: ARNAB)5. The focus of the
Asian Cropping Systems Network was later expanded to include crop-
livestock systems in 1984 and rice-fish systems in 1987.
Through the years, IDRC has been incorporating the lessons learned
from previous projects, and the scope and coverage of these projects have
been evolving. IDRC has established support to systems research as one
of its top priorities, even while recognizing that most national agricultural
research systems are organized along commodity lines. The role of
networks has been instrumental (1) in creating a critical mass of
researchers following the same objectives and methodologies, (2) as a
forum for the exchange of experiences, and (3) as a source of training and
mutual backup support. The purpose of this document is to present some
of IDRC's experiences in animal production systems research, highlighting
some of the results, the constraints, and the facilitating factors for this kind
of activities; it will also discuss the outlook for systems research from
IDRC's perspectives.
IDRC ACTIVITIES IN FARMING SYSTEMS RESEARCH
IDRC supports farming systems research and related activities, mostly

through its former Agricultural, Food, and Nutrition Sciences Division.
Other divisions such as Communications, Information Sciences, and
Fellowship and Awards have also financed some activities. The support
provided through the Crop Product'lon Systems Program, Animal
5 PANESA: Pasture Network for Eastern and Southern Africa, ARNAB: African Research
Network for Agricultural By-products.
Production Systems Program, Forestry, and Agricultural Economics, has

been mostly in specific research projects in cropping, animal production,
or agroforestry systems. Also, the following networks are being financed
by IDRC: RISPAL (The Latin American Network for Animal Production
Systems), ARFSN (Asian Rice Farming Systems Network), RIMISP
(International Network for Research Methodology in Production Systems),
SRUPNA (Small Ruminant Production Systems Network). In addition,
some other networks in which IDRC is involved are either following a
systems perspective or have on-farm activities. That is the case of the
CIAT-coordinated RIEPT (International Network for the Evaluation of
Tropical Pastures) and the ILCA-coordinated AFRNET (African Feed
Resources Network).
Through the years, support to systems-type research has been

increasing in the different regions where IDRC operates. This trend has
been more noticeable in the case of Latin America and Asia, where it is
becoming the main priority in cropping, animal, and mixed systems (Figs.
1 and 2). In Africa, the evolution towards systems has been slower;
however, emphasis on that type of research is increasing. The importance
IDRC places on systems research is further illustrated by the fact that
diagnosis, problem identification, and some kind of systems analysis is
emphasized in many of the so-called commodity projects. Furthermore,
lately IDRC has been promoting a whole-commodity system approach in
some projects in order to link together the environment, production,
processing, marketing, and utilization processes and to identify bottlenecks
and opportunities for improvement.
Regional differences
Although there are some common elements in the types of institutional

problems faced in different parts of the world, as well as in the
characteristics of some ecosystems, IDRC has recognized the need for
regionally differentiated programs to tackle the problems of the prevailing
agricultural systems in Asia, Africa, and Latin America. Thus, in Asia,
considering land constraints and the prevalence of mixed (crop-animal)
systems, a higher priority has been given to those target systems as well
as to the utilization of small ruminants in plantations in order to maximize
output of animal products through the utilization of undergrowth. In Africa,
given the importance of pastoral systems in the semiarid lands and the
need to intensify land use in the more favorable highland areas of Eastern
Africa, IDRC has chosen to concentrate efforts to address those systems.
In the dry/humid tropical areas of Latin America, considering that land

is not a serious constraint, efforts have concentrated on the improvement
of pasture-based dual-purpose cattle production systems. In the low
humid tropics, due to the fragility of the ecosystems, IDRC has been
supporting agroforestry and silvopastoral systems research. In the more
unfavorable areas, such as the highlands of the Andean region, the
emphasis has been on the improvement of mixed systems.
Fig. 1. IDRC's total investment in crop and animal production by

region.
60 -
ft AFRICA 0 ASIA *LATIN AMERICA
so _-
40
30
20
10
0
1971-1976 1977-1982 1983-1990
Fig. 2. Support to systems projects, according to regions 1971-1990

Strategies
Research approach. The main activity of IDRC is to support specific

research projects addressing the problems of target systems in specific
geographic areas. These target systems are usually selected on the basis
of an analysis of the system's different hierarchies in order to judge the
relative importance of the region and the specific target system within the
country context (Fig. 3). For financial and administrative convenience,
projects are supported in 3-year phases. However, generally they are
financed for periods of more than 9 to 12 years, usually not exceeding
CAD$100,000 per year.
COUNTRY
ECONOMIC SOCIAL-
REGIONS POLITICAL
SYSTEM
SYSTEMS
REGION
MARKETING, NON-
CREDIT AND FARMING
AGRICULTURAL
INFORMATION SYSTEMS SYSTEMS
NT RS
FARM SYSTEM
SOCIO-ECONOMIC
SUB-SYSTEM
CROP ANIMAL
AGROECO-SYSTEM AGROECO-SYSTEM
CROP AGROECOSYSTEM , ANIMAL AGROECOSYSTEM
WEEDS PLAGUES SOILS WEEDS

SOILS
CROPS DISEASES PASTURES ANIMALS
Fig. 3. A diagram illustrating systems hierarchy.
Adapted from Hart (1979).
Projects follow the general systems methodology, beginning with area

selection and system characterization (rapid rural appraisal and static and
dynamic surveys). Simultaneously, some exploratory trials are established.
These trials lead to the design and on-farm testing of alternatives (Fig. 4).
Training activities for researchers, extensionists, and farmers are
established in all projects. Projects are directly administered by the
grantees, as IDRC believes strongly in the need to create managerial

capacity within national institutions. Collaboration and backup by
international or regional institutions is encouraged. Most projects
participate in networks.
Agricultural systems research for the improvement of small farms is a

relatively new field. Very few institutions offer organized training in
systems. There are relatively few opportunities to present research results
in the more organized scientific media (journals, forums, conferences).
Systems researchers often have to work in isolation from the scientific
community. One of the strategies followed by IDRC to fulfill some of the
previously mentioned limitations is the organization of networks. These
networks organize training activities and provide both technical backup to
projects and a forum for the discussion and exchange of experiences; they
also serve to promote research approaches and methodologies and help
in their institutionalization.
AREA SELECTION
I
SYSTEM CHARACTERIZATION TARGET SYSTEM JJt7
DIAGNOSIS
i
I
DATA
BANK DESIGN
EVALUATION
I
TRANSFER OF TECHNOLOGY
I
Fig. 4. Systems methodological phases.

Choice of institutions. Priority is given to the support of national

institutions and universities. Support is also provided to regional and
international organizations, especially for networking activities or for
strategic research.
Technical manpower. Most projects are executed directly by local

researchers who are staff members of the recipient institutions. Only
occasionally, at the request of those institutions, IDRC assigns project
advisors to fill specific needs. Otherwise, external support is provided
through short-term consultants and monitoring by the Program Officers.
Monitoring and evaluation. Projects are monitored by Program Officers

who are occasionally complemented by consultants and network
coordinators. Formal evaluations are conducted in certain cases.
However, most of the time informal evaluations take place, especially at the
end of a three-year phase.
RESULTS
Bioeconomics
The results of projects supported by IDRC have been significant. They

include the characterization of production systems, the development of
methodologies, and the design of appropriate technologies and alternative
systems. Some of the results have been presented in this meeting as well
as in other ones. For the purpose of this presentation, the following are
mentioned:
Design of altemative systems in Botswana. Livestock production in

Botswana is an important activity. The country exports beef, but is a net
importer of milk. In an effort to promote milk self-sufficiency as well as the
agricultural sector in general, the government has assigned a high priority
to the development of the dairy industry in the periurban areas. As part of
this effort, the government is promoting dual-purpose (milk and beef)
systems by crossing the native Tswana cattle with the Simmenthal. IDRC
has been supporting the Department of Livestock Research in this type of
activities since 1984. A diagnostic study was conducted in 1985 in order
to identify farmers' constraints. The main limitations identified included
inadequate feeding during the dry season and low productivity of the
Tswana cattle. The project introduced the use of Lablab purpureus hay
and supplementation with sorghum bran to feed cattle during the dry
season. The feeding package was tested on farms with both native and
crossbred cattle. It was found that the milk yield of crossbred animals
more than doubled that of the native cattle (Fig 5.). An economic analysis
has not been done; however, farmers outside the project are showing
interest in these technologies.
Dairy-beefproduction systems in Guatemala. This project is working on

medium-scale farms in several areas of the country characterized by a high
rainfall during the wet season (over 1,200 mm) and an extended dry
season (6-8 months). Traditionally, farmers practice dual-purpose
production systems based mainly on Brown Swiss x Zebu crossbreds;
feeding is based on the use of jaragua grass (Hyparrhenia rufa) and African
star grass (Cynodon nlemfuensis). The project is being implemented by
a group of institutions, including the Instituto de Ciencia y Tecnologia
Agricola (ICTA), the Universidad de San Carlos, the Direccion General de
Servicios Pecuarios (DIGESEPE), and the Inter-American Institute for
Cooperation on Agriculture (IICA).
In addition to a complete characterization of the systems practiced, the

project has generated various alternatives, especially for cattle feeding
during the wet and dry seasons. Activities have included research on the
introduction of improved species of grasses and legumes, pasture
management, feed conservation for the dry season, cut-and-carry pastures
for the dry season, the introduction of legume trees for dry season feeding,
health diagnosis, and mineral and protein supplementation.
160
asX+TS
140
120
100
04
60 1
4 0
20
0
1 2 3 4 5 E 7 B 0 1 0
Fig. 5. Response of Simmenthal-Tswana (SX) and Tswana (TS)

milking cows to an improved feeding system based on lablab
hay and sorghum bran.
The results have been detailed in the projects' annual reports; for the
purpose of this document, only the results of a technology adoption survey
conducted by DIGESEPE are presented. The sample included 200
farmers. Eighty percent reported the use of cut-and-carry forages, mainly
sugar cane; 30%, the use of silage; and 16.5%, the use of leucaena. These
technologies have been promoted by the project (Table 1).
Table 1. Technology adoption in Guatemala
Technology Percentage of farmers

adopting it (n = 200)
Sugarcane 80
Corn/bean silage 30
Leucaena 16.5
Source: IICA (1991).
Amazonian production systems in Pucallpa, Peru. The Amazonian area

of Peru constitutes the majority of the country's territory (approximately
60%). The region is still underpopulated; however, there is tremendous
pressure for its utilization, since it constitutes the main agricultural frontier
for the impoverished populations of the highlands and the coast. Migrants
usually bring their animals and their own technologies. Land is being
opened close to the highways and rivers to ensure transportation of inputs
and products. Shifting cultivation is practiced. After the forest has been
cut and burned, annual crops (corn, rice) are established taking advantage
of the built-up soil fertility. After a few years, when fertility has been
depleted, pastures are sown and animals are introduced.
Livestock production is the stabilizing force in the Amazonian system.

For more than two decades the Instituto Veterinario de Investigaciones
Tropicales y de Altura (IVITA) has been conducting research on tropical
pastures and animal production at its station in Pucallpa. Given the
relatively limited impact during the earlier years, IVITA approached IDRC
for support in systems research. The cooperative project started in 1980.
A diagnostic study was conducted to characterize the prevailing systems
and identify their limitations. A strong component research program was
carried out with emphasis on the search for low-input pastures that fit the
system as well as on research in animal health, reproduction, management,
and economics. Seed production activities were established. Contacts

with the Development Corporation of Ucayali were also established, leading
to the implementation of a livestock development project.
IVITA's work has led to the development of some improved

technologies, such as the use of Brachiaria decumbens pastures (Table 2).
The project leader was supported for his Ph.D. degree in agricultural
economics. As part of his dissertation, he conducted a survey of 128
farmers in Pucallpa in order to develop econometric models of technology
adoption. It was found that there was wide adoption of improved pastures,
especially B. decumbens and Kudzu (Pueraria phaseoloides) as shown in
Table 3.
Table 2. Liveweight gain per animal in different pastures, Pucallpa (on-

station trials)
Pasture Stocking rate Liveweight gain

AU/ha g/day/animal
Native grass 1.8 227

2.6 160
Brachiaria decumbens 1.8 503

2.7 433
Source: (VITA (1989), and Santhirasegaram et al. (1975) cited by Riesco (1990).
Table 3. Pasture land use in Pucallpa, Peru
Pasture Proportion of Frequency of farms

total, % where present, %
Brachiaria alone 44 80
Kudzu alone 12 28
Natural grasses 22 100
Kudzu associations 20 51
Total 100 100
Source: Riesco (1990).

Andean farming systems in Puno, Peru. IDRC, with the financial support
of the Canadian International Development Agency (CIDA), is executing,
jointly with the Instituto Nacional de Investigacion Agraria y Agroindustrial
(INIAA), a research and development project to improve the well-being of
peasants in five communities in the department of Puno. The department
is in the Peruvian high plateau (Altiplano) bordering Bolivia and is the
poorest in the country. Most of the communities are located at altitudes
over 3,900 meters above sea level. The climatic pattern is erratic.
Droughts, floods, hail, and frosts are common occurrences, making
cropping rather risky except in the more protected areas surrounding Lake
Titicaca where a more benign microclimate prevails. Livestock production
is very important in most communities; cattle and minor species are raised
in the lower altitude areas and sheep and South American camelids at
higher altitudes. The project made a full characterization of the different
communities, leading to an excellent understanding of the tremendously
complex systems from the ecological, biological, economic, and social
points of view. The project also conducted considerable research on
components, especially in crops, pastures, and animal production.
Simulation models were also designed and validated for alpaca, sheep,
cattle and potatoes. Training and development activities were also
conducted. A rotational seed fund was established with approximately 200
ha of seed nurseries of the main Andean crops and pastures. .
One of the common practices in the communities surrounding the lake

is the fattening of cattle, based on the grazing of Totora (Scirpus totora)
and Llacchu (Helodea potamogetum), which grow in the lake. This is done
during the dry months (May-August). Farmers also harvest these forages
from the lake and offer them fresh or spread on the ground; this causes
losses of 20% to 40% (dry and wet season, respectively) due to trampling,
urination, and defecation. The project determined, through the use of a
simulation model and the climatic data of the last 50 years, that cattle heat
losses due to the cold weather could reach up to 25% of the metabolizable
energy consumed in the coldest months (May-August). Also observed was
the fact that the high water content of the aquatic forages diminished
consumption and therefore limited the cattle's performance in spite of their
relative high quality. This led to the proposal of a cattle fattening scheme
in which animals would be kept in rustic shelters and fed dried aquatic
forages. Drying the forages for 48 hours could reduce their water content
from 85-90% to 40%, while drying for four hours could result in a 70%
reduction which could be appropriate for small quantities (R. Quiroz 1991,
personal communication). Ex-ante analysis showed the practice to be
feasible; a validation trial showed the considerable advantages of this
fattening system (Table 4).
Table 4. Alternatives for cattle fattening based on Llacchu and Totora

forage, in Carata, Puno, Peru
Alternative Weight gain

g/head/day
Housing + fresh forage 470
Housing + dry forage 1050
No housing + fresh forage 323
Source: INIAA (1991).
Institutional strengthening
Very few organizations have actually institutionalized the systems

approach. However, considerable resources have been invested by the
participating institutions in sustaining this approach, and the degree of
commitment in many cases has been increasing. Another indirect indicator
has been the additional funding and participation from other donors and
institutions. Furthermore, IDRC has played an important role in supporting
researchers for graduate training (Table 5); this could be tied to the impact
that is being obtained through networking.
Table S. IDRC's involvement in animal production systems training
Level of training
Region Ph.D. M.Sc. Short courses'
Africa 5 7 8
Asia -- 2 60
Latin America 7 7 110
1 in person-months
Source: IDRC files.

Network development
One of the outstanding characteristics of IDRC since its inception has

been the promotion and support of research networks. IDRC participated
in the development of the research network concept and the creation of
many networks. The first one financed by DRC was the Cropping Systems
I
Network in Asia which was started in the mid-1970s, and which later
became the ARFSN. The oldest network specifically addressing animal
production systems research is RISPAL. IDRC also finances RIMISP in
Latin America for the development of farming systems methodology. The
African Feed Resources Network (AFRNET), based on two other networks
that were supported by IDRC (PANESA and ARNAB), is coordinated by
ILCA and supported by IDRC. The main emphasis of AFRNET is the
development of adequate feeding systems for ruminants; however, a
systems approach and a strong emphasis on on-farm testing is followed.
In Asia, IDRC recently agreed to support the Small Ruminant Production
Systems Network for Asia (SRUPNA). The number of countries and
institutions involved, as well as the participating researchers is included in
Table 6.
Table 6. IDRC-sponsored networks in animal production systems

research, 1991
Network Countries Institutions Researchers
Systems networks
RISPAL (Latin America) 12 20 153

RIMISP (Latin America) 10 15 n/a
SRUPNA (Asia) 12 20 85
Systems-related networks
RIEPT (Latin America) 18 51 104

ARNAB (Africa) 9 29 40
AFRNET (Africa) 10 23 25
Source: IDRC files.

ASSESSMENT OF IMPACT
Institutional
As mentioned, very few institutions have actually adopted systems

research within their structures. The long and continuous operation of
some systems projects, and the strong support provided by the oldest
networks (i.e., RISPAL and ARFSN) have resulted, however, in a strong
group of systems practitioners who are having an impact within their
respective institutions and countries. In some cases, those researchers are
playing a leading role within their institutional environment and have been
able to successfully link research, training, and extension activities with a
systems perspective.
International research systems
Stronger impact has been achieved in the case of international centers,

which either have been coordinating networks for a long time or have had
long-term support of specific research projects. In spite of some efforts to
promote systems experiences in international forums and in professional
associations, the impact at those levels has been limited. However, an
increasing number of commodity researchers are more conscious of the
need to evaluate commodities within a systems perspective, or at least to
validate their results at the farm level.
Human resource development
Impact on human resources has been obtained through the training of

staff associated with the projects. They have benefited not only from
hands-on experience but in many cases from specific formal and in-service
training organized by networks. As shown in Table 5, a large number of
researchers have been trained in different parts of the world although the
major impact has been achieved in Latin America and Asia, in great part
due to the stronger and continuous support provided, and the role played,
by networks.
CONSTRAINTS AND FACILITATING FACTORS
Institutional
Research organization and focus. It is a fact that most institutions are

still organized along commodity lines and that the disciplinary focus
prevails. This is a reflection of the type of training received by most
professionals in both the developed and developing world. The choosing
of the small farmer as a target population has resulted in a most difficult
challenge for researchers and research managers, who have been formed
along disciplinary lines, often with an orientation to different environmental
conditions. The complexity of some of the problems to be solved requires
not only a holistic understanding but a high analytical capacity. Often,
institutions assign the younger and less trained researchers to conduct this
type of research, while older, more specialized researchers are given
managerial positions, conduct highly specialized research, or are reluctant
to get involved in this kind of activity. However, in several of the projects
supported by IDRC, this limitation has been removed, as strong teams
have been assembled within the circumstances prevailing in the different
countries.
Long-term research, staff stability, and donor fatigue. Animal

production systems research, especially with animals of long reproductive
cycle, is of a long-term nature. On the other hand, national institutions
often become overly bureaucratized or politicized and researchers are not
well rewarded, all of which leads to a high turnover rate of staff and
directors, thus disturbing the research process. At the same time,
overexpectations about the impact of FSR have been created, and the lack
of greater and faster impact and the difficulties encountered are leading to
a situation of donor fatigue. Other issues are appearing in the agendas of
donor agencies, making support to systems projects increasingly more
difficult.
Research and development linkages
In most developing countries, the linkages between research and

development efforts are weak. In a very few cases, development projects
are formulated from the results of research activities, as the pressures for
change are high. On the other hand, experiences from development
projects are seldom utilized, and valuable feedback opportunities are lost.
The need for impact and the scarcity of available resources demand a
more efficient linkage between research and development.
Methodology
The extrapolation of cropping systems methodology to tackle animal

production or mixed systems research has not been possible in the case
of small farms. Difficulties associated with sample size, random sampling,
animal mobility, number of experimental units, statistical design, variability
of experimental units, and others have been encountered. Therefore, on
many occasions case studies have had to be used. Simulation has also
helped to screen possible alternatives, and creativity has helped in finding
new technologies. One has to recognize that the methodology is still
evolving; therefore, new projects should benefit from the experience gained
by older projects.
Human resources
There is a scarcity of trained technical staff in many institutions in the

developing world. Formal training in FSR is rather limited, which is a clear
constraint for the implementation of many systems projects.
OUTLOOK
Expansion of systems research
Given the emergence of new development concerns, systems

researchers will have to face new challenges. The implications of this need
to be discussed widely and practical experiences need to be developed.
Since it would be impractical to form very complete technical teams to
encompass all the necessary disciplines within a single institution,
complementary efforts may have to be sought among key institutions.
Rather than developing a large number of projects working on the same
issues, most probably resources will need to be concentrated on fewer, but
more solid research undertakings, involving highly experienced personnel
with high analytical capacity. The mistakes made in earlier systems
projects, by perceiving an oversimplified set of problems calling for
simplistic solutions, need to be avoided. Moreover, new issues (such as

impact on environment and sustainability, gender, and equity) need to be
analyzed in a systematic, serious way, and openly discussed with donors
and international organizations.
Given a situation of constrained resources, research will need to be

carefully prioritized. In this respect, the study of whole-commodity
systems, in order to understand the relation between environment,
production, processing, marketing, and utilization, as well as to identify
bottlenecks and impact opportunities, may help in the prioritization
process.
From research to development
The linkages between research and development must be strengthened.

A common understanding of the goals, procedures, and roles of the
different actors needs to be reached. There is a tendency in some
institutions to think' that because of the relatively limited impact of micro-
level research, this should be deleted in order to concentrate on macro-
issues. The temptation to pursue policy issues exclusively, as well as to
look for rapid, simplistic solutions, is great. The danger of that route is that
badly formulated policies could cause serious damage and the time
required for recovery may significantly delay the development process.
The value of micro-level research and its use to formulate policies need to
be recognized. At the same time, the impact of macro-policies on the rural
situation needs to be determined.
The role of the private sector
Given the constraints faced by most governmental institutions in the

developing world, the need to streamline and improve institutions is leading
to a reevaluation of the role of the private sector in all aspects (i.e.,
research, extension, promotion). In the case of Latin America, there is a
proliferation of nongovernmental organizations, which are competing with
governmental institutions and universities for research and development
funds. Their objectives and modes of operation vary considerably;
therefore, impact assessment must be based on a careful analysis of the
role of the different institutions, and their comparative advantages need to
be weighed so as to avoid a dilution effect in the use of resources. One
alternative that could be followed is the channeling of governmental efforts
to the less-affluent sectors of society, while the private sector could finance
and conduct its own research. This, of course, is not the only alternative;
what to recommend depends on the situation of the different countries and
the objectives of the different institutions.
The approach to international cooperation
Resources are becoming more scarce, and competition for them

tougher. Also, competition among donors sometimes hampers
development efforts, since human resources are scarce and agendas are
often conflicting. International cooperation should be based on mutual
understanding of the needs, aspirations, and possibilities of developing and
developed countries. Coordination among donors may be needed to avoid
duplications and misuse of resources.
The development of human resources
There is a continuous need to form more and better professionals and

practitioners in developing countries. It must be recognized that the
orientation provided in developed countries is not necessarily the most
suitable for the conditions in developing countries. A more careful analysis
of the needs of, and the type of training required by, developing countries
needs to be made. The development of analytical capacity should be
stressed in the formation of professionals.
IDRC's experiences while supporting systems research in developing

countries have been fruitful. They have permitted: (1) a better
understanding of the great variability of situations encountered in the
different countries and the difficulties often faced by local researchers, (2)
an understanding of the long-term nature of the process, and (3) the
recognition of the progress achieved through a "learning by doing"
process. Solutions are not always simple; however, it is in the hands of
the people directly involved to work toward meeting their own aspirations.
IDRC has learned to recognize its supporting role in the achievement of
those objectives.
LITERATURE CITED
HART, R.D. 1979. Agroecosistemas: Conceptos basicos. Turrialba, C.R.

IICA (INTER-AMERICAN INSTITUTE FOR COOPERATION ON

AGRICULTURE). 1987. Proyecto Mejoramiento de Sistemas de
Producci6n Bovina de Doble Prop6sito en Guatemala. Segundo
Informe Tecnico de Progreso. Guatemala, Gua.,
IICA/ICTA/DIGESEPE/USAC. 70 p.
IICA. 1991. Propuesta Fase III: Proyecto Mejoramiento de Sistemas de

Producci6n Bovina de Doble Proposito en Guatemala. Guatemala,
Gua., IICA/ICTA/DIGESEPE/USAC. 26 p.
INIAA (INSTITUTO NACIONAL DE INVESTIGACION AGRARIA Y

AGROINDUSTRIAL). 1991. Proyecto de Investigaci6n de Sistemas
Agropecuarios Andinos. Informe Anual 1990-1991. Puno, Peru,
Proyecto INIAA/CIID/ACDI. 347 p.
(VITA (INSTITUTO VETERINARIO DE INVESTIGACION TROPICAL Y DE

ALTURA). 1987. Sistemas de Producci6n Amaz6nicos. Proyecto IVITA-
CIID Fase 111, Informe Tecnico. Pucallpa, Peru, IVITA. 109 p.
IVITA. 1989. Sistemas de Produccion Amazonicos. Proyecto (VITA-CIID

Fase 111. Informe tecnico del proyecto. Pucallpa, Peru, IVITA. 31 p.
LI PUN, H.H.; BOREL, R. 1986. La investigaci6n en componentes en el

proceso de investigacion en sistemas de producci6n animal. In
Reunion de Trabajo sobre Sistemas de Producci6n Animal (6., Pan.,
1986). Informe. H.H. Li Pun, N. Gutierrez (Eds.). Bogota, Col., IDRC
MR139s. p. 10-43.
LI PUN, H. H., ESTRADA, R.D.; SERE, C. 1990. International cooperation

for the development of dairy production in Third World countries: The
IDRC strategy. In Symposium on Alternatives for Improving Traditional
Dairying in Developing Countries, 23rd International Dairy Congress,
Montreal, Can. p. 208-227.
LI PUN, H. H. 1991. Dairy production research in developing countries:

IDRC experiences. Paper presented at the Expert Consultation Meeting
on Dairy Development organized by FAO. Rome, Italy, FAO. July 2-4,
1991.
MINISTRY OF AGRICULTURE. 1987. Livestock and range research in

Botswana. Gaborone, Botswana. Animal Production Research Unit.
105 p.
RIESCO, A. 1990. Adoption of improved livestock production practices in

the Peruvian Amazonia: An econometric analysis. Ph.D. dissertation.
Ames, Iowa, Iowa State University. 205 p.
LIVESTOCK SYSTEMS RESEARCH AT CIRAD
Bernard Rey', Philippe Lhoste2

and Vincent Dolle3
INTRODUCTION
CIRAD (Centre de Cooperation Internationale en Recherche

Agronomique pour Ie Developpement) is the French agency in charge of
research on tropical agriculture. Because the government is its major
donor, CIRAD's programs illustrate the French policy towards agricultural
research in the tropics, specifically towards livestock systems research.
CIRAD is a consortium of eleven research departments dealing with major
food and cash crops in the tropics as well as with livestock and forestry.
Two departments are involved in livestock systems research: IEMVT and
DSA.
IEMVT (Institut d'Elevage et Medecine Veterinaire des Pays Tropicaux)

has extensive experience in the West African countries. The Livestock
Systems Program is one of IEMVT's five main programs; it deals with
the analysis and understanding of livestock systems, the development
of reference data sets for indigenous livestock flock productivity, the
crop-livestock relationship, and the economics of livestock with a
"whole-commodity system" approach. IEMVT has been involved in
research conducted on-station and began on-farm research in the
1970s. IEMVT carries out research with a systems perspective in
several African countries, the West Indies, New Caledonia, and Reunion
island.
IEMVT-CIRAD seconded to ILCA/CIPEA, P.O. Box 5689, Addis Ababa, Ethiopia.
2 IEMVT-CIRAD Place Viala, 34060 Montpellier CEDEX, France.
3 DSA-CIRAD; BP 5035, 34032 Montpellier CEDEX, France.

DSA is the Department for Agrarian Systems. Its mandate is to carry

out research with a development perspective. DSA was created along
the lines of the first R&D programs, particularly the Sine Saloum
Experimental Units in the late 1970s in Senegal (Benoit-Cattin 1986).
It is staffed with socioeconomists as well as biological scientists. All
DSA research activities follow a systems approach. Livestock systems
research becomes the dominant research activity when livestock is the
key component in the farming system, as is the case in three projects
in Brazil and Venezuela.
One research project by IEMVT in Senegal and another by DSA in

Venezuela will illustrate CIRAD activities in livestock systems research.
A MULTIDISCIPLINARY RESEARCH MODEL FOR LIVESTOCK

SYSTEMS (SENEGAL)
The research model
Landais and Faugere (1990) presented a research model developed

from the analysis of the difficulties encountered with diagnostic studies of
livestock systems. They subsequently discussed its implications for data
management and multidisciplinarity work.
There is an increasing awareness of the need for reference data sets

for livestock productivity in Africa, particularly to monitor progress and
identify domains of improvement. On-farm monitoring techniques and
procedures have been developed by IEMVT-CIRAD and other institutions
over the last 10 years, and there are now many such data sets, although
analysis and reporting still lacks homogeneity.
Although monitoring, based on long and costly diagnoses4, produces

accurate information, it does not lead per se to the proposal of alternatives.
To justify the search for appropriate innovations, the diagnostic stage must
explain and assess the performances of livestock and quantify room for
improvement in animal production. Livestock systems research needs a
4 As a consequence, on-farm monitoring livestock performances has been severely

criticized. Paradoxically, everyone recognizes that on-farm testing of alternatives
requires on-farm monitoring of the same performances.
model to combine the experience gained in livestock monitoring with the

contribution of various disciplines at different stages of the research. Since
each individual scientist does not need all the information on the livestock
system at the same time, data management must be flexible.
To insure multidisciplinarity, a common basis for communication

among scientists has to be found. Data on the demography of the
population under study constitutes the core of the research system and
needs to be shared by all scientists. An individual monitoring procedure
provides excellent information on population demography, resulting from
an aggregate of data at an inferior hierarchical level.
The productivity and pathology project in Senegal
The aforementioned philosophy has been used to develop the software

PANURGE in the framework of the Pathology and Productivity of Small
Ruminants Research Project, a project jointly run by the Institut Senegalais
de Recherches Agronomiques (ISRA) and IEMVT.
PANURGE was developed on a modular basis with demography as the

core data set. Given the objectives of the research, two additional
modules were initially developed: (1) growth and adult weight monitoring
and (2) health monitoring. In subsequent years, interrelated modules were
developed to cater to one-shot surveys (feeding practices, milking
practices, etc.), multiple-shot surveys (e.g., parasitemia, serology, housing),
and permanent monitoring (mainly economic aspects), which are being
applied to all or to a segment of the flocks under study, depending of the
type of information required. Experimental variables are also being
monitored in the data base at the individual level.
The field data collection is organized to insure a proper and rapid flow
of information from the flocks to the centralized data management
operator. Data validity is checked before entering the data base. In
Senegal, three areas have been under study in different ecozones since
1984. In each site, more than 1,000 sheep and 1,000 goats are
permanently monitored.
The comprehensive data base offers many analytical possibilities, but

only a few of these will be stressed. Results have highlighted the seasonal
and interannual variations in flock size: the number of adult females varies
by more than 20% and males by more than 50 percent. This program
consequently allows an unbiased estimation of the flock productivity per
female per year, thus avoiding the bias which would appear with methods
based on a single enumeration. The establishment of relationships
between health status and flock productivity allows for new epidemiological
research developments on chronic and subchronic diseases. The
economic data base allows the translation of the biological flock
performances into monetary terms and the ex-ante analysis of the
interventions proposed in the diagnostic stage.
This research setup may be also used to test innovations on a variety

of sites and species. So far, improvements in the feeding system and
health treatments have been tested; in fact, treatment with a broad-
spectrum antihelminthic has been shown to have a series of nonspecific
positive effects on mortality, age at first kidding, abortions and other
aspects, which have been quantified and combined to improve flock
productivity in biological and economic terms (Faugere et al. 1987).
Discussion
A comprehensive research system based on continuous gathering of

information provides a good understanding of livestock systems in a
variable environment like that of the Sahel. The contribution of different
disciplines to this understanding can be enhanced by software conceived
to promote interdisciplinary communication. Such is the case of livestock
systems models envisaged with a unique data base for future development.
The experimental capacity of a research system model is worth

comment. Like in a station flock, all animals' backgrounds are known.
This allows the design of experiments with randomly or purposely selected
sheep and goats, with a precision similar to the one which could be
obtained on station. Furthermore, the analysis can address before-versus-
after comparisons of the same flock, as well as treatment-versus-control
comparisons. Moreover, most of the data necessary to analyze the trial
are already part of the system; any supplementary data needed could be
collected at minimum cost.
Three points are frequently discussed in relation to on-farm livestock

monitoring: number of animals, duration, and cost.
Landais and Faugere (1990) argue that such a research system is not
justified for monitoring less than 1000 heads of a given species per
site.
The same authors observed that meaningful results for small ruminants
could be obtained with three to five years of monitoring, which allowed
compensation for high environmental variability.
Roughly, proportional costs (except researchers' salaries) were

estimated at 3,000 CFA5 per head monitored per year6. Although
expensive,' it may be argued that such a research system is competitive
with the cost of keeping a flock of a similar size on a station, or with
the cost per scientist involved.
The publication of a user's manual (Faugere et al. 1988) permitted the

implementation of PANURGE in Burkina Faso and in Cote d'Ivoire for the
monitoring of large numbers of animals over several years.
A NETWORK OF REFERENCE FARMS FOR SYSTEMS DEVELOPMENT

(VENEZUELA)
A research program set in a development context
The Venezuelan research institute, FONAIAP7, started a research and

development project in the Aroa and Bajo Tocuyo valleys in 1983. DSA
has been associated with this project since 1985 when technological
alternatives were first tested on farm.
The farming systems of the area were characterized during the first
phase of the project (1983-1986). An environmentally adapted, dual-
purpose cattle production system was being.used in small- and middle-
sized farms. The diagnosis of such a system provided important clues for
improving its productivity.
5
Or Franc CFA, a monetary unit (Communautr Francophone Africaine).
6
The set up of on-farm monitoring is expensive (identification of farmers and animals,
training of enumerators, etc.). To be acceptable, such costs should be spread over
several years of monitoring.
7 See the List of Acronyms and Abbreviations.

Research objectives were twofold: (a) to test and evaluate

technological alternatives for improving productivity of dual-purpose cattle
in the farming systems which had been characterized, and (b) to build up
a data set of the technical and economic aspects of the livestock systems,
to be used by the extension services. A spin-off of these objectives was
the development of methods for data collection and research on dual-
purpose livestock systems.
An approach based on a network of reference farms
In the 1970s, under farmers' pressure, the French extension services

developed monitoring and analytical procedures to compare farms and
identify prospects for their development. An underlying principle is that
proposals to improve farming systems need validation of research results
under local condition and at the farm level. The farms where this validation
takes effect are called "reference farms," a network of which was developed
by the FONAIAP project, following the French experience (Bonnal et al.
1988).
In Venezuela, the methods of data monitoring and feedback to farmers

were tested in a few "pilot farms." In a second phase, the monitoring
technique was reduced as much as possible and implemented in a larger
number of reference farms. Finally, "satellite farms," linked with each
reference farm permitted the diffusion of the innovations tested in the
reference farms. Researchers provided their professional experience to
validate the information gathered, as well as the framework for its analysis
in a systems perspective. This was expected to generate a dynamic in
which research could be phased out after some years, and farmers and
extensionists could carry on the process.
Choice of reference farms. Reference farms must represent the

diversity of farming situations and farmers' objectives, which is not
synonymous with statistical representativeness. Consequently, in
Venezuela the 53 reference farms were not randomly chosen, but
purposely selected on the basis of the knowledge gained during the
diagnostic stage.
Data collection. Data are gathered to provide information on the major

constraints identified. Farmer participation in this task is sought to
minimize the recording frequency. Feedback on the information gathered
is provided to the farmers each month, each six months, or each year,
according to the topic.
Data analysis and advice to farmers. The objective of data analysis is

to advise farmers on technical innovations which can improve their dual-
purpose cattle systems; recommendations are specific to each "reference
farm" where they are tested. The data base is used to compare farms
among themselves. Fifteen indices were used to describe the system
performance and to serve as a basis for comparison among farms. The
index with the lowest relative value (compared to other farms) will
represent the area that should have priority for improvement in the system.
Recommendations will then address an improvement of this index. The
economic information in the data base permitted a financial ex-ante
analysis.
Over the years, such a comparison can also be carried out to analyze
the evolution of the system and the ex-post, effect of the innovations. The
methods which have been developed will lead to a national network of
reference farms in Venezuela.
OTHER FEATURES OF LIVESTOCK SYSTEMS RESEARCH AT CIRAD
CIRAD researchers and other French scientists have made an effort to

define livestock systems (Lhoste 1984; Landais et al. 1987), which has
provided a guideline for the diagnostic process, a framework for the
modeling of livestock systems, and a theory for comparative development
of livestock systems.
Many other research methods have been used to look at the diversity
of livestock systems at different levels of analysis, in addition to the
succession of diagnostic and experimentation phases described in this
paper. For example, multivariate and cluster analyses have been used
extensively to define farm types (Cervantes et al. 1986; Salas et al. 1986);
they have proved very useful to researchers and policy makers for
understanding the diversity of farmers' objectives, constraints,
performances, and needs.
Some farmers' practices appear to vary, while others tend to be

relatively constant in a region; attempts have been made to explain the
variation in productivity of farmers' herds as a result of the different
management strategies followed. Similarly, plot analysis, developed for
annual crops, has been adapted for fodder crops (Cesar and Coulibaly
1990). An ecopathological approach has been developed to understand
why severity of chronic and subchronic pathologies varies among herds

within a similar environment, and how this can be related to farmers'
management (Lefevre 1989).
Crop-livestock interactions need to be addressed with a systems

perspective. Several CIRAD scientists collaborate with other institutions'
multidisciplinary teams in Burkina Faso, Cameroon, Senegal, and Brazil
(Landais et al. 1990; Cahiers de la Recherche Developpement 1985, 1986).
CONCLUSIONS
While both are implementing livestock systems research, the contrast

between the two projects presented shows the complementarity of IEMVT
and DSA. Livestock systems research, and more generally systems
research, is still supported by CIRAD's main donor. However,
shortcomings of past experiences are pushing CIRAD scientists to develop
new tools to improve their research efficiency within a systems perspective.
Is there a French approach? Although the general philosophy of

farming systems research does not vary between the English-speaking and
French-speaking "worlds," some aspects could be singled out as being
more specific to the French work: a greater interest is placed on the time
frame and dynamics of the system; a greater role is assigned to
socioeconomics in the research process; and the farm level (rather than
the herd level) is usually the focus when conducting livestock systems
research.
LITERATURE CITED
BENOIT-CATTIN, M. (Ed.). 1986. Recherche et developpement agricole:

Les unites experimentales du Senegal. ISRA, CIRAD, FAC. 500 p.
BONNAL, P.; CASTILLO, J.; DOLLE, V. 1988. Un reseau de fermes de

references comme outils d'observation et de gestion du milieu rural le
project Aroa-Bajo Tocuyo au Venezuela. Les Cahiers de la Recherche
D6veloppment No. 19. p. 59-74.
CAHIERS DE LA RECHERCHE DEVELOPPMENT. 1985. Relations

Agriculture-Elevage No. 7. France, CIRAD.
CAHIERS DE LA RECHERCHE DEVELOPPMENT. 1986. Relations

Agriculture-Elevage Nos. 9 and 10. Montpellier, France, CIRAD.
CERVANTES, N.; CHOISIS, J.P.; BOUCHIER, A.; LHOSTE, Ph. 1986. Une
typologie des elevages bovins de I'etat de Colima au Mexique:
premiere etape du diagnostic. Revue d'Elevage et de Medecine
Veterinarie des Pays Tropicaux 39 (1): 21-28.
CESAR, J.; COULIBALY, Z. 1990. Le role des jacheres et des cultures

fourrageres dans le maintien de la fertilite des terres. In Savanes
d'Afrique, terres fertiles?. Actes des rencontres internationales, 10-14
December, 1990. Montpellier, CIRAD. p. 271-287.
FAUGERE, O.; LEFORBAN, Y.; NERCY, C.; NDIAYE, M. 1987. Essai de

traitement des affections respiratoires des petits ruminants du Sine
Saloum (Senegal) a ('aide dune oxytetracycline a longue action.
Revue d'Elevage et de Medecine Veterinaire des Pays Tropicaux 40 (1):
21-32.
FAUGERE, 0.; LANDAIS, E.; FAUGERES, B. 1988. PANURGE Manuel.

