Geoderma Regional 37 (2024) e00772

Contents lists available at ScienceDirect

Geoderma Regional
journal homepage: www.elsevier.com/locate/geodrs

Biological soil health indicators are sensitive to shade tree management in a

young cacao (Theobroma cacao L.) production system
Anna M. Visscher a, *, Eduardo Chavez b, Carlos Caicedo c, Leider Tinoco c, Mirjam Pulleman a, d
a
Soil Biology Group, Wageningen University, P.O. Box 47, 6700 AA Wageningen, the Netherlands
b
Escuela Superior Politécnica del Litoral, ESPOL, Facultad de Ciencias de la Vida, Campus Gustavo Galindo Km. 30.5 Vía Perimetral, P.O. Box 09-01-5863, Guayaquil,
Ecuador
c
Central Experimental Station of the Amazon of INIAP (National Agricultural and Livestock Farming Research Institute), Ecuador
d
International Center for Tropical Agriculture (CIAT), AA6713, 763537 Cali, Colombia

A R T I C L E I N F O A B S T R A C T

Keywords: Cacao agroforestry systems can offer important benefits, such as greenhouse gas mitigation, microclimate
Alfisol regulation and improved soil health. The selection of tree species for cacao agroforestry systems is a critical step
Amazon impacting cacao yields, as well as the environmental and economic sustainability of the production system.
Cadmium
However, the effects of different tree species on soil processes and functions have been poorly studied. We
Ecuador
assessed a series of soil health indicators in a five-year-old agroforestry trial located in the Ecuadorian Amazon.
Nutrient cycling
Soil carbon The following treatments: “control” (cacao monoculture), “timber” (cacao with Cedrelinga cateniformis Ducke; a
Soil macrofauna leguminous tree), “fruit” (cacao with Bactris gasipaes), “N-fix” (cacao with Erythrina velutina Wild) and “mixed”
Soil respiration (cacao with C. cateniformis + E. velutina) were replicated 3 times in randomized blocks. The experiment was
managed organically with low levels of external inputs. We collected soil, litter and leaf samples of cacao trees at
two different distances (~2 m and ~ 6 m) from the shade tree. Biological (potential soil respiration, and mac­
rofauna abundance, richness and Shannon diversity), chemical (pH, CEC, total C and N, macro and micro­
nutrients and Cd), and physical (bulk density and water holding capacity) soil health indicators were measured.
Additionally, we analysed nutrients and Cd in cacao leaves and litter. Results showed positive effects of the
“timber” and “mixed” treatments on biological soil health indicators, primarily earthworm abundance and po­
tential respiration, as compared to cacao monocultures. Treatment effects on potential respiration showed
different trends in litter versus soil, indicating that effects of shade trees on litter respiration do not transfer
directly to soil respiration. Physical and chemical soil indicators, including available Cd which is irrelevant in the
context of food safety regulations, did not show any differences among the treatments. Five years after estab­
lishment, no significant differences in yields were found among the control and any of the shade tree treatments.

1. Introduction cacao production practices (Franzen and Mulder, 2007). Widespread

impoverishment of soil health (Giller et al., 1997), is threatening cacao
Cacao (Theobroma cacao L.) is an important global commodity, supply, the wellbeing of smallholder cacao producers and contribute to
grown mainly by smallholder farmers throughout the tropics. Most deforestation and biodiversity loss (Tscharntke et al., 2005; Pretty,
cacao is produced in West Africa and Ecuador is the largest cacao pro­ 1997).
ducer in Latin America. Despite growing market demand, global total Theobroma cacao is an understory tree species native to humid
cacao production is declining (Fountain and Hütz-Adams, 2022). tropical regions of northern South America, and according to some re­
Numerous factors have contributed to this trend, declining soil fertility, ports, central America (Motamayor et al., 2002). Most cacao cultivation
increases in pests and diseases and climate change are commonly re­ has traditionally taken place in agroforestry systems in the lower story of
ported (Richard and Ræbild, 2016; Clough et al., 2009). Furthermore, the evergreen forest (Boyer, 1973). Agroforestry management involves
social factors such as low profit for smallholder farmers and shortage of the combination of different vegetation strata and species (including
labour and technical assistance limit farmers’ engagement in sustainable shade trees) and is gaining advocacy in the context of global crises such

* Corresponding author.
E-mail address: [email protected] (A.M. Visscher).

https://doi.org/10.1016/j.geodrs.2024.e00772
Received 15 October 2023; Received in revised form 23 January 2024; Accepted 27 January 2024
Available online 29 January 2024
2352-0094/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

as climate mitigation, adaptation and biodiversity loss. Agroforests can microbial parameters. This study highlighted the importance of inte­
also contribute to the restoration and maintenance of soil health in cacao grating trees that provide quality litter for enhanced cocoa nutrient
production systems (Jose, 2009). In the second half of the 20th century availability and accelerated soil carbon storage. Further supporting this
monocrop plantations have been introduced as the main production approach, a study by Koné and Yao (2021) in Ivory Coast demonstrated
model for cacao. However, nowadays cacao AFS receive renewed in­ that mixed tree species stands, like agroforestry systems, can parallel
terest (Blaser et al., 2018; Niether et al., 2020). Shading is especially natural forests in maintaining soil organic carbon stocks and microbial
crucial in the establishment phase of cacao production (de Almeida and activity, underscoring their potential in the development of sustainable
Valle, 2007), but higher yields have been reported when shade is pro­ timber systems in West and Central Africa. These studies collectively
gressively removed during production plateau (Ofori-Frimpong et al., emphasize the utility of biological indicators in detecting variations in
2007; Blaser et al., 2017) as cacao trees have a high self-shading capacity soil health due to agroforestry, meanwhile highlighting a pivotal shift
once mature (Hartemink, 2005). Nonetheless, agroforestry systems are towards more sustainable cocoa farming practices that prioritize both
nowadays increasingly promoted by supply chain actors to improve the crop yield and environmental conservation.. However, these studies
sustainability of cacao production systems (Blaser et al., 2018; Niether place less emphasis on the impact of shade tree arrangement and young
et al., 2020). The selection of shade tree species for cacao agroforestry agroforestry trials on soil health, leaving a knowledge gap.
systems is a critical step determining the performance of the cacao Our study was aimed at quantifying the effect of different shade tree
production system but has been poorly studied, especially when it comes species and species arrangements on soil health and its relations to cacao
to effects on soil health (Niether et al., 2020). Both Steffan-Dewenter production in a young plantation situated in the Ecuadorian Amazon.
et al. (2007) and Niether et al. (2020) highlight that the effects of shade Special attention was placed on the role of different shade tree species
trees are highly context-dependent and influenced by species’ specific (representing legume, fruit or timber trees, or combinations) and effect
traits that affect interactions with the main crop due to resource of distance of the shade trees in affecting: i) litter composition and soil
competition and complementary. Therefore, although shade trees offer a health at different distance from the shade tree in cacao agroforestry
number of key services, little is known about the effects of the presence systems, ii) the spatial tree diversity within the productions systems and
and species of shade trees in cacao cultivation on soil functions and the iii) indicators with relevance for the following soil functions were
net impacts on crop productivity (Marshall, 2004; Niether et al., 2020). prioritized in this context: nutrient cycling, carbon concentration,
The term soil health refers to “the continued capacity of a soil to contaminant mitigation and cacao pod yield. The objective of this study
function as a vital living system, within ecosystem and land-use boundaries, was to evaluate a set of soil health indicators, that include biological,
to sustain plants, animals and humans” (Doran and Zeiss, 2000). A healthy chemical, and physical aspects, and are linked to one or more soil
soil supports multiple soil functions: carbon cycling and storage, functions in cacao agroecosystems. Another key objective was to
nutrient cycling, contaminant mitigation, pest control, water regulation determine which indicators are sensitive to changes in management, in
and plant production (Creamer et al., 2022). Soil health is an abstraction particular different shade tree arrangements in a young agroforestry
of a holistic concept that cannot be measured directly (Janzen et al., system, and to assess whether the proximity to these shade trees in­
2021). Instead, we assess soil health based on sets of indicators that are fluences these indicators. We hypothesized that: i) the early effects of
proxies of soil processes relevant to the abovementioned soil functions. shade tree species on key soil functions are most prominently captured
Indicator sets and assessments allow monitoring of soil health within by soil biological indicators; ii) the inclusion of shade trees, especially
agroecosystems, in response to different management strategies, and leguminous trees, in a low–input organic cacao production system has a
inform interventions to be implemented to restore and maintain soil positive impact on cacao yields; iii) soil health indicator values change
health (Bünemann et al., 2018). Soil health integrates biology, physics, with distance from the shade trees and this effect can be related to dif­
and chemistry, therefore any indicator set aimed at assessing soil health ferences in shading, litter quality and quantity.
should ideally reflect all three aspects (Creamer et al., 2022). Most in­
dicator schemes to assess soil health focus on temperate climates and 2. Materials and methods
annual cropping systems or pastures, yet few existing approaches for
monitoring soil health have been applied to cacao (or other tropical 2.1. Site description
perennial crops) rising to contextualise them for applicability in these
production systems (Amponsah-Doku et al., 2022). Tropical perennial The study was carried out in the Northeastern part of the Ecuadorian
systems, such as cacao, distinctively rely on the substantial recycling of Amazon region in an existing replicated field experiment established in
tree biomass into surface litter, which is crucial for maintaining soil 2015 at the Central Amazon Research Center (ECCA) of the Instituto
health by enhancing organic matter content and nutrient cycling Nacional de Investigaciones Agropecuarias (INIAP). The trial is located
(Asigbaase et al., 2021). at Joya de los Sachas in the province of Orellana (0◦ 21′31.9”S,
Over the past decade, a number of studies have assessed biological 76◦ 51′51.1”W; 265 m asl.). Mean annual temperature is 24.0 ◦ C with
aspects of soil health in cacao plantations, expanding beyond traditional variation of 1.3 ◦ C and a mean annual precipitation of 3217 mm, cor­
chemical and physical indicators. For example, Rousseau et al. (2012) responding to an Köppen-Geiger climate classification of tropical wet
explored soil macrofauna assemblages as a soil health indicator within (Aw). The soil at the site has been classified as an Alfisol, according to
cacao agroforestry systems in Costa Rica. Alfaro-Flores et al. (2015) USDA Soil Taxonomy (López et al., 2021). The soil texture was char­
reported that in Bolivia’s “Alto Beni” region, agroforestry and fallow acterized as a clay loam (22.7% sand, 37.5% clay) in the top layer of the
plots exhibit higher soil microbial biomass than monoculture cocoa A horizon, changing to silty clay (15.5% sand, 42.3% clay) in the AB
farms, with no significant difference between organic and conventional horizon (Table 1). The soil was slightly acidic, with an average pH of 5.4
management. In Peru, Buyer et al. (2017) found that soil microbial and SOM content of 87.6 g kg− 1 in the A horizon. The trial includes four
communities in cacao plantations are significantly affected by agrofor­ different treatments (low input organic, high input organic, high input
estry management and cover cropping. Natural agroforestry systems conventional and medium input conventional). The current study only
altered less the microbial community structures compared to soils in focused on the low-input organic treatment. Organic inputs for this
slash-and-burn methods, and specific cover crops, like centro (Cen­ treatment are mostly derived from internal resources such as litter fall
trosema macrocarpum) and perennial peanut (Arachis pintoi), were found and prunings from cacao and shade trees, which is representative of
to alter bacterial ratios and fungal biomass. Yao et al. (2021) conducted smallholder practices in the area.
a study in West Africa assessing carbon and macronutrient cycling in The shade treatments were established in a domized Block Design (n
cocoa and teak (Tectona grandis) plantations using indicators such as leaf = 3) in 2015, in an area that was previously under oil palm production.
litterfall, carbon and macronutrient inputs, and soil chemical and All shade trees were recently planted and carefully chosen for their

