Priscila Sanjuan de Medeiros-Sarmento
Leandro Valle Ferreira2
Markus Gastauer- 1Instituto Tecnológico Vale, Belém, Pará, Brazil
- 2Museu Paraense Emílio Goeldi, CBO, Belém, Pará, Brazil
In the Amazon, shifting cultivation has historically shaped the landscape and remains a key land-use system practiced by traditional communities. The ecological sustainability of this system depends on the regeneration capacity of forests between cultivation cycles. In this study, we investigated mechanisms regulating the recovery of soil macrofauna during natural regeneration of areas previously managed via slash-and-burn agriculture, evaluating the effects of fallow age, environmental variables, and trophic interactions on macrofauna diversity and community composition. We sampled 40 plots along a successional gradient from 1 to >80 years and collected soil and vegetation data. We used structural equation modeling, generalized additive models, and multivariate analysis to understand observed patterns. Fallow age did not directly affect macrofauna but exerted an indirect effect through vegetation. We also observed a top-down effect of predators on herbivores, detritivores, and geophages, highlighting the role of trophic interactions in structuring soil communities. These findings reinforce that sustainability of shifting cultivation should not be assessed solely on fallow age or vegetation cover but on the capacity of the regenerated system to sustain ecological functions. Given effective vegetation regeneration and a landscape favoring ecosystem resilience, this traditional land-use system can contribute to biodiversity and soil functionality restoration.
1 Introduction
Soil invertebrates play a key role in ecosystem functioning. They regulate biological interactions, promote soil structure formation, and mediate processes such as litter decomposition and nutrient cycling, which are directly linked to forest productivity and regeneration (Lavelle et al., 2006; Serra et al., 2021). Among them, soil macrofauna stand out as sensitive bioindicators of environmental change and serve as valuable proxies for monitoring soil functions in regenerating ecosystems (Demetrio et al., 2022). These attributes are particularly relevant in tropical landscapes subject to shifting agriculture, also known as slash-and-burn agriculture, a traditional land-use system used for thousands of years by communities in the Amazon (Roosevelt, 2013). In this system, small-scale farmers clear forests by fire, cultivate for a few years, and then abandon the area, allowing it to regenerate during fallow periods (Coomes et al., 2021). This cyclical land use creates a mosaic of natural, agricultural, and fallow areas, where regeneration yields secondary forests with varying ages and structural complexity.
Fallow age plays a central role in shaping soil macrofaunal communities, both directly and indirectly. Direct effects can emerge from stochastic processes such as ecological drift and dispersal, which operate independently of environmental filtering (Peniston et al., 2024). These processes are most pronounced in the earliest successional stages, when populations are typically smaller and more isolated, increasing demographic fluctuations and the likelihood of local extinction and thereby reducing richness (Schoereder et al., 2004; Ebenman et al., 2004). Indirect effects arise from gradual shifts in vegetation structure and soil properties that accompany secondary forest development. As fallows age, aboveground biomass and floristic diversity increase, canopies close, litter inputs accumulate, and soils gain organic matter, moisture retention and horizon differentiation (Lintemani et al., 2020; Villa et al., 2021; Mertz et al., 2021). These coupled changes act as environmental filters, favoring species with traits adapted to deeper litter layers, higher habitat heterogeneity and more stable microclimates, thereby progressively reshaping soil macrofaunal communities (Lavelle et al., 2006; Brown et al., 2024).
Building on this temporal template, secondary forests that emerge after fallow periods of 3 to more than 20 years provide important environmental and social benefits, including biodiversity conservation and support for rural livelihoods (Mertz et al., 2021). However, their ecological resilience is far from uniform and varies with management intensity, fallow duration and local environmental conditions (Mukul et al., 2020; Vasconcelos et al., 2020; Serra et al., 2021; Ramos et al., 2025). These sources of variation are directly reflected in soil macrofauna, which are widely used as indicators of ecosystem services and land-use sustainability (Brown et al., 2024) but whose responses to shifting cultivation remain heavily dependent on landscape context and study design.
Finally, as vegetation structure and resource availability change along succession, trophic interactions are also reorganized, with predators potentially regulating the densities of herbivores and detritivores through top-down control within soil food webs (Scheu and Schaefer, 1998; Ponsard et al., 2000). In tropical systems, groups of predators such as ants can strongly influence the abundance of other functional guilds, thereby shaping community assembly and key soil processes, including litter decomposition, bioturbation, soil aggregation and nutrient cycling (Schuldt and Staab, 2015; Stemmelen et al., 2022). Given this interplay among time since fallow (disturbance history), environmental filtering (vegetation and soils), and top-down control, clarifying their relative contributions is essential for understanding soil macrofauna recovery in regenerating forests (Figure 1).
Figure 1. Conceptual framework for soil macrofauna recovery in Amazonian secondary forests. Pathway diagram showing the expected relationships between fallow age and environmental changes (vegetation and soil properties) in the soil macrofaunal community. Regeneration age can directly influence macrofaunal density and diversity, as well as indirectly through changes in vegetation and soil properties. Trophic interactions (predators, herbivores and detritivores) can also modulate the community response to the succession process.