1-Le suivi sur le terrain et la tenue des fichiers manuels., 134 p. 2-Le
fichier informatique, 200 p., 3-Annexes, 260 p. Maisons Alfort, France,
Institut d'Elevage et de Medecine Veterinaire des Pays Tropicaux.
LANDAIS, E.; LHOSTE, Ph.; MILLEVILLE, P. 1987. Points de vue sur la

zootechnie et les systemes d'elevage tropicaux. Cahiers des Sciences
Humaines, ORSTOM 23 (3-4): 421-437.
LANDAIS, E.; FAUGERE, O. 1990. Un modele illustre de systeme

d'investigation pour ('etude pluridisciplinaire des systemes d'elevage en
milieu traditional Africain. Les Cahiers de la Recherche Developpment
No. 25. p. 75-93.
LANDAIS, E.; LHOSTE, Ph.; GUERIN, H. 1990. Systemes d'elevage et

transferts de fertilite. In Savanes d'Afrique, Terres fertiles. Actes des
rencontres international es, 10-14 December, 1990. Montpellier, CIRAD.
p. 219-270.
LEFEVRE, P.C. 1989. Problematique des enquetes d'ecopathologie dans

les pays du Tiers-Monde. Capricorne (France) 2:4, 7-9.
LHOSTE, Ph. 1984. Le diagnostic sur le systeme d'elevage. Cahiers de

la Recherche D6veloppment 3-4: 84-88.
SALAS, M.; PLANCHENAULT, D.; ROY, R. 1986. Etude des systemes

d'elevage bovin traditionnel en Guadeloupe: Typologie d'elevage.
Revue dtlevage et de Medecine Veterinaire des Pays Tropicaux 39(1):
59-71.
ANIMAL PRODUCTION SYSTEMS RESEARCH:
GTZ's EXPERIENCES AND PROSPECTS
Gunter Kleemann'
INTRODUCTION
The government-owned Deutsche Gesellschaft fur Technische

Zusammenarbeit mbH (GTZ) was founded in 1975 with the legal status of
a private company and acts on commission of the Ministry for Economic
Cooperation of the Federal Republic of Germany. It has responsibility in
the planning, execution, and evaluation of technical cooperation projects
and programs with developing countries on a nonprofit basis. GTZ
generally uses its own staff and service facilities, but to some extent also
subcontracts with specialized consultancy firms and private consultants.
Besides carrying out official government tasks, GTZ can also act as an
independent consulting firm and execute projects and programs for other
donor agencies, the World Bank, regional development banks, the
European Community, or national governments, against payment.
Since its reorganization in 1989 based on a regionally oriented

structure, agricultural development and research have formed part of the
Planning and Development Department and the correspondent
Subdepartment 42: Agriculture, Forestry and Conservation of Natural
Resources. Theoretically, farming systems research can be promoted
through all four of the Subdepartment's divisions: Agricultural Production
Systems; Animal Production, Veterinary Services and Fishery; Plant
Production and Plant Protection and International Agricultural Research;
and Forest Resources Management. In 1990 GTZ was in charge of 2278
projects in 104 countries employing directly or indirectly a total staff of
7933 persons. GTZ and its predecessors have gained experience in
livestock development and the execution of relevant projects during the
last three decades in Africa, Asia, and Latin America.
1
GTZ, Postfach 5180, 65726 Eschborn, Germany.
WHY SUPPORT SYSTEMS RESEARCH?
The importance of agricultural systems research in general and animal

production systems research (APSR) in particular has been recognized by
the German Ministry for Economic Cooperation and the predecessors of
GTZ since the beginning of the 1970s. This was due in part to the financial
and personal support given since that time to the international agricultural
research centers by the German government and to the feedback
produced by this engagement.
The experiences of GTZ projects over the last decades lead to the
conclusion that existing animal production systems must be considered in
all their complexity and that intensification or change can only take place
after the main limiting factors have been determined. It became clear that
the understanding of production systems with their weak and strong points
is an essential requirement for successful project work and that criteria of
success should be not only economic ones but should also include
sustainability, as well as social and ecological compatibility. This is in
conformity with the principle of German development policy that considers
the human being as the focus of its objectives (Grosse-Herrenthey 1991).
It must also be made clear that in order to be successfully executed,

research activities on animal production systems within the framework of
a GTZ livestock or rural development project must be initiated as a result
of the participative planning procedure known as ZOOP (Objective-
Oriented Project Planning). Additionally, they must have the following
logistical and material support:
Specially trained personnel and interdisciplinary groups
Scientific feedback from a research institution in Germany or elsewhere

interested in the same subject
Interest and support from a national research institution which

recognizes the importance of on-farm research
Interest, commitment and creativity of the project leaders
Communication with other institutions working in similar areas and

exchange of experiences
Resources (time, money, personnel, literature, farms, methodologies,

instruments, microcomputers, etc.)
Although the development of animal production systems is one of the

main objectives of the corresponding GTZ department, the support of
APSR, as described in this paper, has not been centrally planned and
executed. It is, rather, the result of the requirements of specific project
types, enjoying national and personal support and interest on all levels,
inside and outside of GTZ. In this context, it is important to know that one
of the main principles of GTZ project policy is the high grade of
independence and responsibility of the German team leader or single
expert.
As no special policy paper on APSR exists in GTZ, the statements made

here are based on headquarters communications related to the subject,
publications of GTZ staff, and personal appraisals and experiences of the
author.
EVOLUTION OF SUPPORT TO AGRICULTURAL PRODUCTION

SYSTEMS RESEARCH AND THE RELATIVE IMPORTANCE
OF ANIMAL PRODUCTION SYSTEMS
International agricultural research has always been supported by the

Federal Ministry for Economic Cooperation BMZ and GTZ. Traditionally,
farming systems has been a main subject for German students of
agricultural science, and research on farming systems in the tropics has
had a stronghold in Germany (thanks to outstanding figures like Professor
Hans Ruthenberg). In the early 1970s, the German Council for Tropical
and Subtropical Agricultural Research (ATSAF) was founded. In 1971 the
Federal Republic of Germany became a member of the Consultative Group
of International Agricultural Research (CGIAR) and the major donor to the
International Centers after the USA and the World Bank. In 1979, the
World Bank charged GTZ with the foundation of the International Service
for National Agricultural Research (ISNAR). The budget for international
agricultural research executed through GTZ increased sharply from 6.5
million Deutsche marks (Dm) in 1988 and 6 millon Dm in 1989 to 19.7
million Dm in 1990. This corresponds to 1.4% of the total financial supply
received by GTZ in 1990 (GTZ 1991).
The information available to the author does not permit the

quantification of the exact amounts and the evolution of the support given
by GTZ to agricultural research in general and to APSR, in particular, due
to the different ways and strategies through which GTZ supports relevant
activities.
A directory elaborated by Weniger and Miescke (1981) on animal

production research in the tropics and subtropics carried out by German
scientific institutions indicated six projects related to animal production
systems from a total of 72 registered. Support to traditional agricultural
research still prevails, as shown in a directory of current projects in
agricultural research for the tropics and subtropics carried out by research
institutes in the Federal Republic of Germany (Weniger 1988). From a total
of 657 projects registered, just 11 are mentioned under the keywords farm
systems, farming systems, and farming systems research, and 64 research
projects correspond to the area of animal production, excluding fisheries
and veterinary science. Thirteen research projects related to farming and
animal production systems conducted by German research institutions with
funding through BMZ and GTZ could be identified from this directory; these
are listed along with others in the next section.
TYPES OF PROJECTS SUPPORTED
The types of support BMZ and GTZ are offering to APSR are of broad
variance. They can be subdivided into two groups: (1) support given by
the BMZ/GTZ headquarters and (2) support given through a specific
project. In any case, such research is normally financed from project
funds and conducted in cooperation with national or international research
institutes. Another minor source of financing can be GTZ funds proper.
BMZ/GTZ headquarters can offer funding to support the following:
Research-oriented projects (national institutions, universities) (e.g.,

Colombia until 1989)
Sector-oriented projects (e.g., Ecuador, Colombia, Sri Lanka, Malaysia,

Burundi, Pakistan)
Integrated rural development projects

International centers (e.g., CIAT, ILCA)2
Scholarships for postdoctoral studies at international research centers

or projects (e.g., CIAT, Colombia)
Specific research involving GTZ, German universities and international

research institutes (e.g., the ETES2 project with CIAT, Technical
University of Berlin, and GTZ participation)
Scientific feedback given by German universities (Technical University

of Berlin, Universities of Gottingen, Stuttgart, and Hohenheim)
Membership in research networks such as REAPER (Reseau Euro-

Africain sur le Petit Ruminants) (REAPER 1991)
Organization and co-organization of conferences and workshops (e.g.,

the seminar on Coordination of Research for the Development of Small
Ruminants in Africa in Montpellier, France) (CTA 1986)
Interdisciplinary three-month field studies carried out by the Center for

Advanced Training in Agricultural Development from the Technical
University of Berlin (Peters et al. 1979; Kleemann et al. 1983;
Maldonado et al. 1987; Valle-Zarate et al. 1987)
Assignment of integrated-focus experts, placed by the Center for

International Migration and Development and under contract with a
local institution.
The kind of support that can be given to APSR by a GTZ livestock

sector project has been described in detail by Kleemann (1990) for the
case of the Colombo-German ICA/GTZ3 projects "Intensification of animal
disease control" (1980-1989) and "Introduction of a system of integrated
technical assistance in livestock" (1989-1994). The main activities involved
in this support are given below:
Development of monitoring schemes for cattle and pig farms (Otte et

al. 1989) and application of such schemes to more than 300 farms in
15 different regions of Colombia (Lezzaca et al. 1991)

Development of a software (MONTY) for monitoring production and

economic parameters of cattle and mixed farms (Kleemann et al. 1991;
Reyes 1991)
Conduction of specific farm surveys (cattle, pigs, goats, epidemiology,

feeding systems, and production systems)
Training of counterparts abroad at the University of Reading and within

Colombia at the International Training Center for Livestock Development
(CICADEP) in farm monitoring, data evaluation, and information
management using of microcomputers (Villamil 1986; Navarrete 1988;
Kleemann et al. 1989; Lobo and Kleemann 1991)
Elaboration of computer models to assess productive efficiency of cattle

production systems (Villamil 1988); relations between seasonality,
reproduction, and production (Kleemann et al. 1988); and economic
viability of supplementary feeding in cattle
Training of national and German students leading to diploma, M.Sc. and

Ph.D. degrees (Schellenberg and Weniger 1985; Villamil 1986; Navarrete
1988; Reule 1989; Velasco 1989; Rodriguez et al. 1991)
Organization of international courses and symposia on herd health and

production management in dairy and dual-purpose systems and on
goat production (Colombia, 1986 and 1987)
Collection of literature summaries on livestock production systems

research in Colombia (Lopez 1991) and fodder tree and bush research
in Latin America
Consultancies
Visits to similar projects
Project meetings and exchange of experiences
Participation in seminars and congresses
Elaboration of respective publications
When it comes to putting the project strategies into practice, a variety

of mechanisms are employed due to the complexity and variability of the
type of support given by GTZ headquarters or a single project. Relations
with international research institutions are usually channelled through GTZ

headquarters, whereas the single project develops its own mechanisms of
relations within the respective country. Feedback and contacts are
provided on request by the GTZ Department of Animal Production. Many
of the project leaders maintain special links to interested investigators at
their "home" university. GTZ promotes human resource development in
agricultural research both through scholarships for German doctorate
programs (on request of interested projects), and through a project
assistance program. GTZ is implementing a pilot monitoring and
evaluation scheme in several projects. All GTZ projects are submitted to
the standard planning and evaluation scheme requested by BMZ. Projects
with their own farm monitoring schemes (Malaysia, Ecuador, Colombia,
and Costa Rica) are theoretically able to perform the evaluation of their
impact at the farm level.
BRIEF DESCRIPTION OF TYPE OF PROJECTS SUPPORTED DURING

THE PERIOD OF 1970-1990
The number of projects carried out by GTZ during 1970-1990, as well

as the lack of precise and complete information on support given to APSR,
and of a precise definition of where and when actual research starts and
ends, make it somewhat difficult to give brief descriptions of all types of
projects supported so far. Two lists of projects with activities that
include(d) farming and/or APSR are given below. One list refers to
projects executed by GTZ in three continents (incomplete); the other refers
to projects conducted by other German institutions, mostly universities,
with financial support from BMZ/GTZ (complete).
Some GTZ projects with activities related to APSR
Pattoki Livestock Production Project (PLPP), Phase (1978-1981),

1
Pakistan: Farming systems research and land-use patterns, economic

evaluation of various livestock farming systems, establishment of
physical models, dairy production, buffalo keeping.
Pattoki Livestock Production Project (PLPP), Phase II (1982-1992),

Pakistan: Artificial insemination, veterinary and animal health care
program, supply of farm inputs, village extension service, promotion of
women, village milk procurement system, milk processing plant, and

milk distribution network.
Promotion of Dual-Purpose Cattle Improvement (PROMEGAN) (1975-

1981), Colombia: Research on dual-purpose and beef cattle farm
systems, artificial insemination with Simmental, technical assistance in
breeding, feeding and health care assistance for pilot farms, animal
nutrition and reproduction laboratories at a local university.
Intensification of Animal Disease Control (1980-1989), ICA/GTZ,

Colombia: Development of microcomputer-based farm-monitoring
methodologies and instruments, health and production surveys in cattle,
pig and goat production systems, training, and establishment of
information systems.
Integrated Technical Assistance

in Livestock Production (1989-1994),
ICA/GTZ, Colombia: Nationwide technology transfer and monitoring of
over 300 farms involving cattle and pigs, continued education in
methods and instruments of integrated technical assistance,
development of "software" modules for economic and mixed-farm
monitoring, research in cattle feeding systems.
Promotion of Cattle Production (PROFOGAN) (1986-1993), Ecuador:

Nationwide research on cattle production systems, economic analysis
of production systems, pasture management and improvement, cattle
marketing, national livestock sector planning.
Livestock Planning Unit (LPU) (1983-1988), Sri Lanka: Data collection

for economic analyses of dairy production systems, project benefit
evaluation of different livestock development projects, baseline survey,
impact analysis survey, development of a data acquisition system,
livestock statistics.
Smallholder Integrated Livestock Extension Project (SILEP), Sri Lanka

(in preparation): Baseline survey to identify farming systems and the
socioeconomic situation of farm families, monitoring system,
development of integrated crop and livestock systems.
Promotion of Integrated Animal Production (1987-1993), Ruyigi,

Burundi: Agroforestry and pastoral integration and diversification in
smallholder farms, baseline survey, small ruminants, rabbits, use of
dung, erosion control.
Range Management Handbook (RMH) (1987-1994), Kenya: Elaboration

of a data base on ecological and socioeconomic conditions for
rangelands in and and semiarid land districts, surveys, data collection
on vegetation, soil, climate, wafer, infrastructure, human and livestock
population, socioeconomics, land use, marketing and production
systems, elaboration of handbook.
Research projects related to farming and animal production systems

carried out by German universities and sponsored by BMZ/GTZ,
according to Weniger (1988) and the author's own information
Research on milk and meat production in different farming systems in

the tropical lowlands of Northern Colombia (1974-1976). Technical
University of Berlin, CIAT, GTZ (Schellenberg and Weniger 1985).
Extensive cattle production systems: Brazil, Colombia, Venezuela (1978-

1982). (ETES Project: Technical and Economic Studies on Animal
Production Systems). CIAT, Technical University of Berlin, GTZ
(Weniger 1981; Vera and Sere 1985).
Milkproduction in dual-purpose cattle farms in Pie del Monte Uanero

(Meta, Colombia) (1982). Technical University of Berlin, Center for
Advanced Training in Agricultural Development (Kleemann et al. 1983).
The role and outlook of swine production in improving small farmers of

the Gran Chaco Province, Bolivia (1986). Technical University of Berlin,
Center for Advanced Training in Agricultural Development (Valle-Zarate
et al. 1987).
Agricultural systems in four regions of southern Ecuador (1987).

Technical University of Berlin, Center for Advanced Training in
Agricultural Development (Maldonado et al. 1987).
Research on dual-purpose systems and the role of improved grass-

legume pastures for milk and beef production on acid soils of tropical
America, Colombia (1989-1991). Technical University of Berlin, CIAT,
GTZ (Weniger 1988).
Goat production in low-income economic units in West Malaysia.

Technical University of Berlin, Center for Advanced Training in
Agricultural Development (Peters et al. 1979).
Development of adapted farming systems on the basis of draft animal

utilization in the Northwest Province of Cameroon (1980). University of
Gottingen, GTZ.
Comparative economic analysis of hoe and bullock farming systems in

Northern Ghana (1982-1986). University of Gottingen, Crops Research
Institute (Ghana), GTZ.
Small ruminant production in a smallholding agricultural area in a dry

subtropical environment (Egypt) (1985-1987). University of Gottingen,
Cairo University, BMZ, DFG (Peters et al. 1979).
Traditional pig production in smallholdings of Zimbabwe (1986-1988).

University of Gottingen, Pig Industry Board of Zimbabwe, BMZ/DFG
(Peters et al. 1979).
Analysis and planning of farm households in the tropics (1981).

University of Hohenheim, ICRISAT, ICARDA, BMZ/GTZ.
Intensification and stabilization in farming systems (Africa, Asia) (1983

to present). University of Hohenheim, BMZ/GTZ.
Rabbit production in the Sudan Zone of West Africa (Burkina Faso)

(1985-1987). University of Hohenheim, GTZ.
Traditional goat production in the Santander Province of Colombia

(1985). University of Hohenheim, ICA/GTZ (Reule 1989).
Small ruminants in West Africa (Cote d'Ivoire) (1986-1988). University

of Hohenheim, GTZ.
Agroforestry systems in the tropics (1977). Bundesforshungsanstalt fur

Forst- and Holzwirtschaft Hamburg, GTZ, ICRAF, National Institutions
in Indonesia, Fiji and Senegal.
RESULTS AND ASSESSMENT OF IMPACT
The activities promoted by GTZ in APSR have had positive and multiple
results at all levels. Most important seems to be the rising awareness of
all participants involved of the importance of research and development
activities with a farming systems perspective. The new livestock

development concepts of GTZ take into account the respective lessons
learned in the past.
Another result is the growing recognition that animal production in most

farms (smallholder, traditional) is recognized as only one component of
more complex and already fairly adapted farming systems, and also as a
component of rural development. As a founding member of REAPER in
1986, and in charge of its secretariat, GTZ is aware that small-ruminant
research must be integrated into the general development system and that
such research must take into account, among others, the following
determining factors (REAPER 1991):
Farming systems and their foreseeable evolution
The change of land-use patterns in quest of integrated sustainable

systems
The relationship between production systems and the economic

environment
GTZ and other collaborating German and international institutions have

been strengthened through network participation and the feedback
received from the results of APSR. At the same time, the number and
quality of human resources able to contribute to APSR in Germany has
increased considerably. There are many examples showing a positive
impact on national research and human resource development, as well as
others where the expected positive effects could not be sustained or
achieved at all.
More precise quantitative and qualitative information on impact

assessment of research in animal production systems within GTZ is not yet
available; however, aware of the need, several projects are starting to
develop and implement methodologies to study the impact of its activities,
including research.
CONSTRAINTS AND FACILITATING FACTORS
The statements on general constraints in this chapter are based on the

author's own experiences and do not necessarily reflect the experiences
of other GTZ projects.
National counterpart research institutions may neither give adequate

priority to research in farming and animal production systems, nor be able
to form the necessary interdisciplinary groups and guarantee their
continuity. There are just a few (and many times insufficiently trained)
researchers available in the developing countries, and they have
deficiencies in the areas of methodology, data analysis, and publishing.
Under these circumstances, the relatively high costs of scientific

backstopping from abroad can become a main constraint as project
budgets become tighter year after year. In general, the lack of an
organized exchange of experiences between projects, institutions, and
researchers within and between countries and continents is still a mayor
limiting factor, despite progress made through network activities.
Some of the constraints are internal to GTZ and are related to the fact
that research is usually considered a side activity in a typical development-
oriented project. Under these conditions time, human resources, and
funding become even more limiting factors. Moreover, traditional projects
and research approaches focusing on just one or two components of
animal production are still quite common, creating an inappropriate
environment for research with a systems perspective.
An article by Steiner and Werner (1988) reviewed and analyzed GTZ

experiences with on-farm research in rural development projects. The
authors arrived at the following conclusions about constraints:
Cooperation with national agricultural research institutes is often so

difficult that projects frequently have to introduce on-farm research
entirely on their own.
On-farm research within the project requires an expert to deal solely

with this field. The success of this research depends on the quality of
training provided to the personnel involved.
On-farm research cannot develop new extension programs overnight.

In particular, meeting the necessary preconditions can be very time-
consuming and difficult.
As in many other cases, there are fewer factors facilitating activities than
there are constraining them. Still, these are the most important ones when
it comes to assessing the potentials and future of research in animal
production systems. First of all, GTZ's growing interest in and awareness
of tropical farming systems research has been essential in achieving the
existing level of participation and the results obtained. Equally decisive has
been the personal interest shown by all participants in this process, such
as farmers, students, researchers, and extension workers. The feedback
originated has resulted in increased motivation and creativity and a better
understanding of each others' needs and expectations, facilitating future
work as well as improving results and project efficiency.
OUTLOOK
The prospects of GTZ's continued support and contribution to APSR in

the future are quite promising in light of the new strategies, concepts, and
priorities documented in recent internal policy papers (Grosse-Herrenthey
1991). Priority will be given to projects in the area of animal production in
and regions, integration of animal production into agricultural farming
systems, and market-oriented animal production. New vertical project
strategies that are systems-oriented and multidisciplinary will include and
emphasize aspects such as those listed below:
Recognition of traditional production systems
Ecological compatibility through protective and sustainable use of

natural resources
The role of women in animal production
New target groups such as the urban and landless population
Marketing and distribution of animal products
Training and participation in networks and conferences
Applied research projects
Although a substantial increase of financial aid in general, and research

in animal production systems in particular, is not very likely to take place
in the near future, GTZ is willing to and capable of improving its
performance in this area through coordination, communication, and
institutional and network strengthening.
LITERATURE CITED
CTA (CENTRE TECHNIQUE DE COOPERATION AGRICOLE ET RURALE).

1986. Coordination de la recherche pour developpement -des petits
ruminants en Afrique. Montpellier, France. Centre technique de
cooperation agricole et rurale, compte rendu du seminaire 13-17
Octobre 1986. Convention ACP-CEE de Lome. Wageningen-Ede, The
Netherlands.
GROSSE-HERRENTHEY, E. 1991. Uberlegungen zur Forderung der

Tierhaltung and Tiergesundheit. Strategien and Ansatze der Deutschen
Gesellschaft fur Technische Zusammenarbeit (GTZ) GmbH. GTZ-
internes Positionspapier. Eschborn, Federal Republic of Germany. 12
p.
GTZ (DEUTSCHE GESELLSCHAFT FOR TESCHNISCHE

ZUSAMMENARBEIT). 1991. Jahresbericht 1990. GmbH. Eschborn,
Federal Republic of Germany. 66 p.
KLEEMANN, G.; KRAIN, E.; KUCHENREUTHER, R.; OTERO, A.; SAUTER,

H.; THAYSEN, J.; WARNER, A. 1983. Situacion actual y potencial de
la produccion lechera en explotaciones de doble propbsito en el
Piedemonte Llanero (Meta, Colombia). Serie: Studien, Nr. IV/40,
Seminar fur Landwirtschaftliche Entwicklung, Teschnische Universitat
Berlin. 309 p.
KLEEMANN, G.; PETERSEN, S.; MONCADA, H. Modelo

1988.
computarizado para determinar relaciones entre estacionalidad,
reproduccion y produccion doble. In Reunion de la Asociacion
Latinoamericana de Produccion Animal (11., 1988, La Habana).
Resumenes. p. 163.
KLEEMANN, G.; LOBO, C.A.; GONZALEZ, C.A. 1989. The International

Training Center in Livestock Development CICADEP (Colombia):
Experiences and prospects. Conference book 6th International
Conference of Institutes for Tropical Veterinary Medicine. 28 August -
1 September 1989, International Agriculture Centre, Wageningen, The
Netherlands. (Abstract).
KLEEMANN, G. 1990. Aportes del Proyecto ICA-GTZ a la investigaci6n

en sistemas de producci6n. In Seminario-Taller sobre Sistemas de
Producci6n Pecuarios. Memorias. Tibaitata, Col. ICA-CIID. Feb. 8-9,
1990.
KLEEMANN, G.; LOBO, C.A.; SERRANO, G.; HANKS, J. 1991. MONTY,

a program for monitoring of cattle farms. Bogota, Col. Document
Proyecto Colombo-Aleman ICA-GTZ. 8 p.
LEZZACA, M.G.; KLEEMANN, G.; LOBO, C.A. 1991. Development of

monitoring schemes for cattle and pig farms: Experiences in Colombia.
Bogota, Col. Document Proyecto Colombo-Aleman ICA-GTZ. 7 p.
LOBO, C.A.; KLEEMANN, G. 1991. A model for continued training in

integrated technical assistance for livestock development. World
Veterinary Congress. (24., 1991, Rio de Janeiro, Bra). Abstracts.
LOPEZ, C.A. 1991. Sistemas de producci6n pecuarios en Colombia.

Bibliografia. Bogota, Col. Instituto Colombiano Agropecuario, Proyecto
Colombo-Aleman ICA-GTZ. 90 p.
MALDONADO, C.; BITSCH, R.; DOMS, U.; HAHN, H.; MEJIA, G.; PREUSS,
S.; SCHUCHT, M. 1987. Sistemas de producci6n agropecuarios en
dos zonas del sur del Ecuador. Schriftenreihe des Fachbereichs No.
103, Seminar fur LandwirtschaflicheEntwicklung, Technische Universitat
Berlin. 351 p.
NAVARRETE, M.G. 1988. The development of an information system for

an advisory scheme on cattle health and productivity in the humid
tropics of Colombia. Reading, U.K. University of Reading, Ph.D.
Thesis. 221 p.
OTTE, M.J.; OTTE, E.; NAVARRETE, M.; SANCHEZ, J. 1989. Monitoreo

de la salud y productividad animal en Colombia. Bogota, Col.
Proyecto Colombo-Aleman ICA-GTZ, Informe Tecnico No. 5. 87 p.
PETERS, K.; DEICHERT, G.; DREWES, E.; FICHTNER, G.; MOLL, S.;
CHAVARRIA, F.; DIAKITE, B. 1979. Goat production in low income
economic units of selected areas in West-Malaysia. Technische
Universitat Berlin. Reihe: Studien Nr. IV/27, Seminar fur
Landwirtschaftliche Entwicklung. 179 p.
REAPER (RESEAU EURO-AFRICAIN SUR LE PETIT RUMINANTS). 1991.

Research on small ruminants in Africa. A contribution towards strategy.
GTZ, Dpt. 422, Eschborn, Federal Republic of Germany. 10 p.
REULE, M. 1989. Produccion caprina en Colombia. Parte II: Produccion

caprina tradicional en Santander. Bogota, Col. ICA-GTZ, Proyecto
Colombo-Aleman. Informe Tecnico No. 6. p. 19-113.
REYES, E. 1991. An illustration of the use of a computerized recording

system in monitoring financial performance of cattle farms in Colombia.
Ottawa, Can. International Symposium on Veterinary Epidemiology and
Economics. Proceedings. 12-16 August 1991.
RODRIGUEZ, R.; MOLANO, G.; KLEEMANN, G.; GONZALEZ, C.A. 1991.

Evaluacion de sistemas de produccion y alimentacibn bovina (Quindio,
Colombia). Bogota, Col. Proyecto Colombo-aleman ICA-GTZ. Informe
Tecnico No. 11. 106 p.
SCHELLENBERG, R.; WENIGER, J.H. 1985. Sistemas de produccion de

leche y carne en fincas ganaderas en la costa Atlantica de Colombia.
Eschborn, Federal Republic of Germany. Proyecto PROMEGAN.
Deutsche Gesellschaft fur Technische Zusammenarbeit GTZ. Informe
Tecnico No. 5. 218 p.
STEINER, K.G.; WERNER, J. 1988. On farm research in landlichen

Entwicklungsprojekten. Entwicklung + landlicher raum, 22. Jahrgang
Heft 3/88.
VALLE-ZARATE, A.; HEYMELL, G-F.; JEEBE, M.; LENGEFELD, K.;

SANDHAGEN, H.; SZYSKA, M. 1987. Condiciones actuales y potencial
de la produccion porcina para mejorar la situacion del pequeno
productor en la provincia Gran Chaco-Bolivia. Technische Universitat
Berlin. Schriftenreihe des Fachbereichs Nr. 95, Seminar fur
Landwirtschaftliche Entwicklung. 261 p.
VELASCO, C. 1989. Caracterizacion epidemiologica de explotaciones

porcinas del Valle del Cauca. Bogota, Col. Facultad de Medicina
Veterinaria y Zootecnia, Universidad Nacional. Trabajo de Grado.
VERA R.R.; SERE, C. 1985. Sistemas de produccion pecuaria extensiva.

Brasil, Colombia, Venezuela. Informe final proyecto ETES. Cali, Col.
CIAT.
VILLAMIL, L.C. 1986. The application of information technology in the

development of livestock services in Colombia. Reading, U.K.,
University of Reading. Ph.D. Thesis. 187 p.
VILLAMIL, L.C. 1988. Information technology in the development of

livestock services in Colombia. Bogota, Col. Colombo/German Project
on Animal Diseases Control, ICA-GTZ. 97 p.
WENIGER, J.H. 1988. Agricultural research for the tropics and subtropics:
Current projects of research institutes in the Federal Republic of
Germany. Berlin, ICT-International Cooperation and Transfer GmbH,
German Council for Tropical and Subtropical Agricultural Research
(ATSAF). 741 p.
WENIGER, J.H.; MIESCKE, B. 1981. Aktualisierung and Orientierung der

Forschungaktivitaten wissenschaflicher Institutionen in der
Bundesrepublik Deutschland uber Fragen der Tierproduktion am
tropischen and subtropischen Standort. DSE Bericht, Feldafing, Federal
Republic of Germany.
IN THE TROPICS: A REVIEW OF
DUTCH EXPERIENCES
Johannes B. Schiere''2
INTRODUCTION
Animal production systems research (APSR) has no clear definition and

includes many activities. Some APSR can be considered as a special
branch of farming systems research (FSR), which itself has many
definitions (Simmonds 1986). Many FSR projects are really crop systems
research (CSR) rather than FSR, which, in its purest form, should
encompass the entire agricultural enterprise, including cropping, livestock,
fisheries, forestry, and even off-farm activities and consumption.
Sometimes APSR is added as a component of CSR or vice versa. The
variety of projects that can claim to have APSR components epitomizes the
imprecision of the definitions for APSR. This paper will use the term
APSR/FSR for the aggregate of livestock and farming systems research
work. Although APSR/FSR may seem a relatively recent activity for the
research or donor community, the concept or components of it are not
new.
In Dutch literature around 1860, Staring distinguished landbouwstelsels,

or systems of farming, based on differences in soil and farm type, as
opposed to purely geographical criteria. That work was reviewed by
Veldink (1970), who also described farmer-led, on-farm research during the
last century. The indirect application of Staring's concept in the case of
y
Department of Tropical Animal Production, Agricultural University Wageningen, P.O. Box
338, 6700 Wageningen, The Netherlands.
2 This paper could not have been written without the interaction with numerous unnamed
farmers, farm workers, students, scientists, teachers, and extension counterparts from all
over the world. Special thanks are due to many colleagues from The Netherlands for
supplying and checking much of the information in this paper in this paper.
dairy development in The Netherlands is that applied research and

demonstration is done at decentralized regional research stations in five
distinct agroecological zones in addition to work on central locations
(Osinga 1990). The Netherlands Agricultural Economic Institute (LEI) and
the European Economic Community (EEC) categorize agricultural
enterprises according to types and size of production (EEC 1989). That
classification was originally designed before the Second World War in
mutual discussions between farmers, research, and extension workers, i.e.,
a rapid rural appraisal (RRA) avant-la-lettre.
Agriculturists in the former Dutch colonies recognized the need for FSR
well before World War 11, as reviewed by Fresco (1986). APSR was done
by government veterinarians like Kok (1921), Aalfs (1934), Hoekstra (1948),
Huitema (1982) and Merkens (1982), in a descriptive but not necessarily
participatory form.
The early Dutch interest in various aspects of FSR persists (Fresco 1986;
Fresco and Wesphal 1988; Vereijken 1992; Stroosnijder and Rheenen 1993),
with attention to APSR by Korver and Arendonk (1988), Osinga (1990) and
Roeleveld (1995). Continued interest in APSR/FSR could indicate the
relevance of the core concepts. More realistically, it is also due to the fact
that the term farming systems is a convenient umbrella under which to
justify a host of activities. Shifts of emphasis are natural and likely to
occur. The increased emphasis on sustainability of livestock production
represents such a shift of attention.
Livestock may not be the only cause of environmental problems, but

they certainly play a part (Durning and Brough 1991). Problems of sustain-
ability are manifested as pollution in high-input systems (Korevaar and Boer
1990; Aarts et al. 1992) or mining of soils and vegetation in low-input
systems (fireman 1990; Stoorvogel and Smaling 1990; Pol 1992). Recent
work such as that from Conway and Barbier (1990) underlines the need for
a systems approach for work on sustainability.
The preparation of the present document was difficult, given that:
definitions of APSR and FRS are still unclear, which means that working
definitions not caught up in semantics are still required
most of the APSR-related experience is poorly reported because it is

done by people with an interest in the practical application of the results
rather than in scientific publications
most of the work is done with counterparts from all over the world; to
claim these experiences as "Dutch" is presumptuous, to say the least
the classification of systems work into tropical and temperate categories

is useful but difficult to maintain strictly (Schiere and Wit 1993).
Therefore, this paper also makes reference to some temperate climate
experiences, from The Netherlands, that are relevant for the tropics.
This article reviews only issues that are specific for APSR; issues that are
general for FSR are not discussed. Gender issues in livestock develop-
ment are important but will be dealt with by Leesberg (1995) in the present
Proceedings of the Global Workshop on Animal Production Systems. In
this same publication, a case study of Dutch APSR/FSR in the tropics is
discussed by Roeleveld (1995). Dutch involvement in participatory work
is best reviewed in ILEIA (1991).
A WORKING DEFINITION OF APSR/FSR
Theoretical discussion of definitions and components of APSR/FSR has

been left to others, but a working description of APSR/FSR elements is
given for purposes of the discussion in this paper (Table 1). The compo-
nents are ill-defined, as are their scale and level of analysis: for example,
where does the holistic approach stop? at the farm gate or at a
macroeconomic level? in the short term or in the long term? (Fresco and
Westphal 1988).
The use of surveys, on-farm research demonstration units, and other

methods associated with FSR was known before formalized FSR methodo-
logies came into being. Many development projects had APSR/FSR
components avant-la-lettre. Even "a good look around by an experienced
observer" can be considered as being an RRA. The isolated application of
APSR/FSR components may have introduced errors when they did not
follow the logical sequence or subjected to systems analysis; however, it
must also be recognized that perfection was not achieved after the
introduction of formalized APSR/FSR.
Simmonds (1986) refers to A and B as the deep analysis of farming

systems as they exist (essentially an academic activity) and which this
paper considers as upstream activities that, together with disciplinary
information on issues such as feeding standards, soil analysis and
veterinary sciences, are indispensable for the following phases. The work
of Staring fits best in C1, but could not have been done without A and B.
The work of the colonial veterinarians, mentioned in the Introduction, fits
best in A and C1. The stages C1 through C5 are a personal interpretation
of farming systems research and extension (FSR/E) (Knipscheer and De
Boer, personal communication).
Table 1. A working definition of the operational phases of APSR/FSR

(see text for explanation)
A Compilation of basic information and data concerning field situations
B Development of systems research methodologies (and software)
C1 Farming systems description and constraint identification (includes

surveys, rapid rural appraisal, etc.)
C2 Technology identification and design of research programs; relevant

component research
C2/3 Modeling and design of farming systems (or farming system

synthesis).
C3 Technology testing on-station and in pilot units
C4 Technology testing on-farm with emphasis on farmers participation
C5 Demonstration and extension of results and monitoring
In Simmonds' concept, the phases C1 -C5 would be called OFR/FSP (on-

farm research with farming systems perspective), a practical adjunct to
research which seeks to test the socioeconomic suitability of research
ideas on farm before recommending extension. The distinct phase C2/3
illustrates the struggle with formal definitions that cannot precisely cover
all concepts, and it stresses that this phase is essentially a part of all
phases. Simmonds would call C2/3 the NFSD (new farming systems
development), which seeks to develop complex radical changes rather

than the stepwise change characteristics of OFR/FSP. The NFSD concept
is.probably defined as what the Dutch call farming system synthesis. It is
probably 'old hat' for economists, but a new experience for discipline-
oriented agricultural scientists.
A REVIEW OF DUTCH APSR/FSR ACTIVITIES
The following listing is a sample of Dutch activities concerning

APSR/FSR, organized in alphabetical order and with no pretension of
being comprehensive. The acronyms stand for projects as well as
institutions and are used throughout the text as easy reference, and
without further specification.
Table 2 includes a few additional projects (with their references) and

indicates roughly which APSR/FSR elements (or phases) are part of the
projects. It is evident that not all projects involve all phases. Omission of
essential steps such as C1 or C2/3 may lead to errors in C3, C4 and C5;
also, insufficient information from A or B leads to inefficient operation of
C1, and so on. This is one reason why many projects start to work on one
phase and end up doing another. For example, many development
projects end up doing elaborate surveys without reaching their goal of
actual intervention.
BIOCON. The joint Indo-Dutch project started with research on the

improvement of straw quality for animal feed. Subsequently it included
work on farming systems characterization, particularly of cropping and
livestock systems, as a means to determine the relevance of extension
messages and research programs. A follow-up phase will include more
APSR/FSR components with emphasis on the use of crop residues in
livestock feeding. The project includes multidisciplinary teams of animal
nutritionists, economists and extension specialists located at government
as well as NGO centers. Models with standard software and on-station
pilot units are essential components to guide both research and extension.
Information on transferable technologies will be made available with
specification of their suitability for each farming system (Kiran Singh and
Schiere 1991).
G)
Table 2. A schematic presentation of some Dutch APSR/FSR projects with an approximate indication of the G1
N
methodological phases involved 1,2,3
Melhodo ogical phase (see Table 1)
Project A B C1 C2 C2/3 C3 C4 C5 References
*BIOCON + + + + + + ++ + + + + + Kiran Singh and Schiere (1993)

*CABO + + ++ + + + + Wolf et al. (1991)
*COLIBRI ++ + ++ + + ++ + + + + Perez et al. (1989)
Baayen et al. (1991)
*DTAP/SOW + + + + + + Udo and Brouwer (1991)
*ENVIRONMENT + + + + + ++ ++ + + + + + + + Korevaar and Boer (1990)
Aarts et al. (1992)
*EXTINT + + + + ++ ++ Schiere (1992)
*INRES + ++ + ++ + + + + + + ++ Stroosnijder and Rheenen (1993)
*KIT + ++++ ++++ ++++ ++++ ++ Roeleveld (1995)
*NDDP + + ++ +++ +++ ++ ++ ++++ Voskuil (1989)
PASTO + + + + + ++ + + + + + + + + + + ++ Koeslag and Urbina (1985)
Engel (1990)
*PRODERM + ++ + + + ++ ++ + + + + ++ Kaasschieter (1990)
Leeuwen (1991)
SLNLDP-SUP/FFUP + + + ++ + + + + + + ++ Ibrahim and Schiere (1986)
SLNLDP-SHDDP + + + + + + + ++++ + + + + ++ Jong et al. (1992)
*SUDAN + ++++ Struif Bontkes (1991)
*WADG ++ ++ + + ++ + + + + + + ++ + Smith and Bosman (1988)
RLIP +++ +++ +++ +++ + ++ ++ ++ YAR (1989)
Based on discussions with R. De Jong and A.J. Nell.

