Genetic diversity and population structure of Alternaria species from tomato and potato in North Carolina and Wisconsin

Early blight (EB) caused by Alternaria linariae or Alternaria solani and leaf blight (LB) caused by A. alternata are economically important diseases of tomato and potato. Little is known about the genetic diversity and population structure of these pathogens in the United States. A total of 214 isolates of A. alternata (n = 61), A. linariae (n = 96), and A. solani (n = 57) were collected from tomato and potato in North Carolina and Wisconsin and grouped into populations based on geographic locations and tomato varieties. We exploited 220 single nucleotide polymorphisms derived from DNA sequences of 10 microsatellite loci to analyse the population genetic structure between species and between populations within species and infer the mode of reproduction. High genetic variation and genotypic diversity were observed in all the populations analysed. The null hypothesis of the clonality test based on the index of association \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left( {\overline{r}_{d} } \right)$$\end{document}r¯d was rejected, and equal frequencies of mating types under random mating were detected in some studied populations of Alternaria spp., suggesting that recombination can play an important role in the evolution of these pathogens. Most genetic differences were found between species, and the results showed three distinct genetic clusters corresponding to the three Alternaria spp. We found no evidence for clustering of geographic location populations or tomato variety populations. Analyses of molecular variance revealed high (> 85%) genetic variation within individuals in a population, confirming a lack of population subdivision within species. Alternaria linariae populations harboured more multilocus genotypes (MLGs) than A. alternata and A. solani populations and shared the same MLG between populations within a species, which was suggestive of gene flow and population expansion. Although both A. linariae and A. solani can cause EB on tomatoes and potatoes, these two species are genetically differentiated. Our results provide new insights into the evolution and structure of Alternaria spp. and can lead to new directions in optimizing management strategies to mitigate the impact of these pathogens on tomato and potato production in North Carolina and Wisconsin.

www.nature.com/scientificreports/ Several fungicides are currently available to control Alternaria spp. Due to its short life cycle and rapid growth rates via asexual reproduction 12 , the reduced sensitivity of strains of Alternaria spp. to quinone outside inhibitor (QoI) and succinate dehydrogenase inhibitor (SDHI) fungicides has been reported 10,13 . The planting of resistant varieties is the most economical, sustainable, and environmentally friendly approach to controlling these plant pathogens. However, the effectiveness of resistant varieties and fungicides inextricably depends on knowledge of the evolutionary potential and population genetic structure of Alternaria spp. 14,15 . Furthermore, investigating the population biology of plant pathogens can help us understand the origin, sources of inoculum, migration, and phylogeography of the pathogen 15 . It also provides useful insights into the patterns of genetic diversity, which can assist in selecting target pathogen isolates for screening germplasm for resistance and developing durable resistant tomato and potato varieties 14,16 . Information on the genetic diversity, genetic structure, and reproductive biology of Alternaria spp. populations are necessary to understand the impact of evolutionary forces, such as mutation, selection, genetic drift, recombination, and migration, on the evolutionary trajectory of pathogen populations over time 16,17 . Understanding the factors that are responsible for creating pathogen variability and structure is of practical significance in epidemiological studies and in optimizing pathogen management strategies. For example, plant pathogenic fungi with a mixed reproduction system of clonality and recombination have a higher potential for mutation and gene flow than asexually reproducing fungi 16 . The recombination process creates genetic diversity and enhances the propensity of pathogens to evolve rapidly in response to changes in management strategies 14,16 .
Currently, the genetic variation and reproductive strategy of Alternaria spp. in tomato and potato are unknown in NC and WI. Alternaria alternata, A. solani, and A. linariae are excellent candidates for investigating the model of genetic divergence and speciation because of their intimate association with tomato and potato. Alternaria spp. are known to reproduce primarily by asexual reproduction, and a high degree of clonality is dominated by a few genotypes, low gene diversity, and a significant degree of nonrandom association among unlinked alleles 15,16 . Although no sexual stage has been found in Alternaria spp., signatures of recombination in clonal lineages of A. alternata (the citrus brown spot pathogen) and high levels of genetic and genotypic diversity have been reported in a few Alternaria spp. [18][19][20] . Sexual reproduction in most plant pathogenic fungi, including Alternaria spp., are controlled by two mating type (MAT) idiomorphs: MAT1-1 and MAT1-2 21,22 . Using a polymerase chain reaction (PCR)-based approach, the MAT1-1 and MAT1-2 idiomorphs of approximately 1.9 and 2.2 kb, respectively, were cloned and sequenced in A. alternata 21 . Subsequently, MAT1-1 and MAT1-2 idiomorphs have been identified in several Alternaria spp. 15,[21][22][23] . This has prompted the analysis of population genetics using molecular markers to indirectly infer the hypothesis of clonality or recombination 15,16,[24][25][26] .
