Design and analysis of unreplicated plant breeding trials

点击翻译按钮获取中文摘要

Early generation plant breeding trials are often laid out according to unreplicated designs. Replicated checks may be used for error control, for example in augmented designs based on an incomplete block design for checks which are augmented with unreplicated entries. Traditional augmented designs require considerable resources to be spent on genotypes that are not themselves of interest. Therefore, it has been suggested to replace checks with partially replicated entries, leading to so-called prep designs. In the present proposal we suggest combining both ideas to develop what we call augmented prep designs. A non-trivial design problem arises when trials are to be performed at multiple locations. The main challenge then is how to augment the blocks so as to balance the number of pairwise concurrences. This task can be tackled in different ways based on the use (-arrays which are well-known as the basis for (-designs in fully replicated experiments. Furthermore, we also explore the refinement of designs when analysis by spatial models is envisioned. Robustness of the designs to the presence of genotype-environment interaction is also investigated.