Operationalizing the Riverscape Concept Using Bayesian Hierarchical Models: A Fish Case Study

The riverscape concept has gained popularity as a useful framework to understand how local and regional environments affect the distribution and abundance of stream organisms, yet there has been limited progress in advancing this concept from theoretical abstraction to quantitative realization.  Recent developments in quantitative ecology provide an opportunity to further explore the utility of the riverscape concept in freshwater ecology.  In this study, we used the riverscape perspective to guide the development of a Bayesian hierarchical model that relates multi-scaled environmental variables to the distribution and abundance of a single fish species.  The model simultaneously quantified the hierarchy of environmental determinants of species’ spatial distribution and spatiotemporal variation in abundance as well as the extent of longitudinal connectivity among sampling locations.  We illustrate the model with a small-bodied species, the Empire gudgeon (Hypseleotris galii), in the Mary and Albert Rivers, Queensland, Australia.  Sampling occurred at 17 and 11 relatively undisturbed locations in the Mary and Albert Rivers respectively.  Local-scale environmental variables used in the model included hydraulics, instream structural habitat and recent antecedent flow variables, and landscape-scale variables included descriptors of catchment position and long-term hydrologic variation.  Predictive performance of the model was high, accounting for 60% of variation in the distribution and abundance of Empire gudgeon.  The species displayed a distinct spatial distribution within each catchment with a higher probability of presence in mid-elevation reaches than lowland or headwater streams.  However, being a highly mobile species, there was considerable temporal variation in the species’ distribution and abundance that could be explained by local-scale environmental variation including depth and stream gradient.  The findings of this study suggest that Bayesian hierarchical models offer a way to operationalize the riverscape concept into quantitative models of species distribution and abundance.  The development of such quantitative ecological models can provide an opportunity to further advance riverscape ecology.