Agriculture, Ecosystems and Environment 252 (2018) 83–92
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment
journal homepage: www.elsevier.com/locate/agee
Research paper
Agroforestry systems can mitigate the severity of cocoa swollen shoot virus MARK
disease
Christian Andresa,b,⁎, Wilma J. Blaserb, Henry K. Dzahini-Obiateyc, George A. Ameyawc,
Owusu K. Domfehc, Moses A. Awiagahc, Andreas Gattingera, Monika Schneidera,
Samuel K. O dffei , Johan Sixb
a Research Institute of Organic Agriculture (FiBL), Ackerstrasse 113, Postfach 219, 5070, Frick, Switzerland
b Department of Environmental Systems Science, Swiss Federal Institute of Technology, ETH Zurich, Tannenstrasse 1, 8092 Zürich, Switzerland
c Cocoa Research Institute of Ghana, P. O. Box 8, New Tafo-Akim, Eastern Region, Ghana
d University of Ghana, P.O. Box LG 25, Legon, Accra, Ghana
A R T I C L E I N F O A B S T R A C T
Keywords: Currently, the only effective treatment for cocoa (Theobroma cacao L.) infected with the cocoa swollen shoot
Theobroma cacao virus disease (CSSVD) is to cut and replant infected trees. Hence, the development of preventive control mea-
Cocoa swollen shoot virus disease sures and strategies to mitigate the severity of the disease are of utmost importance. While past research has
Agroforestry mainly focused on resistance breeding, mild strain cross protection and vector control, diversification measures
High-light stress such as agroforestry have received relatively less attention, despite their potential to mitigate CSSVD severity.
Soil fertility
Yield 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.
1. Introduction anywhere in the world (Thresh et al., 1988). Latest estimates put the
total number of trees claimed by the disease at over 300 Mio cocoa trees
Crop diseases are an increasing problem worldwide, and have been (Dzahini-Obiatey, personal communication), which in monetary terms
estimated to cause yield losses ranging between 20%–40% of global may amount to losses of several Mio US dollars for the government.
agricultural productivity (Savary et al., 2012). In Ghana, the second Furthermore, the high prevalence of CSSVD has curtailed many cocoa
largest producer of cocoa (Theobroma cacao L.) worldwide, the liveli- farmers of their regular source of income, as whole fields could be lost
hoods of around 800,000 families depend on revenues from the crop to the disease (Dzahini-Obiatey et al., 2010).
(Danso-Abbeam, 2014). Since more than eight decades, cocoa pro- Therefore, the development of efficient preventive control measures
ductivity in West Africa has been severely limited by the Cocoa Swollen for CSSVD that are economically feasible for farmers is of utmost im-
Shoot Virus Disease (CSSVD), especially in Ghana (Dzahini-Obiatey portance. Despite a lot of research on preventive control measures
et al., 2010). (Andres et al., 2017) during the last eighty years (mainly focused on
The only effective treatment for CSSVD is to cut infected trees and resistance breeding, mild strain cross-protection and the control of
replant with disease free planting material. Launched in 1946, the of- mealybug vectors), however, farmers have not implemented those
ficial eradication campaign in Ghana has been considered the most measures consistently, and the disease is still widespread throughout
ambitious and costliest of its kind to control a plant viral disease Ghana (Ameyaw et al., 2014).
⁎ Corresponding author at: Research Institute of Organic Agriculture (FiBL), Ackerstrasse 113, Postfach 219, 5070, Frick, Switzerland.
E-mail address: christian.andres@fibl.org (C. Andres).
http://dx.doi.org/10.1016/j.agee.2017.09.031
Received 28 June 2017; Received in revised form 25 September 2017; Accepted 27 September 2017
Available online 05 November 2017
0167-8809/ © 2017 Elsevier B.V. All rights reserved.
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
On the other hand, diversification measures such as agroforestry or trees reduce both CSSVD symptom severity and capsid damage of
barrier (strip) cropping systems show some potential to reduce CSSVD CSSVD-infected cocoa aged 16 years and above, and thus contribute to
spread (Schroth et al., 2000; Domfeh et al., 2016), but have received lowering the yield-reducing effect of CSSVD as compared to cocoa
less attention in past research (Andres et al., 2017). Shade trees can stands without shade tree canopy cover. Taking into account that most
provide a multitude of ecosystem services (Beer, 1987; Andres and cocoa farmers in Ghana do not prune their shade trees, we also hy-
Bhullar, 2016; Schneider et al., 2016). If the species are chosen delib- pothesised that soil fertility is not different in full-sun and shaded plots.
erately and managed well, they can have beneficial effects on the de- Consistent with shade recommendations by existing manuals for good
velopment and health of cocoa trees and thus help them to cope with agronomic practices in cocoa cultivation, we expected a medium shade
external stressor such as CSSVD. These include the i) prevention of level of 40%–50% to be optimal to balance CSSVD symptom severity
high-light stress (Rice and Greenberg, 2000), ii) improvement of water versus reduced cocoa yield, and hypothesised that shade per se (and its
and nutrient recycling (Buresh et al., 2004), iii) contribution to the potential causal factors) is more important to explain CSSVD symptom
control of pests and diseases (Ahenkorah et al., 1974; Schroth et al., severity than agro-biodiversity or soil fertility.
2000 Wood and Lass, 2001), iv) maintenance of soil fertility (Götsch,
1994; Schroth et al., 2000; Buresh et al., 2004; Bos et al., 2007; Fonte 2. Materials and methods
and Six, 2010; Gama-Rodrigues, 2011; Tscharntke et al., 2011), and v)
regulation of micro-climatic conditions for vigorous crop growth 2.1. Site selection
(Schroth et al., 2016). These effects may contribute to the maintenance
of long-term sustainable cocoa yields (Rice and Greenberg, 2000). This study was performed in four operational districts of the Cocoa
Furthermore, agroforestry systems were found to be equally or more Health and Extension Division (CHED), subsidiary of the Ghana Cocoa
profitable than full-sun systems under ideal management conditions Board (COCOBOD), in the Eastern Region of Ghana. The study region
(Armengot et al., 2016). lies within the semi-deciduous rain forest zone with a bi-modal rainy
However, shade trees may also have adverse effects on the devel- season, annual rainfall ranging between 1200 mm and 1700 mm. Tall
opment and health of cocoa trees, as some are alternative hosts for trees with evergreen undergrowth characterize the vegetation.
