FIGURE 4 Venn diagram illustrating the partitioning of explained variance in community composition among the three sets of explanatory variables (forest habitat type, spatial and topoclimatic predictors) and their intersections. Variance partitioning was conducted on the best-fit dbRDA models, which only included significant variables after a forward selection procedure (Table 2). Analyses were performed on the Simpson dissimilarity index (βSIM) community dissimilarity matrices at ASV and OTU levels. Percentages refer to adjusted coefficients of determination (R 2ADJ). Percentages below 0.25% are not shown.