Original Article

Innovation Diffusion in Conservation Agriculture: A Network Approach

Julio Daz-Josa, Roberto Rendn-Medelb,*, Bram Govaertsc, Jorge Aguilar-vilab and
Manrrubio Muoz-Rodriguezb
a

Cornell International Institute for Food, Agriculture and Development (CIIFAD), Cornell University,
Ithaca, New York, USA.
b
Center of Economic, Social and Technological Research on Agribusiness and World Agriculture
(CIESTAAM), Universidad Autnoma Chapingo, Mxico.
c
International Maize and Wheat Improvement Center (CIMMYT), Mxico.
*E-mail: rendon.roberto@ciestaam.edu.mx

Abstract While discussions promoting networks as a means of enhancing innovation have increased
in the past decade, methods for analysing real-world networks for developing countries remain largely
undeveloped. In this study, data from stakeholders involved in conservation agriculture (CA) were used to
analyse innovations and innovators using a network approach. Results indicate that farmers learn mainly
from other farmers, whereas practices are learned from different sources or pathways and adopted step by
step. CA is not a package and principles are applied depending on the local context. Actors in the network
play different roles in knowledge creation and acquisition. A distinction can be made between prescribed
networks where formal relations and vertical structures predominate, and emerging networks based mainly
on informal relations and a horizontal knowledge structure. The main conclusion is that applying network
analysis helps to identify key stakeholders and facilitate assertive and effective network orchestration.
Tandis que le nombre des discussions mettant en avant les rseaux comme moyen de booster linnovation
sest accru ces dix dernires annes, les mthodes danalyse pour les rseaux rels des pays en
dveloppement restent sous-dveloppes. Dans cette tude, les donnes des acteurs impliqus dans
lAgriculture de Conservation (AC) sont utilises pour analyser les innovations et innovateurs qui utilisent
une approche rseau. Les rsultats montrent que les fermiers apprennent essentiellement dautres fermiers,
alors que les pratiques agricoles de conservation sont apprises partir de diffrentes sources ou
cheminements et sont adoptes pas pas. LAC nest pas un tout uniforme et les principes sont appliqus
selon le contexte local. Les acteurs du rseau jouent des rles diffrents quant la cration et lacquisition du
savoir. On peut faire une distinction entre les rseaux prescrits o les relations formelles et les structures
verticales prdominent, et les rseaux mergents, bass essentiellement sur des relations informelles et sur
une structure horizontale du savoir. La conclusion principale est que lapplication de lanalyse rseau aide
identier les acteurs cls et favorise une orchestration de rseau assure et efcace.
European Journal of Development Research advance online publication, 5 March 2015;
doi:10.1057/ejdr.2015.9
Keywords: innovation
orchestration; Mexico

networks;

conservation

agriculture;

social

network

analysis;

network

Introduction
Current trends in population growth suggest that food insecurity is likely to increase under
climate change (CC) scenarios (Reynolds et al, 2012). Over the coming decades, the most
profound and direct impact of CC will be in agriculture, and food production will increase or
decrease depending on the responsiveness of farmers, researchers, extension workers and all
those involved in food production (Brown and Funk, 2008; Lobell and Burke, 2010). To address
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116
www.palgrave-journals.com/ejdr/

Daz et al
these issues, there is clearly an urgent need to establish networks that focus on creating pathways
for sharing knowledge, information and practices among actors at different levels (Kennel, 2013).
One alternative for addressing CC-related problems in agricultural production is conservation
agriculture (CA). According to Kassam et al (2009), CA increases yields, reduces greenhouse gas
emissions, improves overall ecosystem functioning and reduces vulnerability to CC. CA is based
on three main principles: (i) minimal soil disturbance (minimum tillage) to promote biological
soil health, such as increased microbial diversity that enhances nutrient cycling, nitrogen xing
and bio control of plant pathogens; (ii) permanent soil cover (mulch) to protect the soil from
climatic stress and water erosion; and (iii) crop rotation to protect crops from weeds, diseases and
pests (Hobbs, 2007; Hobbs et al, 2008).
CA can be complemented by other good agronomic and management practices (GAMP) that
improve yield, such as quality seeds, soil fertility management (biofertilizers), and integrated
weed and pest management methods. In combination with GAMP, CA offers important
opportunities to improve production systems (Gupta and Seth, 2007; Laik et al, 2014). In some
cases, special machinery or implements are necessary to adopt reduced till practices for land
preparation. Tools adapted to local conditions are needed at three levels of power usage (man
power, animal traction and mechanized equipment).
However, CA diffusion is limited among small-scale farmers in tropical countries (Scopel
et al, 2013). Farmers in countries where CA is not practised face intellectual, social and
technological problems that hinder technology adoption (Friedrich and Kassam, 2009). The
spread of CA depends on strategies that help disseminate and adapt practices to local knowledge
and conditions (Thierfelder and Wall, 2011; Johansen et al, 2012).
Knowledge generation, exchange and use are included in the Technological Change trilogy
proposed by Joseph Schumpeter: invention (science), innovation (generation and development of
new ideas and their rst use) and diffusion (the spread of these new technologies to potential
users) (Stoneman and Diederen, 1994). However, the diffusion process can also be viewed from a
network perspective, in which individuals and institutions play different roles as key conduits of
information and potential catalysts in the diffusion of innovations (Jackson and Yariv, 2011).
Therefore, understanding network-based innovation diffusion could help close agricultural gaps
to ensure food security in the face of CC.
Several authors have studied the diffusion of innovations through networks from different
perspectives (Wejnert, 2002) and using different models (Burt, 1987; Valente, 1996). In this
article, we use the agricultural innovation systems approach in which a network is dened as an
innovation space where actors (individuals and organizations) interact with one another and are
connected in some way (Spielman et al, 2011). Thus, in this study CA innovation networks are
made up of individuals and institutions involved in knowledge development, dissemination and
use as a way of managing the CA diffusion process through prescribed and emerging relationships. Spielman et al (2009) suggest a number of methods and tools such as social network
analysis1 (SNA) to understand how agents interact within the technological change process.
In agriculture, there are examples of theoretical and empirical network analysis applications
(Conley and Udry, 2001; Monge et al, 2008; Matuschke and Qaim, 2009; Spielman et al, 2011;
Maertens and Barrett, 2012; Hermans et al, 2013); all agree that networks are important in innovation
diffusion. Studying innovation diffusion through networks is essential when adapting CA systems to
local conditions (Thierfelder and Wall, 2011). Diffusion occurs from two perspectives: (i) diffusion
into a population by introducing an innovation and (ii) diffusion within a population by contagion,
social inuence or through a social learning process2 (Strang and Meyer, 1993; Young, 2009).
In this study, external sources of knowledge are dened as research institutes, universities, and
government and extension services, whereas internal sources are dened as other farmers
2

