ETH Library

Transdisciplinary Systems
Research to Reduce the Cocoa
Swollen Shoot Virus Disease in
Ghana

Doctoral Thesis

Author(s):
Andres, Christian

Publication date:
2018

Permanent link:
https://doi.org/10.3929/ethz-b-000271975

Rights / license:
In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection.
For more information, please consult the Terms of use.
 DISS. ETH NO. 25049

TRANSDISCIPLINARY SYSTEMS RESEARCH TO REDUCE THE

COCOA SWOLLEN SHOOT VIRUS DISEASE IN GHANA

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH
(Dr. sc. ETH Zurich)

presented by
CHRISTIAN ANDRES

MSc ETH Agr

born on 22.07.1986
citizen of Aarwangen, BE

accepted on the recommendation of

Prof. Dr. Johan Six, examiner

Prof. Dr. Andreas Gattinger, co-examiner
Dr. Philipp Weckenbrock, co-examiner
Dr. Christian Pohl, co-examiner

2018
ii
Table of Contents

List of Figures....................................................................................................................................... iv
List of Tables ........................................................................................................................................ vi
Abbreviations ..................................................................................................................................... vii
Abstract ............................................................................................................................................... ix
Résumé ................................................................................................................................................ xi
Acknowledgements ........................................................................................................................... xiii
1. Introduction ................................................................................................................................. 1
1.1 “Ghana is cocoa, cocoa is Ghana” – the economic and social importance of cocoa in
Ghana .................................................................................................................................. 1
1.2 Cocoa swollen shoot virus disease in West Africa and in Ghana ........................................ 4
1.3 Preventive control measures for CSSVD.............................................................................. 6
1.4 The sustainability of cocoa production systems along a gradient of increasing diversity .. 7
1.5 Motivation of the research................................................................................................ 14
1.6 Overview of the structure of the thesis ............................................................................ 14
1.7 Research questions............................................................................................................ 15
1.8 Research design ................................................................................................................. 16
2. Combatting cocoa swollen shoot virus disease: what do we know? ........................................ 19
2.1 Introduction ....................................................................................................................... 19
2.2 Materials and methods ..................................................................................................... 22
2.3 Results ............................................................................................................................... 23
2.4 Discussion .......................................................................................................................... 26
2.5 Conclusion ......................................................................................................................... 29
3. Agroforestry systems can mitigate the severity of cocoa swollen shoot virus disease ............ 31
3.1 Introduction ....................................................................................................................... 31
3.2 Materials and methods ..................................................................................................... 33
3.3 Results ............................................................................................................................... 37
3.4 Discussion .......................................................................................................................... 43
3.5 Conclusion ......................................................................................................................... 45
4. Social barriers to the implementation of measures to prevent the spread of cocoa swollen
shoot virus disease in Ghana ..................................................................................................... 47
4.1 Introduction ....................................................................................................................... 47
4.2 Materials and methods ..................................................................................................... 49
4.3 Results and discussion ....................................................................................................... 53
4.4 Conclusion ......................................................................................................................... 58
5. General discussion, conclusions and outlook............................................................................ 59
5.1 Prevention is better than cure .......................................................................................... 59
5.2 Should diseased farms be cured? The prospect of agroforestry to cope with CSSVD ...... 60
5.3 The need for information to tackle CSSVD ........................................................................ 61
5.4 Changing the perspective – is CSSVD the problem we need to solve? ............................. 62
5.5 The need for multi-dimensional innovation to address the CSSVD issue ......................... 62
5.6 The need for transdisciplinarity in future cocoa research ................................................ 67
References ......................................................................................................................................... 69
Appendix............................................................................................................................................ 83
Appendix A: List of references included in the meta-analysis ...................................................... 83
Appendix B: Reducing the spread of CSSVD – the importance of landscape connectivity ........... 85
Appendix C: Developing a novel mapping tool for CSSVD occurrence ......................................... 89
Appendix D: Personal motivation .................................................................................................. 93

iii
List of Figures
Figure 1: Development of cocoa beans production of the top six producer countries (FAO, 2018). ..... 1
Figure 2: Main stakeholders in the cocoa value chain of Ghana; CHED: Cocoa Health and Extension
Division, CMC: Cocoa Marketing Company, COCOBOD: Ghana Cocoa Board, CRIG: Cocoa Research
Institute of Ghana, LBCs: Licensed Buying Companies, MoFA: Ministry of Food and Agriculture, SPD:
Seed Production Division, QCC: Quality Control Company; →: support (institutional, technical and
organizational), : cocoa beans flow (adapted from Glin et al. (2015)). ............................................... 2
Figure 3: A cocoa tree marks the one Cedi coin on the left and a cocoa pod is shown on the new
(since 2007) 20 Pesewa coin on the right, reflecting its importance in the country. ............................. 3
Figure 4: Current prevalence of cocoa swollen shoot virus disease in Ghana (Data CHED, 2016) ......... 5
Figure 5: Cocoa cultivation in a monoculture (left) and in a dynamic agroforestry system (right) ........ 8
Figure 6: Example of crops in a dynamic agroforestry system in the humid tropics. In the majority of
cases, all species are planted/sown at the same time, and non-crop species are integrated to
accumulate biomass. Crops are grouped according to their lifespan into pioneer, secondary, and
primary species. An optimal succession is achieved by considering the different growth dynamics of
each species (Götsch, 1994; Andres et al., 2016). Source: Figure based on own research. ................. 10
Figure 7: Upper panel: development cycle of plants from vegetative growth to flowering (floración),
maturity (maduración) and withering (marchitamiento,). Middle panel: withering of pioneer species
(maíz = maize) slows down the development of secondary (papaya) and primary species (cacao =
cocoa). Lower panel: optimal successional dynamics characterized by removing species at the right
time in order to create best growing conditions for plant species of the next successional phase (Milz,
1997)...................................................................................................................................................... 12
Figure 8: Localisation of the eligible studies used for meta-analysis .................................................... 24
Figure 9: Mean difference in reduction of infection with CSSVD of different improved practices vs.
control (see section 2.2.1). Horizontal bars show 95% confidence intervals. The numbers displayed
before and in brackets (after the name of the improved practices) represent the number of
comparative data pairs and number of sub-studies, respectively. ....................................................... 24
Figure 10: Plots selected in the Eastern Region .................................................................................... 34
Figure 11: Relationship between shade tree canopy cover and A: CSSVD symptom severity (n=839
trees on 26 plots (13 pairs of full-sun/shade plots)), B: cocoa yield in the main harvest season 2016
(n=438 trees on 22 plots (11 pairs of full-sun/shade plots)). Each point represents the mean value of
all observations from 20 individual trees per plot. Vertical lines indicate optimal shade levels (solid)
and respective 95% confidence intervals (dotted). Red dotted lines indicate the suggested range of
optimal shade levels to balance CSSVD symptom severity versus reduced cocoa yield, based on our
dataset. .................................................................................................................................................. 41
Figure 12: Regression analyses of potential causal factors for shade effects on CSSVD symptom
severity. Individual panels depict the regression analyses between CSSVD symptom severity and a)
number of shade trees, b) number of shade tree species, c) pH, d) organic C, e) total N, f) available P,
g)/h)/i) exchangeable Ca/Mg/K, respectively. Open (○) and closed (●) symbols represent full-sun and
shade plots, respectively. Trend lines and respective Pearson’s correlation coefficients (r) are

iv
displayed for fixed predictors that were statistically significant. Horizontal dotted lines indicate no
statistical significance. ........................................................................................................................... 42
Figure 13: Left panel: vein clearing, a typical symptom of cocoa swollen shoot virus disease (CSSVD,
severe strain 1A) in mature cocoa leaves. Right panel: a group with representatives from all the
major stakeholders in the cocoa value chain of Ghana discussing possible solutions to tackle CSSVD.
Pictures: author. .................................................................................................................................... 48
Figure 14: Operational districts of the Cocoa Health and Extension Division (CHED) in which the study
was carried out: Oyoko (high prevalence of cocoa swollen shoot virus disease (CSSVD)) and Nkawkaw
(low CSSVD prevalence) in the East, Boako (high CSSVD prevalence) and Sefwi Bekwai (low CSSVD
prevalence) in the West. ....................................................................................................................... 50
Figure 15: Most significantly influencing factors on the adoption of any of the four cocoa swollen
shoot virus disease (CSSVD) prevention measures. The size of the circle is proportional to the number
of cases in which the respective factors had a significant effect on the adoption of any of the four
CSSVD prevention measures. ................................................................................................................ 53
Figure 16: Schematic ecosystem health from soil to human. ............................................................... 65
Figure 17: The forest protection strip marked with a red cross seems to have served as a physical
barrier preventing the spread of CSSVD. While all the cocoa areas in front of the strip were infected,
there was no CSSVD behind the forest strip. ........................................................................................ 85
Figure 18: Upper panel: block of about 300 hectares in Brong Ahafo region, where all the cocoa areas
are separated by at least 10 m and no CSSVD occurs (SO: scattered outbreaks). Lower panel: sector
(made up of several blocks) of about 2’000 hectares in Western North region, where the cocoa areas
are continuous and fully affected by CSSVD (AMI: area of mass infection). ........................................ 86
Figure 19: Percentage of block area covered with cocoa related to cocoa area infected by CSSVD. .. 86
Figure 20: Concept to investigate the potential of fragmented cocoa landscapes to reduce CSSVD
using raster-based connectivity analysis and inverse distance weighting (IDW).................................. 87
Figure 21: Spectrum of visible and near-infrared light and absorption patterns of a healthy plant
(green curve), and unhealthy plant (orange curve) and soil (red curve) .............................................. 89
Figure 22: NDVI of sick (left) and healthy potato plant (right). No difference is visible between these
plants with RGB or human eye. ............................................................................................................. 90
Figure 23: Setup for the lab scale analysis of developing the novel mapping tool. Upper panel:
copystand (German “Reprostativ”) with a non-reflecting base surface and two light bulbs. Lower
panel: resulting picture of an infected leaf when using this setup. ...................................................... 91
Figure 24: Upper panel: drone model “GreenCopter® Tropic-Quadro” by RUCON Engineering
(Germany) we planned to use for developing a novel mapping tool for CSSVD occurrence. Middle
panel: taking pictures in the field using a telescopic stick. Lower panel: two magic arms mounted on
top to fix the cameras. A script on the SD card of the cameras ensures that pictures are taken
automatically. ........................................................................................................................................ 92

v
List of Tables
Table 1: Potential benefits provided by cocoa agroforestry systems reported in the literature ........... 9
Table 2: Overview and explanation of preventive control measures for CSSVD and of improved
practices (IP) presented in eligible studies (see section 2.2.1) ............................................................. 21
Table 3: Overview of the studies on preventive control measures for cocoa swollen shoot virus
disease included in the meta-analysis (Appendix A). ............................................................................ 25
Table 4: Characterization of plots compared for CSSVD symptom severity and cocoa yield in the
Eastern Region of Ghana in two seasons in 2016. ................................................................................ 38
Table 5: Soil fertility parameters of plots compared for CSSVD symptom severity and cocoa yield in
the Eastern Region of Ghana in two seasons in 2016. .......................................................................... 40
Table 6: Model results from binary logistic regression analyses about the adoption of the four CSSVD
prevention measures............................................................................................................................. 54

vi
Abbreviations
AF: agroforestry
AFS: agroforestry systems
AMI: areas of mass infection
CHED: Cocoa Health and Extension Division
CMC: Cocoa Marketing Company
COCOBOD: Ghana Cocoa Board
CRIG: Cocoa Research Institute of Ghana
CSSV: cocoa swollen shoot virus
CSSVD: cocoa swollen shoot virus disease
DAFS: dynamic agroforestry system
GIS: geographical information systems
ICT: information and communication technologies
IDW: inverse distance weighting
IP: improved practice
LBC(s): licensed buying company(ies)
MD: mean difference
MoFA: Ministry of Food and Agriculture
NDVI: normalized difference vegetation index
NIR: near-infrared
PAR: photosynthetic active radiation
PC: purchasing clerk
PCR: polymerase chain reaction
QCC: Quality Control Company
qPCR: real-time polymerase chain reaction
RGB: red, green, blue
SO: scattered outbreaks
SPD: Seed Production Division
TAM: technology acceptance model
TCA: true cost accounting
VSS: voluntary sustainability standard

vii
viii
Abstract
Cocoa swollen shoot virus disease (CSSVD) is one of the major factors limiting cocoa (Theobroma cacao
L.) productivity in West Africa. The only effective treatment for CSSVD is to cut infected trees and
replant with disease free planting material. The official eradication campaign in Ghana has cut out
more than 300 million trees since 1946. Hence, the development of preventive control measures and
strategies to mitigate the severity of the disease are of utmost importance. 80 years of research on
preventive control measures have mainly focused on resistance breeding, mild strain cross-protection
(inoculation of cocoa seedlings with a mild strain of the virus to protect against the severe strain) and
control of mealybug vectors. In addition, research has recommended cordon sanitaire, barrier (strip)
cropping and removing alternative host trees as further prevention measures. However, diversification
measures such as agroforestry (for shading) or barrier (strip) cropping have received less attention in
past research, despite their potential to mitigate CSSVD symptom severity or to reduce the spread of
CSSVD. In spite of promising results from research on preventive control measures, CSSVD is more
prevalent in the field than ever before.
The large body of knowledge on preventive control measures for CSSVD is fragmented and many
publications are not easily accessible. Furthermore, the literature has never been systematically
evaluated and quantitatively assessed. Hence, we consolidated this knowledge with an extensive
literature review followed by meta-analysis to identify the pertinent research gaps. Out of 423
publications on CSSVD-related issues, I selected 34 studies, which contained 52 datasets on seven
different preventive control measures. Results showed that resistance breeding and mild strain cross-
protection may reduce CSSVD infection by 30 percent, while the potential of diversification measures
(shading/agroforestry and barrier (strip) cropping) seems to be considerably higher (40 and 85 percent,
respectively). However, there is a lack of evidence because of a low number of studies about
diversification measures, indicating that our results have to be interpreted with care and calling for
more research in this area.
Furthermore, I studied the effects of shade on CSSVD symptom severity, capsid damage and cocoa
yield along a gradient of increasing shade tree abundance in smallholder cocoa farms in Ghana. I also
measured photosynthetic active radiation and assessed soil fertility in order to elaborate on potential
causal factors for possible shade effects on CSSVD symptom severity. Both CSSVD symptom severity
and cocoa yields followed quadratic curves, and were found to be lowest and highest in plots with 54%
and 39% shade, respectively. The simulated optimal shade levels for CSSVD symptom severity and
cocoa yield overlapped between 45% – 53%, indicating that agroforestry systems with around 50%
shade cover may be an optimal coping strategy to balance CSSVD symptom severity versus reduced
cocoa yield until diseased cocoa is replaced with more resistant varieties. Furthermore, my results
suggest that rather than increased soil fertility, reduced high-light stress may have been responsible
for the shade effects on CSSVD symptom severity.
Finally, I evaluated the current adoption of the prevention measures cordon sanitaire, barrier (strip)
cropping, using partly tolerant hybrid seedlings and removing alternative host trees, and identified
their adoption constraints for farmers. I conducted a quantitative survey with 396 farmers in the
Eastern and Western Regions of Ghana, held six focus group discussions with eight to eleven
participants each, and hosted a multi-stakeholder validation workshop with 31 key actors from the
cocoa value chain of Ghana.

ix
Our results show that the adoption of prevention measures remains limited. Farmers with a more
extensive social network, a larger farm size, more secure land tenure rights and more knowledge about
the prevention measures were more likely to adopt them, especially barrier (strip) cropping, hybrid
seedlings and removing alternative host trees. Lack of knowledge about the measures was the single
largest barrier for adoption, with 51% of all the participating farmers not even being aware of the
prevention measures. My results suggest that a farmers’ social network has the highest potential to
be an effective source of information, which was in line with the finding that the information flow
between farmers and other stakeholders in Ghana is a critical factor affecting knowledge spread and
consequently the adoption of CSSVD prevention measures.

The various issues around CSSVD are highly complex, touching different stakeholders and requiring
ecological, technical (biological, mechanical and organisational), socio-political and economic
innovations to be addressed. These include the reshaping of both the design and the management of
cocoa production systems at farm and landscape levels, as well as the development of resistant cocoa
varieties, the production of sufficient barrier (strip) crop and shade tree seedlings and biological
control options for the mealybug vectors. Decentralised nurseries are also desperately needed to
provide farmers with seedlings in a timely manner.

x
Résumé
La maladie du swollen shoot du cacaoyer (CSSVD) est l’un des principaux facteurs limitant la
productivité du cacaoyer (Theobroma cacao L.) en Afrique de l’Ouest. Le seul traitement efficace
contre le CSSVD consiste à couper les arbres infectés et à replanter avec de matériel végétal exempt
de maladie. Au Ghana, la campagne officielle d’éradication a coupé plus de 300 millions d’arbres depuis
1946. Par conséquent, l’élaboration de mesures de lutte préventive et des stratégies pour atténuer la
sévérité de la maladie sont d’une importance capitale. 80 années de recherche sur les mesures de lutte
préventives ont porté principalement sur la sélection de variétés résistantes, la protection croisée avec
des souches atténuées (inoculation des plants de cacao avec une souche bénigne du virus pour
protéger contre la souche sévère) et le contrôle des cochenilles (les vectrices de la maladie). En outre,
la recherche a recommandé cordon sanitaire, barrière (bande) de culture et arracher les arbres
alternatif d’hôtes comme les mesures préventives supplémentaires. Cependant, les mesures de
diversification telles que l’agroforesterie ou barrière (bande) de culture ont reçu moins d’attention
dans des recherches antérieures, malgré leur potentiel pour atténuer la gravité des symptômes CSSVD
ou pour réduire la propagation du CSSVD. En dépit de résultats prometteurs de la recherche sur les
mesures de lutte préventives, CSSVD est plus répandu dans le domaine que jamais auparavant.
Le grand corpus de connaissances sur les mesures de lutte préventives contre CSSVD est fragmenté et
nombreuses publications ne sont pas facilement accessibles. En outre, la littérature n’a jamais été
systématiquement et quantitativement évaluée. Par conséquent, nous avons consolidé cette
connaissance avec une analyse documentaire approfondie suivie de méta-analyse afin d’identifier les
lacunes de la recherche pertinente. De 423 publications sur les questions liées à CSSVD, nous avons
sélectionné 34 études, qui contenait 52 ensembles de données sur sept différentes mesures
préventives. Les résultats ont montré que l’élevage de la résistance de la souche protection croisée
peuvent réduire l’infection CSSVD de 30 pour cent, alors que le potentiel des mesures de diversification
(ombrage/agroforesterie et barrière (bande) de culture) semble être beaucoup plus élevé (40 et 85
pour cent, respectivement). Cependant, il y a un manque de preuve en raison d’un faible nombre
d’études portant sur des mesures de diversification, indiquant que nos résultats doivent être
interprétés avec précaution et en appelant à davantage de recherches dans ce domaine.
En plus, nous avons étudié les effets de l’ombrage sur la sévérité des symptômes du CSSVD, sur les
dégâts des capsides et sur le rendement cacao selon un gradient d’ombrage sur de petites
exploitations agricoles au Ghana. Aussi, nous avons mesuré le rayonnement photosynthétiquement
actif et évalué la fertilité des sols afin de mieux comprendre les facteurs potentiels de causalité des
effets de l’ombrage sur la sévérité des symptômes du CSSVD. Les courbes de sévérité des symptômes
du CSSVD et celle du rendement du cacao ont des tendances quadratiques. Elles se sont révélées plus
basses et plus élevées dans les parcelles ombragées à 54 % et 39 %, respectivement. Les niveaux
simulés d’ombrage optimal pour la sévérité des symptômes du CSSVD et le rendement du cacao se
sont chevauchés entre 45 % et 53 %, indiquant que les systèmes agroforestiers avec environ 50 %
d’ombrage peuvent être une stratégie d’adaptation optimale pour un compromis entre la sévérité des
symptômes du CSSVD et la réduction du rendement de cacao jusqu’à ce que le cacaoyer malade soit
remplacé par des variétés plus tolérant. Nos résultats suggèrent aussi que, plutôt que la fertilité du sol,
c’est le stress dû à la forte luminosité qui est été responsable des effets de l’ombrage sur la sévérité
des symptômes du CSSVD.
Enfin, nous avons évalué l’adoption actuelle des mesures préventives telles que le cordon sanitaire,
les barrières (bande) de culture, l’utilisation de matériel végétal partiellement tolérant (hybrides) et
xi
l’arrachage des arbres hôtes alternatifs et avons identifié les contraintes d’adoption de ces mesures
par les agriculteurs. Nous avons mené un sondage quantitatif avec 396 agriculteurs dans les régions
Est et Ouest du Ghana, effectué six groupes de discussion avec huit à onze participants chacun, et nous
avons organisé un atelier de validation multi-acteurs avec 31 acteurs clés de la filière cacao du Ghana.
Nos résultats montrent que l’adoption des mesures préventives reste limitée. Les agriculteurs avec un
réseau social plus vaste, une plus grande exploitation, des droits de jouissance des terres plus sûrs et
avec plus de connaissances sur les mesures préventives étaient plus susceptibles de les adopter,
surtout la barrière (bande) de culture, les semences hybrides et l’arrachage des arbres hôtes
alternatifs. Le manque de connaissances des mesures préventives a été le plus grand obstacle à leur
adoption, avec 51 % de tous les agriculteurs ayant participé au sondage qui n’étaient pas informés de
leur existence. Nos résultats suggèrent que le réseau social des agriculteurs a le plus grand potentiel
d’être une source efficace de l’information, ce qui était conforme à la conclusion que le flux
d’informations entre les agriculteurs et les autres intervenants au Ghana est un facteur important
affectant la diffusion des connaissances et, par conséquent, l’adoption des mesures préventives.
Les différentes questions à traiter autour de CSSVD sont très complexes, touchant différentes parties
et exigeant des innovations écologiques, techniques (biologiques, mécaniques et organisationnelles),
sociopolitique et économiques. Citons notamment la refonte de la conception et la gestion des
systèmes de production de cacao aux niveaux de la ferme et du paysage, ainsi que le développement
de variétés de cacao résistantes à la maladie, la production d’une quantité suffisante de plantes
barrières (bande) et d’arbres d’ombrages et les options de lutte biologique pour le contrôle cochenilles
(vectrices de la maladie). Des pépinières décentralisées sont aussi désespérément recherchées pour
offrir des pépinières aux agriculteurs en temps opportun.

xii
Acknowledgements
Having had the opportunity to share the last three years of my life with so many lovely people, I
consider myself very fortunate. This thesis would not have been possible without your support,
guidance, patience, and love, so I want to say THANK YOU to all of you! There are too many of you to
count, let alone name, but here I will try to put the most influential people on paper to repay your
kindness for all that you have done for me.

First, I would like to thank Johan Six for giving me the opportunity to deepen my knowledge in this
exciting field of study. I appreciate your support and encouragement to develop my own ideas and
projects. Thanks for accepting to be my Professor on this thesis and welcoming me in the Sustainable
Agroecosystems Group. I appreciate the time I was able to spend with the group and thank Janine
Gremion for her administrative support. I would also like to thank Dr Wilma Blaser for giving very useful
advice concerning my research, and for her constructive inputs and collaboration for two manuscripts.

My sincere acknowledgement goes to Dr Urs Niggli and Beate Huber for their continued commitment,
keeping me engaged at FiBL. For me, the last 6 ½ years at FiBL have been a journey of immense
learning, during which you gave me the necessary freedom to joyfully pursue my passion of agricultural
research for development. I am deeply grateful for this. I am also very happy to be part of FiBL’s
Department of International Cooperation. The many nice activities, great walks, talks and laughs, as
well as inspiring retreats and working sessions were a real pleasure. Thank you Dr Noah Adamtey,
Dr Laura Armengot Martinez, David Bautze, Dr Thomas Bernet, Dr Gurbir Bhullar, Tobias Eisenring,
Dr Salvador Garibay, Dr Irene Kadzere, Nora Kägi, Gian L. Nicolay, Dr Amritbir S. Riar, Dr Toralf Richter,
Monika Schneider, Tetiana Sigg and Paul van den Berge for the lovey time we spent together!

I wish to express my gratitude to both Dr Gilbert Anim-Kwapong (former Executive Director of the
Cocoa Research Institute of Ghana (CRIG) and Dr Franklin Amoah (current Executive Director of CRIG)
who warmly welcomed me at CRIG and provided all the necessary facilities to make my stay a pleasant
and unforgettable experience. The days I spent at CRIG, recovering from hard fieldwork under the
scorching sun, gave me the necessary energy to keep on going. Their support enabled successful data
collection and gathering of the vast body of literature used for this study. Thank you very much!

Special thanks go to my main supervisor in Ghana, Dr Henry Dzahini-Obiatey, as well as to Dr George

Akumfi Ameyaw and Dr Owusu Domfeh (all three CRIG) for their continuous support and inspirational
talks, as well as for their constructive inputs to all three manuscripts. Thanks also to Samuel Larbi
Gyamfi for his assistance in digitalizing grey literature from the CRIG library. I enjoyed the meetings
with my second supervisor in Ghana, Dr Samuel Offei (University of Ghana), especially the talks about
spirituality. Thank you very much for your time and care.

My sincere acknowledgement goes to both Dr Francis Baah (former Executive Director of the Cocoa
Health and Extension Division (CHED), COCOBOD) and Dr Emmauel Opoku (current Executive Director
of CHED), and to Dr Emmanuel Ahia Clottey and Faruk Nyame (both CHED) for ensuring smooth
collaboration with CHED staff, and for providing spatial data that will be used in future projects.

I am very grateful to all the CRIG and CHED staff who assisted me with both field and desktop work,
especially Moses Awiagah, Mr Clifford, Mr Enoch and Mr Ashitey (all CRIG), as well as Mr Nyarko, Mr
Klu and Mr Dickson (all CHED). I also give thanks to Dr Godfred Awudzi (CRIG) for the methodological
inputs to assess capsid damage and to Dr Amos Quaye (CRIG) for support in the soil analysis. I will not
forget the heart-warming discussions and advice I received from Dr Kofi Acheampong (CRIG, retired).

Thank you Dr Andreas Gattinger and Dr Philipp Weckenbrock (both University of Giessen) for being my
co-supervisor and committee members, respectively, and for being great friends. The many happy
reunions we had in the course of the last couple of years, be it at FiBL, in Zurich, or more recently in
Giessen, were among the highlights of my PhD period. Feeling understood and working together for a
greater goal of the common good always left me inspired and full of energy. "Hasta la victoria siempre"
xiii
I would also like to thank Dr Beni Rohrbach (Lucerne University of Applied Sciences and Arts), Dr Philip
Claudio Jörg (University of Zurich) and Andri Baltensweiler (Swiss Federal Research Institute WSL) for
the helpful discussions about the concept of the forthcoming landscape study. Thanks Beni for
preparing the map used for Figure 7 of this thesis, and thanks to Matthew Cooper (University of
Maryland) for getting the first results from the spatial data, looking forward to proceeding on this!
Thanks also to Dr Frank Liebisch (ETH) for the helpful discussions about the concept of the potential
mapping (NDVI) study.

Thanks to Raphael Hörler (Zurich University of Applied Sciences) for your MSc thesis and to Laura
Kuonen (Bern University of Applied Sciences) for her term paper. Raphael, it was a pleasure to spend
two months with you in Ghana, and I am sure we will meet at the beach of Busua again any time soon.

Thanks to Dr Robert Lockwood (University of Reading) and the late Prof Dr Mike Thresh (University of
Reading) for compiling the EndNote library I used for the meta-analysis, as well as to Dr Nicholas Cryer
(Mondelēz International) for making it available. Thanks also to Dr Adrian Müller (FiBL) for the help in
conducting the meta-analysis, and to Fränzi Korner for professional assistance with statistical analysis
and data interpretation. Thanks also to Dr Christian Pohl for being on my committee and making the
effort to be present at my defence, as well as to Dr Alexander Widmer for chairing my PhD defence.

I would like to extend my deep gratitude and admiration to all the cocoa farmers I have met on this
journey, which includes Brigitte Cuendet. Thanks Brigitte for critical thoughts on the topic, and for your
kind hospitality. It is through the continuous work of farmers that we researchers get to make advances
in the development of sustainable cocoa production systems. Thank you, and ayekoooooooooooo!

Thanks to the Sawiris Foundation for Social Development who funded this project through the E4D
scholarship programme of ETH Global, and to Patricia Heuberger and Barbara Becker (both ETH Global)
for their administrative support. Thanks also to World Food System Centre and the University of Basel
for the smaller additional grants we received.

I cannot give enough thanks to my friends and family who supported me all the way, kept me grounded
and sane. Thank you to my family for understanding my drive to accomplish this goal, even though it
meant frequent travelling and living far away for a bit every now and then. Mum, Dad, Dichä and Evi,
my everlasting gratefulness, admiration and love to all of you. To my friends, thank you for your true
friendships and care. Distance makes you stronger, and so it was always a truly enriching experience
when we had the opportunity to meet after a longer period of separation. Wherever I go, I make new
friends, and Ghana was no exception to this. Thank you Israel, Kwame, Moses, Zack and Bobo for
enriching my experience in Ghana. OPC is just being born, so the adventure is just about to start. Fyah!

Last, but not least, I would like to thank my Spiritual Guide, venerable Geshe Kelsang Gyatso Rinpoche
(Geshe-La), as well as my root guru Gen Losang Kelsang and the whole Sangha community at KMC
Zurich from the depths of my heart. I met the holy Buddhadharma of the New Kadampa Tradition in
spring 2014, just around the time I was starting to set up the project that later became my PhD and
finally led to this thesis. The amount of blessings that I have received through the perfect presentation
of Je Tsongkhapa’s doctrine that are Geshe-La’s inspiring modern teachings of Buddhadharma is
inconceivable. Never before had I met people that dedicate their life so fully to making the path of
wisdom and compassion flourish in this world, and thereby make a significant contribution to the noble
goal of World Peace. I pray that you may all have very long and healthy lives so that you can continue
this excellent unmistaken path. NAMO GURUBÄ, NAMO BUDDHAYA, NAMO DHARMAYA, NAMO
SANGHAYA. I dedicate this dissertation to the flourishing of Kadam Dharma throughout the world.

With the sincerest of thanks,

xiv
1. Introduction
1.1 “Ghana is cocoa, cocoa is Ghana” – the economic and social importance of cocoa in Ghana
1.1.1 History
Tetteh Quarshie, a farmer who is said to have brought Amelonado cocoa pods from Fernando Po – a
small island in the gulf of Guinea – to Ghana in 1879, started the rapid evolution of cocoa production
in Ghana (Leiter and Harding, 2004). The average temperature of over 20°C, an average annual rainfall
of at least 1500mm and good soils make Ghana a favourable place for cocoa production (Hütz-Adams,
2011). As cocoa – compared to the traditional alternatives like palm oil or rubber – grows very fast, is
easy to handle, and demands less labour, farmers in current day Eastern Region of Ghana quickly
adopted its cultivation on a large scale.

Under the British colonial government, cocoa production spread to current day Ashanti Region, which
led to Ghana becoming the single biggest cocoa producer in the world by 1910 (Leiter and Harding,
2004). Cocoa production continued to increase until 1964, making Ghana also the world’s largest
exporter of cocoa beans with 581’000 tons per year. After this record figure, however, cocoa
production fell to an all-time low of 166’700 tons in 1984. The reasons for this decline include the
collapse of the cocoa prize in 1965, bad weather conditions, inflation of the Ghanaian currency and
yield losses due to pests and diseases (Kuonen, 2016). Figure 1: shows the history of cocoa production
in Ghana from 1961 until 2014.

Figure 1: Development of cocoa beans production of the top six producer countries (FAO, 2018).

1
1.1.2 The most important actors in the cocoa value chain of Ghana
Today, the main actor in the cocoa value chain of Ghana is undoubtedly the Ghana Cocoa Board
(COCOBOD). Established in the course of the Economic Recovery Programme in 1983 as the successor
of the Cocoa Marketing Board (established in 1947), COCOBOD facilitates the production, processing
and marketing of good quality cocoa. This is done by its various subsidiaries: i) the Cocoa Health and
Extension Division (CHED), ii) the Cocoa Marketing Company (CMC), iii) the Cocoa Research Institute
of Ghana (CRIG), iv) the Seed Production Division (SPD), and v) the Quality Control Company (QCC).
COCOBOD ensures a producer price of more than 50% of the Free On Board (FOB) prize1, and employs
around 60’000 people (Lundstedt and Pärssinen, 2009). A reform in 1993 partially liberalized the
internal marketing, allowing private licensed buying companies (LBCs) to purchase cocoa from the
producers at the farm gate (Dormon et al., 2004). These changes led to the recovery of cocoa
production. Today, Ghana produces around 800’000 tons of cocoa beans (FAO, 2018). The main
stakeholders in the cocoa value chain of Ghana are shown in Figure 2.

Figure 2: Main stakeholders in the cocoa value chain of Ghana; CHED: Cocoa Health and Extension
Division, CMC: Cocoa Marketing Company, COCOBOD: Ghana Cocoa Board, CRIG: Cocoa Research
Institute of Ghana, LBCs: Licensed Buying Companies, MoFA: Ministry of Food and Agriculture, SPD:
Seed Production Division, QCC: Quality Control Company; →: support (institutional, technical and
organizational), : cocoa beans flow (adapted from Glin et al. (2015)).

Farmers grow, harvest, ferment and dry the beans before they sell them to the purchasing clerk (PC).
The PC is a member of the local community employed by an LBC who organizes the purchase and
delivery of cocoa to the LBC warehouse. The LBC is responsible for storing and selling the beans to the
CMC, which stores, markets and exports the cocoa (Kooijmans, 2016).

