Predicting genomic traits in ammonia-oxidizing archaea using phylogenetic signals

Abstract

Phylogenetic conservatism of microbial traits has paved the way for phylogeny-based predictions, allowing us to move from descriptive to predictive functional microbial ecology. Here, we applied phylogenetic eigenvector mapping, an approach not previously used for microorganisms, to predict key traits of ammonia-oxidizing archaea (AOA), which are important players in nitrogen cycling. Using 168 nearly complete AOA genomes and metagenome assembled genomes from public databases, we predicted the distribution of 18 ecologically relevant genes across an updated amoA gene phylogeny, including a novel variant of an ammonia transporter found in this study. Of the selected genes, 94% displayed a significant phylogenetic signal and gene presence was predicted with >88% accuracy, >88% sensitivity, and >80% specificity. The phylogenetic eigenvector approach performed equally well as ancestral state reconstruction of traits. We implemented the predictive models on an amoA sequencing dataset of AOA soil communities and show key ecological predictions, e.g., that AOA communities in nitrogen rich soils have capacity for ureolytic metabolism while those adapted to low pH soils have the high affinity ammonia transporter (amt2). Predicting genomic traits can shed light on the potential functions that microbes perform across earth biomes, further contributing to a better mechanistic understanding of their community assembly.