Does microbial community structure matter for predicting ecosystem function? Use of statistical models to examine relationships between the environment, community and processes

Abstract:

Microorganisms control all major biogeochemical cycles, yet the importance of microbial community structure for ecosystem function is widely debated. Indeed, few nutrient cycling models directly account for variation in community structure, leading some researchers to speculate that this information could provide important and missing explanatory power to predict ecosystem function. However, if variation in environmental variables strongly correlates with variation in microbial community composition, then information on microbial community composition may not improve models. Here, we use a data synthesis approach to ask when and where information on the microbial community matters for predictions of ecosystem function. We collated data from approximately 100 different studies and used statistical approaches to ask if models with data on microbial community composition significantly improved models of ecosystem function based on environmental data alone. We found that only 25% of models of ecosystem processes were significantly improved with the addition of data on microbial community composition. Specifically, we found that for phylogenetically broad processes, diversity indicators yielded more significant increases in explanatory power than abundance data. Our results also demonstrate that for phylogenetically narrow processes, qPCR data on functional genes yielded higher explanatory power than for broad processes. Further, we found that all types of data on microbial community composition explained more variation in obligate processes compared to facultative processes. Overall, our results suggest that trait distributions both within communities and within individuals affect the relative importance of microbial community composition for explaining ecosystem function.