nMDS and clustering analyses can essentially show you how similar or dissimilar groups are. One example is in studies assessing community structure in benthic habitats, nMDS and clustering analyses, coupled with PERMANOVA/ANOSIM can distinguish sites or locations (what you are calling 'stations') that are similar or dissimilar in terms of their community structure. One could then look for potential correlates that may be influencing the trends expressed in the nMDS/cluster analysis (e.g., grain size, pH, anoxia/hypoxia, etc.). It can also be used in the same way for studies of biodiversity, by which you can distinguish areas that are similar or different in terms of biodiversity.
Totally agree with previous opinions. nMDS plots are very handy to show distribution patterns, however people normally tend to forget that its just a graphical representation (2D) and has to be followed by some sort of statistical analysis in order to confirm what the ordination plots show. In this regard, ANOSIM, PERMANOVA among others, are very useful.
Alternatively if you have some environmental data a DCA, RDA, PCA might be useful to to show how your sample/species are correlated with the different variables.
As previously stated, nMDS is an exploratory analysis which does not allow you to explain the similarities or dissimilarities among samples/sites/stations etc. The output graphics can show you some patterns in your data and you can try other analysis to try to explain them.
In the cluster analysis, not only the choice of the index is important, but also the decision regarding the clustering strategy (UPGMA or WPGMA).
Finally, although the topology of a dendrogram which resulted of a clustering analysis is similar to a phylogenetic tree, it is important to clarify that this analysis should never be used to search for phylogenetic relationships, since it is based in similarities rather than homologies and parsimony.
Very nice explanations of the value of NMDS. May I refer you all to a paper we published, which uses NMDS to show the time course of benthic annelid assemblages that have been manipulated by detrital addition.
Kelaher, B., Levinton, J.S., 2003. Variation in detrital-enrichment causes changes in spatio-temporal development of soft-sediment assemblages. Marine Ecology Progress Series. 261: 85-97.
This paper can be downloaded from Research Gate.
What you can see is a variety of time courses of responses but an eventual convergence of annelid composition as resources are used up from spring to summer on a salt marsh mud flat in the northeast United States. The paper also demonstrates stronger responses of surface feeding annelids to detrital additions, relative to deeper feeding annelids, which show less spatial differentiation in time course and spatial variation of abundance.
Cheers,
Jeff Levinton
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software