Sampling design and statistical considerations for the commercial groundfish fishery of Oregon

Abstract

Fisheries management is often based on data collected through various sample survey programs. At a minimum, commercial fisheries must be monitored to determine the species and age compositions of the landings; this provides the baseline data on which to assess the stocks. An equally important objective, which is often ignored, is the determination of the variability associated with derived estimates. This paper presents measures of dispersion for landing estimates generated from a two-stage sampling design used for commercially harvested groundfish species landed at Oregon ports; particular attention is given to the relative magnitudes of variability at the first and second stages of sampling designs used for estimating species and age compositions. At least two-thirds of the total amount of rockfish landed in each port/quarter stratum consisted of estimates of species composition that had small coefficients of variation (CV < 10%). For each species sampled for age composition, at least three-fourths of the total landings included estimates for individual ages with CVs < 25%. For the majority of the landings, the variation at the first stage of sampling contributed at least 63 and 90% to the variance of the estimates for the species and age compositions, respectively.