Influence of Soil Physical and Hydraulic Properties on Cacao Productivity Under Agroforestry Systems in the Amazonian Piedmont

Abstract

In the Amazonian piedmont, cacao-based agroforestry systems (cAFSs) were significantly influenced by the soil’s physical, hydraulic, and structural characteristics, which largely determined agricultural productivity. A total of 122 plots with cocoa-based agroforestry systems measuring 1000 m² were randomly selected from different farms located in the Amazonian foothills in the department of Caquetá. Different variables related to soil physics and hydrology, as well as production, were determined for each plot. Soil characteristics explain 33% of the total variance in cocoa yield. Sand content (71.2%) correlated positively with yield, while clay (22.62%) and silt (23.99%) correlated negatively. Three soil types were identified: sandy loam (high productivity, yield 1129.07 g) and two variants of sandy clay loam (lower yield, 323.97 g). Hydraulic properties were important, with total porosity of 56.04% and hydraulic conductivity of 20.45 mm h⁻¹. The CCN-51 and ICS-60 clones performed better in sandy loam soils, while ICS-95 and TSH-565 adapted better to sandy clay loam soils with medium stability. The physical and hydric soil properties are crucial factors that directly influence cocoa productivity in agroforestry systems of the Amazon piedmont, where the appropriate selection of clones according to soil characteristics is fundamental to optimize crop productivity and sustainability.

1. Introduction

Cacao (Theobroma cacao L.) is a commercially significant crop that plays a vital role in the country’s agricultural economy [1]. In Colombia, approximately 65,341 families are engaged in cacao farming [2]. This species is closely linked to environmental conservation, as its production system is typically associated with forestry and other shade-providing crops such as plantains and bananas [3]. The successful implementation of cacao cultivation requires careful consideration of edaphic factors, including soil physical and chemical characteristics, which are closely related to fertility. Environmental factors such as precipitation, temperature, solar radiation, wind, and relative humidity significantly influence cacao yield. Additionally, the genetic composition of planting material plays a crucial role in the biological aspect of cultivation [4,5].

The dominant cultural practice of cacao cultivation in the Colombian Amazon has evolved significantly over time, resulting in considerable genetic diversity in current plantations. Historically, the establishment of the crop has followed a characteristic pattern that involves slash-and-burn of forests or fallows, followed by the planting of temporary shade species such as plantain (Musa × paradisiaca L.-Musaceae) and banana (Musa × sapientum L.-Musaceae), along with the establishment of timber trees for permanent shade. Regarding the genetic composition of the plantations, there is a clear distinction between old and more recent plantations [6,7]. Plantations over 30 years old are primarily composed of hybrid trees, resulting from cross-pollination between introduced universal clones and wild trees in the region. As part of technological innovation in the sector, new plantations have been established mainly with specific clones such as TSH-565, ICS-1, ICS-39, ICS-60, ICS-95, EET-8, and CCN-51, all of which are grafted onto hybrid rootstock of clone IMC-67 [8]. This genetic diversity present in the production systems is a crucial factor that must be considered when evaluating soil–plant interactions and the expression of specific characteristics in different growing areas of the region.

Soil water content plays a fundamental role in cacao cultivation. García’s [9] research confirms that after eight weeks without rainfall, soil moisture content decreases significantly, and stress symptoms appear by the tenth week. In this context, the analysis of water use in agriculture must carefully consider the interaction between various sources, highlighting the soil’s role as a moisture reservoir and the soil-water interaction [10]. However, Rodríguez et al. [11] note that inadequate soil drainage can cause significant damage to cacao plants, substantially reducing their production. Most interactions between water, soil, and plant roots occur under unsaturated conditions, where empty pores are present [10]. Therefore, it is crucial to consider changes in land use, as these modify soil physical properties such as bulk density, porosity, coverage, structure, and texture [11]. In this context, Souza et al. [12] establish that cover crops offer benefits by significantly reducing penetration resistance, thus favoring root flow and development. Furthermore, in areas where cacao is integrated with agroforestry systems (AFSs), the nutrients stored in the AFS biomass exceed those present in the soil. This creates a beneficial association for cacao cultivation, as when leaves fall and decompose, the leaf litter enriches the soil and provides essential nutrients available for cacao uptake [13]. Studies have been conducted on the relationship between hydraulic conductivity and porosity, as well as the effect of root systems in AFSs [14].

Therefore, the objective of this study is to investigate the influence of soil physical and hydraulic properties on cacao productivity under agroforestry systems in the Amazonian Piedmont. The research seeks to understand how fundamental soil characteristics, including texture, structure, density, water retention capacity, drainage, and porosity, directly affect the yield and development of cacao crops in this specific region. This analysis is particularly relevant in the context of agroforestry systems, where the interaction between different system components can significantly influence soil properties and, consequently, cacao productivity. The study aims to generate knowledge that will allow optimization of management practices and improve site selection for establishing new crops, thus contributing to the sustainability and efficiency of cacao production in the Amazonian Piedmont region. Soil physical and hydraulic properties, particularly sand content, structural stability, and total porosity, are expected to have a significant influence on cocoa productivity in agroforestry systems of the Amazonian Piedmont, where sandy loam soils with high drainage capacity will show superior yields (>1000 kg ha⁻¹ year⁻¹) compared to sandy clay loam soils with low structural stability, due to their better conditions for aeration, infiltration, and root development.

2. Materials and Methods

2.1. Study Area

The study was conducted between 2020 and 2023 in rural areas of the municipalities of San Jose de Fragua, Belén de los Andaquíes, El Paujil, and El Doncello in the Department of Caquetá, southeastern Colombia (Figure 1). The study area is characterized by an annual mean temperature of 25 °C and annual precipitation of up to 3600 mm, concentrated between April and November, with a slightly dry season from December to March. A total of 122 plots under cacao-based agroforestry arrangements were randomly selected from 59 farms belonging to the Cacao Producers Association of Caquetá Department. These were distributed as follows: 14 from the Organic Agricultural and Cacao Association of San José del Fragua municipality (its acronym in Spanish is “ASOACASAN”), 15 from the Alternative Agroforestry Producers Association of Belén de los Andaquíes (its acronym in Spanish is “ASPROABELÉN”), and 30 from the Cacao Growers Committee of El Paujil and El Doncello (its acronym in Spanish is “COMCAP”). Within each farm, productive cacao plots larger than 1.5 ha were identified. Subsequently, a sampling plot measuring 50 × 20 m (total area of 1000 m²) was established in the central portion of each plot. These plots were used to determine soil physical and hydraulic parameters, as well as the productive component and yield of cacao trees.

