Speciation of fungal plant pathogens has been associated with host jumps, host domestication, clonal divergence, and hybridization. Although we have substantial insight into the speciation histories of several important plant pathogens, we still know very little about the underlying genetics of reproductive isolation. Studies in Saccharomyces cerevisiae, Neurospora crassa, and nonfungal model systems illustrate that reproductive barriers can evolve by different mechanisms, including genetic incompatibilities between neutral and adaptive substitutions, reinforcement selection, and chromosomal rearrangements. Advances in genome sequencing and sequence analyses provide a new framework to identify those traits that have driven the divergence of populations or caused reproductive isolation between species of fungal plant pathogens. These traits can be recognized based on signatures of strong divergent selection between species or through the association of allelic combination conferring hybrid inferiority. Comparative genome analyses also provide information about the contribution of genome rearrangements to speciation. This is particularly relevant for species of fungal pathogens with extreme levels of genomic rearrangements and within-species genome plasticity.
This review aims to address speciation processes in fungal plant pathogens. The key questions are: Which are the traits that drive speciation of fungal plant pathogens? Which are the underlying mechanisms of speciation and what is the genetic architecture of reproductive isolation? Is speciation mediated by the effect of single genes or a combined effect of many genes and which genes are these? What role does the host play during speciation of pathogens and to what extent are host specialization and speciation connected?
Fungi have relatively small genomes, and many species are amenable to experimental and molecular manipulation. These features make them excellent models with which to unravel the genetic, molecular, and cellular aspects of speciation. New avenues for studying speciation in fungi have been opened by advances in genome analysis and systems biology. Pioneering work has been conducted in the two model species Saccharomyces cerevisiae and Neurospora crassa. In these two systems, integrated studies using classical genetics, genomic resources, and experimental approaches have shed light on mechanisms conferring reproductive isolation between experimentally evolved lineages or divergent species (Dettman et al., 2008; Anderson et al., 2010; Parreiras et al., 2011; Turner et al., 2011).
For several important plant pathogens, experimental evolution, genetic mapping, and molecular manipulation can be considerably more challenging. Their life cycle is tightly bound to that of their host, and furthermore many cannot be crossed artificially or have a predominately asexual lifestyle. Yet insight into speciation genetics is essential to understand the emergence of new pathogenic species. Several fundamental questions relating to speciation genetics can now be addressed using the growing resource of genome data. Population genomics and comparative genomics approaches offer the possibility not only of recapturing the evolutionary history of species, but also of identifying signatures of selection associated with speciation events. Traits that have been instrumental in speciation will stand out because of an increased fixation of adaptive mutations or because they bear the signature of a selective sweep, that is, the depletion of variation within genomic regions encompassing a strongly selected gene. Ecological speciation in particular will leave such footprints of divergent selection in traits that have been a prerequisite during divergent adaptation. Persistent questions in speciation research are: how many genes are responsible for speciation and do single speciation genes exist? In a few cases, linkage of traits involved in local adaptation and reproductive isolation has been documented, suggesting that one or a few genes can drive speciation (Hawthorne & Via, 2001; Parreiras et al., 2011), but this requires that divergent selection on the particular speciation trait is very strong (Nosil et al., 2009).
In addition to ecological divergence and speciation, reproductive isolation can also arise as a consequence of chromosomal rearrangements that are fixed between diverging populations. There is little evidence that rearrangements per se reduce fitness when they are heterozygous. However, inversions or translocations will create chromosomal regions in which homologous recombination is suppressed. These regions will be ‘protected’ from gene flow, allowing mutations to accumulate between populations. Rearranged regions can thereby have an indirect effect on reproductive isolation through the maintenance of regions with elevated levels of divergence in isolated and linked genes (Rieseberg, 2001).
Chromosomal rearrangements have been described in a number of closely related plant pathogenic fungi; however, the impact on recombination has not been studied in any plant pathogen system. A particular case of chromosomal rearrangement is described as ‘genome plasticity’ in several plant pathogenic species. This involves not only chromosomal rearrangements but also the presence of accessory chromosomes and rapidly evolving genomic islands where pathogenicity-related traits can be located. The sequencing of many fungal pathogen genomes has certainly confirmed that structural reorganization between closely related fungal pathogens and between isolates of the same species is the rule rather than the exception. It has also been documented that adaptive evolution can be accelerated by an increased dynamic in repetitive regions and accessory chromosomes (see reviews by Raffaele & Kamoun (2012) and Croll & McDonald (2012)). Whether such plasticity may confer reproductive isolation has not been considered to date. However, this can now be addressed by relating observations of extreme genome plasticity to speciation by dissecting the type and origin of structural variation and the effect of rearrangements on recombination.
Below, I will summarize and discuss current literature on speciation genomics and recent genome studies of pathogenic species addressing species formation. I will finally discuss future directions in our research into fungal plant pathogen evolution and the genetics of speciation.
II. Speciation genomics
Genomic resources of various organisms have been analyzed to study speciation genetics and the genomic consequences of speciation. Several methods relying on divergence estimates, single nucleotide polymorphism (SNP) data, or estimates of nonsynonymous and synonymous divergence have been developed to identify outlier loci in the genome (Fig. 1). Measures that reflect divergent adaptive evolution include the fixation index Fst, Tajima's D, statistics based on the site frequency spectrum, and, for coding sequences, estimates of nonsynonymous and synonymous divergence or polymorphism.
