Franziska Bauchinger David Seki David Berry ^*

• Centre for Microbiology and Environmental Systems Science CeMESS, University of Vienna, Vienna, Austria

Keystone species are thought to play a critical role in determining the structure and function of microbial communities. As they are important candidates for microbiome-targeted interventions, the identification and characterization of keystones is a pressing research goal. Both empirical as well as computational approaches to identify keystones have been proposed, and in particular correlation network analysis is frequently utilized to interrogate sequencing-based microbiome data. Here, we apply an established method for identifying putative keystone taxa in correlation networks. We develop a robust workflow for network construction and systematically evaluate the effects of taxonomic resolution on network properties and the identification of keystone taxa. We are able to identify correlation network keystone species and genera, but could not detect taxa with high keystone potential at lower taxonomic resolution. Based on the correlation patterns observed, we hypothesize that the identified putative keystone taxa have a stabilizing effect that is exerted on correlated taxa. Correlation network analysis further revealed subcommunities present in the dataset that are remarkably similar to previously described patterns. The interrogation of available metatranscriptomes also revealed distinct transcriptional states present in all putative keystone taxa. These results suggest that keystone taxa may have stabilizing properties in a subset of community members rather than global effects. The work presented here contributes to the understanding of correlation network keystone taxa and sheds light on their potential ecological significance.