2.2. Characterization of Soil Physical Properties

The characterization of soil physical properties involved the evaluation of eight fundamental physical variables to obtain a comprehensive understanding of soil physical quality. Soil samples were collected during November–December 2020 and April–May 2023. In each plot, three undisturbed soil samples were taken at depths of 0–10, 10–20, and 20–30 cm using a metallic ring with a volume of 98.44 cm³. Five subsamples were collected to form a composite sample per depth for each plot, taken from the center and four corners of the plot. In the laboratory, undisturbed samples were weighed and oven-dried at 105 °C for 48 h until reaching constant weight. Bulk density (BD, Mg m⁻³) was calculated as the ratio of total dry soil mass to cylinder volume, while particle density (PD) was determined using the pycnometer method [15]. Total porosity (TP) was calculated using the equation [TP = (1 − (BD/PD)) × 100]. Soil texture analysis was performed using the hydrometer method following the USDA classification system. According to this system, soil particles are classified into three size fractions: i. sand (particles between 50 and 2000 µm), ii. silt (particles between 2 and 50 µm), and iii. clay (particles smaller than 2 µm). The results were expressed as percentages of sand, silt, and clay, which were then plotted on the USDA texture triangle to determine the texture class. The USDA texture triangle classifies soils into 12 texture classes based on these particle size percentages [16].

Soil aggregate stability was assessed using undisturbed soil samples collected at each depth (0–10, 10–20, and 20–30 cm) from three sampling points per plot. Samples were gently broken along their natural breakage lines and passed through an 8 mm sieve. Prior to water aggregation determination, samples were dried at 45 °C. Two 40 g subsamples per plot were processed using a Yoder apparatus, utilizing a nest of five sieves (4, 2, 1, 0.5, and 0.25 mm mesh sizes). The sieves were agitated in water at 30 oscillations per minute for 30 min. The resulting aggregates from each sieve were oven-dried at 105 °C for 48 h and weighed to calculate structural stability percentage, mean weight diameter (MWD), aggregation state (%), and aggregation index, following Val Bavel [17].

Soil aggregate morphology was determined following Velásquez et al. [18] methodology. A soil block (10 × 10 × 10 cm) was carefully collected and stored in a plastic box to avoid compaction before laboratory analysis at the Universidad de la Amazonia. Visual separation of macroaggregate components included biogenic aggregates (BA), which are dense formations showing clear biological activity; physical aggregates (PA), which are geometric shapes resulting from soil wetting and drying processes; root aggregates (RA), formed through root-soil interactions; non-aggregated soil (NA), consisting of soil particles (<5 mm); and organic matter aggregates (OA), including plant residues such as leaves, loose roots, and wood pieces. All categories were identified, separated, dried, and weighed to calculate the percentage of each macroaggregate component.

Field measurements included soil penetration resistance (SPR) and visual soil evaluation. The SPR measurements were taken around the soil sampling pit to a depth of 30 cm using a manual penetrometer (Eijkelkamp) with a 60° cone angle and 2 cm² surface area. The SPR values were recorded in triplicate for each sampling point per plot (n = 15). The Visual Evaluation of Soil Structure (VESS) was conducted following Guimaraes et al. [19] methodology. At each sampling point, a soil pit (30 × 30 × 40 cm depth) was excavated to extract an intact soil block (20 × 10 × 30 cm depth, ~6000 cm³ volume) using a spade. The sample was placed on a light-colored plastic tray for examination.

The evaluation involved gentle manipulation to reveal structural characteristics, identification of contrasting layers based on color, assessment of root distribution and biological activity, and assignment of VESS scores (Sq_scores) using the visual interpretation chart. Scores 1–2 indicate high physical quality with favorable plant growth conditions; score 3 represents good physical quality but at a critical threshold requiring management attention; and scores 4–5 indicate poor physical quality, suggesting an urgent need for management modifications. Photographs were taken to validate field-assigned scores [20]. Then, the overall weighted Sq_score for each sample was calculated based on the individual score and thickness of each contrasting soil layer, according to the following equation:

$$VESS {Sq}_{score}= \sum _{i=1}^n{\displaystyle \frac{{Sq}_i{T}_i}{TT}}$$

(1)

where VESS Sq_score is the overall sample score, Sq and T are the score and thickness of each identified soil layer, and TT is the total depth of the soil sample (i.e., 30 cm).

Similarly, weighted VESS scores were determined for soil layers (0–10, 10–20, and 20–30 cm) to compare and correlate VESS data with other soil parameters evaluated in these same soil layers [21]. It is important to note that all VESS evaluations were conducted by a single trained individual. This measure was implemented to avoid potential variations attributable to different evaluators, thereby ensuring standardization in score assignment and minimizing the method’s subjectivity [21].

2.3. Soil Hydraulic Characteristics

Soil infiltration was evaluated using the method established by USDA [22]. Stainless steel cylinders with a diameter of 15 cm were used. Three repetitions per plot were performed (adjacent to the soil sampling points) at a depth of 0–10 cm. A known volume of water (300 mL) was added, and the height of the water column was measured in centimeters. Subsequently, the time taken for the water to completely infiltrate into the soil was recorded to estimate the infiltration rate (IR) in (ml min⁻¹). To determine the water retention capacity of different soils in cacao agroforestry systems, a moisture retention curve method was developed. An undisturbed soil block (10 × 10 × 10 cm) was collected adjacent to the other sampling sites and stored in a plastic box. To calculate each soil’s water retention capacity, each soil block was placed in 2000 mL plastic pots with anti-thrip mesh at the bottom to prevent soil loss. The pots were then placed in plastic boxes to completely saturate the soil with water for 12 h. After 12 h, the pots were removed and allowed to drain under ambient conditions, protected from weather elements. Subsequently, soil weight was recorded every 24 h until day 20. After 20 days, the soil was removed from the pots and dried in paper bags in a drying marquee until the samples were completely dry, being weighed until reaching constant weight. Soil moisture (%) was determined using the equation: soil moisture = (1 − (soil dry mass/soil wet mass)) × 1000 [23], and soil field capacity (FC) was determined at 48 h. Based on texture and bulk density (ρb) and using the pedo-transfer function of ROSETTA software 1.0 [24], initial assessments were derived to estimate soil hydraulic parameters (T_r: residual volumetric water content, T_s: saturated volumetric water content, α and n: appropriate parameters of soil water characteristic curve). This function predicts saturated hydraulic conductivity (K_s) in the saturation curve.