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Innovation Diffusion in Conservation Agriculture

(relatives, neighbours and friends). An example of external and internal sources of knowledge
complementing each other is if a farmer hears about a new CA practice from a research institute
and talks about it to his/her neighbours. External sources can play different roles such as:
(i) promoting a dialogue between the developers and users of an innovation through knowledge
facilitation; (ii) network brokerage, for example, to organize a platform for a meeting or help
obtain public subsidies to promote activities; or (iii) managing the innovation process to maintain
network stability and coordination and enhance technological change (Howells, 2006; Klerkx
and Leeuwis, 2008; Klerkx and Aarts, 2013).
Three key features dene internal sources and the roles they play: (i) prestige (enjoyed by
people who have been identied by their peers as someone to turn to for knowledge) is likely to
be a true source of inuence in the diffusion process (Iyengar et al, 2011); (ii ) spatial proximity,
given that new adopters are usually found near farmers who have already adopted an innovation
(Rogers, 1995; Nyblom et al, 2003); and (iii) weak ties, a concept proposed by Granovetter
(1973), who argues that individuals or groups with strong relationships share the same
information and therefore have little to contribute to each other, while weak ties, which are
difcult to identify, add new information to the network.
From the network perspective, relationships can be differentiated as being vertical (imposed or
formal structure) or horizontal (interdependency relationships) (Hadjikhani and Thilenius,
2005). Within external/vertical and internal/horizontal sources and perspectives, there are
different forms of knowledge and innovation. Explicit knowledge refers to recent or infrequent
knowledge, whereas tacit knowledge is created through individual experience (Chilton and
Bloodgood, 2008). There are two modes of innovation: (i) the Science, Technology and
Innovation (STI) mode that is based on a formal process for generating and using explicit
knowledge; and (ii) the Doing, Using and Interacting (DUI) mode in which tacit knowledge or
know-how is acquired through an informal learning process (Jensen et al, 2007). Combining both
modes leads to improved innovation capabilities.
Lastly, innovation diffusion includes two main and interrelated factors: (i) innovations, that is,
the creative use of various forms of knowledge in response to economic or social demands
(OECD, 1999); and (ii) innovators, persons or institutions that create techniques to integrate the
developments, production and marketing activities at an early stage of the development work.
Innovators can also inuence the probability that an innovation will be adopted (Freeman, 1991;
Wejnert, 2002).
From the above dichotomies, two network structures may be distinguished: (i) prescribed
networks consisting of formally specied relationships between superiors and subordinates
and (ii) emerging networks with informal patterns of interaction, in which individuals have a
particular interest (Ibarra, 1992). Figure 1 combines multiple dichotomies to clarify the diffusion
process. Actors play different roles and knowledge dissemination takes different paths within these
two structures.
The research reported in this article aims to analyse the diffusion of CA innovations through a
network approach, by addressing two questions: (i) Which CA innovations have farmers adopted
and where did they learn them? and (ii) Who are the key innovators in the network and what roles
do they play in the diffusion of CA practices? These questions were explored from a network
perspective with the objective of identifying barriers and opportunities concerning the promotion
of a more sustainable agricultural system.
Information from maize producers involved in Mexicos Sustainable Modernization of
Traditional Agriculture3 (MasAgro) initiative was used to analyse innovations and innovators
through a network approach. The article concludes by stressing the importance of establishing
networks focusing on CA practices as an alternative strategy for mitigating the impact of CC.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

Daz et al

Figure 1: Multiple dichotomies between prescribed and emerging networks.

Note: The arrows represent the vertical and horizontal dimensions of the relationships.

Methodology
Data Collection
Data collection took place in the Mexican state of Chiapas, where a survey was conducted
involving 545 maize producers. This was complemented by 15 semi-structured interviews with
key actors in the network and 6 focal group meetings with research institutions and producer
organizations, as well as participant observation. Network data were collected using an egocentric approach to capture direct personal information from stakeholders, which is ideal for
studying the effects of social networks on individuals (Wejnert, 2010). Network analysis uses
specic terms that are dened in Table 1.
To obtain information on learning relationships among CA practitioners, we asked each farmer
to list the CA and GAMP practices he/she applies and indicate from whom he/she learned them. A
questionnaire was administered using snowball sampling to select participants (Wasserman and
Faust, 1994). This type of sampling has advantages for capturing network properties (Maertens and
Barrett, 2012) and saves resources in large networks such as the ones included in this case study.
The disadvantage of this sampling method is that it can miss specic groups of actors if they are not
well connected, or if the sample does not include a large portion of the network (Scott, 2000; Ekboir
et al, 2006).
Data Analysis
Centrality measures are one of the most commonly used tools to identify key nodes in SNA. The
Centrality of a node captures all communications activity and the people who control these
communications, and also identies the people who occupy critical positions in the network
(Freeman, 1978; Valente et al, 2008).
However, Borgatti (2005) proposes Key Player measures that select a small set of population
members to use as seeds for the diffusion of practices (www.analytictech.com/). This study
applied several Centrality principles and Key Player measures to detect communication activity
and its control in the network (Table 2).
4