CSSV, harbouring the virus and serving as a source of infection for Humidity and temperature are generally high ranging between
cocoa stands planted nearby (Posnette et al., 1950). Depending on the 70%–80% and 20 °C–32 °C, respectively. Soils of the study region are
architecture of their root systems, they may also be competing for dominated by Acrisols and Lixisols (Ghana Statistical Service, 2014).
water, which may cause drought stress in cocoa trees, especially during For the selection of suitable study sites, we collaborated closely with
the dry season (Carr and Lockwood, 2011). Furthermore, some shade CHED. Plots were located in the Eastern Region of Ghana because of the
tree canopies may provide too much shade, increasing relative hu- presence of gradients in both CSSVD prevalence (from areas of mass
midity in cocoa stands and thus favouring fungal diseases (Schroth infection (Southeast) to areas of only scattered outbreaks (Northwest))
et al., 2000; Babin et al., 2010; Oro et al., 2012). and shade tree canopy cover. Traditional agroforestry systems with
Several studies showed that shade trees can decrease pest popula- shade trees are more prevalent in this region as compared to for ex-
tions (Beer et al., 1998; Jaramillo et al., 2009; Thorlakson and Neufeldt, ample the Western Region of Ghana, where large-scale cocoa mono-
2012), and favoured natural pest antagonists (Opoku et al., 2002). culture are more common. With the objective to compare shade
Bigger (1981) showed that mealybug vectors of CSSVD were more (agroforestry systems) to full-sun (monoculture) production systems,
abundant under full-sun conditions than under shade. In addition, the we selected 23 pairs of one shade and one full-sun plot each (46 plots)
number of mealybug predators and parasitoids was higher on shaded along a Southeast–Northwest transect (between 6°3′0.23″N,
plots (Bigger, 1981; Mani and Shivaraju, 2016). Also, infection with 0°29′28.38″W and 7°26′28.10″N, 1°9′0.03″W, Fig. A.1 in
mistletoes, which indirectly encourage the spread of CSSVD, was shown Supplementary material). Thereby, we covered thw two gradients
to be higher under full-sun conditions in West Africa (Schroth et al., mentioned above. Our target was to be close to 0% and above 40% of
2000). While these results underline the potential of agroforestry sys- shade for full-sun and shade plots, respectively. All plots were located
tems to reduce CSSVD spread, the potential of shade trees to reduce the in cocoa areas infected with CSSVD that were at least 10 ha in size and
severity of CSSVD symptoms has received less research attention in the planted with Amazonia type cocoa aged between 25 and 28 years
past. (average 26.5 years).
Here we explore the hypothesis that in a CSSVD-infected cocoa Each plot measured 50 m × 50 m (0.25 ha). To avoid border effects,
stand, symptom severity may be related to the general health status of we obtained all data in inner sub-plots measuring 30 m × 30 m. Paired
the plantation. While it is established that shade reduces the physio- plots were similar in terms of exposure, slope, etc. and located as close
logical stress on cocoa trees and their susceptibility to diseases (Beer as possible (max. 100 m apart) in order to avoid confounding effects
et al., 1998 Wood and Lass, 2001; Babin et al., 2010; Oro et al., 2012), such as differing soil types, etc. At the same time, each pair of shade/
only one study carried out under controlled research conditions has full-sun was located in a maximum distance (up to 45 km) to the next
suggested the potential of shade to regulate CSSVD symptom severity pair in order to cover potential differences in local conditions such as
(Legg, 1982). However, the extent to which shade influences CSSVD microclimate, landscape structure and soils.
symptom severity and consequently cocoa yield has not been in-
vestigated under actual farmers’ field conditions so far. Furthermore, 2.2. Data collection
while existing manuals for good agronomic practices in cocoa cultiva-
tion recommend between 40%–50% shade (Beer et al., 2004; SAN, 2.2.1. Plot characterization
2005; Opoku-Ameyaw et al., 2010; Schroth et al., 2016), the question To characterize our research plots, we obtained various soil and
whether this degree of shading is also optimal to cope with CSSVD plant parameters at plot and tree level. We counted number of cocoa,
remains unanswered. Moreover, while there are many factors that may plantain and shade trees in each plot. With the number of cocoa trees,
indirectly affect CSSVD symptom severity (e.g. agro-biodiversity, soil we calculated a plot-specific cocoa density (“Spacing”), which we used
fertility, etc.), research has not addressed the question which of those as a random factor for statistical analyses in order to correct for varying
factors are the most influencing ones. cocoa tree densities in different plots (see Section 2.3). We identified
Thus, we conducted on-farm research in Ghana in two seasons in shade tree species with the help of staff from the Cocoa Research In-
2016 (dry season from December to March, wet season from April to stitute of Ghana (CRIG), farmers and a photo-guide for the forest trees
mid-November) to elaborate the effects of shade on CSSVD symptom of Ghana (Hawthorne and Gyakari, 2006). In order to characterize the
severity, capsid damage and cocoa yield. We hypothesised that shade agro-biodiversity of the research plots, we calculated the Shannon
84
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
85
Table 1
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 of Number of Number of Number of Shade tree species Shannon Visual shade tree Photosynthetic active
infestation damage cocoa trees plantain trees shade trees shade tree index of canopy cover radiation [% full-sun]
[per ha] [per ha] [per ha] species [per diversity estimation [%] (μmol m−2 s−1)
plot]
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 floribunda, 0.35 0.53 0.56 (1120)
Terminalia ivorensis, Terminalia superba
14 Poor Very high 1.65 989 389 33 3 Elaeis guineensis, Milicia excelsa, Persea americana 0.72 0.00 0.97 (1906)
15 Poor High 1.53 833 11 67 4 Azadirachta indica, Holarrhena floribunda, 0.42 0.11 0.34 (669)
Spathodea campanulata, Terminalia ivorensis
16 Poor Medium 1.32 1′378 0 22 2 Allanblackia parviflora, Rauvolfia vomitoria 0.09 0.00 0.89 (1747)
17 Very poor Medium 1.58 600 0 33 3 Ceiba pentandra, Elaeis guineensis, Spathodea 0.26 0.03 0.93 (1255)
campanulata
18 Good Low 1.35 878 11 178 8 Alstonia boonei, Elaeis guineensis, Holarrhena 0.80 0.60 0.39 (440)
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, Samanea 1.06 0.46 0.69 (420)
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 pterocarpoides, Cola 0.62 0.62 0.59 (426)
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, 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 sinensis, 0.86 0.41 0.44 (694)
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
(continued on next page)
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
index (Hs) of diversity (Shannon and Weaver, 1949; Kindt and Coe,
S
2005) using the equation HS = −∑ pi ln pi, where: S= number of
i=1
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 1). 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 sig-
nificantly 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 ra-
diation (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, re-
spectively (Table 1). 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)) ac-
cording 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 detec-
tion 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 stan-
dard 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 pre-
sence in 26 of the 46 plots, so we only used these 26 plots for further
data collection (see Section 2.2.2), and only present the data of these
plots here. All 26 plots were infected with the severe CSSVD strain 1A.
