Welcome to Robust Design Mark-Recapture

Hey Lily!

This site is for you and introduces the fundamentals of robust design mark-recapture modelling that you will need for your honours project. I have tried to write it to be accessible in terms of your project, but I also want to give you a solid statistical foundation that will hopefully be useful after your undergraduate.

If anything here is wrong or confusing, just let me know. I would be genuinely surprised if everything made sense on the first read, so do not be alarmed if parts need to be talked through in person. The stats here are complex. I think you can understand them, but it might take a bit of time. So ask for help if you get stuck.

Suggested order

I’d recommend reading in the following order:

1. A refresher on GLMs
   You will be building on generalised linear models throughout (you learnt these in BI3010), so this page makes sure the core ideas are fresh before we introduce anything new.

2. The detection problem
   The central problem that motivates needing fancy methods: non-detections are ambiguous, and ignoring that ambiguity biases every estimate you care about.

3. The Lincoln-Petersen estimator
   The simplest mark-recapture method. Two trapping occasions that gives one estimate of population size. Important to wrap your hear around.

4. The Cormack-Jolly-Seber model
   The model that first separated survival from detection. Developed, as it happens, by two people working in Aberdeen.

5. The robust design
   How nesting “closed sampling occasions” inside an “open population framework” lets you estimate population size and survival simultaneously, from the same data.

6. Temporary emigration
   The final layer of the model: animals that are alive but temporarily absent from your study area, and the two parameters that handle them.

7. Implementing the robust design in RMark
   The practical page. How to organise your data, write out the full model, fit it, and produce figures that communicate what you found.

Enjoy! (Maybe)