LAUSR

Biological nitrogen fixation performed by diazotrophs forms a cornerstone of Earth's terrestrial ecosystem productivity. However, the composition, diversity and distribution of soil diazotrophs are poorly understood across different soil ecosystems. Furthermore, the biological potential of the key diazotroph species in relation to key environmental parameters is unknown. To address this, we used meta-analysis approach to merge together 39 independent diazotroph amplicon sequencing (nifH gene) datasets consisting of 1988 independent soil samples. We then employed multiple statistical analyses and machine-learning approaches to compare diazotroph community differences and indicator species between terrestrial ecosystems on a global scale. The distribution, composition and structure of diazotroph communities varied across seven different terrestrial ecosystems, with community composition exhibiting an especially clear effect. The Cyanobacteria were the most abundant taxa in crust ecosystems (accounting for ~45% of diazotrophs), while other terrestrial ecosystems were dominated by Proteobacteria, including Alpha-, Beta- and Gamma-Proteobacteria (accounting for ~70% of diazotrophs). Farmland ecosystems harboured the highest and crust ecosystems the lowest alpha and phylogenetic diversities. Azospirillum zeae, Skermanella aerolata and four Bradyrhizobium species were identified as key indicator species of potential diazotroph activity. Overall, diazotroph abundances and distribution were affected by multiple environmental parameters, including soil pH, nitrogen, organic carbon, C:N ratio and annual mean precipitation and temperature. Together, our findings suggest that based on the relative abundance and diversity of nifH marker gene, diazotrophs have adapted to a range of environmental niches globally.