To assess genetic diversity, molecular markers such as amplified fragment length polymorphisms (AFLPs), isozymes, random amplified polymorphic DNA (RAPD), and random amplified microsatellites (RAMSs) have been used in A. alternata and A. solani from tomato and potato 8,18,19,27,28 . However, these markers have limited discriminatory power to differentiate closely related isolates within populations of the pathogen. Microsatellite loci or simple sequence repeats (SSRs) and single nucleotide polymorphisms (SNPs) are preferred markers 17 because they are co-dominant, reproducible, and are located in known sequences in the genome compared to RAPDs, AFLPs, and other markers. SSRs are tandem repeat motifs 29 that occur mostly in the intergenic and non-coding regions of the genome 30,31 . Microsatellite loci have been widely used in population genetic studies to analyse the genetic structure and evolutionary biology of several pathogenic fungi [32][33][34][35] . The use of SNPs and high-throughput sequencing to investigate the genetic diversity of plant-pathogen populations has gained attention in recent years, partly owing to the reduced cost of DNA sequencing, the proliferation of tools for assembly and annotation, and the possibility of simultaneously genotyping thousands of molecular markers in multiple individuals 36,37 . Importantly, several limitations of gel electrophoresis and capillary-based methods 37,38 can be overcome by the sequencing of microsatellite loci (SSR-seq). Subsequently, SSR-seq offers a more precise and statistically powerful tool to resolve the genetic diversity and population structure of pathogen populations 38,39 .
In this study, we examined SNPs from 10 microsatellite loci to detect genetic variation in three Alternaria spp. sampled from tomato and potato in NC and WI. We tested the hypothesis of population subdivision to determine whether local populations are genetically structured by geographic location or tomato variety and the evolutionary factors that shape the population structures of different populations. We further tested the hypothesis of recombination within Alternaria spp. to discern evolutionary relationships among isolates in each population and to examine the role of random mating in creating diversity in these species. Stratified by geographic location or by tomato variety, we conducted analyses to answer the following biological questions. Do SNPs from microsatellite loci detect the genetic structure within a species and are the populations of Alternaria spp. from NC and WI genetically differentiated? If so, which of the species is more diverse? What are the main factors (geographic location or tomato variety) structuring populations within a species? Is there genotype sharing between species and between populations within a species, which could be an indicator of gene flow? What is the implication of the observed high genetic and genotypic diversity in disease epidemics? We also explored evidence of random mating by examining MAT idiomorphs distributed between species and between populations within a species. Does clonal or random mating occur or is there a mixed reproductive system? What are the implications of investigating this for pathogen management strategies?

Results
Morphological characteristics. Based on the characteristics of single spores developed on acidified potato dextrose agar (A-PDA) (Difco Laboratories, Detroit, MI) plates that were amended with two antibiotics, ampicillin (0.06 g/L of medium) and rifampicin (0.024 g/L of medium), to minimize bacterial growth, all Alternaria isolates were classified into two groups. The isolates of group I were characterized by the formation of obclavate, dark brown, and short septate conidia with a short beak and branched chains. These morphological Molecular determination of species. Neither PCR assay amplified genomic DNA isolates of A. alternata. The primer pairs OAsF7 and OasR6 amplified a 164 bp PCR product from all DNA samples specific to A. solani, whereas the primer pairs OatF4O and atR2 produced a 483 bp amplicon from all A. linariae isolates. In all, 61, 96, and 57 isolates were confirmed as A. alternata, A. linariae, and A. solani, respectively, and were included in genetic analyses ( Fig. 1 and Supplementary Table S1).
Microsatellite locus sequence variability. Missing data were minimal, and thus, there was no variability in the dataset. Comparatively, A. linariae populations had more variations in each microsatellite locus than A. alternata and A. solani populations. In all, 220 SNPs were identified from 10 microsatellite loci across Alternaria spp. The number of SNPs per locus varied from nine for SSR534 to 38 for SSR271 (Supplementary Table S2). The genotype rarefaction curve showed a linear increase as the number of microsatellite loci increased, signifying that these loci were polymorphic and useful to assess the genetic diversity between species and between populations within species (Supplementary Fig. S1).
