Morphological differentiation of two cichlid species in Sri Lanka using truss networks

Abstract

Morphometric characterization using truss network systems has been widely used for fish identification and stock identification. In Sri Lanka, differentiation of Oreochromis mossambicus and O. niloticus, the two major species in reservoir fishery, using conventional morphometric characteristic and phenotypic variations is problematic, due to the reasons that both species share somewhat similar body shape characteristics and that the populations of the two species have mixed characteristics as a result of introgressive hybridization. In the present study, an attempt was made to employ truss network analysis to differentiate O. mossambicus and O. niloticus. Truss measurements of 100 specimens of O. mossambicus from Negombo lagoon and 100 specimens of O. niloticus from Baira lake were analyzed using principal component analysis (PCA). Before the analysis, two approaches of data transformation were employed to minimize to eliminate any size effect in the data set of truss measurements. It was concluded that the data standardization method using a geometric mean regression techniques on the logarithms of truss length and total length was more effective for morphometric differentiation of the two cichlid species than using ratio of truss length to standard length for data transformation.