1
Price invoiced or quoted by a seller that includes all charges up to placing the goods on board a ship at the
port of departure specified by the buyer. Also called collect freight, freight collect, or freight forward.
2
1.1.3 Cultural value of cocoa
Cocoa cultivation in Ghana dates back more than 100 years, playing a crucial role for the country ever
since its introduction in the late 19th century. It is estimated that the cocoa sector now employs around
3.2 million Ghanaians (Lundstedt and Pärssinen, 2009), or 18% of the country’s population above 15
years (Countrymeters, 2017). Gockowski et al. (2011) even estimate that cocoa cultivation supports
more than six million livelihoods. Cocoa delivers 30% of Ghana’s total export earnings, making it the
most important agricultural crop (Monastyrnaya et al., 2016). The long history of cocoa exports also
helped Ghana to maintain a continuous flow of foreign exchange earnings and government revenues.
Rural communities benefitted from new roads, healthcare infrastructure and cocoa scholarships.
Furthermore, the various NGOs involved in the cocoa sector also had beneficial effects for the
development of rural communities. Kuapa Kokoo or the Cadbury Cocoa Partnership for example, invest
millions of dollars into building mobile clinics, school buildings and boreholes (Opoku, 2012). It is
therefore not surprising that the saying “Ghana is cocoa, cocoa is Ghana” is widely used in rural
communities and by CHED officers to start a farmer’s group meeting. Ghanaian cocoa farmers are
proud of their cocoa, which is reflected by their many statements that they love seeing their hard-
produced cocoa beans turned into a valuable product and that cocoa is the backbone of Ghana’s
economy. Cocoa can be seen as the heartbeat of Ghana’s socio-economic development and is given
credit on the old one Cedi coins (since 1967), and the new 20 Pesewa coins (introduced in 2007) shown
in Figure 3.

Figure 3: A cocoa tree marks the one Cedi coin on the left and a cocoa pod is shown on the new (since
2007) 20 Pesewa coin on the right, reflecting its importance in the country.

3
1.2 Cocoa swollen shoot virus disease in West Africa and in Ghana
1.2.1 Pests and diseases in cocoa worldwide and in West Africa
With estimated yield losses of 20% – 40% of global agricultural productivity, crop diseases are an
increasing problem worldwide (Savary et al., 2012). Cocoa (Theobroma cacao L.) is no exception to this
general rule. Pests and diseases common to all three production regions (Africa, Asia and Latin
America) include different species of cocoa mirids (Distantiella theobroma and Sahlbergella singularis
in Africa, Helopeltis bergrothi in Asia and Monalonion dissimulatum in Latin America), as well as the
black pod disease (Phytophthora megakarya and Phytophthora palmivora). Among the major region-
specific pests and diseases are the cocoa pod borer (Conopomorpha cramerella) and Vascular Streak
Dieback (Ceratobasidium theobromae) in Asia, as well as the Witches Broom (Moniliophthora
perniciosa) and the Frosty Pod Rot (Moniliophthora roreri) in Latin America. In Africa, the single biggest
problem is undoubtedly the cocoa swollen shoot virus disease (CSSVD). CSSVD is caused by the cocoa
swollen shoot virus (CSSV), which is a member of the family Caulimoviridae, genus Badnavirus (Muller
et al., 2018) and is the most damaging virus affecting cocoa (Wetten et al., 2016).

1.2.2 Origin of CSSVD and prevalence in West Africa

Ivory Coast, Ghana, Nigeria, Sierra Leone and Togo have reported cases of CSSVD (Muller, 2016). The
disease has been particularly severe in Ghana (Ameyaw et al., 2014) where currently about 17% of all
cocoa areas are affected by CSSVD (Dzahini-Obiatey, personal communication). CSSVD was first
described in Ghana at Effiduase in the New Juabeng district of the Eastern region in 1936 (Steven,
1936) although the disease was probably present in the nearby Nankese township of Ghana from 1922
(Paine, 1945).

As CSSVD was the first disease of cocoa shown to be caused by a virus (Posnette et al., 1950a), it is
likely that it had spread to cocoa from plants indigenous to West Africa. Therefore, Posnette started
to work on the host range of CSSV back in 1940. He tested several plants that are indigenous to West
Africa, and found that trees of the species Ceiba pentandra and Cola chlamydantha were naturally
infected in Ghana. He concluded that Cola chlamydantha trees were an important source of CSSV for
cocoa trees, and should therefore be removed along with other alternative hosts growing near cocoa
whenever possible (Posnette et al., 1950b). Today, however, the spread of CSSVD between cocoa trees
is much more important than the spread from alternative hosts to cocoa trees (Ollennu, personal
communication). While the removal of alternative hosts remains an important aspect of current
intervention strategies, the focus should be put on how to prevent the spread in production landscapes
that are mainly characterized by cocoa trees (Ollennu, personal communication). Some of the
disadvantages of removing alternative host trees are the high costs associated with materials
(chainsaws, fuel, etc.) and labour (see chapter 4), and the reduction of shade, which is vital to reduce
the severity of CSSVD symptoms (see chapter 3).

1.2.3 CSSVD genetic diversity, symptoms and spread

The high genetic diversity with many different strains of CSSV (Muller et al., 2018), as well as different
host genotypes lead to different symptoms such as transient red veins and mottling in young leaves,
different shades of chlorosis in mature leaves and pods, root atrophy and stunting, as well as root and
stem swellings. Highly pathogenic strains cause severe leaf chlorosis, which may result in rapid death
of cocoa trees (Dzahini-Obiatey et al., 2010).

4
Figure 4: Current prevalence of cocoa swollen shoot virus disease in Ghana (Data CHED, 2016)

CSSVD spreads through different species of mealybugs (Hemiptera: Coccoidea: Pseudococcidae). There
are at least 61 mealybug species found on cocoa, of which 19 have been reported in West Africa, and
to date 16 of those are thought to act as vectors of CSSV (Roivainen, 1980). The most important vectors
on cocoa are thought to be Formicococcus njalensis (Laing) and Planococcus citri (Risso), as they are
generally the predominant mealybugs on the crop in the Afrotropical region (Wetten et al., 2016).
Within the vector, the virus is thought to be semi-persistent, with reported differences between the
two species (Posnette and Robertson, 1950). However, novel results indicate that the virus might be
circulative rather than semi-persistent2 in mealybug vectors (Wetten, personal communication). If
these preliminary results were to be confirmed, this would have important implications for the design
of potential CSSVD prevention measures such as barrier (strip) cropping (see section 1.3.4). CSSVD
spread happens in two ways: “radial spread” occurs when relatively immobile adult mealybugs move
along interlocking branches of adjacent trees, while the more mobile first instar nymphs cause “jump
spread” when transported by wind, thereby creating new outbreaks (Jeger and Thresh, 1993).

2
Semi-persistent viral transmission involves the virus entering the foregut of the insect. Viruses that manage to
pass through the gut into the haemolymph and then to the salivary glands are known as persistent. Circulative
is a sub-class of persistent viruses that are protected inside the vectors by a protein, but are not able to
replicate in both the plant and the insect.
5
1.2.4 CSSVD treatment
The only effective treatment for CSSVD is to cut infected trees and replant with disease free planting
material (eradication). Thereby, symptomatic trees are removed together with all adjacent, apparently
healthy trees. When large outbreaks occur, extensive areas are cleared and replanted (Thresh and
Owusu, 1986). In 1946, Ghana launched an official eradication campaign to try to contain the disease.
Since its inception, the campaign cut an estimated number of 300 million cocoa trees, which may
amount to losses of several billion US dollars for both the affected farmers and the government
(Dzahini-Obiatey, personal communication). Ghana’s campaign is said to be the costliest of its kind
worldwide (Thresh et al., 1988; Ampofo, 1997). Despite these efforts, the campaign failed. Reasons for
the failure include the campaign being discontinued several times (Ollennu et al., 1989a; Dzahini-
Obiatey et al., 2006), as well as deficient eradication procedures (Thresh et al., 1988). Consequently,
the disease continues to spread into new areas (Domfeh et al., 2011). Since eradication is the only
effective treatment, CSSVD is a major factor limiting the productivity of cocoa farmers, threatening the
livelihoods of the farmer families. The high prevalence of CSSVD has curtailed many cocoa farmers of
their regular source of income, as whole plantations were lost (Dzahini-Obiatey et al., 2010).

1.3 Preventive control measures for CSSVD

Several institutes in Nigeria, Togo and particularly in Ghana have carried out extensive research on
CSSVD in virology, epidemiology and genetic improvement of cocoa germplasm. Those insights are
vital to better understand the problem, and form the basis for the development of feasible solutions
to the CSSVD menace. However, considering the huge challenges associated with trying to eliminate
the disease outlined above, the development of feasible preventive control measures for CSSVD is of
utmost importance. The last eighty years of research on prevention measures focused mainly on
resistance breeding, mild strain cross-protection (inoculation of cocoa seedlings with a mild strain of
the virus to protect against the severe strain) and control of mealybug vectors (see chapter 2).
However, today we have no CSSVD-resistant variety, nor a feasible approach to implement mild strain
cross-protection at large scale, and attempts to control mealybug vectors failed.

1.3.1 Breeding
Many viral diseases in other crops such as potato, tobacco and tomato have effectively been managed
by increasing varietal resistance (Swaminathan, 1993; Moury et al., 2010). However, a recent
publication by Padi et al. (2013) shows no significant differences in resistance of cocoa trees to CSSVD
between existing and new varieties, underlying the limitations inherent in recycling the minimal
resistance available within a narrow genetic base, and calls for a re-appraisal of variety
recommendations. Resistance breeding may reduce CSSVD infection by up to 30% (see chapter 2), but
the high genetic variability of the virus (Muller and Sackey, 2005; Oro et al., 2012; Abrokwah et al.,
2016) impedes the development of resistant varieties.

1.3.2 Mild strain cross-protection

Just as humans can be vaccinated against viral diseases by injections of a non-pathogenic strain of a
particular virus to protect against the pathogenic strain, so can cocoa trees. The technique of mild
strain cross-protection involves inoculation of cocoa seedlings with a mild strain of the virus to protect
against the severe strain (Ollennu et al., 1999). The potential of mild strain cross protection is just as

6
high as breeding (see chapter 2). While this approach is certainly promising, implementing it on a large
scale is very laborious and costly, and entails several technical and logistical challenges such as
inoculating and distributing several millions of cocoa seedlings to farmers (Domfeh, personal
communication). Furthermore, the high genetic diversity of CSSV makes it difficult to find appropriate
mild strains for different severe strains, and simultaneous protection against several different severe
strains is practically not feasible (Ollennu et al., 1999; Dzahini-Obiatey et al., 2010; Muller, 2016).

1.3.3 Control of mealybug vectors

Controlling mealybug vectors also bears a high potential to reduce CSSVD (see chapter 2). While
systemic insecticides were found to be effective under researcher-managed conditions (Hanna et al.,
1952), a number of problems associated with their use have been reported, such as their high cost,
toxicity to mammals and residues in cocoa beans (Domfeh, personal communication). Most studies on
vector control (e.g. Hanna et al., 1952; Nicol, 1952) have only reported the effect of a particular
insecticide on the abundance of mealybugs, and not on CSSVD infection. However, reporting CSSVD
infection is crucial, as lower numbers of mealybugs do not necessarily translate into reduced CSSVD
infection, especially due to the very patchy distribution of mealybugs in the field (Campbell, 1990).
Therefore, such studies should be repeated under field conditions and in addition to the effects on
mealybug populations, numbers of symptomatic trees should also be reported.

1.3.4 Cordon sanitaire and barrier (strip) cropping

Further prevention measures include cordon sanitaire (Vanderplank 1947) and barrier (strip) cropping
(Domfeh et al. 2016). Those measures involve leaving a 10 m-wide cocoa-free zone or planting a 10 m-
wide barrier of non-host crops (such as citrus, oil palm, coffee, or rubber) around newly planted cocoa.
In both cases, mealybug vectors become non-infective when moving into new plantations, because of
the semi-persistent nature of CSSV in mealybugs mentioned above (see section 1.2.3). Although
cordon sanitaire is effective, it also results in loss of income due to fewer cocoa trees planted.
Furthermore, when farmers leave land in fallow, exposure to direct sunlight may dry out the soil and
subject adjacent cocoa trees to increased water stress, and there is a risk that neighbouring farmers
claim the fallow portion as part of their farm. Barrier (strip) cropping has the advantages of potentially
reducing CSSVD infection by up to 85% (see chapter 2), while providing additional income through
diversified production. However, the seeds of barrier crops (especially the ones of oil palm) attract
rodents who also feed on cocoa pods, and citrus may attract white flies (see chapter 4). Another
disadvantage is the potential loss of income due to lower prices for barrier crops.

1.4 The sustainability of cocoa production systems along a gradient of increasing diversity
The global sustainability of cocoa is at stake (ICCO, 2012). More than half of the world’s supply is
produced in unsustainable monocultures (Figure 5, left) in Ivory Coast and Ghana (Ruf, 2011; Tondoh
et al., 2015; FAO, 2018). There, pressures arise from various ecological, economic and social issues,
such as ageing tree stocks, degraded soils and prevalence of pests and diseases (Wessel and Quist-
Wessel, 2015), as well as vulnerability to climate change (Schroth et al., 2016b), low productivity (van
Vliet and Giller, 2017), low farm gate prices and an ageing farming population (ICCO, 2012).

7
Figure 5: Cocoa cultivation in a monoculture (left) and in a dynamic agroforestry system (right)

1.4.1 Ecological background of the cocoa tree and current production systems
Cocoa originates from the lower strata of the Amazonian forests (Wood and Lass, 2001). Traditionally,
farmers grew cocoa beneath shade tree canopies of primary or secondary forest (Purseglove, 1968,
Rice and Greenberg, 2000). In these so-called agroforestry systems, producers often substitute
external inputs by system-inherent resources such as nutrient cycling through pruning of shade trees
(Vaast and Somarriba, 2014). In line with this, existing manuals for good agronomic practices in cocoa
cultivation recommend between 40% – 50% shade (Beer et al., 2004; SAN, 2005; Opoku-Ameyaw et
al., 2010; Schroth et al., 2016a). However, there has been a shift from shaded agroforestry systems
towards monoculture-like full-sun systems (Ruf, 2011). Using agro-chemicals, monoculture systems
quite rapidly (i.e., within eight to ten years) attain higher yield levels than agroforestry systems (Beer
et al., 1998). However, they often decline in productivity and profitability after about 15 – 20 years
(Ahenkorah et al., 1987; Rice and Greenberg, 2000) due to higher physiological stress (Beer et al.,
1998), epidemic breakouts of pests and diseases (Ahenkorah et al., 1987; Clough et al., 2009a;
Tscharntke et al., 2011) and soil fertility depletion (Rice and Greenberg, 2000; Dawoe et al., 2014).
Therefore, there is an urgent need for more sustainable cocoa production systems (Degrande et al.,
2006; Steffan-Dewenter et al., 2007).

1.4.2 Agroforestry Systems: An Alternative?

Concerning solution to this complex crisis, some hope rests on agroforestry systems (AFS) (Jose, 2009).
In AFS, farmers can produce timber, fruits, fodder, firewood, construction material, ornamentals, and
plants used in medicine and rituals along with cocoa and food crops (Cerda et al., 2014, Jagoret et al.,
2014, Somarriba et al., 2014, Sonwa et al., 2014). AFS could therefore make an important contribution
to the livelihoods and food security of smallholders by decreasing their vulnerability towards price
fluctuations on global markets or outbreaks of pests and diseases (Tscharntke et al., 2011, Duguma et
al., 2001, Bentley et al., 2004, Cerda et al., 2014, Schroth et al., 2000a, Bos et al., 2007, Sonwa et al.,
2007). In addition, agroforestry systems could provide multiple benefits and contribute to a wide array
of ecosystem services as outlined in Table 1. However, the mere association of shade trees and cocoa
may not be sufficient to realize the much-quoted sustainability benefits of AFS (Blaser et al., 2017;
Wartenberg et al., 2017; Andres et al., 2018). Indeed, without adequate management, cocoa
agroforestry (AF) may even be less resilient to sub-optimal and extreme climates than cocoa grown
under full-sun conditions (Abdulai et al., 2017).
8
While there is evidence for the high ecological and social potential of agroforestry systems (Clough et
al., 2009a, Tscharntke et al., 2011, Jacobi et al., 2014), recent literature suggests that under current
market conditions they are often not economically viable in the short term compared to monocultures
(Vaast and Somarriba, 2014). Consequently, the improvement of market conditions for agroforestry
systems is a key factor for their implementation at a larger scale. Furthermore, it is important to
understand the dynamics of economic benefits at farm level in agroforestry systems versus
monocultures (Schneider et al., 2016).

Table 1: Potential benefits provided by cocoa agroforestry systems reported in the literature
Potential benefit Study
Improved pollination (De Beenhouwer et al., 2013)
Long-term stable cocoa (Rice and Greenberg, 2000, Obiri et al., 2007, Bisseleua et al.,
yields 2013)
Longer lifespan of cocoa (Obiri et al., 2007, Ruf and Zadi, 1998)
plantations
Control of pests and (Tscharntke et al., 2011, Smith Dumont et al., 2014, Bieng et al.,
diseases, erosion control 2013, Gidoin et al., 2014, Sperber et al., 2004, Lin, 2011)

Biodiversity conservation (Rice and Greenberg, 2000, Clough et al., 2009b, Fonte and Six,
and enhancement 2010, Sonwa et al., 2007)
Climate change mitigation (Schroth et al., 2013, Somarriba et al., 2013, Somarriba et al.,
through C sequestration 2014, Jacobi et al., 2014, Fonte et al., 2010b, Verchot et al., 2007,
Saj et al., 2013, Clough et al., 2010)
Nutrient cycling (Buresh et al., 2004, Gama-Rodrigues, 2011)
Soil fertility maintenance (Fonte et al., 2010a, Isaac et al., 2007, Tscharntke et al., 2011,
or enhancement Mbow et al., 2014)
Watershed protection (Garrity, 2004)
Reduction of deforestation (Asare, 2006, Clough et al., 2011, Tscharntke et al., 2012)

The definition of agroforestry by the National Agroforestry Centre, United States Department of
Agriculture (2017) is “the intentional combination of agriculture and forestry to create integrated and
sustainable land-use systems”. Thus, AF practice needs to be systematic in order to avoid negative
effects of trees on crops, and to realize potential benefits. Key to realizing the potential of an AFS is
that practitioners chose the shade tree species deliberately, and manage them well (Bai et al., 2017).
Cocoa farmers may know about shade trees functions in cocoa systems, and prefer species that cover
their specific needs (Graefe et al., 2017). Scientists may use this knowledge to recommend companion
trees to smallholders (Gram et al., 2017). However, such sound AF practices have seen low adoption
rates, and systematic approaches to AF are lacking. An example of a systematic approach is the
dynamic agroforestry system (DAFS) (Figure 5, right) (Goetsch, 1992; Götsch, 1994; Andres et al.,
2016). The movie “Life in Syntropy”3 offers insights into the application of this system in Brazil. Practical
experience shows that well-managed DAFS have the potential to enhance resilience (Yana and
Weinert, 2001; Schnatmann, 2006; Vieira et al., 2009).

3
https://www.youtube.com/watch?v=gSPNRu4ZPvE
9
1.4.3 Dynamic Agroforestry: Principles and Examples
“Dynamic”, “successional” and “analog” agroforestry systems are cross-cutting concepts based on
principles of plant density and diversity (Analog Forestry Network RIFA, 2012, Schulz, 2011, Milz, 2012).
Dynamic agroforestry systems are based on the understanding of the succession and structure of
natural ecosystems. The main features of dynamic agroforestry systems are i) high planting densities
and diversity, stratification, and a high energy flow usually without the use of external inputs, and ii)
management practices such as different types of pruning interventions, e.g. rehabilitation, formative,
maintenance pruning, selective weeding or grafting, and selection of healthy, productive planting
material (Götsch, 1994).

Like in natural species sucession, crops are grouped according to their lifespan into pioneer, secondary,
and primary species. All the species are planted or sown at the same time, leading to a “crop
succession” (Figure 6) enriched by the regeneration of native plants (Götsch, 1994). Pioneer species
include rice, cassava, or pigeon peas. These are subsequently replaced until the system is characterized
by secondary species such as pineapple, papaya, and banana, as well as slower growing secondary and
primary tree species which simultaneously develop in their shade. The cocoa tree is a primary species
with a potential life span of more than 100 years (Wood and Lass, 2001). After about 10-15 years, the
secondary species dominate the system, and are eventually replaced by the primary species. Plants
which have completed their life cycle are either harvested or cut down, chopped up, and left to
decompose in the plots to help maintain soil fertility.

Figure 6: Example of crops in a dynamic agroforestry system in the humid tropics. In the majority of
cases, all species are planted/sown at the same time, and non-crop species are integrated to
accumulate biomass. Crops are grouped according to their lifespan into pioneer, secondary, and
primary species. An optimal succession is achieved by considering the different growth dynamics of
each species (Götsch, 1994; Andres et al., 2016). Source: Figure based on own research.

10
The higher, emergent to canopy strata of the system may be occupied by rubber and timber trees, the
middle, low canopy to understory strata by fruit trees including oil palm, and the lower, understory to
forest floor strata, by cocoa trees. This stratification and the planting of tree species with
complementary root systems aims at minimizing the competition for light, water, and nutrients, by
assuring that different ecological niches are occupied (Götsch, 1994). This way, synergies between the
subsequent successional phases are enhanced, and each successional phase creates the necessary
conditions for the plant species of the next successional phase (Milz, 1997, Figure 7).

In a dynamic agroforestry system, every plant is potentially useful. Ernst Götsch (1994) described for
the Brazilian context how invasive pioneer plants can be highly beneficial for the system, as they may
play a vital role in enhancing soil fertility. The continuous addition of organic material from pruning
maintains soil fertility, and the complexity of the system may lead to a natural self-regulation of pests
and diseases (Andres et al., 2016). Götsch (1994) observed more vigorous plant growth and higher
productivity when crops were introduced during an adequate successional phase of the overall system.
The author further describes how secondary species developed well under pioneer species, but not
vice versa, and concluded that productivity depends on sucessional dynamics of the system which can
be stimulated by pruning (Götsch, 1994). Schulz et al. (1994) point out that dynamic agroforestry
farmers oberved a kind of allelopathic effect of maturing plants, reducing the growth of their
neighbouring plants at the end of their life cycle, and a growth stimulating effect of young plants,
increasing the vegetative growth of their neighbouring plants.

A major advantage of dynamic agroforestry systems is that the high crop diversity allows for harvests
and income already during the first year of plantation establishment (Andres et al., 2016). This allows
dynamic agroforestry farmers to avoid the “hunger gap” that occurs in cocoa monocultures, which
only start to produce after four to six years.

The few empirical studies on dynamic agroforestry systems conducted to date compared dynamic
agroforestry plots to monocultures in Bolivia, and show higher productivity and soil fertility in dynamic
agroforestry systems. Todt et al. (2010, 2009) found significantly higher nutrient concentrations,
thicker Ah horizons, and higher organic matter content in dynamic agroforestry systems which were
cultivated for more than 20 years. Milz (2010) studied the damage of fruit flies (Anastrepha spp. and
Ceratitis captitata) in citrus trees. He found more than twice the amount of aborted fruits due to fruit
fly damage in monocultures than in dynamic agroforestry systems. Productivity per orange tree was
significantly higher in dynamic agroforestry systems and the sugar content of the fruit was not
significantly different beween the two systems (Milz, 2010). Gruberg (2011) assessed pests, diseases,
and productivity in dynamic agroforestry vs. low-diversity cocoa and citrus systems, and found less
incidence of witches’ broom and black pod rot, similar cocoa productivity, and a multitude of different
products in dynamic agroforestry systems. Schulz (2011) and Vieira et al. (2009) described how heavily
degraded castor bean (Ricinus communis) monocultures were successfully restored with dynamic
agroforestry systems in Bahia, Brazil. In one study, castor bean production increased by 90%, and total
productivity increased fourfold after the implementation of dynamic agroforestry systems (Schulz,
2011).

11
Figure 7: Upper panel: development cycle of plants from vegetative growth to flowering (floración),
maturity (maduración) and withering (marchitamiento,). Middle panel: withering of pioneer species
(maíz = maize) slows down the development of secondary (papaya) and primary species (cacao =
cocoa). Lower panel: optimal successional dynamics characterized by removing species at the right
time in order to create best growing conditions for plant species of the next successional phase (Milz,
1997).

12
1.4.4 Advantages of shade trees
In cocoa systems, shade trees can have multiple benefits that foster cocoa growth, provided the
species are chosen deliberately and well managed (Bai et al., 2017). Growing cocoa under the canopy
of shade trees reduces the physiological stress on cocoa trees, which fosters their healthy development
(Beer et al., 1998; Wood and Lass, 2001). Some studies have shown i) an improvement of water and
nutrient recycling (Buresh et al., 2004; Smith Dumont et al., 2014), ii) pest and disease control (Schroth
et al., 2000a), iii) maintenance of soil fertility (Schroth et al., 2000b; Bos et al., 2007; Gama-Rodrigues,
2011; Tscharntke et al., 2011), and iv) regulation of micro-climate (Schroth et al., 2016a). Blaser et al.
(2017) found localized positive effects of individual shade trees on soil carbon and nitrogen content,
as well as soil aggregation. These potential positive effects may help maintaining sustainable cocoa
yields in the long-term (Rice and Greenberg, 2000). Furthermore, shade trees were shown to sequester
high amounts of carbon (Abou Rajab et al., 2016), and AFS were found to be equally or more profitable
than full-sun systems under ideal management (Armengot et al., 2016) as well as under farmer-
managed conditions (Rosalien et al., 2017).
In terms of pests and diseases, shade trees can be equally beneficial. For example, they can indirectly
increase the capacity of cocoa trees to defend themselves against diseases such as CSSVD (Babin et al.,
2010; Oro et al., 2012), or contribute directly to pest and disease control by favouring populations of
beneficial organisms (Ahenkorah et al., 1974; Schroth et al., 2000a; Wood and Lass, 2001; Opoku et
al., 2002). Several studies have also shown that shade trees can decrease pest populations (Beer et al.,
1998; Jaramillo et al., 2009; Thorlakson and Neufeldt, 2012). Bigger (1981), for instance, showed that
mealybug vectors of CSSVD were more abundant under full-sun conditions than under shade. In
addition, the number of mealybug predators and parasitoids was higher on shaded plots (Bigger, 1981;
Mani and Shivaraju, 2016). Furthermore, Schroth et al. (2000a) reported that infection with mistletoes,
which indirectly encourage the spread of CSSVD, were higher under full-sun conditions in West Africa.
These results underline the potential of diversification measures such as agroforestry systems or
barrier (strip) cropping to reduce the spread of CSSVD (Schroth et al., 2000a; Domfeh et al., 2016).

1.4.5 Disadvantages of shade trees

However, shade trees may also have adverse effects on the development and health of cocoa trees, as
some are alternative hosts for CSSV, harbouring the virus and serving as a source of infection for cocoa
stands planted nearby (Posnette et al., 1950b). Furthermore, some shade tree canopies may provide
too much shade, increasing relative humidity in cocoa stands and thus favouring fungal diseases
(Schroth et al., 2000a; Babin et al., 2010; Oro et al., 2012).
The mere association of shade trees and cocoa may neither be sufficient to mitigate soil degradation
after deforestation (Wartenberg et al., 2017), nor to enhance soil fertility and soil carbon sequestration
(Blaser et al., 2017; Andres et al., 2018). Depending on the architecture of their root systems, they may
also be competing for water, which may cause drought stress in cocoa trees, especially during the dry
season (Carr and Lockwood, 2011). Without adequate management, shade trees may even decrease
cocoa growth and yields, as well as the resilience of a cocoa plantation to sub-optimal and extreme
climates (Abdulai et al., 2017). Therefore, it is of utmost importance that agroforestry practitioners
chose and manage shade trees well in order to avoid negative effects of trees on crops, and to realize
potential benefits.

13
1.5 Motivation of the research
When trying to tackle a problem as complex as CSSVD, it is important to have a very good idea about
the point of departure, i.e., asking yourself the question: what do we already know? Therefore, I first
embarked on consolidating the large body of knowledge about preventive control measures for CSSVD,
as this information was fragmented and had never been systematically evaluated and quantitatively
assessed before. Not only did this activity give me a comprehensive overview about the knowledge in
the field, it also provided a basis for further research I carried out during my PhD. I found that there is
a significant knowledge gap about the potential of agroforestry systems to reduce the severity of
CSSVD, and that the adoption of promising CSSVD prevention measures by farmers remains limited.

However, only one study carried out under controlled research conditions has indicated the potential
of agroforestry to reduce the severity of CSSVD symptoms (Legg, 1982). This is why I decided to
investigate this potential and see whether reduced CSSVD symptom severity translates into higher
cocoa yield under actual farmers’ field conditions. Furthermore, I wanted to determine the optimal
range of shading to reduce CSSVD. This led to the questions about the definition of “shade” and about
the real processes behind observed differences in shaded compared to non-shaded systems. “Shade”
as such could be effective but also really “just” a proxy for agro-biodiversity, soil fertility, soil moisture
and many other factors. However, studying the underlying causal factors of shade, I found limits to the
feasibility of such research. Nevertheless, I was able to determine which of the factors agro-
biodiversity, soil fertility and high-light stress may have had the strongest influence. The results of this
work are outlined in chapter 3.

However, even if researchers develop effective preventive control measures for CSSVD, these can only
have a real impact if farmers adopt them. I also investigated this factor because I found that the
adoption of promising CSSVD prevention measures by farmers remains limited, and that the main
reasons for this are largely unknown. The results of this work are outlined in chapter 4.

1.6 Overview of the structure of the thesis

The thesis is divided into three main parts with five chapters. The first part consists of a general
introduction, including the research questions and the research design (chapter 1).

The second part of the thesis consists of three self-contained articles. The first article in chapter 2
consolidates the large body of knowledge on preventive control measures for CSSVD with an extensive
literature review followed by meta-analysis to identify the pertinent research gaps. Since the results
reveal both a high potential of shading to reduce CSSVD, as well as a knowledge gap in this area, the
second article (chapter 3) shows results of field research conducted in the Eastern Region of Ghana on
the extent to which agroforestry systems can mitigate the severity of CSSVD. While the results of
chapter 3 suggest that agroforestry systems can indeed mitigate the severity of CSSVD to a certain
extent, CSSVD is still highly prevalent in Ghana and it is known that the adoption of promising CSSVD
prevention measures by farmers remains limited. Therefore, I investigated the social barriers to the
implementation of CSSVD prevention measures in the Eastern and Western Regions of Ghana in the
third article (chapter 4) in order to elaborate targeted policy recommendations and appropriate
knowledge transfer.

14
The last part of the thesis consists of chapter 5, which summarizes the results of the previous chapters
and elaborates implications for policymakers, discusses the research approach and gives an outlook,
suggesting avenues for future research based on the findings.

The Appendix of the thesis presents two concepts for studies that came up during the course of the
research, but have not been brought to completion so far. These are the importance of landscape
connectivity for reducing the spread of CSSVD (Appendix B), and the development of a novel mapping
tool using drone imagery and normalized difference vegetation index (Appendix C). Finally, Appendix
D outlines my personal motivation for the project.

1.7 Research questions

I carefully checked the adequateness of the research questions, which I had formulated in the initial
proposal, as well as the feasibility of their implementation during an extensive induction phase that
lasted from spring to autumn 2015. In this period, I exchanged with various stakeholders from
policymakers through researchers and extension officers to farmers in order to make sure that the
research priorities I set were relevant and targeted at bringing about change for the benefit of all. This
transdisciplinary process was an excellent opportunity to develop a deeper understanding of the
situation the various stakeholders on the ground were facing, and to appreciate the wickedness of the
issues around CSSVD.

A lot of information on CSSVD is available (see review by Dzahini-Obiatey et al., 2010) from published
and unpublished sources (grey literature), but no quantitative assessment of this vast body of data had
been done so far. Therefore, the guiding research question for the first part of the study was:

1. “What is the effectiveness of the main CSSVD control options?”

The influence of diversification on plot scale (shading by agroforestry trees) on CSSVD had not been
studied so far. Therefore, the guiding research question for the second part of the study was:

2. “What are the effects of diversification (at plot scale) on CSSVD symptom severity?”

Finally, the guiding research question to investigate the issue of farmers’ adoption of research-based
recommendations for CSSVD control was:

3. “How can we enhance the adoption of CSSVD control options in Ghana?”

By covering those aspects, I aimed at a holistic understanding of the issues in cocoa production in order
to come up with feasible solutions.

15
1.8 Research design
The research was delineated into the following activities and sub-activities:

Activity 1: Literature review and meta-analysis

Activity 1.1: Literature review about the main CSSVD control options
There is a lot of anecdotal knowledge about the main CSSVD control options, especially in the library
of African institutions such as CRIG. However, this information was fragmented and had never been
systematically evaluated to provide evidence. Therefore, I conducted an extensive review of the
relevant literature from both peer-reviewed and non-peer-reviewed sources as well as grey literature.
Activity 1.2: Preparing the data matrix for meta-analysis
A set of improved practices (e.g. use of resistant varieties, barrier (strip) cropping, etc.) and the effect
size (i.e., reduction of infection due to a particular improved practice) was defined. After that, the
eligible studies were evaluated, followed by parameterization and integration of the information into
a common data matrix for meta-analysis as outlined by Borenstein et al. (2009).
Activity 1.3: Comprehensive meta-analysis
Finally, a comprehensive meta-analysis as outlined by Petrokofsky and Mills (2009) was carried out.
Meta-analysis is a powerful statistical technique to combine and compare results from a range of
independent studies, by weighing the results according to their different precisions, reflected in their
standard deviations and underlying replications. Meta-analysis has rapidly gained popularity as it
offers an opportunity to review the exponentially growing body of knowledge in a way that reduces
bias associated with traditional narrative reviews (Petrokofsky and Mills 2009).

Activity 2: In-depth study of diversification at plot scale

The influence of diversification at plot scale (i.e., shading by agroforestry trees, barrier (strip) cropping,
etc.) on CSSVD symptom severity had not been studied so far. Therefore, I conducted an in-depth study
on this factor.
I estimated shade tree cover as described by Somarriba (2002), identified shade tree species using the
method of Hawthorne and Gyakari (2006) and assessed CSSVD symptom severity on a categorical scale
from 1 – 10 using a slightly adapted method from the common practice described by Padi et al. (2013).
Subsequently, I established the relationships between CSSVD symptom severity, shade levels and
diversification indices (e.g. Shannon Index, etc.), taking into account important biophysical and socio-
economic co-factors (e.g. soil fertility parameters, management practices and capsid damage).