Nucleotide diversity and neutrality test. Statistics of nucleotide diversity and neutrality tests varied with microsatellite loci and populations within a species (Supplementary Table S3). In general, the values for S, h, H d, Pi, and θ w were greater in A. solani than in A. alternata or A. linariae. Most nucleotide diversity estimates in three microsatellite loci, SSR271, SSR400, and SSR511, were high across populations.  Table S3). The four microsatellite loci SSR201, SSR327, SSR391, and SSR534 yielded low (0.0 to 0.01) Pi and θ w in A. linariae, whereas microsatellite loci SSR186, SSR201, and SSR457 showed low (0.0 to 0.01) Pi and θ w in A. alternata. Tajima's D values were negative for most of the microsatellite loci across populations (Supplementary  Table S3  The colour reflects the precise location from which isolates were collected. The number (sample size) of isolates collected from each location is indicated. Population differentiation. Comparisons of three Alternaria species showed high levels of genetic differentiation (e.g., WAU-As, Jost's D = 0.984) ( Table 2). Stratification of populations by geographic location revealed low pairwise differences in genetic differentiation between populations within a species. For example, Jost's D values between the BHT-Aa and STO-Aa populations and between the MAC-Al and HAY-Al populations were closely related (Jost's D = 0.034 to 0.035). These populations were not significantly differentiated (P < 0.05) ( Table 2). Similarly, stratification of populations based on tomato varieties revealed limited genetic differentiation between populations within the same species, with Jost's D values ranging from 0.000 to 0.083 (Table 2). AMOVA analyses revealed that 1 to 5% of genetic variation was observed between populations. In contrast, the highest (> 85%) genetic variation came from individuals within populations ( The membership probability suggested that similar levels of admixture were shared between populations within a species; however, no admixture was shared between any two species (Fig. 5). The MSN based on Nei's genetic distance revealed no clustering of geographic populations within species (Fig. 6). The demographic history of A. alternata and A. linariae in the sampled locations was investigated using approximate Bayesian computation. These scenarios supported our DAPC results, which showed the closeness and lack of clustering of three A. alternata populations (PSW-Aa, STO-Aa, and BHT-Aa). Scenario 3 was proposed to be the PSW-Aa population as the most likely progenitor of the STO-Aa and BHT-A populations ( Fig. 7A-C). Similarly, A linariae Linkage disequilibrium. We calculated the standardized index of association (r d ) in clone-corrected data to infer the LD. We found nonsignificant LD (P > 0.05) between species and most populations within a species (Table 4 and Supplementary Figs. S11, S12, and S13). Intriguingly, LD values in two populations, STO-Aa and Heirloom-Aa, significantly (P < 0.05) deviated from the hypothesis of recombination.  www.nature.com/scientificreports/ www.nature.com/scientificreports/ Mating type determination. As expected, the mating-type primers amplified either an 800-bp MAT1-1 fragment or a 280-bp MAT1-2-specific product from any DNA sample ( Supplementary Fig. S14a, b). The frequencies of the two mating types were in equilibrium for the A. alternata and A. solani populations (P ≤ 0.05); www.nature.com/scientificreports/ however, skewing towards MAT1-1 was observed in A. linariae ( Table 4). The geographic population STO-Aa within A. alternata and the tomato variety population Heirloom-Aa did not deviate significantly (P ≤ 0.05) from a 1:1 ratio. The null hypothesis of random mating cannot be rejected for these populations. However, the mating type ratios deviated from 1:1 for the remaining geographic location populations and tomato variety populations within a species. The mating type ratios in seven populations (PSW-Aa, HAY-Al, MAC-Al, MAD-Al, Hybrid-Aa, Picus-Al, and Tasti Lee-Al) were skewed towards MAT-1-1, whereas in two populations (SWA-Al and Plum Regal-Al), it was skewed towards MAT1-2 (Table 4).

