Which statistical method to use?

We have recently done an experiment on the affect of grazing pressure on plant height and coverage, but now I'm going round in circles trying to work out which statistical method I should be using to analyse the data.

We had four groups, each of which had 9 trays: 3 trays with no caterpillers, 3 trays containing 2 caterpillars, and 3 trays containing 5 caterpillers. Each tray contained 4 plant species and the height and coverage for each species was measured 3 times at one-week intervals.

Therefore for the trays with 0 caterpillars i have 12 measurements of height and coverage for each type of plant for each of the three week's sampling. Likewise for the trays with 2 and 5 caterpillars.

1. Is it reasonable to assume that I should use the mean of the data for each type of tray before doing any statistical analysis?
2. Am i right in thinking that the data is non-parametric?
3. What would be the best statistical analysis to perform and, most importantly, why?
4. As we only have three measurements over time, how reliable can i expect any statistical analysis to be?

Many thanks,