基于DUS测试的标准品种形态性状稳定性和重要性分析

为了进一步提高水稻品种特异性(Distinctness)、一致性(Uniformity)和稳定性(Stability)(简称DUS)测试的准确性，以水稻DUS测试指南中列出的49份水稻标准品种为材料，应用AMMI模型对水稻品种50个DUS测试性状进行了稳定性分析，并采用随机森林算法对测试性状的重要性进行了评价。结果表明，不同性状的稳定性差异较大，以质量性状和假质量性状为主的目测性状稳定性高，而以数量性状为主的量测性状稳定性较低。不同性状的重要性参数差异甚大，总体上，以数量形式表示的量测性状的重要性参数较大，而以分级代码表示的目测性状重要性参数相对较小。对于茎节包露、穗类型、穗立形状、外颖茸毛和叶茸毛等稳定性较低、重要性参数不高的性状，可从水稻DUS测试指南中去除；对成熟期、抽穗期等稳定性较低但在生产实践中很重要的农艺性状，宜安排在申请者所在地参试点测试，使其表达状态真实表现；或者采用标准品种进行校正，从而提高测试的准确性。

Stability and Importance of Morphological Characteristics in Example Varieties Based on DUS Test

In order to further improve the accuracy of distinctness, uniformity and stability (ab. DUS) test of rice varieties, the AMMI model was used to analyse the stability and Random Forest algorithm was used to evalue the importance for 50 charcacteristics of 49 rice example varieties listed in DUS test guideline. The results showed that stability is largely different amony different characteristics, and visual characteristics are highly stable for main qualitive characteristics and pseudo-qualitive characteristics, and measurement characteristics had low stability for quantitative characteristics . On the whole, the importance parameter difference was very big among different characteristics, which is larger for measurement characteristics, and relatively smaller for visual characteristics. Some characteristics with low stability and low importance could be excluded from DUS guideline, for example “panicle: exsertion”, “panicle: type”, “panicle: attitude”, “Spikelet: pubescence of lemma”, “leaf: pubescence”, while those with low stability, but very important in practice, for example “time of maturity” and “time of heading” could be only in the location of applicants tested to avoid misjustice and use example varieties for correction.