Environmental drivers
Abiotic environmental variables correlated with elevation to a varying degree in the three sampled regions. As expected, MAT showed strong negative correlation with elevation in all three regions. MAP generally was not correlated with elevation except in Argentina, where it was highest in the lower montane forest zone (Fig. S4). Soil pH tended to decrease with elevation in all regions, although in Panama the trend was not significant. In Borneo, pH increased again above 3500 masl, representing the sparsely vegetated summit zone of Mt. Kinabalu. In a related manner, soil organic material (OM), and to a lesser extent N, were correlated positively with elevation in general, again with change of trend in the summit zone of Mt. Kinabalu (Fig. S4).
GNMDS ordinations revealed strong structuring of fungal communities according to elevation in all three regions (Pearson’s correlations values: r 2 = 0.8633,r 2 = 0.9406, and r 2 = 0.9661; all p = 0.001) in Argentina, Borneo, and Panama, respectively (Fig. 1). With respect to abiotic variables, MAT and soil pH were correlated strongly with fungal community composition in all three regions (all p = 0.001), whereas other climatic and edaphic variables differed in their importance among the regions. Partial Mantel tests indicated that abiotic environmental variables were linked robustly to community structure in all sampling areas (Argentina:r = 0.636, p < 0.001; Borneo: r = 0.444,p < 0.001; Panama r = 0.576, p< 0.001), when spatial proximity was accounted for (control matrix), whereas spatial proximity was not significant (all p ˃ 0.1) when abiotic variables were controlled.
In the PerMANOVA analyses, MAT and pH tended to explain the greatest variation in the total fungal community composition and remained significant contributors in the combined model in all three regions when correlation among parameters was accounted for. The contribution of other abiotic variables varied among functional groups and geographic regions, with P and OM explaining significant proportions of variation in fungal community structure in at least two regions and MAP being especially important in Argentina (Table 1). These results are consistent with Hypothesis 1, particularly with respect to MAT, pH, and to a lesser extent MAP, as drivers of soil fungal community composition.