Environmental Covariates

[kdecollibus]

Hi,

I'm trying to estimate monthly survival for house finches over the last 11 years. I have 132 encounter occasions and I would like to incorporate monthly precipitation into my analysis. I prepared the input file with the binary and number of birds with that particular encounter history like usual. I listed the 132 monthly precipitation values afterwards separated by spaces and followed by a semicolon. I got an error when I tried to run any model. The prompt said it was a point-read error of 0.02540.1778. Now, those numbers 0.02540 and 0.1778 are two of my precipitation values and they are separated by a space in the input file for sure.

What could be causing this?

Any feedback would be greatly appreciated.

Thanks.
-Karin

[jlaake]

Have you read the electronic book? Environmental/occasion specific covariates are not individual covariates. Only the latter are entered with the capture history data. Also when you enter individual covariates, they are entered on the same line as the capture history and not afterward. It was trying to read in the covariate values as capture history values. Occasion specific values are entered directly into the design matrix.

--jeff

[kdecollibus]

I had them on the same line like they were individual covariates.

Thanks again.

[jlaake]

In that case either there was an error in the input or you did not specify the correct number of individual covariates on the input line.

If the individuals experience different values of the environmental covariates then you can either use groups and specify the covariate values in the design matrix or you can treat them as individual covariates that differ across individuals. In general your models will run faster if you avoid individual covariates.

--jeff

[ELR]

Hi Karin,
Coincidentally, I just finished re-reading Chapter 7 of the MARK book this morning. Have a look, or should I say, have another look, for yourself and be sure to read the "constraining with real covariates" section. If you have only 1 measurment of precipitation for each sample period (so rainfall is the same for all individuals within a group) then you will want to add this info to the design matrix and not in the input file, as Jeff mentioned.
I am about to do the same sort of analysis myself.
Good luck to both of us!
Erin

[cooch]

In addition to previous recommendations - especially pointers to sections in Chapter 7 - you might also spend some 'quality time' reading Chapter 12 - especially section 12.6, which discusses how individual covariates can under some circumstances be used to specify group associations. There are pros and cons to using individual covariates in this manner, but understanding the basic idea is critical in deciding how to approach the problem in general.

[kdecollibus]

Thanks for all the advice!

-Karin