2.4. Cocoa Productive and Yield Parameters

The productive component assessment was conducted during two periods: November-December 2020 (Yield1) and April-May 2023 (Yield2), which represent peak production periods. In each 1000 m² plot, the number of cocoa trees and associated plants in the agroforestry system was counted. For each individual tree, reproductive organs were counted, including the number of pods (ripe and unripe) and green cherelles, to estimate the number of pods per tree (NPT). Similarly, the incidence of pests (Monalonium dissimulatum) and diseases (Moniliophthora spp. and Phytophthora spp.) was determined by counting all pods with visible signs and symptoms [25,26]. To determine yield parameters, 25 cocoa pods were collected from each plot to measure pod length (PL), pod diameter (PD), and pod fresh weight (PFW). Additionally, the number of beans per pod (NGP) and fresh bean weight per pod were recorded to later estimate dry bean weight per pod (WGP). The following indices were calculated: i. Bean index (GI) = WGP/NGP, ii. Pod index (PI) = 1000/WGP, iii. Harvest index (HI) = NPT/PI. Cocoa bean yield (GY) was calculated using the following formula: GY (kg ha⁻¹ year⁻¹) = (average NPT) × (average WGP) × (number of cocoa trees per hectare (counted in 1000 m² × 10)) [27].

2.5. Data Analysis

To evaluate the relationship between soil physical and hydraulic attributes and agronomic variables in cocoa agroforestry systems, cluster analyses were performed using physical and hydraulic variables. These analyses were conducted using Ward’s method and Euclidean distance. After identifying typologies, Linear Mixed Models (LMMs) were fitted, considering typologies as fixed effects and plots as random effects. Normality and homoscedasticity were validated through exploratory residual analysis using QQ-plots and adjusted-plots. When significant statistical differences between typologies were detected, the DGC means comparison test was applied at 5% significance level. Additionally, a Principal Component Analysis (PCA) was performed on the complete dataset, complemented with a Monte Carlo test (999 permutations) to evaluate statistical significance (α = 0.05) between the established typologies. Furthermore, Pearson correlation analysis was conducted to study the relationship between yield and physical, hydraulic, and productive variables across different typologies. Finally, co-inertia analyses were performed, accompanied by Monte Carlo tests (999 permutations), to evaluate the significance of covariation between datasets. The LMMs were implemented using the lme function from the nlme package [28] in R software, version 3.6.1 [29], and the InfoStat interface, version 2020 [30]. For visualizing plot distribution according to soil texture in relation to typologies, the ggplot2 [31], soiltexture [32], and ggtern [33] packages were used in R. The Principal Component Analysis (PCA), Monte Carlo tests, and co-inertia analyses were performed using the ade4 package [34] and factoextra package [35] in R software, version 4.4.2 [36].

3. Results

3.1. General Soil Characteristics in Cocoa-Based Agroforestry Systems in the Amazonian Piedmont

According to the results obtained from the Principal Component Analysis (PCA), it was found that cocoa-based agroforestry systems (cAFSs) were significantly separated (p < 0.001; 33% of explained variance) (Figure 2). Axis 1 (32% of explained variance) separated cAFSs in relation to soil compaction and stability gradient (Figure 2c). Axis 2 (18.81% of explained variance) separated cAFSs with higher yield, pod size, and grain (Figure 2d). When analyzing the relationships between variables and grain yield, it was observed that sand content (S) showed a significant positive correlation (r = 0.36), indicating that this variable positively affects productivity, possibly due to better soil infiltration and aeration. However, clay content (C) had a significant negative correlation (r = −0.31), suggesting that an excess of it limited yield due to problems such as compaction or poor drainage. Silt (Si) also showed a significant negative correlation (r = −0.24), indicating that a higher silt content negatively affected yield.

The weighted mean diameter of aggregates (WMD) showed a significant positive correlation (r = 0.32), highlighting the importance of aggregate stability in productivity. Similarly, structural stability (SSA) showed a significant positive correlation (r = 0.25), emphasizing the relevance of stable soil structure. Macro-aggregates (MaG) were also positively correlated with yield (r = 0.31), while meso-aggregates (MeG) and micro-aggregates (MiG) showed significant negative correlations (r = −0.27 and r = −0.25, respectively), indicating that an excess of these was detrimental. Regarding biogenic aggregates (BA%), they showed a significant negative correlation (r = −0.27), while non-aggregates (NA%) had a significant positive correlation (r = 0.29), which might have reflected soil characteristics that favored productivity. Finally, visual evaluation of soil structure (VESS) had a significant negative correlation (r = −0.26), suggesting that poorer structure reduced yield.

Regarding hydric variables, bulk density (BD) showed a significant negative correlation (r = −0.34), indicating that higher density limited yield. In contrast, total porosity (TP) had a significant positive correlation (r = 0.37), highlighting its importance for water retention and aeration. Penetration resistance (PR) showed a significant negative correlation (r = −0.31), suggesting that higher resistance impaired root growth. Residual volumetric water content (T_r) had a significant positive correlation (r = 0.37), indicating that better water retention favored yield. Additionally, saturated hydraulic conductivity (K_s) also showed a significant positive correlation (r = 0.36), associated with better infiltration and water availability. Productive variables showed significant correlations with yield: pod fresh weight (r = 0.45), pod diameter (r = 0.48), and wet grain weight per pod (r = 0.48) correlated positively, while pod index showed a negative correlation (r = −0.45). In conclusion, soil physical and hydric characteristics, along with productive variables, played a crucial role in agricultural yield. Proper management that optimized structural stability, porosity, and water retention, while minimizing the negative effects of compaction and excess of certain aggregates, maximized productivity. Furthermore, pod characteristics, such as weight and diameter, were key factors in achieving high yields.

