Interannual and Spatial Variability in Maturity of Walleye Pollock Gadus chalcogrammus and Implications for Spawning Stock Biomass Estimates in the Gulf of Alaska

Abstract

Catch quotas for walleye pollock Gadus chalcogrammus, the dominant species in the groundfish fishery off Alaska, are set by applying harvest control rules to annual estimates of spawning stock biomass (SSB) from age-structured stock assessments. Adult walleye pollock abundance and maturity status have been monitored in early spring in Shelikof Strait in the Gulf of Alaska for almost three decades. The sampling strategy for maturity status is largely characterized as targeted, albeit opportunistic, sampling of trawl tows made during hydroacoustic surveys. Trawl sampling during pre-spawning biomass surveys, which do not adequately account for spatial patterns in the distribution of immature and mature fish, can bias estimated maturity ogives from which SSB is calculated. Utilizing these maturity data, we developed mixed-effects generalized additive models to examine spatial and temporal patterns in walleye pollock maturity and the influence of these patterns on estimates of SSB. Current stock assessment practice is to estimate SSB as the product of annual estimates of numbers at age, weight at age, and mean maturity at age for 1983-present. In practice, we found this strategy to be conservative for a time period from 2003–2013 as, on average, it underestimates SSB by a 4.7 to 11.9% difference when compared to our estimates of SSB that account for spatial structure or both temporal and spatial structure. Inclusion of spatially explicit information for walleye pollock maturity has implications for understanding stock reproductive biology and thus the setting of sustainable harvest rates used to manage this valuable fishery.

Effects of climate and demography on reproductive phenology of a harvested marine fish population

“…We used only data sampled from March 12 to April 1 to ensure consistency among years. Note that Williams et al () model maturity, which includes fish in pre‐spawning state, thus the response variable differs from this study.…”

“…Historically, the pre‐spawning survey for pollock in Shelikof Strait (late March) has collected information on gonad maturity of sampled fish (see Williams, Kruse, & Dorn, and Stienessen et al, ). These data cannot be used to reconstruct the distribution of spawning times because sampling occurs over a short‐time window (6–10 days) and macroscopic maturity classification is imprecise; however, a rough indication of relative spawn timing can be estimated from the proportion of fish that are in spawning or spent stages at the time of the survey.…”

“…Only fish age 3 and older were included. Following Williams et al, , we fit a GAM with a binomial response SP (0 = classified as immature, developing, or pre‐spawning, 1 = classified as spawning or spent) of the following form:…”

“…We use analysis of these data as an external validation of our approach, expecting that in years with early spawn timing, a higher proportion of fish would have been in spawning or spent stages during the late March survey. Rather than use raw proportions of fish in spawning and spent stages, we corrected for sampling day of year effects (later sampling is more likely to encounter spawning or spent fish), and geographic effects (e.g., due to spatial variation in the maturity status of fish; Williams et al, ). Only fish age 3 and older were included.…”

“…Only fish age 3 and older were included. Following Williams et al, 2016, we fit a GAM with a binomial response SP (0 = classified as immature, developing, or pre-spawning, 1 = classified as spawning or spent) of the following form: We used a degree-day calculation to estimate temperature-dependent time to hatch of pollock eggs based on results from three laboratory studies. The best model indicated that it takes 97.9…”