Age-specific survival in a long-lived species

[William1]

I’m hoping someone can give me a quick clarification on classifying age groups. I have two age categories: adults and subadults. Originally, to create the groups I used a “first marked as” classification. A subadult that was first marked as a subadult was included in this group even if it was later encountered as an adult.

I questioned whether it made sense to include the recaptures as an adult into the likelihoods. After some investigation, I noticed that others (Zablan, Braun and White, 2003) for example, would reclassify a subadult as an adult if it was later recaptured as an adult.

I restructured my groups accordingly; subadults included individuals that were only encountered as subadults. Subadults that were later recaptured as adults were included in the adult group.

The two classification methods yielded significantly different survival estimates. “First marked as” models detected no difference in survival, the former method estimated much lower survival for the subadults than adults.

I understand how the discrepancy could arise, since the former method means animals only encountered as subadults would never have a recapture (as adults) thus decreasing apparent survival; whereas using a “first marked as” method results in their adult encounters being included in the likelihood.

For a long-lived species, such as whales, how should age categories be classified to get an accurate comparison of survival?

Many thanks in advance.


Estimation of Greater Sage-Grouse Survival in North Park, Colorado Author(s): Marilet A. Zablan, Clait E. Braun, Gary C. White Source: The Journal of Wildlife Management, Vol. 67, No. 1 (Jan., 2003), pp. 144-154 Published by: Allen Press

[Jochen]

Hi,
assuming you are after a simple CJS model for survival and not knowing the paper you refer to, I recommend to group your animals according to age at first capture, and then take account of age-dependent variation of p and Phi by structuring the PIMs for the subadult group(s). You may then start your analysis with an age and time dependent general model. The number of subadult groups and related parameters in the PIMs usually depends on how long the animals remain subadults.
Hope this helps
Jochen

[William1]

Thanks for the reply Jochen. I had never considered this approach. I have tranisents in the population and have accounted for them using the 2 age-class model. Would your approach still be feasible considering I have to use age classes to distinguish transients?