Evaluation of the effect of morphological traits on fish growth by comparison using ridge and ordinary least squares regression

Mehmet Fatih CAN,Cemil KARA
DOI: https://doi.org/10.61326/jofbs.v4i1.03
2024-06-29
Journal of Biometry Studies
Abstract:Growth in fish is characterized by length and weight and studies that encompassing relationships among the fish length and weigh with some morphometric traits provide crucial information in the field of fish biology. High correlations (or multicollinearity) among the morphometric traits in fish morphology studies is a well-known phenomenon. If the relationship between growth and morphometry is modeled using the ordinary least squares estimator (OLS), the parameter estimates are likely to be too large in absolute value and possibly have the wrong sign due to the problem of multicollinearity. The ridge regression (RR) estimator has been proposed to avoid the adverse effects of multicollinearity among the regressors. In this study, therefore, multiple linear RR was used to model relationship fish length and fish weight to some predictive metric traits. Predictive traits were predorsal length, head length, post dorsal length, head height, and eye diameter. Data were derived from a total of 126 Capoeta damascina individuals sampled from the Euphrates, Türkiye. The ridge optimal k-parameter, producing acceptable variance influence factor (VIF<10), was determined as 0.021 for both of growth indices considering a combination of ridge trace and VIF trace plots and MSE and VIF values produced from an array of k values (0.0≤ k≤1.00). Although, small decreases were observed in the adjusted R-square values (ARS) obtained by RR compared to the ARS values obtained by OLS, a significant decrease in VIF values outweighs this drawback, indicating that the models are more stable. The methodological approach and findings in this study may contribute to filling a gap in the literature regarding the relationships between fish growth and morphology. Additionally, it could enhance better growth predictions for different fish species, aiding sustainable fisheries management and the selective cultivation of desired fish traits in aquaculture by improving the understanding of morphometric features.
What problem does this paper attempt to address?