应用主成分分析和聚类分析的水稻源库特性研究

Application of Principal Component Analysis and Cluster Analysis of Source-sink Characteristics of Rice

为明确水稻源库类型变化对产量的影响,在海南三亚、广东深圳和湖北荆州3个环境,考察引自国际水稻研究所的79份水稻种质资源的剑叶叶面积、倒2叶叶面积、比叶重、有效穗数、每穗总粒数、千粒重、结实率和单株产量等8个源库相关性状,对3个环境数据联合进行主成分分析和聚类分析。结果表明:主成分分析将8个源库相关性状简化为彼此互不相关的4个主成分,其所提供的信息量占全部信息量的82.14%;根据4个主成分的特征值及载荷矩阵,得出各主成分的得分表达式,并筛选出得分较高和较低的品种;利用所选取的4个主成分因子得分进行系统聚类,将79份水稻品种(系)划分为源小库小互作型、源限制型、库限制型、源库互作型4大类群,其中源库互作型的材料综合性状好,产量高,是水稻育种拓宽亲本多样性的首选种质材料。

For the purpose of making clear the effect of source sink type on the yield of rice, principle and cluster analyses on eight rice source and sink-related traits including flag leaf area(FLA), area of top second leaf(ATSL), specific leaf weight(SLW),productive panicle number per plant(PPN), spikelet number per panicle(SNP), thousand grain weight(TGW), seed setting rate(SSR) and grain yield per plant(GYP) of 79 germplasms introduced by IRRI were conducted based on the data collected in Sanya of Hainan, Shenzhen of Guangdong and Jingzhou of Hubei environments in 2013-2014. The results showed that eight source and sink traits were classified into 4 independent principle components, which provided 82.14% accounting for the total information. Expression of respective principle components was deduced based on eigenvalue and load matrix of the four principle components, and some accessions with higher and lower scores were screened out. The cluster analysis indicated that the 79 germplasms were classified into four groups, i.e., small source and sink interaction type, source limited type, sink limited type and source-sink interaction type. Among them, the source-sink interaction type was comprehensively better traits with higher yield, which is the preferred to enlarge genetic diversity of breeding parents in rice breeding program.