Discussion
Our study aimed to analyse the genetic diversity and population structure of three Alternaria spp. in NC and WI. To the best of our knowledge, this is the first attempt to use SNPs from 10 SSR-seq in three Alternaria spp. sampled from populations of tomato and potato to test the hypothesis of population subdivisions and determine whether they are in linkage equilibrium. Our analyses identified high genotypic diversity and low levels of genetic diversity in all Alternaria spp. and across populations within a species. To test for clonality, both r d and two mating types provide evidence for random mating in Alternaria species. Although difficult to observe directly, the evidence for recombination was unexpected in our haploid organisms 11 . These findings are consistent with a pathogen that has a mixed reproductive mode of both clonality and sexuality, which has implications for the efficacy of fungicides and durability of host plant resistance due to increased pathogen variability. High gene and genotypic diversity have been reported for A. brassicicola 15,40 , A. alternata 23,41-43 and A. solani 18,28,44 . Furthermore, high genetic and genotypic diversity has been ascribed to the random association between alleles at different loci 24 , genetic drift, gene flow, and parasexual mating leading to nonrandom association of alleles 16 . Our analysis revealed 162 unique MLGs across species, and the same MLGs were shared between some populations within a species. The presence of the same MLGs between populations within a species suggests the potential for gene flow 16 . We hypothesize that gene flow may be attributed to three possible reasons that have also been demonstrated in other pathosystems 45,46 . First, conidia of Alternaria spp. can spread by wind and with seeds 47,48 causing EB and LB epidemics during the growing seasons. Second, these fungi can overwinter in the plant debris and lesions on shoots, which may serve as inoculum reservoirs in the absence of hosts 47 and subsequently spread to geographically close farms where susceptible varieties are grown. Third, human-mediated movement through contaminated vehicles and farm machinery, infected tomato transplants, and potato tubers distributed or shipped to producers could contribute to the short-and long-distance dispersal of these fungi. However, the potential long-distance dispersal of Alternaria spp. inoculum within the United States deserves further investigation.
Multilocus linkage disequilibrium (LD) tests suggested potential recombination in two A. alternata geographic location populations (BHT-Aa and PSW-Aa) and four A. linariae geographic populations (HAY-AI, MAC-AI, MAD-AI, and SWA-Al). In contrast, the A. solani population (WAU-As) and two A. alternata populations (STO-Aa and Heirloom-Aa) exhibited clonality. A few geographic location populations (STO-Aa and WAU-As) and www.nature.com/scientificreports/ one tomato variety population (Heirloom-Aa) contained the two mating types, and their distributions were not significantly different from a 1:1 ratio, suggesting random mating. The occurrence of both mating types along with high levels of genetic diversity, the large number of singleton MLGs, and random mating provide evidence for most populations undergoing sexual recombination. Our findings corroborate previous studies that also reported a similar conclusion in asexual and sexual fungi 15,40,[49][50][51][52][53][54] . We also found an unequal frequency of mating types or a biased distribution in some populations within a species. These results could be due to crop management practices such as the application of fungicide or the use of resistant cultivars that can favour one or the other mating type, as demonstrated in other studies 23 . Alternatively, nonrandom mating 15,24,55 and the selective advantage of the isolate virulence in the host might contribute to causing skewed mating type ratios 56,57 . The findings of a mixed reproductive mode and high genetic diversity of Alternaria spp. based on SNPs are parallel to those of other pathosystems 15,50 , where the impact of mixed reproduction is well documented. The assessment of population differentiation using Jost's D index revealed that genetic differences were found between species but not between populations within a species. Our findings based on comparative genealogical analyses using DIYABC indicated that Alternaria spp. are part of a panmictic population due to a lack of population clustering based on Bayesian analysis or possibly due to low sequence variations within a species 5,30,58 . We found a lack of genetic differentiation between the populations within A. alternata or A. linariae from tomatoes in NC. This result indicated that these two fungi can coexist in the same agricultural systems and microclimates www.nature.com/scientificreports/ suitable for tomato production. The development of management strategies and breeding for resistance must be species-specific. Our analysis further showed that the populations between A. linariae and A. solani did not follow a pattern of differentiation by geographic distance. Population genetics-based inference of both genetic diversity and population differentiation of Alternaria spp. may be governed by their effective population sizes, immigration from other populations and effects exerted by evolutionary processes such as random genetic drift and mutation rates 59,60 . The total genome sizes of three Alternaria spp. reported previously were A. alternata = 33.2 mega-bases (Mb); A. linariae = 32.9 Mb, and A. solani = 34 Mb 30 . Furthermore, these genomes are comparable to recently sequenced and annotated genomes of several other species of Alternaria 30 . A larger genome size usually generates higher genotypic diversity because more alleles could emerge through mutation and random genetic drift 61 . One of the limitations of our study is the reliance on small sample sizes of A. alternata populations collected from tomatoes in NC. Similarly, the A. solani population (WAU-As) was collected from a potato from one location in WI. To overcome these limitations, geographically more diverse and larger samples are needed to collect and analyse population genetics parameters. We expect that the distribution of sample sizes may not be representative of the populations being sampled and is likely affected by the accuracy of genotypic diversity and population differentiation analyses. Our results must be interpreted cautiously when comparing genetic diversity and population structure with global populations of A. alternata and A. solani.
Our analyses revealed three genetic clusters corresponding to the individual species. The SWA-Al population from Swain County was slightly different from the other three populations HAY-Al, MAC-Al, and MAD-Al. Likewise, A. solani populations causing EB on tomato and potato were not differentiated from each other 2,3 . The hypothesis of a lack of host specialization needs to be confirmed by cross-inoculation of A. linariae and A. solani isolates on tomato and potato varieties. We also found a few signals of admixed members attributed to the movement of the common MLGs through the dispersal of airborne conidia and infected planting material 16,62 . Although the founder sample history is unknown, our neutrality test showed a signature of population expansion. When populations were stratified by geographic location or tomato variety, > 85% of the genetic variation www.nature.com/scientificreports/ was distributed in individuals within a population. Our findings of high genetic diversity in three Alternaria spp. were similar to previous studies 63,64 , who also reported little or no evidence of population differentiation based on geographic origin or host of origin. The centre of origin of any pathogen is likely to possess a population with higher genetic variability than recently established populations 54 . In this study, the A. alternata and A. linariae populations from tomatoes in NC had higher genotypic diversity than the A. solani population from potatoes in WI. It appeared that the populations from tomatoes could have been established earlier than the A. solani population from potatoes. The A. solani population (WAU-As) was sampled from EB-resistant potato genotypes such as Atlantic, Russet Burbank, a wild relative S. berthaultii, F 1 hybrids, advanced breeding lines, and parental genotypes 8 . These genetically diverse potato genotypes might have some effects on the pattern of differentiation within the A. solani population from WI. All A. alternata populations analysed were sampled from hybrids and heirlooms with unknown genes for resistance. In contrast, A. linariae populations were collected from commercially grown tomato varieties such as Picus (U. S. Patent # 7,807,886 B2), Testi Lee 65 , and Plum Regal 66 , and these varieties have some levels of resistance to A. linariae. These resistant varieties could exert selection pressure on the pathogen populations, thus promoting high genotypic diversity in A. linariae populations in NC. It is worth noting that the SWA-Al population was more variable than other populations and was collected from 'Plum Regal' . This variety was developed from a cross between the EB-resistant tomato breeding lines NC 25P and NC 30P 66 .
In conclusion, we used SNPs from 10 microsatellite loci to gain insights into Alternaria species diversity and population structure with respect to species, geographic location, and affected hosts in NC and WI. Our analysis revealed three distinct genetic clusters that corresponded to each species. The effect of geographic location and tomato variety on populations within a species was low. We used several population genetics tools to test for the occurrence of random mating to further infer the mode of reproduction of the studied populations. The presence of high genotypic diversity, high and diverse MLGs, linkage equilibrium based on the r d and both mating-type ratios supported the hypothesis of random mating. These findings suggest that Alternaria spp. can maintain the sexual pathway in the studied populations 23 . Plant pathogen populations with a mixed reproductive system (i.e., both clonal and recombination) are considered the highest risk 16 . The process of random mating may increase the frequency of the evolution of new MLGs through the random association of alleles, whereas asexual reproduction maintains the propagation of well-adapted MLGs. Thus, gene flow may occur through the movement of these MLGs between geographic locations or through infected tomato transplants. Importantly, careful intervention must be exercised when applying different fungicides and deploying single-gene resistance varieties because pathogen populations with high evolutionary potential are more likely to overcome host genetic resistance 16 . Tomato and potato breeding strategies that combine quantitative resistance should be a valuable approach for developing durable resistance to these fungal pathogens in NC and WI.  For A. alternata, a haphazard sampling strategy was used in geographic regions and spanned ~ 370 km from south to north and ~ 650 km from east to west in NC. Alternaria alternata isolates were collected and isolated (as described above) and grouped according to counties within a similar geographic region and tomato type ( Fig. 1 and Supplementary Table S1). Among them, the BHT-Aa population (n = 10) was collected from grapetype tomatoes 66 from Buncombe, Henderson, and Transylvania (BHT) counties in the western mountain region, NC. The PSW-Aa population (n = 11) was collected from the commercial hybrid tomato cv. Florida 47 (Fla 47) and Roma from Pender, Sampson, and Wake (PSW) counties in the inner coastal plain and Piedmont region in NC. The Stokes (STO)-Aa population (n = 40) was sampled from Stokes county in the northwestern Piedmont region in NC. The STO-Aa population was collected mainly from heirloom tomatoes such as Aunt Gertie's Gold, Cherokee Purple, German Pink, Mortgage Filter, Mr. Stripey, Pink Brandy Wine, Snowstorm, and Verna Orange. Alternaria alternata populations were grouped according to tomato type and represented by the Grape-Aa population (n = 10) from BHT in southwestern NC; the Hybrid-Aa population (n = 10) was from PSW in central and eastern NC, and the Heirloom-Aa population (n = 38) was from STO in northwestern NC.