A substantial body of work has examined how deforestation and subsequent fallow periods affect soil macrofauna in tropical shifting-cultivation landscapes (Franco et al., 2019). Along chronosequences, fallows older than 20 years often recover species richness and higher taxa comparable to primary forests (e.g., Yoshima et al., 2013; Serra et al., 2021), yet changes in abundance and community composition are markedly less consistent among studies. This context dependence likely reflects the joint influence of increasing litter inputs and chemical diversity with stand age (Cole et al., 2020) and the activity of ecosystem engineers such as earthworms, termites and ants, which enhance soil microhabitat heterogeneity and feedback on soil structure and function (Lavelle et al., 2016). At the same time, repeated burning and land-use intensification reduce the resilience of secondary forests, depleting seed banks and leading to structurally simplified, low-diversity stands (Jakovac et al., 2015; Mesquita et al., 2015), with expected negative consequences for soil macrofauna (Franco et al., 2019). In the Amazon, however, most studies have focused on broad comparisons among land uses or agroforestry systems (e.g., Rousseau et al., 2021; Villa et al., 2020, 2021), and few explicitly disentangle how fallow age, vegetation structure and soil conditions jointly shape macrofaunal richness, density and composition within shifting-cultivation mosaics.
To address these knowledge gaps, we investigated how soil macrofaunal communities respond to forest regeneration following slash-and-burn agriculture along a 40-year chronosequence and in old-growth forest in the Caxiuanã National Forest (CNF), Pará, Brazil. We evaluated the combined effects of fallow age, vegetation structure, and soil properties on the composition and functional structure of macrofauna. Specifically, we tested three hypotheses: (i) macrofauna density and richness increase with fallow age, mediated by vegetation and soil changes; (ii) macrofauna community composition shifts over time, and these shifts are associated with changes in soil fertility and vegetation patterns; and (iii) trophic interactions shape macrofaunal structure, with predators exerting top-down control. By disentangling the relative roles of fallow age, vegetation and soil complexity, and trophic interactions in shaping soil macrofaunal recovery, our study aims to identify the key ecological drivers of belowground biodiversity regeneration and to provide evidence-based guidance for sustainable land-use practices in tropical shifting cultivation systems.
2 Methods
2.1 Study site
This study was conducted in the Caxiuanã National Forest (CNF), Pará, eastern Amazonia (1°13′86″S; 48°17′41.18″W) (Figure 2), a 330,000-ha terra firme tropical forest dominated by yellow oxisols. The region has a hot and humid climate (Am – Köppen), with annual rainfall ranging from 2,000 to 2,500 mm, a well-defined dry season from July to November (monthly rainfall <100 mm), and mean annual temperatures of approximately 27 °C (Costa et al., 2018). Established in 1961, the CNF is one of Brazil’s oldest federal conservation units, designated for sustainable use and integrating biodiversity conservation, long-term ecological research, and regulated resource extraction. Traditional riverine communities inhabit the area, relying on small-scale shifting cultivation – primarily cassava (Manihot esculenta Crantz) – in addition to fishing and forest-based livelihoods (Lisboa et al., 2013). The landscape is characterized by a mosaic of old-growth forests and secondary forests spanning 1 to 40 years of age, formed through natural regeneration following slash-and-burn agriculture.
Figure 2. Geographic location of the study area in the Caxiuanã National Forest and the sampling plots. The classification into stages of natural regeneration refers to the age after abandonment of shifting agricultural activities. Stage I refers to areas up to 10 years old, Stage II refers to areas 11–25 years in age, Stage III refers to areas 26–40 years in age, and Stage IV refers to old-growth forests without clearcutting for at least 80 years.
2.2 Data collection
We sampled in 40 permanent plots (20 × 20 m), comprising 37 areas under natural regeneration and 3 patches of old-growth forest. These plots represent a chronosequence, differing only in time since the start of fallow, allowing us to infer successional dynamics without long-term monitoring (Chazdon et al., 2023). For descriptive purposes, we categorized the plots into four stages: Stage I (1–10 years; 18 plots), Stage II (11–25 years; 8 plots), Stage III (26–40 years; 11 plots), and Stage IV (>80 years; 3 plots) (see Figure 2 for plot positions and stage classification). Plot ages were confirmed through interviews with local residents. While these categories help describe successional patterns, fallow age was used as a continuous predictor in statistical models to capture more nuanced trends.
Soil macrofauna were sampled bimonthly from July 2013 to May 2014 (six field campaigns) using the Tropical Soil Biology and Fertility Programme (TSBF) method (Anderson and Ingram, 1994), which involves manual extraction of litter and soil monoliths (25 × 25 cm) at 0–10 cm and 10–20 cm depths. Three replicates per layer were collected from each plot in each campaign. Collected fauna were preserved in 70% ethanol, sorted under a stereomicroscope, and identified into 29 taxonomic groups. Functional classification followed predominant feeding habits, as outlined in Brown et al. (2024) (Supplementary Table S1). Vegetation structure was assessed in July 2013 using nested subplots: 20 × 20 m for trees (DBH ≥ 10 cm) and 5 × 5 m for shrubs (DBH < 10 cm), recording species richness, density of individuals, mean DBH, height, and basal area. Canopy openness was quantified using hemispheric photos taken under diffuse light at five points per plot, aligned north, and processed in ImageJ software (Schroeder et al., 2021) to estimate percent openness from pixel contrast.
Soil chemistry followed standard protocols (Teixeira et al., 2017): total nitrogen (Kjeldahl); available phosphorus and potassium (Mehlich-1); exchangeable calcium and magnesium (1 mol L−1 KCl); soil organic matter (Walkley–Black); and pH in H₂O (1:2.5). Analyses were conducted at Embrapa Amazônia Oriental.
2.3 Statistical analysis
All statistical analyses were conducted using RStudio 2022.12.0. As an initial descriptive step, we compared macrofauna density and richness across successional stages (I–IV) using one-way ANOVA followed by Tukey’s post hoc test (vegan package), after confirming normality and homogeneity of variances. This categorical approach was used only to illustrate average differences among stages. In all subsequent analyses, fallow age was treated as a continuous predictor to more accurately capture successional trends.
