Plasmodiophora brassicae, causal agent for Clubroot, has a wide host range including non-cruciferous plants (Ren et al., 2016). Clubroot has caused severe cruciferous crop yield loss of up to 50% in Canada (Strelkov and Tewari, 2005) and 20–30% in China (Chai et al., 2014). The pathogen physiology, progress in efforts to combat this disease, and an update on its epidemiology in Canada, Europe and China, have recently been the subject of various reviews (Chai et al., 2014; Diederichsen et al., 2014; Dixon, 2014; Strelkov and Hwang, 2014; Ríčarová et al., 2016). Furthermore, sources of P. brassicae R genes and genetic mapping studies have been reviewed by Piao et al. (2004); Rahman et al. (2014). To date, at least 19 R genes/QTL have been mapped for Clubroot resistance in B. napus, mostly from the study by Werner et al. (2007). However, only two Clubroot resistance genes have been cloned, Crr1a and CRa, both in B. rapa (Table 2 and Figure 1).

Sources of P. brassicae R genes in B. rapa have been particularly promising as there are many race-specific, single, dominant R genes present, as compared with B. oleracea which has a more continuous resistance profile involving various combinations of both major and minor genes (Diederichsen et al., 2009; Rahman et al., 2011). Due to the different physiological resistance mechanisms toward P. brassicae in B. rapa and B. oleracea, complete resistance is often reported in the former but not in the latter (Tanaka et al., 2006; Donald et al., 2008; Hwang et al., 2012). Recent screening efforts in China have identified B. rapa to have good resistance against Canadian pathotype 3 (Zhang et al., 2015). Hence, it is not uncommon that breeding efforts make use of B. rapa, in particular using the European fodder turnip as the recurrent parent, to introduce resistance into B. rapa, B. napus, and B. oleracea (Diederichsen et al., 2009). The introgressed regions can be characterized using SNP genotyping and next generation sequencing in order to identify and characterize these resistance genes and determine if they are the same gene or distinct. They could then be deployed in a more targeted manner.

In Canada, most oilseed B. napus shows little, or at best only moderate, resistance to Williams pathotypes 3 and 5, while B. napus rutabaga lines have better resistance (Jakir Hasan et al., 2012). However, in Australia, the majority of B. napus was found to be resistant (Donald et al., 2006), reflecting the influence of the local abiotic environments on the pathogenicity of the fungus, or the origin of host and/or pathogen influencing the resistance outcome (Donald et al., 2006; Donald and Porter, 2014). Further QTL or Genome-Wide Association Study (GWAS) studies should be performed in B. napus in order to identify these resistance genes.

Traditionally, the AB genome amphidiploid B. juncea does not show high resistance. It often shows low, moderate (Peng et al., 2014) or intermediate (Sharma et al., 2012) resistance. However, more recently, one accession of B. juncea was found to be highly resistant to the local pathotype 4 under field conditions in China (Ren et al., 2016). Another good source of resistance from the B genome is diploid B. nigra, which has been reported to display high resistance against all local pathotypes in Canada (Jakir Hasan et al., 2012; Chu et al., 2013; Peng et al., 2014). There is therefore the potential for all A, B and C genomes of the Brassica species to harbor R genes against P. brassicae.

Similar to the Clubroot pathogen, the Sclerotinia Stem Rot pathogen Sclerotinia sclerotiorum also has a wide host range, including members outside the crucifer family (Zhao et al., 2004). In the Brassica species, this pathogen can cause up to 80% yield losses in China (Mei et al., 2011) and has previously been reported to cause a loss of AU$ 23 million to the canola industry in Western Australia (Dafwa, 2015). The resistance studies on Sclerotinia Stem Rot have mainly focused on the inheritance type at different developmental stages; namely seedling leaf and adult stem stages of the host plant. Thus far, no R genes have been cloned for this pathogen (Wu et al., 2016b). However, candidate R genes against S. sclerotiorum have been identified and physically mapped on the B. napus genome (Li et al., 2015). These are NBS-LRR genes with gene copies found on chromosomes A6 and C7 (Li et al., 2015). Candidate R genes against S. sclerotiorum, belonging to the CC-NBS-LRR class, were also identified on chromosome A9 in B. napus (Mei et al., 2013). Yet, the majority of the QTL for S. sclerotiorum have been identified in the C genome (C9 and C6) (Li et al., 2015), indicating that B. oleracea is a good source of R genes for S. sclerotiorum. This was recently confirmed for many B. oleracea var. capitata genotypes from China (You et al., 2016).

There can be strong correlation between leaf and stem resistance across some specific Brassica genotypes (Mei et al., 2011; Wu et al., 2013). For example, a few Chinese B. oleracea var. capitata genotypes were found to express extremely high level stem and leaf resistances (You et al., 2016). However, when diverse sets of Brassica genotypes from different parts of the world were evaluated, for most genotypes there was generally no correlation between leaf and stem resistances (Uloth et al., 2013; You et al., 2016). Where there is a strong correlation, such cases share common genes (Li et al., 2015). This contrasting finding could reflect that high resistance against Sclerotinia Stem Rot in both stem and leaf is highly specific at the subpopulation level of the host species. The QTL region for stem resistance on chromosome C2 in B. napus was found to be homoeologous with that on chromosome A2, and falls within the syntenic region of their respective progenitor genomes (Wei et al., 2014), suggesting that the R genes could be inherited from a common ancestral chromosome, that is largely conserved.

Although, monogenic control of single dominant genes against S. sclerotiorum has been found outside the Brassica genus, for example in common bean (Phaseolus vulgaris; Schwartz et al., 2006), it is highly likely that monogenic control is present in Brassicas where only a HR and/or lesions ≤2 mm diameter have been observed on stems (Uloth et al., 2013, 2015a; You et al., 2016) or cotyledons (Garg et al., 2008, 2010b; Uloth et al., 2014; Ge et al., 2015). The genetic control of resistance genes is further complicated by the influence of inoculation methods (Zhao et al., 2004; Yin et al., 2010) and physiological traits, such as flowering time, as their QTL were found to be in the same genomic regions (Zhao et al., 2006; Mei et al., 2013; Wei et al., 2014). Despite this, Li et al. (2007) both defined the importance of the type and time of inoculation and how expression of field stem resistance becomes independent of flowering time provided assessment is delayed until 3 weeks after inoculation. Identification of resistance genes against S. sclerotiorum has been hampered by inconsistencies in time and type of inoculation. Lines that have been shown to be highly resistant in one study can be susceptible in a different study. The use of common isolates and techniques will assist in moving this research forward.

B genome Brassicas are commonly reported to have high levels of Sclerotinia Stem Rot resistance (Navabi et al., 2010). There is a range in resistance across the Australian B. napus rapeseed varieties carrying one or more B. carinata introgressions (Uloth et al., 2013). The Australian B. juncea varieties were shown to be generally more susceptible overall to Sclerotinia Stem Rot (Uloth et al., 2015a), but the Indian (Barbetti et al., 2015) and Chinese (Li et al., 2006a, 2009) B. juncea varieties displayed high levels of host resistance. Wild weedy Brassicaceae species have shown particular promise for high level resistance to Sclerotinia Stem Rot (Garg et al., 2010a; You et al., 2016), including genera such as Raphanus (Uloth et al., 2013). In particular, B. napus and B. juncea with introgression(s) from wild crucifers, including Erucastrum cardaminoides, Diplotaxis tenuisiliqua, and E. abyssinicum showed high levels of stem resistance (Garg et al., 2010a). In addition, resistance sources from the wild B. incana and B. insularis have been identified (Mei et al., 2011, 2013). Characterisation of these introgression regions using genomic technologies, and identification of candidate Sclerotinia R genes, through sequencing the introgression regions, will be the first step toward understanding the interaction between the host and pathogen.

The characterization of R genes for resistance against S. sclerotiorum in B. napus is particularly challenging. The interaction in this pathosystem is more complex than previously thought (Amselem et al., 2011). Host resistance studies in B. napus and other Brassicaceae genotypes have also demonstrated that not only did S. sclerotiorum cause pathotype-specific HR in B. napus cotyledons (Garg et al., 2008, 2010c; Barbetti et al., 2014), it also caused non-pathotype-specific resistance response in some genotypes of B. napus, Raphanus sativus, B. juncea, and B. nigra, where the host was resistant against different pathotypes tested (Barbetti et al., 2014; Ge et al., 2015). These two resistance outcomes against S. sclerotiorum rely on the specific combination between host genotype-pathogen isolate. The question is how do we distinguish these two resistances using molecular tools so that we can exploit the R genes? An enabler in the search for effective high level resistance to Sclerotinia Stem Rot has been the ability to define the sub-specific variation of S. sclerotiorum and to characterize pathotypes of this pathogen (Ge et al., 2012). This has not only provided explanation for many conflicting findings in search for effective new resistances but, more importantly, allowed identification of the first Brassica resistances to Sclerotinia Stem Rot that are truly pathotype-independent (Barbetti et al., 2014). Utilisation of this information and careful selection of pathotypes, inoculation methods and use of introgression lines, combined with technologies such as gene expression profiling and genome sequencing will be the first step to identification of resistance genes.

The pathogen Hyaloperonospora parasitica is responsible for up to 90% yield loss of Chinese cabbage (B. rapa) in China (Yu et al., 2009). Although, five R genes from Arabidopsis that encode both TIR- and CC-NBS-LRR have been cloned for resistance to H. parasitica; namely RPP2 (Sinapidou et al., 2004) and RPP5 (Parker et al., 1997) on chromosome 4, RPP8 (McDowell et al., 1998), RPP1 (Botella et al., 1998) and RPP13 (Bittner-Eddy et al., 2000) on chromosome 3 (Botella et al., 1998), no R gene has been cloned from Brassica species. However, Ppa3, a single dominant R gene against this pathogen was mapped in B. oleracea using molecular markers (Singh et al., 2012). The major locus Pp523 on chromosome C8 has also been identified to confer resistance at the adult stage in B. oleracea (Farinhó et al., 2004; Carlier et al., 2012), with a syntenic locus on chromosome 1 in A. thaliana (Farinhó et al., 2007). Another major gene RPP31 for adult resistance against Downy Mildew was genetically mapped on chromosome 5 in A. thaliana (McDowell et al., 2005). Since many more R genes have been cloned or mapped in Arabidopsis, the orthologous genes in Brassica can be investigated through genomics approaches, for example comparative analysis of genomes (Yu et al., 2014) and pan-genome analysis (Golicz et al., 2016).

Genetic inheritance of specific resistance to H. parasitica has been reported in B. oleracea (Natti et al., 1967), B. napus (Lucas et al., 1988; Nashaat et al., 1997), and B. juncea (Nashaat and Awasthi, 1995; Nashaat et al., 2004; Chattopadhyay and Séguin-Swartz, 2005). Among these reports, Natti et al. (1967) was the only one that reported the existence of separate races (1 and 2) in broccoli (B. oleracea var. italica). Sherriff and Lucas (1990) used 33 isolates, collected from four different Brassica species (B. napus, B. juncea, B. oleracea, and B. campestris) and geographic locations to infect a set of Brassica accessions (including B. napus, B. juncea, B. oleracea, B. carinata, B. nigra, B. campestris and Raphanobrassica). However, the study failed to delineate pathotypes of H. parasitica, which might explain variability in the resistance outcome. More recently, Mohammed et al. (2017) tested 131 Brassicaceae lines which highlighted excellent resistance to Downy Mildew, and was the first study to demonstrate the existence of very high levels of pathotype-independent resistance in Australian B. napus rapeseed varieties. However, no molecular analysis of these lines has been performed in order to identify resistance genes.

The resistance loci expressed in the cotyledon does not necessarily translate to adult plant resistance in B. oleracea against H. parasitica, as the resistance outcome can be different between the growth stages (Dickson and Petzoldt, 1993; Coelho et al., 1998). However, in some instances, there was association of resistance at cotyledon stage with that expressed at the adult stage (Jensen et al., 1999a; Wang et al., 2000). Similarly, screening tests on seedling cotyledons across B. napus cultivars were claimed to provide an accurate estimation of expression of field susceptibilities/resistance on seedling and more mature leaves for Downy Mildew (Ge et al., 2008). These contrasting findings indicate that the resistance outcome in the host against Downy Mildew is likely to be QTL-specific where the QTL is associated with developmental stages of the plant. Identification of such QTL should be particularly useful for breeding purposes, especially where both seedling and adult plant resistances together are required.

