| Crop varieties are routinely evaluated in regional trials organized by administration of government in China, and the data from such trials is used as the key basis for variety registration. The system of rice regional trials has been regarded as the most important method for evaluation varieties. In recent years, many concerns were focused on the experimental precision of rice regional trials. However, on this evaluation system, some problems should be mentioned. Firstly, no proper standard being used for estimation experimental precision in rice regional trials at present; Secondly, in many studies, location effects were generally considered fixed in models of variance analysis, and the comparison between fixed effects and random effects was lacked. Thirdly, yield stability parameters were used to assess the interaction of G E (genotype environment), but the analytical methods were generally limited and no suitable for evaluation yield stability. The objectives of this experiment were (1) to evaluate degree of experimental precision of rice regional trials; (2) to determine the application of model in variance analysis and yield stability evaluation; (3) to discuss the approaches for improving experimental precision degree.The coefficient of error variation ( CEV), relative least significant difference (RLSD) and genetic coefficient variation (GCV) were used to analysis the experimental precision and location precision according to the data of regional trials for medium japonic a rice and late medium japonica rice in Jiangsu province, and the conclusions were analysed between fixed effects and random effects on location in variance analysis. Meanwhile, the results were compared for yield stability by LR (Liner Regression) model and AMMI model (Additive Main Effects and Mutiplicative Interaction) based on the data of regional trials for medium japonica rice in Jiangsu province.The results were as follows:1. Experimental precision of rice regional trialIn one-year-one-location trials, theCEV ranged from 0.88 to 10.0% in rice regional trials. The percentage of CEV < 5% accounted for 85.0% and 90.0% of total trials, respectively, in regional trials of medium japonica rice and late medium japonica rice in Jiangsu province. The results also showed that only 45.7% of total trials reached theprecision degree that can identify yield differences of 5%. In one-year-multi-location trials, the averaged of CEV was 3.78% and 3.44%, respectively, the averaged of RLSD0.05 was 3.38% and2.54%, respectively, in regional trials of medium japonica rice and late medium japonica rice. Therefore, the experimental precision was more accurate, but variety comparison precision will be improved on one-year-one-location trials.Meanwhile, the results indicated both CEV and RLSD0.05 were essential for evaluation precision of rice regional trials.2. Appraisal accuracy of experiment stationsThe GCV is important standard for evaluation experiment stations. The study indicated that GCV ranged from 4.54 to 11.33% in regional trials of medium japonica rice, Rudong experiment station and Ganyu experiment station showed higher accurate, Huaian experiment station and Donghai experiment station showed lower accurate. The GCV ranged from 2.91 to 6.64% in regional trials of late medium japonica rice. Nantong experiment station showed higher accuracy, and Taizhou experiment station showed lower accurate.In different locations of rice regional trials, Taizhou experiment station has special performance on CEV, RLSD0.05 and GCV. Taizhou experiment station showed lower precision.3. Study for models of variance analysis in rice regional trialsIn one-year-multi-location trials, experiment stations were usually considered as fixed effects, In this case, the significant level was lower than that of random effects. For example, the significant standard that can identify yield differences for varieties was 10 Kg and13 Kg at 0.05 and 0.01 level in rice regional trials when locations were considered as fixed effects. When locations were considered as random effects, the... |
PDF Full Text Request