To test whether macrofaunal diversity and density increase with fallow age, mediated by vegetation and soil properties, we performed structural equation modeling (SEM) via the piecewiseSEM package (Lefcheck et al., 2016). The variables were summarized through principal component analysis (PCA) (Hasan and Abdulazeez, 2021), allowing soil (PC1_soil) and vegetation (PC1_veg) to be represented by their first principal components, explaining 47.7 and 71% of the variance, respectively. Only PC1 was retained due to the sharp decline in explained variance after the first axis. Variables were centered and scaled; the main contributors to each axis are indicated by the biplot vectors in Supplementary Figures S1A,B. We tested for spatial autocorrelation in residuals using Moran’s I (spdep package; Bivand et al., 2017); finding no significant spatial structure, we proceeded without spatial terms. SEMs were fitted using linear mixed-effects models (LMEs) with plot identity as a random factor, based on ecological assumptions underlying the hypotheses (Fox and Weisberg, 2002). Model fit was evaluated using Fisher’s C and its associated p-value when independence claims were present, as well as the Akaike information criterion (AIC) and the marginal coefficients of determination (marginal R2) for each response variable, which were used as indicators of model adequacy (Lefcheck et al., 2016). For transparency, all hypothesized paths were retained in the SEM diagram, with non-significant paths shown grey arrows to distinguish the hypothesis space (Figure 1) from estimated effects (Figure 3).
Figure 3. Structural equation model (SEM) of the drivers of soil macrofauna attributes along forest regeneration. The diagram summarizes the relationships between fallow age, vegetation structure, soil properties, and soil macrofaunal community attributes. Arrows indicate the direction and magnitude of standardized path coefficients. Statistically significant relationships (p < 0.05) are shown in color, with darker colors indicating higher significance, whereas gray arrows indicate nonsignificant paths (p ≥ 0.05) that were retained to display the full hypothesized model. Numerical values on the arrows are standardized estimates. The final SEM was saturated (Fisher’s C not applicable), with AIC = −9,612.8.
To assess whether macrofaunal community structure changes along succession, we used NMDS ordination (Bray–Curtis distance, metaMDS, k = 2) and PERMANOVA (adonis2, 9,999 permutations) (Anderson, 2017) via the vegan package (Oksanen et al., 2007). Community composition (density by taxonomic group) was analyzed in relation to vegetation age, and PC1 scores of soil and vegetation. NMDS scores were used to visualize successional differences.
To identify taxonomic and functional groups that were disproportionately represented in the successional stages, we constructed contingency tables of group occurrence by stage and applied chi-squared tests of independence. Pearson residuals were extracted for each taxonomic group × stage combination, allowing us to visualize deviations from expected frequencies under the null hypothesis of independence.
Finally, to evaluate trophic interactions, we applied generalized additive models (GAMs) using the mgcv package (Wood and Wood, 2015). These models tested whether predator density exerts top-down control on herbivores, detritivores, geophages, and fungivores. Separate models were built for each group with vegetation age, soil quality, vegetation structure, and predator density as explanatory variables. Plot identity was included as a random effect to control for spatial pseudoreplication. Models were fitted using REML and evaluated via adjusted R2.
3 Results
A total of 46,070 individuals were recorded and classified into 29 taxonomic groups. Isoptera comprised 48% of the total, followed by Formicidae (25%) and Oligochaeta (9%). As succession advanced, macrofauna density converged toward values observed in mature forests. The density of geophages, phytophages and fungivores also increased across the successional stages (Supplementary Figure S2). At the taxonomic level, however, groups, responded idiosyncratically to succession (Supplementary Table S2). For example, Oligochaeta densities were significantly higher in old-growth forests (Stage IV) than in all fallow stages, whereas Isoptera and Araneae peaked at intermediate stages (II–III), with lower values in early fallows (Stage I) and intermediate densities in the oldest secondary forests (Stage IV). Several predator and mesopredator taxa also showed distinct successional peaks: Chilopoda were more abundant in Stages II–IV than in Stage I, Diplura and Scorpionida reached their highest densities in Stage III, and Mantodea occurred exclusively in Stage IV. In contrast, Blattodea exhibited the opposite pattern, with higher densities in early fallows and a marked decline toward old-growth forest, while Symphyla were more frequent in Stage IV than in the younger stages. Other groups showed weak or non-significant trends along the gradient. These heterogeneous responses highlight that soil macrofauna do not follow single, uniform successional trajectory.
Environmental gradients along the chronosequence were reflected in coordinated shifts in soil and vegetation. The first soil axis (PC1_soil, 47.7% of the variance) decreased with fallow age, with highest scores in Stage I, intermediate values in Stage II and the lowest scores in Stages III–IV. This axis represents a gradient from sites with higher pH and base cations (Ca: Mg) towards sites richer in organic matter and total N. In contrast, the first vegetation axis (PC1_veg, 71.4%) increased monotonically from Stage I to Stage IV, summarizing stand development, with greater tree density, species richness, DBH and height, together with lower canopy openness (Supplementary Figure S1). Consistently, Supplementary Figure S3 shows that individual soil and vegetation variables follow similar trends, with early fallows characterized by higher pH and base cations and lower biomass and canopy closure, whereas older fallows and old-growth forests exhibit higher organic matter, tree structural attributes and reduced canopy openness. Together, these ordinations depict a coordinated trajectory in which soil PC1 decreases while vegetation PC1 increases with fallow age.