2 The projects marked with an asterisk (*) are elaborated in the text.
3 The number of ++++ indicates the level of involvement of that phase.
CABO. The CABO activities with APSR/FSR phases cover many

activities, such as those listed below:
comparisons of carrying capacity, desired production levels and required

input levels (Breman 1990)
use of simulation models for pasture, crops, and animal production

(Penning de Vries and Djiteye 1982; Alberda 1984)
use of linear programming models that look at agricultural and

sustainability problems (which can be severe both in temperate and in
tropical production systems), as well as at problems of competition
between agricultural sectors (Duivenbooden et al. 1991)
preparation of manuals for evaluation of pastures and livestock in the

Sahel
identification of potentials and bottlenecks in rural development and

formulation of methodologies, options, and strategies for development
purposes (Ayyan and Keulen 1987)
animal production studies for Niger, Burkina Faso, and Mali (Breman et
al. 1990; Keulen and Breman 1990)
conduction of a recent project (Project Production Soudano-Sahelienne:

exploitation optimale des elements nutritifs en elevage). This work aims
to reduce the "trial and error" from FSR and development programs. An
overview of the developed methodologies is given in Wolf et al. (1991).
COLIBRI. This is a Costa Rican project with a veterinary orientation; it

collects information concerning animal and pasture production, health and
related factors for farm management, extension, training, research
programs and provides feedback to the farmer. It is a multidisciplinary and
decentralized livestock information system; system validation takes place
with intensive monitoring of dual-purpose and beef cattle, dairy cattle, and
swine. Farm data are collected regularly on personal computers mostly
through privately run stations. Single registration and entry of data, low
hardware requirements and privatization are important elements in the
development of COLIBRI, which is now playing an increasing role in the
regional standardization of livestock data collection and processing in
Central America (Perez et al. 1989; Baayen et al. 1991).
DTAP/SOW. The initial aim of DTAP/SOW was the development of a

mainframe simulation model for Southeast Asian livestock systems, along
the lines of the Texas A&M models (Sanders and Cartwright 1979, quoted
by Udo and Brouwer 1991). The attempted precision was frustrated by the
lack of biological information and the complex use of the program.
Nevertheless, indirectly the activities helped to identify research needs.
Spin-offs of this effort were the following:
a data base to record and interpret animal production parameters
of the data base and of modules on

a package for statistical analysis
herd composition, feed intake and feed allocation
an experimental study on compensatory intake and growth regulation
Recently, user-friendly PC-simulation models of animal production as a

function of feed quality and quantity have been developed (Udo and
Brouwer 1991; Zemmelink et al. 1992).
EXTINT. This is an undertaking at the author's department, where the

crop-livestock interactions are studied mainly from literature sources
supported by project experiences. The emphasis is on definition and
design of sustainable farming systems, taking into consideration the
changing functions of livestock in farming systems due to decreasing land
and associated biomass availability (Schiere 1992). Indirectly related to
this work is the involvement of the Department with FAO activities on
sustainable agriculture (Kaasschieter et al. 1992).
INRES. The goals of INRES include the build up of an interdisciplinary

research capacity, in relation to the quantitative analyses of agricultural
farm-units and directed to sustainable agricultural development in limestone
areas of East Java. The emphasis is on quantitative work (Stroosnijder and
Rheenen 1993). The hypothesis is that present agricultural farm-units in
the limestone areas of East Java are not sustainable, primarily because of
decreasing soil fertility.
KIT. Dutch APSR/FSR with formalized methodological phases C1

through C5 (Table 1) are few. Except for the BIOCON project (Kiran Singh
and Schiere 1991) most are in Africa and are coordinated by the Royal
Tropical Institute (KIT). They are described as farming systems research
and development (FSR/D) and are situated in countries where institutional
setups for FSR are well known (i.e., Mali, Benin, Zambia and Tanzania).
Livestock research in all four programs is carried out in the context of
mixed farming systems where animal production is secondary to, though

closely integrated with crop production. In three out of four projects
animal traction plays a crucial role in agricultural development (Roeleveld
1995).
ENVIRONMENT. This category covers a variety of activities in The

Netherlands that are concerned with the reduction of the environmental
damage inherent in high-input systems. This work is not based on formal
FSR concepts dealing with on-farm problems (phases C1 through C5,
Table 1), but rather it originated from an analysis of macroenviron mental
needs.
The seemingly wide divergence between tropical and temperate

problems can lead to unique converging concerns; for example, the same
research which seek the reduction of N losses does so for different
purposes: in Europe, to spare the environment, while in Africa, to spare the
farmers' economy (S. Schukking, personal communication 1991). A
national model farm (in this paper called a pilot unit, phase C3) is
established based on a farming systems design (phase C2/3) (Korevaar
and Boer 1990; Aarts et al. 1992). Similar work is being done by Hermans
and Vereyken (1992), where aspects of nature conservation are included
in the farm synthesis.
NDDP. The National Dairy Development Project (NDDP), under the

Ministry of Livestock Development in Nairobi, restricts itself to the high-
potential areas of many districts all over Kenya. The major objectives of
the project are listed below:
introduction and promotion of the zero-grazing system among mixed

small-scale farms
collection and analysis of data regarding technical and socioeconomic

aspects of the dairy enterprise within existing farming systems at the
farm level
creation and maintenance of a two-way flow of information between

research and extension regarding intensive dairy production within the
relevant farming systems (Voskuil 1989).
PRODERM. Some development-oriented livestock projects with APSR

operational phases were cofunded by the Dutch Government in Latin
America. An example is the PRODERM project which started as a "credit
program" for crop and livestock production. Later on, the focus changed
to improvement of production conditions (drainage and irrigation), training,
and community development. The livestock-development program aimed
to improve the productivity of cattle, sheep, and alpaca at a small farmer
level. Applied research was done for improvement and redefinition of the
program on issues like those listed below:
land ownership and use patterns for cropping and livestock
animal feed and natural vegetation resources
livestock ownership patterns
fertility parameters and marketing of anirnals
credit for small farmers
economic effects of livestock development programs on dairying, fatte-

ning, and wool production by sheep and alpaca
The project's goals to increase physical output of products such as milk

and meat was not achieved, but from the farmers' point of view this
program was a success because an investment in animals under
conditions of high inflation was quite useful. Repayment of credit was
satisfactory. The problem is that the objectives of a program may differ for
the project and for the target group. A goal such as increasing saving
accounts (many animals at low cost) may even conflict with the production
goal of milk, meat or improvement of the environment. Improvement of
production conditions such as irrigation, drainage, and disease control
appeared to be the first steps to improvement of livestock production
(Kaasschieter 1990; Leeuwen 1991).
SUDAN. Based on data and experiences of a Sudan-Dutch rural

development project in southern Sudan (which originated with a purely
developmental approach), a comprehensive systems analysis of the rural
area of the Bor District was carried out. The aim was to develop an
understanding of (1) the dynamics of the system over a period of 20 years
and (2) the possible effects of a number of interventions (such as
veterinary services, medical services, agricultural innovations, and
infrastructure) on population size, quality of life, and animal and crop
production. Modeling proved helpful in obtaining a consistent picture of
a system and its dynamics and for discussion of the effects of inter-
ventions. Some interventions proved to be disastrous after a number of
years, and processes which were disregarded in periods of stability
appeared to become important when the system became unstable. Due
to lack of data, it was difficult to validate the model properly. The model
should therefore not be considered as a tool for prediction but rather as
a tool for a better understanding of the system and for identifying research
questions that are relevant to the solution of problems in the area (Struif
Bontkes 1991).
WADG. The first phase of the WADG project on Management of the

West African Dwarf Goat in the Humid Tropics (1981-1987), was aimed to
create upstream research possibilities and on-station research (especially
nutrition, housing, and management). Much was already known about the
research priorities, the socioeconomic setting of goat production, and
constraints and possible solutions through earlier work at ILCA. Only a few
pilot farms were established during the end of the first phase. Upon the
insistence of the Dutch donor, during the second phase (1988-1992)
emphasis shifted to on-farm recording of data relevant to the farming
system as a whole (including economic data) and detailed data on the
goat component in the system. A package with several components in the
field of nutrition, disease control, and housing and management was
offered to the farmers, who could adopt components and modify them to
their own needs. This approach has some advantages; for example, it
gives good data on adoption rate and especially on the modifications
apparently required for the different groups of households. However, data
are not collected according to a neat experimental design, so a statistical
analysis of the effects of adopted changes is problematic (Smith and
Bosman 1988)
SPECIAL APSR ISSUES
The topics discussed below are limited to issues that are specific for the
livestock sector in APSR/FSR and follow the sequence of the operational
phases shown in Table 1. The difficulty of poor reporting of informal field
APSR/FSR experiences is overcome to some extent by informal interviews
with colleagues and the inclusion of much "grey" literature. Wherever
possible, reference is given to the most recent literature on the subject.
General aspects of institution building, FSR methodology, data collection,

choice of software and the intricacies of interdisciplinary work are delibera-
tely excluded. Many of those aspects are not typical of APSR and are
discussed in other publications (Fresco 1986; Collinson 1987). Recent
reviews on typical APSR problems may be found in Nordblom et al. (1985)
and Udo and Brouwer (1991).
THE SEARCH FOR BASIC DATA AND THE STUDY OF METHODOLOGY
Data collection and methodology development are easier in homogenous

high-input systems than in heterogenous low-input systems. Except for
industrial-type pig and poultry production and for areas with good milk
marketing, livestock is likely to be concentrated on marginal and variable
lands. The literature and this paper focus mainly on large ruminants. Not
much is known on the role of backyard animals (pigs, poultry, rabbits,
guinea pigs), although some studies have been done by Hendrikx (1985),
Veluw (1987), Boering (1987), Oegema (1989), Kuit (1990), and Verschuur
(1991). This is a problem, since yard animals fulfill a role in family nutrition
and provide extra income for women.
Special problems for collection of livestock data include those listed

below:
Measurement of feed intake is difficult because feed can be transported

between systems and closed systems are rare. Furthermore, feed intake
of animals from range or stall feeding is difficult to estimate, partly
because of selective consumption. This means that the main feed inputs
can hardly be qualified or quantified.
Measurement of herd dynamics is difficult due to the fluidity of the herds

where livestock is considered the equivalent of a savings account or
because small herds of often-landless peasants are not representative of
the average herd in the region. A village bull takes care of haphazard
breeding, while calves for dairy or bulls for draft may be sold to or
bought from other areas. To consider the whole village as a single herd
creates problems of aggregation due to divergent socioeconomic
classes of livestock owners and production purposes of the individual
animals.
Livestock can, within limits, adjust their metabolism to the type and level
of feeding, making it difficult to set a specific production level as an
objective in optimization programs. High individual production levels are
not synonymous with high yie:ds per unit area or farm income (Schiere
and Wit 1993).
Livestock production is often multipurpose (e.g., milk, meat, draft, status,

savings account, or a hedge against inflation). The value of many of
these products is difficult to assess. Livestock cannot be seen in
isolation from cropping, although in many situations it seems that
cropping decisions are given primary importance by farmers. The
purpose of production, and hence its value, may differ between and
within socioeconomic groups of farmers, making it difficult to put proper
values on production and costs for inputs.
The measurement of changes over time is also a problem. Such

changes may be caused by market forces and by changing infrastructure
and farmers' knowledge, but are increasingly due to a decreasing
resource base (Schiere 1992). The results of thorough studies are likely
to be outdated by the time they are published, often too late for the
original objectives of the project. Long-term effects are more important
in APSR than in CSR/FSR because of long generation intervals and
carry-over effects between seasons.
Development of methodology, basic data collection, and improved

production functions in phases A and B (Table 1) are essential to diminish
the trial and error nature of phases C1 through C5. Especially, the applied
nature of projects which center on phases C1-C5 allow little time to
increase precision of prediction and must rely on projects concentrating
on phases A and B, as well as on pure on-station research, for better
information. The time required for data processing and reporting is
generally grossly underestimated and especially the applied or
development-oriented projects resort to "best bets" or "guestimates" rather
than elaborate data bases. The dynamic character of farming systems is
often not captured sufficiently by static surveys. The variability and
unreliability of many data bases prevent their use as supports of common
sense in practical field work. Sensitivity analysis with "guestimates" can
yield information as useful as the detailed information from elaborate
surveys.
User-friendly software for such purposes in APSR is increasingly available

to supplement versatile standard packages such as Lotus, LP88 and
DBase. Some of these were developed under DTAP/SOW and originated
from the frustration of attempts to make large universal models which

required detailed information (Udo and Brouwer 1991). COLIBRI offers a
range of possibilities for management programs, and more complicated
programs such as multigoal linear programming are already in practice,
together with dynamic models that predict primary production, such as
CABO.
Unfortunately, the use of models suffers from either one of two maladies:
Agreat reluctance to accept modeling as an aid to common sense.

Many are afraid to use computers or see modeling as the "alchemists'
illusion."
A lack of interpretation of the results. Too many people take the

computer-generated and operator-inspired printouts as if they were the
ultimate truth from holy scripture.
With proper interpretation, and as an aid to common sense, modeling

can be useful in the training of extension workers to understand the conse-
quences of innovations, the design of extension messages, and in the
determination of research priorities. Modeling may aid, but can never
replace, common sense and interaction with the target group. Experience
in BIOCON is that simple models, as a common platform of discussion,
assist in getting extension and research workers to interact (Neil 1986; Wit
et al. 1993).
Description of farming systems and identification of constraints
Depending on the purpose of the classification and on local conditions,

farming systems can be characterized by socioecological, physical,
climatological, and/or topographical criteria, leading to transects such as
illustrated by Chambers et al. (1989) and Conway and Barbier (1990).
Examples of Dutch work of this type are the soil classification studies by
Stomph et al. (1994).
Labeling the applicability of extension messages or research priorities as

tropical or temperate is too rigid and superficial (Schiere and Wit 1993).
This does not imply that "temperate" technology and concepts can solve
"tropical" problems (after partly creating them!), but increased interaction
can be mutually beneficial. ENVIRONMENT shows that high- and low-input
systems have a shared interest in plugging the leaks in systems analysis,
although for different motives. Many characteristics of the low-input
farming found in Europe in past centuries are also found in modern-day

tropical countries, though the conditions for European and North American
agricultural development in the 19th century were different than conditions
for the development of tropical agriculture today.
A paper on livestock policy, by the Dutch Government (DGIS 1990), uses

four ecological zones (highlands, humid, semiarid densely populated, and
semiarid sparsely populated) superimposed on a classification of pastoral,
mixed, and specialized systems. While preparing that document, it was
found that population pressure was an was overriding factor and was often
correlated with soil quality, transport facilities, and nearness to markets
(i.e., availability and demand patterns) as discussed below.
For planning agricultural, development, the best criteria for classification

of farming systems are probably based on resource availability. This is
implied by Chambers et al. (1989) when referring to industrial, Green
Revolution, and resource-poor farming or, in more detailed fashion, for
livestock in the work of economists by Crotty (1980). Schiere (1993)
explains how applicability of extension messages depends on specific
resource combinations for each farming system.
Technology identification and design of research programs
Most of the work on participatory research is unreported. The author

himself has had experience with grassland development in Peru, rabbits in
Indonesia and straw feeding techniques in Sri Lanka, but they are only a
small part of innumerable, unreported field experiences.
Economic modeling and sensitivity analysis can determine which farming

systems are most likely to benefit from an innovation (Nell 1986; Kiran
Singh and Schiere 1991). Such a process is a typical example of
technology-driven extension with its inherent risks (Ruling 1989), but is
defendable provided it is followed up with a need analysis (Collinson 1987).
This modeling and data collection can even lead to a reversal of the
procedure (Wit et al. 1993) as experienced in the COLIBRI and KIT
projects. BIOCON and KIT also aim to develop a "handbook of
transferable technologies" (or "menu of options") which will indicate which
technology in the field of animal nutrition might be useful for which farming
system, with possible side effects. Publications such as this differ from
scientific publications, because of the target group, whose interest is not
in elaborate scientific references, but in the "how-to" and where to apply
the procedure.
Modeling and design of farming systems

(or farming systems synthesis)
Farming systems synthesis is a form of modeling that decides priorities

for research and extension (Simmonds 1986) or that is required for the
design of new farming systems as done by Schiere (1992) and Vereijken
(1992), where systems are subject to drastic changes. The process of
modeling can be called management support (Huirne 1990), and can be
divided into three types of decisions:
strategic decisions that consider the design and concept of the entire
farm in the long term
tactical decisions, where medium-term decisions (such as cropping

patterns) are made
operational decisions, that accept the farm as given but study the effect
of day-to-day decisions.
It isrelevant to note here the difference between the Francophone and

the Anglophone approach to FSR (Fresco 1986), where the first is
concerned with drastic changes of system (a decision of the strategic
type), and the second more with incremental change (tactical or
operational decisions). The incremental type of change requires
techniques such as gross margin analysis; drastic change will generally
depend more on techniques such as linear programming.
Synthesis of new systems is difficult due to the multitude of possible

alternatives. It requires multidisciplinary work and reconsideration of
preconceived ideas such as the need for livestock, the need of high
individual production, the concept that livestock is the only form of capital
(not recognizing that land or trees might also be used in that manner), and
the idea that only milk or meat, (and not maintenance or draft) are the only
valid production outputs from animals.
On-station technology testing
The problems of on-farm research with livestock, as compared with on-

farm research with crops, have been reviewed by Gryseels (1989). They
include the indivisibility of animal herds as compared with a plot of crops,
the difficulty of measuring feed intake in-stable and on-range, the mobility
of the animals, and the emotional values attached to animals.
For livestock development, the establishment of pilot units (with

prototype systems) on-station is required to avoid failures in the field,
though much depends on whether the introduced technology is a
vaccination or a new feeding practice. Work with pilot units is described
by Osinga (1990), Jong and Meinderts (1992) and Jong et al. (1992); and
information is also available from the NDDP, WADG, and NDDP-Kenya
projects (see Table 2). Work with pilot units does not exclude farmer
participation. The PASTO project has published little, but accumulated
much experience on every methodological phase, from C1 to C5, from
work with pilot units to the use of model farmers (Jong and Engel,
personal communication, 1991). Engel (1990) described aspects of on-
farm work and the research-extension linkage.
The final objective of a pilot unit may be that the farmer eventually
adopts the concept with his or her own modifications. But the purpose of
the unit is not lost when the farmer takes only certain aspects of a set of
recommendations, such as better dung collection, a better feeder design,
or other roof material. Pilot units can give only a very rough estimate on
labor needs, farm economy, etc., because at best they approach, but never
equate, actual conditions and the system variability is too large to study
with only a few, if any, replicates. Pilot units are not meant to be static
models, but should be constantly modified to include suggestions from
farmers, extensionists, and scientists.
On-farm technology testing
Much of what has been said about on-station testing also applies to on-
farm testing. This phase cannot be discussed without considering the type
of message that is being delivered (Schiere 1993). A vaccination needs
much less demonstration than a new way of growing fodder or processing
milk. In Sri Lanka the urea-ammonia treatment of straw was tested
extensively on-farm with the presence of both farmer and extensionist; the
"problem" was that farmers asked about the milk price and the artificial
insemination service, which implied that either the message did not appeal
to them, that this message would only benefit them if other problems were
also solved, that farmers pretended to accept the message as a way to
obtain additional favors, or that the farmers were satisfied with the method,
but thought it was wore important to continue with other problems.
Besides choosing farmers with the proper position in the community, the
message to be demonstrated or farm-tested should be in harmony with the
supply and demand conditions of the farming system in question.
Extension of results
APSR has not generated the spectacular results of the Green Revolution.
This does not mean that results are not there (Schiere 1993). The problem
is that much extension effort is wasted because of ill-conceived
government programs which try to cover the whole country with messages
that are highly farming-system specific in nature, but have not been
developed on the basis of APSR/FSR (Schiere 1993). While a lack of
APSR/FSR is more likely to cause failure than too much APSR/FSR,
elaborate APSR/FSR takes too much time for impatient donors and
recipient governments. The extensionist should be prepared to accept
unexpected impacts; for example, a project has not failed when the
farmers obtain a higher price for a crossbred than for the milk (R.K. Patel,
personal communication 1991), and a project aimed at the treatment of
straw should consider it an achievement if people stop burning straw and
consider it as feed more than they did previously. When the problems are
complex, APSR/FSR may not provide answers because there are no
simple solutions to the problems.
Some examples of research results that eventually reach the farmers are
given by Schiere (1993) and include (whether or not based on explicit use
of APSR/FSR):
the use of biogas, straw treatment, and dairy housing in Sri Lanka
(subsidized) (SLNLDP)
reduction of NH3 emissions in Dutch livestock production

(ENVIRONMENT)
improvements in the draft animal situations (KIT)
dairy producers cooperative (PASTO)
dairy development with zero grazing (NDDP).
DONOR FATIGUE?
Donor fatigue with APSR/FSR per se is not very obvious in The

Netherlands. The new livestock policy document stresses the need for
more systems work, though this does not directly result in increased
funding. Donor fatigue with livestock projects, or agriculture in general, is

a fact that needs to be accepted for the time being. Other areas of
interest, such as environment, tend to receive more attention (Pronk 1990).
This is partly caused by ill-; onceived programs and exaggerated
expectations in the past, based on insufficient understanding of livestock
production systems (DGIS 1987). Also, it is partly due to the uphill battle
of APSR/FSR to help provide food for an ever-increasing population in
conditions of declining resources.
Maybe it can be said that funding has returned to a more appropriate

level as many projects were being overfunded. Many projects were
euphorically, rather than realistically, conceived. On the other hand,
APSR/FSR must raise expectations if it is to receive continued funding. By
virtue of its systems orientation, APSR/FSR has to make clear that:
Not all results can or should be measured in direct biological or

economic parameters.
In view of the variability of farming systems (Conway and Barbier 1990),

no blanket solution can be found. Regional solutions will be the rule
rather than the exception and the best that one can aim for. This is
likely to have implications for institutional setups in terms of required
flexibility.
Reduction rather than reversal of a downward trend may be an

achievement in marginal areas.
The lessons from past livestock projects supported by The Netherlands

Government were synthesized in a new policy document (DGIS 1990).
This document recognizes that different systems need different policies.
It actually tries to formulate policies for each ecological zone, while
considering livestock for rural development as an integral part of existing
farming patterns. A similar view is given in a recent World Bank report on
dairy development in the Sub Sahara (Walshe et al. 1991).
CONCLUSIONS
APSR/FSR is ill-defined and covers many activities. APSR/FSR is not

new on the Dutch scene, but only few livestock projects are known that
explicitly followed a formalized approach in the sense of FSR/D or FSR/E.
Nevertheless, many livestock projects traditionally incorporated certain

operational phases of the FSR approach into their activities, though
sometimes with a lack of essential elements. Future analyses of project
results will be helped by explaining which concepts are meant when the
term APSR/FSR is used in these projects.
A substantial amount of APSR/FSR work goes unreported, because field

workers are often development-oriented and, therefore, write few
publications. Scientifically oriented workers take too long to produce
results because the task of interpreting and publishing of data is generally
grossly underestimated.
Decreased funding for livestock, or agriculture in general, does not

necessarily mean donor fatigue for APSR. In fact, APSR is receiving
increased recognition; however it will also have to stand up to current
expectations. Under conditions of declining resource availability, this is not
an easy task. Livestock is essential in many rural economies though its
role is likely to change with time. The interactions between crops and
livestock, as well as attention to aspects of sustainability, demand a
systems approach.
Neither extension nor research can put their priorities right without a
better understanding of systems. A lesson from blueprint livestock projects
is that livestock development cannot be undertaken isolated from
environmental concern. Livestock will probably become a more integrated
part of future farming systems where crop residues will play an increasingly
important role as livestock feed.
The classification of livestock farming systems for development purposes

should focus on the resource availability. If well done, it will reflect the
availability of production factors, parallel with aspects of land classification,
climate and socioeconomic criteria. Differences between tropical and
temperate systems should not obscure common interests between the
systems. The farming systems and their resource availability are not static
over locations nor over time, and increasing stress is put on the
environment to produce more with less.
The change in systems requires additional research to supplement

farmers' knowledge, as well as good communication and listings of
available technologies, to avoid duplication. Recognition that resource
patterns change with time should force the APSR/FSR to take a dynamic
approach.
Modeling is an imperfect, but useful aid for planning extension work and
setting research priorities. The lack of field data and production factors
should discourage the development of sophisticated software except for
basic research. Simple but versatile standard software that allows for
sensitivity analysis can cope sufficiently with problems of development
projects.
There is a conspicuous lack of systematic work on the role of small

livestock, especially pigs and poultry. It contrasts with the importance
usually placed on the role of livestock to provide petty cash and nutrition
for the rural family.
LITERATURE CITED
AALFS, H.G. 1934. De rundveeteelt op het eland Bali (Animal Husbandry

on the island of Bali). Rijks, Universiteit Utrecht. Ph.D. Dissertation. 174
P.
AARTS, H.F.M.; BIEWINGA, E.E.; KEULEN, H. VAN. 1992. Dairy farming

systems based on efficient nutrient management. Netherlands Journal
of Agricultural Science 40:285-299.
ALBERDA, Th. 1984. Production and water use of several food and fodder
crops under irrigation in the desert area of Southwestern Peru.
Wageningen, The Netherlands, PUDOC, Agricultural Research Report
928. 50 p.
AYYAN, M.A.; KEULEN, H. VAN. 1987. Mariut- Project, Final Report Sub-
mitted to DGIS. CABO, Wageningen, The Netherlands. 80 p.
BAAYEN, M.; PEREZ, E.; MULLER, E.; CAPPELLA, E. 1991. Development

of a livestock information system in Costa Rica. Proceedings of the XXIV
World Veterinary Congress, Brazil.
BOERING, S. 1987. Cabras, ahorrar o producir, un estudio sobre el papel

de la cabras en sistemas de produccion campesina en Santander del
Sur. Wageningen, The Netherlands, Agricultural University, Department
of Developmental Economics/Bucaramanga, Col., SENA-CAAC. MSc
thesis. 25 p.
BREMAN, H. 1990. Integrating crops and livestock in southern Mali: rural

development or environmental degradation? In Theoretical production
ecology: Reflections and prospects. R. Rabbinge (Ed.). Wageningen,
The Netherlands, CABO-DLO. p. 277-294.
BREMAN, H.; KETELAARS, J.J.M.H.; TRAORE, N. 1990. Un remedie

contre le manque de terre? Bilan des elements nutritifs, production
primaire et elevage au Sahel. Secheresse 2: 109-117.
CHAMBERS, R.; PACEY, A; THRUPP, L.A. (Eds.). 1989. Farmer first:

Farmer innovation and agricultural research. London, U.K. Intermediate
Technology Publications. 218 p.
COLLINSON, M.P. 1987. Farming systems research: Procedures for

technology development. Experimental Agriculture 23:365-386.
CONWAY, G.R.; BARBIER, E.B. 1990. After the green revolution:

Sustainable agriculture for development. London, U.K., Earthscan
Publications. 205 p.
CROTTY, R.D. 1980. Cattle, economics and development. Slough,

England. Commonwealth Agricultural Bureaux. 253 p.
DUIVENBOODEN, N. VAN; GOSSEYE, P.A.; KEULEN, H. VAN (Eds.).

1991. Plant, livestock and fish production. Competing for limited
resources: The case of the Fifth Region of Mali. Wageningen, The
Netherlands, CABO-DLO/Mopti, Mali, ESPR. Report 2.
DGIS (DIRECTORATE GENERAL FOR INTERNATIONAL COOPERATION).

1987. Assisting Livestock Development: Evaluation Report. The Hague,
Netherlands Development Cooperation.
DGIS. 1990. Veehouderij in ontwikkelingslanden (Animal husbandry in

developing countries). The Hague, Netherlands Development
Cooperation.
DURNING, A.B.; BROUGH, H.B. 1991. Taking stock: Animal farming and
the environment. WorldWatch Paper 103 (ISBN. 1-878071-04-1).
EEC (EUROPEAN ECONOMIC COMMUNITY). 1989. Farm accountancy

data network, an A-Z of methodology. Office for Official Publications of
the European Communities, Brussels/Luxembourg. ISBN - 92-826-0096-
3
ENGEL, P. 1990. The impact of improved institutional coordination on

agricultural performance: The case of the Narino Highlands in Colombia.
The Hague, Netherlands. International Service for National Agricultural
Research (ISNAR). Linkages discussion paper No. 4. 23 p.
FRESCO, L.O. 1986. Cassava in shifting cultivation: A systems approach

to agricultural technology development in Africa. Amsterdam, The
Netherlands. Royal Tropical Institute, Floris Publications. 240 p.
FRESCO, L.O.; WESTPHAL, E. 1988. A hierarchical classification of farm

systems. Experimental Agriculture 24:399-419.
GRYSEELS, G. 1989. Role of livestock on mixed smallholder farms in the

Ethiopian highlands - A case study from the Baso and the Worena
Wereda near Debre Berhan. Wageningen, The Netherlands, Agricultural
University. Ph.D. Dissertation. 249 p.
HENDRIKX, J. 1985. Erfdierhouderij van de kleine boer in de ontwikke

lingslanden, (Keeping of backyard animals by the tropical small farmer).
Wageningen, The Netherlands, Agricultural University, Department of
Tropical Animal Production. M.Sc. Thesis.
HERMANS, C.; VEREIJKEN, P. 1992. Integration of animal husbandry and

nature conservation on grassland farms. Netherlands Journal of
Agricultural Science 40:301-314.
HOEKSTRA, P. 1948. Paardenteelt op het eland Soemba (Horse

husbandry on the island of Soemba). Indonesia, Departement van
Landbouw en Visserij, eeartsenijkundige mededelingen no. 90.
HUIRNE, R.B.M. 1990. Basic concepts of computerized support for farm

management. European Rural Agricultural Economics 17: 69-84.
HUITEMA, H. 1982. Animal husbandry in the tropics, its economic

importance and potentialities: Studies in a few regions of Indonesia.
Amsterdam, Royal Tropical Institute, Department of Agricultural
Research. Communication Series No. 73. 313 p.
IBRAHIM, M.N.M.; SCHIERE, J.B. (Eds). 1986. Rice straw and related
feeds in ruminant rations. Proceedings of an international workshop,
Kandy, Sri Lanka, March 24-28, 1986. Kandy, Sri Lanka, Straw Utilization
Project. Wageningen, The Netherlands, Agricultural University,
Department of Tropical Animal Production. Publication No. 2. 407 p.
ILEIA (INFORMATION CENTRE FOR LOW EXTERNAL-INPUT

AGRICULTURE). 1991. Participatory technology development in
sustainable agriculture: An introduction. Leusden, The Netherlands,
ILEIA.
JONG, R. DE; KURUPPU, L.G.; JAYAWARDENA, M. 1992. Management

and results of smallholder crop-stock in Sri Lanka. In Livestock and feed
development in the tropics. M.N.M. Ibrahim, R. De Jong, J. Van
Bruchem, H. Purnomo (Eds.). Malang, Indonesia. p. 357-366.
JONG, R. DE; MEINDERTS, J.H. 1992. Different approaches to dairy

stock development in Sri Lanka. In Livestock and feed development in
the tropics. M.N.M. Ibrahim, R. De Jong, J. Van Bruchem, H. Purnomo
(Eds.). Malang, Indonesia, Brawijaya University. p. 367-382.
KAASSCHIETER, G.A. 1990. La ganaderia en la sierra: Experiencias del

PRODERM. Cusco, Peru. Nederlands Economisch Instituut, PRODERM.
113 p.
KAASSCHIETER, G.A.; JONG, R. DE; SCHIERE, J.B.; ZWART, D. 1992.

A sustainable livestock production and health in developing countries,
Veterinary Quarterly. (in press)
KEULEN, H. VAN; BREMAN, H. 1990. Agricultural development in the

West African Sahelian region: A cure against land hunger? Agriculture,
Ecosystems and Environment 32: 177-197.
KIRAN SINGH; SCHIERE, J.B. (Eds.). 1993. Feeding of ruminants on

fibrous crop residues. Proceedings of a workshop, 4-8 February 1991.
Krishi Bhavan, New Delhi, India, Indian Council of Agricultural Research,
National Dairy Research Institute. 486 p.
KOESLAG, J.H.; URBINA, N.R. 1985. Produccibn de leche: Zonas de

ladera fria. Pasto, Instituto Colombiano Agropecuario, Pasto Project. 74
P.
KOK, J. 1921. Het rund en de rundveeteelt op Madoera (the bovines and

bovine production on Madoera). Rijks, Universiteit Utrecht, Ph.D.
Dissertation. 187 p.
KOREVAAR, H.; BOER, D.J. DEN. 1990. Practical measures to reduce

nutrient losses from grassland systems. Paper presented at the Fertilizer
Society, London, 13 December 1990. 30 p.
KORVER, S.; ARENDONK, J.A.M. VAN. (Eds.). 1988. Modeling of livestock

production systems. In Seminar of the EEC program for the
coordination of agricultural residues. Proceedings. Dordrecht, The
Netherlands, Kluwer Academic Publishers. 215 p.
KUIT, H.G. 1990. Ganaderia campesina: Estudio de caso y aportes para

el debate. Juliaca, Puno, Peru. CEPIA. 182 p.
LEESBERG, J. 1995. Gender issues in animal production systems

research. In Animal Production Systems: Global Workshop, San Jose,
C.R., 16-21 September 1991. M.E. Ruiz, C. Sere, H. Li Pun (Eds.). San
Jose, C.R., International Development Research Centre and Latin
American Animal Production Systems Research Network. IICA-RISPAL,
IDRC. p. 417-435.
LEEUWEN, N. VAN. 1991. Ganaderia en la sierra: Posibilidades y

limitaciones. Cusco, Peru. PRODERM, NEI. 90 p.
MERKENS, J. 1982. Bijdrage tot de kennis van den karbouw en de

karbouwenteelt in Nederlands Oost-I ndie (Contributions to the knowledge
of the buffalo and its husbandry in the Netherlands East-Indies). Utrecht,
The Netherlands. Ph.D. Dissertation. 191 p.
NELL, A.J. 1986. Feed cost in different farming systems. In Rice straw
and related feeds in ruminant rations. Proceedings of an international
workshop, Sri Lanka, 24-28 March 1986. M.N.M. Ibrahim, J.B. Schiere
(Eds.). Kandy, Sri Lanka, Straw Utilization Project. p. 326-336.
NORDBLOM, T.L.; AWAD EL KARIM HAMID AHMED; POTTS, G.R. 1985.

Research methodology for livestock on-farm trials. Proceedings of a
workshop, Aleppo, Syria, 25-28 March 1985. Ottawa, Can., International
Development Research Centre.
OEGEMA, S. 1989. La cunicultura en la zona influencia del CLEM - una

impresion. Dronten, The Netherlands, Agricultural Superior School/Tulua
(Valle), Col., SENA-CLEM. B.Sc. Thesis.
OSINGA, A. 1990. Dairy farming systems research in the Netherlands.

Paper presented at the Livestock Farming Systems Symposium,
Toulouse, INRA, France, 9-12 July 1990. 6 p.
PENNING DE VRIES, F.W.T.; DJITEYE, M.A. 1982. La productivite des

paturages saheliens. Textes du Cours PPS: Tome 1, theorie; Tome 2
travaux practique. Inv. Agronomique Wageningen. 237 p. + 139 p.
PEREZ, E.; BAAYEN, M.T.; CAPELLA, E.; BARKEMA, H. 1989.

Development of a livestock information system for Costa Rica. In
Livestock production and diseases in the tropics. Proceedings of the 6th
International Conference of Institutes for Tropical Veterinary Medicine.
H.K. Kuil, R.W. Paling, J.E. Huhn (Eds.). University of Utrecht, Faculty
of Veterinary Medicine. 411 p.
POL, P. VAN DER. 1992. Soil mining: An unseen contributor to farm

income in Southern Mali. Amsterdam, The Netherlands, Royal Tropical
Institute (KIT), Bulletin 325. 47 p.
PRONK, J. 1990. Een wereld van verschil, nieuwe leaders voor ontwikke-
lingssamenwerking in de jaren negentig. (A world of difference, new
frameworks for development cooperation in the nineties). The Hague,
The Netherlands, Government Printers and Publishers. 385 p.
ROELEVELD, A. 1995. Selected issues on adaptive animal production

research in Africa. In Animal Production Systems: Global Workshop,
San Jose, C.R., 16-21 September 1991. M.E. Ruiz, C. Sere, H. Li Pun
(Eds.). San Jose, C.R., International Development Research Centre and
Latin American Production Systems Research Network. IICA-RISPAL,
IDRC. p. 355-370.
ROLING, N. 1989. The agricultural research-technology transfer interface:

Knowledge systems perspective. The Hague, The Netherlands,
International Service for National Agricultural Research (ISNAR).
Linkages Theme Paper No. 6. 27 p.
SCHIERE, J.B. 1992. Aspects of modeling for sustainable crop-livestock

integration. In Livestock and feed development in the tropics. M.N.M.
Ibrahim, R. De Jong, J. Van Bruchem, H. Purnomo (Eds.). Malang,
Indonesia, Brawijaya University. p. 433-443.
SCHIERE, J.B. 1993. Research and extension in livestock development.

In Animal production in developing countries. M. Gill, E. Owen, G.E.
Pollett, T.L.J. Lawrence (Eds.). Wye, U.K., British Society of Animal
Production. Occasional Publication No. 16. p. 135-145.
SCHIERE, J.B.; WIT, J. DE. 1993. Feeding standards and feeding systems
(a review). Animal Feed Science and Technology 43:121-134.
SIMMONDS, N.W. 1986. A shcrt review of farming systems research in

the tropics. Experimental Agriculture 22:1-14.
SMITH, O.B.; BOSMAN, H.G. (Eds.). 1988. Goat production in the humid
tropics. Proceedings of a workshop held at the University of Ife. Ile-Ife,
Nigeria, 20-24 July, 1987. Wageningen, The Netherlands, PUDOC. 187
p.
STOMPH, T.J.; FRESCO, L.O.; KEULEN, H. VAN. 1994. Land use system
evaluation: Concepts and methodology. Agricultural Systems 44:243-
255.
STOORVOGEL, J.J.; SMALING, E.M.A. 1990. Assessment of soil nutrient

depletion in Sub-Saharan Africa: 1983-2000, Vol. I: Main report. 2 ed.
Wageningen, The Netherlands, Staring Centre. 137 p.
STROOSNIJDER, L.; RHEENEN, T. VAN. 1993. Making farming systems

analysis a more objective and quantitative research tool. In Systems
approaches for agricultural development. F.W.T. Penning de Vries et al.
(Eds.). The Netherlands, Kluwer Academic Publishers. p. 341-353.
STRUIF BONTKES, Tj. 1991. On dinka, cattle and grain, an interdiscipli

nary simulation study on a rural area in Southern Sudan. Wageningen,
The Netherlands, Wageningen Agricultural University papers 0169-345 x;
91-3. 102 p.
UDO, H.M.J.; BROUWER, B.O. 1991. Systematic analyses of the livestock

component of farming systems: Concepts and practices. Wageningen,
The Netherlands, Agricultural University, Department of Tropical Animal
Production.
VELDINK, J.G. 1970. W.C.H. Staring 1808 - 1877 Geoloog en

Land bouwkund ige. Wageningen, The Netherlands, PUDOC. Ph.D.
Dissertation. 260 p.
VELUW, C. VAN. 1987. Traditional poultry keeping in Northern Ghana.

ILEIA 3(4):12-13.
VEREIJKEN, P. 1992. A methodic way to more sustainable farming

systems. Netherlands Journal of Agricultural Science 40:209-223.
VERSCHUUR, C.M. 1991. La capricultura, una alternativa para el

campesino!? un estudio sobre las explotaciones caprinas en el Valle del
Cauca. Wageningen, The Netherlands, Agricultural University,
Department of Developmental Economics/Tulua (Valle), Col., SENA-
CLEM. M.Sc. Thesis. 35 p.
VOSKUIL, G.C.J. 1989. Zerograzing in Kenya. Livestock Newsletter (The

Netherlands) 2(1):24-28.
WALSHE, M.J.; GRINDLE, J.; NELL, A.J.; BACHMANN, M. 1991. Dairy

development in Sub-Saharan Africa: A study of issues and options.
Washington, D.C. World Bank, World Bank technical paper 0253-7494,
No. 135. Africa Technical Department series. 94 p.
WIT, J. DE; DHAKA, J.P.; SUBBA RAO, A. 1993. Relevance of breeding

and management for more or better straw in different farming systems.
In Feeding of ruminants on fibrous crop residues. Kiran Singh, J.B.
Schiere (Eds.). New Delhi, India. Indian Council of Agricultural
Research, National Dairy Research Institute. p. 404-414.
WOLF, J.; BREMAN, H.; KEULEN, H. VAN. 1991. Bio-economic capability

of West-African drylands. A report prepared at the request of the U.N.
Sudano-Sahelian Office, New York. Wageningen, The Netherlands,
CABO-DLO. 83 p.
YAR (YEMEN ARAB REPUBLIC). 1989. List of titles of range and livestock
improvement project communications with abstracts. Dhamar, Yemen
Arab Republic, RLIP/Amersfoort-Leersum, The Netherlands, DHV-RIN.
15 p.
ZEMMELINK, G.; BROUWER, B.O.; IFAR SUBAGIYO. 1992. Feed

utilization and the role of ruminants in farming systems. In Livestock and
feed development in the tropics. M.N.M. Ibrahim, R. De Jong, J. Van
Bruchem, H. Purnomo (Eds.). Malang, Indonesia, Brawijaya University.
p. 444-451.
SELECTED ISSUES ON ADAPTIVE ANIMAL
PRODUCTION RESEARCH IN AFRICA
A. Roeleveld'
INTRODUCTION
Dutch assistance to farming systems research and development

(FSR/D) programs having a livestock component is presently applied to
four sub-Saharan country-programs: Mali (DRSPR-Sikasso), Benin (RAMR-
Mono), Zambia (ARPT-WP) and Tanzania (Lake Zone). Table 1 presents
some basic parameters of the research area and the livestock component
of the programs.
Livestock research in all four programs is conducted within the context

of mixed farming systems. Animal traction plays a crucial role in
agricultural development in three out of the four programs (the Benin
program is the exception).
Research agendas of national FSR/D programs are not restricted to one

specific sector, theme, area, or client group. Their scope is broad in that
they have the task of identifying and testing technologies for crop and
livestock development and indicating environmental (institutional,
economic, legislative) aspects conducive to agricultural development.
Results and experiences of these programs differ because of location-

specific circumstances and their variable implementation periods.
However, they also share many themes with respect to technological,
institutional, and organizational issues. The following paragraphs highlight
four issues based on experiences which are largely shared by the
programs: priority setting, diagnostic research, research on crop-livestock
interactions, and the sustainability perspective of animal production
systems research. These issues have been chosen not only because they
FSRD/D Animal Husbandry Specialist, Agricultural Development Program, Royal Tropical

Institute (KIT), 63 Mauritskade, 1092 AD Amsterdam, The Netherlands.
are shared by the different programs but also because they have, as
different as the topics may appear, a common denominator which
repeatedly proves critical in adaptive research: priority setting.
FSR/D teams face a difficult task; the research area is often large and
very heterogeneous, creating numerous possibilities for agricultural
(livestock) research. Choices have to be made as to what research
themes to address, what methods to apply and how to organize the
research. The four issues presented in this paper are discussed in this
context.
Table 1. Characteristics of the research area and the livestock section of 4 FSR/D
programs in Africa
Program and countrya
ARPT-WP DRSPR-Sikasso Lake Zone RAMR-Mono

Zambia Mali Tanzania Benin
Research area
(103 km2) 126 96 20 4
No. of agroecological
zones 3 3 4 5
Average annual
rainfallb (mm) 760-1,140 600-1,200 700-2,000 1,000-2,000
Average population
density (cap/km2) 3-5 15-50 49-56 19-215
Farming systems mixed mixed mixed mixed
pastoral
Major livestock
species cattle cattle, sheep, cattle, sheep goats
goats
Program livestock
sections
o staff 2 3 - 2
u technicians 1 2 1 1
a
The programs are only identified by their acronyms or short names.
o
The range in different agroecological zones is given.
` Locally, much higher densities may occur; e.g., 100 cap/km2 along rivers and floodplains in Zambia, and
200-500 cap/km2 in Tanzania.
" Does not include fieldworkers.
PRIORITY SETTING
FSR/D teams can succeed only if they develop a coherent and well-
focused research program. This requires strategic decision making with
regard to client groups, priority zones, and research/development themes.
Two factors that have proved to be important when making strategic
choices, or setting priorities, are discussed below.
A common vision of agricultural development and the demand for

technology
Choices need to be made on the basis of a shared vision of the

constraints and potentials of the different farming activities (including
nonagricultural activities) in the different agroecological zones covered by
the program. The development of a common vision and its implementation
in a coherent research program is difficult and remains a challenge for the
four programs cited in Table 1.
A number of activities aimed at achieving this common vision and at

developing a clearer research strategy have been carried out by the
different programs:
State-of-the-art papers on agriculture in the different agroecological

zones (e.g., Dicko and Heemskerk 1990) were prepared. In addition to
facilitating interdisciplinarity, the drafting of state-of-the-art papers
requires the inventory and synthesis of earlier research and development
results, which are generally overlooked.
Diagnostic surveys in new research areas are critical, but demand

manpower and time; to solve this problem and to promote
interinstitutional interaction, other services (e.g., the Regional
Development Organization, the Extension Service, and the Livestock
Development Service) have been invited to participate.
Participatory planning workshops were organized in Mali and Benin.

