Multistate occupancy modelling

[Prashant_mahajan]

Hi- I am working on multistate occupancy modeling in PRESENCE. I need some help in designing the matrix for the same. I have been entering the data in the following way:

a1 a2 a3 a4
psi 1 0 0 0
R 0 1 cov1 cov2

b1 b2 b3 b4

p1(1) 1 0 int 0
p1(2) 1 0 int 0
p1(3) 1 0 int 0
p2(1) 0 1 0 int
p2(2) 0 1 0 int
p2(3) 0 1 0 int

c1
delta(1) 1
delta(2) 1
delta(3) 1

where occupancy is kept constant, state 2 (R) is varied by two covariates (cov1 and cov2), and both the detection probabilities (p2 & p2) are varied by the same covariate (int) while delta is kept constant. I just wished to know If the above design matrix is correct or there is some error in designing the matrix. Also, I have a doubt in two parameters, delta (probability of correcting identifying breeders given presence) and p12 (probability of observing species in state 2, given the true state being 1). Are both delta and p12 different or the same? Can some please explain both the parameters in brief?

[jhines]

Hi, your design matrix for psi and R looks correct. Psi (prob. site is occupied) is constant for all sites and estimated using beta parameter a1. R (prob. site is in 2nd state, given it is occupied) varies according to 2 covariates and an intercept and is estimated using beta parameters, a2, a3 and a4.

Detection matrix looks good as well. Both p1 and p2 are constant across surveys, but vary according to the covariate 'int'. The number in parentheses next to the p1 or p2 is the survey number.

Delta matrix is OK too, with delta constant.

P1(i) is the probability that the species is detected in survey i, given the occupancy state is 1 (eg., occupied but no reproduction). p2(i) is the probability that the species is detected in survey i, given the occupancy state is 2 (eg., occupied with reproduction).

If the occupancy state is 2, you might detect the species being there, but not detect evidence of reproduction. In that case, the detection-history would record a '1' for that site/survey, even though the true state is '2'. The probability associated with that event is p2(i)*(1-delta(i))... species was detected, but evidence of state=2 was not detected. If evidence of reproduction was detected, the detection-history would be '2' for that site/survey and the probability would be p2(i)*delta(i).

[Prashant_mahajan]

Thank you for the response and the clarification in the delta parameter.