Polymorphic Microsatellite Development, Genetic Diversity, Population Differentiation and Sexual State of Phytophthora capsici on Commercial Peppers in Three Provinces of Southwest China

ABSTRACT

Pepper blight, caused by the oomycete pathogen Phytophthora capsici (P. capsici), is one of the most destructive soilborne diseases worldwide. Between 2019 and 2020, 235 single spore isolates of P. capsici were collected from 36 commercial pepper planting areas in Sichuan, Chongqing, and Guizhou provinces in China. A novel full set of 323 high-quality polymorphic microsatellites was obtained by resequencing 10 isolates. In total, 163 isolates with two alleles per microsatellite locus were used for population analysis and resulted in 156 genotypes on 10 microsatellite loci. The genetic diversity, population differentiation, principal component, genetic structure, and genetic relationships analyses showed an extensive variety of the P. capsici in Sichuan and Guizhou with clonal lineages, two shared genotypes, and no geographic differentiation. The population from Chongqing was differentiated from that of Sichuan and Guizhou and had the highest genetic diversity. There was no significant distinction between the populations of the two sampling years, but there was a small differentiation between the populations from bell peppers and hot peppers. The isolates from Southwest China were largely distant from the two reference isolates from the USA. The analysis of molecular variance showed that the major variance of the populations was within populations. The linkage equilibrium test, mating type composition, and oospore detection indicated that only P. capsici from the Jiulongpo district of Chongqing had appeared in sexual recombination. Overall, this study revealed that the high and complex genetic diversity population of P. capsici in Sichuan, Chongqing, and Guizhou with uneven geographic variation and limited sexual reproductive behavior in Chongqing, potentially driven by differences in the geographical environment, reproductive patterns, different cultivars, and artificial long-distance transfers.

IMPORTANCE Phytophthora capsici, a notorious soilborne and rapidly evolving pathogen with a wide range of hosts, is a huge threat to pepper production worldwide. However, the detailed genetic structure and dynamics of P. capsici in most Chinese provinces are still unclear, even though China is the world's largest producer and consumer of peppers. Here, a novel full set of high-quality polymorphic microsatellites, obtained by genome resequencing data of 10 isolates from Southwest China, was provided for future population analyses. In this study, we further investigated and established the genetic structure, sexual recombination, geographic subdivisions, interannual stability, differentiation in different types of host peppers, and member relationships of P. capsici from three provinces in Southwest China. These results reveal the genetic structure and dynamics of P. capsici in three provinces of Southwest China and help us to execute more effective management strategies in the future.

INTRODUCTION

Pepper blight is a serious and global occurring disease caused by Phytophthora capsici (P. capsici), which is a soilborne plant pathogen with a wide host range (1, 2). This pathogen can produce foliar blighting, damping off, wilting, and root, stem, and fruit rot in many vegetable crops (3), including peppers, tomatoes, eggplants, lima beans, and all cucurbits (pumpkins, squash, melons, cucumbers, etc.) (4). Phytophthora capsici has two mating types (called A1 and A2), by which the pathogen can mate to produce thick-walled oospores. Oospores can survive in the soil for several years (5, 6). Epidemics of pepper blight are often driven by the asexual life cycles of P. capsici under warm and wet conditions (7), especially in the rainy season because P. capsici can produce large numbers of deciduous sporangia under such conditions and release a huge number of zoospores in water to secondary infect hosts (7). At present, P. capsici is widely distributed around the world, causing huge economic losses to multiple important vegetables, especially peppers (7, 8).

In the field, disease management strategies for the control of pepper blight primarily rely on the application of fungicides and cultivation measures (3), including raised beds, plastic mulch, and drip irrigation. However, the efficacy of such disease management is limited for that P. capsici has a broad host range, long-lived dormant sexual spores, extensive genotypic diversity, and an explosive asexual disease cycle (9). Resistant varieties and fungicides are quickly being overcome in the field by P. capsici (10) because of its frequent sexual reproduction, rapid variation and extensive genotypic diversity (7). Thus, a better understanding of the population genetic structures and dynamics of P. capsici is essential for us to develop more effective management strategies (11). Except for Gansu province (8), the detailed population structures of P. capsici in most provinces of China are still unknown, despite being widespread (12), even though China is the world's largest producer and consumer of bell and chili peppers.

