Pradel Lambda and Groups

[npseudacris]

Hi:

In going through chapers 13 and 14 of the Gentle Introduction, I had a question about interpreting models where there is group dependance on lambda. The top of page 14-37 in reference to sensible model comparisons states "Because the lambda values are population growth on a per capita basis and include both survival and recruitment, models with group - and time - dependence in lambda may not be indicitive of changes in recruitment between the two groups."

Does this mean one has to be careful in comparing fit between models with and without group dependence because of differences in the parameters that are able to be estimated or does it refer to the real parameter estimates (those that can be) and whether they are accurate for each group or am I missing the point altogether.

My best model has group and time dependence on lambda, phi, and p. Excluding parameters that are confounded and cannot be estimated, I want to have Lambda estimates for each group for each year but based on the above I am a little confused on if I can interpret it that way.

Thanks,
Nicole

[cooch]

Hi:

In going through chapers 13 and 14 of the Gentle Introduction, I had a question about interpreting models where there is group dependance on lambda. The top of page 14-37 in reference to sensible model comparisons states "Because the lambda values are population growth on a per capita basis and include both survival and recruitment, models with group - and time - dependence in lambda may not be indicitive of changes in recruitment between the two groups."

Means what it says. Lambda is an omnibus measure reflecting the combined influences of survival, and recruitment. As such, variation in lambda could reflect differences in survival, or recruitment, or both. This is analogous to differences in recovery rates in dead recovery models (where the differences could be kill rate, retrieval rate, or reporting rate, or any combination thereof).

You *can* parition sources of variation in lambda though - as discussed in various places in Chapter 13 and Chapter 14. I'd suggest having a look at the Nichols et al. Ecology paper referred to in the last few pages of chapter 13, and the relevant sections of the Williams, Nichols and Conroy text.

Does this mean one has to be careful in comparing fit between models with and without group dependence because of differences in the parameters that are able to be estimated or does it refer to the real parameter estimates (those that can be) and whether they are accurate for each group or am I missing the point altogether.

My best model has group and time dependence on lambda, phi, and p. Excluding parameters that are confounded and cannot be estimated, I want to have Lambda estimates for each group for each year but based on the above I am a little confused on if I can interpret it that way.

Thanks,
Nicole[/quote]