Fig. 1

Responses of functional gene alpha diversity and compositional turnover to elevation. The relationships between Shannon diversity and elevation (a, c) were examined by linear models, and the model significances were determined with F-statistics (P < 0.05). For kingdoms (a), we considered linear and quadratic terms and selected the best models, i.e. those that minimized the corrected Akaike’s information criterion. The adjusted R² values were 0.394, 0.181 and 0.672 for archaea, 0.346, 0.394 and 0.829 for bacteria, and 0.428, 0.346 and 0.844 for fungi, in Norway, Spain and China, respectively. The relationships between Bray-Curtis dissimilarity and elevational Euclidean distance (b, d) were calculated by linear models, and the model significances were obtained by a Mantel test (1000 permutations, P < 0.05). For kingdoms (b), the Pearson r values were 0.212, 0.361 and 0.643 for archaea, 0.199, 0.28 and 0.700 for bacteria and 0.283, 0.25 and 0.692 for fungi, in Norway, Spain and China, respectively. Across functional categories, we show the slope values from LMs assessing the gene family alpha diversity-elevation relationships (c) and the gene family compositional turnover-elevational distance relationships (d). For kingdoms (a, b), significant and non-significant models are shown as solid and dashed lines, respectively. The violin boxplots for the Shannon diversity (c) and Bray-Curtis dissimilarity (d) depict the median and the first and the third quartiles of the slopes of the gene families with significant (P < 0.05) models. Across functional categories, differences among mountains in the model slopes for the Shannon diversity (c) and Bray-Curtis dissimilarity (d) were examined with a Bonferroni-corrected pairwise t-test (P < 0.05) and are indicated with the symbols + and ▲. The elevation (a) and elevational distances (b) are shown as raw data for visualization purposes (z-transformed for the analyses). C, carbon; N, nitrogen; P, phosphorus; S, sulphur; NO, Norway; SP, Spain; CH, China. Details on the models can be found in methods