2
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Table 1 newly planted at a density of 3 × 3 m. Net plots contained 6 cacao plants

Data from a soil profile at the INIAP cocoa agroforestry trial in the Ecuadorian and were surrounded by 3 rows of cacao plants as borders (Fig. 1 and 2).
amazon. Field work in this regard was performed by dr. E. Chavez (ESPOL) and The cacao cultivars in the trial are two national clones EET 95 and EET
dr. A. Margenot (University of Illinois) in 2018 at the same research site. 103, alternated in strips (Fig. S1). Differences between cacao genotype
Horizon Top Bottom Thickness Sand Clay Silt USDA textural were not determined in this study. The cacao trees received 0.5 kg
depth depth class Ecoabonaza” (organic fertilizer with chicken manure) and 0.35 kg lime
(cm) (cm) (cm) (%) (%) (%) per tree before planting. No pesticides were applied. Pruning of the
A 0 8 8 22.7 37.5 39.8 clay loam
cacao trees is carried out biannually, with the clippings subsequently
AB 8 13 5 15.5 42.3 42.3 silty clay redistributed across the soil. Shade trees were pruned once per year and
Bt 13 20.5 7.5 16.5 58 25.5 clay weed was controlled with a weed wacker monthly. Harvest took place
Bt 20.5 28 7.5 18.3 53.7 28 clay manually once per month.
Bt 28 49 21 15.6 58.6 25.8 clay
Bt 49 62 13 12.4 64.6 23.1 clay
Bt 62 103 41 15.6 66.2 18.2 clay 2.2. Selection of soil health indicators
C 103 165 62 62 22.3 15.6 sandy clay loam
C 165 170 5 33.1 36.9 30 clay loam
The selection of indicators was driven by both practical and scientific
C 170 177 7 55.1 18 26.9 sandy loam
C 177 215 38 81.5 16.2 2.3 sandy loam considerations as the predictor should be meaningful, robust, inter­
pretable, sensitive, low-cost and relatively easy to measure considering
local field and laboratory conditions. We identified indicators that
functional significance in the local context, including fruit production, included at least one or more biological, chemical, or physical soil in­
nitrogen fixation, and wood production. The shade treatments consist of dicators, with a particular emphasis on those that provide information
four agroforestry systems: “timber” (Cedrelinga cateniformis Ducke; a about the following soil functions: nutrient cycling, carbon concentra­
leguminous timber tree also/locally known as tornillo or cedrorana), tion, contaminant mitigation (cadmium) and primary production. We
“fruit” (Bactris gasipaes; peach palm), “N-fix” (Erythrina velutina Wild; a conducted measurements of the biological indicators (potential soil and
widely used leguminous shade tree) and “mixed” (C. cateniformis + litter respiration, total macrofauna and earthworm abundance, and
E. velutina). A fifth “control” treatment is included (monoculture without taxonomic richness and Shannon diversity of macrofauna); chemical
shade; MONO). Shade tree spacing in the agroforestry systems was 12 × indicators (pH, cation exchange capacity (CEC), total soil organic carbon
12 m in timber and fruit, and 6 × 6 m in Erythrina. Cacao trees were (SOC), total nitrogen (N), macro and micronutrients and cadmium (Cd));

Fig. 1. Visual representation of the experimental trial in northern the Ecuadorian Amazon, which aimed to evaluate the performance of different shade tree ar­
rangements on predictors of soil health and cacao productivity. The satellite image indicates the location of the experimental sites where the agroforestry systems
were amended traditionally, while the tree icons indicate the type of agroforestry system within the sampled plot. The treatments consist of five agroforestry systems
which are distinguished as: “monoculture” (without shade trees; control), “timber” (Cedrelinga cateniformis Ducke; a leguminous timber tree also/locally known as
tornillo or cedrorana), “fruit” (Bactris gasipaes; peach palm), “N-fix” (Erythrina velutina Wild; a widely used leguminous shade tree) and “mixed” (C. cateniformis
+ E. velutina).

3
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Fig. 2. Layout of the sampling strategies in the different agroforestry treatment.

and physical indicators (bulk density and water holding capacity). moist soil was stored in a cooler with ice for transport to the laboratory.
Furthermore, we assessed nutrient and trace metal levels in cacao leaves The remaining soil, leaf and litter samples, were transported at ambient
and litter. temperature. Pod yield data for the years 2017, 2018, 2019 were
collected as follows: Pods were collected once every month during the
period January–September for 9 out of 144 cacao trees per plot. Fresh
2.3. Soil, plant sampling, and yield data collection pod weight in kg per plant was recorded. Shade cover in the different
shade trees (at two distances) was estimated using a spherical densi­
Soil, litter and leaf samples were collected in October 2019. In total, tometer. Readings took place in the week of 21–26 October 2019, always
15 plots were sampled (5 shade tree treatments x 3 replicate blocks). To by the same person to avoid bias. The device was held level above the
account for the effects of distance from the shade trees on litter and soil cacao tree and the number of uncovered blocks on the mirror were
characteristics, samples were taken around the cacao trees at two counted. On each location four readings were made facing north, east,
different locations within the net plot (~2 m and ~ 6 m from the shade south, and west. Later, the average was calculated to determine over­
tree). In the monoculture “control” plot, samples were taken in the story density (%).
center of the net plot (Fig. 2). Soil and litter samples were taken on
opposite sides of the selected cacao tree (in an angle of 180◦ ). Sampling
was conducted within the rooting zone of the tree, at 60 cm from the 2.4. Measurement of biological soil health indicators
trunk, as closer to the trunk of the cacao tree roots are not abundantly.
First, surface litter was collected using a 20.3 cm diameter circular Soil macrofauna community abundance and diversity were assessed
sampling frame after removal of living biomass. Then, the underlying using the Tropical Soil Biology and Fertility (TSBF) method (Anderson
soil was collected at depths of 0–7.5 cm and 7.5–15 using a spade. Litter and Ingram, 1993), with a modified sampling design considering small
and soil samples from the same tree were individually combined, experimental plots. A soil monolith (25 × 25 × 20 cm deep) was exca­
yielding one composite soil sample and one composite litter sample per vated with a flat shovel adjacent to each sampled tree (two per plot) and
layer for each cacao tree. Additionally, the second pair of mature leaves put in a bag (Fig. 2). All macrofauna visible with the naked eye (>2 mm)
on a flowering branch was collected in four cardinal directions, to obtain was hand-sorted in a nearby barn and stored in 70% ethanol (or 4%
a total of 8 leaves per plant (2 leaf samples per plot) for foliar analysis. formalin for earthworms). Blocks 1 and 2 of the trial had extremely high
Immediately after sampling, field moist soil samples from the first soil densities of ants dominated by the species Nylanderia fulva; commonly
depth were homogenized and split into two subsamples. About 250 g of known as “crazy ants”. Therefore, for practical reasons, we collected

