I need to conduct a pesticide efficacy trial on avocadoes. I have been asked two describe two different trial approaches: replicated and not replicated within field.
Option 1: strip plot trial. 3 treatments, no replication. Repeated across 3/4 locations. 500-1000 fruit per treatment assessed for pest. Grower can apply treatments, so much cheaper.
Option 2: small plot replicated trial. 3/4 treatments, 10 reps, 2/3 trees per plot. Repeated at 2 locations. 50-100 fruit per plot assessed for pest. Treatments applied with motor and handgun, so labour intensive.
Is it possible to get good data from Option 1? If it were 2 treatments I could use a unpaired t test to analyse. But since there is no treatment replication (i.e. treatments applied to one experimental unit only (1/3 block)) I cannot use ANOVA? Am i right to think that the trees cannot be viewed as replications? Can I do multiple t tests with bonferroni correction? For example, 3 treatments = 0.05/3 = 0.017.
Can I run Option 1 as an ANOVA if I repeat trial at 3/4 locations, so locations are my replications? If so, can I use any location or do they need to be in close proximity to one another?
Thanks in advance for the help!