occuRN - output of r

[Nathan B]

From the formula p_ij = 1 - (1 - r_ij) ^ N_i, is there command for unmarked to output r, the detection probability of an individual?

Thanks,

Nathan

[Jeffrey Royle]

hi Nathan,

 in the occuRN function the natural parameterization of the model in unmarked is in terms of r. i.e., all the output is for individual level detection.

regards

andy

--
You received this message because you are subscribed to the Google Groups "unmarked" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Nathan B]

Andy: I'm not sure if you misunderstood the question or if I misunderstood your answer. I can get estimates of the coefficients, and I can backtransform if I need to get estimates of p. Furthermore I can get estimates of Ni using the ranef function. But the r_ij appears to be a low level parameter. Since it's included in the liklihood function if I look at occuRN (as r.ij), I'm wondering if it is possible for me to "see" those values. I tried editing the occuRN function by dumping ests into umfit, but it yells at me for having an invalid name (a dependencies issue I guess). I'm not even sure if that would work or if what I'm asking even makes sense. What am I missing here?
Cheers,
Nathan

[Jeffrey Royle]

hi Nathan,

 Here is the example from occuRN, follow this along here and see if this answers your question (I added some bits at the end):

> data(birds)
> woodthrushUMF <- unmarkedFrameOccu(woodthrush.bin)
> # survey occasion-specific detection probabilities
> (fm.wood.rn <- occuRN(~ obsNum ~ 1, woodthrushUMF))

Call:
occuRN(formula = ~obsNum ~ 1, data = woodthrushUMF)

Abundance:
 Estimate    SE    z P(>|z|)
    0.792 0.155 5.09 3.5e-07

Detection:
            Estimate    SE      z  P(>|z|)
(Intercept)   -1.833 0.345 -5.305 1.12e-07
obsNum2        0.427 0.429  0.994 3.20e-01
obsNum3       -0.144 0.455 -0.317 7.51e-01
obsNum4        0.463 0.424  1.093 2.74e-01
obsNum5        0.779 0.414  1.879 6.02e-02
obsNum6        0.801 0.415  1.928 5.39e-02
obsNum7        1.057 0.419  2.523 1.17e-02
obsNum8        0.805 0.415  1.938 5.26e-02
obsNum9        0.878 0.414  2.121 3.39e-02
obsNum10       0.937 0.417  2.245 2.47e-02
obsNum11       0.706 0.417  1.696 8.98e-02

AIC: 636.2771
>
> # Empirical Bayes estimates of abundance at each site
> re <- ranef(fm.wood.rn)
Warning message:
In .local(object, ...) :
  You did not specify K, the maximum value of N, so it was set to 50
> plot(re)
>

This is the part I added right here. If you use predict with type="det" then these predictions should be predictions of r_{ij}. (I hope....).   I believe there should be one row of this matrix for EACH "site x rep" combination"

 
> predict(fm.wood.rn,type="det")
    Predicted         SE      lower     upper
1   0.1378947 0.04107050 0.07515917 0.2394386
2   0.1968553 0.05101559 0.11522030 0.3156923

[Nathan B]

Ah predict!!! most excellent thank you! I was using backTransform type="det"....

Thank you!

Nathan

[Nathan B]

Ok I see what happened now. I got confused between p in occupancy and r in the RN model. doh!