4
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

100 ants from each monolith with >100 ants present, which refers to an copper (Cu), iron (Fe), cadmium (Cd). Total soil carbon (TSC, also
equivalent of >1000 (ind. m− 2) . All macrofauna was transported in a known as TOC) and total soil nitrogen (TSN) were quantified using dry
cooler to the lab. Specimens from each sampling point were identified, combustion. This process involved using 150 mg of dry soil, which was
generally, to the order level (depending on the taxonomic group), for homogenized through a 250 μm sieve to create a uniform and testable
calculation of diversity indices and further analyses. Macrofauna sample. We employed a CN Analyzer (Elementar Analyzer, Germany) for
abundance and diversity were expressed as taxonomic richness (S, the analysis.
number of distinct taxonomic groups) and Shannon Index (H; Shannon,
1948). 2.6. Litter and leaf characterization
Potential soil respiration (CO2 evolution during incubation under
standardized conditions, homogenized soil sample, optimal tempera­ The total weight of the litter samples collected in the field was
ture, and moisture) is used as indicator for microbial activity. This in­ measured when still moist and corrected for moisture content (see
dicator was measured following a modified method from Hess and below) to calculate dry litter mass per m2 present in the field at the time
Schmidt (1995). Air-dried soil samples from the first soil layer were of sampling. One part of the moist litter sample was directly stored in the
homogenized by passing 50 g of soil through a 2 mm sieve for removal of fridge for 30 d until further analysis. Potential litter respiration was
debris and large particles. The moisture content of dry soil was then measured following the same protocol mentioned for soil samples
gradually adjusted to 60% of the water holding capacity (determined by (Section 2.4), but adjusted for fresh organic material using the method
adding water to an approx. 50 g dry soil sample with a plant sprayer described in Pulleman and Tietema (1999). A fixed amount of moist
until the soil appeared shimmering and water was noticeably accumu­ litter, corresponding to 4.4 g of dry weight, was brought to the desired
lated in the openings). The re-wetted soil was subsequently transferred initial gravimetric water content of 272% by adding demineralized
to polyethylene bottles (300 mL), sealed with parafilm and stored in a water using a plant sprayer. Measurements of respiration lasted 10 min
dark place at constant temperature (23 ◦ C). The bottles were opened for as described above for soil respiration measurements.
10 min prior to each measurement to refresh the headspace. A Go The cacao leaves were cleaned in a bath of deionized water prior to
Direct® CO2 Gas Sensor was used to determine increase in gaseous CO2 drying and processing. The remaining part of the litter sample and the
concentration in the headspace (ppm CO2s1) of the bottles. Measure­ cacao leaf samples were oven-dried at 60 ◦ C until stable weight. Dry
ments were repeated at 24, 48, 72 and 96 h time intervals and lasted for litter and leaf samples were ground and sieved (850 μm) before deter­
10 min. To avoid the disturbance effect due to rewetting, we excluded mination of total element concentrations. In short, 350 mg of tissue was
the measurements taken after 24 and 48 h from the data set and focused weighed in a 50 mL reaction tube. Then, samples were digested with 6
on respiration measurements once the CO2 curve had stabilized mL of concentrated HNO3 and 1 mL of H2O2. The samples were left to
following the first flush of gas (Pulleman and Tietema, 1999). The C digest for 12 h at 80 ◦ C in an open digestion system (DigiPREP, SCP
respiration in μg C-CO2 per gram dry soil per hour was calculated Science). Macro and micronutrients and trace metals were than quan­
applying the following formula: tified by ICP-OES. 50 mg of pulverized material (< 2 mm) were used to

[change CO2 in headspace (mg.m− 3 .h)*headspace volume (m− 3 ) ]

C respiration = 0.273*
air dry sample weight (g)

where 0.273 is the conversion factor from C to CO2 in mg. prepare a homogeneous testable sample to determine total carbon (TC)
and total nitrogen (TN) content through dry combustion using an CN
analyzer (Elementar Analyzer, Germany).
2.5. Measurement of chemical and physical soil health indicators
Lignin concentrations in litter was measured using the Klason
method in which lignin is determined as ‘klason lignin’ by quantifying
Bulk density (BD) and gravimetric soil moisture content (W) at the
the wood component insoluble in a 72% sulphuric acid solution (Kirk
time of sampling were measured at a depth of 10 cm by inserting PVC
and Obst, 1988). No pre-treatment, such as ethanol-benzene or 80%
rings (diam. 7.5 cm; height 7.8) into the center of this layer capturing
aqueous ethanol (1:2), was applied for lignin determination as the ef­
undisturbed soil using one of the sides of the pit following excavation of
fects of different pre-treatments are still widely disputed (Toda et al.,
the monoliths. This resulted in a total of two measurements per plot (one
2015). The concentrations of extracted total polyphenol were deter­
per cacao tree). The ring was then placed in an air-tight plastic bag
mined according to method described by (Anderson and Ingram, 1993),
within a protective container for transport. Soils in the PVC rings were
after extraction with 20 mL methanol. Briefly, 0.75 g of litter was
carefully removed by cutting the edge of the ring with an iron saw. The
weighted into a 50 mL glass tube. Then, 20 mL of a 50% methanol so­
diameter of each ring was measured (in cm) prior to its removal. The
lution was added. Accordingly, the tube was covered with parafilm and
moist soil held by de ring was then weighed and a representative sub­
placed in a water bath at 70–75 ◦ C for 1 h. The extract was then filtered.
sample (40 g) was used to determine the gravimetrical moisture content
In a new flask were then Folin-Denis reagent, sodium 17% carbonate,
by oven drying at 105 ◦ C until constant weight. The subsamples were
the extract and water well mixed and put to rest for 20 min. Afterwards
used to determine dry weight of the moist samples in the rings. Rocks
the absorbance of the extracts was measured at 761 nm using a UV-VIS
and large roots from each ring were dried and weighed separately. Then,
spectrophotometer (Biotek synergy HT).
the calculated dry soil weight (g) was divided by the soil volume (cm3)
to calculate bulk density (g.cm− 3).
The disturbed soil samples (see Section 2.2) were air-dried, manually 2.7. Statistical analysis
crushed and sieved (< 2 mm) prior to determination of nutrient avail­
ability and other chemical soil properties. Soil pH was measured in All analyses were carried out using JMP pro 14.0.1 software package
deionized water (1:2.5) after 30 min of equilibration. Availability of soil (SAS Institute 2015). Data analyses involved all soil, leaf and litter
macro-, micro- nutrients and Cd was quantified by ICP-OES after Meh­ samples taken from the 5 different agroforestry treatments and 3 blocks,
lich III extractions (Mehlich, 1984). Analyses included phosphorus (P), at 2 distances nested within the main plots (n = 33). Before starting any
potassium (K), calcium (Ca), magnesium (Mg), sulphur (S), zinc (Zn), analyses, an ANOVA was used to rule out any block effect on leaf, litter,

5
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

and soil indicators. Then, a mixed model approach was applied to the shade cover followed the order “N-fix” > “fruit” = “mixed” =
examine the main and interaction effects of the treatments (Treat.; fixed “timber” > “control”. There was no interaction effect of treatment and
factor) and distance from the shade trees (Dist; a random factor to ac­ sampling distance on shade cover.
count for spatial autocorrelation) on the different soil health indicator Litter dry weight, C/N ratio and macronutrient content were not
values, litter quality and quantity, shade cover and leaf nutrient content. significantly affected by treatment or sampling distance (Table 2). The
Additionally, ANOVA was used to see if distance had an immediate ef­ average litter dry weight was 11.1 Mg ha− 1 and litter contained 275C,
fect on the C and N contents in the soil samples from different depths. 19 N, 1.5 P, 5.3 K (g kg− 1). Concentrations of Zn (p = 0.027) and Cd (p =
This was not the case. Two-way ANOVA was used to explore relation­ 0.038) in litter differed significantly among treatments with concen­
ships between soil C and N contents, Treat. and soil depth, to understand trations being highest in the “control” with only cacao trees (178 and
the impact of shade tree arrangement on these soil health indicators. 17.5 mg kg− 1, for Zn and Cd respectively) and lowest in “N-fix” (130 and
Fixed factors included Treat., soil depth and Treat. by soil depth in­ 5.5 g kg− 1, respectively), but no distance or interaction effects were
teractions. Where there were significant effects of Treat., interactions found.
between Treat.*Dist. and Treat.*depth, Tukey post-hoc analysis (α = Potential litter respiration taken as point measurement after 96 h
0.05) was carried out. Another mixed model approach was applied to differed (p = 0.047) among treatments and was marginally significant
examine the main and interactive effects of treatment (Treat.) on cacao after 168 h (p = 0.062; Fig. 3 & S2). Moreover, distance affected po­
yield of 2017, 2018 and 2019 and block effect (as random factor). A tential litter respiration after 168 h (p = 0.013). Overall, potential litter
principal component analysis (PCA) was performed to identify which respiration rates were higher in samples collected closer to the shade
soil health indicators and litter characteristics are most strongly asso­ trees than in the samples that were taken at larger distance. There was
ciated with each other, determine which of those variables contribute no interactive effect of shade tree and distance on potential litter
most to the variation in the data, and how they related to the different respiration.
agroforestry arrangements of the trial. When needed, data was log
transformed to comply with statistical assumptions of homoscedasticity 3.2. Soil health indicators
and normality of the residuals. This was done for i) Litter: K, Cd, poly­
phenols, respiration (96 h), ii) Leaf: P, Fe, Zn and Cd and iii) Soil: K, P, 3.2.1. Biological soil health indicators
Zn, Cd and respiration (76 h), species richness, total macrofauna, macro A total of 2899 macro-invertebrates were collected, representing 15
fauna diversity and earthworm abundance. different taxonomic groups (Table S1). Samples from two out of the
three blocks were infested by ants with counts of more than >1000 in­
3. Results dividuals per m2. This specific ant was identified as Nylanderia fulva
Mayr (1862), commonly known as crazy ant. When excluding crazy ants
3.1. Shade and litter characteristics of the different agroforestry systems from the total macrofauna count, we counted a total of 376 specimens of
arthropods and earthworms. The most abundant groups were Annelida
Shade cover at the time of sampling varied from 0 to 65%, depending (earthworms, 61.6%) and Hymenoptera (wasps, bees and ants, 15.9% of
on shade tree treatment (p < 0.001; Table 2). A clear effect of distance on total individuals). All other taxonomic groups, including Isopoda
shade cover of the cacao trees was observed, except for “N-fix”. Esti­ (woodlice), Chilopoda (Centipedes) and Coleoptera (Beetles) counted
mated shade cover at 2 m from the shade tree was largest for the “N-fix” each for <5% of total abundance.
and “fruit” treatments (51% and 22%, respectively), and further Earthworm abundance was affected by shade trees (p = 0.043), such
decreased in the order “timber” = “mixed” > “control”. At 6 m distance that earthworm abundance was a factor 3.2 higher in “mixed”

Table 2
Means and standard errors for shading and litter data observed in the different cocoa agroforestry systems (n = 5). P-values for the effect of treatment (Treat.), Distance
(Dist.) and their interaction (Int.) are provided. Samples were taken in October 2019 in a cocoa agroforestry trial in Coca, Amazon, Ecuador. Means followed by the
same letter in the column do not differ as per the Tukey test (α = 0.05). Shaded headers indicate treatments that include legume trees.
Cocoa agroforestry systems p-values

FRUIT MIX MONO N-FIX TIMBER Treat. Dist. Int.

