COS 276-4 - Exploring fitness traits tradeoffs and microbial life history strategies using genome-scale data

Abstract: Identifying microbial life-history strategies and trait tradeoffs under environmental change is argued to improve our understanding of the role of microbial processes in ecosystem functioning. In this work, we explore microbial life strategies and related fitness-traits tradeoffs of grassland and shrubland soil microbial communities from the Loma Ridge Global Change Experiment in Southern California using the R tool Microtrait. Microtrait uses hidden Markov models to estimate fitness traits from metagenome-assembled genomes (MAGs). Following the YAS framework, estimated traits are categorized as resource acquisition, resource use, and stress tolerance traits. Traditionally, trait-tradeoff analyses use annotated microbial genomes from databases that might not incorporate emergent genome signatures under the environmental pressures that MAGs can offer. We ran Microtrait with 500 MAGs and derived 200 putative functional guilds with around 70% of similarity within the guilds. The fitness traits derived from Microtrait suggest that the putative guilds mainly invest in resource acquisition traits for substrate breakdown and transport of metabolites into the cell. In guilds with lower investment in resource acquisition, we did not observe trait tradeoffs against stress tolerance or resource use traits. This could indicate that more traits are playing a role in defining microbial life strategies. Within the resource acquisition traits category, guilds mainly invest in free amino acids uptake and protein degradation traits, and some tradeoffs are observed against monosaccharide transport and xyloglucan degradation traits. The results of this work will be used to parameterize a trait-based model to explore the effects of these microbial life strategies on soil functions such as litter decomposition and soil respiration. Thus, our work constitutes the first important step in organizing and understanding the existing biodiversity before any modeling effort. Additionally, our approach using genome-scale data opens up new possibilities for data-driven model development that could ultimately strengthen our understanding of microbial processes and ecosystem functioning.