Prediction of year-class strength by calibration regression analysis of multiple recruit index series

Abstract

The analysis of multiple time series of indices of recruitment to fish stocks by means of calibration regression is discussed, together with the use of the relationships so fitted for the prediction of year class strength. A simple method for the combination of the estimates derived from different index series using inverse variance weighted averages is proposed, and methods for the estimation of the overall error in the prediction are discussed. The method used has been shown to perform well in simulation tests, and is well adapted for use on real datasets with time series of variable length and missing data. It has been implemented in a computer program (RCT3, superseding RCRTINX2) which is available for operational use, and has been endorsed by the ICES Working Group on the Methods of Fish Stock Assessment as being satisfactory for operational use until more complex methods have been shown to have superior performance.