Poisson log normal for temporary and permanent marks

[SBeatham]

Hi,

I'm a rookie MARK user, I have read through a lot of the relevant chapters in 'A gentle introduction' but unfortunately I am still unsure how to analyse my data. I have been trapping badgers from six populations 4 times a year for the last two years. Three have been treated with a fertility control drug and three are controls. I am therefore interested in changes in population size and birth rates over the two years. On each capture individuals are uniquely tattooed and given a unique fur clip. They were then camera-trapped. I wanted to compare mark recapture and mark resight techniques. I have used Bowden's estimator on the mark resight however number of marked animals is very low, <6, meaning I have very large CIs in some cases. I was wondering if there was any merit in analysing it using the Poisson log normal closed population model. I read in Chapter 18 it was possible to look at changes over time as I have permanent IDs however can I use the robust design if my field marks are temporary (I haven't been able to devise a way of getting unique permanent field marks on badgers) meaning a previously caught animal will be seen as unmarked if its fur clip is not refreshed.

I have read previous post where it has been said that capture mark recapture N estimates from open populations are inaccurate, I was looking at JS but now I'm a little confused. I have small populations with high recapture rates and I also have the issue of temporary immigration/emigration.

I hope you can help (I'm starting to think I'm beyond help!)

Many thanks

Sarah

[bmcclintock]

Hi Sarah,

It might require some additional information about your sampling design to explore all of your options, but (assuming you have “closed” camera-trapping sessions and the temporary marks are not lost within these sessions) you can use the Poisson log normal closed population model to estimate abundance. Ignoring survival and temporary emigration (which you cannot estimate using mark-resight unless tags are permanent and field-readable), you could perform an analysis where each closed session constitutes a “group”. The advantage of this approach over Bowden’s Estimator would be the ability to “borrow” resighting rate information across sessions to improve precision. See Section 18.4.1 in The Book.

Cheers,
Brett

[SBeatham]

Hi Brett,

Thank you for the quick response. I assumed this would be the best way of analysing the data but from chapter 18 I could not work out the format the capture history would have to be. With my 8 closed population mark resight primary intervals for each of the six badger groups, would I enter this as 8 groups and 1 primary interval and enter each badger group as a separate file? Then I presume I would analyse differences between treated and control using a separate stats package.

| understand the method above would be more accurate but I would like to try a capture mark recapture as well as a comparison, I've agreed to do a talk comparing the two so I'm committed to it now! Would POPAN be the best model to use for this?

Thanks again

Sarah

[bmcclintock]

As described, you would have 6x8 = 48 "groups" with 1 primary interval (i.e. 48 abundance estimates). The idea then is to investigate more parsimonious models that constrain resighting rate parameters by population, treatment, etc. I'm not sure what you mean by "analyse differences between treated and control using a separate stats package," but you can look at treatment effects within MARK by keeping track of which of these "groups" belong to each treatment.