Staff members of other research institutions and agricultural services
also attended the workshops, in the course of which the participants
exchanged ideas and discussed research priorities. This has resulted
in a shared, overall vision of agricultural constraints and development
objectives, and has substantially contributed to mutual understanding
and cooperation among services. Follow-up activities are necessary to
elaborate specific work plans and to adjust or elaborate upon earlier

findings. These activities are defined in conjunction with the annual
review of research results and formulation of the workplan.
A vital attribute of adaptive research teams is an interdisciplinary work

style, without which a common focus cannot be developed. Unfortunately,
however, multidisciplinary teams do not have a natural tendency to strive
for interdisciplinarity. Activities as described above are both an end in
themselves and a means to forge interdisciplinarity.
Research protocols and reports
Research programs are frequently hindered by deficient research

protocols and poor reporting of research findings. This is often due to the
fact that research programs are too demanding, leaving very little time for
proper analysis, timely reporting of trial results, and well-elaborated
research protocols.
In recent years several programs have noted the imbalance between the
time spent on field work and the time spent on the preparation, analysis,
and reporting of the trials and surveys. To solve this problem, in-service
training courses have been organized in several programs. Training
modules have been developed to facilitate these courses. More attention
has also been given to the discussions of trial results and to the
formulation of new research proposals with the participation of the teams
working in the villages, the on-station research centers, and the
extension/development organizations. So far, results have been
encouraging:
The quality of the reports and protocols have been substantially

improved and national coordinators in Mali and Zambia have decided
that the formats applied should become the standards for their national
FSR/D programs.
Interdisciplinary teamwork is improving.
Farmer assessment of trial results gained importance and provided

essential feedback to the research program.
Linkages with other services and research centers have been improved,
resulting in: (1) improved dissemination of tested technologies, (2)
monitoring of the impact of the disseminated technologies, and (3)
changes in on-farm and on-station research agendas (these changes

cannot, however, be fully ascribed to improved reporting and
presentation of,trial results alone).
DIAGNOSTIC LIVESTOCK RESEARCH: IN SEARCH OF QUICKER

IDENTIFICATION OF ON-FARM EXPERIMENTS
In the diagnostic phase, surveys were conducted to monitor basic

productivity parameters of cattle, goats, and sheep. The studies aimed at
providing more detailed information on reproduction, mortality, growth, and
changes in herd composition, in order to define constraints on livestock
production and identify opportunities for innovations. Table 2 presents
characteristics of herd monitoring surveys conducted in Benin, Zambia,
and Mali.
Team approaches used to gain farmer participation varied considerably

among the respective programs. In the earliest surveys, carried out in Mali
(Bengaly et al. 1993), field workers did most of the measuring, farmers
were not asked for much additional information, and feedback of survey
results was minimal. Having no clear idea of the purpose of these surveys,
let alone an interest in their implementation, farmers' attitudes towards the
surveys remained passive. Not much information on the farmers' decision-
making process was gathered, which complicated data interpretation.
Monitoring surveys recently conducted in Benin and Mali adopted a

more participatory and action-oriented approach and were less extensive
in design (both in duration and in the number of animals monitored). After
each visit, productivity and management data were discussed with the
participants, both individually and jointly. In the course of the group
discussions, research themes or development activities (e.g., disease
control campaigns) were identified. In village meetings, once or twice a
year, research results were reviewed and new activities were planned.
These meetings were also attended by representatives of the local
veterinary and livestock service; thus, they were directly involved in the
planning of development activities and even in the execution of research
activities. This improved collaboration between the adaptive research team
and the services involved.
Table 2. Characteristics of herd monitoring studies conducted in three African countries
Program and country
ARPT-WP DRSPR-Sikasso RAMR-Mono

Zambia Mali Benin
No. of agroecological zones studied 11 2 2

No. of villages involved in fieldwork 212 8 3
No. of households involved > 100 20-30 80-100
Average number of animals in survey
c. Cattle 1,500 800

Sheep 1,700
c. Goats 1,000 500
Frequency of observations (weeks) 2-4 8 4-5

Duration of study (months)
c Fieldwork 24 24-48 18
t. Analysis, reporting 4 10 5
Time spent on monitoring studies (%)
P. livestock staff ±50 ±40 25-40

o Technicians - 50 -
c. Fieldworkers 50 10-15 25-40
Subdivided into four subsystems.

2 Kraals, each comprising cattle belonging to 3 to 14 owners.
With respect to the survey design, the research programs observed four
different aspects which merit particular attention:
Survey designs were generally based on vaguely defined research

concerns (i.e., obtaining more precise information on constraints in
animal production).
Data gathering in all three cases (Table 2) was very labor intensive and
time consuming for field workers as well as for staff. Consequently, for
a long time (two to three years), relatively few on-farm experiments were
conducted.
Time and skills required for data processing and analysis were often
underestimated.
The wealth of productivity data does not imply a smooth translation of

survey results into concrete experimental programs.
Preliminary information with regard to farmer participation indicates the

following:
Considerable time is devoted to the implementation of this type of

survey.
Farmer participation has been increased by the involvement of both

survey participants and village communities, the latter being crucial in
program activities focused on management of natural resources
(rangeland) at the village level.
Discussion of survey results with farmers resulted in better

understanding of constraints and potentials for animal production.
Whether or not the participatory approach followed is efficient and

effective in the identification of constraints and potentials still needs to be
assessed. It was found that without sufficient in-service training of field
workers (and the supervisory staff), the approach is doomed to fail
because it will quickly fall back into the usual gathering of productivity
data.
It has been concluded that adaptive research programs must carefully

consider the necessity of monitoring surveys as a diagnostic tool;
adequate survey design requires clear research questions (hypotheses).
The rapid identification of on-farm experiments must receive more
attention; in fact, research managers must see to it that on-farm
experiments quickly become more important than diagnostic activities.
RESEARCH ON CROP-LIVESTOCK INTERACTIONS:

THE NEED TO FOCUS ON NEW OPPORTUNITIES
Research programs analyzed in this paper focus on mixed farming

systems, whose improvement depends heavily on a clear understanding
of crop-livestock interactions. Research into the field of crop-livestock
interactions has been quite successful with respect to animal traction in

Mali (Kleene et al. 1989), through studies on adaptation of farm
implements, training of oxen, and management of oxen (housing and
supplementary feeding with crop residues and agroindustrial by-products).
Also, the first technology recommendations produced in Zambia were
related to animal traction.
Possibilities for strengthening research on crop-livestock interactions are

generally underestimated, while the possibilities for more intensive
exploitation of feed resources (crop residues and rangeland) are often
overestimated.
Lack of knowledge of present husbandry practices usually cause the first

error. The limited scope for improvement is illustrated by the role of
manure in the restoration of soil fertility. Manure is generally used
intensively by farmers in Mali and Zambia (Penninkhoff 1990). Research
has focused on increasing the quantity of manure by improving kraaling
(confinement) and by adding straw to fresh manure in the kraal. Partly
because of methodological problems, few quantitative data on the effect
of manure on soil properties and crop yields are available; however,
farmers in South Mali are increasingly inclined to invest in the improvement
of their kraals and stables and in transportation of crop residues and
manure. Possibilities for further improvement in manure production and
use are limited; even if all available manure is used to fertilize cropland, soil
fertility (in many areas) cannot be maintained under the present farming
practices (Keulen and Breman 1990; Pol 1992).
The frequent overestimation of carrying capacity of village territory is

illustrated by the use of crop residues, which are often intensively grazed,
or stocked and fed to the animals in the dry season. Farmers in Mali have
started to refuse to allow transhumant pastoralists to graze crop residues
because they require the increasingly scarce forage for their own cattle.
The increased pressure on land not only results in competition for scarce
forage resources between two different animal production systems, but
also results in increasing competition between animal and crop production
(expansion of cropland has deprived cattle of the best rangelands, while
high stocking rates and shortened fallows do not allow for the restoration
of soil fertility).
Where the traditional themes for strengthening crop-livestock interaction

have little scope for further development, new opportunities present
themselves. The only possible strategy for agricultural development lies in
the increased productivity of land (arable and range). In the context of
those areas where the carrying capacity of the natural resources is

achieved or even exceeded, measures to increase productivity have to go
hand in hand with measures that aim at judicious resource management,
such as erosion control and land regeneration (Keulen and Breman 1990).
Research programs will increasingly focus on the role of perennials in
cropping systems, improved fallow, intercropping with food legumes,
optimum application of chemical fertilizers, and improved rangeland
management. However, many "innovations" are conditioned by better land-
use regulations and improved prices for both livestock and crops.
THE SUSTAINABILITY PERSPECTIVE: A FATAL ATTRACTION?
The context of sustainable land use
In much of the semiarid and semihumid climatic zones of Africa,

agriculture is subject to considerable pressure from an increasingly
intensive exploitation of natural resources. Expansion of arable land has
resulted in cultivation of soils less suitable for agriculture and often
susceptible to erosion, and in a shortening of the natural fallow period
required to restore soil fertility. In addition, a growing number of animals
depend on crop residues for fodder supply as the area of natural grassland
decreases. Consequently, agricultural practices in many areas have largely
exceeded the environmental carrying capacity, resulting in depletion of the
soil, degeneration of natural vegetation, and erosion of unprotected soils.
For the development of sustainable farming systems, the balance

between production and natural resource conservation must be restored.
However, increased population pressure generally requires a production
level higher than, or at least equal to, the present production level. As
economic restrictions often hamper the use of external resources,
possibilities for increased productivity are limited. Under these conditions,
optimal use of available means is an absolute must.
The sustainability perspective in adaptive animal production research
Interest in sustainable development of agricultural systems has

increased, as reflected in a growing number of publications and increased
research on sustainable technological innovations and systems
development. Hart and Sands (1991) applied the concept to land use and
elaborated a conceptual framework for sustainable land-use systems

research. Posner and Gilbert (1991) discussed whether or not research on
sustainable agriculture fits within FSR/D programs and will prove to be a
"fatal attraction."
The inclusion of criteria for sustainability, referring to ecologically sound

and economically viable land use, undoubtedly complicates the research
agendas of FSR/D programs, due to increasingly complex research issues,
expansion of the time horizon, the level of research (from farm to village),
the level of the decision-making process (from farm household to village
community), and the methodologies required.
Nevertheless, the most recent operational plans of the four FSR/D

programs mentioned earlier pay considerable attention to (mostly
biophysical) sustainable land-use and sustainability as an additional
criterion of technology performance. This approach will have far-reaching
implications for research; for example, the development of a coherent,
interdisciplinary research program, the mix of short- and long-term
research themes, the development of methodologies that suit the new
requirements, and the stimulation of collective efforts at the village level are
among the issues that need attention. It goes without saying that these
FSR/D programs can only gradually incorporate sustainability as a criterion
for the evaluation of system performance.
Although the gap between what needs to be done and what has been
achieved is likely to widen, the sustainability issue should not be
considered as a "fatal attraction" to FSR/D programs. Moreover, as
illustrated above, FSR/D programs have no choice but to address the
sustainability issue. Rather, the question is how and to what extent. Not
all programs can nor should immediately embark upon a full-fledged land
management endeavor. However, ongoing and new research themes
should explicitly address the possible medium- to long-term impact of
technology on the environment, and the question of whether technology
should be accompanied by other technical, economic or organizational
measures in order to become (or remain) environmentally sound. An
interesting, difficult subject in animal production is the impact of
supplementary feeding (nitrogen supplementation) in the dry season on the
exploitation rate of rangeland.
In the field of animal production, sustainable land use is closely related

to soil fertility of cropland and the exploitation of rangeland by cattle and
small ruminants. Two aspects with a clear focus on sustainable land use
are manure and improved fallow, as previously discussed.
The most complicated research focus is the one that looks into the
natural resource management of an entire village. This comprises both
arable and rangeland, and takes into account the different forms of
exploitation (crop and animal production, wood collection) of these
resources, whether by the villagers themselves or by "outsiders."
In recent years, research on land management at the village (watershed)

level has received considerable attention in sub-Sahelian countries, many
of which face serious overexploitation of their natural resources. The
DRSPR-Sikasso program, located in southern Mali, is the only one of the
four programs mentioned which is conducting research in land
management at the village level, and is one of the earliest of its kind in
West Africa; therefore, this program merits more discussion.
Community-based land management in southern Mali
The origin of the DRSPR-Sikasso program lies in the technical and

methodological approach developed between 1980 and 1986 by Hijkoop
et al. (1991) to combat soil degradation. The Malinese Rural Development
Organization (CMDT) decided to introduce the approach on a large scale
in the cotton-growing zone. It therefore created an erosion control
program known as the "Projet Lutte Anti-Erosive" (PLAE) in 1986 (Campen
1991). PLAE refined the approach and developed training courses for
extension workers.
In some of the villages, the establishment of physical erosion control

measures such as stone rows and grass strips was very successful.
Residents of these villages complained, however, that nonresident cattle
owners and woodcutters damaged their work and made further
improvements in the management of their resources virtually impossible.
The interest displayed by the farmers in developing more activities to stop
the alarming rate of resource degradation in the area resulted in a pilot
program in 1989 that aimed at developing an approach for better land
management at the village level (Djouarma et al. 1991). The concept of the
program was based on the hypotheses that (1) villagers regard
management of their resources and their sustainable use not only as
individuals, but also as a collective responsibility and priority; (2) the State
creates a favorable socioeconomic environment and officially delegates
management responsibility to the villagers. Whether these hypotheses are
correct remains to be proven.
Land management is very difficult because it is affected by many

different levels of the decision-making process. Fortunately, the Malinese
government and the CMDT have given a high priority to gestion de terroir
programs. The development of these programs cannot be addressed by
the adaptive research institutes or any other single institute. A
collaborative, long-term program is a prerequisite. Therefore, a technical
committee has been set up, where PLAE/CMDT, DRSPR, PROFED
(Women's Development Program), the Livestock Service, the Forestry
Service and local administration officers participate.
In identifying the boundaries of the study area, the program immediately

ran into the first major difficulty with regard to land management: land
rights. In order to avoid land feuds, village elders preferred not to establish
village boundaries. In the end, an area of 12,000 ha, demarcated by three
roads and comprising land of five neighboring villages, was identified as
the project area.
Initially, the services and programs involved communicated with the

villagers largely through the villages' administrative heads. It turned out
that villagers were insufficiently involved and informed. Discussions
between the technical committee and village administrators on this matter
have resulted in the creation, by the villages, of an intervillage committee.
Presently, PROFED and DRSPR are developing methods to increase
women's participation in the program.
Constraints and development possibilities were discussed in village

meetings, in which an interdisciplinary group of representatives of the
technical committee participated. Major constraints identified by the
farmers were the following:
Exploitation of trees by nonresident woodcutters
Grazing of "their" rangeland by large herds owned by nonresidents (rich

traders and transhumant pastoralists)
Bush fires, soil and rangeland degradation, and forestry services taxes
Fig. 1 presents a diagram of the constraints and solutions as they have

been identified by the farmers and the technicians.
DRSPR supported the process of the identification of problems and

activities by conducting specific studies on land rights, on women's
participation in erosion control activities, and on the actual stocking rate
(an extensive carrying capacity study was conducted in 1989), among

other issues. The stocking rate study illustrated the severe pressure
exerted on the rangeland by the livestock of nonresidents.
After almost two years, the changes in the use and management of the
land consisted mainly of a reduction of the pressure exerted by
nonresidents. Due to legislative restrictions, residents cannot forbid
nonresidents to graze their animals on the land used by the six villages.
Important achievements have been made in the organization of the users
(i.e., villages) and the supporting services, as well as in the cooperation of
services and local administrative and political authorities. It is too early yet
to expect any changes in the actual manner and intensity of land
exploitation by the villagers themselves. Hunter and Weaver (1991),
synthesizing 11 years of experience with a rangeland management area in
Lesotho, concluded that the exclusion of nonresident cattle is one of the
few tangible results as far as intensive land exploitation is concerned.
Creativity will be required in looking for ways to increase the productivity
of good arable land and decrease exploitation of crop and rangeland. To
this end, the adaptive research team should provide the necessary
information to decide on the required socioeconomic, institutional, and
legislative conditions for improved land use.
IDENTIFIED PROBLEMS
1 Damage to
Overgrazing erosion control
devices
Grazing by Villagers Own Free roaming

Grazing by herds of cattle, sheep,
herds of urban transhumant too much and goats in
traders \ pestorelis;s cattle the dry season
Solutions proposed by villagers

Sensitization of Sensitization Selling of non- Application of
I traders to productive animals sanctions
leave the zone Regulations for
III\ passing herds Introduction of Avallabllity of
Report tress meat In local diet concentrates
passers to the
authorities Cattle fattening
authorities
Solutions proposed by technical
committee
Regulations of Tax levying on Destock herds of Concentrate
access to village grazing according to large cattle distribution
rangeland species and owners according to
management area forage,
Cattle tracks practices Savings notes crops, and
improved kraals
Stabling of cattle Use of forage
crops
Fig. 1. Diagram on the analysis of overgrazing in Siwaa, Southern Mali

Preliminary results of the southern Mali case study led to the following
conclusions:
Research and development activities in land management at the village

level require a concerted approach by the development services
involved, the adaptive research team, the local authorities and the
organized villagers.
Adaptive research plays a supporting role in this process. Important

activities are participatory diagnostic research, the identification of
promising technologies, and the identification of important contextual
factors that impede, or are conducive to, sustainable land use.
Progress in the development of sustainable land use is slow due to the

large number of contextual factors involved and the substantial changes
in farming brought about by relatively small changes in land-use
patterns.
In view of the above, research in land-use systems, at the village level,

is likely to have more impact if it is integrated with development activities
of other institutes, services, or projects.
Effective changes in land use can only be brought about by a set of

interventions at different levels (i.e., farm, village, local, and national
authorities).
Resource conservation and agricultural production must be taken as a

unified concern if any adoption of technology is to take place.
PROFITING FROM COMMON EXPERIENCES
Exchange of, and building upon, experiences with new issues and
approaches are considered important tools to improve the performance
and efficiency of research programs. The KIT Agricultural Development
Program has started a project that will capitalize on the experience gained
by other FSR/D programs it has supported. The programs containing the
animal production component have helped in the design of methodology
in three areas: (1) diagnostic research with particular attention to the role
of monitoring surveys, (2) the role of improved fallow, and (3) priority-
setting with respect to research on crop-livestock interactions.
LITERATURE CITED
BENGALY, K.; MEURS, C.B.H.; BERCKMOES, W.M.L. 1993. La

productivite des bovins et des ovins dans les exploitations agricoles de
Fonsebougou. Rapport de recherche. Sikasso, Mali. lnstitut
d'Economie Rurale, Departement de Recherche sur les Systemes de
Production Rurale de Sikasso.
CAMPEN, W. VAN. 1991. The long road to sound land management in

southern Mali. In A. Juijsman and H. Savenije (Eds.). Making haste
slowly: Strengthening local environmental management in agricultural
development. Development-oriented research in agriculture.
Amsterdam, The Netherlands. Royal Tropical Institute. p. 131-148.
DICKO, M.; HEEMSKERK, W. 1990. The wetland agricultural system in

Western Province. State of the art paper nr. 1. Mongu, Zambia, Ministry
of Agriculture, Department of Agriculture, Adaptive Research Planning
Team Western Province.
DJOUARMA, H.; JOLDERTSMA, R.; KOOIJMAN, M. 1991. Gestion de

terroir villageois: Un pari difficile. L'experience de la Recherche-Systeme
au Mali-Sud. Paper presented at the "11th Annual Farming Systems
Research and Extension Symposium in the 1990's: Critical issues and
future directions", Michigan State University, 5-10 October.
HART, R.D.; SANDS, M.W. 1991. Sustainable land-use systems research

and development. AFSRE Newsletter 2 (1):1-6.
HIJKOOP, J.; VAN DER POEL, P.; KAYA, B. 1991. Une lutte de longue
haleine... amenagements anti-erosifs et gestion de terroir. lnstitut
d'Economie Rurale (Bamako, Mali), lnstitut Royal des Tropiques
(Amsterdam, Pays Bas). Systemes de Production Rurale au Mali, Vol.
2. 154 p.
HUNTER, J.P.; WEAVER, L.C. 1991. Development of grazing associations

in Lesotho: The search for sustainability. Paper prepared for
presentation to the International Symposium on Management Systems
for Sustainable Agriculture for Sub-Saharan Africa. Amsterdam, The
Netherlands. Royal Tropical Institute.
KEULEN, H. VAN.; BREMAN, H. 1990. Agricultural development in the

West African Sahelian region: A cure against land hunger?. Agriculture,
Ecosystems and Environment 32: 177-197.
KLEENE, P.; SANOGO, B.; VIERSTRA, G. 1989. A partir de Fonsegougou

... presentation, objectifs et methodologie du Volet Fonsebougou (1977-
1987). Institut d'Economie Rurale (Bamako, Mali), Institut Royal des
Tropiques (Amsterdam, Pays Bas). Systemes de Production Rurale au
Mali, Vol. 1. 145 p.
PENNINKHOFF, P. 1990. Cattle manure in the farming systems of

Western Province: Practices and possibilities. Mongu, Zambia. Ministry
of Agriculture, Department of Agriculture, Adaptive Research Planning
Team, Western Province.
POL, P. VAN DER. 1992. Soil mining: An unseen contributor to farm

income in southern Mali. Amsterdam, The Netherlands. Royal Tropical
Institute. Bulletin 325. 47 p.
POSNER, J.L.; GILBERT, E. 1991. Sustainable agriculture and farming

systems research teams in semiarid Africa: A fatal attraction?. Journal
of Farming Systems Research and Extension 2(1): 71-86.
NEW ISSUES
THE ROLE OF LIVESTOCK IN THE DESIGN OF
Robert D. Hart' and Michael W. Sands2
INTRODUCTION
Many environmentalists are convinced that sustainable livestock system

is an oxymoron. The word livestock brings to mind overgrazed hillsides
or compacted tropical soils. In countries with well-developed economies,
livestock are often associated with human health problems that are a result
of overconsumption of animal fats. In short, few people have made the
argument that more research needs to be done to improve livestock
production systems as a means of developing more sustainable land-use
systems. But that is precisely one of the key conclusions of this paper.
Livestock are blamed for both their direct and indirect negative impact
on the environment. Repetto (1987), a senior economist at the World
Resources Institute, states over 70% of the Third World's rangelands are
now moderately or severely desertified" and that an indirect negative effect
of the livestock sector on the environment has occurred because
"governments, especially in Latin America, have offered generous fiscal and
financial support" (to the livestock sector). He concludes that this has
reduced agricultural employment opportunities and increased cropping
pressures, which "aggravates soil erosion and losses in soil fertility."
But even livestock critics admit that, if managed correctly, livestock

systems can play an important role in agricultural development without a
negative impact on the environment. Some authors have even argued that
a specific livestock technology, such as the use of animal manure, can
have a positive impact. However, it is seldom suggested, as it is argued
' Executive Director of INFORM, 611 Siegfridals Rd., Kutztown, PA 19530, USA.
2 Director of the International Division, Rodale Institute, 611 Siegfridals Rd., Kutztown, PA
19530, USA.
in this paper, that livestock systems can play an important role in the
development of more sustainable land-use systems. This paper is
organized as follows: first, the concept of a sustainable land-use system is
defined; examples of different constraints to sustainability are identified; the
potential role of livestock technology to ameliorate sustainability constraints
is discussed; and finally, the potential contribution of livestock systems to
the development of sustainable land-use systems is analyzed.
Many definitions of sustainability have been proposed. Brklacich et al.

(1991) have suggested that conceptual inconsistencies are a result of
different perspectives, such as environmental accounting and sustained
yield, among others, taken by different authors. Geng et al. (1990) point
out that conceptual inconsistencies can be clarified by distinguishing
between system components, which are then subdivided into resources,
technologies, and activities, which are analyzed according to their
objectives. Lyman and Herdt (1988) view sustainability as "... the latest
twist in the continuing elaboration of criteria by which agricultural
development is defined and agricultural technology is evaluated."
One factor that contributes to both the semantic and conceptual

differences regarding sustainability is the use of the term as both a system
property and as a type of agricultural and rural development. FAO (1991)
defines sustainable agriculture as "...the management and conservation of
technological and institutional change in such a manner as to insure the
attainment and continued satisfaction of human needs for present and
future generations." In spite of the differences between the way the
concept of sustainability is applied, most approaches have in common the
objective of maintaining present natural resource productivity.
In this paper, it has been assumed that sustainability is a characteristic

of the relationship between a resource and its use. This implies that a
system cannot be described as sustainable or nonsustainable unless the
limits of the system include all of its critical resources. This leads to the
conclusion that the concept of sustainability is more applicable at a
macroscale and should be applied to larger systems such as a watershed
or a geographic region rather than at the farm or farm enterprise level.
New Issues 375
This concept has no applicability at the component technology level since

technologies in and of themselves cannot be either sustainable or
nonsustainable.
The emphasis in this paper is on land-use systems. The term "land-use

systems" will be used as a name for a system that could be defined more
precisely as a "resource-use systems." In effect, "land" will be used as an
inclusive label for a set of resources that includes biological, social and
economic resources and not just the physical resources usually associated
with the word "land." While agriculture is only one type of land use, the
land-use systems discussed in this paper will be agricultural land-use
systems.
While farming is only one type of agriculture (forestry and ranching are
two other types), the examples discussed in this paper will be farm-
oriented. The concept of a "sustainable land-use system" and the graphic
representation of the concepts used in this paper evolved from a workshop
on this topic sponsored by the Indian Council of Agricultural Research, the
United States Department of Agriculture, and the Rodale Research Center
(Hart and Sands 1990).
The farm system in the diagram depicted in Fig. 1 interacts with the
socioeconomic environment through the purchase of inputs (seed, fertilizer,
etc.) and the sale of outputs (e.g., grain, employment of family labor off-
farm). It interacts with the biophysical environment through the
degradation of natural resources caused by some farm outputs (pesticides,
manure) and through the use of resources (such as water and nutrients)
as inputs for farm production processes.
Research to develop sustainable land-use systems requires explicit
consideration of all the farm/environment relationships depicted in Fig. 1.
Inputs or farm technologies that do not negatively affect the environment,
but are not economically viable, will not be adopted by farmers. The
development of economically viable systems that degrade the resource
base to the point that natural-resource productivity decreases cannot
sustain system production longer than society is willing to subsidize it by
purchasing external inputs. Extensive production systems that are
managed in such a way that their production exceeds the productivity of
the natural resource bases (e.g., cutting forest faster than a forest can
produce trees) may be economically viable in the short run, but the income
produced by these systems is, in effect, being borrowed from future
generations.
Value of system production exceeds value of inputs
Inputs do
PURCHASED
INPUTS I I SYSTEM
PRODUCTION
not degrade 7 production does

UM bass
resource not our ea
RESOURCE RESOURCE
b
DEGRADATION 1 PRODUCTIVITY resource
productivity
BIOPHC
ENV MIE
Resource degradation does not exceed

resource productivity
Fig. 1. Sustainable land use system.
The diagram depicted in Fig. 1 helps to clarify concepts, but is too

general to have value as a conceptual framework for developing
sustainable land-use systems. Fig. 2 is a first attempt toward the
development of a general systems framework that could be used by a
multidisciplinary team to develop a site-specific system. The framework
subdivides the socioeconomic and biophysical environment into local and
regional systems. Farm systems are conceptualized as including the farm
household (which is also a component of the local socioeconomic system)
and the associated crop and/or livestock production systems. Crop and
livestock systems are depicted as interacting with the soil (which is also a
component of the local biophysical system) and with the pests and
diseases that affect them.
While there is general agreement on many of the concepts discussed

above, there is no agreed-upon methodology to develop and promote
sustainable land-use systems. However, there is considerable experience
and methodology available for the development and promotion of many of
the technologies often associated with sustainable systems (e.g., minimum
tillage, multispecies cropping systems, and integrated pest management).
New Issues 377
Also, methodologies that have evolved from agroforestry research,

crop/livestock systems research, and farmer-participatory research can
definitely contribute to the development of a sustainable land-use systems
research and development methodology.
EG ON O
LOCAL SOCIOECONOMIC ENVIRONMENT
INPUT purchased outputs URBAN

PRODUCERS Inputs sold MARKETS
C
HOUSEHOLD 0
CROPS
I pests &
diseases 1C
y LIVESTOCK
F
A
R
M
M R
U
N
E
I
N
T
SOIL Y N
NATURAL soil water J IRRIGATION

ECOSYSTEMS pollutants nutrients SYSTEMS
LOCAL BIOPHYSICAL ENVIRONMENT
v Aquatic
ec"yo
i \ar
EG ON BIORHYSI 4 ENVIRONMEN VA
Fig. 2. A first attempt towards the development of a general systems

framework.
SUSTAINABILITY STRATEGIES
To develop more-sustainable land-use systems, the first key question

that must be addressed is: What system/resource relationship appears to
be nonsustainable? Clearly, a different approach will be taken if the
relationship between groundwater and irrigated grain production is at risk
than if the problem is the relationship between local markets and vegetable
producers. In many cases, the sustainability of a given land-use system
may be affected by the nonsustainability of more than one
system/resource relationship.
The next step in the development of an alternative land-use system is to

identify more specifically the constraints that make the system/resource
relationship nonsustainable. A nonsustainable groundwater-irrigated grain
production relationship may be caused by increasing salinization. The
vegetable producers/local markets relationship may be constrained by
increasing product quality demands made by local consumers. When
sustainability constraints have been identified, potential technologies can
be identified that can ameliorate these constraints. For example, better
water management technology can be developed for the grain producers,
and better post-harvest processing systems can be developed for the
vegetable producers. One way of organizing sustainability constraints and
the potential technologies to address these constraints is by analyzing
different types of system/resource relationships. Four types of
relationships included in the conceptual framework are depicted in Fig. 2:
(1) farm/physical resource relationships, (2) farm/biological resource
relationships, (3) farm/economic resource relationships, and (4)
farm/social resource relationships. Examples of different sustainability
constraints and potential technologies to address these constraints are
listed in Table 1 and are briefly discussed below.
Constraints related to farm/physical resource relationships are often the

key to land-use sustainability, and they are the first examples that come to
mind when considering the issue of sustainability. It is difficult to disagree
with the premise that the maintenance of natural resource productivity,
more specifically soil productivity, is a prerequisite for any sustainable land-
use system. Soil productivity is tightly linked to plant production.
Technologies to ameliorate these sustainability constraints, such as the use
of erosion control barriers and the recycling of crop residues, usually
tighten soil/plant cycles and decrease nutrient losses so that they do not
have to be replaced by external inputs.
Maintaining farm/biological resource relationships is also critical to land-

use system sustainability. Both crop and livestock germplasm are key
biological resources. The evolution of pests and diseases requires the
constant development of new varieties and breeds. A less obvious
biological resource that affects farm system sustainability is the flora and
fauna found in natural ecosystems. Researchers are becoming
increasingly aware of the importance of beneficial insects and
microorganisms for the biological control of pests and diseases. Soil
microorganisms, earthworms, and termites, to name a few, are key
biological resources, and it is becoming increasingly clear that they are
critical to the sustainability of agriculture.
New Issues 379
Table 1. Examples of different sustainability constraints and potential technologies that

could address the constraints
System/resource Sustainability Potential technologies

relationship constraints
Farm/physical resources Farm resources (such as Decreasing soil fertility
(such as soil and water local and social service Decreasing soil
and regulatory institutions) moisture
Farm/biological resources
(such as crop and lives- Increasing pest
tock germplasm, bene- damages
ficial organisms) Increasing disease
resistance
Farm/economic resources
(such as credit-purchased Decreasing net farm
inputs, markets) income
Leguminous cover crops Increasing quality
Decreasing access to land Use of animal manure requirements
Use of living mulches

Decreasing social services Water harvesting
techniques
Release of biocontrol
organisms
Species-specific pesticides
Diverse crop rotations New

crop varieties and animal
breeds
Decrease use of
purchased
higher-value crops and
livestock
Improved postharvest
handling
More heterogeneous
varieties
Changes in land tenure

policy
More intensive production
systems
Formation of local coope-

ratives
Changes in government
policies
Economic viability depends on sustaining the relationship between a

farm and its economic resources such as credit, purchased inputs
(fertilizer, pesticides, etc.), and markets. This relationship can be
susceptible to changes in either the economic environment or the farm
system. Changes in input availability or market demand usually require
changes in the farming systems. Changes in farming practices can lead
to changes in the characteristics of the economic resources, such as when
increased production leads to decreasing prices in the market.
Farm/social resource relationships are just as critical to land-use system

sustainability as the relationships with the physical, biological, and
economic resources. Local social service institutions affect farmer access
to education, health care, and so on. Other institutions regulate access to
important resources such as land. Policies affect land tenure, while
subsidies of inputs and price supports for specific commodities greatly
affect farm system sustainability. As in the case of farm/economic
resource relationships, farms can impact institutions and institutions can
impact farms. This relationship is, of course, greatly affected by different
types of political systems that control local empowerment.
Livestock can affect land-use system sustainability in two very different

ways: (1) livestock technology can ameliorate specific farm/resource
sustainability constraints, and (2) livestock systems can affect different
farm/resource relationships. Both these potential roles are discussed
below.
THE ROLE OF LIVESTOCK TECHNOLOGY
Table 1 includes examples of different sustainability constraints and

potential technologies that could ameliorate these constraints. The
examples cited include different types of agricultural technologies, but
many of them are associated with the production of livestock feed,
livestock herd management, or livestock wastes. Since inappropriate
management of feed production systems, livestock herds, and animal
wastes can have a serious negative environmental impact, both the
potential negative and positive roles of livestock with regard to the four
types of system/resource relationships are discussed below.
New Issues 381
Livestock/physical resource relationships can be impacted both

positively and negatively by livestock technology. The most common
negative impacts of livestock on the physical resource base are from
overconcentration of manure (affecting both surface and ground water),
soil compaction, and the indirect effect of overgrazing on soil erosion.
However, livestock technology can have a positive impact on the physical
resource base through the use of legume-based feed production systems
and through prudent management of manure to improve soil structure and
fertility.
Livestock can have a negative effect on biological resources through

their impact on natural flora and fauna. Emphasis is usually placed on the
role of livestock in displacing megafauna such as elephants and other high-
profile animals, but impact on soil microflora and microfauna may be much
more important in the long run. On the other hand, livestock feed
production technologies, such as those present in agroforestry systems
and other high-diversity systems, can contribute positively to the
maintenance of biological resources. Animal manure can be managed in
such a way as to improve soil biodiversity.
Farm/economic resource relationships can be negatively impacted by

the use of technology that leads to overproduction or unacceptable quality
of livestock products; notwithstanding, livestock products often have
characteristics that give them advantages over crop products. Some
animal products can be sold directly; in other cases, processing
technology can be used to add value to the product (e.g., milk or cheese).
Livestock technology that adds flexibility with regard to the timing of the
purchase and use of inputs or the timing of product sales can have a
positive impact on farm/market relationships.
Livestock technology can also affect livestock/social resource

relationships. In many cultures, livestock are a form of insurance; animals
are sold only when cash is needed in a emergency. Livestock can also
play a critical role in defining and maintaining social relationships. Large
livestock herds may be maintained for purposes of social status even
though they are overgrazing the range and causing environmental damage.
However, there may be situations where livestock infrastructure or livestock
technologies (such as new breeds) can replace large herds that are
maintained as insurance or as a status symbol; these new technologies
may have a less negative environmental impact.
THE ROLE OF LIVESTOCK SYSTEMS
It is often difficult to separate the impact of livestock component

technologies on specific system/resource relationships from the role of
livestock systems within the complex set of many different
system/resource relationships that make up a farm's environment.
However, the characteristics of livestock systems (as opposed to livestock
technology) give them key advantages over other types of systems in the
development of sustainable land-use systems.
Livestock systems are both resilient and resistant to changes in the

ecological and the economic environments. They are resistant to (i.e.,
relatively unaffected by) changes in climate, as compared with many crop
production systems; for example, many types of livestock can tolerate long
periods of drought. They are resilient (i.e., relatively flexible) when
subjected to changes in prices; for example, a decrease in the price of milk
can be met by changing the herd structure to decrease milk production or
by marketing meat instead of milk.
A key question is What are the characteristics of some livestock systems

that give them the ability to buffer changes in both the ecological and
economic environments? Specific properties are difficult to separate, but
for the purpose of this analysis the following four properties, characteristic
of many livestock systems, can be identified.
Input flexibility. A key characteristic of many livestock systems is that

they tolerate changes in sources of inputs. Feed mixes can be formulated
from many different feeds, either produced on the farm or purchased from
off the farm.
Optional outputs. Livestock systems typically have many different types

of outputs and the market can alter the relationships among these. For
example, when prices go down in one product (milk), emphasis can be
shifted to others (cheese, meat, live animals).
Spatial disaggregation. Many livestock systems have multispecies feed

production systems and multispecies herds. The herds are often managed
as age and sex subunits (male calves, cows, etc.). When these
components are separated spatially, they interact with different sets of
physical, biological, economic, and social resources. This decreases the
risk to the system since different system components function within
different environments.
New Issues 383
Temporal discontinuity. Livestock systems are typically multitrophic with

outputs from one component acting as inputs to another to form chains.
Changes in the soil affect plant growth, plant growth affects animal
performance, animal performance affects output production, and output
production affects product processing. Since each of these linked
processes functions on a different time schedule, changes in one process
may take months or even years to affect another. Processes like feed
storage (of hay, silage, etc.) break up time sequences even more. Risk is
decreased since an unpredicted stress may affect one process, but the
output from this process may not be needed until a future date, and a
different source may be substituted, leaving the next link in the chain
unaffected.
CONCLUSIONS
While there is no doubt that inappropriate livestock technology and

livestock systems can have a serious negative impact on resources and
can reduce land-use system sustainability, livestock systems have unique
properties that allow them to make an important contribution to the
development of sustainable land-use systems. Properties such as input
flexibility, optional outputs, spatial disaggregation, and temporal
discontinuity make livestock systems resilient and resistant to changes in
either the ecological or socioeconomic environment.
Mixed livestock-cropping systems can be dangerous. The complexity

of mixed farms and farm/resource relationships makes it difficult predict
whether a new technology will have a net positive or net negative impact
on the sustainability of land-use system. For example, changes in the
social system such as increasing a price subsidy for a commodity (e.g.,
milk) can lead to changes in livestock management that require changes
in feed production that can affect the production of food crops that, in turn,
can increase soil erosion. The possibility of introducing new technology
with unforeseen consequences is high.
It is customary to end this type of paper with a call for more research.
Given both the important, potentially positive role for livestock in the
development of sustainable land-use systems and the potentially negative
impact from introducing unpredictable changes, the call for more research
in this case is more than a traditional conclusion; it is a demand on the
behalf of future generations that can potentially benefit enormously from

livestock-based land-use systems or suffer the consequences of
nonsustainable development.
LITERATURE CITED
BRKLACICH, M.; BRYANT, C.R.; SNIT, B. 1991. Review and appraisal of

concept of sustainable food production systems. Environmental
Management 15(1):1-14.
FAO (FOOD AND AGRICULTURE ORGANIZATION). 1991. The Den

Bosch declaration agenda for action on sustainable agriculture and rural
development; report on the conference. Conference on Agriculture and
the Environment. S-Hertogenbosch, The Netherlands, 15-19 April, 1991.
GENG, S.; HESS, C.E.; AUBURN, J. 1990. Sustainable agricultural

systems: Concepts and definitions. Journal of Agronomy and Crop
Science 165:78-85.
HART, R.D.; SANDS, M.W. 1991. Sustainable land-use systems research

and development. In Sustainable Land-use Systems Research.
Workshop Proceedings. New Delhi, India. R.D. Hart, M.W. Sands
(Eds.). Kutztown, Pa. Rodale Institute. p. 1-12.
LYMAN, J.K.; HERDT, R.W. 1989. Sense and sustainability: Sustainability

as an objective ininternational agricultural research. Agricultural
Economics 3(4):381-398.
REPETTO, R. 1987. Economic incentives for sustainable production.

