R help - Apr 2013 - vegan -varpart is bigger than 100% in total?

I am trying to find the percentage of the parameters explaining the bacterial
community composition. I have one data matrix with relative abundance of OTUs
and one with environmental parameters. I used varpart in vegan package but the
values in the venn diagram is bigger than 100% in total.How is it possible? What
might be the reason? Thank you

library(vegan)
gotud <- decostand(gotu,"hell") 
G.var <-
varpart(gotud,~Depth,~CH4_oxidation,~d13C_DIC+Temperature+Spec.Conductivity+DO+CH4+N2O+pH,
data=ge)
plot(G.var) 


Sample	 depth	T	SpCond DO pH Chl CH4	CH4oxidation	N2O	TA	POC	BP	SO4 
Gi25Feb12	25	24.46	1104	4.42	8.94	1.6	40.00174578	1.1	6.683260809	12.723
41.98069543	0.008282072	150.1168
Gi27.5Fb12	27.5	24.42	1105	4.36	8.93	1.2	68.53538846	1.4	7.431459607	12.711
31.4397919	0.020730967	188.4166
Gi30Fb12	30	23.71	1128	2.62	8.81	1	10.53158545	0.9	8.415644664	12.817
23.06912467	0.007716205	199.9561
Gi32.5Fb12	32.5	23.58	1135	1.63	8.79	1.2	11.77550422	0	7.069811212	12.916
26.80357225	0.00245392	170.9078
Gi35Fb12	35	23.54	1138	1.42	8.78	-0.4	11.57631611	0	6.976269751	12.978
31.80221025	0.006877572	176.4877
Gi37.5Fb12	37.5	23.5	1138	1.37	8.78	0.6	17.19312361	0	7.382521684	12.975
19.67401691	0.007159828	164.5442

OTU matrix 

Group	Otu04518	Otu04569	Otu04598	Otu04636	Otu04671	Otu04672	Otu04795	Otu04821
Otu04891	Otu04930	Otu05062	Otu05118	Otu05148	Otu05440	Otu05520	Otu05555	Otu05690
Otu06033 Otu06037	Otu06194	Otu06354
Gi25Fb12r	0.000567	0	0	0	0.001134	0.000567	0.002834	0.000567	0.000567	0	0	0	0
0.000567	0	0	0	0	0	0	0
Gi275Fb12r	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0 
Gi30Fb12r	0	0	0.003145	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0 
Gi325Fb12r	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.0048	0.0208	0	0	0	0	0 
Gi35Fb12r	0	0	0	0.001114	0	0	0	0	0	0	0.001114	0	0	0	0	0	0	0	0	0	0 
Gi375Fb12r	0	0.001447	0	0	0	0	0	0	0	0	0	0	0.001447	0	0	0	0	0	0	0	0

Ozgul Inceoglu <Ozgul.Inceoglu <at> ulb.ac.be> writes:
> 
> I am trying to find the percentage of the parameters explaining the
bacterial community composition. I
> have one data matrix with relative abundance of OTUs and one with
environmental parameters. I used
> varpart in vegan package but the values in the venn diagram is bigger than
100% in total.How is it possible?

Dear Ozgul Inceoglu,

This is not possible and this is not the case. What is possible, and
what does happen is that you can have negative components in your
diagram. These negative components are mind boggling, but they do
appear regularly. Legendre & Legendre (2012) "Numerical Ecology"
give
some reasons for their occurrence.

In you Venn diagram you have obviously selected an option of not 
displaying negative terms. Get the full sums of the components,
including negative ones, and I bet the sums will match.

Cheers, Jari Oksanen