Impact of ecological filters within habitats
Our study design aimed to investigate the impact of ecological processes on soil food web structure across altitudinal gradients and diverse environmental conditions. Although the GDM explained a relatively low percentage of variance when quantifying changes in the structure of local food webs (Fig. 5a), it’s important to note that typical deviance explained for models of compositional dissimilarity ranges from 20% to 50%, and this percentage can decrease with an increasing number of sites, which was relatively high in our study (Ferrier et al.2007; Mokany et al. 2022). Furthermore, the resolution of our soil food webs using trophic groups instead of species led to low average dissimilarities (0.27 and 0.48 in forests, and 0.30 and 0.53 in grasslands for groups and interactions respectively), reducing model predictability (Mokany et al. 2022). Importantly, we found that trophic link structure significantly influenced changes in trophic groups abundance composition (Fig.4). However, current models cannot account for this factor, and incorporating network structure in diversity models may enhance predictability but requires further implementation (Poggiato et al. 2022).
Spatial processes, environmental filtering, and changes in plant community composition explained variation in food web structure within habitats. In forests, spatial distance played a significant role, explaining a substantial portion of trophic group (31%) and interaction dissimilarity (58%), even when accounting for confounding environmental changes. Limited dispersal opportunities in closed forest habitats, particularly for organisms dispersed through wind (anemochory), might influence soil food web structure in mountain forests. Conversely, open grassland habitats subject to strong winds (Tackenberg & Stöcklin 2008) and high levels of herbivory, both domestic and wild, can facilitate the dispersal of soil organisms. Furthermore, mountain massifs and large distances between forested areas (up to 250 km) can impact meso and macro fauna, susceptible to ecological drift due to their limited dispersal capacity or smaller population sizes (Arribas et al.2021; Kang et al. 2022). This is in line with our results showing that food web dissimilarity increased with greater spatial distances (Fig.5b). Yet, the spatial distance used in the model may not fully capture the complexity of spatial, and further investigation using experimental setups to clarify spatial connectivity and dispersal limitation effects is required.
Environmental filtering played a significant role in both habitats, primarily influenced by soil edaphic properties. In the French Alps, the soil C/N ratio emerged as the most influential abiotic factor in shaping soil food web composition. It reflects organic matter decomposability and resource availability, as discussed above, impacting decomposer groups, their consumers, and the entire feeding channel Conversely, variables related to total energy or resources, such as the NDVI and SOM, had a smaller impact on soil food web dissimilarity. While this contrasts with previous studies emphasizing the role of these variables in diversity patterns of soil groups in mountains (Zinger et al.2011; Calderón‐Sanou et al. 2022), it highlights the emergence of different ecological drivers at different levels of biodiversity organization. In our study, resource quality, particularly litter quality, seemed to dominate food web composition rather than quantity. Additionally, environmental factors associated with harsh conditions, such as pH and frozen degree days, known drivers of soil taxonomic diversity and composition (Fierer & Jackson 2006; Decaëns 2010), also played a major role in shaping soil food webs.
Plant-soil interactions likely play a crucial role in shaping grasslands food webs in the French Alps. Although we did not directly measure these interactions, the taxonomic and/or functional composition of plant communities had a major impact on soil food web structure. Mountain grasslands exhibit high variability in plant species richness and the presence of functionally important taxa such as Fabaceae, Brassicaceae or Poaceae across different altitudes and massifs. Changes in plant diversity can influence the composition of soil food webs (Eisenhaueret al. 2013) , and altering plant identity can affect different soil taxa (Zinger et al. 2011). Interestingly, taxonomic dissimilarity had a larger effect than plant functional dissimilarity, possibly because above-ground traits used in our study may have a lesser role in plant-soil interactions compared to root traits (Kardol & De Long 2018; Wilschut et al. 2019). Surprisingly, in forests, plant composition had a minor impact on food web structure. Soil variables like pH and C/N ratio may better capture variations in forest type (e.g., broadleaf vs coniferous forests), directly influencing soil food webs.