Evaluation of Three Methods for Estimating Numbers of Steelhead Smolts Emigrating from Great Lakes Tributaries

Abstract

We compared three techniques for estimating the number of steelhead Oncorhynchus mykiss smolts emigrating from the Betsie River in northwestern Michigan and evaluated the effect of sampling effort on estimates of annual production of smolts. In May and June 1993–1996, we used visual observations, time-lapse videography, and mark–recapture to estimate the number of emigrating hatchery and wild steelhead smolts. Based on the proportions of each species captured by trapping, hatchery and wild steelhead smolts composed 30–61% of all emigrating juvenile salmonids, which also included brown trout Salmo trutta, coho salmon Oncorhynchus kisutch, and chinook salmon Oncorhynchus tshawytscha. Annually, wild steelhead composed 12–52% of all steelhead smolts. Mark–recapture estimates and 95% confidence intervals (CIs) for steelhead smolts ranged from 13,836 (13,219–14,495) in 1995 to 56,660 (50,721–63,834) in 1993. Mark–recapture estimates yielded results that were up to nine times greater than those from observational methods. Estimates of smolt abundance (±2 SEs) from visual observation and videography were similar to one another and ranged from 3,097 (±690) in 1996 to 11,118 (±3,882) in 1994. A random sampling protocol with different levels of effort for nightly sampling demonstrated a trend for estimates that exceeded the confidence intervals when less than 75% of the nights were sampled. With additional refinements, surface view videography is a promising technique for use in monitoring emigrating smolts. We recommend that assumptions on marked-fish behavior be confirmed to avoid overestimating fish abundance in mark–recapture applications, such as for trapping efficiency or estimates of smolt abundance. When considering a study design for estimating smolts, the desired levels of precision for the data should be considered in determining the costs and benefits of sampling with high intensity.