The structural equation model was consistent with this interpretation of the successional gradients. Fallow age had a weak negative standardized effect on soil PC1 and a strong positive effect on vegetation PC1, but showed no direct path to macrofaunal attributes (AIC = −9,612.8). Soil PC1 had a marginal negative effect on macrofauna density, whereas vegetation PC1 had no significant direct effect on density. In contrast, richness increased significantly with vegetation PC1 and also significant positive relationship with density. Accordingly, Supplementary Table S3 indicates that richness responds positively to both vegetation PC1 and macrofauna density, while density is slightly constrained by soil PC1, suggesting that structurally more complex stands indirectly promote higher macrofauna richness even when overall density is limited by soil conditions. Overall, the model explained approximately 70% of the variation in vegetation structure and 13% of macrofauna richness (Figure 3).
Regarding community composition, the constrained ordination of macrofauna against fallow age, PC1_soil and PC1_veg (Figure 4A) showed extensive overlap among stages and relatively short environmental vectors, indicating that only a small fraction of the variation in community composition is aligned with these broad gradients (in agreement with the PERMANOVA, which detected a significant but low explanatory power of the full model; R2 = 0.0786, F = 3.31, p = 0.0001; Supplementary Table S4). The unconstrained NMDS (Figure 4B) revealed significant differences among stages, with the global PERMANOVA detecting a stage effect (p < 0.001; Supplementary Table S4) and pairwise tests indicating that Stage I differs significantly from Stages II and IV, whereas Stages II–IV show stronger overlap in ordination space. At the taxon level, association plots based on Pearson residuals (Figure 4C) showed that only a subset of groups contributes strongly to these differences, with predators, Isoptera and Oligochaeta displaying the most pronounced positive or negative residuals across stages. Frequency analyses indicated that predators were overrepresented in Stage I and underrepresented in Stage IV, whereas termites were underrepresented in Stage I and overrepresented in Stage III; earthworms were less frequent than expected in Stages II–III and more frequent in Stage IV (Figure 4C). At the level of functional groups, predators were again overrepresented in Stage I, while geophages and detritivores tended to be overrepresented in the older stages (Figure 4D). Across all consumer guilds, predator density emerged as the dominant predictor in the GAMs, exerting strong and highly significant positive effects on herbivores, detritivores, geophages and fungivores. Supplementary Table S5 shows that predator density is the only explanatory variable with significant smooth terms for all consumer guilds (edf ≈ 1, F = 51.8–53.9, p < 0.001), whereas smooths for fallow age, PC1_soil and PC1_veg have low F-values and non-significant p-values (p > 0.06). In the model for predator density, only PC1_soil showed a marginal effect (edf = 1.16, F = 3.02, p = 0.063), while age and PC1_veg were not significant (Supplementary Table S5).
Figure 4. Multivariate responses of soil macrofauna and environmental gradients across successional stages. (A) PCA biplot showing plots across successional stages (I–IV) based on soil properties (PC1_soil), vegetation (PC1_veg), and fallow age. (B) NMDS ordination of macrofaunal community composition based on Bray–Curtis dissimilarity; pairwise PERMANOVA (9,999 permutations) indicates significant differences (p < 0.05). (C,D) Over- and underrepresented taxonomic (C) and functional (D) groups across stages. Blue = overrepresented; red = underrepresented.
4 Discussion
Our findings indicate that macrofaunal recovery is driven chiefly by indirect pathways, through vegetation rebuilding and soil reorganization, rather than by a simple, monotonic effect of fallow age. Gains in richness track increases in vegetation structure and habitat heterogeneity, whereas density remains constrained by soil conditions, revealing a partial decoupling between community size and diversity. At the compositional level, most turnover was not captured by the measured gradients, indicating contributions beyond environmental filtering. Taken together, these patterns show that time since fallow alone is an incomplete indicator of soil macrofauna recovery and should be interpreted jointly with vegetation attributes and edaphic conditions.
4.1 Successional declines in soil nutrients do not constrain macrofauna
The observed non-linear changes in soil properties along the successional gradient, challenge the assumption that nutrient-poor soils necessarily hinder macrofaunal recovery (Lavelle et al., 2006; Chamorro-Martínez et al., 2022). According to Supplementary Figure S3, early fallows combine higher pH with lower soil organic matter and N and P contents, intermediate stages exhibit increased organic matter and nutrient availability, and the oldest forests (Stage IV) tend to stabilize or show a slight decrease in these variables. Thus, in our PCA, higher PC1_soil scores in early stages represent soils with higher pH and base cations but lower organic matter and N, whereas lower PC1_soil scores in older stages reflect more acidic, organic-rich soils with greater N availability.
Consistent with this interpretation, higher PC1_soil values in early stages coincide with low vegetation biomass and limited nutrient uptake, whereas lower PC1_soil values in older stages are associated with denser, structurally more complex stands with higher PC1_veg. As vegetation structure develops and PC1_veg increases, nutrients become progressively immobilized in plant tissues and litter, leading to lower PC1_soil scores in older stages. This pattern directly links the soil ordination axis to vegetation nutrient uptake, indicating that the decline in PC1_soil along the chronosequence reflects nutrient transfer into biomass rather than an impediment to macrofaunal recovery. Such conditions are typical of Amazonian forests, where nutrients are concentrated in living and detrital biomass and soils remain acidic and comparatively nutrient-poor (Sayer et al., 2024). Consequently, the negative correlation between PC1_soil and macrofauna density is better interpreted as emerging from vegetation-driven redistribution of nutrients, while vegetation continues to enhance soil macrofauna richness through greater structural and resource heterogeneity. In these secondary forests, lower PC1_soil scores during succession can therefore be read as a signal of vegetation build-up rather than simply as soil degradation; thus, vegetation structure, litter inputs and biomass may serve as more informative early indicators of soil macrofauna recovery than soil chemistry alone.