16
Activity 3: Enhancing the adoption of CSSVD control options
This task can be tackled in two ways: bottom-up or top-down, i.e., starting from the farmers (study on
socio-economic adoption constraints) or from the government/industry (study of the political context).
An effective integration of the two perspectives is most likely to create impact.
Therefore, in a first step I obtained data on socio-economic adoption constraints using semi-structured
interviews and focus group discussions with farmers as proposed by Anim-Kwapong and Frimpong
(2006). Such information included farm sizes, land tenure, as well as the farmers’ i) knowledge and
management of CSSVD and improved practices, ii) educational level, iii) affiliation to a
cooperative/farmer group, iv) inclusion in extension programs, v) access to information, vi)
perceptions of current production and willingness to cut and replant cocoa plantations, and vii)
farmers’ age.
Secondly, a study of the political context helped me to understand the complexity in which the CSSVD
problem is embedded, and to identify how challenges in the farmers’ context can best be addressed.
For example, hindering factors for farmers to engage in sustainable cocoa production systems were
elaborated in a transdisciplinary stakeholder workshop that was informed by the previously
established farmers’ perspectives/needs. This led to information that was useful to suggest future
steps toward the elaboration of an implementation action plan to enhance the dissemination of
feasible CSSVD control options (adaptation of existing CHED extension programs).
I aimed at fostering adoption by farmers by applying participatory approaches such as on-farm
research, farmer field days and exchange meetings that emphasized on education about CSSVD
(spread, feasible control options, sources of information, etc.). Furthermore, products relevant for
farmers’ education such as illustrated leaflets and videos on best practices of CSSVD management that
can be shared by farmers on mobile phones using Bluetooth technology were elaborated and
distributed. These activities stimulated mutual learning by integrating different forms of knowledge
and making it visible to different stakeholders.

Activity 4: Publication of Research Findings and Final Thesis Writing

I ensured dissemination of the results among the scientific community by publishing articles in peer-
reviewed journals and through participation in international conferences. Furthermore, I provided my
results to CRIG so that they can share them with organizations active in cocoa research and extension
in other producer countries of West Africa (e.g. CNRA in Cote d’Ivoire, CRIN in Nigeria, IRAD in
Cameroon and CRA-F & ITRA in Togo), which may lead to enhanced adoption by smallholders.

17
18
2. Combatting cocoa swollen shoot virus disease: what do we know?
Andres, C., Gattinger, A., Dzahini-Obiatey, H.K., Blaser, W.J., Offei, S.K., Six, J.4

CSSVD is one of the major factors limiting cocoa productivity in West Africa. The only cure for CSSVD
is to cut out visibly infected trees and the official eradication campaign in Ghana has cut out more than
300 million trees since 1946. 80 years of research on preventive control measures have mainly focused
on resistance breeding, mild strain cross-protection (inoculation of cocoa seedlings with a mild strain
of the virus to protect against the severe strain) and control of mealybug vectors. Meanwhile
diversification measures such as agroforestry (for shading) or barrier (strip) cropping have received
less attention. Despite promising results, CSSVD is more prevalent in the field than ever before. The
large body of knowledge on preventive control measures for CSSVD is fragmented and many
publications are not easily accessible. Furthermore, the literature has never been systematically
evaluated and quantitatively assessed. Hence, we consolidated this knowledge with an extensive
literature review followed by meta-analysis to identify the pertinent research gaps. Out of 423
publications on CSSVD-related issues, we selected 34 studies, which contained 52 datasets on seven
different preventive control measures. Results showed that resistance breeding and mild strain cross-
protection may reduce CSSVD infection by 30 percent, while the potential of diversification measures
(shading/agroforestry and barrier (strip) cropping) seems to be considerably higher (40 and 85 percent,
respectively). However, there is a lack of evidence because of a low number of studies about
diversification measures, indicating that our results have to be interpreted with care and calling for
more research in this area. Future testing is needed to evaluate the efficacy of barrier (strip) cropping
to reduce CSSVD, and address the effect of shade on CSSVD symptom severity. Furthermore, the
practical relevance of different preventive control measures for farmers needs to be assessed, and
shade should be considered in current breeding programs for CSSVD resistance.

2.1 Introduction
CSSVD has plagued cocoa production in West Africa for more than eight decades (Dzahini-Obiatey et
al., 2010). . CSSVD is caused by CSSV, which belongs to the genus Badnavirus (Lot et al., 1991). CSSVD
only occurs in West Africa, and has been reported in Ivory Coast, Ghana, Nigeria, Sierra Leone and Togo
(Figure 8, (Muller, 2016)). The disease has been particularly severe in Ghana, where it was already
observed in 1922 (Paine, 1945) and described in 1936 (Steven, 1936).
Several institutes in Nigeria, Togo and particularly in Ghana have done extensive research on CSSVD in
virology, epidemiology and genetic improvement of cocoa germplasm. CSSV is genetically diverse,
different isolates/strains can cause different symptoms, including transient red veins and mottling in
young leaves, different shades of chlorosis in mature leaves, and pods, root atrophy and stunting, as
well as root and stem swellings. Most of these symptoms are unique to different strains and symptom
expression depends on host genotype. Highly pathogenic strains cause severe leaf chlorosis, which
may result in rapid death of cocoa trees (Dzahini-Obiatey et al., 2010).

4Published as: Andres, C., Gattinger, A., Dzahini-Obiatey, H.K., Blaser, W.J., Offei, S.K., Six, J. (2017) Combatting Cocoa
Swollen Shoot Virus Disease: What do we know? Crop Protection 98: 76-84.
doi:https://doi.org/10.1016/j.cropro.2017.03.010
19
Different species of mealybugs (Pseudococcidae spp.), the most important ones being Formicococcus
njalensis (Laing) and Planococcus citri (Box, 1945; Cornwell, 1953, 1958, 1960; Thresh and Tinsley,
1959; Roivainen, 1971; Bigger, 1981) spread CSSVD. The spread of CSSVD can happen in two ways:
either through “radial spread”, caused by relatively immobile adult mealybugs that move along
interlocking branches of adjacent trees, or through “jump spread”, caused by more mobile first instar
nymphs transported by wind, which create new outbreaks (Jeger and Thresh, 1993).
The only treatment for CSSVD-infected cocoa is to cut out visibly infected trees (eradication). Thereby,
symptomatic trees are removed together with all adjacent, apparently healthy trees. When large
outbreaks occur, extensive areas are cleared and replanted (Thresh and Owusu, 1986). The official
eradication campaign in Ghana, launched in 1946, has cut out more than 300 million trees by 2010,
and was called the costliest campaign of its kind anywhere in the world (Thresh et al., 1988; Ampofo,
1997). However, the eradication campaign failed due to various reasons, such as the campaign being
disrupted and discontinued several times (Ollennu et al., 1989a; Dzahini-Obiatey et al., 2006), and
because of deficient eradication procedures (Thresh et al., 1988). Consequently, the disease continues
to spread into new areas (Domfeh et al., 2011).
Past research has had some focus on the development of preventive control measures to be
implemented along with eradication (Table 2). However, farmers have only to a limited extend
implemented those measures, which may be due to their perspective and situation not sufficiently
being taken into account. For example, farmers said they would not do barrier (strip) cropping with oil
palm because it attracts rodents, which can damage cocoa pods (Wetten, personal communication).
The selection of the suitable trees that match farmers’ preferences and expectations is most likely to
be successful when done in a participatory manner together with farmers, so that their views and
knowledge are taken into account (Vaast and Somarriba, 2014).
There is a lot of anecdotal knowledge about preventive control measures for CSSVD, especially in the
library of African institutions such as CRIG. However, this information is fragmented and has never
been systematically evaluated and quantitatively assessed. In this study, we consolidated this
knowledge in order to provide a basis for future research and development. We conducted an
extensive literature review followed by meta-analysis to elaborate the relative effectiveness of
different preventive control measures for CSSVD. We hypothesised that preventive control measures
related to diversification (barrier (strip) cropping, shading/agroforestry) reduce CSSVD infection
significantly more than preventive control measures related to breeding and mild strain cross
protection.

20
Table 2: Overview and explanation of preventive control measures for CSSVD and of improved practices (IP) presented in eligible studies (see section 2.2.1)
Measure Measure/IP Explanation Advantages Disadvantages Included in References
category meta-analysis?
Breeding/mild Breeding Breeding for varietal resistance to Slows CSSVD by up Limited availability of seedlings Yes (Kenten and Lockwood, 1977;
strain cross CSSVD to 20% Lockwood, 1981a, b; Posnette, 1981;
protection Firempong, 1984; Campbell, 1990;
Dzahini-Obiatey et al., 2006; Dzahini-
Obiatey et al., 2010; Padi et al., 2013;
Ameyaw et al., 2014; Muller, 2016)
Mild strain cross Inoculating cocoa beans/seedlings Adverse effects of Difficult to find appropriate mild Yes (Ollennu et al., 1999 ; Dzahini-Obiatey
protection with mild CSSV isolate/strain to severe strain can isolates/strain for different severe et al., 2010; Muller, 2016)
protect from severe strain be prevented for strains due to high genetic diversity of
several years CSSV, simultaneous protection against
different severe strains impossible,
high cost
Diversification Barrier (strip) Lining newly planted cocoa with 10m Non-host crops Additional income generated may be Yes (Ollennu et al., 2003; Ollennu et al.,
cropping barrier of non-host crop (e.g. citrus, generate less than what could be generated by 2005 ; Dzahini-Obiatey et al., 2006;
oil palm, coffee, rubber, etc.) additional income cocoa Dzahini-Obiatey et al., 2010)
Shading / Shading of cocoa seedlings and See Andres and Lower short-term cocoa yields Yes (Adegbola, no date)
agroforestry plantations (agroforestry) Bhullar (2016)
Eliminating Cordon sanitaire Leaving 10m-wide cocoa-free zone Effective method Less cocoa planted means less income No (Vanderplank, 1947; Dzahini-Obiatey et
sources of around newly planted cocoa for farmers, leaving land under fallow al., 2006; Dzahini-Obiatey et al., 2010)
inoculum could lead to loss of user rights for land
Removing Removing alternative host trees Potential source of Shade for young cocoa plants may be No (Dzahini-Obiatey et al., 2006; Dzahini-
alternative host which naturally harbor the virus infection and missing, and chance to generate Obiatey et al., 2010; Ameyaw et al.,
trees habitat for vectors additional income may be lost 2014; Muller, 2016)
removed
Screening Antiviral Cocoa leaf and embryo extracts as n.a. n.a. Yes (Adomako and Owusu, 1974; Dzahini-
method for substance well as quercetin used to inhibit Obiatey et al., 2005)
resistance CSSV transmission in seeds
breeding (screening methods for resistance
breeding)
Somatic Tissue culture technique to generate n.a. n.a. Yes (Quainoo et al., 2008)
embryogenesis* disease-free clonal propagules from
visibly infected trees (screening
methods for resistance breeding)
Vector control Insecticide Controlling mealybug vectors with Easy application Inefficient due to mealybugs’ water- Yes (Tscharntke et al., 2012 ; Forster et al.,
insecticides repelling (waxy) body and behaviour 2013; Ameyaw et al., 2014; Muller,
(hiding), adverse effects on natural 2016)
antagonists (disrupting functional
biodiversity), no practically feasible
method for farmers available
*
Somatic embryogenesis is of no practical relevance for farmers, but may be engaged to support breeding efforts; n.a. = not applicable

21
2.2 Materials and methods
2.2.1 Data collection
To identify studies reporting preventive control measures for CSSVD we searched the databases of
Web of Science, Scopus and Google Scholar using the following keywords: „cocoa swollen shoot virus
disease“, „cocoa swollen shoot virus“, „swollen shoot virus“, „swollen shoot disease“, „swollen shoot“,
„CSSVD“ and „CSSV“. We gathered further articles from the reference sections of already collected
articles, and from an Endnote Library on CSSVD-related literature, which was compiled by researchers
from CRIG (Ameyaw, personal communication), and the University of Reading, UK (Cryer, personal
communication). These were often technical or annual reports, which were only available as
hardcopies in the CRIG library (grey literature). This procedure led to a total number of 423 articles
related to CSSVD. In the following, preventive control measures will be referred to as “improved
practices”.
We screened all articles and selected those for meta-analysis (in the following referred to as “eligible
studies”) that met the following criteria: i) an improved practice was compared to a control under the
same pedo-climatic conditions, ii) the data presented values of visible CSSVD symptoms (of
seedlings/trees) in the improved practice and control treatment, respectively, and iii) the mean values
of improved practice and control were reported together with a measure of variability (standard
deviation, standard error of the mean or p value) and number of replications (n). This led to the
selection of 34 studies with 52 sub-studies (Table 3) that reported data on seven different improved
practices (Table 2). A full reference list of the papers included in the meta-analysis is provided in
Appendix A. In this paper, the term “study/studies” refers to individual papers/publications, while the
term “sub-study/sub-studies” refers to different experiments/trials presented in the same study.
Furthermore, we used visible CSSVD symptoms as a proxy for CSSVD infection. Actual CSSVD infection
can only be confirmed by serology or nucleic acid-based testing. However, for ease of understanding
we used the term “CSSVD infection” in the text.

2.2.2 Data analysis

We extracted the data from the 52 eligible sub-studies manually and compiled them in a common
Microsoft Excel data matrix. We calculated the reduction of CSSVD infection as follows:

𝐼𝑛𝑓𝑒𝑐𝑡𝑖𝑜𝑛𝐼𝑃
𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑖𝑛𝑓𝑒𝑐𝑡𝑖𝑜𝑛 [%] = (1 − ( )) ∗ 100
𝐼𝑛𝑓𝑒𝑐𝑡𝑖𝑜𝑛𝑐𝑜𝑛𝑡𝑟𝑜𝑙

Where InfectionIP and Infectioncontrol = absolute (number of seedlings/trees) or relative (percentage) values of
CSSVD infection in improved practice and control treatment, respectively.

Thereby, each data pair corresponded to one comparison of an improved practice with the control.
For example, if one study presented data for five partially resistant varieties versus a susceptible
control, this resulted in five data pairs (comparison of each variety with the control). This led to a total
number of 400 data pairs (see legend in Figure 9).
In some cases, there were multiple comparisons of different levels of improved practices with the same
control in one study (e.g. five partially resistant varieties each time versus the same susceptible
control). Since these data are not independent from each other, creating an individual data pair from
each level of the improved practice with the same control led to an overestimation of the control
value’s confidence through artificial repetitions in the dataset. This overestimation was avoided by
22
aggregating the data pairs to sub-study level at the cost of losing some information from the individual
values of the different levels of the improved practices. We chose this procedure because the number
of replications (n) was equal among different levels of the improved practices and the control in most
studies. Therefore, we did not risk to loose information coming from the weighing of the mean values
of the different levels of the improved practices (which depends on the value of n).
Among the eligible studies, there were two groups of studies: studies, which were carried out at the
greenhouse-/lab-scale, and studies, which were carried out at the plot/farm-scale. To account for the
compilation of all studies in one common data matrix, we run two separate meta-analyses for the two
subgroups. This allowed us to check whether their outcomes would differ from the outcome of the
single meta-analysis performed on the pooled dataset. Since the outcomes did not differ, we pursued
the single meta-analysis of the pooled dataset at sub-study level.
We calculated the magnitude of effects (“effect sizes”) as the raw mean difference (MD) of
improved practices over the control using a random effects meta-analysis model as described by
Borenstein et al. (2010). The raw MD was calculated as follows:
𝑀𝑒𝑎𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 (𝑀𝐷) = 𝑚𝑒𝑎𝑛𝑖𝑚𝑝𝑟𝑜𝑣𝑒𝑑 – 𝑚𝑒𝑎𝑛𝑐𝑜𝑛𝑡𝑟𝑜𝑙

MD is used when the scale of the outcomes is equal among all the studies, which was true in our case.
We represented the precision of the effect sizes by weighing the means of each sub-study using their
respective standard deviations and numbers of replications (n). We chose the random-effects model
because it assumes that single effect sizes depend on the study context (differences in methods and
sampling), which was very much the case in our heterogeneous dataset. We estimated the average
true effect and variance Ƭ2 with the restricted maximum-likelihood estimator (REML), and used the
Knapp-Hartung adjustment to account for uncertainties when estimating (residual) heterogeneity. We
found no outliers in our dataset by DFBETAS value analysis (Viechtbauer, 2010). We defined the
weighted MDs of the improved practices to be significantly different from zero or from each other
when the 95% confidence interval (CI) did not overlap zero or the 95% CI of another improved practice.
We conducted all analyses with the “metafor” package of the R statistical software, version 3.3.1
(Viechtbauer, 2010).
2.3 Results
From the five countries with reported CSSVD cases, only three had eligible studies that we included in
the meta-analysis (Figure 8), and the vast majority of the eligible studies were conducted in Ghana
(Table 3). Most of the studies focused on the two improved practices breeding and mild strain cross
protection (56% and 26%, respectively), both of which showed a potential to reduce CSSVD by about
30% (Figure 9). There were only two studies concerning diversification measures (shading/agroforestry
and barrier (strip) cropping). Both improved practices showed a higher potential to reduce CSSVD by
40 % (shading/agroforestry) and 85 % (barrier (strip) cropping), respectively (Figure 9).
Further, we only found one study about the effect of insecticide (vector control) on the number of
visibly infected trees, which was also the only farm-scale study we included in the meta-analysis. The
mean value of 86% (of the 11 consecutive field observations conducted on a monthly interval), shows
a trend towards a high potential of this improved practice (Figure 9). Regarding screening methods for
resistance breeding, we had only one eligible study for somatic embryogenesis and two for antiviral
substance. With a mean value of 89%, the potential of somatic embryogenesis was clearly superior to
antiviral substance (4%).

23
Figure 8: Localisation of the eligible studies used for meta-analysis

Figure 9: Mean difference in reduction of infection with CSSVD of different improved practices vs.
control (see section 2.2.1). Horizontal bars show 95% confidence intervals. The numbers displayed
before and in brackets (after the name of the improved practices) represent the number of
comparative data pairs and number of sub-studies, respectively.

24
Table 3: Overview of the studies on preventive control measures for cocoa swollen shoot virus disease included in the meta-analysis (Appendix A).
Reference Institution and location of study Improved practice (IP) Type of Scale of Number of Age IP c Duration of
publication a comparison sub-studies b measurements d
Adomako and Owusu, 1974 CRIG, Tafo, Ghana Antiviral substance 0 Greenhouse/lab 1 n.a. 1 yr
Dzahini-Obiatey et al., 2005 CRIG, Tafo, Ghana Antiviral substance 1 Greenhouse/lab 1 n.a. 1 yr
Ollennu et al., 2005 CRIG, Bunso, Ghana Barrier (strip) cropping 1 Plot 1 5 yr 11 yr
Adomako, 1977 CRIG, Tafo, Ghana Breeding 1 Greenhouse/lab 1 n.a. 84 d
Adu-Ampomah et al., 1996 CRIG, Tafo, Ghana Breeding 1 Plot 2 3.5 yr 3 yr
Cilas et al., 1988 IRCC, Kpalime, Togo Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Dale, 1957 CRIG, Tafo, Ghana Breeding 1 Greenhouse/lab 1 n.a. 2 yr
Djiepkor et al., 1994 IRCC, Kpalime, Togo Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Kenten and Legg, 1970 CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Kenten and Lockwood, 1977 CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Legg and Kenten, 1970 CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Legg and Kenten, 1971a CRIG, Tafo, Ghana Breeding 0 Plot 1 3.5 yr 9 yr
Legg and Kenten, 1971b CRIG, Tafo, Ghana Breeding 1 Plot 1 4 yr 10 yr
Legg and Lockwood, 1977a CRIG, Tafo, Ghana Breeding 0 Plot 1 3 yr 9 yr
Legg and Lockwood, 1977b CRIG, Tafo, Ghana Breeding 1 Plot 1 n.a. 1 yr
Legg and Lockwood, 1981 CRIG, Tafo, Ghana Breeding 0 Plot 7 3 yr 4 yr
Lockwood, 1981a CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Lockwood, 1981b CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Longworth et al., 1965 CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 1 n.a. 1 yr
Paulin et al., 1994 IRCC, Kpalime, Togo Breeding 1 Greenhouse/lab 3* n.a. 1 yr
Posnette and Todd, 1951 CRIG, Tafo, Ghana Breeding 0 Greenhouse/lab 4 n.a. 1 yr
Tsatsu and Bekou, 2003 IRCC, Kpalime, Togo Breeding 1 Plot 3** 15 yr 15 yr
Hanna and Heatherington, 1957 Chesterford Research Park, Insecticide (vector control) 0 Farm 1 2 yr 2 yr
Atukrum, Ghana
Crowdy and Posnette, 1947 CRIG, Tafo, Ghana Mild strain cross protection 0 Plot 1 6 yr 3 yr
Ollennu and Owusu, 1997 CRIG, Tafo, Ghana Mild strain cross protection 1 Plot 1 6 yr 7 yr
Ollennu and Owusu, 2002 CRIG, Tafo, Ghana Mild strain cross protection 1 Plot 1 0.5 yr 8 yr
Ollennu and Owusu, 2003 CRIG, Tafo, Ghana Mild strain cross protection 0 Plot 1 4 yr 3.5 yr
Ollennu et al., 1996 CRIG, Tafo, Ghana Mild strain cross protection 0 Plot 1 6 yr 7 yr
Ollennu, 1989 CRIG, Tafo, Ghana Mild strain cross protection 1 Plot 1 1.5 yr 1 yr
Ollenu et al., 1999 CRIG, Tafo, Ghana Mild strain cross protection 0 Plot 1 4 yr 1 yr
Owusu et al., 1999 CRIG, Tafo, Ghana Mild strain cross protection 1 Greenhouse/lab 3* n.a. 1 yr
Owusu, 1971 CRIG, Tafo, Ghana Mild strain cross protection 1 Plot 2*** 17 yr 1 yr
Adegbola, -- CRIN, Ibadan, Nigeria Shading/agroforestry 1 Greenhouse/lab 1 n.a. 56 d
Quainoo et al., 2008 University for Development Somatic embryogenesis 0 Greenhouse/lab 2 n.a. 1 yr
Studies, Tamale, Ghana
a
0 = scientific journal article, 1 = books chapter/proceeding/technical report; b Sub-studies are different experiments/trials presented in the same study; c Time period since
adoption of improved practice in years (yr); d Time period of measurements in days (d)/years (yr); CRIG = Cocoa Research Institute of Ghana, IRCC = Institut de Recherches du
Cafe et du Cacao, CRIN = Cocoa Research Institute of Nigeria; n.a. = not applicable (in greenhouse/lab studies improved practice is implemented directly); *two greenhouse/lab
experiments, one plot comparison; **two plot comparisons, one greenhouse/lab experiment; ***one greenhouse/lab experiment, one plot comparison

25
2.4 Discussion
We first discuss our results’ implications and knowledge gaps for each improved practice, then outline
the limitations of our dataset and finally provide an outlook on solving the CSSVD problem.
2.4.1 Best practices and knowledge gaps in the combat of CSSVD
Our results show that all prevention measures included in the meta-analysis significantly reduced
CSSVD infection under researcher managed conditions, except for antiviral substance (Figure 9).
Furthermore, they depict that the focus of past research activities has been on breeding and mild strain
cross protection, while the other improved practices, which could be very much of practical relevance
for farmers (i.e., barrier (strip) cropping, shading/agroforestry and insecticide (vector control)), are
only represented by a single study each. According to Posnette (1981), no other feasible measures
would be as effective in reducing the losses caused by CSSVD as an increase in varietal resistance,
which may decrease the rate of spread by up to 20%. Our results for breeding confirm this, and even
show a higher potential for this improved practice (30%) than claimed by the author. Many viral
diseases in other crops such as potato, tobacco and tomato have effectively been managed by
increasing varietal resistance (Swaminathan, 1993; Moury et al., 2010). However, a recent publication
in this field of research by Padi et al. (2013) shows no significant differences in resistance between
existing and new varieties, underlying the limitations inherent in recycling the minimal resistance
available within a narrow genetic base, and calls for a re-appraisal of variety recommendations.
Our results suggest that the potential of mild strain cross protection is just as high as breeding.
Preventing the adverse effects of severe virus strains by inoculating cocoa beans or seedlings with mild
strains is certainly promising (Ollennu et al., 1999; Dzahini-Obiatey et al., 2010; Muller, 2016).
However, implementing this improved practice on a large scale is very laborious and costly, and entails
several technical and logistical challenges such as inoculating and distributing several millions of cocoa
seedlings to farmers every year (Domfeh, personal communication). Furthermore, the high genetic
diversity of CSSV makes it difficult to find appropriate mild strains for different severe strains, and
simultaneous protection against several different severe strains is practically not feasible (Ollennu et
al., 1999; Dzahini-Obiatey et al., 2010; Muller, 2016).
In contrast, diversification measures (barrier (strip) cropping and shading/agroforestry) showed a
higher potential in our study. However, our results have to be interpreted with care due to the limited
number of studies, which is also reflected in the wide confidence intervals of these improved practices,
and explains why the positive error bar of barrier (strip) cropping goes beyond 100% reduction
infection (Figure 9). Nevertheless, the most recent publication in the field of barrier (strip) cropping
confirms our results (Domfeh et al., 2016). The promising trend we show here, coupled with the fact
that the implementation of these improved practices by smallholders may be relatively easy (i.e., no
need for costly inputs), calls for further research in this area. Furthermore, it would be important to
consider shade as a factor in currently ongoing breeding programs for CSSVD resistance.
There is a knowledge gap about the potential of more diversified production systems such as
agroforestry or strip crop systems to reduce the spread and/or severity of CSSVD. In general, the
effects of shade trees on incidences of pests and diseases in cocoa production systems are complex
and ambiguous (Beer et al., 1998; Staver et al., 2001; Bedimo et al., 2012). Many studies have shown
that self-regulation, the natural control of pests and diseases, is enhanced in diversified cocoa
production systems (Sperber et al., 2004; Isaac et al., 2007; Lin, 2011; Tscharntke et al., 2011; Bieng et
al., 2013; Gidoin et al., 2014; Mbow et al., 2014; Vaast and Somarriba, 2014). However, the effects of

26
different commonly used shade tree species on mealybug populations and CSSVD infection have not
been investigated so far, and they may vary depending on individual shade tree species (Franzen and
Mulder, 2007). Furthermore, it is difficult to identify adequate shade levels and tree species
compositions that minimize mealybug populations and thus likelihood of CSSVD infection while
ensuring favourable growing conditions for cocoa trees. This is because both optimal shade levels for
cocoa trees and mealybug populations vary in the course of the year. Therefore, more research on
effective agroforestry designs to combat CSSVD is needed. In this respect, it might be worthwhile to
facilitate exchange between CSSVD specialists and ongoing agroforestry systems research in the humid
tropics such as the project “agroforestry for food security” (AFS4Food) by CIRAD. AFS4Food assesses
the performance of agroforestry systems to understand the compromises farmers have to strike
between the products and the different services provided by such systems.
Interestingly, the result from the only study we included in the meta-analysis about vector control,
suggests a high potential of a systemic insecticide to control the mealybug vectors and thus reduce
CSSVD. This is particularly interesting, as it is the only farm-scale study (i.e., conducted under actual
farmers’ field conditions) we included in the meta-analysis. Systemic insecticides have also been found
to be effective under researcher-managed conditions. However, a number of problems associated with
their use have been reported, such as their high cost, toxicity to mammals and residues in cocoa beans
(Domfeh, personal communication). Unfortunately, most studies on vector control (e.g. Hanna et al.,
1952; Nicol, 1952) have only reported the effect of a particular insecticide on the abundance of
mealybugs, and not on CSSVD infection. However, reporting CSSVD infection is crucial, as lower
numbers of mealybugs do not necessarily translate into reduced CSSVD infection, especially due to the
very patchy distribution of mealybugs in the field (Campbell, 1990). Therefore, such studies should be
repeated under field conditions and in addition to the effects on mealybug populations, symptomatic
trees numbers should also be reported.
Regarding screening methods for resistance breeding, our results suggest that somatic embryogenesis
is superior to antiviral substances (quercetin, cocoa leaf and embryo extracts). However, care has to
be taken in the interpretation of our results due to the limited number of studies, which also explains
the high confidence intervals of these improved practices and the fact that the negative error bar of
antiviral substance and the positive error bar of somatic embryogenesis are below 0% and beyond
100% reduction infection, respectively (Figure 9). This is because in a meta-analysis, the confidence
intervals of each mean are weighted with the number of replications and the measure of variability
presented in the original studies. Hence, the credible intervals may go below or beyond a certain
meaningful value (like in our case) because of estimates done by the software.
Besides the effect of individual improved practices, the factor age (i.e., the period since adoption of an
improved practice) may correlate with the reduction of infection. We checked this for the 16 studies
where data on “Age IP” (Table 3) was available. The results showed that there is a slight trend towards
lower reduction of infection with increasing time since adoption, but this was statistically not
significant. However, other works suggest that both breeding and mild strain cross protection may be
effective in delaying the spread of CSSDV, but not to completely prevent it (Posnette, 1981; Ameyaw
et al., 2016).

27
2.4.2 Limitations of the Dataset
It is important to note that the effect size presented in this meta-analysis (reduction of infection as
percentage) is a relative measure. As such, it may lead to an under- or overestimation of the potential
of a particular improved practice to reduce CSSVD infection under field conditions. We accounted for
this by aggregating the data pairs to sub-study level as explained in section 2.2.2. Finding an adequate
effect size that would allow for comparison of the heterogeneous set of studies while accounting for
the pooling of comparisons at different scales (greenhouse/lab vs. plot/farm) was challenging, and the
percentage reduction of infection we chose represented the best possible compromise to satisfy both.
In this respect, we acknowledge that our meta-analysis is rather a descriptive analysis to identify
overall patterns in the currently available dataset rather than an inferential statistical method
delivering exact results. Another shortcoming of the dataset is that many of the studies are quite old
(average year of publication was 1983). This is not to say that there are no recent publications on
CSSVD. However, none of the more recent publications has looked at the effect size presented in this
study, which is of practical relevance to farmers.
2.4.3 Solving the CSSVD problem: an outlook
80 years of disciplinary research and 70 years of official, eradication campaign has been a very cost-
intensive endeavour, which has not solved the CSSVD problem in Ghana (Dzahini-Obiatey et al., 2010).
While the currently ongoing breeding efforts for CSSVD resistance and subsequent replacement of
susceptible cocoa with resistant varieties may be part of the solution, there is certainly a long way to
go until widespread implementation on the ground will be achieved, especially because resistance to
the official eradication campaign by farmers remains strong.
However, the question about alternative, feasible options remains unresolved as of yet. As for
research, the ongoing standardization of methodologies and protocols in breeding for CSSVD
resistance is an important step towards resolution of the CSSVD problem (Padi et al., 2013). This is
especially so, as we are becoming more aware of the major genetic variations in the virus across the
West Africa regions (Muller and Sackey, 2005; Oro et al., 2012; Abrokwah et al., 2016). In addition, all
CSSVD-related research should see to ensure that they report their results in a way that enables
quantitative consolidation of several studies through meta-analytical tools.
In practice, on the other hand, an immediate strategy to deal with CSSVD infection is needed. As there
are no short-term solutions to the problem, we need to find ways of living with CSSVD, especially in
parts of the Eastern and Western Regions of Ghana, which were declared areas of mass infection
where complete eradication and replanting is not feasible. Diversifying production systems and
landscapes, which are not infected yet, while subsequently replacing diseased cocoa with more
resistant varieties may be an interesting option for farmers.
Another interesting avenue for future research is biological control of the vectors (Padi, 1997). After
several initiatives in the 1950s had failed due to several reasons, such as the presence of secondary
natural enemies, inefficient rearing and release methods, as well as introduction of ineffective host-
specific parasitoids of other mealybug species, this topic was temporarily abandoned. However, efforts
in biological control of the vectors were recently revived at CRIG, and trials with the exotic predator
Cryptolaemus montrouzieri and the pathogenic fungus Beauveria bassiana were conducted (Plantwise,
2016). Moreover, more attention should be given to the suggestion of farmers to control mealybugs
by altering ant-complexes in the field, shifting the dominance from ant species, which are mealybug-
tendering (Crematogaster spp.) to those who are not (Oecophylla spp.) (Ayenor et al., 2004).

28
2.5 Conclusion
This is the first time the effectiveness of different preventive control measures for CSSVD are
summarized and weighed versus each other in a quantitative way. Since only a fraction of the large
body of literature on CSSVD investigated the efficacy of prevention measures, this meta-analysis is a
stepping-stone for consolidating what we know about the combat of CSSVD, and for setting future
research priorities. Our study clearly shows that there has been a strong focus on resistance breeding
and mild strain cross protection. While these approaches certainly have their potential, they should be
combined with other approaches to form an ensemble of solutions, which is needed to solve a problem
as extensive as CSSVD. Strip crop (barrier (strip) cropping) and agroforestry systems (shading) might
be well suited to complement breeding and mild strain cross protection, as they showed greatest
potential to reduce CSSVD in our study. Moreover, these approaches could be implemented directly
by farmers themselves. However, there is a severe knowledge gap about these approaches’ potential
to reduce the spread and/or severity of CSSVD, which has to be addressed by future research.
Furthermore, shade should be considered in breeding programs for CSSVD resistance and the practical
relevance of different preventive control measures for farmers needs to be studied.

29
30
3. Agroforestry systems can mitigate the severity of cocoa swollen shoot virus disease
Andres, C., Blaser, W.J., Dzahini-Obiatey, H.K., Ameyaw, G.A., Domfeh, O.K., Awiagah, M.A.,
Gattinger, A., Schneider, M., Offei, S.K., Six, J. 5

The only effective treatment for CSSVD is to cut infected trees and replant with disease free planting
material. Hence, the development of preventive control measures and strategies to mitigate the
severity of the disease are of utmost importance. While past research has mainly focused on resistance
breeding, mild strain cross protection and vector control, diversification measures such as agroforestry
have received relatively less attention, despite their potential to mitigate CSSVD symptom severity.
Therefore, we studied the effects of shade on CSSVD symptom severity, capsid damage and cocoa yield
along a gradient of increasing shade tree abundance in smallholder cocoa farms in Ghana.
Furthermore, we measured photosynthetic active radiation and assessed soil fertility in order to
elaborate on potential causal factors for possible shade effects on CSSVD symptom severity. Both
CSSVD symptom severity and cocoa yields followed quadratic curves, and were found to be lowest and
highest in plots with 54% and 39% shade, respectively. The simulated optimal shade levels for CSSVD
symptom severity and cocoa yield overlapped between 45% – 53%, indicating that agroforestry
systems with around 50% shade cover may be an optimal coping strategy to balance CSSVD symptom
severity versus reduced cocoa yield until diseased cocoa is replaced with more resistant varieties.
Furthermore, our results suggest that rather than soil fertility, high-light and possibly also soil moisture
stress may have been responsible for the shade effects on CSSVD symptom severity.

