Path Analysis on Effects of Main Economic Traits on the Yield of YU6, A Japonica x Indica Hybrid Rice Line

: To further explore the effect of major factors on the yield of a line of japonica x indica hybrid rice, to clarify technical approaches to high efficient high yield production, the author analyzed 237 groups of data from 9 years in 6 counties of Taizhou. With correlation, regression, and path analyses, effects of major economic traits on the yield were determined. An effective approach to high efficient high yield production is to increase a seed setting rate and to produce large panicles while ensuring the number of productive panicles.


Path Analysis on Effects of Main Economic Traits on the Yield of YU6, A Japonica x Indica Hybrid Rice Line
Weiming Liu To cite this version: In recent years, along with the successful creation and application of YU6 hybrid rice, studies on the cultivation of this type of rice started to erupt [1][2][3][4][5][6][7][8][9]. In order to explore the effect of its main economic traits on the yield, clarify the technical approach for high efficiency, high yield production, the author has analyzed data from the past on economic trait and yield, conducted regression, and path analysis before [10]. Now, the author has collected more data from additional years and sites, conducted further analysis on the influences of major economic traits on the yield of japonica x indica hybrids, using YU6 as a representative. Table 1 shows the data for panicles and grains of YU6 cultivated in 6 counties, districts or cities.

Analysis Method
Using YU6 as a representative, the author performed correlation and regression analyses, and finally, path analysis to determine the relationship between each of the major economic traits and yield, the relationship among these traits in the japonica x indica rice line, to clarify the approaches for high yield and high efficiency.

Correlation analysis
As shown in Table 2

Stepwise regression analysis
A regression equation was obtained from a stepwise regression analysis on 237 groups of data, based on the principle of maximizing the correlation coefficient: where, 1 x is the number of productive panicles, 2 x the number of grains per panicle, 3 x the number of filled grains, 4 x seed setting rate, 5 x the 1000-seed weight. Their ranges were, respectively, 1108.5~2161.5 thousand/hm 2  x ) = 0.1871) and the seed setting rate ( r ( y , 4 x ) = 0.1101). However, these traits were interdependent to a certain extent. Because multivariate regression analysis has multiple co-linearity, it is not easy to judge contribution of each trait. The path analysis has effectively determined the direct effect of each variable on the result, allowed estimating the indirect effect of an independent variable, thus enabled the direct comparison of the importance of each trait on the yield [8,9,10]. Therefore, the author also performed a path analysis to determine effects of the major economic traits on yield in japonica x indica rice.

Path analysis
A path analysis on five major economic traits selected from the stepwise regression analysis was conducted. Results indicates that the number of productive panicles had the greatest contribution, followed by the number of filled grains per panicle, the number of grains per panicle, the seed setting rate and the 1000-seed weight (Table 3). The number of productive panicles had the greatest direct effect, with a direct path coefficient 0.8171. But it had negative effects through the other four traits, especially through the number of filled grains per panicle, with the path coefficient -0.4041. Thus, an increase in the number of productive panicles affected not only the number of grains per year, seed setting rate and the 1000-seed weight, but largely the number of filled grains. The overall effect of the number of productive panicles on the yield was r = 0.1948.
The number of grains per panicle had little effect on the yield, with a direct path coefficient 0.2531. It had a large indirect effect through the number of productive panicles negatively and through the number of filled grains per panicle positively, with coefficients -0.5598 and 0.5252. Its positive indirect effect through seed setting rate and negative indirect effect through the weight of ones thousand grains were small, being 0.0134 and -0.0023. The overall effect on yield from the number of grains per panicle was r = 0.2295.
The number of filled grains had a large direct effect on the yield, with a direct path coefficient 0.6447. Its negative indirect effect was large through the number of productive panicles, with a coefficient -0.5122. Its indirect positive effects through the number of grains per panicle and through the seed setting rate were smaller, with coefficients 0.2062 and 0.1291. Its negative effect through the 1000-seed weight was weak. The number of filled grains per panicle had not only a large direct effect, but also a large overall effect on the yield, being 0.4584. Therefore, while considering other traits especially the number of productive panicles, attention should be paid to increase the number of grains per panicle, especially the number of filled grains per panicle.
The seed setting rate had a direct path coefficient 0.2157, a large positive indirect effect coefficient 0.3859 through the number of filled grains per panicle. Its negative indirect effects through the number productive panicles and the 1000-seed weight, positive indirect effect through the number of grains per panicle were small. The overall effect of the seed setting rate on the yield was 0.4973.
The 1000-seed weight had the smallest direct effect on the yield, with a direct path coefficient 0.1240. It had not only small negative values for indirect effects through other traits, but also a negative overall effect on the yield. Thus, it is not advisable to increase the 1000-seed weight towards the goal of high efficient production.

Discussion and Conclusion
Economic traits affect not only directly the yield, but also indirectly through their influence on other traits. Through the above analysis on YU6, a representative of japonica x indica hybrid rice lines, it can be seen that the technical road to high efficient high yield production is to increase seed setting rate and produce large panicles while ensuring the number of productive panicles.
The stepwise regression analysis retained 5 factors that affect the yield. Path analysis results indicate that the number of productive panicles has the greatest influence on the yield, followed by the number of filled grains per panicle, the number of grains per panicle, seed setting rate and the 1000-seed weight. The previous stepwise regression analysis [7] removed the number of grains per panicle due to its insignificant effect. It indicates that the number of filled grains per panicle contributed the greatest influence, followed by the number of productive panicles, the 1000-seed weight and the seed setting rate. Although the first two most import factors are the same between the current analysis and the previous analysis, the order of these two factors has changed. This is due to the retention of the number of grains per panicle in the stepwise analysis in the current analysis. It is clear that the number of filled grains per panicle is highly dependent on the number of grains per panicle, and they were collinear. The current identified technical approach to high efficient, high yield production is the same as previously reported. The current study is more reliable since it had more data and these data covered a longer period.