4.2 Landscape connectivity and dispersal underpin richness gains without density increases
SEM results indicate that fallow age affects macrofauna indirectly via vegetation, offering partial support for an increase in diversity through succession. Although richness and density show similar successional patterns, with lower values in early fallows and higher values in older stages (Supplementary Figure S2), their responses are not proportional: richness increases more consistently along the chronosequence, whereas density remains highly variable and shows no clear age-related. In line with this interpretation, our SEM indicates that PC1_soil exerts a negative effect on macrofauna density, whereas vegetation structure promotes richness, reinforcing the idea that density is constrained by local edaphic conditions while richness continues to accumulate through time.
Such a pattern is compatible with species accumulation via immigration and establishment, whereby longer successional periods increase the likelihood of colonization and persistence even when local abundance approaches a carrying capacity imposed by resource availability and soil conditions. Under such constraints, additional species can be added through turnover and habitat differentiation without major changes in total community size, as predicted by island biogeography theory (MacArthur and Wilson, 2001). The landscape context of our study, secondary patches embedded within an extensive, low-disturbance matrix of old-growth forest, likely facilitates continual immigration from source populations, thereby increasing richness without parallel gains in local density.
The clear separation in macrofaunal community composition between the early and late stages indicates a gradual process of species replacement over time, possibly reflecting both changes in environmental conditions and stochastic processes of colonization and local extinction. However, given that environmental predictors explained only a small fraction of the compositional variation (PERMANOVA; F = 3.313, R2 = 0.08, p = 0.0001), the unexplained portion is most plausibly attributed to dispersal-related processes (immigration, drift, local extinction) acting together with environmental filtering, rather than to a single dominant mechanism. Because the study landscape is highly connected to old-growth forest, this joint signal of environment and dispersal is expected; in more isolated or repeatedly disturbed landscapes, weaker compositional shifts and slower richness gains would be likely.
4.3 Predator-driven trophic effects shape belowground succession
An interesting pattern that deserves attention is the higher-than-expected frequency of predators at the early stages of succession. Although this may seem counterintuitive, it likely reflects the opportunistic behavior of mobile predators that exploit more open habitats with greater exposure of potential prey (Brahma et al., 2022). Consistent with this, our GAMs indicate top-down regulation (Schuldt et al., 2017): predator density significantly influenced herbivores, detritivores, geophages, and fungivores. This aligns with trophic-cascade theory (Wardle and Yeates, 1993), whereby top-down forces constrain lower trophic levels, and is compatible with the foraging plasticity of soil predators, many of them generalists that shift diets with prey availability across microhabitats (Briones, 2018). These interactions highlight the role of trophic complexity in shaping soil community assembly and suggest that predation, in addition to environmental filtering, fulfills a central role in structuring belowground food webs throughout succession.
Within this context, our analytical framework distinguishes between abiotic and biotic components: the structural equation model summarizes how fallow age, soil properties and vegetation structure jointly affect overall macrofauna density and richness, whereas the trophic GAMs explicitly resolve how predators influence the different functional groups along these same environmental gradients. Together, these two components provide a coherent picture of both abiotic control and biotic interactions in belowground succession. Although our environmental predictors did not detect strong direct effects of vegetation or soil on predator density, empirical and theoretical work indicate that structurally complex forests tend to support more diverse and functionally effective predator assemblages by providing a greater variety of hunting substrates and refuges (Schuldt and Staab, 2015; Stemmelen et al., 2022). Thus, vegetation complexity most likely influences soil macrofauna both directly, via habitat and resource heterogeneity, and indirectly, by modulating predator–prey interactions and trophic cascades. Taken together, our results suggest that vegetation structure represents a major abiotic context within which predator-driven top-down effects operate to shape belowground community organization through succession.
4.4 Shifting cultivation in low-degradation, well-connected settings is not inherently degrading
Although soil macrofauna at the order level are widely applied as bioindicators in tropical ecosystems, including secondary forests, pastures, and agroforestry systems (Velasquez and Lavelle, 2019; Lavelle et al., 2022; Lugo et al., 2024), our findings highlight that their responses to regeneration are not uniform. This heterogeneity challenges the establishment of a single successional model for soil fauna recovery in shifting cultivation systems and suggests that indicator value must be assessed on a case-by-case basis. Some groups may serve as reliable proxies of successional progress, where others show inconsistent trajectories that limit their predictive power, underscoring the need for caution when adopting soil macrofauna as recovery indicators, given that their performance is strongly mediated by ecological traits and local environmental conditions. Nonetheless, order-level identification remains a pragmatic compromise, balancing feasibility and ecological informativeness: it reduces the demand for taxonomic expertise and processing time, while still capturing meaningful successional patterns when combined with vegetation and soil metrics. This approach aligns with the criteria for effective indicators (Serra et al., 2021) and provides valuable insights when integrated into broader monitoring framework.
The rapid recovery of soil macrofauna observed in this study aligns with broader evidence of swift regeneration of key ecosystem properties in tropical and neotropical land-use systems (de Medeiros-Sarmento et al., 2021; Rozendaal et al., 2019), thereby supporting the view that, shifting cultivation is not intrinsically degrading to soil macrofauna when practiced under favorable conditions. Several authors have advocated extended fallow periods, spanning decades, to ensure the full recovery of soil macrofauna, tree diversity, and carbon stocks (Cole et al., 2020; Villa et al., 2020), yet our results indicate that, under conditions of limited prior degradation and strong landscape connectivity to mature forests, shorter fallow periods may suffice for substantial macrofaunal recovery. We also observed that soil organic matter and N peak at intermediate stages and decline slightly in the oldest forests (Supplementary Figure S3), indicating that extending fallow duration does not necessarily lead to continuous gains in these soil properties. In our system, such recovery is contingent upon the reestablishment of vegetation, which plays a central role in mediating soil community responses. Therefore, fallow duration should be guided not by vegetation age or appearance alone, but by its effective structural, floristic, and biomass restoration. These attributes are critical for reestablishing the ecological conditions needed for macrofaunal functioning and trophic interactions. Taken together, these findings suggest that shifting cultivation cannot be considered inherently degrading; rather, its effects depend on management intensity, cycle frequency and landscape configuration. When practiced with low intensity and appropriate fallow intervals, it can support biodiversity conservation at the landscape scale by fostering secondary forest regeneration and sustaining soil faunal communities.