3.1 Introduction
Crop diseases are an increasing problem worldwide, and have been estimated to cause yield losses
ranging between 20% – 40% of global agricultural productivity (Savary et al., 2012). In Ghana, the
second largest producer of cocoa worldwide, the livelihoods of around 800’000 families depend on
revenues from the crop (Danso-Abbeam, 2014). Since more than eight decades, cocoa productivity in
West Africa has been severely limited by CSSVD, especially in Ghana (Dzahini-Obiatey et al., 2010).
The only effective treatment for CSSVD is to cut infected trees and replant with disease free planting
material. Launched in 1946, the official eradication campaign in Ghana has been considered the most
ambitious and costliest of its kind to control a plant viral disease anywhere in the world (Thresh et al.,
1988). Latest estimates put the total number of trees claimed by the disease at over 300 million cocoa
trees (Dzahini-Obiatey, personal communication), which in monetary terms may amount to losses of
several billion US dollars for the government. Furthermore, the high prevalence of CSSVD has curtailed
many cocoa farmers of their regular source of income, as whole fields could be lost to the disease
(Dzahini-Obiatey et al., 2010).
Therefore, the development of efficient preventive control measures for CSSVD that are economically
feasible for farmers is of utmost importance. Despite a lot of research on preventive control measures
(Andres et al., 2017) during the last eighty years (mainly focused on resistance breeding, mild strain
cross-protection and the control of mealybug vectors), however, farmers have not implemented those
measures consistently, and the disease is still widespread throughout Ghana (Ameyaw et al., 2014).

5Published as: Andres, C., Blaser, W.J., Dzahini-Obiatey, H.K., Ameyaw, G.A., Domfeh, O.K., Awiagah, M.A., Gattinger, A.,
Schneider, M., Offei, S.K., Six, J. (2018) Agroforestry systems can mitigate the severity of cocoa swollen shoot virus disease.
Agriculture, Ecosystems and Environment 252: 83-92. doi:https://doi.org/10.1016/j.agee.2017.09.031
31
On the other hand, diversification measures such as agroforestry or barrier (strip) cropping systems
show some potential to reduce CSSVD spread (Schroth et al., 2000a; Domfeh et al., 2016), but have
received less attention in past research (Andres et al., 2017). Shade trees can provide a multitude of
ecosystem services (Beer, 1987; Andres and Bhullar, 2016; Schneider et al., 2016). If the species are
chosen deliberately and managed well, they can have beneficial effects on the development and health
of cocoa trees and thus help them to cope with external stressor such as CSSVD. These include the i)
prevention of high-light stress (Rice and Greenberg, 2000), ii) improvement of water and nutrient
recycling (Buresh et al., 2004), iii) contribution to the control of pests and diseases (Ahenkorah et al.,
1974; Schroth et al., 2000a; Wood and Lass, 2001), iv) maintenance of soil fertility (Götsch, 1994;
Schroth et al., 2000a; Buresh et al., 2004; Bos et al., 2007; Fonte and Six, 2010; Gama-Rodrigues, 2011;
Tscharntke et al., 2011), and v) regulation of micro-climatic conditions for vigorous crop growth
(Schroth et al., 2016a). These effects may contribute to the maintenance of long-term sustainable
cocoa yields (Rice and Greenberg, 2000). Furthermore, agroforestry systems were found to be equally
or more profitable than full-sun systems under ideal management conditions (Armengot et al., 2016).
However, shade trees may also have adverse effects on the development and health of cocoa trees,
as some are alternative hosts for CSSV, harbouring the virus and serving as a source of infection for
cocoa stands planted nearby (Posnette et al., 1950b). Depending on the architecture of their root
systems, they may also be competing for water, which may cause drought stress in cocoa trees,
especially during the dry season (Carr and Lockwood, 2011). Furthermore, some shade tree canopies
may provide too much shade, increasing relative humidity in cocoa stands and thus favouring fungal
diseases (Schroth et al., 2000a; Babin et al., 2010; Oro et al., 2012).
Several studies showed that shade trees can decrease pest populations (Beer et al., 1998; Jaramillo et
al., 2009; Thorlakson and Neufeldt, 2012), and favoured natural pest antagonists (Opoku et al., 2002).
Bigger (1981) showed that mealybug vectors of CSSVD were more abundant under full-sun conditions
than under shade. In addition, the number of mealybug predators and parasitoids was higher on
shaded plots (Bigger, 1981; Mani and Shivaraju, 2016). Also, infection with mistletoes, which indirectly
encourage the spread of CSSVD, was shown to be higher under full-sun conditions in West Africa
(Schroth et al., 2000a). While these results underline the potential of agroforestry systems to reduce
CSSVD spread, the potential of shade trees to reduce the severity of CSSVD symptoms has received
less research attention in the past.
Here we explore the hypothesis that in a CSSVD-infected cocoa stand, symptom severity may be
related to the general health status of the plantation. While it is established that shade reduces the
physiological stress on cocoa trees and their susceptibility to diseases (Beer et al., 1998; Wood and
Lass, 2001; Babin et al., 2010; Oro et al., 2012), only one study carried out under controlled research
conditions has suggested the potential of shade to regulate CSSVD symptom severity (Legg, 1982).
However, the extent to which shade influences CSSVD symptom severity and consequently cocoa yield
has not been investigated under actual farmers’ field conditions so far. Furthermore, while existing
manuals for good agronomic practices in cocoa cultivation recommend between 40% – 50% shade
(Beer et al., 2004; SAN, 2005; Opoku-Ameyaw et al., 2010; Schroth et al., 2016a), the question whether
this degree of shading is also optimal to cope with CSSVD remains unanswered. Moreover, while there
are many factors that may indirectly affect CSSVD symptom severity (e.g. agro-biodiversity, soil
fertility, etc.), research has not addressed the question which of those factors are the most influencing
ones.

32
Thus, we conducted on-farm research in Ghana in two seasons in 2016 (dry season from December to
March, wet season from April to mid-November) to elaborate the effects of shade on CSSVD symptom
severity, capsid damage and cocoa yield. We hypothesised that shade trees reduce both CSSVD
symptom severity and capsid damage of CSSVD-infected cocoa aged 16 years and above, and thus
contribute to lowering the yield-reducing effect of CSSVD as compared to cocoa stands without shade
tree canopy cover. Taking into account that most cocoa farmers in Ghana do not prune their shade
trees, we also hypothesised that soil fertility is not different in full-sun and shaded plots. Consistent
with shade recommendations by existing manuals for good agronomic practices in cocoa cultivation,
we expected a medium shade level of 40% – 50% to be optimal to balance CSSVD symptom severity
versus reduced cocoa yield, and hypothesised that shade per se (and its potential causal factors) is
more important to explain CSSVD symptom severity than agro-biodiversity or soil fertility.

3.2 Materials and methods

3.2.1 Site selection
This study was performed in four operational districts of CHED, subsidiary of COCOBOD, in the Eastern
Region of Ghana. The study region lies within the semi-deciduous rain forest zone with a bi-modal rainy
season, annual rainfall ranging between 1200 mm and 1700 mm. Tall trees with evergreen
undergrowth characterize the vegetation. Humidity and temperature are generally high ranging
between 70% – 80% and 20°C – 32°C, respectively. Soils of the study region are dominated by Acrisols
and Lixisols (Ghana Statistical Service, 2014).
For the selection of suitable study sites, we collaborated closely with CHED. Plots were located in the
Eastern Region of Ghana because of the presence of gradients in both CSSVD prevalence (from areas
of mass infection (Southeast) to areas of only scattered outbreaks (Northwest)) and shade tree canopy
cover. Traditional agroforestry systems with shade trees are more prevalent in this region as compared
to for example the Western Region of Ghana, where large-scale cocoa monoculture are more common.
With the objective to compare shade (agroforestry systems) to full-sun (monoculture) production
systems, we selected 23 pairs of one shade and one full-sun plot each (46 plots) along a Southeast –
Northwest transect (between 6°3’0.23’’N, 0°29’28.38’’W and 7°26’28.10’’N, 1°9’0.03’’W, Figure 10).
Thereby, we covered the two gradients mentioned above. Our target was to be close to 0% and above
40% of shade for full-sun and shade plots, respectively. All plots were located in cocoa areas infected
with CSSVD that were at least 10 hectares in size and planted with Amazonia type cocoa aged between
25 and 28 years (average 26.5 years).
Each plot measured 50m × 50m (0.25 ha). To avoid border effects, we obtained all data in inner sub-
plots measuring 30m × 30m. Paired plots were similar in terms of exposure, slope, etc. and located as
close as possible (max. 100 m apart) in order to avoid confounding effects such as differing soil types,
etc. At the same time, each pair of shade/full-sun was located in a maximum distance (up to 45 km) to
the next pair in order to cover potential differences in local conditions such as microclimate, landscape
structure and soils.

33
Figure 10: Plots selected in the Eastern Region

3.2.2 Data collection

3.2.2.1 Plot characterization
To characterize our research plots, we obtained various soil and plant parameters at plot and tree
level. We counted number of cocoa, plantain and shade trees in each plot. With the number of cocoa
trees, we calculated a plot-specific cocoa density (“Spacing”), which we used as a random factor for
statistical analyses in order to correct for varying cocoa tree densities in different plots (see section
3.2.3). We identified shade tree species with the help of CRIG staff, farmers and a photo-guide for the
forest trees of Ghana (Hawthorne and Gyakari, 2006).

34
In order to characterize the agro-biodiversity of the research plots, we calculated the Shannon index
(Hs) of diversity (Shannon and Weaver, 1949; Kindt and Coe, 2005) using the equation 𝐻𝑆 =
− ∑Si=1 𝑝𝑖 ln 𝑝𝑖 , where: S = number of species in a production system, and pi = share from one category
in the total number of categories. The plot characterization revealed that we chose suitable sites to
conduct our research, as the number of shade trees, number of shade tree species, and Shannon index
of diversity differed significantly between shade and full-sun plots (Table 4). We estimated the general
management condition (maintenance, pruning, etc.) and mistletoe infestation of cocoa trees with the
help of CHED staff using simple categorical classifications from 1 to 4 (from Very Poor/Low – Good/Very
High, respectively) (Awudzi, 2014). In addition, we recorded management intensities by interviewing
farmers about their agronomic procedures, which revealed that all plots were under low-intensity
management, and none of the selected plots differed significantly in management intensity.
To assess shade, we used two different methods: visual estimation of the shade tree canopy cover
following the method described by Somarriba (2002), as well as recordings of photosynthetic active
radiation (PAR) measured between shade tree and cocoa canopies using QSO-S PAR Photon Flux
sensors and data loggers (Decagon Devices, USA). We calculated the relative PAR by relating recordings
done within the plots to simultaneous ones (one record per minute) done in an open area close-by
using a permanent logger (representing 100% of PAR). No shade was cast on cocoa trees by plantain
trees, as they were all equally or less tall than the cocoa trees. We achieved our aim to be close to 0%
and above 40% of shade for full-sun and shade plots, respectively (Table 4). Visual estimation of shade
tree canopy cover correlated well with direct PAR measurements (r = 0.77, p < 0.001), which shows
that the rather simple, cheap and quick method of Somarriba (2002) delivers adequate results.
We sampled the soil in all four corners and the centre of each plot to a depth of 0 – 25 cm using a soil
auger. We took five sub-samples per plot at a distance of 10.6 m along two transects running diagonally
through the inner sub-plots, and bulked them prior to analysis. We measured standard soil fertility
parameters (pH, organic C (Corg), total N (Ntot), available P (Pav), exchangeable Ca (Caex), Mg (Mgex) and
K (Kex)) according to standard methods of the International Soil Reference and Information Centre (van
Reeuwijk, 2002). All analyses were done at the Ecological Laboratory of the University of Ghana in
Accra, Ghana.
We verified CSSVD presence in the plots by visual symptom detection with the help of local CSSVD
specialists, and grafted bud wood samples on susceptible (Amelonado) scions at CRIG to double-check
CSSVD presence by verifying symptom expression on the obtained saplings. Furthermore, we
confirmed CSSVD presence using the standard PCR procedure described by Muller et al. (2001). Even
though all selected plots corresponded to locations, which according to CHED maps were infected with
CSSVD, we could only confirm CSSVD presence in 26 of the 46 plots, so we only used these 26 plots for
further data collection (see section 3.2.2.2), and only present the data of these plots here. All 26 plots
were infected with the severe CSSVD strain 1A.

3.2.2.2 CSSVD symptom severity, cocoa yield and capsid damage

Within each of the 26 CSSVD-infected plots, we selected 20 cocoa trees along two transects running
diagonally through the plots (520 trees in total). We assessed the CSSVD symptom severity of those
520 trees based on a categorical scale from 1 – 10, which we slightly adapted from the common
practice described by Padi et al. (2013). The numbers on the scale represent the following: 1 = no
symptom, 2 = red vein banding, 3 = chlorotic vein flecking, 4 = chlorotic vein clearing, 5 = green vein

35
banding, 6 = diffused flecking, 7 = fern pattern, 8 = swollen stem, 9 = dying plant, 10 = dead plant. In
order to capture the variability of CSSVD symptom expression occurring in the course of the year, we
assessed CSSVD symptom severity twice, once during the dry season (March/April) and once during
the rainy season (September) in 2016.
We assessed cocoa yield on the same 520 trees by counting the number of pods shortly before the
major harvest season in September 2016. Furthermore, we assessed capsid damage on flush leaves
during the dry season in March and April 2016 with the help of local capsid specialists, by rating each
cocoa tree on a categorical scale from 1 – 5 where the numbers represent the percentage of damage
on flush leaves: 1 = 1 – 20%, 2 = 21 – 40%, 3 = 41 – 60%, 4 = 61 – 80%, 5 = 81 – 100% (Awudzi, 2014).

3.2.3 Data analysis

In the dataset of CSSVD symptom severity, we classified all trees with severity scores 9 and 10 (“dying
plant” and “dead plant”) as outliers, and removed them from the dataset, because these symptoms
could also be caused by other phenomena such as mistletoe or capsid damage, and may therefore not
solely be attributed to CSSVD. This led to the removal of 155 (30%) and 46 (9%) data points from the
data sets of full-sun and shade, respectively (total number of cocoa tree observations = 839). However,
because tree death is arguably at least partly caused by CSSVD, the removal of these data points may
have led to an underestimation of the average damage, particularly under full-sun conditions.
Consequently, we rather underestimated overall differences between shade and full-sun plots,
indicating that the data we present here are a conservative measure of the actual effects on the field.
In the dataset of cocoa yield, we excluded the data from four entire plots (number of plots analysed
(n) = 22, total number of cocoa tree observations = 440), because the farmers had applied a heavy
rejuvenation pruning to their cocoa trees, which severely influenced the yield data.
We analysed our data in two different ways. First, we compared all outcome variables shown in Table
4 and Table 5 between shade and full-sun plots (i.e., treating the explanatory variable Shade as a
categorical variable with two levels), using a normal linear mixed effect model with Shade as fixed
predictor and District (n=4), Pair (n=13), and Spacing (n=26) as random factors. The random factors in
our models accounted for variabilities that may exist due to regional (District) or local agro-ecological
conditions (Pair), or plot-specific growing conditions (Spacing). Secondly, we assessed the relationship
of CSSV symptom severity and cocoa yield (the two outcome variables of main interest for our study)
with shade tree canopy cover by regression analyses (i.e., treating the explanatory variable Shade as a
continuous explanatory variable). To do this, we used a similar model, but included a quadratic term
of Shade (Shade2) as additional fixed predictor, as well as Plot (n=26) as additional random factor. The
random factor Plot accounted for variability that may exist due to pseudo-replication of individual trees
within each plot. In the case of cocoa yield, we additionally included Capsid Damage (n=438) as a
random factor in the model, with the aim to discern the effect of CSSVD symptom severity by
separating it from the effect of capsid damage on yield. We encountered no violations of model
assumptions by graphical residual analysis (normal Q-Q and Tukey-Anscombe plots). We used the lmer
function of the lme4 package in R for all linear mixed effect model analyses (Bates et al., 2015), and
Pearson’s correlation coefficients (r) to measure correlations between individual parameters.

36
We calculated the marginal R2 values shown in Figure 11 as a measure of the proportion of explained
variance by subtracting the ratio between the explained variance (total variance/residual variance)
from the total variance of the response variable (CSSVD symptom severity or cocoa yield) as explained
by Nakagawa and Schielzeth (2013). To identify optimal shade levels (vertical solid lines in Figure 11),
we used the minimum/maximum of the quadratic regression lines as calculated by setting its derivative
to zero. As uncertainty measurements for these optima we reported the 95% confidence intervals,
which we obtained via Monte Carlo simulation based on 5’000 simulations from the joint posterior
distributions of the model parameters from which we calculated 5’000 values that constituted random
samples from the posterior distribution of the optimal shade values. We used the 2.5% and the 97.5%
quantiles of these values as lower and upper limits of the 95% confidence intervals of the optima.
To elaborate on potential causal factors for shade effects on CSSVD symptom severity, we performed
regression analyses between CSSVD symptom severity and all possible and meaningful fixed predictors
(number of shade trees, number of shade tree species, Shannon index of diversity, pH, Corg, Ntot, P, Ca,
Mg and K), using District (n=4), Pair (n=13), Plot (n=26) and Spacing (n=26) as random factors. We
conducted all analyses with the statistical software R, version 3.3.1 (R Core Team, 2017).

3.3 Results
Shade and full-sun plots showed no significant differences in any of the standard soil fertility
parameters, except for slightly higher Ca values in full-sun plots (Table 5). CSSVD symptom severity
was significantly lower under shade (by 56% on average, Figure 11A). The quadratic model showed a
decrease of CSSVD symptom severity from full-sun conditions until an optimal shade level of 54%
shade, after which the severity score increased again. The 95% confidence interval of the optimal
shade level for CSSVD symptom severity lay between 45% – 76% (dashed/dotted lines in Figure 11A).
There were significant positive correlations between CSSVD symptom severity and mistletoe
infestation (r = 0.32, p < 0.01) as well as capsid damage (r = 0.32, p < 0.01).
Similarly, cocoa yield per tree was significantly higher under shade (by 42% on average, Figure 11B)
until an optimal shade level of 39% shade, after which the yield decreased again. The 95% confidence
interval of the optimal shade level for cocoa yield lay between 34% – 53% (dotted/dashed lines in
Figure 11B). The optimal shade levels for CSSVD symptom severity and cocoa yield overlapped
between 45% – 53% (dashed lines in Figure 11).
The regression analyses of potential causal factors for shade effects on CSSVD symptom severity
revealed that there were significant negative correlations between CSSVD symptom severity and
number of shade trees as well as number of shade tree species, while there was a significant positive
correlation with Caex (Figure 12). In addition, the correlation between number of shade trees and
number of shade tree species was much stronger (r = 0.91) than the correlation between number of
shade tree species and Shannon index of diversity (r = 0.47). There were no significant correlations
between CSSVD symptom severity and any of the other factors: Shannon index of diversity, pH, Corg,
Ntot, Pav, Mgex or Kex (dashed lines in Figure 12).

37
Table 4: Characterization of plots compared for CSSVD symptom severity and cocoa yield in the Eastern Region of Ghana in two seasons in 2016.
Plot Condition Mistletoe Capsid Number Number Number Number of Shade tree species Shannon Visual shade Photosynthetic
infestation damage of cocoa of of shade shade tree index of tree canopy active radiation
trees plantain trees species diversity cover [% full-sun]
[per ha] trees [per [per ha] [per plot] estimation [%] (µmol m-2 s-1)
ha]
9 Average Medium 1.05 918 67 33 2 Albizia zygia, Terminalia ivorensis 0.36 0.28 0.84 (962)
10 Poor Medium 1.45 698 333 0 0 n.a. 0.60 0.00 1.00 (1263)
11 Poor Medium 1.40 368 133 11 1 Terminalia ivorensis 0.75 0.08 0.96 (1196)
12 Average Low 1.26 293 100 33 3 Alstonia boonei, Spathodea campanulata, 0.94 0.28 0.77 (906)
Sterculiua tragacantha
13 Average High 1.80 1’167 0 89 4 Allanblackia parviflora, Holarrhena 0.35 0.53 0.56 (1120)
floribunda, Terminalia ivorensis,
Terminalia superba
14 Poor Very high 1.65 989 389 33 3 Elaeis guineensis, Milicia excelsa, Persea 0.72 0.00 0.97 (1906)
americana
15 Poor High 1.53 833 11 67 4 Azadirachta indica, Holarrhena 0.42 0.11 0.34 (669)
floribunda, Spathodea campanulata,
Terminalia ivorensis
16 Poor Medium 1.32 1’378 0 22 2 Allanblackia parviflora, Rauvolfia 0.09 0.00 0.89 (1747)
vomitoria
17 Very poor Medium 1.58 600 0 33 3 Ceiba pentandra, Elaeis guineensis, 0.26 0.03 0.93 (1255)
Spathodea campanulata
18 Good Low 1.35 878 11 178 8 Alstonia boonei, Elaeis guineensis, 0.80 0.60 0.39 (440)
Holarrhena floribunda, Persea americana,
Petersianthis macrocarpus, Rauvolfia
vomitoria, Terminalia ivorensis,
Tetrapleura tetraptera
19 Average Very high 2.45 800 0 33 2 Persea americana, Terminalia ivorensis 0.19 0.04 0.89 (983)
20 Poor Medium 0.58 544 156 111 5 Khaya grandifolia, Rauvolfia vomitoria, 1.06 0.46 0.69 (420)
Samanea dinklagei, Spathodea
campanulata, Terminalia ivorensis
21 Very poor Medium 1.30 756 167 22 2 Citrus sinensis, Persea americana 0.59 0.09 1.00 (844)
22 Average Medium 1.53 1’033 56 100 6 Alstonia boonei, Amphimas 0.62 0.62 0.59 (426)
pterocarpoides, Cola nitida, Khaya
grandifolia, Persea americana, Spathodea
campanulata
23 Very poor Very high 3.25 378 67 11 1 Voacanga africana 0.53 0.00 0.94 (878)
24 Average High 2.35 533 78 144 9 Allanblackia parviflora, Alstonia boonei, 1.18 0.29 0.47 (670)
Amphimas pterocarpoides, Ceiba
pentandra, Citrus sinensis, Cola nitida,

38
 Elaeis guineensis, Rauvolfia vomitoria,
Spathodea campanulata
25 Good Medium 1.65 600 100 122 7 Allanblackia parviflora, Khaya grandifolia, 1.05 0.49 0.61 (1129)
Petersianthis macrocarpus, Rauvolfia
vomitoria, Samanea dinklagei, Spathodea
campanulata, Terminalia ivorensis
26 Very poor High 2.16 478 156 22 2 Holarrhena floribunda, Ficus capensis 0.71 0.01 0.94 (1845)
27 Good Low 0.40 1’044 300 122 4 Holarrhena floribunda, Khaya grandifolia, 0.87 0.22 0.71 (1070)
Rauvolfia vomitoria, Terminalia ivorensis
28 Good Low 1.05 633 67 278 13 Acacia spp., Alstonia boonei, Amphimas 1.48 0.47 0.38 (659)
pterocarpoides, Ceiba pentandra, Elaeis
guineensis, Holarrhena floribunda, Khaya
grandifolia, Petersianthis macrocarpus,
Rauvolfia vomitoria, Samanea dinklagei,
Spathodea campanulata, Terminalia
ivorensis, Voacanga africana
29 Good Low 0.90 799 111 144 7 Alstonia boonei, Ceiba pentandra, Citrus 0.86 0.41 0.44 (694)
sinensis, Elaeis guineensis, Spathodea
campanulata, Terminalia ivorensis,
Terminalia superba
30 Very poor High 2.37 395 100 78 6 Ceiba pentandra, Elaeis guineensis, Khaya 0.95 0.06 0.75 (1076)
grandifolia, Spathodea campanulata,
Terminalia ivorensis, Terminalia superba
31 Good Low 1.70 722 100 133 6 Allanblackia parviflora, Amphimas 0.96 0.76 0.18 (118)
pterocarpoides, Ceiba pentandra,
Holarrhena floribunda, Rauvolfia
vomitoria, Terminalia ivorensis
32 Poor High 3.26 456 111 33 2 Alstonia boonei, Terminalia ivorensis 0.72 0.07 0.30 (134)
33 Good Low 0.95 1’089 122 100 5 Holarrhena floribunda, Persea americana, 0.67 0.48 0.51 (274)
Rauvolfia vomitoria, Terminalia ivorensis,
Terminalia superba
34 Very poor Medium 2.16 811 322 44 4 Amphimas pterocarpoides, Holarrhena 0.79 0.05 0.88 (580)
floribunda, Musanga cecropioides,
Rauvolfia vomitoria
Mean 3.31 a 1.77 b 1.36 b 772 a 75 b* 118 a 6.0 a n.a. 0.83 a 0.44 a 0.52 b
S
sem 0.21 0.23 0.13 69 13 18 0.8 n.a. 0.09 0.05 0.05
Mean 1.85 b 2.62 a 1.90 a 704 a 160 a 36 b 2.4 b n.a. 0.60 b 0.05 b 0.86 a
FS
sem 0.32 0.27 0.23 84 38 9 0.4 n.a. 0.07 0.02 0.05
CHED: Cocoa Health and Extension Division; OSN: Osino, TAF: Tafo, OYO: Oyoko; Age class C: 16 – 30 years; sem: standard error of the mean; Condition/Mistletoe infestation: Very
poor/Low = 1, Poor/Medium=2, Average/High = 3, Good/Very high = 4; n.a.: not applicable; different superscript letters after mean values indicate significant differences.
39
Table 5: Soil fertility parameters of plots compared for CSSVD symptom severity and cocoa yield in the Eastern Region of Ghana in two seasons in 2016.
Plot Pair CHED district Shade pH [H2O, 1:1] Corg [g kg-1] Ntot [g kg-1] Pav [mg kg-1] Caex [cmol+ kg-1] Mgex [cmol+ kg-1] Kex [cmol+ kg-1]
(S)/Full-Sun
(FS)
9 5 OSN S 6.8 13.6 4.50 14.8 2.44 0.25 0.32
10 5 OSN FS 7.2 17.6 3.00 18.2 2.53 0.22 0.29
11 6 OSN FS 6.2 15.9 4.10 15.0 3.14 0.22 0.22
12 6 OSN S 6.5 16.7 3.10 14.4 2.02 0.26 0.15
13 7 TAF S 6.6 6.40 4.10 16.4 1.18 0.22 0.05
14 7 TAF FS 6.7 18.3 4.60 26.8 2.56 0.19 0.13
15 8 TAF S 6.2 5.60 4.10 16.6 1.36 0.17 0.11
16 8 TAF FS 6.4 11.9 3.90 18.8 2.14 0.21 0.07
17 9 TAF FS 6.9 19.9 3.70 18.0 2.61 0.25 0.15
18 9 TAF S 6.4 22.3 3.90 18.8 2.58 0.23 0.14
19 10 TAF FS 6.8 23.1 4.70 17.8 2.60 0.24 0.16
20 10 TAF S 6.7 17.6 3.90 16.6 2.44 0.78 0.06
21 11 TAF FS 6.7 19.9 4.10 14.3 2.61 0.26 0.21
22 11 TAF S 6.8 24.7 3.90 18.4 2.59 0.22 0.24
23 12 OYO FS 6.5 10.4 5.00 12.6 2.57 0.22 0.35
24 12 OYO S 6.7 15.9 4.30 34.6 2.57 0.25 0.16
25 13 OYO S 6.1 14.4 3.60 23.2 2.49 0.20 0.16
26 13 OYO FS 6.6 12.8 3.50 30.8 2.64 0.21 0.20
27 14 OYO FS 6.1 18.3 4.70 22.4 2.50 0.19 0.17
28 14 OYO S 6.2 20.7 4.60 19.0 2.62 0.24 0.16
29 15 OYO S 7.3 13.6 4.90 29.4 2.45 0.18 0.21
30 15 OYO FS 6.0 11.9 4.10 17.8 2.46 0.20 0.19
31 16 OYO S 6.3 13.6 5.40 24.0 2.42 0.21 0.19
32 16 OYO FS 6.0 11.2 4.50 19.4 2.59 0.19 0.33
33 17 OYO S 6.1 12.8 3.70 16.4 1.88 0.19 0.11
34 17 OYO FS 5.1 7.10 3.90 21.2 2.38 0.18 0.16
Mean S n.a. n.a. S 6.5 a 15.2 a 4.20 a 20.2 a 2.23 b 0.26 a 0.16 a
sem n.a. n.a. n.a. 0.1 1.5 0.20 1.7 0.13 0.04 0.02
Mean FS n.a. n.a. FS 6.4 a 15.3 a 4.10 a 19.5 a 2.56 a 0.21 a 0.20 a
sem n.a. n.a. n.a. 0.1 1.3 0.20 1.4 0.06 0.01 0.02
CHED: Cocoa Health and Extension Division; OSN: Osino, TAF: Tafo, OYO: Oyoko; Age class C: 16 – 30 years; sem: standard error of the mean; different superscript letters after
mean values indicate significant differences

40
Figure 11: Relationship between shade tree canopy cover and A: CSSVD symptom severity (n=839 trees
on 26 plots (13 pairs of full-sun/shade plots)), B: cocoa yield in the main harvest season 2016 (n=438
trees on 22 plots (11 pairs of full-sun/shade plots)). Each point represents the mean value of all
observations from 20 individual trees per plot. Vertical lines indicate optimal shade levels (solid) and
respective 95% confidence intervals (dotted). Red dotted lines indicate the suggested range of optimal
shade levels to balance CSSVD symptom severity versus reduced cocoa yield, based on our dataset.

41
Figure 12: Regression analyses of potential causal factors for shade effects on CSSVD symptom severity. Individual panels depict the regression analyses
between CSSVD symptom severity and a) number of shade trees, b) number of shade tree species, c) pH, d) organic C, e) total N, f) available P, g)/h)/i)
exchangeable Ca/Mg/K, respectively. Open (○) and closed (●) symbols represent full-sun and shade plots, respectively. Trend lines and respective Pearson’s
correlation coefficients (r) are displayed for fixed predictors that were statistically significant. Horizontal dotted lines indicate no statistical significance.
42
3.4 Discussion
Since complete eradication and replanting of all CSSVD-infected cocoa trees is not feasible, a coping
strategy is urgently needed (Andres et al., 2017). Our data suggests that agroforestry systems with
around 50% shade may be optimal to balance CSSVD symptom severity versus reduced cocoa yield
(dashed lines in Figure 11), which may help mitigating CSSVD symptom severity until current cocoa is
replaced with more resistant varieties. These results are in line with shade levels recommended by
existing manuals for good agronomic practices in cocoa cultivation (Beer et al., 2004; SAN, 2005;
Opoku-Ameyaw et al., 2010; Schroth et al., 2016a). Our results from the regression analyses of
potential causal factors for shade effects on CSSVD symptom severity indicated that shade per se was
a relatively more important potential causal factor for shade effects on CSSVD symptom severity,
compared to agro-biodiversity (measured by both Shannon index as well as shade tree diversity) or
soil fertility. In the following, we discuss the possible reasons for higher CSSVD symptom severity and
lower cocoa yields in full-sun plots (i.e., the meaning of the term “shade per se“).
3.4.1 Possible reasons for higher CSSVD symptom severity and lower cocoa yields in
full-sun plots
3.4.1.1 High-light and moisture stress
Cocoa is a heliophobe preferring low levels of incident radiation (Vinod, 2012). Studies have shown
that exposure to light intensities higher than 1800 μmol m-2 s-1 induced high-light stress responses
(Lichtenthaler and Burkart, 1999) such as reduction in chlorophyll Ievels and photosynthetic rate
(Galyuon et al., 1996), or even damages to the leaf photosynthetic mechanisms of cocoa (Raja Harun
and Hardwick, 1988). With a maximum measurement of 1983 μmol m-2 s-1 (data not shown) and mean
values for individual plots as high as 1906 μmol m-2 s-1 (Table 4), our data from full-sun plots were
above this critical value, which indicates that cocoa trees in full-sun plots indeed may have suffered
from high-light stress. This may have led to low availability of photo-assimilates which suppresses
flower production, reduces yield (Asomaning et al., 1971) and tree vigour, and leads to higher
susceptibility towards attack by pests and diseases (Owusu, 1980; Schroth et al., 2000a; Wood and
Lass, 2001). These effects may be supported by the significant positive correlations between CSSVD
symptom severity and mistletoe infestation as well as capsid damage we found in our study (see
section 3.3). These effects may partly explain the higher CSSVD symptom severity and lower cocoa
yield we observed below the optimal shade levels of 54% and 39%, respectively. Our results are in line
with the ones by Legg (1982), who observed virus symptoms to be particularly serious in cocoa trees
under stress, and reported tolerance of virus infection to be at a maximum under shaded conditions.
However, excessive shading leads to low availability of photo-assimilates, which may have similar
effects as the ones described above (lower flower production, yield, tree vigour, and more pests and
diseases (Asomaning et al., 1971; Owusu, 1980; Schroth et al., 2000a; Wood and Lass, 2001)). Hence,
these effects may partly explain the observed increase in CSSVD symptom severity and decrease in
cocoa yield above the optimal shade levels of 54% and 39%, respectively. The upper boundary of the
95% confidence interval of the optimal shade level for CSSVD symptom severity (76% in Figure 11A) is
uncertain, because the Monte Carlo simulations in this area of the curve were based on only one data
point. Thus, more data on CSSVD symptom severity under heavy shading regimes (>50%) are needed.
Our results are in line with the quadratic model proposed by Vernon (1967) to explain the relationship
between cocoa yield and available light, and suggest that some degree of shading may be desirable to
maintain the productivity of CSSVD-infected cocoa aged 16 years and above. However, more data is
needed to confirm Vernon’s (1967)) quadratic equation under conditions of CSSVD infection.
43
Besides direct high-light stress, the removal of shade trees may indirectly cause moisture stress.
Excessive leaf transpiration and increased soil evaporation were reported to be major factors
contributing to the yield decline some ten years after removal of shade trees in several trials in Ghana
and elsewhere (Cunningham and Burridge, 1960; Asomaning et al., 1971; Ahenkorah et al., 1974; Babin
et al., 2010; Oro et al., 2012). Thus, the integration of suitable shade trees species that are adequately
pruned and managed by farmers is vital to maintain cocoa tree health by potentially reducing both
high-light and moisture stress.

