ear-tagged and culled deer - population estimation

[slmayhew]

One of my co-workers would like to estimate the density of a white-tailed deer population on a hunt club, if possible, using some data from tagged deer, but there are some confounding issues that I am not sure how to address.

The scene: The club is many, many square miles of good deer habitat surrounded by a high fence, but with regularly spaced low sections to allow deer to cross the fence. So throughout the scenario there may be some immigration/emigration, although likely minimal.

Daily trapping (traps closely but irregularly spaced and covering almost all of the club land) and ear-tagging of deer occured over approximately 3 weeks during the winter, followed by a 5 day no-trapping interval during which a cull occured, followed by approximately 4 weeks of almost daily trapping and ear tagging. (The trapping and tagging were not for the purposes of a mark-recapture study, hence the difficulty in trying to analyse the data in this manner.) There were many recaptures during the trapping and 2 recoveries of tagged deer during the cull. (One problem, some of the deer appear to be trap-happy.)

I can generate capture histories for the ear-tagged deer, and I know how many deer were culled and which were ear-tagged.

The question: can we estimate the population density/size of the deer on the club before and/or after the cull?

I have considered several options, but none seem to be appropriate or to take advantage of all of the data. I considered Lincoln-Petersen using the first trapping effort as the first capture period and the cull as the recapture, but that does not take advatage of the data from the deer of the second trapping period or the daily capture/recapture data. I considered change-in-ratio because the cull was almost entirely does, but trapping tends to be biased toward fawns and does, violating the equal-detection probability assumption. I considered robust design with the two trapping efforts as the primary occasions, but I'm not sure how I can account for the known number of removals.

I would greatly appreciate any suggestions anyone might have on how I should approach this problem.

Sarah Mayhew
Michigan DNR

[ganghis]

Hi Sarah,

Sounds like the 'Barker Robust Design' data type is what you need. Idea is to use a robust design before and after culling season, and to model survival and recovery between closed periods. I don't think there's any documentation of this model directly in MARK, but you should be able to figure out what each of the parameters mean by reading relevant sections in the Gentle Introduction. Alternatively, there wouldn't be anything wrong with ignoring the recoveries and doing a straight robust design (don't censor these animals' captures), but you would lose a little information.

Cheers,
Paul Conn