Indicator Unit mean (s.e) mean (s.e) mean (s.e) mean (s.e) mean (s.e)

Shading
Shade cover % 22.2b (11.05) 14.53bc (3.51) 0c (0.00) 51.08a (6.05) 13.42bc (5.05) <0.001 <0.001 0.109
Litter
− 1
Litter dry weight Mg ha 9.8 (1.89) 11.1 (1.16) 12.3 (1.74) 12.4 (1.54) 9.9 (1.12) 0.645 0.570 0.525
C g kg− 1 271.6 (34.80) 280.9 (21.80) 275.4 (38.40) 289.1 (37.60) 264.4 (28.90) 0.975 0.374 0.417
N g kg− 1 18.0 (1.40) 20.0 (1.50) 19.3 (2.00) 22.4 (2.30) 19.0 (1.80) 0.412 0.504 0.082
P g kg− 1 1.6 (0.10) 1.5 (0.10) 1.6 (0.10) 1.4 (0.10) 1.5 (0.10) 0.668 0.953 0.944
K g kg− 1 6.0 (0.90) 4.6 (0.50) 5.7 (1.10) 4.2 (0.60) 6.3 (0.80) 0.260 0.713 0.590
Mg g kg− 1 4.1 (0.10) 4.0 (0.20) 3.6 (0.30) 3.8 (0.20) 3.8 (0.40) 0.734 0.348 0.115
Zn mg kg− 1 166.1ab (7.50) 155.6ab (9.20) 177.8a (11.90) 130.1b (5.80) 144.2ab (10.80) 0.027 0.912 0.449
Fe mg kg− 1 16,698 (3282.90) 16,286 (1435.30) 14,726 (2361.50) 14,715 (2904.10) 16,819 (2273.90) 0.905 0.638 0.675
Cd mg kg− 1 8.2ab (1.30) 11.7ab (2.90) 17.5a (4.10) 5.5b (0.60) 6.8ab (1.40) 0.038 0.592 0.452
Total Polyphenols μg ml− 1 18.7 (1.75) 17.0 (1.40) 24.5 (2.90) 17.9 (1.80) 21.4 (3.20) 0.114 0.914 0.246
Klason lignin
Residue % SR* 49.3 (1.30) 48.6 (2.20) 50.5 (1.60) 44.8 (2.10) 48.8 (1.40) 0.375 0.289 0.782
Litter Respiration ug CO2-C g− 1 dry
96 h soil hr− 1 84.5ab (14.83) 60.2ab (8.38) 62.8ab (12.44) 39.4b (4.20) 89.9a (17.90) 0.047 0.091 0.552
Litter Respiration ug CO2-C g− 1 dry
69.9 (6.30) 54.2 (8.80) 62.2 (13.00) 35.7 (4.60) 70.6 (12.80) 0.062 0.013 0.397
168 h soil hr− 1

* Solid Residue.
**Ca, Co, Mn, Cu and S were collected in this study but were not displayed in the main analyses as focus was placed on the macro and common micronutrient dis­
cussion, there were no significant differences found for these nutrients in this mixed model analyses. Al was not reported as result returned by ICP, furthermore Al
results were also not reliable according to the reference sample. Na and Mo were reported but mostly below detection limit. (n.d = non defined).

6
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

[A] Earthworm abundance [B] Soil respiraon a

a ab
ab
bc abc
c
ab ab

[C] Lier respiraon a

ab

ab
ab

Fig. 3. Biological indicators of soil health and litter characteristics analysed under different cocoa agroforestry systems in Coca, Amazon Ecuador. The bar graph
shows the means with std. err. From left to right: A) showing mean values of earthworm abundance sampled in the trial, B) showing mean values of soil respiration
after 96 h of incubation in the lab, C) showing mean values of litter respiration after 96 h of incubation in the lab.

agroforests as compared to "control" monocultures. Earth worm abun­ treatment. The C/N ratio of the first soil layer was a factor 1.28 higher in
dance was not affected by distance from the shade trees nor by the “fruit” agroforestry systems as compared to “mixed”. Moreover, higher
interaction between the two (Fig. 3; Table 4). C/N ratios were found in the upper soil layer (11.2 on average)
Treatment and distance did not influence macrofauna diversity, compared to the deeper layer (7.9 on average) in all treatments. Tree
taxonomic richness, or other indicators of macrofauna abundance (p > species did not affect C/N ratio at the 7.5–15 cm depth.
0.05). Taxonomic richness (number of orders) ranged from 4 to 6 taxa Total soil N was on average 4.7 (±0.13) g N kg− 1 in the first soil layer
per monolith, while diversity, based on the Shannon Index (H), ranged (Table 4), and Mehlich 3-extractable macronutrient concentration was
from 0.57 to 0.90 per monolith. on average 29.1 (±1.63) g P kg− 1, 0.54 (±0.04) g K kg− 1 and Fe and Zn
Potential soil respiration, as measured in lab conditions, was signif­ were 156.4 (±3.7) and 4.7(±0.32) mg kg− 1, respectively, with no
icantly affected by shade tree after 72 h (p < 0.01) and 96 h (p < 0.01; treatment effect for any of the measured nutrients (Table 3). Average
Fig. 3). After 96 h, soil respiration was a factor 1.43 higher in the Mehlich 3-extractable Cd in the soil across all treatments was 0.3
“timber” treatment compared to the “control”. These findings are (±0.01) mg Cd kg− 1. Nutrients and Cd were not significantly affected by
consistent with respiration rates at 72 h. There was no effect of distance treatment, distance, nor their interaction.
or its interaction with tree species on soil respiration (p > 0.05). Bulk density in the first soil layer was on average 0.73 g cm− 3 and the
water holding capacity was 48% (gravimetric) across treatments
3.2.2. Physical and chemical soil health indicators (Table 3). Bulk density and the water holding capacity were not
Carbon and N contents and C/N ratios were analysed in two soil significantly affected by treatment, distance, nor their interaction.
layers and showed significant depth effects (P < 0.01; Table 4). The C
and N content in the first soil layer was two-fold greater than in the
3.3. Relationships between litter characteristics and soil health indicators
deeper soil layer in all treatments, but we found no effect of treatment on
soil C and N content in both 0–7.5 and 7.5–15 cm layers.
Relations between the litter characteristics, soil health indicators and
The soil C/N ratio varied (p = 0.021) with soil depth (D) and with
shade trees were explored through principal component analysis (PCA).

7
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Table 3
Means and standard errors for biological, chemical and physical soil health indicators observed in the different cocoa agroforestry systems (n = 5). P-values for the
effect of treatment (Treat.), Distance (Dist.) and their interaction (Int.) are provided. Samples were taken in October 2019 in a cocoa agroforestry trial in Coca, Amazon,
Ecuador. Means followed by the same letter in the column do not differ as per the Tukey test (α = 0.05). Shaded headers indicate treatments that include legume trees.
Cocoa agroforestry systems p-values

FRUIT MIX MONO N-FIX TIMBER Treat. Dist. Int.

Indicator Unit mean (s.e) mean (s.e) mean (s.e) mean (s.e) mean (s.e)

Soil Biological
ug CO2-C g− 1 dry
Soil Respiration 72 h 6.34ab (0.30) 6.07b (0.39) 5.69b (0.18) 5.90b (0.41) 8.53a (1.04) 0.009 0.462 0.462
soil hr− 1
ug CO2-C g− 1 dry
Soil Respiration 96 h 5.72ab (0.34) 4.70bc (0.33) 4.22c (0.20) 4.90abc (0.27) 6.02a (0.53) 0.007 0.112 0.112
soil hr− 1
Macrofauna diversity Shannon-index 0.75 (0.18) 0.90 (0.20) 0.57 (0.23) 0.71 (0.20) 0.78 (0.21) 0.920 0.838 0.838
Species Richness # 4 (0.43) 5 (0.96) 4 (0.88) 4 (0.42) 6 (1.28) 0.355 0.941 0.941
Annelida ind m− 2 213ab (62.49) 267a (37.33) 83b (26.83) 159ab (33.65) 145ab (35.45) 0.043 0.865 0.865
Total macrofauna abundance 2
ind m− 282.0 (44.9) 366.7 (49.1) 304.0 (184.7) 215.3 (40.8) 272.7 (56.9) 0.820 0.486 0.486
(without Nylanderia fulva)
Soil chemical & physical *
pH – >5 5.57 (0.12) 5.36 (0.09) 5.41 (0.14) 5.27 (0.12) 5.65 (0.16) 0.332 0.965 0.874
− 1
CEC cmol kg – 9.22 (1.26) 8.00 (0.49) 7.22 (0.83) 7.23 (0.72) 9.32 (1.20) 0.354 0.930 0.679
avP g kg− 1 >25 29.21 (4.46) 29.87 (3.37) 29.82 (4.55) 27.86 (3.94) 28.27 (3.01) 0.997 0.676 0.835
K g kg− 1 >0.15 0.42 (0.06) 0.59 (0.11) 0.43 (0.06) 0.51 (0.06) 0.69 (0.17) 0.503 0.745 0.647
Mg g kg− 1 >3 0.16 (0.01) 0.15 (0.01) 0.17 (0.02) 0.17 (0.02) 0.19 (0.02) 0.651 0.886 0.168
Zn mg kg− 1 >1 6.40 (0.86) 3.81 (0.29) 4.49 (0.36) 4.26 (0.72) 5.16 (1.05) 0.087 0.904 0.456
Fe mg kg− 1 >4.5 171.08 (6.64) 149.90 (3.92) 156.34 (8.59) 159.74 (14.68) 147.94 (6.62) 0.370 0.787 0.816
melich.av Cd mg kg− 1 n.a 0.36 (0.03) 0.32 (0.03) 0.37 (0.04) 0.30 (0.04) 0.31 (0.03) 0.322 0.487 0.345
Bulk Density g cm− 3 0.75 (0.02) 0.73 (0.02) 0.69 (0.04) 0.75 (0.02) 0.75 (0.03) 0.586 0.590 0.753
WHC % 50.22 (0.40) 47.8 (0.86) 47.82 (1.52) 49.12 (0.72) 47.88 (1.29) 0.187 0.303 0.809

*Indicative reference values for available soil nutrients to support production (van Vliet and Giller, 2017; Wessel, 1970).
**Ca, Co, Mn, Cu and S were collected in this study but were not displayed in the main analyses as the focus was not mainly on chemical soil analyses, there were no
significant differences found for these nutrients in this mixed model analyses. Al was not reported as result returned by ICP, furthermore Al results were also not
reliable according to the reference sample. Na and Mo were reported but mostly below detection limit.