3.2. Soils in Cocoa-Based Agroforestry Systems in the Amazonian Piedmont

Different soil types with contrasting characteristics in terms of cocoa production were identified using physical and hydrological variables in cAFSs. The soil types were i. Sandy loam, with high drainage capacity and productivity (HDP), ii. sandy clay loam with medium structural stability (MSS), and iii. sandy clay loam with low structural stability and productivity (LSP). Generally, regarding physical properties, MSS and HDP soils have similar sand content (69.35% and 71.04%, respectively), while LSP soils contain less sand (53.39%) and more clay (22.62%) and silt (23.99%). Figure 3 shows the distribution of plots according to soil texture.

Texture directly influences soil structure. The HDP soils exhibit higher weighted mean diameter (WMD: 2.17 mm) and aggregate structural stability (SSA: 66.95%) compared to LSP soils (WMD: 1.02 mm, SSA: 44.78%,

Table 1. Physical and hydraulic variables as well as productive variables in soils of cacao-based agroforestry systems implemented in the Amazonian foothills.

Parameters	Variables	Acronym	MSS	HDP	LSP	General	p-Value
Physical	Sand	S	69.49 ± 1.15 a	71.2 ± 1.42 a	53.39 ± 0.88 b	66.22 ± 0.96	<0.0001	***
	Clay	C	20.57 ± 1.01 b	19.14 ± 1.01 b	22.62 ± 0.67 a	20.62 ± 0.59	0.0128	**
	Silt	Si	9.95 ± 0.66 b	9.67 ± 0.74 b	23.99 ± 0.75 a	13.17 ± 0.69	<0.0001	***
	Weighted mean diameter	WMD	0.84 ± 0.05 a	0.95 ± 0.04 a	0.47 ± 0.04 b	0.79 ± 0.03	<0.0001	***
	Structural stability of aggregates	SSA	84.75 ± 0.95 b	88.5 ± 0.97 a	63.08 ± 2.79 c	80.79 ± 1.24	<0.0001	***
	Aggregate aggregation index	AAI	2.05 ± 0.13 a	2.08 ± 0.12 a	1.78 ± 0.11 a	2 ± 0.07	0.0852	ns
	Macro-aggregates	MaG	22.12 ± 0.98 b	24.95 ± 0.94 a	12.27 ± 1.16 c	20.66 ± 0.74	<0.0001	***
	Meso-aggregates	MeG	11.94 ± 0.68 a	10.45 ± 0.64 b	12.97 ± 0.43 a	11.73 ± 0.39	0.0056	**
	Micro-aggregates	MiG	6.07 ± 0.39 b	4.62 ± 0.39 c	14.77 ± 1.11 a	7.67 ± 0.52	<0.0001	***
	Root aggregates %	RA	5.57 ± 0.53 b	6.05 ± 0.71 b	8.46 ± 0.93 a	6.39 ± 0.44	0.0232	*
	Physical aggregates %	PA	3.71 ± 0.61 a	5.04 ± 0.97 a	2.69 ± 0.54 a	3.87 ± 0.43	0.093	ns
	Biogenic aggregates %	BA	28.57 ± 2.32 b	23.23 ± 2.52 b	53.08 ± 3.01 a	32.72 ± 1.82	<0.0001	***
	Organic aggregates %	OA	1.78 ± 0.13 a	1.69 ± 0.15 a	1.37 ± 0.12 b	1.65 ± 0.08	0.034	*
	Non aggregates %	NA%	60.41 ± 2.61 a	63.97 ± 3.01 a	34.4 ± 2.57 b	55.36 ± 1.94	0.023	*
	Visual Evaluation of Soil Structure	VESS	2.62 ± 0.05 b	2.69 ± 0.07 b	2.92 ± 0.06 a	2.71 ± 0.03	0.0011	***
Hydric	Bulk density	BD	1.26 ± 0.03 b	1.11 ± 0.03 c	1.61 ± 0.02 a	1.34 ± 0.02	<0.0001	***
	Real density	RD	2.53 ± 0.01 a	2.54 ± 0.01 a	2.52 ± 0.01 a	2.53 ± 0.01	0.7121	ns
	Total porosity	TP	50.26 ± 1.11 b	56.04 ± 1.21 a	36.29 ± 0.68 c	48.72 ± 0.93	<0.0001	***
	Penetration resistance	PR	217.87 ± 6.12 b	201.27 ± 6.12 b	264.48 ± 9.71 a	223.82 ± 4.59	<0.0001	***
	Residual volumetric water content	Tr	0.14 ± 0.0008 a	0.14 ± 0.00096 a	0.13 ± 0.00043 b	0.13 ± 0.00062	<0.0001	***
	Saturated volumetric water content	Ts	0.49 ± 0.0028 a	0.49 ± 0.0029 a	0.47 ± 0.0022 b	0.49 ± 0.0018	<0.0001	***
	Relationship between matrix pressure and water content	a	0.01 ± 0.00011 a	0.01 ± 0.0001 a	0.01 ± 0.0002 b	0.01 ± 0.00014	<0.0001	***
	Shape of the water retention curve	n	1.22 ± 0.0017 b	1.22 ± 0.0019 b	1.26 ± 0.0031 a	1.23 ± 0.0021	<0.0001	***
	Saturated hydraulic conductivity	Ks	19.73 ± 0.53 a	20.45 ± 0.62 a	14.45 ± 0.44 b	18.71 ± 0.39	<0.0001	***
	Gravimetric humidity	HG	28.13 ± 0.51 b	30.98 ± 0.66 a	22.16 ± 0.38 c	27.53 ± 0.44	<0.0001	***
	Infiltration	Infil	2.49 ± 0.06 a	2.44 ± 0.02 a	2.33 ± 0.03 b	2.44 ± 0.03	0.0014	***
	Structural stability index	St	3.11 ± 0.21 a	3.32 ± 0.21 a	2.82 ± 0.16 a	3.11 ± 0.12	0.1779	ns
	Soil erodibility	K	0.16 ± 0.0047 b	0.16 ± 0.01 b	0.23 ± 0.0041 a	0.18 ± 0.004	<0.0001	***
	Field capacity	FC	53.12 ± 1.32 a	57.05 ± 1.42 a	38.32 ± 0.92 b	50.83 ± 1.01	<0.0001	***
Productive	Pod fresh weight	PFW	1008.69 ± 51.13 b	1705.16 ± 72.64 a	689.48 ± 14.43 c	1144.28 ± 48.26	<0.0001	***
	Pod length	PL	20.31 ± 0.2 b	22.36 ± 0.26 a	21.01 ± 0.29 b	21.09 ± 0.16	<0.0001	***
	Pod diameter	PD	86.64 ± 0.75 b	97.22 ± 0.84 a	88.74 ± 1.07 b	90.33 ± 0.65	<0.0001	***
	Weight of grain plus placenta	WGPP	159.65 ± 4.46 c	236.71 ± 5.55 a	206.94 ± 8.48 b	194.09 ± 4.53	<0.0001	***
	Weight of grains per pod	WGP	143.17 ± 3.62 c	211.88 ± 4.8 a	185.69 ± 7.11 b	173.96 ± 3.91	<0.0001	***
	Dry grain weight	WGD	57.27 ± 1.45 c	84.75 ± 1.92 a	74.28 ± 2.84 b	69.59 ± 1.56	<0.0001	***
	Number of grains per pod	NGP	39.12 ± 0.59 b	44.13 ± 0.94 a	44.74 ± 0.99 a	41.96 ± 0.51	<0.0001	***
	Pod husk weight	PHW	865.52 ± 51.7 b	1493.28 ± 71.88 a	503.79 ± 9.93 c	970.32 ± 47.04	<0.0001	***
	Grain index	GI	1.46 ± 0.03 c	1.93 ± 0.04 a	1.65 ± 0.04 b	1.65 ± 0.03	<0.0001	***
	Pod index	PI	18.05 ± 0.45 a	12.03 ± 0.29 c	14.12 ± 0.65 b	15.3 ± 0.36	<0.0001	***
	Harvest index1	HI1	0.46 ± 0.04 c	1.07 ± 0.1 a	0.76 ± 0.09 b	0.71 ± 0.05	<0.0001	***
	Grain yield1	Yield1	372.86 ± 30.97 c	863.82 ± 84.43 a	618.16 ± 74.53 b	579.14 ± 38.97	<0.0001	***
	Harvest index2	HI2	0.71 ± 0.06 b	1.39 ± 0.09 a	0.79 ± 0.07 b	0.93 ± 0.05	<0.0001	***
	Grain yield2	Yield2	571.13 ± 47.15 b	1129.07 ± 71.37 a	640.31 ± 55.33 b	756.18 ± 40.12	<0.0001	***