Divergent selection is reflected as significantly increased levels of sequence differentiation. When coding sequences of more than two species are compared, it is possible to test the strength of positive selection that has affected a particular locus in the different species. Such branch-specific estimates can elucidate the different selection pressures that have acted on the divergent species. Within a population, strong selection of a particular trait or an advantageous allele can cause a selective sweep and a depletion of variation within the genomic region spanning the selected locus. Genetic hitchhiking will cause allelic variants linked to the selected locus to increase in frequency. Comparative population genomics approaches have the power to contrast variation within and between species and thereby uncover the present effect of natural selection in current populations and the past effect of natural selection during species divergence. Ultimately, such studies can unravel speciation scenarios and provide candidate genes involved in ecological speciation and reproductive isolation. Such comparative genome approaches have been explored in nonfungal model systems and demonstrate the power of the use of genomic data to study speciation. For example, in the flowering plant Mimulus gattus, a hybrid lethality allele hitchhiked with a strongly selected allele for copper tolerance (Fig. 2) (Wright et al., 2013). The linkage between copper tolerance and hybrid lethality confers reproductive isolation between copper-tolerant and copper-nontolerant populations. In M. gattus, hybrid incompatibility is thereby evolving as an incidental by-product of divergent adaptation in distinct environments.
1. Lessons from stick insects, primates, and fruit flies
The genomic data of 161 individuals of the stick insect Timema cristinae sampled from 28 populations have been used to elucidate the processes of population divergence and the evolution of reproductive isolation during incipient speciation (Nosil et al., 2012). Timema cristinae feeds on two distinct host species. The populations of T. cristinae feeding on these two host species show strong patterns of divergent selection and increased reproductive isolation. Reproductive isolation is expected to evolve as a consequence of ecological specialization. However, in addition to ecological specialization, allopatric and sympatric populations show varying levels of reproductive isolation, which suggests a role of reinforcement selection. Reinforcement selection acts to increase reproductive isolation between species. Among the 28 populations, Nosil and colleagues identified outlier loci – defined as loci with an exceptionally high level of between-population divergence (Nosil et al., 2012). By conducting pairwise comparisons between populations, the authors were able to show that the number of outlier loci increased with geographic distance as a consequence of weaker gene flow and demographic variability between geographically separated populations. Furthermore, some outlier loci could be associated directly with host-specific selection, which supports the importance of ecological adaptation in population divergence. Finally, a considerable number of outlier loci were exclusively present in sympatric or adjacent populations, between which the level of reproductive isolation was stronger. The authors propose that these outlier loci could play a role in the observed reproductive isolation between sympatric populations, and they conclude that incipient speciation in T. cristinae is a product of multiple factors, including divergent host selection, geographic isolation between allopatric populations, and reinforcement between sympatric populations.
For decades, a large number of studies have aimed to unravel the speciation history of humans. The availability of full genome sequences of several primate species has allowed extensive analyses of genome evolution and evolutionary relationships between species (Locke et al., 2011; Prüfer et al., 2012; Scally et al., 2012). These large genomic data sets provide information on patterns of speciation, including speciation times and demographic changes, such as alterations in effective population sizes and levels of introgression. While these parameters can be assessed as averages for the species, they can also be inferred locally across genome alignments. Local estimates of divergence, effective population size, recombination rate, and incomplete lineage sorting (ILS) supply information on the selection that acted on the genomes in the past and was associated with speciation (Dutheil & Hobolth, 2012). ILS has been used to explore evolutionary forces during the speciation period of human, chimpanzee, and gorilla (Scally et al., 2012). The amount of ILS is a function of the ratio of ancestral effective population size over the time between two speciation events. For a given time-span between two speciation events, a large amount of ILS reflects a large ancestral effective population size (Dutheil & Hobolth, 2012). Genomic regions that have experienced strong selection during speciation can be recognized as regions devoid of ILS. Species-specific traits and genes that have been affected strongly by natural selection during speciation are expected to be located in such regions where ancestral variation is absent. In fungi, genome-wide estimates of ILS have so far only been obtained in the wheat pathogen Zymoseptoria tritici (synonym Mycosphaerella graminicola) and its closest relatives, as will be described below in section V.3 ‘Host-driven divergence and ecological speciation’ (Stukenbrock et al., 2011).
Another important model in speciation genetics is Drosophila melanogaster and closely related species. Earlier studies aimed to identify single speciation genes, while more recent studies have focused on the impact of chromosome rearrangements on reproductive isolation between closely related species.
Chromosomal inversions are present in the closely related parapatric species Drosophila pseudoobscura and Drosophila persimilis (Kulathinal et al., 2009). Hybrids of D. pseudoobscura and D. persimilis are rarely found in nature, and male hybrids are sterile. Kulathinal and colleagues have investigated the genomic consequences of three large paracentric chromosomal inversions and their role in speciation (Kulathinal et al., 2009). Inverted chromosomal regions are characterized by significantly higher divergence and strong suppression of recombination, which suggests that gene flow during the initial species divergence affected the inverted regions to a much smaller extent (Kulathinal et al., 2009; McGaugh & Noor, 2012). Recombination is suppressed in these regions as a result of the fatal consequences of meiotic crossing over between heterokaryons, as 50% of the products of recombination will be either acentric or dicentric (Fig. 3). Consequently, genes located inside the inversions are ‘protected’ from gene flow with other populations and will diverge to a greater extent. In D. pseudoobscura and D. persimilis, genes located in the inverted regions include those involved in male sterility, male mating success, female species preferences, and hybrid inviability. Higher divergence in these genes may indeed promote reproductive isolation between the two Drosophila species.
To what extent chromosomal rearrangements can lead to hybrid inviability of fungal pathogens is a crucial and important question. Unexpectedly high levels of structural variation within sexual species show that individuals can mate despite variation in their chromosomal composition. Recombination may, however, be suppressed in affected genomic locations, as found in Drosophila, leaving these regions more prone to accelerated evolution.
