I am trying to compare distances matrices belonging to two species matrices. One is bacteria and the other is ectomycorrhizal fungi. Both species matrices were generated from the same plant roots using different extraction methods.
I want to compare the species composition between the two matrices to determine if the bacterial and ectomycorrhizal compositions are correlated in different plant roots. The problem I am facing is that the bacterial matrix is much larger than the ectoymycorrhizal matrix and thus they are asymetric.
I know that Mantel's test does not require symmetric matrices in theory, however the packages I have found that run mantel tests all do require symmetrical matrices. I have tried vegan, ade4 and ape and all have given issues when trying to run the mantel test.
Does anybody have any ideas as to how I can modify my data to run in one of these packages, or perhaps another package that can run a mantel test on asymmetric data?
Thanks!
I have tried vegan, ade4 and ape and all have given issues when trying to run the mantel test.
You are trying to compare two matrices of different sizes:
However, the Mantel test requires that the two matrices that are being compared are of the same size, as it calculates correlations over the corresponding positions in each matrix.
The Mantel test is normally applied to distance or (dis)simmilarity matrices, which are generally symmetric. Under the assumption that your abundance matrices contain the same samples, the two distance matrices are of equal size. If your matrices are of unequal size, they do not contain the same samples and you cannot apply the Mantel test. If this is the case, you should limit the analysis to only those samples that occur in both matrices.
If you have not yet calculated distance matrices from your abundance matrices, you can use the
vegdistfunction in theveganpackage.