Inferring riverscape dispersal processes from fish biodiversity patterns

Dispersal patterns are recognized as determinants of biodiversity structure, particularly in rivers, where dendritic organization, waterflow direction, large distance immigrants from the outlet and fragmentation by dams combine to produce a complex dispersal scenario. Unravelling the role, magnitude and spatial scale at which these dispersal sources determine metacommunity diversity is challenging and requires a large amount of spatiotemporal information, which is rarely available. Here, we incorporate alternative dispersal hypotheses into metacommunity models, contrasting their predictions with the observed pattern of fish diversity (58 sampled sites) in the Negro River basin of Uruguay. Evidence supports: (i) a dispersal constrained by the river network, sharply decaying in upstream but not in downstream river directions; (ii) an outlet as a source of individuals that affects diversity even at distant communities; and (iii) a nonconclusive effect of dams, in which models with or without dam barriers are similarly supported. Observed alpha and beta diversity were well predicted by the metacommunity model (r = 0.55 and r = 0.56, respectively). Variation in diversity among simulations systematically decreased from headwaters to the outlet, evidencing a poorly recognized change in processes stochasticity along the landscape. Even without considering the well-recognized role of local filters in the assembly of the fish community, dispersal mechanisms were able to explain riverscape diversity. Dispersal patterns are made of several dispersal sources operating at different spatial scales, which are more complex than the arrival of individuals from species pool or than dispersal exchanges between neighbouring communities only. The non-conclusive effect of dams might stem from the long time lag of biotic relaxation following river fragmentation. Massive fragmentation of rivers challenges the preservation of their diversity and functioning due to disruptions in the different dispersal processes. However, demonstrating the actual and potential effect of dispersal disruption is limited by available information and the long time lags involved in faunal relaxation. Combining empirical information with the modelling of hypotheses emerges as a compelling approach for unravelling metacommunity mechanisms. Dispersal is here evidenced as a complex multi-scale phenomenon, a point that might be considered in theoretical and empirical studies and in ecosystem management.