Analysis of Variance in Vegetation Surveys Based on the Beta-binomial Distribution

Abstract

SHIYOMI et al. (2000) proposed a new method for surveying grassland vegetation and analyzing the surveyed data, based on the beta-binomial distribution. Using this method, we were able to estimate (1) the number of occurrences of each species observed per unit area (quadrat) in a grassland, (2) the degree of spatial heterogeneity for each species and for the whole community, and (3) the species diversity and its spatial distribution. In this paper, we develop a method for comparing the number of occurrences and the spatial heterogeneity between species which occur in the surveyed grassland. For this purpose, an analysis of variance (ANOVA) based on a random model was applied (adequate data-transformations are necessary for calculating statistical quantities). Using the error variance in the ANOVA, we can apply DUNCAN's method, SCHEFFE'S method, TUKEY's method for multiple comparisons between species and STUDENT'S t-test. This ANOVA will make it practical to analyse vegetation in grassland surveys.