Annals of Regional Science 21(3): 44-59.
POLICY IMPACT ASSESSMENT IN
SMALL-RUMINANT PRODUCTION SYSTEMS:
THE PERUVIAN CASE
Corinne Valdivia'
INTRODUCTION
Agricultural development projects usually share a common goal; to

increase productivity as well as the income and welfare of the farmers
targeted. However, after a technology is developed, it is often introduced
with little success, or when successfully adopted, the returns are low or
negative. In farming systems research (FSR) various exogenous factors
are acknowledged (Norman 1991); these may affect the system or the
outcome of a project but in many cases how the results are impacted by
such external factors has not been assessed.
Some of these exogenous factors are what has been referred to as

higher hierarchy systems (Hart 1995); these higher systems include
markets, the domestic and world economy, and government policies aimed
not only at the producers but also at the economy as a whole. The
economic approach to development has changed in the past forty years.
It has gone from sector analysis of agriculture and its contributions to
development, to a micro-approach where the focus has been placed on
the individual producers and their technology. In the last 25 years, the
induced innovation theory has proposed that producers will introduce
innovations, and that markets (specifically their price signals), will
determine the adoption of technology.
1
The author was a Post-Doctoral Research Associate of the Small Ruminant Collaborative
Research Support Program (SR-CRSP), University of Missouri-Columbia (UMC) and
Winrock International when this paper was paper was presented at the Animal
Production Systems Global Workshop, Costa Rica, September, 1991. The constructive
comments of Domingo Martinez and Jere L. Gilles at UMC on an earlier draft of this
paper, are much appreciated.
After the failure of the belief that technology was readily available and
only needed to be transferred, the farming systems approach was adopted
to develop more appropriate technologies or innovations. It has become
widely recognized that innovations are transferred successfully only if the
system and its external effects are well understood. FSR has been
successful at understanding the interactions within a system and also at
developing new alternatives that require an additional commitment of
resources.
If the external conditions that impact the system, such as markets and
government policies are unfavorable, producers will not be willing to adopt
these innovations. These conditions must be modified in order for the
adoption to take place. Thus, it is important to take the external factors
into account because rational producers observe their effects when
adoption decisions are made.
One approach to understanding the impact of government policies and

markets on producers is multimarket analysis. Markets and their signals
(such as prices) impact farmers to varying degrees. These signals are
determined by their level of competition, integration, and production
structure. Prices are also determined by the behavior of the consumers of
farm products, at any level of processing. The industry, its technology and
competitors, and the demand (domestic and international) all form a
system of interrelated markets, and this system has an impact on the
producer. When government policy intervenes in the system, producers
are affected.
This paper describes how higher hierarchy systems are introduced and
used for policy impact assessment in small-ruminant production. The
macroeconomic environment and its influence on policy formulation is
discussed first. The economies of several Latin American countries are
moving towards free trade and market-oriented policies that will eliminate
the distorting effect of government intervention on factor prices, such as
interest rates, wages, and exchange rates; the impact of this movement is
analyzed.
The system of markets within which farmers operate is described and

the impact of market linkages and technology on the producer's income
is presented. Finally, the model used to assess policy impact for the case
of the small-ruminant producers of Peru is explained, identifying those
policies that were relevant in the last 25 years.
New Issues 387
THE MACROECONOMIC ENVIRONMENT: AGRICULTURE

AND GOVERNMENT POLICIES
Developing countries look to improvements in the agricultural sector as

part of the process of becoming developed nations. Many different models
of development have been tried since the 1950s. Sometimes the priorities
were set in the industrial sector, while at others times the Japanese model
of the beginning of this century was followed, emphasizing a strong
agriculture as the basis for the development of the industrial sector.
The Small-Ruminant Collaborative Research Support Program (SR-

CRSP) was designed to understand the system in which small-ruminant
production takes place. Several research components were considered.
Agropastoral communities were studied in Peru from a multidisciplinary
perspective. This required the participation of scientists from range
management, animal breeding, health, agronomy, sociology and
economics. The goal was to develop effective technologies through an
understanding of the production system and the economic and social
interrelations of agropastoral communities raising small ruminants and
camelids.
However, there is an additional need to understand the policies that

affect producers in order to propose effective alternatives. In so doing, it
is necessary to keep in mind what the government goals and priorities are,
as well as the limitations it faces. For example, Peru has been formulating
development plans since the late 1950s in order to obtain aid from US
lending agencies; however, many plans emphasizing development of a
domestic industry were not successful. During the 1970s, a plan to change
the structure of agriculture was undertaken through an agrarian reform
process that lasted ten years. The 1980s saw a new view of agricultural
development based on small private landholdings, promoted by national
and foreign institutions. Notwithstanding, the government development
goals for agriculture were set aside because the short-term problems they
faced outweighed the long-term goals of development. Though the
government is concerned with the process of agricultural development, it
is not willing to pay the political price required (like unpopular measures
such as higher food prices).
De Janvry's perspective of the agrarian question in Latin America relies

on the existence of two economies in conflict, an export economy and an
infant industrial economy, which require opposite government policies to
grow (De Janvry 1981). Which one is given priority will determine where
the development emphasis is placed. In the past, most of Peru's

investment in agriculture was geared towards the export of raw materials
such as cotton, sugar and coffee, a policy backed by the well-to-do social
classes, who derived a benefit from it.
According to Timmer et al. (1983), the actual prices for factors of

production faced by rural decision makers (wages, the cost of capital, and
imported equipment) are influenced significantly by macropolicies. Hayami
and Ruttan (1985), analyzing the gap created between the developed and
developing countries with respect to agriculture, point to the distortion of
factor prices introduced by governments as the cause for this gap. In
Latin America, capital was subsidized believing that industry could be
developed through distortion in the real factor prices; the result was to
favor the use of capital-intensive technologies.
Timmer et al. (1983) indicated that low interest rates, overvalued

exchange rates2, and subsidies to labor-saving technologies can
counteract the growth of demand for labor in agriculture when a policy of
high agricultural prices is introduced. Timmer et al. (1983) also stated that
interest rates, wage rates, exchange rates, land rental rates, and the rural
and urban terms of trade are influenced by macroeconomic policies.
Overall price3 policies (those affecting interest rates, wages, and the
exchange rate) may counteract the incentive policies introduced at the
producer level.
Gardner (1987) stated that overvalued and undervalued exchange rates

impact the farm goods that are usually traded (imports and exports) by
affecting their prices relative to those goods that are not traded. Quoting
Schuh, Gardner (1987) stressed the need to adopt border policies to
insulate the domestic market from exchange rate fluctuations that can
swamp the domestic commodity-market policies. Internationally traded
agricultural goods are affected by these policies. In Peru's case, the
government's control of the exchange rate has resulted in overvaluation
2 The term exchange rate in this paper corresponds to the amount of domestic currency
paid for each US dollar. An overvaluation of the exchange rate occurs when the
domestic currency is overvalued or overpriced: more dollars can be bought with a set
amount of local currency.
3
Macro prices are the prices of factors set for the economy as a whole. These are prices
determined by the macroeconomy, and sometimes by the government through policies
affecting the interest rate, exchange rate, and wage rate (especially the minimum wage
rate).
New Issues 389
because the exchange rate has remained fixed despite inflation. This
overvaluation of the domestic currency with respect to the dollar, and not
its fluctuation,. has undermined domestic commodity-market policies.
As Reardon (1984) and Gardner (1987) stated for the Peruvian policy of
the 1980s, the budgetary outlays for subsidies, such as for urban food, can
impose significant pressures on national debt payments. Overvalued
exchange rates make imports cheap. The presence of currency controls,
import tariffs and barriers, and allocation of foreign exchange to preferred
importers indicate the existence of an overvalued national currency.
Overvalued exchange rates act as implicit taxes on agriculture because
internal prices received by farmers must be low to compete with cheap
imports. Thus, the terms of trade are unfavorable for agricultural
producers. The prices of nontraded goods are relatively high compared
with the prices of food and imported goods.
The governments of developing countries tend to subsidize interest

rates. According to Timmer et al. (1983), and Hayami and Ruttan (1985),
this subsidy distorts the efficient choice of technology. High interest rates,
on the other hand, increase domestic saving and reduce unemployment
by encouraging investment in labor-intensive technologies.
Minimum-wage laws affect both the choice of technology and job

creation by encouraging the use of capital-intensive technologies. Timmer
et al. (1983) stated that macro-prices have an effect on specific policies
designed to stimulate growth of a particular sector ("sectoral efforts to
design incentive price policies for particular commodities are partly or even
totally vitiated by contrary pressures from foreign exchange rate, alternative
subsidy policies or high internal inflation").
In order to understand what will happen to the prices received by the

producers, it is necessary to consider the direction that government
policies are taking. With (1) structural adjustment, whereby prices are
determined by the market and not by government subsidies or controls
and (2) market-oriented policies that encourage free trade and the
elimination of government controls, producers may or may not be worse
off, depending on the type of product they generate. If production relies
on the use of subsidized capital or imported inputs, the implementation of
these policies will cause production costs to rise and producers will not be
likely to introduce or continue practices that are too costly. On the other
hand, with production of traded goods, those that may be imported or
exported, a policy of structural adjustment that reduces overvaluation of
the local currency (exchange rate) will increase the prices received by
farmers. At present, because devaluation has not been successful in many

Latin American countries, imports are still cheaper than domestic products.
A move towards market-oriented policies under these adverse price
conditions, will depress the economy of domestic agricultural producers.
AGRICULTURAL POLICY ANALYSIS AND THE ROLE OF INDUSTRY

AND INTERNATIONAL MARKETS
According to Gardner (1987), the best way to make conjectures about

policy is- to use ,a simulation model. This forces the analyst to be
systematic about inferences and conjectures made, explicit about facts
used and assumptions maintained, leading naturally to a quantitative
statement of results.
There are several approaches to policy evaluation; single-market partial

equilibrium, multimarket, and general equilibrium analysis are frameworks
that can be used to analyze policy. Each approach has its advantages and
disadvantages (Tolley et al. 1982; Braverman et al. 1987; De Janvry and
Sadoulet 1987; Gardner 1987). Linkages between the commodities, groups
of commodities, or sectors,4 and the rest of the economy, are crucial in
deciding which approach to use. It is also important to understand the
limitations of each approach to policy analysis.
Partial equilibrium analysis
In single-commodity markets, partial equilibrium models have been used

to analyze policies. With this type of approach the welfare analysis
concentrates on the impacts on consumers and producers through the
changes in consumer and producer surplus. This method assumes that
all other prices and parameters remain constant, i.e., they do not influence,
and are not influenced by, the rest of the economy.
4 Sector is defined as a group of markets that are interrelated through supply and demand
of outputs and inputs; the small ruminant sector of Peru includes sheep producers, the
textile industry and the meat processing industry of both sheep and South American
camelids (i.e., more than one market).
New Issues 391
This type of analysis does not take into account the impact on related
markets that a policy distortion introduces, such as the changes in demand
for inputs, and the input prices. Because these linkages to other markets
are ignored, the final effects on the output prices, that are used as a
reference for welfare or cost-benefit analysis, are overestimated, unless
there are abundant input supplies such that an increase in demand will not
change the input prices (De Janvry and Sadoulet 1987; Hertel and Tsigas
1988). Partial equilibrium analysis is appropriate when the supply of inputs
is unlimited, hence costs will not change and the output supply curve will
remain the same (Just et al. 1982; Gardner 1987; Hertel and Tsigas 1988).
General equilibrium analysis
General equilibrium analysis, on the other hand, takes into account all
the market interrelations. The supply and demand of outputs and inputs,
as well as all the resource constraints and flow of payments to factors are
considered. Consumption and savings as well as taxes and government
expenditures (e.g., subsidies) are considered relevant to the economic
cycle. This approach takes into account the relations between industry
and agriculture, and allows for analysis of welfare and income redistribution
between social classes. With this approach, De Janvry and Sadoulet
(1987) identified social classes by type of consumption goods that they
purchased, and the price change of these goods were used to calculate
the impact on their real income.
Hertel and Tsigas (1988) stated that the benefits of computable general
equilibrium models are theoretical consistency, accounting consistency,
treatment of interindustry effects, welfare analysis and view of the total or
global economy. It explains how the economy as a whole has an impact
on agriculture and vice versa. This approach is appropriate for aggregate
analysis.
Multimarket analysis
A third approach takes place when the commodity or sector being

analyzed is so small relative to the rest of the economy that the demand
for inputs from that sector will have no or very small impact on the prices
of the rest of the economy. Thus, if the objective of the analysis is to
measure only the impacts within the sector, a market analysis that takes
into account the interrelated markets would be enough to address policy
questions (Hertel and Tsigas 1988). Multimarket analysis acknowledges
the existence of cross-price effects between outputs, and addresses the

problem of substitution between outputs and inputs without the
complications of building a computable general equilibrium model.
If the questions to be answered relate only to the sector that is being

studied, and the effects to the rest of the economy are negligible,
resources would be better used in a multimarket analysis. In Peru, the size
of the small-ruminant sector with respect to agriculture and to the rest of
the economy is small, but interrelations can be traced to the textile sector,
making this multimarket approach attractive for policy analysis.
The multimarket method is a simulation tool that incorporates the

interactions across markets, and the economic, political and technical
constraints which policy makers face. As any simulation tool does, it helps
systematize ideas, consistently emulates how the system works, and
provides information for policy decision; it does not predict exact
outcomes. The advantage over larger models is that it is not necessary to
have complete information about market interdependence to make good
use of the multimarket approach. It is_ important to decide which
interlinkages are critical and which indirect effects are insignificant.
Policy analysis, in the framework of interrelated markets, yields results

that would not be found in single-market partial equilibrium analysis.
Governments apply different policy instruments, the final effects of which
are determined by the interrelations and structure of the markets being
studied.
In agriculture it is necessary to determine the processing chain that farm

products go through, the technologies used in processing, the ability to
substitute the farmers' products sold in the industry, and the type of
demand for the final products. Identification of the interrelationships a
commodity has with industries and final markets will define the system
through which policies are felt, and will determine what will happen to
prices when innovations are introduced. If markets are competitive, the
technology of each industry and the demand for its final products
determine the industries' ability to purchase the producers' output. If the
demand for the final product is very stable and will not increase much
when the price falls, and if the possibility of substituting other inputs for the
farm product in the production process is very small, the price of the farm
output falls at a greater pace than the increase in demand, leaving the
producers worse off.
New Issues 393
If products are tradeable, which means that they are traded in the
international market (either imported or exported), any government policy
that affects the exchange rate will have an impact on producers. Any
policy used by the government to promote exports or to promote or
control the importation of products competing with the ones being
produced in these markets will also influence producers.
Market linkages and market distortions: their impact

on the producer's income
Determining the linkages between producers and industry is necessary

to understand what influences the prices paid to producers. If demand
varies with price change, a technological innovation that reduces the cost
of production per unit or increases the productivity at a given cost will
increase the use of this farm product. At the same time, the characteristics
of this product in the industrial production process will determine if more
will be purchased when the price falls. If this product can take the place
of other factors in the production process, producers will be able to sell
more. On the other hand, if the structure of production is fixed, and this
product cannot take the place of other inputs, a decrease in its price,
without a change in the price of other inputs, will not determine an
increase in the demand for farm output, resulting in an even greater drop
of producers' income.
Another factor to take into account is the production function at the farm
level. In the case of sheep in the Peruvian highlands, meat and wool are
jointly produced. Another important output as a source of fertilizer and
energy is manure. In the case of meat and wool, a multimarket system
approach is appropriate because the effects of policy and technological
innovations filter through the system, making it possible to assess the
impact of both final markets on producers. If retail meat prices fall, there
will be consequences at the producer level that will also affect the textile
industry because meat and wool are jointly produced.
Finally, the degree of government intervention in the markets, at the

producer or consumer level, will influence prices and producer income. If,
as has been traditional in the past, a government decision determines the
amount of meat imported (by subsidizing the exchange rate and allocating
import quotas), domestic producers will be forced to compete at a
disadvantage with producers of other countries. The domestic processing
industry will also be forced to compete with the importers, who are usually
monopolists due to the size of the local markets. When this happens, the
gains from trade go to the importers and consumers, while revenues of
domestic producers drop.
Distortions of macro-prices, such as exchange rates, through

government intervention, subsidize imported goods by making them
cheaper than goods produced domestically. This policy has depressed
export markets of such goods as textiles, depressing demand for farm
products by the domestic industry.
Technological innovations: who benefits?
Economic theory has shown that technological innovations in a

competitive setting benefit consumers in any given society. They also
benefit innovators (those who first adopt the technology) who will increase
the supply of farm products, thus driving the price down. The high-cost
producers that do not adopt this technology will be forced, to sell their
products at lower prices, which means that their retained income will drop.
Developers should realize that increasing production will depress

agricultural prices. Because this is the case they should try to determine
what the demand for the increased production looks like. If the industry
cannot absorb the product and there are no alternative uses or markets
developed simultaneously, producers will face a decrease of their income.
If the demand for the final goods using this farm product as an input does
not increase to offset the fall of the farm product's price, producers'
income will fall. A multimarket system identifies these market linkages and
industry performance and helps explain what the results of an increased
supply of the farm products will be.
the equations that define all interrelated markets are identified, as well
If
as the markets' equilibrium conditions, then a set of simultaneous
equations can be derived. In these equations, endogenous and exogenous
variables are identified, and parameters such as cost shares at the industry
level, elasticity coefficients of processing technology, and market shares
(domestic and international) are used to simulate alternative production
scenarios. An increase of production of farm inputs (exogenous) is
introduced into the system to capture its effect on the use of those inputs
and on prices received by farmers. If two or more commodities are jointly
produced, as sheep meat and wool, they both must be taken into account
to predict the effects on all markets.
New Issues 395
FINDINGS AND RESULTS OF SMALL-RUMINANT RESEARCH

IN PERU: ISSUES THAT NEED TO BE CONSIDERED
IN PRODUCER-IMPACT ASSESSMENT
In developing the research for policy impact assessment on the case of

small ruminants in Peru, several hypothesis were considered. The first, that
macroeconomic policies and the accompanying macro-prices (in this case,
specifically exchange rates) have a significant impact on producers. These
and market structure (monopoly, oligopoly, and vertical integration are
some examples) determine the final impact of policies and innovations on
farmers. The influence of government's macropolicies can override the
effects of specific policies designed to benefit producers.
With respect to the effect of policies on the livestock sector, agricultural

producers receive little protection and are discriminated against by
subsidies on competitive imports and by taxes on their exports. The
government's import policy on meat, designed to maintain low food prices
and low labor costs for the industrial sector (De Janvry 1981; Reardon
1984), has adversely affected sheep producers by lowering the internal
price of meat. This has also decreased wool production, given that wool
and meat are produced jointly.
Policies that affect producers, processors, and consumers of meat and

wool were analyzed. A multimarket model of the livestock sector for policy
simulation was developed. The set of markets and the interrelationships
of the system can be seen in Fig. 1. As in many developing countries, in
Peru inadequate data availability imposes limitations and restricts the
selection of methods to analyze effects.
In the Peruvian Andes, a large part of the rural population raises small
ruminants. Sheep, alpacas and llamas are raised in agropastoral and
pastoral peasant communities, and some production cooperatives that
remain from the agrarian reform of 1969. These peasant producers have
been largely forgotten by governments in their development plans. Only
25% were commercial producers5 (members of the production
cooperatives). The peasant producers trade their products whose value
is influenced by market conditions.
5 These are defined as commercial producers because their production is specifically

destined for the markets, whereas individual producers sell only their surplus.
PRODUCTION PEASANT
COOPERATIVE PRODUCERS
I
SHEEP
PRODUCTION
I
TAXES
MEAT WOOL EXCHANGE
RATE
QOTAS
I\17
11
I=J
Fig. 1. The sheep multimarket system of Peru.
The government's support of agriculture has centered on the

development of irrigation projects in the arid coastal region where export
crops and products for urban consumption (such as rice and sugar) are
grown. Colonization projects to develop cocoa and coffee production
were encouraged through credit incentives in the Amazon region; also,
through the 1980s, emphasis was placed on crops for industrial use,
especially to support the poultry industry. Historically, small-ruminant
producers, particularly those in peasant communities, have been left out
of the country's political and economic arena. This is reflected in the lack
of studies on small-ruminant producers and the surrounding conditions and
policies that have made this group one of the poorest of the country.
New Issues 397
The poor understanding of the conditions determining the levels of

poverty faced by highland producers prompted the necessity to analyze
the way they are integrated to the country's economy and politics, and the
channels through which political and economic conditions are felt. This
study identified mechanisms through which producers relate to the
economy and the means by which the trend of worsening poverty could
be stopped and reverted. This study also estimated the costs to the
government in order to propose policy alternatives that will be successful.
For an economy in crisis, with poor investment capacity, attractive policies
are those that can stop the process of impoverishment without major
investments and without endangering the government's political stability.
Thus, the study consisted of an analysis of a small-ruminant sector

comprised of three major markets: two retail ones (meat and textiles) and
the farm-level market for sheep and alpacas, sheep carcass meat, and
wool and fiber. Policies and the linkages through which producers were
affected were identified. The study analyzed policy formulation in this
sector of the economy. It also identified interrelationships and effects of
the macroeconomy on the farmers. The policies and scenarios were
designed with intentions of correcting market signal distortions (which
historically have been against the interests of sheep and alpaca producers
of the highland peasant communities) and stopping the process of
impoverishment. The expectation of integrating FSR with multimarket
analysis is that combined research efforts at the farm and global level in
the SR-CRSP will help to understand interactions, give a range of possible
innovation responses, and help meet the goal of economic welfare for
small-ruminant producers.
Small-ruminant production in the context

of government policy formulation
A first step in the formulation of policy is to understand the role and the
importance of the sector being analyzed within the economy and politics
of a country. Governments try to maximize their income and invest in
those sectors that respond most quickly. They also try to maintain their
political stability investing in those areas that will help achieve this goal.
Policy becomes effective when the governments have a possibility of
gaining from it. To implement policies that will develop or at least protect
the producers' income, it is important to show how the government
benefits from these policies. The rural population has the disadvantage of
not being able to exercise political pressure at the government level,
therefore their interests are rarely defended. An alternative way to attain
political stability in the rural areas (while favoring rural farmers and
deferring impoverishment) is through increasing foreign exchange reserves
by increasing exports of wool and alpaca fiber, and reducing imports of
meat. In the short run, an easy way out of food scarcity is imports, which
directly compete with domestic production. In many cases, due to an
exchange rate overvaluation, imports become inexpensive and this will put
products of rural areas at a competitive disadvantage, thus contributing to
impoverishment.
Profile of the livestock system in the Peruvian economy
Peru is, today, one of the poorest countries in South America.

Characterized as an open economy, this country exports between 20 and
25% of its gross domestic product (GDP). Around 80% of the exports are
primary products. Fifty percent of these exports are oil and copper.
Traditional agricultural exports are cotton, sugar, coffee, wool and fiber.
The economy is very sensitive to international market cycles (Thorp and
Bertram 1978).
Agriculture's contribution to the GDP has fluctuated between 14% (1970)

and 11% (1984), and employs 31% of the total work force. While the
country's population has increased at a rate of 2.8% per year, Lima has
grown at a rate of 3.7 percent. Population in the highlands, where
livestock production is concentrated, increased at a rate of 1.3% (Martinez
1986); thus, migration to the cities is an acute problem. The impact that
government policy has on targeting the poorest of the poor (the small-
ruminant producers), needs to be assessed (Gilles 1980). Political unrest,
land invasions and violence are manifestations of this poverty.
Both the agropastoral and the pastoral (livestock) peasant communities

produce sheep. Altitude and diversity of resources determine the
proportions of crop and livestock production. Jamtgaard (1984), using the
1977 census of the officially recognized peasant communities, developed
a typology that classified 33% of the communities as pastoral and 27% as
agropastoral. These communities own approximately 50% of the sheep in
the country. The rest is owned by cooperatives and independent
producers. The latter may include producers in peasant communities that
are not officially recognized by the government.
Government intervention in the agricultural sector of Peru increased in

1969 with the agrarian reform. Subsidies, taxes and credit policies, as well
as intervention in marketing channels, and import and export regulations
New Issues 399
were frequently used. Reardon (1984) typified these government policies

as "cheap food policy" (low prices for food) and protection to the Infant
Industry (cheap imports of inputs and protectionist measures). In 1980, a
populist group defended cheap food policies and native industries, while
confronted with a monetarist team linked to the International Monetary
Fund interested in liberalizing policies and trade favoring farmers. In 1985,
price controls, zero-interest credit for poor peasant producers, crop
insurance, and target prices for food crops were introduced. Livestock
cooperatives in the south were reorganized by transferring lands to the
peasant communities.
The small-ruminant system: Results of policy simulations
The small-ruminant (sheep-alpaca) sector consists of three interrelated

markets: the sheep producers that sell slaughtered animals and raw wool,
the meat retailers, and the wool textile producers who purchase their inputs
from sheep and alpaca producers. International trade is essential to sheep
meat consumption and wool and fiber products exported. Government
intervention policies affect all of these actors (Fig 1).
The meat market depends on administered procurement prices

(Reardon 1984); import licenses are allowed to vary in order to clear the
market. In the case of lamb, prices tend to be regulated indirectly through
imports from New Zealand and other countries. Wool which is exported
to the international market can take many forms and is affected by several
government policies. Some raw wool may be exported directly with no
processing, but producers must pay export taxes. The government also
may intervene to limit or forbid the export of greasy wool to benefit the
domestic processing sector. This sector obtains tax credits when they
export the processed wool and fiber; that is, the government encourages
value-added exports.
Reardon (1984) characterized the manufacturing sector, including

textiles, as an oligopoly, which implies that the benefits from technological
improvements at the farm level are captured by this industry. Some wool
processing cooperatives have performed the role of price regulators, with
the intention of improving the prices paid to producers, but finally failed.
Meat processing consists mainly of slaughter houses located in urban
areas, where producers or middlemen bring the sheep. The government
intervenes in this market by controlling the price of beef and subsidizing
poultry production. Imports are consumed mainly in the capital city.
The peasant communities sell part of their wool to the middlemen, and
sell meat to the local markets. In agropastoral communities, a portion of
both wool and meat are sold locally. Government intervention at the
producer level consists of redistribution of pastures from cooperatives to
peasant communities. Export taxes and quantity controls on wool mainly
affect the price-responsive producers.
Food policy impacts the meat market. The policies adopted by the
government to increase imports and satisfy the national demand, while
keeping prices low, have discouraged domestic producers.
Simultaneously, the textile industry has been favored by protectionist
policies that forced producers to sell the wool and alpaca fiber to the
domestic industry. Food and industrial policy impact on the farmer,
discouraging production.
The policies analyzed in the Peruvian case consisted of exchange rate

changes (brought about by either elimination of subsidies or devaluation
of the domestic currency), the move towards a market-oriented economy
and an increase in food availability. The way that these policies link with
the effects that increased farm production has in the multimarket system
was also analyzed (some policies are highlighted by circles in Fig.1).
Several issues made the small-ruminant sector interesting. The first was
the joint nature of wool and meat production. The second was that both
products are sold to the manufacturing and processing industries. The
third is that technologies used in the two industries differ in the ability to
use these farm products. The fourth was that both products are affected
by international trade and by the exchange rate (i.e., the macro-price)
policy.
FINDINGS
About the joint nature of wool and meat production
The joint nature of wool and meat production implies that the retail meat
and the textile markets are not independent; therefore, policies formulated
for one market will have an impact on the other. In general, given the
substitution of inputs in industry, and the response of supply and demand,
a policy that increases the demand of one product (meat or textiles), be
it domestic or international, increases the price of that farm product and
decreases the price of the other farm product with which it is jointly
produced. This drop in the price of the second farm output takes place
because there is an increase of its supply, while its demand stays the
New Issues 401
same, thus driving the price down. Nevertheless, all policy simulations
show an increase in the average revenue to farmers (taking into account
both meat and textiles) when the increase in production is demand-driven.
About the effect of technology in industry
The elasticity of substitution is a coefficient that indicates how readily

one input in a given industry will substitute for another when the prices of
these inputs change, relative to each other. A greater elasticity of
substitution results in a greater demand for the cheaper input. If joint
production exists, the size of the elasticity of substitution determines the
size of the fall of the second farm product; the greater the elasticity of
substitution is, the smaller will be the fall of the other farm product price,
because producers of the other industry will be induced to demand the
input that has become relatively cheaper. The various simulations show
that there are effects in all markets, the size depending on the values of the
elasticity coefficients.
Given that response in terms of sheep supply is small, an increase in

demand for both meat and textiles has a greater impact on their prices
than on quantities produced. Because inputs in textile production can be
more easily substituted, wool prices will not rise as much as carcass price.
It is important to determine the value of the substitution coefficients in the
processing industries to learn about the magnitude of change of input
prices and quantities with more precision.
One important problem in determining the elasticities of substitution is

the lack of reported statistics at the industry level. Secondary sources can
give a feel for the behavior of technology, and can help determine the
probable value of these coefficients. When such sources are used, a
sensitivity analysis should be conducted to determine the consistency of
the results when the values of these parameters change. The value of
these coefficients is also crucial when new technologies that increase
production are introduced at the farm level.
About producer and consumer response
Valdivia (1990) estimated consumer-demand responses (elasticities of

demand) at the retail level for sheep meat and calculated the demand for
textiles using theoretical economic assumptions due to lack of statistics.
These consumer goods are responsive (i.e., elastic) to both price and
income changes. The demand response to price of retail sheep meat was
estimated at -1.16, which means that if meat prices fall off 1 %, demand will
increase by 1.16 percent.
At the producer level, collection of data is a problem. Sheep producers

are not very responsive to price; thus, the only price coefficient obtained
was not statistically different from zero: .08 in the short run and .11 in the
long indicating that if prices would rise by 1 % production would rise
run6,
by. 1 % Given these data restrictions, a sensitivity analysis was conducted,

changing the parameters, to determine how robust the results were with
respect to these changes. In general, the results did not change.
About technology that increases farm production
Economic theory has shown that the benefits of technological

innovations are eventually passed to the consumers (Schuh 1990).
Producers are better off when prices fall only if they are able to sell more
in proportion to the price drop. Increases of meat and wool at the farm
level are brought about by the elimination of technological and market
constraints. An important constraint in the Peruvian highlands is the
availability of feed sources. Many technological alternatives have been
developed, but few are being adopted because external conditions have
not changed. Given the depressed meat prices, introduction of cultivated
pastures is a technology that will be sitting on the shelf for some time.
This is so because producers are not willing to adopt a technology that
requires additional resources, whose costs will not be covered by
increased prices.
Other constraints exist and FSR contributes by identifying them and by

generating technology to remove these constraints. Consider, for example,
a simulated scenario where a technological change results in increased
feed availability and, consequently, in increased sheep production. Given,
that other market conditions remain the same, and the elasticities of
substitution in the industry are very small, the result will be a decrease of
the price of both outputs (producers receive less per unit produced). This
price drop is greater when imports are not reduced by the government,
6 The price of sheep was not available. Weighted estimates of real prices of meat and
wool (based on prices from 1970 to 1987) were used to estimate the supply response.
A one-year lag was included that was significant. For a review of sheep supply
elasticities, see Valdivia (1990).
New Issues 403
which controls the import licenses. If imports of meat fall, the retail and
producer price of sheep meat will not decrease as much as with the fixed
quota. In other words, an increase in the domestic supply must be
accompanied by a reduction of the meat import quota to avoid a sharp fall
of sheep output prices.
Producers will be better off if imports fall, and if the elasticity of

substitution is high, because industry will substitute the cheaper (and in
this case, local) input. The value of the elasticity of substitution is crucial
when measuring the effects of technological change. The greater the
possibilities of substituting one input for another in a production process,
the more the cheapest one will be used. In this case, an innovation that
increases supply of the farm product will also increase demand. The price
fall, resulting from the increased supply, will encourage use of this product
in a larger proportion than the relatively more expensive inputs. When the
quota is in place, a drop in sheep price will be greater than the rise in its
production; as a result, total revenue will fall, leaving producers worse off.
About direct versus macro (indirect) policies
Meat and wool are products regulated under different government

policies. Sheep meat is a commodity of the food industry, and is affected
by the government's food policies. Wool is used in the textile industry to
produce yam and fabrics as final products, and as inputs for the apparel
industry. Thus, wool is affected by textile policies which, in turn, are
influenced by the government's attitude toward protecting or developing
a domestic industry. The net impact of these two policies is uncertain,
given that production is joint and because the weight and effect of each
factor are unknown. Therefore, all policies that impact the small-ruminant
multimarket system need to be modeled in order to determine the impact
on sheep producers.
Policies that directly influence the small-ruminant sector are generated

in the retail meat market and in the textile industry. In Peru, the prevailing
policy of the meat market from 1985 to 1989 was to regulate the meat
prices through import quotas at a highly subsidized exchange rate. The
objective was to keep meat prices stable when aggregate demand for
sheep meat increased. The main policy affecting the textile sector was
aimed at exports. The tax credit and the export subsidy incentives were
designed to ensure supply of wool. Today, the subsidies have been
eliminated, although the government has been unable to devaluate the
domestic currency.
The effectiveness of these policies should be determined, assuming

other economic measures that have an effect on the whole economy are
in place. These other economic measures are called macro-price policies,
in this case exchange-rate policies. Such policies are constantly used by
the government of Peru, and counteract other policies specifically intended
to benefit small-ruminant producers. The overvalued exchange rate and
the cheap-labor policy (based on wage controls) have favored cheap
imports of food and inputs for industry. The recession in the early 1980s
caused the government to shift its policy to increase aggregate demand
through higher income levels and food price controls.
Past policies have distorted market signals, contributing to the

stagnation of agriculture and making the traditional forms of production
unsustainable. Prices have been distorted with food and input subsidies
and with overvalued and subsidized exchange rates for food imports, all
of which has discouraged commercial sheep production. Given that this
sector includes foreign trade (imports of sheep meat from abroad and
exports of wool and fiber textiles and apparel), policies that influence the
prices of tradeable commodities also have an impact on both commodities
of this sector.
Government policies over the last 20 years have had a negative impact
on sheep production; both wool and meat production have declined
according to official statistics. There have been two land tenure changes
in the last twenty years. During the 1970s, an agrarian reform turned the
large sheep-producing estates into cooperatives; following this, a
restructuring process, originated by peasant land invasions, destroyed
many of the cooperatives and their commercial production capacity in
southern Peru. There have been only two policies specifically aimed at
sheep producers: an export tax on raw wool throughout the 1970s and
1980s and a policy implemented during the 1980s to prohibit the export of
raw wool until domestic demand was satisfied, while prohibiting the
importation of raw wool.
Meat policies. During the 1970s and 1980s, two meat policies, opposite
in nature, directly affected the retail market but not the farm; these were
the rationing of red meats during the 1970s, and imports to satisfy the
domestic demand during the 1980s. These two different approaches had
the same goal: to keep consumer prices low and stable. A subsidized
exchange rate for meat imports was crucial for reaching this objective
under the second policy. The subsidy was roughly 50% of the value of the
exchange rate in national currency.
New Issues 405
Real meat prices did not increase during this period. Because prices
were stable, real income grew in 1986 and 1987, and imports increased
significantly to satisfy the demand in Lima. However, poultry is the only
meat industry that grew in the last 20 years because imported inputs have
been subsidized. Thus, poultry became the most important source of meat
in the country. The price of sheep meat relative to poultry was kept lower
than the relative beef-poultry price. While Lima's consumption grew at a
rapid rate in the late 1980s, officially recorded sheep slaughter fell
drastically, indicating the substitution of imported sheep meat for domestic
production.
Textile policies. Policies aimed directly at the textile industry were

formulated to protect its development. During the 1970s and 1980s, such
policies attempted to produce low stable input prices of wool, labor, and
credit; also, they sought a secure supply of inputs with export incentives
during periods of depressed domestic demand. Export was considered
the solution to the depressed domestic market for textiles during recession
periods, so tax credits were used to promote exports, especially when the
domestic currency was overvalued.
The price paid to producers for wool increased from the mid-1970s to
the beginning of the 1980s, as the international market expanded, and then
fell during the early 1980s recession. New export incentives and the real
income growth of the mid-1980s increased both the demand for wool
textiles and the real price of wool. A steady demand for wool, combined
with the importation of sheep meat, increased the importance of wool (as
compared to meat) in the revenue of producers, although the total real
revenue was lower.
About market-oriented policies and restructuring
Two scenarios were modeled: one with government quotas for retail
meat imports, which established a wedge between the domestic and
international price; the other with free trade in both textiles and meat, with
retail prices determined by the international market. A quota on meat
imports reflected government policy throughout the 1980s, while free trade
and promotion of market-oriented policies is the new policy of the 1990s.
Quota scenario. A decrease of the meat import quota, an increase of

textiles prices (through export subsidies), income, and poultry price were
simulated. Table 1 shows the direction of change in each market: textiles,
meat and farm production. In general, the results show that sheep
producers benefited from a reduction of meat imports and an increase in

the prices of textiles, sheep meat, and other meats such as poultry.
Income increase is effective only if the import quota does not change, i.e.,
when restrictions on imports exist, as occurred in 1986. Income rose and
with it the import levels of meat. In all cases, a policy introduced in one
final market had repercussions in the other final market. Low sheep supply
response, difficulties to substitute inputs, and final demand response to
prices, determined the size of the change of input use (farm outputs), and
prices.
Table 1. Free trade of textiles when sheep meat import quotas exist Effects of reducing
meat import quota by 1% or increasing income, price of poultry, feed
availability and textile price, also by 1 %1
Policy action
Affected Meat quota Income Poultry Animal Textile

quota increase price feed price
decrease increase availability increase
Textile market
Retail demand = + _
Retail supply + + + + +
Exports + - + + +
Other input quantity + + + + +
Other input price + + + + +
Wool demand + + + + +
Wool price - - - - +
Meat market
Retail demand - + + + +
Retail supply + + + + +
Meat price + + + - -
Carcass price + + + - -
Carcass quantity + + + + +
Other quantity + + + + +
Other input price + + + + +
Farm producers
Sheep herd + + + + +
2
Sheep revenue + + + 3 +
1 A positive sign (+) indicates increase in value; a negative sign (-) indicates a decrease, and an equal sign
2 (_) indicates no change.
3 Best outcome.
Revenue fall greater than production increase
New Issues 407
When import quotas increase, as they did in 1986 and 1987, consumers
and importers benefit and sheep producers are worse off. Given the data
and the setting of the simulation model, price or revenue per sheep fell
.51 % for each 1 % increase of the import quota, and numbers of sheep
increased .06% (Valdivia 1990). Wool production fell and its price rose
.01 % during the late 1980s scenario. Different elasticities of substitution for
the textile and meat industries were used under the assumption that
substitution of inputs is greater in textile production than in meat
processing.
Trade liberalization. As in the previous model, the same policies and

changes in the exchange rate were simulated assuming free trade in both
markets; prices were determined in the international market. Table 2
shows the direction of change in the free trade context.
Table 2. Free trade in meat and textiles markets: Effects of a 1% increase in the price
of meat, textiles, feed availability, income and the exchange rate devaluation
Meat Feed Textile Income Exchange

price availability price General Specific
Textile market
Retail demand
Retail supply
Exports
Other input quantity
Other input price
Wool demand
Wool price
Meat market
Retail demand + + -
Retail supply + + + _ + +
Imports - - - + - -
Wholesale meat quantity + + + _ + +
Wholesale meat price + _ + +
Other input quantity + + + _ + +
Other input prices + + + _ + +
Farm producers
Sheep herd + + + _ + +
Sheep price + - + _ + +
A positive sign (+) Indicates increase in value; a negative sign (-) Indicates a decrease, and an equal sign
(=)indicates no change.
General: consists of a change in the exchange rate affecting imports of meat, exports of textiles, and
imports of inputs for poultry production.
Specific: Considers only a change in the exchange rate affecting textiles and meat as set by the
international market.
Results differ when policies are introduced into a market system where
both retail market prices are determined exogenously (by the international
market). Table 3 indicates percent changes in the different markets when
exchange rates change. The extent of the change is a result of the values
of the parameters identified in each market.
Table 3. Free trade for meat and textiles: Impact of a 5% devaluation of the selective
exchange rates used for textile exports and sheep meat imports
Endogenous Sheep retail meat Textile exchange exchange

variable exchange rate2 rate3
Textile sector
Retail demand 0.00
Retail supply 0.75 0.40
Exports 2.00 25.80
Other input quantity 0.45 6.50
Other input price 0.20 1.45
Wool demand 0.75 0.40
Raw wool price -0.15 8.00
Sheep meat sector

Retail demand -5.80
Retail supply 0.85 0.40
Imports -13.55 -0.50
Wholesale carcass 13.95 -0.60
Sheep herd 0.75 0.40
Sheep produce's
average revenue 6.90
Devaluation of the domestic currency with respect to the US dollar.

Devaluation of the exchange rate for the export of textiles. This is equivalent to promoting exports with tax
credits.
Elimination of subsidies to the exchange rate for meat importation.
Trade liberalization with adjustment of the exchange rate. Includes an impact on poultry price, sheep retail
meat and textile exchange rates.
When the price of meat increases, producers benefit. For example,

given the market shares of 1986-1987 and industry production functions
with different elasticities of substitution, the model showed that producer
price of sheep meat rose 2.8% with a 1 % increase of the retail meat price
paid in national currency7. Domestic production substituted for imports
7 These values were obtained from the simulation model when the production and market
parameters of the 1980s were used.
New Issues 409
of meat, so the resulting fall in consumption was felt in the import sector,
but not in the retail meat industry. There was also a positive effect in the
textile industry, because the increase in meat supply (due to fixed joint
production) also increased wool supply by.2%, making its price fall by.03
percent. When the textile price increased, the effects were similar to the
quota scenario.
Both wool and meat supply increased, and more domestic retail meat
was available because of the textile policy. When the domestic currency
was overvalued, an increase in real income in the free trade scenario was
not good for sheep producers. The income increase induced more
imports, and textile exports shifted to the domestic market, as actually
happened in 1986 and 1987. This happened because free trade was
introduced in a context where imports were made relatively cheaper with
respect to domestic meat as a result of the subsidized exchange rate. The
results differed when subsidies to the exchange rate were eliminated, as it
took place in 1990. The increase in the price of poultry also encouraged
more sheep meat imports under free trade but had no impact on the rest
of the economy. This happened because prices of imports were cheaper
in relation to domestic products (domestic currency overvaluation).
Comparisons. Under both scenarios, a policy that increases the price

of textiles has a positive effect on the textile industry and benefits sheep
producers. The increase in the supply of meat results in lower prices of
carcass and retail meat. When import quotas are fixed, domestic demand
for textiles falls and the consumption of domestic meat rises. An incentive
policy on textiles has a positive effect on the retail meat market.
Policies that increase real income will only have a positive effect on
production in both industries if the quota on imports remains fixed. A
decrease in poultry subsidies will not only benefit the retail meat market
but also will benefit textile producers when imports of meat are restricted,
either by quotas or by a devalued (closer to a real) exchange rate.
Table 4 shows that meat and textile price increases benefit farmers more
in the free trade than in the meat quota scenario. The change in total
revenue (sheep revenue multiplied by the number of animals) is stronger
in the free trade scenario. Income increase at the retail level benefits
farmers in the meat quota scenario because imports cannot grow more.
The strongest policy benefiting producers takes place with the devaluation
of the domestic currency because it affects both final markets, increasing
textile exports and reducing meat imports.
Table 4. Sheep producer revenue under a free trade setting and under the meat quota
setting: Increase in prices, income, and farm inputs (%) and domestic currency
devaluation1
Free trade setting
Meat Feed Income Textile Exchange

price quantity price rate
devaluation2
Sheep herd 0.15 0.25 0.00 0.08 0.23

Sheep revenue 1.38 -0.19 0.00 0.75 2.13
Meat quota setting
Quota Feed Income Textile Poultry

decrease quality price price
Sheep revenue 0.51 -0.34 0.85 0.70 1.28
' Figures shown are total elasticities.