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 to 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 banding, 6 = diffused flecking, 7 = fern pat-
tern, 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
86
Table 1 (continued)
Plot Condition Mistletoe Capsid Number of Number of Number of Number of Shade tree species Shannon Visual shade tree Photosynthetic active
infestation damage cocoa trees plantain trees shade trees shade tree index of canopy cover radiation [% full-sun]
[per ha] [per ha] [per ha] species [per diversity estimation [%] (μmol m−2 s−1)
plot]
31 Good Low 1.70 722 100 133 6 Allanblackia parviflora, Amphimas pterocarpoides, 0.96 0.76 0.18 (118)
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 floribunda, 0.79 0.05 0.88 (580)
Musanga cecropioides, Rauvolfia vomitoria
Mean S 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
sem 0.21 0.23 0.13 69 13 18 0.8 n.a. 0.09 0.05 0.05
Mean FS 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
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.
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
(September) in 2016. full-sun plots (i.e., treating the explanatory variable Shade as a cate-
We assessed cocoa yield on the same 520 trees by counting the gorical variable with two levels), using a normal linear mixed effect
number of pods shortly before the major harvest season in September model with Shade as fixed predictor and District (n = 4), Pair (n = 13),
2016. Furthermore, we assessed capsid damage on flush leaves during and Spacing (n = 26) as random factors. The random factors in our
the dry season in March and April 2016 with the help of local capsid models accounted for variabilities that may exist due to regional (Dis-
specialists, by rating each cocoa tree on a categorical scale from 1 to 5 trict) or local agro-ecological conditions (Pair), or plot-specific growing
where the numbers represent the percentage of damage on flush leaves: conditions (Spacing). Secondly, we assessed the relationship of CSSV
1 = 1–20%, 2 = 21–40%, 3 = 41–60%, 4 = 61–80%, 5 = 81–100% symptom severity and cocoa yield (the two outcome variables of main
(Awudzi, 2014). 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
2.3. Data analysis a quadratic term of Shade (Shade2) as additional fixed predictor, as well
as Plot (n = 26) as additional random factor. The random factor Plot
In the dataset of CSSVD symptom severity, we classified all trees accounted for variability that may exist due to pseudo-replication of
with severity scores 9 and 10 (“dying plant” and “dead plant”) as individual trees within each plot. In the case of cocoa yield, we ad-
outliers, and removed them from the dataset, because these symptoms ditionally included Capsid Damage (n = 438) as a random factor in the
could also be caused by other phenomena such as mistletoe or capsid model, with the aim to discern the effect of CSSVD symptom severity by
damage, and may therefore not solely be attributed to CSSVD. This led separating it from the effect of capsid damage on yield. We encountered
to the removal of 155 (30%) and 46 (9%) data points from the data sets no violations of model assumptions by graphical residual analysis
of full-sun and shade, respectively (total number of cocoa tree ob- (normal Q–Q and Tukey-Anscombe plots). We used the lmer function of
servations = 839). However, because tree death is arguably at least the lme4 package in R for all linear mixed effect model analyses (Bates
partly caused by CSSVD, the removal of these data points may have led et al., 2015), and Pearson’s correlation coefficients (r) to measure
to an underestimation of the average damage, particularly under full- correlations between individual parameters.
sun conditions. Consequently, we rather underestimated overall dif- We calculated the marginal R2 values shown in Fig. 1 as a measure
ferences between shade and full-sun plots, indicating that the data we of the proportion of explained variance by subtracting the ratio be-
present here are a conservative measure of the actual effects on the tween the explained variance (total variance/residual variance) from
field. In the dataset of cocoa yield, we excluded the data from four the total variance of the response variable (CSSVD symptom severity or
entire plots (number of plots analysed (n) = 22, total number of cocoa cocoa yield) as explained by Nakagawa and Schielzeth (2013). To
tree observations = 440), because the farmers had applied a heavy identify optimal shade levels (vertical solid lines in Fig. 1), we used the
rejuvenation pruning to their cocoa trees, which severely influenced the minimum/maximum of the quadratic regression lines as calculated by
yield data. setting its derivative to zero. As uncertainty measurements for these
We analysed our data in two different ways. First, we compared all optima we reported the 95% confidence intervals, which we obtained
outcome variables shown in Table 1 and Table 2 between shade and
Table 2
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 Shade (S)/Full-Sun pH [H2O, C [g kg−1] N [g kg−1] P [mg kg−1] Ca Mg + −1org tot av ex ex Kex [cmol kg ]
district (FS) 1:1] [cmol+ kg−1] [cmol+ kg−1]
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.
