repeated measures subject with n>1?

#1
I am analyzing a repeated-measures dataset in SAS using proc mixed, and I'm stumped about the repeated measures subject. Here's the design of the experiment: I have forest plots, each containing 100 trees, and randomly assigned to contain 1, 2, 4, or 8 different genetic clones per plot ("genetic diversity"). The plots are replicated 1-3 times each, giving a total of 49 plots (4900 trees). Individual trees were marked and measured 6 different years to follow growth, and I want to test how genetic diversity affected plant growth over time. So, time is the repeated factor, genetic diversity is my independent factor, and height is the dependent factor. I'm struggling to decide wither the repeated subject is plot given that the genetic diversity treatment is assigned at the level of plot, or whether tree nested within plot should be the repeated subject. If I use plot (which is what seems most appropriate), can I include all the trees sampled within plots (in other words, can SAS handle n>1 within the repeated subject), or do I need to generate plot level means as my data set (n=1 for the repeated subject)? I have run the models both ways and get different answers, and I'm not sure what's best.

Anyone have any ideas??
Thank you!