Fisher's LSD in Systat10

#1
I have a multivariate data as detailed following. There are 3 group of plants; each group has five plants (biological replicates). I have quantities of 20 compounds for all 15 plants. For each of the twenty compounds I want to see if the mean for each of the three groups is different from that of other. I don't know much of the statistics but I guess LSD test (Fisher's) would be most appropriate to apply in this case. Can anyone comment also on in which other ways can such data be analyzed statistically ? (I am already done with standard deviation :cool:)

I have Systat10. Can anyone comment on is what I am doing correct? (see below)
For doing Fisher's LSD, I would go to Statistics>General Linear model>Estimate model. There I will select the variables (compounds, in my case). What information should I look in the output after this? Why we should "Estimate mode" first and not directly do the LSD ?
Then I will go to Statistics>General Linear model>Pairwise comparison. Here I am asked to enter something in the "Group" field. What is this??
Where exactly in the output can I find if the means are equal or not after selecting Fisher's LSD and clicking OK?
Thanks!