Annual survival probability when S is constant

[quokkasRcool]

Hi,

I have used a multi state recaptures only procedure in MARK to analyse the survival rates of quokkas that I trapped every season over a period of two years ( 8 encounter occasions).

My best fitting model is one where survival is constant (S(.)p(age.sex)psi(.)) and the value for S has been given as 0.8451.
What I am confused about, is whether I need to cube the survival rate from the output, to get an estimate over a year, or whether I use the survival rate as given (because survival is constant).
I am just trying to figure out the Px values for my Lefkovitch matrix (time step 1 year), and was wondering if you could advise me as to whether I need to modify (multiply) the survival rate MARK has given at all.

Thanks so much

Veronica Phillips

[cooch]

Hi,

I have used a multi state recaptures only procedure in MARK to analyse the survival rates of quokkas that I trapped every season over a period of two years ( 8 encounter occasions).

My best fitting model is one where survival is constant (S(.)p(age.sex)psi(.)) and the value for S has been given as 0.8451.
What I am confused about, is whether I need to cube the survival rate from the output, to get an estimate over a year, or whether I use the survival rate as given (because survival is constant).
I am just trying to figure out the Px values for my Lefkovitch matrix (time step 1 year), and was wondering if you could advise me as to whether I need to modify (multiply) the survival rate MARK has given at all.

Thanks so much

Veronica Phillips

Well, you don't specify what a 'season' is, but I'll assume its some fraction of the 12-month calendar year. Say, every 4 months (3 samples within the calendar year). If that is the case, then yes, you simply take that estimate and raise it to the 3rd power. This assumes that your estimate is a constant. If the estimates vary of season, then the simple product of the seasonal estimates (e.g., S(Jan -> April) x S(May -> August) x S(September -> December).

If you want variances for the annual estimates, you could use the Delta method (Appendix B in the book).

[jlaake]

That assumes the default time interval of 1 was used between occasions. If fractions of a year were specified then S is already an annual rate.

[cooch]

That assumes the default time interval of 1 was used between occasions. If fractions of a year were specified then S is already an annual rate.

Good point. In that event, see the - sidebar - starting on ~ p. 24 in Chapter 4 of the book.

[quokkasRcool]

That's great. Thank you so much for your help!