Predicting fish recruitment from juvenile abundance and environmental indices

Prediction of year-class strength is a critical challenge for fisheries managers. Theoretically, predictions of recruitment should be better when they are based on estimates of cohort size taken close to the age of recruitment and may improve if the effects of environmental factors that influence pre-recruit mortality are accounted for. In practice, measurement error and difficulties in establishing robust recruitment-environment relationships complicate the picture. For 5 fish stocks of 4 species in 3 ecosystems, we examined the usefulness of indices of juvenile abundance relative to larval abundance for predicting recruitment. Further, we examined whether the use of environmental covariates improved predictions. For 2 of 4 stocks with sufficient data (1 stock did not have larval data), juvenile abundance was a better predictor of recruitment compared to larval indices. For the 2 other stocks, we found that juvenile indices were not superior to larval indices, possibly because of error in the measurement of juvenile abundance. In all 5 of these stocks, regression analysis showed that inclusion of environmental correlates contributed significantly to explaining recruitment variation compared to models based on juvenile indices alone. Further, cross validation showed that forecasts of future recruitment were either improved or qualitatively unchanged by including environmental correlates. This was despite apparent nonstationarity in the recruitment-environment relationships; most of the environmental variables and pre-recruit abundance indices were significantly correlated with recruitment for only parts of the studied period. Such complex responses to environmental changes are difficult to anticipate, yet the environmental information should not be ignored altogether.