Molecular genetic markers are a key tool in genetic research regarding structures and dynamics. Among various molecular genetic markers, simple sequence repeats (SSRs), also known as microsatellites, are currently popular and powerful molecular markers in genetic population studies, because of their good reproducibility, codominance, low DNA template restriction, ease of scoring, low technical difficulty and assumed neutrality (13–15). Simple sequence repeats have reliably been proven to be a more efficient marker system than restriction fragment length polymorphism (RFLP) and random amplified polymorphic DNA (RAPD) markers (16, 17). Different sets of SSRs, which have been developed by expressed sequence tags (18, 19), initial draft genomes (18), or transcriptomes (20), have been used for genetic diversity analyses of P. capsici in previous studies. Recently, the genome of P. capsici was significantly updated by long fragment sequencing, and the genome size even was changed from 64-Mbp to 95.2 or 100.5-Mbp (21, 22). This reassembled genome of P. capsici provides the possibility of obtaining more accurate and complete SSR locus information. Accurate and complete information regarding SSR loci is essential for future research on the genetic structures and quantitative trait locus mapping of P. capsici. In recent years, next-generation sequencing (23) has become increasingly affordable and provides the ability to apace excavate pleomorphic SSR markers by calculating the genome resequencing on the computer (24, 25), which could avoid time consumption on a large number of PCR experiments (15, 26).

Sichuan, Chongqing, and Guizhou are the main hot pepper production and consumption provinces in southwest China. There are many commercial pepper planting areas with different varieties. Pepper blight has occurred frequently in these areas under rainy weather conditions in recent years. We even found that some hot peppers with high resistance, such as pod peppers were infected with P. capsici in some farms. Here, we provided a full set of pleomorphic SSR loci with a calculation pipeline by genome resequencing data of 10 isolates from Southwest China, and insights into the genetic diversity, population differentiation, and genetic structure of P. capsici in commercial peppers from these three provinces. We also assessed the genetic variation of isolates from two different types of pepper, population stability between two consecutive years, the possibility of sexual recombination, and the genetic relationships of isolates from these three provinces with those of other regions. Taken together, this study provided a comprehensive understanding of the population diversities and dynamics of P. capsici in Southwest China, which are important for developing effective disease management practices.

RESULTS

Isolation and mating type determination of P. capsici.

For two consecutive years (2019 to 2020), a total of 235 single spore isolates of P. capsici were isolated from 36 commercial pepper sites in 21 counties from Sichuan, Chongqing, and Guizhou (Fig. 1, Table S1 in Supplemental File 2). All the isolates were confirmed by PCR and agarose electrophoresis with specific detection markers as previously reported (27) and were examined for mating type (Table S1 in Supplemental File 2). The mating type compositions of the 21 sampled counties are also shown in Fig. 1. Overall, there were 36 (15.3%) A1 and 199 (84.7%) A2 mating types (Table 1). The A1 mating type accounted for the majority (32/36) in Chongqing. Three A1 mating types were found in Guizhou and one A1 mating type was found in Sichuan. The χ² test for mating type (isolates ≥ 10) at all the sampling sites showed that only the numbers of A1 and A2 in Xipeng town, Jiulongpo county, Chongqing met the ratio of 1:1 (χ² = 0.04, df = 1, P < 0.01).

FIG 1.

Geographic sources of the pepper blight pathogen P. capsici in three southwest China provinces from 2019 to 2020. The inset shows the locations of sampling counties. In the pie charts, the total numbers of isolates at each sampling county are indicated and the relative proportions of A1 and A2 are indicated as green and blue, respectively. LZ, Lizhou; PZ, Pengzhou; PA, Pengan; XD, Xindu; JY, Jianyan; YaJ, Yanjiang; LX, Luxian; TN, Tongnan; HC, Hechuan; TL, Tongliang; DZ, Dazu; JLP, Jiulongpo; SZ, Shizhu; BN, Banan; BS, Bishan; YC, Yongchuan; SY, Suiyang; WL, Weiling; DF, Welling; HX, Huaxi; XX, Xixiu.

