2.1 Long-Term Field Experiment

2.1.1 Study Site, Experimental Design and Soil Sampling

The experimental site is located at the Changchun Agroecology Observation Station (44°00′ N, 125°24′ E) of the Northeast Institute of Geography and Agroecology. This region features a continental monsoon climate, with a mean annual temperature of 4°C and a mean annual precipitation of 550 mm. The soil is derived from loamy loess and is classified as a black soil (Hapudoll according to USDA Soil Taxonomy; Soil survey staff 1999). The long-term experiment began in autumn 2011, with initial soil pH at 7.3 and soil C and nitrogen (N) contents at 17.1 g C kg⁻¹ and 1.5 g N kg⁻¹, respectively. Prior to the trial, the field had been continuously cropped with maize (Zea mays L.) for 30 years.

The long-term parent field experiment followed a randomized complete block design, with three tillage treatments, including no-till, reduced till, and conventional till, and two cropping systems such as maize-soybean rotation and continuous maize cropping. Each treatment had four replicates with each plot area of 7.8 × 25 m. In this study, only the plots with maize monoculture under conventional till and no-till, combined with straw return to the field, were considered (Figure 1A).

For the no-till treatment, there was no soil disturbance except for planting, which was performed using a no-till planter (KINZE-3000NT, Williamsburg, IA, USA). For the till treatment, the soil was annually fall moldboard ploughed to a depth of 20 cm after maize harvest, followed by discing (7.5–10 cm in depth), harrowing, and ridge-building during the seedbed preparation process in the following spring. After harvest, maize straw in all treatments was returned to the soil to prevent water and wind erosion during the winter and early spring. In the no-till plots, maize straw was cut into about 30 cm pieces and manually placed among 30–35 cm standing stubble. The straw in the till plots was removed before and manually replaced on the soil surface after fall moldboard ploughing, and then mixed with the plow layer by discing and cultivation in the following spring. The treatments were applied each year since the initiation of the parent experiment in 2011. Based on the maize yield and harvest index, the annual straw return amount to the field was approximately 11 Mg ha⁻¹ in both no-till and till treatments. Consequently, this straw amount was used for the follow-up in situ ¹³C straw tracing experiment (Section 2.2).

Soil samples were collected in late April of 2017 and 2018, as this timing best reflects the impact of tillage practices without being influenced by the growing season of maize. Five soil cores were randomly taken from the center of each plot using a hand auger (4.8 cm inner diameter) to a depth of 20 cm. These cores were then combined to form one representative soil sample for each plot.

2.2 ¹³C In Situ Tracing Experiment

2.3 Data Analysis

All statistical analyses were performed in R 3.6.2, except for the structural equation model (SEM) analysis, which was performed using IBM SPSS Amos 22.0.

To examine the relationships between omnivores-predators and other nematode and microbial groups from a bottom-up resource-driven perspective, a functional group-based interaction network was built using data from the field experiment (Potapov et al. 2023). Seven functional groups were classified based on traits, resource use, and predation: primary consumers (fungi, actinomycetes, and general bacteria), secondary consumers (fungivores, bacterivores, and plant-parasites), and tertiary consumers (omnivores-predators). Plant-parasites were classified as secondary consumers because some genera (e.g., Boleodorus, Neothada, and Tylenchus; refer to Table S1) feed on both plant roots and fungal hyphae (Yeates et al. 1993). For consistency, all taxa were converted to their respective C content (ng C g⁻¹ dry soil) prior to network construction (conversion details are provided in the Data S1).

Each functional group was represented as a matrix, and Bray–Curtis dissimilarity was calculated across all plots within the same treatment (Rzanny and Voigt 2012). Mantel tests were performed for each pair of functional groups across trophic levels, and significant correlations (p < 0.1) were used to create directional links from lower to higher trophic groups. Interaction complexity and strength were quantified using three metrics: connectivity, interaction diversity, and interaction strength, as described by Rzanny and Voigt (2012). Detailed formulae are provided in the Data S1.

To evaluate the feeding strategy of omnivores-predators under different tillage systems, two approaches were employed. First, δ¹³C signatures in organisms (refer to Table S3) were used to assess predator–prey relationships. A significant linear correlation between the δ¹³C signature of omnivores-predators and a particular group (e.g., microbivores, plant-parasites and microbial groups) indicates a strong predator–prey relationship, as the δ¹³C signature in an organism's body is responsive to the δ¹³C signature of its dietary resources (Potapov et al. 2018; Ruess and Chamberlain 2010).

