Cucurbit Genetics Cooperative
Other Crop Genetics Cooperatives
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Cucurbit Genetics Cooperative Report 17:27-29 (article 8) 1994

Analysis of Component Traits for Early Yield in Cucumber

Mingan Yin, Hongwen Cui

Department of Horticulture, Northwestern Agricultural University, Yangling, Shaanxi 712100, P.R. China

Early maturity and high yield is an important objective in cucumber (Cucumis sativus L.) breeding programs. The early yield of cucumber is complicated by its quantitative nature and is affected by many other horticultural traits (e.g. epistasis, pleiotropy). The study of horticultural traits which affect early yield can help plant breeders recognize the component factors of early yield and provide for theoretical guidance in breeding. Thus, an experiment was designed to analyze the relationship between early yield and 25 horticulturally important traits in cucumber using stepwise regression. It was hoped that an optimum set of functions could be identified for early yield and its component traits.

Methods. Experimental plant material consisted of 41 F1 hybrids of 4 cultivars (numbered 1-45). Field work was conducted at the vegetable station, Northwestern Agricultural University, China, from early April to mid-June, 1991. Seeds were sown in plastic tunnels and arranged according to number on April 6. There were 30plants in each plot, spaced on 0.6 m row centers and positioned 0.4m apart in each row (4.4m2 area). Ten plants in each plot were sampled for observation and measurement.

Sixteen traits observed were: 1) days from sowing (DFS) to first staminate flowering (x1); 2) dfs to first staminate flowering on 50% plants (x2); 3) dfs to first pistillate flowering (x3)); 4) dfs to first pistillate flowering on 50% plants (x4); 5) node of first pistillate (x5); 6) leaf area index (x6); 7) fruits set on main vine (x8); 9) fruiting percentage on main vine (x9); 10) fruit branches per plant (x10)); 11) total branches per plant (x11); 12) fruiting branch percentage (x12); 13) mean pistillate flowers per node on main vine (x13); 14) fruits harvested in early stage (x14); 15) mean weight in jin per fruit (x 15); and, 16) early yield in jin (y).

Multiple stepwise regression was performed using early yield as the dependent variable and other traits as independent variables using the ANALYST program. Standard determination coefficient, R2 80%, was used in the analysis.

Results.The F value in the first step (regression) was used as the critical value (i.e., F=2.05) while regressing 15 independent variables. In each step, an independent variable whose F value was not significant and was the least value in partial regression tests was eliminated. Then the next regression was performed until all remaining variables had significant F values. According to this principle, x1, x7 , x6, x5 , x8 , x13 , x10 and x11 were eliminated in order from the first step to the eighth. At the ninth step an optimum regression equation was identified whose F value distribution of independent variables is shown in Table 1 as:

y = -21.36513 - 0.4417x2 - 0.4147x3 + 1.51552x4 - 0.05135x9 + 0.00869x12 + 0.20941x14 + 5.72047x15 .

Data suggest that cucumber early yield was mainly composed of x2 , x3 , x4 , x9 , x12 , x14 , and x15 . In the regression interval, the y value varied directly to x14 and x4 , and inversely to x2 .

The coefficient of determination, R2 , was used as a critical value for evaluation of the regression equation's value over sampling dates. It was calculated that R2 was 0.83 when the regression equation was constructed at the ninth step {i.e., square sum of variation determined by the 7 independent variables (Uy/x) was 83% of square sum of the total y variation (y2)]. This value was more than the expected standard R2 (80%). Thus, the regression equation used accurately estimated sample dates.

Table 1. F value of regression parameters in the ninth step during the determination of early yield in cucumber.

F value

Table 2. F test of optimum regression equation in early yield determination of cucumber.
Critical value
F0.01(7,37) = 3.18

From Table 2 it can be seen that the linear relationship between seven horticultural strains and early yield was significant in the equation. Therefore, the relationship between cause and effect, as expressed by the regression function, was creditable.

The standard errors of the regression estimate was S(70) = 1.50984. when the confidence coefficient (1- ά) was 0.95, then ∆ = t0.05(n-3)S(y0 )) - 2.0211 x 1.50984 = 3.05160. The interval estimate of the mean value of y(E) was y0+ 3.05160.

The standard error of observation (D*) (y0) = 3.40887. When the confidence coefficient is 1- ά = 0.95, then ∆ = t0.05 (n-3)S(y0 ) = 3.40887 x 2.20211 = 6.88967. The interval estimate of the sample value of y (y*) was: y0 + 6.88967.

Discussion. Seven traits remained in the regression equation and eight traits were eliminated. The traits which were eliminated could only be thought of as not significant in their linear relationship with early yield. Their relationship to other traits could be interpreted (e.g. x3 should have certain relationship to x5 , and x12 contained x10 . Such complicated relationships can best be understood by path analysis.

Considering the effect of the environment on trait expression, one should analyze the heritability and genotype correlation between the seven traits. Such a treatment would characterize their genetic basis.

The optimum regression equation was based on the mean of the observed value for each trait. Thus, only when a trait value is near to its mean can correct predictions be made. If the trait value is not close to its mean, prediction errors cannot be avoided.

The experiment used only early maturing hybrids and cultivars. If middle and late maturing cultivars were evaluated, the main component of traits for early yield in cucumber would be better understood.

Literature Cited

  1. Want, Y. 1985. Path analysis of main horticultural traits and early yield of cucumber. J. Northeastern Agric. College 2:54-58.
  2. Ma, Yuhua. 1984. Basica of quantitative Genetics in Plant Breeding. Jiang-Su Science and Technology Press.
  3. Mo, H. 1992. Statistics in Agricultural Test. Shanghai science and Technology Press.
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Department of Horticultural Science Box 7609North Carolina State UniversityRaleigh, NC 27695-7609919-515-5363
Page citation: Wehner, T.C., Cucurbit Genetics Cooperative;
Created by T.C. Wehner and T. Ng, 1 June 2005; design by C.T. Glenn;
send questions to T.C. Wehner; last revised on 11 December, 2009