Testing Predictions of Marine Fish and Shellfish Landings from Environmental Variables

Abstract

Previous studies by Sutcliffe and co-workers using exploratory analysis found correlations between environmental variables and lagged annual catch for several Gulf of St. Lawrence and Gulf of Maine fish and shellfish stocks. The present study tests these relationships using recent data. For 6 of the 13 stocks investigated, correlations between the 9–14 yr of new catch data and that predicted from the environmental models remained high (r > 0.5) and of the same sign; however, individually none was statistically significant (p > 0.05) after accounting for the loss of degrees of freedom due to the high autocorrelation in the data. The hypothesis of an overall environmental effect on the landings was considered. It could not be substantiated as the correlation coefficient for 5 of the 13 stocks reversed sign using the new data, but changes in fishing effort are believed to mask detection of environmentally induced variability in the landings of many stocks. The utility of environmentally based predictions was also tested. Overall, the mean deviations of the predicted catch based on environmental regressions were similar to predictions based on the long-term means but were higher than predictions using lagged catch. Environmentally based predictions of landings for invertebrate stocks were generally more accurate than those for fish stocks.