What is the evidence for sexual reproduction of Phytophthora infestans in Europe?
The biology of late blight of potato and tomato, caused by Phytophthora infestans, changed when sexual reproduction by the pathogen became possible in many parts of the world, including Europe. In northern Europe, especially Scandinavia, there is increasing evidence that the pathogen is reproducing sexually on a regular basis, although in other regions further south or to the west it appears to reproduce primarily in a clonal manner. The presence of both mating types, the production of viable oospores, and observations of fields with soilborne sources of inoculum are consistent with sexual reproduction. Studies with different marker systems have revealed a population structure without any dominating clonal lineages in Scandinavia, and that is most easily explained by sexual reproduction. Phytophthora infestans recovered from the soil can also be linked to parental genotypes using likelihood-based methods when codominant markers are used. A synthesis of all the available data points to a second centre of sexual reproduction in northern Europe.
The pathogen Phytophthora infestans and the disease it causes (late blight of potato and tomato) are some of the veterans in the history of plant pathology. Most plant pathologists are aware of the historical significance of this disease in potatoes and the social impacts of P. infestans becoming a chronic feature of potato production (Large, 1940). The disease first appeared in Europe in the 1840s, and since then it has been a major problem wherever potatoes are grown. The pathogen that caused the earliest epidemics (up to the 1970s) was identified as a single clonal lineage (Spielman et al., 1991), although analysis of herbarium specimens (Ristaino et al., 2001) and other data (Gómez-Alpizar et al., 2007) indicate that several different mitochondrial haplotypes were present in the early collections. Regular sexual reproduction of the pathogen was unknown until it was reported by Smoot et al. (1958) and Galindo & Gallegly (1960). Their studies revealed that P. infestans is heterothallic, and requires two mating types (referred to as A1 and A2) in order to form oospores. The clonal lineages responsible for the epidemics that took place before the 1970s were all of the A1 mating type, and this explained the absence of oospores in most potato production areas, with the exception of the Toluca Valley in Mexico.
A second emigration of P. infestans (Spielman et al., 1991; Fry et al., 1992) enabled sexual reproduction of the pathogen, and this new emigration has displaced the older A1 population that was dominant in most of the world. In many regions, including most of North America, South America, Asia and Africa, P. infestans still reproduces primarily in a clonal manner. Clonal survival is so common that one popular textbook on plant pathology states that ‘sexual reproduction is rare in nature’ (Agrios, 2005) when referring to P. infestans. If the pathogen survives as a clonal lineage, this means that disease control measures should be directed towards pathogen-free planting material and reducing the impact of sporangia that may come from other infected plants, whether these plants are from the same or another field. However, increasing evidence from northern Europe is consistent with a second centre of regular sexual reproduction by the pathogen outside of the Toluca Valley, and this evidence is reviewed here. Regular sexual reproduction by P. infestans directly affects disease management strategies. An immediate consequence is that there is a source of initial inoculum in the soil, thus restricting the frequency with which potato can be grown. A more long-term effect of sexual recombination is that the evolutionary potential of the pathogen increases (McDonald & Linde, 2002). Given that host-plant resistance using specific resistance genes has received renewed attention, the ability of the pathogen to overcome these resistance genes may be linked to its evolutionary potential. Thus, sexual reproduction may limit the usefulness of this disease control strategy, although no data that can confirm this phenomenon are currently available.
A general summary covering the evidence that sexual reproduction of P. infestans occurs in Europe is lacking. One presentation that began to summarize these data was given at the Global Initiative on Late Blight (GILB) Conference in 2008, and focused on the way in which the epidemiology of late blight could be affected by oospores (Andersson et al., 2009). The presentation covered aspects such as mating distribution, visual observation of oospores formed in plant tissue, and the evidence for a soilborne source of inoculum, but did not include any of the evidence based on genetic analyses of the population structure of the pathogen. A presentation was also made at the 10th International Epidemiology Conference, and the present paper has grown out of that presentation. Given the importance of sexual reproduction of P. infestans, its absence in many potato production systems, and the threat that it represents to late blight control, a review of the evidence that this pathogen reproduces sexually is timely. Such a review would also organize and categorize the clues that are left by sexual reproduction of an organism, and this would assist those studying other pathosystems to decide if sexual reproduction of the pathogen is present or not.
