Leveraging environmental covariates to predict crop variety performance in targeted environments

经常进行多环境试验，以评估待登记的候选作物品种的种植和使用价值。虽然这些试验的主要目的是品种登记，但数据也可用于向农民提出建议。品种注册的决定通常基于所有试验环境中的品种平均值，而向农民提出的建议则需要对品种性能进行更有针对性的估计。这种更有针对性的方法可以基于使用协变量信息的目标环境的表征。在这个项目中，我们将使用新开发的“环境学”的概念，以利用环境协变量的信息来预测在任何环境中的品种性能，从多环境试验的环境中选择的目标群体。该预测框架允许将多环境试验的有价值的结果投射到目标环境群体中的农民环境。

Multi-environment trials are routinely conducted in order to assess the value for cultivation and use of candidate crop varieties for registration. While the primary purpose of these trials is variety registration, the data may also be used to derive recommendations to farmers. Whereas decisions for variety registration are usually based on variety means across all trial environments, recommendations to farmers require more targeted estimates of variety performance. Such more targeted approaches can be based on a characterization of targeted environments using covariate information. In this project, we will use the newly developed "enviromics" concept in order to make use of environmental covariate information to predict variety performance in any environment picked from the target population of environments for which multi-environment trials were performed. This prediction framework allows projecting the valuable results of multi-environment trials to farmers' environments in the target population of environments.