Z Devaluation of the domestic currency had the strongest impact.
About indirect or macropolicies: The exchange rate and income
A devaluation of the exchange rate (more Peruvian currency per dollar)

increases the price of tradeable goods (in this case, retail sheep meat and
textiles); thus, consumers are worse off. However, it should be kept in
mind that both products are price-elastic and that substitution possibilities
exist: this means that neither of these consumer goods are necessities in
Peru. In this context, non-tradeable products, usually produced in the
peasant economy, become somewhat cheaper. This may stimulate the
demand for non-tradeable agricultural products, moving consumption away
from products containing imported inputs.
The greatest price increase in sheep products takes place when

domestic currency is devaluated 1 %, benefiting both retail markets;
domestic meat is substituted for imported meat, and textile producers
export more wool to the international market. These are important results,
given that the real exchange rate is overvalued (BCR 1988), and that food
subsidies are expensive for the government.
Income is increased by raising the minimum wage rate (which is set by

the government). This wage increase affects sheep production in more
than one way. The first is that the demand for both commodities (retail
New Issues 411
meat and textiles) will rise. This will benefit farm producers if the demand
is of domestic production. In the meat market, there are two ways to
ensure demand of this type: by restricting imports and by eliminating
subsidies to the exchange rate used for food imports. Even if only one of
these measures is taken, the benefits are received by the importers of
meat.
A wage increase raises the relative price of other inputs with respect to
wool and fiber in the textile industry, and raises the price of labor with
respect to carcass meat in the meat industry. The result is ambiguous in
this case. On one hand, substitution induces the use of wool and carcass
meat; on the other, production drops because costs of production have
increased.
Effective policies. In a free-trade context, sheep producers' revenues

are increased more effectively by policies that increase the domestic retail
meat market than by similar policies in the textile industry. This is because
substitution of inputs in these two industry production functions are
different (with input substitution in meat production being smaller than in
textile production). If single-price policies are introduced, these are more
effective in the retail meat market (either under the free-trade or the quota
system).
The Peruvian government traditionally maintains low food prices by

using an overvalued exchange rate, thus putting pressure on the use of
scarce foreign-exchange resources. If meat imports do not receive a
preferential, heavily subsidized exchange rate, sheep producers and both
industries benefit.
Given that retail sheep meat consumption is price-elastic and can be

substituted with other meats, the burden to consumers is not extreme: So
far only importers and some consumers (mainly the middle-income class
of Lima), have benefited from government policies.
Macro-price policy, specifically the exchange-rate policy, has a great

impact on the whole small-ruminant sector. The overvalued exchange rate
policies have a negative impact on this sector, while moving toward a real
exchange rate will benefit all producers. A devaluation of the exchange
rate, or a decrease in the subsidies on the exchange rate to import sheep
meat, will decrease, and even stop, imports (as occurred in 1980 and
1990), benefiting the small-ruminant sector at greater savings to the
government, rather than imposing import quotas. Import quotas are costly,
and do not generate government revenues.
LESSONS AND IMPLICATIONS
This research was initiated with the idea that government policies
favoring the development of agriculture may be adversely affected by other
policies that the government designs for the whole economy. The
simulation scenarios showed that the positive effects a direct policy may
have on sheep producers may be counteracted and sometimes nullified by
macro or global policies. Policies favoring development of sheep
production per se do not exist, but existing policies affect the final markets
for sheep meat and wool. The exchange-rate policy was found to be most
effective in terms of sheep revenue increase and government foreign
exchange earnings. The effect of this policy is stronger than the effect of
any individual direct policy. A subsidized or overvalued exchange rate will
have a greater negative impact on the sector than a single policy directed
at promoting the consumption of either retail product. In this context, an
open economy, market-oriented policies and structural adjustments will
benefit producers of exported, as well as imported, commodities. In the
case of sheep producers, demand for both.wool and meat would grow if
the government eliminated the exchange rate overvaluation.
There are no specific policies directed to increasing consumption of

domestically produced sheep meat, The cheap-food policy implemented
by the government has been harmful to the farmers (although it is now
being eliminated). A subsidized exchange rate encouraged sheep meat
imports at a time when the government was increasing real income to
stimulate aggregate demand. Policies in retail meat were designed to
subsidize the consumption of imported meat and the production of
substitutes such as poultry. As a result, domestic meat has been
substituted by poultry and imported sheep meat. The textile industry, a
demander of wool, was favored by export tax credits. Wool export taxes
and export prohibition were enacted to protect the development of this
sector. The exchange-rate overvaluation that existed discouraged textile
exports.
The nature of the existing policies such as cheap-food price policy is

explained by the need for government stability. Meanwhile sheep
producers, a very large part of the rural population, live in extremely
difficult and uncertain conditions. The reduction of sheep imports will not
reduce the stability of government, but will help stop the process of
impoverishment that sheep producers face.
Studies to understand farm-household reactions to the market signals

are necessary to accurately ascertain responses to policy. Case studies
contribute to this understanding. The survey of rural homes conducted by
New Issues 413
the Ministry of Agriculture (1986) should provide the necessary information

to meet this end. Consistent statistics are also necessary to understand
the behavior of the aggregate and for policy formulation.
This study shows that Peru's policies have been diverse, in many cases
in conflict witheach other, and usually unpredictable. Producers have
survived by reinforcing their behavior rather than by adapting to these
confusing signals. It is necessary to design strategies to implement
technologies in a given policy context.
The opening of developing economies will have diverse impacts

depending on the characteristics of each country. Policy impact
assessment allows an ex-ante determination of how things will change,
therefore guiding the farming systems researcher towards an efficient use
of resources.
The policies analyzed for Peru reflect the shifts in government strategy
and are the reason why production of meat and wool have declined. This
uncertain policy setting must betaken into account when technological
alternatives are being developed: alternatives should be flexible to help
farmers respond to market signals in uncertain conditions. Being able to
understand the interactions of the system and the effects of policy on the
various components of the system allows for an active role in developing
policy support systems (Norman 1991).
On the other hand, with a move toward market-oriented policies and

structural adjustment, the economies of developing countries will be
affected by international markets and macroeconomic policy. In many
countries, especially those of Latin America, the possibilities of influencing
policy by agriculture are very slim. Therefore, policy assessment, as
illustrated in this paper, should help in designing technologies to be tested.
The simulation of policy scenarios helps explain why producers have not
been willing to adopt new alternatives.
The "external conditions" must change for them to be willing to modify

their production practices and allocation of resources. FSR contributes to
the welfare of producers by increasing the flexibility of the system through
the development of several alternatives that become valid as external
conditions change. Policy analysis, on the other hand, contributes to
efficient investments in FSR, by pointing out those areas in the market
(processing industries are an example) that are weak and need to be
developed.
The move toward market-oriented economies implies the elimination of

controls on and subsidies to marketed commodities. Multimarket policy-
analysis shows that sheep producers would benefit from an open economy
and structural adjustment if the government is capable of devaluating the
domestic currency.
LITERATURE CITED
BCR (BANCO CENTRAL DE RESERVA DEL PERU). 1988. Boletin del

Banco Central de Reserva del Peru. Lima, Peru. BCR.
BRAVERMAN, A.; HAMMER, J.; GRON, A. 1987. Multimarket analysis of

agricultural price policies in an operational context: The case of Cyprus.
The World Bank Economics Review 1:337-356.
DE JANVRY, A. 1981. The agrarian question and reformism in Latin

America. Baltimore, Johns Hopkins Press. 311 p.
DE JANVRY, A.; SADOULET, E. 1987. Agricultural price policy in general

equilibrium models: Results and comparisons. American Journal of
Agricultural Economics 69:230-246.
GARDNER, B. 1987. The economics of agricultural policies. New York,

McMillan Publishing Co. 387 p.
GILLES, J. 1980. Introduction. In Andean peasant economics and

pastoralism. Columbia, Missouri, University of Missouri, Small-Ruminant
CRSP Publication 1. 135 p.
HART, R.D. 1995. Role of livestock in the design of sustainable land-use

systems. In Animal Production Systems: Global Workshop. M.E. Ruiz,
C. Sere, H. Li Pun (Eds.). San Jose, C.R. International Development
Research Centre and Latin American Animal Production Systems
Research Network. IICA-RISPAL, IDRC. p. 373-384.
HAYAMI, Y.; RUTTAN, V. 1985. Agricultural development: An international

perspective. Baltimore and London, The Johns Hopkins Press. 506 p.
HERTEL, T.; TSIGAS, M. 1988. General equilibrium analysis of supply

control in U.S. agriculture. Unpublished paper. Indiana, Purdue
University, Department of Agricultural Economics. 34 p.
New Issues 415
JAMTGAARD, K. 1984. Targeting development: A production system

analysis. In Social sciences in international agriculture research. C.
McCorkle (Ed.). Boulder, Colorado, Lynne Riener Pub. p. 195-212.
JUST, R.; HUETH, D.; SCHMITZ, A. 1982. Applied welfare economics and
public policy. New Jersey, Prentice Hall. 491 p.
MARTINEZ, D. 1986. La ganaderfa en el Peru. Lima, Peru, CIDEP. 78 p.
MINISTRY OF AGRICULTURE. 1986. Encuesta nacional de hogares

rurales: Resultados definitivos. Lima, Peru. Ministry of Agriculture,
Oficina Sectorial de Estadistica. PADI AID-Peru Project 527-0238.
NORMAN, D.W. 1991. Household economics and community dynamics.

In Crop-livestock systems for the Bolivian highlands. C. Valdivia (Ed.).
Lubbock, Texas. p. 1-12.
REARDON, T. 1984. Agricultural price policy in Peru. Berkeley, University

of California, University Microfilms International, Ph.D. dissertation.
SCHUH, G.E. 1990. Sustainability, marginal areas and agricultural

research. Development (Journal of the Society for International
Development) 1990(3/4):138-143.
THORP, R.; BERTRAM, G. 1978. Peru 1890-1977: Growth and policy in

an open economy. New York, Columbia University Press. 475 p.
TIMMER, P.; FALCON, W.; PEARSON, S. 1983. Food policy analysis.

Baltimore, The Johns Hopkins University Press. 301 p.
TOLLEY, G.; THOMAS. V.; MING WONG, C. 1982. Agricultural price

policies and the developing countries. Baltimore, The John Hopkins
University Press. 242 p.
VALDIVIA, C. 1990. Impact of government policies on the small-ruminant

sector of Peru. Unpublished Ph.D. dissertation. Columbia, University
of Missouri, Department of Agricultural Economics. 316 p.
GENDER ISSUES IN ANIMAL PRODUCTION
SYSTEMS RESEARCH
July Leesberg'
INTRODUCTION
Women play an important role in animal production as well as in crop

production. In Botswana, data show a 100% involvement. in routine
management, input purchase, and sale of small ruminants (Beerling 1989),
while in Yemen women spend 6.5 to 10 hours daily caring for cattle and
small ruminants (Maarse and van Schoot 1989). Trivelli Frenzolini (1983)
mentions that 26% of the farm income in Chile is brought in by women
through the sale of milk, cheese, chickens and eggs alone.
Farming systems research (FSR) is of recent date, and its focus has
been more on crop production than on animal production. Animal
production systems research (APSR), as used in this paper, refers to
animal production research done in a systems context, rather than per se,
and offers excellent opportunities for including women's interests in
livestock research for development purposes.
This paper does not provide a complete overview, but is meant as a

discussion paper for the Animal Production Systems Workshop. It aims
first at showing the importance of gender issues in APSR, and second, at
discussing some methodological issues and formulating guidelines that are
of crucial importance for gender-sensitive APSR. A short overview of the
growing gender awareness in APSR is given by means of a literature
search on the importance of gender roles in livestock production. Later,
the variability of gender roles in livestock production is illustrated by means
' Department of Gender Studies in Agriculture, Agricultural University, Wageningen, The

Netherlands. Presently at the International Agricultural Centre, Wageningen, The
Netherlands.
of three case studies selected to show varying impacts of specific gender

issues on APSR. From these three examples and other experiences,
issues of particular relevance to gender-sensitive APSR are discussed in a
broader context.
WOMEN AND LIVESTOCK IN THE FARMING

SYSTEMS CONTEXT
Although a considerable amount of information exists on "women in

livestock production," women do not seem to be systematically included
in the animal production systems component of FSR. Official censuses
rarely represent correctly women's participation in either crop or livestock
production. The invisibility of women in statistics has been demonstrated2,
pointing to the need to improve census methods and to develop methods
that measure women's actual participation in production more objectively
(Deere and Leon de Leal 1982; Safilios-Rothschild 1985).
Fresco (1985) stressed the crucial role that women play in food
security, including crop and animal production. She conferred a role to
FSR in the investigation of this matter3. However, Fresco remarked that,
to date, there has been little systematic thinking within FSR, either on the
identification of intrahousehold resource allocation or on methods to take
into account women's production patterns and knowledge of agriculture
and methods to screen technology for negative effects on women.
Experience in the development of gender-sensitive methodologies for

research, monitoring, and evaluation has been, gained and documented.
2
The 1972 Peruvian census includes 2.6% of rural women of Cajamarca in the
occupational category of agriculturalists. In the 1976 Cajamarca Peasant Family survey,
only 4% of the principal female household members declared their occupation to be
that of agriculturalist; animal care fared somewhat better, with 12% of the women
declaring that they were shepherdesses or milk maids. Improved inquiry methods used
in Cajamarca indicated that in 46% of the households women were the primary or the
secondary person charged with crop production, and in 95% with animal care (Deere
and Leon de Leal 1982).
3
Farming systems research, since it has shown a new appreciation of the traditional
farming techniques, may potentially overcome some of the inadequacies of past
agricultural research, especially with respect to women. There is no doubt that the
utilization of research results will always affect women, either directly as agricultural
producers or indirectly through the impact of the division of labor and the allocation of
resources and income within and between households (Fresco 1985).
New Issues 419
A collection of interesting findings related to gender issues in FSR was

published by Poats et al. (1988). A framework for the analysis of gender
roles in development projects in developing countries was designed by
Overholt et al. (1985), and was eiaborated in more detail for the FSR
context by Feldstein and Poats (1989).
Most of the production systems research thus far documented is on the

subject of women in crop production. In contrast, the issue of women in
livestock has long been neglected. However, since 1980 a great number
of studies outside the FSR context have focused on "women and livestock"
(Chavangi and Hanssen 1983; Martins 1994). Interesting observations have
been made on women's activities in livestock production and on the
inclusion of women in livestock research (Fernandez 1994). These
publications offer promising opportunities for the formulation of some
guidelines for future APSR. However, before this is done, some case
studies will be discussed.
GENDER ISSUES IN ANIMAL PRODUCTION RESEARCH
In this section, different roles of men and women smallholders in animal

production are presented based on two experiences: one documented by
the International Livestock Center for Africa (ILCA) in Nigeria, and the other
by the Small-Ruminant Collaborative Research Support Program (SR-
CRSP) in Peru. The implications for research will be considered in each
case.
Modernizing milk production in Nigeria
The ILCA experience with settled Fulani in Nigeria showed the need to
pay attention to the differential access to and control of resources for men
and women, and to the production objectives of the different livestock
producers. Examination of this case will show that small farmers do not
always opt for higher milk production and that men and women may have
conflicting views on the topic.
Implications for research. ILCA's trials concentrated on purchasing

supplementary feeds and/or paying for the establishment of small
improved pastures (Waters-Bayer 1985). The objective was to improve
milk production, a goal considered to be to the benefit of women, since

they control milk processing and marketing. A collection center for fresh
milk was established near the village being studied.
ILCA originally recommended that only cows in their last third of

gestation and lactating cows should be given the supplementary feeding
or improved grass at the end of the dry season. However, most of the
collaborating Fulani included other animals in the trial, especially the
weaker animals (Taylor-Powell and Suleiman 1984, cited by Water-Bayer
1985). Milk offtake from the supplemented animals did not increase signifi-
cantly, but there was a highly significant increase in calf survival rates
(Otchere 1984, cited by Waters-Bayer 1985).
The pattern of resource control and decision making within the Fulani
household is such that innovations requiring cash expenditures for herd
inputs are not likely to result in increased milk production. Even when
animals are owned by the wives, men are responsible for the herd and for
its feeding, including investments in supplements, expecting women to
make little or no contribution to herd expenditures. As herd managers, the
men wanted to see results from their investment in terms of herd
improvement and growth. Since men control milk allocation, they can
direct any increase in milk yield into animal production, by leaving more
milk for calves. The rest of the milk is sent to the women, who control milk
processing and sales of milk and dairy products, as well as the income
derived therefrom.
Consequently, any additional income gained from increased animal

sales would accrue primarily to the men. This did not necessarily mean
that women and children could not benefit from the investments made by
men. With the technology promoted by ILCA, it was expected that milk
production would drop less in the dry season; in addition, lower mortality
rates and higher fertility rates promised more lactating cows in a few years,
which could benefit women and children in the medium term (Waters-Bayer
1986). The innovations, however, may not actually bring the expected
increase in dairy production nor the benefits foreseen for women.
Another aspect of the trial was highlighted by Waters-Bayer (1985):

although the milk collection center was close enough, the women did not
use it; they expressed no interest in selling the milk to the center as long
as it paid such low prices for their milk. By making butter and nono (sour
milk), and selling it with fura (a spicy dumpling usually made of millet), the
women could make up to four times the price offered for their milk at the
collection center. They explained that they would benefit more from the
New Issues 421
introduction of grain processing mills, which would save time in the

preparation of fura, as well as from good roads, which would facilitate their
marketing activities (Waters-Bayer 1986).
Waters-Bayer concluded that the promotion of smallholder dairy

development should aim, not at replacing the existing processing and
marketing system, but at increasing productivity of the pastoral herds.
Given the difficult environmental situation of Fulani producers, it will take
several years before any increase in herd productivity will result in substan-
tially higher milk offtake levels. Research must be designed with this in
mind.
If dairy development is intended to benefit small-scale producers and

low-income consumers, it should be designed on the basis of existing milk
production, processing and marketing systems. Smallholder dairy
schemes must be built on knowledge of how the existing production
systems work and on their underlying rationale. All household members
with decision-making capabilities should be taken into consideration and
be actively involved; in the Fulani case, this includes women.
Simple techniques for dairy production improvement were proposed to

women. However, they expressed more urgent needs for better roads (so
that milk products could be sold at more profitable markets) and the need
to ease the strenuous work of pounding the cooked fura (Waters-Bayer
1985, 1986).
On-farm alley farming research in Nigeria
ILCA's experience in alley farming (arable crop-based farming with

fodder trees and food crops interplanted) illustrates the advantages of a
holistic approach to animal production research, including such issues as
the access to and the control of land and trees, the need for the definition
of subgroups within a recommendation domain, the division of labor within
a production system, and finally, what daily and seasonal labor allocation
patterns are used by each of the different participants in livestock
production. It is concluded that the inclusion of a woman on the research
team has greatly improved the active participation of farming women in the
research.
Implications for research. The objective of ILCA's research in the

humid zone of Nigeria is the improvement of smallholder sheep and goat
production systems. Much of this work has been based on the alley
farming system, which seeks to improve the availability and quality of feed
produced. In the extension phase, the project aimed to demonstrate new
techniques associated with alley farming and to supply interested farmers
with seeds necessary to establish the crops.
It was mostly men who were present at the early meetings. Although
women were also specifically requested to attend, there was little response.
Women could not leave their productive and reproductive tasks as easily
as men could. Of those women who appeared at the meetings, few
planted alley farms (Okali and Cassaday 1985, cited by Francis and Atta-
Krah 1990). At the end of the first planting season in 1984, only 12 women
(27% of the female population) had established alley farms, as compared
with 56 men. This was surprising, since it was shown that alley farming
would produce more feed for animals and it was known that a large
proportion of the small ruminants in the zone belonged to women (Okali
1979, cited by Francis and Atta-Krah 1990); in addition, the women, as
dairy-food processors, are responsible for much of the supplementary
feeding of the animals.
ILCA changed its methodology for research and extension by

employing a female research associate. She investigated women's time
allocation, budget, decision-making processes within the household,
ownership of the animals, access to land, the use of family and hired labor,
and the time spent outside the village for trading purposes. Alley farms
managed by men and women were regularly monitored.
The research showed that women were strongly involved in agriculture,

the majority managing their own farms independently. However, they
seldom possessed land. They usually grew annual crops, since planting
of trees gave ownership of these trees to women even if the land was
owned by men. On the other hand, although tree cropping generally has
been a male preserve, if women asked their husband's permission to grow
the leguminous fodder trees promoted by the project, the men usually did
not object. Safilios-Rothschild (1987) suggested that this may be explained
by the interest of men in leguminous crops as a source of mulch and
fodder in the.dry season.
Women have a more severe labor constraint than men, since women
often have little cash for hiring labor and their own time has to be shared
between the household, trading, and the farm. While in 1984 only 17% of
the successfully established alley farms belonged to women, this
percentage rose to 50% in the second year. ILCA concluded that the
special extension efforts aimed at women had been effective and that, in
New Issues 423
spite of the severe labor and cash constraints on women, they were able
to manage the alley farms as efficiently as men.
This experience underlines the need to distinguish different subgroups

within a recommendation domain. Each subgroup should be represented
in the on-farm experimentation. The extension methodologies should be
adapted according to the needs and availability of each subgroup,
including the adaptation of meetings to the working schedules of the
different target groups (Francis and Atta-Krah 1990).
Participatory research with women's committees in Peru
The Peruvian example was selected because it shows very clearly how
different the interests of men and women can be. Men organize their work
around crop production, while women do so around animal production.
Furthermore, the case highlights the influence that organized women can
have on research. It demonstrates the possibility of having the research
goals formulated by the target group itself and the decisive role that
women can play in the design and execution of on-farm research.
Implications forresearch. The Small-Ruminant Collaborative Research

Support Program, the Peruvian National Institute for Agricultural Research
and Extension, and the University of Missouri have had a joint program in
the high altitudes of Jauja, Peru, since 1983 (Fernandez 1994). Their aim
is to identify technological alternatives based on farmers' knowledge and
an understanding of the traditional production system, which includes both
crop and livestock production components.
In Jauja, men are responsible for cropping; they are sometimes assisted
by the women. Although men are responsible for the health care of the
animals, livestock production is the responsibility of the women, and
children assist in overseeing the grazing animals. Farmers' organizations
consisted mainly of men, and therefore were more oriented towards crop
improvement; women were more interested in solutions for animal
production problems.
The women decided to organize themselves separately and formulated

their priorities for research in animal production along the following lines:
internal and external parasite control in sheep, forage production for use
in the dry season, improved management of communal rangelands, and
information on animal selection criteria. The women showed their interest
in working together toward the solution of these problems and organized

themselves into "Women's Agricultural Production Committees" as legal
entities within community structures. After definition and prioritization of
their production problems, the women worked towards a solution following
these steps:
gathering traditional knowledge on the problem
consulting with external specialists to complement the information
evaluation of the possible alternatives
planning of tests with the most promising alternatives
implementation and evaluation of tests
An example of how the women's committees worked through these

solutions may be seen with the first production problem, the animal health
component. The veterinarian discussed with the women which treatments
should be included in the test. They decided on two traditional herbal
treatments, to be compared with the available chemical medicine. During
one and a half years, the groups carried out tests on parasite control. It
was found that the treatment with tobacco leaf was just as effective as the
chemical product tested. The availability of the plant, however, was too
limited for dipping all the community flocks at the required intervals. The
women decided to start a plant multiplication project on a community plot.
Plans were developed to start research on the other priorities formulated
by the women (Fernandez 1994).
GUIDELINES FOR GENDER-SENSITIVE RESEARCH

IN ANIMAL PRODUCTION
Based on considerations put forth by Safilios-Rothschild (1983)4, the

basic elements needed for gender-sensitive production systems research
4 Technical innovations and improvements concerning a particular crop or livestock

cannot be examined in isolation from other crops and livestock and from the dynamics
between men and women that determine labor allocation, access to agricultural services
and resources, the nature and extent of agricultural investments and the allocation of
resulting profits (Safi ios-Rothschi ld 1983).
I
New Issues 425
may be summarized as follows: a holistic view of men's and women's crop

and livestock activities (as in the ILCA examples), an understanding of
decision-making processes among men and women (the first ILCA
example), and an insight into women's and/or men's access to and control
of production resources, including the control of income, as discussed in
the three preceding examples.
Five important issues for the development of gender-sensitive APSR will

be elaborated in this section: definitions, unit of analysis, research
objectives, methodological issues, and training and education of women.
Definitions
A correct definition of what is understood as animal husbandry is of

crucial importance. This is necessary to prevent the exclusion of women's
differential labor contribution and responsibilities in livestock production
from censuses and surveys5. Livestock activities should therefore take
into consideration all activities, from feeding and grazing the animals to
health management, processing or transformation of animal products, and
marketing of animals and their products.
Livestock researchers tend to prefer research on large animals like

camels and cattle, with a strong tendency to concentrate on improving milk
production (Safilios-Rothschild 1983). This may bias the information
obtained on women's participation in animal husbandry; although in certain
regions women have an interest in large animals, they are also found
actively involved in rearing small animals: goats, sheep, pigs, guinea pigs,
rabbits, turkeys, chickens, pigeons, and even bees. The choice of the type
of animal may be decisive in terms of who will benefit from the research
results.
In general, researchers should beware of interpreting important

concepts like animal husbandry, livestock, and farm work in such a limited
sense that it might exclude the spheres of interest of women.
5
A furthersource of underestimation of female agriculture is due to the ambiguity in the
census of what actually constitutes agricultural work.... If the other activities associated
with agricultural production, besides field work of crop production are taken into
account - such as animal care, cooking for field hands, agricultural processing and
marketing - the measure of women's agricultural involvement increases significantly
(Deere and Leon de Leal 1982).
Unit of analysis
The household, headed by individual males, who are the major decision
makers for all productive activities, has long been considered as the basic
unit for analysis (Chase 1988). However, it has now been widely
recognized not only that interhousehold relations should be taken into
account in studying the production system, but also that intrahousehold
relations may be of fundamental importance for understanding decision-
making processes at the household level (Collinson, cited by Feldstein and
Poats 1990).
The household cannot be seen as a homogeneous unit. Instead, in

many cases, individual household members (both men and women) have
productive subsystems that are clearly identifiable as their own and have
considerable autonomy with regard to labor allocation and income
utilization (Safilios-Rothschild 1987). Men and women hold not only
shared, but also conflicting interests in production and consumption.
Gender must be seen as an important category, just like age, caste, class,
and ethnic affiliation.
The household must be regarded as a system that is subdivided into

subsystems. Household members responsible for different subsystems try
to maximize their own benefits through negotiations with other members
of the household (Overholt et al. 1985; Fresco 1985; Safilios-Rothschild
1987).
The importance of shared and conflicting interests within the household

is illustrated by the Jauja example in Peru. The Women's Agricultural
Production Committee decided to carry out trials with fodder crops for the
improvement of animal production. Since cropping is a responsibility of
men, they assisted the women in the establishment of the trial. At the end
of the season, however, men considered the fodder production as theirs.
They harvested the crop and sold it on the market for their own benefit.
Women were unable to do anything about it since men controlled the
crops (Maria Fernandez 1991, personal communication).
It can be concluded that the basic unit of analysis should be the

productive subsystems; different subsystems may exist at the level of
different members of the household, and those household subsystems
have to be considered as having shared as well as conflicting interests.
New Issues 427
Research objectives
Any research must take into account both the goals and the daily reality
of the livestock producers. The majority of the livestock in the developing
countries are kept in small herds owned by smallholders (Chavangi and
Hanssen 1983). Consequently, research must deal with the specific and
complicated characteristics and dynamics of smallholder production
systems; thus, a holistic view of the production system is needed.
When introducing new techniques or tools, the important questions to

be asked (Aarnink and Kingma 1991) are: Who will control the new
techniques or tools? How will labor allocation at the household level be
affected? Who will control the output?
Usually it takes considerable time and effort for researchers to become

fully aware of the rationale behind the decision-making processes of small
farmers. Therefore, letting men and women smallholders participate in the
formulation of the research objectives may improve the relevance and
efficiency of research in the smallholder context.
Gauch and David (1983) mention the reasons why women keep
animals: for marketing; for rites, fetishes, and sacrifices; for hospitality; for
manure; and for consumption. Safilios-Rothschild (1983) explained that
women keep animals for cash expenditures to meet family needs such as
school fees, health care and social obligations. They also see their animals
as kind of life insurance and as an investment that can be cashed in when
facing adverse situations. Milk production is usually not the primary
objective for women who keep sheep and goats. Subsistence activities are
not as motivating to women as those that provide them with an income
they can control or that permit the accumulation of assets that they can
use of in times of need or take along with them in case of divorce (Safilios-
Rothschild 1983).
It has been argued by different authors that the alternatives to be tested

in research should fit the developmental trend of the existing system and
should not try to introduce sophisticated novelties, particularly if the
consequences of this introduction cannot be foreseen. For example, Wa-
ters-Bayer (1986) stated that for the settled Fulani who are modernizing
milk production, smallholder dairy development should not aim at replacing
the existing processing and marketing system, but rather at increasing the
productivity of the pastoral herds according to the goals that cattle keepers
pursue.
For sheep and goats, Safilios-Rothschild (1983) argued that while there
may be great gains to be obtained in breed selection and crossbreeding,
many times this strategy ignores the fact that unimproved small ruminants
are important for smallholders partly because of their low input and low
labor demands. Thus, in the case of resource-poor farmers, profitability
tends to decrease when attempts are made to raise productivity through
schemes calling for increased inputs (Safilios-Rothschild 1983). Such
innovations may have additional consequences since increased labor
demands (associated with the innovation) often imply greater labor efforts
for women (Safilios-Rothschild 1983); this does not necessarily mean that
women will have greater control of the income generated by these efforts.
In general, it can be concluded that research objectives cannot be

adequately defined without a proper understanding of the existing producti-
on system as a whole. Research should aim, as much as possible, at the
improvement of the existing systems and make use of local, indigenous
knowledge. The objectives should consider not only the introduction of the
techniques as such, but also important questions concerning their effects
on (a) changes in labor input, (b) access to and control of the techniques,
and (c) control of the output.
Methodological issues
It has been stated that a holistic view of the production system is

needed in order to understand its dynamics. FSR offers a methodology for
the development of this view. Ten different stages in the iterative research
process can be defined as:
1. Diagnostic phase (rapid rural appraisal)
2. Identification of constraints
3. Prioritization of constraints
4. Screening of alternatives
5. Selection of alternatives
6. Design of tests
7. Execution of tests
New Issues 429
8. Evaluation of tests
9. Formulation of recommendations
10. Extension
In FSR, once the target group and the research area are selected, the
different recommendation domains have to be defined. Recommendation
domains are groups of roughly homogeneous farmers with similar
socioeconomic circumstances, for whom similar recommendations can be
made (Shaner et al. 1982). As men and women constitute different
categories of farmers, they have to be considered as different
recommendation domains (Due 1988; Sutherland 1988). Even within the
recommendation domain concerning women, differences must be
recognized between women of different ages, as well as between women
as heads of households, as polygamous or monogamous wives, or as
unmarried women, among other groups.
Once the target groups have been decided upon, actual research may
begin. Since the study of intrahousehoid dynamics forms an integral part
of FSR (and consequently of APSR), gender analysis should be included
in the research framework used. Feldstein and Poats (1989) have
developed a useful framework for gender analysis to be applied in all
phases of FSR and APSR projects.
Gender analysis includes analyses of both men and women farmers in

terms of (1) their activities, (2) their access to and control of resources, (3)
their decision-making processes, (4) equal participation of men and women
(both farmers and professionals) in all phases of research, (5) equality of
benefits and incentives, and (6) external factors that may influence the
productive activities of men and women differently (such as religion,
culture, history, and economy).
Why include gender analysis in all research phases? No intervention is

gender neutral; each intervention generates changes in gender relations.
Continuous analysis in FSR and in APSR allows the adjustment of changes
in such manner that they do not cause disadvantages to women's
positions and do promote equal opportunities and rights among men and
women. Inclusion of gender analysis implies that interviews must be done
with both men and women (Wahyuni et al. 1987). Women, like men,
should be treated as experts in all domains that, according to the activity
analysis, appear to be their occupations (George 1990).
A cost/benefit analysis that takes into consideration the opportunity

costs of men's and women's labor investments in other types of activities
is needed for each alternative being tested (Safilios-Rothschild 1983).
Both men and women should be included in all meetings and in the
research process. Sometimes this means that separate meetings will have
to be organized, since men and women have different interests (Fernandez
1994), different networks, and different working schedules (Francis and
Atta-Krah 1990).
The opportunities created by the inclusion of female members on the

research team is also recognized by Safilios-Rothschild (1983), who stated
that it may be of help if some of the project's expatriate and local staff are
women, in order to ease communication with the farmers. Several positive
effects can be obtained from working with groups of women. In addition
to improving research aims, the groups that were established in Jauja,
Peru (Fernandez 1994) obtained legal status, thus creating an opportunity
for women to operate at the public level in the village and to receive
services that until then had been only available to men since, they
performed the public representation in general.
Livestock production has some specific characteristics which make

animal production research more difficult than research in crop production.
Livestock production is a complex matter, and men and women owners
are not eager to give information on the topic. Time allocation studies may
reflect in some degree the differential importance of certain crops for men
and women. This is not the case for animal production; the person who
takes care of the animals (and who supplies the labor) very often is not the
owner of the animals.
In Mali, the author found that farmers spent hardly any time on livestock
production, although their herds were of considerable importance
(Leesberg et al. 1990). The animals were entrusted to different Fulani
shepherds, who took them out in transhumance during most of the year.
A farmer would never tell anyone how many cattle he or she owned. The
animals were kept mainly as savings, used for occasions when cash was
needed (birth, funerals, taxes, marriage). Thus, animal production is a
complicated matter to deal with and should be considered within the
context of the whole production system, including crop production, off-
farm and household activities, and a continuous gender analysis.
New Issues 431
Training and education of women
On the training of farming women, Safilios-Rothschild (1983) remarked

that the comprehensive training of women plays a crucial role in the
organization of cooperatives, increasing their ability to control their projects
and to gain institutional legitimacy; thus, training should be taken into
consideration in development planning. A higher level of education should
be available to disadvantaged rural women. This would include animal
health care, bookkeeping, development of managerial and organizational
skills, marketing, handling of credit, and other areas of expertise.
At the professional level, Oxby (1983) argued that one of the reasons
why women are not able to participate fully in the design and
administration of projects is that educational qualifications in a branch of
animal sciences are required, or at least advantageous. Relatively few
women possess such qualifications. This is why it is particularly important
to make sure that women are being trained in the relevant disciplines.
Beerling (1991) stressed the importance of female staff members for

reaching women in the area of livestock production. She remarked that in
the Northwest Frontier Province of Pakistan almost all rural households
own some animals, and that most of the everyday management of
livestock is done by women. Extension efforts on livestock management,
however, have never reached the female farmers.
In the religious and sociocultural context of the Northwest Frontier

Province, women cannot have direct contacts with men, yet the
Department of Livestock and Dairy Development (DLDD) has no female
field staff. Women sufficiently qualified to fulfill functions in the DLDD are
not available; therefore, it is recommended that an entirely new training
curriculum be developed for the education of female staff members and
that the extension program should be adapted to the needs and
possibilities of the rural women, who also need training that will enable
them to participate fully in research concerning animal production.
In conclusion, all members of the research teams need adequate

training in gender-sensitive methodology. Under certain conditions (for
example-in the Moslim context of Yemen, Pakistan and Bangladesh),
trained women are indispensable for reaching women farmers. However,
few women professionals in livestock sciences are available, and
specialized training is required.

FOR FUTURE RESEARCH
The FSR methodology, including gender analysis, offers a good

conceptual and methodological framework for gender-sensitive APSR.
Methodological issues that are specific to the animal production systems
context have been discussed, a summary of which is given below.
First, it has been argued that it is important to redefine the activities

included when the term animal husbandry is used. The definition should
include all animal-related activities that men and women farmers conduct:
feeding, health management, daily care, processing and marketing of
animal products. These activities should be specified according to the
different species of animals present at the farm.
The predominant interest of animal production scientists in favor of

larger animal species raises the question of whether it is necessary to
establish new curriculums in education of animal production scientists,
focusing on the socioeconomic .importance of small animal species.
In the context of smallholders, research should be oriented towards the

improvement of existing systems, while avoiding the introduction of exotic
technologies. The participation of both men and women and careful
monitoring are required when new techniques are being introduced, since
they often impose an extra labor demand on women. This is only
acceptable if women can decide for themselves whether or not they want
to supply the extra effort.
For the formulation of relevant research objectives, it is important to

understand what goals the different livestock producers pursue. It may not
be useful to improve milk production by means of considerable labor and
capital investment if animals are mainly kept as life insurance rather than
for milk production. Women often do not possess the necessary funds
and time for the establishment of a complete change in their animal
production system.
Special attention should be given to the composition of the APSR

research team. The presence of female researchers can greatly facilitate
the communication with women producers, especially in areas where
women cannot have direct contacts with men (e.g., Islamic countries).
This by no means implies that male colleagues should now confine
New Issues 433
themselves to research with male producers only. Both men and women
scientists urgently need good training in gender issues and gender-
sensitive research methodologies.
Women lag behind at the educational level. There is a need for women
animal scientists. Women at the village level should have access to
training in all kinds of skills, as well as opportunities to organize themselves
and to participate in research activities. Investments should be made in
order to overcome this deficiency.
LITERATURE CITED
AARNINK, N.; KINGMA, K. 1991. The Shamba is like a child. Leiden,

The Netherlands, Leiden University, Women and Autonomy Center.
36 p.
BEERLING, M.L. 1989. Thematic evaluation on the integration of

women in rural development: Report on a field mission to Botswana
Sheep and Goat Development Project, Botswana. EEC. The Hague,
The Netherlands. BMB Management Consulting for Development BV,
Arnhem and Femconsult, Consultants on Women and Development. 68
p. + annexes.
BEERLING, M.L. 1991. Livestock extension through women workers

in NWFP (Pakistan). Zeist, The Netherlands. RDP Livestock Services
BV. 53 p. + annexes.
CHASE, V. 1988. Farming systems research in the Eastern Caribbean:

An attempt at analyzing intra-household dynamics. In Gender issues
in farming systems research and extension. S.V. Poats, M. Schmink,
A.Spring (Eds.). Boulder, Colorado, Westview Press. p. 171-183.
CHAVANGI, N.A.; HANSSEN, A. 1983. Women in livestock production

with particular reference to dairying. Expert consultation on women
and food production. Rome, Italy, FAO. 14 p.
DEERE, C.D.; LEON DE LEAL, M. 1982. Women in Andean

agriculture. Geneva, Switzerland, International Labour Office. Women,
Work and Development 4. 172 p.
DUE, J.M. 1988. Intra-household gender issues in farming systems in

Tanzania, Zambia and Malawi. In Gender issues in farming systems
research and extension. S.V. Poats, M. Schmink, A. Spring (Eds.).
Boulder, Colorado, Westview Press. p. 331-344.
FELDSTEIN, H.S.; POATS, S.V. 1989. Working together: Gender

analysis in farming systems research. Vol. I: Case studies, Connecticut,
USA, Kumarian Press. 271 p.
FERNANDEZ, M.E. 1994. Women's agricultural production

committees and participative-research-action approach. In Tools for the
field: Methodologies handbook for gender analysis in agriculture. H.
Feldstein Sims, J. Jiggens (Eds.). Hartford, Connecticut, U.S.A.,
Kumarian Press. p. 239-243.
FRANCIS, P.A.; ATTA-KRAH, A.N. 1990. .Incorporating gender

concerns into on-farm research: The household and alley farming in
Southwest Nigeria. Ibadan, Nigeria. ILCA, Humid Zone Program.
Manuscript.
FRESCO, L. 1985. Food security and women: Implications for

agricultural research. In Femmes et Politiques Alimentaires: Actes du
Seminaire International sur la Place des Femmes dans I'Autosufficance
et les Strategies Alimentaires. J. Bisiliat (Ed.). Paris, France.
ORSTOM-Centre International de I'Enfance. 12 p.
GAUCH, A.; DAVID, A.B. 1983. The role of women in food production
with particular reference to small animals at village level. Expert
Consultation on Women in Food Production, Rome, Italy, FAO.
ESM:WIFP/83/15. 18 p.
GEORGE, S. 1990. The female of the species: Women and dairying

in India. In Development perspectives for the 1990s. Singer,
Prendergast (Eds.). Basingstoke and London, UK, Macmillan.
LEESBERG, J.; SANAGO, M.K.; DIALLO, 0.; CISSE, K.; DIONI, K.;
DEMBERE, K.; THIERO, M.; COULIBALY, M. 1990. La recherche sur
les systemes de production a ['Office du Niger: Production et
intensification dans les secteurs de Niono et Koky. Niono, Mali, Office
du Niger, ARPON Projet. 40 p. + annexes.
New Issues 435
MAARSE, L.M.; VAN SCHOOT, C.M.E. 1989. Rural women in livestock

production and rangeland use: Observations in 5 selected villages of
the Damar Montane Plains, Yemen Arabic Republic. Ministry of
Agriculture and Fisheries, Agricultural Research Authority, Yemen Arabic
Republic; and DHV Consulting Engineers, Amersfoort, The Netherlands.
BLIP Communication No. 34. 60 p.
MARTINS, C. 1990. Zur Rolle von Frauen in der Tierproduktion in der

Dritten Welt. Literature bersicht. Arbeitspapier zum GTZ-Projekt
90,9127.3-91-100 "Frauenforderung in der Tierproduktion," Germany.
OVERHOLT, C.; ANDERSON, M.B.; CLOUD, K.; AUSTEN, J.E. (Eds.).

