CONICET | Buscador de Institutos y Recursos Humanos

A total of 46 samples from coastal and shelf sectors from the northern Argentine Sea (34° -41° S) were digitized to compare the performance of the new ZooImage analysis method for copepod abundance and biovolume estimations. A training set of 1437 objects were used for the automatic discrimination using Random Forest algorithm with a general accuracy of 83.92%. A total of 11 taxa were automatically classified. Copepods were divided in three categories: Large calanoids, Small calanoids and Cyclopoids and identified with an accuracy of 83.15%, 79.5% and 85.7% respectively.The discriminant analysis revealed both the equivalent circular diameter (ECD) and the area were the best variables to differentiate the three copepod categories. Samples were previously quantified by optical methods in order to compare with automatic results. Automatic copepod biovolume measurements were estimated from individual calculations applying new ZooImage allometric parameters, and were compared with manual calculations using specific size/biovolume equations. It was demonstrated that ZooImage can potentially be used as a tool for abundance and biovolume estimations of calanoid and cyclopoid copepods and allow us to obtain results more rapidly reducing the time lag involved using the traditional method.