Crop Science 41:792-796 (2001)
© 2001 Crop Science Society of America
CROP ECOLOGY, PRODUCTION & MANAGEMENT
Diallel Analysis of Cultivar Mixtures in Winter Wheat
E.R. Gallandt*a,
S.M. Dofingb,
P.E. Reisenauerc and
E. Donaldsonc
a Univ. of Maine, Dep. of Plant, Soil, and Environmental Sciences, 5722 Deering Hall, Orono, ME 04469-5722
b Pioneer Hybrid International, Maize Research, 21888 North 950th Rd., Adair, IL 61411
c Dep. of Crop and Soil Sciences, Washington State Univ., Pullman, WA 99164-6420
* Corresponding author (gallandt{at}maine.edu)
 |
ABSTRACT
|
|---|
Cultivar mixtures have been suggested as a means to achieve increased crop productivity. By choosing cultivars that complement each other for performance of important traits, mixtures could be formulated to meet specific production requirements. The objective of this study was to evaluate the performance of wheat mixtures and their pure line component cultivars across a wide range of environmental conditions. Six winter wheat (Triticum aestivum L.) pure line cultivars and the 15 mixtures obtained by mixing seed of pairs of cultivars in equal proportions were evaluated in 33 environments in eastern Washington. Averaged across all environments, mixtures were 1.5% higher yielding than the mean yield of their pure line cultivar components. There was no difference in protein between mixtures and pure line cultivars. For both grain yield and protein, performance in mixtures was highly correlated with the average of the two component pure lines. Diallel analysis of mixing ability, analogous to genetic analysis of combining ability, demonstrated that pure lines differed in their ability to determine both grain yield and protein in mixtures. The ability to predict mixture performance based on pure line performance, together with the potential for above average grain yield, suggested that mixtures can be formulated to achieve specific production requirements.
|
INTRODUCTION
|
|---|
VARIETAL MIXTURES are common in subsistence farming systems, offering growers diversity of diet, stability of income, and reduced losses to pests (Smithson and Lennè, 1996). In developed countries, mixtures of genetically pure cultivars are rarely planted, despite considerable research suggesting that intraspecific diversity may contribute to producers' goals related to yield, risk, and pest management. Most research on cultivar mixtures has focused on mixing component cultivars that differ in their disease resistance to manage foliar pathogens (Jeger et al., 1981; Stuke and Fehrmann, 1988; Finckh and Wolfe, 1997). When intraspecific diversity of disease resistance is deployed in a given field, there is a tendency for disease incidence and yield to match that of the most resistant component (Finckh and Wolfe, 1997). The genetic diversity of varietal mixtures may also contribute to higher grain yield in the absence of pests. Gustafsson (1953) examined three barley (Hordeum vulgare L.) cultivars and all possible two-way mixtures under a range of fertility and spacing treatments. Mixtures were found to be superior to the average of the two variety components in every situation, with an average yield advantage of about 4%. Furthermore, one component cultivar was identified that caused the mixture to exceed the highest yielding component of that mixture.
Sage (1971) studied the yield of mixtures obtained by blending seed of cultivars in equal proportions. Yields of mixtures were higher than those of pure lines only at low seeding rates, and it was concluded that the advantage of mixtures would only be realized in unfavorable environments. There is some evidence that high-yielding mixtures must have a high-yielding component. Shorter and Frey (1979) studied pure line and mixtures derived from composite crosses in oat (Avena sativa L.). General mixing ability predominated and was statistically significant, while specific mixing ability was not significant. Also working with oat mixtures, Pfahler (1965) found that yield stability was improved in mixtures relative to the pure line components. The yield benefit realized with mixtures may be a function of greater niche exploitation or complementary resource utilization, mechanisms which have been studied in great depth in interspecific mixtures, e.g., intercropping systems (Fukai and Trenbath, 1993), but not, to our knowledge, in intraspecific mixtures. Therefore, mixture performance could be influenced by numerous site-specific conditions including fertility, precipitation, seeding dates and rates, as well as possibly weed pressure.