Presence of both mating types
The first requirement for a sexually reproducing population is the occurrence of both mating types. While the original reports of the A2 mating type were confined to Mexico, the new emigration of P. infestans took a new population, which included the A2 mating type, to Europe, probably sometime in the late 1970s. The first reports of the A2 mating type in Europe came from Switzerland (Hohl & Iselin, 1984), followed by reports from many European countries, such as those of Malcolmson (1985), Shaw et al. (1985), Tantius et al. (1986) and Kadir & Umaerus (1987). See Drenth et al. (1993) for a more comprehensive list of reports of the A2 mating type.
However, the presence of both mating types does not necessarily mean that the pathogen population is reproducing sexually, although it obviously is a prerequisite. It is possible for P. infestans to reproduce clonally even if both mating types are present in a country or region. Two or more separate clonal lineages could exist in parallel, surviving on potato tubers during the winter, as when there was only a single mating type.
Production of oospores
For a population to reproduce sexually, both mating types would have to be present in the same field, infect the same leaf, and give rise to oospores. If one mating type is present in only a small proportion, and if it is found only in a limited area, production of oospores would be more limited than if there were equal numbers of the two mating types and they were evenly distributed. This density dependence, often called the Allee effect, has been modelled for some heterothallic plant pathogens (Garrett & Bowden, 2002), but not specifically for P. infestans. Thus, the distribution of the different mating types can provide information on the probability of formation of oospores and sexual reproduction. Equal numbers of A1 and A2 would lead to larger amounts of oospore production. An even distribution would also result from a sexual population. In a study in the Netherlands, Zwankhuizen et al. (2000) reported that 25% of isolates and 44% of genotypes were of the A2 mating type (out of a total of 1712 isolates and 1048 genotypes). In a Finnish study, 23% of isolates were of the A2 mating type (Lehtinen et al., 2007), although there was substantial variation from year to year, with A2 being dominant in 2000. A Nordic collection made in 2003 revealed a 60:40 distribution of A1:A2 (Lehtinen et al., 2008).
After reports of the occurrence of both mating types of P. infestans in Europe, a limited number of studies have surveyed individual fields for both the presence of both mating types, and the presence of oospores in leaf tissue (Götz, 1991; Drenth et al., 1993; Andersson et al., 1998; Hanson & Shattock, 1998). Detailed studies of organic potato production fields in Sweden and Norway, not treated with any fungicides, revealed oospore production in approximately one-third of the fields, albeit only after incubation of the Norwegian samples (Dahlberg et al., 2002; Hermansen et al., 2002). Similar results were obtained in a Swedish survey the following year (Hjelm, 2003). A Finnish study was directed towards fields with both a history of late blight and that also showed early infection. Oospores were observed in stems collected from the field, and in leaf tissue after incubation (Lehtinen & Hannukkala, 2004). In a Dutch study, factors such as potato cultivar, crop rotation and soil type affected production of oospores in leaves showing two or more lesions in the field, after a 3-week incubation period (Kessel et al., 2002).
The production of oospores in plant tissue does not necessarily mean that they are a source of inoculum for future late blight epidemics. They would have to survive and be able to infect plants in the following growing season. Stains and plasmolysis tests have been used to check the viability of oospores (Pittis & Shattock, 1994), although this may not indicate if the oospores can germinate and infect. Even under laboratory conditions, it is difficult to germinate oospores. Smoot et al. (1958), Pittis & Shattock (1994) and Strömberg et al. (2001) reported maximum germination rates of 10%, 35% and 10%, respectively. A number of studies tested the viability of oospores by burying them in soil and using a bioassay (Drenth et al., 1995; Turkensteen et al., 2000) in which the soil was saturated with water and potato leaves were floated on the surface. Infected leaves could be then recovered and the pathogen isolated from them. These studies indicated that oospores could survive in the soil for up to 4 years.
Inoculum sources in the soil
Complementary studies have indicated the possibility of soilborne sources of inoculum. A study conducted by Andersson et al. (1998) indicated that there was a soilborne source of inoculum that was able to survive two winters and an intervening crop of barley. The fact that the distribution of infected plants in the second crop matched that of the first crop suggested that there were a large number of localized inoculum sources. A study in Denmark indicated that disease onset has become earlier with more frequent potato production (Bødker et al., 2006), possibly involving inoculum sources in the soil. In a Finnish study (Hannukkala et al., 2007), historical data on the occurrence of late blight from the period 1938–1992 were analysed. In the period 1998–2002, late blight occurred on average 24 days earlier than in other periods. Analysis using weather data could only partially explain the earlier outbreaks of late blight, and the authors concluded that the shift could also have been the result of oospores in the soil.