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Innovation Diffusion in Conservation Agriculture

Table 1: Basic terms used in SNA
Terms

Denitions

Network/Graph

A graph or network is a pair (N,g), where g is an nn adjacency matrix on the set of

nodes and gij indicates the relationship between i and j. A graph displays points that
represent nodes (individuals and institutions) and lines that represent edges
(knowledge relationships)
Also referred to as a vertex, agent, actor or player, a node can be a producer,
extension worker or research centre, among others
Interest node within a network (for example, maize producers who promote CA)
Node connected to an ego (for example, an individual or institution connected to a
maize producer)
Network edge among egos and alters. A relationship between two nodes i and j.
Edges are also referred to as links or ties, for example, knowledge relationships
about CA practices
Number of knowledge edges that a node has to or from other nodes. Degree of a
node i in a network (N,g) is the number of neighbours that i has in the network, so
that di(g)=|Ni(g)|
Pathway where each node and edge is used only once
Shortest path connecting two nodes (individuals or institutions)
A graph is directed if g is required to be asymmetric so that gijgji, and is undirected
otherwise. Unilateral relationships are important in directed networks, which, for
example, can be used to answer the question: Who do maize producers learn from?
Density represents the proportion of potential relationships in a network that are
actually present. In a directed graph, density=l/g(g1), where the total number of
possible lines is equal to the total number of pairs of points
Network index that measures the difference in Centrality (relative importance of a
node within a graph) between the most central node and all others
When there is a tie from i to j, and also from j to h, then there is also a tie from i to h;
in other words, friends of my friends are my friends

Node
Ego
Alter
Edge (link/tie)
Degree
Path
Geodesic distance
Directed graph
Density
Centralization
Transitivity

Source: Borgatti (2005), Hanneman and Riddle (2005), Jackson (2011).

The interviews and focus group discussions were transcribed and systematically analysed to
contrast the questionnaire results. This ensured the reliability and validity of the data through
triangulations of information. Innovation adoption patterns were analysed using ggplot
(Wickham, 2009). For network analysis the statistical package R was used, and the extensions
sna (Butts, 2010) and igraph (Csardi and Nepusz, 2012) were used to calculate Centrality
measures. The Key Player software (Borgatti, 2006) was used to identify strategic nodes in
harvest and diffuse indicators.

Results
Innovation Adoption and Knowledge Sources
Figure 2a details the adoption patterns of eight CA and GAMP practices, in which 21.5 per cent
of the producers adopted ve to seven practices, 69.2 per cent two to four and 9.4 per cent one or
none. Figure 2b shows the adoption index of these practices. Weed management, use of quality
seed and minimum tillage are the practices most adopted by farmers, as they seek to reduce labour
costs, increase yields and improve soil fertility.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

Table 2: Graph indices and Centrality Key Player measures in the CA innovation network
Descriptiona

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Graphlevel indices
Density
Measure that relates the number of observed
relationships to the number of possible
relationships
Centralization Measure that exposes the difference in Centrality
(relative importance of a node within a graph)
between the most central node and others
Transitivity
Measure that indicates the transitive relationships.
For i, j and h actors. When there is a tie from i to j,
and also from j to h, then there is also a tie from i
to h
Modularity
Networks naturally divide into communities or
modules. Modularity is the degree of network
grouping into subgroups (Newman, 2006)
Centrality and Key Player measures
Betweenness Measure that identies nodes with paths most
travelled. A node is central if it lies between other
nodes and connects the path between them
(Wasserman and Faust, 1994)
Leverage
Measure that assesses the degree of Centrality of a
node, with respect to the degrees of its immediate
neighbours with which it interacts. A node is
locally important as it relates to nodes that have a
high degree of Centrality (Joyce et al, 2010)
Alpha Centrality Measure that assesses nodes that are connected to
many nodes, which in turn are connected to many
other nodes (Bonacich and Lloyd, 2001)
Harvest
Diffuse

Utility

Indices

Mean

Standard
deviation

Identies the level of interaction in the network

0.003

When analysing knowledge interactions it is

important to know who is referenced in the
learning of new practices
A high transitivity indicates stronger relationships
between network members. Friends of my friends
are my friends

0.192

0.470

It is important to identify integrated groups as

they correspond to sets of producers who perform
agricultural practices and have similar behaviours

0.782 (21
communities)

Finds nodes that act as bridges or connectors for

communicating information

2.809

4.228

Identies local nodes that are important for

disseminating information. In this study, these
local nodes are producers who are referred to by
other producers as learning sources

0.031

0.076

0.511

0.146

0.221

0.146

0.202

Identies nodes that are cited by many others and

are therefore the most important information
resources for innovation. In the present study, this
measure captures only external sources of
learning
Measure that locates key nodes that receive
Identies key nodes that are connected and
learning relationships from others (Borgatti,
embedded within the network to be used as seeds
2006)
for the diffusion of practices or attitudes
Measure that locates key nodes that send learning Identies key information collecting nodes
relationships to other nodes or information
seekers (Borgatti, 2006)

For theoretical and mathematical development of each measure, see the corresponding references.

