Viral but not bacterial community succession is characterized by extreme turnover shortly after rewetting dry soils

Abstract

As central members of soil trophic networks, viruses have the potential to drive substantial microbial mortality and nutrient turnover. Pinpointing viral contributions to terrestrial ecosystem processes remains a challenge, as temporal dynamics are difficult to unravel in the spatially and physicochemically heterogeneous soil environment. In Mediterranean grasslands, the first rainfall after seasonal drought provides an ecosystem reset, triggering microbial activity during a tractable window for capturing short-term dynamics. Here, we simulated precipitation in microcosms from four distinct, dry grassland soils and generated 144 viromes and 84 metagenomes to characterize viral, prokaryotic, and relic DNA dynamics over 10 days. Vastly different viral communities in each soil followed remarkably similar successional trajectories. Wet-up triggered a significant increase in viral abundance and richness, followed by extensive compositional turnover. While temporal turnover in prokaryotic communities was much less pronounced, differences in the relative abundances of Actinobacteria (enriched in dry soils) and Proteobacteria (enriched in wetted soils) matched those of their predicted phages, indicating viral predation of dominant bacterial taxa. Rewetting also rapidly depleted relic DNA, which subsequently re-accumulated, indicating substantial new microbial mortality in the days after wet-up, particularly of the taxa putatively under phage predation. Production of abundant, diverse viral particles via microbial host cell lysis appears to be a conserved feature of the early response to soil rewetting, and results suggest the potential for "Cull-the-Winner" dynamics, whereby viruses infect and cull but do not decimate dominant host populations.

Viromes vs. mixed community metagenomes: choice of method dictates interpretation of viral community ecology

“…Hypergeometric testing is an effective way to infer functional enrichment from differentially abundant genes across groups [99–101], and it has been applied to the study of viral communities as well [84,102]. Our hypergeometric test results revealed several gene functional categories that were over and underrepresented in viromes and metagenomes.…”

“…Viral communities can also be assembled if extra measures are taken to enrich for virus-like particle DNA during extraction to generate viromes. Although viromes can generally offer a more focused view of viruses in a sample compared to metagenomes [23], the consequences of choosing one sampling method over the other have relatively unexplored and limited to individual study ecosystems [5,6,29]. Here, we applied the same analytical methods to collections of paired virome and metagenome sequence reads to directly infer the unique and shared results gained from each sample method.…”

“…While there have been a handful of studies in the past that have examined viral community data resulting from viromes in comparison to metagenomes [6,11,29,84,85], even fewer have taken a closer look at specific genome-level differences that result across the two methods. Here, we examined the effects that choice of extraction method has on resulting viral genome assemblies.…”

“…With this, our next aim was to assess whether enrichment of genes in either viromes or metagenomes led to enrichments in any particular function across sample methods. Hypergeometric testing is an effective way to infer functional enrichment from differentially abundant genes across groups [99][100][101], and it has been applied to the study of viral communities as well [84,102]. Our hypergeometric test results revealed several gene functional categories that were over and underrepresented in viromes and metagenomes.…”

Benchmarking Bioinformatic Virus Identification Tools Using Real-World Metagenomic Data across Biomes

“…(3) the persistence of DNA in dead microbial matter 83,84 , which should be removed during the DNase step in the virome preparation but may persist in the microbiomes. Seawater datasets had the best assembly quality with mean percentages of properly paired reads of 79% and 45% for viral and microbial datasets, respectively.…”