Year Class Strength Estimation With Mixed Effect Linear Models

Monday, September 9, 2013: 1:00 PM
Harris Brake (The Marriott Little Rock)
David Staples , Minnesota Department of Natural Resources, Forest Lake, MN
Donald Pereira , Division of Fish and Wildlife, Minnesota Department of Natural Resources, Saint Paul, MN
Fishery managers often want to quantify a cohort's year class strength (YCS) from survey data; however, sampling variability or past years without sample data can be problematic.  To address these issues, we demonstrate YCS estimation with fixed and random effect linear models using gillnet survey data from Minnesota walleye populations.  The simplest method is a fixed effect model of age-specific catch rates as a function of age and cohort.  The YCS estimate is the least square mean value for each cohort; i.e., a predicted mean catch for the cohort averaged over all ages in the model.  A random term can be added to represent annual variation in catchability, a concern for passive gears such as trapnets or gillnets that rely on fish movement.  For data series with missing years, a random cohort effect can be used instead of the fixed cohort effect to allow YCS estimation with fewer model degrees of freedom; this also allows estimation of temporal variability in YCS and facilitates more straightforward comparisons of YCS between populations.