Effect of rotation of canola (Brassica napus) cultivars with different complements of blackleg resistance genes on disease severity

Authors


E-mail: bhowlett@unimelb.edu.au

Abstract

Blackleg disease (phoma stem canker) caused by the fungus Leptosphaeria maculans is a major disease of canola (oilseed rape, Brassica napus) worldwide. Canola plants in pots were exposed to blackleg-infested stubble of canola with different complements of resistance genes and then assessed for disease. Plant mortality was reduced when plants were exposed to stubble from a cultivar with a different complement of resistance genes compared to stubble of a cultivar with the same resistance gene. These findings were consistent with 7 years of field surveys, which showed that changes in selection pressure as a result of extensive sowing of cultivars with major-gene resistance, termed ‘sylvestris resistance’, dramatically influenced the frequency of virulent isolates in the population towards particular resistance genes, and therefore disease severity. All these data were supported by PCR-genotyping surveys of fungal populations whereby the frequency of virulence alleles of avirulence genes AvrLm1 and AvrLm4 changed significantly depending on the resistance gene present in the cultivar from which the isolates were cultured. This is the first example of a study showing that sowing of canola cultivars with different complements of resistance genes in subsequent years, i.e. rotation of resistance genes, minimizes disease pressure by manipulating fungal populations. This approach provides a valuable disease management strategy for canola growers and is likely to be applicable to other plant diseases.

Introduction

Breeding for disease resistance is an important strategy for controlling plant disease. However, the deployment of cultivars with resistance conferred by major or single genes can result in ‘boom and bust’ cycles (McDonald & Linde, 2002a). The ‘boom’ occurs when a single major resistance gene with large effect is sown widely over a large geographical area. The ‘bust’ then comes when the pathogen population adapts to the presence of the major resistance gene and evolves to overcome or break down the resistance. Examples of boom and bust cycles have been documented for diseases such as rusts and powdery mildews of cereals (McIntosh & Brown, 1997; Eversmeyer & Kramer, 2000; MacKay et al., 2011), scab of apples (Guerin & Le Cam, 2004) and more recently downy mildew of grapevines (Peressotti et al., 2010).

An example of a boom and bust cycle is the breakdown of resistance in canola (oilseed rape, Brassica napus) in France (Rouxel et al., 2003a) and in Australia to blackleg (phoma stem canker) caused by the fungus Leptosphaeria maculans (Sprague et al., 2006a). Blackleg disease is responsible for major yield losses of canola worldwide (Fitt et al., 2006). The interaction between L. maculans and canola is a typical gene-for-gene relationship (Balesdent et al., 2005). More than 12 major genes (Rlm1–9, RlmS, LepR1-3 and BLMR1-2) conferring resistance to L. maculans have been identified (Yu et al., 2005, 2008; Delourme et al., 2006; Van de Wouw et al., 2009; Long et al., 2011). Nine of these genes have been mapped to linkage groups on N2, N7 and N10 (Delourme et al., 2006; Rimmer, 2006; Yu et al., 2008; Long et al., 2011), but none have been cloned. Twelve corresponding L. maculans avirulence genes have been identified, seven of them mapped to two gene clusters (Balesdent et al., 2005; Yu et al., 2005; Van de Wouw et al., 2009). Three avirulence genes, AvrLm1, AvrLm6 and AvrLm4-7, have been cloned (Gout et al., 2006; Fudal et al., 2007; Parlange et al., 2009). AvrLm1 and AvrLm6 confer avirulence towards Rlm1 and Rlm6, respectively, whilst AvrLm4-7 confers avirulence towards two different resistance genes, Rlm4 and Rlm7 (Gout et al., 2006; Fudal et al., 2007; Parlange et al., 2009). Mutations/deletions in the coding regions of these avirulence genes can lead to virulence towards the corresponding resistance genes (Gout et al., 2006; Fudal et al., 2007; Parlange et al., 2009; Van de Wouw et al., 2010a).

In Australia, L. maculans undergoes annual sexual reproduction on canola stubble. Large numbers of airborne ascospores are released following autumn rains in the following season and are primary inocula for blackleg disease in newly sown crops. Under these conditions L. maculans populations evolve quickly because of prolific sexual and asexual reproduction, high potential for gene flow and large population sizes, as well as a high propensity for mutations caused by repeat induced point (RIP) mutation (McDonald & Linde, 2002a; Van de Wouw et al., 2010a; Rouxel et al., 2011; Travadon et al., 2011). Thus, the frequency of virulence alleles of avirulence genes can rapidly increase leading to breakdown of resistance as seen in France in 2000 (Rouxel et al., 2003a) and Australia in 2003 (Sprague et al., 2006a). In the latter example, cultivars harbouring major-gene resistance introgressed from Brassica rapa subsp. sylvestris (Crouch et al., 1994) were commercially released in Australia in 2000. These cultivars, hereafter referred to as having ‘sylvestris resistance’, were widely grown during the next 2 years in regions near Bordertown and on the Eyre Peninsula in southeastern Australia and high yields were obtained. However, in 2003 the resistance became ineffective, resulting in 90% yield losses and substantial economic losses (Sprague et al., 2006a; Van de Wouw et al., 2010b). Initially it was thought that the sylvestris cultivars contained a single resistance gene, named LepR3 (Yu et al., 2005). However, two independent studies later showed that these cultivars contain two resistance genes, denoted as Rlm1 and RlmS by Van de Wouw et al. (2009) and BLMR1 and BLMR2 by Long et al. (2011). It is unknown which of these latter genes LepR3 corresponds to and the gene nomenclature is unresolved. This paper uses the Rlm1 and RlmS terminology.

Disease management strategies can minimize the risk of resistance being overcome (McDonald & Linde, 2002b; Aubertot et al., 2006). Rotations with other crops, stubble removal and tillage are all deployed to decrease incidence of blackleg through reduced exposure to inoculum (Aubertot et al., 2006). Separation of canola crops from stubble sources by at least 500 m is recommended in Australia, as beyond that distance the amount of inoculum does not cause enough disease to significantly affect yield (Marcroft et al., 2004).

This study tested the hypothesis that rotation of cultivars with different resistance genes can increase the durability of disease resistance. The effect of sowing canola cultivars with different complements of resistance genes in subsequent years, i.e. rotation of resistance genes, on the severity of blackleg disease was investigated. Furthermore, the possibility of reintroducing resistance genes whose effectiveness had previously been overcome was explored.

Materials and methods

Brassica cultivars

Brassica napus cultivars were selected on the basis of their different complements of blackleg resistance genes (Rlm1–9 and RlmS) expressed at the seedling stage. The resistance genes for each cultivar were identified from glasshouse experiments where isolates with known avirulence alleles were inoculated onto cotyledons of individual cultivars and resistance was assessed after 2–3 weeks as described by Balesdent et al. (2005) (S. Marcroft & A. Van de Wouw, unpublished data, University of Melbourne, Melbourne, Vic, Australia). These cultivars may have additional seedling resistance genes that are as yet uncharacterized. Brassica napus cultivars termed ‘sylvestris’ have blackleg resistance in part derived from B. rapa subsp. sylvestris. Cultivars termed ‘non-sylvestris’ share no common resistance genes with the cultivars derived from B.  rapa subsp. sylvestris. Cultivars termed ‘other’ share one or more resistance genes with either non-sylvestris or sylvestris cultivars. Brassica juncea cultivars with so-far unidentified resistance genes were also used (Table 1).

Table 1.   Blackleg (Leptosphaeria maculans) resistance genes in Brassica napus cultivars
CultivarResistance genesaResistance categoryReferences
  1. aThese cultivars may have additional resistance genes that are as yet uncharacterized.

  2. bATR-Beacon is heterogeneous for Rlm3 and Rlm4. Rlm3 is present in 60% of seed and Rlm4 is present in 75% of seed (A. Van de Wouw & S. Marcroft, unpublished data).

  3. c Brassica juncea.

WestarNo R genesNone Rouxel et al., 2003b
Surpass400 Rlm1, RlmS Sylvestris Van de Wouw et al., 2009
Surpass501TT Rlm1, RlmS SylvestrisVan de Wouw & Marcroft, unpublished
Surpass603CL Rlm1, RlmS SylvestrisVan de Wouw & Marcroft, unpublished
45Y77CL Rlm1, RlmS SylvestrisVan de Wouw & Marcroft, unpublished
ATR-Beacon Rlm3, Rlm4 b Non-sylvestrisVan de Wouw & Marcroft, unpublished
AG-Emblem Rlm4, Rlm9 Non-sylvestrisVan de Wouw & Marcroft, unpublished
Skipton Rlm4, Rlm9 Non-sylvestrisVan de Wouw & Marcroft, unpublished
AV-Garnet Rlm1, Rlm9 OtherVan de Wouw & Marcroft, unpublished
Columbus Rlm1, Rlm3 Other Rouxel et al., 2003b
DunecUnknown Brassica juncea  

Leptosphaeria maculans isolates

A total of 133 L. maculans isolates were cultured from individual ascospores discharged from infected stubble of B. napus cultivars from nine sites across Victoria, New South Wales and South Australia as described by Sprague et al. (2006b) (Table S1). Sixty-seven of these were derived from cultivars with sylvestris resistance (Rlm1 and RlmS) – termed ‘sylvestris isolates’. Sixty-six isolates were derived from cultivars with non-sylvestris resistance and were termed ‘non-sylvestris isolates’. These cultivars contain Rlm4 but lack Rlm1 and RlmS resistance genes. All isolates were maintained on 10% Campbells V8 juice agar.

Disease assessment

Disease was assessed as either plant mortality or disease severity as follows. Plant mortality was determined either by counting the number of healthy plants compared to dead plants that had visible stem cankers caused by L. maculans, or by counting the number of seedlings to emerge and then, in the same plot, re-counting the number of plants alive at plant maturity. Blackleg severity was determined by removing plants from the soil and cutting them at the crown. The cross section of the crown was visually inspected and the percentage of the crown discoloured by L. maculans infection was estimated. Disease severity was scored as 0, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90 or 100% internal infection (Marcroft et al., 2004).

Field survey of blackleg severity

In 2004, seven National Variety Trial (NVT) yield evaluation sites in high rainfall areas of southern Australia (Fig. 1a) were identified where farmers had sowed mostly canola cultivars with sylvestris resistance and, in particular, where cv. Surpass501TT (sylvestris resistance: Rlm1 and RlmS) had high disease severity (Fig. 1b). In that same year, cultivars with sylvestris resistance were withdrawn from sale in Australia. From 2004–2010, cv. Surpass501TT (Rlm1 and RlmS) and cv. ATR-Beacon (non-sylvestris resistance: Rlm3 and Rlm4) were sown in each trial site in triplicate plots (approx. 1 m × 10 m) and monitored for blackleg severity. The locations of the trial sites differed between paddocks each year but were within 5 km of plots sown in the previous year. For both cultivars ATR-Beacon and Surpass501TT, 60 plants (20 from each triplicate plot) were chosen at random and assessed for blackleg severity as described above (percentage internal infection). To take into account environmental factors between growing seasons that influence disease pressure, blackleg severity of the cultivars with sylvestris (Rlm1 and RlmS) or non-sylvestris resistance (Rlm3 and Rlm4) was compared. The disease severity score of the cultivar with non-sylvestris resistance was subtracted from that of the sylvestris cultivar. For instance, in 2004 at Arthurton, the cultivar with sylvestris resistance had 78% disease severity and the cultivars with non-sylvestris resistance had 25%. The adjusted disease severity score for the Arthurton site was therefore 53 (78 minus 25) (Table 2). The total area in South Australia sown to sylvestris cultivars was estimated by consultation with South Australian agronomists, representatives of seed companies and farming systems groups.

Figure 1.

 Blackleg (Leptosphaeria maculans) disease severity of Brassica napus cultivars at seven National Variety Trial (NVT) sites in South Australia. (a) Locations of sites; the grey shaded areas indicate canola-growing regions. (b) Differences in disease severity of B. napus cultivars with (Surpass501TT: Rlm1 and RlmS) or without (ATR-Beacon: Rlm3 and Rlm4) sylvestris resistance (Rlm1 and RlmS) sown at these sites in relation to acreage of cultivars with sylvestris resistance sown in South Australia, averaged across all sites for each year (see also Table 2 for data from individual sites). Positive values indicate situations where there was more disease in the sylvestris than the non-sylvestris cultivar. Disease severity was not measured until 2004.

Table 2.   Blackleg (Leptosphaeria maculans) disease severity in Brassica cultivars at National Variety Trial (NVT) sites in South Australia
NVT siteAdjusted disease severitya each year
2004200520062007200820092010
  1. aAdjusted disease severity scores were calculated by subtracting the disease severity (%) of a Brassica cultivar with non-sylvestris resistance (ATR-Beacon: Rlm3 and Rlm4) from the disease severity of a cultivar with sylvestris resistance (Surpass501TT: Rlm1 and RlmS). Negative values indicate situations where the cultivar with non-sylvestris resistance has more disease than the cultivar with sylvestris resistance. n.a. indicates no data collected at that site for that year.

  2. *Indicates a significant (< 0·05) change in average internal infection compared to the previous year.

