(no subject)
Previous survival analysis models were run using either the SURVIVE or MARK programs. For this paper it was decided to run the models in SAS to be able to manipulate the survival and recovery probabilities better. The first task was then to translate the previous survival programs into SAS code. A ‘skeleton’ program was devised for the main model and then survival and recovery parameters were defined as macros. These macros could then change depending on how the survival and recovery were defined for each model. In order to keep the estimates for survival and recovery within probability boundaries (0 ≤ prob ≤ 1), the macros defined each parameter as 1 / 1 + efunction. This way, as the function defining the parameter increases positively, the probability goes towards 0, and as the function defining the parameter increases negatively, the probability goes towards 1. There are two functions used to define the survival and recovery parameters. In the time-based models the function is a0 + a1*&year, where &year is the recovery year. This function is used in the macro in three different manners: when the parameter is constant the function simply becomes a0; when the parameter follows a linear trend with time the function remains a0 + a1*&year; and when the parameter has a first-year value and then remains constant the function becomes a0 + a1*&year=1. In the size-based models the function is based on the Von Bertalanffy growth model and is a0 + a1*(Linf - (Linf- rellen)*exp(-k*&time)), where Linf = 88.7550 (found using PROC NLIN in SAS), rellen is the length upon release, k = 0.2603 (found using PROC NLIN in SAS), and &time is the number of years that the animals were free. The function then becomes a0 when the parameter is constant, a0 + a1*(Linf - (Linf- rellen)*exp(-k*&time)) when the parameter follows a linear trend, or a0 + a1*(Linf - (Linf- rellen)*exp(-k*&time=0)) when the parameter has a first-year then constant probability. The ‘skeleton’ program can be found in Appendix A and macros for each model can be found in Appendix B.