S.C.A.-AGRICULTURAL RESEARCH STATION
Turda, 3350, str. Agriculturii 27 Jud Cluj, Romania.
V. Moldovan, Maria Moldovan, and Rozalia Kadar.
Winter wheat provides a substantially larger amount of the world's wheat production than does spring wheat because winter wheat is more productive in those areas where both types can be grown. Thus, winter wheat usually is preferred over spring wheat in the regions where the climate permits production. The limits of winter wheat adaptation are established primarily by winter temperature. Thus, the winter survival temperature determines the northern limit and the winter temperature that is sufficiently low to permit vernalization gives the southern limit of the cropping area. From this point of view, winter wheat cultivars must have a high enough winter hardiness in northern regions and low requirements for vernalization in southern regions to be acceptable to producers.
Improved cultivars substantially contribute to increase wheat production. However, wheat yields in most production regions seem to be no more than one-half of the potential yield of the cultivars and far below the theoretical maximum yields. This difference reflects powerful production constraints that prevent the true genetic potential for yield to be expressed by the grown cultivars.
Although wheat-breeding programs have some priorities in common, the major objective of increasing the genetic potential of yield for most, if not for all, can be achieved via breeding for higher yield potential or by diminishing or eliminating hazards that reduce yield. Actually, wheat breeding seeks to remove yield constraints by developing cultivars with resistance to disease, insects, lodging, cold, heat, and drought. Other yield constraints can be best dealt with through improved cultural practices and management. Obviously, some yield constraints are fixed by the environment and cannot be manipulated.
As a breeding objective, yield represents an extreme example of a quantitative trait being polygenically inherited and subject to environmental influence to a large extent. Studies have shown that the environmental variation associated with yield often exceeds genotypic variation, which leads to confounding the genotype mean performance with its true value.
Among breeding priorities, stability of performance may be as important as high yield potential. Therefore, 'genotype x environment' interactions are of major importance, because they provide information about the effect of different environments on cultivar performance and have a key role for assessment of performance stability of the breeding materials.
Developing a wheat cultivar generally results from the selection of valuable recombinants found in manageable hybrid populations. During the breeding process, they will be grown in a limited set of environments. Evaluation of breeding material in a wide range of environments seldom is possible, not to mention the multiple environments encountered by new cultivars released for commercial production. Testing over as wide a range of environments generally is essential if widely adapted cultivars are to be identified.
Environments are seldom, if ever, duplicated. Variation in an environment at a single location over years can be as great as those between locations in one year. Therefore, variations in the 'genotype x environment' interaction that are pertinent to wheat breeding problems are those associated with 'cultivar x year', 'cultivar x location', and 'cultivar x year x location'.
We have discussed stability and adaptation of winter wheat. Yield stability has been defined as the ability of a cultivar to produce an expected yield at the level of productivity of a certain environment (i.e., the cultivar that has no 'genotype x environment' interactions). In practice, the wide variation in yield stability are related to the range in adaptation and response to production inputs. Therefore, wheat cultivars must have sufficient potential to maintain competitive yields in various environments and react favorably to conditions or increased production inputs.
That practical wheat breeding can make increases in genetic yield potential without substantial loss in yield stability and adaptation is of question. Some believe that yield potential and yield stability are more or less independent. Others say that yield stability is inversely proportional to the sum of squares for the 'genotype x environment' interaction attributable to that cultivar. The fact that one cultivar has significantly superior mean yields than another over a wide range of environments denotes genetic differences in the behavior of different genotypes. However, high mean yield alone is not necessarily indicative of high stability and wide adaptation.
Finlay and Wilkinson (1963) pointed out that the desired genotype is the one that produces a high mean yield over a range of environments and has average yield stability in comparison with other genotypes in the same conditions. They suggested that each nursery mean yield can be considered as a measure of an environment and, thus, an array of low- to high-yielding environments becomes available from a given set of ecological trials. The response of a particular cultivar to this range of environments can be estimated by the regression of yield of each cultivar on the mean yield of the nursery. The regression coefficient (b) is considered as a parameter of yield stability. So, b = 1 denotes cultivars with average stability; b > 1 are less stable cultivars, and b < 1 denotes very stable cultivars.
