Another MCMCglmm question - fixing correlations

Hi,

is it possible in MCMglmm to fix the (co)variance matrix so that the
resulting variances could vary, but covariances had always the value
ensuring that r=1? It would be useful in testing cross-sex rG where
the H0 should be rG=1 rather than rG=0 (as defined in idh()). I looked
through available (co)variance structures but none seems suitable.

best,
sz.