^{a, b, c}: Means with a common letter between rows are not significantly different at 5% probability. The * indicates the level of significance: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), and not significant (ns). i. Sandy loam, with high drainage capacity and productivity (HDP). ii. Sandy clay loam with medium structural stability (MSS). iii. Sandy clay loam with low structural stability and productivity (LSP).

). Macroaggregates predominate in HDP soils (24.95%), while microaggregates are more abundant in LSP soils (14.77%). Hydrologically, HDP soils display lower bulk density (1.11 g cm⁻³) and higher total porosity (56.04%), favoring infiltration (2.44 mm h⁻¹) and saturated hydraulic conductivity (20.45 mm h⁻¹, Table 1). The LSP soils show higher bulk density (1.61 g cm⁻³), lower porosity (36.29%), and field capacity (38.32%) compared to HDP soils (57.05%). In terms of productivity, HDP soils register superior fresh pod weight (1705.16 g) and grain yield (1129.07 g), contrasting with lower values in LSP soils (538.33 g and 323.97 g, respectively). The harvest index is also superior in HDP soils (HI1: 1.07, HI2: 1.39) versus LSP soils (HI1: 0.63, HI2: 0.92, Table 1). Below is a detailed description of each soil type found in cocoa-based agroforestry systems in the Amazonian piedmont:

3.2.1. Sandy Loam with High Drainage and Productivity (HDP)

The analyzed HDP soils exhibit optimal physical, hydraulic, and productive characteristics for sustainable agricultural activities. The textural composition shows predominance of sand (71.23 ± 1.41%), with moderate proportions of clay (19.14 ± 1.01%) and low silt content (9.67 ± 0.74%) (Table 1). Structural stability is high, as evidenced by the weighted mean diameter (WMD = 0.95 ± 0.04 mm) and aggregate structural stability (SSA = 88.52 ± 0.97%) (Table 1). Hydraulic properties show low bulk density (BD = 1.11 ± 0.03 g cm⁻³) and adequate total porosity (TP = 56.04 ± 1.21%) (Table 1). Field capacity (FC = 57.05 ± 1.44%) and saturated hydraulic conductivity (K_s = 20.45 ± 0.61 mm h⁻¹) indicate good water retention and movement (Table 1). Productive indicators are significant, with high fresh pod weight (PFW = 1705.16 ± 72.64 g) and grain yield (Yield2 = 1129.07 ± 71.37 g) (Table 1). Harvest indices (HI1 = 1.07 ± 0.14; HI2 = 1.39 ± 0.09) suggest high efficiency in biomass conversion (Table 1).

3.2.2. Sandy Clay Loam with Medium Structural Stability (MSS)

The MSS soils exhibited specific textural and structural characteristics that directly influence their functionality. Particle size analysis revealed a composition dominated by sand (69.49 ± 1.15%), followed by clay (20.57 ± 1.01%) and silt (9.95 ± 0.66%), configuring a sandy clay loam texture that favors soil drainage and aeration (Table 1). The structural evaluation demonstrated moderately favorable values, with a weighted mean diameter (WMD) of 0.84 ± 0.05 mm and an aggregate structural stability (SSA) of 84.75 ± 0.95%. The aggregation index (AAI = 2.05 ± 0.12) indicates adequate particle cohesion. Aggregate distribution showed predominance of macro-aggregates (22.12 ± 0.98%) and meso-aggregates (11.94 ± 0.68%), with significant presence of biogenic aggregates (28.57 ± 2.32%) (Table 1). Physical properties evidenced favorable conditions for plant development, with a bulk density of 1.26 g cm⁻³ and particle density of 2.53 g cm⁻³. Total porosity of 50.26% suggests an adequate balance in the water-air relationship. Hydric parameters showed residual volumetric water content of 0.14 and saturated of 0.49, with field capacity of 53.12% and saturated hydraulic conductivity of 19.73 mm h⁻¹ (Table 1).

