Estimates of α and β diversity at ASV and OTU levels
The resulting fully-filtered ASVs were clustered into operational taxonomic units (OTUs) based on patristic pairwise distances, using a threshold of 3% genetic divergence, which is commonly applied on arthropod COI metabarcoding datasets (Dopheide et al., 2019; Yu et al., 2012). Patristic distances were calculated based on a UPGMA tree, reconstructed using F84 model-corrected pairwise distances and a mafft FFT-NS-i alignment (Katoh & Standley, 2013). These analyses were conducted using the adegenet (Jombart, 2008), phangorn (Schliep, 2011) and ape (Paradis & Schliep, 2019) packages in R (R Core Team, 2020).
Using the ASV and OTU datasets, we calculated α diversity for each soil layer, sampling site and habitat type. We also calculated community dissimilarity among sampling sites and between soil layers using the Sørensen index (βSOR), which we decomposed in its additive components (Simpson dissimilarity index or spatial turnover without the effect of variation in richness, βSIM, and nestedness, βSNE; Baselga, 2010). Then, we used these community dissimilarity matrices to calculate the local contribution to β diversity (LCBD), a comparative indicator of the uniqueness of each sample in terms of community composition (Legendre & De Cáceres, 2013). We calculated the above diversity metrics using the betapart (Baselga & Orme, 2012) and adespatial (Dray et al., 2020) R packages.