3.4.1.2 Soil fertility

Several studies have shown that the loss of nutrients, and especially exchangeable bases (Owusu,
1980), in full-sun systems leads to yield decline (Cunningham and Burridge, 1960; Asomaning et al.,
1971; Ahenkorah et al., 1974; Babin et al., 2010; Oro et al., 2012). Our findings seem to be in line with
those results, as the soil in our plots was severely deficient in both Ca and Mg when compared to the
minimum values proposed by Wessel (1971), which also means that the significant positive correlation
of CSSVD symptom severity and Ca (r = 0.14, p < 0.05) is of no ecological significance. However, since
soil nutrient status was about the same in shade and full-sun plots, it is more likely that high-light and
moisture stress were responsible for the observed effects on CSSVD symptom severity and cocoa yield.
Another main result of our study is the lack of increased soil fertility under shade trees compared to
full-sun systems, which is in line with the results of others who showed no differences in soil carbon
between shade and full-sun cocoa systems (Gockowski and Sonwa, 2011; Jacobi et al., 2014;
Mohammed et al., 2016). Recently, Blaser et al. (2017) reported limited benefits of shade trees for soil
fertility in cocoa agroforests in Ghana. A common feature of our study and the one by Blaser et al.
(2017) was that farmers did not prune shade tree canopies on the research plots of both studies.
However, the results of both studies contrast sharply with the experiences of agroforestry
practitioners, who deliberately prune and thus increase the amount of biomass applied to the soil
considerably, which improved soil fertility significantly (Götsch, 1994; Fonte and Six, 2010). These
contrasting results may indicate that the limited amount of biomass recycled from deeper soil layers
to the top soil through mere litter fall (Buresh et al., 2004; Gama-Rodrigues, 2011; Tscharntke et al.,
2011) may have limited effects for soil fertility in cocoa agroforests. Thus, shade tree pruning may be
a necessity to stimulate tree growth and increase organic matter input for soil fertility enhancements
in cocoa agroforests.
If soil fertility and soil moisture are favourable for vigorous crop growth, cocoa yields positively
correlate with light intensity (Cunningham and Burridge, 1960; Ahenkorah et al., 1987; Wood and Lass,
2001; Babin et al., 2010; Oro et al., 2012). In poor soils with little or no fertilization, however, higher
yields are obtained under shade (de Almeida and Valle, 2007; Vinod, 2012). Thus, with shade removal
comes the necessity of continuous fertilizer applications to maintain yields. Therefore, keeping, and
especially managing shade trees may be an important strategy to maintain cocoa productivity under
current conditions of smallholder-grown cocoa in Ghana with rare or no application of external
nutrients.

44
3.4.2 Avenues for future research
In order to better understand the causal factors for the higher CSSVD symptom severity and lower
cocoa yield we observed under full-sun conditions, future research should i) measure moisture stress,
and ii) quantify viral load in plant tissue by qPCR analyses, and relate it to CSSVD symptom severity,
taking into account the size of infectious mealybug populations in diseased fields. In addition, shade
should be incorporated into ongoing breeding programs for CSSVD resistance (Andres et al., 2017).
Many studies have shown that natural control of pests and diseases is enhanced in cocoa agroforests
(Sperber et al., 2004; Tscharntke et al., 2011; Bieng et al., 2013; Gidoin et al., 2014; Vaast and
Somarriba, 2014), and it is established that cocoa needs shade to remain healthy and productive
(Owusu, 1980; Wood and Lass, 2001; Carr and Lockwood, 2011). However, there is a knowledge gap
about the effects of different shade tree species on CSSVD, and about suitable shade tree species for
effective agroforestry designs to combat CSSVD under different pedo-climatic and socio-economic
conditions (Owusu, 1980; Andres et al., 2016), which should be addressed by future research.
Despite agroforestry systems being a potential coping strategy to mitigate CSSVD symptom severity,
they may not be able to prevent CSSVD spread. This is because the canopies of tall shade trees are
likely not to limit the movement of mealybug vectors that spread the disease at the forest floor (Jeger
and Thresh, 1993). If the continuity of cocoa plantations was broken with occasional strips (barriers)
of non-host crops such as citrus or oil palm, mealybug movement would likely be restricted and thus
CSSVD spread reduced (Domfeh et al., 2016). Hence, future research should investigate the effect of
cocoa landscape connectivity on CSSVD spread.

3.5 Conclusion
In this paper, we address the need for more on-farm research about the potential of agroforestry
systems to mitigate the severity of CSSVD reported by Andres et al. (2017). Here, we show that shade
reduced CSSVD symptom severity and capsid damage in cocoa stands aged 16 years and above, which
may have helped lowering the yield-reducing effect of CSSVD. High-light and possibly also soil moisture
stress were more likely to be explanatory factors for those shade effects rather than soil fertility. Our
results indicate that agroforestry systems with around 50% shade seem to be optimal to balance CSSVD
symptom severity versus reduced cocoa yield, and may thus offer an optimal coping strategy until
diseased cocoa is replaced with more resistant varieties.

45
46
4. Social barriers to the implementation of measures to prevent the spread of cocoa
swollen shoot virus disease in Ghana
Andres, C., Hoerler, R., Home, R., Jörin, J., Dzahini-Obiatey, H.K., Ameyaw, G.A., Domfeh, O.K., Blaser,
W.J., Gattinger, A., Offei, S.K., Six, J.6

CSSVD is a major factor limiting the productivity of cocoa for farmers in West Africa. The only effective
treatment for CSSVD is to cut infected trees and replant with disease free planting material. Research
has recommended the following prevention measures: i) cordon sanitaire, ii) barrier (strip) cropping,
iii) using partly tolerant hybrid seedlings, and iv) removing alternative host trees. Here, we evaluate
the current adoption of these prevention measures and identify their adoption constraints for farmers.
We conducted a quantitative survey with 396 farmers in the Eastern and Western Regions of Ghana,
held six focus group discussions with eight to eleven participants each, and hosted a multi-stakeholder
validation workshop with 31 key actors from the cocoa value chain of Ghana. Our results show that
the adoption of prevention measures remains limited. Farmers with a more extensive social network,
a larger farm size, more secure land tenure rights and more knowledge about the prevention measures
were more likely to adopt them, especially barrier (strip) cropping, hybrid seedlings and removing
alternative host trees. Lack of knowledge about the measures was the single biggest barrier for their
adoption, with 51% of all the participating farmers not even being aware of them. Our results suggest
that a social network has the highest potential to be an information source for farmers, which was in
line with the key finding that the information flow between farmers and other stakeholders in Ghana
is a critical factor affecting knowledge spread and consequently the adoption of CSSVD prevention
measures. Our results may contribute to the elaboration of an implementation action plan to boost
the dissemination of feasible CSSVD prevention measures in Ghana in order to cover farmers’ needs
for information (e.g. technical advice) and inputs (e.g. access to hybrid seedlings) efficiently.
4.1 Introduction
CSSVD is a major factor limiting the productivity of cocoa for farmers in West Africa. CSSVD is caused
by CSSV, which belongs to the genus Badnavirus (Lot et al., 1991). CSSV causes symptoms such as
transient red veins and mottling in young leaves, chlorosis in mature leaves and pods, root atrophy
and stunting, as well as root and stem swellings. Highly pathogenic strains cause severe leaf chlorosis,
which may result in the rapid death of cocoa trees (Dzahini-Obiatey et al., 2010). Different species of
mealybugs spread CSSVD by serving as vectors for CSSV (Pseudococcidae spp.), with the most
important being Formicococcus njalensis (Laing) and Planococcus citri (Bigger, 1981).
The disease is particularly relevant to Ghana, the second largest cocoa producer country worldwide,
where about 17% of all cocoa areas are affected by CSSVD (Dzahini-Obiatey, personal communication).
Around 800’000 farmer families produce Ghana’s cocoa (Danso-Abbeam, 2014), which means that
approximately 5 million people (20% of the population) are directly dependent on cocoa. Since the
only effective treatment for CSSVD is to cut infected trees and replant with disease free planting
material, CSSVD threatens the livelihoods of these people. The Ghanaian government has
implemented a national eradication program with the objective to control the disease (Thresh and
Owusu, 1986). Since its initiation in 1946, the program has cut out more than 300 million trees, with
immense economic costs for both the government and cocoa farmers (Dzahini-Obiatey et al., 2010).
Thus, CSSVD is a major issue for all stakeholders in the cocoa value chain of Ghana (Figure 13).

6 Submitted to Agronomy for Sustainable Development, Dec. 22nd, 2017

47
Figure 13: Left panel: vein clearing, a typical symptom of cocoa swollen shoot virus disease (CSSVD,
severe strain 1A) in mature cocoa leaves. Right panel: a group with representatives from all the major
stakeholders in the cocoa value chain of Ghana discussing possible solutions to tackle CSSVD. Pictures:
author.

Research into CSSVD has progressed and several prevention measures have been recommended,
including:
1) Cordon sanitaire (Vanderplank, 1947), which involves leaving a 10 m-wide cocoa-free zone
around newly planted cocoa, so that mealybug vectors become non-infective when moving
into new plantations. Although this measure was shown to be effective, it also results in loss
of income due to less cocoa being planted. Furthermore, when land is left fallow, exposure to
direct sunlight may dry out the soil and subject adjacent cocoa trees to increased water stress,
and there is a risk that neighboring farmers may claim the fallow portion as part of their farm.
2) Barrier (strip) cropping (Domfeh et al., 2016), which involves lining newly planted cocoa
orchards with a 10 m barrier of non-host crops, such as citrus, oil palm, coffee, or rubber, so
that mealybug vectors become non-infective when moving into new plantations. Barrier (strip)
cropping has the advantages of potentially reducing CSSVD infection by up to 85% (Andres et
al., 2017), while providing additional income through diversified production cropping (Domfeh
et al., 2016). However, the seeds of barrier crops (especially the ones of oil palm) attract
rodents who also feed on cocoa pods, and citrus may attract white flies (see section 4.3.1.2).
Another disadvantage is the potential loss of income due to lower prices for barrier crops.
3) Planting seedlings of partly tolerant hybrid cocoa varieties (Padi et al., 2013) involves
resistance breeding, which can reduce CSSVD infection by up to 30% (Andres et al., 2017).
However, high genetic variability of the virus impedes the development of resistant varieties.
4) Removing alternative host trees (Posnette et al., 1950b) involves removing large shade trees
which naturally harbor CSSV, thereby destroying potential sources of infection and habitats of
the mealybug vectors. The disadvantages of this measure are the high costs associated with
materials (chainsaws, fuel, etc.) and labor, and it causes reduced shade, which is vital to reduce
the severity of CSSVD symptoms (Andres et al., 2018).
However, the adoption of the recommended prevention measures remains limited and CSSVD is still
highly prevalent in Ghana (Ameyaw et al., 2014). The reasons for the reluctance of farmers to
implement prevention measures remain largely unknown and there has been little investigation on

48
the barriers to their adoption. Therefore, the aims of this study are to identify both the extent to which
these prevention measures have been adopted, and the main constraints that prevent cocoa farmers
in Ghana from adopting them. To address these aims, we investigated the state of farmers’ knowledge
about CSSVD in general, and about the prevention measures in particular, and the feasibility of the
individual prevention measures for the farmers. Furthermore, we identified the actions that could be
taken to increase adoption, and the stakeholders who are most likely to be the drivers of change
towards increased adoption of prevention measures.
Based on the structure of the cocoa growing industry in Ghana, a review of the literature and the
results of previous research, we hypothesized that the number of farmers who adopt recommended
CSSVD prevention measures is significantly higher for farmers:
i) with bigger cocoa farms, as the implementation of barrier (strip) cropping, for instance,
would claim a smaller percentage of their land than it would for smaller farms;
ii) who know more people who have adopted the measures (social network size) because it
has been shown that early adopters tend to be those with better information networks
(Aguilar-Gallegos et al., 2015);
iii) who had more years of experience in cocoa farming because more experienced farmers
are more likely to have received information about prevention measures by extension
services;
iv) who had been trained by a voluntary sustainability standard (VSS) certification scheme
(e.g. Fairtrade, Rainforest Alliance or UTZ Certified) because VSS teaches farmers about
prevention measures and have been shown to trigger behavioral change (Ruben and
Zuniga, 2011);
v) who have had previous experiences with CSSVD infection because exposure to a
production challenge can trigger behavioral change (Brondizio and Moran, 2008);
vi) with more knowledge about the measures because knowledge about prevention
measures is key to their implementation; and
vii) who are the owners of the land they cultivate (as opposed to sharecroppers) because
landowners have higher freedom of action than sharecroppers, since they do not face the
fear of job loss when adopting certain measures (Lambrecht and Asare, 2015).
4.2 Materials and methods
4.2.1 Selection of study sites and participants
We conducted our study in the Eastern and Western Regions of Ghana. These areas have a long history
of CSSVD occurrence (Domfeh et al., 2011), and thus many farmers are familiar with the disease and
may have considered implementing prevention methods. Furthermore, the regions show biophysical
as well as socio-economic differences: Eastern Ghana has soils that are more suitable for cocoa growing
and the land is mostly owned by families. The Western Region has soils that are marginal for cocoa
growing and there are more migrants growing cocoa in sharecropping agreements. In each region, we
selected two operational districts of CHED: Oyoko (high CSSVD prevalence) and Nkawkaw (low CSSVD
prevalence) in the East, as well as Boako (high CSSVD prevalence) and Sefwi Bekwai (low CSSVD
prevalence) in the West (Figure 14).

49
Figure 14: Operational districts of the Cocoa Health and Extension Division (CHED) in which the study
was carried out: Oyoko (high prevalence of cocoa swollen shoot virus disease (CSSVD)) and Nkawkaw
(low CSSVD prevalence) in the East, Boako (high CSSVD prevalence) and Sefwi Bekwai (low CSSVD
prevalence) in the West.

CHED assisted us in the selection of farmers for the survey and focus group discussions. Since we relied
on their recommendations and had to consider logistical aspects (i.e., accessibility), our selection
method was not random. However, we ensured representativeness by selecting communities based
on spatial data about the prevalence of CSSVD, certification and farm size, and then selecting individual
farmers in each community.
The median values characterizing the participating farmers were: 50 years old, living in a household
with seven members, had been cultivating cocoa for 15 years and had two cocoa plantations of about
1.2 hectares each. About a quarter of the farmers were women, and two thirds were landowners, while
one third were sharecroppers. Our sample is in line with those by Hainmueller et al. (2011) and

50
Muilerman (2013), who reported the median values of: farmer age (50), household size (5 and 6 in
Hainmueller et al. (2011) and Muilerman (2013), respectively), number of farms per farmer (2), farm
size (1.2 hectares), proportion of women farmers (19%), and proportion of landowners (70%). 43% of
the participating farmers produced cocoa certified by a VSS such as Fairtrade, Rainforest Alliance or
UTZ Certified, which is considerably higher than the 15% reported by Deppeler (2014).

4.2.2 Data collection

4.2.2.1 Farmer survey
Between January and March 2017, we performed a face to face quantitative survey with 396 farmers
in the four CHED operational districts mentioned above (see section 4.2.1), using a questionnaire with
18 closed single-choice and three open-ended questions, which we had pre-tested and adapted in
collaboration with experts from the Research Institute of Organic Agriculture (FiBL). While the closed
questions were aimed at investigating the factors that might hinder the adoption of recommended
CSSVD prevention measures, the open questions were aimed at getting a better understanding of the
state of farmers’ knowledge about CSSVD in general and about the prevention measures in particular.
Responses were recorded directly on pre-programmed tablet computers. With the help of CHED
facilitators, we conducted all interviews in the local language (Twi) and attempted to create a common
baseline knowledge among all the participants by providing information about the virus, its symptoms
and prevention measures using illustrated information leaflets. The information leaflet was provided
after initial questions on farmers’ knowledge of the prevention measures had been asked, and before
we had a more detailed discussion on the individual prevention measures.
We collected data about age, gender, household size, total acreage of cocoa farms and number of
fields, social network size, number of years in cocoa farming, certification (binary, yes/no), experience
with CSSVD infection (binary, yes/no), knowledge about prevention measures (binary, no/some
knowledge), and land tenure (binary, landowner/sharecropper). Regarding the variable knowledge, we
coded it 1 (‘some knowledge’) if a farmer was able to mention a prevention measure, whereas we
coded it 0 (‘no knowledge’) if a farmer mentioned any other measure (e.g. cutting trees as a treatment
for CSSVD infections) or nothing at all. We obtained the acreage per field by dividing the total acreage
of each cocoa farm by the number of fields. The social network size was defined as the number of
family members or close friends who had already adopted a particular prevention measure. We
assessed adoption of CSSVD prevention measures by asking farmers whether they knew the measures,
and, if yes, whether they had tried them before, or were still practicing them.

4.2.2.2 Focus group discussions and transdisciplinary multi-stakeholder workshop

With the goal to complement the farmer survey, we performed six semi-structured focus group
discussions, between January and March 2017, using a questionnaire with main, additional and
clarifying questions as proposed by Laforest (2009). The questions were pre-tested and adapted in the
same way as described above (see section 4.2.2.1). To facilitate free talking, we selected farmers who
belonged to the same farmer group or originated from the same community. The focus groups
consisted of eight to eleven farmers. Again, we conducted all the discussions in the local language (Twi)
with the help of local moderators.

51
With the aim to validate the results from the farmer survey and the focus group discussions, we
conducted a workshop with representatives of the major stakeholders in the cocoa value chain of
Ghana in April 2017. A total of 31 people representing the six major stakeholders participated in the
workshop: i) CHED extension officers, ii) COCOBOD managers, iii) CRIG scientists, iv) farmers, v)
licensed buying company (LBC) representatives, and vi) Non-Governmental Organizations (NGOs)
representatives. Through interactive exercises, participants elaborated stakeholder-specific needs and
challenges concerning CSSVD. Furthermore, they discussed the necessary future steps towards
effective implementation of the most promising solutions, and gave opinions about the responsibility
of individual stakeholders to take these steps.
4.2.3 Data analysis
Qualitative data from the open-ended questions in the farmer survey, the focus group discussions, and
the multi-stakeholder validation workshop were translated into English, and analyzed for common
themes or categories using the guidelines by Eliot & Associates (2005). The identified themes were
then used to confirm or support the interpretations of the quantitative results from the farmer survey.
To analyses our quantitative data, we used binary logistic regression models for each of the four
prevention measures (dependent variables) to identify any significant effects on the adoption of CSSVD
prevention measures. To decide which explanatory variables to include in the logistic regression
analyses, we started from the model with all explanatory variables and two-way interactions, and
removed non-significant interaction terms as well as separate variables from the model if they did not
improve the model fit (or pseudo R2) significantly. This led to four measure-specific models (Table 6),
which together included the continuous variables: i) ‘acreage per field’, ii) ‘social network size’, iii)
‘number of years in cocoa farming’; and categorical (binary) variables: iv) ‘certification’, v) ‘experience
with CSSVD infection’, vi) ‘knowledge’, and vii) ‘land tenure’, as well as some influential two-way
interactions of these explanatory variables. In addition, we included the variables ‘age’, ‘gender’ and
‘district’ in all four measure-specific models to account for confounding effects.
We used the Wald chi-square statistic (Wald χ2) to test the statistical significance of individual
regression parameters. To assure that the model assumptions for binary logistic regression were not
violated, we checked for linear relationships (linearity of logit) using the model fit statistic and R2
values. We also checked for multicollinearity by verifying correlations among explanatory variables
with the help of a correlation matrix. As none of the variables showed correlations of r ≥ 0.8,
multicollinearity was considered to be negligible in our dataset, which was confirmed by the finding
that no variable had a greater Variance Inflation Factor than 2.5 (Allison, 2012).
Since the number of hypotheses we tested together was relatively small (five to six), we corrected for
multiple testing to minimize the likelihood of Type 1 error accumulation (rejecting the null hypothesis
when it is actually true) using the Bonferroni-Holm method, which is slightly less conservative than the
Bonferroni correction (Abdi, 2010). We performed all statistical analyses with the statistics software
IBM SPSS Statistics, Version 23.
Odds are defined as the likelihood of an event occurring versus the likelihood of the event not
occurring. It can be expressed with the following formula: odds = p/(1-p). Of particular interest are the
odds to achieve a certain outcome conditional on a variable (e.g. gender). As an example for gender,
after the odds to achieve an event for males and females have been calculated, the division of these
odds creates the odds ratio. This ratio states how much more likely it is for males (or females) to
achieve a certain outcome compared to females (or males).

52
The results were interpreted in light of the technology acceptance model (TAM) (Davis, 1989) as a
theoretical framework to study and interpret farmers’ behavior in the adoption of recommended
CSSVD prevention measures. TAM proposes that ‘perceived ease of use’ and ‘perceived usefulness’
combined, create an intention and lead to behavior. In practical terms, this means whether people
have the skills to use recommended CSSVD prevention measures, and whether they perceive their use
to be of benefit to them. In our study, however, behavior is further influenced by whether people have
access to recommended CSSVD prevention measures, so we adapted TAM by including the factor
‘perceived access’, which, in this case, implies knowledge of the prevention measure. Thus, the
theoretical framework guiding our study is based on the understanding that perceived access,
perceived ease of use, and perceived usefulness combined, create an intention and lead to behavior.

4.3 Results and discussion

4.3.1 Adoption of the four CSSVD prevention measures
The adoption of any of the four CSSVD prevention measures was most significantly influenced by the
factors ‘social network size’, ‘acreage per field’, ‘land tenure’, ‘knowledge’, ‘certification’, ‘experience’
(includes factors ‘number of years in cocoa farming’ and ‘experience with CSSVD infection’) and
‘gender’. These factors, or their interaction with another one of these factors, had significant effects
on the adoption of any of the four CSSVD prevention measures in 5, 4, 2, 2, 2, 2 (1 & 1) and 2 cases,
respectively (Figure 15).

Figure 15: Most significantly influencing factors on the adoption of any of the four cocoa swollen shoot
virus disease (CSSVD) prevention measures. The size of the circle is proportional to the number of cases
in which the respective factors had a significant effect on the adoption of any of the four CSSVD
prevention measures.
53
The model results from binary logistic regression analyses about the adoption of the four CSSVD
prevention measures are shown in Table 6. Of particular interest is column Exp(B) (odds ratio), which
states – in the case of gender, for example – how much more likely it is for males (or females) to adopt
a particular prevention measure. To calculate the odds ratio for the interaction terms, the parameter
estimate (B) and standard error of the mean (sem) are needed, therefore, they were included in Table
6. Furthermore, it is important to know the reference category from which the odds ratio is calculated
for all categorical variables. These are ‘no knowledge’ (knowledge), ‘small social network’ (social
network), ‘sharecropper’ (land tenure), ‘some infection’ (CSSVD infection), ‘certified’ (certification),
and ‘female’ (gender).

Table 6: Model results from binary logistic regression analyses about the adoption of the four CSSVD
prevention measures
Variables (dependent/explanatory) B sem Wald χ2 P value Exp(B) 95% CI for Exp(B)
Lower Upper
Adoption cordon sanitaire
Social network size 4.517 1.420 10.112 .018* 91.524 0.001 0.177
Knowledge 3.780 1.300 8.451 .040* 43.802 0.002 0.292
Social network size × number of years in
0.102 0.051 3.942 .047* 0.903 1.001 1.224
cocoa farming
Adoption barrier (strip) cropping
Knowledge 3.397 0.836 16.511 .001** 30.303 0.007 0.172
Social network size 1.955 0.480 16.618 .001** 7.042 0.055 0.362
Acreage per field 0.528 0.167 10.023 .020* 1.695 1.223 2.350
Land tenure 3.176 1.047 9.206 .029* 23.961 3.079 186.465
Acreage per field × land tenure -0.484 0.176 7.574 .006* 0.617 0.437 0.870
Adoption hybrid seedlings
Acreage per field × experience with
0.324 0.116 7.761 .005* 1.382 1.101 1.735
CSSVD infection
Certification × social network size -0.092 0.033 7.836 .005* 0.913 0.856 0.973
Adoption alternative host tree removal
Social network size 0.046 0.017 7.142 .023* 1.047 1.012 1.084
Gender 2.512 1.068 5.535 .037* 12.334 1.521 100.030
Gender × land tenure -2.943 1.113 6.994 .008* 0.053 0.006 0.467
Acreage per field × certification -0.248 0.116 4.587 .032* 0.781 0.622 0.979
B: parameter estimate; sem: standard error of the mean; Wald χ2: Wald chi-square statistic; Exp(B): odds ratio;
CI: confidence interval; *, **: significant at P < 0.05, < 0.01, respectively (after Bonferroni-Holm correction).

The percentages of surveyed farmers who had already adopted prevention measures were: 9% for
cordon sanitaire, 12% for barrier (strip) cropping, 41% for hybrid seedlings, and 40% for alternative
host tree removal. In reality, however, the adoption rates may be even lower because hybrid seedling
and alternative host tree removal are associated with other perceived benefits, and thus may not have
been adopted to prevent CSSVD.
The prevention measures ‘hybrid seedlings’ and ‘alternative host tree removal’ have a high adoption
potential because farmers perceived them to be useful. However, barriers to adoption were found to
be difficulties in accessing the hybrid seedlings, and that removal of alternative host trees is not
perceived to be easy. Farmers perceived cordon sanitaire and barrier (strip) cropping to be less useful,
which highlights the need for education campaigns to stress the effectiveness of these prevention
measures (Domfeh et al., 2016; Vanderplank, 1947). The participating farmers consistently mentioned
the need for technical advice and access to hybrid seedlings as challenges, which participants of the
validation workshop also stressed to be of vital importance for potential solutions to prevent CSSVD.
54
4.3.1.1 Cordon sanitaire
‘Social network size’, ‘knowledge’ and the interaction between ‘social network size’ and ‘number of
years in cocoa farming’ had significant effects on the adoption of cordon sanitaire Table 6). The results
suggest that being relatively new to cocoa farming combined with having a social network of two or
more who had already adopted cordon sanitaire were associated with more adoption of this
prevention measure. This may be explained by the responses to the open-ended questions in the
interviews, which indicated that cocoa farming, in general, is a traditional livelihood activity of local
families, where children learn about farming practice from their parents. Farmers with more than 15
years of experience may have firmly established habits in their practice, and may therefore be
reluctant to accept and adopt new recommendations that are unfamiliar to them, such as leaving a
cordon sanitaire around the entire farm. Similarly, farmers with some knowledge about cordon
sanitaire were much more likely to adopt the measure than farmers without any knowledge (Table 6).
Farmers stated that cordon sanitaire is not feasible because it would lead to water stress for adjacent
cocoa trees, higher weed pressure and thus labor requirements, and that neighboring farmers would
claim the fallow land. Thus, although farmers may have been convinced of the usefulness of cordon
sanitaire, many perceived it to be difficult to implement, which violates a requirement of the
technology acceptance model (Davis, 1989). Of the farmers who mentioned specific challenges with
cordon sanitaire, 62% said that their farm is too small to implement this preventive measure.

4.3.1.2 Barrier (strip) cropping

‘Knowledge’, ‘social network size’, ‘acreage per field’, ‘land tenure’ and the interaction between
‘acreage per field’ and ‘land tenure’ had significant effects on the adoption of barrier (strip) cropping
(Table 6). Similar to the results for cordon sanitaire, the two strongest predictors for adoption were
‘knowledge’ and ‘social network size’. With a median farm size of about 1.2 hectares, implementing
barrier (strip) cropping may claim up to 30% of the land, which may be another explanatory factor for
the limited adoption. Furthermore, potential revenues from barrier crops may not fully compensate
for reduced cocoa yields (Ameyaw et al., 2014), so there is a need for more in depth evaluation of the
economic potential of different barrier crops (Domfeh et al., 2016).
Our results suggest that landowners were more likely to adopt barrier (strip) cropping than
sharecroppers were, but adoption increased significantly when the farm size was larger than 1.2
hectares. This was found to be especially the case for sharecroppers. While this result suggests that
landowners may be more open to considering requests from their sharecroppers with larger farm sizes,
it also points to the limited freedom of action for sharecroppers, which is associated with the fear of
job loss when adopting measures such as cutting out infected cocoa trees (Lambrecht and Asare,
2015). Focus group participants confirmed the threat of landowners taking back their land in such
cases. Thus, the root cause of the problem may yet again be a lack of knowledge, but on the side of
the landowners. Until the landowners are convinced of the usefulness of barrier (strip) cropping,
adoption of it will be minimal, which violates the ‘perceived usefulness’ requirement of the technology
acceptance model (Davis, 1989). Educating landowners about CSSVD may help them to make informed
decisions about future land use together with their sharecroppers, and thereby remove this barrier to
adoption. Farmers mentioned access to seedlings as the most important challenge to adopt barrier
(strip) cropping. Furthermore, barrier crops may compete with cocoa for light and water, which may
favor black pod disease (Phytophthora spp.), and they can attract pests such as rodents (oil palm) or
white flies (citrus).
55
4.3.1.3 Hybrid seedlings
The two interactions between ‘acreage per field’ and ‘experience with CSSVD infection’, as well as
between ‘certification’ and ‘social network size’ had significant effects on the adoption of hybrid
seedlings (Table 6). Farmers who had individual farms of more than 1.2 hectares in size were less likely
to use hybrid seedlings when their farm was infected by CSSVD. This may be explained by the greater
financial resources needed to replant bigger farms with hybrids compared to smaller farms. Moreover,
farmers with a larger social network who already adopted hybrid seedlings, and especially those who
had been trained by a VSS certification scheme, showed higher adoption rates compared to farmers
with a smaller social network. This may be due to larger social networks of certified farmers being
more efficient in spreading knowledge about hybrids. We therefore encourage organizations that are
active in certification to exchange up to date information with extension services to profit from each
other’s experiences. When asked about their challenges, farmers mainly stated access and associated
transport costs as well as timely delivery of seedlings (i.e., during the rainy season). This was especially
the case for farmers living in remote villages. The primary barrier to adoption of this prevention
measure appears to be related to access to the seedlings, which is essentially an issue of access to the
technology (Davis, 1989).

4.3.1.4 Alternative host tree removal

‘Social network size’ and ‘gender’, as well as the two interactions between ‘gender’ and ‘land tenure’,
and between ‘acreage per field’ and ‘certification’ had significant effects on the adoption of alternative
host tree removal (Table 6). The strongest predictor indicated that farmers with a larger social network
were more likely to remove alternative host trees.
Our results suggest that once the individual farms were larger than 1.2 hectares in size, farmers who
had been trained by a VSS certification scheme were more likely to remove alternative host trees
compared to non-certified farmers. This may be due to certification agents educating farmers about
alternative host trees. The increased adoption on larger farms may be explained by farmers disposing
of more financial resources, which enabled them to fell the trees or simply by a higher frequency of
alternative host trees on larger surfaces. While the adoption rates of prevention measures did not
differ between male or female landowners, male sharecroppers removed alternative host trees
significantly more often than their female counterparts did. This indicates that less secure land tenure
imposes more insecurities for women than for men, which affects their willingness to risk adopting
alternative host tree removal. An alternative explanation is that women sharecroppers may simply
have access to fewer resources than men have, or lack the necessary physical force to fell alternative
host trees by themselves.
When asked about their challenges, farmers mentioned the damage caused by felling shade trees, the
need to provide shade for cocoa trees and the associated costs for machines and labor to remove the
trees. The primary barrier to adoption of this prevention measure appears to be the difficulties (and
costs) in implementing the prevention measure, which the technology acceptance model would
describe as an issue of ‘perceived ease of use’, although the ‘perceived usefulness’ (Davis, 1989)
appears to be also diminished by the prevention measure producing another problem that needs to
be solved.

56
4.3.2 General needs and challenges of different stakeholders, and proposed solutions
Different stakeholders expressed contrasting views on how to tackle CSSVD. While farmers stressed
that lack of information and education presented the biggest challenge to preventing CSSVD and was
therefore their greatest need, other stakeholders mainly stressed that they are facing resource
constraints to address the CSSVD challenge. Furthermore, participants of the validation workshop
mentioned improving information flow, existing policies and land tenure systems as key issues to tackle
the CSSVD problem.
The participants of the validation workshop agreed that more collaboration and participation among
stakeholders is needed in aspects such as research, information sharing, quality assurance and
policymaking. They suggested that increased collaboration would facilitate information exchange
while also building mutual respect and trust, and may be achieved through the establishment of
demonstration farms across the cocoa growing regions of Ghana by collaborative efforts of COCOBOD,
NGOs, LBCs and farmers. It would be especially important to involve the traditional leaders (chiefs) in
those platforms, as their consent is needed to ensure that the word will spread to the farmers who are
affected by CSSVD.
Furthermore, the chiefs are key actors when it comes to the revision of Ghana’s land tenure system.
The current system has led to small farm sizes, as inheritance and sharecropping practices continue to
fragment the cocoa landscape into ever-smaller cocoa farms (Dzahini-Obiatey et al., 2010). This is not
only problematic for the rehabilitation of farms destroyed by CSSVD, but it also impedes the adoption
of cordon sanitaire and/or barrier (strip) cropping. To effectively increase farm size, the proposition to
set up land banks should be revisited by the central government (Dzahini-Obiatey et al., 2006).
4.3.3 Knowledge and information flow
Lack of knowledge about CSSVD prevention measures was the single biggest barrier for their adoption,
with 51% of all the participating farmers not even being aware of them. Lack of awareness means that
the prevention measures are not accessible to farmers, thus a requirement of the technology
acceptance model is not fulfilled (Davis, 1989). Prior to showing the farmers the information leaflets,
only 12% mentioned that they knew at least one of the four prevention measures. Farmers who had
been shown the leaflets reported knowledge of cordon sanitaire (37%), barrier (strip) cropping (41%),
use of hybrid seedlings (69%) and alternative host tree removal (71%). This shift indicates a significant
lack of knowledge in that most farmers were not aware of the measures being at least partly effective
in CSSVD prevention, even when many farmers knew about them. The lack of knowledge was also
reported in the focus group discussions and the multi-stakeholder workshop.
Furthermore, the results suggest that the social network has the most potential as a source of
information about CSSVD prevention measures that may lead farmers to adopt them. However, the
median values of social network size was 0 people for cordon sanitaire, 0 for barrier (strip) cropping, 6
for hybrid seedlings, and 5 for alternative host tree removal, indicating that more than half of the
farmers did not know anyone who had already adopted cordon sanitaire or barrier (strip) cropping.
While we were able to evaluate the size of social networks, a limitation of our study was that we were
unable to check the accuracy of information spread by social networks. Furthermore, we were not able
to conduct follow up interviews with participating farmers to confirm our interpretation of their
responses. These remain topics for future research.
Baah (2008) showed that the media (radio, TV) was the preferred information source of farmers, with
extension officers and social networks ranking second and third. Thus, CSSVD-specific radio programs
57
could be a potential solution to addressing the identified lack of knowledge about prevention
measures. Specific videos that could be transmitted via television or shared among farmers via
Bluetooth on 3G mobile phones, a technology whose potential has been demonstrated for West Africa
(Sousa et al., 2016), may also enable transfer of knowledge about the prevention measures.
Furthermore, the focus group discussions revealed that farmers would like to be visited more
frequently by extension officers but the high farmer to extension officer ratio of 2’500/1 challenges
CHED to physically reach out to farmers, which underlines the reported inadequacy of extension
support for Ghanaian cocoa farmers (Baah, 2008). The participants of the validation workshop
collectively stated the opinion that this ratio should be lowered to at least 500/1.
The current knowledge on the barriers to the adoption of CSSVD prevention measures for farmers is
very limited. Here, our study offers novel insights, showing that more than half of the participating
farmers were not even aware of what they can do to prevent CSSVD. Insufficient knowledge spread is
the key constraint that has to be overcome to increase the currently limited adoption. Based on our
data, we suggest to capitalize on farmers’ social networks to enhance information flow in order to
achieve this. Furthermore, the revision of Ghana’s land tenure system is imperative to address the
issues of small farm sizes and insecure land tenure rights, both of which are hindrances to increasing
the adoption of CSSVD prevention measures. With this study, a first step has been taken towards a
potential action plan that enhances the adoption of CSSVD prevention measures in Ghana to efficiently
remediate the CSSVD menace.

