How belowground microbial communities will respond to multiple climate change factors such as higher ambient temperatures and altered precipitation events and what the impacts of their activities will be on soil carbon cycling processes remain uncertain, limiting predictive projections of soil carbon stocks. To provide insights into these issues, we have been performing experimentally warming 2 to 4 °C above ambient temperature and precipitation manipulations, in-situ, at several sites including an Alaskan tundra permafrost (AK), an Oklahoma temperate grassland (OK), a Eurasian steppe (China), and a tropical forest soil in Puerto Rico (PR). By coupling respiration data and soil indices with well-replicated whole-community shotgun metagenomic sequencing from control and warming plots we hope to improve our understanding of temperature sensitivity of soil organic matter decomposition by soil microorganisms. To enable this research, we have been also developing bioinformatics tools for metagenome analysis and comparison as well as for data integration, and we have been making these tools available for online analysis or download for standalone applications through a dedicated webserver (http://enve-omics.ce.gatech.edu/).
Metagenomes collected after 1 and 5 years of warming yielded near-complete representation of microbial community ‘sequence richness’ at AK, and revealed that AK communities are ~10 times less diverse than their temperate and tropical counterparts. A custom-made assembly and contig binning strategy allowed for the recovery of many near-complete bacterial population genomes from all locations. In particular, populations recovered from AK soils collectively made up to ~20% of the total microbial community, which is remarkable for soil, and allowed us to robustly assess the biogeography of these populations across our sites. For instance, we found that the AK dominant populations were present, and often abundant, in geographically distant (~100-530 kilometers apart) tundra habitats (>98% genome-derived average nucleotide identity) but not in OK or PR samples. Warming alone, or in combination with precipitation, induced significant shifts in pathways related to SOM-decomposition in all sites, resulting in less organic carbon in soil, albeit the magnitude of the shifts depended on the site considered. AK tundra clearly showed the largest shifts among all sites, with dominant bacterial populations shifting in abundance by as much as 80% in response to the warming treatment. Therefore, even such mild warming conditions induce detectable changes in microbial communities and these changes mediate mostly positive feedback responses, while the tundra ecosystem is the most vulnerable one among those studied.