Abstract The complex and interactive effects of genotype (G), environment (E), and management (M) can be a barrier to the development of sound agronomic recommendations. We hypothesize that long-term variety performance trials (VPT) can be used to understand these effects and improve regional recommendations. Our objective was to explore long-term VPT data to improve management and variety-selection recommendations using winter wheat (Triticum aestivum L.) in the U.S. central Great Plains as a case-study. Data of grain yield, variety, and trial management were collected from 748 wheat VPT conducted in the states of Colorado, Kansas, and Oklahoma over nineteen harvest years (2000–2018) and 92 locations, resulting in 97,996 yield observations. Using 30-yr cumulative annual precipitation and growing degrees days, we partitioned the study region into 11 contiguous sub-regions, which we refer to as growing adaptation regions (GAR). We used variance component analysis, gradient boosted trees, and conditional inference trees to explore the management and variety trait effects within each GAR. For the variety trait analysis, the VPT dataset was reduced to account for varieties for which 17 agronomic traits and 11 disease/insect reaction ratings were available (65,264 yield observations). GAR accounted for 46 % of the total variation in grain yield, M for 32 %, residuals (including interactions) for 13 %, year for 7 %, and G for 2 %. Conditional inference trees identified interactions among management practices and their effects on yield within each GAR. For instance, water regime was the most important practice influencing wheat yield in the semi-arid western portion of the study region, followed by sowing date and fungicide. In dryland trials, there was typically an interaction between fungicide, sowing date, and tillage system, depending on GAR. Other management practices (e.g. dual-purpose management, crop rotation, and tillage practice) also significantly affected yield, depending on GAR. The main variety trait associated with increased yields depended on region and management combination. For instance, drought tolerance was the most important trait in dryland trials while stripe rust tolerance was more relevant in irrigated trials in the semi-arid region. In this research, we demonstrated an approach that uses widely available long-term VPT data to improve management and variety selection recommendations and can be used in other regions and crops for which long-term VPT data are available.

中文翻译：

---

探索长期品种性能试验以改善环境特定基因型 × 管理建议：冬小麦案例研究

摘要 基因型 (G)、环境 (E) 和管理 (M) 的复杂交互影响可能成为制定合理农艺推荐的障碍。我们假设可以使用长期品种性能试验 (VPT) 来了解这些影响并改进区域建议。我们的目标是使用美国大平原中部的冬小麦 (Triticum aestivum L.) 作为案例研究，探索长期 VPT 数据，以改进管理和品种选择建议。从在科罗拉多州、堪萨斯州和俄克拉荷马州 19 个收获年（2000-2018 年）和 92 个地点进行的 748 次小麦 VPT 收集粮食产量、品种和试验管理数据，产生 97,996 次产量观察。使用 30 年累积年降水量和生长度日，我们将研究区域划分为 11 个连续的子区域，我们将其称为不断增长的适应区域 (GAR)。我们使用方差分量分析、梯度提升树和条件推理树来探索每个 GAR 内的管理和品种特征效应。对于品种性状分析，减少了 VPT 数据集以考虑可提供 17 个农艺性状和 11 个疾病/昆虫反应评级（65,264 个产量观察值）的品种。GAR 占谷物产量总变异的 46%，M 占 32%，残差（包括相互作用）占 13%，年份占 7%，G 占 2%。条件推理树确定了管理实践之间的相互作用及其对每个 GAR 内产量的影响。例如，在研究区域的半干旱西部地区，水分管理是影响小麦产量的最重要的措施，其次是播种日期和杀菌剂。在旱地试验中，杀菌剂、播种日期和耕作系统之间通常存在相互作用，具体取决于 GAR。其他管理实践（例如双重目的管理、作物轮作和耕作实践）也显着影响了产量，这取决于 GAR。与产量增加相关的主要品种性状取决于地区和管理组合。例如，耐旱性是旱地试验中最重要的性状，而条锈病耐受性在半干旱地区的灌溉试验中更为重要。在这项研究中，