Dear all

I would like to compare the community composition of a certain region, against the composition of the neighbouring regions. This is an attempt to validate the ecoregion itself (its species composition is significantly different from that of its neighbours, or needs to be reassessed (merged with others, or boundaries redrawn?).
The original tread to do so (comparing regions according to their spp composition to see if they were distinctive units) was an article by VanSickle, and its method is implemented in the R package vegan (mrpp). Reading further, I found that adonis is more recommended for such questions (also read Anderson's paper about the technique).
What I am interested to know is if the focal region is different from its neighbours (I will test all regions against their neigb one by one), and NOT if the whole set of neighbours are significantly different from each other. So I re-classed my data entries into only two categories, as "focal" and "other", as opossed to "focal", "neigb1", "neigb2","neigb3"...

My qst is If is that valid? Reclassing all the other neighbouring regions as "other" without distinguishing them? or am I violating some test assumption that I am failing to see.

I have ran both tests (focal vs. other and focal vs neigb1, neigb2, etc) using the same data, and results are obviously different Sum of squares, R2, etc., but so far the significancy is consistent (p<0.5) for both tests. I double tested doing and NMDS and assessing visually and those regions that are significantly different, look like they are... Haven't found [so far] a case where one test is significant and the other is not. But I have <50 regions to test, for five vertebrate groups which renders more than 250 permanovas (x 2 if I have to double test)... I have no problem programming that, but I would like to know if I am in the correct path, since still running the script will be very time consuming, and also because I want to do what is statistically correct.

My main interest is to identify which regions do not differ from their neighbours, because those will need a closer look and maybe future reassignment, those significantly different from their neighbours will be left as they are , and all people in my team will be pleased and happy!

Thanks a lot for your help and for reading this looong post (if you ever reached this bottom line).

A