A study of variation in the Bryozoan Fistulipora decora (Moore and Dudley) from the Beil Limestone Member of the Lecompton Limestone of Kansas

Abstract

An increasing number of bryozoan systematists are recognizing the need for a more quantitative treatment of variation in taxonomic studies of Bryozoa. The principal objective of this study was to isolate, as far as possible, the relative contributions of inter- and intracolony variation to the total variation of a population. Failure of the assumption of homogeneity of variances, for the data as a whole, precluded use of a nested analysis of variance model. This assumption was met by subsets of the data for several localities, and single classification anovas were applied in those instances.

Analysis of between locality differences by the Kruskal-Wallis anova analog revealed highly significant differences between localities. A nested anova (for which the assumptions had not been met) was performed for comparison, and resulted in nonsignificant values for between locality differences. This emphasized the hazard of using anova where assumptions are not met.

Analysis of within locality variance revealed highly significant differences between colonies. Significant intercolony variation within a locality can reasonably be attributed to a high degree of genetic diversity as the broad environment within the small area of the collecting site was seemingly relatively uniform. Partitioning of the total variance by single classification anova revealed that the greatest proportion of within locality variation is explained by differences within one colony. This is attributed to differences in the microenvironment affecting individuals within a single colony. Comparison of coefficients of intra- and intercolony variation, as suggested by Oliver (1968), was not appropriate in this study due to the significant overlap of 95% confidence limits for most cases. The independence of characters used in this study (as a measure of the degree of redundancy of information they provided) was evaluated by calculating a matrix of Pearson product-moment correlation coefficients. At least two independent characters could be recognized.