Does SAS have a procedure that allows one to analyze the following?

I have a nested-factorial design (Ponds within nested within herbivory, ponds crossed with plant type. Each pondXplant has 3 replicate plots. Two species of each plant are planted within each plot). Within each plot, 3 measures will be taken, these measures are of an ordinal scale (0 to 7, roughly corresponding to the log-odds). Each measure will describe the proportion of a specific plant (or open water) within each plot. For example, within each plot there are 2 species of plant and open water (3 categories), if each occupied 33% of the plot, we would record 3, 3, 3 for each category within that plot. Each plot will be measured each quarter for 2 years, suggesting a repeated measures component. The questions of interest relate to species composition within plots through time based on competition, and treatment differences associated with herbivory.

If I had a single measure for each plot, I would think in terms of Proc MIXED, but I have 3 measures per plot per pond per time. The relatedness of the measures within the plot would seem to negate MIXED, but may not if someone could advise me. Because I have mutually exclusive and exhaustive categories within each plot, I thought maybe CATMOD would be the procedure to use, but my measures are not multinomial, they are ordinal. I could use GLM, but am concerned because it appears I have less ability to alter the covariance matrix.

Any suggested directions would be appreciated.

Warren Schlechte Heart of the Hills Research Station HC7 Box 62 Ingram, Texas 78025 Phone: 830.866.3356 Fax: 830.866.3549