We used stratified random sampling 67 to collect A. linariae isolates from four major tomato-growing counties in NC: Haywood, Macon, Madison, and Swain. These counties spanned 165 km from east to west. Each site was divided into strata or rows approximately 250 m long. In each row, three-leaf samples were collected ~ 40 m apart. At least 10 rows were sampled, and up to 60 samples were collected from each county (Supplementary  Table S1). In all, 96 A. linariae isolates were collected and represented by HAY-Al (n = 21), MAC-Al (n = 33), MAD-Al (n = 21), and SWA-Al (n = 21) from Haywood, Macon, Madison, and Swain counties, respectively ( Fig. 1 and Supplementary Table S1). Alternaria linariae isolates collected from western NC were mainly from hybrid tomatoes, and tomato variety populations were represented by the Picus-Al population (n = 21), Tasti Lee-Al population (n = 54), and Plum Regal-Al population (n = 21) (Supplementary Table S1).
We used random sampling to collect A. solani isolates (n = 57) between 2008 and 2009 at the Hancock Agricultural Research Station, Waushara County, WI ( Fig. 1 and Supplementary Table S1). Potato leaves showing EB lesions were collected from breeding lines and families. Three tissue samples (4 mm 2 ) were cut from the margin of EB lesions on the same leaflet and placed onto clarified V-8 (CV8) medium (100 mL of clarified V8 juice, 1.5 g CaCO 3 , 900 mL distilled H 2 O, and 12.7 g agar) amended with streptomycin (25 μg/mL) and kanamycin (50 μg/ mL). Some of the isolates of A. solani from potatoes represented by the WAU-As population were also characterized previously by RAPD markers 8  Morphological characteristics. Conidia were taken from sporulating colonies cultured on A-PDA for 7 days using an inoculation loop and then streaked onto 1% water agar plates. The plates were incubated for 24 h at 28 °C to allow for conidial germination. The germinating conidia were transferred under a dissecting microscope to a new A-PDA plate. All fungal isolates sampled from tomato and potato were cultured on A-PDA for 7 days and examined for morphological characteristics. For each isolate, two replicates of A-PDA plates were used. These cultures were incubated at 25 °C under cool white, fluorescent light with an 8 h photoperiod and 16 h darkness. For each of the tested isolates, colony colour, pigment production in A-PDA medium, conidial shape and size, the presence or absence of beaks on conidia, branching, spore colour, and the presence of horizontal and vertical septa were examined under a compound microscope as described previously 1,4 . To maintain the fungal isolates, an 8 mm diameter mycelial plug was cut from each A-PDA plate after 7 to 10 days of incubation using a cork borer. Mycelial plugs were dried under laminar flow overnight, transferred into 2 mL screw-cap cryogenic vials, and stored at − 80 °C. In total, 214 isolates of three Alternaria spp., A. alternata (n = 61), A. linariae (n = 96), and A. solani (n = 57) ( Fig. 1 and Supplementary Table S1), were analysed to address the biological hypotheses laid out in the introduction. www.nature.com/scientificreports/ DNA extraction. To extract genomic DNA, isolates were revived on A-PDA plates and incubated at 25 °C for 7 d. Fungal tissues were scraped with a sterile razor blade and placed into 2 mL screw-capped tubes. The tubes were freeze-dried (Virtis Genesis 25 ES Freeze dryer, Virtis Company, Gardiner, NY) for 72 h with the following cycle: − 25 °C for 40 min; − 30 °C for 60 min; − 20 °C for 600 min, and − 10 °C for 600 min with a pressure of 400 mTorr. The lyophilized freeze-dried fungal tissues were ground to a fine powder using a microtube homogenizer (Model D1030-E, Beadbug, Benchmark Scientific Inc., Edison, NJ) at 400 rpm for 3 min. The glass beads were removed, and genomic DNA was extracted using the DNeasy Plant Mini kit (Qiagen, Valencia, CA) following the manufacturer's recommended protocol. The DNA concentration was quantified using a fluorometer (Qubit 2.0, Invitrogen by Life Technology, Carlsbad, CA). The final concentration of DNA at 10 ng/µL was prepared for each sample for all PCR assays and SSR-seq.  Table S2). In our preliminary studies, the genomic DNA of five representative isolates of each species was arbitrarily selected to examine the utility of the SNPs from SSR-seq as sequence markers. A 25 µL reaction volume contained 8.5 µL of sterile distilled water, 2 µL of 10 ng/ µL genomic DNA, 1 µL each of 10 µM reverse and forward primers, and 12.5 µL of GoTaq Green PCR mix (Promega Inc., Madison, WI). The PCR conditions for all microsatellite loci included initial denaturation at 94 °C for 2 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 58 °C to 60 °C for 1 min and extension at 72 °C for 2 min, and a final extension step at 72 °C for 10 min. The quality and expected size of the PCR product of each sample were confirmed by electrophoresis on a 1.5% agarose gel (w/v www.nature.com/scientificreports/ These included (1) between species collected from tomato and potato in NC and WI, (2) between geographic populations within species, and (3) between tomato variety populations within A. alternata or A. linariae collected in NC (Supplementary Table S1).

Molecular determination of species. Although
Statistics of sequence diversity. Data for population genetic analyses were based on the haploids and the presence and frequency of SNPs from 10 microsatellite loci. We computed the number of segregating sites (S), the number of haplotypes (h), haplotypic diversity (H d ), nucleotide diversity (Pi), and Watterson's Theta (θ w ) per site from the total number of mutations (Eta) 69 between species and populations within species in DnaSP ver. 6 70 . We used SNP data sets and calculated Tajima's D (TD) 71 and probability (P) values to test the hypothesis of deviation under a standard neutral model of molecular evolution. The significance of these values was tested using 10,000 random permutations. A negative TD value indicates population expansion 71 . A genotype accumulation curve was produced with the R package poppr 72,73 to determine whether SNPs were sufficiently informative to discriminate between individual multilocus genotypes (MLGs). The average gene diversity (H T ) 74 for each locus was calculated using GeneAIEx ver. 6.5 68 .
Genotypic diversity. We assume that populations of Alternaria spp. are functionally asexual and exhibit a clonal structure. Therefore, we used clone-corrected data to calculate genotypic indices between species and between geographic location populations and tomato variety populations within species. Genotypic diversity is a measure of genotypic richness and evenness 75 . Estimates of the number of MLGs increases for populations with larger sample sizes and thus comparing genotypic richness among the populations with unequal sample sizes could be biased and inappropriate. To facilitate comparison between species and populations within species, the number of expected multilocus genotypes (eMLGs) based on rarefaction was calculated from numbers of MLGs expected at a sample size equivalent to the population with a minimum of 10 samples using 999 bootstrapping in the R package poppr ver. 2.8.1 72,75 . Genotypic evenness (E5) is a measure of the relative abundance of each genotype in the populations 76 . To assess the distribution of genotypes within each species and between populations within species, E5 was calculated as the ratio of the number of abundant genotypes to the number of rarer genotypes 75 . E5 values ranged from 0 to 1, with 0 indicating that a population is dominated by a single MLG and 1 indicating that all genotypes occur at the same frequency 76 78 , and Stoddart and Taylor's genotypic diversity index (G) 79 . To compare the genotypic diversity within a population of different sample sizes, we analysed the indices E5, G, and λ following random sampling of the data using rarefaction in poppr ver. 2.8.1 72,75 with 1000 jackknife replicates as described previously 52 . Unbiased expected heterozygosity was defined as H exp = (1 − ∑pi 2 ) × (n/n − 1), where p is the SNP frequency at a locus, and n is the number of observed SNPs at each locus 74,75 and was calculated between species and between populations within species using the poppr package ver. 2.8.1 72,75 in R 72,80 . For further analyses, sequence data were clone-corrected on all population hierarchical levels using the clone correct function in poppr ver. 2.8.1 72,75 , and isolates that shared the same SNPs were regarded as clones.
