Allowing for Time-Varying Productivity in Fisheries Stock-Recruitment Relationships.

Marshall, Rachel C.¹ *, Jeremy S. Collie¹, Richard J. Bell², ¹Graduate School of Oceanography, University of Rhode Island, Narragansett, RI, ²The Nature Conservancy, Narragansett, RI

Productivity, defined in the present study as the number of recruits expected per unit biomass of spawners, is frequently incorporated into fisheries stock-recruitment models as a constant value representing the average over time. Research has shown that for some commercial fish stocks, productivity in reality actually follows a trend with time. Instances where the average productivity is greater than the actual present-year productivity can lead to overestimates of expected recruits, and subsequently, fisheries reference points and catch limits that are too high for the stock. In the present study, a dynamic Ricker stock-recruitment model with a Kalman filter is used to analyze over 80 stocks in five regions around the US and identify stocks that show time-varying trends in productivity. The Kalman filter model accounts for the variability in recruitment by dividing it into observation error (noise), and process error (signal) explained by variation in productivity. To account for challenges in estimating noise and signal variances for individual stocks, some of which may have shorter recruitment and spawner time series than are preferable in the model, process and measurement variances for each stock are estimated assuming a common signal to noise ratio estimated for each region. Preliminary results suggest that several stocks around the US show time-varying trends in productivity, and future research will explore potential drivers of these trends, possible explanations for differences between regions or life-history strategies, and applications of these results for fisheries management.