中国长江流域冬麦区小麦品种试验精确度回溯分析

为全面评价长江流域小麦品种区域试验精确度的发展水平，分析了2011—2021年度中国长江流域冬麦区单年单点和一年多点小麦区域试验的试验精确度和品种比较精确度，对产量、穗数、穗粒数、千粒重、生育期、株高、基本苗和最高茎数等性状的试验精确度差异进行了比较。结果表明，长江流域冬麦区单年单点品种试验精确度较好，试验误差变异系数(CEV)在5%和10%以下的试点数占比分别在60%和95%以上，但品种比较精确度(RLSD_0.05)在3%以下的试验比例不到10%，长江上游和中下游单点品种试验RLSD_0.05在10%以下的试点数分别占比约60%和90%。一年多点区域试验精确度和品种比较精确度显著优于单年单点试验，上游和中下游试验平均CEV分别在8%和5%以下。采用试点固定模型时，长江上游和中下游试验平均RLSD_0.05分别为2.19%和1.49%，符合国家小麦品种审定标准对品种比较精确度的要求；但采用试点随机模型时，小麦品种试验尚无法鉴别出品种间3%的产量差异。试验对生育期、株高、基本苗、千粒重和穗数的RLSD_0.05优于产量性状，而最高茎数和穗粒数的RLSD_0.05较低。

Retrospective Analysis of Experiment Precision of the Regional Yield Trials for Winter Wheat Varieties in the Yangtze River Valley

The precision of crop variety regional trials is the foundation of candidate variety evaluation and accurate selection. This study analyzed the experiment precision (EP) and variety comparison precision (VCP) of single and multiple-location wheat regional trials in the winter wheat region of the Yangtze River Valley (YRV) from 2011 to 2021, and compared EP andVCP of multiple traits including grain yield, spike number, kernel number per spike, 1 000-kernel weight, growth duration, plant height, basic seedling number, and maximum stem and tillers. Results showed that: EP of the single wheat trials in the winter wheat region of YRV performed well enough. The proportion of single trials with the coefficient of error variation (CEV) less than 5% and 10% was above 60% and 95%, respectively. However, the proportion of single trials with the relative least significant difference at 0.05 significant level (RLSD_0.05) below 3% is less than 10%, and those with RLSD_0.05 below 10% in the upper, and middle and lower reaches of YRV accounted only for about 60% and 90%, respectively.EP and VCP of one-year multi-location wheat trials were significantly better than those of single trials. The average CEV of multi-location trials in one year in the upper, and middle and lower reaches of YRV was below 8% and 5%, respectively. Under the site fixed ANOVA model, the average RLSD_0.05 in the upper reaches and the middle-lower reaches of YRV were 2.19% and 1.49%, respectively, which met the requirements of the national wheat registration standards in terms of variety comparison accuracy. Nevertheless, even the multi-location wheat trials in YRV were not sensitive enough to identify 3% yield difference among varieties under site random ANOVA mode.The estimated RLSD_0.05 values of growth duration, plant height, basic seedling number, 1 000-kernel weight, and spike number were found better than that of grain yield, while RLSD_0.05 of maximum stem and tillers, and kernel number per spike was relatively lower.