Population differentiation. Genetic differentiation between pairs of populations was assessed in clonecorrected data using Jost's D differentiation index 81 implemented in R using the mmod 1.3.3 package 82 . The statistical significance of Jost's D was calculated by a randomization procedure in which all isolates in the respective populations were reassigned randomly to the populations, and Jost's D was recalculated from the randomized dataset. Jost's D was estimated under the null hypothesis of no differentiation between populations within a species. Analysis of molecular variance (AMOVA) 83 was calculated using Arlequin ver. 3.5 83 and hierfstat ver. 0.5-7 84 implemented in the R package to test the following null and alternative hypotheses. In this context, we computed three F-statistics to address the biological questions: (1) F ST partitions the genetic variation among populations within a species. If the F ST value = 0, then there was no genetic differentiation, whereas if the F ST value > 0, then the populations were subdivided. (2) F IS partitions the genetic variation among individuals within populations of a species, and (3) F IT indicates the genetic variation within individual isolates in a population. F IS and F IT may be positive or negative, with a negative value being indicative of excess heterozygosity. The significance of F-statistics at each level was evaluated based on 1000 random permutations, and significant genetic differentiation was declared when P < 0.05 in the R package ade4 ver.1.7.5 85 . Alternaria solani (n = 57) has only one population, and thus, this species was not included in AMOVA for comparison between and within populations within a species.
Population structure. We assessed the genetic structure of the combined populations of Alternaria spp.
using clone-corrected data to test the hypothesis of population subdivision between species and populations within a species. Population structure was analysed by conducting a discriminant analysis of principal components (DAPC) and using the function xvaldapc in the R package adegenet ver. 2.1.2 86 . Because DAPC assumes no prior knowledge of group membership, our main goal in using this analysis was to display clustering of genetically related populations while minimizing variation within clusters. Another advantage of using this approach is that DAPC does not make any assumptions regarding the data structure or underlying population genetics model. To estimate the optimal number of genetic clusters of MLGs in the data sets, the K-means clustering algorithm procedure was implemented in the function 'find. clusters 87 . This model relies on data transformation using principal component analysis (PCA) to delineate clusters and then uses discriminate analysis based on the number of principal components retained by the user to maximize the separation between populations 87 . We identified the alleles that best discriminate a set of defined clusters of Alternaria spp. A minimum spanning network (MSN) was constructed with clone-corrected data to calculate the genetic relatedness among MLGs in www.nature.com/scientificreports/ populations within a species using the imsn function in poppr 73,80 and Nei's 74 genetic distance. The network was visualized using the igraph package ver. 1.2.1 88 . The DAPC clusters were used to inform and develop historical scenarios describing the evolutionary relationships among populations of Alternaria spp. These scenarios were investigated using the approximate Bayesian computational (ABC) approach and conducted in the clone-corrected dataset using DIYABC v 2.10 89 . The observed MLG data were compared with large numbers of simulated datasets (one million per scenario) based on evolutionary scenarios (models). Possible scenarios were set with conditions as follows: t4 > t3 > t2 > t1. The posterior distribution of each of the three scenarios was estimated by conducting linear regressions on the 1000 simulated datasets after a logit transformation. The posterior probability of each scenario was computed by direct and logistic regression methods to identify the scenario with the highest posterior probability.
Linkage disequilibrium (LD) tests. We calculated the standardized index of association (ř d 2 ) 90 on clonecorrected data using the poppr package in R to infer the mode of reproduction (e.g., clonal or recombination) between species and between populations within a species. The significance of ř d was tested by comparing the observed ř d values against 10,000 randomized data sets to the null hypothesis of no LD; that is, alleles observed at different loci are unlinked and in random association in these populations 90 . Sexual populations are expected to have linkage equilibrium owing to no linkage among loci, whereas clonal or asexual populations are expected to have significant LD owing to linkage among loci 90 . Mating type determination. We carried out two PCR assays to amplify the MAT1-1 or MAT1-2 allele to infer reproductive modes. MAT1-1 was amplified using the forward primer CHO13 22 and reverse primer Amat1r: (5′-GAC CAG GCT TTC GYC ATC) 91 . MAT1-2 was amplified with primer pairs MCHMG1 and MCHMG2 22 . The 25 µL PCR mixture used for amplification was prepared as described above. The PCR conditions and cycles for mating type analysis were conducted as described previously 22,91 . The amplified PCR products were separated in a 1.5% agarose gel in 1 × TAE buffer stained with 0.001% (v/v) GelRed (Biotium, Inc., Union City, CA). A 100-bp DNA marker (Invitrogen Life Technologies, Carlsbad, CA) was loaded in each gel to determine the PCR product size. Clone-corrected data were used to estimate the frequency of the mating types between species and between populations within species according to the formula χ 2 = ∑[(o − e)2/e], where o is the observed value of the mating type, and e is the expected value. The null hypothesis was that the observed mating types to the expected show equal frequencies (1:1 ratio) under random mating. The chi-square values for goodness-of-fit were calculated in SAS V9.4 (SAS Institute, Cary, NC).