0.2

Daz et al

Concept/
measure

Innovation Diffusion in Conservation Agriculture

Figure 2: Pattern and adoption index of CA practices. (a) Adoption pattern; (b) Adoption index: (1)
Minimum tillage, (2) Machinery, (3) Mulch, (4) Improved seed, (5) Crop rotation, (6) Biofertilizers, (7)
Weed control, (8) Pest control.
Note: Shaded areas in Figure 2a show the three groups that conform to adoption patterns of CA and
GAMP.

Some CA principles such as the use of mulch and crop rotation have low adoption, while
minimum tillage is practised by 51.2 per cent of farmers, probably due to previous initiatives
promoting minimum tillage in Mexico. As an example, in 1998 producers in Chiapas formed a
minimum tillage club as part of an FIRA4 project, which was highly successful in reducing the
practice of burning residue.
The use of mulch is a major constraint for some producers because they have to apply direct
seeding in plots with residue cover. This study found that 52 per cent of farmers rent machinery
and 42.6 per cent have no access to any type of equipment. To overcome this problem, it is
necessary to adapt machinery at the three levels of power usage (man power, animal traction and
mechanized equipment) to local conditions. The fact that just a small number of farmers use
specialized machinery (Figure 2b) implies that there are limitations for adopting a comprehensive
CA package. We found that although a farmer may have the interest, often they do not have the
resources to access specialized machinery.
Mulch can also constrain crop rotations. Many farmers produce crops that have different
growth habits, which can make handling residue challenging. For example, rotating maize and
peanuts with full residue retention makes the peanut harvest difcult, if not impossible.
With respect to sources of information about CA and GAMP, these were classied based on
the type of actor and whether they are external or internal. Results indicate that farmers learn
mainly from other farmers, and that the importance of information sources differs depending on
the type of practices they use (Figure 3).
Practices such as biofertilizer use are learned from research institutions. Those relating to
marketing or intellectual property rights, such as quality seeds, are learned from input suppliers.
Mechanization has a low adoption index (Figure 2b), and farmers learn how to use machinery
from other farmers (Figure 3) because there is no strategy that provides funding and training for
this important component.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

Daz et al

Figure 3: Information sources. Producers learn each CA practice from a different source and at a different
level. Example: considering only those producers who apply biofertilizers, 84.6 per cent learned the
technique from a research institute.
External source: Government, input supplier, research institute and organization; Internal source: Producers.

Key Actors in Innovation Diffusion

Density describes the general connectedness in a graph but is lower in directed graphs, as in this
case. Centralization close to 20 per cent indicates a high domain of some nodes about
relationships. Transitivity as a proportion indicates that there may be strong ties in the network.
Network modularity indicates a high degree of grouping and the existence of 21 groups in the
interior network that exhibit strong ties in learning relationships (Table 2).
Centrality and Key Player measures were calculated for each actor. External sources were
detected by the alpha Centrality score in statistical analysis, which was complemented by data
gained through participant observation. On the other hand, internal sources were detected by
identifying the highest Centrality and Key Player scores and then combining those scores with the
theoretical denition of each measure.
Network visualization gives an overview of the relationships in the diffusion process. Figure 4
shows the presence of star-like congurations, in which actors concentrate knowledge relationships from others. This indicates that certain actors control the knowledge activity.
The presence of key actors at the centre of star-like congurations within the network is
important because the information they exchange and the role they play inuences the probability
of innovation adoption. This study performed a selection of key actors through SNA, as a set of
network members that can function as innovation champions and seeds for the diffusion of CA
and other GAMP practices.
The Role of Key Actors in Innovation Diffusion
A typology of key actors is shown in Table 3. Stakeholders of different types are identied as
internal and external sources of learning. External resources are systemic instruments5 for
8

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Innovation Diffusion in Conservation Agriculture

Figure 4: CA network in Chiapas, Mexico. Vertex labels indicate nodes with the highest Centrality and
Key Player scores: Alpha Centrality (A); Leverage (L); Betweenness (B); Harvest (H); Diffuse (D);
Stakeholders with low scores (S).

research support that play different roles, such as formulating policy design strategies, providing
innovations and generating technology roadmaps. Public innovation consultants are divided into
two groups: (i) public innovation consultants related to individual farmers who work to link
systemic instrument efforts and (ii) public innovation consultants related to organized farmers
using public funding. Finally, private consultants focus on marketing their products.
There are three internal sources of learning: (i) local bases for innovation, who are ampliers
via prestige and concentrate learning relationships from other actors; (ii) peer managers, who are
bridges that use weak ties to provide new information to the network; and (iii) collectors, who
need spatial proximity to learn and share information.
From these elements we identied two structures: (i) prescribed networks, where relations
among superiors and subordinates are hierarchical, and characterized by the predominance of
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

Daz et al
formal relations, external sources of information, emphasis on explicit knowledge and
innovations as key elements; and (ii) emerging networks where peer relations are more
egalitarian, characterized by the predominance of informal relations, internal sources of
information for learning, an emphasis on tacit knowledge and innovators as key actors who
spread information.

