1Department of Plant Biology, Royal Veterinary and Agricultural University, Frederiksberg, Denmark; 2Danish Institute of Agricultural Sciences, Research Center Flakkebjerg, Denmark; 3Department of Agricultural Sciences, Royal Veterinary and Agricultural University, Tåstrup, Denmark

Background and objectives
Mixtures of barley cultivars or lines containing different race-specific resistance genes to powdery mildew (Erysiphe graminis f.sp. hordei) efficiently reduce disease level. The reducing effect on powdery mildew may be partitioned into a direct effect due to the resistance genes and an indirect effect due to mixing the components with different resistance genes. As expected, the level of resistance of a mixture (mean of resistance levels of components in pure stands) is the principal factor determining observed disease level. Also, the indirect effect of mixing, expressed as the difference between expected (mean of components) and observed disease level divided by expected disease level, increases with increasing resistance level of the mixture. In mixtures of near-isogenic lines, it has been shown that the observed disease levels can be described as a non-linear function of the expected relative disease levels of the mixtures, whereas the positive relationship between the mixing effect and mean resistance level of the mixture is linear [1]. To look further into the importance of the resistance level of a mixture for the observed powdery mildew level and mixing effects, experiments were conducted with two-component mixtures of barley lines, and a model was developed.

Materials and methods
Six two-component mixtures, composed of a susceptible cultivar Pallas, a resistant, near-isogenic Pallas-line P11 (Mla13) [2], and four two-component mixtures composed of two intermediate resistant lines,(P17 (Mlk) and P08B (Mla9), were grown giving a range of mean resistance levels. The 0 and 100% levels were obtained by including Pallas and P11 in pure stand. The experimental design was a randomized complete block with four replicates. Powdery mildew assessments were carried out three times during the growth season by counting colonies on the upper five leaves of the main tiller of 50 randomly selected plants per plot. An analysis of variance on transformed data (colony numbers) was applied.

Results and conclusions
The line P11 was 100% resistant (Mla13-virulence frequency=0) during the whole growth season, and consequently the mean resistance level of the Pallas/P11mixtures equalled the proportion of Pallas. At all three sampling dates, the relationships among the relative observed powdery mildew levels and the mean resistance levels of the mixtures were non-linear and fitted a second-degree polynomial. The nature of the relationship did not change over time from the start to the peak of the epidemic. In absolute values (colony number) the reduction was largest in the intermediate resistant mixtures. A positive linear relationship was found between the mixing effect and the mean resistance levels of the mixtures, i.e. the advantage of mixing was largest in the most resistant mixtures. The mixing effect was already seen 10 days after the first powdery mildew colonies were observed and increased over time. The very simple design of mixtures composed of one resistant and one susceptible component was ideal for studying the mechanisms, due to the density of the resistant line. The near-isogenic lines P17 and P08B were included in order to study whether induced resistance due to avirulent pathotypes was of any significance in mixtures composed of intermediate resistant lines. Under the given field conditions, the line P08B was unexpectedly close to 100% resistant, and the results were therefore inconclusive. In mixtures composed of more than two components differing in resistance level, other levels of mixing effects may be expected. Hovewer, the basic relationships are still valid, as seen in other studies [1]. It is therefore concluded that the powdery mildew resistance level of mixtures and the efficiency of mixing selected components can be predicted from the resistance of the components in pure stands.

1. Kølster P, Munk L, Stølen O, 1989. Crop Science 29, 1459-1463.
2. Kølster P, Munk L, Stølen O, Løhde J, 1986. Crop Science 26, 903-907.