eps=1-gam parameterization; missing output

[ekalies]

I am using the models:
psi(fire severity),gam(yr),eps=1-gam,p(site) AND
psi(fire severity),gam(.),eps=1-gam,p(site)
in a multi-season occupancy model with 4 seasons.

After running any one of these models, PRESENCE gives me psi for all sites in year 1 ("Individual Site Estimates of Psi 1"), then gamma for site 1 for years 2,3,4 ("Individual Site estimates of <gam1>"). In addition, it gives me lambda for almost all sites in year 2 (275/280 sites) plus site 1 in years 2-4 ("DERIVED parameters - lam2,lam3,lam4,..."), as shown below (truncated).
So, I can't get any information on additional sites for years 3 and 4, and this is important in calculating occupancy based on my psi covariate. I've tried to figure out how to get at the calculations from every which way, but I need at least gamma or lambda for all sites in years 3 and 4. Any suggestions (or am I missing something)? it seems like the output truncates before it is finished with all sites/years. I tried to changes "set # estimates to print by default" to no avail. THANKS!

Site lam(t) Std.err 95% conf. interval
1 B1-1 lam( 2): 1.1301 0.1561 0.8241 - 1.4362
1 B1-1 lam( 3): 0.9210 0.0953 0.7343 - 1.1078
1 B1-1 lam( 4): 0.6394 0.0954 0.4523 - 0.8264
2 B1-2 lam( 2): 1.5744 0.3142 0.9585 - 2.1902
3 B1-3 lam( 2): 1.1301 0.1561 0.8241 - 1.4362
6 B1-6 lam( 2): 1.4188 0.3013 0.8282 - 2.0094
7 B1-7 lam( 2): 1.1301 0.1561 0.8241 - 1.4362
9 B1-9 lam( 2): 1.5744 0.3142 0.9585 - 2.1902
10 B1-10 lam( 2): 1.1301 0.1561 0.8241 - 1.4362
14 B1-14 lam( 2): 1.4188 0.3013 0.8282 - 2.0094
15 B1-15 lam( 2): 1.1301 0.1561 0.8241 - 1.4362
17 B1-17 lam( 2): 1.5744 0.3142 0.9585 - 2.1902
18 B1-18 lam( 2): 1.1301 0.1561 0.8241 - 1.43
ETC- ends at 275 sites for lam(2) (out of 280 total sites).

[darryl]

PRESENCE is set up so that you get estimates of all sites for a real parameter if that parameter has a covariate in it. If there is no covariate, only a single value is reported for each season (which applies to all sites). For the derived parameters (like lambda) PRESENCE does things a little bit differently to try and avoid overly-long output. It only reports a value if it is different from the last one it reported, so in year 2 there must be a few double ups (consecutive sites that have the same lambda value) that don't get reported. The reason you only get lamdas for year 3 and 4 for site 1, is that all the other sites have the same lambda value.

If this is puzzling, then think about the model you're trying to fit here. By using the constraint eps=1-gam, then gamma1, gamma2 and gamma 3 are actually psi2, psi3, and psi4. So you're fitting a model where first year occupancy varies across sites according to some covariate, but it doesn't for years 2-4. Under this parametrisation, there's no 'memory' in the system so the variation in the first year, doesn't get carried through to later years.

Hope this helps
Darryl