TABLE 1.

Summary of isolates genotyped from commercial pepper in three southwest China provinces

Province/Municipality	County/county-level city	No. of isolates in 2019	No. of isolates in 2020	A1 mating type	A2 mating type	Total no. of isolates
Chongqing		54	27	19	62	81
	Banan	3	6	5	4	9
	Bishan	3	4	0	7	7
	Dazu	8	3	4	7	11
	Hechuan	9	4	2	11	13
	Jiulongpo	10	0	5	5	10
	Shizhu	6	0	0	6	6
	Tongliang	5	3	1	7	8
	Tongnan	5	1	1	5	6
	Yongchuan	5	6	1	10	11
GuiZhou		34	3	3	34	37
	Dafang	9	3	0	12	12
	Weiling	10	0	0	10	10
	Huaxi	8	0	2	6	8
	Suiyang	3	0	0	3	3
	Xixiu	4	0	1	3	4
Sichuan		39	2	1	40	41
	Jianyang	16	0	1	15	16
	Pengzhou	12	0	0	12	12
	Xindu	6	2	0	8	8
	Lizhou	1	0	0	1	1
	Luxian	1	0	0	1	1
	Yanjiang	2	0	0	2	2
	Pengan	1	0	0	1	1

Identification of polymorphic SSR loci by genome resequencing.

To quickly obtain SSR loci with polymorphism in our population, genomic resequencing was performed on 10 isolates, which were randomly selected from different sampling locations in Southwest China (Table S2 in Supplemental File 3). Our genomic sequencing generated 25.3 Giga bases (28) clean paired-end reads (Table S2 in Supplemental File 3). A total of 10 de novo assemblies were generated with an average length of 99.33 million bases (Mb), a GC content of 50.90%, an N₅₀ value of 2,302 bp, and a maximum length of 767,975 bp (Table S3 in Supplemental File 4). Then, a total of 84,894,875 bp were aligned against the updated reference genome (21) with an average 90.25% overall alignment rate (Table S4 in Supplemental File 5). We then identified 13,884 high-quality indels in the 10 isolates, including 6,212 deletions and 7,516 insertions, with an average of 2.24 indels per 10 kb. MIcroSAtellite identification tool (MISA) was used to predict SSRs in the whole updated reference genome. A total of 4241 SSRs were identified through gene structural annotation, most of which (51.4%) were intergenic. To obtain polymorphic SSRs in our population, sequence data of the 10 isolates were calculated using CandiSSR. As a result, 799 polymorphic SSRs were obtained. Then, these candidates of polymorphic SSRs were further polished using the following filter criteria. (i) There were no indels in the 150 bp flanking sequence, and (ii) the read depth of the same SSR allele in each isolate was greater than 5. Finally, only 323 polymorphic SSRs with high confidence were obtained (Table S5 in Supplemental File 6). All the previously calculated pipelines are shown in Fig. 2A. Among these SSRs, dinucleotide repeats were the most abundant (176, 54.48%), followed by trinucleotide repeats (122, 37.77%), tetranucleotide repeats (19, 5.88%), pentanucleotide repeats (4, 1.23%), and hexanucleotide repeats (2, 0.61%) (Fig. 2B and Table S6 in Supplemental File 7). A total of 69 repeat motifs were discovered and no single motif class dominated overwhelmingly (Table S6 in Supplemental File 7). We also found that most polymorphic SSRs were in intergenic regions (165, 51.1%), followed by SSRs in 2 kb downstream of the genes (108, 33.4%), in exons (35, 10.8%), in introns (14, 4.3%), and finally, one SSR assigned to 2 kb upstream of the genes (Fig. S1A in Supplemental File 1), which was similar to the whole SSRs in the P. capsici genome (Fig. S1B in Supplemental File 1).

FIG 2.

