survival results ?

[branta]

Hi,
I am analyzing 4 years of band recoveries. Group 1 is adults and group 2 is young. The best model I ran turns out to be Brownie et al. (1985) H02 model (i.e., survival rates constant year to year and recovery rates year specific, and rates are dependent for the first year of life only). I have ran my numbers multiple times and keep getting the same results. My adult survival (1:s = 0.523) is lower than the young survival (2:s = 0.67). However, direct recovery rates for the young are greater; therefore, how can survival for the adults be less? I have read the entire "A Gentle Introduction" and am not quite sure what the problem is. Below are the numbers I have been using. Any advise would be greatly appreciated. Thanks.
recovery matrix group = 1;
42 23 14 10 ;
40 23 20 ;
75 45 ;
80 ;
316 235 504 461;
recovery matrix group = 2;
107 85 30 16 ;
98 58 38;
77 71;
129;
694 521 545 563;

[ganghis]

I suspect two processes might be operative here. First, the recovery model assume that tag recoveries come from anywhere; that is, no permanent emigration from the population at risk of recovery. In this case, survival (S) should really be interpreted as apparent survival (phi). Is there any evidence to suggest that adults disproportionately emigrate?

Second, the recovery parameter in the Brownie models (f) should be interpreted as the probability that an individual is reported given it was alive immediately prior to the survival interval. Thus it is a conglomeration of processes of survival, harvest, and reporting (assuming harvest is what generates your data). At least in waterfowl species, there is alot of evidence to suggest that young birds are more vulnerable to the gun. Thus they often have higher hunting mortality than adults. However, if natural mortality was higher in adults, this might also explain the pattern you are seeing.

[cooch]

Hi,
I am analyzing 4 years of band recoveries. Group 1 is adults and group 2 is young. The best model I ran turns out to be Brownie et al. (1985) H02 model (i.e., survival rates constant year to year and recovery rates year specific, and rates are dependent for the first year of life only). I have ran my numbers multiple times and keep getting the same results. My adult survival (1:s = 0.523) is lower than the young survival (2:s = 0.67). However, direct recovery rates for the young are greater; therefore, how can survival for the adults be less? I have read the entire "A Gentle Introduction" and am not quite sure what the problem is. Below are the numbers I have been using. Any advise would be greatly appreciated. Thanks.
recovery matrix group = 1;
42 23 14 10 ;
40 23 20 ;
75 45 ;
80 ;
316 235 504 461;
recovery matrix group = 2;
107 85 30 16 ;
98 58 38;
77 71;
129;
694 521 545 563;

First, resist the temptation to use the Brownie nomenclature for model naming - its easier (for all concerned) if you use the somewhat stnadard linear models notation to describe your models. Most of us would have no recollection whatsoever about what Model H02 might be.

Second, I did a very quick analysis of your recovery matrices (very quick - I fit 6 or 7 models), and your suggestion that direct recovery rates for young are 'greater' than for adults is not particularly strongly supported by the data. I'd re-check your model fitting.

Third, as Paul notes in his reply (and as noted in several places in the 'Gentle Guide') recovery rate is the product of 3 events: getting shot, getting retrieved, and getting reported - so, part of the 'survival story' is inextricably tied up in recovery rate - but, mortality can occur for 'non-hunting' reasons also. Stare at the fate diagram on page 2. With a little thought, its not hard to come up with scenarios such as the one you describe. Interpreting variation in recovery rate in the absence of other information can sometimes be tricky.

Fourth, unless you're interested in trying to partition 'hunter' from 'natural' mortality, its worth running your data through the Seber parameterization, also described in some detail in Chapter 10. Especially read (or re-read) pp. 17-18.