[R-sig-eco] variance partitioning using 'varpart' in vegan

[ckellogg]

Hello,
I have been using the 'varpart' function in vegan. For one of my datasets, I
get many "negative" variances.  I have looked at the equations provided in
partitioning.pdf in vegandocs, and I see how this can physically happen -
but I guess I do not fully understand why or what exactly this means.  If
anyone can provide any insight or point me in the direction of a helpful
resource, it would be much appreciated.  One other item I have been
struggling with is in the output for 'varpart'.  If the 

Also, in testing the significance of the variances explained, I have
following the code shown in the 'varpart' documentation.  If I am using 4
explanatory variables - A, B, C, D - and want to figure out the significance
of the variance shared by all variables, would this be correct
implementation of this code:

rda.result <- rda(ANEEAbulk.z2 ~ A + B + C + D)
anova(rda.result, step=999, perm.max=999)  

But isn't this more the total variance explained - not the shared variance? 
How would I code to find the significance of the shared variance among all
variables?  Is this significance even testable?  

and if I want to figure out the significance of the shared variance
explained by just A and D, would this be the correct code:

rda.result <- rda(ANEEAbulk.z2 ~ A + D + Condition(B) + Condition(C))
anova(rda.result, step=999, perm.max=999)

I hope my questions make sense.  I just want to make sure I am stepping
through all of these procedures correctly.  

Thank you very much,

Colleen

.....
Colleen Kellogg
Postdoctoral Scientist
UMCES - Horn Point Laboratory     
Cambridge, MD 21613

--
View this message in context: http://r-sig-ecology.471788.n2.nabble.com/variance-partitioning-using-varpart-in-vegan-tp6972644p6972644.html
Sent from the r-sig-ecology mailing list archive at Nabble.com.