From classical to nonparametric growth models: Towards comprehensive modelling of mussel growth patterns.

Understanding biological processes, such as growth, is crucial to development management and sustainability plans for bivalve populations. Von Bertalanffy and Gompertz models have been commonly used to fit bivalve growth. These models assume that individual growth is only determined by size, overlooking the effects of environmental and intrinsic conditions on growth patterns. The comparison between classical models and nonparametric GAM (generalized additive models) fits conducted in this work shows that the latter provide a more realistic approach of mussel growth measured in terms of shell length, and dry weight of hard and soft tissues. GAM fits detected a reduction in growth during the cold season, under unfavourable nutritional conditions. These fits also captured the decoupling between hard and soft tissue growth, widely addressed in the literature but not incorporated in growth models. In addition a GAM fit of condition index allowed us to explain annual changes in resources allocation, identifying the asymptotic growth of shell and the effects of the reproductive cycle on soft tissue fluctuations.
(Copyright © 2017 Elsevier Ltd. All rights reserved.)