Dataset including some intervals without new marks

[laportepierre]

Variation in survival estimates with the various time intervals

I am puzzling with the result of my analysis of my “unusual” datasets with MARK.
I hope that somebody could give me indication of what I should do.

I have a data set of 8 years (time encounters) where birds have been banded as young or adults in 5 regions. I build a model general using two class age configuring the PIM as when both young and adults are marked. The fives regions where coded as a multistrata model to get estimates of migrations between regions.

In the dataset, the last two years, no anymore birds were banded but effort have been done to catch birds to increase the return birds.

If I run the model with only with the 6 first years where birds all adult and young birds where banded, I get, for each regions the highest adult survival rate. If I had the seventh years (where no new birds where banded, adults survival estimates decrease by around 20% for all regions,. If I add the 8th years, I find another 20% decrease in adults survival estimates; at the same time for youngs survivals remains write quite stable

I not understand why the adult survival decreases due to the fact the last two years of data set doesn’t contains new marked birds. Is somebody have any explanations or suggestions

My perception is to use the first 5 years for adults survivals estimates and the full 8 years for survival and migration rate, as the sampling size increase in this case.

Is I am wrong?

P. Laporte