[R-sig-eco] adonis once again

[Alfredo Tello]

Hi Jens,

"strata" is a stratifying parameter that constrains the permutations for
significance testing within the level you assign to it. You would use
"strata" when you have nested factors in your design. In you're first adonis
model you are testing whether richness explains a statistically significant
fraction of the variation observed in reho.dist. Since you are stratifying
by env$orgTab, what the output is telling you is that env$RICHNESS explains
a significant percent of the variation observed in reho.dist when
controlling for differences in env$orgTab. adonis is basically permuting
within env$orgTab and not across the entire dataset. In this case,
specifying "strata" seems appropriate to me because it wouldn't make much
sense to permute across the whole dataset, which in your case being a lit
review, probably implies the use of different methodologies, etc... Anyway,
the specification of "strata" depends entirely on what you want to test. The
same applies to your second example. What adonis is then telling you is that
there is no significant relation between env$AREA and reho.dist when
controlling for env$orgTab.

Hope this helps,

A







--
View this message in context: http://r-sig-ecology.471788.n2.nabble.com/adonis-once-again-tp6232010p6233574.html
Sent from the r-sig-ecology mailing list archive at Nabble.com.