creating residual variable at proc glimmix

[sld]

Yes, DHARMa is an R package, not SAS.

The "23/45" location does not direct me to the text that I think you are referring to (the document does not have page numbers, and the numbering depends on your particular computer). DHARMa functions are able to accommodate and assess overall overdispersion, but from my reading of it, I do not think it assesses overdispersion that varies depending by levels of predictor variables.

Although GLIMMIX has the ability to specify heterogeneous variances, the SAS code in your message does not specify heterogeneous variances in any way (note what you have commented out using /* ... */). For distributions other than the normal, I think great care would need to be taken with specification of heterogeneous variances. I am not in the position to advise you about that coding: I have not done much of that sort of modeling, and when I undertake it at some time in the future, I would do it only in the framework of a lot of simulation exercises--where you know the answer, and see how well you can recover it.

But maybe you are not trying to specify heterogeneous variances in combination with a gamma distribution? It is not clear from your message.

I hope this helps.