www.phidot.org

Hi All,

I've tried to use RMark to fit Burnham models with individual covariates to my data and it works well when the covariate is used to model only one parameter [eg. S(~weight)p(~1)r(~1)F(~1)]. Yet when the covariate is used to model two or more parameters [eg S(~weight)p(~1)r(~weight)F(~1)] or when two or more covariates are used to model one parameter [eg S(~weight+temperature)p(~1)r(~1)F(~1)] I get the following error message.

Error in real.vcv[i, ] = readBin(z, numeric(), nlogit) :
replacement has length zero

However, the models seem to be successfully fit in MARK since an output with the estimates is generated in the directory. I was interested in overcoming this error in order to extract the output in R since this would help me a lot in expediting my analyses. Has anyone encountered a similar problem and can provide me with any suggestions? Thanks a lot!

Edu

This is hopefully due to having incompatible RMark and MARK versions. Last October, there was a change in MARK with the storage of the binary file (*.vcv). I made subsequent changes to the code in RMark. Please upgrade to 1.8.8 for RMark and newest MARK version. If you currenly have the newest versions then contact me offlist and we'll work out the problems. If upgrading works, please post a resolution to the list.

thanks --jeff

Hi Jeff,

Sorry for taking so long to reply. I thought I was gonna receive a response to my post in my e-mail and only today it hit me that it hadn't been sent to my e-mail but I could find a reply in the forum website. I just did what you suggested me to do and it worked just fine! Thanks you very much! It will make things much easier and faster for me now.

Cheers,

Eduardo