Are diatom community assembly processes scale invariant in streams?

Biological communities are structured by deterministic and stochastic factors often regarded to as community assembly processes. Deterministic factors comprise environmental filtering and biotic interactions while stochastic factors include random dispersal and ecological drift. Earlier research suggests that the relative importance of assembly processes may vary with scale. This is because biotic interactions typically occur at fine spatial scales, whereas heterogeneous environment filters species more efficiently at larger scales. The aim of this study was to investigate whether stream diatom community assembly processes varied among three spatial scales, ranging from within-site scales to across drainage-system scales in Finland. The smallest spatial scale was a single riffle where a local community was defined as diatoms living on a single stone. The next larger spatial scale was a drainage system while the largest spatial scale comprised the pooled data from the eight drainage systems. To test whether the diatom communities were stochastically or deterministically assembled, the null model approach based on Raup-Crick dissimilarities (beta RC) was used. Redundancy analyses were also run to visualise diatom community variation and to investigate the main environmental variables associated with community variation. At the within-riffle scale, mean and median beta RC values were approaching -1, suggesting that environmental filtering shaped the communities. At the within-drainage system scale, mean and median beta RC values were also negative, but notably larger than within riffles. Across drainages, beta RC values were at similar level to the within-drainage system scale. At the riffle scale, water depth and light had the strongest influence on diatom community variation followed by current velocity along the stream. At the drainage scale, conductivity, pH, latitude, total phosphorus, and water colour had the largest influence on diatom community variation. Overall, negative Raup-Crick dissimilarities indicated that environmental filtering dominated diatom community assembly at all scales but especially at small, within-riffle scales. There was no consistent support for limiting similarity even at the smallest scales and the effect of stochastic, dispersal-related factors became stronger at larger scales. These results support the idea that chemical conditions in streams are decisive factors for freshwater diatom biodiversity especially at regional scales.