87
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
Fig. 1. 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/
dashed). Dashed lines indicate the suggested range of
optimal shade levels to balance CSSVD symptom
severity versus reduced cocoa yield, based on our
dataset.
via Monte Carlo simulation based on 5,000 simulations from the joint the severity score increased again. The 95% confidence interval of the
posterior distributions of the model parameters from which we calcu- optimal shade level for CSSVD symptom severity lay between 45%–76%
lated 5,000 values that constituted random samples from the posterior (dashed/dotted lines in Fig. 1A). There were significant positive cor-
distribution of the optimal shade values. We used the 2.5% and the relations between CSSVD symptom severity and mistletoe infestation
97.5% quantiles of these values as lower and upper limits of the 95% (r = 0.32, p < 0.01) as well as capsid damage (r = 0.32, p < 0.01).
confidence intervals of the optima. Similarly, cocoa yield per tree was significantly higher under shade
To elaborate on potential causal factors for shade effects on CSSVD (by 42% on average, Fig. 1B) until an optimal shade level of 39% shade,
symptom severity, we performed regression analyses between CSSVD after which the yield decreased again. The 95% confidence interval of
symptom severity and all possible and meaningful fixed predictors the optimal shade level for cocoa yield lay between 34%–53% (dotted/
(number of shade trees, number of shade tree species, Shannon index of dashed lines in Fig. 1B). The optimal shade levels for CSSVD symptom
diversity, pH, Corg, Ntot, P, Ca, Mg and K), using District (n = 4), Pair severity and cocoa yield overlapped between 45%–53% (dashed lines in
(n = 13), Plot (n = 26) and Spacing (n = 26) as random factors. We Fig. 1).
conducted all analyses with the statistical software R, version 3.3.1 (R The regression analyses of potential causal factors for shade effects
Core Team, 2017). on CSSVD symptom severity revealed that there were significant ne-
gative correlations between CSSVD symptom severity and number of
3. Results shade trees as well as number of shade tree species, while there was a
significant positive correlation with Caex (Fig. 2). In addition, the cor-
Shade and full-sun plots showed no significant differences in any of relation between number of shade trees and number of shade tree
the standard soil fertility parameters, except for slightly higher Ca va- species was much stronger (r = 0.91) than the correlation between
lues in full-sun plots (Table 2). CSSVD symptom severity was sig- number of shade tree species and Shannon index of diversity
nificantly lower under shade (by 56% on average, Fig. 1A). The quad- (r = 0.47). There were no significant correlations between CSSVD
ratic model showed a decrease of CSSVD symptom severity from full- symptom severity and any of the other factors: Shannon index of di-
sun conditions until an optimal shade level of 54% shade, after which versity, pH, Corg, Ntot, Pav, Mgex or Kex (dashed lines in Fig. 2).
88
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
Fig. 2. 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)–(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. Dashed lines indicate no statistical significance.
4. Discussion plots as high as 1906 μmol m−2 s−1 (Table 1), our data from full-sun
plots were above this critical value, which indicates that cocoa trees in
Since complete eradication and replanting of CSSVD-infected cocoa full-sun plots indeed may have suffered from high-light stress. This may
trees is not feasible, especially in areas of mass infection in the Eastern have led to low availability of photo-assimilates which suppresses
and Western Regions of Ghana, a coping strategy is urgently needed flower production, reduces yield (Asomaning et al., 1971) and tree
(Andres et al., 2017). Our data suggests that agroforestry systems with vigour, and leads to higher susceptibility towards attack by pests and
around 50% shade may be optimal to balance CSSVD symptom severity diseases (Owusu, 1980; Schroth et al., 2000; Wood and Lass, 2001).
versus reduced cocoa yield (dashed lines in Fig. 1), which may help These effects may be supported by the significant positive correlations
mitigating CSSVD severity until current cocoa is replaced with more between CSSVD symptom severity and mistletoe infestation as well as
resistant varieties. These results are in line with shade levels re- capsid damage we found in our study (see Section 3). These effects may
commended by existing manuals for good agronomic practices in cocoa partly explain the higher CSSVD symptom severity and lower cocoa
cultivation (Beer et al., 2004; SAN, 2005; Opoku-Ameyaw et al., 2010; yield we observed below the optimal shade levels of 54% and 39%,
Schroth et al., 2016). Our results from the regression analyses of po- respectively. Our results are in line with the ones by Legg (1982), who
tential causal factors for shade effects on CSSVD symptom severity in- observed virus symptoms to be particularly serious in cocoa trees under
dicated that shade per se was a relatively more important potential stress, and reported tolerance of virus infection to be at a maximum
causal factor for shade effects on CSSVD symptom severity, compared under shaded conditions.
to agro-biodiversity (measured by both Shannon index as well as shade However, excessive shading also leads to low availability of photo-
tree diversity) or soil fertility. In the following, we discuss the possible assimilates, which may have similar effects as the ones described in the
reasons for higher CSSVD symptom severity and lower cocoa yields in paragraph above (lower flower production, yield, tree vigour, and more
full-sun plots (i.e., the meaning of the term “shade per se”). pests and diseases Asomaning et al., 1971; Owusu, 1980; Schroth et al.,
2000; Wood and Lass, 2001). Hence, these effects may partly explain
why we observed an increase in CSSVD symptom severity and a de-
4.1. Possible reasons for higher CSSVD symptom severity and lower cocoa crease in cocoa yield above the optimal shade levels of 54% and 39%,
yields in full-sun plots respectively. The upper boundary of the 95% confidence interval we
reported for the optimal shade level with regard to CSSVD symptom
4.1.1. High-light and moisture stress severity (76% in Fig. 1A) is rather uncertain, because the Monte Carlo
Cocoa is a heliophobe preferring low levels of incident radiation simulations in this area of the curve were based on only one data point.