In other studies, soil taken from fields where oospores were observed has also been able to give rise to a limited number of infections when analysed with the previously mentioned bioassay to determine the viability of oospores (Drenth et al., 1995; Turkensteen et al., 2000; Lehtinen & Hannukkala, 2004). While these studies probably indicate that P. infestans has survived in the soil, the conditions of the bioassay are different from those that usually occur under natural conditions. Thus, they are only an indicator that viable oospores are present in the soil, not that they can germinate and initiate late blight epidemics under natural conditions.
Population studies with various marker systems
The use of marker systems has made it easier to study whether sexual reproduction of P. infestans takes place under natural conditions. A sexually reproducing population would give rise to a large number of different genotypes, whereas one reproducing clonally would produce similar, if not identical, individuals. A number of genetic markers have become available which can be used to study P. infestans. The first of these markers were two polymorphic allozyme loci, referred to as Gpi (glucose phosphate isomerase) and Pep (peptidase) (Spielman et al., 1990), which enabled researchers to study a total of three stable markers when mating type was included.
The identification of a moderately repetitive nuclear restriction fragment polymorphism led to the development of the first molecular marker (called RG57) that could be used to study P. infestans populations (Goodwin et al., 1992). The use of the RG57 probe provides many more loci and enabled a wide number of population genetic studies to be carried out. Initially, 25 fingerprint bands could be identified by this probe (although two bands were inconsistent), and subsequently an additional four bands were also identified. Another marker that has been used is mitochondrial DNA (Carter et al., 1990; Griffith & Shaw, 1998), although this is not inherited in the same manner as traits on the nuclear chromosomes. Another approach is to use the method referred to as amplified fragment length polymorphisms (AFLP) (Vos et al., 1995).
More recently, simple sequence repeats (SSR microsatellites), developed by Knapova & Gisi (2002) and also by the James Hutton Institute (JHI) (Cooke & Lees, 2004; Lees et al., 2006), have also been documented for use as markers. These seem to be the markers of choice, and have been used in a large number of population studies. These markers seem to be stable within clonal lineages and, in a recent study based on a collection from Nicaragua (Blandón-Diaz et al., 2012), only limited variation was seen at some loci, despite considerable variation in virulence phenotype and metalaxyl resistance. Different population studies of P. infestans from Europe, along with the marker system used and the general conclusion about clonality and variability, are presented in Table 1. The most recent studies were performed exclusively with SSR markers, and the large number of studies from northern Europe is probably a reflection of the suspected sexual reproduction of the pathogen in this area.
Table 1. European population studies of Phytophthora infestans using different marker systems, and general conclusions about population structure
|Grönberg et al. (2012)||SE||SSR||No||Yes|
|Brurberg et al. (2011)||SE/NO/DK/FI||SSR||No||Yes|
|Widmark et al. (2011)||SE||SSR||No?||Yes|
|Gisi et al. (2011)||Europe||SSR||Yes||Yes|
|Kildea et al. (2010)||IE||SSR||Yes|| |
|Widmark et al. (2007)||SE||SSR||No?||Yes|
|Montarry et al. (2010)||FR||SSR||Yes||Yes|
|Cooke et al. (2009b)||IE/UK (N. Ireland)||RG57||Yes|| |
|Lees et al. (2009)||UK||SSR||Yes||Yes|
|Cooke et al. (2009a)||Europe/UK||SSR||Yes 13_A2||Yes|
|Cooke et al. (2007)||Europe/UK||SSR||Yes 13_A2||Yes|
|Evenhuis et al. (2007)||NL||AFLP||No||Yes|
|Shaw et al. (2007)||UK||RG57/mtDNA||Yes||Yes|
|Lebecka et al. (2007)||PL||SSR||?||Yes|
|Flier et al. (2007)||FR/UK/NO/CH||AFLP||Yes||Yes|
|Deahl et al. (2006)||UK (Jersey)||RG57||Yes|| |
|Cooke et al. (2006)||UK (N. Ireland)||RG57/mtDNA/allozymes||Yes||Yes|
|Day et al. (2004)||UK||RG57/mtDNA||Yes|| |
|Cooke et al. (2003)||UK (Scotland)||AFLP/RG57||?||Yes|
|Griffin et al. (2002)||IE||RG57||Yes||Yes|
|Knapova & Gisi (2002)||CH/FR||RAPD/AFLP/SSR||Yes||Yes|
|Elansky et al. (2001)||RU||RG57/mtDNA||Yes||Yes|
|Purvis et al. (2001)||UK||AFLP/RG57||Yes||Yes|
|Carlisle et al. (2001)||IE||RAPD/RG57/allozymes||Yes|| |
|Zwankhuizen et al. (2000)||NL||RG57||Yes|| |
|Brurberg et al. (1999)||NO/FI||RG57||No||Yes|
|Lebreton & Andrivon (1998)||FR||RG57||Yes||Yes|
|Sujkowski et al. (1994)||PL||RG57/allozymes||Yes||Yes|
|Drenth et al. (1994)||NL||RG57||No||Yes|
|Fry et al. (1991)||NL||Allozymes||?||?|
One obvious conclusion is that if sexual reproduction is not taking place, then there must be an extremely large number of clonal lineages that exist in parallel in some of these regions. There is also a clear trend, in that the Nordic countries seem to have the largest number of genotypes, with little or no dominance by specific clonal lineages. In the study by Brurberg et al. (2011), 169 different genotypes were found among 191 isolates where six or more loci could be scored. In that study, the authors were unable to reject the hypothesis that the loci were at Hardy–Weinberg equilibrium. In the UK and France however, there seem to be dominant lineages, although they can be displaced. Montarry et al. (2010) concluded that there were two clusters within the French population, both with a strongly clonal structure, evidenced by the low genotypic diversity and the presence of many isolates with the same multilocus genotype. The Netherlands seems to be somewhat intermediate, with some clonal lineages, as well as more diverse populations in some sampling areas (such as organic potato production or home gardens) (Zwankhuizen et al., 2000).
