The Role of Forest Conversion and Agroecological Practices in Enhancing Ecosystem Services in Tropical Oxisols of the Amazon Basin

Abstract

This study evaluates the effectiveness of agroecological practices—organic fertilization and biofertilization—in enhancing ecosystem services in agroforestry and pasture systems. A field experiment was conducted over three years, comparing these practices to a control treatment and a natural ecosystem as a reference. Soil chemical, physical, and biological parameters were assessed, including soil organic carbon (SOC), microbial respiration, root density, and gene abundances of key microbial groups (Archaea, Bacteria, and Fungi). Organic fertilization resulted in a significant increase in SOC, phosphorus, microbial biomass, and root density, indicating improved soil structure and fertility. Biofertilization showed selective effects, promoting archaeal abundance but reducing bacterial and fungal diversity. Seasonal variation influenced nutrient cycling, with organic fertilization buffering against dry-season declines in microbial activity and nutrient availability. Aboveground dry biomass and litter deposition were highest in the natural ecosystem, followed by organic fertilization treatments in agroforestry and pasture systems. Despite improvements under agroecological management, the natural ecosystem consistently maintained superior soil quality and biological resilience. The findings highlight that organic inputs and diversified cropping systems enhance soil health but do not fully replicate the ecological benefits of undisturbed forests. In conclusion, agroecological practices provide viable alternatives to mitigate soil degradation and sustain ecosystem services in tropical Oxisols. Organic fertilization emerges as the most effective strategy, fostering long-term improvements in soil fertility and microbial dynamics. However, continued research is needed to optimize these practices for greater resilience and sustainability in Amazonian agroecosystems.

1. Introduction

The Amazon Basin, renowned for its unparalleled biodiversity and vast carbon reserves, has experienced significant deforestation due to the conversion of forests into agricultural lands [1]. Agriculture stands as the second leading cause of forest conversion in the Amazon, leading to substantial soil erosion, river siltation, and contamination from agrochemicals [2]. In 2023, the Amazon rainforest experienced a deforestation rate of less than 4000 km², primarily in Brazil, Peru, and Bolivia. This extensive deforestation not only diminishes biodiversity but also disrupts essential ecosystem services, including carbon sequestration and climate regulation [3].

Implementing agroecological practices, such as organic fertilization and biofertilization, within agroforestry systems and pastures has been shown to mitigate soil quality decline. Agroforestry systems, which integrate trees with crops or livestock, have been found to enhance soil fertility and physical properties [4]. For instance, the establishment of agroforestry systems over areas previously occupied by extensive pasturelands has been shown to enhance soil fertility by 12.79% and improve soil physical indicators by 6.5% [5]. These practices promote nutrient recycling, improve soil structure, and increase biodiversity, thereby countering the adverse effects of conventional agriculture [6].

Seasonal variations in tropical regions, characterized by distinct dry and rainy periods, significantly influence aboveground biomass production, litter dynamics, soil fertility, soil physical properties, and microbial gene diversity [7]. During the rainy season, increased moisture availability typically enhances plant growth, leading to higher biomass accumulation and litter deposition [8]. Conversely, the dry season may reduce microbial activity and nutrient mineralization rates, affecting soil fertility and structure [7]. These seasonal fluctuations necessitate adaptive management strategies to maintain ecosystem functionality and productivity throughout the year [6].

Over time, agroecological practices can lead to substantial improvements in both provisioning and regulating ecosystem services [9]. Provisioning services, such as crop yield and timber production, benefit from enhanced soil fertility and structure, while regulating services, including carbon sequestration, erosion control, and water regulation, are bolstered by increased vegetation cover and biodiversity [4]. Long-term studies have demonstrated that agroforestry systems can significantly reduce soil erosion rates and improve soil health indicators compared to monocultures, thereby sustaining agricultural productivity and environmental quality [10].

We aimed to investigate the impact of agroecological practices—control, biofertilization, and organic fertilization—on ecosystem services (provisioning and regulating services) in three distinct agroecosystems: an agroforestry system, a natural ecosystem (Amazon rainforest), and pasture. This study evaluated how these practices influence both provisioning services (e.g., aboveground dry biomass and litter deposition) and regulating services (e.g., soil carbon, nutrient cycling, and erosion control) in Tropical Oxisols within the Amazon Basin. We hypothesized that (a) the conversion of natural ecosystems to pasture significantly reduces both provisioning and regulating ecosystem services in the Amazon basin; (b) agroforestry systems that mimic the belowground root diversity of natural ecosystems can enhance regulating services, which in turn increases provisioning services compared to pasture; and (c) agroecological practices can mitigate the negative impacts on both regulating and provisioning services in simplified ecosystems such as pasture. These hypotheses are grounded in three key cause–effect pathways: (i) provisioning services are an outcome of effective regulating services [11]; (ii) degradation of regulating services diminishes provisioning capacity [12]; and (iii) overexploitation of provisioning services weakens regulating services [5].

2. Materials and Methods

2.1. Experimental Sites, Soil Classification, and Climate Characterization

The experiment was carried out in three sites across the Amazon Basin (Cruzeiro do Sul, AC, 7°37′17″ S 72°42′43″ W, 192 m.a.s.l; Boca do Acre, AM, 8°45′12″ S 67°23′09″ W, 104 m.a.s.l; and Cerejeiras, RO, 13°10′07″ S 61°14′29″ W, 194 m.a.s.l.) from 2022 to 2024, with soil sampling during the dry and rainy seasons (e.g., following a semestral schedule of sampling). The soil in the three study sites was described as reddish Oxisol with a uniform sandy loam texture and fine granular structure [13]. Details about the climate classification, average temperature, annual maximum temperature, annual precipitation, and relative humidity are provided in

Table 1. Climate profiles of study sites across the Amazon Basin.

Site	Climate Type 1	Average Temperature	Annual Maximum Temperature	Annual Precipitation	Relative Humidity
Cruzeiro do Sul, AC	Af	High, generally above 25 °C, little variation	Stable, generally between 30 °C and 33 °C	Over 1800 mm, no dry season	Very high, generally above 80% year-round
Boca do Acre, AM	Am	Similar to Af, around 25°, slight seasonal variation	Similar to Af, slight peaks during dry season	1800–2500 mm, short dry season	High, slight decrease during dry season
Cerejeiras, RO	Aw	High, between 20 °C and 30C, more seasonal variation	Greater variation, can exceed 35 °C during the dry season	1000–1800 mm, well-defined dry season	Variable, high during wet season, lower (50–60%) in dry season

¹ According to Köppen–Geiger’s climate classification.

.

2.2. Studied Ecosystems—Classification and Characterization

In each studied site, we analyzed the effects of three agroecological practices in two agroecosystems (Agroforestry system and pasture) and an additional treatment by considering a natural ecosystem. The basal soil physicochemical properties of the studied ecosystems in the three studied sites are provided in

Table 2. Baseline physico-chemical properties of soils in each studied ecosystem prior to the experiment (2022). Values are presented as means (n = 15 per site).

Ecosystems		Site
	Cruzeiro do Sul	Boca do Acre	Cerejeiras
	Bulk Density (g/cm3)
Agroforestry system	1.06	1.05	1.06
Pasture	1.16	1.09	1.12
Natural Ecosystem	0.99	0.99	1.02
		Soil Porosity (mm3/mm3)
Agroforestry system	35.4	34.9	35.6
Pasture	36.1	36.7	35.9
Natural Ecosystem	60.3	61.9	59.3
		Soil pH
Agroforestry system	5.1	5.5	5.3
Pasture	4.9	5.2	5.1
Natural Ecosystem	4.8	4.6	4.5
		Soil Organic Carbon (g/kg)
Agroforestry system	20.1	22.3	19.8
Pasture	19.9	22.3	24.9
Natural ecosystem	249.6	259.7	255.6

.

