An Assessment of the Geographic Closure Assumption in Mark–Recapture Abundance Estimates of Anadromous Steelhead Populations

Abstract Closed population models are commonly used to estimate stream salmonid abundances using mark–recapture information collected during electrofishing surveys. To meet the model assumption of geographic closure, block nets are often used to prevent emigration and immigration of fish during the survey. Increased sampling and tagging efforts in an open site may be an appealing trade-off given the time it takes to properly deploy block nets, but it also increases an abundance estimate's vulnerability to bias. We assessed the extent of geographic closure violation from emigration in open sites between mark and recapture passes utilizing PIT antennas as virtual block nets. This allowed us to quantify emigration rates of juvenile steelhead Oncorhynchus mykiss from 60 fish surveys across multiple seasons and watersheds. Our goals were to determine how season and site length influence emigration, examine how the life history of an anadromous salmonid may induce bias on mark–recapture abundance estimates, and provide recommendations on minimizing the bias associated with violating the geographic closure assumption. Average emigration rate was low across all surveys and watersheds (5.1%), with higher emigration rates correlating with larger fish, shorter site length, and the season in which the site was sampled. We concluded that the bias associated with violating the geographic closure assumption in an open site can be minimized by avoiding times of migration and by sampling sites up to 650 m long depending on fish density and capture efficiency. Our findings provide useful information for planning mark–recapture studies that have multiple sampling objectives. Received July 27, 2016; accepted May 24, 2017Published online August 4, 2017