Penetration resistance (217.87 kPa) and structural stability index (3.11) indicate moderately favorable physical-mechanical conditions. Erodibility (0.16) suggests low susceptibility to erosion, while the water retention curve (n = 1.22) shows gradual water release for plants (Table 1). In terms of productivity, significant results were observed in pod development, with an average fresh weight of 1008.69 ± 51.13 g and dimensions of 20.31 ± 0.24 cm in length and 86.64 ± 0.75 mm in diameter. Pod grain weight reached 143.17 ± 3.62 g, with an average of 39.12 ± 0.59 grains per pod. Productive efficiency indices showed harvest values (HI1 = 0.46 ± 0.04; HI2 = 0.71 ± 0.06) that reflect variable biomass to yield conversion (Table 1). Grain yield showed two scenarios: Yield1 with 372.86 ± 30.97 g and Yield2 with 571.14 ± 47.15 g, suggesting a differential response to management conditions. The grain index (1.46 ± 0.03) and pod index (18.05 ± 0.45) indicate moderate efficiency in grain production relative to total pod biomass (Table 1).

3.2.3. Sandy-Clay Loam with Low Structural Stability and Productivity (LSP)

The soil exhibited a textural composition of 53.39 ± 0.88% sand, 22.62 ± 0.67% clay, and 23.99 ± 0.75% silt, with a weighted mean diameter of 0.47 ± 0.04 mm (Table 1). Structural analysis revealed moderate aggregate stability (SSA = 63.08 ± 2.79%) and aggregation index (AAI = 1.78 ± 0.14), with predominant distribution in micro-aggregates (14.77 ± 1.11%) and meso-aggregates (12.97 ± 0.43%). Biogenic aggregates constituted 53.08 ± 3.01% of the total (Table 1). Physical properties indicated significant compaction with bulk density of 1.61 ± 0.02 g cm⁻³ and limited total porosity (36.29 ± 0.68%). Penetration resistance reached 264.48 ± 9.71 kPa (Table 1). Hydrological characteristics showed moderate water retention capacity (FC = 38.32 ± 0.92%) with saturated hydraulic conductivity of 14.45 ± 0.44 mm h⁻¹. The structural stability index (2.82 ± 0.16) and erodibility (0.23 ± 0.0041) indicated susceptibility to degradation (Table 1). Agricultural productivity parameters demonstrated moderate performance with fresh pod weight of 689.48 ± 14.43 g and pod length of 21 ± 0.29 cm. Grain indices (GI = 1.65 ± 0.04) and harvest indices (HI1 = 0.76 ± 0.09; HI2 = 0.79 ± 0.07) reflected suboptimal resource utilization, resulting in yields of 618.16 ± 74.53 g (Yield1) and 640.31 ± 55.33 g (Yield2) (Table 1).

3.3. Productivity and Its Relationship with Physical and Hydrological Variables in Amazonian Piedmont Soils

Soil texture showed significant correlations, where sand content demonstrated positive correlation in MSS (r = 0.36), HDP (r = 0.34), and LSP (r = 0.45) soils, suggesting that higher sand content favors productivity through better infiltration and aeration. In contrast, clay content showed negative correlation in general soils (r = −0.31), MSS (r = −0.33), and HDP (r = −0.33), indicating that excess clay could limit productivity through compaction or poor drainage (

Table 2. Pearson correlation between grain yield (kg ha⁻¹) and the different physical, hydrological and productive variables.

Parameters	Acronym	General	MSS	HDP	LSP
Physical	S	0.36 ***	0.34 **	0.45 **	0.27
	C	−0.31 ***	−0.33 **	−0.33 *	−0.002
	Si	−0.24 **	−0.09	−0.44 *	−0.31
	WMD	0.32 ***	0.16	0.37 *	0.11
	SSA	0.25**	−0.07	0.63 ***	0.07
	AAI	0.07	−0.26 *	0.35 *	0.22
	MaG	0.31 ***	0.08	0.51 ***	0.12
	MeG	−0.27 ***	−0.13	−0.38 *	−0.16
	MiG	−0.25 **	0.06	−0.63 ***	−0.07
	RA%	−0.12	0.02	−0.25	−0.24
	PA%	−0.05	−0.22	−0.15	−0.08
	BA%	−0.27 ***	−0.24	−0.23	0.0017
	OA%	−0.04	−0.14	0.08	−0.1
	NA%	0.29 ***	0.27 *	0.29	0.1
	VESS	−0.26 ***	−0.26	−0.42 *	−0.38 *
Hydric	BD	−0.34 ***	−0.07	−0.41 **	−0.06
	RD	0.21*	0.19	0.45*	0.02
	TP	0.37 ***	0.16	0.52 ***	0.09
	PR	−0.31 ***	−0.29 *	−0.44 **	0.15
	Tr	0.37 ***	0.35 **	0.49 ***	0.29
	Ts	0.34 ***	0.34 **	0.33 *	0.13
	a	0.22 *	0.13	0.29	0.36
	n	−0.27 ***	−0.26	−0.42 **	−0.35
	Ks	0.36 ***	0.38 ***	0.38 *	0.19
	HG	0.28 ***	−0.17	0.35 *	0.08
	Infil	−0.05	−0.13	0.28	−0.15
	St	0.02	−0.25	0.38 *	−0.26
	K	−0.24 **	−0.11	−0.37 *	−0.32
	FC	0.19 *	−0.11	0.17	0.13
Productive	PFW	0.45 ***	−0.14	0.38 *	0.55 ***
	PL	0.37 ***	0.16	0.07	0.17
	PD	0.48 ***	−0.01	0.33	0.42 *
	WGPP	0.49 ***	0.25	0.09	0.43 *
	WGP	0.48 ***	0.18	0.16	0.46 **
	WGD	0.48 ***	0.18	0.14	0.46 **
	NGP	0.29 ***	0.07	0.18	0.33
	PHW	0.42 ***	−0.15	0.38 *	0.47 **
	GI	0.46 ***	0.19	−0.11	0.46 **
	PI	−0.45 ***	−0.17	−0.08	−0.44 *
	HI1	0.18 *	−0.06	−0.47 ***	0.64 ***
	Yield1	0.18 *	−0.06	−0.47 ***	0.64 ***
	HI2	1 ***	1 ***	1 ***	1 ***

The values of the correlation coefficients are presented according to the significance level: p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***). i. Sandy loam, with high drainage capacity and productivity (HDP). ii. Sandy clay loam with medium structural stability (MSS). iii. Sandy clay loam with low structural stability and productivity (LSP). Meaning of the acronyms in Table 1.

