What can and cannot be predicted from stock and recruitment curves

Understanding the relation between the size and characteristics of spawners and subsequent recruitment is central to fishery biology and management, yet a century of stock–recruit (S-R) curves has not delivered reliable recruitment predictions. I argue that this is because we have concentrated on the wrong feature of recruitment. Recruitment is far more buffered than the underlying randomness of individual encounters, feeding, and mortality would lead us to expect, and it is this buffering, not the residual variance around a fitted curve, that is the phenomenon that is the most important to explain. A statistical S-R relationship estimates the buffer from a stock’s own history as a fixed parameter; it does not explain the buffer’s mechanism and therefore cannot anticipate when the buffer itself changes, which under anthropogenic forcing is exactly what is at stake. I present three components of a mechanistic, trait-based alternative. Firstly reducing complexity through traits, so that asymptotic body size gives a first prediction of both the size and the source of a stock’s recruitment buffer. Secondly, recognizing the stabilizing effect that the recruitment buffer has on at least two mechanistic sources acting at different life stages, with traits deciding which dominates, and finally showing, through the act of first feeding, how stability can emerge from a stochastic process. Related to Johan Hjort’s insight from 1914, I review the mechanistic stochastic theory for first food encounter. This essay is intended as food for thought for seasoned scientists and as an appetizer for early-career researchers to combine new observations with theory towards understanding, and eventually predicting, the source of recruitment buffering/stability.