Progress in modelling herring populations: an individual-based model of growth

tock assessment may gain from taking into account individual variations in growth, because size is a key predictor of survival and reproduction. In trying to understand patterns in empirical observations, a major challenge is to model the changes in the size dis- tribution of a cohort with age. We introduce an individual-based growth model that is founded on the use of a stochastic class of processes called subordinators. This modelling approach has several desirable features, because it (i) can take account of both indi- vidual and environmental sources of random variations, (ii) has the property of letting size increase monotonically, and (iii) ensures that the mean size-at-age follows the widely accepted von Bertalanffy equation. The parameterization of the model is tested on two Atlantic herring (Clupea harengus) datasets collected from the stocks of North Sea autumn spawners (ICES Divisions IVa, IVb, and IVc) and western Baltic spring spawners (ICES Subarea III). The size distributions obtained from the subordinator model largely match the observed size distributions, suggesting that this approach might be successfully implemented to support the assessment of commercial fish stocks, such as when modelling of size variability is required.