5 Conclusion
Our findings demonstrate that vegetation complexity, rather than fallow age alone, is the primary driver of soil macrofauna recovery in Amazonian secondary forests Vegetation regeneration strongly enhances macrofauna richness, while fallow age exerted only indirect effects through its influence on vegetation and soils, underscoring that forest recovery depends not merely on time but on ecological conditions that promote succession. In contrast, total macrofauna density appears to reflect local soil constraints and thus functions more as a site-level attribute than as a robust indicator of recovery. Predator-driven top-down control further shapes community structure, regulating herbivores, detritivores, and geophages across stages. Importantly, soil fauna did not respond uniformly: taxonomic and functional groups followed distinct trajectories, reinforcing the need for caution when generalizing their role as recovery indicators. From an applied perspective, simple metrics centred on richness, complemented by density where appropriate, and interpreted together with vegetation attributes, can serve as cost-effective tools for restoration monitoring, even at the order level of identification. Overall, our results highlight that effective vegetation regeneration is essential to reestablish soil biodiversity and ecosystem functions after shifting cultivation and they challenge the generalized view that this practice is inherently degrading. When practiced with low management intensity, adequate fallow regimes and within well-connected, low-degradation landscapes, shifting cultivation can instead contribute to the recovery and maintenance of essential ecosystem functions.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethics statement
The manuscript presents research on animals that do not require ethical approval for their study.
Author contributions
PM-S: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing. LF: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing – original draft. MG: Conceptualization, Formal analysis, Methodology, Resources, Supervision, Validation, Visualization, Writing – review & editing.
Funding
The author(s) declare that financial support was received for the research and/or publication of this article. PM-S is grateful for the doctoral scholarship provided by the Coordination for the Improvement of Higher Education Personnel (CAPES). MG acknowledges support from the Brazilian National Council for Scientific and Technological Development (CNPq) through a productivity scholarship (grant no. 310865/2022-0). The authors thank the Program of Long-Term Ecological Research (PELD) in Estação Científica Ferreira Penna, Caxiuanã, Amazônia Oriental (process no. 441224/2016-4) for the resources provided during field work, the Museu Paraense Emílio Goeldi for logistic support, and Instituto Tecnológico Vale for additional financial support.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.
Generative AI statement
The authors declare that no Gen AI was used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/ffgc.2025.1713966/full#supplementary-material
References
Anderson, J.M., and Ingram, J., (1994). Tropical soil biology and fertility: A hand book of methods. Soil Science, 157:265. Wallingford UK: C-A-B International.
Bivand, R., Altman, M., Anselin, L., Assunção, R., Berke, O., Bernat, A., et al. (2017). Package ‘spdep’. Spatial Depend. 1:192. doi: 10.32614/CRAN.package.spdep
Brahma, S., Sharma, D., Banerjee, S., Saha, G. K., and Aditya, G. (2022). Influence of habitat complexity on the prey mortality in IGP system involving insect predators (Heteroptera) and prey (Diptera): implications in biological control. PLoS One 17:e0264840. doi: 10.1371/journal.pone.0264840,
Briones, M. J. (2018). The serendipitous value of soil fauna in ecosystem functioning: the unexplained explained. Front. Environ. Sci. 6:149. doi: 10.3389/fenvs.2018.00149
Brown, G. G., Demetrio, W. C., Gabriac, Q., Pasini, A., Korasaki, V., Oliveira, L. J., et al. (2024). Soil macrofauna communities in Brazilian land-use systems. Biodivers. Data J. 12:e115000. doi: 10.3897/BDJ.12.e115000
Chamorro-Martínez, Y., Torregroza-Espinosa, A. C., Pallares, M. I. M., Osorio, D. P., Paternina, A. C., and Echeverría-González, A. (2022). Soil macrofauna, mesofauna and microfauna and their relationship with soil quality in agricultural areas in northern Colombia: ecological implications. Rev. Bras. Ciênc. Solo 46:e0210132. doi: 10.36783/18069657rbcs20210132
Chazdon, R. L., Norden, N., Colwell, R. K., and Chao, A. (2023). Monitoring recovery of tree diversity during tropical forest restoration: lessons from long-term trajectories of natural regeneration. Philos. Trans. R. Soc. B 378:20210069. doi: 10.1098/rstb.2021.0069,
Cole, R. J., Selmants, P., Khan, S., and Chazdon, R. (2020). Litter dynamics recover faster than arthropod biodiversity during tropical forest succession. Biotropica 52, 22–33. doi: 10.1111/btp.12740
Coomes, O. T., Cheng, Y., Takasaki, Y., and Abizaid, C. (2021). What drives clearing of old-growth forest over secondary forests in tropical shifting cultivation systems? Evidence from the Peruvian Amazon. Ecol. Econ. 189:107170. doi: 10.1016/j.ecolecon.2021.107170
Costa, A., Rodrigues, H., Júnior, J. F. S., Nunes, L. R. C., de Moraes, B. C., Costa, A. E., et al. (2018). Variabilidade Horária, Diária e Sazonal da Frequência e Intensidade de Precipitação em uma Floresta Tropical Chuvosa na Amazônia Brasileira. Rev. Bras. Geogr. 11, 1290–1302.