Discussion
The Adoption of CA Together with GAMP
CA principles are applied together with GAMP depending on the local context. In this study, the
STI innovation mode is based on the effectiveness of external information sources and explicit
knowledge, whereas the DUI innovation mode considers strong internal information sources
where tacit knowledge or know-how determines how producers adapt, adopt and respond to
innovation. Both modes work simultaneously in the diffusion process, but knowledge takes
different trajectories (Figure 3).
There are practices, such as mulch, biofertilizers and quality seeds that require external support
(public or private). On the other hand, internal sources adapt knowledge to local conditions, learn
from other producers and only involve external sources to improve certain techniques. However,
bottlenecks are created when practices/crops are simultaneous, such as maize and peanut
production, or are reliant on each other, such as mulch and minimum tillage. According to
Andersson and DSouza (2014) minimum tillage without residue retention can be more harmful
to agro-ecosystem productivity.
It is important that public policymakers consider that CA practices involve a cost to producers,
for they must invest time in learning and buying capital goods (Pretty, 2008). The presence of
interrelated innovations is an aspect that should also be considered, as many simultaneous
decisions may be involved (Gershon et al, 1985) that require a long-term vision. CA principles
require reorganizing the production process (Scopel et al, 2013). Because CA is knowledgeintensive and a complex system to learn, the initial phase of its adoption is a slow process
(Kassam et al, 2009), which Lahmar (2010) qualies as step by step in the adoption pattern,
as shown in Figure 2a. This pattern is seen in other parts of the world as well. In Southern Africa
Andersson and DSouza (2014) found that CA practices were adopted in partial or incremental
form, with minimum tillage being the rst component to be adopted. In addition, in Central Asia
adoption of CA in temperate zones is limited to one or two components rather than a combination
of all practices (Kienzler et al, 2012). Adapting machinery to local conditions is another challenge
to overcome. It has been shown that adapting machinery design to new farming systems, can
facilitate the practices of minimum tillage and the seeding into residues. (Hobbs, 2007; Kienzler
et al, 2012; Kirkegaard et al, 2014).
The Spread of CA
If the evolutionary approach (Nowak and Higheld, 2011) is used to analyse innovations, an
innovation that has a selective advantage would spread faster, and one with a selective
disadvantage would spread more slowly. In CA, a selective advantage means that a producer
generates prots by applying one practice (b/c>1) in the short term. A selective disadvantage
would mean that a producer would make a great investment in learning and capital goods, but
would only generate a return in the long term. This is one of the greatest impediments to the
10

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Innovation Diffusion in Conservation Agriculture

adoption of CA, as Thierfelder et al (2013) show that under CA, increases in crop yields take
between three and ve seasons to be signicant. However, there is a greater probability that an
innovation will reach the entire population if producers have more selective advantages. This
is known as the xation probability of innovation.
How do networks help to increase the xation probability of an agricultural innovation? First,
network mapping helps to develop a strategic plan to promote ampliers and hinder suppressors;
key actors as star-like congurations (see Figure 4) are ampliers (whenever there is reciprocity
in relationships), whereas suppressors are nodes connected to a single root (Lieberman et al,
2005) where information is concentrated. Second, the networks help detect individuals who
practise innovations with selective advantages that can promote the diffusion of practices that
give good results, that is, in a triadic structure, a non-adopter tied to two adopters or promoters of
an innovation will end up adopting it (Monge et al, 2008).
Kienzler et al (2012) states that both a better diffusion of information, and more awareness
and learning between farmers and policymakers about the advantages of CA are necessary to
achieve adoption of CA practices. This demands the creation of ties between farmers, markets
and organizations that provide services. The degree of network integration will inuence the
probability of innovation xation, the ability to detect groups or communities through
modularity and the design of specic strategies to promote better agronomic practices.
Furthermore, in accordance with Marcoux and Lusseau (2013), the network modularity,
as well as the size of the clusters or communities, has an effect on the cooperation within the
network.
Integrating Networks to Promote CA
Given the varied actors, roles and beliefs in a network, certain mechanisms are needed to align
strategies towards a common end. Klerkx and Aarts (2013) addressed the need to shape the
direction and system of innovation under the concept of orchestration, dened as the
establishment of a set of strategies to implement an action. Valente (2012) uses the term
network intervention when describing how social networks are used to accelerate the
diffusion of innovations and improve organizational efciency. Examples of these forms of
intervention are:
(i) Identication of individuals who act as champions. In Table 3, the local bases for
innovation play this role through prestige and are called referents. However, those actors
are not always the best means of spreading information. Peer managers who act as
bridging individuals can create more cohesive networks (Valente and Fujimoto, 2010)
and improve new information through the use of weak ties.
(ii) Segmentation of the network, which refers to identifying groups of people (through
modularity) who change how something is done at the same time; a clear example of this
is the minimum tillage club located in Chiapas. This is a group of people (a community
of practice) who share an interest and work to deepen their knowledge (Wenger et al,
2002) but also establish their own rules.
(iii) Induction of behaviours, which refers to the identication of actors with high reach
(coverage) in the network who can help implement media campaigns promoting CA benets.
(iv) Changing relationships in the network, in which the alteration strategy takes a dynamic
perspective and adds/deletes nodes, adds/deletes ties or rewires existing ties. For example,
in this study collectors can be linked to local bases for innovation to promote
knowledge exchange or delete nodes to deter suppressors.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

11

12

Table 3: A typology of the role of external and internal sources of knowledge

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

1. Diffusion into a
population
External sources
Explicit knowledge
STI mode
Formal relationships

Inuence level

(2.1) Farm level

Network measure

Role

Observations

Stakeholders

Innovation process
management

Catalyse innovation,
promote networks.
Provide infrastructure

International and
national research
institutes, government,
universities

Demand articulation

(a) Public consultants Alpha Centrality

(individual producers)

Demand articulation
/network brokerage

Platform meetings,
knowledge exchange
Knowledge
facilitation

Research institutes,
universities
Public extensionists as
individuals

(b) Public consultants Alpha Centrality

(organized producers)

Network brokerage

Public extensionists as
a team/Despachos

(c) Private consultants Alpha Centrality

Network brokerage

(a) Local bases for

innovation

Leverage
Centrality/harvest

Prestige/spatial proximity
for learning

Knowledge
facilitation, help
provide public
subsidies
Commercialize their
products
Combine explicit and
tacit knowledge.
Ampliers for
innovation

(b) Peer managers

Betweenness

Weak ties

(c) Collectors

Diffuse

Spatial proximity

(1.1) Policy and innovation (a) Systemic actor for

support level
innovation support

(1.2) Brokerage level

2. Diffusion within a
population
Internal sources
Tacit knowledge
DUI mode
Informal relationships

Type of actor

Source: Klerkx and Leeuwis (2008) using Borgatti (2006), Howells (2006).

