pooling capture occassions in a Robust Design

[fredfrick]

I am working on an analysis of 16 years (15 survival intervals) of mark-recap data for a bat colony. The data are organized with numerous repeated capture occasions during the colony breeding season each year, making a Robust Design appropriate. During the peak of the breeding season (June-July), an assumption of demographic closure seems reasonable as site fidelity and survival during the breeding season is high at bat maternity colonies, so I am planning on using the closed RD.

In addition to secondary sampling occasions (range 5-16) within each June-July period in each year, there were a few additional trapping sessions in May and August, but during these times bats are still either arriving or are starting to depart from the colony site. There are only a few capture occasions in May (3-6 a year, in 5 years) and a few capture occasions in August (1-4 a year, in 11 years).

I’ve Read Chapter 15 of the Gentle Intro as well as Kendall ‘99 (etc) and I see that a potential solution is to collapse all May capture events into a single encounter occasion at the start of my primary occasion and collapse all August capture events into a single occasion at the end (pg. 15-11 of the Intro). Kendall ‘99 points out that results will be biased if there is trap dependence or individual heterogeneity in recapture rates with this strategy, but it will be unbiased if only time-dependent.

My question is: Given that I have many multiple recapture events in June and July (5-16, depending on year), will my results likely be biased if I incorporate trap dependence or heterogeneity into recapture probabilities if I collapse a few occasions in May and August at both ends of the closed period (June-July)?

Other suggestions?

Thanks for your help.

Cheers,
Winifred