Table 4
C/N ratio, C & N content in the 0–7.5 cm, 7.5–15 layers in soils under different cocoa agroforestry systems in Coca, Amazonia, Ecuador (means). P-values for the effect
of treatment (Treat.), Soil depth and their interaction (Int.) are provided. In each treatment, lowercase letters compare means on the columns. Means followed by the
same letter in the column do not differ as per the Tukey test (p < 0.05). Shaded headers indicate treatments that include legume trees.
Cocoa agroforestry systems p-values

FRUIT MIX MONO N-FIX TIMBER Treat. Soil depth Int.

Indicator Soil layer (cm) mean (s.e) mean (s.e) mean (s.e) mean (s.e) mean (s.e)

C content % 0–7.5 5.43 (0.26) 4.61 (0.33) 5.01 (0.18) 5.40 (0.13) 5.25 (0.27)
7.5–15 1.99 (0.03) 2.48 (0.19) 2.60 (0.33) 2.40 (0.15) 2.33 (0.20) 0.296 <0.001 0.411
N Content % 0–7.5 0.45 (0.03) 0.51 (0.03) 0.44 (0.02) 0.45 (0.01) 0.46 (0.02)
7.5–15 0.25 (0.01) 0.30 (0.02) 0.33 (0.03) 0.30 (0.01) 0.29 (0.01) 0.548 <0.001 0.069
C/N ratio – 0–7.5 12.08a (0.39) 9.38b (0.94) 11.33ab (0.38) 11.85ab (0.26) 11.23ab (0.22)
7.5–15 7.79c (0.28) 8.02c (0.19) 7.71c (0.22) 8.12c (0.21) 7.92c (0.26) 0.043 <0.001 0.021

Each principal component (PC) was interpreted according to the 3.4. Cacao productivity, leaf nutrient, and Cd content
magnitude of their eigenvalues (Fig. 4). The first principal component
(PC1) explained 34.4% of the variance in the data and was mainly A significant variation in foliar nitrogen (N) content was observed
associated with litter parameters, such as litter respiration (0.97), lignin across different treatments (p < 0.01, Table 5). Specifically, lower foliar
content (0.91), N content (− 0.96), C content (− 0.91) litter dry weight N contents were recorded in both the “control” (20.3 g kg− 1) and “fruit”
(− 0.98), but at the same time also strongly associated to chemical soil treatment (20.0 g kg− 1) as compared to the “N-fix” treatment (25.7 g
health indicators such as pH (0.92) and CEC (0.91). Litter dry weight, C kg− 1). No significant single or interactive effects of treatment and dis­
and N concentrations clustered mostly around "N-fix", while soil bio­ tance were found for other nutrients. Average foliar nutrient contents
logical parameters seem to cluster in the opposite direction of the cacao were 1.18 (±0.03) g P kg− 1 and 79.11 (±4.99) mg Fe kg− 1, which is
monocultures ("control"). Cadmium content in soil and litter appears to under the optimal value of 2 g kg− 1 and 175 mg kg− 1 as indicated by van
cluster with the cacao monocultures. The second principal component Vliet and Giller (2017, Table 14). The foliar Cd content, averaging 10.87
(PC2) explained 27.2% of the variance in the data and was mainly (± 0.83) mg kg− 1, exhibited no variation across treatments (p = 0.556).
associated with soil health indicators, mainly positively with biological No difference was observed between the agroforestry systems (p =
indicators such as macrofauna diversity (0.95), earthworm abundance 0.667; Table 6), nor did we find a block (p = 0.338) effect on cacao bean
(0.73) and macrofauna richness (0.71) and negatively with soil Cd yield of all 3 years establishment. Pod counts between year 2 and 5 after
(− 0.81). Litter characteristics and soil health indicators do not seem to planting was variable in time, with 2018 showing the highest yields.
cluster in clear separated groups. Average pod production was 47 kg ha− 1, 515 and 142 kg ha− 1 in 2017,
2018 and 2019, respectively.

8
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Table 5
Means and standard errors for leaf data observed in the different cocoa agroforestry systems (n = 5). P-values for the effect of treatment (Treat.), Distance (Dist.) and
their interaction (Int.) are provided. Samples were taken in October 2019 in a cocoa agroforestry trial in Coca, Amazon, Ecuador. Means followed by the same letter in
the column do not differ as per the Tukey test (α = 0.05). Shaded headers indicate treatments that include legume trees.
Cocoa agroforestry systems p-values

FRUIT MIX MONO N-FIX TIMBER Treat. Dist. Int.

Indicator Unit mean (s.e) mean (s.e) mean (s.e) mean (s.e) mean (s.e)

Leaf *
C g kg− 1 n.d 442.2 (5.13) 445.2 (3.30) 445.8 (3.90) 452.8 (3.60) 449.4 (3.00) 0.458 0.472 0.932
N g kg− 1 >20 20.0b (0.72) 22.1ab (1.20) 20.3b (0.70) 25.7a (0.80) 21.4ab (1.40) 0.019 0.771 0.795
P g kg− 1 >2 1.2 (0.10) 1.1 (0.04) 1.1 (0.06) 1.4 (0.10) 1.1 (0.00) 0.105 0.195 0.730
K g kg− 1 >12 11.7 (1.33) 12.3 (0.87) 11.9 (0.95) 15.2 (1.60) 11.6 (0.70) 0.245 0.517 0.907
Mg g kg− 1 >2 5.9 (0.44) 4.9 (0.26) 4.9 (0.47) 4.8 (0.30) 5.9 (0.70) 0.308 0.718 0.964
1
Zn mg kg− >80 254.4 (49.05) 211.3 (31.80) 227.4 (15.80) 182.8 (24.90) 191.3 (21.70) 0.631 0.973 0.881
1
Fe mg kg− >175 85.9 (15.01) 70.4 (10.40) 88.8 (13.00) 68.9 (5.20) 85.9 (10.80) 0.653 0.594 0.086
1
Cd mg kg− <1.68 10.6 (1.14) 13.1 (2.55) 10.9 (1.29) 9.4 (1.22) 9.3 (1.33) 0.556 0.965 0.936

* Indicative reference values for nutrients in plant leaves to support cocoa production P/K according to (van Vliet and Giller, 2017;table 14), C/N/Mg/Zn/Fe according
to Chávez et al. (2019; slide 27), Cd calculated from Argüello et al., 2019 (Section 4.3).
**Ca, Co, Mn, Cu and S were collected in this study but were not displayed in the main analyses as focus was placed on the macro and common micronutrient dis­
cussion, there were no significant differences found for these nutrients in this mixed model analyses. Al was not reported as result returned by ICP, furthermore Al
results were also not reliable according to the reference sample. Na and Mo were reported but mostly below detection limit. (n.d = non defined).

Table 6
Means and standard errors for cacao pod yield data observed in the different cocoa agroforestry systems (n = 5). P-values for the effect of treatment (Treat.), block and
their interaction (Int.) are provided. Samples were taken in October 2019 in a cocoa agroforestry trial in Coca, Amazon, Ecuador. Means followed by the same letter in
the column do not differ as per the Tukey test (α = 0.05). Shaded headers indicate treatments that include legume trees.
Cocoa agroforestry systems p-values

FRUIT MIX MONO N-FIX TIMBER Treat. Block Int.

