Introduction

The mlo gene is the only known gene for mildew resistance in barley against which no highly virulent mildew pathotypes have yet been found in agricultural environments. The gene is therefore widely used by breeders throughout Europe. About 100 commercial mlo cultivars have been registered in Europe and approximately half the spring barley growing area in Europe is presently sown with mlo-carrying cultivars. Although it is often believed that the mlo gene of barley provides a durable and race non-specific resistance to powdery mildew (i.e. Hückelhofen et al., 2001, Schulze-Lefert et al., 1994), mildew strains that partially overcome the mlo-mediated resistance are known from laboratory experiments (Schwarzbach, 1979), from Japan (Lyngkjaer et al., 1995) and from European environments (Schwarzbach 1987, 1998, Slater & Clarkson, 2001a,b). Since only approximate figures and limited data on the extent of compatibility between mlo-hosts and partially mlo-adapted mildew isolates are available, this paper attempts to describe in quantitative terms the range and extent of compatibility between barley cultivars with and without the mlo gene and powdery mildew isolates collected recently from European agricultural environments.

Materials and Methods

Terminology

The ability of the mlo-adapted mildew pathotypes to reproduce on mlo hosts is specific for mlo and specific host-parasite interactions can be observed between the particular host and pathogen strains (Lyngkjaer et al., 1995). Therefore the term "partial virulence" is used, according to Schwarzbach (1998), rather than "mlo-aggressivity" preferred originally by Jorgensen (1992). The term "virulence" itself is used in its broad original sense as the specific ability of a pathotype to reproduce on a host with a particular resistance gene, regardless of the mechanisms involved, and the term "avirulence" for the lack of this ability. On the molecular level, in contrast, pathotypes are characterised by avirulence genes which interact specifically with matching resistance genes causing hypersensitive reactions. However, not all resistance is based on hypersensitivity. In studies of pathogen populations, it is essential to use a common term describing the specific ability of any pathotype to reproduce on hosts with matching resistance genes, since otherwise different or opposite terms would be necessary for different mechanisms of pathogenicity. Therefore we prefer the term "virulence" for this specific ability of the pathogen.

Test varieties

The degree of partial mlo-virulence was tested on the cultivars shown in Table 1. Golden Promise, an X-ray mutant of cv. Maythorpe, was grown extensively in Scotland for more than twenty years from 1966, despite being very susceptible to mildew. Diamant is a Czech cultivar from 1964, without known genes for mildew resistance, susceptible under field and laboratory conditions. HL70-8 (=SZ5139), a chemically induced mutant of Diamant, backcrossed to Diamant, carries the mlo9 allele. Apex is a Dutch cultivar, which carries the mlo-11 allele, derived from an Ethiopian landrace, plus Mlg. It is more susceptible to mildew than many other mlo-cultivars (Schwarzbach, 1987, 2001; Slater & Clarkson, 2001a) but is usually resistant under field conditions. Chalice, Riviera, Landlord and Chariot, all currently grown in the UK, are also usually resistant in the field.

Mildew isolates

The isolates used in this study included the four highest ranking isolates from a larger collection of partially mlo-virulent isolates collected in 1999 from the air spora in the Czech Republic and Lower Austria with a jet spore trap and four partially mlo-virulent isolates collected from the air spora or infected plants at Cambridge, UK, in 1998 and 1999 (Slater & Clarkson, 2001a,b). For comparison, the isolates HL-3 with high partial mlo-virulence (Schwarzbach, 1979), PV-47 with low partial mlo-virulence from Moravia (unpublished) and a non mlo-virulent isolate CC/1 collected from Cambridge in 1970 (Slater & Clarkson, 2001a) were included. The isolates used and their virulence spectra are listed in Table 2. The virulence spectra were determined by artificial infection tests on detached leaf segments of standard differential cultivars used at NIAB Cambridge (Slater & Clarkson 2001b).

Experimental conditions

The experiments were conducted using the standard methods of the authors. At NIAB Cambridge these methods were: 25 mm leaf segments detached from 10-12 day old seedlings of the test cultivars grown in a spore-proof glasshouse were placed on the surface of benzimidazole agar (0.5% agar, 100ppm benzimidazole) in rectangular polystyrene boxes and inoculated using a settling tower. Following incubation for 10 days at 15 - 18°C with 16 h photoperiod, the area of the leaf segments infected with mildew was assessed on a 0 -10 scale. The data were summarised for eight replications and expressed as percentages of the infected area on segments of mlo cultivars, relative to the infected area of leaf segments of the Mlo cultivars. The experiment was repeated four times on different dates.