The genetic mechanism of resistance in the Brassica-H. parasitica pathosystem needs further clarification, as a HR has been observed in both host and non-host species (Li et al., 2011). Further, there is no evidence of clear-cut differences between Brassica host species which displayed a hypersensitive, partially resistant or susceptible reaction, compared with non-host species, during the early stages of infection, at spore germination and host penetration (Li et al., 2011). This may simply mean that the expression of the important components of host resistance in Brassicas relates to resistance mechanisms occurring during later stages of infection. Besides, H. parasitica infects a wide host range including Sinapis and Raphanus species and even species outside of the Brassicaceae, with these latter hosts mostly resistant to H. parasitica (Dickinson and Greenhalgh, 1977). Another intriguing aspect of the host resistance against H. parasitica is that the genetic control can be recessively inherited (Carlsson et al., 2004). Controlled infection studies using the wider host range, combined with molecular analysis such as gene expression and GWAS studies could be utilized as a first step toward understanding of the genetic mechanisms of resistance.

The sources of isolate, either geographic location from which the isolate was sourced or the host from which the isolate was found, influence the phenotypic outcome. Isolates sourced from different locations within the U.S.A. produced consistent disease scores within a test host genotype (Dickson and Petzoldt, 1993). By contrast, in Australia, significant sub-specific variation occurs and this needs to be considered in host-pathogen studies on resistance. Similarly, resistance was isolate specific whereby an English isolate was more virulent than a Danish isolate on B. oleracea cultivars (Jensen et al., 1999a). Unfortunately, none of these studies allows comparison of H. parasitica pathotype structure over time, or across different crops, different countries or continents. To date, it has not been possible to universally characterize H. parasitica pathotypes due to unavailability of host differentials used. Hence, there is clearly a need to develop and deploy a group of universal host differentials such that pathotypes of H. parasitica can be delineated, otherwise progress toward developing new varieties with resistance to the existing H. parasitica populations will be curtailed.

The potential for L. maculans to cause severe damage was highlighted by the widespread total collapse of susceptible oilseed rape varieties in Western Australia in the early 1970s (Sivasithamparam et al., 2005) and subsequently in the early 2000s with the collapse of single dominant gene-based resistance from B. rapa ssp. sylvestris in various regions of Australia. L. maculans also causes severe damage to oilseed rape in the UK (Barnes et al., 2010), Canada and mainland Europe (Hwang et al., 2016). The less aggressive blackleg pathogen, L. biglobosa, is more prevalent in China (Liu Z. et al., 2014) and was recently reported to be present in New Zealand (Lob et al., 2013). Although no severe B. napus yield loss has been reported from L. biglobosa, researchers have taken precautions to prevent the spread of L. maculans into China (Zhang X. et al., 2014; Cai X. et al., 2017).

We focus on the host resistance against L. maculans in this review. Published work on sources of Blackleg resistance genes in Brassica species have been increasing over the years, with the focus more recently toward quantitative resistance (Jestin et al., 2011; Huang et al., 2016; Larkan et al., 2016a), and the need to bridge this historical knowledge gap has been highlighted (Delourme et al., 2006, 2012; Rimmer, 2006). Although 12 Blackleg qualitative race specific R genes have been genetically mapped and more have been recognized (Table 2), it remains uncertain if some of these genes are actually the same gene with different nomenclature or allelic variants of the same gene. This uncertainty arises from researchers using different crosses, different isolates and different marker systems. For example the genes LmR1, cRLMm, Rlm4, cRLMrb, and LEM1 were all mapped on chromosome A7 in B. napus using Australian cultivars Shiralee, Maluka, Skipton and the French cultivar Major, respectively (Ferreira et al., 1995; Mayerhofer et al., 1997; Rimmer et al., 1999; Balesdent et al., 2001; Rimmer, 2006), with different molecular markers and therefore, until one is cloned, it is unknown if they are the same gene. This uncertainty has implications in breeding as breeders do not know whether they are using the same or different sources of resistance. Furthermore, it is recommended in Australia to rotate the source of resistance gene grown annually so the pathogen population cannot build up to cause severe yield loss (Van De Wouw et al., 2016b). The lack of knowledge of the number of resistance genes makes this more difficult to achieve and often means a pool of fewer resistance genes is used.

Another example relates to the LepR3 gene in B. napus which has been shown to interact with AvrLm1 (Larkan et al., 2013), however phenotypic studies suggest that AvrLm1 may also be recognized by Rlm1, thus raising the question of the relationship between these two R genes (Rouxel and Balesdent, 2013). The relationship between Rlm4 (Raman et al., 2012b) and Rlm7 (Balesdent et al., 2002), implicated to be responsible for the HR outcome on B. napus during attack by L. maculans carrying the allele AvrLm4 or AvrLm7 respectively, is also unknown. Cloning studies on these effectors revealed that AvrLm4 and AvrLm7 are two distinct alleles of the same gene, termed AvrLm4-7, and can induce resistance in B. napus that harbors either Rlm4 or Rlm7. However, the relationship between the two R genes, is yet to be determined (Parlange et al., 2009). It is also possible that Rlm3, Rlm4 and Rlm7 may all be allelic variations of the same gene, based on evidence showing the location of these genes on a clustered region on chromosome A7 in B. napus and not being found together in the same B. napus pure lines, with a recent study showing an unusual interplay between AvrLm4-7 and AvrLm3 (Delourme et al., 2004; Larkan et al., 2016b; Plissonneau et al., 2017a). Clearly there is a need to reconcile different nomenclatures for these genes and the relationship between them. Toward this, a roundtable discussion amongst experts in the field was conducted during the Brassica2016 conference and a consensus reached to join research efforts to bridge this gap (Plissonneau et al., 2017b). The success story of clarifying AvrLmJ1 and AvrLm5 are indeed the same Avr gene (Plissonneau et al., 2017a), is an example that should be followed by Brassica and blackleg researchers to clarify nomenclature for other genes.

