Burnham parameter estimates and GoF testing

questions concerning analysis/theory using program MARK

Burnham parameter estimates and GoF testing

Postby THOMO386 » Tue Sep 28, 2021 11:03 pm

Hello,
I am new to MARK and trying to conduct an analysis for first year survival estimates for sea lions using a Burnham joint live encounter dead recovery model. The dataset in question is long term from 1994-2021 for a population in which the first pup was born in 1994 and is slowly growing from there. New born pups have been tagged every year roughly February-April but it has not been confined to a specific occasion or timeframe each year. Similarly re-capture (re-sighting) data has been collected continuously through the years, with inconsistent effort. Recoveries are from the same area as are re-sightings so I have fixed the fidelity parameter to 1.

My approach then is to bin re-sightings into yearly occasions to create an annual capture matrices including males and females. This raises obvious issues with scarcity of data as the early years have very few individuals to sample, and some years there are no re-sightings. From my understanding this results in non-estimable parameters? In contrast, some years have recapture probabilities of 1 (with very small SE) and checking the dataset this seems to be correctly estimated so I would think this is ok? Does this affect the model in any way?

I created a general model with S, p and r varying by time which resulted in a model that was over parameterized, and on further thought this make sense given the non-estimable nature of some parameters. From reading, I have gathered that the general approach is to run GoF on the most parameterized model. Is there a reason that GoF tests aren't just run on the model from which inferences are to be made (e.g. the top candidate model), or do I need to fit a "general model" which isn't over-parameterized first for GoF?

Finally, I would think that the boot strap and median c-hat would be the preferable tests for this type of model (once I have one that is properly estimating parameters)?

Sorry for the small essay of questions, any and all help/suggestion would be greatly appreciated!
Thanks very much
THOMO386
 
Posts: 2
Joined: Sun Jun 13, 2021 10:38 pm

Return to analysis help

Who is online

Users browsing this forum: No registered users and 9 guests

cron