Elevational patterns of fungal functional groups
All functional groups exhibited significant (p < 0.05) differences in community composition among the three elevational forest types in all geographic regions, consistent with our general Hypothesis 2. GNMDS plots of the datasets corresponding to functional groups showed similar correlations with elevation and edaphic variables as detailed above. The GNMDS ordinations of the four largest functional groups in terms of OTU richness are shown in Fig. 2, with richness of taxonomic orders displayed as vectors. In all regions, forest type was a significant source of variation in community composition in almost all functional groups, explaining ca. 20% of the variation in all fungi and between 11.5% and 26.5% in the functional groups in Argentina, between 15.3% and 32.6% in Borneo, and between 13.0% and 31.5% in Panama (Table 1). We did not find support for greater community turnover among elevational forest types in plant-associated than in non-plant-associated functional groups, in contrast to Hypothesis 2a.
Differences in richness among elevational forest types also were observed in all functional groups in at least one region. Total fungal richness was generally similar in all three elevational forest types, except in Panama, where lower montane forests harbored significantly more fungal OTUs than the other forest types (Fig. 3). We found three functional groups, all associated with plants (ECM fungi, plant pathogens, and wood decomposers), for which observed OTU richness as well as proportional richness differed significantly among elevational forest types in all three regions (Fig. 3, Fig. S5). Root-associated fungi, the only remaining plant-associated guild, and animal pathogens, differed significantly in richness in two regions, while richness of mycoparasites and generalist saprotrophs generally did not differ meaningfully among the elevational forest types. Greater differences among vegetation types in plant-associated fungi compared to non-plant-associated guilds are consistent with Hypothesis 2a.
Observed richness and proportional richness of plant pathogens and wood decomposers, and, to a lesser extent, animal pathogens and generalist saprotrophs, tended to be higher in lowland and/or lower montane forests than in upper montane forests, largely consistent with Hypothesis 2b. In contrast, ECM fungi, mycoparasites, and non-ECM root-associated fungi showed mixed patterns (Fig. 3, Fig. S5). For example, in Neotropical forests, ECM fungi peaked in richness in the upper montane forests, but they were most diverse in the lower montane forests in Borneo. Similar patterns also were observed with respect to proportional richness and to some extent in proportional abundance, with the notable exceptions of a pronounced abundance peak of mycoparasitic fungi and the abundance minimum of saprotrophs in the lower montane forest in Borneo, and the lack of significant differences in the proportional abundance of wood decomposers among elevation zones in all three regions (Fig. S6).