Fig. 2

Relative contributions of climatic and local non-climatic predictors in shaping the functional gene alpha diversity and assemblage composition. The independent effects of the selected predictors on the Shannon diversity (a) and assemblage composition (c) were examined by hierarchical partitioning, and their significance (P < 0.05) was tested through a 1000 randomization-based procedure. The R² and P values above the plots in (a) and (c) were calculated by linear models. The variances in Shannon diversity (b) and assemblage composition (d) associated with the climatic and local non-climatic predictors were obtained using variation partitioning, based on adjusted R² and significances tested with analysis of variance. As local predictors were not selected in Spain, the variance associated with climatic predictors was determined through linear models for the Shannon diversity and a Mantel test (1000 permutations) for the assemblage composition, with test significance based on F-statistic and Pearson r value, respectively. The significance levels in the variation partitioning are indicated by *P < 0.05, **P < 0.01, ***P < 0.001. The assemblage composition was estimated using first axis coordinates from principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity matrices (see Fig. S3 in Additional file 1 for details on the axis-explained variance). Green and yellow bars represent climatic and non-climatic local predictors, respectively. Details on predictor selection are presented in “Methods” and in Additional file 1 (Fig. S14). TWQ, mean temperature of the warmest quarter; TAP, total annual precipitation; PCQ, mean precipitation of the coldest quarter; TAR, temperature annual range; IST, isothermality (relationship between air temperature diurnal and annual ranges); TSE, temperature seasonality; Chl-a, chorophyll-a; Cveloc, current velocity; Shad, shading; TN, total nitrogen