III. Speciation scenarios in fungi
1. To be or not to be a species
Species are reproductively isolated lineages that have emerged from a common ancestor through divergent adaptation, genomic changes, or hybridization. Reproductive isolation is instrumental in speciation as it is necessary for the genetic integrity of emerging species. Fungi do not always adhere to this biological species concept, which defines species according to their reproductive boundaries. Several defined fungal species have offspring through interspecific mating, and the existence of cryptic species complexes and asexual lineages further complicates the definition of species based on the presence of reproductive barriers.
Genomics should be mentioned here as a new approach to assess lineage divergence and species boundaries in fungi. Comparative genomics studies of two or more species allow the identification of species-specific traits, and comparisons of genome content and structure. This is particularly relevant for the distinction of closely related clonal species, where genetic isolation is more difficult to assess. Species-specific traits may be genomic regions that are unique and fixed to a particular species or strongly modified compared with the homologous loci in close relatives.
Comparative genomics also allows us to infer parameters related to species histories. Whole-genome coalescence approaches are new tools for estimating genomic relationships and speciation times (Hobolth et al., 2007), as well as ancestral recombination maps and ILS (Dutheil et al., 2009). Recently, a method has been developed for inferences of genome-wide ancestral migration rates allowing fine-scale insights into the emergence of reproductive boundaries and past introgression during the divergence of species (Mailund et al., 2012). These coalescence approaches provide insights into species evolution but also serve as tools with which to characterize species delimitations (Fujita et al., 2012). Speciation times can be computed with posterior probabilities, providing robust and standardized measures of the evolutionary independence of species.
2. The origin of fungal plant pathogens
Studies based on coalescence theory and phylogeographic analyses have addressed the origin of fungal plant pathogens and asked where the species come from and how they have emerged (Stukenbrock & McDonald, 2008). These studies have used sequence data from multiple loci to reconstruct the evolutionary history of species and to infer migration routes and population structures of ancestral and contemporary populations. Most studies have focused on pathogens associated with agricultural crops, but a few have addressed speciation patterns of fungal plant pathogens in natural ecosystems.
Wild plant species have been and are reservoirs of fungal pathogens that emerge on cultivated plant species. Host shifting and speciation have been associated with crop domestication of the rice blast pathogen Magnaporthe oryzae and the wheat pathogen Z. tritici (Couch et al., 2005; Stukenbrock et al., 2007). Although host domestication has driven the emergence of these two pathogens, their patterns of speciation differ greatly. The emergence of M. oryzae on rice was associated with host shifting, the loss of sexual reproduction, and clonal speciation (Couch et al., 2005). Speciation of Z. tritici has been associated with strong host specialization, for which frequent sexual recombination has been instrumental by driving rapid adaptive evolution (Stukenbrock et al., 2011).
The Microbotryum species complex of smut pathogens has been a model system for pathogen speciation in natural ecosystems (Giraud et al., 2008). The two closely related species Microbotryum lychnidis-dioicae and Microbotryum silenes-dioicae co-exist within the same geographic area. A focus of research in this system has been the origin of co-existing yet divergent smut species on distinct Silene hosts. Using microsatellite data and coalescence approaches, the speciation history of M. lychnidis-dioicae and M. silenes-dioicae has been inferred (Gladieux et al., 2011). These species initially diverged in allopatry but recently have been brought into secondary contact, where they recurrently hybridize. The beginning of hybridization between the two Microbotryum species appears to coincide with the onset of agriculture and may have been mediated by human activities.
In summary, speciation patterns of fungal plant pathogens and endophytes have been associated with changes in ecology, for example, host shifting or host domestication, or with large changes in genome composition, such as through hybridization. A general and important finding is that new species can emerge and spread rapidly. The underlying genetics that allows the rapid establishment of reproductive barriers or their breakdown when hybridization occurs remains elusive.
IV. The genetics and genomics of reproductive isolation in fungi
A focus in speciation research has been the genetics of reproductive barriers. Several reviews discuss theoretical and empirical studies on speciation genetics (see e.g. Barton & Charlesworth, 1984; Coyne & Orr, 1998; Orr & Smith, 1998; Gavrilets et al., 2000; Rieseberg, 2001; Turelli et al., 2001; Rundle & Nosil, 2005; Rieseberg & Willis, 2007; Via, 2009). Among these is a seminal review by L. Kohn on speciation mechanisms in fungi (Kohn, 2005). Kohn in particular considers speciation in the light of the distinct genetic compositions and lifestyles of fungal species. Some of the particularities of fungal speciation are that the fungi can undergo prolonged periods of asexual reproduction, be homothallic or pseudohomothallic, consist of heterokaryotic mycelia, and exhibit different forms of vegetative incompatibility systems. These diverse life-forms promote distinct routes of speciation. Furthermore, the fungal lifecycle of pathogenic species is associated with one or in some cases more hosts, and the evolution of reproductive barriers and speciation must therefore be closely linked to the pathogenic lifestyle of the species.
Reproductive isolation arises from pre- or post-zygotic reproductive barriers. Pre-zygotic reproductive barriers include differences in ecology, such as distinct host species, differences in the timing of reproduction, and gametic incompatibility (Giraud et al., 2010). Post-zygotic reproductive barriers occur when hybrids are unfit or inviable. Hybrids are intermediate phenotypes and therefore less fit in the environments of their parents. In addition to suboptimal combinations of adaptive traits, genetic incompatibilities between parental genomes can have negative effects on hybrid viability. The genetics of hybrid incompatibilities have been studied in species of Drosophila, Saccharomyces, and Mimulus, documenting that both adaptive and neutral variation can clash in hybrids (Presgraves et al., 2003; Tao et al., 2003; Kao et al., 2010; Parreiras et al., 2011; McGaugh & Noor, 2012; Wright et al., 2013).