Alpha Centrality

Bridges between peers

and external sources.
Capable of articulating
or disrupting the
network

Private consultants
Producers as strategic
nodes for the diffusion
of innovations in local
environments

Producers
participating in
decision making
levels, linking other
producers and
organizing the
network at local level
They tell others about Producers that
new practices and the mention many other
good or bad outcomes producers

Daz et al

Predominant
components

Innovation Diffusion in Conservation Agriculture

Conclusions
The whole network structure shown in Figure 4, and the network analysis carried out, helps to
identify key stakeholders and facilitate assertive and effective network orchestration. Allies of the
network orchestrator must identify both the brokerage and farm levels (Table 3), as they play
different roles and imply different paths for information. Locating key actors at the organizational
level can be relatively simple if you have a basic knowledge of the environment. However, in
complex networks and at the farmer level, this task can be complicated, and use of the SNA tool
can provide valuable information.
In the vertical dimension of the diffusion process (prescribed network), intervention strategies
are dened by the role of systemic actors for innovation as well as public and private consultants.
However, this process is truncated where producers have informal methods for functional
behaviour (emerging networks). In a horizontal sense, the diffusion process may begin from the
bottom levels, where internal sources act as ampliers, bridges or collectors for knowledge
exchange. This latter process is truncated by the top levels when formal structures are dened to
make things happen.
Finally, the results demonstrate that studies on innovation diffusion must consider two
dimensions: (i) innovation as technological practices, whose adoption patterns and sources of
information play different roles in learning; and (ii) innovators at different levels of the network
structure that produce, manage, articulate, adapt, adopt or replicate an innovation as part of a
dynamic process. Future research should consider a dynamic analysis of the behaviour of
learning relationships given that they change over time; trust is built and broken, new
technologies and thus new sources of learning emerge, and those producers and institutions that
were at one time inuential may not be in the future. Dynamic analysis will allow better
assessment of the innovation adoption pattern and the role of innovators as change agents.

Acknowledgements
This work was carried out under the project Mapeo de Redes de Innovacin TTF 2013-019 by the Center
of Economic, Social and Technological Research on Agribusiness and World Agriculture (CIESTAAM) in
collaboration with International Maize and Wheat Improvement Center (CIMMYT). Financial support of the
Mexicos Ministry of Agriculture, Livestock, Rural Development, Fisheries, and Food (SAGARPA) is
acknowledged, as well as two anonymous reviewers for their helpful comments.

Notes
1.
2.

3.

4.

SNA is used to analyse ties between actors, and the patterns and implications of these ties within a social
structure.
Contagion: people adopt when they come into contact with others who have already adopted; social
inuence: people adopt when enough people in the group have adopted; social learning: people adopt
once they see enough empirical evidence to convince them that an innovation is worth adopting; see
Young (2009).
MasAgro, an initiative of Mexicos Ministry of Agriculture, Livestock, Rural Development, Fisheries
and Food (SAGARPA) and the International Maize and Wheat Improvement Center (CIMMYT), brings
together national and international organizations in partnership with innovative Mexican farmers to
obtain higher and more stable crop yields.
Trust Funds for Rural Development (FIRA) is a second-tier development bank that offers credit and
guarantees, training, technical assistance and technology-transfer support to the agriculture, livestock,
shing, forestry and agribusiness sectors in Mexico.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

13

Daz et al
5.

The systemic instruments of innovation are actors and/or institutions that construct and organize the
systems of innovation, and provide technological platforms for learning and experimentation, as well as
infrastructure. Moreover, they promote demand articulation, market strategy and development vision;
see Smits and Kuhlmann (2004).