These feeding proportions were subsequently subjected to a repeated measures analysis of variance (RMANOVA) to test for differences in feeding behavior of omnivores-predators between tillage practices (no-till vs. till).

To explore the differences in microbial necromass under contrasting tillage practices, Student's t-test and RMANOVA were used to analyze data from the 2017 to 2018 integrated dataset and the 1-year ¹³C in situ tracing experiment. To explore the role of omnivores-predators in explaining the variation in microbial necromass, Pearson correlation and random forest model analyses were conducted. Pearson correlation was first performed to assess the correlation between the fungal-to-bacterial necromass ratio and different soil nematode and microbial groups in both the field and ¹³C in situ tracing experiments. To further identify the relative importance of all soil nematode and microbial groups to microbial necromass within different tillage systems, a random forest analysis was performed based on data from the ¹³C in situ tracing experiment. Group importance was determined based on the percentage increase in mean square error in the random forest model.

To understand how omnivores-predators affect microbial necromass via top-down control within the food chain, general linear models (GLMs) and structural equation modeling (SEM) were applied based on the ¹³C enrichment rates of organisms. GLMs were used to examine the relationships between omnivores-predators, microbivores, and microbes. The analysis was initially performed on the full dataset and then on subsets grouped according to different tillage practices, with only p-values < 0.05 reported. Positive linear regression suggests that the rate at which higher trophic groups acquire ¹³C by consuming lower trophic groups is closely related to the ¹³C enrichment rate of the lower trophic groups. Negative linear regression indicates that the increased predation pressure from higher trophic groups decreased the ¹³C enrichment rate in the lower trophic groups. To verify the extent of top-down control exerted by omnivores-predators on adjacent trophic groups (microbivores), best-fit models were used to assess how the predation influences the temporal trajectories of ¹³C content in microbivores during the tracing period. Additionally, the relationships between various microbial groups (i.e., fungi, bacteria, and actinomycetes) and the fungal-to-bacterial necromass ratio were examined using the same GLMs approach. To improve model linearity, the ratio was ln-transformed. The negative linear correlation between the ¹³C enrichment rates of microbial groups and the ¹³C-fungal to ¹³C-bacterial necromass ratio was expected, as active microbes are the precursors of necromass formation (Liang et al. 2017).

SEM was conducted to partition the top-down control mechanisms of omnivores-predators on microbial necromass across distinct tillage systems. To increase the model fit, paths from plant-parasites to general bacteria or actinomycetes were included, as suggested by modification indices. These paths are justified because plant-parasitic nematodes can subsidize bacteria by releasing C and other nutrients through penetrating root cells (Yeates et al. 1993). Such modifications do not detract from the overarching regulatory role of omnivores-predators in exerting top-down control on other nematode and microbial groups. The final model was selected based on the p-value, comparative fit index (CFI), goodness-of-fit (GFI) and chi-square fit statistics/° of freedom (CMIN/DF).

3.1 Role of Omnivores-Predators in Different Tillage Practices

The nematode community had clear changes at the community level in the no-till soil compared to the till soil, characterized by a 46% increase in abundance and a 67% increase in richness of omnivores-predators (Figure S1 and Table S4). Functional group networks displayed a distinct role of omnivores-predators in maintaining the structure of the soil food web between till and no-till treatments (Figure 2A). In the no-till soil, omnivores-predators, acting as tertiary consumers, interacted more diversely and strongly with secondary consumers of fungivores, bacterivores, and plant-parasites, as well as primary consumers of fungi, general bacteria, and actinomycetes (as indicated by the interaction diversity, strength, and connectivity; Figure S2). These changes led to a notable increase in the structural complexity of the no-till soil food web compared to the till soil food web. Specifically, interaction diversity, interaction strength, and connectivity in the no-till soil food web increased by 105%, 28%, and 200%, respectively (Figure 2B).

The structural differences in the soil food web between till and no-till treatments were reflected in corresponding changes in the ¹³C content and enrichment rate across food web components (Figure S3). Specifically, during the tracing period, the ¹³C content and enrichment rate increased by 310% and 402% for omnivores-predators and by 64% and 50% for fungi under the no-till treatment compared to the till treatment. However, the ¹³C content and enrichment rate of fungivores decreased by 73% and 90%, respectively, under the no-till soil (Figure S3).

3.2 Importance of Omnivores-Predators in Explaining Microbial Necromass

Tillage practices did not affect fungal and bacterial necromass as indicated by statistical analyses in both the long-term field and ¹³C in situ tracing experiments (Figure 3A,B). However, the fungal-to-bacterial necromass ratio was notably affected by tillage, with no-till treatment increasing this ratio by 20% in the long-term field experiment and by an average of 29% in the ¹³C in situ tracing experiment (Figure 3A,B).