At the laboratory in the CR, 25 mm long leaf segments of Diamant (Mlo) and Apex (mlo) seedlings, grown mildew- free under continuous light, were placed in 9 cm Petri-dishes on the surface of a medium containing 0.4% agar, 20 ppm benzimidazole and 50 ppm of a mineral nutrient solution, as described in detail earlier (Schwarzbach, 2001). The segments were arranged in 5 triplets of Apex-Diamant-Apex and uniformly inoculated in a settling tower with the test isolates at densities below 1000 spores per cm². After 6 days of incubation at 17-18°C under continuous dim fluorescent light, the number of mildew colonies on the segments was counted and converted to number of colonies per Apex segment relative to number of colonies per Diamant segment. The experiments were repeated four times on different dates.

Statistical evaluation

Standard statistical methods were used where possible. The Analysis of Variance was omitted, since the variances were not homogeneous between the groups of data. For the evaluation of incomplete data sets, the Least Squares (LSQ) procedure was used.

Results

Virulence tests

The virulence spectra of the test isolates, determined at Cambridge, are given in Table 2.

The virulence spectrum of isolates originating from Moravia and Lower Austria was more complex than that of the isolates from the UK. This corresponds with the observed west-east trend of increasing virulence complexity, due to the continuous selection of the airborne pathogen population during its migration across Europe in the prevailing west - east wind direction (Limpert et al., 1999).

Czech experiments

In the Czech Republic (CR), the experiment was repeated four times on the dates given in Table 3. As there was one value missing, the means were adjusted using a Least Squares procedure. The standard deviation of the isolate means decreased with decreasing partial mlo-virulence, i.e. the variances were not homogeneous. The data could not, therefore, be evaluated by ANOVA. However, there is a significant range of partial mlo-virulence among the field isolates, as can be seen from the means and their standard deviations.

The existence of differences in partial mlo-virulence can also be demonstrated by the correlation between the individual experiments. Since the highly virulent PV-1 is not a field isolate, it was excluded from the calculations. Nevertheless, the mean correlation coefficient between tests of r=0.57 at 42 degrees of freedom is highly significant, indicating a range of different levels of partial virulence, expressed as number of mildew pustules per leaf segment, among the remaining isolates. The individual correlation coefficients are summarised in Table 4.

Cambridge experiments

At Cambridge, the results with the three Mlo cultivars and the six mlo cultivars were summarised. The mlo cultivars, except HL 70-8, have other resistance genes in addition to mlo that could distort the estimation of partial mlo virulence. Therefore, a table of background compatibility between the host cultivars and mildew isolates was derived from both the virulence spectra and the results obtained with the mlo varieties. Compatibility was indicated if at least a few colonies of type IV developed on the leaf segments. For isolate CC/1, with its very low partial mlo virulence and avirulence for all resistance genes except Mla12, the distinction was questionable. The data of CC/1 were therefore excluded from the calculations. The background compatibility data are summarised in Table 5.

Only results from background compatible variety x isolate interactions were used for the calculations. Since the data obtained thus from the mlo varieties were non-orthogonal, the LSQ technique was used to summarise the mlo results in each experiment. The results of all four experiments are summarised in Table 6.

The results obtained in Cambridge confirm the existence of differences in partial mlo-virulence, as indicated by the means and their standard deviations. The reliability of the results is also confirmed by the correlations between the individual experiments. Results obtained with the highly virulent PV-1 and the virtually avirulent CC/1 isolates were excluded from the calculations. Nevertheless, the mean correlation coefficient between tests of r=0.59 at 36 degrees of freedom is highly significant, indicating a range of different levels of partial virulence, expressed as the proportion of infected area of the leaf segments. The correlation coefficients are summarised in Table 7.

Summary of results

The results of the experiments from both the Cambridge and Czech laboratories are summarised in Table 8.

Discussion

The results obtained independently in both laboratories confirm that the tested field isolates have a considerable level of partial mlo virulence which varies between the isolates. The difference in ranking of the field isolates in the two laboratories was not unexpected as leaf area infected was assessed in the UK, while in the Czech Republic the number of colonies per leaf segment was determined. The same infected leaf area can be covered by a small number of large colonies or by a large number of small colonies. The growth of colonies and the number of colonies are both components of partial mlo virulence, probably controlled by different genes. Therefore, several genes must be implicated both in the differences in infection levels between the partially mlo-virulent field isolates and in the differences in ranking. This is also supported by the finding that all the tested mildew isolates have a different virulence spectrum, indicating their independent origin from different mutation or recombination events.

