More sensitive microbial responses to the interactive effects of warming and altered precipitation in subsoil than topsoil of an alpine grassland ecosystem.

Subsoil is a large organic carbon reservoir, storing more than half of the total soil organic carbon (SOC) globally. Conventionally, subsoil is assumed to not be susceptible to climate change, but recent studies document that climate change could significantly alter subsoil carbon cycling. However, little is known about subsoil microbial responses to the interactive effects of climate warming and altered precipitation. Here, we investigated carbon cycling and associated microbial responses in both subsoil (30-40 cm) and topsoil (0-10 cm) in a Tibetan alpine grassland over 4 years of warming and altered precipitation. Compared to the unchanged topsoil carbon (β = .55, p = .587), subsoil carbon exhibited a stronger response to the interactive effect of warming and altered precipitation (β = 2.04, p = .021), that is, warming decreased subsoil carbon content by 28.20% under decreased precipitation while warming increased subsoil carbon content by 18.02% under increased precipitation.Furthermore, 512 metagenome-assembled genomes (MAGs) were recovered, including representatives of phyla with poor genomic representation. Compared to only one changed topsoil MAG, 16 subsoil MAGs were significantly affected by altered precipitation, and 5 subsoil MAGs were significantly affected by the interactive effect of warming and precipitation. More than twice as many populations whose MAG abundances correlated significantly with the variations of carbon content, components and fluxes were observed in the subsoil than topsoil, suggesting their potential contribution in mediating subsoil carbon cycling. Collectively, our findings highlight the more sensitive responses of specific microbial lineages to the interactive effects of warming and altered precipitation in the subsoil than topsoil, and provide key information for predicting subsoil carbon cycling under future climate change scenarios.