reverse time multi-strata robust design

[sharpnw]

I wish to employ the method described by Nichols et al 2000 whereby the seniority parameter (gamma) is calculated for 2 age classes (adult and young) by using the multi-state robust design model. When reversing my capture history, should I reverse it in totality or should I reverse only the order of the primary periods while maintaining forward time over the secondary periods for estimation of the capture probabilities?

Thanks
Nick Sharp
Auburn University