(Vinod, 2012). Studies have shown that exposure to light intensities Thus, more data on CSSVD symptom severity under heavy shading re-
higher than 1800 μmol m−2 s−1 induced high-light stress responses gimes above 50% are needed. Our results seem to be in line with the
(Lichtenthaler and Burkart, 1999) such as reduction in chlorophyll Ie- quadratic model proposed by Vernon (1967) to explain the relationship
vels and photosynthetic rate (Galyuon et al., 1996), or even damages to between cocoa yield and available light, and suggest that some degree
the leaf photosynthetic mechanisms of cocoa (Raja Harun and of shading may be desirable to maintain the productivity of CSSVD-
Hardwick, 1988). With a maximum measurement of infected cocoa aged 16 years and above. However, more data is needed
1983 μmol m−2 s−1 (data not shown) and mean values for individual
89
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
to confirm Vernon’s (1967) quadratic equation under conditions of (Andres et al., 2017).
CSSVD infection. Many studies have shown that natural control of pests and diseases
Besides direct high-light stress, the removal of shade trees may in- is enhanced in cocoa agroforests (Sperber et al., 2004; Tscharntke et al.,
directly cause moisture stress. Excessive leaf transpiration and in- 2011; Bieng et al., 2013; Gidoin et al., 2014; Vaast and Somarriba,
creased soil evaporation were reported to be major factors contributing 2014), and it is established that cocoa needs shade to remain healthy
to the yield decline some ten years after removal of shade trees in and productive (Owusu, 1980; Wood and Lass, 2001; Carr and
several trials in Ghana and elsewhere (Cunningham and Burridge, Lockwood, 2011). However, there is a knowledge gap about the effects
1960; Asomaning et al., 1971; Ahenkorah et al., 1974; Babin et al., of different shade tree species on CSSVD, and about suitable shade tree
2010; Oro et al., 2012). Thus, the integration of suitable shade trees species for effective agroforestry designs to combat CSSVD under dif-
species that are adequately pruned and managed by farmers is vital to ferent pedo-climatic and socio-economic conditions (Owusu, 1980;
maintain cocoa tree health by potentially reducing both high-light and Andres et al., 2016), which should be addressed by future research.
moisture stress. Despite agroforestry systems being a potential coping strategy to
mitigate CSSVD symptom severity, they may not be able to prevent
4.1.2. Soil fertility CSSVD spread. This is because the canopies of tall shade trees are likely
Several studies have shown that the loss of nutrients, and especially not to limit the movement of mealybug vectors that spread the disease
exchangeable bases (Owusu, 1980), in full-sun systems leads to yield at the forest floor (Jeger and Thresh, 1993). If the continuity of cocoa
decline (Cunningham and Burridge, 1960; Asomaning et al., 1971; plantations was broken with occasional strips (barriers) of non-host
Ahenkorah et al., 1974; Babin et al., 2010; Oro et al., 2012). Our crops such as citrus or oil palm, mealybug movement would likely be
findings seem to be in line with those results, as the soil in our plots was restricted and thus CSSVD spread reduced (Domfeh et al., 2016). Hence,
severely deficient in both Ca and Mg when compared to the minimum future research should investigate the effect of cocoa landscape con-
values proposed by Wessel (1971), which also means that the sig- nectivity on CSSVD spread.
nificant positive correlation of CSSVD symptom severity and Ca
(r = 0.14, p < 0.05) is of no ecological significance. However, since 5. Conclusion
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 In this paper, we address the need for more on-farm research about
observed effects on CSSVD symptom severity and cocoa yield. the potential of agroforestry systems to mitigate the severity of the
Another main result of our study is the lack of increased soil fertility cocoa swollen shoot virus disease (CSSVD) reported by Andres et al.
under shade trees compared to full-sun systems, which is in line with (2017). Here, we show that shade reduced CSSVD symptom severity
the results of others who showed no differences in soil carbon between and capsid damage in cocoa stands aged 16 years and above, which
shade and full-sun cocoa systems (Gockowski and Sonwa, 2011; Jacobi may have helped lowering the yield-reducing effect of CSSVD. High-
et al., 2014; Mohammed et al., 2016). Recently, Blaser et al. (2017) light and possibly also soil moisture stress were more likely to be ex-
reported limited benefits of shade trees for soil fertility in cocoa agro- planatory factors for those shade effects rather than soil fertility. Our
forests in Ghana. A common feature of our study and the one by Blaser results indicate that agroforestry systems with around 50% shade seem
et al. (2017) was that farmers did not prune shade tree canopies on the to be optimal to balance CSSVD symptom severity versus reduced cocoa
research plots of both studies. However, the results of both studies yield, and may thus offer an optimal coping strategy until diseased
contrast sharply with the experiences of agroforestry practitioners, who cocoa is replaced with more resistant varieties.
deliberately prune and thus increase the amount of biomass applied to
the soil considerably, which improved soil fertility significantly Funding sources
(Götsch, 1994; Fonte and Six, 2010). These contrasting results may
indicate that the limited amount of biomass recycled from deeper soil This work was supported by the E4D scholarship programme of ETH
layers to the top soil through mere litter fall (Buresh et al., 2004; Gama- Global, which is funded through the Sawiris Foundation for social de-
Rodrigues, 2011; Tscharntke et al., 2011) may have limited effects for velopment. The sponsors have no role in either study design, collection,
soil fertility in cocoa agroforests. Thus, shade tree pruning may be a analysis or interpretation of data, writing of the report and decision to
necessity to stimulate tree growth and increase organic matter input for submit the article for publication.
soil fertility enhancements in cocoa agroforests.
