Bioinformatics
Sequences were sorted according to samples and adapters (identification tags) were removed in Galaxy (https://main.g2.bx.psu.edu/root). Primer sequences were removed and poor-quality ends were trimmed based on a 0.02 error probability limit in Geneious Pro 5.6.1 (BioMatters, New Zealand). Sequences were filtered with USEARCH v.8.0 (Edgar 2010) with the following settings: all sequences were truncated to 200 bp and sequences with expected error > 1 were discarded. For each sample, identical sequences were collapsed into unique sequence types while preserving their counts. The quality-filtered sequences from all samples were grouped into operational taxonomic units (OTUs) at 97% sequence similarity in USEARCH, and global singletons and putative chimeric sequences were removed. We assigned sequences to taxonomic groups based on pairwise similarity searches against the curated UNITE+INSD fungal ITS sequence database (released August 22, 2016), containing identified fungal sequences with assignments to Species Hypothesis groups based on dynamic similarity thresholds (Kõljalget al . 2013). We excluded OTUs with < 80% similarity or < 150 bp pairwise alignment length to a fungal reference sequence. To minimize artifactual OTUs that may have been generated during the molecular work, in subsequent analyses we only included OTUs that occurred in at least two samples. The final dataset contained 20 620 fungal OTUs, representing a total of 3 962 204 high-quality sequences.
We assigned fungal OTUs to functional groups in two steps. First, we assigned putative functional guilds using FunGuild (Nguyen et al . 2015). We recognize the limitations of functional inference based on partial ITS sequences, and here use these guilds as hypothetical functional groups. Secondly, for genera that comprise species from multiple functional guilds (e.g., Amanita , Entoloma ,Ramaria , Sebacina and related genera in the Sebacinales), we assigned ecological function to each OTU on a case by case basis with available ecological information for the matching Species Hypothesis in the UNITE database. Fungi that can grow as saprotrophs as well as animal or plant pathogens were designated as animal or plant pathogens, respectively, to acknowledge the ecological distinction from purely saprotrophic fungi. Similarly, we separated fungi capable of degrading complex carbohydrates of wood from generalist saprotrophs that generally can utilize only simple sugars. In total, functional assignments were made for 13 356 OTUs (ca. 65%). The quality-filtered and rarefied dataset contained 8720, 7829, and 4498 fungal OTUs in the samples from Argentina, Borneo, and Panama, respectively, of which 5917, 5078, and 2896 OTUs were assigned to functional groups, respectively. Targeted Locus Study projects corresponding to the three study regions have been deposited at DDBJ/EMBL/GenBank under accessions KDPX00000000 (Argentinian Yungas), KDPY00000000 (Malaysian Borneo), and KDPZ00000000 (Panama). The versions described in this paper are the first versions, i.e. KDPX01000000, KDPY01000000, and KDPZ01000000, respectively.