Empirical modelling of interannual trends in abundance of squid (Loligo forbesi) in Scottish waters

This study compares alternative empirical models for forecasting interannual variation in squid (Loligo spp.) abundance around Scotland (UK). Annual abundance indices (1970-1996) were derived from fishery data, for stocks in coastal waters (northern North Sea and West Coast) and the Rockall area. Abundance in coastal waters was correlated. with several annual environmental indices, including the winter North Atlantic Oscillation index and the average sea surface temperature (SST) and sea surface salinity (SSS) in the northern North Sea. Regression models, relating fishery abundance to annual environmental indices, and time-series models, predicting abundance from the previous year's abundance, could both provide reasonable fits to observed abundance data. Model fits obtained for coastal waters were generally better than those for the Rockall area. Regression models produced consistently poor forecasts whereas time-series models produced satisfactory forecasts for the years 1997-1999. Forecast accuracy for the time-series models could be improved by taking into account environmental variables. Thus, adding effects of SST in May in the northern North Sea in the current and previous years improved a time-series model of squid abundance in coastal waters. Improved forecasts were also achieved by fitting time-series models to monthly abundance data. The apparent influence of May SST in the northern North Sea on the coastal fishery could indicate that SST affects recruitment strength. Results for the offshore Rockall fishery suggest that large-scale and longer-term oceanographic processes may be involved. (C) 2002 Elsevier Science B.V. All rights reserved.