Methodological Bias in Estimates of Strain Composition and Straying of Hatchery‐Produced Steelhead in Lake Michigan Tributaries

Abstract

Steelhead Oncorhynchus mykiss were first introduced into the Great Lakes in the late 1800s. Subsequently, natural recruitment of steelhead from spawning runs in streams across the basin has been regularly supplemented by hatchery production of strains derived from widely dispersed locales within the species’ native range. Estimates of hatchery contributions to the spawning runs of naturalized populations may be underrepresented by observations of marked fish, as not all hatchery fish are marked prior to release. To assess the potential bias in estimates of the hatchery contribution to steelhead spawning runs in four major rivers in Michigan, we used scale pattern analysis (SPA) to identify nonmarked hatchery fish and multilocus genotypes to estimate the proportional contributions of each hatchery strain to spawning runs. The four hatchery strains currently stocked are significantly genetically distinct (mean FST = 0.077), making it possible to identify specific strains by use of likelihood‐based assignment tests. The differences between direct (mark observations) and indirect (SPA and genetic analysis) estimates of hatchery contribution were mainly due to variations in the percentage of hatchery fish marked by states prior to release and the potential for confusion between certain marks and injuries. By combining direct and indirect assessment methodologies, we estimated that the percentage of hatchery fish returning to the four rivers ranged from 13% to 31% of total spawning runs. The large contribution of hatchery fish to nonstocked rivers differed significantly from expectations of strain‐specific stocking rates across the Lake Michigan basin and for individual streams, indicating high amounts of straying into Michigan streams.