Indicator Unit mean (s.e) mean (s.e) mean (s.e) mean (s.e) mean (s.e)

Cocoa pod yield

1
Yield 2017 kg ha− 29.6 (29.62) 48.9 (30.71) 50.6 (30.76) 79.91 (46.97) 29.4 (29.39) 0.667 0.338 0.964
1
Yield 2018 kg ha− 631.7 (116.12) 627.7 (189.67) 545.1 (237.34) 436.6 (74.59) 336.3 (47.06) 0.636 0.501 0.227
1
Yield 2019 kg ha− 79.1 (32.27) 260.2 (111.09) 122.0 (68.37) 116.1 (63.59) 132.5 (58.26) 0.806 0.218 0.508

4. Discussion therefore presence for earthworms as observed in the “mixed”

treatment.
4.1. Shade tree species in cacao agroforestry systems influenced biological Potential soil respiration was a factor 1.4 higher in agroforests that
soil health indicators contained only C. cateniformis or B. gasipaes trees compared to "control,
indicating that shade tree species may provide better quality litter inputs
Our findings showed that agroforestry influenced some of the bio­ than cacao only, thus stimulating soil microbial activity and nutrient
logical soil health indicators, mainly earthworm abundance and soil cycling compared to unshaded cacao. Lori et al. (2022) found similar
respiration, even at early stage of establishment (5 years). Yet, other results in a 15-year-old cacao trial in Boliva, where organic management
important physical and chemical soil health indicators were not affected within a long-term agroforestry trial significantly enhanced the micro­
by agroforestry systems. This confirms our first hypothesis that the early bial community diversity in comparison to conventional monoculture.
effects of shade tree species on soil health are mostly captured through It’s plausible that soil respiration rates are elevated in agroforestry
biological soil indicators. Earthworms’ abundances were a factor three systems due to enhanced temperature regulation coupled with a greater
higher in cacao intercropped with C. cateniformis + E. velutina than in the quantity and diversity of litter, these factors collectively foster an
cacao monoculture. Similar findings were found in a coffee agroforestry optimal niche, promoting a more varied microbial community (Araújo
in Costa Rica, where earthworm abundance was lowest in the organic et al., 2014; Rodrigues et al., 2015). A more diverse microbial commu­
monoculture system compared to the organic agroforestry systems with nity with more abundant microbial community can result in higher
either Erythrina poepiggiana or Terminalia amazonia trees (Johnson- respiration rates in agroforestry systems. It would therefore be recom­
Maynard and Lugo-Perez, 2006). This positive effect of shade trees on mended to test in a future study the total microbial biomass and com­
earthworms and soil respiration may be explained by a higher litter munity composition to better understand the dynamics between
quality in the system as compared to monoculture cacao. For instance, agroforestry systems and varying respiration rates in soil and litter layer
Cesarz et al. (2007) demonstrated in their study that earthworm density within this study. We expected that N-fixing species (C. cateniformis and
and diversity is generally positively correlated with tree species di­ E. velutina) would have a greater positive effect on biological soil in­
versity, reflecting the importance of litter quality attributes for decom­ dicators, compared to other treatments, but this was not confirmed in
poser fauna. Despite the minimal variation in litter quality across our study. While potential respiration trends somewhat aligned between
treatments observed in our study, possibly because of the homogeneous litter and soil in response to treatments, an anomaly arose with the ‘n-
litter layer dominated by cacao leaves, our data indeed suggests that fix’ and ‘control’ treatment. This suggests that the influence of shade
shade trees can influence litter respiration. For instance, the highest rate trees on litter respiration in some cases may directly impact soil respi­
of litter respiration was recorded in the plot with C. cateniformis, a ration. Such linkage carries implications for soil health, as the litter layer
leguminous timber tree species. This species may provide a more plays a pivotal role as an interface connecting above-ground and below-
nutritionally rich litter layer, potentially enhancing the food source and ground ecosystems.

9
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Fig. 4. Bi-plot showing how different agroforestry systems relate to litter characteristics and a set of biological, chemical and physical soil health indicators. Samples
were collected within a cacao agroforestry trial within the Ecuadorian Amazon. Principal component 1 and 2 explain 34.4% and 27.2% of total variation,
respectively.

Shade trees had no effect on the selected chemical and physical soil amount of litter (5–10 t/ha/y; van Vliet and Giller, 2017). However,
health indicators, except C/N ratio in the upper soil layer (0–10 cm). Our litter samples closer to the shade tree showed overall higher microbial
results demonstrate that agroforestry systems influence soil health first respiration compared to amples taken further away, after one week of
through changes in biological indicators which over time could poten­ incubation. A higher litter respiration rate implies greater microbial
tially lead to alter soil physical-chemical parameters and impact agro­ activity and decomposition activity within the litter layer. This may
nomic traits in cacao. For instance, a continued high earthworm density affect nutrient cycling, ecosystem productivity, carbon emissions, and
during a prolonged period of time may greatly influence soil organic soil health, contingent upon the broader ecological context and pre­
matter dynamics and soil structure and associated physical conditions in vailing environmental conditions. As a result, it raises the possibility
a plot which positively affects nutrient cycling and plant uptake (Barois that, over time, proximity to certain shade trees (such as C. cateniformis
et al., 1999). However, the impact of shade trees on soil health can vary and E. velutina) may increase nutrient availability in the soil. For
greatly between scientific studies. For instance, our results align with instance, in a global meta-analysis on agroforestry, Ma et al. (2020)
finding of Ayres et al. (2009) who confirmed in their study conducted in highlight that beside tree-biomass also the soil organic carbon stock in
the Southern Rockey Mountains, North America, that tree species traits agroforestry systems (relative to cropland or pasture) increases with tree
mainly influenced leaf litter quality, soil respiration, and macrofauna age and that such patterns vary with tree species richness and regional
abundance. Contrastingly, de Souza et al. (2012), who compared a 13- climate. However, this finding was not observed in the soil samples
year-old coffee monoculture with agroforestry systems with differing analysed in our study .
shade-tree arrangements in the Atlantic Forest biome of Brazil, found
that chemical and biological soil health in agroforestry systems did not
differ significantly from full-sun coffee production systems. 4.3. Shade tree species affected nutrient content, but not Cd in cacao
leaves

4.2. Distance from shade tree does not influence litter quality or soil The planting density of E. velutina within the “N-fix” treatment was
indicators higher than in the other treatments, which explains the elevated shade
level observed within the treatment which could cause competition ef­
We hypothesized that soil health indicator values differ with distance fects with cacao for other sun and essential nutrients (Blaser et al.,
from the shade trees and this effect can be related to differences in 2018). Nonetheless, agroforestry treatments did not affect cacao yields,
shading, litter quality and quantity. Recent studies confirm that soil however the shade tree species arrangements impacted foliar nutrient
health indicators could be affected by tree species in the agroforestry content of cacao trees. For instance, foliar N content of cacao trees
systems and the distance of the cacao to the shade tree (Niether et al., grown in the agroforest with only E. velutina trees was overall a factor
2020). Nevertheless, our results showed that distance from the shade 1.3 higher than foliar N content of cacao trees in the monoculture
tree did not s change leaf, litter or soil health indicators. This finding (“control”) or agroforests with the fruit tree B. gasipaes. This might result
could be attributed to the fact that cacao trees produce a substantial from biomass inputs of the E. velutina trees. For instance, the higher N

10
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

content in the leave tissue could be caused by the N-fixing capacity of considerations of nutrient inputs and decomposition. Finally, moni­
the E. velutina shade trees in this system and the higher density of these toring Cd levels, crucial for food safety compliance, addressed a specific
trees compared to other arraignment. Oelbermann et al. (2004), who concern but was somewhat less directly related to the broader assess­
studied the decomposition of E. poeppigiana in alley cropping systems in ment of soil health. These findings collectively provided a nuanced
Costa Rica, observed that N inputs ranged from 73.84 kg ha− 1 in understanding of the complex interactions within agroforestry systems
younger productions system (< 5 years) to 220.85 kg ha− 1 in full grown and their potential benefits for cacao cultivation.
production systems (> 15 years). Hence, tree species from the Erythrina
genus can contribute significantly to both SOM storage and to the N 5. Conclusion
capital of soils leading to higher N available for uptake compared to
cacao monocultures (Heuveldop et al., 1988). This study contributes to a better understanding of the effects of
Foliar nutrient analyses indicated a deficiency of P and Fe in the shade tree management and tree species on soil health indicators and
cacao leaves of all treatments, based on indicative references provided cacao yields in young cacao production systems. Our findings revealed
by van Vliet and Giller (2017). To achieve high productivity, cacao re­ biological indicators, mainly earthworm abundance and soil respiration,
quires a soil with adequate levels of nutrients, however there is still a were most sensitive to early impacts of different agroforestry arrange­
lack of knowledge on the impact of nutrient concentration on plant ments. Distance from shade tree or cacao genotype did not influence soil
yields in cacao, particularly Fe. Soil pH affects the availability of nu­ health indicators or litter characteristics. Litter respiration exhibited
trients, especially P, the optimum soil pH recommended to sustain variations among the different shade tree arrangements, with the
adequate cacao growth is ~6.7 (Hartemink, 2005). However, most “timber” treatment, featuring the N-fixing tree species C. cateniformis,
cacao soils, including the one analysed in our study, are more acidic showing higher rates. This suggests a potential improvement in litter
mainly because the pedogenesis of soils in hot and humid environments quality in this treatment. Cadmium levels in litter, leaves and soil were
(such as the one in the Amazon). Moreover, cacao cultivation tends to high and above threshold level mentioned in related literature. We
result in soil acidification because cacao trees, including their leaves, conclude that biological soil health indicators are more sensitive to
contain significant amounts of organic acids (Ahenkorah et al., 1987). detect differences between shade tree species than chemical or physical
These organic acids gradually contribute to soil acidity over time. Re­ indicators. However, this was not yet expressed in pod yields nor foliar
searchers have reported that cacao can tolerate acid soil pH provided nutrient levels, except foliar N content which varied among the different
that the supply of nutrients should be adequate (Wood, 1985). We shade tree treatments, being highest within the “N-fix” and lowest in the
further found that foliar Cd levels were not affected by shade trees and “Fruit” systems. Further monitoring of this trial is recommended to
are too high which poses a risk for cacao farmers. Cadmium in cacao capture the long-term effect of agroforestry systems in soil health,
leaves is considered a good proxy for bean Cd (Shanying et al., 2017; nutrient cycling, and cacao yield. Our data analysis revealed consis­
Barraza et al., 2017). The elevated Cd content in the beans poses chal­ tently high cadmium levels across all shade tree arrangements. How­
lenges for smallholder farmers in selling their produce, primarily due to ever, we did not observe any discernible effects, which is possibly
food safety regulations (Argüello et al., 2019). attributed to the relatively young age of the trial. Hence, for a more
Shade tree treatments did not affect cacao yields. Agroforestry sys­ thorough understanding of the agroforestry system’s influence on cad­
tems have been reported to differ in yield by Blaser et al. (2018) and de mium accumulation in cacao, it is advisable to conduct a comprehensive
Heuvels et al. (2012), both studies indicated that this is mostly due to assessment in the future when the trial is old er or in similar settings with
differences in shading intensity in the plots. Nevertheless, the manage­ established shade trees. This research will help address the question of
ment of nutrients can significantly influence the relationship between whether agroforestry can contribute to the mitigation of cadmium
shade tree density and cacao yields, a pattern that has also been contamination in cacao plantations.
observed in coffee cultivation (Pulleman et al., 2023). The latter is also
in line with a study performed by Jacobi et al. (2015), who found that CRediT authorship contribution statement
organically managed successional or simple agroforestry systems did not
impact annual cacao production compared to monoculture systems in Anna M. Visscher: Conceptualization, Data curation, Formal anal­
Alto Beni, Bolivia. Plot age in their study varied from 9 to 14 years. The ysis, Investigation, Methodology, Software, Visualization, Writing –
lack of a yield effect in our study might also be explained by the young original draft, Writing – review & editing. Eduardo Chavez: Concep­
age (5 years) of the plantation. Cacao yields usually stabilize when the tualization, Investigation, Methodology, Resources, Supervision,
plants are more mature (>5 years; Lass and Wood, 1985), meaning that Writing – review & editing, Data curation, Writing – original draft.
effects of the added shade trees on the cacao yields might become more Carlos Caicedo: Resources, Writing – review & editing. Leider Tinoco:
prominent over the years (de Heuvels et al., 2012). We observed that Resources, Writing – review & editing. Mirjam Pulleman: Conceptu­
cacao yields were very variable over the last 3 years (2017–19), with alization, Data curation, Formal analysis, Investigation, Methodology,
2018 being the peak year. Several factors determine the actual yield of Supervision, Visualization, Writing – original draft, Writing – review &
cacao, causing that yields generally strongly differ from site to site even editing.
within plantations (van Vliet and Giller, 2017). Further research is
therefore needed to understand in greater details which mechanism Declaration of competing interest
influence cacao yield in agroforestry systems over time.
In our study, we assessed various indicators to gauge their relevance The authors declare that they have no known competing financial
in understanding the impact of shade trees on soil health and cacao interests or personal relationships that could have appeared to influence
production. At the forefront of usefulness, we report that earthworm the work reported in this paper.
abundance and soil respiration provided critical insights. These in­
dicators showed the ecological conditions of the soil and microbial ac­ Data availability
tivity for better understand nutrient cycling and soil structure.
Additionally, foliar nutrient content analysis played a pivotal role in Data will be made available on request.
optimizing cacao productivity by offering insights into nutrient avail­
ability. Chemical and physical soil health indicators, while not showing Acknowledgements
significant differences between treatments, contributed to a compre­
hensive view of soil health. Despite minimal variation in our study, the We would like to thank INIAP for its collaboration in this project,
assessment of litter quality and quantity held relevance for long-term special thanks go to the field crew at the Estacion Experimental Central

