Computer Analyses of Petersen's Original Data on Bottom Communities

Publisher Website
Google Scholar
Abstract
Petersen—type bottom communities (a concept based on Petersen's 1914 data) are characterized by constant species which are dominant in terms of numbers and weight. Some workers recently have questioned the existence of these communities, in which quantitative data upon very few of the species present have been analyzed subjectively, and other potentially important species have been neglected. Various computer techniques were used to analyze Petersen's data. For the majority of species only presence—absence data can be used; for quantitative data 88 species from 193 stations were processed. The final methods involved transformed data of numbers and weights, respectively, the Canberra metric coefficient, and a combination of flexible and group—average sorting followed by allocation. The presentation of results is difficult because, even after classification, the complexity of the data–for example, the lists of species of different grades of constancy and abundance which characterize site groups–lies beyond human ability to comprehend adequately. Results are compared with Petersen's and with each other. Petersen's classification compares most closely with ours involving binary data. Our "numbers" and "weights" classifications both reveal Petersen—type communities, but they differ markedly from each other and more markedly still from Petersen's results

This publication has 0 references indexed in Scilit: