Population genetic analysis reveals cryptic sex in the phytopathogenic fungus Alternaria alternata

Reproductive mode can impact population genetic dynamics and evolutionary landscape of plant pathogens as well as on disease epidemiology and management. In this study, we monitored the spatial dynamics and mating type idiomorphs in ~700 Alternaria alternata isolates sampled from the main potato production areas in China to infer the mating system of potato early blight. Consistent with the expectation of asexual species, identical genotypes were recovered from different locations separated by hundreds of kilometers of geographic distance and spanned across many years. However, high genotype diversity, equal MAT1-1 and MAT1-2 frequencies within and among populations, no genetic differentiation and phylogenetic association between two mating types, combined with random association amongst neutral markers in some field populations, suggested that sexual reproduction may also play an important role in the epidemics and evolution of the pathogen in at least half of the populations assayed despite the fact that no teleomorphs have been observed yet naturally or artificially. Our results indicated that A. alternata may adopt an epidemic mode of reproduction by combining many cycles of asexual propagation with fewer cycles of sexual reproduction, facilitating its adaptation to changing environments and making the disease management on potato fields even more difficult.

associations amongst neutral loci indicate that the pathogen population may undergo regular sexual recombination [21][22][23] . For many heterothallic fungi, sexual reproduction is controlled by a single regulatory locus containing two alternate idiomorphs (MAT1-1 and MAT1-2) 17,18,24 and requires the interaction between individuals from distinct mating types. Heterothallic pathogens with regular cycles of sexual reproduction are expected to have the two mating types present in equal frequencies in a population as a result of frequency dependency 22,25 , and low genetic differentiation and no phylogenetic associations between isolates in the two mating type groups attributed to the frequent exchange of genetic material. In this study, we used population genetic approach to investigate the occurrence of cryptic sexual reproduction in the plant pathogenic fungus Alternaria alternata, which is associated with the early blight disease on potatoes.
Potato early blight, characterized by the formation of dark-colored spots that are necrotic in the centre with a pattern of concentric rings on leaves, is among the most destructive emerging pathogen worldwide 26 . It is a polycyclic disease and its epidemic can be rapidly built up when environments are conducive 27 . Considerable economic losses caused by potato early blight have been reported in many countries 28 . In the recent years, the frequency and scale of early blight occurrences have increased, possibly attributed to reduced nitrogen supplies, increased air temperature, withdrawal of some effective fungicides and the changes of farming systems such as the adoption of irrigation and minimum tillage for resource and environment conservations 29 . In China, early blight has been observed in all potato production areas, suggesting environmental conditions in the country are favorable for its epidemic, and the disease can occur in any stages of potato development.
Both Alternaria alternata and A. solani can cause early blight on potatoes in the world 26,[30][31][32][33] . A. alternata produces densely turfy colonies in dark grey to black brown color on PDA plates. Its conidium is usually pale to light brown containing ~15 septa in length with numerous secondary and occasionally tertiary chains branching from apical and median cells 31 . On the other hand, A. solani produces densely dark grey to black colonies with sparse aerial mycelium. Its conidia usually has 9-11 transverse septa and 1-2 longitudinal septa with one long to ovoid beak 26,31 . In China and Europe, the majority of previous studies indicate that A. alternata out-competed A. solani though in importance the relative proportions of these two species may vary in populations 31 . Our survey from China is consistent with this result (unpublished data).
