METODOS NUMERICOS PARA LA CLASIFICACION DE POLEN ACTUAL Y FOSIL

Edgardo Juan Romero, Ricardo Maronna

Abstract


A method of classification, based on Multivariate Analisis is presented. It brought good results in living and fossil Nothofagus pollen, and can be used in other systematic problems with dubious taxonomic boundaries. The possibility of numerical identification in living species was previously checked, through a discrimination procedure. The methods separates the data in an arbitrary number of groups, through an iterative process which assigns them to the previous round baricenter. The best partition is evaluated with a specially developed index (Maronna 1971). Results are discussed. Further incidental applications are quoted, in order to improve other classifications and to confront living and fossil taxa. Rigurous mathematical explanations are given in a final appendix.

Full Text:

 Subscribers Only