Growth model selection for juvenile blacklip abalone (Haliotis rubra): assessing statistical and biological validity

Abstract

Accurate estimates of marine organism growth are important for modelling the dynamics of populations and rely on the selection of an appropriate growth model. However, there is no assurance that the statistically optimum model will also be biologically plausible. Three growth models (von Bertalanffy, Gompertz and a linear model) were fitted to a dataset consisting of two cohorts of juvenile size classes of blacklip abalone (Haliotis rubra). Results show that the non-seasonal Gompertz was statistically better than the non-seasonal von Bertalanffy and linear models. There was a persistent seasonal signal through the juvenile size range, with slow growth in winter and fast growth during summer. When a seasonal term was formally incorporated, the model fits were greatly improved, particularly for the linear and von Bertalanffy models. The seasonal-Gompertz predicted growth rates that were biologically implausible for juveniles of 2 mm shell length; 107 μm day–1 for one cohort and 24 μm day–1 for the other. These rates are inconsistent with published growth rates observed under both controlled and wild conditions. In contrast, the seasonal-linear model predicted growth rates of 60 μm day–1 for animals of 2 mm shell length, consistent with published findings. The selection of a growth model based solely on statistical criteria may not take into account the complex processes that influence growth of juveniles.