Factors Affecting Variability of Trapnet Catches

Abstract

We summarized the literature relating to sampling characteristics of trapnets, and examined the variability of trapnet catches from Long Point Bay, Lake Erie. In Long Point Bay the frequency distributions of numbers of fish caught per trapnet lift were very skewed, peaking at zero for all species. Standard deviations of raw catches of even the most commonly taken species (Ambloplites rupestris, Perca flavescens, Aplodinotus grunniens) were high, about twice as large as their means. From 50 to 65% of the variability of transformed catches could be attributed to differences in soak time, sampling location, season, and year. Precision of estimating abundances of fish from catch-per-effort data is low for trapnets as well as for other sampling gear. Our fishing effort of nine trapnet lifts per sampling design cell sufficed to estimate, with 95% confidence, geometric mean catches in Long Point Bay to within a factor of 2 of their true values. To estimate them to within 20–25% would have required about 100 lifts. Regarding possible future changes of abundance, it would take 3 yr of observation to establish statistical significance of a decrease of arithmetic mean catches to half the levels observed in 1971–78.