Stochasticity Shapes Microbial Communities in High-Altitude Lakes, Whereas Species Selection and Homogenization Dispersal Are More Important in Lowland Lakes: Case of Benthic Diatoms in Alpine Lakes.

Lakes are fundamental ecosystems for human well-being, and understanding the ecological processes shaping their biological communities is of prime importance. Among microalgae, diatoms are often dominating the biomass of their benthic habitats, especially in European alpine lakes. We hypothesized that, because of their contrasted hydro-morphometric characteristics, high-altitude and lowland lake communities may have contrasted diversities and may be shaped differently by ecological processes. We sampled benthic diatoms in 46 high-altitude and 12 lowland lakes in the framework of several projects between years 2013 and 2019. Each lake was sampled in several places along their shoreline depending on lake size, for a total of 514 samples. Samples were analyzed with DNA metabarcoding and specific diatom primers. We used a naïve taxonomic approach based on amplicon sequence variant and a method based on null modelling and phylogenetic indices, together with variance partitioning and co-occurrence networks. Lowland lakes had higher alpha diversities, and the irdiatom ecological networks were more connected than high-altitude lakes. These distinct structural patterns between lake types were explained by different contributions of ecological processes. Although dispersal was similarly restricted in both lake types, diatom communities in high-altitude lakes were mainly governed by stochastic processes, which was not the case in lowlands. This pattern can be explained by the extreme conditions at high altitudes, which create major bottlenecks in diatom communities and thus favour stochasticity. Conversely, homogenizing dispersion was greater in lowlands, confirming the hypothesis of greater connectivity via anthropogenic and biotic vectors in lowland lakes. These findings suggest that high-altitude lakes could be less resilient to environmental changes due to lower diversity and network connectivity compared with lowlands. Finally, we advocate the combined use of diversity metrics, statistical analyses to assess ecological processes, and inference of networks because they provide complementary information to understand lakes vulnerabilities.