Arthurton538*−22*11*−250*13*
Frances224−16*−292*n.a.32
Keith5146−11*6n.a.n.a.29
Minlaton503*−66029*4
Mt Hope342*110438*18*
Riverton4217*−12*1−342*32*
Yeelanna51302911*−31410

Rotation of canola cultivars with different complements of blackleg resistance genes

The effect of rotation of resistance genes on disease severity was determined using a novel high-throughput system whereby plants in pots were exposed to natural inoculum from different stubbles and then disease was assessed. In 2003, a commercial crop of cv. Surpass603CL (sylvestris resistance: Rlm1 and RlmS) with severe blackleg symptoms was identified at Bordertown, South Australia. In 2004, cv. Surpass400 (sylvestris-resistance: Rlm1 and RlmS) was sown in pots in a shadehouse, where environmental conditions were not controlled. The shadehouse was at least 5 km from any canola crop. Susceptible cv. Westar (no resistance genes) was grown in the shadehouse each year and displayed no blackleg disease when assessed at maturity (data not shown). When plants were at the cotyledon growth stage, five pots (four plants per pot, 20 plants in total) were placed on the cv. Surpass603CL stubble in the Bordertown crop and were thus infected with natural inocula. When lesions were visible on any plant (usually 4–6 weeks), plants were moved into a shadehouse and mortality was scored at maturity. Also in 2004, a 1-ha plot of cv. ATR-Beacon (Rlm3 and Rlm4) was sown into the same paddock at Bordertown that had the cv. Surpass603CL stubble from the previous year’s crop. In 2005, pots sown with cv. Surpass400 were placed on the ATR-Beacon stubble, infected with natural inoculum and then moved into the shadehouse and assessed for plant mortality. Some Surpass603CL stubble from the 2003 commercial crop was still present at this site. Each year, a susceptible control, cv. Westar (no resistance genes), was exposed to the same stubble as cv. Surpass 400. In both years, all Westar plants died (data not shown).

A large-scale experiment was then established in Victoria. In 2007 two sites at Nurcong and Kalkee, and in 2008 one site at Horsham were sown with cultivars 45Y77CL (sylvestris resistance: Rlm1 and RlmS), ATR-Beacon (non-sylvestris resistance: Rlm3 and Rlm4), AV-Garnet (other resistance: Rlm1 and Rlm9) and B. juncea cv. Dune in 0·5-ha plots. The sites were isolated by at least 2 km from any other canola stubble and each 0·5-ha plot was isolated by at least 200 m from each other. In 2008 at Nurcong and Kalkee, and in 2009 at Horsham, triplicate rows of each cultivar (45Y77CL, ATR-Beacon, AV-Garnet and Dune) were sown into stubble from each cultivar. Seedlings that emerged were counted. In all rows, at least 50 plants of each cultivar grew. At maturity, surviving plants were counted in each triplicate row and the average mortality determined. Average plant mortality was only compared within a site rather than between sites to account for any differences in climatic conditions between sites. The sites were assessed for other diseases and pests to ensure that disease severity scores were caused solely by blackleg.

In addition, cvs 45Y77CL (sylvestris resistance: Rlm1 and RlmS), Surpass501TT (sylvestris resistance: Rlm1 and RlmS), ATR-Beacon (non-sylvestris resistance: Rlm3 and Rlm4), AV-Garnet (other resistance: Rlm1 and Rlm9) and B. juncea cv. Dune were propagated in pots and placed onto the stubble at Nurcong. These cultivars were assessed for internal infection at maturity. Average disease severity was determined from 20 plants for each cultivar.

When analysing these data, treatments were classified as either ‘exposed to stubble of cultivars with the same resistance genes’ or ‘to stubble of cultivars with a different complement of resistance genes’. In the former situation, the cultivar used to generate the stubble may not have been the same cultivar, but had the same characterized resistance genes as the cultivar exposed to it. In the latter situation, the cultivar used to generate the stubble had a different complement of resistance genes to that of the cultivar exposed to it. However, in one case there was a common resistance gene, Rlm1, in Surpass501TT (Rlm1 and RlmS) and AV-Garnet (Rlm1 and Rlm9).

Pathogenicity testing of L. maculans isolates

Five plants of cvs ATR-Beacon (Rlm3 and Rlm4), Columbus (Rlm1 and Rlm3) and Surpass400 (Rlm1 and RlmS) were infected with each of 20 sylvestris isolates and 20 non-sylvestris isolates selected from the 133 isolates. Plants at the cotyledon growth stage were sprayed with conidia (5 mL; 106 conidia mL−1) suspended in 0·05% Tween 20. The plants were placed in a humidity chamber for 48 h and then transferred into a shadehouse. Disease severity was determined at plant maturity as described above.

