Which statistical test/model does the R package adespatial use during forward selection to acquire the p-value?

Which statistical test/model does the R package adespatial use during forward selection (forward.sel) to acquire the p-value? The documentation only describes a p-value is returned, which can be corrected with the Holm method. Is the same method applied as with the vegan package (anova.cca)? Thank you.