Richness and uniqueness of soil microarthropod communities
After denoising and chimera filtering, the removal of putative spurious sequences by metamate, followed by additional filtering of community tables, generated a fully-filtered dataset comprising 907 ASVs which clustered into 386 OTUs (putative species) across the three taxonomic groups. Specifically, we retrieved a total of 353 ASVs and 154 OTUs of Coleoptera, which was the most diverse group at both across- and within-habitat scales. The Acari and Collembola datasets comprised 237 and 317 ASVs which clustered into 139 and 93 OTUs, respectively (Figure 2).
The average α diversity of haplotypes per sampling site differed among habitat types, with the Cyprus cedar (Cb) habitat showing significantly higher richness than the other four forest types (Tukey’s test: p-value <0.028 in all comparisons involving Cb; Figure 2). At OTU level, the average α diversity per site was higher in the forest types distributed at low and mid altitudes (PbQa and Cb) than in those habitats restricted to higher elevations (Pn and Jn Figure 2). However, the significance of this pattern was not preserved after post-hoc tests (Tukey’s test: all p-values >0.123). When soil substrates were independently compared, the leaf litter layer showed significantly higher richness (α diversity) than the deep soil substrate in all comparisons (Figure S2). The average local contribution to β diversity (LCBD) per sampling site differed significantly among habitats, with the Calabrian Pine (Pb) and Golden Oak (Qa) communities showing significantly higher uniqueness at both ASV and OTU level when compared to the high-altitude Pn and Jn communities (Tukey’s test: p-value <0.05 in most comparisons involving Pb or Qa; Figure 2). When independently analyzed, deep soil communities showed significantly higher LCBD estimates, indicating a more unique composition than the leaf litter communities (Figure S2, inset plots), a pattern particularly evident in the high-altitude Pn and Jn communities (Figure S2).
Regression analyses (GLMMs/GLMs; see Supplemental Information) showed a significantly negative relationship between average richness (α diversity) of ASVs per site and longitude (Lon), indicating that community richness decreased towards the east of the Troodos mountain range (Table 1; Table S4). To ensure that this relationship was not biased by the Cyprus cedar (Cb) sites, which are geographically restricted to the westernmost part of the study area and have the highest ASV richness (Figures 1-2), additional analyses excluded these sites. These analyses consistently supported the significant effect of longitude on ASVs richness across forest habitats (95% CI: [-34.968] - [-4.032]). Conversely, the richness (α diversity) of OTUs per site was only explained by the topoclimatic predictor ENVPC2 (Table 1; Table S4). Similarly, we found that LCBD estimates at both ASV and OTU levels were significantly correlated with the topoclimatic variables, as summarized with the ENVPC1 and ENVPC2 predictors (Table 1; Table S4).