If soil fertility and soil moisture are favourable for vigorous crop Acknowledgments
growth, cocoa yields positively correlate with light intensity
(Cunningham and Burridge, 1960; Ahenkorah et al., 1987; Wood and We wish to express our gratitude to Dr Gilbert Anim-Kwapong
Lass, 2001; Babin et al., 2010; Oro et al., 2012). In poor soils with little (former Executive Director, CRIG) who through his continuous support
or no fertilization, however, higher yields are obtained under shade (de enabled successful data collection. Our sincere acknowledgements go to
Almeida and Valle, 2007; Vinod, 2012). Thus, with shade removal all the staff from CRIG and CHED who assisted us with both field and
comes the necessity of continuous fertilizer applications to maintain desktop work, especially Mr Clifford, Mr Enoch and Mr Ashitey (all
yields. Therefore, keeping, and especially managing shade trees may be CRIG), as well as Mr Nyarko, Mr Klu and Mr Dickson (all CHED).
an important strategy to maintain cocoa productivity under current Thanks are due to Dr Godfred Awudzi (CRIG) for his methodological
conditions of smallholder-grown cocoa in Ghana with rare or no ap- inputs to assess capsid damage and to Dr Amos Quaye (CRIG) for
plication of external nutrients. support in the soil analysis. We gratefully acknowledge the continuous
discussions and advice we received from Dr Kofi Acheampong (CRIG,
4.2. Avenues for future research retired). Thanks to Fränzi Korner for professional assistance with sta-
tistical analysis and data interpretation. Finally, we would like to ex-
In order to better understand the causal factors for the higher tend our gratitude to all the cocoa farmers who through their con-
CSSVD symptom severity and lower cocoa yield we observed under full- tinuous work enable us to advance in the development of sustainable
sun conditions, future research should i) measure moisture stress, and cocoa production systems.
ii) quantify viral load in plant tissue by qPCR analyses, and relate it to
CSSVD symptom severity, taking into account the size of infectious Appendix A. Supplementary data
mealybug populations in diseased fields. In addition, shade should be
incorporated into ongoing breeding programs for CSSVD resistance Supplementary data associated with this article can be found, in the
90
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
online version, at https://doi.org/10.1016/j.agee.2017.09.031. structure and pod production explain frosty pod rot intensity in cacao agroforests,
Costa Rica. Phytopathology 104, 275–281.
Gockowski, J., Sonwa, D., 2011. Cocoa intensification scenarios and their predicted im-
References pact on CO(2) emissions biodiversity conservation, and rural livelihoods in the
Guinea rain forest of West Africa. Environ. Manag. 48, 307–321.
Hawthorne, W., Gyakari, N., 2006. Photoguide for the Forest Trees of Ghana: A Tree-
Ahenkorah, Y., Akrofi, G.S., Adri, A.K., 1974. End of first cocoa shade and manurial ex- spotter's Field Guide for Identifying the Largest Trees. Oxford Forestry Institute,
periment at Cocoa Research Institute of Ghana. J. Hortic. Sci. Biotechnol. 49, 43–51. Oxford, UK.
Ahenkorah, Y., Halm, B.J., Appiah, M.R., Akrofi, G.S., Yirenkyi, J.E.K., 1987. 20 years Jacobi, J., Andres, C., Schneider, M., Pillco, M., Calizaya, P., Rist, S., 2014. Carbon stocks,
results from a shade and fertilizer trial on Amazon cocoa (Theobroma cacao) in Ghana. tree diversity, and the role of organic certification in different cocoa production
Exp. Agric. 23, 31–39. systems in Alto Beni, Bolivia. Agrofor. Syst. 88, 1117–1132.
Ameyaw, G.A., Dzahini-Obiatey, H.K., Domfeh, O., 2014. Perspectives on cocoa swollen Jaramillo, J., Chabi-Olaye, A., Kamonjo, C., Jaramillo, A., Vega, F.E., Poehling, H.-M.,
shoot virus disease (CSSVD) management in Ghana. Crop Prot. 65, 64–70. Borgemeister, C., 2009. Thermal tolerance of the coffee berry borer Hypothenemus
Andres, C., Bhullar, G.S., 2016. Sustainable intensification of tropical agro-ecosystems: hampei: predictions of climate change impact on a tropical insect pest. PLoS One 4.
need and potentials. Front. Environ. Sci. 4. Jeger, M.J., Thresh, J.M., 1993. Modeling reinfection of replanted cocoa by swollen shoot
Andres, C., Comoé, H., Beerli, A., Schneider, M., Rist, S., Jacobi, J., 2016. Cocoa in virus in pandemically diseased areas. J. Appl. Ecol. 30, 187–196.
monoculture and dynamic agroforestry. In: Lichtfouse, E. (Ed.), Sustainable Kindt, R., Coe, R., 2005. Tree Diversity Analysis. A Manual and Software for Common
Agriculture Reviews 19. Springer International Publishing, Switzerland, pp. 121–153. Statistical Methods for Ecological and Biodiversity Studies. World Agroforestry
Andres, C., Gattinger, A., Dzahini-Obiatey, H.K., Blaser, W.J., Offei, S.K., Six, J., 2017. Centre (ICRAF), Nairobi.
Combatting cocoa swollen shoot virus disease: what do we know? Crop Prot. 98, Legg, J.T., 1982. The Cocoa Swollen Shoot Research Project at the Cocoa Research
76–84. Institute, Tafo, Ghana, 1969–78. Overseas Development Administration, London.
Armengot, L., Barbieri, P., Andres, C., Milz, J., Schneider, M., 2016. Cacao agroforestry Lichtenthaler, H.K., Burkart, S., 1999. Photosynthesis and high light stress. Bulg. J. Plant
systems have higher return on labor compared to full-sun monocultures. Agron. Physiol. 25, 3–16.
Sustain. Dev. 36, 70. Mani, M., Shivaraju, C., 2016. Mealybugs and Their Management in Agricultural and
Asomaning, E.J.A., Kwakwa, R.S., Hutcheon, W.V., 1971. Physiological studies on an Horticultural Crops. Springer India, New Delhi.
