Prediction of Morphometric Characteristics of Pomadasys Maculatus from the Karachi Coast, Pakistan, Using Principal Component Analysis (PCA)

Abstract

Among the fisheries resources of Pakistan, Pomadasys maculatus is an important species due to its economic potential and export value. This study aimed to identify a new set of variables for P. maculatus using multivariate statistical methods. Morphometric data of male and female fish were collected from the Korangi Creek area, located at 24°47?N latitude and 67°05?E longitude along the Karachi coast, from January 2013 to December 2014. During the study period, a total of 1,056 specimens, including 503 males and 553 females, were investigated. The dataset included important morphometric variables such as total length (TL, mm), standard length (SL, mm), body weight (Bwt, g), fork length (FL, mm), and other related measurements. Principal Component Analysis (PCA) was applied as a statistical tool to reduce the dimensionality of the data. The first three components, PC1, PC2, and PC3, accounted for 90% of the total variation. PC1 explained 83.96% of the variance and was mainly influenced by TL, Bwt, and SL, representing the overall size of the fish. PC2 explained 3.58% of the variance and was associated with strong swimming muscles, particularly caudal peduncle length (CPL). PC3 contributed 2.5% of the variance and was linked to eye diameter (ED). By applying PCA, 90% of the data variance was preserved while reducing dimensionality. Regression models using factor scores as independent variables were developed to predict future fish characteristics. These models were validated using root mean square error (RMSE) and mean absolute percentage error (MAPE), which showed minimum error values. This study highlights the effectiveness of PCA, and factor score projections in predicting the morphometric characteristics of P. maculatus along the Karachi coast.