国家大豆品种区域试验精确度研究

以2000 2002年国家大豆品种区域试验的6 239个试验数据为依据,在一年一点、一年多点、一点两年和两年多点等四个层次上对国家大豆品种区域试验精确度进行研究,主要结果如下:⑴就目前国家大豆品种区域试验水平和管理体制,在单年单点试验、一点两年试验和两年多点试验等三个试验层次上的试验精确度,均无法满足国家大豆品种审定的需要,而只能依据一年多点的试验结果作为国家大豆品种审定的试验依据。⑵新的国家大豆品种审定标准规定,新品种比对照增产5%且差异显著是国家大豆品种审定的必要条件。由此可以看出,目前国家大豆品种区域试验西北春大豆组、南方春大豆组和北方春大豆晚熟组的试验精确度需要进一步提高。⑶影响一年多点试验的品种比较精确度的因素主要有试验误差、重复数、参试品种数、试验点数和品种×试点互作效应。提高一年多点试验的品种比较精确度的途径主要有:一是通过优化试验方案、提高人员素质、科学统计分析降低试验误差;二是合理安排试验重复数、参试品种数和试验点数;三是合理化分区试组以降低品种×试点互作效应。

Studies on the Accuracy of National Soybean Variety Regional Trials

Based on 6239 National Soybean Variety Regional Trial(NSVRT) data from 2000 to 2002,the index of experimental accuracy of NSVRT,coefficients of error variation(CEV) and relative LSD0.05,were investigated at four experiment levels:one-year/one-location,one-year/multi-locations,one-location/two-year,multi-locations/two-year.Only one-year/multi-location level can be used for National Soybean Variety Approval at present.The accuracy of regional trial in Northwest spring-sowing area,south spring-sowing area and North spring-sowing area should be improved.Experiment error,replications,varieties,locations and variety-location interaction were factors affecting the accuracy of one-year/multi-location experiments.Several approaches for one-year/multi-location experiments were: (1) to optimize experiment design,train more skillful technicians,and analyze data correctly;(2) to arrange the number of replications,varieties and locations reasonably;(3) to divide the experiment groups reasonably in order to reduce the variety-location interaction.