The agroforestry system consisted of integrating Coffea arabica L. (coffee), Hevea brasiliensis L. (rubber tree), Zea mays L. (maize), and Phaseolus vulgaris L. (common bean). This system is classified as a diversified land use approach that optimizes both economic and ecological functions. In this system, C. arabica was grown under partial shade provided by H. brasiliensis, which not only supports coffee production by moderating microclimate conditions but also offers latex as an additional income source. P. vulgaris was intercropped between the rows, contributing to nitrogen fixation through its symbiosis with Rhizobium bacteria, improving soil fertility for the other crops. We expected that this agroforestry system would enhance biodiversity, promote nutrient cycling, and minimize soil erosion, while also providing multiple products, such as coffee beans, rubber, maize, and legumes. The shade-tolerant nature of coffee in this context helps improve soil structure and water retention, enhancing ecosystem resilience [5].

For the pasture ecosystem, we studied well-conserved pastures of Urochloa brizantha (Hochst. ex A.Rich.) R.Webster in the Amazon Basin. U. brizantha is a robust and drought-tolerant grass species, commonly used due to its capacity to produce high-quality forage while maintaining soil structure and preventing erosion. In well-managed systems, rotational grazing practices help sustain the pasture’s productivity by allowing time for the grass to recover, reducing the risk of overgrazing and land degradation. U. brizantha also contributes to regulating services, such as soil carbon sequestration, by storing carbon in its root systems and improving soil organic matter content. These pastures, when well-maintained, can be resilient to both drought and heavy rainfall, offering a sustainable forage solution while reducing negative environmental impacts typical of poorly managed pastures [14].

Finally, we considered in each study site a primary forest fragment of the Amazon tropical rainforest as a reference treatment. The primary forest fragments considered in this study were characterized by rich biodiversity and endemic tree species occurrence, such as Handroanthus serratifolius (Vahl) S.Grose, Buchenavia tetraphylla (Aubl.) R.A.Howard, Hura crepitans L., Albizia niopoides (Benth.) Burkart, Apuleia leiocarpa (Vogel) J.F.Macbr., Barnebydendron riedelli (Tul.) J.H.Kirkbr., Copaifera multijuga L., Hymenaeae courbaril L., Parkia paraensis Ducke, Schizolobium parahyba (Vell.) Blake, Eschweilera grandiflora (Aubl.) Sandwith, Ceiba pentandra Gaertn., Ceiba samauma (Mart.) K.Schum., Sterculia apetala (Jacq.) H.Karst., Cedrela odorata L., Castilla ulei Warb., Ficus insipida Willd., Astronium lecointei Ducke, Eschweilera bracteosa (Poepp. ex O.Berg) Miers., Bertholletia excelsa Humb. & Bonpl., Euterpe precatoria Mart., and Dipteryx odorata Willd. These endemic species play key roles in the provision of both provisioning services (e.g., fruits, nuts, and timber) and regulating services (e.g., climate regulation, water cycling, and carbon storage). The complex stratification of the rainforest, with its canopy, understory, and ground layers, supports a highly diverse community of flora and fauna, contributing to nutrient cycling and the maintenance of soil fertility. This ecosystem is essential for stabilizing the regional climate and sustaining hydrological cycles, while its dense vegetation cover helps prevent soil erosion and promote biodiversity conservation. The ecological interactions within this system are fundamental for the long-term resilience and functioning of the Amazon Basin, making it a critical focus for conservation efforts [15].

2.3. Agroecological Practices—Brief Characterization

As agroecological practices, we considered three treatments: foliar biofertilization, organic fertilization with cow manure, and a control treatment (non-fertilization). Foliar biofertilization (a mixture of fresh cattle manure, unrefined sugar, fresh milk, and yeast in a 20:1:1:1 × 10⁻³ ratio) in the Amazon Basin is a common agroecological practice aimed at enhancing plant nutrition by applying nutrient-rich solutions directly to the leaves. In our study, farmers applied foliar biofertilizers at doses of 2 L ha⁻¹. The application rate was once every two weeks during the growing season, with a focus on critical growth stages, such as flowering and fruiting. The main objective of foliar biofertilization was to improve nutrient uptake efficiency, increase plant resistance to stress (e.g., drought, pests), and boost crop yields without relying on synthetic fertilizers [16]. Organic fertilization with cow manure is a widely adopted agroecological practice in the Amazon Basin, aimed at improving soil fertility and structure. In the experiment, farmers applied cow manure at doses of 25 tons per hectare. The application was performed once a year by incorporating it into the soil during land preparation. The primary objectives of using cow manure are to replenish soil nutrients, increase soil organic matter, and promote sustainable crop production, particularly in tropical soils that are prone to nutrient depletion [17]. Finally, the control treatment provided a baseline comparison, against which the effects of biofertilization and organic fertilization can be measured. All agroecological practices were applied and monitored from 2022. The chemical composition of foliar biofertilizer and organic fertilizer are shown in

Table 3. Chemical composition of foliar biofertilizer and organic fertilizer used in the field experiment. Values are stated as mean (n = 20).

		Organic Fertilizer
C/N Ratio	N (g kg−1)	P (g kg−1)	K (g kg−1)
21.13	20.85	16.18	31.18
Biofertilizer
pH	Ca2+ (mmolc L−1)	Mg2+ (mmolc L−1)	Na+ (mmolc L−1)
5.1	27.5	64.4	6.0
K+ (mmolc L−1)	C (mmolc L−1)	P (mmolc L−1)	NH4+ (mmolc L−1)
15.7	130.0	270.7	0.8
SO42− (molc L−1)	B (mg L−1)	Cu (mmolc L−1)	Cu (mmolc L−1)
7.6	0.7	0.2	44.9
Mn (mmolc L−1)	Zn (mmolc L−1)	Moisture (%)	CO32− (mmolc L−1)
11.3	1.0	0.7	0.0
HCO3− (mmolc L−1)
160.0

.

2.4. Experimental Design

A field experiment was conducted at three independent sites from 2022 to 2024, with soil sampling considering the seasonal variation between the dry and rainy seasons. We analyzed the effects of agroecological practices in an agroforestry system, and a pasture ecosystem on both provisioning and regulating services. The experiment in each site followed a randomized block design using a factorial scheme of 3 × 2 × 2 + 1, with agroecological practices (control, biofertilization, and organic fertilization), agroecosystems (agroforestry system and pasture), seasons (dry and rainy), and an additional reference treatment (natural ecosystem) as the main factors, replicated in five blocks. The studied plots measured 50 × 60 m. Within each plot designated for ecosystem service measurements, we conducted sampling during both the dry and rainy seasons from 2022 to 2024. Sampling activities included the following:

(i) Plant material collection for estimating aboveground dry biomass and yield: In the natural ecosystem and for perennial plants (e.g., C. arabica and H. brasiliensis) in the agroforestry system, aboveground biomass was estimated using species-specific allometric equations. In the pasture system, aboveground fresh biomass was measured by clipping plant material within a 1.5 × 1.5 m area and later drying it to estimate the dry biomass and productivity.

(ii) Soil sampling for physical and biochemical properties: For physical properties, we collected undisturbed soil samples using volumetric cylinders (5.57 cm diameter × 4.1 cm height). For biochemical analyses, disturbed soil samples were taken from the top 20 cm using a soil auger.

(iii) Litter sampling was conducted using a metallic square frame (1.5 × 1.5 m) placed on the soil surface. Collected litter was used to estimate litter deposition and litter quality, including lignin, carbon (C), nitrogen (N), and phosphorus (P) contents. In pasture areas, litter consisted of dead leaves, stems, and other plant debris accumulated on the surface within the same 1.5 × 1.5 m area.

(iv) Soil monoliths (20 × 20 × 20 cm) were collected at the end of the dry and rainy seasons each year, in a randomly selected location within each plot. These samples were used to assess root density and the abundance of soil bacteria, fungi, and archaea via qPCR.