1. Genetic and genomic incompatibilities
Genetic incompatibilities can be mediated by allelic differences between combinations of homologous genes. The Dobzhansky–Muller (DM) model considers the effect of multiple divergent alleles on reproductive isolation (Bateson, 1909; Dobzhansky & Dobzhansky, 1937; Muller, 1942). The model suggests that the phenotypic effect of an allele depends on the genomic background and thereby assumes a strong contribution of epistatic interactions and co-evolution between unlinked loci for a given phenotype. Epistasis refers to the nonallelic interaction between unlinked genes and can be either positive or negative, depending on whether the actual phenotype is greater or smaller than the expected additive phenotype. According to the DM model, individual genes alone do not account for the evolution of reproductive isolation, but rather the antagonistic interaction between alleles from diverged genomes. Studies in species of Neurospora and Sacharomyces have confirmed a determining role of negative epistatic DM interactions in the evolution of reproductive barriers (Anderson et al., 2010; Dettman et al., 2010; Turner et al., 2011).
Anderson and co-workers have addressed the importance of genetic incompatibilities during the incipient stages of speciation in S. cerevisiae (Dettman et al., 2007; Anderson et al., 2010). They experimentally allowed the evolution of S. cerevisiae strains in two distinct environments to simulate ecological divergence. Strains from the evolved populations (500 generations) were crossed to investigate the impact of strong divergent selection on hybrid fitness. The authors not only identified the genetic determinants of adaptation but also recognized positive epistatic interactions between evolved alleles and one negative epistatic DM interaction between acquired adaptive mutations in the parental genomes. This negative DM interaction involves the evolved allelic variants of the proton-efflux pump PMA1 (Plasma Membrane ATPase) and a global regulator of mRNA encoding the mitochondrial protein MKT1 (Maintenance of K2 Killer Toxin) (Parreiras et al., 2011). Interaction of differently adapted alleles of these two genes leads to a decrease in intracellular pH and a reduced rate of glucose uptake by the yeast cells and thereby negatively affects growth. These elegant studies demonstrate the evolution of genetic incompatibilities from the population level to the cellular level and illustrate how only a few adaptive mutations can confer significant hybrid inferiority.
Speciation in yeast has also been studied in comparisons of already divergent species. Saccharomyces cerevisiae and its closest relative Saccharomyces paradoxus can mate, but hybrids are sterile. Sterile hybrids of S. paradoxus and S. cerevisiae can be generated experimentally to assess the patterns of genome content in the hybrids. DM incompatibilities responsible for hybrid sterility would be detectable as underrepresented or excluded gene combinations. Kao and co-workers screened the meiotic products of S. paradoxus and S. cerevisiae and found no major DM incompatibility but rather several loci with a small effect on complex gene interactions (Kao et al., 2010). Another study using full chromosome transfer between S. paradoxus and S. cerevisiae yielded results that also support the theory that major DM incompatibilities or speciation genes do not account for hybrid inviability between the two yeast species (Greig, 2007). The transferred chromosomes confer reduced growth of the yeast cells but never the inviability or sterility observed between real hybrid cells. Instead of major DM incompatibilities, simple DNA sequence divergence between S. paradoxus and S. cerevisiae may prevent recombination and the formation of meiotic products.
Why do DM incompatibilities play a determining role during incipient speciation of yeast but not in hybrid sterility between more distantly related species? The experimentally diverged yeast strains diverged in the presence of a strong selection pressure, and the accumulated variation reflects only adaptive variation. However, the genome-wide divergence of S. paradoxus and S. cerevisiae reflects thousands of generations of evolution and includes both adaptive and neutral variation. Hybrid sterility can thereby evolve by different means depending on the patterns of divergence and the strength of natural selection on the diverging species.
For fungal plant pathogens, distinct host species obviously impose strong selection pressures on pathogenicity-related genes. Clearly, specialized pathogens adapt to the immune system of their particular host. However, generalist pathogens must also adapt to overcome host defenses. Generalist pathogens commonly produce many different genetic variants, some of which become associated with different host species (Woolhouse et al., 2001); for example, Sclerotinia sclerotiorum and Verticillium dahliae (Carbone & Kohn, 2001; de Jonge et al., 2012). However, whether genes involved in pathogenicity also mediate an effect on hybrid sterility or inferiority remains to be shown. The ideal model systems to address this are groups of cryptic species or complexes of closely related species occurring on distinct hosts. These groups of species can include differently adapted pathogens with reduced gene flow between the species. The underlying genetics of reproductive barriers can be investigated using full genome sequencing of parental and hybrid strains to detect nonrepresented allele combinations in the hybrids and to correlate hybrid fitness to the presence or absence of certain pathogenicity-related genes.
2. Genomic divergence of clonal and selfing species
Some fungi can propagate for extended periods of time without sexual recombination (see review by Taylor et al., 1999). Consequently, genetic and phenotypic variation can accumulate in diverging lineages. As will be discussed below (section V.1 ‘Chromosomal speciation’), repetitive elements have been shown to contribute to rapid genome changes and adaptive evolution in asexual species (de Jonge et al., 2012; Xue et al., 2012). Accordingly, repetitive elements may play a crucial role in accelerating the divergence of genomes in clonal lineages. The lack of sexual recombination furthermore enhances the accumulation of genomic variants by genetic drift and mutational order, and reproductive isolation potentially evolves as a by-product of genomic divergence. The same scenario is relevant for homothallic and selfing species. These undergo meiosis; however, the parental genomes are identical to the progenitor genomes. Consequently, homothallic and selfing species can, like clonal species, accumulate increased levels of genetic variation, which subsequently can act as a barrier to sexual recombination with nonself individuals. Future comparative genomics studies of closely related homothallic and heterothallic species would provide insights into the contribution of homothallism and heterothallism to genetic differentiation of lineages.
