Fig. 3

Overview of the temporal dynamics of total and active DNA and RNA viral communities. A Distribution of all DNA (blue) and RNA (green) vOTUs across metagenome and metatranscriptome datasets. The vOTUs and samples are clustered based on vOTU relative abundances (log-transformed RPKM). Color bars above the heatmap indicate the location, depth, season, and type of each dataset. The left bar (category) indicates the vOTUs which are classified as DNA or RNA virus. B–G Beta-diversity of total RNA (B), total DNA (C), and active DNA (D) viral communities across the 4 seasons. For each group of viruses, non-metric multidimensional scaling (NMDS) ordination plots, representing the (dis)similarities between samples based on vOTU relative abundance, are presented in the left panels (B, C, and D). Individual samples are colored based on season: September (yellow), March (blue), May (red), and June (green). Stress values associated with two-dimensional ordination and PERMANOVA results describing the variance in community composition explained by season are reported for each plot. The RPKM of total RNA (E), total DNA (F), and active DNA (G) vOTUs predicted to infect putative host groups (for RNA vOTUs) or bacterial class (for DNA vOTUs) is indicated for each. “Other” represents the remaining host classes (representing less than 0.1% of hosts). For “active” DNA, only vOTUs identified as active were considered (see the “Methods” section), and the RPKM from metatranscriptome read mapping was used as an estimation of the relative abundance instead of RPKM from metagenome read mapping