Help with data on swans

[AndyKelly]

Hi all,

This is my first posting. I am hoping that someboday can give me some advice on how to proceed with MARK to analyse data on post-release survival of orphaned mute swans compared to those ringed and released on or near the nest.

Between 1999 and2008, we released 290 marked (ringed) swans and between 1999 and 2008 we have resightings of 154 and dead recoveries for 50. For the re-sighted birds, resighting events range from 1 to 10 (and for the dead birds we have some resighting data prior to them being found dead). We did not re-capture the birds.

Data for birds ringed at the nest will be available for comparison. So essentially I have 2 groups (rehabilitated orphans and ringed on nest). I want to compare the survival for the two groups.

I have read many of the 'help' documents available but really am no further forward. There just seems to be too many options. I think I need to use a Burnham model, but what aabout Brownie parameterisation and 'contrstaints'. All very confusing.

Can anybody help?

Thanks
AndyKelly

[cooch]

Hi all,

This is my first posting. I am hoping that someboday can give me some advice on how to proceed with MARK to analyse data on post-release survival of orphaned mute swans compared to those ringed and released on or near the nest.

Between 1999 and2008, we released 290 marked (ringed) swans and between 1999 and 2008 we have resightings of 154 and dead recoveries for 50. For the re-sighted birds, resighting events range from 1 to 10 (and for the dead birds we have some resighting data prior to them being found dead). We did not re-capture the birds.

Data for birds ringed at the nest will be available for comparison. So essentially I have 2 groups (rehabilitated orphans and ringed on nest). I want to compare the survival for the two groups.

I have read many of the 'help' documents available but really am no further forward. There just seems to be too many options. I think I need to use a Burnham model, but what aabout Brownie parameterisation and 'contrstaints'. All very confusing.

Can anybody help?

Thanks
AndyKelly

Sorry - there is no reasonable substitute for actually learning how to use MARK. You can't simply 'pick a model' and fit your data. You are strongly urged to read/study Chapters 1 -> 7 of 'the book':

http://www.phidot.org/software/mark/docs/book/

Then, have a look at the dead recoveries chapter, which you'd need to understand before looking at Burnham models.

[AndyKelly]

Thanks for the advice - I have read Chs 1 - 7 and I am no further forward. I am not a statistician and the book assumes a knowledge that occassional users like me will not have. I want to use MARK as a tool to answer a question. Do rehabilitated orphaned mute swans show survival that is comparable to swans ringed and released on the nest? Unfortunately in my job I do not have the time to spend getting to grips with MARK and hoped to get some advice - that's what I though the forum was for. The problem is that I will have nobody to confirm that what I do is right.

I'll go back to plan B which was to get somebody else to do the analysis.

[cooch]

Thanks for the advice - I have read Chs 1 - 7 and I am no further forward. I am not a statistician and the book assumes a knowledge that occassional users like me will not have. I want to use MARK as a tool to answer a question. Do rehabilitated orphaned mute swans show survival that is comparable to swans ringed and released on the nest? Unfortunately in my job I do not have the time to spend getting to grips with MARK and hoped to get some advice - that's what I though the forum was for. The problem is that I will have nobody to confirm that what I do is right.

I'll go back to plan B which was to get somebody else to do the analysis.

Plan B then. The forum is designed to help people who will be using MARK to do the analyses - not to (in effect) do the analyses for you.

[dhewitt]

AndyKelly,

This sounds like a promising data set, and the question is straightforward and intriguing. I'd venture a guess that you can get a decent answer with your data. I'd encourage you to stick with this and, in the process, learn some of the power of MARK for capture-recapture analysis.

In order to help, we need a little more information about the study. Specifically,

(1) how many swans in each group (nest, rehabbed) were ringed in each season?

(2) what are the intervals during which the resightings and dead recoveries can occur? (e.g., do you only resight when they return to breed, but dead recoveries come from all over the place?)

In any case, you were on the right track -- a Burnham or Barker model that accommodates both the resighting encounters and the dead recovery information should be the way to go.

[AndyKelly]

Thanks dhewitt

My most recent advice was to go on a workshop for a week at a cost of £900 - I work for an animal welfare charity so that was not an option.

The data for rining on the nest is still being comllated by the Ringing (Banding) Group and will be with me very soon.

Numbers of orphaned / rehabilitated swans:

1999 34
2000 37
2001 14
2002 22
2003 23
2004 24
2005 55
2006 39
2007 42
Total 290

Numbers ringed on the nest each year will be similar.

Re-sightings were ad-hoc and were over a large area as were dead recoveries. The re-sightings were basically opportunistic - swan study group members reported any sightings to us. There were no 're-capture' events.

For the 290 released, we have live resightings for 154 and dead recoveries for 50. For the live re-sightings we have between 1 and 10 resighting events.

Looking into it I may need to call the first month of release in 1999 'month 1' and then have resighting events every 3 months or so and look to see if each swan was sighted within that 3 month interval - that will allow me to sort my data into LDLD format, but I am anot sure how to deal with the different years - if all 290 had been rreleased in 1999 it would be more straightforward!

Any advice would be gratefully recieved...

Best regards
Andy Kelly

[dhewitt]

Andy,

The ad-hoc nature of the resightings is a bit troublesome. You need to be able to define an interval over which you will estimate survival. Off the bat, with numbers of releases like you have, my guess is that you cannot estimate a survival rate over 3-month periods like you suggest. You'd like to have a nice short window when you are resighting birds (perhaps at the nesting area) that will act as "recaptures" and then a much longer interval between when more resightings and dead recoveries can occur. Do you have a sense of the intervals you could define, or the distribution of the resightings and recoveries throughout the years?

- Dave