Pipeline of polymorphic SSRs development and high-quality polymorphic SSR repeats count. (A) The pipeline of polymorphic SSRs development. It is divided into three parts: preparation of input data, variant calling (steps 1 to 3), and development of SSR (steps 4 to 6), indicated by yellow, red, and blue fillings, respectively. (B) Histogram of polymorphic SSR counts in our resequencing isolates.

Genetic diversity in different populations.

Six frequently used (Pcap2, Pcap5, Pcap7, Pcap8, SSR14, and SSR17) (18, 20) and four novel polymorphic SSR markers (PcSSR_394, PcSSR_155, PcSSR_127, and PcSSR_554), which were randomly selected from Table S5 in Supplemental File 6, were used to estimate the population diversity and structure through PCR with fluorescently labeled primer and capillary electrophoresis on the 3730XL sequencer in our study. All these 10 SSR markers were in different scaffolds and their corresponding primers are annotated in Table 2. Dynamic extreme aneuploidy with three or more alleles in some SSR loci was also found as previously reported (20, 29). Considering that this dynamic extreme aneuploidy in some SSR loci severely hinders genomic studies and population analysis, the population analysis in our study was also calculated with diploid isolates similar to previous research on Phytophthora infestains (30) or P. capsici (20). In total, 159 isolates (Table 1) with no more than two alleles at each SSR loci and four isolates from the outgroup, were retained for further analyses (Table S6 in Supplemental File 7). The genotype accumulation curve indicated that these 10 SSR markers had sufficient power to distinguish the genotypes of our isolates (Fig. 3). A total of 150 unique multilocus genotypes (MLGs) were discriminated using the 10 SSR loci, of which 145 MLGs were present in 1 isolate and the remaining 5 MLGs were distributed in 14 isolates, where each MLG was distributed in 2 to 4 isolates (Fig. S2 in Supplemental File 1).

TABLE 2.

SSR primers for population analysis in our study

SSR name	SSR repeats	Location	F primer (5′–3′)	R primer (5′–3′)	Allelic size
PcCap2 (18)	(AG)n	scaffold182	GTGGCAACGGCAACCACATATAG	GAATTCGATTTGGCCACGTGATAACG	233-305bp
PcCap5 (18)	(AAG)n	scaffold169	GCTTGCATCAATTTATCGCAG	ATTGTGAACGGTCATCACTG	272-300bp
PcCap7 (18)	(GAA)n	scaffold145	CCCATTTGAAGATGATGCACACT	ATATACGCGTGCTTGTCAGTCT	339-383bp
PcCap8 (18)	(TTC)n	scaffold43	TAGGGTTCAGGACCAGCATGT	ATGGGTGGTGCTATGGATGG	287-305bp
SSR14 (19)	(AAG)n	scaffold17	CAGAAACACACGTCTCCGGA	GTTCGAACTGCTCCTGCTCT	199-208bp
SSR17 (19)	(AAG)n	scaffold77	TATCGGACGTTCTCGCCATG	TGAGCGGTTTCTGCTCGAAT	78-108bp
PcSSR_394	(GAA)n	scaffold338	CCTGACGCCCAACAGTTCTTCCG	AGGCAGTGCTGAGTAGGTCAA	249-273bp
PcSSR_155	(AT)n	scaffold18	CCCTGGTACACATCCACCAA	GGGTGATGTGTGCCCGTAAA	116-152bp
PcSSR_127	(AGA)n	scaffold51	CGCCGACATCCTCGTACTTA	GCCACTCGTTGTAGCTCTCC	161-170bp
PcSSR_554	(GTAT)n	scaffold162	AGGCTCGACCCGTCAATTTT	GTCTTCCTGTGACGCTAAGCT	183-215bp

FIG 3.

Genotype accumulation curve for the P. capsici population from three provinces (Chongqing, Sichuan, and Guizhou), in the southwest of China.

