Habitat selection drives diatom community assembly and network complexity in sediment-laden riverine environments.

ABSTRACT

Microbial communities assemble stochastically and deterministically, but how different assembly processes shape diatom community structure across riverine habitats is unclear, especially in sediment-laden environments. In this study, we deciphered the mechanisms of riverine diatom community assembly in the water column and riverbed substrate with varying sediment concentrations. Water and sediment samples were collected from 44 sampling sites along the Yellow River mainstream during two seasons. Diatom communities were characterized based on high-throughput sequencing of the 18S ribosomal RNA genes coupled with multivariate statistical analyses. A total of 198 diatom species were taxonomically assigned, including 182 free-living and particle-attached species and 184 surface-sediment species. Planktonic communities were structurally different from benthic communities, with Cyclotella being dominant mainly in the middle and lower reaches of the river with higher sediment concentrations. Both stochastic and deterministic processes affected diatom community assembly in different habitats. Species dispersal was more important in the water than in the substrate, and this process was strengthened by increased sediment concentration across habitats. Diatom communities exhibited lower network complexity and enhanced antagonistic or competitive interactions between species in response to higher sediment concentrations compared with lower sediment concentrations mainly in the source region of the river. Differences in the species composition and community diversity of planktonic diatoms were closely correlated with the proportion of bare land area, nitrogen nutrients, precipitation, and sediment concentration. In particular, particle-attached diatoms responded sensitively to environmental factors. These findings provide strong evidence for sediment-mediated assembly and interactions of riverine diatom communities.