2.5. Aboveground Dry Biomass and Yield

To estimate total aboveground dry biomass (_tADB) at the plot level, we tailored our approach to the traits of each agroecosystem (agroforestry system, natural ecosystem, and pasture). We began by selecting fifty plants from the middle of each plot. Annual plant species were harvested at the aboveground level, while perennial plant species had their biomass estimated using allometric equations. Fresh biomass of annual plants was air-dried, then oven-dried at 65 °C for 24 h. For yield calculations, grains were dried and standardized to 14% moisture. _tADB estimates were converted from kg ha⁻¹ to t ha⁻¹. Below, we detail the methods used for estimating the total aboveground dry biomass across the studied agroecosystems:

• In the agroforestry system, aboveground dry biomass and yield were measured at the physiological maturity of P. vulgaris and Z. mays. Biomass was estimated using plant material, while the grain data were used for yield estimates. For H. brasiliensis and C. arabica, aboveground biomass was calculated with allometric equations: ADB_{(H. brasiliensis)} = exp(−3.31 + 0.95(lnD²H)) × 1.02 and ADB_{(C. arabica)} = 0.117 × D^1.732 × H^0.760. Here, ADB is aboveground dry biomass (kg/plant), D is diameter at breast height (m), and H is plant height (m). Total aboveground dry biomass (t ha⁻¹) was calculated as follows: _tADB = [(PD × ADB_{(H. brasiliensis)}) + (PD × ADB_{(C. arabica)}) + (PD × (ADB + yield_{(P. vugaris)}) + (PD × (ADB + yield_{(Z. mays)})] × 0.00334, where PD is plant density (plants/plot), and 0.00334 converts kg/plot to t/ha.
• In the natural ecosystem, aboveground dry biomass (t ha⁻¹) was estimated using the following equation: ADB = −8.261 × (D × 1.737) × (L × 0.891) × (P × 0.969), where ADB is aboveground dry biomass, D is diameter at breast height (cm), L is commercial stem length (m), and P is basic wood density (g cm⁻³) [18].
• In the pasture ecosystem, aboveground dry biomass (t ha⁻¹) was estimated by collecting fresh biomass using metallic squares (1 × 1 m), air drying the fresh biomass until a constant weight, estimating the dry biomass, and applying the dry biomass results in the following equation: ADB = PDB × 10, where ADB is aboveground dry biomass (t ha⁻¹), PDB is the plant dry biomass (kg m⁻²), and 10 converts kg/m² to t/ha [19].

2.6. Litter Characterization

To determine the litter deposition, litter was collected in the rainy and dry seasons from 2022 to 2024. Three metallic squares (1 × 1 m) were placed on the soil surface in each plot, with sampling points randomly selected using digital maps and geographic coordinates. Litter material within the squares was gathered into plastic bags, then air-dried at 60 °C for 48 h until a constant dry biomass was achieved [20]. Nutrient contents (C, N, and P) were analyzed following [21], and lignin content was determined using Klason’s method [22].

2.7. Soil Physical and Biochemical Characterization

To physically characterize soil samples, we determined the bulk density, soil moisture, soil porosity, soil resistance, geometric mean diameter (GMD), and weighted mean diameter (WMD). Bulk density was estimated using the soil dry weight and cylinder volume, as described by [23]. To determine soil moisture, and soil porosity, we followed the protocols described by [24]. Soil resistance was estimated using an electronic bench pen [24]. The GMD and WMD were estimated following the protocol described by [25].

For the soil biochemical properties, we analyzed soil pH, soil organic carbon, total nitrogen, available P, microbial respiration, microbial carbon, and microbial nitrogen. In each plot, we collected five samples, and these samples collected at each sampling point were mixed. However, the samples for each plot were measured separately. The soil pH was measured in a suspension of soil and distilled water (1:2.5, v:v, soil: water suspension), as described by [26]. The soil organic carbon was determined by the rapid dichromate oxidation method [27]. The total nitrogen was determined using sulfuric acid and potassium sulphate digestion, as described by Kjeldahl [26]. Soil P was estimated colorimetrically using a spectrophotometer at 882 nm by extraction with Mehlich-1. The microbial respiration was determined by the incubation method, while for the microbial carbon and nitrogen, we used the fumigation-extraction method [28].

2.8. Density of Fine Roots and Abundance of Soil Bacteria, Fungi, and Archaea via qPCR

To characterize the root density of each agroecosystem, we collected three soil monoliths (20 × 20 × 20 cm) at each plot. The monoliths were wrapped in plastic film and transported with minimal disturbance until analysis. To estimate the density of fine roots, we just collected fine roots (diameter less than 2 mm) from the soil monoliths. These roots were washed using a 0.5 mm nylon mesh bag. Root dry biomass (g) was determined after drying the samples for 48 h at 65 °C. Root density was calculated by dividing the root dry biomass by the monolith volume.

To characterize the abundance of soil microorganisms via qPCR, we consider the abundance of Archaea, Bacteria, and Fungi. Archaea indicate microorganisms that play a crucial role in nutrient cycling, particularly in extreme or anaerobic environments. Bacteria signify a diverse group of microorganisms essential for decomposing organic matter, fixing nitrogen, and promoting plant health. Fungi represent organisms that decompose complex organic materials, form symbiotic relationships with plants (mycorrhizae), and contribute to soil structure and fertility. We separated bulk from rhizospheric soil using a standardized protocol from [29]. The dsDNA was extracted from frozen rhizosphere soil samples using the protocol described by [30]. The dsDNA was quantified with the QuantiFluor^TM system using 485 nm excitation and 520 nm emission in the FLUOstar Omega microplate reader (BMG Labtech, Ortenberg, Germany). To account for DNA loss during extraction, dsDNA content and marker gene abundances were corrected by dividing the data by the extraction efficiency, as described in [31]. Archaea, Bacteria, and Fungi were quantified via quantitative real-time PCR (qPCR), targeting ITS1 for fungi and the 16S rRNA gene for archaea and bacteria, using the LightCycler 480 SYBR Green I Master and Probes Master in the LightCycler 480 Instrument II. Primers used were the following: (i) NSI1 and 58A2R for Fungi [32,33]; (ii) BAC338F and BAC805R with probe BAC516F for Bacteria; and (iii) ARC787F and ARC1059R with probe ARC915F for Archaea [32]. Reaction mixtures, cycling conditions, and cloning fragments for qPCR standards were used as proposed by [30]. Standard preparation was performed as described by [33].

2.9. Models for Provisioning and Regulating Services

We developed two primary artificial models: (i) Provisioning Services (PS), representing the direct products obtained from the studied plots, including above- and belowground biomass and litter; and (ii) Regulating Services (RS), which encompass processes that maintain environmental balance and soil health, such as nutrient cycling, erosion control, and soil quality. In constructing these models, we considered the following: (i) PS is a function of aboveground dry biomass, litter deposition, and fine root density; and (ii) RS is a function of litter quality and soil physicochemical properties. The models were developed using the “stepwise” procedure. Microbial gene diversity and dsDNA content were excluded from these models, as they are associated with supporting services, which fall outside the aim of this manuscript. Additionally, we identified three key factors influencing the artificial models for provisioning (PS) and regulating services (RS): seasonal dependency, where PS and RS are influenced by seasonal variation (SV); ecosystem dependency, where PS and RS are shaped by the ecosystem type (ECO); and management dependency, where PS and RS are affected by agroecological practices (AP). To deepen our understanding, we also analyzed interactions between these factors, considering PS and RS as functions of SV and ECO; SV and AP; AP and ECO; and the combined influence of SV, AP, and ECO.

2.10. Statistical Analysis

First, all data about aboveground dry biomass, litter deposition, litter quality, soil physical properties, soil chemical properties, root density, and microbial gene diversity were tested for normality and homoscedasticity using the “shapiro.test” and “bartlett” functions in the “stats” and “dplyr” packages, respectively. Using the “decostand” function provided in the “vegan” package, we log- transformed all data to meet the required criteria. To address the complexity of our experimental design, we initially performed a comprehensive ANOVA that accounted for all relevant sources of variation, including study sites (df = 2), years (df = 2), seasons (df = 1), agroecological practices (df = 2), ecosystems (df = 2), and their interactions. This full model was applied to evaluate treatment effects on aboveground dry biomass, litter deposition, litter quality, soil physical and chemical properties, root density, and microbial gene abundance.

