Search code examples
rmatrixcorrelationvegan

Partial Mantel test in R vegan package gives error

I'm trying to run Partial Mantel tests using R vegan package. I'm getting the following error in the output even on a very simple dummy data.

My code:

  m1 = matrix(c(0, 1, 2,
                1, 0, 3,
                2, 3, 0),nrow = 3)
  m2 = m1 * 2
  m3 = m1 * 3
  mtest = mantel.partial(xdis = m1, ydis = m2,zdis = m3)
  print(mtest)

The output:

Partial Mantel statistic based on Pearson product-moment correlation 

Call:
mantel.partial(xdis = m1, ydis = m2, zdis = m3) 

Mantel statistic r:   NaN 
      Significance: NA 

Error in quantile.default(x$perm, c(0.9, 0.95, 0.975, 0.99)) : 
  missing values and NaN's not allowed if 'na.rm' is FALSE
Solution

  • The example matrices are linearly dependent; m2 is a simple linear transformation of m1 so they are perfectly correlated. As is m3. Once you account for the correlation between one pair of matrices, there is nothing left over to correlate with, hence the weird NaNs and NAs and the ultimate error.