Correlation analysis demonstrated a consistent positive correlation between omnivores-predators and the fungal-to-bacterial necromass ratio in both experimental settings (Figure 3C). Further random forest analysis identified the key components of the soil food web (Figure 3D) that influenced microbial necromass in the ¹³C in situ tracing experiment. Fungi and general bacteria were the two most important predictors of variation in fungal and bacterial necromass under till treatment (Figure 3D). In contrast, omnivores-predators emerged as the principal predictor of variation in both fungal and bacterial necromass under no-till treatment.

3.3 Feeding Strategy of Omnivores-Predators in Different Tillage Practices

In the till soil, the δ¹³C of omnivores-predators was closely associated with the δ¹³C of fungi (Figure 4A and Figure S4). Conversely, the δ¹³C of omnivores-predators in the no-till soil was primarily linked to the δ¹³C of fungivores (Figure 4B and Figure S4). The feeding proportions of omnivores-predators displayed distinct differences between till and no-till treatments (Figure 4C). In the till soil, omnivores-predators fed mainly on plant-parasites (46%) and fungi (14%), whereas in the no-till soil, fungivores and plant-parasites accounted for 51% and 29% of their diet, respectively. Among various food sources in the omnivores-predators' diet, fungivores and fungi were strongly affected by tillage practice (Figure c and Table S5). Compared to the till treatment, the proportion of fungivores in the diet increased by 292% and the proportion of fungi decreased by 64% under the no-till treatment (Figure 4C). This intense feeding by omnivores-predators on fungivores in the no-till soil reduced the feeding intensity of fungivores on fungi, with a 99% reduction compared to their feeding intensity in the till soil (Figure S5).

3.4 Top-Down Control of Omnivores-Predators on Microbial Necromass

In a top-down direction in the food chain, a relationship between omnivores-predators and fungivores was only found in the no-till soil, where the ¹³C enrichment rate of fungivores decreased as the ¹³C enrichment rate of omnivores-predators increased (Figure 5A). This inverse relationship was further highlighted by the opposite temporal changes in the ¹³C content of omnivores-predators and fungivores under the no-till soil, a pattern not present under the till soil (Figure 6A). For the relationship between fungivores and fungi, a positive correlation was found only in the till soil. For the relationship between omnivores-predators and fungi, a positive correlation was revealed in both the till and no-till soils (Figure 5A). For the relationships between microbial groups and fungal-to-bacterial necromass ratio, the strongest correlation was found between fungi and the fungal-to-bacterial necromass ratio under the no-till treatment (R² = 0.44; Figure 5B).

SEMs revealed that the regulatory pathways underlying the top-down effects of omnivores-predators on the fungal-to-bacterial necromass ratio differed between the two tillage systems (Figure 6B). In the no-till soil, an inverse relationship (path coefficient = −0.55) was revealed between omnivores-predators and fungivores, disrupting the direct pathway from fungivores to fungi. Instead, a strong pathway from omnivores-predators to fungi was established (path coefficient = 0.63). In the till soil, no such inverse relationship was found and the pathway from fungivores to fungi remained significant (path coefficient = 0.38). Moreover, a pathway from omnivores-predators to fungi was also present in the till soil (path coefficient = 0.66). As a result, the path coefficient from fungi to the fungal-to-bacterial necromass ratio was 14% higher under the no-till treatment compared to the till treatment.

4.1 Omnivores-Predators Serve as Key Regulators of Microbial Necromass in No-Till System

Microbial necromass formation is typically influenced by microbial activity (Buckeridge et al. 2020; Wang et al. 2023; Yang et al. 2022), as shown in the till soil, where general bacteria and fungi emerged as the primary predictors of necromass (Figure 3D). In contrast, omnivores-predators were identified as the most important predictors of both fungal and bacterial necromass in the no-till soil (Figure 3D). This shift suggests that higher trophic levels may play a critical role in regulating microbial necromass formation under no-till conditions with reduced soil disturbance.