Despite numerous examples in which mixture performance was superior to that of pure line components, there are exceptions. Rajeswara Rao and Prasad (1982) compared grain yield of spring wheat mixtures and their pure line components, and found no advantage of mixtures. Likewise, Finckh and Mundt (1996) examined mixtures of five winter wheat varieties in Oregon and found that yield did not differ between mixtures and pure lines. Baker and Briggs (1984) found no significant differences in yield between the average performance of 10 barley cultivars and the 45 possible two-component mixtures. Patterson et al. (1963) found no yield advantage of oat mixtures over pure line components, but noted that mixtures had greater lodging resistance.
Overall, there appears to be a slight, but significant advantage to the genetic diversity afforded by varietal mixtures. Jensen (1988) summarized literature concerning mixtures and concluded that the grain yield of mixtures was generally close to the average of the components, with a small skew towards the higher-yielding component. He suggested the most appropriate uses of mixtures would be to (i) identify the occasional mixture that performed better than the best component, and (ii) utilize other benefits of mixtures such as lodging resistance, stability, or specialty use. Likewise, Trenbath (1974), in an extensive review of literature in which forage yield of mixtures was measured, found that yield of mixtures was greater than that of pure lines in more than half the experiments examined. Smithson and Lennè (1996) summarized data from more than 60 studies involving cereals (barley, wheat, and oat) and also concluded that mixtures offer a small but significant advantage over the yield of the components. The greatest advantage over pure line components (5.4%) was observed with wheat mixtures. The authors felt that the most informative methods to study mixtures were those analogous to the diallel and combining ability analyses developed for genetic studies.
Approximately 17% of the 644 000 ha of soft white common winter wheat seeded in Washington in 1999 consisted of mixtures (WASS, 1999). The mixture of Madsen and Rod was the fourth most popular cultivar of winter wheat seeded in 1999. Most studies of mixtures have been conducted with a relatively small number of environments, or a relatively homogeneous set of environmental conditions. The objective of this study was to evaluate the performance of winter wheat mixtures and their pure line component cultivars across a large number of environments encompassing a wide range of environmental conditions.
|
MATERIALS AND METHODS
|
|---|
The experimental material consisted of six soft white winter wheat cultivars and the 15 mixtures obtained by mixing seed of pairs of the cultivars in equal proportions. The six cultivars (hereafter referred to as pure lines), considered to be the elite lines for Washington growers at the inception of this study, were Eltan, Hill 81, Lewjain, Madsen, Rod, and Stephens. These pure lines differ in straw strength, emergence, winter survival, and their resistance to several important pathogens.
Field trials were conducted at 13 locations for three winter cereal growing seasons during 1994 to 1997 in eastern Washington. Twelve locations were grown under dryland production practices; one site was irrigated. The locations used in this study represent climatologically diverse growing conditions, with annual precipitation ranging from approximately 28 to 61 cm. Management practices such as previous crop, seedbed preparation, pest control, and fertilizer application varied across sites, and were representative of prevalent practices at individual sites. A total of 33 environments (location x year combinations) were analyzed. The experimental design for all trials was a randomized complete block with four replications. Plots consisted of seven rows, 3.7 m long, spaced 15 cm apart. Plots were harvested with a plot combine and the weight of grain recorded. Near infrared spectroscopy (Infratec 1226, Foss North America, Eden Prairie, MN) was used to estimate protein concentration in each grain sample.
A diallel analysis as outlined by Hallauer (1981) across environments was performed. Genotypes (pure lines and mixtures) were considered fixed effects and environments were considered random effects. The analysis was similar to Griffing's (1956) Method 2, Model 1 that uses parents and crosses without reciprocals, except that it allows for the partitioning of specific combining ability into variation attributable to (i) among parents and (ii) parents vs. crosses sources. Pure lines and mixtures in this study were considered analogous to parents and crosses, respectively, of diallel crosses used in genetic applications. This was justified by the fact that plants from crosses in genetic applications contain equal contributions of both parents, while each mixture in this study contained equal contributions of its two pure line components.