Different studies possible with different markers
One benefit of SSR markers is that they are codominant, and are thus able to distinguish between heterozygotes and homozygotes, an important feature when analysing the populations of diploid organisms such as P. infestans. The marker regions can be treated as genes, and allele frequencies within the population studied can be estimated. The probability of observing a particular multilocus genotype, which is in turn dependent on allele frequencies and the assumption of Hardy–Weinburg equilibrium, was calculated in a French study (Montarry et al., 2010). A wide range of genetic indices, such as genetic richness, Simpson's evenness index, various F-statistics, and index of multilocus linkage disequilibrium, were calculated in that study.
Marker systems also permit detailed studies at the field level. In one study, Widmark et al. (2007) were able to examine specific foci in the field, and concluded that some foci, because of the large numbers of different genotypes within them, were probably the result of sexual reproduction. Other foci contained only single genotypes. Another study that used the codominant nature of microsatellite markers was reported by Widmark et al. (2011). In that study, the population of P. infestans in a potato field was characterized using a large number of samples and SSR markers. The following year, soil samples were taken and P. infestans isolates were obtained using the method described by Drenth et al. (1995). These isolates were then characterized using microsatellite markers. None of the soil isolates were identical to those that infected the crop the previous year. Likelihood-based methods for assigning parentage (Marshall et al., 1998; Kalinowski et al., 2007) could then be used to link the isolates from the soil to candidate parental isolates, with corresponding probabilities.
A summary of the evidence
The available evidence, when taken together, points to sexual reproduction of P. infestans in Scandinavia. Taken separately, much of the evidence derived from individual studies given in support of sexual reproduction of P. infestans can also be explained by other mechanisms. For example, the presence of both mating types can be explained by two clonal lineages, each of a different mating type. The general types of evidence and possible alternative explanations are summarized in Table 2. However, some of the alternative explanations require a number of unlikely events to occur. For example, in the study reported by Widmark et al. (2007), it is hard to reconcile the data with clonal survival of the pathogen. If there was no oospore germination in that field, a single tuber would have to have been infected with a large number of clonal lineages, or tubers with different clonal lineages planted in close proximity to each other. Alternatively, each seed tuber could have been infected with a different clonal lineage, but with suitable conditions for disease development only where disease foci developed.
Table 2. Observations consistent with sexual reproduction of Phytophthora infestans in northern Europe and possible alternative explanations of these observations
|Both mating types in 50:50 ratio||Co-existing clonal lineages|
|Oospores readily found in the field||No germination or infection|
|Experiments (Andersson et al., 1998) indicating soilborne inoculum||Infected volunteers survive 2 years|
|Fingerprinting studies indicating different genotypes||A large number of co-existing clonal lineages|
|Large number of genotypes in foci||Infected seed tubers (some with many genotypes, other with only one)|
|Isolates from soil recombinants from previous epidemic||Contamination by clonal lineages in laboratory and greenhouse that just happen to be recombinants of isolates used in epidemic study the previous year|
The appearance of different clonal lineages of different mating types or different mitochondrial haplotypes in many cropping situations is conclusive evidence that new genotypes are being generated, possibly on a local basis. These may be generated locally (assuming that both mating types are present in an area) or may be the result of dispersal from other areas, either naturally or via distribution of infected material. In this respect, it can be useful to compare Scandinavia, which shows extreme genotypic variation, with Great Britain and France, where there seem to be dominant clonal lineages superimposed over a background of genetic variation. In this case, it can be useful to consider how clonal lineages can dominate. One possibility is that oospore infection takes place earlier than infection from tubers in Scandinavia (Yuen, 2012) and that these early infections prevent the later, possibly more aggressive lineages from dominating. Combining the information in Table 2 with what is known about the earlier (pre-1970s populations) of P. infestans, one can hypothesize how information about variability and clonal lineages can be used to determine how frequently sexual reproduction takes place. Thus, where there are dominating clonal lineages but no background variability, there is little or no sexual reproduction. Variability together with dominating clonal lineages could result from limited sexual reproduction. Genotypic variability with no dominating clonal lineages would result from frequent, regular sexual reproduction.
