我国棉花品种区域试验精确度的演变分析

Evolution Analysis of the Precision of Regional Cotton Variety Trials in China

农作物区域试验精确度分析是品种和试验环境科学评价的基础。本研究分析了2000―2014年期间长江流域、黄河流域和西北内陆棉区国家棉花区试的试验精确度分布与演变动态,并分析比较了棉花12个主要性状的精确度差异,旨在全面分析和评价我国棉花品种区域试验精确度的发展水平和演变规律,为全国棉花品种区域试验的合理布局和优化设计提供参考依据。研究结果表明:(1)单年单点棉花区域试验能鉴别出5%、8%、10%、12%、15%和20%品种间差异的比率分别约为20%、50%、70%、83%、92%和98%。(2)单年多点棉花区域试验精确度呈逐年提高的演变趋势,RLSD0.05(RLSD,relative least significant difference)由2000年的11.43%下降到2014年的7.1%,其中长江流域的RLSD0.05已经连续9年、黄河流域棉区的RLSD0.05连续4年稳定在5%以下,可以稳定地鉴别品种间5%的差异;西北内陆棉区的RLSD0.05也呈逐年下降趋势,目前可以可靠鉴别品种间10%的差异。(3)棉花12个主要性状的试验精确度差异显著,其中皮棉产量和结铃数的精确度较低,霜前花率、衣分、纤维长度和生育期的精确度较高,其余性状精确度中等。

Evaluation of experimental precision is the foundation of scientific assessment to candidate varieties and test environ- ments in regional crop trials. The distribution of one-year-one-site test precisions, the evolution dynamics of one-year-multi-loca- tion trials and the accuracy comparison among 12 major cotton characteristics were implemented to explore and evaluate the de- velopmental status and evolution trends of the experiment precision in national cotton regional trials, and thus to provide a theo- retical guidance for the reasonable layout and optimal design of national cotton regional trials＇ scheme, by using the datasets collected from the national cotton regional trials in the Yangtze Valley, Yellow River Valley and the Northwest Inland cotton planting regions from 2000 to 2014. Research results showed that： （1） the percentage of one-year-one-site tests, those could dis- criminate the differences among cultivars up to 5%, 8%, 10%, 12%, 15% and 20%, were about 20%, 50%, 70%, 83%, 92% and 98%, respectively. （2） The precision evolution trend of one-year-multi-location trial increased year by year. The relative least significant difference at 0.05 level （RLSDo.os） declined from 11.43% in 2000 to 7.1% in 2014 in nationwide, among of which RLSDo.oswas continuously lower than 5% in recent successive 9 years in the Yangtze Valley, and 4 years in the Yellow River Valley, so that the trials in the two regions were accurate enough to reliably identify the 5% difference among cultivars. Howev- er, national cotton regional trials in the Northwest Inland region could reliably discriminate about 10% difference currently, al- though RLSDo.o5 value was also on the decline these years. （3） The experiment precision of 12 major cotton characteristics were significantly differenct, in which lint cotton yield and bolls per plant were lower in test precision, yield ratio before frost, lint per- cent, fiber length and growth period were of relativly high precision, while the rest of the traits were of medium accuracy.