st3 <- system.time(ofit_glmmadmb_mcmc <-
                   glmmadmb(NCalls~(FoodTreatment+ArrivalTime)*SexParent+offset(logBroodSize),
                            random=~1|Nest,
                            data=Owls,
                            start=list(fixed=coef(ofit_glmmadmb),pz=ofit_glmmadmb$pz),
                            zeroInflation=TRUE,
                            mcmc=TRUE,
                            mcmc.opts=mcmc.control(mcmc=5000),
                            family="poisson"))
## effective size on short trial run seems to be about 20% of real size
##  (but could be worse than that) so try running for 5000 steps ...