Beta estimates for Pradel model

[sixtystrat]

I am evaluating a count index by analyzing capture-recapture data and using the index as a covariate to see if the model is improved over the global model. As such, I need to estimate beta to describe how the index and population growth (lambda) relate to one another. I am using the Pradel (lambda) model and a process described by Nichols and Hines (2002). When I try to run the best model with the identity matrix, I get non-convergence. How do I "un-transform" those beta values by hand? It says in MARK that the original beta values are based on the logit link but in the help file it says they are based on the log link. At any rate, I am unsure how to un-transform the betas and estimate the SEs. The method shown in Cooch and White doesn't seem to work with my model structure. Any help would be appreciated!!!
Joe Clark
USGS, Knoxville, TN