|Maas, Tisha - UNIVERSITY OF NEBRASKA|
|Marx, David - UNIVERSITY OF NEBRASKA|
Submitted to: Applied Statistics In Agriculture Conference Proceedings
Publication Type: Proceedings
Publication Acceptance Date: December 16, 2002
Publication Date: March 5, 2003
Citation: MAAS, T., MARX, D., PEDERSEN, J.F. UNREPLICATED SPATIAL DESIGNS COMPARED USING OPTIMALITY CRITERIA. APPLIED STATISTICS IN AGRICULTURE CONFERENCE PROCEEDINGS. 2003. Interpretive Summary: Germplasm collections for many crop species may contain thousands of lines. While having large numbers of resources is desirable, the task of searching through (or screening) thousands of lines for unique traits can become difficult in practice because of the large numbers. Researchers are aware of natural variability in fields that can influence their ability to measure unique traits and account for that variability by repeating or replicating experiments. However, in large germplasm screening endeavors it is often impossible (due to small amounts of seed) or impractical (due to the large numbers involved) to replicate entire experiments. In such cases, a single line (or check) may be replicated. This experiment examined five different spatial arrangements of checks using computer-based optimality criteria and actual field trials. The results showed that the optimality criteria were not necessarily comparable to what was actually observed in the field, indicating need for further refinement of field designs for such applications.
Technical Abstract: In large variety field trials, it is often impossible or impractical to replicate each variety. In these situations, the researcher may choose to use only one replicate of each test variety and to include a ¿check¿ variety every so often so that the spatial variability of the field may be determined. Five different check patterns were purposefully designed, each possessing distinct characteristics. The purpose of this study is to determine which spatial patterns for the check variety are more able to identify the spatial structure in a field and to rank the experimental varieties accurately. The problem was approached in two ways. First, the check patterns were compared using optimality criteria. Then, the patterns were applied to an actual field experiment, and the data collected was used to identify the spatial structure of variation in the field and to test for experimental variety differences. It is shown that the results from the optimality criteria were not necessarily comparable to what was actually observed in the field.