Paper Details

Abstract

Cultivar-mixture studies often rank mixtures relative to their component pure stands, whereas a deployment decision may instead seek the treatment with the greatest absolute grain yield. We asked whether those objectives identified the same treatment when only earlier field environments were available. In a secondary decision analysis of a randomized Swiss winter-wheat experiment conducted at three sites during 2021–2023, the candidate menu contained eight pure cultivars, all 28 binary mixtures and one eight-cultivar mixture. Using only 2021, the pure-only and full-menu absolute-yield rules both selected Colmetta, whereas the parent-referenced rule selected Campanile × Molinera. After those choices were frozen, 2022–2023 grain yield averaged 77.77 and 67.32 dt ha−1, respectively, giving R−P = −10.45 dt ha−1 (−13.4%). All six site-year contrasts were negative, and the conditional environment-resampling interval from the analysis plan frozen before validation was −12.31 to −7.95 dt ha−1. A post-result block-level check found 17 of 18 randomized-block contrasts negative (exact two-sided randomization p = 9.9 × 10−5). A 2023-only chronological sensitivity retained the direction, and the two frozen sensitivity analyses did not change the parent-referenced selection. The selected mixture nevertheless had higher protein, Zeleny sedimentation value and hectoliter weight, indicating a trade-off rather than general mixture inferiority. Supporting rice and maize archives showed the same criterion-dependent selection pattern but variable later effects. Thus, a parent-referenced biological metric can remain useful for mixture assessment while being misaligned with a later absolute-yield deployment objective.

To Cite This Article