de Medeiros-Sarmento, P. S., Ferreira, L. V., and Gastauer, M. (2021). Natural regeneration triggers compositional and functional shifts in soil seed banks. Sci. Total Environ. 753:141934. doi: 10.1016/j.scitotenv.2020.141934,
Demetrio, W., Cavalieri-Polizeli, K. M. V., Guimarães, R. M. L., de Almeida Ferreira, S., Parron, L. M., and Brown, G. G. (2022). Macrofauna communities and their relationship with soil structural quality in different land use systems. Soil Res. 60, 648–660. doi: 10.1071/SR21157
Ebenman, B. O., Law, R., and Borrvall, C. (2004). Community viability analysis: the response of ecological communities to species loss. Ecology 85, 2591–2600. doi: 10.1890/03-8018
Fox, J., and Weisberg, S. (2002). “Mixed-effects models in R” in An R companion to applied regression (Thousand Oaks, CA, USA: SAGE).
Franco, A. L. C., Sobral, B. W., Silva, A. L. C., and Wall, D. H. (2019). Amazonian deforestation and soil biodiversity. Conserv. Biol. 33, 590–600. doi: 10.1111/cobi.13234,
Hasan, B. M. S., and Abdulazeez, A. M. (2021). A review of principal component analysis algorithm for dimensionality reduction. J. Soft Comput. Data Mining 2, 20–30. doi: 10.30880/jscdm.2021.02.01.003
Jakovac, C. C., Pena-Claros, M., Kuyper, T. W., and Bongers, F. (2015). Loss of secondary-forest resilience by land-use intensification in the Amazon. J. Ecol. 103, 67–77. doi: 10.1111/1365-2745.12298
Lavelle, P., Decaëns, T., Aubert, M., Barot, S., Blouin, M., Bureau, F., et al. (2006). Soil invertebrates and ecosystem services. Eur. J. Soil Biol. 42, S3–S15. doi: 10.1016/j.ejsobi.2006.10.002
Lavelle, P., Mathieu, J., Spain, A., Brown, G., Fragoso, C., Lapied, E., et al. (2022). Soil macroinvertebrate communities: a world-wide assessment. Glob. Ecol. Biogeogr. 31, 1261–1276. doi: 10.1111/geb.13492
Lavelle, P., Spain, A., Blouin, M., Brown, G., Decaëns, T., Grimaldi, M., et al. (2016). Ecosystem engineers in a self-organized soil: a review of concepts and future research questions. Soil Sci. 181, 91–109. doi: 10.1097/SS.0000000000000155
Lintemani, M. G., Loss, A., Mendes, C. S., and Fantini, AC. (2020). Long fallows allow soil regeneration in slash-and-burn agriculture. J. Sci. Food Agric. 100, 1142–1154. doi: 10.1002/jsfa.10123
Lisboa, P. L. B., Bezerra, M., and de Rezende Cardoso, A. L. (2013). Caxiuana: História natural e ecologia de uma Floresta Nacional da Amazônia. Belém, Brazil: Museu Paraense Emílio Goeldi.
Lugo, E. H., Ibañez, E. V., and Lavelle, P. (2024). A global indicator of soil macroinvertebrate community composition, abundance and diversity. Appl. Soil Ecol. 193:105138. doi: 10.1016/j.apsoil.2023.105138
MacArthur, R. H., and Wilson, E. O. (2001). The theory of island biogeography, vol. 1: Princeton University Press. doi: 10.1515/9781400881376
Mertz, O., Bruun, T. B., Jepsen, M. R., Ryan, C. M., Zaehringer, J. G., Hinrup, J. S., et al. (2021). Ecosystem service provision by secondary forests in shifting cultivation areas remains poorly understood. Hum. Ecol. 49, 271–283. doi: 10.1007/s10745-021-00236-x
Mesquita, R. D. G., Massoca, P. E. D., Jakovac, C. C., Bentos, T. V., and Williamson, G. B. (2015). Amazon rain forest succession: stochasticity or land-use legacy? Bioscience 65, 849–861. doi: 10.1093/biosci/biv108
Mukul, S. A., Herbohn, J., and Firn, J. (2020). Rapid recovery of tropical forest diversity and structure after shifting cultivation in the Philippines uplands. Ecol. Evol. 10, 7189–7211. doi: 10.1002/ece3.6419,
Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M. H. H., Oksanen, M. J., et al. (2007). The vegan package. Commun. Ecol. 10:719. doi: 10.32614/CRAN.package.vegan
Peniston, J. H., Backus, G. A., Baskett, M. L., Fletcher, R. J., and Holt, R. D. (2024). Ecological and evolutionary consequences of temporal variation in dispersal. Ecography 2024:e06699. doi: 10.1111/ecog.06699
Ponsard, S., Arditi, R., and Jost, C. (2000). Assessing top-down and bottom-up control in a litter-based soil macroinvertebrate food chain. Oikos 89, 524–540. doi: 10.1034/j.1600-0706.2000.890312.x
Ramos, A. P. D., Lima, S. S. D., Ferreira, C. D. S., Pinto, L. A. R. D. S., Ferreira, R., Dias, A., et al. (2025). Macrofauna and soil properties in agroforestry system and secondary forest. Rev. Bras. Ciênc. Solo 49:e0240091. doi: 10.36783/18069657rbcs20240091
Roosevelt, A. C. (2013). The Amazon and the Anthropocene: 13,000 years of human influence in a tropical rainforest. Anthropocene 4, 69–87. doi: 10.1016/j.ancene.2014.05.001