A. alternata is an opportunistic haploid pathogen affecting various important crops 34 . Though carrying a functional MAT1 gene 18 , it is considered to be an asexual fungus (Ascomycota) because teleomorph (the sexual form) has not been observed yet in nature. Conflicting with theoretical expectation for an asexual pathogen, previous population genetic surveys revealed high genetic variation in A. alternata populations 26,35,36 . However, the majority of these surveys focused the descriptions of within-population variation using virulence, vegetative compatibility or genetic markers with less reliability and resolution such as RAPD. Population genetic analysis on the spatial dynamics of genetic variation in A. alternata using the combination of highly polymorphic neutral markers such as SSR and functional genes such as the mating type gene is scarce, particularly for A. alternata from potato. Knowledge from this type of analysis is important for understanding the reproductive mode and adaptive potential of the pathogen, therefore, necessary in designing an effective and sustainable disease management program for controlling this emerging disease worldwide. It is reported that A. infectoria, a species which occupies a basal position in the phylogeny of Alternaria genus, connects to a Lewia teleomorph 37,38 , suggesting that, in fact, most of the Alternaria species may have a sexual ancestor 39 . Furthermore, some evidence of sexual recombination was reported in A. alternate from citrus by sequence approach 40 . Hence, the objectives of this study were: 1) to screen specific primers for the determination of mating types in A. alternata; 2) to determine spatial distribution of the two mating types in A. alternata populations; and 3) to infer the occurrence of sexual reproduction in the field populations of A. alternata.

Results
Spatial distribution of mating type frequency in A. alternate. A total of 725 Alternaria alternata isolates from different parts of China ( Fig. 1) were assayed for mating type frequencies with the two pairs of primers (Table 1). All produced a unique amplification of either 661 or 494 bp, corresponding to MAT1-1 or MAT1-2, respectively (Fig. 2). After clonal correction, 496 isolates were included in the analysis of mating type frequency ( Table 2). Both mating types were detected in all 17 field populations sampled across China (Table 2). MAT1-1 frequency in the field populations ranged from 0.30 in Suizhou to 0.63 in Changle with an average of 0.45 when data from the study were combined whilst MAT1-2 frequency ranged from 0.37 in Changle to 0.70 in Suizhou with an average of 0.55 for combined data ( Table 2). The hypothesis of 1:1 ratio between MAT1-1 and MAT1-2 frequencies was not rejected within all but one field population. Except for one case (Suizhou population), no significant differences in mating type frequency were detected amongst field populations (Table 2) and regional populations (p = 0.154).
Genetic diversity in A. alternata populations. The isolates were assigned to 253 distinct genotypes.
Among these genotypes, 134 (52.96%) were detected only once (Fig. 3A). The most common genotype was detected 24 times and the most widespread genotype was detected in 9 out of the 17 field populations (Fig. 3B). Looking at the temporal scale, 203 (80.2%), 45 (17.8%) and 5 (2.0%) out of the 253 genotypes were detected within a sampling year, in two and three sampling years (Fig. 3C), respectively. Therefore, the frequency of shared genotypes decreased from the first to third sampling year. Clonal fraction in the field populations ranged from 0.00 to 0.37 with a total clonal fraction of 0.27 in the pooled population. Correspondingly, standardized Shannon index in the field populations ranged from 0.84 to 1.00 with a grand Shannon index of 0.78 when isolates from different fields were pooled together (Table 3). Average Nei's gene diversity in the 17 field populations ranged from 0.19 in PAS6 to 0.79 in PAS2 with an grand mean of 0.38 and average allele number in the 17 field populations ranged from 2.2 in PD8 to 8.9 in PAS2 with an grand mean of 3.7 (Table 4). Test for gametic equilibrium. In the 17 field populations, the hypothesis of random association was not rejected in 60.7-100% locus pairs with an average of 85.1%, and 80.3-96.7% allele pairs with an average of 88.6% (Table 3). When isolates from different fields were pooled together, the hypothesis of random association was retained in 78.4% locus pairs and 88.1% allele pairs. The observed variance of heterozygosity did not exceed either L 95 or U 95 values in five out of nine field populations with a sample size > 30 and the corresponding I A was not significantly different from theoretical expectation for random mating also in five of these populations (Table 3).