Based on the significant effects identified in the full model, we then conducted three focused explanatory ANOVAs to better represent the variability in specific variables:

• Year × Agroecological practice:

Applied to aboveground dry biomass; litter deposition; litter lignin; litter C, N, and P contents; root density; bacteria abundance; fungi abundance; archaea abundance; and ds DNA. Seasons and ecosystems were used as replicates, as they did not show significant effects on these variables.

2.

Seasons × Agroecological practice × Ecosystem:

Applied to soil pH, SOC, total N, P, microbial C, and microbial N. Years were used as replicates, as they were not statistically significant for these variables.

3.

Season × ecosystem:

Applied to litter lignin and litter nutrients (N, P, and C). Years and agroecological practices were used as replicates.

4.

Ecosystem × Agroecological practice:

Applied to bulk density, soil moisture, soil porosity, soil resistance, GMD, and WMD. Years and seasons were used as replicates.

In all cases, study sites were treated as replicates, as there were no significant differences among them. Bonferroni’s test was used as the post hoc test (p < 0.05). To construct the predictive models for provisioning and regulating services, we used multiple linear regression analyses. The step() function from the stats package was used for stepwise model selection based on the Akaike Information Criterion (AIC), allowing automatic inclusion or exclusion of predictors to identify the most parsimonious models. Prior to model fitting, all predictor variables were checked for multicollinearity using the vif() function from the car package, and variables with VIF > 5 were either excluded or replaced by alternative predictors to reduce collinearity. Model assumptions (normality of residuals, homoscedasticity, and independence) were validated using diagnostic plots from the ggplot2 and performance packages. Adjusted R² and F-values were calculated to assess model fit and explanatory power. Interaction terms for ecosystem, seasonal, and management dependency were added using manually defined categorical variables, and their effects were incorporated into separate models. All visualizations were produced using ggplot2, and statistical outputs were formatted using broom and dplyr packages. All statistical analyses were conducted using R software, 3.4.0 [34].

3. Results

3.1. Effect of Agroecological Practices on Aboveground Biomass and Litter Deposition over Time

Significant interactions between year and agroecological practices were observed for both total aboveground dry biomass and litter deposition (Figure 1). For these variables, we did not observe significant differences between seasons and sites, thus they were used as replicates. The natural ecosystem exhibited the highest values for total aboveground dry biomass (Figure 1A) and litter deposition (Figure 1B). Notably, litter deposition in organic fertilization after the second year of the experiment was statistically comparable to that of the natural ecosystem. Over three years, total aboveground dry biomass increased exponentially by 19.31% and 14.78% under organic fertilization and biofertilization, respectively, while the control showed a 33.98% exponential decline. Compared to the natural ecosystem, agroecological practices resulted in significantly lower total aboveground dry biomass values (Figure 1A). Litter deposition increased exponentially by 30.25% under organic fertilization and decreased by 28.30% in the control, with no significant changes observed in biofertilization or the natural ecosystem over time (Figure 1B).

3.2. Variation in Litter Quality Across Ecosystems and Seasons

The two-way ANOVA revealed significant interactions between ecosystems and seasons for litter lignin and C, N, and P contents (

Table 4. Litter lignin and nutrient (N, P, and C) contents (mean ± standard deviation) as influenced by the interaction among seasons and ecosystems across the Amazon Basin.

Ecosystems	Litter Lignin Content (%)	Litter C Content (g/kg)	Litter N Content (g/kg)	Litter P Content (g/kg)
		Dry season
Agroforestry system	57.4 ± 2.8 Ab 1	379.0 ± 2.7 Ab	9.05 ± 0.24 Bb	0.85 ± 0.02 Ba
Pasture	66.8 ± 2.9 Aa	380.0 ± 2.6 Ab	12.80 ± 0.59 Ba	0.84 ± 0.07 Ba
Natural ecosystem	44.8 ± 1.1 Ac	391.0 ± 6.2 Aa	9.87 ± 0.04 Ab	0.69 ± 0.02 Ab
		Rainy season
Agroforestry system	53.6 ± 3.6 Bb	372.0 ± 3.3 Bb	11.5 ± 0.45 Ab	0.97 ± 0.03 Aa
Pasture	65.6 ± 2.4 Aa	375.0 ± 3.2 Bb	14.76 ± 0.70 Aa	0.97 ± 0.03 Aa
Natural ecosystem	42.8 ± 2.1 Ac	386.0 ± 3.6 Ba	10.8 ± 0.21 Ab	0.72 ± 0.04 Ab

¹ Bold values represent the highest significant values. Lowercase letters indicate significant differences among ecosystems within each season, while capital letters denote significant differences between seasons within each ecosystem, as determined by Bonferroni’s test (p < 0.05).

). Litter lignin and C contents were consistently highest during the dry season, whereas litter N and P contents peaked during the rainy season. The interaction between seasons and ecosystems showed that the highest litter lignin values occurred under pasture during the dry season, while the natural ecosystem exhibited the highest litter C content during the same season. In contrast, the pasture ecosystem during the rainy season recorded the highest litter N and P contents. Although ecosystems had a significant positive effect on litter P content in the rainy season, no significant differences were observed between the agroforestry system and pasture for this parameter (Table 4).

Our results revealed significant interactions among agroecological practices over time for litter lignin content and litter C, N, and P contents (

Table 5. Litter lignin and nutrient (N, P, and C) contents (mean ± standard deviation) as influenced by the agroecological practices over time across the Amazon Basin.

Agroecological Practices		Litter Lignin Content (%)
	2022	2023	2024
Biofertilization	59.0 ± 0.04 a 1	59.4 ± 0.10 a	59.5 ± 0.05 b
Control	58.6 ± 0.05 a	61.9 ± 0.03 a	65.0 ± 0.02 a
Organic fertilization	62.9 ± 0.13 a	62.2 ± 0.18 a	58.8 ± 0.10 b
Natural ecosystem	42.0 ± 0.01 b	44.6 ± 0.07 b	44.8 ± 0.06 c
		Litter C content (g/kg)
Biofertilization	380. ± 7.06 c	380. ± 6.24 c	381. ± 6.25 c
Control	376. ± 7.39 c	354. ± 9.76 d	311. ± 1.28 d
Organic fertilization	395. ± 5.72 a	401. ± 4.76 a	410. ± 4.48 a
Natural ecosystem	387. ± 5.48 b	387. ± 4.10 b	391. ± 2.78 b
		Litter N content (g/kg)
Biofertilization	9.62 ± 0.58 c	10.8 ± 1.20 b	11.4 ± 1.46 b
Control	9.13 ± 0.90 c	8.10 ± 0.92 c	5.46 ± 1.12 c
Organic fertilization	13.7 ± 3.36 a	18.7 ± 6.46 a	20.7 ± 2.76 a
Natural ecosystem	10.2 ± 0.50 b	10.4 ± 0.69 b	10.4 ± 0.83 b
		Litter P content (g/kg)
Biofertilization	0.85 ± 0.04 b	0.95 ± 0.10 b	0.93 ± 0.05 b
Control	0.76 ± 0.05 c	0.56 ± 0.03 d	0.40 ± 0.02 d
Organic fertilization	1.10 ± 0.13 a	1.27 ± 0.18 a	1.36 ± 0.10 a
Natural ecosystem	0.68 ± 0.01 c	0.70 ± 0.01 c	0.73 ± 0.06 c

¹ Bold values represent the highest significant values. Lowercase letters indicate significant differences among agroecological practices within each year, as determined by Bonferroni’s test (p < 0.05).

). Litter nutrient contents (C, N, and P) were consistently highest in organic fertilization during the three consecutive years. For litter lignin content, the highest values were recorded under control during the third year of the field experiment (Table 5).

