Pradel phi,f models - problems with derived parameter ests.

[bglenn]

I’m using Pradel survival and recruitment models to look at effects of climate on spotted owl population dynamics. My input file contains capture history data, 4 groups, and a number of individual weather covariates.

I’m having trouble getting derived estimates for lambda when my models contain individual covariates. I get a “ No derived parameters VC matrix available” message when I click on the derived parameter button. I’m using the latest version of MARK (5.1 ), and it does give variance-covariance output for all my models so I can calculate lambda and se(lambda) using the delta method. I’m just wondering why MARK won’t do the derived parameter estimates.

I do get derived parameter output with models built using PIMs (e.g. group*time models), but I don’t get the derived parameters when I build the same model using a design matrix. Anyone have any suggestions?

Thanks!
Betsy

[gwhite]

Betsy:
Lambda is a population-level parameter, and it doesn't make sense to talk about a lambda value generated from an individual covariate. That is, you are saying that the POPULATION increased at some rate given the value of the lambda for a single individual. So, in general, it doesn't make biological sense to generate lambda estimates for models with individual covariates in them.

Gary

[bglenn]

Thanks Gary!

I figured that I was "missing the point" on something in these models. I am interested in looking at lambda estimates for models that have weather effects on survival/recruitment as well as looking at the relative contributions of survival and recruitment to lambda. It would make sense to examine models that have population-level (e.g study area level) weather covariates, wouldn't it?

Betsy

[gwhite]

Betsy:
Yes, it does make sense to model lambda as a function of temporal or spatial covariates, both of which are characteristics of the total population.
Gary