A new approach to the analysis of stock-recruitment relationships: "model-free estimation" using fuzzy logic

In response to the need for overt recognition of uncertainty in management of natural resources, we present a new, innovative, causal approach for ana lysis of stock-recruitment relationships and prediction of recruitment. App lying principles and techniques developed from the theory of fuzzy sets, we demonstrate how heuristic reasoning can be used to define stock-recruitmen t relationships, explicitly characterise vagueness and uncertainty, and pro vide a functional relationship that combines stock size and past recruitmen t to predict future recruitment. The approach is termed model-free estimati on or approximation. Tested on eight stock-recruitment data sets, there was no significant difference between recruitment predicted by the fuzzy appro ximation method and the Ricker or Beverton-Holt recruitment functions. We a ccount for effects of nonstationarity by incorporating rules that relate pa st recruitment to future recruitment in the fuzzy stock-recruitment system. A weighting factor, w, represents the degree of belief in the importance o f past recruitment and stock size in predicting future recruitment. The app roach is robust with respect to the number of fuzzy sets used to define dat a clusters, can be tailored to individual circumstances, and thrives in dat a-poor situations where analytical methods may be inappropriate. It is a si mple and broadly applicable solution with important implications for fish s tock assessment and fisheries management in general.