Population differentiation and phylogenetic relatedness between MAT1-1 and MAT1-2
isolates. Genetic differentiation (G ST ) between the two mating type groups in the 17 field populations ranged from 0.01 to 0.31 (Table 3). Most field populations with a reasonable sample size (> 16) did not have a G ST value larger than 0.05. When all isolates from different fields were pooled together, G ST between MAT1-1 and MAT1-2 isolates was 0.002. Isolates from the two mating types were randomly distributed in phylogenetic clades (Fig. 4), suggesting no genetic association between the isolates from the same mating type group.  Table 1. Sequence, annealing temperature (T) and fragment size of PCR primers designed to amplify mating type idiomorphs of the fungal pathogen A. alternata. Discussion Same genotypes were recovered in the A. alternata populations sampled from different geographic regions and years (Fig.3), consistent with the theoretical expectation for a species presumably reproducing asexually. This result suggests that infected tubers or other materials may be the main mechanisms responsible for long distance dispersal   of conidia of A. alternata on potatoes, as observed on other crops [41][42][43] . However, clonal fraction was considerably low across populations. In addition to asexual reproduction, our results suggested that, like many presumed asexual fungi 15 , cryptic recombination might be occurring and contributing significantly to the population genetic dynamics and epidemics of this emerging potato pathogen. We have several lines of evidence that support the hypothesis of recombination in A. alternata populations from potatoes in China: high genotype diversity, random    association among neutral markers, equal mating type frequency and no genetic differentiation between isolates of two mating type groups.
Populations without recombination will exhibit a high degree of clonality dominated by a few genotypes and low gene diversity. Though some genotypes were detected many times in our study, large genotype diversity was observed in the majority of the 17 field populations: more than 50% of genotypes were detected only once and all isolates had unique genotypes in some populations. This level of genotype diversity is comparable to some fungi with frequent sexual recombination 10,23,44 but much higher than those reproducing primarily in clonally. For example, in an asexual potato pathogen P. infestans, only 49 genotypes were detected in the 531 isolates analyzed with a similar type of molecular markers 45 . High genotype diversity in A. alternata was also found in previous population surveys 26,35,36 .
Populations without recombination are also expected to exhibit a significant degree of non-random association among unlinked alleles. However, in the current study, all gametic equilibrium analyses indicate a low to no non-random associations amongst the SSR markers in most of the 17 field populations ( Table 4). The hypothesis of random associations was retained in ~80% of allele-by-allele and ~90% of locus-by-locus comparisons. Multilocus association analysis also indicated that more than 50% (9/17) of the fungal populations displayed random association in the SSR loci. Analysis of multilocus association is strongly influenced by population sizes and result from populations with a sample size < 30 is less reliable 46 . When we re-analysed the multilocus association only using the populations with ≥ 30 isolates, we also found the hypothesis of random mating in neutral SSR markers was retained in more than 50% of the fungal populations assayed. While many biological and evolutionary processes including genetic drift, asexual reproduction and assorting mating may lead to non-random association, the finding of random association among SSR markers strongly indicate that most of the A. alternata populations sampled from potatoes in China underwent regular recombination.
Genetic differentiation between isolates from MAT1-1 and MAT1-2 was very low when SSR variation in the two groups of the pathogen was compared, suggesting a high degree of genetic exchange between the two sub-populations of the pathogen. G ST between two mating type groups in the majority of the 17 field populations was lower than 0.05. When the isolates from different regions were pooled to form a national population, no genetic differentiation (G ST = 0.00) was detected between two mating type groups. In three pathogen populations (Kunming2, Changle and Qujing), moderate to high genetic differentiations were detected between the two mating type groups, but they were likely attributed to sampling error caused by small population sizes as only 5-8 genotypes were included in these populations for the analysis.
Phylogenetic analyses for the relatedness between the two mating type groups of isolates using both partial (50 isolates, Fig. 4A) and the whole set (Fig. 4B) of SSR data also indicated no genetic convergence existed, suggesting frequent exchange of genetic material through recombination homogenizing MAT1-1 and MAT1-2 genomes.