1985. Gender roles in development projects: A case book. West
Hartford, Connecticut, U.S.A., Kumarian Press. 333 p.
OXBY, C. 1983. Women's contribution to animal production and

husbandry. World Animal Review (FAO) 48:2-11.
POATS, S.V.; SCHMINK, M.; SPRING, A. (Eds.). 1988. Gender issues

in farming systems research and extension. Boulder, Colorado,
Westview Press. 450 p.
SAFILIOS-ROTHSCHILD, C. 1983. Women in sheep and goat

production and marketing. Expert Consultation on Women in Food
Production. Rome, Italy, FAO. ESH:WIFP/83/8. 15 p.
SAFILIOS-ROTHSCHILD, C. 1985. The persistence of women's

invisibility in agriculture: Theoretical and policy lessons from Lesotho
and Sierra Leone. Economic Development and Cultural Change
(U.S.A.) 33(2):229-317.
SAFILIOS-ROTHSCHILD, C. 1987. Farming systems and gender

issues: Implications for agricultural training and projects. Consultancy
for the Ministry of Agriculture and Fisheries of The Netherlands and the
International Agricultural Center, The Netherlands. Report. 34 p.
SHANER, W.W.; PHILIPP, P.F.; SCHMEHL, W.R. 1982. Farming

systems research and development: Guidelines for developing
countries. Boulder, Colorado, Westview Press. 414 p.
SUTHERLAND, A.J. 1988. The gender factor and technology options

for Zambia's subsistence farming system. In Gender issues in farming
systems research and extension. S.V. Poats, M. Schmink, A. Spring
(Eds.). Boulder, Colorado, Westview Press. p. 389-407.
WAHYUNI, S.; VAN KNIPSCHEER, H.C.; GAYLORD, M. 1987.

Women's decision-making role in small-ruminant production: The
conflicting views of husbands and wives. Agricultural Administration
and Extension 24:91-98.
WATERS-BAYER, A. 1985. Dairying by settled Fulani women in

Central Nigeria and some implications for dairy development. London,
UK. Overseas Development Institute, Pastoral Development Network,
Paper 20c. 24 p.
WATERS-BAYER, A. 1986. Modernizing milk production in Nigeria:

Who benefits? Ceres 19(5):34-39.
WORK GROUP
SESSIONS
INSTITUTIONAL ISSUES
A report by Working Group 1'
Terms of reference. This group discussed a range of institutional

issues such as (1) sustainability of the research effort beyond the
termination of external funding, (2) linkages of research and extension, (3)
feedback mechanisms to influence disciplinary and other research, (4)
perceived high cost of farming systems research (FSR), (5) donor fatigue
and possibilities to increase participation of farmers in terms of both
decision making and funding, (6) critical mass, (7) discipline mix and
training level of team members, and (8) the role of systems research vis-a-
vis the commonly found divergence between research and extension.
These issues were addressed using the presented case studies for
reference. The group was requested to produce a set of
recommendations particularly focused on dealing with animal production
systems (APS) projects in a manner appropriate for the new directions,
such as natural resource management and commodity system
perspective.
BACKGROUND
Practitioners of farming systems research (FSR) place high expectations

on this approach to address and resolve some of the problems usually
neglected by conventional commodity- or discipline-oriented biological
research. The balance of opinion to date is that FSR, in spite of its holistic
nature, has not been as successful as expected or claimed, given the
failure of many FSR projects.
In some cases, projects fail because they are actually conventional

projects on which the FSR approach was belatedly imposed. On the other
hand, a number of genuine FSR projects fail because certain crucial issues
E.Thompson (Chairperson), H. Li Pun (Discussion Opener), 0. B. Smith (Rapporteur),

G. Kleemann, B. Kiflewahid, E. Lindarte, F. Romero, L. Setshwahelo.
are not properly considered and addressed: policy/commodity,

methodological, and institutional issues, among others. Institutional issues
were the subject of discussion in Working Group 1. A summary of the
group's views, conclusions, and recommendations is set out below.
INSTITUTIONALIZATION OF SYSTEMS RESEARCH

AND HUMAN RESOURCE DEVELOPMENT
The value of systems research for (1) the characterization of production

systems, (2) the identification of constraints, and (3) the design, evaluation,
validation, and extension of solutions is well known. Thus, the systems
approach needs to be institutionalized, and researchers and development
specialists must be sensitized and equipped to utilize this powerful tool.
Discussion on the ways to achieve this produced the following
recommendations:
International or regional centers should play a crucial role in this

process.
The systems research concept should be introduced into training

programs at the graduate training level.
Seminars, workshops, and short courses should be regularly used to

reinforce this sensitization process.
The initiation of courses on agricultural development and ecology

should be encouraged.
MULTIDISCIPLINARY WORK
The complexity of issues encountered in the field demands a

multidisciplinary approach for systems research to be successful. This,
however, generates a number of problems and constraints that need to be
adroitly addressed if teamwork, so crucial for systems research, is to be
achieved.
Work Group Sessions 441
One of these areas of concern is the size and structure of the team,
which should be dictated by the problems to be solved and the objectives
of project; other areas are team leadership, allocation of tasks, and active
participation of all team members. In this latter respect, it is recommended
that older, experienced team members assume greater responsibilities. An
intractable problem faced when attempting to use the multidisciplinary
concept is the continued organization of research institutions and
universities on a disciplinary or commodity basis, with units spatially
separated.
Under such conditions, it is often difficult to gather a team of dedicated

researchers willing to work together for a common goal, while keeping their
professional biases in abeyance. One way out of this, recommended here
for consideration, is the institutionalization of a systems unit that can enroll
commodity specialists as members of an interdisciplinary team for a well-
identified job, provided that the unit has the means to attract and motivate
participating specialists.
PERCEIVED HIGH COST OF FSR
FSR is not necessarily a more expensive proposition than specialized

on-station, commodity-based research. In fact, systems research is often
cheaper and has a greater impact. Nevertheless, it is recommended that
the cost of monitoring studies be moderated by using existing extension
services. This approach will promote and maintain vital linkages between
research and extension, one of the issues identified as a constraint to a
more effective impact of FSR. In this respect, it is recommended that
private volunteer organizations (PVOs) and nongovernmental organizations
(NGOs) be involved in these extension activities to ensure a greater level
of technology transfer and adoption.
VIABILITY OF THE NARIS
In many developing countries, the ability of the national research

institutions to carry out their mandate, as well as their viability, is being
questioned. Poor funding, a reflection of a general economic malaise, has
led to deteriorating infrastructure, loss of qualified professionals, and a lack
of incentives for the remaining scientists. Financially limited national

institutions may still manage to carry out commodity-based research,
which is rarely demand-driven; they are, however, usually in no shape to
use the FSR approach because of a lack of required multidisciplinary
expertise, motivation, and dedication.
One suggestion that has been made is the strengthening of these

institutions by downsizing, restructuring, rationalizing, and reducing their
number so that scarce available resources can be better utilized by fewer
institutions. Channeling of a part of donor funds, now allocated to
international agricultural research centers (IARCs), to national centers for
capacity building is another suggestion. This is based on the rationale that
a strong upstream-oriented IARC cannot have much impact without the
collaboration of an equally strong national center.
There is no generalized consensus on either proposal. In the last

decade many new institutional actors (NGOs, producers, and other
associations) have come to play a vital role in agricultural systems or
components research, thus increasing the number of options and making
it reasonable and desirable to analyze every case individually in light of
these institutional developments.
SUSTAINABILITY OF DONOR-SUPPORTED PROJECTS
If donor withdrawal occurs, many projects collapse and do not attain

the anticipated impact. The consensus of the group was that many such
projects are designed by donors without taking into account the prevailing
socioeconomic realities of recipient countries; their collapse after the
withdrawal of their only driving factor (donor funds) is no surprise. Another
category is the megaproject that is well beyond the financial resources the
recipient country could justifiably invest (politically or economically) in one
project.
The question of sustainability is irrelevant to other types of projects

which are finite and aimed at solving a particular problem, which once
solved does not require continued research intervention. For several
reasons, other research endeavors turn out to be failures and should be
accepted as such without undue efforts to turn them into positive research
interventions.
GENDER ISSUES
Terns of reference. There is a growing awareness of the need to

address gender issues directly for decisions on concerns such as
agricultural technology and the allocation of rural family savings and labor.
This work group reviewed past experiences in animal production

systems research and development (APSR/D) to identify cases where the
understanding of gender issues contributed significantly to project
success. Furthermore, the group developed a checklist of gender issues
of potential importance in APSRD, on the basis of type of farming system,
ethnic origin of the population, development stage and other factors. The
group was also asked for suggestions as to strategies to incorporate
gender issues in ongoing research programs.
BACKGROUND
Farming systems research and extension (FSR/E) will in the end be

evaluated by its contributions to the development of rural societies in the
Third World. In a changing world, development has expanded to include
new dimensions such as the sustainability of natural resource productivity
and equity (the distribution of the results and benefits of development
among the different social classes and strata). The gender issue is part of
this expanded agenda of development.
GENDER ANALYSIS: WHY?
The working group began by discussing the concept of gender and

agreed that it refers to the sociocultural determination of the role of the
sexes. The incorporation of the gender dimension into FSR/E can help to
achieve two goals:
C. Devendra (Chairperson), J. Turk (Discussion Opener), J. Berdegue (Rapporteur), J.B.

Schiere, J. Leesberg, F. Nyaribo, D. Norman, J. Ferguson, F. Navas, C. Sere.
1. The empowerment of women. Both the processes and the results of

technological and economic development can have an effect on
women's access to the control of resources, their self-esteem and their
opportunities for personal advancement.
research and transfer of technology.

2. Greater efficiency and efficacy in
Several studies show that the participation of agricultural decision
makers (including women) a key ingredient of good research and
is
technology transfer.
GENDER ANALYSIS: WHEN?
Economic growth with greater equity is an infrequent combination. The

promotion of both economic growth and equity requires the investment of
human resources (both genders) as well as material resources. In a
context of scarcity of resources, a conflict or trade-off may exist in many
cases.
Donors, politicians and administrators have always put great pressure

on researchers and extensioniststo produce the greatest possible changes
in family income in the. shortest possible time. This is particularly true of
animal production systems research (APSR). In a context of declining
financial support for APSR, it is important that these groups clearly define
what relative impact is expected from specific development objectives, in
terms of the resources provided.
Not recognizing the trade-off that exists between objectives will lead to
unrealistic demands placed on APSR. The recognition of these trade-offs
means that the proportion of the total effort devoted to gender analysis
should be evaluated by each project, according to objectives and mandate
defined by the donor and national agencies.
GENDER ANALYSIS: HOW?
The group strongly recommends that APSR should take a number of

basic steps to develop the necessary sensitivity to the gender problem:
1. APSR projects should include qualified female researchers on their

staffs. Female members should participate in the project from its
inception and throughout all its phases. Their participation does not
imply that the group as a whole does not share the responsibility for
gender analysis.
2. The diagnostic phase of APSR projects should include the gender

variable so that the project becomes sensitive to problems, constraints,
and opportunities related to this issue. Particular care should be taken
to identify technological opportunities that have both economic and
equity potential. The discussion group believes that animal production,
small-ruminant systems in particular, offers this combined potential in
.many countries.
3. The ex-ante evaluation of technological alternatives should include the

gender variable. Such evaluations should at least estimate the positive
and negative impact of the innovations on gender. At a minimum, no
technological alternative should be developed that is detrimental to
women in absolute terms.
4. The identification of the technological alternatives that have a gender

dimension should include the definition of differential support services
for women in order to help this sector of the population capture the
benefits of the improved technologies. Collaborative relations should be
established with agencies that can provide the required services, such
as literacy campaigns, child care services, women's organizations, and
credit and marketing services for women.
The group was not able to agree on a number of methodological issues.

For example, there was disagreement on determining whether the gender
variable should by itself force a definition of recommendation domains, in
all cases and in all circumstances.
THE ROLE OF NARIS, (ARCS, NGOS1
AND NETWORKS
A report by Working Group 32
Terms of reference. APSR/D has traditionally been positioned within

the context of national agricultural research institutes (NARIs) or in some
cases within official extension organizations. Nevertheless, international
agricultural research centers (IARCs) have also been important players,
frequently in association with NARIs. Bilateral donors have sponsored the
development of separate institutions, generally within the administrative
structure of the Ministry of Agriculture. The evolution of the institutional
scenario of the last decade (deterioration of the publicly funded
institutions, increasing self-organization of farmers and other groups of
society, acknowledgement and support of nongovernmental organizations
(NGOs) and networks by donors, upstream movement of IARCs) has
substantially expanded the range of potential institutional arrangements.
This group reviewed past experiences, assessed strengths and

weaknesses of alternative institutional settings, defined recommended
criteria for choice of institutions for new ventures, and discussed paths for
transition of existing arrangements to more productive ones. The
discussion was oriented to contribute to future action by taking into
account the changing funding situation and demands (e.g., more
sustainable systems).
BACKGROUND
Discussion on animal production systems research (APSR), as related

to the various organizations indicated above, must first consider that APSR
is not solely for the generation of technology, but also for the promotion
1
NARIs: National agricultural research institutes, IARCs: International agricultural research
centers, NGOs: Nongovernmental organizations.
2
E. Alarcon (Chairperson), L. Navarro (Discussion Opener), M. E. Ruiz (Rapporteur), M.
Sarmiento, J. Yazman, R. Hart, A. Roeleveld, E. Muller.
of agricultural development. Given the increasing complexity of agricultural

development as new issues such as sustainability and gender are taken up
in a more explicit manner, deciding which type of organization is best
suited to support and/or implement APSR work, are not easy tasks. An
overview is needed of the structure, behavior, and impact of the
institutional system responsible for research and development of
agricultural technology as an input for agricultural development.
INSTITUTIONAL STRUCTURE
Within any given country, the NARIs, together with universities, NGOs,
and other institutions responsible for agricultural technology research and
development, constitute the national agricultural research system.
Upstream, this system is supported by the international agricultural
research community, particularly the IARCs and the regional agricultural
research centers. Downstream, it is complemented by institutions dealing
with technical assistance, extension, credit, information, and agrarian
reform, as well as agribusinesses of different sorts.
The NARIs were created as an attempt to disconnect the purely

technical mandate implied in research from the bureaucratic context of
state institutions, to increase the efficiency of research funding utilization.
However, with time the NARIs drifted into the same sluggishness as their
parent institutions. Thus, insufficient resources and inefficient use of
funding for research are now recurrent maladies. After the creation of
IARCs, with their high-level technical mandate, research facilities, and high-
powered technical staff, funding tended to be channeled to these
international organizations, with or without demand for IARC-NARI
partnership (supposedly a responsibility of the IARCs).
For a number of well-known reasons3, private technical organizations

(NGOs) also began to appear. A major factor in this process was the need
expressed by donor agencies to place resources in the hands of groups
that promised to be more efficient and effective than the state agencies in
the use of research funding. This shift in donor interest, as well as the
3 Including officialdom's ignorance of the need to address the smallholder's plight, low
salaries received at the national institutions, bureaucratic overburden, lack of technical
recognition, and high personnel turnover.
generalized shortage of national resources and the flight of professionals

to the private sector or the developed world, had a marked negative effect
on the NARIs.
INSTITUTIONAL BEHAVIOR
Coordination of organizations within the local institutional system is

almost nonexistent. Certain past linkages between some regional or
international institutions and the NARIs have disappeared. The creation of
networks by the regional or international organizations is an attempt to
compensate for this weakness. The NARIs, structurally rigid and discipline-
oriented, have hindered, and sometimes openly rejected, the establishment
of farming systems research programs. APSR has been interpreted as a
social endeavor requiring generalists rather than specialists, as a producer
of unpublishable papers, and as a source of organizational confusion.
The systems approach evolved as a response to the lack of sensitivity

to the small farmer's social, economic, and technical framework. This
approach was intended to fill the communication gap between the
researcher and the farmer, as well as the gap left by the lack of feedback
from the extensionist to the researcher. However, this purpose was not
understood properly by the upper hierarchy of the NARIs. In fact, in some
cases farming systems research, with emphasis on small farmers, may
have been seen as a politically oriented movement. This partly explains
why farming systems research has not been institutionalized in many
NARIs and IARCs. Another reason is the fact that this approach has
only
been operational for the last twenty years, and APSR for even less. APSR,
particularly that involving large animal species with long production cycles,
has suffered from a lack of time to prove itself useful. Additionally, in the
case of NARIs, the origin of the support for APSR was largely provided by
donor agencies, with only token and declining contributions by the country
involved; when the external source dried up, the NARIs reverted
to
discipline-oriented, on-station research.
In contrast, the IARCs have evolved from an outright resistance to

the
systems approach (exception is made of IRRI and ICRAF)4 to their present
acceptance. In some cases this acceptance is very clear; in others it may
4 See list of acronyms and abbreviations at the beginning of the book.

be seen indirectly in the embrace of such issues as gender, sustainability,

ecologically friendly production, and so forth. However, the shortage of
research funds has placed the IARCs and the NARIs in competition for
research grants. Separation between NARIs and IARCs in the developing
countries is unfortunate. The networks, created originally for exchange of
information and mutual support, have, in addition, helped to bring back
together not only the NARIs but also the regional and international
research centers working on common topics.
IMPACT
The performance of APSR, within both NARIs and IARCs, has not been
satisfactory; one reason is that farming systems research demands
intensive and interdisciplinary action, work conducted on-farm, and
significant financial resources. However, against this negative backdrop,
two important new organizational options have emerged: the NGOs and
the networks.
The NGOs have shown aggressiveness in the capture of funds, and in

several cases they work more efficiently than the NARIs (their better salary
levels usually allow the hiring of the more qualified professionals).
However, most of the NGOs simply act as administrators of the funds and
resort to subcontracting. Moreover, of those which choose to execute
projects, a few are showing signs of promiscuity in project development,
working on subjects as diverse as pest control, the role of rural women,
and large-scale irrigation schemes, all with the same limited core staff.
The networks, initiated almost exclusively by IARCs or donors, cannot

be defined as homogeneous. Some involve research projects, others
involve NARIs. In general, they have contributed to the strengthening of
linkages between NARIs, and between these and the IARCs, as well as to
the strengthening of the national research programs. When a network acts
in a defined technical fashion, it becomes extremely useful.
Properly managed, the network can efficiently use resources and

achieve impact. The prestige thus gained must then be recognized and
taken advantage of by providing more support and by entrusting the
network with new mandates such as the identification of projects and the
design and administration of projects of regional scope or with an
ecosystem configuration. Networks can also ally themselves with lobbying
groups, not only to obtain continued support, but also to ensure the
utilization of research results from its membership for social and economic
impact.
CONCLUSIONS
1. There is an awareness in the international and national community

about farming systems research, in terms of its philosophy as well as
its methodological requirements, usefulness, and constraints.
2. Due to its multidimensional and interdisciplinary nature, animal

production systems research, as part of farming systems research, is
a facilitator of cooperative initiatives, both at the institutional level and
among researchers from different disciplines.
3. The systems approach reduces the time required from the point of
problem identification to the point of finding the solution or solutions
to that problem.
4. Production systems research is uniquely prepared and provides an

excellent opportunity to face current issues such as use and abuse of
natural resources, gender, social equity, and food security.
5. The commodity- or discipline-oriented structure and inflexible nature

of some of the national research and development organizations have
prevented a wider adoption of farming systems research as a
strategy.
6. The appearance of the NGOs and networks is due, in part, to the

weak linkages among the institutions responsible for research and
development. Another contributing factor has been the great diversity
of technological demands in relation to different agroecozones and
products.
7. The IARCs, due to their pioneering role in farming systems research,

must continue to act in this area. However, their actions must be as
promoters and facilitators, not as competitors of the NARIs, or any
other national or regional organization, including NGOs and the
networks.
8. The IARCs enjoy comparative advantages for upstream research.

However, they should look for closer working linkages with the NARIs
and help them to reach higher levels of understanding of current key
issues and of preparedness to face them.
9. International, national, and regional agricultural research

organizations, along with international organizations for technical
cooperation, should try to dialogue among themselves with a view to
presenting a concerted set of activities designed to solve regional
research and development problems. This high-level cooperation
could be extended to include donor agencies.
10. To improve research cost effectiveness, there is a need for better

collaborative schemes among institutions, both within countries and
between countries, with participation of the IARCs, NARIs, NGOs and
the networks.
11. Itis recognized that several NGOs possess considerable know-how

and that some of them have filled the gap created by the NARIs'
weaknesses in the application of farming systems research
methodology. However, it is suggested that NGOs and public
institutions look for linkages with the industrial sector for in-house
funding and sources of practical areas of work.
12. The networks are an efficient mechanism for cooperation, as

illustrated by RISPAL's performance, as they allow for closer
collaboration between the IARCs and the NARIs. Similar mechanisms
should be established within the countries.
13. The networks should include certain NGOs, not only as members
and sources of funding, but as members of their governing boards as
well.
14. The networks should serve as a forum for intraregional and

extraregional discussions, perhaps oriented to specific topics and
ecosystems. Intercontinental lines of communication and technical
support should be established to promote joint actions in the design
and application of APSR methodology, activities, and results.
Terns of reference. This stage is critical to insure the value of the

whole APSR/D effort. The implementation of very labor-intensive data
collection efforts only marginally utilized in the design of alternatives is
frequently observed. Several issues were presented to this group for
discussion, as follows:
What methodologies led to the selection of interventions in the case

studies presented? How successful were they? With the benefit of
hindsight, would they be used again?
What are the relative roles of statistical analysis, mathematical

modeling, expert systems, participatory farmers consultation, and other
factors.
A critical constraint is the imagination and experience of the analyst

involved. Can techniques be developed to sharpen this skill?
BACKGROUND
In the process of animal production systems research (APSR),

identification of alternatives immediately follows the identification of both
the constraints to be resolved and the opportunities available. Thus, it links
the diagnostic phase with the experimental phase, and is considered
critical for ensuring the success and coherence of the whole research
exercise. The time required to obtain results in livestock research is
usually long. A proper identification of alternatives is therefore important
for the cost effectiveness of APSR.
G. Cubillos (Chairperson), J. Kategile (Discussion Opener), B. Re y (Rapporteur), M.

Salaam, H. Vargas, I. M. Nitis, R. D. Estrada, M. Baayen, F. Holmann, W. Alhassan
It has been argued that a logical link between the information produced
during the diagnostic stage and the alternatives selected is not always
clear. In some cases, a limited number of apparently preconceived
technologies are tested; in others, the management of the hard data
gathered does not lead to any innovative alternative. After reviewing the
experience with livestock in some case studies, Working Group 4
concentrated on the mechanism for the identification and selection of
alternatives as well as on the consequences this mechanism may have on
the conditions under which data should be collected.
LESSONS FROM THE APSR CASE STUDIES
The case studies showed the wide range of ways the alternatives to be
tested have been identified. Contrasting examples are given below:
1. In Indonesia, the shrub and tree foliage alternative was an

improvement on the farmers' current practice for dry season feeding.
This alternative was based purely on observation and on the farmers'
opinions.
2. In Nigeria, the constraint (dry season feeding) and the alternative

technology (fodder bank) were selected on the basis of literature and
expert knowledge. A second collection of data on-farm were used to
tailor the technology to the target group.
3. In Guatemala, a three-year dynamic diagnostic phase led to a range

of innovations discussed among researchers, extensionists and
farmers. Farm data were used to help test these alternatives
biologically and economically by means of a model. Thus, the data
were used not only to determine constraints, but also to judge
alternatives.
On the basis of these examples, the working group questioned the

need for lengthy, labor-intensive data collection techniques to identify
suitable alternative technologies. It generally appears that the identification
of alternatives is, above all, an intellectual exercise which requires
considerable knowledge and understanding of livestock farming in the
target area. If existing knowledge is insufficient, then farm data collection
is justified.
The identification of suitable alternatives, unlike constraint identification,

always involves some ex-ante analysis. Conducting such analysis does not
necessarily imply the use of quantitative information; the information used
may be purely qualitative (as in the Indonesian case). However, ex-ante
analysis often requires some quantitative information, particularly when the
testing to be conducted is costly (as is most on-farm livestock research).
CRITERIA FOR SELECTION OF ALTERNATIVES
Before alternative technologies are chosen, some thought has to be

given to the criteria for their selection, beyond the fact that they must
resolve constraints or include opportunities identified. These criteria will
also be used to prioritize the interventions. Selection criteria depend on
which of the system's hierarchical levels one is positioned. Many times,
multilevel criteria have to be used and, therefore, the selection of
alternatives must consider trade-offs.
In addition, a researcher-perceived constraint is not always equivalent

to a farmer-perceived constraint; this means that the alternative must also
be perceived by the farmer as one that will truly resolve his/her problem,
regardless of the fact that such an alternative may also be solving
problems at other hierarchical levels. In the final analysis, the alternative
must fit the farmer's rationale and values, and be acceptable to him/her.
The microeconomic rationale may be at odds with the macroeconomic

one. Benefits to farmers and benefits to society were found contradictory
in several case studies, such as the Costa Rican experience in Rio Frio.
Environmental issues present the same type of conflict among hierarchical
levels of the system.
The assessment of the direction in which the livestock system will

evolve is speculative and might differ from one analyst to another.
Development policies and their possible amendments should be made
known and taken into account. The clients for research results should also
be taken into account; farmers, researchers, extension workers, and
planners have different ways of judging alternatives.
Once the constraints and opportunities are identified in the diagnostic

stage, alternatives can usually be designed on the basis of available
technologies; sometimes new technology may have to be produced. The
Guatemala example shows how existing technologies were chosen and
combined with new knowledge created at the research station level.
The group agreed on the need to involve farmers in the design of

alternatives to a greater extent than has been done in the past. Drawbacks
during the implementation phase of the technology should be anticipated.
Early testing of promising technology in livestock research is desirable;
however, this will depend on the type of technology, its level of impact on
the system, the time required to obtain results, and the level of risk the
researcher and the farmer are able to assume.
The working group stressed the role and importance of the researcher's
experience in animal production systems diagnosis and design of
alternatives. Experience leads researchers to decrease the volume of
information gathered from farmers. Nevertheless, optimal data collection
will depend on each specific situation. The benefits of experience are
obviously affected by the turnover rate of researchers. They are also
affected by the fact that systems research is often carried out by young
staff members, who lack experience and consequently opt for extensive
data collection in the hope that nothing important will be missed.
HOW MUCH DATA IS NECESSARY TO IDENTIFY ALTERNATIVES?
Lengthy, labor-intensive data collection efforts that are only marginally

utilized in the identification of constraints and alternatives are a concern for
donors and research managers. While the value of ex-ante analysis in
animal production systems research cannot be stressed enough, this
statement contrasts with the concept that researchers should strive for a
minimum data set to conduct the diagnosis and to design alternatives. It
was not possible for the working group to give conclusive
recommendations on the subject of minimum data (number of farms to
visit, number of herds to survey or monitor, quantity and accuracy of data).
Nevertheless, a few suggestions were made:
Before conducting a diagnostic survey, maximum information on the

area and situation under study should be gathered.
Informal data collection and interview techniques should be more widely

used in animal production systems research; however, their success is
strongly linked to experience. These techniques should be used to
identify areas and/or disciplines which require more specific
quantitative information to ascertain the accuracy and relative
importance of the constraints. Objective-driven data collection can also
make data gathering more cost-effective.
As an alternative research process for understanding the system and

identifying alternatives, some scientists give considerable importance to
the variability of livestock management by the farmers themselves. This
variability results from the relatively long production cycles in animals.
POLICY/COMMODITY SYSTEM PERSPECTIVE
Terms of reference. FSR evolved from traditional farm management

and biological sciences as a response to the low adoption of innovations
developed by conventional biological research. It was implicitly
hypothesized that farm level constraints (limited availability and poor
quality of resources, seasonality of supply of labor, etc) were the major
culprits.
A large share of the intellectual input came from researchers from

developed countries, where institutional arrangements (markets, policies,
legal framework) had evolved gradually over a long time span, and were
considered relatively efficient and subject only to gradual change. Under
these circumstances, such factors beyond the farm gate were taken as
parameters, not variables. On the other hand, the failure of many FSR
projects has been blamed on factors such as inappropriate policies and
nonfunctioning markets. Additionally, policies were observed to be very
volatile and amenable to change. All this became particularly obvious in
the 1980s, when the globalization of the world economy and structural
adjustment in many countries induced rapid policy changes. These in
turn have either positively or negatively affected existing production
systems. Thus, consideration of factors beyond the farm level is crucial,
particularly as systems become more commercially oriented and more
value is added to agricultural raw materials (e.g., postharvest processing,
packaging) to make them available to an increasingly urban society.
This work group will review case studies from this perspective,
identifying system constraints in the upper levels of the hierarchy which
have impinged on the results of cases presented. It will discuss
approaches to identify each constraint, methodologies to conduct
experiments to assess the magnitude of such constraints, and the
feasibility of overcoming them. The groups will also make suggestions on
institutional and organizational implications of this broader perspective.
L. Navarro (Chairperson), C. Sere (Discussion Opener), C. Valdivia (Rapporteur), O. B.

Smith, J. Kategiie, F. Navas, D. Martinez.
OBJECTIVES
Changing conditions in the world economy and the impact that they
have on all national economies, as well as the impact of commodity
systems (such as markets and industry), make it necessary to consider the
effects of these changes on farming systems research (FSR). The main
objective of working Group 5's discussion was to address (1) the impact
that conventional government policies and the macroeconomy (mainly
interest and exchange rates, international trade, and structural adjustment
thrusts) have on farming systems, and (2) the demands imposed on small
farms by commodity systems, as these issues relate to FSR in general and
to animal production systems research (APSR) in particular.
THE PROBLEM
Changes in the broad hierarchy of systems have usually been taken as

directive, constraining parameters within which FSR must be developed.
This poses problems for the validity of the research results, especially in
livestock systems research, where long periods are required to yield
results. Short-term government goals and an unpredictable policy horizon
have consistently been at odds with long-term development plans in the
agricultural sector, particularly in APSR.
The extreme volatility of government policies may even result (in some
cases) in regression to subsistence economies, where farmers try to isolate
themselves from the unreliable conditions of the market. This may occur
under conditions of great political turmoil.
The globalization of the world economy and the current opening of

developing economies add to the uncertainty that small farms will face in
the so-called new international order. This is especially true in the process
of transition towards structural adjustment.
Usually, FSR does not dwell on the workings and consequences of

commodity markets; however, changes in these markets have such an
impact that it may become necessary to adjust the objectives of a research
project in order to offer a product appropriate to these changes. Most
farming systems projects seek to increase the output or productivity of
some existing farm enterprise, but this objective will likely fall short of the
new economic goals and conditions brought about by the opening of

domestic markets. In such cases, the development of new production
system prototypes may be a valid alternative to the mere adjustment of an
existing system.
The combination of changes in both in policies and commodity markets

may affect small farms in a more profound manner than could possibly be
accomplished through technological intervention at the farm level, which
is a common strategy followed by farming systems research. If this is not
taken into consideration, the validity of the whole research effort may be
put in peril. In summary, in order to be successful in facing these
changing conditions, it is necessary to incorporate in FSR projects the
demands and constraints imposed by the broader hierarchy systems (i.e.,
the macroenvironment).
Following is a nonprioritized series of comments on some specific

macro- and commodity-systems aspects related to farming systems
research.
Types of output
Small farms engage in the production of three general types of

commodities (in terms of intended destination); these are products for the
global market, the local market, and family consumption. It is possible, of
course, that a particular product may be distributed to more than one
destination. Researchers should recognize that quality and quantity
requirements are more stringent in the broader markets and should
proceed accordingly.
With livestock, it is very common for the same species to provide

several products which face quite different market conditions. A typical
example is that of animals raised for fiber, which face both a world wool
market and a local meat market, and also provide manure for on-farm use.
This implies that the research thrust should be broad enough to improve
one aspect (according to the project's overall objectives) without
jeopardizing the others. This precaution may be less important if there is
a stable market for the alternative product, thus making it possible, in the
long run, for the farmer to use the increased income to purchase those
products which he/she no longer generates.
In open economies, where efficiently produced goods can be traded

freely, careful consideration should be given to the types of products that
should be researched if the aim is to increase output. New, nontraditional
products should be introduced, given the greater positive impact they can
have on the system.
Flexibility
When there is uncertainty regarding the outcome of changes occurring

in the macroenvironment, research projects should add flexibility to the
farming system, enabling it to better withstand several possible scenarios.
This may be particularly relevant when traditional systems are exposed to
new market conditions without being prepared for them, thus risking
disruption, which usually implies high social costs.
Time frame
Given the fact that macropolicies tend to be very volatile, it is necessary

that FSR have a built-in feedback system, allowing for internal changes in
the thrust of the project. This is especially true in livestock research
because of the long time frames necessary for results to be produced.
Past experiences have shown that research is usually planned on the basis
of an initial set of policy conditions; however, when these conditions
change, as was illustrated by the Costa Rican case study on milk, then the
product may no longer find a favorable market. Thus, instead of profuse
data research at the beginning of the project, frequent appraisals are
recommended in order to understand and internalize the changes in
policies and market conditions.
Influencing policy changes
In some cases, it may be desirable, and even necessary, that a

research project report its findings to the appropriate policy-making bodies,
so as to influence policy changes beneficial to the project's objectives. It
has been noted that the diffusion of research findings and results has often
been neglected in policy formulation.
Work Group Sessions
Operational aspects
It would not be realistic to expect that every farming systems project

carry out the needed macroeconomic analyses. However, proposals
should include such considerations. The role of networks could be
broadened to support this type of macroeconomic and commodity
analysis, both through a consultant referral service and through the
building and maintenance of relevant data bases. In countries where the
civil service has continuity, the national agrarian research institutes may be
able to host this research.
In cases where local markets are of great importance, the study of the
marketing channels should be an integral part of the farming systems
research effort. At another level, opportunities to include commodity
processing should be explored. Also, in ongoing research projects,
important lessons could be learned from examining how farmers have
reacted to and adjusted to recent policy and commodity-market changes.
It is very likely that most projects conducting follow-up or dynamic
surveys
will find that information readily available.
SUSTAINABILITY PERSPECTIVES
Terms of reference. Animal production systems have frequently been

associated with nonsustainable development processes (e.g., grazing of
hillsides has frequently led to erosion and silting of watersheds). The
clearing of rain forests in the Amazon and Central America is associated
with the demand for pastures to produce beef. Intensive monogastric
animal production systems close to urban centers can lead to
contamination of surface water and to air pollution. On the other hand,
efficient pasture-based production systems incorporating nitrogen-fixing
legumes enhance physical and chemical soil properties. Such systems
can generate income levels which are competitive with annual crops, thus
reducing the incentives for less sustainable crop production.
Sustainability basically involves incorporating external factors and a

sufficiently long-term perspective into the economic analysis. This
perspective leads to a valuation of alternatives different from the one
generally undertaken by the resource-poor farmer. Does this imply that
participatory approaches will lead to sustainable solutions when such
externalities and differences in time frame exist?
This work group was asked to analyze the extent to which sustainability
considerations might change the alternatives proposed in the case
studies. The group's terms of reference included the discussion of the
issue of how to incorporate the sustainability perspective into participatory
project design and management. Based on this, the group made
recommendations of further changes needed in APS projects to enhance
their contribution to sustainable development, taking into consideration
such aspects as disciplines, evaluation methods, time horizon, basic
research needs, and policy implications).
R. Hart (Chairperson), F. Romero (Discussion Opener), G. Cubillos (Rapporteur), J.

Ferguson, E. Muller, I. M. Nitis, A. Roeleveld, E. Thomson, J. Turk, J. Yazman, H. Li Pun
BACKGROUND
The importance of the sustainability issue stems from the following

points: (1) the population growth rate in the less-developed countries is still
high, and this fact will cause pressure upon resources; (2) some recent
estimates indicate food production has decreased by 6.3% in the last three
years; and (3) there is a noticeable reduction in natural resources (such as
forested areas) affecting cropping cycles, pasture utilization, and other
areas of agricultural production. The above points lead to the need to
increase production in a sustainable manner without degrading the
resource base.
SUSTAINABLE LIVESTOCK SYSTEMS AND THE RESOURCE BASE
In order to reach a common ground of understanding, the group

considered at the outset this definition of sustainability: sustainability is the
maintenance of both the resource base and the production level it is
capable of maintaining. This concept requires a description of the
relationship between the resources and their utilization in a production
function. Some of the points discussed led to the following statements:
Sustainability is dynamic, and as such is affected by several external

factors. In many areas, the population increase may constitute a basic
issue, as human survival may be at stake. The maintenance of the
resource base may be difficult under such circumstances.
The characteristics of the relationship between technology and the

resource base are important. An understanding of this relationship
must take into account the differences between short-term and long-
term objectives.
Since livestock plays an important role in the sustainability of natural

resources, both negative and positive experiences need to be
documented. Little effort has been made to understand the role that
livestock plays in sustainable agriculture, especially at the political and
donor levels.
The group considered it necessary to identify the resources affected by

the process of seeking sustainability, to indicate those considered critical,
and to list the measurements that ought to be taken. Toward this end, the
group analyzed the case studies presented in the workshop, looking at

sustainability from the biological, economic, and social viewpoints. From
the biological angle, most cases appeared to be sustainable for the
conditions under which they were conducted; there was insufficient
information to evaluate economic sustainability, and most cases were
sustainable from a social perspective.
The rate at which resources are used can be greatly affected by

external factors, and this will, in turn, affect the renovation cycle; the
following are the critical resources that have to be measured, considering
that not only the resource in itself is important, but also its renovation
cycle:
Soil: This is the key element, and measurements of fertility, structure

and runoff must be taken.
Vegetation: Animal feeding depends heavily on biomass availability.

Botanical composition, soil cover, and dry matter availability should be
measured.
Climate: Measurements of rainfall and temperature should be made in

areas under study.
Population: Human population should be measured in terms of social

structure, number, health status, and other pertinent characteristics.
Based on the case studies presented, the group considered that, in

order to make adequate measurements, multidisciplinary teams, should be
assembled. The following disciplines were identified as minimal
requirements: soils, systems analysis, ecology, economics, and sociology.
In addition, production-oriented specialists must always be present to
maintain an adequate balance.
The key indicators to define sustainability need to be identified. Since

farming systems and their sustainability are dynamic in nature, sufficient
time is needed to make adequate measurements. Depending on the
attributes of the systems to be measured, the minimum time will be 10
years or longer. However, since it may not be possible to obtain such
long-term financing from the outset, it is suggested that appropriate
intermediate outcomes be clearly identified, so that they will serve as
indicators of progress.
Because of the time involved, projects where sustainability is measured

need to be acknowledged as important by the national institutions.
Multiple funding is also desirable to reduce the risk of terminating the
efforts before results are obtained if a donor fails to renew its support.
Sustainability may be achievable if the farmer feels that the land resource
is secure enough. Thus, land tenure is a critical requirement if real land
resource measurements are to be taken.
RECOMMENDATIONS
1. Available information regarding sustainability issues under different

ecological conditions should be collected, and the key indicators for the
measurement of sustainability under different project conditions should
be defined.
2. Project and research activities dealing with and attempting to measure

sustainability must have a longer life than at present.
METHODOLOGICAL ISSUES
A report of Working Group 7'
Terms of reference. This group discussed a series of methodological

issues which repeatedly arise in APS projects, as follows:
The magnitude of the changes proposed: Crop-oriented FSR has

clearly emphasized marginal changes in existing crop enterprises. In
more complex, small-scale animal production systems (where animals
play a series of roles), the equilibrium may be so multifactor-dependent
that it is difficult to introduce marginal changes in one component.
More radical changes seem necessary (e.g., the use of intensive
manuring and very productive pastures).
The size of the recommendation domains: there is a clear trade-off

involved between the specificity of the recommendation and the cost
effectiveness of the research effort. This trade-off seems to be different
in crops and animal systems.
The effort put into diagnosis versus experimentation: The trend seems
to be to accept our limited success in diagnosis and to increase the
role of actual experimentation to learn about the constraints of a system
and alternatives to improve it.
The role of statistical analysis: This is especially important when units

of experimentation (e.g., cows) are few (on a small farm) and results
can only be measured in the long term (e.g., impact on useful life of a
cow).
Sample size and logistics involved.
Other issues that arose from the presentation of the case studies.
B. Rey(Chairperson), M. Ruiz (Discussion Opener), D. Norman (Rapporteur), A. Ruiz, M.

