Use of numerical frequency and binary data in classification of marine fouling communities

Abstract

A census was taken of marine fouling communities from three study sites by using four different methods. The resulting data types (numerical with 100% subsampling; numerical with 56% subsampling; frequency; and binary) were classified and the information derived from each classification was compared. The results suggest that similar information is produced from classifications using either numerical 100%, numerical 56% or frequency data. Binary data when classified yielded less meaningful information.

A comparison of the information content of frequency and numerical data was performed by calculating correlation coefficients for each sample. It was found that frequency values do not contain any less information than numerical values.

During the census, the time required to complete each method was noted. Collecting frequency data was found to be more time efficient than collecting numerical data. Consequently it is suggested that when both the costs (time expended) and benefits (information gained) of each data type are considered, frequency data is the optimal form of data for use in the classification of sessile invertebrate communities.