The populations of our isolates were subdivided into three different provinces (Sichuan, Chongqing, and Guizhou), eight different counties (Dazu, Hechuan, Jiulongpo, Yongchuan, Dafang, Weiling, Jianyang, and Pengzhou) in which the number of MLGs was greater than or equal to 10, different sampling years (Year_2019 and Year_2020) in which they were sampled in different years in the same counties, and different sampling types of pepper (hot pepper and bell pepper) to carry out further analysis (Table 3). Genetic diversities were estimated by genotypic richness, genotypic diversity, genotypic evenness, and gene (allelic) diversity, as shown in Table 3 for the different populations. Considering the different sample sizes among the different populations, eMLG, which was estimated to be under the maximum shared sample size based on rarefaction, was more suitable for representing genotypic richness than MLG. The values of expected MLG (eMLG) showed that isolates from Chongqing (eMLG = 36.59) had the highest genotypic richness among the different provinces, and there was similar genotypic richness among populations from different sampling counties, years, or types of pepper (Table 3). The Shannon-Wiener index (H), a measure of species diversity in a population, indicated that the Chongqing subpopulation was more diverse than the Guizhou and Sichuan subpopulations. The Simpson index (λ) was used to estimate the probability of the number of individuals sampled from a population two consecutive times being of the same species. Both the Shannon-Wiener index (H) and Simpson’s index (λ) were used to measure the genotypic diversity in the different populations. Considering that the Simpson index can be affected by sample size, λ was corrected for sample size and multiplied by N/(N−1), in which N was the sample size for each subpopulation. Our computations showed that isolates from Chongqing (H = 4.36, corrected λ = 0.999) had higher genotypic diversity than those from Guizhou (H = 3.536, corrected λ = 0.997) or Sichuan (H = 3.578, corrected λ = 0.993). Genotypic evenness, represented by E.5, described the distribution of genotypes in a population. Based on the E.5 measurements, the evenness of the populations from the different provinces was significantly different, namely, Chongqing (0.986) > Guizhou (0.972) > Sichuan (0.882). The sample evenness of Jianyang County (0.885) in Sichuan province was the lowest among all the counties. Unlike the geographically divided groups, genotypic diversity and evenness were similar for the two different sampling years and the two different types of peppers. We also found that all the populations had high levels of genotypic diversity and genotypic evenness in our analysis for the values of both corrected λ and E.5 were all nearly 1. The mean number of observed alleles (AO), observed heterozygosity (H_O), expected heterozygosity (H_E), and mean allelic richness for all the SSR loci were used to describe the gene diversity (Table 3). The unbiased expected heterozygosity (H_E) value of Chongqing (0.6161) was higher than that of Guizhou (0.4916) or Sichuan (0.4964), that of Jiulongpo (0.649) was the highest among all the counties, that of the year 2020 (0.6147) had higher than that of the year 2019 (0.5714), and that of hot peppers (0.611) had higher values than that of bell peppers (0.5565). Furthermore, the allelic richness of each population obtained by sparse computation and taking the average of distinct alleles per locus, showed that Chongqing had the highest value, the year 2020 had higher values than the year 2019, and hot peppers had higher values than those of bell peppers. These results indicated that the genetic diversity of the Chongqing population was higher than that of the Sichuan and Guizhou populations. In particular, the samples from Jiulongpo county in Chongqing had the highest genetic diversity among all the detected counties (Table 3).

TABLE 3.