References
Andersson, J.A. and DSouza, S. (2014) From adoption claims to understanding farmers and contexts: A
literature review of conservation agriculture (CA) adoption among smallholder farmers in Southern
Africa. Agriculture, Ecosystems & Environment 187: 116132.
Bonacich, P. and Lloyd, P. (2001) Eigenvector-like measures of centrality for asymmetric relations. Social
Networks 23(3): 191201.
Borgatti, S. (2005) Centrality and network ow. Social Networks 27(1): 5571.
Borgatti, S. (2006) Identifying sets of key players in a social network. Computational & Mathematical
Organization Theory 12(1): 2134.
Brown, M.E. and Funk, C.C. (2008) Food security under climate change. Science 319(5863): 580581.
Burt, R.S. (1987) Social contagion and innovation: Cohesion versus structural equivalence. American
Journal of Sociology 92(6): 12871335.
Butts, C.T. (2010) sna: Tools for social network analysis. R package version 2.20, http://CRAN.Rproject.
org/package=sna, accessed 24 May 2013.
Chilton, M.A. and Bloodgood, J.M. (2008) The dimensions of tacit & explicit knowledge: A description and
measure. International Journal of Knowledge Management 4(2): 7591.
Conley, T. and Udry, C. (2001) Social learning through networks: The adoption of new agricultural technologies in Ghana. American Journal of Agricultural Economics 83(3): 668673.
Csardi, G. and Nepusz, T. (2012) The igraph software package for complex network research, http://igraph.
sf.net, accessed 13 May 2013.
Ekboir, J.M., Muoz, M., Aguilar, J., Rendn, R., Garca, J. and Altamirano, R. (2006) On the Uneven
Distribution of Innovative Capabilities and Why that Matters for Research, Extension and Development
Policies. Washington DC: International Food Policy Research Institute (IFPRI). ISNAR, Discussion
Paper 7, 23 pp.
Freeman, C. (1991) Networks of innovators: A synthesis of research issues. Research Policy 20(5):
499514.
Freeman, L.C. (1978) Centrality in social networks conceptual clarication. Social Networks 1(3): 215239.
Friedrich, T. and Kassam, A. (2009) Adoption of conservation agriculture technologies: Constraints and
opportunities. In: Proceedings of the 4th World Congress on Conservation Agriculture; 4-7 February,
New Delhi, India: Lead Papers, pp. 257264.
Gershon, F., Just, R.E. and Zilberman, D. (1985) Adoption of agricultural innovations in developing
countries: A survey. Economic Development and Cultural Change 33(2): 255298.
Granovetter, M.S. (1973) The strength of weak ties. American Journal of Sociology 78(6): 13601380.
Gupta, R. and Seth, A. (2007) A review of resource conserving technologies for sustainable management of
the rice Wheat cropping systems of the Indo-Gangetic plains (IGP). Crop Protection 26(3): 436447.
Hadjikhani, A. and Thilenius, P. (2005) The impact of horizontal and vertical connections on relationships
commitment and trust. Journal of Business & Industrial Marketing 20(3): 136147.
Hanneman, R.A. and Riddle, M. (eds.) (2005) Introduction to Social Network Methods. Riverside, CA:
University of California.
Hermans, F., Stuiver, M., Beers, P.J. and Kok, K. (2013) The distribution of roles and functions for upscaling
and outscaling innovations in agricultural innovation systems. Agricultural Systems 115: 117128.
Hobbs, P.R. (2007) Conservation agriculture: What is it and why is it important for future sustainable food
production? The Journal of Agricultural Science 145(2): 127137.
Hobbs, P.R., Sayre, K. and Gupta, R. (2008) The role of conservation agriculture in sustainable agriculture.
Philosophical Transactions of the Royal Society B: Biological Sciences 363(1491): 543555.
Howells, J. (2006) Intermediation and the role of intermediaries in innovation. Research Policy 35(5):
715728.
Ibarra, H. (1992) Structural alignments, individual strategies, and managerial action: Elements toward a
network theory of getting things done. In: N. Nohria and R.G. Eccles (eds.) Networks and Organizations. Boston, MA: Harvard Business School Press.
14

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116

Innovation Diffusion in Conservation Agriculture

Iyengar, R., Van den Bulte, C. and Valente, T.W. (2011) Opinion, leadership and social contagion in new
product diffusion. Marketing Science 30(2): 195212.
Jackson, M.O. (2011) An overview of social networks and economic applications. In: A.B. Jess Benhabib
and O.J. Matthew (eds.) Handbook of Social Economics. North-Holland, the Netherlands: Elsevier B.V.
Chapter 12.
Jackson, M.O. and Yariv, L. (2011) Diffusion, strategic interaction, and social structure. In: A.B. Jess
Benhabib and O.J. Matthew (eds.) Handbook of Social Economics. North-Holland, the Netherlands:
Elsevier B.V. Chapter 14, pp. 645678.
Jensen, M.B., Johnson, B., Lorenz, E. and Lundval, B.. (2007) Forms of knowledge and modes of innovation. Research Policy 36(5): 680693.
Johansen, C., Haque, M.E., Bell, R.W., Thierfelder, C. and Esdaile, R.J. (2012) Conservation agriculture for
small holder rainfed farming: Opportunities and constraints of new mechanized seeding systems. Field
Crops Research 132: 1832.
Joyce, K.E., Laurienti, P.J., Burdette, J.H. and Hayasaca, S. (2010) A new measure of centrality for brain
networks. PLoS ONE 5(8): e12200.
Kassam, A., Friedrich, T., Shaxson, F. and Pretty, J. (2009) The spread of conservation agriculture: Justication, sustainability and uptake. International Journal of Agricultural Sustainability 7(4): 292320.
Kennel, C.F. (2013) Knowledge action networks and regional climate change adaptation. Technovation
33(45): 107.
Kienzler, K.M. et al. (2012) Conservation agriculture in Central Asia What do we know and where do we
go from here? Field Crops Research 132: 95105.
Kirkegaard, J.A., Conyers, M.K., Hunt, J.R., Kirkby, C.A., Watt, M. and Rebetzke, G.J. (2014) Sense and
nonsense in conservation agriculture: Principles, pragmatism and productivity in Australian mixed
farming systems. Agriculture, Ecosystems & Environment 187: 133145.
Klerkx, L. and Aarts, N. (2013) The interaction of multiple champions in orchestrating innovation networks:
Conicts and complementarities. Technovation 33(67): 193210.
Klerkx, L. and Leeuwis, C. (2008) Matching demand and supply in the agricultural knowledge infrastructure: Experiences with innovation intermediaries. Food Policy 33(3): 260276.
Lahmar, R. (2010) Adoption of conservation agriculture in Europe: Lessons of the KASSA project. Land
Use Policy 27(1): 410.
Laik, R. et al. (2014) Integration of conservation agriculture with best management practices for improving
system performance of the rice Wheat rotation in the Eastern Indo-Gangetic plains of India. Agriculture, Ecosystems & Environment 195: 6882.
Lieberman, E., Hauert, C. and Nowak, M.A. (2005) Evolutionary dynamics on graphs. Nature 433(7023):
312316.
Lobell, D. and Burke, M. (2010) Economic impacts of climate change on agriculture to 2030. In: M.P.
Reynolds (ed.) Climate Change and Crop Production. Wallingford, UK: CABI, v.1, pp. 3849.
Maertens, A. and Barrett, C.B. (2012) Measuring social networks effects on agricultural technology adoption. American Journal of Agricultural Economics 95(2): 353359.
Marcoux, M. and Lusseau, D. (2013) Network modularity promotes cooperation. Journal of Theoretical
Biology 324: 103108.
Matuschke, I. and Qaim, M. (2009) The impact of social networks on hybrid seed adoption in India. Agricultural Economics 40(5): 493505.
Monge, M., Hartwich, F. and Halgin, D. (2008) How Change Agents and Social Capital Inuence Adoption
of Innovations among Small Farmers: Evidence from Social Networks in Rural Bolivia. Washington
DC: International Food Policy Research Institute (IFPRI), Discussion Paper 00761.
Newman, M. (2006) Modularity and community structure in networks. Proceedings of the National Academy of Sciences of the United States of America 103(23): 85778582.
Nowak, M.A. and Higheld, R. (2011) Super Cooperators: Altruism, Evolution, and Why We Need Each
Other to Succeed. New York: Free Press.
Nyblom, J., Borgatti, S., Roslakka, J. and Salo, M.A. (2003) Statistical analysis of network data An
application to diffusion of innovation. Social Networks 25(2): 175195.
OECD (1999) Managing National Innovation Systems. Paris, France: OECD, p. 188.
Pretty, J. (2008) Agricultural sustainability: Concepts, principles and evidence. Philosophical Transactions
of the Royal Society B: Biological Sciences 363(1491): 447465.
Reynolds, M.P., Hellin, J., Govaerts, B., Kosina, P., Sonder, K. and Braun, H. (2012) Global crop
improvement networks to bridge technology gaps. Journal of Experimental Botany 63(1): 112.
2015 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research 116