Eberhart and Russel (1966) developed this concept of stability and suggested the use of two stability parameters when describing the performance of one cultivar over a range of environments. They proposed that the regression of each cultivar on an environmental index and a function of the squared deviations from regression would provide more useful estimates of yield stability parameters. The environmental index is a coded deviation of each environment from the grand mean over a given range of environments. Environmental index is obtained for each environment by subtracting the grand mean of all cultivars over all environments from the mean of all cultivars in each environment. This forces the regression of the mean of all cultivars on the environmental index to have unit slope (b = 1). Therefore, a stable cultivar can be defined as one that have above average performance in all environments, a unit regression coefficient (b = 1), and a deviation from regression as small as possible (Sd^2^ = 0).
We evaluated the data from 22 ecological yield trials to examine the contributions of the ARDS Turda winter wheats to increases in yield and stability of performance in wheat production. The trials are from seven locations and three years (19982000) plus one location with one year (1998); a total of 22 trials. These locations are representative of the diverse environmental conditions in Romania. Experimental cultivars in each trial did not exceed 25, including the long-term check Bezostaia 1, used for comparison to newer cultivars and nursery performance over the years. The trials usually were evaluated in a RCB with six replications. Previous crop, seeding date, and fertilization were different at each location and conformed to local practices. Because part of cultivars in the nursery are changed annually and may influence stability parameters, we chose only 11 cultivars that remained in all trials over the experimental years. The 22 trials included 10 Romanian winter wheats plus Bezostaia 1, the check considered to have had a fairly stable yield and satisfactory adaptation. Five of the 10 Romanian cultivars analyzed were released by the ARDS Turda wheat-breeding program.
'Cultivar x year', 'cultivar x location', and 'cultivar x year x location' interactions were significant, indicating that the yield performance of the cultivars varied with the environments.
Stability parameters, computed according to the Eberhart and Russel model, were used to describe the performance of cultivars over environments. According to the model, the environmental index as an independent variable (x) was obtained for each of 22 environments as the mean of those 11 cultivars minus the grand mean (mean of the 11 cultivars in all 22 environments). The mean yield of each cultivar in each environment (y) was than regressed upon the environmental index. The statistical mean yield, regression coefficient (b), and coefficient of determination (r2) are currently used to evaluate the stability of yield over environments. We prefer coefficient of determination instead of deviation from regression because it directly gives predictability of a cultivar in relation to the environmental index. Although the deviation from regression must be as small as possible (approaching 0), the desired coefficient of determination is one that approaching 1 when considerable confidence can be attributed on one environment's measurement of a cultivar's performance and adaptation.
Stability parameters for the yield of the ARDS Turda winter wheats in the 22 ecological trials and the check Bezostaia 1 are presented in Table 1. According to the statistical model, the mean yields correspond to an environmental index value of 0. Directly evaluating the percentage gain in yield attributed to cultivar improvement is relative to Bezostaia 1. In our case, this value was between 8 % for Transilvania and 17 % for Ariesan. At the same time, in comparison with Bezostaia 1, our cultivars have had regression coefficient of 1 or slightly higher, except for Turda 95, which has a lower value. In addition, coefficient of determination values equal to or higher than Bezostaia 1 show that the cultivar response to environments is predictable to a considerable degree. Turda 95, which has a lower slope of regression (b = 0.87), seems to be well adapted to suboptimal environmental conditions.
| Cultivar | Mean yield | Regression coefficient (b) | Coefficient of determination (r^2^) | |
|---|---|---|---|---|
| q/ha | % | |||
| Transilvania | 54.4 | 108 | 1.00 | 0.91 |
| Ariesan | 58.7 | 117 | 1.07 | 0.93 |
| Apullum | 57.9 | 115 | 1.11 | 0.90 |
| Turda 95 | 58.2 | 116 | 0.87 | 0.86 |
| Turda 2000 | 58.2 | 116 | 1.07 | 0.86 |
| Bezostaia 1 | 50.3 | 100 | 0.96 | 0.86 |
A higher regression coefficient is desirable for high-yielding cultivars because they must be responsive to favorable conditions or increased cultural input. Above average performance in all types of environments must be maintained. The regression of the yield of Transilvania (released in 1982) and Turda 2000 (released in 2000) on environmental indexes compared with the Bezostaia 1 check are shown in Figure 1. Differences in the mean yield of Transilvania and Turda 2000 relative to Bezostaia 1 demonstrate a continuous yield advance achieved by our wheat-breeding program during the last 30 years. The regression lines of the two cultivars are nearly parallel with that of Bezostaia 1, indicating that their superiority is maintained across a wide range of environments. The slope of Bezostaia 1 is b = 0.96, whereas the slope of Transilvania is b = 1 and Turda 2000 is b = 1.07. These cultivars tend to be slightly more favorable to environments. For these two cultivars, breeding progress to improve yield potential was accompanied with improved stability of performance.