). Structural parameters showed that the Weighted Mean Diameter correlated positively in general soils (r = 0.33) and HDP (r = 0.37), while Aggregate Structural Stability showed highly significant correlation in HDP soils (r = 0.63). The Aggregation Index presented positive correlation in HDP (r = 0.35) and general soils (r = 0.07), but negative in MSS soils (r = −0.26) (Table 2).

Aggregate distribution indicated that macroaggregates showed significant positive correlation in general soils (r = 0.31) and HDP (r = 0.51), while microaggregates correlated negatively in general soils (r = −0.25) and HDP (r = −0.63). Biogenic aggregates showed negative correlation in general soils (r = −0.27) and MSS (r = −0.24) (Table 2). Physical properties showed that bulk density correlated negatively in general soils (r = −0.34) and HDP (r = −0.41), indicating that greater compaction limited yield. Particle density showed positive correlation in general soils (r = 0.21) and HDP (r = 0.46). Total porosity presented significant positive correlation in general soils (r = 0.37) and HDP (r = 0.58), highlighting its importance for water infiltration and storage (Table 2). Penetration resistance correlated negatively in general soils (r = −0.31), MSS (r = −0.29), and HDP (r = −0.44), evidencing the adverse effect of compaction. Hydric parameters indicated that residual volumetric water content correlated positively in general soils (r = 0.37), MSS (r = 0.35), and HDP (r = 0.49), while saturated water content showed positive correlation in general soils (r = 0.34), MSS (r = 0.34), and HDP (r = 0.33), highlighting the importance of water retention for productivity (Table 2). Visual Evaluation of Soil Structure (VESS) showed significant negative correlation across all soil types, being more pronounced in LSP soils (r = −0.43), suggesting that degraded structure negatively affects yield (Table 2).

The correspondence analysis conducted in cacao agroforestry systems revealed significant patterns of association between specific genotypes and edaphic characteristics (R = 0.59; p < 0.0001). The total system variability was primarily explained by two dimensions in Figure 4, where the first dimension (Dim1) represented 90.9% of the variance, while the second dimension (Dim2) explained the remaining 9.1%. This variance distribution indicates a strong structure in genotype-environment association patterns. Three clearly differentiated soil typologies were identified: HDP (sandy loam soils with high drainage capacity), MSS (sandy clay loam soils with medium structural stability), and LSP (sandy clay loam soils with low structural stability). The spatial distribution of clones in relation to these soil typologies demonstrated specific preferential associations. The CCN-51 and ICS-60 clones exhibited a marked affinity with HDP soils, suggesting optimal adaptation to high drainage conditions. On the other hand, the ICS-95, Yellow Hybrid, Red Hybrid, and TSH-565 clones showed a preferential association with MSS soils, indicating their better adaptation to intermediate structural stability conditions. Particularly notable was the association of the FEC-2 clone with LSP soils, evidenced by its peripheral position in the ordination diagram, suggesting an adaptive specialization to conditions of low structural stability. The spatial arrangement of clones in the correspondence analysis not only reflects specific adaptation patterns but also provides valuable information for optimizing cacao agroforestry systems through directed selection of genetic material based on predominant edaphic characteristics.

Figure 5 shows the relationships between the matrices of physical and hydrological variables, as well as between these and productive variables. The relationship between physical and hydrological variables was high with an RV of 0.73 (p < 0.001, Figure 5a). In the case of agronomic variables in relation to hydraulic and physical variables, the RV was 0.15 and 0.13 (p < 0.001, Figure 5b,c).

4. Discussion

Cacao cultivation in the Amazon has had great social importance in recent years [37] as it has allowed for increased family well-being. However, due to the climate variability situation in the Amazon, cacao producers have developed different arrangements of cacao-based agroforestry systems (cAFSs) to address this situation [6,7], especially due to drought conditions that have occurred. It is important to mention that cacao is a highly drought-sensitive species, making high soil water content essential, especially in regions with marked dry seasons [38]. In this context, cAFSs stand out for their sensitivity to climate change and water deficit [39], but also for their key role in sustainable land use under drought conditions [40]. Furthermore, these systems do not generate undesirable environmental impacts [41] and contribute to resilience against water stress in tropical regions [42].

Our results show that soil physical and hydraulic variables explained 33% of the variance, which allowed the identification of three soil typologies differentiated mainly by their level of compaction and structural stability, based on the results obtained from principal component analysis (PCA). These differences were reflected in grain and pod size, as well as in productive yield. In the cacao cultivation areas of the Amazonian foothills, three soil types were identified. i. Sandy loam with high drainage capacity and productivity (HDP): These soils have a high sand content, which favors drainage and aeration, along with notable structural stability. These characteristics translate into high productivity and greater pod weight and size. ii. Sandy clay loam with medium structural stability (MSS): with a balanced texture, these soils offer moderate conditions for cultivation, although they present limitations in water retention, which can affect productivity in dry periods. iii. Sandy clay loam with low structural stability and productivity (LSP): these soils, with higher clay and silt content, show compaction problems, low porosity, and lower yield. Their water retention capacity and aeration are limited, which negatively impacts plant development.

Among soil properties, physical aspects are fundamental for agricultural development, as they directly influence the soil’s capacity to retain water, infiltration rate, root-reachable depth, and consequently, crop yield. Within these physical properties, soil texture, defined by its granulometric composition, is one of the most important attributes, as it affects parameters such as bulk density (BD), porosity, structure, and water retention capacity [43]. For example, sandy soils, with low water retention capacity and high leaching potential, compromise cacao productivity, especially in dry years [44]. On the other hand, highly clayey soils can favor flooding and anaerobic conditions, which also negatively affects productivity [45]. These observations are explained by the inverse relationship between BD and clay, and between BD and total porosity (TP), as clay fraction minerals are essential for maintaining soil particles flocculated and stable in aggregates [46]. However, the soils in which cacao-based agroforestry systems were implemented showed higher grain yields where sand content was significantly higher.

