Temporal changes in community structure were further examined
by multivariate analyses. Similarity of assemblage was calculated
using the Bray–Curtis dissimilarity index for all pairs of the 10
groups by combining data from 5 replicate samples for each group.
Two types of similarity matrices were developed using nontransformed
abundance data and presence/absence data to
examine whether relative abundance of component taxa and taxa
composition vary similarly among years for each site and vegetation
type. Variation in similarity was graphed using a nonmetric
multidimensional scaling (nMDS) ordination method based on 20
iterations of data. (Clarke and Warwick, 2001).