overestimation of bear abundance, heterogeneity of capture

[bear16]

Someone mentioned that Gary made a statement in his talk at the Wildlife Society conference indicating that bear abundance is always overestimated while other estimates tend to be underestimated. Why is that? Genotyping error? Something to do with the model used? How do we minimize this problem? Is there a good source on this topic?

Some people have suggested to me that the Huggins heterogeneity model tends to overestimate abundance, but others have indicated that there is definitely heterogeneity in capture of bears (in hair snares) hence the need to use this model for single year data. Boulanger (2004, 2008, etc.) has found that females are more likely to visit snares while males visit tree rubs more frequently and that distance from the grid edge and previous physical capture may be influential. Are there other sources that suggest heterogeneity across individuals? Is it more appropriate to just use the regular Huggins model with a covariate for distance from the grid edge? Or would that likely not include all the heterogeneity?

Thank you for your help! I am just learning about the various models, so I would appreciate any help in the matter.

[gwhite]

I've been mis-quoted. My comment was that DNA studies (often on bears) that had mis-classification are one of the few ways that mark-recapture estimates are biased high. In general, mark-recapture studies are biased low because of individual heterogeneity. The Pledger model helps correct for heterogeneity and give unbiased estiamtes, although the precision of the estimate suffers. I don't believe there is a general conclusion about whether the Pledger estimator is biased either way because of all the ways that individual heterogeneity might occur.

Gary

[bear16]

Thank you Gary! I will try out the Pledger model.

[cooch]

Thank you Gary! I will try out the Pledger model.

In chapter 14 of the MARK book, 'Pledger models' are referred to more generally as 'mixture models', since the approach is based on a finite mixture approximation (first developed in a likelihood framework by Shirley Pledger and colleagues). Be sure to read the relevant sections of chapter 14 carefully. The mixture models are relatively straightforward to execute, but it is important to understand the assumptions underlying the models.