I'm fairly certain this has been posted before but when I did a search I couldn't find it. In working with an individual offlist, it became apparent that the newest version of MARK 8.1 is failing if the individual covariate name is > 10 characters. I'm using June 2014 version of MARK and that is not happening. I guess I should update.
Anyhow, anytime you see the following:
STOP ERROR
MARK did not run properly. If error message was not shown, re-run MARK with invisible=FALSE
Model output is not available
You should check the most recent .out file in your working directory. RMark is telling you that something is wrong with the .out file that MARK created such that it can't find the values it expects. Any R errors after that point are not relevant. I'm going to change the message so it specifically says to check the .out file. In this case the .out ended with
INPUT --- design matrix constraints=10 covariates=11;
INPUT --- 1 allgrass400 0 0 0 0 0 0 0 0 0;
ERROR -- Error with floating point read of ALLGRASS400
MARK gave the error while reading in the design matrix. MARK doesn't allow very long variable names. When the user changed the variable name to have fewer than 10 characters the problem went away. I thought I had a check in the RMark code for this but if I do it is not working. I'll check and add it to avoid this issue. Regardless always use individual covariate names with <10 characters. Variable names in the design data can be of any length because R doesn't have this restriction and they are used in model.matrix to create the DM and the names don't appear in the DM.
regards --jeff