A regression approach to identify discriminating locations

Selection of planting locations is an important issue in plant breeding and may significantly impact the probability correctly identifying desired genotype based on noisy data. This especially true for early-stage experimentation where a set experimental genotypes observed very few locations. study aims to identify that are most helpful discriminating main effects two or more genotypes. A regression-based model proposed characterizes sensitivity location effect, with significant positive termed discriminative. Comparing ranking using these results exaggerated expected differences between effects, allowing accurate comparisons presence noise. Based model, expression making correct selection derived given known sensitivities. The then used characterize optimal maximizes selecting better effect. These independent being compared only depend number considered sensitivities effects. theoretical implications validated data obtained from commercial soybean program. show by made 76% time, versus 67% all 63% time non-discriminating