When combined with other measures of population genetic parameters such as genotype diversity and gametic equilibrium/disequilibrium 47 , frequency distribution of mating types in heterothallic fungi with a single MAT gene controlling two idiomorphs 17 can be used to infer mating history of fungi 44,[48][49][50] . While skewed ratio between the two mating types in a population could be caused by many biological and evolutionary events, an observed coexistence and equal frequency of both mating types would be consistent with regular occurrences of recombination. In our study, the null hypothesis of a 1:1 ratio between the two mating types was retained in all but one populations and this result remained the same regardless whether clone corrected or clone-uncorrected data were used. The observation of significant departure from a 1:1 ratio in one field population could be due to a false rejection of the null hypothesis. As the probability of the type I error for tests conducted at the P = 0.05 level is 0.05, the likely number of falsely rejected null hypotheses in our 24 tests (17 fields + 4 regions + 3 years) is 1.2.
Both meiotic recombination in sexual reproduction and mitotic recombination in parasexual reproduction could explain the observed patterns of population genetic structure in A. alternata. Parasexual reproduction was first described in Aspergillus nidulans 51 and since then has been reported in many pathogenic fungi 40,[52][53][54] . It first involves the fusion of hyphae (anastomosis) from different strains in the same vegetative compatibility group to form heterokaryons (heterokaryosis) and then fusion of the unlike nuclei in the cell of the heterokaryon (karyogamy) to form diploid nuclei. Homologous chromosomes in the diploid nuclei undergo synapsis and mitotic recombination (cross-over), generating recombinant chromosomes if the homologs carry different genetic information. The diploid nucleus begins to lose chromosomes gradually through cycles of mitotic divisions and return to a stable haploid state eventually. Each of these events is believed to be relatively rare 55 . For example, the frequency of chiasma formation in parasexual reproduction is much lower than that in sexual reproduction because chromosomes in the former do not pair in a regular arrangement. Furthermore, due to non-disjunctive division in diploid nuclei, parasexual reproduction is expected to form a series of aneuploid either with additional or missing copies of chromosomes, generating multiple or null amplifications by PCR amplification. In our study, we did not detect such multiple or null amplifications in any SSR amplifications of the isolates. Therefore, while we cannot exclude the possibility of parasexual reproduction, we believe the main contribution to the observed genetic variation in A. alternata is meiotic recombination through sexual reproduction. It is likely the sexual stage of the pathogen occurs off-season on alternate hosts, making it difficult to observe. Sexual reproduction in A. alternata has also been documented preciously from a small area (a single citrus grove) in North America 40 . Our results suggest sexual reproduction may commonly occur in the pathogen by using a large collection involved many population originated from wide geographical locations.
In conclusion, our results indicate that, like many eukaryotic pathogens, A. alternata in potatoes may have evolved to have an epidemic mode of reproduction 56 . This reproducing strategy allows the pathogen to preserve allelic combinations that are well adapted to existing hosts and environments while retain its ability to generate new allelic combinations that may offer an advantage on novel hosts and in changing environments 57 , making it more difficult to manage. In this case, using evolutionary and ecological knowledge of A. alternata and its hosts to design a dynamic disease management program that combine all available approaches such as host resistance, cultivar rotation, disease sanitary and fungicide is important to manage potato early blight effectively and durably 58 . The deployment of single gene resistance varieties and high risk fungicides should be done carefully because the pathogen has high evolutionary potential 59,60 .

