Time Series Analysis and the Forecasting of Menhaden Catch and CPUE

Catch and catch per unit of effort (CPUE) data for Atlantic menhaden (Brevoortia tyrannus) and the gulf menhaden (Brevoortia patronus) of the Gulf of Mexico were analyzed with autocorrelation analysis to test for time lags and to develop forecasting equations. Analysis of data for two closely related species and for CPUE, as well as catch, provides an opportunity to evaluate time-series methods. The results for the two species and for catch and CPUE data were considerably different. A 6-year time lag was found for both the Atlantic menhaden catch and the gulf catch but the correlation was positive in the Atlantic and negative in the Gulf. A 3-year time lag, corresponding to the age at maturity, was found for the Atlantic menhaden CPUE. No significant time lags were found for CPUE in the Gulf of Mexico. Environmental factors such as changes in sea temperature could have caused the observed delayed density dependence but the time lags also could have arisen from a reproductive delay. The autocorrelation structure of the catch and CPUE data were applied to develop autoregressive models for year-ahead forecasts and confidence intervals. The coefficients of determination for the catch series were high for both the gulf menhaden and the Atlantic menhaden but the confidence intervals for year-ahead predictions were wide.