Hi Ree,

nMDS and ANOSIM are not directly related and you can use them alone. With ANOSIM, you statistically test if the groups differ in species composition. With nMDS, you are not performing any test, you (the algorithm in fact:) just arrange plots in low number of dimensions (typically 2) - usually in order to display the overall compositional pattern in your data.

However, both approaches are often used together, because:

1) ANOSIM tells you the test statistics and associated p-value, but it is hard to imagine how compositionally different are the groups. Therefore, nMDS is used to evaluate visually how compositionally different are groups, which are statistically compared with ANOSIM.

2) ANOSIM statistics is based on rank order of dissimilarities and nMDS algorithm tries to display rank order of dissimilarities. Therefore - form all ordination methods - the nMDS is the most related to ANOSIM.

Hope this helps (a little bit)

Best,

martin

Martin has answered quite well above - nMDS is simply a way of visualising the results - it is separate from the ANOSIM test. You can still perform an ANOSIM if the stress value for your nMDS is high. To put it simply, a high stress value just indicates that the algorithm has had trouble visually representing the spread of your data in 2 dimensions. Therefore if the stress is very high there there is probably little point including the nMDS plot in your report.

Thanks so much! This is really helpful!