What is needed, however, is way to combine all the information into a single conclusion as to whether sexual reproduction takes place and how often this is the case. Being able to formulate hypotheses as to how dominant clones arise and persist could also be useful. If there were quantitative estimates for the different probabilities, Bayesian methods (Yuen & Hughes, 2002) could be used to calculate some probabilities. A decision about sexual reproduction of P. infestans, though, is relatively personal, so that the important probabilities are not objective numbers, but reflect the perceived strengths of individual pieces of evidence. The relative strengths of belief in different pieces of evidence may vary from person to person, and thus this method of combining information can only yield a personal probability distribution regarding sexual reproduction of P. infestans.
Viewing data in a Bayesian framework, however, can be useful in that it allows for prior beliefs to be incorporated with new knowledge to form what is called the posterior. In this respect, people use a somewhat Bayesian approach for most decisions that they have to make. For this particular example, a person may have doubts as to whether P. infestans reproduces sexually outside of the Toluca Valley (the prior). Then, after being confronted with a piece of evidence relating to sexual reproduction (e.g. the presence of both A1 and A2 mating types), the prior is combined with this new data to form the posterior. The resulting posterior may still be ‘doubtful’ if the prior beliefs are much stronger than the data. On the other hand, the resulting posterior might be ‘sexual reproduction is taking place’ if the prior beliefs about sexual reproduction are not so strong and are outweighed by the data. Thus, the strength of the prior beliefs will affect how new information is interpreted. Bayesian methods also allow for some sorts of data to be stronger than others. For example, evidence for genotypic diversity using 12 microsatellite markers, each with several alleles together with mitochondrial DNA haplotype data, would be stronger than mitochondrial DNA haplotype data alone. A Bayesian view of how new information is combined with priors can also explain how different conclusions are reached about other issues related to P. infestans. For example, information is available that certain resistance genes derived from Solanum bulbocastanum can be overcome by some isolates of P. infestans (Förche et al., 2010). If this information is combined with a strong prior that these resistance genes are unique and durable, the conclusion would be reached that their use may be a viable disease control strategy. However, if the prior is that vertical resistance genes will be overcome by P. infestans, then the conclusion reached could be that their use would not be a good disease control strategy. In either case, a Bayesian-oriented structure allows new and old information to be compared and combined.
Many people have formed their priors about sexual reproduction of P. infestans based on what was learned decades ago. Thus, it is harder to accept data that says otherwise, although in practice, the posterior is shifted in small steps as the evidence accumulates. In Scandinavia, one could reverse the question and ask what evidence there is for clonal reproduction of P. infestans. If the prior is ‘we don't know if there is sexual reproduction’ and this is combined with the recent studies of P. infestans, one would see (in Bayesian terminology) that the available data only moves the posterior towards a conclusion that sexual reproduction takes place. Thus, only a prior of ‘sexual reproduction does not take place’ (however weak or strong) could contribute support to a conclusion that the pathogen reproduces in a clonal manner in this region.
The available evidence from northern Europe strongly supports regular sexual reproduction of P. infestans in this region. The presence of sexual reproduction changes the epidemiology of this important disease, and thus changes the way in which disease control must be approached. Increased sexual reproduction leads to increased variability in the pathogen population, with immediate implications for the use of host-plant resistance and fungicides. The presence of an inoculum source in the soil, accompanied by shifts in when initial infections can take place, further complicates disease control strategies. Alternative explanations of the data supporting sexual reproduction may be plausible for some types of studies, but as a whole it is easier to accept that there is a second centre of sexual reproduction for this important pathogen in northern Europe, with corresponding complications for disease control methods.
The authors would like to acknowledge the discussions and consultations they have had with their late blight colleagues in Scandinavia, the rest of Europe and throughout the world.