4.4 Conclusion
Here, we identified significant barriers to the adoption of the four CSSVD prevention measures i)
cordon sanitaire, ii) barrier (strip) cropping, iii) using partly tolerant hybrid planting material, and iv)
removing alternative host trees by cocoa farmers in the Eastern and Western Regions of Ghana. In
general, we found a severe lack of knowledge, and hence low adoption rates, among the participating
farmers with respect to CSSVD and prevention measures. Farmers with wider social networks, larger
farms, more knowledge and those who were landowners were more likely to adopt the measures than
farmers with smaller social networks and farms, less knowledge and being sharecroppers.
Furthermore, farmers with fewer years of experience in cocoa farming, along with those who had
previous experience with CSSVD infection showed higher adoption rates.
Our results suggest that the four CSSVD prevention measures partly satisfy the prerequisites of the
technology acceptance model (Davis, 1989), showing that access (awareness) was the most important
factor influencing adoption. Since a combination of locally adapted prevention measures may provide
the best defense against CSSVD, complex information needs to be conveyed to smallholders in an
understandable way. One solution could be the development of a manual for sustainable cocoa
production, which participants of the validation workshop stressed to be of vital importance. A further
challenge for future research is to explore knowledge transfer methods that are appropriate in the
Ghanaian context because we found the information flow among stakeholders in the cocoa value chain
of Ghana to be a key factor affecting knowledge spread, and consequently the adoption of prevention
measures. Lastly, we recommend that anyone interested in providing agricultural information to
Ghanaian cocoa farmers should consider a dissemination strategy based around radio programs and
videos on TV and mobile devices as an alternative to engaging a large numbers of extension officers.

58
5. General discussion, conclusions and outlook
To discuss the CSSVD issue holistically, I would like to draw the analogy of a sick cocoa tree to a sick
human being: prevention is the key to staying healthy. There are many examples how regular exercise,
healthy food and enough sleep keeps human beings healthy and makes them resilient to diseases.
Once one contracts a disease, however, there are different types of remedies the diseased person can
take or apply in order to cure the disease, or at least mitigate its’ severity. Whether prevention or cure,
one thing is essential: information. If we do not know possibilities to prevent, cure or mitigate the
disease, we risk falling victim to its’ full adverse effects. Thus, prevention is better than cure and
information is key to both prevent and treat successfully.

It is not much different with cocoa trees and CSSVD. There are preventive control measures for CSSVD
(see section 1.3) that, if applied correctly, have the potential to prevent a cocoa plantation from
becoming infected. I discuss these in section 5.1. Once the farm is infected, however, the only thing
that can be done is to cure or mitigate the disease, i.e., by cutting and replanting (cure), or by, for
example, diversifying production systems to sustainable AFS (mitigation). I discuss the question if we
should cure diseased farms, or try to cope with CSSVD through sound agroforestry approaches in
section 5.2. Whether cutting, replanting or diversifying, farmers need to be well informed about
CSSVD. They need information about preventive control measures and curative or mitigating strategies
so that they can keep their farms healthy. I discuss the need for information to tackle CSSVD in section
5.3. An alternative way of looking at the CSSVD “problem” is that CSSVD is only a symptom of a much
deeper lying problem, that is, unsustainable production systems and landscapes. I change the
perspective to discuss the question if CSSVD is the actual problem we need to solve in section 5.4.
Finally, section 5.5 outlines the need for multi-dimensional innovation to address the CSSVD issue and
section 5.6 gives an outlook, discussing the need for transdisciplinarity in future cocoa research.

5.1 Prevention is better than cure

Prevention is better than cure, but how to prevent CSSVD? Undoubtedly, past research has put the
biggest focus on resistance breeding. However, the huge amount of resources spent by research
institutes in West Africa during the last 80 years on breeding and related research such as virology and
epidemiology have only led to partial successes and have not had the desired effects on cocoa
productivity so far, as today we have no CSSVD-resistant variety. Varietal resistance surely is an
approach with a great inherent potential. However, the varieties that are available today may reduce
CSSVD infection only by up to 30% (see chapter 2), and the problems outlined above (see section 1.3.1)
complicate the development of resistant varieties and call for concerted efforts among the breeding
communities across continental borders.

The case of mild strain cross-protection is not much different. Even though past research spent a
considerable amount of resources on this technique, it may only reduce CSSVD infection by as little as
up to 30% (see chapter 2). The challenges outlined above (see section 1.3.2) have not been resolved,
and resolving them in the future seems hardly feasible. In addition, there is no feasible approach to
implement mild strain cross-protection at large scale. Both resistance breeding and mild strain cross
protection certainly have potential. To solve a problem as complex as CSSVD, they should be combined
with other approaches.

59
The approach of controlling mealybug vectors also bears a high potential (see chapter 2). However,
efforts with this technique have also proved futile so far. In the case of contact insecticides,
applications were inefficient due to the mealybugs’, waxy, water-repelling body and its cryptic
behaviour, hiding between cocoa pods and cocoa tree stems (Ameyaw et al., 2014). The use of
systemic insecticides, though effective (Hanna et al., 1952), was found problematic because of high
cost, toxicity to mammals and residues in cocoa beans (Domfeh, personal communication).
Furthermore, contact insecticides have adverse effects on natural antagonists and pollinators,
disrupting functional biodiversity and thereby lowering the self-regulation capacity of agro-ecosystems
(Forster et al., 2013). On the other hand, biological control of the vectors still bears an unexploited
potential to remediate the problem (Padi, 1997). After several failed initiatives in the 1950s, the
recently revived efforts at CRIG should be pursued, and the suggestion of farmers to control mealybugs
by altering ant-complexes in the field should be incorporated into the ongoing activities (Ayenor et al.,
2004).

The last, but not the least, prevention measure on offer is barrier (strip) cropping (Domfeh et al. 2016).
In light of the research presented above, I think this is the single most promising approach to bring
about rapid change in the farmers’ context, because it does not involve the use of costly inputs, can
be implemented directly by farmers themselves on their farms, and has a very high reduction potential
(up to 85% CSSVD reduction, see chapter 2). Thus, barrier (strip) cropping systems might be well suited
to complement breeding and mild strain cross protection. Breaking the continuity of the cocoa
landscape is of utmost importance to address the underlying root cause of the problem, i.e., large,
uniform cocoa landscapes characterized by very low species diversity. However, this measure does
cost land in terms of acreage farmers could otherwise allocate to cocoa, so unless farmers are able to
make a benefit through diversified production the adoption of this technique may also remain limited.

5.2 Should diseased farms be cured? The prospect of agroforestry to cope with CSSVD
In view of the scope of the CSSVD problem, it becomes clear that there are no short-term solutions.
Different types of innovation in different dimensions are required to address the issue holistically (see
section 5.5). Cocoa farmers in areas of mass infection (AMI, parts of the Eastern and Western Regions
of Ghana), on the other hand, need an immediate strategy to deal with the problem. Ways of living
with CSSVD must be identified, especially as complete eradication and replanting of AMI at once is not
feasible. Diversifying production systems and landscapes that are not yet infected, while subsequently
replacing diseased cocoa with more resistant varieties may be an interesting option for farmers. Here,
agroforestry can offer solutions.

Addressing the need for more on-farm research about the potential of agroforestry systems to mitigate
the severity of the cocoa swollen shoot virus disease, I found that, indeed, shade reduced CSSVD
symptom severity and capsid damage in cocoa stands aged 16 years and above, which may have helped
lowering the yield-reducing effect of CSSVD. In other words, AFS can be a great coping strategy to live
with CSSVD until farmers are able to cut and replant diseased cocoa, provided one dedicates a lot of
energy into farm management and applies good agronomic practices. The example of “Sronko
(Unique) Farms” in Eastern Ghana shows that even if they are fully infected with CSSVD severe strain
1A, farms can still be fairly productive.7 While moderate shade seems to be beneficial, excessive

7
https://www.youtube.com/watch?v=7UCY33zG9RE&feature=youtu.be
60
shading may have adverse effects. My research indicates that agroforestry systems with around 50%
shade seem to optimally balance CSSVD symptom severity and reduced cocoa yield, which is in line
with existing recommendations for sustainable cocoa cultivation (Beer et al., 2004; SAN, 2005; Opoku-
Ameyaw et al., 2010; Schroth et al., 2016a).

The way a cocoa agroforestry system is set up may influence its long-term resilience. For reasons of
feasibility, and due to legal obstacles for farmers to acquire the rights to use timber trees in their farm
(Clientearth, 2013), West African farmers often plant cocoa into thinned secondary forest rather than
felling existing trees and establishing temporary shade such as plantain (Musa spp.). Such systems, in
most cases, result mainly from farmers knowing about the necessity of shade for proper establishment
of young cocoa. While this practice is comprehensible from a socio-economic point of view, it may lead
to problems of resource competition, as the roots of already established plants may significantly
reduce the growth of plants that are introduced at a later point in time (Faget et al., 2012). Thus, in a
cocoa plantation established from thinned secondary forest, the remaining shade trees that have
already colonized the topsoil with their lateral roots may induce water stress for cocoa trees, especially
under extreme environmental conditions such as drought (Abdulai et al., 2017). A similar mechanism
might also lead to nutrient competition, explaining the limited increase in soil fertility recent studies
observed in common cocoa agroforests (Wartenberg et al., 2017; Blaser et al., 2017; chapter 3). Given
the contradictory results about the soil fertility effects of shade trees reported in different studies
(positive soil fertility effects are mentioned in section 1.4.4), the question about suitable management
practices in AFS warrants further investigation. Future research should study the necessary
adaptations of management practices to different environmental and socio-economic contexts in West
Africa in order to reap AFS’s full range of potential benefits.

The underlying reasons for the differences observed in shaded compared to non-shaded systems also
warrant further investigation. Shade could be a proxy for agro-biodiversity, soil fertility, soil moisture
and many other factors. My results suggest that high-light stress, and not soil fertility, was the main
explanatory factor for the shade effects I observed. Moisture stress is another potential explanatory
factor, which I was not able to cover in my research and which should thus be analysed in future
research. Furthermore, shade should be considered in breeding programs for CSSVD resistance, as
adapted varieties that are productive and healthy under medium shading regimes are an integral part
of the much called-for sustainable cocoa production systems that are necessary in order to safeguard
the future of chocolate.

5.3 The need for information to tackle CSSVD

My study on the practical relevance of different preventive control measures for farmers in the Eastern
and Western Regions of Ghana revealed a severe lack of knowledge, and hence low adoption rates,
with respect to both CSSVD in general and prevention measures in particular. Farmers with wider social
networks and more knowledge were more likely to adopt the measures than farmers with smaller
social networks and less knowledge. My results suggest that awareness about preventive control
measures was the most important factor influencing adoption. As information flow among
stakeholders in the cocoa value chain of Ghana is a key factor affecting knowledge spread, targeted
innovations in information and communication technologies (ICT) are needed to enhance the adoption
of prevention measures (see section 5.5).

61
5.4 Changing the perspective – is CSSVD the problem we need to solve?
An alternative way of looking at the CSSVD “problem” is that CSSVD is only a symptom of a much
deeper lying problem, that is, unsustainable production systems and landscapes. More than half of the
world’s cocoa grows in unsustainable full-sun monocultures in Ivory Coast and Ghana, which rely
heavily on the use of agrochemicals and improved genetic material specifically developed and
optimized for these systems (Ruf, 2011; Tondoh et al., 2015; FAO, 2018). The various ecological,
economic and social issues associated with these systems (Tscharntke et al., 2011) include old tree
stocks, degraded soils and prevalence of pests and diseases (Wessel and Quist-Wessel, 2015), as well
as vulnerability to climate change (Schroth et al., 2016a), low productivity (van Vliet and Giller, 2017),
low farm gate prices and an ageing farming population (ICCO, 2012).

Why are there so many problems with full-sun systems? Explanations inevitable link us back to the
ecological background of the cocoa tree (see section 1.4.1), which suggests that the current production
systems are not in line with the cocoa tree’s ecology. Taken from their natural environment, the dense
humid Amazon rainforest, and placed into full sunlight on extensive monocultures in West Africa and
beyond, the fragile cocoa trees are not in their ideal environment. The main issue at stake here is the
question of entropy vs. syntropy. In contrast to entropy, syntropy (also known as negentropy) is a
result of retro-causality leading to persistent and more complex organization. Cocoa production
systems that are mainly characterized by entropy, however (i.e., full-sun systems), are characterized
by simpler organization (i.e., reduced systems complexity) and thus less persistent, or subject to more
changes. These considerations from theoretical physics may be good to think about, but for as long as
we are stuck in the reductionist economic paradigm by which the world is currently ruled, we should
not expect large-scale changes towards long-term resilience of global cocoa production.

Current policies and market dynamics favour unsustainable agronomic practices by stimulating the
production of single commodities in large quantities. Such commodities are sold at very low prices at
the cost of the environment, the producers and, ultimately, humankind as such. If these costs were
internalized (true cost accounting (TCA)), produce coming from unsustainable production systems
would become more expensive and sustainable produce more competitive. Here, science has to
develop feasible solutions for all stakeholders, which may translate into a shift towards higher
sustainability of agricultural production. However, there are major research gaps in the quantification
of the true environmental, social, and health costs of different agricultural production systems.
Furthermore, we lack a common, feasible and scalable TCA framework, and the practical
implementation of such accounting systems is understandably complex, requiring dedicated efforts by
policy institutions based on comprehensive research findings (Andres and Bhullar, 2016). It is about
time that we create the much needed win-win-win scenarios where farmers, the rest of the (cocoa)
value chain and nature benefit. We should find ways out of the self-destructive behaviour of eroding
our very resources.

5.5 The need for multi-dimensional innovation to address the CSSVD issue
The context of CSSVD is highly complex. It concerns different stakeholders and requires a range of
innovations in different dimensions. These include ecological, technical (biological, mechanical and
organisational), socio-political and economic innovations that are further detailed and discussed
below.

62
In terms of ecological innovation, there is a need to reshape both the design and the management of
cocoa production systems at farm and landscape levels, taking into account the social context we are
dealing with. On the farm level, this means feasible, productive systems that combine cocoa and shade
trees in synergistic relationships. A basic research question we need to address in this respect are
optimal DAFS designs (cocoa – tree species associations) for different biophysical and socio-economic
contexts. A DAFS that works in one place may not work in another place where either the biophysical
(soil and climatic factors) or the socio-economic context (land tenure rights, history and knowledge of
the farming population, presence and proximity of relevant institutions, etc.), or both are different.
For example, we need to know which shade tree species and annual intercrops work best on different
soil types with different physicochemical conditions and in different climates. Another question we
need to address is which types of shade tree species are easiest for farmers to propagate themselves,
and for which ones they need to rely on specialized nurseries that supply them with the respective
seedlings. The goal of such research should be the development of a modelling tool that facilitates
decision-making of what shade tree species and annual intercrops to use in a given biophysical and
socio-economic setting.

The likelihood of successfully establishing such highly diverse and productive systems is increased if
cocoa is not established in thinned secondary forests, but from plain fields, because of root systems
architecture of different tree species, as well as differing growth dynamics of pioneer, secondary, and
primary species (Götsch, 1994; Andres et al., 2016). The concepts of species succession and the
structure of natural ecosystems can serve as sources of inspiration in this respect. However, it is also
possible to establish DAFS in thinned secondary forests, as illustrated in a video from Eastern Ghana.8
Starting a new cycle of succession after felling at least some of the existing trees, and applying a thick
layer of mulch rather than burning the plant material, DAFS practitioners protect and build up the soil.
They capitalize on the dynamics of succession to exploit both the air and the soil volume fully through
stratification of above- and belowground parts of plants. By prioritizing practices such as high-density
planting and regular pruning, DAFS practitioners transform cocoa plantations into diverse “food
forests”. However, the main drawback of DAFS is undoubtedly their high labour- and knowledge-
intensity. Here, we need new projects that address the feasibility of DAFS for local producers, as well
as their environmental and socio-economic resilience.

As for the design of production landscapes, the importance of landscape connectivity should not be
underestimated. CSSVD can only spread if there are uniform production landscapes, i.e., vast acreages
of connected cocoa farms that facilitate the movement of the mealybug vectors. Some preliminary
results suggest that dense cropping correlates with CSSVD prevalence (unpublished data), which
makes sense when considering the behaviour of the mealybug vectors. As the treatment of vast
acreages of cocoa farms is challenging insofar as it involves the persuasion of many farmers to cut and
replant their farms, policy reform is urgently needed in this area. Policies should be formulated which
direct farmers to delimit their cocoa plantation with a strip of at least 10 m of a non-host crop (e.g.
coffee) when they renew their plantation. Another policy should aim at preventing that a new cocoa
plantation connects two existing cocoa farms and at maintaining a distance of at least 10 m between
cocoa areas. However, changing Ghana’s cocoa production landscape from uniformity to a complex
agro-ecological matrix of cocoa agroforestry systems (Perfecto and Vandermeer, 2010) is a gradual
process that will require time. Appendix B outlines a concept for a study on the importance of

8
https://www.youtube.com/watch?v=7UCY33zG9RE&feature=youtu.be
63
landscape connectivity for reducing the spread of CSSVD that came up during the course of the
research, but has not been brought to completion so far.

There are three main categories of technical innovations: biological, mechanical and organisational.
Biological innovations needed are resistant cocoa varieties, production of sufficient barrier (strip) crop
and shade tree seedlings for farmers, as well as biological control of the vectors, all of which I have
already explained in section 5.1.

In terms of mechanical innovations, the mechanisation of cocoa production will be very important to
attract young people to go into cocoa farming. The trend towards a lack of youth involvement in
current cocoa plantations is bound to continue unless cocoa farming is rendered more attractive. Many
young Ghanaians look for regular and decent income earning opportunities. Thus, cocoa farming needs
to reinvent itself, leaving the reputation of being a poverty trap of parents and grandparents behind,
and becoming a viable business alternative that combines well with innovative approaches such as
ecotourism. Success stories from Africa and other continents9 can serve as a source of inspiration for
young Ghanaians.

Another mechanical innovation with a high potential is a novel mapping tool for CSSVD occurrence
using drone imagery and normalized difference vegetation index (NDVI). If mapping CSSVD from the
air would become possible, this would be a major contribution towards increasing efficiency in the
Ghanaian cocoa sector by reducing the workload on CHED extension officers. This way, the wish for
more frequent visits by extension officers, which farmers expressed during various interviews and
focus group discussions, may be met. At the same time, more resources are needed to reduce the gap
in the farmer to extension officer ratio from currently 2’500/1 to the targeted ratio of 500/1 (collective
statement of validation workshop participants). Appendix C outlines a concept for a study on the
development of this novel mapping tool that came up during the course of the research, but has not
been brought to completion so far.

Finally, decentralised nurseries are an extremely important mechanical innovation, which is needed to
provide farmers with seedlings of resistant cocoa varieties, barrier (strip) crops and shade trees in a
timely manner. The workshop participants unanimously stressed this as one of the major shortcomings
in the Ghanaian cocoa sector that has to be overcome if Ghana’s production level is to rise again.

As for organisational innovations, ICT solutions are vital to improve the communication with farmers.
I found the information flow among stakeholders in the cocoa value chain of Ghana to be a key factor
limiting the adoption of CSSVD prevention measures. Given that a combination of locally adapted
prevention measures may provide the best defence against CSSVD, complex information needs to be
conveyed to smallholders in an understandable way, while making sure that knowledge transfer
methods are appropriate for the local context. This is a challenge future research should tackle in
collaboration with CHED and farmers using a transdisciplinary approach to target rapid
implementation. As we know that the media (radio, TV) is the preferred information source of farmers
(Baah, 2008), CSSVD-specific radio programs could be a potential solution to addressing the identified
lack of knowledge about prevention measures. Specific videos that could be transmitted via television
or shared among farmers via Bluetooth on 3G mobile phones – a technology whose potential has been
demonstrated for West Africa (Sousa et al. 2016) – may also enable knowledge transfer about
prevention measures. COCOBOD, as well as everyone else interested in providing agricultural

9
https://www.youtube.com/watch?v=gSPNRu4ZPvE
64
information to Ghanaian cocoa farmers, should carefully consider a dissemination strategy based
around radio programs and videos on TV and mobile devices as an alternative to engaging a large
number of extension officers.

While farmers easily understand videos on agricultural innovations, much is to be learned about how
to optimize such videos. For example, should the videos be participatory or should a scientist present
them as a lecture? Future projects should address this question, uniting agronomists, communication
scientists, sociologists and film-makers in transdisciplinary ways to create videos on CSSVD prevention.
After sharing the videos with farmers, project areas should be revisited to evaluate both video spread
and the adoption of portrayed innovations, as well as their real life impact. Such an approach may also
provide insights about the potential of videos to overcome existing gender gaps in terms of access to
agricultural information (Gurumurthy, 2006).

Another issue for future research is how to stimulate interest among young people to go into farming.
Farming could be portrayed as a beautiful work that reconnects humans to nature and thus lead to
ecosystem health from the soil through plants, animals, landscapes and humans (Costanza, 2012;
Figure 16). We can learn a lot from NGOs that are active in the field such as Organic India10. Such
approaches may help to overcome the image in the mind of African youths of farming being a tedious
and unprofitable endeavour, and thus help to reduce rural-urban migration and farm abandonment
(Odiaka, 2015). Reaching the youth through videos that are adapted to their way of communication
and language may be interesting in this respect. Preliminary versions such as the video “Cocoa
Sasabroah”11 (cocoa rheumatism, the name for CSSVD in the local language Twi) may be adapted in
order to maximize impact after resolving the question how youths gain their knowledge for farming
activities.

Figure 16: Schematic ecosystem health from soil to human.

10
https://us.organicindia.com/
11
https://www.youtube.com/watch?time_continue=2&v=QxVcgh0DKr4
65
Socio-political innovations carry arguably the most powerful potential to bring about large-scale
change towards more nature-friendly cocoa production. When making management choices, farmers
face trade-offs between productivity and other ecosystem services that may not directly contribute to
their livelihoods (Vaast and Somarriba, 2014). AFS usually do not produce as much cocoa as full-sun
systems during the establishment phase (Schneider et al., 2016). As cocoa is the main cash crop,
farmers will tend to continue going for the system that provides the quickest return on their
investment, i.e., full-sun systems. An incentive scheme for nature-friendly cocoa production could
reverse this trend. As the most suitable DAFS design and the best practices for enhancing productivity
and ecosystem services may vary under different environmental and socio-economic conditions,
understanding and supporting the development of such systems requires an incentive scheme that is
based on the analyses of local knowledge, strategies and practices of different stakeholders in the
cocoa value chain.

Another very important and urgently needed socio-political innovation is the revision of Ghana’s land
tenure system. Dzahini-Obiatey et al. (2006) have stressed the importance of this measure more than
a decade ago. Many farmers in Ghana lease their land, growing cocoa in informal sharecropping
agreements (Yelsang, 2013). The current system has led to small farm sizes, as inheritance and
sharecropping practices continue to fragment the cocoa landscape into ever-smaller cocoa farms
(Dzahini-Obiatey et al. 2010). This is not only problematic for the rehabilitation of farms destroyed by
CSSVD, but it also impedes the adoption of cordon sanitaire and/or barrier (strip) cropping. To increase
farm size effectively, the proposition to set up land banks (Dzahini-Obiatey et al., 2006) should be
revisited by the central government. It is of utmost importance to involve the traditional leaders
(chiefs) in this process, as their consent is needed to ensure local acceptance by the farmers who are
affected by CSSVD.

Finally, an economic innovation is also vital to address the CSSVD issue, especially with regard to the
flexibility of farmers. The fact that farmers are calling upon COCOBOD for the provisioning of adequate
equipment to manage their plantations (e.g. pruning shears for cocoa and high-altitude saws for shade
trees) shows their dependency on the authorities in thought and deed. Often, they cannot get access
to credits because they are too poor and cannot provide collateral for a potential loan. This impedes
their ability to develop a sense of entrepreneurship, viewing cocoa farming as a business rather than
a family tradition that is to be followed at all costs, even when running at a loss. According to Yunus
(2000), failure of traditional financial institutions to extend credit to the poor is the single most
important reason for the perpetuation of poverty. Providing money without beneficiaries having to
provide collateral and pay interests has worked in Bangladesh (Yunus and Weber, 2007). However,
bringing forward microcredit schemes for smallholders in West Africa calls for studies of the necessary
adaptations for making it work in the local context. Asamoah and Amoah (2015) showed in a study
with Ghanaian cocoa farmers involved in organised farmer associations that a microfinance model
helped them to mobilize savings, even if they did not have any savings culture before joining the
schemes. Key to this success was the group concept using peer support, group guarantee and social
capital generated through the formation of associations. We need more such studies, as well as further
implementing schemes and committed institutions to stimulate entrepreneurial thinking among cocoa
farmers further.

66
5.6 The need for transdisciplinarity in future cocoa research
Designing agroecosystems similar to natural ecosystems may be the only way to cultivate cocoa
sustainably (Milz, 2012). Ideally, the objective should be to optimize systems for productivity,
biodiversity, and food sovereignty (Patel, 2009) in the long term, rather than short-term maximization
of yield. However, the main constraints for large-scale adoption of approaches such as DAFS are their
knowledge- and labour-intensive nature. Not only do interested farmers need to understand the
underlying principles of these approaches, they also need technical support, as well as help in
establishing farmer-to-farmer knowledge and exchange networks, and fair prices for their produce.
Independent organizations such as Ecotop Consult12 are thus pivotal for initiating these processes on
the ground in order to stimulate bottom-up learning approaches.
While this might resolve the knowledge constraints, labour intensity remains a challenge. Smallholders
optimize opportunity costs and invest labour into different activities accordingly. Hence, the problem
is again on the market side, i.e., producers are paid low prices for their cocoa, which underpins yet
again the need to improve market conditions for produce from AFS. The establishment of a new label
and certification scheme for sound agroforestry is a long-term goal that can be interesting in this
respect. However, relocating the addition of value from consumer to producer countries is, arguably,
a much more powerful and swift way to empower local communities involved in cocoa production,
trade and processing. Novel Start-Ups such as fairafric13 are taking a flying leap to improve market
conditions for local cocoa producers.
One of the key factors future cocoa research needs to address is the design of sustainable production
systems. Social learning processes through transdisciplinary research are required to develop decision
support tools that facilitate the determination of adequate cocoa – tree species associations in order
to achieve effective regulation of pests and diseases while ensuring favourable growing conditions.
Similarly, the adoption of DAFS needs not only knowledge transfer, but social learning processes and
transdisciplinary approaches. Capacity building, knowledge exchange networks and continuous
technical support are important means of enhancing the adoption of DAFS, especially as socio-
economic factors (e.g. farmers’ age, share of on-/off-farm income, etc.) affect the success of farmers
on the ground, which, in turn, affects the adoption of DAFS.
As agroecosystems are complex, the above challenges need to be tackled with complexity, i.e.,
diversity. This refers not only to diversity in production, but also to the ways in which knowledge is
produced and shared between people and institutions. We as researchers need to reflect this diversity
and the processes involved in research, integrating social and natural sciences in the design of our
projects and taking into account different forms of knowledge while regarding phenomena from a
perspective that goes beyond specific disciplines and is based on broad participation (Hirsch-Hadorn
et al., 2006). Fortunately, a growing number of researchers seem to understand that their knowledge
is complementary rather than superior to that of farmers and other stakeholders.
Researchers need to pursue actively the path of transdisciplinary and participatory action research that
allows for the conservation and application of local knowledge, while enabling knowledge co-
production and mutual learning among farmers, researchers, and other stakeholders such as
consumers and policymakers. As Vaast and Somarriba (2014) pointed out: “The selection of tree
species and combinations is likely to be most effective where farmers participate, so that their goals

12
http://www.ecotop-consult.de/
13
https://fairafric.com/de/
67
and aspirations are taken into account and their local agroforestry knowledge is incorporated into the
design and management of the system.” The lack of such approaches in the Ghanaian cocoa sector
may explain why farmers have implemented available CSSVD prevention measures only to a limited
extent (Ameyaw et al., 2014). For example, farmers mentioned that they would not do barrier (strip)
cropping with oil palm because this attracts rodents, which can damage cocoa pods. Integrative
approaches are also more likely to help identify and implement strategies to adapt to multiple
stressors. Adaptation to climate change impacts, for example, means much more than identifying and
planting resistant crops (Pohl et al., 2010). The fact that innovation platforms were used to bring about
institutional change in Ghana’s cocoa sector (Adu-Acheampong et al., 2017) shows that researchers
are taking steps towards transdisciplinary and participatory action research, making research more
solution-oriented and relevant for the livelihoods of cocoa producers throughout the tropics.
Certification standards hold a certain potential to influence the future design of cocoa production
systems. While there is a need to better assess the long-term effects of implementing good practices
(including AF), developed by eco-certification schemes across a wide range of ecological and socio-
economical contexts (ICCO, 2014), certification bodies also need to be open to continuous adaptation
of their standards according to research results derived from projects with farmer involvement.
Organic certification alone, for instance, may not lead to a diversification of cocoa production systems
or the implementation of DAFS (Jacobi et al., 2013). Therefore, organic and other certification schemes
should explicitly emphasize the need to diversify in order to foster the resilience of cocoa production
systems to factors of global change. In addition, policymakers should address the costs associated with
certification schemes, as these may present a major constraint for farmers wishing to obtain
certification, and build incentives for organic and AF produce on the consumer side.
Sood and Mitchell (2006) found the attitude of farmers towards agroforestry systems to be the most
important factor of adoption, which highlights the importance of extension programmes for
knowledge sharing between agroforestry experts and farmers. Experience shows that especially in
areas where farmers face big challenges in their own plantations, the willingness to adopt agroforestry
or dynamic agroforestry is high (Andres et al, 2016). However, as the perceptions of farmers about
different production systems change according to their underlying motivation of engaging in them (e.g.
expected income or knowledge gain on management practices in AFS), the way experts interact with
farmers and farmer organizations (e.g. on the principles of DAFS) needs to be improved. Experiences
from Ivory Coast show that this basic principle is often not applied adequately (Andres et al, 2016). I
advocate for interactive knowledge sharing methods such as farmer field schools, which stimulate
farmers’ participation and give scientists the role of facilitators who integrate different forms of
knowledge and make them visible to different stakeholders (Pohl et al., 2010). Much can be learnt
from the Latin American agro-ecological movement, such as the farmer-to-farmer (“campesino a
campesino”) movement and its learning approaches in which researchers and external consultants are
facilitators rather than instructors (Holt-Giménez, 2006).

68
References
Abdi, H., 2010. Holm's Sequential Bonferroni Procedure. In: Salkind NJ (ed) Encyclopedia of Research
Design. Thousand Oaks, California. doi:10.4135/9781412961288
Abdulai, I., Vaast, P., Hoffmann, M.P., Asare, R., Jassogne, L., Van Asten, P., Rötter, R.P., Graefe, S.,
2017. Cocoa agroforestry is less resilient to sub-optimal and extreme climate than cocoa in
full sun. Global Change Biology.
Abou Rajab, Y., Leuschner, C., Barus, H., Tjoa, A., Hertel, D., 2016. Cacao Cultivation under Diverse
Shade Tree Cover Allows High Carbon Storage and Sequestration without Yield Losses. Plos
One 11.
Abrokwah, F., Dzahini-Obiatey, H., Galyuon, I., Osae-Awuku, F., Muller, E., 2016. Geographical
Distribution of Cacao swollen shoot virus Molecular Variability in Ghana. Plant Disease 100,
2011-2017.
Adegbola, M.O.K., no date. Effect of shading on cacao seedlings infected with the cocoa swollen-
shoot virus (CSSV). Annual Report Cocoa Research Institute of Nigeria (CRIN) Ibadan, Nigeria.
Adomako, D., Owusu, G.K., 1974. Studies on the mechanical transmission of cocoa swollen shoot
virus: Some factors affecting virus multiplication and symptom development in cocoa. Ghana
Journal of Agricultural Sciences 7, 7-15
Adu-Acheampong, R., Jiggins, J.L.S., Quartey, E.T., Karikari, N.M., Jonfia-Essien, W., Quarshie, E., Osei-
Fosu, P., Amuzu, M., Afari-Mintah, C., Ofori-Frimpong, K., Sakyi-dawson, O., 2017. An
innovation platform for institutional change in Ghana's cocoa sector. Cahiers d'Etudes et de
Recherches Francophones. Agricultures 26.
Aguilar-Gallegos, N., Muñoz-Rodríguez, M., Santoyo-Cortés, H., Aguilar-Ávila, J., Klerkx, L., 2015.
Information networks that generate economic value: A study on clusters of adopters of new
or improved technologies and practices among oil palm growers in Mexico. Agric Syst
135:122-132. doi:https://doi.org/10.1016/j.agsy.2015.01.003
Ahenkorah, Y., Akrofi, G.S., Adri, A.K., 1974. End of first cocoa shade and manurial experiment at
Cocoa Research Institute of Ghana. J. Horticul. Sci. Biotech. 49, 43-51.
Ahenkorah, Y., Halm, B.J., Appiah, M.R., Akrofi, G.S., Yirenkyi, J.E.K., 1987. 20 years results from a
shade and fertilizer trial on Amazon cocoa (Theobroma cacao) in Ghana. Exp. Agric. 23, 31-
39.
de Almeida, A.-A.F., Valle, R.R., 2007. Ecophysiology of the cacao tree. Braz. J. Plant Physiol. 19, 425-
448.
Allison, P., 2012. When Can You Safely Ignore Multicollinearity? Statistical Horizons. Available:
https://statisticalhorizons.com/multicollinearity. Accessed 27 February 2018.
Ameyaw, G.A., Dzahini-Obiatey, H.K., Domfeh, O., 2014. Perspectives on cocoa swollen shoot virus
disease (CSSVD) management in Ghana. Crop Prot. 65, 64-70.
Ameyaw, G.A., Domfeh, O., Dzahini-Obiatey, H., Ollennu, L.A.A., Owusu, G.K., 2016. Appraisal of
Cocoa swollen shoot virus (CSSV) Mild Isolates for Cross Protection of Cocoa Against Severe
Strains in Ghana. Plant Disease 100, 810-815.
Ampofo, S.T., 1997. The current cocoa swollen shoot virus disease situation in Ghana. Proceedings of
INCOPED 1st International Seminar on Cocoa Pests and Diseases, 1997, 175-178.
Analog Forestry Network (RIFA), 2012. Field Guide to Analog Forestry.
Andres, C., Bhullar, G.S., 2016. Sustainable intensification of tropical agro-ecosystems: Need and
potentials. Front. Environ. Sci. 4, 1-10.

