teeny tiny weeny confidence intervals....nonsense surely?

[lionmonkey]

Ooooooohhhhhhhh why ofh why

I am trying to generate population estimates from wallabies coaxed into cage traps and given pit tags. We trap them for four consecutive nights.

The data per animal looks like this

0010 1;
etc etc

I feed it lovingly to MARK and plump for closed capture and run predefined models where only N is not time dependent i.e. 4 models.

I then learnedly peruse the AICc and nod knowledgeably at my colleagues as I pluck the "best" model with a model likelihood of 1 and the smallest negative AICc. The only problem is the bnloody estimate has a confidence interval of 0 around an estimate of 90....could our experiment have been so damned good. Of course not, we are mere mortals....what the !@#$ am I missing?