were conducted using the Plymouth Marine Laboratories PRIMER v6 (Plymouth Routines in Multivariate Ecological Research) suite of programs (Clarke and Warwick, 2001; Clarke and Gorley, 2006).
5.
Faunal data for multivariate analysis were imported into PRIMER and initially subjected to fourth root transformation to reduce the influence of any highly abundant taxa allowing less abundant species a greater role in driving the emergent multivariate patterns. The transformed data were then subjected to hierarchical clustering to identify sample groupings based on the Bray Curtis index of similarity. This process combines samples into groups starting with the highest mutual similarities and then gradually lowers the similarity level at which groups are formed. The process ends with a single cluster containing all stations and is best expressed as a dendrogram showing the sequential clustering of stations against relative similarity. To best describe the ecological differences between sites, the groups were identified on the basis of a slice at 20% similarity for the Infaunal communities and a slice at 48% similarity for the Epifaunal communities.
6.
The MDS (Multi-dimensional Scaling) procedure uses the same similarity matrix as that used by the cluster analysis to produce an ordination of stations which is multidimensional. This is carried out to satisfy the between samples relationships indicated by the similarity matrix. This multi-dimensional ordination is then reduced to a 2 or 3 dimensional representation that is a more accessible and useable representation. The representativeness of these low dimensional versions, in comparison to the multi-dimensional array, is indicated by a stress level. The closer this stress level is to zero, the better the representation.
10.6.4 Results
10.6.4.1 Infaunal communities 7.
Some data rationalisation was undertaken before performing multivariate analysis on the full grab sample dataset. Only the enumerated components of the species recorded in the grabs was included. Were a presence or number of species per volume had been recorded a value of 1 was used in the data.
8.
16 distinct infaunal communities were identified at a 20% similarity slice. At this level communities at six stations did not fit within any of the groups nor did any of these six group fit with one another. Therefore, these stations were termed outliers.
9.
The resultant dendrogram was very large and therefore it is not possible to display within this report, the MDS plots are displayed below. The stress revealed by 2– dimensional representation (Diagram 10.6.1) is given as 0.24 (top right corner of the