69
Andres, C., Comoé, H., Beerli, A., Schneider, M., Rist, S., Jacobi, J., 2016. Cocoa in monoculture and
dynamic agroforestry, in: Lichtfouse, E. (Ed.), Sustainable Agriculture Reviews 19. Springer
International Publishing, Switzerland, pp. 121-153.
Andres, C., Gattinger, A., Dzahini-Obiatey, H.K., Blaser, W.J., Offei, S.K., Six, J., 2017. Combatting
cocoa swollen shoot virus disease: what do we know? Crop Prot. 98, 76-84.
Andres, C., Blaser, W.J., Dzahini-Obiatey, H.K., Ameyaw, G.A., Domfeh, O.K., Awiagah, M.A.,
Gattinger, A., Schneider, M., Offei, S.K., Six. J., 2018. Agroforestry systems can mitigate the
severity of cocoa swollen shoot virus disease. Agric Ecosyst Environ 252:83-92.
doi:https://doi.org/10.1016/j.agee.2017.09.031
Anim-Kwapong G. J. and Frimpong E. B. (2008). Vulnerability of agriculture to climate change –
impact of climate change on cocoa production. Cocoa Research Institute of Ghana, Tafo.
Armengot, L., Barbieri, P., Andres, C., Milz, J., Schneider, M., 2016. Cacao agroforestry systems have
higher return on labor compared to full-sun monocultures. Agron. Sustain. Dev. 36, 70.
Asamoah, M., Amoah, F., 2015. Microcredit Schemes: A Tool for Promoting Rural Savings Capacity
among Poor Farm Families: A Case Study in the Eastern Region of Ghana. Open Journal of
Social Sciences, 3, 24-30. doi: 10.4236/jss.2015.31003.
Asare, R., 2006. A review on cocoa agroforestry as a means for biodiversity conservation. World
Cocoa Foundation Partnership Conference, Brussels, Belgium.
Asomaning, E.J.A., Kwakwa, R.S., Hutcheon, W.V., 1971. Physiological studies on an Amazon shade
and fertilizer trial at the Cocoa Research Institute of Ghana. Ghana. J. Agric. Sci. 4, 47-64.
Awudzi, G.K., 2014. Development of an information system for mirid attack on cocoa based on the
crop status and mirid population monitoring. The School of Agriculture, Policy and
Development, University of Reading, Reading UK, PhD thesis: 209.
Ayenor, G.K., Roling, N.G., Padi, B., Van Huis, A., Obeng-Ofori, D., Atengdem, P.B., 2004. Converging
farmers' and scientists' perspectives on researchable constraints on organic cocoa
production in Ghana: Results of a diagnostic study. NJAS – Wageningen Journal of Life
Sciences 52, 261-284.
Baah, F., 2008. Cocoa Farmer Characteristics and Access to Research-based Information in Two
Districts of Ashanti, Ghana. J Sci Technol (Ghana) 28 (3):10-18.
doi:http://dx.doi.org/10.4314/just.v28i3.33095
Babin R., Ten Hoopen G.M., Cilas C., Enjalric F., Yédé, Gendre P., Lumaret J.P., 2010. Impact of shade
on the spatial distribution of Sahlbergella singularis in traditional cocoa agroforests. Agric.
For. Entomol. 12, 69-79.
Bai, S.H., Trueman, S.J., Nevenimo, T., Hannet, G., Bapiwai, P., Poienou, M., Wallace, H.M., 2017.
Effects of shade-tree species and spacing on soil and leaf nutrient concentrations in cocoa
plantations at 8 years after establishment. Agriculture Ecosystems & Environment 246, 134-
143.
Bates, D., Maechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using lme4. J.
Stat. Softw. 67, 1-48.
Bedimo, J.A.M., Dufour, B.P., Cilas, C., Avelino, J., 2012. Effects of shade trees on Coffea Arabica pests
and diseases. Cahiers Agricultures 21, 89-97.
Beer, J., 1987. Advantages, disadvantages and desirable characteristics of shade trees for coffee,
cacao and tea. Agrofor. Syst. 5, 3-13.
Beer, J., Muschler, R., Kass, D., Somarriba, E., 1998. Shade management in coffee and cacao
plantations. Agrofor. Syst. 38, 139–164.
70
Beer, J., Ibrahim, M., Somarriba, E., Barrance, A., Leakey, R., 2004. Establecimiento y manejo de
árboles en sistemas agroforestales, in: Cordero, J., Boshier, D. (Eds.), Árboles de
Centroamérica: un manual para extensionistas. Oxford Forestry Institute (OFI), Centro
Agronómico Tropical de Investigación y Enseñanza (CATIE). Oxford, UK, Costa Rica, pp. 197–
242.
Bentley, J.W., Boa, E., Stonehouse, J., 2004. Neighbor trees: Shade, intercropping, and cacao in
Ecuador. Human Ecology, 32, 241-270.
Bieng, M.A.N., Gidoin, C., Avelino, J., Cilas, C., Deheuvels, O., Wery, J., 2013. Diversity and spatial
clustering of shade trees affect cacao yield and pathogen pressure in Costa Rican agroforests.
Basic Appl. Ecol. 14, 329-336.
Bigger, M., 1981. Observations on the insect fauna of shaded and unshaded Amelonado cocoa. Bull.
Entomol. Res. 71, 107-119.
Bisseleua, H.B.D., Fotio, D., Yede, Missoup, A.D., Vidal, S., 2013. Shade Tree Diversity, Cocoa Pest
Damage, Yield Compensating Inputs and Farmers' Net Returns in West Africa. Plos One, 8.
Blaser, W.J., Oppong, J., Yeboah, E., Six, J., 2017. Shade trees have limited benefits for soil fertility in
cocoa agroforests. Agric. Ecosyst. Environ. 243, 83-91.
Borenstein, M., Hedges, L.V., Higgins, J.P.T., Rothstein, H.R., 2010. A basic introduction to fixed-effect
and random-effects models for meta-analysis. Research Synthesis Methods 1, 97-111.
Bos, M.M., Steffan-Dewenter, I., Tscharntke, T., 2007. Shade tree management affects fruit abortion,
insect pests and pathogens of cacao. Agric. Ecosyst. Environ. 120, 201-205.
Box, H.E., 1945. Insect transmission of the swollen-shoot virus in West African cacao. Nature 155,
608-609.
Brondizio, E.S., Moran, E.F., 2008. Human dimensions of climate change: the vulnerability of small
farmers in the Amazon. Philos Trans R Soc B – Biol Sci 363 (1498):1803-1809.
doi:10.1098/rstb.2007.0025
Buresh, R.J., Rowe, E.C., Livesley, S.J., Cadisch, G., Mafongoya, P., 2004. Opportunities for capture of
deep soil nutrients, in: van Noordwijk, M., Cadisch, G., Ong, C.K. (Eds.) Below-ground
Interactions in Tropical Agroecosystems: Concepts and Models with Multiple Plant
Components. CABI Publishing, Wallingford, pp. 109-123.
Campbell, C.A.M., 1990. The susceptibility of cocoa to mealybugs (Pseudococcidae) and other
honeydew-producing Homoptera in Ghana. Bull. Entomol. Res. 80, 137-151.
Carr, M.K.V., Lockwood, G., 2011. The water relations and irrigation requirements of cocoa
(Theobroma cacao L.): a review. Exp. Agric. 47, 653-676.
Cerda, R., Deheuvels, O., Calvache, D., Niehaus, L., Saenz, Y., Kent, J., Vilchez, S., Villota, A., Martinez,
C., Somarriba, E., 2014. Contribution of cocoa agroforestry systems to family income and
domestic consumption: looking toward intensification. Agroforestry Systems, 88, 957-981.
Clientearth, 2013. Understanding the legality of rights, permits and certificates to harvest naturally
occurring timber in Ghana. Clientearth, London. Available:
https://www.documents.clientearth.org/wp-content/uploads/library/2013-04-01-
understanding-the-legality-of-rights-permits-and-certificates-to-harvest-naturally-occuring-
timber-in-ghana-ce-en.pdf. Accessed 27 February 2018.

71
Clough, Y., Abrahamczyk, S., Adams, M.-O., Anshary, A., Ariyanti, N., Betz, L., Buchori, D., Cicuzza, D.,
Darras, K., Putra, D.D., Fiala, B., Gradstein, S.R., Kessler, M., Klein, A.-M., Pitopang, R., Sahari,
B., Scherber, C., Schulze, C.H., Shahabuddin, Sporn, S., Stenchly, K., Tjitrosoedirdjo, S.S.,
Wanger, T.C., Weist, M., Wielgoss, A., Tscharntke, T., 2010. Biodiversity patterns and trophic
interactions in human-dominated tropical landscapes in Sulawesi (Indonesia): plants,
arthropods and vertebrates. In: Tscharntke, T., Leuschner, C., Veldkamp, E., Faust, H.,
Guhardja, E., Bidin, A. (Eds.) Tropical Rainforests and Agroforests under Global Change:
Ecological and Socio-Economic Valuations.
Clough, Y., Barkmann, J., Juhrbandt, J., Kessler, M., Wanger, T.C., Anshary, A., Buchori, D., Cicuzza, D.,
Darras, K., Putra, D.D., Erasmi, S., Pitopang, R., Schmidt, C., Schulze, C.H., Seidel, D., Steffan-
Dewenter, I., Stenchly, K., Vidal, S., Weist, M., Wielgoss, A.C., Tscharntke, T., 2011.
Combining high biodiversity with high yields in tropical agroforests. Proceedings of the
National Academy of Sciences of the United States of America, 108, 8311-8316.
Clough, Y., Faust, H., Tscharntke, T., 2009a. Cacao boom and bust: sustainability of agroforests and
opportunities for biodiversity conservation. Conservation Letters, 2, 197-205.
Clough, Y., Putra, D.D., Pitopang, R., Tscharntke, T., 2009b. Local and landscape factors determine
functional bird diversity in Indonesian cacao agroforestry. Biological Conservation, 142, 1032-
1041.
Countrymeters, 2017. Ghana Population. Available: http://countrymeters.info/en/Ghana. Accessed
27 February 2018.
Cornwell, P.B., 1953. Mealybug population, distribution and migration. Proceedings of the West
African International Cacao Research Conference, 1953, West African Cacao Research
Institute, Tafo, Gold Coast, 8-17.
Cornwell, P.B., 1958. Movements of the vectors of virus diseases of cacao in Ghana. I. – Canopy
movement in and between trees. West African Cocoa Research Institute, Tafo, Ghana.
Cornwell, P.B., 1960. Movements of the vectors of virus diseases of cacao in Ghana. II. – Wind
movements and aerial dispersal. West African Cocoa Research Institute, Tafo, Ghana.
Costanza, R., 2012. Ecosystem health and ecological engineering. Ecological Engineering 45, 24-29.
Cunningham, R.K., Burridge, J.C., 1960. The growth of cacao (Theobroma cacao) with and without
shade. Ann. Bot. 24, 458-462.
Danso-Abbeam, G., 2014. Modelling farmers investment in agrochemicals. The experience of
smallholder cocoa farmers in Ghana. Res. Appl. Econ. 6, 1-16.
Davis, F.D., 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information
Technology. MIS Q 13 (3):319-340. doi:10.2307/249008
Dawoe, E.K., Quashie-Sam, J.S., Oppong, S.K., 2014. Effect of land-use conversion from forest to
cocoa agroforest on soil characteristics and quality of a Ferric Lixisol in lowland humid Ghana.
Agroforestry Systems, 88, 87-99.
De Beenhouwer, M., Aerts, R., Honnay, O., 2013. A global meta-analysis of the biodiversity and
ecosystem service benefits of coffee and cacao agroforestry. Agriculture Ecosystems &
Environment, 175, 1-7.
Deppeler, A., 2014. Who benefits from certification? Analysis of third-party audited certification
schemes in the cocoa sector in Ghana. MSc thesis, HAFL, Zollikofen, Switzerland.

72
Domfeh, O., Dzahini-Obiatey, H.K., Ameyaw, G.A., Abaka-Ewusie, K., Opoku, G., 2011. Cocoa swollen
shoot virus disease situation in Ghana: A review of current trends. African Journal of
Agricultural Research 6, 5033-5039.
Domfeh, O., Ameyaw, G.A., Dzahini-Obiatey, H.K., Ollennu, L.A.A., Osei-Bonsu, K., Acheampong, K.,
Aneani, F., Owusu-Ansah, F., 2016. Use of immune crops as barrier in the management of
cacao swollen shoot virus disease (CSSVD) – long-term assessment. Plant Dis. 100, 1889-
1893.
Dormon, E.N.A., Huis, A.V., Leeuwis, C., Obeng-Ofori, D., Sakyi-Dawson, O., 2004. Causes of low
productivity of cocoa in Ghana: farmers' perspectives and insights from research and the
socio-political establishment. NJAS Wageningen Journal of Life Sciences 52, 237-259.
Duguma, B., Gockowski, J., Bakala, J., 2001. Smallholder Cacao (Theobroma cacao Linn.) cultivation in
agroforestry systems of West and Central Africa: challenges and opportunities. Agroforestry
Systems, 51, 177-188.
Dzahini-Obiatey, H.K., Aculey, P.C., Ollennu, L.A., 2005. Inhibition of cocoa swollen shoot Badnavirus
(CSSV) by an antiviral substance. Proceedings of the 14th International Cocoa Research
Conference, 2003, Accra, Ghana, 1365-1371.
Dzahini-Obiatey, H.K., Ameyaw, G.A., Ollennu, L.A., 2006. Control of cocoa swollen shoot disease by
eradicating infected trees in Ghana: A survey of treated and replanted areas. Crop Prot. 25,
647-652.
Dzahini-Obiatey, H.K., Domfeh, O., Amoah, F.M., 2010. Over seventy years of a viral disease of cocoa
in ghana: from researchers' perspective. Afr. J. Agric. Res. 5, 476-485.
Eliot & Associates, 2005. Guidelines for Conducting a Focus Group. Massachusetts Technical
Assistance Partnership for Prevention (MassTAPP). Available:
http://masstapp.edc.org/guidelines-conducting-focus-group. Accessed 27 February 2018.
Faget, M., Liedgens, M., Feil, B., Stamp, P., Herrera, J.M., 2012. Root growth of maize in an Italian
ryegrass living mulch studied with a non-destructive method. European Journal of Agronomy
36, 1-8.
FAO, 2018. FAOSTAT Database on Agriculture Available: http://faostat.fao.org. Accessed 27 February
2018.
Firempong, S., 1984. Laboratory and field evaluation of cocoa progenies for resistance to mealybug
vectors (Hemiptera, Pseudococcidae) of swollen shoot virus. Bull. Entomol. Res. 74, 97-102.
Fonte, S.J., Six, J., 2010. Earthworms and litter management contributions to ecosystem services in a
tropical agroforestry system. Ecol. Appl. 20, 1061-1073.
Fonte, S.J., Barrios, E., Six, J., 2010a. Earthworm impacts on soil organic matter and fertilizer
dynamics in tropical hillside agroecosystems of Honduras. Pedobiologia 53, 327-335.
Fonte, S.J., Barrios, E., Six, J., 2010b. Earthworms, soil fertility and aggregate-associated soil organic
matter dynamics in the Quesungual agroforestry system. Geoderma 155, 320-328.
Forster, D., Adamtey, N., Messmer, M.M., Pfiffner, L., Baker, B., Huber, B., Niggli, U., 2013. Organic
agriculture - Driving innovations in crop research, In: Bhullar, G.S., Bhullar, N.K. (Eds.),
Agricultural Sustainability. Academic Press, San Diego, 21-46.
Franzen, M., Mulder, M.B., 2007. Ecological, economic and social perspectives on cocoa production
worldwide. Biodiversity and Conservation 16, 3835-3849.
Galyuon, I.K.A., McDavid, C.R., Lopez, F.B., Spence, J.A., 1996. The effect of irradiance level on cocoa
(Theobroma cacao L.). 2. Gas exchange and chlorophyll fluorescence. Trop. Agric. 73, 29-33.

73
Gama-Rodrigues, A.C., 2011. Soil organic matter, nutrient cycling and biological dinitrogen-fixation in
agroforestry systems. Agrofor. Syst. 81, 191-193.
Garrity, D.P., 2004. Agroforestry and the achievement of the millennium development goals.
Agroforestry Systems, 61-2, 5-17.
Ghana Statistical Service, 2014. District Analytical Report - New Juaben Municipal, Population &
Housing Census 2010. Available:
http://www.statsghana.gov.gh/docfiles/2010_District_Report/Eastern/NEW%20JUABEN.pdf.
Accessed 27 February 2018.
Gidoin, C., Avelino, J., Deheuvels, O., Cilas, C., Bieng, M.A.N., 2014. Shade tree spatial structure and
pod production explain frosty pod rot intensity in cacao agroforests, Costa Rica. Phytopathol.
104, 275-281.
Glin, L.C., Oosterveer, P., Mol, A.P.J., 2015. Governing the Organic Cocoa Network from Ghana:
Towards Hybrid Governance Arrangements? Journal of Agrarian Change 15, 43-64.
Gockowski, J., Afari-Sefa, V., Sarpong, D.B., Osei-Asare, Y.B., Dziwornu, A.K., 2011. Increasing Income
of Ghanaian Cocoa Farmers: Is Introduction of Fine Flavour Cocoa a Viable Alternative?
Journal of International Agriculture 50, 175-200.
Gockowski, J., Sonwa, D., 2011. Cocoa intensification scenarios and their predicted impact on CO(2)
emissions, biodiversity conservation, and rural livelihoods in the Guinea rain forest of West
Africa. Environ. Manag. 48, 307-321.
Goetsch, E., 1992. Natural succession of species in agroforestry and in soil recovery. Fazenda Três
Colinas, Agrosilvicultura Ltda., 45436 Pirai do Norte, Bahia, Brazil. Available:
www.agrofloresta.net/static/artigos/agroforestry_1992_gotsch.pdf. Accessed 27 February
2018.
Götsch, E., 1994. Break-through in agriculture. Agricultura familiar e agroecologia (ASPTA), Rio de
Janeiro.
Graefe, S., Meyer-Sand, L.F., Chauvette, K., Abdulai, I., Jassogne, L., Vaast, P., Asare, R., 2017.
Evaluating Farmers' Knowledge of Shade Trees in Different Cocoa Agro-Ecological Zones in
Ghana. Human Ecology 45, 321-332.
Gram, G., Vaast, P., van der Wolf, J., Jassogne, L., 2017. Local tree knowledge can fast-track
agroforestry recommendations for coffee smallholders along a climate gradient in Mount
Elgon, Uganda. Agroforestry Systems.
Gruberg, H., 2011. Sostenibilidad de la Agroforestería Sucesional en Bolivia - Una evaluación
económica, sociocultural y ecológica en tres estudios de caso en la zona del Alto Beni,
Editorial Académica Española.
Gurumurthy, A., 2006. Promoting gender equality? Some development-related uses of ICTs by
women. Development in Practice, 16(6), 611-616.
Hanna, A.D., Heatherington, W., Judenko, E., 1952. Control of the mealybug vectors of the swollen
shoot virus by a systemic insecticide. Nature 169, 334-335.
Hainmueller, J., Hiscox, M.J., Tampe, M., 2011. Sustainable Development for Cocoa Farmers in
Ghana. MIT and Harvard University, Cambridge, MA. Available: https://www.theigc.org/wp-
content/uploads/2015/02/Hainmueller-Et-Al-2011-Working-Paper.pdf. Accessed 27 February
2018.
Hawthorne, W., Gyakari, N., 2006. Photoguide for the forest trees of Ghana: a tree-spotter's field
guide for identifying the largest trees. Oxford Forestry Institute, Oxford, UK.

74
Hirsch-Hadorn, G., Bradley, D., Pohl, C., Rist, S., Wiesmann, U., 2006. Implications of
transdisciplinarity for sustainability research. Ecological Economics, 60, 119-128.
Hoerler, R., 2017. What are the main constraints, and how to enhance adoption of recommended
CSSVD prevention measures by cocoa farmers in Ghana? MSc thesis, ETH Zurich, Switzerland.
Holt-Giménez, E., 2006. Campesino a Campesino: Voices from Latin America's Farmer to Farmer
Movement for Sustainable Agriculture, Food First Books.
Hütz-Adams, F., 2011. Ghana: Vom bitteren Kakao zur süßen Schokolade. SÜDWIND e. V., SÜDWIND
e.V., Institut für Ökonomie und Ökumene, Siegburg.
ICCO, 2012. Global Cocoa Agenda. World Cocoa Conference 2012 Abidjan. Available:
https://www.icco.org/about-us/international-cocoa-agreements/doc_download/307-global-
cocoa-agenda-english.html. Accessed 27 February 2018.
ICCO, 2014. Zurich Certification Workshop finds common ground. Available:
http://www.icco.org/about-us/icco-news/253-zurich-certification-workshop-finds-common-
ground.html. Accessed 27 February 2018.
Isaac, M., Ulzen-Appiah, F., Timmer, V., Quashie-Sam, S., 2007. Early growth and nutritional response
to resource competition in cocoa-shade intercropped systems. Plant Soil 298, 243-254.
Jacobi, J., Andres, C., Schneider, M., Pillco, M., Calizaya, P., Rist, S., 2014. Carbon stocks, tree
diversity, and the role of organic certification in different cocoa production systems in Alto
Beni, Bolivia. Agrofor. Syst. 88, 1117-1132.
Jacobi, J., Schneider, M., Bottazzi, P., Pillco, M., Calizaya, P., Rist, S., 2015. Agroecosystem resilience
and farmers’ perceptions of climate change impacts on cocoa farms in Alto Beni, Bolivia.
Renewable Agriculture and Food Systems 30, 170-183.
Jagoret, P., Kwesseu, J., Messie, C., Michel-Dounias, I., Malézieux, E., 2014. Farmers’ assessment of
the use value of agrobiodiversity in complex cocoa agroforestry systems in central
Cameroon. Agroforestry Systems, 88, 983-1000.
Jaramillo, J., Chabi-Olaye, A., Kamonjo, C., Jaramillo, A., Vega, F.E., Poehling, H.-M., Borgemeister, C.,
2009. Thermal tolerance of the coffee berry borer Hypothenemus hampei: predictions of
climate change impact on a tropical insect pest. Plos One 4.
Jeger, M.J., Thresh, J.M., 1993. Modeling reinfection of replanted cocoa by swollen shoot virus in
pandemically diseased areas. J. Appl. Ecol. 30, 187-196.
Jose, S., 2009. Agroforestry for ecosystem services and environmental benefits: an overview.
Agroforestry Systems 76, 1-10.
Kenten, R.H., Lockwood, G., 1977. Studies on possibility of increasing resistance to cocoa swollen-
shoot virus by breeding. Ann. Appl. Biol. 85, 71-78.
Kindt , R., Coe, R., 2005. Tree diversity analysis. A manual and software for common statistical
methods for ecological and biodiversity studies. World Agroforestry Centre (ICRAF), Nairobi.
Kooijmans, L., 2016. Space for deliberation in Ghana's cocoa sector. MSc thesis, University of
Amsterdam. Available: https://inclusivevcc.files.wordpress.com/2015/07/ids_thesis_-
_space_for_deliberation_in_ghanas_cocoa_sector_-_lars_kooijmans.pdf. Accessed 27
February 2018.
Kuonen, L., 2016. How do land tenure agreements influence the adoption of research based
recommendations to control Cocoa Swollen Shoot Virus Disease (CSSVD) in Ghana? Term
paper, HAFL Zollikofen, Switzerland.

75
Laforest, J., 2009. Safety Diagnosis Tool Kit for Local Communities. Guide to Organizing Semi-
Structured Interviews With Key Informants. Institut national de santé publique du Québec,
Québec. Available:
https://www.inspq.qc.ca/sites/default/files/publications/1437_guideorgaentretienssemidirig
informcles2eed_va.pdf. Accessed 27 February 2018.
Lambrecht, I., Asare, S., 2015. Smallholders and land tenure in Ghana: Aligning context, empirics, and
policy. IFPRI Discussion Paper 1492 Washington, DC: International Food Policy Research
Institute (IFPRI). Available: http://www.ifpri.org/publication/smallholders-and-land-tenure-
ghana-aligning-context-empirics-and-policy. Accessed 27 February 2018.
Legg, J.T., 1982. The cocoa swollen shoot research project at the Cocoa Research Institute, Tafo,
Ghana, 1969-78. Overseas Development Administration, London.
Leiter, J., Harding, S., 2004. Trinidad, Brazil, and Ghana: three melting moments in the history of
cocoa. Journal of Rural Studies 20, 113-130.
Lichtenthaler, H.K., Burkart, S., 1999. Photosynthesis and high light stress. Bulg. J. Plant Physiol. 25,
3–16.
Lin, B.B., 2011. Resilience in agriculture through crop diversification: Adaptive management for
environmental change. Bioscience 61, 183-193.
Lockwood, G., 1981a. Genetic aspects of resistance to cocoa swollen-shoot virus in Ghana. Ann. Appl.
Biol. 98, 131-141.
Lockwood, G., 1981b. Resistance of cocoa to swollen-shoot virus in Ghana. II. Assessment of seed
inoculation experiments. Ann. Appl. Biol. 97, 91-97.
Lot, H., Djiekpor, E., Jacquemond, M., 1991. Characterization of the genome of cacao swollen shoot
virus. J. Gen. Virol. 72, 1735-1739.
Lundstedt, H., Pärssinen, S., 2009. Cocoa is Ghana, Ghana is Cocoa - Evaluating Reforms of the
Ghanaian Cocoa Sector. MSc thesis, University of Lund. Available:
http://lup.lub.lu.se/student-papers/record/1511300/file/1647134.pdf. Accessed 27 February
2018.
Mani, M., Shivaraju, C., 2016. Mealybugs and their management in agricultural and horticultural
crops. Springer India, New Delhi.
Mbow, C., Smith, P., Skole, D., Duguma, L., Bustamante, M., 2014. Achieving mitigation and
adaptation to climate change through sustainable agroforestry practices in Africa. Current
Opinion in Environmental Sustainability 6, 8-14.
Milz, J., 1997. Guía para el Establecimiento de Sistemas Agroforestales. DED Servicio Alemán de
Cooperación Social-Técnica, La Paz.
Milz, J., 2010. Producción de Naranja (Citrus sinensis) en sistemas agroforestales sucesionales en Alto
Beni, Bolivia - Estudio de caso. In: Beck, S. (Ed.) Biodiversidad y Ecología en Bolivia. La Paz,
Bolivia: Instituto de Ecologia, Universidad Mayor de San Andrés (UMSA).
Milz, J., 2012. The gloomy outlook for cocoa production in The Ivory Coast and strategies for
sustainable solutions for recovery and improvements of productivity. La Paz: Ecotop Consult.
Mohammed, A.M., Robinson, J.S., Midmore, D., Verhoef, A., 2016. Carbon storage in Ghanaian cocoa
ecosystems. Carbon Balance Manag. 11, 1-6.

76
Monastyrnaya, E., Joerin, J., Dawoe, E., Six, J., 2016. Assessing the resilience of the cocoa value chain
in Ghana. Swiss Federal Institute of Technology (ETH), Zurich & Kwame Nkrumah University
of Science and Technology, Kumasi. Available:
https://www.ethz.ch/content/dam/ethz/special-interest/usys/ias/enhancing-resilience-
dam/documents/Final%20Report.%20Cocoa%20in%20Ghana.pdf. Accessed 27 February
2018.
Moury, B., Fabre, F., Montarry, J., Janzac, B., Ayme, V., Palloix, A., 2010. The adaptation of plant
viruses to varietal resistances. Virologie 14, 227-239.
Muilerman, S., 2013. Occupational Safety and Health on Ghanaian cocoa farms. Baseline report.
Sustainable Tree Crops Program, International Institute of Tropical Agriculture (IITA), Accra,
Ghana. Available: http://www.cocoainitiative.org/wp-
content/uploads/2017/10/Occupational-safety-and-health-on-Ghanaian-cocoa-farms.pdf.
Accessed 27 February 2018.
Muller, E., Jacquot, E., Yot, P., 2001. Early detection of cacao swollen shoot virus using the
polymerase chain reaction. J. Virol. Methods 93, 15-22.
Muller, E., Sackey, S., 2005. Molecular variability analysis of five new complete cacao swollen shoot
virus genomic sequences. Arch. Virol 150, 53-66.
Muller, E., 2016. Cacao swollen shoot virus (CSSV): History, biology, and genome, In: Bailey, B.A.,
Meinhard, L.W. (Eds.), Cacao Diseases - A History of Old Enemies and New Encounters.
Springer International Publishing, Switzerland.
Muller, E., Ravel, S., Agret, C., Abrokwah, F., Dzahini-Obiatey, H., Galyuon, I., Kouakou, K., Jeyaseelan,
E.C., Allainguillaume, J., Wetten, A., 2018. Next generation sequencing elucidates cacao
badnavirus diversity and reveals the existence of more than ten viral species. Virus Research
244, 235-251.
National Agroforestry Center, 2017. Agroforestry Practices. United States Department of Agriculture
(USDA), National Agroforestry Center (NAC). Available:
https://www.fs.usda.gov/nac/practices/index.shtml. Accessed 27 February 2018.
Nakagawa, S., Schielzeth, H., 2013. A general and simple method for obtaining R2 from generalized
linear mixed-effects models. Methods Ecol. Evol. 4, 133-142.
Nicol, J., 1952. Systemic insecticides and the mealybug vectors of swollen-shoot virus of cacao.
Nature 169, 120-120.
Obiri, B.D., Bright, G.A., McDonald, M.A., Anglaaere, l.C.N., Cobbina, J., 2007. Financial analysis of
shaded cocoa in Ghana. Agroforestry Systems, 71, 139-149.
Odiaka, E. (2015). Perception of the influence of home videos on youth farmers in Makurdi, Nigeria.
Journal of Agricultural & Food Information, 16(4), 337-346.
Ollennu, L.A., Owusu, G.K., Thresh, J.M., 1989a. The control of cocoa swollen shoot disease in Ghana.
Cocoa Growers Bulletin 42, 25-35.
Ollennu, L.A.A., Owusu, G.K., Thresh, J.M., 1989b. Spread of cocoa swollen shoot virus to recent
plantings in Ghana. Crop Protection 8, 251-264.
Ollennu, L.A., Owusu, G.K., Dzahini-Obiatey, H.K., 1999. Recent studies of mild strain cross-protection
with cocoa swollen shoot virus. J. Ghana Sci. Assoc. 2, 5-9.
Ollennu, L.A., Osei Bonsu, K., Acheampong, K., Aneani, F., Ackonor, J.B., 2003. The use of immune
crops as barrier between cocoa plantings. Progress Report 2002-2003, Cocoa Research
Institute of Ghana, 180-181.

77
Ollennu, L.A., Osei-Bonsu, K., Aneani, F., Acheampong, K., 2005. Preliminary studies of the control of
cocoa swollen shoot disease by the use of immune crops as barrier. Proceedings of the 14th
International Cocoa Research Conference, 2003, Accra, Ghana, 839-843.
Opoku, J., 2012. Why Ghana is cocoa and cocoa is Ghana. Available:
https://www.modernghana.com/news/398737/1/why-ghana-is-cocoa-and-cocoa-is-
ghana.html. Accessed 27 February 2018.
Opoku-Ameyaw, K., Baah, F., Gyedu-Akoto, E., Anchirinah, V., Dzahini-Obiatey, H.K., Cudjoe, A.R.,
Aquaye, S., Opoku, S.Y., 2010. Cocoa Manual - A source book for sustainable cocoa
production. Cocoa Research Institute of Ghana, New Tafo-Akim.
Opoku, I.Y., Akrofi, A.Y., Appiah, A.A., 2002. Shade trees are alternative hosts of the cocoa pathogen
Phytophthora megakarya. Crop Prot. 21, 629-634.
Oro, F., Mississo, E., Okassa, M., Guilhaumon, C., Fenouillet, C., Cilas, C., Muller, E., 2012.
Geographical differentiation of the molecular diversity of cacao swollen shoot virus in Togo.
Arch. Virol 157, 509-514.
Oro, Z.F., Bonnot, F., Ngo Bieng, M.A., Delaitre, E., Dufour, B.P., Ametefe, K.E., Mississo, E., Wegbe,
K., Muller, E., Cilas, C., 2012. Spatiotemporal pattern analysis of Cacao swollen shoot virus in
experimental plots in Togo. Plant Pathol. 61, 1043-1051.
Owusu, J.K., 1980. Light requirements of cocoa: a review. Proceedings of the International
Conference on Cocoa and Coconuts 1978, Kuala Lumpur, Malaysia, 112-121.
Padi, F.K., 1997. Prospects for the control of cocoa mealybugs. Proceedings of INCOPED 1st
International Seminar on Cocoa Pests and Diseases, 1997, 249-263.
Padi, F.K., Domfeh, O., Takrama, J., Opoku, S., 2013. An evaluation of gains in breeding for resistance
to the cocoa swollen shoot virus disease in Ghana. Crop Prot. 51, 24-31.
Paine, J., 1945. Report of Agronomy Division, West African Cocoa Research Institute, Tafo, Ghana.
Patel, R., 2009. Food sovereignty. The Journal of Peasant Studies 36, 663-706.
Perfecto, I., Vandermeer, J., 2010. The agroecological matrix as alternative to the land-
sparing/agriculture intensification model. Proceedings of the National Academy of Sciences
of the United States of America 107, 5786-5791.
Plantwise, 2016. West African cocoa mealybug (Planococcoides njalensis), in: Bank, P.K. (Ed.),
Plantwise Technical Factsheet.
Pohl, C., Rist, S., Zimmermann, A., Fry, P., Gurung, G.S., Schneider, F., Speranza, C.I., Kiteme, B.,
Boillat, S., Serrano, E., Hadorn, G.H., Wiesmann, U., 2010. Researchers' roles in knowledge
co-production: experience from sustainability research in Kenya, Switzerland, Bolivia and
Nepal. Science and Public Policy, 37, 267-281.
Posnette, A.F., Robertson, N.F., 1950. Virus diseases of cacao in West Africa. VI. Vector investigations.
Annals of Applied Biology 37, 363-377.
Posnette, A.F., Robertson, N.F., Todd, J.M., 1950. Virus diseases of cacao in West Africa. V.
Alternative host plants. Ann. Appl. Biol. 37, 229-240.
Posnette, A.F., 1981. Viruses and resistance to virus diseases in cocoa. Proceedings of the 6th
International Cocoa Research Conference, 1977, Caracas, Venezuela, 262-266.
Purseglove, J., 1968. Tropical Crops: Dicotyledons. Harlow, UK: Longman.
Quainoo, A.K., Wetten, A.C., Allainguillaume, J., 2008. The effectiveness of somatic embryogenesis in
eliminating the cocoa swollen shoot virus from infected cocoa trees. J. Virol. Methods 149,
91-96.

