Ability of phabsim to predict chinook salmon spawning habitat

Abstract

AbstractPHABSIM, part of the Instream Flow Incremental Methodology, was used to predict the spawning habitat used by chinook salmon in a 600 m long section of the Nechako River, British Columbia, Canada. Predictions of the model were compared to the location and amount of habitat actually used by adult chinook salmon in 1974, 1980, and 1986. About 3800 m2 (70 per cent) of the spawning area actually used by the population were predicted as unusable by the ‘best’ prediction, while 87 per cent of the area predicted as usable has never had recorded use. The ‘best’ prediction resulted from using close transect spacing, frequent measurements along the transect, river‐specific habitat suitability criteria, and modelling habitat at the fish's position near the stream bottom. Depending upon the spacing of the transects and the habitat suitability criteria used, PHABSIM predicted 210 per cent to 660 per cent more spawning habitat was available than historically had ever been used. Chinook salmon in the Nechako River spawn mainly on the upstream face of dunes, therefore, the assumption in PHABSIM that conditions predicted at the transects remain unchanged upstream and downstream part way to the adjacent transects was false. This assumption resulted in about two‐thirds of the correct predictions being made for the wrong reason. The accuracy of PHABSIM's predictions for spawning might be improved by incorporating an index of river bottom topography or velocity gradient into the model.