Amazon shade and fertilizer trial at the Cocoa Research Institute of Ghana. Ghana J. Mohammed, A.M., Robinson, J.S., Midmore, D., Verhoef, A., 2016. Carbon storage in
Agric. Sci. 4, 47–64. Ghanaian cocoa ecosystems. Carbon Balance Manag. 11, 1–6.
Awudzi, G.K., 2014. Development of an Information System for Mirid Attack on Cocoa Muller, E., Jacquot, E., Yot, P., 2001. Early detection of cacao swollen shoot virus using
Based on the Crop Status and Mirid Population Monitoring. The School of the polymerase chain reaction. J. Virol. Methods 93, 15–22.
Agriculture, Policy and Development, University of Reading, Reading, UK PhD thesis: Nakagawa, S., Schielzeth, H., 2013. A general and simple method for obtaining R2 from
209. generalized linear mixed-effects models. Methods Ecol. Evol. 4, 133–142.
Babin, R., Ten Hoopen, G.M., Cilas, C., Enjalric, F., Yédé Gendre, P., Lumaret, J.P., 2010. Opoku, I.Y., Akrofi, A.Y., Appiah, A.A., 2002. Shade trees are alternative hosts of the
Impact of shade on the spatial distribution of Sahlbergella singularis in traditional cocoa pathogen Phytophthora megakarya. Crop Prot. 21, 629–634.
cocoa agroforests. Agric. For. Entomol. 12, 69–79. Opoku-Ameyaw, K., Baah, F., Gyedu-Akoto, E., Anchirinah, V., Dzahini-Obiatey, H.K.,
Bates, D., Maechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models Cudjoe, A.R., Aquaye, S., Opoku, S.Y., 2010. Cocoa Manual – A Source Book for
using lme4. J. Stat. Softw. 67, 1–48. Sustainable Cocoa Production. Cocoa Research Institute of Ghana, New Tafo-Akim.
Beer, J., Muschler, R., Kass, D., Somarriba, E., 1998. Shade management in coffee and Oro, Z.F., Bonnot, F., Ngo Bieng, M.A., Delaitre, E., Dufour, B.P., Ametefe, K.E., Mississo,
cacao plantations. Agrofor. Syst. 38, 139–164. E., Wegbe, K., Muller, E., Cilas, C., 2012. Spatiotemporal pattern analysis of Cacao
Beer, J., Ibrahim, M., Somarriba, E., Barrance, A., Leakey, R., 2004. Establecimiento y swollen shoot virus in experimental plots in Togo. Plant Pathol. 61, 1043–1051.
manejo de árboles en sistemas agroforestales. In: Cordero, J., Boshier, D. (Eds.), Owusu, J.K., 1980. Light requirements of cocoa: a review. In: Proceedings of the
Árboles de Centroamérica: un manual para extensionistas. Oxford Forestry Institute International Conference on Cocoa and Coconuts 1978. Kuala Lumpur, Malaysia. pp.
(OFI), Centro Agronómico Tropical de Investigación y Enseñanza (CATIE), Oxford, 112–121.
UK Costa Rica, pp. 197–242. Padi, F.K., Domfeh, O., Takrama, J., Opoku, S., 2013. An evaluation of gains in breeding
Beer, J., 1987. Advantages, disadvantages and desirable characteristics of shade trees for for resistance to the cocoa swollen shoot virus disease in Ghana. Crop Prot. 51,
coffee cacao and tea. Agrofor. Syst. 5, 3–13. 24–31.
Bieng, M.A.N., Gidoin, C., Avelino, J., Cilas, C., Deheuvels, O., Wery, J., 2013. Diversity Posnette, A.F., Robertson, N.F., Todd, J.M., 1950. Virus diseases of cacao in West Africa:
and spatial clustering of shade trees affect cacao yield and pathogen pressure in Costa v. Alternative host plants. Ann. Appl. Biol. 37, 229–240.
Rican agroforests. Basic Appl. Ecol. 14, 329–336. R Core Team, 2017. R: A Language and Environment for Statistical Computing. R
Bigger, M., 1981. Observations on the insect fauna of shaded and unshaded Amelonado Foundation for Statistical Computing, Vienna. https://www.r-project.org/ (Accessed
cocoa. Bull. Entomol. Res. 71, 107–119. 26.06.2017).
Blaser, W.J., Oppong, J., Yeboah, E., Six, J., 2017. Shade trees have limited benefits for Raja Harun, R.M., Hardwick, K., 1988. The effects of prolonged exposure to different light
soil fertility in cocoa agroforests. Agric. Ecosyst. Environ. 243, 83–91. intensities on the photosynthesis of cocoa leaves. In: Proceedings of the Tenth
Bos, M.M., Steffan-Dewenter, I., Tscharntke, T., 2007. Shade tree management affects International Cocoa Research Conference 1988. Santo Domingo, Dominican
fruit abortion, insect pests and pathogens of cacao. Agric. Ecosyst. Environ. 120, Republic. pp. 205–209.
201–205. Rice, R.A., Greenberg, R., 2000. Cacao cultivation and the conservation of biological
Buresh, R.J., Rowe, E.C., Livesley, S.J., Cadisch, G., Mafongoya, P., 2004. Opportunities diversity. Ambio 29, 167–173.
for capture of deep soil nutrients. In: van Noordwijk, M., Cadisch, G., Ong, C.K. SAN, 2005. Additional Criteria and Indicators for Cocoa Production. Sustainable
(Eds.), Below-ground Interactions in Tropical Agroecosystems: Concepts and Models Agriculture Network, Rainforest Alliance. http://www.worldagroforestry.org/
with Multiple Plant Components. CABI Publishing, Wallingford, pp. 109–123. treesandmarkets/inaforesta/documents/agrof_cons_biodiv/criteria%20and
Carr, M.K.V., Lockwood, G., 2011. The water relations and irrigation requirements of %20indicators%20for%20cocoa_2005.pdf (Accessed 26.06.2017).
cocoa (Theobroma cacao L.): a review. Exp. Agric. 47, 653–676. Savary, S., Ficke, A., Aubertot, J.N., Hollier, C., 2012. Crop losses due to diseases and
Cunningham, R.K., Burridge, J.C., 1960. The growth of cacao (Theobroma cacao) with and their implications for global food production losses and food security. Food Secur. 4,
without shade. Ann. Bot. 24, 458–462. 519–537.