Materials and Methods
Fungal collections. Infected potato leaves with early blight were collected at random at 1-2 meter intervals from 17 farmer fields across China (Fig. 1) between 2011 and 2013. One infected leaf was collected from each plant. To isolate fungi, infected leaves were rinsed briefly with distilled water, sterilized with 75% alcohol for 20 seconds and incubated on plates filled with 1% water-agar medium. After 24 hours, one conidium from each infected leaf was transferred to potato dextrose agar (PDA, Potato 200 g/L, Glucose 20 g/L and Agar 20 g/L) medium in a Petri dish and cultured at 25 °C for two weeks. Fungal isolates were purified by repeatedly transferring a single conidiophore to a flesh PDA plate for three times and mycelia from the third purification were then harvested for DNA extraction. Genomic DNA was extracted using plant gDNA kit (Promega Biotech. co. LTD., Beijing) according to the manufacturer's instructions. All isolates were checked morphologically with a light microscope and molecularly by PCR amplification of ITS regions with primers 5′-TCCGTAGGTGAACCTGCGC-3′ and 5′ -TCCTCCGCTTATTGATATGC-3′ to confirm they were A. alternata. Mating type determination. Mating types in A. alternata isolates were determined by PCR amplification of genomic DNAs with two pairs of mating type-specific primers ( Table 1). The primers were designed from two complete mating type sequences of A. alternata (GU735423.1 and GU735413.1) downloaded from Genebank and synthesized by Ruiboxingke Biotech. Co. LTD. (Beijing). Primers A amplifies a part of the α -domain, generating a unique 661 bp fragment from the isolates carrying MAT1-1 idiomorph and Primers B amplifies a part of the HMG-box, generating a unique 494 bp fragment from the isolates carrying MAT1-2 idiomorph. PCR amplifications of mating types were performed using a mixture of two pairs of primers (multiplex PCR) and carried out in an 2720 thermal cycler (Applied Biosystems, Foster city, California) in 20 μ L PCR reaction volume that consisted of 1 μ L (10 μ M) of each primer, 2 μ L (2 mM) of dNTPs, 2 μ L of 10 × reaction buffer, 1 unit of Taq polymerase (Transgen Biotech co. LTD., Beijing), 2 μ L of genomic DNA and 9 μ L of sterilized distilled water. The program used for mating type amplification was: initially held at 95 °C for 5 min, followed by 33 cycles of 94 °C for 60 s, 53 °C for 30 s and 72 °C for 30 s; ended with a final extension at 72 °C for 10 min. PCR products were separated by electrophoresis on 1.2% agarose gel at 100 V for 60 min and visualized with a G8140 Golden View I (Solarbio Science and Technology, Beijing).

Data analyses.
Alleles were assigned based on the sizes of PCR amplifications that were generated by each pair of SSR primers using GeneMaker software version 1.31 with a binning procedure. PCR amplification with an identical size generated by the same pair of primers was considered as an allele. Multilocus haplotype for each isolate was formed by joining the alleles at each SSR locus in the same order. Isolates with the same multilocus haplotype and mating type were considered as an individual member of the same clone, the asexual progeny of a genotype. In the estimation of population genetic parameters except for genotype diversity, only one member of clone was included in each population. If the clone was detected in several populations, one representative of the clone was retained for each population. All isolates (clone-corrected) were used to estimate genotype diversity.
Fungal isolates were hierarchically organized into "field" and "regional" populations. Isolates sampled from the same field were considered as the member of a field population and from different fields within the same cropping region were grouped into regional populations. There are four potato cropping regions in China: Northern Single-cropping Region (NSR), Central Double-cropping Region (CDR), Southwestern Multiple-cropping Region (SMR) and Southern Winter-cropping Region (SWR). The null hypothesis of 1:1 ratio between the two mating types within a population was evaluated by a simple χ 2 test and heterogeneity in mating type frequency among populations was evaluated by a contingency χ 2 test 63 Genotype diversity was measured with standardized Shannon index 64 . The clonal fraction, defined as the proportion of fungal isolates in a population originating from asexual reproduction 23 , was calculated as 1-[(number of different genotypes)/(total number of isolates)]. Gene diversity in each SSR locus estimated in Nei's diversity 65 and genetic differentiation between the two mating type groups in each of the 17 field populations was evaluated by G ST 65 using Popgen 1.32. Gametic equilibrium in the field populations was evaluated by multilocus association 46 , locus-by-locus 66 and allele-by-allele comparisons 66 . Standard deviation of I A for each population was generated by 1000 resamples of original data and its difference from theoretical expectation or random mating was evaluated by a t-test. Clone-corrected SSR data were used for these analyses. A phylogenetic tree among 25 genotypes each of the two mating types randomly selected from the total collection was reconstructed using UPGMA and displayed with NTSYS (Version 2.1, 2000, Applied Biostatistics). A further neighbor joining tree was reconstructed using SSR data of all 253 genotypes. Nei's genetic distance 67 in the neighbor joining tree was calculated using GENALEX 6.5 and the phylogenetic tree was displayed using Mega 5.