proceduresThe significance level of 0.05 was adopted for all the analysis.The cropping season was assumed as a random effect, and thedata obtained in different years were pooled. In the comparisonof qualitative variables, the means were compared only with therespective control treatment using F-test and contrasts. The dif-ference in the means of the two PGR (Stance®and Pix®) was nottested because they were planted in different areas and dates. Thedose-dependent effect of each PGR across years was analyzed withmultiple linear regression, and the slope significance was tested byt test. For the regression analysis, all the data points were pooledacross years and spraying time. In the graphs, only the mean value,the standard error of the mean (as vertical bars), the regressionline, the equation, and the R2were presented. When the slopewas not significant, the regression line was omitted. The data oncastor mean seed yield measured in the experiment on harvestaids (ET®, Ginstar®+ Finish®, and paraquat) were compared onlywith the check treatment (frost killed plants) using contrasts andF-test.