Rousseau, G. X., Deheuvels, O., Celentano, D., Arias, I. R., Hernández-García, L. M., and Somarriba, E. (2021). Shade tree identity rather than diversity influences soil macrofauna in cacao-based agroforestry systems. Pedobiologia 89:150770. doi: 10.1016/j.pedobi.2021.150770
Rozendaal, D. M. A., Bongers, F., Aide, T. M., Alvarez-Dávila, E., Ascarrunz, N., Balvanera, P., et al. (2019). Biodiversity recovery of Neotropical secondary forests. Sci. Adv. 5:eaau3114. doi: 10.1126/sciadv.aau3114,
Sayer, E. J., Leitman, S. F., Wright, S. J., Rodtassana, C., Vincent, A. G., Brechet, L. M., et al. (2024). Tropical forest above-ground productivity is maintained by nutrients cycled in litter. J. Ecol. 112, 690–700. doi: 10.1111/1365-2745.14251
Scheu, S., and Schaefer, M. (1998). Bottom-up control of the soil macrofauna community in a beechwood on limestone: manipulation of food resources. Ecology 79, 1573–1585. doi: 10.1890/0012-9658(1998)079[1573:BUCOTS]2.0.CO;2
Schoereder, J. H., Sobrinho, T. G., Ribas, C. R., and Campos, R. B. (2004). Colonization and extinction of ant communities in a fragmented landscape. Austral Ecol. 29, 391–398. doi: 10.1111/j.1442-9993.2004.01378.x
Schroeder, A. B., Dobson, E. T., Rueden, C. T., Tomancak, P., Jug, F., and Eliceiri, K. W. (2021). The ImageJ ecosystem: open-source software for image visualization, processing, and analysis. Protein Sci. 30, 234–249. doi: 10.1002/pro.3993,
Schuldt, A., Bruelheide, H., Buscot, F., Assmann, T., Erfmeier, A., Klein, A. M., et al. (2017). Belowground top-down and aboveground bottom-up effects structure multitrophic community relationships in a biodiverse forest. Sci. Rep. 7:4222. doi: 10.1038/s41598-017-04619-3,
Schuldt, A., and Staab, M. (2015). Tree species richness strengthens relationships between ants and the functional composition of spider assemblages in a highly diverse forest. Biotropica 47, 339–346. doi: 10.1111/btp.12209
Serra, R. T., Santos, C. D., Rousseau, G. X., Triana, S. P., Muñoz Gutiérrez, J. A., and Burgos Guerrero, J. E. (2021). Fast recovery of soil macrofauna in regenerating forests of the Amazon. J. Anim. Ecol. 90, 2094–2108. doi: 10.1111/1365-2656.13506,
Stemmelen, A., Jactel, H., Brockerhoff, E., and Castagneyrol, B. (2022). Meta-analysis of tree diversity effects on the abundance, diversity and activity of herbivores' enemies. Basic Appl. Ecol. 58, 130–138. doi: 10.1016/j.baae.2021.12.003
Teixeira, PC, Donagemma, GK, Fontana, A., and Teixeira, WG, 2017. Manual de métodos de análise de solo, p. 15. 3. ed. rev. e ampl. – Brasília, DF: Embrapa.
Vasconcelos, W. L. F. D., Rodrigues, D. D. M., Silva, R. O. C., and Alfaia, S. S. (2020). Diversity and abundance of soil macrofauna in three land use systems in eastern Amazonia. Rev. Bras. Ciênc. Solo 44:e0190136. doi: 10.36783/18069657rbcs20190136
Velasquez, E., and Lavelle, P. (2019). Soil macrofauna as an indicator for evaluating soil based ecosystem services in agricultural landscapes. Acta Oecol. 100:103446. doi: 10.1016/j.actao.2019.103446
Villa, P. M., Martins, S. V., de Oliveira Neto, S. N., Rodrigues, A. C., Hernández, E. P., and Kim, D.-G. (2020). Policy forum: shifting cultivation and agroforestry in the Amazon: premises for REDD+. Forest Policy Econ. 118:102217. doi: 10.1016/j.forpol.2020.102217
Villa, P. M., Martins, S. V., Diniz, É. S., de Oliveira Neto, S. N., Neri, A. V., Pinto-Junior, H., et al. (2021). Taxonomic and functional beta diversity of woody communities along Amazon forest succession: the relative importance of stand age, soil properties and spatial factor. For. Ecol. Manag. 482:118885. doi: 10.1016/j.foreco.2020.118885
Wardle, D. A., and Yeates, G. W. (1993). The dual importance of competition and predation as regulatory forces in terrestrial ecosystems: evidence from decomposer food-webs. Oecologia 93, 303–306. doi: 10.1007/BF00317685,
Keywords: shifting cultivation, natural regeneration, succession, secondary forest, sustainability agriculture
Citation: de Medeiros-Sarmento PS, Ferreira LV and Gastauer M (2025) Recovery of soil macrofauna in Amazonian secondary forests is driven by vegetation complexity rather than fallow age alone. Front. For. Glob. Change. 8:1713966. doi: 10.3389/ffgc.2025.1713966
Edited by:
Qi Li, Chinese Academy of Sciences (CAS), ChinaReviewed by:
Pingting Guan, Northeast Normal University, ChinaBing Li, Northeast Forestry University, China
Copyright © 2025 de Medeiros-Sarmento, Ferreira and Gastauer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Priscila Sanjuan de Medeiros-Sarmento, cHJpc2NpbGEuc2FuanVhbi5tZWRlaXJvc0BnbWFpbC5jb20=