3.3. Impact of Agroecological Practices on Soil Physical Properties

We found no significant effects of seasons or years on soil physical properties. Instead, all physical variables were influenced solely by the interaction between agroecological practices and ecosystems (p < 0.001). Among the practices, organic fertilization yielded the best results, with lower bulk density and soil resistance and higher soil moisture, porosity, geometric mean diameter, and weighted mean diameter. No significant differences were observed between biofertilization and the control for the following: (i) bulk density and soil porosity in the agroforestry system; (ii) soil resistance and soil moisture in both the agroforestry system and pasture; and (iii) geometric mean diameter in the pasture. Compared to the natural ecosystem, the agroforestry system and pasture exhibited significant differences in all soil physical properties, except for soil moisture under organic fertilization in both systems and soil resistance under organic fertilization in the agroforestry system (

Table 6. Soil physical properties (mean ± standard deviation) as influenced by the agroecological practices and ecosystems in the Amazon Basin.

Agroecological Practices	Agroforestry System	Pasture	Natural Ecosystem
		Bulk density (g/cm3)
Biofertilization	1.05 ± 0.02 Ab 1	1.16 ± 0.01 bA
Control	1.06 ± 0.01 aB	1.19 ± 0.03 aA	0.99 ± 0.01 *
Organic fertilization	1.01 ± 0.01 bB	1.14 ± 0.01 bA
		Soil moisture (kg/kg)
Biofertilization	0.31 ± 0.04 bA	0.28 ± 0.08 bA
Control	0.25 ± 0.07 bA	0.24 ± 0.07 bA	0.40 ± 0.07 *
Organic fertilization	0.40 ± 0.01 aA	0.40 ± 0.01 aA
		Soil porosity (mm3/mm3)
Biofertilization	39.8 ± 2.9 bA	41.1 ± 4.2 bA
Control	36.4 ± 4.2 bA	36.7 ± 5.7 cA	62.7 ± 0.8 *
Organic fertilization	54.5 ± 6.1 aA	55.1 ± 6.4 aA
		Soil resistance (%)
Biofertilization	1.32 ± 0.02 aA	1.29 ± 0.01 aA
Control	1.36 ± 0.03 aA	1.32 ± 0.05 aA	1.16 ± 0.03 *
Organic fertilization	1.16 ± 0.04 bA	1.20 ± 0.04 bA
		GMD (mm)
Biofertilization	2.72 ± 0.01 bA	2.37 ± 0.02 bB
Control	2.62 ± 0.09 cA	2.31 ± 0.08 bB	2.97 ± 0.01 *
Organic fertilization	2.82 ± 0.04 aA	2.67 ± 0.10 aB
		WMD (mm)
Biofertilization	2.99 ± 0.01 bA	2.99 ± 0.01 bA
Control	2.78 ± 0.01 cB	2.88 ± 0.11 cA	3.32 ± 0.04 *
Organic fertilization	3.15 ± 0.04 aA	3.11 ± 0.07 aA

Bold values represent the highest significant values. ¹ Lowercase letters indicate significant differences among agroecological practices within each ecosystem, while capital letters denote significant differences between agroforestry system and pasture, as determined by Bonferroni’s test (p < 0.05). Asterisks signify significant differences between the agroecological practice × ecosystem interactions and the natural ecosystem, used as the reference area.

).

3.4. Changes in Soil Chemical Properties Under Agroecological Practices

We found significant effects of seasons, agroecological practices, and the ecosystem on soil chemical properties. Among the practices, organic fertilization yielded the best results, with higher soil pH, soil organic carbon, total nitrogen, P, microbial respiration, microbial C, and microbial N. No significant differences were observed between the rainy and dry seasons for soil organic carbon, total N, P, and microbial N. Compared to the natural ecosystem, the agroforestry system and pasture exhibited significant differences in all soil chemical properties (

Table 7. Soil chemical properties (mean ± standard deviation) as influenced by the interaction among seasons and agroecological practices across the Amazon Basin.

Variables	Agroecological Practices	Agroforestry System	Pasture	Natural Ecosystem
Soil pH		Dry season
	Control	5.35 ± 0.01 aA 1	5.04 ± 0.02 aA	4.80 ± 0.01 *
	Biofertilization	5.53 ± 0.01 aA	5.19 ± 0.01 aA
	Organic fertilization	5.56 ± 0.01 aA	5.26 ± 0.03 aA
		Rainy season
	Control	5.14 ± 0.10 bB	4.93 ± 0.03 aB	4.74 ± 0.04 *
	Biofertilization	5.39 ± 0.08 aB	5.19 ± 0.02 aA
	Organic fertilization	5.45 ± 0.06 aB	5.19 ± 0.05 aA
Soil organic carbon (g/kg)		Dry season
	Control	22.1 ± 5.35 cA	22.4 ± 4.97 cA	259.0 ± 2.44 *
	Biofertilization	31.3 ± 2.65 bA	33.7 ± 5.24 bA
	Organic fertilization	46.6 ± 7.46 aA	40.6 ± 1.01 aA
		Rainy season
	Control	19.7 ± 5.08 cA	19.8 ± 2.83 cA	242.0 ± 2.17 *
	Biofertilization	30.7 ± 2.59 bA	31.9 ± 4.60 bA
	Organic fertilization	43.9 ± 7.70 aA	37.2 ± 6.57 aA
Total N (mg/kg)		Dry season
	Control	2.03 ± 0.28 bA	2.97 ± 0.12 bA	4.25 ± 0.01 *
	Biofertilization	3.01 ± 0.02 aA	3.34 ± 0.02 aA
	Organic fertilization	3.35 ± 0.26 aA	3.62 ± 0.20 aA
		Rainy season
	Control	1.73 ± 0.27 bA	2.74 ± 0.04 bA	3.95 ± 0.07 *
	Biofertilization	2.92 ± 0.10 aA	3.33 ± 0.04 aA
	Organic fertilization	3.27 ± 0.25 aA	3.59 ± 0.14 aA
P (mg/kg)		Dry season
	Control	12.5 ± 2.30 bA	6.70 ± 1.20 cA	63.6 ± 2.45 *
	Biofertilization	16.9 ± 0.84 bA	9.29 ± 0.63 bA
	Organic fertilization	27.8 ± 6.68 aA	12.80 ± 5.75 aA
		Rainy season
	Control	12.3 ± 3.51 bA	6.59 ± 1.33 cA	66.7 ± 0.99 *
	Biofertilization	16.9 ± 0.12 bA	9.30 ± 0.87 bA
	Organic fertilization	26.1 ± 3.67 aA	13.10 ± 4.64 aA
Microbial respiration (mg/kg h)		Dry season
	Control	0.0183 ± 0.002 bA	0.0177 ± 0.003 cB	0.0623 ± 0.002 *
	Biofertilization	0.0257 ± 0.001 bB	0.0273 ± 0.004 bB
	Organic fertilization	0.0417 ± 0.007 aA	0.0447 ± 0.008 aA
		Rainy season
	Control	0.0273 ± 0.010 bA	0.0253 ± 0.007 cA	0.0853 ± 0.004 *
	Biofertilization	0.0383 ± 0.004 aA	0.0353 ± 0.001 bA
	Organic fertilization	0.0457 ± 0.005 aA	0.0483 ± 0.009 aA
Microbial C (g C/kg)		Dry season
	Control	75.5 ± 3.88 cB	68.8 ± 3.11 cA	500.0 ± 1.38 *
	Biofertilization	85.9 ± 5.01 bB	80.1 ± 5.08 bB
	Organic fertilization	98.0 ± 1.69 aB	97.6 ± 1.56 aB
		Rainy season
	Control	85.4 ± 7.23 cA	72.6 ± 7.96 cA	551.0 ± 9.20 *
	Biofertilization	98.7 ± 3.41 bA	91.3 ± 6.87 bA
	Organic fertilization	126.0 ± 2.14 aA	116.0 ± 2.04 aA
Microbial N (mg N/kg)		Dry season
	Control	88.0 ± 0.91 cA	47.9 ± 3.96 cA	131.0 ± 1.71 *
	Biofertilization	118.0 ± 1.56 bA	72.1 ± 0.79 bA
	Organic fertilization	133.0 ± 1.74 aA	94.2 ± 1.77 aA
		Rainy season
	Control	82.5 ± 1.34 cA	45.1 ± 6.38 cA	130.0 ± 0.93 *
	Biofertilization	121.0 ± 1.77 bA	72.2 ± 1.06 bA
	Organic fertilization	137.0 ± 2.37 aA	96.2 ± 1.41 aA

Bold values represent the highest significant values. ¹ Lowercase letters indicate significant differences among agroecological practices within each ecosystem, while capital letters denote significant differences between seasons within each ecosystem, as determined by Bonferroni’s test (p < 0.05). Asterisks signify significant differences between the agroecological practice × ecosystem interactions and the natural ecosystem, used as the reference area.

