Comparing fish biomass models based on biophysical factors in two northern Murray-Darling Basin rivers: a cautionary tale

Abstract

We used multiple linear regression to test the hypothesis that the productivity of two fish species (Nematalosa erebi and Macquaria ambigua) in two Northern basin MDB rivers (Moonie and Warrego) can be predicted from a suite of physical habitat variables. To explore the importance of hydrology and season, these models were developed for two different sampling periods; no-flow and post-flow. For both catchments, simple models were developed using within-waterhole habitat variables and then compared to infer the transferability of the models between catchments. Then, the explanatory power of variables measured at multiple spatial scales was examined in the Warrego catchment. Poor predictive power was revealed in the simple Moonie biomass model for both species. Conversely, high predictive power was detected in the simple Warrego River biomass model, which was considerably improved by including variables measured at multiple spatial scales (within-waterhole, whole-waterhole and landscape scale). Differences between the models suggest that there is no generic model capable of predicting fish biomass in the two catchments. We conclude that the poor transferability of habitat–biomass models between catchments may be due to fundamental differences in their physical characteristics. Furthermore, the improved predictive power of the expanded Warrego model demonstrates the importance of considering factors affecting fish species at all relevant spatial scales. The more powerful Warrego models also differed between sampling periods, no-flow and post-flow, highlighting the role of hydrological variability in determining fish responses to environmental conditions in the MDB.