Non-parametric multivariate techniques contained in
the PRIMER (Plymouth Routines In Multivariate Ecological Research) software package (Clarke & Warwick
1994) and described in Clarke (1993) and Warwick
(1993) were used because they are powerful tools in the
analysis of non-parametric community-structure data.
Similarity matrices were constructed using the Bray–
Curtis similarity measure on non-standardized, squareroot-transformed crab and gastropod abundance and biomass data. Bivalves and rare, unidentified gastropod species were not included in the analysis, but they are listed
in Appendix 2. Transformations reduced the impact of the
common species for a better interpretation of the data.
Formal significance tests for differences between plots
were performed using the one-way ANOSIM permutation
test (Clarke 1993, Clarke & Green 1988). The species
contributing to the dissimilarities between the plots were
investigated using the similarities percentage procedure
(SIMPER; Clarke 1993). Two-dimensional (2D) ordination of similarity matrices calculated from mean data were
displayed using non-metric, multidimensional scaling
(MDS) (Clarke & Green 1988, Kruskal & Wish 1978). A
stress coefficient reflects the extent to which the rank
order of distances between samples on the ordination
agrees with the rank order from the similarity matrices.
The higher the stress coefficient, the higher the disagreement; for example, for 2D ordinations stress < 0.05 gives
an excellent representation.