COS 88-6
Measuring somatic growth variation: A novel method applied to Pacific groundfish

Wednesday, August 13, 2014: 3:20 PM
Beavis, Sheraton Hotel
Christine C. Stawitz, Quantitative Ecology and Resource Management, University of Washington
Timothy E. Essington, School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA
Trevor A. Branch, School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA
Anne B. Hollowed, Alaska Fisheries Science Center, National Marine Fisheries Service, Seattle, WA
Melissa A. Haltuch, Northwest Fisheries Science Center, National Marine Fisheries Service, Seattle, WA
Paul D. Spencer, Alaska Fisheries Science Center, National Marine Fisheries Service, Seattle, WA
Nathan J. Mantua, Southwest Fisheries Science Center, National Marine Fisheries Service, La Jolla, CA

Somatic growth is an important component of production in marine fish populations. Population-level growth patterns in marine fish are known to vary in response to multiple drivers, and shared patterns across stocks may indicate a shared response to environmental conditions. However, analyzing these patterns in commercially-important fish species is difficult due to inconsistencies and bias caused by aging error and issues related to non-random selection inherent in long-term size-at-age time series. An important initial step is to quantify growth in a robust way that can detect changes attributable to cohort-based effects (e.g. via density-dependence or other drivers of first-year growth) or annual-based effects (e.g. changes in food availability or environmental conditions that affect several cohorts simultaneously). In this study, we developed, tested, and applied a novel method of analyzing growth patterns using state-space models and Bayesian estimation methods. Four alternative models to capture annual, cohort, juvenile, and constant growth patterns, respectively, were tested using a simulation study. Next, these models were used to estimate growth trends from survey and commercial fishery catch data collected by the National Marine Fisheries Service across 30 Pacific groundfish stocks from the Bering Sea and Aleutian Islands, Gulf of Alaska, and California Current marine ecosystems.


Our simulation study results suggest our model can robustly detect growth variation on annual and cohort scales. When applied to groundfish data, the majority of stocks in our analysis were shown to experience primarily annual growth effects. The annual growth effect model was given the most support by the DIC for 83% of stocks in our analysis. When the initial growth effect model (7.5% of stocks) or the cohort effect model (5% of stocks) was selected by DIC, DIC weight was more evenly distributed between the two models. Our results suggest the remaining 5% of stocks experience relatively constant growth. When the annual growth effect model was selected, the estimated growth effects showed considerable interannual variability for the majority of stocks, with some synchrony observed between stocks within each of the three ecosystems. Post-hoc comparison of these annual trends with fishing pressure and changing patterns of gear selectivity suggest that estimated growth effects reflect actual population changes in size-at-age rather than the influence of confounding factors. We conclude that state space models are a robust method for estimating growth variation patterns from error-prone fishery data, and analyzed groundfish stocks experience primarily annual population-level growth variation patterns.