The term general mixing ability (GMA) was used to describe the average performance of a pure line in mixtures, determined on the basis of the performance of mixtures having that pure line as a component. This term is similar in concept to general combining ability used in genetic applications. Similarly, the term specific mixing ability (SMA) was used to describe the deviation in performance of a mixture from that predicted by the GMA of both parents. This term is similar in concept to specific combining ability used in genetic applications. The linear model for the analysis across environments to determine the significance of GMA and SMA effects was:
where X ijk = value of the mixture with component Pure Lines i and j from the kth replicate, µ = overall mean, rk = replication effect of the kth replicate, gi = GMA of Pure Line i, gj = GMA of Pure Line j, sij = SMA of the mixture containing Pure Lines i and j, eijk = residual term of the mixture with component Pure Lines i and j from the kth replicate.
Genotype x environment and other interaction terms were determined by standard methods. GMA and SMA effects were calculated by methods described by Hallauer (1981). These have the property that both the sum of GMA effects, along with the sum of SMA effects, must sum to 0.
|
RESULTS AND DISCUSSION
|
|---|
Average grain yield at the 33 environments in this study ranged from 4.33 Mg ha-1 at Bickleton to 7.94 Mg ha-1 at the irrigated site of Moses Lake (Table 1). Because moisture is the primary factor limiting grain yield in this dryland cropping area, annual precipitation was a major determinant of average grain yield at each location. Average protein ranged from 64 g kg-1 at Bickleton to 123 g kg-1 at Moses Lake.
View this table:
[in this window]
[in a new window]
|
Table 1. Locations, average grain yield and protein for six pure line winter wheat cultivars and all 15 mixtures grown in eastern Washington for three growing seasons during 1994–1997. Soil at the Moses Lake location is classified as a sandy loam; all other locations have silt loam soils.
|
|
Average grain yield and protein of the six pure lines and their 15 possible mixtures are shown in Table 2. Rod was the highest yielding pure line, with an average yield of 7.52 Mg ha-1, and Lewjain the lowest, with an average yield of 6.42 Mg ha-1. Grain yield of the mixtures ranged from 6.5 Mg ha-1 for Eltan-Lewjain to 7.24 Mg ha-1 for Madsen-Rod. Grain yield of mixtures could be approximated by average yield of the two component pure lines. The correlation between grain yield of mixtures and the average of their two component pure lines was 0.93** (data not shown).
View this table:
[in this window]
[in a new window]
|
Table 2. Mean grain yield of six pure line winter wheat cultivars and all 15 mixtures grown in 33 environments in eastern Washington for three growing seasons during 1994–1997.
|
|
Among the pure lines Rod had the lowest protein, with a value of 96 g kg-1, and Madsen the highest, with a value of 104 g kg-1. Among the mixtures, Lewjain-Rod and Eltan-Rod had the lowest protein, with values of 97 g kg-1. Two mixtures, Hill 81-Madsen and Madsen- Stephens, had the highest protein of 103 g kg-1. Thus, Rod, the pure line with the lowest protein, was a component of the two mixtures having the lowest protein. Also, Madsen, the pure line with the highest protein, was a component of both mixtures having the highest protein. This suggests that it would be relatively easy to determine components of a mixture that would result in optimum protein levels based on protein levels of the pure line components. The correlation between protein of mixtures and the average protein of the two component pure lines was 0.88** (data not shown).
Significant difference existed for grain yield among both pure lines and mixtures (Table 3). The average yield of all mixtures was also significantly different than the average of all pure lines. Across all mixtures and environments, mixtures were 0.1 Mg ha-1, or about 1.5%, higher yielding than pure lines (Table 2). This is in agreement with Jensen (1988), and represents a difference of economic consequence that favors the use of mixtures in environments, and with germplasm, similar to the ones tested here. Grain yield in mixtures is influenced by intraspecific competition between component pure lines that begins during early development and continues to physiological maturity. Apparently, complementary relationships among pure lines in mixtures for growth habit, shading, or other factors were responsible for the increased grain yield of mixtures. GMA effects were significant, indicating that some pure lines tended to promote higher yields in mixtures than others. SMA effects were also significant, indicating that the GMA of each combination of pairs of pure lines did not account for the differences in yield observed among mixtures. This differs from results of Shorter and Frey (1979) who found significant GMA but no significant SMA effects in oat.
