Alternative growth models in fisheries: Artificial Neural Networks

Abstract

In this study growth of Atherina boyeri, collected from Süreyyabey Dam Lake, was determination by Artificial Neural Networks (ANNs) along with study of length weight relationships (LWRs). A total of 394 individuals including 32.5% female and 67.5% male specimens were studied collected during the fishing season between May 2015 and May 2016 from the local fisherman. The total length and weight of the specimens were 32–90 mm and 0.225–4.062 g respectively. The relationships were W = 0.01285708 L2.67 (R2 = 0.983) for females, W = 0.00678019 L2.95 (R2 = 0.969) for males and W = 0.00641527 L2.87 (R2 = 0.970) for pooled individuals. Mean Absolute Percentage Error (MAPE) of ANNs (0.182) for all specimens was lower than MAPE value of LWR (1.763). The results of study show that ANNs are superior tool to LWRs for fishes of Süreyyabey Dam Lake.

Keywords

• length-weight relationship
• artificial neural networks
• big-scale sand smelt
• Süreyyabey Dam Lake
• Atherina boyeri
• ANNs