Modelling fish growth and comparison of statistical approaches in model selection in the Beibu Gulf, South China Sea

The performance of seven statistical approaches was analyzed using 61 sets of length-at-age data. The seven approaches include coefficient of determination (R²), adjusted coefficient of determination (adj.-R²), root mean squared error (RMSE), Akaike’s information criterion (AIC), bias correction of AIC (AICc) ,AIC differences ( ) and Bayesian information criterion (BIC). The 61 sets of length-at-age data were adopted from specimens collected from fishery-dependent bottom trawl and gillnet fleets caught in Beibu Gulf of South China Sea from July 2006 to December 2009, which were aged by otoliths, scales or vertebrae. Five-candidate models were fitted to each dataset: von Bertalanffy growth model (VBGM), generalized VBGM, Gompertz growth model, Schnute–Richards growth model, and logistic growth model. The best supported model by the data was identified using each of the seven approaches. The results showed that,  or BIC is recommended for selection of fish growth model. The VBGM was not selected as the best model in the cases, only accounted for 24.6% of the total, while generalized VBGM accounted for 26.2%, Gompertz growth model accounted for 18.0%, Schnute–Richards growth model accounted for 9.8% and logistic growth model accounting for 21.3%.