Interspecific hybridization can occur through the fusion of hyphae or through sexual mating between individuals of opposite mating types. The occasional interbreeding of closely related allopatric species shows that reproductive barriers can be permissive and allow the formation of interspecific hybrids. In the endophytic ascomycete genera Epichloë and Neotryphodium, interspecific hybridization has been shown to occur frequently (Oberhofer & Leuchtman, 2012). Hybridization may even be associated with a selective advantage that allows the exchange of ‘host specificity’ factors between species (Craven, 2012). Similar advantageous outcomes of hybridizations have been proposed for pathogens of crop species. This idea should receive special consideration because the human-mediated dispersal of fungal pathogens has greatly enhanced this potential for allopatric pathogenic species to encounter each other and thereby form new species through hybridization (Fisher et al., 2012). Among examples of interspecific hybridizations mediated by human-introduced pathogenic species are the causal agents of Dutch elm disease Ophiostoma ulmi and Ophiostoma novo-ulmi (Brasier, 2001) and the chytrid fungal pathogen Batrachochytrium dendrobatidis, which causes chytridiomycosis in amphibian species (Farrer et al., 2011).
While hybrid speciation certainly takes place in the fungal kingdom, the genomics of hybrids is poorly understood. Hybridization involves not only the advantageous exchange of ‘host specificity’ traits but also the recombination of parental genomes and the bringing together of gene combinations that have not co-evolved in the same genomic background. Fundamental questions are how hybrids cope with genomic incompatibilities between divergent parental genomes and how natural selection acts to shape hybrid genomes. As will be discussed below in section V.4 ‘The happy rendezvous of old relatives: hybrid speciation in pathogenic fungi’, recent population genomics and comparative genome analyses provide the first insights into the genome architecture of hybrids.
4. Speciation genes
It has been proposed that single speciation genes in fungal plant pathogens may be effector genes, as these are key players in the interaction between plants and pathogens (Giraud et al., 2010). Effectors are proteins crucial for the establishment of pathogens within their hosts (see reviews by Friesen et al., 2008; Stergiopoulos & de Wit, 2009; de Jonge et al., 2011). Accordingly, pathogens with different repertoires of effector genes may be isolated on different hosts and effector genes could thus be drivers of ecological isolation and speciation (Giraud et al., 2010). Signatures of positive selection in many effector genes suggest that they indeed co-evolve with their host targets and that host adaptation is reflected at the amino acid level (reviewed in Stukenbrock & McDonald, 2009). Furthermore, both the acquisition and loss of virulence determinants such as effectors have led to the emergence of new isolated lineages and demonstrate the importance of virulence determinants to speciation in some systems (Couch et al., 2005; de Jonge et al., 2012). Whether effector genes can also play a role in hybrid inferiority is not known. Some interesting questions to address are how many host specific effector proteins are required for successful infection of a compatible host, and does the function of effectors depend on epistatic interactions with other loci? Other genes that could have a strong effect on reproductive isolation are those involved in mating, hyphal fusion or dikaryon formation, and those associated with ecological adaptation.
The above scenarios of speciation consider ecological and genetic divergence as drivers of reproductive isolation. Natural selection may, however, also play a significant role in the establishment of reproductive barriers by acting directly on mating-related traits to prevent interspecific matings. This type of selection is called ‘reinforcement selection’. In several studies of plant, animal, and fungal species, reproductive barriers have been found to be stronger between parapatric and sympatric species than between allopatric species (Anderson et al., 1980; Dettman et al., 2003) (see also review by Servedio & Noor, 2003 and references therein). This discrepancy shows that natural selection plays a role in the establishment and maintenance of reproductive barriers between species that co-exist. The selection pressure is, however, absent in allopatric species. Turner and colleagues have addressed which traits are affected by reinforcement selection in N. crassa using quantitative trait locus (QTL) mapping and genome analyses (Turner et al., 2011). They crossed allopatric and sympatric species of N. crassa and Neurospora intermedia and found an adaptive abortion of fruiting bodies only for interspecific sympatric crosses. Eleven QTL loci, several located on the chromosome determining mating type, are responsible for the observed abortive fruiting body development in N. crassa.
In fungal plant pathogens, reinforcement selection is likely to act on co-existing species that infect distinct host species. In a comparative genome study, we found increased divergence at the mating-type locus of the two grass pathogens Z. tritici and Zymoseptoria pseudotritici; this increased divergence may also result from reinforcement selection on mating determinants (Stukenbrock et al., 2010). Experimental interspecific crossing is, however, not yet possible in pathogenic Zymoseptoria species, which prevents experimental testing of this hypothesis.
V. Genomics of fungal plant pathogen speciation
1. Chromosomal speciation
Comparative genomics allows the analysis of co-linearity of genomes and the identification of rearranged regions. There is ample evidence for chromosomal rearrangements between closely related species of plant pathogens. A phenomenon called mesosynteny has been designated to describe genome evolution in several filamentous ascomycetes (Hane et al., 2011). Mesosyteny refers to a particular synteny pattern by which homologous genes of distinct species are located on homologous chromosomes, but where gene orders and directions are random. Mesosynteny appears to be particularly prominent in the Dothideomycetes. Comparative genomics studies of several Dothideomycete genomes revealed that the location of genes on homologous chromosomes is conserved. However, the order and direction of genes are randomized, creating a mesosyntenic structure between genomes (Ohm et al., 2012). It is hypothesized that mesosynteny originates from a high frequency of inversions, but not translocations during meiosis. Inversions appear to accumulate over time in the Dothideomycetes, where the extent of macro-synteny increases between more closely related species. Evidently a high frequency of inversions can be instrumental in speciation if recombination is suppressed in the affected genomic regions. Population genomic sequencing can provide more insights into the frequency, distribution and gene content of inverted regions in Dothideomycetes genomes. Such genome-based studies would create a detailed picture of the genomic dynamics in this group of fungi, and would provide a model system for further studies of chromosomal speciation in fungal plant pathogens.