The positive correlation between soil nematodes and microbial necromass has been attributed to microbial decomposition of nematode corpses, which stimulates microbial necromass formation (Kou et al. 2023). However, our study indicates that omnivores-predators influence microbial necromass through a downward trophic cascade along the omnivores-predators—fungivores—fungi food chain. In the no-till soil, there was an inverse relationship between omnivores-predators and fungivores, where an increase in omnivores-predators decreased the ¹³C content and enrichment rate of fungivores (Figure 5A and Figure 6A). This apparent reduction of fungivore activity disrupted the direct link from fungivores to fungi (Figures 5A and 6B) and was accompanied by an increase in the ¹³C content and enrichment rate of fungi (Figure S3). This cascade effect contributed to a superficial positive correlation between omnivores-predators and fungi (Figures 5A and 6B), which was crucial for the oscillations within the food chain (Hastings and Powell 1991; Terborgh and Estes 2013). The beneficial influence of omnivores-predators on fungi was further reflected in the fungal-to-bacterial necromass ratio, with fungi being the dominant contributor to this ratio in the no-till soil, surpassing other microbial groups (Figure 5B). By limiting fungivores, omnivores-predators indirectly accelerate fungal growth, leading to a greater contribution of fungi to the fungal-to-bacterial necromass ratio in the no-till soil compared to the till soil (Figure 6B).

Conversely, in the till soil, the inverse relationship between omnivores-predators and fungivores was absent. Instead, a direct trophic relationship between omnivores-predators and fungi was common, similar to the predator–prey relationship between fungivores and fungi. This relationship is supported by the feeding preferences of the omnivores-predators, with fungi being their second most important food resource in the till soil (Figure 4A,C). This is also reflected in the coordinated changes in ¹³C enrichment rates, where both omnivores-predators and fungivores increased in parallel with the fungal ¹³C enrichment rate (Figure 5A). In this case, omnivores-predators feed directly on fungi, shortening the trophic cascade and reducing the contribution of fungi to the fungal-to-bacterial necromass ratio.

Overall, omnivores-predators shaped fungi-fungal necromass dynamics through trophic cascades in the no-till soil. However, the causal interpretation derived from the observational data and the SEM model requires caution, as the direction of these relationships depends on model assumptions (Shipley 2016). To robustly disentangle trophic cascades within the complex soil food web, targeted manipulative experiments (e.g., quantifying fungal necromass formation across three or more trophic levels) are needed (Cao et al. 2024).

4.2 Bottom-Up and Top-Down Forces Jointly Trigger the Trophic Cascade in No-Till Food Web

A classic top-down cascade effect was evident in the no-till soil, where top-level omnivores-predators limited the activity of intermediate-level fungivores, indirectly promoting fungal growth and their necromass formation at the lower trophic level. According to trophic cascade theory, the bottom-up force, driven by resource availability, and the top-down force, driven by the traits of top predators (such as abundance and feeding strategy), are the primary factors triggering the occurrence of trophic cascades (Baum and Worm 2009; Pace et al. 1999; Terborgh and Estes 2013). By integrating a functional group network that reflected bottom-up responses with ¹³C-based predator–prey data, we constructed a framework of how these two forces interact to amplify the trophic cascade in the no-till soil (Figure 7).

Bottom-up resource availability plays a fundamental role in shaping the trophic cascade (Osakpolor et al. 2023; Ripple et al. 2016; Terborgh and Estes 2013; Zhou et al. 2025), providing the necessary energy for omnivores-predators to exert their top-down control. Following the transition from till to no-till practices, bottom-up forces were markedly strengthened (Figures 2 and 7). Omnivores-predators formed stronger connectivity with a range of functional groups, from primary consumers (e.g., fungi and bacteria) to secondary consumers (e.g., fungivores and bacterivores), facilitating a more efficient transfer of energy from the basal resources through the food web (Pascual and Dunne 2006; Rooney and McCann 2012). This intensified energy flow supported the growth and reproductive success of omnivores-predators, as reflected in their increased ¹³C content and enrichment rate in the no-till soil (Figure S3).

The top-down control, in turn, is driven by the increased abundance and activity of omnivores-predators, which regulate lower trophic levels by predation (Figure 7). As predators reduce the abundance of intermediate prey, a top-down cascade is triggered (Baum and Worm 2009; Ripple et al. 2016; Terborgh and Estes 2013). The transition from till to no-till practices resulted in distinct community-level changes for omnivores-predators, characterized by increased richness and abundance (Figure S1 and Table S4), and a shift in the dietary preferences, in particular an increase in consumption of fungivores to 51% (Figure 4C). These changes led to a typical top-down control by omnivores-predators on fungivores, as evidenced by their opposite temporal changes in the ¹³C content over the 1-year tracing period (Figure 6A).