Genetic diversity analysis in different subpopulations^a

Division criterion	Subpopulations name	N	MLG	eMLG (SE)	H	Corrected λ	E.5	AO	HO	HE	Mean allelic richness (SE)
Different provinces	Chongqing	81	79	36.589 (0.567)	4.36	0.999	0.986	6.9	0.685	0.616	5.144 (0.125)
	Guizhou	37	35	35 (0)	3.536	0.997	0.972	4	0.708	0.492	3.456 (0.07)
	Sichuan	41	38	34.390 (0.571)	3.578	0.993	0.882	4.8	0.705	0.496	3.913 (0.093)
Different counties (MLG ≥10)	Dazu	11	11	10 (0)	2.398	1	1	3.7	0.727	0.582	2.795 (0.153)
	Hechuan	13	13	10 (0)	2.565	1	1	4	0.685	0.561	2.752 (0.153)
	Jiulongpo	10	10	10 (0)	2.303	1	1	4.1	0.560	0.649	3.163 (0.192)
	Yongchuan	11	11	10 (0)	2.398	1	1	3.9	0.656	0.623	3.064 (0.180)
	Dafang	12	11	9.318 (0.466)	2.369	0.985	0.958	2.7	0.750	0.478	2.314 (0.094)
	Weiling	10	10	10 (0)	2.303	1	1	2.9	0.680	0.465	2.289 (0.1)
	Jianyang	16	14	9.089 (0)	2.567	0.975	0.885	3.9	0.706	0.507	2.672 (0.146)
	Pengzhou	12	12	10 (0)	2.485	1	1	2.8	0.717	0.461	2.191 (0.087)
Different yrs	yr_2019	53	50	30.772 (0.860)	3.882	0.998	0.948	5.8	0.7	0.571	4.168 (0.129)
	yr_2020	32	32	32 (0)	3.466	1	1	6	0.713	0.615	4.655 (0.152)
Different peppers	Hot pepper	36	35	33.110 (0.312)	3.545	0.998	0.984	5.6	0.661	0.611	4.139 (0.16)
	Bell pepper	34	33	33 (0)	3.486	0.998	0.983	4.8	0.738	0.557	3.442 (0.117)

^aN, number of individuals observed. MLG, the number of multilocus genotypes (MLG) observed. eMLG, the number of expected MLG at the smallest sample size ≥ 10 based on rarefaction. SE, standard error. H, Shannon-Wiener Index of MLG diversity. Corrected λ, corrected Simpson index. E.5, Evenness. A_O, mean number of observed alleles. H_O, Nei's (1973) observed heterozygosity were computed using Levene (1949). H_E, Nei's (1973) expected heterozygosity. F_IS, fixation index. Our isolates could be divided into three different provinces, eight different counties (MLG ≥ 10), two different sampling years (2019 and 2020), and two different sampling hosts (hot pepper and bell pepper) by different criteria. Only the samples collected in the same county were counted in the year_2019 and year_2020 subpopulations to reduce the influence of geographical location.

Sexual and asexual reproduction occurs in different locations.

To determine whether sexual reproduction existed in our sampling areas, we tested the SSR loci for linkage disequilibrium. Considering that sampling during epidemics can lead to the overrepresentation of clones, clone-corrected data were used for the linkage disequilibrium test. Tests of the index of association (I_A) and the standard index of association (rbarD) were used to estimate the linkage equilibrium. All three populations showed significant departure from the linkage equilibrium when the isolates were divided by provinces (Table 4). Considering that the natural transmission of P. capsici was mainly limited by water within a finite propagation distance (31), the linkage equilibrium test on the populations (MLG ≥10) in different counties was done (Table 4), which might be more reasonable. The populations from Jiulongpo, Dafang, Weiling, and Pengzhou counties had no significant deviation from the null hypothesis of no association between loci, implying that these subpopulations may undergo random mating. However, only the isolates from Jiulongpo conformed to the 1:1 ratio of the A1 and A2 mating types, whereas the isolates from Dafang, Weiling, and Pengzhou were all the A2 mating types. Moreover, oospores were only detected in one pepper blight sample at Xipeng town, Jiulongpo county (Fig. S3 and S6 in Supplemental File 1). Thus, we deemed that sexual reproduction only appeared in P. capsici from Xipeng town, Jiulongpo county, Chongqing. On the other hand, high levels of linkage disequilibrium and negative fixation indices support the finding that the populations from Dazu, Hechuan, and Jianyang counties were strictly clonal.

TABLE 4.

Test of linkage disequilibrium in P. capsici subpopulations from different sampling positions^a