View this table:
[in this window]
[in a new window]
|
Table 3. Analysis of variance for grain yield and protein of six pure line winter wheat cultivars and all 15 mixtures grown in 33 environments in eastern Washington for three growing seasons during 1994–1997. GMA = general mixing ability, SMA = specific mixing ability.
|
|
The tendency for mixtures to perform somewhat better than the average of their pure line components was not uniform across environmental conditions (Table 3). When the yield superiority of mixtures over the average yield of its two component pure lines was regressed on grain yield of mixtures, a significant linear relationship was observed (Fig. 1a). Mixtures tended to perform relatively better than the mean of their component pure lines in higher yielding environments. However, in the three lowest yielding environments mixtures performed more poorly than the mean of their component pure lines. Thus, the 1.5% overall average yield increase of mixtures over the average of their pure line components was due to mixture superiority in the highest yielding environments. In only one environment was the yield of mixtures greater than the highest yielding component (Fig. 1b). Thus, in environments similar to these, it is unrealistic to expect that mixtures would perform better than the highest yielding pure line component.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 1. Grain yield advantage of mixtures over the mean of both component pure lines (A) and grain yield advantage of mixtures over the highest yielding pure line (B) for 15 mixtures grown in eastern Washington for three growing seasons during 1994 to 1997. **, * signify a significant linear regression at the 0.01 and 0.05 levels, respectively.
|
|
Genotype x environment interaction for yield was partitioned into effects between (i) environments and pure lines and (ii) environments and mixtures (Table 3). Both were significant, indicating that the relative performance of both pure lines and mixtures changed under different environmental conditions. The environment x mixture interaction was partitioned into environment x GMA and environment x SMA effect. The environment x GMA effect was significant, indicating the relative ability to promote high grain yield in mixtures differed across environments. The environment x SMA effect was also significant, indicating interactions among pure line components in mixtures resulted in grain yields that differed from that predicted by the GMA of each component pure line. The environment x pure lines vs. mixtures effect was significant, indicating that the difference in grain yield between pure lines and mixtures was different in different environments.
Significant differences also existed among pure lines and mixtures for protein (Table 3). Unlike grain yield, however, average protein of pure lines and mixtures was not different. Apparently, the complementary relationships between pairs of pure lines that increased grain yield in mixtures relative to the average of their component pure lines was not present for grain protein.
The environment x GMA effect was significant for protein, indicating that the relative ability of pure lines to influence grain protein in mixtures differed across environments. Also, the environment x SMA effect was significant for protein, indicating the presence of interactions between specific pairs of pure lines. The environment x pure lines vs. mixtures effect was also significant, indicating the difference in protein between pure lines and mixtures was different in different environments.
Rod had the highest GMA for grain yield (Table 4), and was clearly superior to all other pure lines as a component in mixtures. Performance as pure lines was generally indicative, but not completely predictive, of GMA of the pure lines. Rod, for example, had both the highest GMA and the highest yield as a pure line. Eltan, had the lowest GMA, and was nearly equal to the lowest yielding pure line. SMA effects were generally small relative to GMA effects. The largest SMA effect for grain yield was the positive interaction between Eltan and Hill 81.
View this table:
[in this window]
[in a new window]
|
Table 4. General mixing ability (along diagonal) and specific mixing ability (below diagonal) for grain yield for all 15 possible mixtures grown in 33 environments in eastern Washington during 1995–1997.
|
|
Madsen had the highest GMA for protein, and Rod the lowest (Table 5). As with grain yield, GMA for protein was predicted well by pure line performance. For example, Madsen had the highest protein of pure lines, and Rod the lowest. As in the case of grain yield, SMA effects were generally smaller than GMA effects, although several exceptions were observed.