Fungivores were represented by specialist fungivorous nematodes, whose activity was limited by omnivores-predators (Figures 5A and 6A,B). This limitation may increase the likelihood of fungal predation by other fungivore consumers, such as mites and collembolans (Potapov et al. 2022). Mites and collembolans, as typically facultative fungivores, are more mobile than nematodes and can move in and out of the food web more freely (Coleman et al. 2024; Potapov et al. 2022). Consequently, their interactions with fungi are likely to be less stable and more incidental, in contrast to the sustained relationship between fungivorous nematodes and fungi. This is particularly true in the soil food web based on soil clods, which harbor more nematodes but have fewer mites and collembolans (Kladivko 2001; Zhang et al. 2021). If mites and collembolans benefited from competition with fungivorous nematodes, we would expect changes in fungal ¹³C content over time, rather than the smooth temporal trend observed (Figure S3). Additionally, the reduced activity of fungivorous nematodes decreased their feeding pressure on fungi (Figure S5). Thus, the role of these mobile and facultative fungivore consumers is less important in the context of the stable and sustained interactions between fungivorous nematodes and fungi (Duffy et al. 2007; Filip et al. 2014). In this way, fungi benefit from the top-down trophic cascades triggered by omnivores-predators and dominate the transfer of straw-C to fungal necromass in the no-till soil (Figure 7).

In this framework, the bidirectional driving forces of bottom-up resource availability and top-down control increase the complexity and stability of the no-till soil food web structure. By providing a continuous flow of energy through the food web, bottom-up forces reinforce top-down cascading effects exerted by omnivores-predators, ultimately promoting fungal biomass and necromass accumulation in the no-till soil (Figure 7).

Soil microenvironment inhabited by organisms can also influence the occurrence of trophic cascades (Kravchenko et al. 2020; Pausch et al. 2018). For example, soils with high water-holding capacity and good pore connectivity increase the probability of encounters between consumers/predators and food sources/prey (especially invertebrates living in the water film), thereby creating trophic cascades (Erktan et al. 2020a). In our study, no-till over a five-year period allowed the soil environment to stabilize (Six et al. 2004), enabling soil organisms to adapt to the higher water holding capacity and increased inter-pore connectivity (Lavelle and Spain 2001). Such microenvironmental changes in the no-till soil, compared to the till soil, have been well demonstrated (Blanco-Canqui and Ruis 2018; Guo et al. 2020; Zhang et al. 2015). Thus, the feeding strategy of omnivores-predators and their limitation to fungivores most likely reflects inherent trophic relationships shaped by the microenvironment. While microenvironmental influences on trophic cascades were not the focus of this study, future research investigating these effects will further our understanding of soil ecosystem functioning.

4.3 Implication for Harnessing Top-Down Control to Improve Soil C Sequestration

The bottom-up resource availability and top-down predation jointly shape fungi-fungal necromass dynamics in the no-till system, leading to a 14% increase in the contribution of fungi to the ratio of fungal to bacterial necromass relative to the till system (Figures 6B). Notably, despite the same amount of maize straw input, continuous tillage resulted in the direct consumption of the fungi by omnivores-predators (Figure 4A,C). This direct consumption interrupted the trophic cascade and dampened the positive effects of omnivores-predators on fungal biomass and necromass formation (Figure 7). Consequently, the impact of tillage systems on the soil food web extends beyond bottom-up resource availability, and top-down predation regulation seems to play an outsized role in shaping energy flow and C cycling of the entire ecosystem (Coleman et al. 2024; Pascual and Dunne 2006).

Microbial necromass, particularly fungal necromass, is the most important contributor to soil C persistence because its components are slowly decomposable compared to bacterial necromass and can be adsorbed on minerals and form the most stable soil C pool—mineral-associated organic C (MAOC) (Klink et al. 2022; Sokol et al. 2022; Whalen et al. 2022). The ¹³C in situ tracing experiment revealed a strong linear correlation (R² = 0.73) between the fungal-to-bacterial necromass ratio and MAOC-¹³C (Figure S6), confirming the dominant contribution of fungal necromass to soil C stabilization. Given the longer turnover time of MAOC (mean 129 years) compared to other soil C pools (Zhou et al. 2024), the beneficial influence of omnivores-predators on fungi-fungal necromass dynamics may disproportionately increase long-term soil C sequestration by reinforcing fungal-derived C stabilization.

Overall, top-down predation regulation, rather than bottom-up resource availability, appears to be crucial in determining the efficiency of soil C sequestration. These findings provide a novel perspective for agricultural management, highlighting the importance of protecting and maintaining top predators within the soil food web. By doing so, it is possible to boost fungal necromass accumulation, thereby strengthening and stabilizing the long-term soil C pool, contributing to sustainable agricultural practices and climate change mitigation efforts.