Viral communities vary more across spatial scales than across annual sampling in soils recovering from decades long heating from an underground coal mine fire in Centralia, PA
This work is published.
Both the raw read data and metagenome assemblies for this study are available through NCBI under bioproject PRJNA974462
Barnett SE, Shade A (2025). Soil viral community dynamics over seven years of heat disturbance: spatial variation exceeds temporal in annually sampled soils. Soil Biology and Biochemistry, 203:109741. DOI: 10.1016/j.soilbio.2025.109741
Viruses are important components of the soil microbiome, influencing microbial population dynamics and the functions of their hosts. However, the relationships and feedbacks between virus dynamics, microbial host dynamics, and environmental disturbance are not understood. Centralia, PA, USA, is the site of an underground coal seam fire that has been burning for over 60 years. As the fire moves along the coal seam, previously heated soils cool to ambient temperature, creating a gradient of heat disturbance intensity and recovery. We examined annual soil viral population dynamics over seven consecutive years in Centralia using untargeted metagenome sequencing. Viral communities changed over time and were distinct between fire-affected and reference sites. Dissimilarity in viral communities was greater across sites (space) than within a site across years (time), and cumulative viral diversity more rapidly stabilized within a site across years than within a year across sites. There also were changes in number of CRISPR arrays per genome as soils cooled, corresponding to shifts in viral diversity. Finally, there were also differences in viral-encoded auxiliary metabolic genes between fire-affected and reference sites. Thus, despite high site-to-site soil viral diversity, there was surprising viral community consistency within a site over the years and shifting host-viral interactions in soils recovering from disturbance. Overall, this work provides insights into the interannual dynamics of soil viruses and their host communities, as well as how they collectively respond to long-term warming.
Code is split up into two directories: Sequence_processing and Analysis.
Code used for sequence processing including viral sequence filtering, vOTU clustering, taxonomy assignment, and annotation can be found under Sequence_processing. Scripts were run using SLURM on the MSU HPCC using slurm batch files with suffix .sb and are numbered by their order in the processing workflow. Outputs such as logs, warnings, or errors if any, are designated by the suffix .out and named in accordence with the library, run number, and slurm batch file.
Formal analysis can be found under Analysis. All analysis was run with R and code was run in Rmarkdown. In the analysis directory you'll find the raw Rmarkdown files (.Rmd), a github friendly markdown rendering (.md) and the associated figure files from the rendering in separate sub-directories. The analysis was broken down into multiple chunks in separate Rmarkdown files:
- Main viral analysis: The majority of the viral analysis included in the manuscript.
- Viral vs. Bacterial analysis: Compairing the viral community to the bacterial community (Mantel test and Procrustes analysis mainly).
This work was supported by the U.S. National Science Foundation CAREER award 1749544. This work was supported in part by Michigan State University through computational resources provided by the Institute for Cyber-Enabled Research.