Do rearing salmonids predictably occupy physical microhabitat?

Abstract Microhabitat suitability models are commonly used to estimate salmonid habitat abundance and quality with unknown accuracy or reliability. When tested, the metrics used to evaluate these models are often limited by the methods used to develop them. More generalized bioverification strategies that transcend methodology are therefore needed in ecohydraulics. This study further developed and applied such a generalized bioverification framework to four approximately 1-m-resolution rearing salmonid microhabitat suitability models. Water depth and velocity habitat suitability criteria (HSC) functions were developed for two size classes of rearing Oncorhynchus tshawytscha and O. mykiss using snorkel survey data collected over three years at seven sites along the lower Yuba River in California, USA. An expert-based cover HSC function was modified from previous studies. HSC functions were applied to previously validated, approximately 1-m-resolution two-dimensional hydrodynamic models and cover maps of the river. Mann–Whitney U tests confirmed that suitability values were significantly higher at utilized locations compared to randomly generated, non-utilized locations for all four models. Bootstrapped forage ratios demonstrated that microhabitat suitability models accurately predicted both preferred and avoided habitat beyond the 95% confidence level. This generalized bioverification framework is recommended for evaluating and comparing the accuracy and reliability of ecohydraulic models used in habitat management worldwide.