Thank you yuling, that’s pretty straightforward.
I think I was overcomplicating it!
I have also computed the weighted average of the log-likelihoods from the posterior and fed that back into loo to obtain a WAIC value that I compare to the individual models’.