The role of habitat connectivity in the assembly of the Golden Oak metacommunity
Habitat fragmentation has been shown to alter the relative importance of spatial vs. environmental processes as drivers of metacommunity structure (Jamoneau, Chabrerie, Closset-Kopp, & Decocq, 2012), but the way we measure spatial distances among habitat patches and account for the effects of the surrounding matrix may affect our interpretation of the predominant processes (Resasco & Fletcher, 2021; Watling, Nowakowski, Donnelly, & Orrock, 2011). In the case of the highly fragmented Golden Oak (Qa) habitat, circuit theory-based connectivity modelling demonstrated that isolation estimates accounting for fragmentation and matrix resistance consistently performed better at explaining turnover than those based on topography, with the latter performing better than null models assuming a homogeneous matrix (Table 3; Table S7). This finding aligns with recent studies that have documented the ecological importance of dispersal corridors for community assembly across different taxonomic groups (Firmiano et al., 2021; Marrec et al., 2021; reviewed in Fletcher, Burrell, Reichert, Vasudev, & Austin, 2016). However, this approach relies on the assumption that all species of the metacommunity respond similarly to landscape heterogeneity. Future studies integrating metabarcoding with morphological information derived from local ‘voucher’ reference collections could facilitate the implementation of species-specific analyses accounting for ecological and trait variation (Brodie & Newmark, 2019; Hartfelder et al., 2020). Despite these limitations, our results provide empirical evidence of an important effect of habitat fragmentation on soil microarthropod metacommunity structure across the Golden Oak forest patches, with the role of environmental filtering remaining equally significant (Table 3). Interestingly, we also obtained equivalent distance-decay of community similarity curves based on connectivity or on topoclimatic distances (Figure 5), although the topoclimatic variables were not spatially autocorrelated (Mantel test, r <0.103, p-value >0.188) and their shared variance with connectivity-based predictors was very low (<2%; Table 3). The similar slopes of the decay curves for ASVs and OTUs based on spatial distances are compatible with a neutral dispersal-constrained model (Baselga et al., 2015), while the equivalent pattern based on topoclimatic distances could be generated under certain scenarios of high dispersal and narrow ecological niches (Baselga et al., 2013). Taken together our results suggest that dispersal limitation and niche-based processes have jointly shaped turnover patterns across the Qa habitat, although each process may affect different fractions of the metacommunity (e.g., some species and/or intraspecific entities may have low dispersal propensity and wide topoclimatic niches, while others might be good dispersers with narrow niches). Future species-specific analyses may help to refine these conclusions.