The high level of partial mlo virulence of isolate PV-1 is surprising. This isolate, known as HL-3, was developed 25 years ago by mass screening for nearly 40 generations in the laboratory from the isolate GE-3. At that time, GE-3 exhibited partial mlo virulence on mlo cultivar leaves, measured as colony counts, of only 0.2 % compared to 10% with HL-3, and 100% on Mlo leaves (Andersen 1991). Since mildew isolates have to be re-inoculated at least once per month, HL-3 must have been maintained in different laboratories for more than 300 generations, an unknown number of which was on mlo hosts. During that time the partial mlo virulence of HL-3 possibly evolved further. The level of partial mlo virulence observed in most of the tested field isolates here is very similar to the original level detected in HL-3 25 years ago. However, while HL-3 is avirulent for almost all resistance genes except Mlg and Mla6 and thus harmless to much of the barley grown in Europe, some of the tested isolates appear to pose a much greater threat. Thus, PV 3 from Moravia, which was virulent for all tested resistance genes, and PV-32 from Lower Austria, which was virulent for almost all the genes, have a large selective advantage in the agricultural environment over pathotypes with a smaller virulence spectrum. Such pathotypes are likely to increase in frequency even in the absence of selection pressure from mlo varieties. Currently, the level of partial mlo virulence of the most adapted known field isolates is not yet sufficiently high to pose a threat to mlo varieties. However, as adaptation progresses, it is very likely that pathotypes such as PV-3 with a broad virulence spectrum and a level of mlo virulence similar to or higher than PV-1 (=HL-3) will emerge. Since such a scenario would be disastrous for European barley growing, counter-measures should be taken.

We suggest the following measures:

1. New barley varieties with the mlo gene should be released only if they do not have a very susceptible background (as, for example, have Apex and Chalice)

2. Systematic monitoring of the frequency and level of partial mlo virulence in the pathogen populations in relevant European countries should be established, to facilitate early warning of problems for breeders and growers.

References

Andersen L, 1991. Mlo aggressiveness in European barley powdery mildew. In: Jorgensen J.H.(Ed.), Integrated Control of Cereal Mildews: Virulence Patterns and their Change, pp 187-196, Riso National Laboratory, Roskilde, 1991.

Hückelhoven R, Trujillo M, Kogel K-H, 2001. Mutations in Ror1 and Ror2 genes cause modification of hydrogen peroxide accumulation in mlo-barley under attack from the powdery mildew fungus. Molecular Plant Pathology 1, 287-292.

Jorgensen JH,1992. Discovery, characterisation and exploitation of Mlo powdery mildew resistance in barley. Euphytica 63,141-152.

Limpert E, Godet F, Muller K, 1999. Dispersal of cereal mildews across Europe. Agricultural Forest Meteorology 97, 293-308.

Lyngkjaer MF, Jensen HP, Ostergard H,1995. A Japanese powdery mildew isolate with exceptionally large infection efficiency on Mlo-resistant barley. Plant Pathology 45, 786-790.

Schulze-Lefert P, Freialdenhoven A, Scherag B, Görg R. 1994. Dissection of resistance pathways in barley to powdery mildew attack. M.J. Daniels et al. (eds.), Advances in Molecular Genetics of Plant-Microbe Interactions 3, Kluwer Academic Publishers, 309-312.

Schwarzbach E,1979. Response to selection for virulence against the ml-o based mildew resistance in barley, not fitting the gene-for-gene hypothesis. Barley Genetics Newsletter 9, 85-88.

Schwarzbach E, 1987. Shifts to increased pathogenicity on mlo varieties. In: MS Wolfe and E Limpert (Eds.), Integrated Control of Cereal Mildews: Monitoring the Pathogen, pp. 5-7. Publ. Martinus Nijhoff, 1987.

Schwarzbach E, 1998. The mlo based resistance of barley against mildew and the response of mildew populations to the use of varieties with the mlo gene. Czech Journal of Genetics and Plant Breeding 34, 3-10.

Schwarzbach E, 2001. Heat induced susceptibility of mlo-barley to powdery mildew (Blumeria graminis D.C. f.sp. hordei Marchal). Czech Journal of Genetics and Plant Breeding. 37, 82-87.

Slater SE, Clarkson JDS, 2001a. Reaction of spring barley cultivars carrying the mlo resistance to infection by powdery mildew isolates in the UK. Cereal Rusts and Powdery Mildews Bulletin [www.crpmb.org/] 2001/0724slater.

Slater SE, Clarkson JDS, 2001b. Mildew of barley. UK Cereal Pathogen Virulence Survey 2000 Annual Report, pp 50-59.