Sarmiento, B. Kiflewahid, J. Berdegue, J. Leesberg, M. Baagen
BACKGROUND
Working Group 7 discussed methodological issues related to farming

systems research and extension (FSR/E), concentrating in some detail on
three areas: the clients, the process, and the experimental stage; the focus
of the discussion was on animal product systems research (APSR) as a
subset of FSR/E.
CLIENTS OF FSR/E
Although the group recognized that the major customer or beneficiary of

FSR/E is the farmer himself, it also acknowledged that there are other
major clients who contribute to the process of agricultural development:
Researchers on experiment stations who produce potential improved

technologies. They benefit not only from receiving feedback from farm-
level work, which contributes to the evaluation of the technologies they
have developed, but also in receiving ideas on what would constitute
relevant research priorities in the future.
Planners who are responsible for developing the relevant policy/ support
systems. They benefit in ways analogous to researchers (i.e., by both
feedback on the suitability of the present policy/support system, and by
ideas as to what might constitute suitable systems in the future).
Extension and development staff who are responsible for disseminating

the improved technologies and sometimes implementing the policy/
support systems.
Such a heterogeneous set of clients means that FSR/E "products" need

to be developed to suit their specific needs and different methods need to
be used to satisfy these needs. For example, Table 1 differentiates trials
according to whether the researcher (R) or farmer(F) manages (M) and/or
implements (1) them. Trials that are undertaken at the researcher managed
and implemented (RMRI) level are likely to have results which are more
acceptable to experiment station-based researchers than are those derived
from farmer-managed and implemented (FMFI) trials. The latter are not
amenable to sorting out cause-effect relationships and exhibit high
coefficients of variation; they are, however, much more likely to appeal to

the farmers who are intimately involved in managing and implementing
them.
Table 1. The FSR/E researchers themselves who need to receive enough credit and
recognition for their work to be able to progress satisfactorily in their careers.
item Researcher- Researcher- Farmer-

managed and managed and managed and
researcher- farmer- farmer-
implemented implemented implemented
(RMRI) (RMFI) (FMFI)
Experimental Stage Design2 1St stage testing 2nd stage

Design: testing
Complexity Most Less
Type Standard Simpler Least
Replication Within and Usually only With and
between sites between sites, without
but can also be Between sites
within only
Levels of treatment Many Less Few

Standardized level
of non-experimental Most Less Least
variables
Plot size Smallest Larger Usually largest
Who selects Researcher Researcher/ Farmer

technology? farmer
Who shoulders risk? Mainly researcher Researcher/ Mainly farmer

farmer
Main discipline of Mainly technical Technical/ Mainly social
researcher social
Participation by:
o Farmer Least More Most

o Researcher Most Less Least
P. Numbers of
farmers None Some Most
o Farmer group Least more most
-
Item Researcher- Researcher- Farmer-

managed and managed and managed and
researcher- farmer- farmer-
implemented implemented implemented
(RMRI) (RMFI) (FMFQ
Potential
P "Yield" Most Less Least
v Measurement
errors Least More Most
v Degree of
precision Most Less Least
v Data:
"Hard" (objective) Most Less Least
"Soft" (subjective) Least More Most
v Determination of
cause/effect
relationships Easiest Harder Hardest
v Incorporation into
farming system Least More Most
Evaluation
c. By whom? Mainly researcher Researcher and Mainly farmer
farmer
. Nature of test Technical feasibility Technical feasibility Validity for

plus economic farmers,
evaluation practicality,
acceptability
Appeal to:
o Researchers Most Less Least
o Extension staff Usually least More Most
o Farmers Least More Most
Results more easily Researcher Researcher/farmer/ Farmer

accepted by extension
' There is a degree of subjectivity in some of the entries in the table, but in general they do reflect reality. In a
sense. these expectations also reflect the reasons why the different types of trials are undertaken.
2 Standard multilocational trials are also RMRI.
The clients defined so far collectively interact in the agricultural

development process. However, the group felt that two other types of
clients of FSR/E should also be kept in mind:
Those responsible for obtaining support for FSR/E (e.g., from a donor
or national program). In such a situation the FSR/E approach must be
able to demonstrate that it can potentially improve the process of
agricultural development in the area under consideration in a way that
is compatible with national priorities.
The FSR/E researchers themselves who need to receive enough credit

and recognition for their work to be able to progress satisfactorily in their
careers.
THE FSR/E PROCESS
In preliminary discussion on this topic, the group made the following

general points:
FSR/E is a process that consists of a number of stages that can, and

often should, overlap with each other (i.e., descriptive, diagnostic,
design, testing and dissemination).
All disciplines involved in FSR/E have roles to play at all stages of the
process, although the relative significance of their contributions may
differ between steps.
The FSR/E process starts with a systems perspective, becomes

reductionist in terms of trial work, and then evaluates the results from a
systems perspective.
The methodology with respect to crops in FSR/E is better developed

than that with respect to animals.
Unfortunately, because of the different characteristics of plants and

animals (e.g., degree of mobility, length of life cycle, degree of variability,
cost of an animal compared with a plant, etc.), much of the
methodology developed for cropping systems research is not suitable
for research in animal production.
The design stage
The design stage has received less methodological attention than the
other stages, and was perceived by the group to be one of the weakest
links in the whole research process. This stage is particularly critical in
APSR because, once a decision is made in the design stage as to what

should be done in livestock work, it is not easy to revise or change plans.
The reasons for this are as follows:
The gestation period for getting results from livestock work is much
longer than that for annual crops, which are usually the focal point of
cropping systems research; thus, costs invested prior to receiving a
return are usually much higher for livestock than for crops.
Livestock work, in contrast to crop work, often involves so-called "lumpy"

inputs (e.g., a cow compared with limitless possible levels of a crop
such as 1 ha or 0.001 ha).
In the case of an annual crop, experimental results from one cropping

cycle can immediately be fed back into design work, and if necessary may
bring about a modification of the next season's testing work. Feedback
from testing work on livestock is likely to take a much longer time period.
The point was made however, that such feedback, although less timely,
can facilitate decision making in future design activities. It is particularly
valuable in this regard, during the dissemination stage, not only to monitor
whatever adoption that may have occurred, but also to evaluate the
adoption process by exploring why results were not as expected.
the factors mentioned above, the general consensus was that

In view of
in systems research on animals considerable attention should be paid to
the design stage. The potential "costs" of not doing so are likely to be
much greater for animal rather than cropping systems research. In
undertaking this design stage the following points were made:
Preliminary design-type activities should start as early as possible after

the commencement of the descriptive or diagnostic activities. Such
preliminary design activities can help identify missing information in
descriptive or diagnostic work as well as give some idea as to what
variables are potentially critically important, for which accurate
measurements are required.
As a result of the descriptive or diagnostic exercise, a number of

problems should be identified. Some prioritization of these problems is
then necessary. A useful way to do this is to rank them according to
their potential positive impact on the farming system as a result of
solving them (i.e., their degree of leverage). Whether or not the most
pressing problems will be addressed by the FSR/E team will depend on

a number of factors, for example:
Whether a potential solution exists to address the problem identified;

if not, the problem should be fed back to station-based researchers.
Whether the solution lies in changing the policy/support system to

enable the adoption of a particular technology; if it does, change
should not be attempted if those responsible for the policy/support
system indicate that changes to encourage the technology's adoption
are not possible.
Whether the solution to the problem lies outside the mandate of the
institution in which the FSR/E team is located.
Whether the FSR/E team has instructions to address equity issues

(e.g., find ways of helping improve the position of women vis-a-vis men
within households, in which case a low-leverage technology may be
the only one potentially feasible).
Design activities should initially result in making a decision as to the

generally desired outcome (e.g., improved production of goat milk). This
should then be followed by considering a number of trials that might
lead to this desirable outcome (e.g., improving breed of goat, growing
higher-quality forage, feeding of concentrates).
In thinking through these strategies, it is important to evaluate them ex-

ante in terms of their potential compatibility, not only with the expected
changes in the enterprise (e.g., goat) under consideration, but also with
the rest of the farming system (i.e., other livestock, crops and off-farm
enterprises). The evaluation should also take into account the needs
(e.g., demand for goat milk) and requirements (e.g., ecological
sustainability) of society.
In order to undertake such an exercise, it is important not only that

FSR/E researchers participate in the design exercise, but that farmers,
station-based researchers, extension/development staff and planners are
also consulted. In the light of potential significance of the macropolicies
in determining what farmers can or cannot do, it is particularly important
that planners are consulted. During such an exercise, the potential for
"endogenizing" specific exogenous variables can be explored. For
example, consultations with planners can help determine whether a
submissive (i.e., accept the policy/support system as fixed) or potentially
interventionist (i.e., be able to change the policy/support system) approach

can be taken with respect to policy parameters. For example, is the plan
to increase goat milk production consistent with the anticipated needs of
the society, and is there a possibility for introducing adjustments in the
policy/support system to encourage goat milk production if the
technologies evaluated for increasing milk production look particularly
promising? This micro- and macro-level analysis is particularly important
in APSR, especially in view of the "lumpy" nature of the inputs and the long
production cycles.
Modeling
Some discussion took place on the role of models in APSR. Modeling

was thought to have more potential in the design stage rather than in
testing-stage activities. With reference to testing, the consensus was that
modeling cannot replace for testing in an on-farm setting. Models are only
a simplistic approximation of reality and cannot simulate the large
coefficients of variation found when technologies are validated under
practical farming conditions. However, the group agreed that under certain
circumstances modeling did have a role to play in design-type activities.
The following points were made:
Models have advantages; for example: (a) they can help structure
thinking; (b) they can help give some indication of the data required; (c)
they facilitate communication within the FSR/E team; and (d) they can
be useful in determining the treatments and levels of treatments to be
used in experimental work.
Modeling can more accurately reflect relationships of component

research and becomes more approximate as more components and
relationships are added.
In early work in a particular location, efforts to specify appropriate

models and "operationalize" them with appropriate hard data will usually
limit their usefulness. However, such models could provide a useful
input into later design-type activities in the same general location.
There does appear to be some potential usefulness for models designed

to investigate the potential for modifying biological parameters, such as
herd/flock dynamic models, optimum calving interval models, and
reproductive models.
Much of the above discussion was specifically geared to mathematical

modeling. However, two other types of modeling were also discussed by
the group. These were:
Nonmathematical models designed to elicit information: Two types of

models that the group thought had potential in structuring and
broadening thinking in design-type activities were Decision Trees and
Expert Systems.
Physical models to investigate possible strategies: There are a number

of possible variations on these, including model farms on experiment
stations and farm units in which selected farmers implement activities
(usually under researcher supervision). Only in very rare circumstances
can such prototypes be usefully transferred as is" to practical farming
conditions. For example, such potential prototypes may be useful when
a new area is being opened up to agricultural activities.
More often, physical models have limited objectives, such as: a)

indicating/ demonstrating to administrators, farmers, and others what is
possible; b) providing a learning "tool" for researchers; and c) providing
a way of obtaining accurate measurements of specific component
technologies under researcher-managed conditions. The component
technologies, if promising, can be tested by farmers within their own
farming systems as a potential incremental strategy.
THE EXPERIMENTAL STAGE OF FSR/E
The group made a number of points about the experimental or testing

stage of FSR/E, with specific reference to APSR. Some of the major
points were as follows:
Allthree types of trials listed in Table 1 have important roles to play in

systems research with livestock.
Whenever possible, RMRI-type activities should be undertaken by

scientists based on experiment stations, using the types of, and levels
of, nonexperimental variables that are likely to exist under practical
farming conditions. If these cannot be simulated properly, then such
work should be undertaken on farms.
Trials at the FMFI level (possibly organized through farmer groups which
are either research- or extension-oriented) provide the most realistic test
of a proposed technology and should be considered part of the testing
stage rather than as a dissemination activity.
RMFI trials, which are not always easy to implement in animal production
systems research, provide an important mode for testing a technology.
Some specific points were noted:
o Where the researcher can control the treatment (e.g., parasite control
versus no control), animals in a single herd/flock can be divided, and
results from the treated and untreated animals compared. However,
where the farmers implement the treatment (e.g., feeding supplements),
then control versus treated animals has to occur across herds/flocks.
t> Replication becomes a particularly difficult issue when treated versus

untreated herds/flocks are used. In such cases, it is important to be
very sure that the resources required to implement such work at the
RMFI level are justified in relation to the extra insights that will be
obtained. It may be more practical, under certain circumstances, to
move straight from the RMRI to the FMFI level with careful monitoring
and evaluation as to what happens at the latter level.
t> In deciding the relevant size of sample required in RMFI testing, the
coefficients of variation of the relevant parameters can be obtained
from the diagnostic or descriptive work, and combined with testing
treatments at a probability level of 20-25% rather than at the more
conventional level of 1-5 percent.
RMFI work is much simpler to undertake with penned (e.g., guinea

pigs, rabbits) than with "mobile" (e.g., goats, cattle) animals. Particular
problems arise with larger-framed animals (e.g., cattle) since there are
usually a smaller number that can be compared.
. Because of the considerable time period over which data collection

may be required in RMFI trials, cost-reducing strategies need to be
employed whenever possible. For example, after satisfactory initial
instruction and monitoring, it may be possible for members of the
farming family (including women or children) to record data. Accuracy
in the data collected is likely to be better if the farmer can be shown
ways in which analysis of such data can help improve the management
of his/her farm.
o Inon-farm research with animals, regression analysis is often more

appropriate than analysis of variance.
CONCLUDING COMMENTS
Although the group recognized that much can be learned from the
experience accumulated in cropping systems research, it also recognized
that much of the methodology cannot be applied directly to research on
animal production systems. More attention needs to be focused on
developing methodologies for APSR, and efforts need to be made to
convince top administrators that this evolutionary process should continue
for some time to come.
IMPACT ASSESSMENT
Terms of reference. APSR has used substantial amounts of public

funds, both in less developed and developed countries. APSR competes
with alternative uses for these funds. To make the allocation of funds
more efficient, it is necessary to document the economics of investing in
APSR/D. Beyond this contribution to the decision makers in charge of
allocating funds, adoption and impact studies can also play a crucial
feedback role for the project team and related researchers. Until now, few
efforts have been made to address these aspects of funding. The working
group reviewed the impact of the case studies presented and the
methodologies employed; based on these analyses, it made a series of
recommendations.
BACKGROUND
The area of impact assessment is a relatively new one to many farming

systems research and extension (FSR/E) practitioners. Impact assessment
of commodity-oriented research, with its monodisciplinary approach, is less
complicated than impact assessment of systems research, which by its
very nature includes many parameters of analysis that need to be ranked
according to their relative importance to researchers, donor agencies, and
users of the technologies generated. Ranking of these parameters in turn
leads to recognizing the trade-offs involved.
J.B. Schiere (Chairperson), R.D. Estrada (Discussion Opener), F. Nyaribo (Rapporteur),

F. Holmann, M. Saleem, H. Vargas, W. Alhassan, G. Kleemann
SETTING OBJECTIVES
When a project is conceived, it must allow the necessary flexibility to

adjust the objectives (and therefore the anticipated impacts) to changing
circumstances. This could be viewed as a "Materials and Methods" section
in the project document. Both clearly stated objectives and indicators of
impact at the beginning of the project will enable project evaluators to use
consistent criteria during the life of the project, as well as in ex-post
evaluation exercises.
INDICATORS
Parameters and indicators need to be clearly defined and should be set

by donor agencies, researchers, and target groups who will use
technologies generated by animal production systems research and
development (APSR/D). Examples of indicators are gain, yields, or
income. Social indicators such as improved welfare and equity can be
used. Other indicators include better nutrition leading to reduced infant
mortality, increased life expectancy, efficiency, sustainability, and training.
The kinds of parameters chosen depend on the nature of the supporting

organization (governmental, nongovernmental, or donor), the type of
project (development, research, or network), and the life span of the
project (long-term or short-term). Many projects require a time lag or
maturity period before impact is felt. The selection of parameters also
depends on the target group or issue (for example, producers versus
consumers, gender). Additionally, the parameters need to be analyzed at
one or all of the following levels: individual, farm, region, national and/or
world. Impact assessment at all levels will require a larger budget, more
personnel, and longer maturation periods.
IMPACT ASSESSMENT IN CASE STUDIES
With respect to the case studies presented, it was pointed out that many
did not demonstrate a wide societal impact, although it was acknowledged
that macrolevel impact analysis is difficult (many variables, large data
requirements). The question of how far APSR/D should go without

overextending itself was also raised. One case study demonstrated the
trade-offs involved in avoiding social discontent versus maintaining a
particular class of producers. Another observation from the case studies
was that the African and Latin American studies differed with respect to the
level of analysis.
The Latin American studies focused on macroeconomic (demand,

consumer welfare) impacts, while the African studies analyzed microlevel
(supply) impacts. These differences were attributed to the different
objective functions facing the farmers where case studies were conducted.
The group felt that few impact assessment studies have been done so far
because researchers view evaluation as a threat to themselves, rather than
as a feedback mechanism.
STRESS ON POSITIVE EFFECTS
Often, evaluators and organizations supporting APSR/D think of impact

only in monetary terms; however, projects may have intangible benefits
that are difficult to quantify. This fact is frequently overlooked. These
intangibles include education, welfare, and happiness. Stakeholders
commonly stress the positive effects of technology. However, these effects
are nearly impossible to demonstrate; such is the case of many APSR/D
projects that work in environments with marginal, limited, and/or degraded
resources with static or even negative marginal productivity. A project
operating in such an environment can, at best, only begin to slow down
(but not reverse, at least in the short term) environmental degradation, and
will have a very weak impact in terms of productivity.
CONCLUSIONS
While realizing that training is an indicator for judging the effectiveness

of APSR/D, the group concluded that researchers themselves need
information and training in impact assessment methodologies. Training
and closer farmer-researcher collaboration can assist in impact assessment
and improve farmer-researcher feedback flows.
An impact-assessment matrix at the beginning of the project is

recommended as a useful tool that can serve as a checklist of key
parameters. This matrix could be used throughout the life of the project
as well as in ex-post impact analysis. A sample matrix is shown in Table
1.
Table 1. Impact assessment matrix
Term
Levels Short Medium Long
Productivity Sustainability Sustainability

Farm Income Environment
Adoption
Adoption
Community Income
Employment
Market establishment Equity

Region Income Consumption
Production
Foreign exchange
Country Consumption
I
National income
To summarize, the key issues in impact assessment as seen by the

group are, first, that the attempt at impact assessment must be made at
the beginning of the projects, and, second, that there is a need to assess
impacts both in the short as well as the long term. In undertaking impact
assessment, the systems approach creates sensitivity to future problems
and opens pathways to future opportunities.
CONCLUSIONS AND
RECOMMENDATIONS
Hugo Li Pun', Manuel E. Ruiz2 and Carlos Sere3
1. Animal productions systems research is a dynamic, continuously

evolving reality, its methodology being streamlined by using past
lessons and incorporating additional concerns such as macroeconomic
analysis and technology adoption studies.
2. Networks are seen as an effective mechanism facilitating this evolution

and enhancing the cost-effectiveness of individual projects.
3. The continuity in farming systems research and the long-term external

support have allowed national programs to persist with this approach
and iteratively improve it. They are obtaining promising results in
systems which require a long-term research perspective to take into
account both long productive cycles in animals and cyclical economic
phenomena.
4. This long-term support has also allowed for the training of a substantial
number of specialists in livestock systems research.
5. In some cases, continuity has also been made possible through the
blending and complementation of efforts of various international,
regional and national technical institutions and donor agencies.
6. The opening of world economies has increased the need for an

expanded systems approach incorporating policy dimensions. The
case studies presented have helped in the understanding of past
performance of production systems and in the reviewing of hypotheses
concerning systems research in the future.
1
International Development Research Centre (IDRC), Canada.
2
Coordinator of RISPAL, Inter-American Institute for Cooperation on Agriculture.
3 Economist, IDRC, Montevideo, Uruguay.

7. National agricultural research centers are having difficulties in the

institutionalization of systems research, a fact that causes serious
concern about the sustainability of this type of research effort beyond
donor support. New institutional arrangements are needed, including
the participation of a wider range of stakeholders (nongovernmental
organizations, farmer organizations, commodity development
institutions, etc.).
8. There is a growing need to document positive and negative impacts of

farming systems research projects in terms of technology adoption and
its contribution to the welfare of the target groups. These target
groups include a range of stakeholders beyond farmers (e.g.,
consumers, environmentalists).
9. The presentations have demonstrated a commonly shared concern

about new issues, such as environmental quality and sustainability,
gender, and income and employment generation. As animal
production is seen by many as an activity detrimental to natural
resources, there is now declining support for animal production
research and development.
10. In view of the above, the Global Workshop participants fully agreed
that, in addressing the new development issues, an assessment of both
the role of animals and the appropriateness of the systems approach
are of utmost importance to the achievement of such understanding.
Therefore, a task force was created to prepare a new research agenda
for animal production systems to explicitly address the new concerns.
11. Addressing the new development issues as they relate to animal

production, particularly the documentation of the trade-offs involved
and the cost and benefits associated with those options, will require
creative research approaches and methodologies.
12. Careful planning and stronger cooperation among various interested

parties are needed in order to efficiently use the limited resources and
to successfully face the new challenges.
APPENDICES
LIST OF PARTICIPANTS
BARBADOS CHILE
John Ferguson Julio Berdegue

British Develop. Division RIMISP
in the Caribbean Box 228-Correo 32
P. O. Box 167 Santiago, Chile
Collymore Rock Phone: (56-2) 223-2423
St. Michael, Barbados Fax: (56-2) 223-2423
Phone: (809) 436-9873
Fax: (809) 426-2194 Gustavo Cubillos
Quillay 2580
Depto. 502
BOLIVIA Providencia
Santiago, Chile
James Yazman Phone: (56-2) 234-2951
Instituto Boliviano de Fax: (56-2) 233-7915
Tecnologia Agropecuaria
Programa SR-CRSP
Texas TECH University COLOMBIA
P. O. Box 5783
La Paz, Bolivia Ruben Dario Estrada
Phone: (591-2) 392-551 Box 7984
Bogota, Colombia
Phone: (57-1) 283-3268
CANADA Fax: (57-1) 282-8947
Hugo Li Pun Stella Feferbaum

IDRC Carrera 15 #80-52
250 Albert Street Apartado Aereo 102198
P. O. Box 8500 Bogota, Colombia
Ottawa, Canada K1 G3H9 Phone: (57-1) 256-6701
Phone: (613) 236-6163 Fax: (57-1) 612-0601
Fax: (613) 567-7749
COLOMBIA
Raul Vera Arnoldo Ruiz

Leader Savanna Program P. 0. Box 532
CIAT 2200 Coronado, Costa Rica
P. 0. Box 6713 Phone: (506) 229-1984
Cali, Colombia Fax: (506) 229-1984
Phone: (57-23) 675-050
Fax: (57-23) 647243 Manuel E. Ruiz
IICA/RISPAL
P. 0. Box 55-2200
COSTA RICA Coronado, Costa Rica
Phone: (506) 229-0222
Enrique Alarcon Fax: (506) 229-4741
I ICA
P. O.Box 55
2200 Coronado, Costa Rica ETHIOPIA
Phone: (506) 229-0222
Fax: (506) 229-4741 M. A. Mohamed-Saleem
ILCA
Mees Baayen P. O. Box 5689
Escuela Medicina Veterinaria Addis Ababa, Ethiopia
Universidad Nacional Phone: (251) 613-215
Heredia, Costa Rica Fax: (251) 611-892
Phone: (506) 237-7833
Fax: (506) 238-1298 Bernard Rey*
ILCA
Federico Holmann P. O. Box 5689
CATIE Addis Ababa, Ethiopia
P. O. Box 74 Phone: (251) 613-215
Turrialba, Costa Rica Fax: (251) 611-892
Phone: (506) 556-6431, ext. 231
Fax: (506) 556-1533
GERMANY
Francisco Romero
Esc. Centroamericana de GUnter Kleemann
Ganaderia GTZ
Apartado 7-4013 Postfach 5180
Atenas, Costa Rica 65726 Eschborn, Germany
Phone: (506) 446-5050 Phone: 49(6196) 79-3281
Fax: (506) 446-5788 Fax: 49(6196) 79-6103
Appendices 493
GHANA KENYA
Walter Alhassan Luis Navarro

Animal Research Institute C.S.I.R. IDRC
P. O. Box 20 P. O. Box 62084
Achimota, Ghana Nairobi, Kenya
Phone: 773551, Accra, Ghana Phone: (245-2) 330-850
Fax: (245-2) 214-583
GUATEMALA Fanny Nyaribo

SR-CRSP
Fernando Navas P. O. Box 252
IICA Office in Guatemala Maseno, Kenya
1 a. Avenida 8-00 Zona 9 Phone: (254) 355-1214
Guatemala, Guatemala Fax: (254) 354-4515
Phone: (502-2) 316-304,362-306
Fax: (502-2) 326-795
MALAYSIA
Hugo Vargas
IICA Office in Guatemala C. Devendra
la. Avenida 8-00 Zona 9 ILRI
Guatemala, Guatemala 8 Jalan 9/5
Phone: (502-2) 326306 46000 Petaling Jaya
Fax: (502-2) 326795 Selangor, Malaysia
Phone: (603) 755-7986
Fax: (603) 757-4493
INDONESIA
IMade Nitis PANAMA

Dept. of Nutrition and Tropical
Forage Science Miguel Sarmiento
Udayana University IDIAP
Denpasar, Bali P. O. Box 58
Indonesia Santiago, Provincia de Veraguas
Phone: (62-361) 23791, ext. 114 Republica de Panama
Fax: (62-361) 36890 Phone: (507) 968-763
Fax: (507) 968-474
SENEGAL
Ola Smith A. Roeleveld

IDRC Royal Tropical Institute
P. 0. Box 11007 Mauritskade 63
Dakar, Senegal 1092 AD Amsterdam
Phone: (221) 244-231 The Netherlands
Phone: (31-20) 568-8423/374
Fax: (31-20) 568-8444
SYRIA
Johannes B. Schiere
Euan Thomson Dept. of Tropical Animal
ICARDA Production
P. O. Box 5466 University of Wageningen
Aleppo, Syria P. O. Box 388, 6700 AH
Phone: (0963) 21-213433 Wageningen, The Netherlands
Fax: (0963) 21-213490 Phone: (31-8370) 83120, 9111
Fax: (31-8370) 83962
TANZANIA
URUGUAY
Jackson A. Kategile
P. 0. Box 1755 Carlos Sere
Mongoro, Tanzania IDRC/LARO
Phone: (255) 562-038 P. O. Box 6379
Plaza Cagancha 1335, Piso 9
Louise Setshwaelo 11.100 Montevideo
SACCAR Uruguay
P. O. Box 3030 Phone: (598-2) 922031 /34
Arusha, Tanzania Fax: (598-2) 920223
Phone: (65) 8383/4
Fax: (65) 8285
U.S.A
THE NETHERLANDS Robert D. Hart

INFORUM
July Leesberg 611 Siegfriedals Rd.
Wageningen Agricultural Kutztown, PA 19530
University USA
Hollandsewg 1, 6706 KN Phone: (215) 683-6383
Wageningen, The Netherlands Fax: (215) 683-8548
Phone: (31-8370) 83374
Fax: (31-8370) 84763
Appendices 495
U.S.A.
Domingo Martinez Joyce Turk

IAP-UMC AID
221 Gentry Hall Washington, D.C. 20523-1809
Columbia, MO 65211 USA
USA Phone: (703) 875-6414
Phone: (314) 882-4746 Fax: (703) 875-5344
Fax: (314) 882-5127
Corinne Valdivia
David Norman Dept. of Agricultural Economics
Kansas State University University of Missouri
Dept. of Agricultural Economics Columbia, MO 65211
Manthattan, KS 66506-4011 USA
USA Phone: (314) 882-4020, 882-6085
Phone: (913) 532-6702 Fax: (314) 882-3958
Fax: (913) 532-6925
LIST OF ACRONYMS AND ABBREVIATIONS
INSTITUTIONS, PROGRAMS AND PROJECTS
ACSAD Arab Center for Studies of Arid Zones and Dry Lands
(Syria)
AFRNET African Feed Resources Network
ARFSN Asian Rice Farming Systems Network
ARNAB African Research Network for Agricultural By-products
ARPON Amelioration de la Riciculture Paysanne a ('Office du Niger

(Mali)
ARPP Nepal Agricultural Research and Production Project
ARPT-WP Adaptive Research Planning Team-Western Province

(Zambia)
ATSAF German Council for Tropical and Subtropical Agricultural

Research
ATSP USAID Agroenterprise and Technology Systems Project
BCCR Banco Central de Costa Rica
BCR Banco Central de Reserva (Peru)
BMZ Bundesministerium fur wirtschaftliche Zusammenarbeit

(Germany)
CAAC Centro Agropecuario "Aguas Calientes" (Tulua, Colombia)
CABO Centre for Agro Biological Research (Wageningen, The

Netherlands)
CARDI Caribbean Agricultural Research and Development

Institute (Trinidad & Tobago)
CATIE Tropical Agricultural Research and Training Center (Costa

Rica)
CE&DAP Centro de Estudios y Desarrollo Agrario del Peru
CENIP Centro de Investigaciones Pecuarias (Dominican Republic)
CGIAR Consultative Group for International Agricultural Research
CIAT International Center for Tropical Agriculture (Colombia)
CICADEP Centro Internacional de Capacitacion en Desarrollo

Pecuario (Colombia)
CIDA Canadian International Development Agency
CIE Centre International de I'Enfance (France)
CIMMYT Centro Internacional de Mejoramiento de Maiz y Trigo

(Mexico)
CIPREDA Centro de Cooperacion Internacional para la Preinversion

Agricola (Guatemala)
CIRAD Centre de Cooperation Internationale en Recherche

Agronomique pour le Developpement (France)
CLEM Centro Latinoamericano de Especies Menores (Tulua,

Colombia)
CMDT Malinese Rural Development Organization (Mali)
CTA Centre Technique de Cooperation Agricole et Rurale

(France)
DGIS Directorate General for International Cooperation

(Wageningen, The Netherlands)
DHV-RIN Dwars Hederik Verhey - Rijks Instituut Natuurbeheer (The

Netherlands)
DIGESEPE Direccibn General de Servicios Pecuarios (Guatemala)

Appendices 499
DLDD Department of Livestock and Dairy Development (Pakistan)
DLO Dienst Landbouwkundig Onderzoek (The Netherlands)
DLS Department of Livestock Services (Nepal)
DOA Department of Agriculture (Nepal)
DRSPR Departement de Recherche sur les Systemes de

Production Rurale (Mali)
DSA Departement Systemes Agraires, CIRAD (France)
EEC European Economic Community
ESPR Etude des Systemes de Production Rural (Mali)
ETES Estudio Tecnico y Economico de Sistemas de Produccion

Pecuaria (a GTZ-CIAT project, Colombia)
FAO United Nations Food and Agriculture Organization
FERTICA Fertilizantes de Centro America (Costa Rica)
FONAIAP Fondo Nacional de Investigaciones Agropecuarias

(Venezuela)
FRMP Farm Resource Management Program (ICARDA, Syria)
FSP Farming Systems Program (ICARDA, Syria)
FSRDD Farming Systems Research and Development Division

(NARC, Nepal)
GO Governmental organization
GTZ Deutsche Gesellschaft fur Teschnische Zusammenarbeit

(Germany)
GUAT Guatemala
HMG His Majesty's Government (Nepal)

[ARC International agricultural research center
ICA Instituto Colombiano Agropecuario (Colombia)
ICAR Indian Council of Agricultural Research (Tulu[a, Colombia)
ICARDA International Center for Agricultural Research in the Dry

Areas (Syria)
ICP Integrated Cereals Project (Nepal)
ICRAF International Council for Research in Agroforestry (Kenya)
ICRISAT International Crops Research Institute for the Semi-Arid

Tropics (India)
ICTA Instituto de Ciencias y Tecnologia Agricolas (Guatemala)
IDA Instituto de Desarrollo Agrario (Costa Rica)
IDB Inter-American Development Bank
IDIAP Instituto de Investigacion Agropecuaria de Panama
IDRC International Development Research Centre (Canada)
IEMVT Institut d'Elevage et de Medecine Veterinaire des Pays

Tropicaux (France)
IICA Inter-American Institute for Cooperation on Agriculture

(Costa Rica)
ILCA International Livestock Centre for Africa (Ethiopia)
ILEIA Information Center for Low External-Input Agriculture (The

Netherlands)
ILRAD International Laboratory for Research on Animal Diseases

(Kenya)
ILRI International Livestock Research Institute (Kenya and

Ethiopia)
Appendices 501
INDECA Instituto Nacional de Comercializaci6n Agricola

(Guatemala)
INIAA Instituto Nacional de Investigaci6n Agraria y Agroindustrial

(Peru)
INIFAP Instituto Nacional de Investigaci6n Forestal, Agricola y

Pecuaria (Mexico)
IRRI International Rice Research Institute (Philippines)
ISABU Institut des Sciences Agronomiques du Burundi
ISAPLAC Information System for Animal Production in Latin America

and the Caribbean (IICA, Costa Rica)
ISNAR International Service for National Agricultural Research

(The Netherlands)
ISRA Institut Senegalais de Recherches Agronomiques
IVITA Instituto Veterinario de Investigaciones Tropicales y de

Altura (Peru)
KAKI Kenya Agricultural Research Institute
KIT Royal Tropical Institute (Amsterdam, The Netherlands)
LEI The Netherlands Agricultural Economics Institute
LPU Livestock Planning Unit (Sri Lanka)
MAG Ministerio de Agricultura y Ganaderia (El Salvador)
MOA Ministry of Agriculture (Nepal)
MTSS Ministerio del Trabajo y Seguridad Social (Costa Rica)
NARC National Agricultural Research Center (Nepal)
NARI National agricultural research institute
NABS National agricultural research systems

NDDP National Dairy Development Project (Kenya)
NDRI National Dairy Research Institute (India)
NEI Nederlands Economisch Instituut
NGO Nongovernmental organization
NLPD National Livestock Projects Division (Federal Livestock

Department, Federal Ministry of Agriculture, Nigeria)
NTSFS Non-Three-Strata Forage System
ORSTOM Institut Frangais de Recherche Scientifique pour le

Developpement en Cooperation (France)
PANESA Pasture Network for Eastern and Southern Africa
PFLP Pasture, Forage and Livestock Program (ICARDA, Syria)
PISA Proyecto de Investigaci6n de los Sistemas Agropecuarios

Andinos (INIAA, Peru)
PLPP Pattoki Livestock Production Project (Pakistan)
PRODERM Proyecto de Desarrollo Rural en Microregiones (Peru)
PROFED Women's Development Program (Mali)
PROFOGAN Programa de Fomento Ganadero (Ecuador)
PROMEGAN Programa de Mejoramiento de la Ganaderia de Doble

Prop6sito (Colombia)
PUDOC Centre for Agricultural Publishing and Documentation

(Wageningen, The Netherlands)
PVO Private volunteer organization
REAPER Reseau Euro-Africain sur Ie Petit Ruminants
RIEPT Red Internacional de Evaluaci6n de Pastos Tropicales

(CIAT, Colombia)
Appendices 503
RIMISP Red Internacional de Metodologfa de investigacibn de

Sistemas de Produccibn (IICA, Chile)
RISPAL Latin American Research Network for Animal Production

Systems (IICA, Costa Rica)
RLIP Range and Livestock Improvement Project (Yemen Arab

Republic)
SACCAR Southern Africa Centre in Cooperation in Agricultural

Research (Tanzania)
SENA Servicio Nacional de Aprendizaje (Bucaramanga,

Santander, Colombia)
SILEP Smallholder Integrated Livestock Extension Project (Sri

Lanka)
SNES Service National des Etudes et Statistiques (Burundi)
SR-CRSP Small Ruminant Collaborative Research Support Program

(USA)
SRD Regional Society for Development (Burundi)
SRUPNA Small Ruminant Production Systems Network
UCCH Universidad Catolica de Chile
UCV Universidad Central de Venezuela
UNDP United Nations Development Program
USAID (Also, AID) US Agency for International Development
USC Universidad de San Carlos (Guatemala)
WANA West Asia and North Africa
WINROCK Winrock International (USA)

TECHNICAL TERMINOLOGY
APS Animal production systems
APSR Animal production systems research
APSR/D Animal production systems research and development
APSR/FSR Animal production systems research and farming systems

research
AU animal unit (400 kg liveweight)
cap capita
CFA Or Franc CFA, Communaute Francophone Africaine
cm centimeter
CSR Crop systems research
d day
Dm Deutsche mark
DM Dry matter
DMI Dry matter intake
DPG Dual-purpose goat
e.g. exempli gratia (for example)
FMFI Farmer-managed, farmer-implemented
FSR Farming systems research
FSR/E Farming systems research and extension
GDP Gross domestic product
GNP Gross national product

Appendices 505
g gram
ha hectare
K potassium
KDPG The Kenya dual-purpose goat
kg kilogram
km2 square kilometer
I liter
LP linear programming
m meter
m2 square meter
meq milliequivalent
min minute
ml milliliter
m.a.s.l. meters above sea level
MONTY A software developed by GTZ for herd monitoring
MT metric ton
n number
N nitrogen
NFSD New farming systems development
OFR On-farm research
OFR/FSP On-farm research with farming systems perspective
OSR On-station research

P phosphorus
PANURGE A software developed in Senegal
PC personal computer
RMFI Researcher-managed, farmer-implemented
RMH Range Management Handbook (Kenya)
RMRI Researcher-managed, researcher-implemented
RRA Rapid rural appraisal
SAS Statistical Analysis System (North Carolina, USA)
SEAG Small East African goat
TLU Tropical livestock unit
TSFS Three-Strata Forage System
vs. versus
WADG West African dwarf goat
yr year
ZOOP Objective-Oriented Project Planning

This book was printed at
IICA Headquarters
in Coronado, San Jose, Costa Rica
in March, 1995
with.a press run of 625 copies.
ISBN 92-9039-267 3
Livestock plays several important roles in rural development. It is a

source of income, employment and traction for the rural poor; it
benefits the environment through nutrient cycling; and it serves as
a buffer against climatic and economic instability. Unfortunately,
livestock in extensive systems in the tropics, especially beef
ranching, has been associated with an increased rate of
deforestation that has high economic and environmental costs for
society. Finding solutions to these problems must break away from
long-established paradigms. Animal agriculture can no longer be
production-driven, but must be environmentally-conscious, natural-
resource sparing, and internationally competitive.
Important knowledge that will contribute to finding appropriate

solutions has come from holistic systems-oriented research and
development initiatives in animal agriculture. This book contains
contributions from well-recognized leaders in research,
development, training and international cooperation. Experiences
from Africa, Asia and Latin America illustrate the application of the
systems approach to animal agriculture and the technical and
socioeconomic results obtained. The views from international
agriculture centers and donor institutions provide an overall, in-
depth analysis of the effectiveness of supporting animal production
work, of the future orientations that this activity should take, and of
the new strategies and reforms the institutions should adopt.
AU
CANADA RISPAL

Animal: Global

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Animal: Global

Uploaded by

Copyright:

Available Formats

ANIMAL PRODUCTION SYSTEMS

ANIMAL PRODUCTION SYSTEMS

Proceedings of the Global Workshop

Inter-American Institute for Cooperation on Agriculture

C Inter-American Institute for Cooperation on Agriculture (IICA) / Latin

All rights reserved. Reproduction of this book, in whole or in part, is prohibited

Global Workshop on Animal Production Systems (1991 : San Jos6,

1. Produccion animal. 2. Sistemas de produccion. I. Ruiz,

Research and Development Collection No. 26

The Research and Development Collection is designed to disseminate

San Jose, Costa Rica, 1995

CASE STUDIES ................................... 23

FODDER BANK: IMPROVING THE NUTRITION OF

A CASE STUDY ON DUAL-PURPOSE GOAT

INTEGRATION OF CROSSBRED DAIRY GOATS INTO

THE THREE-STRATA FORAGE SYSTEM FOR

TECHNOLOGY ADOPTION AND COMPETITIVENESS IN

DUAL-PURPOSE CATTLE PRODUCTION IN GUATEMALA:

EXPERIENCES OF THE INTERNATIONAL CENTERS 181

ANIMAL PRODUCTION SYSTEMS RESEARCH IN NEPAL:

EXPERIENCES WITH LIVESTOCK SYSTEMS RESEARCH AT ICARDA

RESEARCH AT CIAT ON RUMINANT PRODUCTION SYSTEMS

ASIAN RICE FARMING SYSTEMS NETWORK: COLLABORATIVE

RISPAL: THE LATIN AMERICAN RESEARCH NETWORK

DONOR EXPERIENCES AND PERSPECTIVES ....... 267

LIVESTOCK PRODUCTION SYSTEMS RESEARCH:

IDRC EXPERIENCES IN THE SUPPORT OF ANIMAL PRODUCTION

LIVESTOCK SYSTEMS RESEARCH AT CIRAD

ANIMAL PRODUCTION SYSTEMS RESEARCH: GTZ's

ANIMAL PRODUCTION SYSTEMS RESEARCH IN THE TROPICS:

SELECTED ISSUES ON ADAPTIVE ANIMAL

NEW ISSUES ..................................... 371

THE ROLE OF LIVESTOCK IN THE DESIGN OF

POLICY IMPACT ASSESSMENT IN SMALL-RUMINANT

GENDER ISSUES IN ANIMAL PRODUCTION

WORK GROUP SESSIONS ......................... 437

CONCLUSIONS AND RECOMMENDATIONS ................. 487

LIST OF ACRONYMS AND ABBREVIATIONS ................. 497

Finding solutions to these problems must break away from long-

to make these institutions more responsive to social and environmental

Important knowledge that will contribute to finding appropriate solutions

Manuel E. Ruiz Hugo Li Pun

For the past two decades, different agricultural research and

Why systems research?

Research with a systems approach was developed in response to

The main concerns in systems-oriented projects included the perceived

Evolution of systems research

Systems research in general is not a new field. It has been conducted

The application of systems theory to the problems of small farms,

other disciplines was realized. Therefore, sociological studies were

In many instances, the complexity of the problems is of such magnitude

The consumption of animal protein in such countries is associated with

Objectives of the present meeting

To analyze case studies of animal production systems research in

To discuss new development concerns and how they can be

To identify constraints on animal production systems research and to

To promote the exchange of information across regions, as well as the

It is hoped that the conclusions to be derived from this meeting will be

Table la. Protein consumption levels, Africa, 1979-81 and 1986-89

Total Animal Total AnimalAnimal protein

Algeria 64.1 11.4 70.4 14.6 2.8

Total Animal Total Animal Animal protein

NE (1) Egypt 72.9 9.4 78.6 10.7 1.4