Global forage fish recruitment dynamics: A comparison of methods, time-variation, and reverse causality

Abstract

The population dynamics of forage fish are often ‘boom or bust’, and variation in recruitment may be a contributing factor to changes in abundance. Here we applied several methods for identifying stock recruit relationships (SRR) to 52 forage fish stocks: a time-invariant Ricker model and two time-varying methods (dynamic linear models and regime-based models). A positive relationship between spawning biomass and recruitment existed for 28 stocks, with 14 of those being best fit by a Ricker curve. Time-varying models were preferred for 26 stocks over a time-invariant Ricker curve. We also showed through simulation that spurious, yet significant stock-recruit relationships can be generated if recruitment was environmentally driven, the lag between recruitment and spawning biomass was small, and autocorrelation in recruitment was high—a combination of characteristics that many forage fish stocks display. Based on our results, observed time-invariant SRRs can result from a reversed causality of ‘boom and bust’ dynamics in situations in which life histories are short (i.e., changes in recruitment are driving changes in spawning biomass). ‘Hidden’ SRRs are also possible because of time-variation in SRR parameters. Understanding when each of these potential failings of fitting SRRs can occur is a key challenge in forage fish management.