).

3.5. Effects of Agroecological Practices on Root Density and Microbial Gene Abundance

Significant differences were observed in root density, microbial gene abundance (Archaea, Bacteria, and Fungi), and dsDNA content across agroecological practices and the reference area over the three studied years (

Table 8. Root density (A—g/cm³), Bacteria (B—n × 10⁸ g⁻¹ soil), Fungi (C—n × 10⁷ g⁻¹ soil), Archaea (D—n × 10⁶ g⁻¹ soil), and dsDNA (E—µg g⁻¹ soil) of different agroecological practices vs. a natural ecosystem (reference area) from Amazon basin over three consecutive years.

Agroecological Practices		Root Density (g/cm3)
	2022	2023	2024
Biofertilization	2.03 ± 0.02 c	1.97 ± 0.03 c	2.12 ± 0.02 c
Control	1.94 ± 0.03 c	1.68 ± 0.05 d	1.33 ± 0.03 d
Organic fertilization	2.41 ± 0.04 b	2.56 ± 0.03 b	2.76 ± 0.05 b
Natural ecosystem	3.66 ± 0.06 a	3.68 ± 0.05 a	3.68 ± 0.04 a
		Bacteria (n × 108 g−1 soil)
Biofertilization	65.9 ± 1.17 b	65.7 ± 0.98 b	64.9 ± 1.02 a
Control	67.8 ± 1.21 b	64.3 ± 1.13 b	57.1 ± 0.87 b
Organic fertilization	72.9 ± 1.28 a	70.4 ± 1.23 a	55.6 ± 1.12 b
Natural ecosystem	37.9 ± 1.17 c	37.3 ± 0.56 c	37.4 ± 0.98 c
		Fungi (n × 107 g−1 soil)
Biofertilization	4.05 ± 0.27 b	3.52 ± 0.31 b	3.55 ± 0.25 c
Control	4.15 ± 0.30 b	3.78 ± 0.33 b	3.35 ± 0.29 c
Organic fertilization	3.88 ± 0.17 b	3.90 ± 0.15 b	4.25 ± 0.21 b
Natural ecosystem	5.50 ± 0.21 a	5.35 ± 0.17 a	5.40 ± 0.12 a
		Archaea (n × 106 g−1 soil)
Biofertilization	9.07 ± 1.13 a	13.10 ± 0.97 a	13.80 ± 1.41 a
Control	9.68 ± 0.65 a	8.49 ± 0.79 c	8.18 ± 0.98 b
Organic fertilization	7.72 ± 0.52 b	11.10 ± 0.49 b	7.18 ± 0.62 b
Natural ecosystem	3.23 ± 0.33 c	3.93 ± 0.42 d	3.54 ± 0.23 c
	dsDNA (µg g−1 soil)
Biofertilization	8.05 ± 0.21 c	8.04 ± 0.09 c	8.29 ± 0.12 b
Control	7.78 ± 0.19 c	7.03 ± 0.11 c	5.91 ± 0.08 c
Organic fertilization	9.59 ± 0.29 b	11.50 ± 0.21 b	13.90 ± 0.09 a
Natural ecosystem	12.80 ± 0.18 a	12.50 ± 0.17 a	13.10 ± 0.21 a

Bold values represent the highest significant values. Lowercase letters indicate significant differences among agroecological practices within each year, as determined by Bonferroni’s test (p < 0.05).

). The natural ecosystem maintained stable root density, microbial gene abundance, and dsDNA content over three consecutive years. In the control, there was a consistent exponential decline in root density, microbial gene abundance, and dsDNA content. Organic fertilization led to an exponential increase in root density, fungal gene abundance, and dsDNA content, while causing an exponential decrease in bacterial gene abundance and a polynomial increase in archaeal gene abundance. Biofertilization slightly increased root density, archaeal gene abundance, and dsDNA content, while reducing bacterial and fungal gene abundance exponentially (Table 8).

3.6. Models of Provisioning and Regulating Services

Based on the results obtained in the stepwise procedure to understand the multivariate factors that improved both provisioning and regulating services, we have built main, specific, and interaction models considering multiple regressions. The proposed models showed the following functions: (1) provisioning services being influenced by fine root density, litter deposition, and total aboveground dry biomass; (2) regulating services being influenced by litter N content, litter P content, bulk density, geometric mean diameter, soil pH, soil organic carbon, and soil P; (3) in the seasonal dependency model, we observed that the rainy season showed a higher influence on both ecosystem services when compared with the dry season; (4) for ecosystem dependency, we observed that the natural ecosystem accounts for 47.2%, the agroforestry system for 22.4%, and pasture for 7.8%, while the control contributes the least at −42.4% in its specific model; (5) for management dependency, the analysis shows that the reference area (natural ecosystem) holds the highest importance at 55.8%, followed by organic fertilization at 10.1%, biofertilization at 5.6%, and control, with the lowest importance, at −28.5% (

Table 9. Predictive models to estimate provisioning and regulating services based on total aboveground biomass (tADB), litter deposition (LD), litter N content (LNC), litter P content (LPC), fine root density (FRD), bulk density (BD), geometric mean diameter (GMD), soil pH, soil organic carbon (SOC), and soil P, using a database obtained in our field study across the Amazon Basin.

Predictive Model	F-Value	adj.R2	p-Value
Main models
PS 1 = −7.32 + (1.12 × tADB) + (1.18 × LD) + (1.49 × FRD)	189.93	0.98	<0.001
RS = −7.32 + (1.50 × LNC) + (1.48 × LPC) + (2.98 × BD) + (8.94 × GMD) + (1.05 × pH) + (5.75 × SOC) + (5.96 × P)	281.31	0.99	<0.001
Specific models
Seasonal dependency = PS_RS + (8.081 × β)	21.37	0.61	<0.05
Ecosystem dependency = PS_RS + (3.657 × α)	13.14	0.94	<0.001
Management dependency = PS_RS + (2.55 × γ)	11.54	0.95	<0.001
Interaction models
Seasonal/Ecosystem dependency = PS_RS + (3.254 × β) + (8.06 × α)	12.62	0.94	<0.001
Seasonal/Management dependency = PS_RS + (1.667 × β) + (15.31 × γ)	10.88	0.96	<0.001
Management/Ecosystem dependency = PS_RS + (1.157 × α) + (14.11 × γ)	7.01	0.98	<0.01
Seasonal/Ecosystem/Management dependency = PS_RS + (3.092 × β) + (1.669 × α) + (18.848 × γ)	5.73	0.98	<0.05

¹ PS_RS is calculated as the average of PS and RS values. The β value for the dry season is 1.234, while for the rainy season, it is 2.983. The α value for natural ecosystem is 1.424, agroforestry system is 0.676, pasture is 0.234, and control is −1.284. The γ value for control is −2.783, biofertilization is 0.548, organic fertilization is 0.982, and reference area (natural ecosystem) is 5.438.

).