Soil structure influences gas exchange, porosity, bulk density, and water retention [47]. Bulk density is a key indicator for evaluating soil structure, as it reflects compaction and porosity, being influenced by agricultural management practices [48]. Porosity, in turn, conditions essential processes such as gas exchange, water dynamics, penetration resistance, and consequently, root growth and water and nutrient absorption [49]. It is important to note that soil pores are divided into macropores, which facilitate drainage and ventilation, and micropores, essential for water retention. The analyzed HDP soils present a notable balance between their physical, hydraulic, and productive properties, making them a valuable resource for sustainable agriculture. While their high sand proportion ensures good drainage, the low proportion of micro-aggregates and high percentage of non-aggregated particles highlight the need to implement sustainable management practices, such as organic matter incorporation, to improve water and nutrient retention. Additionally, the hydraulic and productive characteristics reflect high agricultural yield potential, although it would be ideal to optimize cultivation conditions to maximize grain production efficiency and improve long-term sustainability.

It is important to mention that production was positively related mainly to sand content, which affects better infiltration and aeration; however, excess clay and silt affected compaction and drainage. At the soil structure level, it is important to highlight the Weight Mean Diameter (WMD) and Aggregate Structural Stability (SSA) on water retention and erosion resistance, consequently affecting pod filling. Additionally, the presence of macro-aggregates and micro-aggregates had positive and negative effects on productivity, respectively. In turn, the visual soil evaluation (VESS) explained that poorer visual soil structure resulted in reduced yield.

Wang et al. [50] emphasized that agroforestry systems increase vertical preferential flow and reduce subsurface lateral flow, thus improving water retention capacity. Furthermore, it has been observed that soil quality and infiltration slightly improve in AFSs compared to monocultures, a difference that could be related to plot age after conversion to agroforestry systems [51]. With cacao-based agroforestry systems, a significant effect on soil hydrological properties has been reported, such as that found by Arévalo-Gardini et al. [52] in the Peruvian Amazon, where a positive impact on plant-available water is related to soil porosity and bulk density. On the other hand, shadeless cacao monocultures are more vulnerable to climate change [53]. In contrast, shade trees reduce surface soil water loss through cacao transpiration [54]. Canopy structure in AFSs significantly influences evapotranspiration (ET), regulating water balance [55].

In subhumid zones, cacao trees face important soil physical limitations, such as low macroporosity, insufficient moisture, and high penetration resistance, which can reduce yield. Correlations between soil physical attributes vary according to the sampled layer and climatic zone, although penetration resistance showed no correlation with other physical attributes. Infiltration has a significant impact on erosion control, runoff, soil moisture content, and groundwater recharge [56]. In AFSs, the topsoil retains more moisture compared to monocultures [54]. However, Sarto et al. [57] noted that AFSs can reduce water availability in the surface layer during rainy seasons, suggesting that trees’ ability to recharge groundwater depends on factors such as climate, soil texture, and fracture depth. For example, research has shown that loam soils offer greater water availability than clay loam or sandy soils [58]. Similarly, infiltration can be lower in sandy soils than in clayey soils [56], highlighting the importance of soil structure and aggregation for adequate infiltration.

Cocoa agroforestry systems in the Amazon Piedmont have emerged as a crucial adaptive response to climate variability, especially under drought conditions. Recent research has identified three soil types that explain 33% of the total variance in productivity. High Drainage and Productivity (HDP) soils, characterized by their sandy loam texture, show superior structural stability and are associated with higher productivity, being particularly favorable for CCN-51 and ICS-60 clones. Medium Structural Stability (MSS) soils, with sandy clay loam texture, present moderate conditions and better adaptation for clones such as ICS-95 and yellow and red hybrids. Lastly, Low Stability (LSP) soils, with higher clay and silt content, face compaction challenges and are commonly used by FEC-2 and CCN-51 clones. Established correlations indicate that productivity increases with higher sand content (r = 0.36) and total porosity (r = 0.37), while decreasing with clay content (r = −0.31) and bulk density (r = −0.34). Agroforestry systems have shown to improve vertical preferential flow and reduce subsurface lateral flow, optimizing water retention capacity. Shade trees play a fundamental role in reducing soil water loss through cocoa transpiration, while canopy structure significantly influences evapotranspiration. Water availability is critical for cocoa yield, with studies showing that annual radiation and precipitation during the dry season can explain up to 70% of variation in annual production. Finally, cacao yield is strongly influenced by water availability. Rainfall scarcity and water table fluctuations affect pod setting and maturation, negatively impacting productivity [38,59]. For example, Moser et al. [60] observed a decrease in yield during periods of intense drought, while simulations conducted by Zuidema et al. [38] indicated that annual radiation and precipitation during the dry season explain up to 70% of the variation in annual cacao yield. Management recommendations include implementing practices that enhance soil structure, specific clone selection based on soil conditions, and maintaining an optimal balance between drainage and water retention, all while carefully considering soil texture in agroforestry system planning.

5. Conclusions

The study results reveal significant findings about soil characteristics in cocoa-based agroforestry systems (cAFSs) in the Amazon Piedmont. Principal Component Analysis (PCA) identified three distinct soil typologies that explain 33% of total variance: sandy loam soils with high drainage capacity and productivity (HDP), sandy clay loam soils with medium structural stability (MSS), and sandy clay loam soils with low structural stability and productivity (LSP). The physical and hydric soil properties proved to be crucial factors directly influencing cocoa yield. Sand content showed a significant positive correlation with yield (r = 0.36), while clay and silt content showed negative correlations (r = −0.31 and r = −0.24, respectively). Structural stability, measured by weighted mean diameter (WMD) and aggregate structural stability (SSA), showed significant positive correlations with yield (r = 0.32 and r = 0.25, respectively). Hydric variables, such as bulk density and total porosity, also proved to be determinant, with correlations of r = −0.34 and r = 0.37, respectively. The correspondence analysis between cocoa clones and soil typologies revealed specific associations, where CCN-51 and ICS-60 clones showed better performance in HDP soils, while ICS-95, Yellow Hybrid, Red Hybrid, and TSH-565 clones adapted better to MSS soils. The results emphasize the importance of considering physical and hydric soil properties in the management of cocoa agroforestry systems, as well as the appropriate selection of clones according to soil characteristics to optimize crop productivity and sustainability in the Amazon region.