PCR assays to determine avirulence allele frequencies

The frequency of AvrLm1 and AvrLm4 was assessed in L. maculans populations as the corresponding resistance genes are present in Australian cultivars. The allele frequencies of the avirulence genes reflect whether the corresponding resistance genes are at risk of breakdown. Genomic DNA was isolated from mycelia as described previously (Sexton & Howlett, 2000). Primers were designed to amplify AvrLm1 (Gout et al., 2006), AvrLm4-7 (Parlange et al., 2009) and mating-type (Cozijnsen & Howlett, 2003) loci (Table 3). Primers for the AvrLm1 locus amplified a 1124-bp product. This indicated the presence of the AvrLm1 allele, while the lack of amplification corresponded to the virulence (avrLm1) allele in most Australian field isolates (Van de Wouw et al., 2010b). Primers for the AvrLm4 locus amplified a 788-bp fragment. This amplicon was then digested with the restriction enzyme HaeIII, which produces different banding patterns that discriminate between the presence or absence of the G120-R amino acid substitution that confers virulence towards Rlm4 (Parlange et al., 2009). Thus the virulence (avrLm4) allele is represented by a band of 324 bp, whilst the AvrLm4 allele is represented by bands of 179 and 145 bp, and bands of 323 and 141 bp are common to both avrLm4 and AvrLm4. A subset of 20 isolates avirulent or virulent towards Rlm1 and Rlm4 were inoculated onto cotyledons of cultivars with these resistance genes. Isolates were designated as virulent or avirulent on the basis of disease symptoms 17 days post-inoculation using the 0–9 scale as previously described (Van de Wouw et al., 2009). These assignments were consistent with the molecular marker profile (data not shown). Fragments of the mating-type alleles, MAT1-1 (688 bp) and MAT1-2 (443 bp), were amplified as a control for quality of DNA and to determine mating-type frequency in populations (Cozijnsen & Howlett, 2003).

Table 3.   Oligonucleotide primers used in PCR analyses
Locus amplifiedSequence (5′ to 3′)Reference
  1. aThree primers were used for the MAT locus to allow amplification of both mating-type alleles in a single PCR reaction.

AvrLm1 CTATTTAGGCTAAGCGTATTCATAAG Gout et al. (2006)
GCGCTGTAGGCTTCATTGTAC
AvrLm4 GATGGATCAACCGCTAACAA Parlange et al. (2009)
GAAGAACCCTGCTAGATAGGTAAGC
MAT a TGGCGAATTAAGGGATTGCTG Cozijnsen & Howlett (2003)
CTCGATGCAATGTACTTGGAGC
CGGAGGTGAAGTTGAAGCCG

Statistical analyses

Disease severity data from all experiments were log10-transformed to achieve normality prior to using analysis of variance techniques. Data were statistically analysed using the statistical package GenStat 5. Significant differences (95% confidence interval) were determined using analysis of variance (anova). The hypothesis tested using anova was that the mean of two alternative treatments was no greater than that caused by normal variation and error in measurement. For all significant (< 0·05) P values, the F statistic was higher than the F-critical value, indicating that the variation was not the result of chance.

All differences in allele frequencies were statistically analysed using chi-square analysis. The hypothesis tested for each gene was that the frequencies of the two alleles were not significantly different for the two isolate populations (sylvestris and non-sylvestris).

Results

Field survey of blackleg severity

The percentage of internal infection (disease severity) in cvs Surpass501TT (sylvestris: Rlm1 and RlmS) and ATR-Beacon (non-sylvestris: Rlm3 and Rlm4) was determined in seven high-rainfall sites for seven sequential years. In 2004 plants at most of the sites had an adjusted disease severity score of between 30 and 50, reflecting a high level of disease in Surpass501TT compared to ATR-Beacon (Table 2; Fig. 1b). By 2005 the adjusted scores at all sites were much lower. By 2006 the sylvestris cultivar had less disease than the non-sylvestris cultivar at five of the seven sites. The levels of disease in 2007 and 2008 were similar in both types of cultivar. However, in 2009 and 2010, disease severity in the cultivar with sylvestris resistance was higher across all sites than in the non-sylvestris cultivar. When the data were averaged across all seven sites, the level of disease within the cultivar with sylvestris-derived resistance decreased during 2004–2008. In 2008, hybrid cultivars 45Y77CL and 46Y78CL containing sylvestris resistance were commercially released. In 2009 and 2010 disease severity was higher in Surpass501TT at the trial sites than in ATR-Beacon. The increase in disease after 2008 coincided with increased acreage of these newly released cultivars containing sylvestris resistance.

Rotation of canola cultivars with different complements of resistance genes

In 2004 at Bordertown, when pots of cv. Surpass400 (sylvestris resistance: Rlm1 and RlmS) were placed onto stubble of cv. Surpass603CL (also with sylvestris resistance) from the previous year’s crop, all plants died (Fig. 2). In the following year, when pots containing Surpass400 were placed onto non-sylvestris ATR-Beacon (Rlm3 and Rlm4) stubble, which also included a small amount of stubble remaining from the 2003 sylvestris crop, only 39% of the plants died (Fig. 2). Thus, blackleg severity of these potted Surpass400 plants was influenced by the resistance gene complement of the stubble to which they were exposed. This finding was confirmed in a larger-scale experiment at Nurcong, where five cultivars (45Y77CL: Rlm1 and RlmS; AV-Garnet: Rlm1 and Rlm9; ATR-Beacon: Rlm3 and Rlm4; Surpass501TT: Rlm1 and RlmS; and B. juncea cv. Dune: unknown resistance) were sown in pots and placed onto the stubble of cultivars with different complements of resistance genes. The highest disease severity scores were recorded when pots were placed on stubble of cultivars with the same resistance gene (Fig. 3), except in the case of B. juncea, where disease severity was <5%, regardless of the stubble to which it was exposed.