4. Discussion

4.1. Agroecological Practices Enhanced Total Aboveground Dry Biomass and Litter Deposition over Three Consecutive Years, with Seasonal Variation and Interactions with Ecosystems

The implementation of agroecological practices influenced total aboveground dry biomass and litter deposition over three consecutive years, with significant interactions between the year, season, and ecosystems [35]. Although the natural ecosystem exhibited the highest total aboveground dry biomass and litter deposition, organic fertilization in agroforestry systems and pastures demonstrated a capacity to enhance these parameters, particularly after the second year of the experiment. The observed exponential increase in the total aboveground dry biomass under organic fertilization (19.31%) and biofertilization (14.78%) suggests that these practices contribute to long-term productivity, while the control treatment experienced a substantial decline (33.98%). These results align with the habitat provision hypothesis, which posits that increased plant biomass under organic fertilization creates a more favorable microhabitat for soil biota and plant interactions, thereby sustaining productivity in managed ecosystems [16].

Litter deposition followed similar trends, with organic fertilization leading to a 30.25% increase over time, whereas the control exhibited a 28.30% reduction. Despite the overall dominance of the natural ecosystem in litter inputs, organic fertilization after the second year yielded values statistically comparable to those of the natural ecosystem. This suggests that agroecological management strategies, particularly organic fertilization, may enhance the nutrient cycling service by maintaining consistent litter inputs, which in turn sustain soil organic matter and fertility [6]. The habitat provision hypothesis further supports these findings by highlighting the role of plant diversity and organic inputs in fostering soil microbial activity, leading to improved decomposition and nutrient release processes [36].

Seasonal variations significantly influenced biomass accumulation and litter deposition, with differential responses observed across ecosystems and agroecological practices. The highest litter deposition under organic fertilization occurred during the dry season, whereas the natural ecosystem exhibited peak values during the rainy season. This pattern indicates that organic inputs may buffer seasonal fluctuations by stabilizing nutrient availability and enhancing microbial decomposition rates under moisture-limited conditions [37]. The nutrient cycling service is particularly relevant in this context, as increased litter deposition contributes to organic matter pools that support long-term soil fertility and plant productivity, mitigating the adverse effects of seasonal constraints [38].

Despite the positive effects of agroecological practices, total aboveground dry biomass and litter deposition remained significantly lower in managed systems (e.g., agroforestry systems and pasture) compared to the natural ecosystem. This discrepancy may be attributed to the structural complexity and functional redundancy of natural ecosystems, which support diverse plant and microbial communities that optimize resource use efficiency [1]. However, the observed interactions among agroecological practices, seasons, and ecosystems highlight the potential for adaptive management strategies to enhance ecosystem services in agroforestry systems and pastures [39]. By leveraging organic amendments and biofertilization, these systems can partially restore key ecological functions, reducing productivity gaps relative to natural ecosystems [9].

Our results indicate that agroecological practices contribute to improved total aboveground dry biomass and litter deposition over time, with strong seasonal influences and ecosystem-dependent interactions. These findings reinforce the importance of habitat provision and nutrient cycling in sustaining productivity and soil health in tropical and managed landscapes [16].

4.2. Seasonal and Ecosystem-Specific Variations in Litter Quality and the Long-Term Influence of Agroecological Practices

Litter quality, as reflected in lignin, C, N, and P contents, varied significantly across ecosystems and seasons, indicating distinct decomposition dynamics and nutrient cycling processes. Our results revealed strong interactions between agroecological practices and seasonal patterns, with litter lignin and C contents reaching peak values during the dry season, while litter N and P contents were highest in the rainy season. These patterns suggest that moisture availability plays a crucial role in regulating litter decomposition and nutrient release in tropical soils [40]. The highest litter lignin content was recorded in pasture during the dry season, likely due to slower decomposition rates under water-limited conditions, whereas the natural ecosystem exhibited the highest litter C content during the same period, reinforcing the importance of ecosystem structure in modulating carbon inputs and turnover [11].

The interaction between seasonality and agroecological practices further emphasized the role of management systems in shaping litter nutrient dynamics [41]. Pasture during the rainy season showed the highest litter N and P contents, likely influenced by increased microbial activity and nutrient mineralization under high soil moisture conditions [42]. However, while ecosystems significantly affected litter P content in the rainy season, no significant differences were observed between the agroforestry system and pasture, suggesting that agroecological practices may buffer seasonal nutrient fluctuations but do not fully restore natural ecosystem functions [43]. These findings align with the soil quality decline in disturbed ecosystems hypothesis, which posits that alterations in plant inputs and decomposition rates in managed systems contribute to shifts in nutrient cycling and soil fertility over time [10].

Long-term assessments revealed significant interactions between agroecological practices and time for litter quality parameters, with organic fertilization consistently maintaining the highest litter C, N, and P contents over three consecutive years. This indicates that organic amendments may mitigate nutrient depletion in agroecosystems, enhancing litter quality and sustaining soil fertility [16]. Conversely, litter lignin content was highest under the control treatment in the third year, likely reflecting reduced decomposition efficiency and the accumulation of recalcitrant organic material [4]. These findings underscore the role of land degradation driven by plant diversity reduction, as lower plant diversity in managed systems can limit the diversity of litter inputs, slowing decomposition rates and nutrient cycling [44].

Despite the benefits of organic fertilization in maintaining litter nutrient quality, overall differences between managed ecosystems and the natural ecosystem highlight persistent challenges in restoring ecological functions. The higher litter C content in the natural ecosystem suggests that undisturbed forests maintain a more stable organic matter pool, supporting long-term carbon sequestration and soil fertility [1]. In contrast, the accumulation of lignin-rich litter under pasture and control treatments points to slower organic matter turnover, potentially leading to declines in soil quality and increased susceptibility to degradation [10]. These findings reinforce the need for diversified agroecological practices that enhance plant functional diversity and optimize litter inputs to sustain ecosystem services [43]. Seasonal variations and long-term trends in litter quality demonstrate the complex interactions between agroecological management, decomposition processes, and nutrient cycling in Tropical Oxisols. The observed patterns support the soil quality decline hypothesis and highlight the need for adaptive management strategies to counteract land degradation in agroecosystems [10].

4.3. Influence of Agroecological Practices and Ecosystems on Soil Physical Properties in the Amazon Basin

Soil physical properties in the Amazon Basin were significantly influenced by the interaction between agroecological practices and ecosystem types, while seasonal and annual variations had no detectable effects. Organic fertilization yielded the most favorable outcomes, with lower bulk density and soil resistance, alongside increased soil moisture, porosity, and aggregate stability. In contrast, biofertilization and the control did not show significant improvements in key physical parameters, particularly in bulk density, porosity, and soil resistance within the agroforestry system and pasture. These findings align with the habitat simplification hypothesis, which suggests that land use intensification leads to structural changes in soil properties due to altered organic matter inputs and reduced biological activity [45].

The observed differences between the agroforestry system, pasture, and the natural ecosystem highlight the long-term impacts of land use on soil physical quality [5]. Both managed systems exhibited significant differences in all soil physical parameters compared to the reference area, except for soil moisture under organic fertilization and soil resistance in the agroforestry system. The lower bulk density and enhanced porosity under organic fertilization suggest improved soil structure due to organic matter inputs, which enhance aggregation and water retention [45]. However, the persistence of compacted soils in biofertilization and control treatments reinforces the role of soil compaction driven by plant diversity reduction, where simplified plant communities contribute to reduced root penetration and biological activity, limiting soil recovery [46].

Soil resistance and moisture dynamics varied among ecosystems, reflecting distinct land use legacies [5]. While organic fertilization improved the soil structure in both the agroforestry and pasture systems, biofertilization and the control exhibited higher soil resistance and lower moisture retention, comparable to degraded landscapes [47]. These patterns are consistent with findings from land use change studies, where pasture and simplified cropping systems are associated with increased compaction and reduced infiltration [48]. In particular, the geometric mean diameter in pasture remained unaffected by biofertilization, emphasizing the persistent challenges of restoring aggregate stability in heavily managed lands [43]. These results suggest that the combination of reduced plant functional diversity and mechanical disturbances in pasture systems can exacerbate soil degradation, reinforcing the negative feedback loop between land use intensity and soil quality [46].

