Detecting Synchrony of Recruitment using Short, Autocorrelated Time Series

Abstract

Synchrony of recruitment to distinct fish stocks is difficult to detect because the available time series are generally short and autocorrelated. The recent introduction of more sophisticated statistical techniques has not been particularly helpful; several contradictory interpretations of similar data sets are discussed in the paper. To help resolve the continuing controversy surrounding the question of synchrony, we review three statistical tests of independence and determine their power using simulated data. The tests are then applied to recruitment data for six cod (Gadus morhua) and three haddock (Melanogrammus aeglefinus) stocks of the northwest Atlantic. Prior to analysis each series was first-differenced to reduce autocorrelation and hence increase statistical reliability in the results. The cod stocks are shown to fluctuate independently of the haddock stocks. There is, however, evidence of synchrony for stocks of the same species; the more widely separated cod stocks have a lower mean correlation [Formula: see text] than the haddock [Formula: see text] but both correlations are significant at the 1% level. The within-species synchrony is not due to fluctuations in our index of egg production and it appears that environmental forcing is probably important.