I am a biologist by training so please excuse my naivety. I generated the ordination diagrams presented here using NMDS on Bray-Curtis dissimilarities. The two diagrams represent communities of moths and birds along an elevational gradient and the labels on the plot are site elevations. I used ANOSIM (R:Vegan) to assess the significance of dissimilarity across the "differently colored" assemblage clusters in the moth data and the result was significant (R = 0.88). Using the same grouping factor (as denoted by the colors in moth NMDS) I checked for dissimmilarity in the bird data and it gives me significant differences (R = 0.82) while I can clearly see the communities along a more or less smooth arc with no apparent abrupt clustering? I read that the NMDS axes are arbitrary and cannot be compared across two independent datasets.
There is clearly something fundamentally wrong in the way I am interpreting these plots, or the ANOSIM results. Any comments would be appreciated.