11
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

de la Amazonía who supported the sampling. We thank the technicians Creamer, R.E., Barel, J.M., Bongiorno, G., Zwetsloot, M.J., 2022. The life of soils:
integrating the who and how of multifunctionality. Soil Biol. Biochem. 166, 108561
at the Soils laboratory of ESPOL for the chemical analyses of the soil and
https://doi.org/10.1016/j.soilbio.2022.108561.
tissue samples. We further thank dr. A. Margenot for sharing data on the de Almeida, A.A.F.D., Valle, R.R., 2007. Ecophysiology of the cacao tree. Braz. J. Plant
soil profile taken at the INIAP trial in 2018. This study is part of the Physiol. 19, 425–448. https://doi.org/10.1590/S1677-04202007000400011.
project “Clima-LoCa” (Fostering low cadmium and climate-relevant in­ de Heuvels, O., Avelino, J., Somarriba, E., Malezieux, E., 2012. Vegetation structure and
productivity in cocoa-based agroforestry systems in Talamanca, Costa Rica. Agric.
novations to enhance the resilience and inclusiveness of the growing Ecosyst. Environ. 149, 181–188. https://doi.org/10.1016/j.agee.2011.03.003.
cacao sectors in Colombia, Ecuador and Peru). The project is coordi­ de Souza, H.N., de Goede, R.G., Brussaard, L., Cardoso, I.M., Duarte, E.M., Fernandes, R.
nated by the Alliance of Bioversity International and CIAT and finan­ B., Pulleman, M., 2012. Protective shade, tree diversity and soil properties in coffee
agroforestry systems in the Atlantic rainforest biome. Agric. Ecosyst. Environ. 146,
cially supported by the DeSIRA Programme of the European Union 179–196. https://doi.org/10.1016/j.agee.2011.11.007.
(grant no. FOOD/2019/407-158). The contents of this document are the Doran, J.W., Zeiss, M.R., 2000. Soil health and sustainability: managing the biotic
sole responsibility of the researchers and can under no circumstances be component of soil quality. Appl. Soil Ecol. 15 (1), 3–11.
Fountain, A.C., Hütz-Adams, F., 2022. Cocoa Barometer. https://cocoabarometer.org/wp
regarded as reflecting the position of the European Union. -content/uploads/2022/12/Cocoa-Barometer-2022.pdf.
Franzen, M., Mulder, M.B., 2007. Ecological, economic and social perspectives on cocoa
Appendix A. Supplementary data production worldwide. Biodivers. Conserv. 16, 3835–3849. https://doi.org/
10.1007/s10531-007-9183-5.
Giller, K.E., Beare, M.H., Lavelle, P., Izac, A.M., Swift, M.J., 1997. Agricultural
Supplementary data to this article can be found online at https://doi. intensification, soil biodiversity and agroecosystem function. Appl. Soil Ecol. 6,
org/10.1016/j.geodrs.2024.e00772. 3–16. https://doi.org/10.1016/S0929-1393(96)00149-7.
Hartemink, A.E., 2005. Nutrient stocks, nutrient cycling, and soil changes in cocoa
ecosystems: a review. Adv. Agron. 86, 227–253. https://doi.org/10.1016/S0065-
References 2113(05)86005-5.
Hess, T.F., Schmidt, S.K., 1995. Improved procedure for obtaining statistically valid
Ahenkorah, Y., Halm, B.J., Appiah, M.R., Akrofi, G.S., Yirenkyi, J.E.K., 1987. Twenty parameter estimates from soil respiration data. Soil Biol. Biochem. 27, 1–7. https://
years’ results from a shade and fertilizer trial on Amazon cocoa (Theobroma Cacao) doi.org/10.1016/0038-0717(94)00166-X.
in Ghana. Exp. Agric. 23, 31–39. https://doi.org/10.1017/S0014479700001101. Heuveldop, J., Fassbender, H.W., Alpizar, L., Enriquez, G., Fölster, H., 1988. Modelling
Alfaro-Flores, A., Morales-Belpaire, I., Schneider, M., 2015. Microbial biomass and agroforestry systems of cacao (Theobroma cacao) with laurel (Cordia alliodora) and
cellulase activity in soils under five different cocoa production systems in alto Beni, poro (Erythrina poeppigiana) in Costa Rica II. Cacao and wood production, litter
Bolivia. Agrofor. Syst. 89, 789–798. https://doi.org/10.1007/s10457-015-9812-z. production and decomposition. Agrofor. Syst. 6, 37–48. https://doi.org/10.1007/
Amponsah-Doku, B., Daymond, A., Robinson, S., Atuah, L., Sizmur, T., 2022. Improving BF02344744.
soil health and closing the yield gap of cocoa production in Ghana–a review. Sci. Afr. Jacobi, J., Schneider, M., Bottazzi, P., Pillco, M., Calizaya, P., Rist, S., 2015.
15, e01075. Agroecosystem resilience and farmers’ perceptions of climate change impacts on
Anderson, J.M., Ingram, J.S.I., 1993. Tropical Soil Biology and Fertility: A Handbook of cocoa farms in alto Beni, Bolivia. Renew Agric. Food Syst. 30, 170–183. https://doi.
Methods, 2nd edn. CAB. International Oxfordshirne, Wallingford, UK. org/10.1017/S174217051300029X.
Araújo, A.S.F., Borges, C.D., Tsai, S.M., Cesarz, S., Eisenhauer, N., 2014. Soil bacterial Janzen, H.H., Janzen, D.W., Gregorich, E.G., 2021. The ‘soil health’metaphor:
diversity in degraded and restored lands of Northeast Brazil. Anton Leeuw. Int. J. G. illuminating or illusory? Soil Biol. Biochem. 159, 108167.
106, 891–899. https://doi.org/10.1007/s10482-014-0258-5. Johnson-Maynard, J., Lugo-Perez, J., 2006. Earthworm populations, microbial biomass
Argüello, D., Chavez, E., Lauryssen, F., Vanderschueren, R., Smolders, E., Montalvo, D., and coffee production in different experimental agroforestry management systems in
2019. Soil properties and agronomic factors affecting cadmium concentrations in Costa Rica. Caribb. J. Sci. 42, 397–409.
cacao beans: a nationwide survey in Ecuador. Sci. Total Environ. 649, 120–127. Jose, S., 2009. Agroforestry for ecosystem services and environmental benefits: an
https://doi.org/10.1016/j.scitotenv.2018.08.292. overview. Agrofor. Syst. 76, 1–10. https://doi.org/10.1007/s10457-009-9229-7.
Asigbaase, M., Dawoe, E., Sjogersten, S., Lomax, B.H., 2021. Decomposition and nutrient Kirk, T.K., Obst, J.R., 1988. Lignin determination Meth Enzymol, 161, pp. 87–101.
mineralisation of leaf litter in smallholder cocoa agroforests: a comparison of organic https://doi.org/10.1016/0076-6879(88)61014-7.
and conventional farms in Ghana. J. Soils Sediments 21, 1010–1023. https://doi. Koné, A.W., Yao, M.K., 2021. Soil microbial functioning and organic carbon storage: can
org/10.1007/s11368-020-02844-4. complex timber tree stands mimic natural forests? J. Environ. Manag. 283, 112002
Ayres, E., Steltzer, H., Berg, S., Wallenstein, M.D., Simmons, B.L., Wall, D.H., 2009. Tree https://doi.org/10.1016/j.jenvman.2021.112002.
species traits influence soil physical, chemical, and biological properties in high Lass, R.A., Wood, G.A.R., 1985. Cocoa Production: Present Constraints and Priorities for
elevation forests. PLoS One 4, e5964. https://doi.org/10.1371/journal. Research. The World Bank. https://agris.fao.org/agris-search/search.do?reco
pone.0005964. rdID=US2012401249.
Barois, I., Lavelle, P., Michel, Brossard, Jérôme, Tondoh, de los Angeles Martinez, M., López, R.E.I., Navarrete, E.F.C., Rosado, J.T.P., García, C.R.S., Margenot, A.J., 2021. Soil
Rossi, J.P., Senapati, B.K., Angeles, A., Fragoso, C., Jimenez, J.J., Thibaud, Decaëns, nitrogen cycling under contrasting management systems in Amazon Coffea canephora
Lattaud, C., Kanyonyo, J., Eric, Blanchart, Lydie, Chapuis, George, Brown, agroecosystems. Soil Sci. Soc. Am. J. 85, 1634–1648. https://doi.org/10.1002/
Moreno, A., 1999. Ecology of earthworm species with large environmental tolerance saj2.20255.
and/or extended distributions. In: Lavelle, P., Brussaard, L., Hendrix, P. (Eds.), Lori, M., Armengot, L., Schneider, M., Schneidewind, U., Bodenhausen, N., Mäder, P.,
Earthworm Management in Tropical Agroecosystems. CABI, Wallingford, pp. 57–85. Krause, H.M., 2022. Organic management enhances soil quality and drives microbial
ISBN 0-85199-270-6. community diversity in cocoa production systems. Sci. Total Environ. 834, 155223
Barraza, F., Schreck, E., Lévêque, T., Uzu, G., López, F., Ruales, J., Maurice, L., 2017. https://doi.org/10.1016/j.scitotenv.2022.155223.
Cadmium bioaccumulation and gastric bioaccessibility in cacao: a field study in Ma, Z., Chen, H.Y., Bork, E.W., Carlyle, C.N., Chang, S.X., 2020. Carbon accumulation in
areas impacted by oil activities in Ecuador. Environ. Pollut. 229, 950–963. https:// agroforestry systems is affected by tree species diversity, age and regional climate: A
doi.org/10.1016/j.envpol.2017.07.080. global meta-analysis. Glob. Ecol. Biogeogr. 29, 1817–1828. https://doi.org/
Blaser, W.J., Oppong, J., Yeboah, E., Six, J., 2017. Shade trees have limited benefits for 10.1111/geb.13145.
soil fertility in cocoa agroforests. Agric. Ecosyst. Environ. 243, 83–91. https://doi. Marshall, E.J.P., 2004. Agricultural landscapes: field margin habitats and their
org/10.1016/j.agee.2017.04.007. interaction with crop production. J. Crop Improv. 12, 365–404. https://doi.org/
Blaser, W.J., Oppong, J., Hart, S.P., Landolt, J., Yeboah, E., Six, J., 2018. Climate-smart 10.1300/J411v12n01_05.
sustainable agriculture in low-to-intermediate shade agroforests. Nat. Sustain. 1, Mehlich, A., 1984. Mehlich 3 soil test extractant: A modification of Mehlich 2 extractant.
234–239. https://doi.org/10.1038/s41893-018-0062-8. Commun. Soil Sci. Plant Anal. 15, 1409–1416. https://doi.org/10.1080/
Boyer, J., 1973. Cycles de la matière organique et des éléments minéraux dans une 00103628409367568.
cacaoyère camerounaise. Café Cacao Thé, 18, pp. 3–30. https://agris.fao.org/agris Motamayor, J.C., Risterucci, A.M., Lopez, P.A., Ortiz, C.F., Moreno, A., Lanaud, C., 2002.
-search/search.do?recordID=US201303265140. Cacao domestication I: the origin of the cacao cultivated by the Mayas. Heredity 89
Bünemann, E.K., Bongiorno, G., Bai, Z., Creamer, R.E., De Deyn, G., de Goede, R., (5), 380–386.
Pulleman, M., 2018. Soil quality–A critical review. Soil Biol. Biochem. 120, 105–125. Niether, W., Jacobi, J., Blaser, W.J., Andres, C., Armengot, L., 2020. Cocoa agroforestry
https://doi.org/10.1016/j.soilbio.2018.01.030. systems versus monocultures: a multi-dimensional meta-analysis. Environ. Res. Lett.
Buyer, J.S., Baligar, V.C., He, Z., Arévalo-Gardini, E., 2017. Soil microbial communities 15, 104085 https://doi.org/10.1088/1748-9326/abb053.
under cacao agroforestry and cover crop systems in Peru. Appl. Soil Ecol. 120, Oelbermann, M., Voroney, R.P., Schlönvoigt, A.M., Kass, D.C., 2004. Decomposition of
73–280. https://doi.org/10.1016/j.apsoil.2017.09.009. Erythrina poeppigiana leaves in 3-, 9-, and 18-year-old alleycropping systems in
Cesarz, S., Fahrenholz, N., Migge-Kleian, S., Platner, C., Schaefer, M., 2007. Earthworm Costa Rica. Agrofor. Syst. 63, 27–32. https://doi.org/10.1023/B:
communities in relation to tree diversity in a deciduous forest. Eur. J. Soil Biol. 43, AGFO.0000049430.52250.87.
61–67. https://doi.org/10.1016/j.ejsobi.2007.08.003. Ofori-Frimpong, K., Asase, A., Mason, J., Danku, L., 2007. Shaded versus unshaded
Clough, Y., Faust, H., Tscharntke, T., 2009. Cacao boom and bust: sustainability of cocoa: implications on litter fall, decomposition, soil fertility and cocoa pod
agroforests and opportunities for biodiversity conservation. Conserv. Lett. 2, development. Symposium on multistrata agroforestry systems with perennial crops.
197–205. https://doi.org/10.1111/j.1755-263X.2009.00072.x. In: CATIE Turrialba, 2007 Costa Rica, pp. 17–21.
Pretty, J., 1997. The sustainable intensification of agriculture. Nat. Res. Forum 21,
247–256. https://doi.org/10.1111/j.1477-8947.1997.tb00699.x.

