BioScience Trends. 2008;2(3):128-132.

Discriminant analysis: A supportive tool for monogenoidean taxonomy.

Agrawal N, Agarwal GG, Tripathi P, Pant R

Data on the measurements of haptoral parts of 47 parasite species of Bychowskyella (1), Cornudiscoides (2) and Thaparocleidus (3), found on Indian freshwater catfish, were gathered from previously reported studies. Based on six morphometric variables concerning haptors, these species were subjected to discriminant analysis in order to more accurately distinguish their generic placement. This paper describes a comparative study of several classification models. Using the original variables in the study, a simple linear discriminant analysis model was constructed and models using principal components (PC) for discrimination have also been explored. The effectiveness of these models is measured in terms of percentage of correct classification. Quadratic discriminant models using original variables and their principal components provided the highest (95%) correct classification for Bychowskyella (1). Species of Thaparocleidus (3) were correctly classified using dorsal and ventral hard-part measurements (100%) as well as hook measurements. The highest percentage (75%) of correct classification for Cornudiscoides (2) was achieved through a quadratic discriminant model using hook measurements.

Full Text: