THE INFLUENCE OF VARIABLE HABITAT SUITABILITY CRITERIA ON PHABSIM HABITAT INDEX RESULTS

The Physical Habitat Simulation System (PHABSIM) still probably remains as the most widespread habitat method used to establish inflow standards or to link habitat temporal variations with fish population dynamics. However, statistical uncertainties around the PHABSIM main output, the weighted usable area (WUA) over discharge curves, are usually ignored. Here, we assess the uncertainty in WUA curves and derived habitat duration curves induced by the variability around the PHABSIM biological model, the habitat suitability criteria, using brown trout Salmo trutta as the model species. Bootstrap analyses showed that the uncertainty around the WUA curves was rather high when bootstrap sample (BS) size was low and differed among age classes, being generally lower for young-of-the-year (YOY). Width of 95% confidence intervals for maximum WUA magnitude increased with decreasing BS size, ranging from 19.3% for YOY trout at the largest BS size (40 transects, 270 habitat use observations) to 146% for juveniles at the smallest BS size (nine transects, 60 habitat use observations). The uncertainty arose primarily from the construction of the channel index variable. Nevertheless, results showed that the uncertainty in WUA values could be reduced down to acceptable levels by using general functional channel index categories. Likewise, the shape of WUA curves was also highly variable when BS was small. These patterns resulted in habitat duration curves being highly uncertain, much more in their amplitude than in their shape. Uncertainty about the flows corresponding to different habitat exceedance values increased with decreasing probability of exceedance. Width of peak flow confidence intervals ranged from 3.3% for YOY trout at the largest BS size to 226% for adults at the smallest BS size. Yet such levels of uncertainty do not necessarily entail critical errors in the decision-making process because large variability in flow peak does not necessarily lead to large variability in WUA magnitude.