Danso-Abbeam, G., 2014. Modelling farmers investment in agrochemicals. The experi- Schneider, M., Andres, C., Trujillo, G., Alcon, F., Amurrio, P., Perez, E., Weibel, F., Milz,
ence of smallholder cocoa farmers in Ghana. Res. Appl. Econ. 6, 1–16. J., 2016. Cocoa and total system yields of organic and conventional agroforestry vs
de Almeida, A.-A.F., Valle, R.R., 2007. Ecophysiology of the cacao tree. Braz. J. Plant monoculture systems in a long-term field trial in Bolivia. Exp. Agric. 53, 351–374.
Physiol. 19, 425–448. Schroth, G., Krauss, U., Gasparotto, L., Aguilar, J.A.D., Vohland, K., 2000. Pests and
Domfeh, O., Ameyaw, G.A., Dzahini-Obiatey, H.K., Ollennu, L.A.A., Osei-Bonsu, K., diseases in agroforestry systems of the humid tropics. Agrofor. Syst. 50, 199–241.
Acheampong, K., Aneani, F., Owusu-Ansah, F., 2016. Use of immune crops as barrier Schroth, G., Jeusset, A., Gomes, A.D., Florence, C.T., Coelho, N.A.P., Faria, D., Laderach,
in the management of cacao swollen shoot virus disease (CSSVD) – long-term as- P., 2016. Climate friendliness of cocoa agroforests is compatible with productivity
sessment. Plant Dis. 100, 1889–1893. increase. Mitig. Adapt. Strateg. Glob. Change 21, 67–80.
Dzahini-Obiatey, H.K., Domfeh, O., Amoah, F.M., 2010. Over seventy years of a viral Shannon, C.E., Weaver, W., 1949. The Mathematical Theory of Communication, fourth
disease of cocoa in ghana: from researchers' perspective. Afr. J. Agric. Res. 5, ed. The University of Illinois Press, Urbana.
476–485. Somarriba, E., 2002. Estimación visual de la sombra en cacaotales y cafetales. Agrofor.
Fonte, S.J., Six, J., 2010. Earthworms and litter management contributions to ecosystem Am. 9, 86–94.
services in a tropical agroforestry system. Ecol. Appl. 20, 1061–1073. Sperber, C.F., Nakayama, K., Valverde, M.J., Neves, F.D., 2004. Tree species richness and
Götsch, E., 1994. Break-through in Agriculture. Agricultura familiar e agroecologia density affect parasitoid diversity in cacao agroforestry. Basic Appl. Ecol. 5, 241–251.
(ASPTA), Rio de Janeiro. Thorlakson, T., Neufeldt, H., 2012. Reducing subsistence farmers' vulnerability to climate
Galyuon, I.K.A., McDavid, C.R., Lopez, F.B., Spence, J.A., 1996. The effect of irradiance change: evaluating the potential contributions of agroforestry in western Kenya.
level on cocoa (Theobroma cacao L.). 2. Gas exchange and chlorophyll fluorescence. Agric. Food Secur. 1, 1–13.
Trop. Agric. 73, 29–33. Thresh, J.M., Owusu, G.K., Ollennu, L.A., 1988. Cocoa swollen shoot: an archetypal
Gama-Rodrigues, A.C., 2011. Soil organic matter, nutrient cycling and biological dini- crowd disease. Z. Pflanzenkrankh. Pflanzenschutz 95, 428–446.
trogen-fixation in agroforestry systems. Agrofor. Syst. 81, 191–193. Tscharntke, T., Clough, Y., Bhagwat, S.A., Buchori, D., Faust, H., Hertel, D., Hölscher, D.,
Ghana Statistical Service, 2014. District Analytical Report – New Juaben Municipal, Juhrbandt, J., Kessler, M., Perfecto, I., Scherber, C., Schroth, G., Veldkamp, E.,
Population &Housing Census 2010. . http://www.statsghana.gov.gh/docfiles/2010_ Wanger, T.C., 2011. Multifunctional shade-tree management in tropical agroforestry
District_Report/Eastern/NEW%20JUABEN.pdf (Accessed 26.06.2017). landscapes – a review. J. Appl. Ecol. 48, 619–629.
Gidoin, C., Avelino, J., Deheuvels, O., Cilas, C., Bieng, M.A.N., 2014. Shade tree spatial
91
C. Andres et al. Agriculture, Ecosystems and Environment 252 (2018) 83–92
Vaast, P., Somarriba, E., 2014. Trade-offs between crop intensification and ecosystem Venkateswarlu, B., Shanker, A.K., Shanker, C., Maheswari, M. (Eds.), Crop Stress and
services: the role of agroforestry in cocoa cultivation. Agrofor. Syst. 88, 947–956. Its Management: Perspectives and Strategies. Springer Netherlands, Dordrecht, pp.
van Reeuwijk, L.P., 2002. Procedures for soil analysis. Technical Paper 9, sixth edition. 45–137.
International Soil Reference and Information Centre, Wageningen. Wessel, M., 1971. Fertilizer Requirements of Cacao (Theobroma Cacao L.) in South-
Vernon, A.J., 1967. Yield and light relationships in cocoa. Trop. Agric. 44, 223–228. Western Nigeria. Koninklijk Instituut voor de Tropen, Amsterdam.
Vinod, K.K., 2012. Stress in plantation crops: adaptation and management. In: Wood, G.A.R., Lass, R.A., 2001. Cocoa, fourth edition. Blackwell Science, Oxford.
92