Compared with other Brassica pathogens, the molecular characterization of avirulence genes in L. maculans is more widely studied (Table 3). We have now been able to study how the pathogen adapts to its new host via strict interaction at the molecular level. For example, the interaction between Rlm3 and AvrLm3 is suppressed by AvrLm4-7 when attacked by the L. maculans isolate that carries this allele (Plissonneau et al., 2016), although this isolate can be uncommon in the Canadian rapeseed fields (Kutcher et al., 2010; Zhang H. et al., 2016). There is positive influence of the frequency of the virulent allele avrLm1 on avrLm6 due to genetic linkage between the two alleles in the pathogen genome (Van De Wouw et al., 2010). This observation is important when decision making is required for which variety to rotate in the field. For example, selecting a variety with Rlm1 directly after growing a variety containing Rlm6 would not be advised as there would be higher infection due to the presence of avrLm1.

The sources of Blackleg R genes were more frequently found and more diversified in the winter type B. napus compared to the spring-type cultivars (Rouxel et al., 2003b), where Rlm4 is the most common R gene in the latter (Rouxel et al., 2003b; Marcroft et al., 2012a). Both B. rapa and B. napus have proven to be good sources of Blackleg resistance genes on the A genome. Some genes from B. rapa have been introgressed into B. napus, such as the LepR-series of R genes from B. rapa ssp. sylvestris (Yu et al., 2005, 2008, 2013). Researchers have also attempted to exploit the B genome for resistance, and R genes/QTL that are highly resistant to L. maculans have been introgressed from the B genome (Rimmer and van den Berg, 1992; Plieske et al., 1998), particularly into B. napus (Sacristan and Gerdemann, 1986; Chevre et al., 1996; Fredua-Agyeman et al., 2014). The resistance level in the introgressed B. napus was the same as that in the donor B genome source, although expression of phenotype was shown to be influenced by temperature (Plieske et al., 1998). These novel sources of R genes introgressed into B. napus will enable continued resistance against the pathogen, through expanding the number of resistance alleles available for breeding, particularly in times of climatic uncertainty.

Both LepR1 and LepR2 are involved in cotyledon resistance and are also associated with resistance at the adult stage in B. napus (Yu et al., 2005). However, the mechanism of resistance at both stages is different because Yu et al. (2005) showed that presence of LepR1 provided complete resistance, while LepR2 only offered partial resistance at the cotyledon stage, yet both genes ensured complete resistance at the adult stage. A very different pattern of resistance exists for the Chinese B. napus lines derived from winter type and spring type parents, where R gene mediated resistance in these lines conferred by both Rlm3 and Rlm4 was effective at both seedling and adult plant stages against L. maculans (Zhang et al., 2016b). Similarly, the Australian winter B. napus lines harboring either Rlm1, Rlm3 or a combination of both display stable resistance against L. maculans where resistance is effective at both seedling and adult plant stages (Light et al., 2011). Such examples, taken together, imply that while R genes can be effective at both cotyledon and adult stages, they can also act interdependently or dependently at different plant developmental stages, and are likely affected by the genotype of host and pathogen, as well as by environmental conditions. The R gene effective at seedling stage can also lie in the same QTL region as that of adult plant resistance (Raman et al., 2012b). Whether this adult plant resistance is mediated by the same R gene or by quantitative non-specific resistance genes is yet to be confirmed. An increased understanding of this molecular basis for resistance, for example through further genome sequencing and gene cloning, will enable the genes to be deployed more effectively. Further, expression of R gene mediated resistance can also be expressed differently across other host plant components; for example, Rlm1 and Rlm4 exhibit major gene resistance on B. napus pods, not only on cotyledons (Elliott et al., 2016).

Besides qualitative resistance, quantitative resistance has also been assessed and reported in B. napus against L. maculans (Huang et al., 2009). Methods to determine quantitative resistance more efficiently such as inoculation at different parts of the leaves under controlled conditions have been evaluated (Huang et al., 2014). Several QTL responsible for quantitative resistance against L. maculans have been identified in French, European and Australian B. napus cultivars (Jestin et al., 2011; Raman et al., 2012b; Huang et al., 2016; Larkan et al., 2016a). However, the genetic control of quantitative resistance against Blackleg is poorly understood as it is very difficult to compare QTL, even with the same molecular markers and lines being utilized, due to environmental differences and a lack of knowledge of the pathogen presence in the field.

The genetic inheritance of pathogen resistance, based either on a gene-for-gene interaction where the R gene is either a single dominant gene (complete resistance) or involves multiple genes (partial resistance) has been established for the four key pathogens of this review (Table 2). The R genes/QTL mapped for the key pathogens are illustrated in Figure 1.

Whilst numerous QTL and R genes have been genetically mapped (Table 2), to date, only four resistance genes have been cloned, for Clubroot and Blackleg resistance. Two Blackleg resistance genes have been cloned; LepR3 was cloned from B. napus cv. Surpass 400 (hybridization between B. rapa ssp. sylvestris and B. oleracea ssp. alboglabra) (Larkan et al., 2013) and Rlm2, an allelic variant of LepR3, from Glacier DH24287 (Larkan et al., 2015). LepR3 is characterized to have a similar function to the Cf-9 gene from tomato that confers resistance toward the biotrophic ascomycete fungus Cladosporium fulvum (Jones et al., 1994), which shares the same mode of attack as L. maculans (Stotz et al., 2014). LepR3/Rlm2 are receptor-like proteins (RLP) that interact specifically with the effectors AvrLm1 and AvrLm2, respectively, at the plant cell surface. This class of disease resistance protein is not the same as the cloned R genes for P. brassicae in B. rapa which are NBS-LRRs, thus raising the question, is RLP considered part of effector triggered immunity (ETI)? In recognizing this “gray area,” a review by Stotz et al. (2014) tried to re-classify plant immunity for apoplastic fungal pathogens such as L. maculans and Cladosporium fulvum, for which the defense mechanism seemed to deviate from the classical NBS-LRR recognition in the host cell cytoplasm where effectors are being secreted. Stotz et al. (2014) proposed the new resistance defense triggered by apoplastic fungal pathogens as “effector-triggered defense” (ETD) that involves RLPs recognizing effectors outside host cell cytoplasm. In Brassica species genome wide R gene identification studies have focused only on NBS-LRR genes, future studies should include RLPs and RLKs in the analysis, as this may pave the way for cloning of further R genes in Brassica species (Rameneni et al., 2015; Sekhwal et al., 2015; Li P. et al., 2016).