The three other cultivars from our program, Ariesan, Apullum, and Turda 95, had different regressions of yield on environmental indices in the same set of trials; graphically illustrated in Figure 2. Ariesan, with the largest mean yield and a reasonable regression coefficient (b = 1.07), has the highest coefficient of determination (0.93) denoting a strong, predictable response to changes in environmental conditions. The combination of increased yield potential with good stability of performance may explain the wide acceptance and popularity of Turda-developed wheats like Ariesan. Turda 95, with a larger mean yield (approaching Ariesan) but low regression coefficient (b = 0.87), sharply contrasts with the smaller mean yield and high regression coefficient of Apullum. The coefficients of determination were nearly similar for the two cultivars. The larger mean yield of Turda 95 clearly is associated with its higher yield in the poorer environments, whereas Apullum with a larger regression coefficient seems to be well adapted in favorable environments. The stability parameters in this study do permit comparisons among cultivars for average yields, stability of performance as a degree of response to changing environments, and the predictability of response to specified environments. Such comparisons would be useful for judging the release of cultivars and making recommendations for suitable production conditions and areas of adaptation for different cultivars.
Conclusions and remarks. The final objective of a winter wheat-breeding program is the release of cultivars combining high yield potential and quality with stability of performance and adaptation. Breeding for resistance to diseases and different others biotic and climatic stress promote such stability. The high level of winter hardiness of wheat cultivars is a major requirement for many winter wheat regions. Many genes condition winter hardiness, but the adaptability and stability represent more complex breeding characters that are controlled genetically and encompass a large number of known and unknown morphological, physiological, and biochemical attributes. Therefore, breeding for adaptation must begin with choosing parents for the crosses. They must be well-adapted genotypes that will give valuable hybrid combinations for the desired cultivar. During the generations of selection, the breeding material needs to be grown in the different biotic and climatic conditions with which they will interact to allow the breeder to make sound judgements of chosen material. In addition, testing breeding material in different simulated conditions such as with pathogen inoculations, aluminum toxicity solutions, and sprouting in a mist cabinet, can help achieve the elements of cultivar adaptation. New techniques of selection or manipulation of genetic material also can aid in developing high-yielding and stable cultivars. Previously, we suggested that the pedigree selection method, with only a few reselections, may conserve some heterogeneity in cultivars and buffer against environmental changes resulting in a good stability of performance (Ann Wheat Newslet 48:113-115). However, we do not exclude the possibility that homozygous genotypes, like pure lines obtained by double-haploid techniques or other methods, may buffer as any other type of population if selection for increased stability is applied.
Breeders agree that testing over a wide range of environments is essential if stable and widely adapted cultivars are to be identified. However, the extensive trial data required for identification stable cultivars becomes available only in advanced generations, when a cultivar is close to or may be already released. Therefore, the methods for evaluating yield stability proposed in this study have had a little significant impact in the early generations of selection regarding breeding wheat for adaptation. Improved evaluation techniques, applied in early generations, should assist in the early identification of those lines having high yield potential associated with good adaptation in highly variable environments or in alerting breeders to possible deficiencies in adaptation for other lines.
Based on our results presented here, trends in cultivar response to environments in regional performance nurseries indicate that breeders must carefully consider the trade-off between maximum yield potential, stability of performance, and ranges in adaptation during cultivar evaluation. However, convincing breeders to sacrifice high yield for increased stability and wide adaptation is difficult. Nevertheless, assessing the 'genotype x environment' interaction as a factor in determining the yield potential in the different production conditions will remain the most important tool in the breeding wheat for yield.
References.
Mss. Rozalia Kadar completed her Ph.D. dissertation in December 2002 under the direction of Prof. Dr. Leon S. Muntean at the University of Agricultural Sciences and Veterinary Medicine Cluj-Napoca. The title of Rozalia's thesis was 'Study of the genotype-environment interaction in achieving of bread-making quality in winter wheat'. Her thesis research underscores the fact that most wheat quality characteristics are heritable traits and more or less influenced by environmental conditions and production inputs. The implications of 'genotype x environment' interactions in development of winter wheat cultivars with improved bread-making quality are discussed.