For the forum:
I would greatly appreciate advice for my survival analysis of trumpeter swan translocation project. I will give you some basic information. It’s gets a bit complicated.
There were 4 years of releases. They caught 78-102 birds each year. For every year they had a control group and a translocated group. They would put equal amount of birds in each group. For the first year they only used neck bands on the birds without radios. The second year they did half of the control and translocated group with neck bands equipped with a radio for telemetry. The third and fourth year also followed this procedure. My resight rate was about 50% each month. It is also possible to lose track of the radioed birds because the birds can fly 200 miles overnight and the radios had a 12 mile range at best. Also each year they used a different radio trying to improve the range. The covariates I am using are condition of the bird, if the bird was alone a majority of the time, parasite load, and weather.
My dilemma is that you could detect mortalities for the radioed birds (control and translocated) and not the non-radioed birds. There was one year were we did find 3 dead non-radioed translocated birds because the poacher was kind enough to leave all the bodies in one area and 2 radioed birds were also poached. This was the last year and the one that I have tried to analyze. I did Cormack Jolly Seber because I was not able to keep track of all of the birds. It gave me a 100% survival for the radioed control birds for all of the top models even though I know 3 birds died. I tried known fate and it gave me a 100% survival for the non-radioed control birds, which I don’t believe is true especially with a SE of 0.
It has been suggested that I do CJS for the non-radioed birds and do known fate for the radioed birds. Assuming that this is the best way to go (unless someone has a better suggestion), should I put the different years for the radioed birds in together where each year is a different group in one dbf file and then do the same for the non-radioed birds? Or run each year in separate models with the control and translocated groups together? Is it okay that for the one year I ran all 4 groups (radioed and non-radioed for the control and translocated group) together in CJS and known fate or do I need to rerun them separating out the radioed and non-radioed birds? I am also debating if I should keep the poached bird in the model or remove them. I want to look at survival with all the factors that could affect them but I didn’t know if it would confound looking at the covariates.
My last question if I look at each year separately and want o compare survival do I need to keep the time period for the data entry the same – I have played with looking at the data monthly and tri-weekly.
Feel free to answer any or all of the questions
I need all the help I can get to finish the ol’ masters.
Cheers
Darlene Kilpatrick