Recent advances in genome sequencing and assembly have allowed us to tap into the rich genetic resources from different crop species and their wild relatives (Bevan et al., 2017). A reference genome is useful to align and compare all the different QTL mapping experiments. By having a reference genome, the physical location of the markers in QTL maps that are linked with candidate genes can be determined (Zhang Y.-M. et al., 2016). Currently, the genome sequence of the Brassica diploids, B. rapa (Wang et al., 2011; Cai C. et al., 2017), B. oleracea (Parkin et al., 2014; Liu S. et al., 2014), and B. nigra (Yang et al., 2016) and the amphidiploids B. napus (Chalhoub et al., 2014; Bayer et al., 2017), B. juncea (Yang et al., 2016) have been published. This along with the advanced genotyping tools such as the 60 K Brassica Infinium SNP array (Clarke et al., 2016) and Genotyping by Sequencing (GBS) (Bus et al., 2012; Bayer et al., 2015; Cheng et al., 2016; Scheben et al., 2017a), have allowed rapid identification of genome-wide QTL and GWAS regions for R genes for Clubroot (Lee et al., 2016; Li L. et al., 2016), Sclerotinia Stem Rot (Wei L. et al., 2016; Wu et al., 2016a) and Blackleg (Raman H. et al., 2016).

Genome sequencing in Brassicas has now been expanded to include the B. oleracea pan-genome where genomes of nine lines were assembled and compared for structural variation (Golicz et al., 2016). Pan-genome sequencing is an important advance, as the diversity of a species is not represented by a single cultivar, and in B. oleracea only 81% genes were present in all lines. Of the 19% variable genes, those associated with disease resistance were found to be among the most prevalent. Therefore, in identification of resistance genes we should look at the variable genes, not only those in the reference genome sequence. Genome resequencing (Cheng et al., 2016) and pan-genomics (Golicz et al., 2016) approaches are promising tools to make identification, characterization and cloning of candidate R genes more achievable and reliable (Figure 2). The pan-genome approach to study structural variation of the genes in Brassica genomes, with whole-genome gene expression and methylation studies has already proved very useful toward uncovering the structure, function and evolutionary origin of R genes (Parkin et al., 2014; Golicz et al., 2016). Copy number variation (CNV) has already been studied for the AP2/ERF superfamily involved in plant stress tolerance in B. napus, CNV of R genes in Brassica is only now being investigated (Batley et al., 2016) and has been implicated in important crops such as cereals (maize and rice), Solanaceae and soybean (Springer et al., 2009; McHale et al., 2012; Xu et al., 2012; Saxena et al., 2014; Wei C. et al., 2016). This gene expansion through structural variation allows the plant to adapt to rapid changes in the environment (Cheung et al., 2009).

Brassica napus is a relatively young crop in evolutionary history that results from an accumulated series of polyploidisation events (Mason and Snowdon, 2016). Prior to hybridization, the whole genome triplication in the diploids (B. rapa and B. oleracea) was preceded by three events of whole genome duplication in the ancestor genome that have further contributed to complicated gene rearrangement, duplication and loss (Town et al., 2006; Chen, 2007). Despite the genome complexity, Brassica species have become models to study evolutionary events and how rich genetic resources can be exploited from the plant (Pires and Gaeta, 2011; Liu S. et al., 2014).

Domesticated B. napus has a lower genetic diversity compared to its diploid progenitors B. rapa and B. oleracea, evidenced by similar gene order and content between the three species due to strong genetic bottlenecks during cultivation (Cheung et al., 2009; Mason and Snowdon, 2016). By contrast, homoeologous rearrangements/recombination occurs more frequently in resynthesized B. napus, thereby contributing greatly toward genetic diversity, which provides a good source of R genes for breeding resistant B. napus cultivars (Gaeta and Pires, 2010). Recently, efforts were made to produce introgressed B. napus in a novel way, from B. juncea (A genome source) and B. carinata (C genome source) (Chatterjee et al., 2016) (rather than from B. oleracea and B. rapa) from where more sources of R genes can then be utilized and identified. While in B. napus, the homoeologous genes are often co-expressed (Chen, 2007), there could also be bias in homoeolog expression, as found in B. juncea (Yang et al., 2016) and B. rapa (Cheng et al., 2012). How the different levels of gene expression affect the stability of R genes remains to be determined. The fate of homoeologous genes, whether they are silenced, expressed, or both copies diverge where the gene function may or may not change (sub- and neo-functionalization) is a result of the genetic and epigenetic interactions between redundant genes (Comai, 2005), biased gene expression (Yang et al., 2016), gene fractionation (Subramaniam et al., 2013) or accelerated amino acid sequence evolution and positive selection (Liu and Adams, 2010). These polyploidy events could potentially result in a selective advantage, for example higher seed yield in B. napus (Osborn et al., 2003).

Although race-specific R genes are easier to manipulate for breeding resistant crops compared to polygenic resistance, which involves many genes and the effect of environmental factors, the identification of R genes in B. napus still remains a challenge due to the complexity of its genome. For example, a major QTL for resistance to S. sclerotiorum was duplicated through homoeologous non-reciprocal transposition (HNRT) in B. napus (Zhao et al., 2006). The quantitative resistance genes against L. maculans are affected by gene duplication in homoeologous regions in B. napus (Fomeju et al., 2014, 2015). L. maculans R genes, Lmr1 and ClmR1 from cultivars Shiralee and Cresor respectively, were mapped on regions that are highly duplicated within and between genomes in B. napus, and originated from the common ancestor before divergence of B. oleracea and B. rapa (Mayerhofer et al., 2005). Evolutionary information regarding resistance genes such as these is useful for breeding purposes, as it highlights specific genera/species likely to be most useful in developing new cultivars with highly effective and durable disease resistances. For H. parasitica, the single dominant gene Pp523 was mapped onto two regions on chromosome C8 and another on C5, reflecting triplicated regions in B. oleracea (Carlier et al., 2011).

The success stories of cloning Crr1a and CRa (Clubroot R genes encoding TIR-NB-LRR) from B. rapa (Hatakeyama et al., 2013) and LepR3/Rlm2 (Larkan et al., 2013) in B. napus, revealed that the R genes are organized as members of allelic variants or tandem repeats. The clustering effect of the R genes for these two pathogens provides further evidence of complexity of the genome organization. Clustering of R genes is also prevalent in association with Clubroot, Blackleg, and Sclerotinia Stem Rot resistance on both B. rapa (Kato et al., 2013) and B. napus (Delourme et al., 2004; Kato et al., 2013; Li et al., 2015). Such gene clustering effects are common in plant genomes for quick adaptive response to pathogens via recombination events (Hulbert et al., 2001; Meyers et al., 2003). Clustering of R genes may potentially enable recognition of different pathogen races carrying different avirulence genes, such as in Clubroot R genes on chromosome A8 (Pang et al., 2014). Epistatic interaction was also observed for Clubroot in B. napus (Manzanares-Dauleux et al., 2000), where a single dominant resistance gene does not explain the resistance outcome in its entirety. It was also reported that the same genomic region in B. napus can act as both QTL and/or major gene based on different pathotypes of Clubroot (Manzanares-Dauleux et al., 2000; Rocherieux et al., 2004). The clustering and duplication of R genes can also cause problems in R gene identification as these genes tend to collapse in genome sequence assemblies. Reliance on the single reference genome sequence can therefore hamper efforts and future work should incorporate novel technologies such as isolated chromosome sequencing or targeted amplification of genomic regions of interest.

Further to the complexity of R genes, resistance levels vary with different host species within B. napus for the key pathogens discussed in this review. Often, intermediate/heterogeneous phenotypes were obtained when using the pathogen isolates either derived from differential sets or from host cultivars. This could be due to genetic variation within the host cultivars and/or even within the pathogen.

With the exception of the recent development of host differentials to delineate pathotypes of Sclerotinia Stem Rot (Ge et al., 2012), host differential sets to delineate pathotypes/races of the other pathogens have generally been established for some time (as early as the 1960s). Examples include Clubroot (Williams, 1966; Buczacki et al., 1975; Somé et al., 1996), Blackleg (Mengistu et al., 1991; Purwantara et al., 2000; Balesdent et al., 2002, 2005; Van de Wouw et al., 2009) and Downy Mildew differentials (Natti, 1958). The usefulness of pathotype classification became obscured when research groups used unknown, different and/or universally unobtainable sources of host differentials that prevented application of a universal and consistent set of host differentials. The “loss” of the Downy Mildew differentials used by Natti (1958) well illustrates these challenges. Application of different host genotype differentials not commonly available across different locations and/or countries has meant findings in terms of definition of pathogen races/pathotypes is, at best, of limited use and then only to that particular locality. Use of pathogen isolates without appropriate pathotype/race classification, can easily lead to misleading R gene identification when it comes to qualitative or quantitative resistance classifications. To overcome this, an additional differential host differential set was established for Clubroot in Japan using local Chinese cabbage cultivars and local field isolates (Kuginuki et al., 1999; Hatakayema et al., 2006) and similarly for Downy Mildew using European isolates from B. oleracea (Coelho et al., 2012). Furthermore, as new pathogen isolates carrying new avirulence genes emerge as a normal consequence of pathogen evolutionary dynamics, host differential sets need to be updated and standardized to enable them to effectively delineate future new pathotypes/races.

Application of different scoring systems could also lead to complications in the identification of R genes. For example, in Downy Mildew, at least three scoring systems have been applied (Dickson and Petzoldt, 1993; Wang et al., 2000). It is also imperative that a robust race/pathotype differentiation system be utilized to separate different and group similar races/pathotypes. While for L. maculans and P. brassicae established race differentiation systems are robust, for H. parasitica and some other pathogens this is not the case. A little utilized but extremely robust system was developed and utilized for the barley scald fungus, Rhynchosporium secalis by Goodwin et al. (1990). The Goodwin et al. (1990) pathogen isolate differentiation system utilizes an octal nomenclature classification that codes pathogen isolates according to their pathogenicity (expressed as disease severity) on the host differentials, mathematically confirming appropriateness of differential host and then grouping these genotypes into sets according to their genetic inheritance. This allows continuity with any previous studies that have different nomenclatures. Its suitability is illustrated by its recent successful application to define the first pathotypes within S. sclerotiorum populations on B. napus within Western Australia (Ge et al., 2012). Despite these differentials being made available worldwide, upon request, there has been little uptake outside of Australia, India and Eastern Europe to date. The Goodwin et al. (1990) pathogen isolate differentiation system applied to H. parasitica isolates would also be highly beneficial for the same reasons.

In B. napus, resistance toward race-specific pathogens such as L. maculans and P. brassicae has been overcome. New virulent pathotypes of P. brassicae that have the potential to cause resistance breakdown in B. napus have been identified (Strelkov et al., 2016; see Table 3). Mass cultivation of cultivars containing particular R genes favors selection of pathogen strains that are better adapted or have greater virulence due to strong selection pressures exerted on the pathogen. Care should be taken that introgressed lines with single dominant gene resistance are not widely grown homogeneously in commercial fields year after year (Marcroft et al., 2012b). Whilst they may be effective in the short term (one to two seasons), they will likely lose their efficacy over a period of time, as occurred in as little as 3–4 years for L. maculans (Li et al., 2003). In France, complete deletion of the Blackleg avirulence gene AvrLm4-7 was found (Daverdin et al., 2012).

