Patterns of Spawning Habitat Selection and Suitability for Two Populations of Spring Chinook Salmon, with an Evaluation of Generic versus Site-Specific Suitability Criteria

Abstract

We evaluated patterns of redd site selection in relation to physical habitat variables (depth, velocity, and gravel size) using logistic regression and developed spawning habitat suitability models for two populations of spring Chinook salmon Oncorhynchus tshawytscha in Idaho. Additionally, we evaluated the validity of published, generic spawning habitat suitability criteria relative to our stream-specific models. In Elk Creek, fish used sites with coarse gravel sizes and shallow water depths; there was no difference in velocity between used and unused sites in this stream. Salmon spawning in Sulphur Creek used sites with deeper and faster water irrespective of gravel size. Spawning habitat suitability in Elk Creek was best modeled as a quadratic function of gravel size; in Sulphur Creek, it was best modeled as a positive function of depth alone. The best model fit to a pooled data set was a hybrid of the two single-stream models. In all cases, models containing all of the habitat variables performed worse than the simplest models. Generic criteria predicted site use poorly in Elk Creek but exceptionally well in Sulphur Creek: in Elk Creek nearly half of all sampled sites were misclassified, while in Sulphur Creek nearly 75% were correctly classified. As river-specific models are likely to outperform generic criteria in most cases, we recommend that researchers carefully weigh the costs of parameterizing such models relative to the costs of the potential misclassification errors from using preexisting generic models.