View this table:
[in this window]
[in a new window]
|
Table 5. General mixing ability (along diagonal) and specific mixing ability (below diagonal) for protein for all 15 possible mixtures grown in 33 environments in eastern Washington for three growing seasons during 1994–1997.
|
|
These results show an overall yield advantage of 1.5% of mixtures compared with the average yield of their pure line components. No differences in protein were found between mixtures and pure lines. The high correlation of grain yield in mixtures and average grain yield of the two component pure lines comprising each mixture indicates that grain yield of mixtures could be accurately predicted from information from pure lines. This was also true of grain protein. Thus, it should be feasible to synthesize mixtures to obtain optimal levels of individual components. For example, high protein may be undesirable for many soft white winter wheat markets. A cultivar such as Madsen, possessing desirable agronomic characteristics but high protein, could be mixed with Rod, a cultivar with high grain yield and low protein. It should be noted, however, that other factors such as maturity and grain uniformity should be considered when determining components of mixtures.
|
NOTES
|
|---|
Washington State Univ. Dep. of Crop and Soil Sciences contribution no. 0108-05.
Received for publication February 25, 2000.
|
REFERENCES
|
|---|
- Baker, R.J., and F.N. Briggs. 1984. Comparison of grain yield of uniblends and biblends of 10 spring barley cultivars. Crop Sci. 24: 85–87.[Abstract/Free Full Text]
- Finckh, M.R., and C.C. Mundt. 1996. Temporal dynamics of plant competition in genetically diverse wheat populations in the presence and absence of stripe rust. J. Appl. Ecol. 33:1041–1052.
- Finckh, M.R., and M.S. Wolfe. 1997. The use of biodiversity to restrict plant diseases and some consequences for farmers and society. p. 203–237. In L.E. Jackson (ed.) Ecology in agriculture. Academic Press, San Diego, CA.
- Fukai, S., and B.R. Trenbath. 1993. Processes determining intercrop productivity and yields of component crops. Field Crops Res. 34: 247–271.
- Griffing, B. 1956. The concept of general and specific combining ability in relation to diallel crossing systems. Aust. J. Biol. Sci. 9:463–493.
- Gustafsson, A. 1953. The cooperation of genotypes in barley. Hereditas 39:1–18.
- Hallauer, A.R., and J.B. Miranda. 1981. Quantitative genetics in maize breeding. 1st ed. Iowa State University Press, Ames, IA.
- Jeger, M.J., D.G. Jones, and E. Griffiths. 1981. Disease progress of non-specialised fungal pathogens in intraspecific mixed stands of cereal cultivars. II. Field experiments. Ann. Appl. Biol. 98:199–210.
- Jensen, N.F. 1988. Plant breeding methodology. John Wiley & Sons, New York, NY.
- Patterson, F.L., J.F. Schafer, R.M. Caldwell, and L.E. Compton. 1963. Comparative standing ability and yield of variety blends of oats. Crop Sci. 3:558–560.[Free Full Text]
- PPfahler, P.L. 1965. Genetic diversity for environmental variability within the cultivated species of Avena. Crop Sci. 5:47–50.
- Rajeswara Rao, B.R., and R. Prasad. 1982. Studies on productivity of seed blends of two spring wheat cultivars under rainfed conditions. Z. Acker. Pflanzenbau 151:17–23.
- Sage, G.C.M. 1971. Inter-varietal competition and its possible consequences for the production of F1 hybrid wheat. J. Agric. Sci. (Cambridge) 77:491–498.
- Shorter, R., and K.J. Frey. 1979. Relative yields of mixtures and monocultures of oat genotypes. Crop Sci. 19:548–553.[Abstract/Free Full Text]
- Smithson, J.B., and J.M. Lennè. 1996. Varietal mixtures: A viable strategy for sustainable productivity in subsistence agriculture. Ann. Appl. Biol. 128:127–158.[ISI]
- Stuke, F., and H. Fehrmann. 1988. Plant pathological aspects in wheat cultivars. J. Plant Dis. Protection 95:531–543.
- Trenbath, B.R. 1974. Biomass productivity of mixtures. Adv. Agron. 26:177–210.
- Washington Agricultural Statistics Service. 1999. Washington agricultural statistics, 1999. USDA National Agricultural Statistics and Department of Agriculture and Washington State Department of Agriculture, Olympia, WA.
TOP
ABSTRACT
INTRODUCTION