12
A.M. Visscher et al. Geoderma Regional 37 (2024) e00772

Pulleman, M., Tietema, A., 1999. Microbial C and N transformations during drying and functioning during tropical rainforest conversion and agroforestry intensification.
rewetting of coniferous forest floor material. Soil Biol. Biochem. 31, 275–285. Proc. Natl. Acad. Sci. 104, 4973–4978. https://doi.org/10.1073/pnas.0608409104.
https://doi.org/10.1016/S0038-0717(98)00116-3. Toda, M., Akiyama, T., Yokoyama, T., Matsumoto, Y., 2015. Quantitative examination of
Pulleman, M.M., Rahn, E., Valle, J.F., 2023. Regenerative agriculture for low-carbon and pre-extraction treatment on the determination of lignin content in leaves.
resilient coffee farms: A practical guidebook. In: Version 1.0. Cali (Colombia): Bioresources 10, 2328–2337. https://ojs.cnr.ncsu.edu/index.php/BioRes/article/vi
International Center for Tropical Agriculture (CIAT), p. 181. https://hdl.handle. ew/BioRes_10_2_2328_Toda_Quantitative_Preextraction_Lignin_Leaf.
net/10568/131997. Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005. Landscape
Richard, A., Ræbild, A., 2016. Tree diversity and canopy cover in cocoa systems in perspectives on agricultural intensification and biodiversity–ecosystem service
Ghana. New For. 47, 287–302. https://doi.org/10.1007/s11056-015-9515-3. management. Ecol. Lett. 8, 857–874. https://doi.org/10.1111/j.1461-
Rodrigues, R.C., Araújo, R.A., Costa, C.S., Lima, A.J., Oliveira, M.E., Cutrim Jr., J.A., 0248.2005.00782.x.
Araújo, A.S., 2015. Soil microbial biomass in an agroforestry system of Northeast van Vliet, J.A., Giller, K.E., 2017. Mineral nutrition of cocoa: A review. In: Adv Agron,
Brazil. Trop. Grassl. 3, 41–48. https://doi.org/10.17138/tgft(3)41-48. 141, 185–270. Academic Press. https://doi.org/10.1016/bs.agron.2016.10.017.
Rousseau, G.X., Deheuvels, O., Arias, I.R., Somarriba, E., 2012. Indicating soil quality in Wessel, M., 1970. Fertiliser experiments on farmers’ cocoa in southwestern Nigeria.
cacao-based agroforestry systems and old-growth forests: the potential of soil Cocoa Growers’ Bull. 15, 22–27. https://www.cabdirect.org/cabdirect/abstract/19
macrofauna assemblage. Ecol. Indic. 23, 535–543. 710305032.
Shannon, C.E., 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27, Wood, G.A.R., 1985. Environment. In: Wood, G.A.R., Lass, R.A. (Eds.), Cocoa, pp. 38-78.
379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x. Longman, London.
Shanying, H.E., Xiaoe, Y.A.N.G., Zhenli, H.E., Baligar, V.C., 2017. Morphological and Yao, M.K., Koné, A.W., Otinga, A.N., Kassin, E.K., Tano, Y., 2021. Carbon and nutrient
physiological responses of plants to cadmium toxicity: a review. Pedosphere 27, cycling in tree plantations vs. natural forests: implication for an efficient cocoa
421–438. https://doi.org/10.1016/S1002-0160(17)60339-4. agroforestry system in West Africa. Reg. Environ. Chang. 21, 1–14. https://doi.org/
Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M.M., Buchori, D., Erasmi, S., 10.1007/s10113-021-01776-0.
Guhardja, E., 2007. Tradeoffs between income, biodiversity, and ecosystem

13