Unintuitive Results

[mjgould]

Hi Murray,

I have some results that I can't quite wrap my head around. I estimated density for the NSC using hair traps. Within the NSC I also estimated density for VPR separately, which accounts for approximately 1/4 of the study area. Unfortunately, we experienced low genotyping success with our genetic samples (~50%) compared to similar studies (~75%). The VPR was especially atrocious. Out of the 100 animals detected on VPR, we only recaptured 5 individuals due to loss data. As expected, VPR's estimates were higher than NSC, their CIs were extremely wide, g0 was incredibly small, and sigma incredible large. I was curious as to whether or not VPR was inflating NSC's estimate so I reanalyzed the data excluding VPR in the mask and any detections or recaptures associated with VPR traps. Surprisingly, the density for NSC without VPR increased. This doesn't seem intuitive to me. I thought that the NSC density would decrease, or at least stay the same. My only guess for the observed results is that VPR's density and CIs were larger because with so few recaptures secr was unable to adequately model the population. Yet, I'm still a bit stumped as to why density for the NSC increased instead of decrease. Any thoughts? Thanks for your time.

Matt

[murray.efford]

Not sure I can say much on the basis of what you've posted. I am a bit suspicious about

...reanalyzed the data excluding VPR in the mask...

as by (apparently) clipping the mask you are excluding some habitat adjoining NSC (whatever that is) that may have actually contributed animals to the sample 'captured' in NSC. This would tend to inflate the estimated density in NSC by a small amount.
Murray

[mjgould]

Sorry for the confusion, Murray! Those two acronyms are long-winded names for two different areas. In my attempt to simplify things I just made it worse. Even so, you assumed correctly that I clipped part of the habitat mask. Also, I did some more digging and thinking after I posted and you're correct. I lost recaptures including ones that turned 2-trap/occasion detection events into a single trap/occasion detection event. I appreciate the insight.