The significant differences between managed ecosystems and the natural reference area underscore the need for sustainable land management strategies to mitigate soil degradation. The positive effects of organic fertilization demonstrate that targeted agroecological practices can enhance soil structure and function, but their effectiveness depends on long-term implementation and the integration of diverse plant species to promote biological activity and organic matter accumulation [12].

4.4. Long-Term Changes in Soil Chemical Properties Under Agroecological Practices and Their Seasonal Interactions

Agroecological practices significantly influenced soil chemical properties over time, with distinct seasonal and ecosystem-dependent variations. Organic fertilization consistently improved key soil parameters, including pH, soil organic carbon (SOC), total nitrogen (N), phosphorus (P), microbial respiration, microbial carbon (C), and microbial nitrogen (N), regardless of the season. In contrast, the control plots exhibited an exponential decline in these parameters, highlighting the impact of soil management strategies on long-term soil fertility [12]. The habitat simplification hypothesis helps explain these trends, as land use intensification and reduced plant functional diversity limit organic matter inputs and microbial activity, leading to a gradual deterioration of soil chemical properties in unmanaged systems [49].

The interaction between seasons, agroecological practices, and ecosystems further emphasized the importance of management in maintaining soil chemical quality [10]. While seasonal variations significantly influenced most soil properties, no significant differences were observed for SOC, total N, P, and microbial N between the dry and rainy seasons. This stability suggests that organic fertilization supports consistent nutrient availability throughout the year, mitigating seasonal fluctuations [9]. However, agroforestry and pasture systems exhibited significant differences in all soil chemical properties when compared to the natural ecosystem, reinforcing the idea that soil quality is closely linked to plant diversity and organic matter inputs [43]. The higher microbial activity and nutrient retention under organic fertilization indicate that diverse plant–soil interactions and continuous organic inputs sustain microbial communities and nutrient cycling, buffering seasonal effects [50].

Over the three years, soil chemical properties followed divergent trajectories, depending on the management strategy. Organic and biofertilization treatments led to an exponential increase in SOC, soil P, microbial C, and microbial N, while the control plots exhibited a steady decline in these parameters [9]. This pattern reflects the role of organic inputs in enhancing soil fertility through microbial-driven nutrient transformations [51]. In contrast, the natural ecosystem maintained stable levels of SOC, soil P, microbial C, and microbial N, suggesting that intact ecosystems inherently sustain their soil quality [5]. However, these values remained significantly different from those in agroecological treatment plots, indicating that even organic amendments do not fully replicate the nutrient dynamics of undisturbed systems. These findings reinforce the importance of plant diversity and organic matter inputs in maintaining soil quality, as reduced plant complexity in managed systems alters nutrient cycling and microbial community dynamics [39].

The decline in soil chemical properties in control plots supports the idea that habitat simplification contributes to long-term soil degradation [48]. Without external organic inputs, soil fertility deteriorates, leading to a loss of microbial activity and reduced nutrient availability [43]. This process is particularly evident in the pasture ecosystems, where lower plant diversity and minimal organic matter return accelerate soil degradation [49]. By contrast, organic fertilization in both agroforestry and pasture systems demonstrated that targeted agroecological interventions can counteract these negative effects, promoting soil fertility and ecosystem resilience [9].

4.5. Variations in Root Density and Microbial Gene Abundance Under Agroecological Practices over Three Consecutive Years and in Comparison to the Reference Area

Agroecological practices significantly influenced root density and microbial gene abundance over time, with distinct responses observed for Archaea, Bacteria, and Fungi. The natural ecosystem maintained stable values for these parameters, highlighting the resilience of undisturbed systems, where plant diversity and organic matter inputs sustain microbial communities [16]. In contrast, the control plots exhibited an exponential decline in root density, microbial gene abundance, and dsDNA content over three years, suggesting that the absence of management interventions leads to progressive soil degradation [48]. This decline aligns with the habitat simplification hypothesis, which posits that land use intensification reduces plant–soil–microbe interactions, leading to biodiversity loss and impaired nutrient cycling [52].

Organic fertilization had the most pronounced effect on soil biological properties, promoting an exponential increase in root density, fungal gene abundance, and dsDNA content. This suggests that organic inputs enhanced rhizosphere conditions, favoring fungal communities known for their role in nutrient acquisition and organic matter decomposition [20]. Interestingly, bacterial gene abundance declined exponentially under organic fertilization, possibly due to competitive interactions between microbial groups or shifts in substrate availability [42]. Archaeal gene abundance exhibited a polynomial increase, indicating a more complex response to organic inputs. These findings reinforce the critical role of organic matter amendments in fostering microbial diversity and improving rhizosphere function, which are essential for soil fertility and plant productivity [50].

Biofertilization influenced root–microbe interactions differently, leading to a moderate increase in root density, archaeal gene abundance, and dsDNA content, while causing an exponential decline in bacterial and fungal gene abundance. This suggests that biofertilizers selectively stimulated certain microbial groups, particularly Archaea, which are known for their roles in methane cycling and nitrogen transformations [53]. The reduction in bacterial and fungal gene abundance under biofertilization may indicate shifts in microbial community structure (e.g., abundance of soil bacteria, fungi, and archaea), possibly due to altered nutrient dynamics or increased microbial competition [9]. These results highlight the need for a nuanced understanding of microbial responses to biofertilizers, as their effects on rhizosphere communities can vary depending on soil conditions and plant–microbe interactions [54].

The contrasting trends between agroecological practices and the reference area suggest that long-term soil health depends on management strategies that support diverse and functional microbial communities [42]. While the natural ecosystem maintained stable microbial and root density dynamics, the control plots exhibited severe declines, underscoring the risks of land degradation in the absence of intervention [54]. Organic fertilization emerged as the most effective strategy for enhancing root–microbe interactions, while biofertilization showed selective microbial stimulation. These findings emphasize the importance of integrating agroecological approaches that promote soil biological activity, ensuring sustainable nutrient cycling and long-term ecosystem resilience [9].

5. Conclusions

Our study demonstrates that agroecological practices significantly influence key ecosystem services in agroforestry systems and pastures, though their effects remain distinct from those observed in the natural ecosystem. Organic fertilization consistently enhanced total aboveground dry biomass and litter deposition over time, with seasonal variations favoring nutrient cycling and habitat provision, whereas the control exhibited a marked decline in these parameters. Soil chemical properties, including soil organic carbon, phosphorus, and microbial biomass, showed an exponential increase under organic fertilization, while the control plots suffered depletion, reinforcing the role of organic amendments in mitigating soil degradation. In contrast, biofertilization yielded moderate improvements, with selective effects on microbial communities and root density. Soil physical properties were strongly influenced by agroecological practices, with organic fertilization reducing bulk density and soil resistance while enhancing porosity and aggregate stability, thus mitigating compaction associated with plant diversity reduction. Litter quality varied seasonally and across ecosystems, with higher lignin and carbon content in the dry season and increased nitrogen and phosphorus during the rainy season, further emphasizing the role of plant diversity and organic inputs in sustaining soil fertility. Rhizobiome analyses revealed complex microbial responses, with organic fertilization favoring fungal proliferation and biofertilization selectively enhancing archaeal abundance, suggesting differential microbial dynamics in response to agroecological management. Despite improvements under organic fertilization and biofertilization, agroforestry systems and pastures still exhibited lower biological and physicochemical resilience than the natural ecosystem, underscoring the importance of maintaining plant diversity and organic matter inputs to sustain long-term soil health and ecosystem functionality. These findings highlight the critical role of agroecological practices in enhancing provisioning and regulating services, offering insights into sustainable land use strategies that balance productivity and ecological integrity in the Amazon basin.