The genetic mechanism of resistance in Brassica is different for each pathosystem. For the B. napus-L. maculans and -P. brassicae pathosystems, it is generally straightforward because distinct major gene resistances and respective cotyledon and root phenotypes have been well established using the Blackleg differential and Williams (1966) systems, making host single dominant R gene effects relatively easy to study. In contrast, in the B. napus-S. sclerotiorum pathosystem, although a hypersensitive reaction was found for S. sclerotiorum on cotyledons (Garg et al., 2008; Ge et al., 2015) and stems (Garg et al., 2010a), it occurred in relation with specific host genotype-S. sclerotiorum isolate or pathotype combinations. In other words, the resistance outcome is pathotype-dependent and genotype expression of resistance is tied to the selected isolate or pathotype (Barbetti et al., 2014). Further, any QTL or gene expression results are tied to these particular host genotype-S. sclerotiorum isolate or pathotype combinations. The fact that seedling vs. adult and leaf vs. stem resistances are generally independently expressed (Uloth et al., 2013, 2014), has inhibited advancement of our understanding at a molecular level in comparison with that for L. maculans. Initially, for S. sclerotiorum, until there is wider utilization of the differentials developed to distinguish distinct pathotypes of S. sclerotiorum, as developed by Ge et al. (2012), it may be productive for molecular studies to focus more on using pathotype-independent resistances (i.e., genotypes showing resistance to all pathotypes) as recently highlighted by Barbetti et al. (2014). Failure to define sub-specific variation in S. sclerotiorum remains an impediment to rapid progress in our understanding at a molecular level with the host genotype-S. sclerotiorum pathosystem. Further, it is evident that gene expression within the pathosystem involving S. sclerotiorum can be differentially expressed across different host genotypes depending upon their sensitivity toward environmental effects, for example small changes in temperature (Uloth et al., 2015b). Such variation further challenges outcomes for molecular studies into resistance against this pathogen, particularly with variations in expression of resistance across different temperatures, especially when temperature-phenotypes occur (Uloth et al., 2015b). Similar challenges occur in relation to the host genotype-H. parasitica pathosystem and likely occur wherever sub-specific pathogen variation occurs but has not been definitively characterized. In addition, comparisons made with known S. sclerotiorum pathotype delineations across approximately 60 Western Australian isolates showed that there was no relationship between pathotype with microsatellite or with IGS haplotype (Clarkson et al., 2017). Clearly there is a need for further molecular work, for instance metabolomics and proteomics, in addition to genomics, in conjunction with appropriate phenotyping across a range of environmental conditions to further investigate and understand these challenges both now and for the future.

Climatic conditions can influence the interaction between plants and pathogens and the pathogen distribution, contributing directly to the disease outcome on the host (Barbetti et al., 2012). Abiotic conditions including temperature and soil nutrient content could influence the resistance spectrum on the host A. thaliana against powdery mildew pathogens (Jorgensen, 2012). In the L. maculans-B. napus pathosystem, temperature effect influences the phenotypic outcome at different plant growth stages. For example, in general, during infection at nearly all growth stages from cotyledon to pod at higher temperature regimes (e.g., 18/24°C), resistance expressed at the seedling stage was positively correlated with the adult stage, as compared to lower temperature regimes (e.g., 11/18°C) (Li et al., 2006b). Warmer temperature and high humidity levels favored L. maculans colonization of the host during a Rlm6-mediated resistance study on B. napus (Huang et al., 2006), confirming a related study on Rlm1-mediated resistance by Badawy et al. (1992). Cold temperature and low soil acidity slow down or inhibit P. brassicae infection on B. napus, while the influence from soil micronutrient stops at high inoculum concentration (Gossen et al., 2014).

Novel sources of R genes for Blackleg can be sourced from other A- and C-genome Brassica species (Chen et al., 2010) or from the B genomes in B. nigra (Chevre et al., 1996), B. juncea (Christianson et al., 2006), and B. carinata (Plieske et al., 1998). Wild Brassicaceae species, such as Sinapis arvensis, also serve as a good source of R genes for Blackleg (Snowdon et al., 2000). However R genes from other wild species remain to be tapped. Besides identifying candidate R genes through comparative genomic analysis, more screening work on wild and land races, including species from the Brassicaceae family, should be performed against the main pathogens to identify resistant plants. This would broaden the gene pool for introgression of novel R genes or alleles into Brassica cultivars. Combining this information with genome sequencing will enable characterization of these novel resistance genes.

Although effector genes are usually unique and do not often share structural similarity but encode very diverse proteins, it is still possible to identify an R gene that could recognize common or near identical effector gene(s) in different pathogen pathotypes/races/isolates, which already exists naturally in some plants/crops (Dangl et al., 1992). This type of R gene is useful and can be transferred between hosts that are closely related or within the same genus or family. Some R genes can also be transferred from host to non-host using GM technologies, providing broad-spectrum immunity against one or more pathogens. One such successful example is the pepper Bs2 gene against bacterial spot disease transferred from its host pepper into non-host tomato (both members of Solanaceae family; Tai et al., 1999). The Bs2 gene encodes the R gene NBS-LRR class (Tai et al., 1999) while the gene avrBs2 in Xanthomonas campestris pathovar vesicatoria, causal pathogen of bacterial spot disease, is essential for virulence activity and is highly conserved between and within the pathovars of X. campestris (Kearney and Staskawicz, 1990). This has opened up new possibilities to protect economically valuable crops, that lack R gene resources, from disastrous pathogens.

Continuous efforts are on-going to breed resistant B. napus cultivars through introgression. B. rapa remains a good source of race specific R genes that often result in high resistance against Clubroot (Strelkov et al., 2006) and Blackleg (Yu et al., 2008). For Clubroot, a single dominant gene from resistant B. rapa turnip morphotypes was successfully introduced into Chinese cabbage (Yoshikawa, 1980). Transfer of a Clubroot resistance gene was also performed from B. napus into B. oleracea, with the resistance gene believed to also be carried in the A genome of B. rapa (Chiang et al., 1980). In Canada, efforts to transfer a Clubroot resistance gene from the B. napus hybrid “Mendel” into a local susceptible spring B. napus cultivar was promising (Rahman et al., 2011). Transfer of a Clubroot R gene was also made from B. rapa Chinese cabbage “Qulihuang” into B. napus “Topas” using marker-assisted selection (Hirani et al., 2016). For L. maculans, R genes have been genetically mapped and identified in the A genome of B. napus (Alamery et al., in press). For S. sclerotiorum, introgression lines of B. juncea carrying genomic segments from the wild Brassicaceae B. fruticulosa demonstrated marker trait association for Sclerotinia Stem Rot resistance (Rana et al., 2017). Within breeding programs, it is important to make informed decisions about the source of novel R genes and which parental genomes to cross, because the R gene from the parents may be masked in the progenies of the resynthesized genomes due to one or more epistatic interactions, such as was found for a Clubroot R gene (Werner et al., 2007). These challenges can be overcome by the latest genome editing techniques, such as CRISPR-Cas9 (Scheben et al., 2017b), as was recently reported in maize with precise mutation on the targeted allele (Svitashev et al., 2016; Liang et al., 2017).