Figure 2.

 Mortality of Brassica napus cv. Surpass400 (Rlm1 and RlmS) when grown in pots and placed into stubble of cultivars with either sylvestris (Surpass 603CL: Rlm1 and RlmS; 2004) or non-sylvestris resistance (ATR-Beacon: Rlm3 and Rlm4; 2005) to blackleg caused by Leptosphaeria maculans. In 2004, cv. Surpass400 was grown in pots and placed into stubble of a cultivar with sylvestris resistance. In 2005, pots of cv. Surpass400 were placed in stubble of a cultivar with non-sylvestris resistance (ATR-Beacon: Rlm3 and Rlm4). Percentage mortality was determined from five replicate pots (each containing four plants). Vertical error bars represent standard error. *Statistically significant differences as shown by anova (< 0·05).

Figure 3.

 Disease severity of Brassica napus cultivars grown in pots and placed on stubble of cultivars with either the same or different complements of blackleg (Leptosphaeria maculans) resistance genes. Five cultivars (45Y77CL: Rlm1 and RlmS; ATR-Beacon: Rlm3 and Rlm4; AV-Garnet: Rlm1 and Rlm9; Surpass501TT: Rlm1 and RlmS; and Dune: unknown resistance) were grown in pots and placed into 1-year-old stubble of three different cultivars with different complements of resistance genes (45Y77CL, ATR-Beacon and AV-Garnet) at Nurcong. Average disease severity was determined from 20 plants. Vertical error bars represent standard error. *Statistically significant differences as shown by anova (< 0·05).

Additionally at the Nurcong, Kalkee and Horsham field sites, cvs 45Y77CL (Rlm1 and RlmS), ATR-Beacon (Rlm3 and Rlm4) and AV-Garnet (Rlm1 and Rlm9) were sown in rows and plant mortality recorded. For all cultivars, fewer plants died when sown into stubble of cultivars with different complements of resistance genes than when sown into stubble of cultivars with the same complement of resistance genes (Fig. 4).

Figure 4.

 Mortality of Brassica napus cultivars sown into stubble of cultivars with either the same or different complements of blackleg (Leptosphaeria maculans) resistance genes. At three field sites in Victoria: (a) Nurcong, (b) Kalkee and (c) Horsham, three cultivars (45Y77CL: Rlm1 and RlmS; ATR-Beacon: Rlm3 and Rlm4; and AV-Garnet: Rlm1 and Rlm9) were sown into 1-year-old stubble of a cultivar with the same or a different complement of resistance genes. Vertical error bars represent standard error. *Statistically significant differences as shown by anova (< 0·05).

Allele frequencies of avirulence genes in fungal populations

The 133 L. maculans isolates (67 sylvestris-derived and 66 non-sylvestris-derived) were screened by PCR to determine the frequencies of AvrLm1 and AvrLm4 alleles (Fig. 5). No isolates cultured from stubble of cultivars with sylvestris-derived (Rlm1 and RlmS) resistance had the AvrLm1 allele, whereas 70% of isolates cultured from stubble of non-sylvestris (Rlm4) cultivars had this allele (Fig. 5a). Conversely, 71% of isolates cultured from stubble of cultivars with sylvestris-derived resistance had the AvrLm4 allele, whereas only 7% of isolates cultured from stubble of cultivars with non-sylvestris resistance had this allele (Fig. 5b). The frequency of the two mating-type alleles (MAT1-1 and MAT1-2) did not differ from the expected 1:1 ratio (Fig. 5c).

Figure 5.

 Allele frequencies of AvrLm1, AvrLm4 and mating-type locus (MAT) in populations of Leptosphaeria maculans collected from stubble of Brassica napus cultivars with different complements of resistance genes. Frequencies of AvrLm1 (a) and AvrLm4 (b) alleles in 133 isolates cultured from stubble of B. napus cultivars with either sylvestris (67 isolates from Rlm1 and RlmS-containing cultivars) or non-sylvestris resistance (66 isolates from cultivars with resistance genes other than Rlm1 and RlmS). *Statistically significant differences as shown by chi-square analysis (< 0·001).

