Issues in the design of unreplicated trials
by
K.T. Stefanova
Department of Agriculture, W.A. 6151, AUSTRALIA
Coauthors: G.P.Y. Clarke

Early generation plant breeding trials involve the evaluation of a large number (greater than 150) varieties at a few trial locations. Due to seed limitations individual field trials often consist of single plots of test varieties augmented with multiple plots of standard commercial ("check") varieties arranged in a grid-like pattern.

The design of the complete series of trials involves several key issues including the layout of the check varieties and their frequency within trials and the use of different randomizations for each trial.

We present the results of simulation studies based on uniformity trials, conducted to investigate some of these issues. Appropriate mixed model analysis turns out to be an important aspect in the interpretation of the results and reflects the relative inability of some designs to allow estimation of spatial parameters.

Date received: August 30, 2001