15

Daz et al
Rogers, E.M. (1995) Diffusion of Innovations, 4th edn. New York: The Free Press.
Scopel, E. et al. (2013) Conservation agriculture cropping systems in temperate and tropical conditions,
performances and impacts: A review. Agronomy for Sustainable Development 33(1): 113130.
Scott, J. (2000) Social Network Analysis: A Handbook. London; Thousand Oaks, CA: Sage Publications.
Smits, R. and Kuhlmann, S. (2004) The rise of systemic instruments in innovation policy. International
Journal of Foresight and Innovation Policy 1(1/2): 432.
Spielman, D.J., Ekboir, J. and Davis, K. (2009) The art and science of innovation systems inquiry: Applications to Sub-Saharan African agriculture. Technology in Society 31(4): 399405.
Spielman, D., Davis, K., Negash, M. and Ayele, G. (2011) Rural innovation systems and networks: Findings
from a study of Ethiopian smallholders. Agriculture and Human Values 28(2): 195212.
Stoneman, P. and Diederen, P. (1994) Technology diffusion and public policy. The Economic Journal
104(425): 918930.
Strang, D. and Meyer, J. (1993) Institutional conditions for diffusion. Theory and Society 22(4): 487511.
Thierfelder, C., Mwila, M. and Rusinamhodzi, L. (2013) Conservation agriculture in eastern and southern
provinces of Zambia: Long-term effects on soil quality and maize productivity. Soil and Tillage
Research 126: 246258.
Thierfelder, C. and Wall, P.C. (2011) Reducing the risk of crop failure for smallholder farmers in Africa
through the adoption of conservation agriculture. In: A. Bationo, B. Waswa, J.M. Okeyo, F. Maina, and
J.M. Kihara (eds.) Innovations as Key to the Green Revolution in Africa. The Netherlands: Springer,
pp. 12691277.
Valente, T.W. (2012) Network interventions. Science 337(6090): 4953.
Valente, T.W. (1996) Social network thresholds in the diffusion of innovations. Social Networks 18(1):
6989.
Valente, T.W., Coronges, K., Lakon, C. and Costenbader, E. (2008) How correlated are network centrality
measures? Connect (Tor), NIHPA Manuscripts 28(1): 1626.
Valente, T.W. and Fujimoto, K. (2010) Bridging: Locating critical connectors in a network. Social Networks
32(3): 212220.
Wasserman, S. and Faust, K. (1994) Social Network Analysis, Methods and Applications. New York:
Cambridge University Press, p. 857.
Wejnert, B. (2002) Integrating models of diffusion of innovations: A conceptual framework. Annual Review
of Sociology 28(1): 297326.
Wejnert, C. (2010) Social network analysis with respondent-driven sampling data: A study of racial
integration on campus. Social Networks 32(2): 112124.
Wenger, E., McDermott, R.A. and Snyder, W. (2002) Cultivating Communities of Practice: A Guide to
Managing Knowledge. Boston, MA: Harvard Business School Press.
Wickham, H. (ed.) (2009) ggplot2: Elegant Graphics for Data Analysis. New York: Springer.
Young, H.P. (2009) Innovation diffusion in heterogeneous populations: Contagion, social inuence, and
social learning. The American Economic Review 99(5): 18991924.

16

2015 European Association of Development Research and Training Institutes 0957-8811

European Journal of Development Research 116