The B genome offers good sources of resistance against two other pathogens besides P. brassicae and S. sclerotiorum; White Rust and White Leaf Spot (A. candida and P. capsellae, respectively). For White Rust, potential sources of resistance genes come from B. nigra and B. juncea (Westman et al., 1999), while resistance against White Leaf Spot potentially comes from B. carinata (Gunasinghe et al., 2016). For White Leaf Spot, in contrast with Clubroot and Blackleg, B. rapa is more susceptible than B. napus (Gunasinghe et al., 2014), but as with Downy Mildew, there is good resistance to White Leaf Spot in B. oleracea. In fact, complete resistance was recorded in B. oleracea var. capitata (Gunasinghe et al., 2016). However, B. juncea originating in India or Australia is highly susceptible to White Leaf Spot (Gunasinghe et al., 2014).

To breed resistant cultivars, it is important to consider the fitness cost to pathogen isolates when they do overcome host resistances. We can study the pathogen avirulence/virulence genes in the field that are adapted to the host cultivar and determine the genetic changes in the pathotype that cause heavy fitness consequences (Vera Cruz et al., 2000; Delmas et al., 2016). It is also important to consider the host R gene as selection pressure on the pathogen can affect the stability of the R gene. In terms of the linked genes in L. maculans, AvrLm1, and AvrLm6, high selection pressure on one Avr gene can cause a change in allele frequency of the other (Brun et al., 2010; Van De Wouw et al., 2010). Understanding the evolutionary potential between the pathogen and the host is essential in developing a durable resistant Brassica crop (McDonald and Linde, 2002).

In the L. maculans-B. napus cultivar “Surpass 400” pathosystem, mixed inoculation of isolates (avirulent and virulent) with contrasting phenotypic effect (resistant vs. susceptible), can render B. napus resistant, even with only a small volume of avirulent isolate in the mixture of the two isolates (Li et al., 2006a). Such “protection” effects are also observed in the field in Australia and have also been observed in the Phytophthora megasperma f.sp. glycinea–soybean pathosystem where a mixed isolate inoculation controlled the lesion spread (lower rate of spread; Ward, 1983). While the mechanism of improved resistance involving mixed isolate inoculations remains unknown, this approach does offer potential for possible “immunization” of host plant and further investigations into such are warranted. However, it should also be noted that the reverse can happen where co-infection results in increased disease susceptibility of the host. For example, co-infection of different pathogens such as A. candida (causal agent for White Rust) and H. parasitica (Downy Mildew) where the two pathogens can grow together within leaf and/or stem tissues of the same host plants (Cooper et al., 2008). Cooper et al. (2008) also showed that pre-infection of virulent A. candida can influence an otherwise avirulence outcome for H. parasitica such that it became virulent in B. juncea, as was also shown by Singh et al. (2002). However, pre-inoculation of the virulent H. parasitica on White Rust-susceptible B. juncea host promotes better resistance to White Rust disease (Singh et al., 2002), suggesting that type of isolate, infection sequence and pathogenicity of the isolate could determine the resistance outcome in the host (Kaur et al., 2011). The molecular basis of these co-infections across multiple pathogens remains unknown.

In addition to the up-regulation of HR-related genes characterized by signaling hormones (Chu et al., 2014; Nováková et al., 2016b; Uloth et al., 2016), regulatory genes such as the transcription factors WRKY (Zhou et al., 1997) and calmodulin-binding transcription activator (CAMTA) (Rahman H. et al., 2016) have also been implicated to regulate the disease response in Brassica pathosystems (Zhou et al., 1997). The WRKY gene was studied in B. napus-S. sclerotiorum and -Alternaria brassicae pathosystems, where gene expression was increased upon pathogen infection alongside defense-related hormones (Yang et al., 2009; Zhao et al., 2009). This gene was shown to enhance resistance against Downy Mildew in the Chinese broccoli B. oleracea var. italica (Jiang et al., 2016). Besides the WRKY gene, for the same host species, a pathogenicity-related defensin gene, which encodes defense-related cysteine-rich protein has also been shown to improve resistance against Downy Mildew (Jiang et al., 2012) and S. sclerotiorum (Zarinpanjeh et al., 2016). Identification of candidate genes for Clubroot resistance revealed Rho proteins may be involved (Kato et al., 2013). In Head Smut in maize, ZmWAK was successfully identified as a regulator gene with quantitative effects on host resistance expression (Balint-Kurti and Holland, 2015; Zuo et al., 2015). Further characterization of these genes should be performed in Brassica species, through genome wide identification and gene expression studies to determine their role in disease resistance.

To control disease outbreaks in crops, it is imperative that pathogen population variability in the field can be quickly identified. This can be achieved by having a full set of genetic markers for different pathogen pathotypes, not only contained within the local fields but universally. For example, initiatives for P. brassicae showed that Cr811 was a good candidate gene to differentiate pathotype 5 from the rest of the pathotypes (Feng et al., 2016).

Molecular characterization of pathogen effectors is gaining momentum. With better genome assemblies and higher gene prediction accuracies, we will soon have better insights into the structures and functions of effector proteins, which allows more efficient screening and identification of R genes (Gibriel et al., 2016). Elicitor candidates for L. maculans (Wilson and Howlett, 2005; Nováková et al., 2016a), and effector candidates for S. sclerotiorum (Guyon et al., 2014), are both responsible for inducing HR. RNAi studies established that the transcription factor LmStuA in L. maculans regulates expression of effector genes (Soyer et al., 2015). Genomic studies researching L. maculans effectors AvrLmJ1 (Van de Wouw et al., 2014), AvrLm1 (Gout et al., 2006), AvrLm2 (Ghanbarnia et al., 2015), AvrLm3 (Plissonneau et al., 2016), AvrLm4-7 (Parlange et al., 2009), AvrLm1-2-6 (Fudal et al., 2007), and AvrLm11 (Balesdent et al., 2013) have provided new understandings on how they are strategically located within regions that are rich in transposable elements, such that they can be lost or inactivated rapidly during sexual reproduction (Rouxel et al., 2011; Howlett et al., 2015). This fungal genomic structure is unique to L. maculans and often the effectors display little or no sequence similarity to each other (Van De Wouw et al., 2010; Rouxel et al., 2011). Specific recognition between AvrLm4-7 and Rlm4 and Rlm7 genes and how the R and Avr effector proteins interact structurally has been examined using crystallography (Blondeau et al., 2015). There should be a focus on studies that provide additional molecular details at the gene level which would further our understanding of the factors leading to virulence and susceptibility associated with R-gene mediated resistance.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.