摘要
以1985～1994年黄河流域棉花区域试验中春棉5轮和夏棉4轮(共计331次单年单点试验)的小区霜前皮棉产量为资料,在单点单年、单点两年和多点两年三个层次上研究棉花区域试验的精确度问题,获得如下主要结果:①品种区域试验应同时采用两项指标来反映其精确度,一为误差项变异系数(CEV),说明试验精确度,另一为品种均值的相对最小显著差数(RLSDa),说明品种比较精确度,②RLSDa和CEV在不同层次上关系不一致,若不考虑试验设计和RLSDa,单纯以CEV的大小决定试验的取舍,缺乏理论与实践依据。③RLSDa能客观反映出试验的综合水平,无量纲,可用于所有以比较处理效应为目的的田间试验,不受作物种类、性状和试验类型限制,建议用作区试中试点评价与建设的一个主要指标。④黄河流域的9轮区试中,春、夏棉仅分别有36.54％和48.78％的单年单点试验的精度足以鉴别出品种间10％的差异,其相应的CEV(v=8,r=4时)为6.8％。有半数以上的试验鉴别不出10％的差异,其精度有待提高。
In order to study and improve experimental precision (EP) and variety compar
ison precision(VCP) in Regional Trials, data of lint yield per plot from 9 rounds of Yellow River
Cotton Regional Trials (YRCRT) were analyzed. Coefficients of error variation (CEV) and a
newly proposed index RLSDa, the least significant difference as a percentage of overall mean
(relative LSDa), were estimated at three experimental levels; one-location-one-year, one-loca
tion-two-year and multi-location-two-year. Following results were obtained:(1) Both CEV and
RLSDa are necessary for evaluating the precision of Regional Trials. They indicate EP and VCP,
respectively. It is inappropriate to remove some data in analysis only according to the value of
CEV(e.g. , CEV>15%), which usually reduce the degree of experimental representation as
well. (2) RLSDa is a dimentionless parameter which can be estimated from experimental data
per se, and can be used as a main index for the evaluation of experiment. (3) For spring and
summer cotton respectively, 36. 54% and 48. 78% of one-location-one-year experiments in YR
CRT reached the precision level that can detect a variety mean difference (expressed as a
percentage of the overall mean, i. e. D%) of 10%. This precision corresponded to a CEV of
6. 8%. Both EP and VCP in YRCRT should be improved. (4) Some methods for improving the
precision of Regional Trials were proposed and discussed finally.
出处
《作物学报》
CAS
CSCD
北大核心
1998年第5期601-607,共7页
Acta Agronomica Sinica
关键词
区域试验
棉花
精确度
品种
Cotton
Regional trial
Precision
CEV
RLSD