78
R Core Team, 2017. R: A Language and Environment for Statistical Computing. R Foundation for
Statistical Computing, Vienna. Available: https://www.r-project.org. Accessed 27 February
2018.
Raja Harun, R.M., Hardwick, K., 1988. The effects of prolonged exposure to different light intensities
on the photosynthesis of cocoa leaves. Proceedings of the tenth International Cocoa
Research Conference 1988, Santo Domingo, Dominican Republic, 205-209.
Rice, R.A., Greenberg, R., 2000. Cacao cultivation and the conservation of biological diversity. Ambio
29, 167-173.
Roivainen, O., 1971. Circulative transmission of cocoa swollen shoot virus by the mealybug
Planococcoides njalensis. Proceedings of the 3rd International Cocoa Research Conference,
1969, Accra, Ghana, 518-521.
Roivainen, O., 1980. Mealybugs. In: Harris, K.F., Maramorosch, K. (Eds.) Vectors of Plant Pathogens.
Academic Press, New York, NY, pp. 15–38.
Rosalien, E.J., Pita, A.V., Maria, J.S., Rene, G.A.B., 2017. Shaded Coffee and Cocoa - Double Dividend
for Biodiversity and Small-scale Farmers. Ecological Economics 140, 136-145.
Ruben, R., Zuniga, G., 2011. How standards compete: comparative impact of coffee certification
schemes in Northern Nicaragua. Supply Chain Manag – Int J 16 (2):98-109.
doi:10.1108/13598541111115356
Ruf, F.O., 2011. The Myth of Complex Cocoa Agroforests: The Case of Ghana. Human Ecology 39, 373-
388.
Ruf, F., Zadi, H., 1998. Cocoa: From Deforestation to Reforestation. CIRAD. Paper prepared for the
Smithsonian Sustainable Cocoa Congress Panama, 1998. Available:
http://nationalzoo.si.edu/scbi/migratorybirds/research/cacao/ruf.cfm. Accessed 27 February
2018.
Saj, S., Jagoret, P., Ngogue, H.T., 2013. Carbon storage and density dynamics of associated trees in
three contrasting Theobroma cacao agroforests of Central Cameroon. Agroforestry Systems,
87, 1309-1320.
SAN, 2005. Additional criteria and indicators for cocoa production. Sustainable Agriculture Network,
Rainforest Alliance. Available:
http://www.worldagroforestry.org/treesandmarkets/inaforesta/documents/agrof_cons_bio
div/criteria%20and%20indicators%20for%20cocoa_2005.pdf. Accessed 27 February 2018.
Savary, S., Ficke, A., Aubertot, J.N., Hollier, C., 2012. Crop losses due to diseases and their
implications for global food production losses and food security. Food Secur. 4, 519-537.
Schnatmann, A., 2006. Análisis Económico de Rentabilidad - Parcelas Agroforestales Multiestrato.
Interinstitucional Alto Beni.
Schneider, M., Andres, C., Trujillo, G., Alcon, F., Amurrio, P., Perez, E., Weibel, F., Milz, J., 2016. Cocoa
and total system yields of organic and conventional agroforestry vs. monoculture systems in
a long-term field trial in Bolivia. Exp. Agric. 53, 351-374.
Schroth, G., Krauss, U., Gasparotto, L., Aguilar, J.A.D., Vohland, K., 2000a. Pests and diseases in
agroforestry systems of the humid tropics. Agroforestry Systems 50, 199-241.
Schroth, G., Teixeira, W.G., Seixas, R., Ferreira da Silva, L., Schaller, M., Macêdo, J.L.V., Zech, W.,
2000b. Effect of five tree crops and a cover crop in multi-strata agroforestry at two
fertilization levels on soil fertility and soil solution chemistry in central Amazonia. Plant and
Soil 221, 143–156.

79
Schroth, G., Bede, L., Paiva, A., Cassano, C., Amorim, A., Faria, D., Mariano-Neto, E., Martini, A.Z.,
Sambuichi, R.R., Lôbo, R., 2013. Contribution of agroforests to landscape carbon storage.
Mitigation and Adaptation Strategies for Global Change, 1-16.
Schroth, G., Jeusset, A., Gomes, A.D., Florence, C.T., Coelho, N.A.P., Faria, D., Laderach, P., 2016a.
Climate friendliness of cocoa agroforests is compatible with productivity increase. Mitig
Adapt Strateg Glob Change 21, 67-80.
Schroth, G., Laderach, P., Martinez-Valle, A.I., Bunn, C., Jassogne, L., 2016b. Vulnerability to climate
change of cocoa in West Africa: Patterns, opportunities and limits to adaptation. Science of
the Total Environment 556, 231-241.
Schulz, B., Becker, B., Götsch, E., 1994. Indigenous Knowledge in a "modern" sustainable agroforestry
system - a case study from Brazil. Agroforestry Systems, 25, 59-69.
Schulz, J., 2011. Imitating natural ecosystems through successional agroforestry for the regeneration
of degraded lands - a case study of smallholder agriculture in northeastern Brazil. In:
Montagnini, F., Francesconi, W., Rossi, E. (Eds.) Agroforestry as a tool for landscape
restoration. New York: Nova Science Publishers.
Shannon, C.E., Weaver, W., 1949. The mathematical theory of communication, fourth edition. The
University of Illinois Press, Urbana.
Smith Dumont, E., Gnahoua, G.M., Ohouo, L., Sinclair, F.L., Vaast, P., 2014. Farmers in Côte d’Ivoire
value integrating tree diversity in cocoa for the provision of ecosystem services. Agroforestry
Systems 88, 1047-1066.
Somarriba, E., 2002. Estimación visual de la sombra en cacaotales y cafetales. Agrofor. Am. 9, 86-94.
Somarriba, E., Cerda, R., Orozco, L., Cifuentes, M., Davila, H., Espin, T., Mavisoy, H., Avila, G.,
Alvarado, E., Poveda, V., Astorga, C., Say, E., Deheuvels, O., 2013. Carbon stocks and cocoa
yields in agroforestry systems of Central America. Agriculture Ecosystems & Environment,
173, 46-57.
Somarriba, E., Suárez-Islas, A., Calero-Borge, W., Villota, A., Castillo, C., Vílchez, S., Deheuvels, O.,
Cerda, R., 2014. Cocoa–timber agroforestry systems: Theobroma cacao–Cordia alliodora in
Central America. Agroforestry Systems 88, 1001-1019.
Sonwa, D., Weise, S., Schroth, G., Janssens, M.J., Howard-Yana, S., 2014. Plant diversity management
in cocoa agroforestry systems in West and Central Africa—effects of markets and household
needs. Agroforestry Systems, 88, 1021-1034.
Sonwa, D.J., Nkongmeneck, B.A., Weise, S.F., Tchatat, M., Adesina, A.A., Janssens, M.J.J., 2007.
Diversity of plants in cocoa agroforests in the humid forest zone of Southern Cameroon.
Biodiversity and Conservation, 16, 2385-2400.
Sood, K.K., Mitchell, C.P., 2006. Importance of human psychological variables in designing socially
acceptable agroforestry systems. Forests, Trees and Livelihoods, 16, 127-137.
Sousa, F., Nicolay, G., Home, R., 2016. Information technologies as a tool for agricultural extension
and farmer to-farmer exchange: Mobile-phone video use in Mali and Burkina Faso. Int J Educ
Dev using ICT (IJEDICT) 12 (3):19-36.
Sperber, C.F., Nakayama, K., Valverde, M.J., Neves, F.D., 2004. Tree species richness and density
affect parasitoid diversity in cacao agroforestry. Basic Appl. Ecol. 5, 241-251.
Staver, C., Guharay, F., Monterroso, D., Muschler, R.G., 2001. Designing pest-suppressive multistrata
perennial crop systems: Shade-grown coffee in Central America. Agrofor. Syst. 53, 151-170.

80
Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M.M., Buchori, D., Erasmi, S., Faust, H., Gerold,
G., Glenk, K., Gradstein, S.R., Guhardja, E., Harteveld, M., Hertel, D., Höhn, P., Kappas, M.,
Köhler, S., Leuschner, C., Maertens, M., Marggraf, R., Migge-Kleian, S., Mogea, J., Pitopang,
R., Schaefer, M., Schwarze, S., Sporn, S.G., Steingrebe, A., Tjitrosoedirdjo, S.S., Tjitrosoemito,
S., Twele, A., Weber, R., Woltmann, L., Zeller, M., Tscharntke, T., 2007. Tradeoffs between
income, biodiversity, and ecosystem functioning during tropical rainforest conversion and
agroforestry intensification. Proceedings of the National Academy of Sciences, 104, 4973-
4978.
Steven, W.H., 1936. A new disease of cacao in the Gold Coast. Gold Coast Farmer 5, 122–144.
Swaminathan, M.S., 1993. Perspectives for crop protection in sustainable agriculture. Ciba
Foundation Symposia 177, 257-272.
Thorlakson, T., Neufeldt, H., 2012. Reducing subsistence farmers' vulnerability to climate change:
evaluating the potential contributions of agroforestry in western Kenya. Agric. Food Secur. 1,
1-13.
Thresh, J.M., Tinsley, T.M., 1959. The viruses of cocoa. West African Cocoa Research Institute, Tafo,
Ghana, 6-28.
Thresh, J.M., Owusu, G.K., 1986. The control of cocoa swollen shoot disease in Ghana - An evaluation
of eradication procedures. Crop Prot. 5, 41-52.
Thresh, J.M., Owusu, G.L.K., Ollennu, L.A., 1988. Cocoa swollen shoot - An archetypal crowd disease.
J. Plant Dis. Prot. 95, 428-446.
Todt, B., 2010. Soil fertility in Monoculture and Succesional Agroforestry Land Use Systems for Citrus
sinensis in Alto Beni, Bolivia. Diplom, Georg-August-Universität.
Todt, B., Kühne, R.F., Gerold, G., 2009. Evaluation of Soil Fertility in Monoculture and Succesional
Agroforestry Land Use Systems for Citrus Sinensis, in Alto Beni, Bolivia. Tropentag
Conference, October 6-8, Hamburg, Germany.
Tondoh, J.E., Kouamé, F.N.G., Martinez Guéi, A., Sey, B., Wowo Koné, A., Gnessougou, N., 2015.
Ecological changes induced by full-sun cocoa farming in Côte d’Ivoire. Global Ecology and
Conservation 3, 575-595.
Tscharntke, T., Clough, Y., Bhagwat, S.A., Buchori, D., Faust, H., Hertel, D., Hölscher, D., Juhrbandt, J.,
Kessler, M., Perfecto, I., Scherber, C., Schroth, G., Veldkamp, E., Wanger, T.C., 2011.
Multifunctional shade-tree management in tropical agroforestry landscapes – A review. J.
Appl. Ecol. 48, 619–629.
Tscharntke, T., Clough, Y., Wanger, T.C., Jackson, L., Motzke, I., Perfecto, I., Vandermeer, J.,
Whitbread, A., 2012. Global food security, biodiversity conservation and the future of
agricultural intensification. Biol. Conserv. 151, 53-59.
Vaast, P., Somarriba, E., 2014. Trade-offs between crop intensification and ecosystem services: the
role of agroforestry in cocoa cultivation. Agrofor. Syst. 88, 947-956.
van Reeuwijk, L.P., 2002. Procedures for soil analysis, sixth edition. Technical Paper 9, International
Soil Reference and Information Centre, Wageningen.
van Vliet, J.A., Giller, K.E., 2017. Mineral Nutrition of Cocoa: A Review. In: Sparks, D.L. (Ed.), Advances
in Agronomy, Vol 141, pp. 185-270.
Vanderplank, J.E., 1947. The relation between the size of plant and the spread of systemic diseases: I.
A discussion of ideal cases and a new approach to problems of control. Ann. Appl. Biol. 34,
376-387.

81
Verchot, L.V., Noordwijk, M.V., Kandji, S., Tomich, T., Ong, C., Albrecht, A., Mackensen, J., Bantilan,
C., Anupama, K.V., Palm, C., 2007. Climate change: linking adaptation and mitigation through
agroforestry. Mitigation and Adaptation Strategies for Global Change, 12, 901-918.
Vernon, A.J., 1967. Yield and light relationships in cocoa. Trop. Agric. 44, 223-228.
Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. Journal of
Statistical Software 36, 1-48.
Vieira, D.L.M., Holl, K.D., Peneireiro, F.M., 2009. Agro-Successional Restoration as a Strategy to
Facilitate Tropical Forest Recovery. Restoration Ecology 17, 451-459.
Vinod, K.K., 2012. Stress in plantation crops: adaptation and management, in: Venkateswarlu, B.,
Shanker, A.K., Shanker, C., Maheswari, M. (Eds.), Crop Stress and its Management:
Perspectives and Strategies. Springer Netherlands, Dordrecht, pp. 45-137.
Wartenberg, A.C., Blaser, W.J., Gattinger, A., Roshetko, J.M., Van Noordwijk, M., Six, J., 2017. Does
shade tree diversity increase soil fertility in cocoa plantations? Agriculture Ecosystems &
Environment 248, 190-199.
Wessel, M., 1971. Fertilizer requirements of cacao (Theobroma cacao L.) in South-Western Nigeria.
Koninklijk Instituut voor de Tropen, Amsterdam.
Wessel, M., Quist-Wessel, P.M.F., 2015. Cocoa production in West Africa, a review and analysis of
recent developments. Njas-Wageningen Journal of Life Sciences 74-75, 1-7.
Wetten, A., Campbell, C., Allainguillaume, J., 2016. High-resolution melt and morphological analyses
of mealybugs (Hemiptera: Pseudococcidae) from cacao: tools for the control of Cacao
swollen shoot virus spread. Pest Manag. Sci. 72, 527-533.
Wood, G.A.R., Lass, R.A., 2001. Cocoa, fourth edition. Blackwell Science, Oxford.
Yana, W., Weinert, H., 2001. Técnicas des sistemas agroforestales MULTIESTRATO: Manual práctico.
PIAF - El Ceibo, Sapecho, Bolivia.
Yelsang, F.D., 2013. Agricultural Land Use Conflict between Landlords and Migrant Farmers in Ghana:
An Examination of Issues Affecting Dagara Migrants in the Brong Ahafo Region. European
Scientific Journal 9, 381-402.
Yunus, M., 2000. How Donor Funds Could Better Reach and Support Grassroots Microcredit
Programmes. Working towards the Microcredit Summit’s Goal and Core Themes. The
Bangladesh Development Studies, 26, 1-13.
Yunus, M., Weber, K., 2007. Creating a world without poverty: Social business and the future of
capitalism. New York: PublicAffairs.

82
Appendix
Appendix A: List of references included in the meta-analysis
Adegbola, M.O.K., no date. Effect of shading on cacao seedlings infected with the cocoa swollen-
shoot virus (CSSV). Annual Report Cocoa Research Institute of Nigeria (CRIN) Ibadan, Nigeria.
Adomako, D., Owusu, G.K., 1974. Studies on the mechanical transmission of cocoa swollen shoot
virus: Some factors affecting virus multiplication and symptom development in cocoa. Ghana
Journal of Agricultural Sciences 7, 7-15
Adomako, D., 1977. A biochemical study of the mechanism of swollen shoot virus resistance in cocoa:
The role of phenolic compounds. Proceedings of the 5th International Cocoa Research
Conference, 1975, Ibadan, Nigeria, 344-350.
Adu-Ampomah, Y., Adomako, B., Owusu, G.K., Ollennu, L.A., 1996. Breeding for resistance to the
cocoa swollen shoot virus in Ghana. Proceedings of the lnternational Workshop on the
Contribution of Disease Resistance to Cocoa Variety lmprovement, 1996, Salvador, Brazil,
173-179.
Cilas, C., Dufour, B.P., Djiekpor, E.K., 1988. Study of resistance to swollen shoot of cocoa (Theobroma
cacao L.) in a practically complete 8 x 8 diallele. Café, Cacao, Thé 32, 105-110.
Crowdy, S.H., Posnette, A.F., 1947. Virus diseases of cacao in West Africa. II. Cross-immunity
experiments with viruses 1A, 1B and 1C. Ann. Appl. Biol. 34, 403-411.
Dale, W.T., 1957. Studies on resistance and tolerance to cacao viruses. Proceedings of the Cacao
Breeding Conference, 1957, West African Cocoa Research Institute, Tafo, Ghana, 3-6.
Djiekpor E.J., Cilas, C., Paulin, D., 1995. Spotlight on cocoa breeding for resistance to swollen shoot.
Proceedings of the International Workshop on Cocoa Breeding Strategies, 1994, Kuala
Lumpur, Malaysia, 38-40.
Dzahini-Obiatey, H.K., Aculey, P.C., Ollennu, L.A., 2005. Inhibition of cocoa swollen shoot badnavirus
(CSSV) by an antiviral substance. Proceedings of the 14th International Cocoa Research
Conference, 2003, Accra, Ghana, 1365-1371.
Hanna, A.D., Heatherington, W., 1957. Arrest of the swollen-shoot virus disease of cacao in the Gold
Coast by controlling the mealybug vectors with the systemic insecticide, Dimefox. Ann. Appl.
Biol. 45, 473-480.
Kenten, R.H., Legg, J.T., 1970. Methods for assessing tolerance and resistance of different types of
cocoa to cocoa swollen-shoot virus. Ann. Appl. Biol. 65, 419-424.
Kenten, R.H., Lockwood, G., 1977. Studies on possibility of increasing resistance to cocoa swollen-
shoot virus by breeding. Ann. Appl. Biol. 85, 71-78.
Legg, J.T., Kenten, R.H., 1970. The resistance and tolerance of different cocoa varieties to cocoa
swollen-shoot virus in Ghana. Ann. Appl. Biol. 65, 425-434.
Legg, J.T., Kenten, R.H., 1971a. Field experiments on the resistance of cocoa to cocoa swollen-shoot
virus. Ann. Appl. Biol. 67, 369-375.
Legg, J.T., Kenten, R.H., 1971b. Selection of cocoa progenies resistant to and tolerant of cocoa
swollen shoot virus. Proceedings of the 3rd International Cocoa Research Conference, 1969,
Accra, Ghana, 503-511.
Legg, J.T., Lockwood, G., 1977a. Evaluation and use of a screening method to aid selection of cocoa
(Theobroma cacao) with field-resistance to cocoa swollen-shoot virus in Ghana. Ann. Appl.
Biol. 86, 241-248.
Legg, J.T., Lockwood, G., 1977b. Tolerance and resistance to cocoa swollen shoot virus. Proceedings
of the 5th International Cocoa Research Conference, 1975, Ibadan, Nigeria, 403-408.
83
Legg, J.T., Lockwood, G., 1981. Resistance of cocoa to swollen-shoot virus in Ghana. I. Field trials.
Ann. Appl. Biol. 97, 75-89.
Lockwood, G., 1981a. Genetic aspects of resistance to cocoa swollen-shoot virus in Ghana. Ann. Appl.
Biol. 98, 131-141.
Lockwood, G., 1981b. Resistance of cocoa to swollen-shoot virus in Ghana. II. Assessment of seed
inoculation experiments. Ann. Appl. Biol. 97, 91-97.
Longworth, J.F., Are, L.A., Freeman, G.H., 1965. Problems in assessment of the response of cacao
progenies (Theobroma cacao L.) to infection with swollen shoot virus in Nigeria. Nigerian
Agricultural Journal 2, 1-6.
Ollennu, L.A., 1989. Swollen shoot disease - New control measures under consideration. Proceedings
of the Cocoa Swollen Shoot Workshop, 1989, Cocoa Research Institute of Ghana, Tafo,
Ghana, 9-12.
Ollennu, L.A., Hughes, J.d.A., Owusu, G.K., 1996. Mild strain cross-protection of cocoa against cocoa
swollen shoot badnavirus. Tropical Science 36, 116-128.
Ollennu, L.A., Owusu, G.K., 1997. Mild virus strain protection studies with cocoa swollen shoot
Badnavirus (CSSV) in Ghana. Proceedings of INCOPED 1st International Seminar on Cocoa
Pests and Diseases, 1997, 204-211.
Ollennu, L.A., Owusu, G.K., Dzahini-Obiatey, H.K., 1999. Recent studies of mild strain cross-protection
with cocoa swollen shoot virus. J. Ghana Sci. Assoc. 2, 5-11.
Ollennu, L.A., Owusu, G.K., 2002. The use of cocoa seedlings for field evaluation of mild strain
protection against cocoa swollen shoot Badnavirus. J. Ghana Sci. Assoc. 4, 14-18.
Ollennu, L.A., Owusu, G.K., 2003. Field evaluation of the protective capability of CSSV mild strain N1
against severe strain New Juaben (1A) isolate. Ghana Journal of Agricultural Science, 3-12.
Ollennu, L.A., Osei-Bonsu, K., Aneani, F., Acheampong, K., 2005. Preliminary studies of the control of
cocoa swollen shoot disease by the use of immune crops as barrier. Proceedings of the 14th
International Cocoa Research Conference, 2003, Accra, Ghana, 839-843.
Owusu, G.K., 1971. Availability of cocoa swollen shoot virus to mealybugs from tolerant and sensitive
cocoa. Proceedings of the 3rd International Cocoa Research Conference, 1969, Accra, Ghana,
512-517.
Owusu, G.K., Ollennu, L.A., Dzahini-Obiatey, H.K., 1999. The prospects for mild strain cross-protection
to control cocoa swollen shoot disease in Ghana. Proceedings of the 12th International
Cocoa Research Conference, 1996, Salvador (Bahia), Brazil, 121-127.
Paulin, D., Djiekpor, E.K., Caplong, P., Cilas, C., 1994. Evaluation of cacao resistance to swollen shoot
virus. Proceedings of the 11th International Cocoa Research Conference, 1993,
Yamoussoukro, Côte d'Ivoire, 269-274.
Posnette, A.F., Todd, J.M., 1951. Virus diseases of cacao in West Africa. VIII. The search for virus-
resistant cacao. Ann. Appl. Biol. 38, 785-800.
Quainoo, A.K., Wetten, A.C., Allainguillaume, J., 2008. The effectiveness of somatic embryogenesis in
eliminating the cocoa swollen shoot virus from infected cocoa trees. J. Virol. Methods 149,
91-96.
Tsatsu, D.K., Bekou, K., 2003. Development and strategy in the fight against swollen shoot disease in
cocoa in Togo. Proceedings of the 14th International Cocoa Research Conference, 2003,
Accra, Ghana, 633-642.

84
Appendix B: Reducing the spread of CSSVD – the importance of landscape connectivity
When I travelled from the Brong Ahafo region to the Western North region during the induction phase
of the project, I was stunned by an observation I made on the fields. While behind the forest protection
strip marked in Figure 17 with a red cross (Goaso district, Brong Ahafo region) there were no incidences
of CSSVD, after I had crossed forest protection strip (Sankore district, last district of Brong Ahafo before
reaching the Western North region) all the cocoa areas were fully infected. Thus, I came up with the
idea of investigating the potential of fragmented cocoa landscapes to reduce CSSVD spread. Figure 18
shows a block of about 300 hectares in Brong Ahafo region, and a whole sector (made up of several
blocks) of about 2’000 hectares in Western North region. While the cocoa areas in Brong Ahafo are all
separated by at least the 10 m that were reported to be effective to reduce CSSVD spread in Ghana
(Ollennu et al., 1989b), the cocoa areas in Western North are continuous and fully affected by CSSVD.

Figure 17: The forest protection strip marked with a red cross seems to have served as a physical
barrier preventing the spread of CSSVD. While all the cocoa areas in front of the strip were infected,
there was no CSSVD behind the forest strip.

Since CSSVD spreads through mealybug vectors, fragmented cocoa landscapes may decrease the
spread of the disease. However, nobody has carried out systematic studies on cocoa landscape
connectivity so far. In its second national mapping campaign, CHED has mapped all the cocoa areas
and CSSVD presence in Ghana between 2010 and 2013. This data can be used to determine the
influence of landscape connectivity on CSSVD spread in cocoa landscapes on a scale of hundreds of
thousands of hectares (> 100’000 ha) or even larger catchments (> 1 Mio ha). Preliminary results from
about 5’000 ha show that high connectivity sites (> 85% of area covered with cocoa) are highly affected
by CSSVD (> 75% of cocoa area diseased) while CSSVD incidences in low connectivity sites (< 60% of
area covered with cocoa) are much rarer (< 2% of cocoa area diseased, Figure 19).

85
Figure 18: Upper panel: block of about 300 hectares in Brong Ahafo region, where all the cocoa areas
are separated by at least 10 m and no CSSVD occurs (SO: scattered outbreaks). Lower panel: sector
(made up of several blocks) of about 2’000 hectares in Western North region, where the cocoa areas
are continuous and fully affected by CSSVD (AMI: area of mass infection).

Figure 19: Percentage of block area covered with cocoa related to cocoa area infected by CSSVD.

86
To further investigate the potential of fragmented cocoa landscapes to reduce CSSVD spread, we have
to compare fragmented cocoa landscapes to more continuous ones using state of the art methods in
spatial data analysis with geographical information systems (GIS). This may include raster-based
connectivity analysis using tool such as inverse distance weighting (IDW, Figure 20). Contrary to similar
approaches in biodiversity conservation, the objective of this activity is to find the minimum distance
by which cocoa areas should be separated in order to minimize CSSVD spread. Data about location of
cocoa areas and CSSVD presence should be integrated with information on the mealy bug behaviour
(speed of movement) and time period of infectivity (for how long can the vector transmit the disease?)
to establish an ecological network model. The expected results may provide crucial insights for
Ghanaian policy makers to design sustainable cocoa landscapes.

Figure 20: Concept to investigate the potential of fragmented cocoa landscapes to reduce CSSVD
using raster-based connectivity analysis and inverse distance weighting (IDW)

87
88
Appendix C: Developing a novel mapping tool for CSSVD occurrence
In its second national mapping campaign, CHED has mapped all the cocoa areas and CSSVD presence
in Ghana between 2010 and 2013, which amounts to a scale of several million hectares. Hundreds of
extension officers carried out this tremendous work, walking around in the cocoa production
landscape carrying handheld GPS devices in order to capture the circumferences of cocoa areas.
Currently, CHED is carrying out the third national mapping campaign to map all the cocoa areas in the
country, and is still relying on the same technology. If mapping CSSVD from the air using drone imagery
and normalized difference vegetation index (NDVI) would become possible, this would be a major
contribution towards increasing efficiency in the Ghanaian cocoa sector.
Since CSSVD leads to leaf chlorosis (Figure 13), the disease-induced leaf yellowing may be used to map
CSSVD occurrence from the air using drone imagery and NDVI. The chlorophyll of plants reflects the
green light. However, it absorbs nearly all the visible red light (which is why plants in daylight look
green), and almost completely reflects the (invisible) near-infrared (NIR) light (Figure 21). The
difference between the reflectance values of the red light and the NIR light determines the NDVI, using
𝑁𝐼𝑅−𝑅𝑒𝑑
the equation 𝑁𝐷𝑉𝐼 = 𝑁𝐼𝑅+𝑅𝑒𝑑. If a plant is unhealthy (such as a chlorotic leaf affected by CSSVD), the
reflectance values of the NIR light is drastically reduced, which reduces the NDVI. Therefore, NDVI is
well suited to serve as a proxy for CSSVD occurrence. Researchers in America have successfully applied
this technique to detect potato virus Y infection accurately and early using drones14. They were able
to detect the virus well before it was visible on the field by the bare eye (Figure 22).

Figure 21: Spectrum of visible and near-infrared light and absorption patterns of a healthy plant (green
curve), and unhealthy plant (orange curve) and soil (red curve)15

14
http://www.capitalpress.com/Research/20160123/drone-research-nets-results-in-idaho-spud-fields
15
Source: https://publiclab.org/notes/warren/10-27-2011/video-tutorial-creating-false-color-ndvi-aerial-
wetlands-imagery
89
Figure 22: NDVI of sick (left) and healthy potato plant (right). No difference is visible between these
plants with RGB or human eye.

The development of the novel mapping tool may be pursued in several stages. In a first step, a lab scale
analysis can deliver the most accurate results as several important parameters such the distance and
the angle between the camera and the infected leaf, as well as the light intensity can be controlled.
This is best done using a copystand (German “Reprostativ”) with a non-reflecting base surface and two
light bulbs as shown in Figure 23. In a second step, standardized on-station trials with standardized
layout, management practices and planting material can serve as a reference to calibrate the two
cameras (one RGB, and one NIR camera) that will need to be used in order to get NDVI pictures. In this
case, the only main factor influencing cocoa leaf colour in should be CSSV infection. Finally, the
technology may be applied on-farm. In this case, however, other factors that may influence cocoa leaf
colour have to be taken into account in order to correct for confounding effects. The two most
important co-factors are soil fertility parameters and capsid damage, which will have to be considered
for statistical analyses. The pictures (raster data) can be analysed using GIS techniques.
There are several options for obtaining the pictures in on-station trials and on farms. A promising
option is the use of a drone (Figure 24), provided one can get the permit to use it from the respective
responsible authority, which proved to be a challenge in our case. The alternative to the drone is to
take the pictures using a telescopic stick16 of 10m in length with two magic arms mounted on top to
fix the cameras, which is what I did. A script on the SD card of the cameras ensures that pictures are
taken automatically. Once the picture are taken, the software ImageJ17 can be used to process the
images. Thereby, one has to crop the pictures to single leaves, apply a threshold colour to select the
pixels of interest, measure the RGB values of RGB images and use the median for NIR pictures. The
skewness then reveals if the curve is left- (-) or right- (+) skewed, indicating more or less green pixels.
Another option is to use Excel to calculate the TGI value, which is the area of the triangle of red, green
and blue in the graph wavelength (x) ~ intensity (y). Here, an image with higher number of yellow cells
will have a lower TGI.

16
http://www.mess-werkzeug-kiste.de/Teleskopmessstab-telefix.html
17
https://imagej.net/Welcome
90
Figure 23: Setup for the lab scale analysis of developing the novel mapping tool. Upper panel:
copystand (German “Reprostativ”) with a non-reflecting base surface and two light bulbs. Lower panel:
resulting picture of an infected leaf when using this setup.

91
Figure 24: Upper panel: drone model “GreenCopter® Tropic-Quadro” by RUCON Engineering
(Germany) we planned to use for developing a novel mapping tool for CSSVD occurrence. Middle panel:
taking pictures in the field using a telescopic stick. Lower panel: two magic arms mounted on top to fix
the cameras. A script on the SD card of the cameras ensures that pictures are taken automatically.
92
Appendix D: Personal motivation
Why did I decide to carry out this particular PhD study? My heartfelt wish is to help sentient beings
around me when- and wherever I can. Cocoa research has a special potential in this respect, as the
product – chocolate – has a global relevance, and improving the sustainability of production systems
may help to support the livelihoods of producer families, and foster biodiversity conservation in an
agro-ecological matrix of cocoa agroforestry systems (Perfecto and Vandermeer, 2010). Reflecting on
my background, here is a little essay why I happened to work on CSSVD in Ghana:
There were two main factors, which influenced my decision to study agricultural sciences at ETH Zurich.
Firstly, plants are my passion. I have been cultivating plants from the days of my childhood, which
influenced me wanting to focus my studies around plants. Secondly, my first authentic experience in a
tropical country (Madagascar in 2004) was fascinating. I was amazed to meet the friendly locals, get to
know what they eat and cultivate, and to see how they go about their daily lives on a materially simpler
level compared to Switzerland. This experience left a strong imprint on me, which developed into a
sincere wish to contribute to the resolution of the food security problem the developing world is facing.
Realising that I wanted to contribute to this problem, I decided to study agricultural sciences (BSc.) and
agro-ecosystem science (MSc.) at ETH Zurich. My studies introduced me to the complexity of agro-
ecosystems, and helped me to better apprehend the factors, which have led to the current challenges,
as well as measures that hold the potential to tackle the pressing issues. By doing my BSc. and MSc.
theses in Ivory Coast and an internship in Sri Lanka, I gained relevant experience in tropical agriculture.
While I started at ETH thinking that I would be going into development aid after my studies, I came to
realize that research is just as relevant for development as direct interventions, and I developed an
enthusiasm for research. Therefore, I applied for a position as Junior Scientist in tropical agriculture at
the Research Institute of Organic Agriculture (FiBL) while finalizing my MSc thesis. My application was
successful, and I have been employed and pursuing agricultural research for development at FiBL with
joy since 2011.
At FiBL, I worked in the project “Long-term farming systems comparisons in the tropics” (SysCom) from
2011 to 2016. One of the project sites is located in Bolivia, where we compare various cocoa production
systems differing in management (conventional/organic) and diversity (monoculture/agroforestry
system) in an on-station trial, and we work with smallholders growing organic cocoa in several on-farm
trials. Through this work, I have focussed my research on the sustainability of cocoa production
systems, and developed the wish to pursue this path in a PhD project. Through my previous work and
travels in the main cocoa production region (Ghana, Ivory Coast), I developed a sincere sympathy for
these countries and the people living there, which strengthened my intention to conduct my PhD study
there.
I was given the opportunity to implement the project through an E4D scholarship, which is financed by
the Sawiris Foundation for Social Development, and managed by ETH Global. Thus, I embarked on the
journey of trying to contribute to the resolution of the CSSVD problem, which I hope will have an
impact on smallholders producing cocoa in West Africa, who live on cocoa and whose livelihoods
depend on the revenues from the crop. Good links to many of the key institutions working on the topic
of sustainable cocoa production systems helped me in developing synergies.

93
94