Twenty of each of the sylvestris and non-sylvestris isolates were spray-inoculated onto cvs ATR-Beacon (Rlm3 and Rlm4), Columbus (Rlm1 and Rlm3) and Surpass400 (Rlm1 and RlmS) at the cotyledon growth stage in a shadehouse. All isolates caused similar levels of internal stem infection on cv. ATR-Beacon (Table 4). However, higher levels of internal stem infection were observed on Columbus or Surpass400 that had been inoculated with sylvestris-derived isolates (17% and 61%, respectively) than on those inoculated with non-sylvestris isolates (7% and 21%, respectively) (Table 4). These results are consistent with the AvrLm1 allele being absent in sylvestris-derived isolates (Fig. 5). There was no significant difference in disease severity in ATR-Beacon (Rlm3 and Rlm4) when it was inoculated with sylvestris or non-sylvestris isolates. Additionally, the level of disease severity was only 17% in cv. Columbus (Rlm1 and Rlm3) compared to 61% in Surpass 400 (Rlm1 and RlmS), despite these cultivars both harbouring Rlm1. These differences in disease severity may be the result of differences in virulence at the AvrLm3 locus, because both ATR-Beacon and Columbus harbour Rlm3. However, virulence at the AvrLm3 locus was not assessed in this study.

Table 4.   Disease severity (%) of cultivars of Brassica napus inoculated with Leptosphaeria maculans isolates cultured from stubble of cultivars with either non-sylvestris or sylvestris resistance
Isolate typeAverage disease severity (%)
ATR-Beacon (Rlm3, Rlm4)Columbus (Rlm1, Rlm3)Surpass400 (Rlm1, RlmS)
  1. Isolate type refers to the stubble from which the isolates were collected [from cultivars with sylvestris (Rlm1 and RlmS) or non-sylvestris resistance (Rlm4)]. Five plants of each cultivar were sprayed with conidia of 40 individual isolates (20 of each isolate type) under controlled environmental conditions. The average disease severity of the five plants was determined at maturity.

  2. *Statistically significant differences as shown by anova (< 0·05).

Non-sylvestris54721
Sylvestris5117*61*

Discussion

The long-term field experiment (7-year survey), which demonstrated the dynamic nature of the L. maculans population, provides valuable data sets for understanding the propensity of resistance breakdown. Specifically, these data showed how quickly (within 2 years) disease severity decreased following decreased selection pressure (smaller acreage of sylvestris cultivars) on the fungal population. MacKay et al. (2011) used historical long-term data sets to show that over time resistance to Puccinia striiformis, the causal agent of yellow rust of wheat, was eroded in some wheat cultivars, despite this fungus presumably not undergoing sexual reproduction. In crops such as wheat, rotation of cultivars with different resistance genes may also be a suitable management strategy to maximize the effectiveness of resistance genes. Furthermore, resistance genes that have been overcome in the past could potentially be reintroduced as part of a management strategy. To test these options, the methods established in the current study, such as the placement of pots containing different canola cultivars on stubble in the field, could be used for other annual and perennial crops to assess the use of cultivar rotation as a method for maximizing durability of resistance genes. This system is considered high-throughput, as hundreds of pots, representing multiple cultivars, can be placed at multiple geographical sites. The alternative to this method is the sowing of small plots for each cultivar at different sites – a much more time-consuming and expensive experiment that requires large areas.

Crop management strategies are essential for long-term effectiveness of resistance genes. McDonald (2010) suggests that quantitative or adult-plant resistance, resistance conferred by multiple minor genes and expressed in the stem of the adult plant, can play an important role in managing durable resistance. For the L. maculans–B. napus interaction, breakdown of major-gene resistance can be delayed when quantitative (adult) resistance is present in combination with major-gene resistance (Brun et al., 2010). These authors showed that resistance conferred by Rlm6 in a canola background that included quantitative resistance was still effective after 3 years. However, in Rlm6 isogenic lines that lacked quantitative resistance, resistance was overcome after only 3 years. The current study only classified cultivars on resistance expressed at the seedling stage, not adult-plant resistance. Adult-plant resistance probably plays a role in disease control in Australian cultivars, and this may explain variations in plant mortality observed at the Nurcong, Kalkee and Horsham trial sites.

Disease severity and fungal isolate genotyping data show that the frequency of virulent isolates in a L. maculans population towards a particular resistance gene can change upon exposure to difference complements of resistance genes in the canola cultivars sown. When cultivars with Rlm1 resistance, for example, are grown widely, isolates that are virulent towards Rlm1 will be selected and therefore increase in number in the population. However, if cultivars with Rlm4 resistance are grown the following year, isolates that are virulent towards Rlm4 will then be selected and increase in frequency, at the expense of virulent avrLm1 isolates. Therefore, rotation of cultivars with different resistance genes should minimize the build-up of isolates virulent to any particular resistance gene. It is worth noting that despite AV-Garnet (Rlm1 and Rlm9) and 45Y77CL (Rlm1 and RlmS) having a common resistance gene (Rlm1), significant differences in mortality were detected for AV-Garnet when grown on stubble of a sylvestris cultivar.