2. Genome plasticity, ‘two-speed genomes’ and adaptive evolution
Comparative genome analyses among related fungal plant pathogens have revealed high levels of interspecific variation in chromosome number and organization. This includes dispensable chromosomes (Coleman et al., 2009; Goodwin et al., 2011), isochores (Rouxel et al., 2011), repetitive genomic islands (Laurie et al., 2012), laterally transferred genes (Friesen et al., 2006), and repeat-driven transduplications (Manning et al., 2013). It has been proposed that this genome plasticity can accelerate adaptive evolution in both asexual and sexual species by the rapid acquisition of new genetic diversity (Croll & McDonald, 2012; Raffaele & Kamoun, 2012).
If mutations mediated by repetitive elements drive speciation, they must confer a large fitness advantage to become fixed, for example by allowing the exploration of a new niche. A population genetic and coalescence-based study of the Magnaporthe grisea species complex has shown that emergence of the rice-infecting species M. oryzae is associated with strong clonal propagation (Couch et al., 2005). Couch and colleagues have also documented the expansion of a transposable element and the loss of a particular avirulence gene that enables the host jump from other grasses to rice. A recent comparative genomics study of M. oryzae underlines that repetitive elements are responsible for rapid evolution, including gain and loss of pathogenicity-related genes (Xue et al., 2012). Indeed, speciation of M. oryzae could have been driven by transposon-mediated gene loss and subsequent clonal propagation and genetic divergence on the new host. Other lineages of asexual pathogens might emerge in a similar manner by the acquisition or loss of pathogenicity-related traits. One example is the wilt pathogen V. dahliae.
Comparative genomics studies of two wilt Verticillium species, V. dahliae and Verticillium albo-atrum, have also revealed several species-specific and repeat-enriched genomic islands in V. dahliae (Klosterman et al., 2011). Genome size, synteny, and gene content are otherwise highly conserved between the two genomes. The genomic islands in V. dahliae were characterized as not only being repeat rich but also having a different sequence composition and a higher number of duplicated genes, which makes them a possible playground for genetic invasions, as suggested by the authors (Klosterman et al., 2011). Verticillium dahliae and V. albo-atrum are asexual species, and dynamic genomic islands may allow the pathogen to adapt more rapidly to environmental changes without the ability to undergo sexual recombination. de Jonge et al. (2012) added evidence to this hypothesis by showing that one genomic island of 50 kb was fixed in race 1 strains of V. dahliae and that this particular region encodes a virulence determinant for infection of tomato (Lycopersicon esculentum). The gene, which was probably acquired recently by horizontal gene transfer, could drive the further divergence between race 1 and race 2 populations by promoting divergent host specialization as in M. oryzae.
In conclusion, fungal pathogens can show substantial genome plasticity, and this plasticity has in several cases been directly associated with virulence-related traits. For asexual species, genome plasticity can enable rapid adaptation to changing environmental conditions and promote the formation of new distinct lineages and clonal speciation. For sexual species, genomic rearrangements appear to be confined to particular genomic regions, such as isochores or accessory chromosomes. These regions may also support the emergence of new virulence-related traits, as observed in the isochores of Leptosphaeria maculans brassicae and the accessory chromosomes of Nectria haematococca. By suppressing recombination in these regions, gene flow between diverging populations is reduced. Consequently, further divergence can accumulate. Reproductive isolation can evolve indirectly as a by-product of suppressed recombination when it increases divergence of genes involved in local adaptation or mating, as observed in Drosophila (McGaugh & Noor, 2012).
3. Host-driven divergence and ecological speciation
Host domestication and ecological specialization
The prominent wheat pathogen Z. tritici originated from a species complex of grass pathogens specialized for infecting a range of different host species (Stukenbrock et al., 2007). This Zymoseptoria species complex includes a group of co-existing, closely related species that infect wild grasses (Stukenbrock et al., 2011). We asked which traits promoted species divergence in an agro-ecosystem and in natural grasslands. Sampling and genome sequencing of ‘domesticated’ and ‘wild’ Zymoseptoria species originating from the Middle East allowed a detailed investigation of the evolutionary history of pathogenic species and the identification of positively selected genes and outlier loci that putatively have played a role during speciation (Stukenbrock et al., 2007, 2010, 2011). To investigate the role of these positively selected genes, we deleted individual genes in Z. tritici (S. Poppe & E. Stukenbrock, unpublished). While all mutants could establish a biotrophic infection in wheat, some were impaired in the ability to reproduce inside the host. A preliminary characterization of these interesting candidates suggests that they do not encode determinants of virulence but rather gene products involved in sporulation. The fixed adaptive mutations in these genes could thus reflect adaptation to asexual spore production in the wheat host. Complementation in Z. tritici with alleles from Z. pseudotritici and Zymoseptoria ardabiliae will allow us to test this hypothesis.
