Abstract

Co-authors: Hélène Morlon, Benoît Pérez-Lamarque, Leo Lahti Understanding the evolution of host-associated microbiota requires accounting for the phylogenetic history of their hosts, because microbiota are transmitted across host generations and because evolutionarily conserved host traits influence microbiota composition. To understand microbiota evolution from large-scale data, we developed an approach to model the changes in microbiota composition through time along the phylogenetic tree of the hosts, where microbiota composition is characterized as the relative abundances of higher-level microbial taxa. We did so by leveraging existing approaches for modelling the evolution of multivariate phenotypic traits. Our approach provides a model-based quantification of phylosymbiosis, i.e., the propensity of microbiota composition to reflect the evolutionary divergence between hosts. It also infers the covariations between microbial taxa abundances through time and reconstructs the microbiota composition of the hosts’ common ancestors along the tree. We can then estimate the influence of covariates such as diet on the retrieved patterns. This approach may be applied over deep evolutionary time, e.g., to the microbiota of all mammals, or on a shorter timescale, e.g., to the microbiota of humans. We applied it to the gut microbiota of mammals and birds (215 and 323 species covering the entire phylogenetic tree, respectively). We found significant levels of phylosymbiosis in both mammals and birds, which cannot be fully explained by diet conservatism, and we reconstructed a microbiota composition consistent with an insectivorous diet for the ancestor of all mammals. We are now developing an extended model that accommodates microbial composition data at a higher taxonomic resolution by grouping microbial taxa into assemblages, and we are investigating the application of our approach to the human gut microbiota.