ITEMS FROM ROMANIA

S.C.A.-AGRICULTURAL RESEARCH STATION

Turda, 3350, str. Agriculturii 27 Jud Cluj, Romania.

Genotypic variability for yield in the common winter wheat collection.

V. Moldovan, Maria Moldovan, and Rozalia Kadar.

One of the main objectives of wheat breeding is to develop high-yielding varieties. In our work, an increase in genetic-yield potential can be expected by simultaneous improvement of the biological yield and the harvest index. One way of obtaining this aim is by using varieties with good combining ability for yield to develop suitable genetic recombination. Previously, we studied some aspects of the genetics of the harvest index in winter wheat that may have implications in practical breeding procedures for improvement of yielding ability (Moldovan et al. 1994). This task is becoming more and more difficult and cannot be successful without knowledge of the available useful variability for yield in the collections of common wheat under the environmental conditions at the breeding center.

Material and methods. For this study, 937 genotypes of common winter wheat of both local and foreign origins and representing a large range of agronomic types were grown in 1997 at ARS, Turda, in a collection-screening nursery. These sources (varieties and lines) were sown on plots of equal size without replication. Each plot consisted of four rows 1 m in length and 30 cm apart. The plot area was 1.20 m2. The local check variety Transilvania was included in 10 plots, which resulted in 104 individual plots. At maturity, plants in all plots were cut with a sickle 5 cm above the ground. In this way, a small part of the straw is lost from the biological yield, resulting in a higher harvest index. The harvested material was weighed to determine biological yield, threshed, then the grain reweighed for grain yield. Harvest index was determined as the ratio of grain yield to total plot yield (biological yield). Plant height was measured for each entry before harvest, because it can be related with harvest index by the effects of the Rht genes on yield.

The large number of observations in this study necessitated computer analyses. Statistical analyses were made for the 937-line wheat collection compared with the 104 plots of the check variety Transilvania. Mean, standard deviation, range, and coefficient of variation (C.V.) were computed for harvest index (%), biological yield (g/plot), grain yield (g/plot), and plant height (cm). Simple correlation coefficients (r) were calculated to provide a measure of the degree of association of these traits with each other. A regression analysis was done to determine more precisely the relationship of harvest index to each studied trait. First, second, and third degree polynomial models were tested. The following equations for these polynomials are:

a) first degree or linear model: y = a + b1x
b) second degree or quadratic model: y = a + b1x + b2x2
c) third degree or cubic model: y = a + b1x + b2x2 + b3x3

where: b = regression coefficients; a = the intercept; x = the independent variable; and y = the dependent variable.

The regression model chosen to represent a particular relationship can be based on the coefficient of determination (r2) values, which are the proportions of the sum of squares of the dependent variable that can be attributed to variation of the independent variables. A maximum value for the coefficient of determination provided the best fit for a particular equation of regression, calculated for two variables.

Results and discussion. Mean, standard deviation, range, and coefficient of variation values for harvest index, biological yield, grain yield, and plant height, of 937 common wheats, comparatively to 104 individual plots of the check variety Transilvania are shown in Table 1.

These results indicate a high variability between the 937 genotypes in the common wheat collection for all four characters studied. The values obtained in this study are the results of phenotypic expression of the characters analyzed. The phenotypic expression of a trait such as those presented can be considered as a linear function of the genotype and the environment in which the genotype was grown. With only one measurement made on each entry, the genotypic effect is confounded completely with the environment and the 'genotype x environment' interaction effects. Thus, part of the variability in the common wheat collection presented here is nongenetic in origin. The parameters of variability of the 104 sites of the check variety Transilvania reflected this nongenetic variability due to the microenvironmental effects plus random error of measurement. Therefore, comparison of the variability of the 937 lines of common wheat with the variability in the 104 sites of Transilvania would indicate that the winter wheat lines studied contain an important amount of genetic variability for harvest index, biological yield, grain yield, and plant height. The magnitude of this variability can be illustrated by comparing the wide range of values for the winter wheat collection with the smaller range obtained for the check. Coefficients of variation also were higher in the case of the common wheat collection compared with the check for all characters studied, revealing the expression of large genetic variability for yield in the collection of winter wheats. Useful genetic variability in the yield characters in the collection is located between the mean values of characters and the upper limits of range values. The results of this study allow tentative identification of wheats likely to be genetically superior in their ability to produce high yield and for use in breeding for high-yielding varieties.

Yield ability is the most complex breeding character, influenced by the many others morphological and physiological traits that are related to yield potential. Correlation coefficients (r) between harvest index, biological yield, grain yield, and plant height are listed in Table 2.

Many of these relationships have been reported previously, but not for wheats representing such a large range of agronomic type, all grown at the same location in the same year. The results indicate that a significant and complex relationships does exist between the characters analyzed. The low correlation in this study between harvest index and biological yield indicates that these two traits are largely independent of each other. Some researchers have reported a negative correlation between these two traits. There also was a significant and negative correlation between harvest index and plant height, suggesting the role of Rht genes for improving yield potential. Biological yield is correlated strongly with grain yield, which together with harvest index can result in a good yielding ability. The harvest index is correlated positively with grain yield, which makes it a useful selection criterion of breeding for improvement yield potential.

In order to explain more precisely the relationships of harvest index with the traits studied, a regression analysis was done. Summaries of the regression models tested are given in Table 3.

Results of regression analysis indicate that quadratic and cubic polynomial models gave the best fit for the regression of biological yield, grain yield, and plant height on harvest index. The coefficients of determination for the linear regressions were small, except those between harvest index and grain yield, which had values nearer to the quadratic and cubic models. Even in this case, nonlinear models better explain the existing relationship, because wheats with the highest harvest index do not necessarily have the highest grain yield. In fact, the opposite can be true. The magnitude of coefficient of determination (r2) for the regression of biological yield, grain yield, and plant height on harvest index indicates that approximately 6 % of the total variation of biological yield, 25 % of the total variation of grain yield, and 12 % of the total variation of plant height can be attributed to variation in harvest index of the wheats analyzed. Wheats with the largest positive deviation of their grain yield values from yield values predicted by the regression equations are the most likely sources of genes for yield ability.

We concluded from this study that the useful genetic variability for yield in common wheat has not been exhausted. As long as usable variation for yield exists, there will continue to be opportunity to increase genetic potential for yield through classical breeding methods. Identification of wheats genetically superior for yield and yield components is the first step of this challenge.

Reference.

• Moldovan V, Botezan V, and Moldovan M. 1994. Investigations on the genetics of the harvest index in winter wheat. Cercet Genet Veget Anim III:45-54. (in Romanian).