We also conducted a detailed analysis of genome evolution using a coalescence hidden Markov model (coalHMM) (Stukenbrock et al., 2011). We used the coalHMM to infer the proportion of sites showing ILS and found that more than 15% of the Z. tritici genome is more closely related to Z. ardabiliae than to Z. pseudotritici, and that for 15% of the genome alignment Z. ardabiliae and Z. pseudotritici are more closely related to each other than to Z. tritici. This high amount of ancestral variation, reflected as ILS, suggests that speciation of the wheat pathogen did not entail a genetic bottleneck. This is surprising taking into account the directional selection pressure imposed by the agro-ecosystem. The speciation scenario of Z. tritici also contrasts with speciation patterns in domesticated crop species, which typically have involved dramatic losses of variation and large reductions in effective population sizes (see review by Glemin & Bataillon, 2009).
The effective population sizes of Z. ardabiliae and Z. pseudotritici are smaller than that of Z. tritici despite their existence and evolution in a heterogeneous wild ecosystem. It is possible that the smaller effective population sizes of Z. ardabiliae and Z. pseudotritici reflect repeated genetic bottlenecks in an environment where the availability of susceptible hosts varies from year to year. Such spatial-temporal variation in host–pathogen systems has been documented in natural ecosystems and confirms that the genetic composition of fungal plant pathogens is dynamic (Tack et al., 2012). Although the Zymoseptoria species currently are reproductively isolated, divergence has occurred in the presence of gene flow (Stukenbrock et al., 2007). This suggests that the incipient speciation of Z. tritici did not involve DM incompatibilities as observed in the experimentally evolved yeast strains (Dettman et al., 2007). Rather, ecological divergence and the continuous accumulation of minor incompatibilities may have contributed to the reproductive isolation of Zymoseptoria species.
Genomic dynamics of alkaloid gene clusters and species diversification of plant-associated Clavicipitaceae species
Ecological divergence is also prominent in members of the ascomycete family Claviciptaceae, including the Epichloë endophytes, and in closely related symbiotic and pathogenic plant-associated species. These fungi produce different alkaloids with diverse neurotrophic effects on vertebrates and invertebrates. The diverse chemotypic diversity of alkaloids in these fungi has been proposed to reflect the broad range of host–fungus interactions. A recent comparative genomics study aimed to characterize the genomic content and divergence of 15 genomes, including sequences from ten Epichloë endophytes, three ergot pathogens, one morning glory symbiont, and one bamboo pathogen (Schardl et al., 2013). The Claviciptaceae genomes vary considerably in their content of repetitive DNA, from 0.1 to 56.9%, which suggests that repetitive elements may have contributed to the divergence of species. Schardl and colleagues focused on the content and organization of alkaloid gene clusters and found a high level of diversification between the Claviciptaceae genomes (Schardl et al., 2013). These gene clusters are highly dynamic and can vary in both copy number and sequence composition. This observed variation in alkaloid gene clusters is driven by repetitive sequences that are highly enriched in these clusters. The authors propose that diversification of alkaloid gene clusters reflects the diversification of Claviciptaceae species and their adaptation to a diverse range of host species and herbivore communities (Schardl et al., 2013). Thereby, the activity of transposable elements directly mediates the diversification of fungal–plant interactions and the divergence of Claviciptaceae lineages. Other genes related to host specialization of Claviciptaceae species have not been identified, but the above-mentioned data set provides a powerful resource for exploring genome-wide adaptive evolution in closely related endophytic, pathogenic, and symbiotic species.
4. The happy rendezvous of old relatives: hybrid speciation in pathogenic fungi
A population genomics study of the wild-grass pathogen Z. pseudotritici has revealed not only species arising from recent hybrid speciation but also a temporal dynamics of pathogenic species in natural vegetation (Stukenbrock et al., 2012). The detailed analysis of five genome sequences of Z. pseudotritici showed a mosaic of large segments of up to 100 kb without a single polymorphism interspersed between segments comprising two distinct haplotypes. Such patterns are consistent with a recent homoploid hybridization event between two haploid individuals. The nonvariable segments represent sequences inherited by the hybrid swarm from only one of the parental strains, while the variable segments represent the sequences passed on to the hybrid swarm from both parental strains. From the size distribution of variable chromosome blocks untouched by recombination in the five genomes, it was possible to infer that the first recombination event, that is, the hybridization event, occurred c. 380 sexual generations ago. The parental species have never been encountered despite repeated sampling at this location; however, both parental haplotypes present in the Z. pseudotritici genome have the same genetic distance to the closest known relative, Z. tritici. This suggests that the common ancestor of the two parental species diverged from Z. tritici before the divergence of the two parental lineages.
In summary, after the divergence of the Z. tritici lineage, the ancestor of Z. pseudotritici diverged into two reproductively isolated lineages. Two individuals from these two diverged lineages recently underwent sexual recombination to form the Z. pseudotritici hybrid. The example illustrates a rapid dynamics of species in the Zymoseptoria species complex (Fig. 4). This dynamics, including divergence and fusion of lineages, may not be unusual, as also exemplified by the Epichloë species that diverge and co-evolve with distinct host species and subsequently undergo interspecific hybridization. However, the association of hybridization in Z. pseudotritici with such a great loss of variation and the absence of backcrossing allows us to elucidate in detail the recombination history of the hybrid.
As a consequence of their relatively small genome sizes and their experimental malleability, fungal hybrids can serve as excellent model systems with which to address fundamental questions of genome evolution of hybrid species: How does natural selection act on hybrid genomes to circumvent severe losses of genetic variation? How are genomic incompatibilities accounted for? To what extent and how is backcrossing with parental species prevented? Experimental studies in yeast have aimed to address these questions and have provided solid evidence that the survival and success of hybrids depend on environmental conditions and the genotype of the hybrid (Greig et al., 2002). Closely related species of fungal plant pathogens may similarly serve as models in experimental evolution and hybridization studies to address how hybrid pathogens acquire new virulence traits, how they perform on old and new hosts, and whether ecological speciation succeeds hybridization by enhancing the isolation between hybrids and their parental species.
