Stock structure analysis of the endemic fish, Barbodes carnaticus (Jerdon 1849), for conservation in a biodiversity hotspot

The population structure of Barbodes carnaticus species was studied using conventional (based on body morphometrics and meristic) and image-based analysis (truss network system) methods. The study was carried out with four stocks, namely Karnataka (KA) and Tamil Nadu (TN) stocks from the River Cauvery, Kerala (KE) stock from the River Chalakudy and farm-reared stock (CI) from Central Institute of Freshwater Aquaculture, Bangalore. A total of 27 morphometric, 9 meristic and 30 truss measurements were used in the study for the stock structure. Fifteen landmarks were used to generate 30 truss distance measurements. The principal component analysis (PCA), factor analysis (FA), discriminant function analysis (DFA) and cluster analysis (CA) were deployed to determine the variation using both the conventional and truss variables. Variations (86.9%) among the morphometric characters were explained by five principal components, while four principal components explain 96.01% of the variation among the truss distances. DFA using conventional method correctly classified 100% of the original grouped classes of the KA, KE and CI and 93.8% of TN stocks. The DFA employed with truss distance was classified into the stocks CI, KA, KE and TN, and the values are 100, 89.1, 8.6 and 6.1%, respectively. Factor analysis based on truss morphometry showed that factor one is related to body shape and factor two is related to head shape. Two clusters were identified in both the conventional and the truss distance analysis. Truss distance-based cluster showed that the KE and CI stocks are similar compared to the TN stock. In contrary, morphometry-based cluster showed the KE and TN stocks are similar compared to CI stock. The multivariate analysis showed that the farm-reared stock (CI) is different from the wild stocks (KA, KE and TN). This study explained that the combination of the conventional and image-based truss network analysis helps to discriminate various stocks of B. carnaticus. Based on the PCA, bilinear data models were generated using R 3.5.3 software for predicting the stock of each individual. Stock discrimination of this species was mainly due to the geographic isolation, river ecology and temperature variations. The stocks of B. carnaticus are highly exploited from the studied rivers, and the species is an important candidate for species diversification to enhance aquaculture production. Within stock variations are found to be minimum in the present morphometric study, hence the gene pool identification and marker study are required for better understanding of the stocks. This stock structure study may help to develop conservation programmes for this endemic species through a more scientific approach.