Statistical analyses.

We used STAMP v2.1.3 (70) and R v3.5.0 to make statistical comparisons between the communities according to both time after wetting and successional stage. Within the R statistical environment, we used the vegan package (71) to explore ecologically meaningful differences between communities according to time after wetting and successional stage. Bray-Curtis dissimilarity matrices were used to explore biocrust community structure after Hellinger transformations of species relative abundance data. PERMANOVA was used to test for differences in community structure on the basis of time after wetting (dry biocrust and biocrust at 3 min, 9 h, 18 h, 42 h, and 49.5 h after wetting) and successional maturity (early, early-mid, late-mid, and late) using the adonis function. We used STAMP to test for significant differences in taxonomic and functional composition between time points and successional stages and used ANOVA with Tukey-Kramer post hoc tests and Bonferroni corrections of the P value statistic.

To obtain viral contigs from biocrust metagenomes, we coassembled samples according to their characteristics and community complexity. The coassemblies were produced by combining the 4 filamentous bundles, the 4 dry biocrust samples, the 5 early successional stage biocrust samples, the 5 early-mid succession biocrust samples, the 5 late-mid succession biocrust samples, and the 5 late succession biocrust metagenomes into single, concatenated metagenomes. This produced 6 coassemblies to complement the 28 individual metagenomes.