Multistate models, hypotheses as a priori models

[Guest]

Hello

I am analysing movements of snow geese between 2 strata; A and B. I tried using MSSURVIV, but people tell me MARK is the way to go. I already figured out how to insert the data and get estimates of S, p and PHI. But the numbers by themselves mean very little to me. Sure, phi ranges from 0.05 to 0.5, but so what? I know from looking at the data that many snow geese (up to a 25% at some banding locations) moved between A and B.

I am testing between 2 hypotheses (see below). In the spirit of a priori hypothesis testing, I want to build models that simulate the 2 hypotheses, and then test them against my data. Can it be done in MARK? If so, how do I start?

The 2 hypotheses are:

1) Snow geese banded in 2 strata are separate populations, i.e. "nobody" moves between A and B. (Capture histories AAAA and BBBB will have the highest probabilities associated with them. In the most extreme cases ABAB, BABA etc. have probability of 0)

2) Snow geese sample both habitats and use both, i.e. they move freely between A and B. (Capture histories ABAB and BABA etc. will have highest probabilities associated with them. In the most extreme cases AAAA and BBBB have probabilities of 0).

best regards,

Jón Einar Jónsson
School of Renewable Natural Resources
Louisiana State University

[cooch]

Hello

I am analysing movements of snow geese between 2 strata; A and B. I tried using MSSURVIV, but people tell me MARK is the way to go. I already figured out how to insert the data and get estimates of S, p and PHI. But the numbers by themselves mean very little to me.

Before asking this sort of question, I'd strongly suggest you download and read (i) the guide to using MARK, and especially (ii) the chapter on MS models. All of your questions are explained there. The guide can be fond at

http://www.phidot.org/software/mark/docs/book/

[Guest]

I spent today reading the multistate chapter.

I take it that if I fix the movement probability (psi) in a model and run it, then that model represents a certain hypothesis, and that model can be used in model selection against other models with other fixed values of psi?

[cooch]

I spent today reading the multistate chapter.

I take it that if I fix the movement probability (psi) in a model and run it, then that model represents a certain hypothesis, and that model can be used in model selection against other models with other fixed values of psi?

Based on your query, I'd suggest (again, strongly) that you do more than just read the one chapter - read (at the minimum) chapters 4 -> 7. If you're questions concern basic model selection, you need to start your learning curve waaay before the MS chapter.