VI. Unraveling pathogen speciation mechanisms: future directions in genome studies
The above-mentioned studies summarize recent investigations of plant pathogenic fungi using comparative genomics and population genomics approaches. They demonstrate the massive amount of information that can be inferred from genome data, including insights into speciation processes and speciation mechanisms. In particular, comparative population genomics studies provide access to outlier loci that show signatures of positive selection between species. Such genes are good candidates for ‘specialization’ genes that have diverged in response to distinct environmental conditions. In addition to signatures of positive selection, patterns of recombination and divergence across genome sequences provide information about putative species-specific regions where recombination is suppressed in interspecific crosses and where divergence accumulates at a higher rate. As demonstrated in Drosophila, such regions can promote species divergence by preventing gene flow in recombination-suppressed regions encompassing genes involved in local adaptation (McGaugh & Noor, 2012).
There is growing evidence that genome plasticity plays a central role in the genome evolution of both sexual and asexual fungal pathogens. While structural variation in several cases has been linked to ecological adaptation, the role of this variation in speciation is still unknown. Population genomic data from several species will allow the distinction between fixed and polymorphic structural variants. Fixed structural variation between species might have played a role in speciation, and these regions might include functionally relevant genes that have contributed to species divergence.
To elucidate the exact role of outlier loci or genes located in rearranged regions, functional studies are required. Transformation approaches have been developed for many fungal plant pathogens to allow the replacement of a candidate gene with a resistance marker. By assessing the phenotype of the deletion mutant, it is possible to infer the functional importance of the relevant gene. Functional studies have in particular focused on pathogenicity-related genes, but speciation-related genes can be studied with similar approaches. Evidently, speciation-related genes could contribute to virulence; however, they can also represent genes associated with mating success and hybrid fitness. It is therefore relevant to consider phenotypic and experimental assays that can allow the evaluation of mating success and hybrid fitness in addition to virulence assays.
Finally, experimental evolution of fungal plant pathogens is a broadly unexplored field; yet, it could contribute substantially to our understanding of incipient speciation. Yeasts as the subject of studies of experimental evolution are advantageous because: (1) the cell cycle is fast; (2) cells can easily be transferred; (3) genetic, molecular, and cellular manipulations have been established; and (4) the genomes are small and well characterized. Although many fungal plant pathogens share similar advantages of axenic growth, their life cycle involves a host, which makes the design of experimental studies considerably more complicated. Nevertheless, many fundamental questions regarding the evolution of plant pathogens could be tested by experimental propagation of the fungus on its host, such as: How many mutations are required to alter virulence? Do adaptive mutations also play a role in the establishment of reproductive barriers, for example, through DM incompatibilities? What is the rate of chromosomal rearrangements, and how rapidly can they be fixed in a population? How does sexual reproduction contribute to population divergence? These questions could be addressed in experiments where populations of pathogens are subjected to divergent selection by propagation on distinct host species. Genome sequencing of progenitors and propagated isolates will allow the characterization and quantification of acquired mutations.
Integrated approaches building on genome analyses provide a powerful framework for unraveling the patterns and mechanisms of speciation in fungal plant pathogens. We can identify signatures of natural selection using comparative population genomics approaches and pick out candidate genes that have been instrumental in speciation. Scans of fungal pathogen genomes have indeed identified pathogenicity-related genes as well as genes involved in plant-associated spore production (e.g. Rouxel et al., 2011; Stukenbrock et al., 2011; de Jonge et al., 2012). These studies confirm the suitability of using evolutionary predictions in studies of speciation and specialization. We need to decipher the contribution of these genes in speciation. Of particular interest is whether virulence-related genes are involved in epistatic interactions, whereby DM incompatibilities can mediate hybrid inferiority.
Candidate genes selected from comparative genomics analyses can be studied using gene deletion and in planta assays. To date, such studies have focused on genes that play a role in virulence. For speciation-related genes, we need to improve phenotyping assays to also recapture the functional importance of candidate genes in mating and hybrid fitness.
Chromosomal speciation in fungal plant pathogens can be mediated either directly through genomic incompatibility between rearranged genomes or indirectly through local suppression of recombination. The occurrence of mesosynteny between several ascomycete pathogens demonstrates a high rate of chromosomal changes and suggests that chromosomal rearrangements such as inversions may potentially contribute to the evolution of new species. In addition, genome plasticity within species also proves to be a mechanism of rapid adaptive evolution and putatively also speciation. Rather than affecting recombination, genome plasticity acts to generate pathogens with altered genotypic and phenotypic characters, which ultimately evolve as independent lineages. Future studies focusing on the evolution of accessory chromosomes and on genomic islands and isochores formed by repetitive elements will allow us to further evaluate the role of genome plasticity in the establishment of reproductive isolation between species.
Lastly, a main focus in plant pathogen research is virulence and the adaptation to host defenses. However, in addition to the host immune system, a fungal pathogen explores a number of other ecological niches. We know very little about the adaptation of pathogens to distinct host tissues, developmental differences in different host species, and reproduction and dispersal outside their host. With the current and future molecular tools and experimental approaches, it will be possible to disentangle other details of the molecular ecology of plant pathogens and to integrate this knowledge into studies of their adaptive evolution and speciation.
I would like to thank Linda Kohn for fruitful discussions, helpful suggestions, and critical reading of a previous version of this manuscript. I am also grateful to Bruce McDonald and Julien Dutheil for their helpful comments and support. The work of E.H.S. is supported by the Max Planck Society, Germany.