It has been reported that the AvrLm4 and AvrLm1 avirulence genes have a role in fitness. Huang et al. (2006, 2010) found that when L. maculans isolates with avrLm1 or avrLm4 alleles were inoculated onto cultivars lacking Rlm1 or Rlm4, significantly fewer and smaller lesions were formed and there was reduced fungal growth down the petiole, than when isolates with AvrLm1 or AvrLm4 alleles were used. The authors suggested that this reflected a fitness penalty associated with virulence. Such a fitness penalty may play an important role in the reintroduction into farming systems of resistance genes that have been overcome.

The frequency of virulence alleles under direct selection can lead to changes in virulence at linked loci not under direct selection. This ‘hitch-hiking’ effect was demonstrated for two loci, AvrLm1 and AvrLm6, in L. maculans, whereby selection towards the AvrLm1 locus resulted in increased frequency of virulence at the AvrLm6 locus (Van de Wouw et al., 2010b). However, selection towards the AvrLm6 locus resulted in a decreased frequency of virulence at the AvrLm1 locus (Brun et al., 2010). These data suggest that cultivars with Rlm1 resistance should not be rotated with cultivars with Rlm6 resistance. Data from the current study suggest that isolates cultured from sylvestris (Rlm1 and RlmS) cultivars have a higher frequency of AvrLm4 alleles. Presumably, this change in frequency is caused by the absence of selection towards AvrLm4, because the sylvestris cultivars do not harbour Rlm4. Therefore, it would be expected that cultivars with Rlm4 resistance would have less disease if rotated with sylvestris (Rlm1 and RlmS) cultivars.

The development of a management strategy whereby cultivars with different complements of resistance genes are rotated requires identification of these resistance genes. With the exceptions of the sylvestris-resistant cultivars and some cultivars sown in previous years that were genotyped in 2003, very little is known about the complements of resistance genes in Australian cultivars (Rouxel et al., 2003b; Van de Wouw et al., 2009). By contrast, in France, CETIOM, the technical centre for oilseed crops and industrial hemp, requires assignment of resistance genes before commercial release of canola cultivars (http://www.cetiom.fr/).

In Canada, the favourable economic return of canola has meant that many farmers are growing canola in the same field every second year (Beckie et al., 2011) and some farmers now grow canola in successive years (C. Jurke, Canola Council of Canada, Lloydminster, Saskatchewan, Canada, personal communication). The experiments carried out in the current study mimic this latter extreme situation, suggesting that the rotation of resistance genes may be an appropriate management strategy for Canada.

Although rotation of resistance genes has been suggested for the control of high-risk pathogens (McDonald & Linde, 2002a) there are no reports of deployment of this strategy to commercial crops. Therefore, the present study apparently describes the first example of how rotation of resistance genes can reduce disease severity and thus the risk of resistance being overcome. Alternative disease management strategies include pyramiding of multiple resistance genes in single cultivars and planting mixtures of crop cultivars with different resistance genes. Pyramiding of multiple resistance genes leads to direct selection towards isolates with all corresponding virulence alleles. Therefore, this type of management strategy can only be applied to pathogens where multiple virulences in one clone are unlikely, such as in Magnaporthe oryzae, the rice blast fungus (McDonald & Linde, 2002a). Resistance gene pyramids lose their effectiveness quickly when faced with a recombining pathogen population (McDonald & Linde, 2002b). As L. maculans falls into this category and is at high risk of overcoming resistance genes, pyramiding resistance genes may not be a cost-effective disease management strategy for controlling blackleg disease. Furthermore, L. maculans has already overcome two different resistance genes simultaneously in the field, Rlm1 and RlmS in sylvestris cultivars, suggesting it has the potential to overcome multiple resistance genes if pyramided (Van de Wouw et al., 2009).

The findings of this study are being adopted into an Australian blackleg management guide (Australian Blackleg Risk Assessor: http://www.grdc.com.au/director/events/factsheets), which highlights that the same canola cultivar should not be grown in close proximity for more than 3 years in a row. Not all resistance genes can be rotated with each other and an understanding of the genomic environment of avirulence genes is required for development of specific rotation strategies. It is likely that these guidelines will be applicable to canola-growing regions in other countries and also for other crops.

Acknowledgements

We thank the Australian Grains Research and Development Corporation for funding this research.

Ancillary