Division criterion	Subpopulations	Ia	P value (Ia)	rbarD	P value (rbarD)	F IS	A1:A2
Different provinces	Chongqing	1.2825	0.0001	0.1413	0.0001	−0.112	Deviate from 1:1
	Guizhou	0.4112	0.0003	0.0544	0.0003	−0.44	Deviate from 1:1
	Sichuan	1.1971	0.0001	0.1388	0.0001	−0.42	Deviate from 1:1
Different counties (MLGs ≥10)	Dazu	2.6252	0.0001	0.3051	0.0001	−0.249	Deviate from 1:1
	Hechuan	0.7870	0.0012	0.0907	0.0009	−0.221	Deviate from 1:1
	Jiulongpo	0.3856	0.0530	0.0444	0.0514	0.137	Meet with 1:1
	Yongchuan	1.5656	0.0001	0.1803	0.0001	−0.051	Deviate from 1:1
	Dafang	0.0016	0.4579	0.0003	0.4567	−0.568	Deviate from 1:1
	Weiling	0.0902	0.3197	0.0164	0.2696	−0.463	Deviate from 1:1
	Jianyang	1.8574	0.0001	0.2166	0.0001	−0.392	Deviate from 1:1
	Pengzhou	−0.1256	0.7120	−0.0221	0.7929	−0.556	Deviate from 1:1

^aIa, the index of association. rbarD, the standardized index of association. P values were estimated using a one-sided permutation test based on 9999 permutations. Linkage disequilibrium tests were done by clone-corrected data. The fixation index (48) was calculated as F_IS = 1 − H_O/H_E. The chi-square calculation was used to test whether the ratio of A1/A2 mating types was significantly different from 1:1 or meet with 1:1 in the same location.

Population differentiation.

The Weir and Cockerham (1984) coefficient (θ) (32) was used to estimate the genetic differentiation of populations (sample size ≥30) between two different types of pepper, two different sampling years, and three distinct sampling provinces (Table S7 in Supplemental File 8). The population of P. capsici in Chongqing was different from that in Sichuan and Guizhou (θ > 0.05, P < 0.01). The population differentiation degree of Guizhou and Sichuan was small (θ < 0.05, P < 0.01). There was no population differentiation between the years 2019 and 2020 (θ < 0.05, P > 0.05). Interestingly, there was also small population differentiation between the hot peppers and bell peppers (θ < 0.05, P < 0.05). Analysis of molecular variance (AMOVA) is another method by which to detect population differentiation utilizing molecular markers (33) without making assumptions about the Hardy-Weinberg equilibrium. The AMOVA of microsatellite genotype data revealed that the major variance in the three different provinces was within populations (87%, estimated variance = 4.362), whereas the remaining variance was among populations (13%, estimated variance = 0.65, F_ST value = 0.13, P = 0.001). The major variance of different counties was also within populations (79%, estimated variance = 3.969), whereas the rest of the variance was among populations (21%, estimated variance = 0.65, F_ST value = 0.207, P = 0.01); the major variance of populations from different sampling years (99%, estimated variance = 4.821) was also within populations, and only a small amount of variance was among populations (1%, estimated variance = 0.034, F_ST value = 0.007, P = 0.196); the major variance of populations from different host peppers was within populations (97%, estimated variance = 4.891), whereas there was small variance among populations (3%, estimated variance = 0.152, F_ST value = 0.030, P = 0.005).

The discriminant analysis of principal components (DAPC) analysis (Fig. 4) based on predefined populations showed that the P. capsici isolates from Chongqing were differentiated from those from Sichuan and Guizhou and that the P. capsici population from Sichuan partially overlapped with that from Guizhou (Fig. 4A). The subpopulation from Jiulongpo county was differentiated from the subpopulations of the other three Chongqing counties, whereas there was a large overlap among the populations of Dafang, Weiling, Jianyang, and Penzhou counties, which present two different provinces (Fig. 4B). There was partial discrimination between the samples of the two different types of peppers, while there was a large overlap between the samples of the two different years (Fig. 4C and D). All these DAPC results are also in line with the Weir and Cockerham (1984) coefficient and AMOVA results.

FIG 4.

DAPC analysis for different subpopulations (MLG ≥ 10). (A and B) Genetic differentiation among different positions (provinces and counties, respectively) by DAPC. Individuals (dots) and groups (colors and ellipses) are positioned on the plane using their values for two variables. Clusters are defined by ellipses and indicate the variance within the clusters whereas dots indicate the positions of individual parasite genotypes within the cluster. (C and D) Individual density plots between two different hosts (C) and two different sampling years (D) are shown for the first discriminant function. Eigenvalues represent the amount of genetic variation captured by the discriminant factors plotted as the x-axis and y-axis.

Genetic structure relationships.

Bayesian clustering was used to examine the population structure of our microsatellite genotype data using STRUCTURE and STRUCTURE HARVESTER software. The K value was determined by the log probability of data (LnP[D]) based on the rate of change in LnP(D) between successive K, using the web-based STRUCTURE HARVESTER version 0.6.92 software. STRUCTURE HARVESTER predicted that K = 3 was the most likely number of clusters for all our isolates from the three provinces (Fig. 5A). The structure of the Chongqing isolates was different from that of the Guizhou and Sichuan isolates (Fig. 5D), which is consistent with the previous results of population differentiation. On the other hand, we also found that the structure of isolates of the two sampling years showed similar constructions (Fig. 5F) while the samples from the two different types of peppers did not show the same constructions (Fig. 5E). All these results are also in line with the Weir and Cockerham (1984) coefficient, AMOVA and DAPC analyses.

FIG 5.

The structure assignments. (A and C) Relationship between delta K (ΔK) and K as revealed by STRUCTURE HARVESTER. Estimation of the number of subgroups for the K values ranging from 1 to 9, by ΔK values. (D) STRUCTURE assignments for P. capsici isolate from three southwest China provinces under the values of K = 3. (E) STRUCTURE assignments for P. capsici isolate from bell pepper and hot pepper in three southwest China provinces under the values of K = 3. (F) STRUCTURE assignments for P. capsici isolates from 2019 and 2020 in the same counties of three southwest China provinces under the values of K = 2.

A minimum spanning network (MSN) based on Bruvo’s distance was further used to visualize the genetic distance between different MLGs in the three different provinces. The result showed that the genetic relationship of the MLGs in Sichuan was closely related to the MLGs in Guizhou (Fig. 6). Two identical MLGs were even found to co-occur in Sichuan and Guizhou (Fig. 6). Furthermore, two isolates (LT263 and LT1534) from America, one isolate (LD) from Qinghai province, China, and one isolate (SD33) from Shandong province, China were used as outgroups for phylogenetic analysis. The Phylogenetic tree (Fig. S4 in Supplemental File 1) showed that the isolates from Chongqing, Sichuan, and Guizhou, were distantly related to the two reference isolates from America and closely related to the two isolates from Qinghai and Shandong provinces, China.

FIG 6.

Minimum spanning network of P. capsici isolates from three southwest China provinces. Each node represents one genotype. Colors indicate population membership. Edge width and shading represent relatedness, and edge length is arbitrary.

DISCUSSION

Phytophthora capsici has a wide range of hosts and is one of the most notorious pathogens that affect pepper worldwide (7). A better understanding of P. capsici population genetics and dynamics will facilitate the development of effective disease management. In this study, a novel full set of polymorphic microsatellites was provided by a pipeline with an updated genome and genome resequencing data of 10 isolates from the Southwest of China. Additionally, the detailed genetic structures and dynamics of P. capsici on commercial pepper in the Sichuan, Chongqing, and Guizhou provinces of China are presented according to genetic diversity, random mating possibility, genetic structure, population differentiation, and member relationships.

SSR markers, which are mined from genome sequences, are widely used in the population genetic analysis of microorganisms. Recently, two updated P. capsici genomes (21, 22) showed a huge difference from the previous draft genome. In this study, we practiced an in silico approach to excavating high-quality polymorphic SSR markers and provided a full set of high-quality polymorphic SSR markers with genomic locus information (Table S5 in Supplemental File 6). A total of 323 high-quality polymorphic SSR markers were successfully curtailed from 4241 candidate SSR loci and four randomly polymorphic SSR markers (PcSSR_394, PcSSR_155, PcSSR_127, and PcSSR_554) were used in subsequent population genetic analysis. These practices suggested that this in silico approach could greatly reduce the experimental labor for polymorphic SSR locus mining and could be applied to identify polymorphic SSR markers in other organisms. These high-quality polymorphic SSRs in the updated genome of P. capsici will be helpful for future P. capsici population research and quantitative trait locus mapping.

The results of the genotypic richness (measured by eMLG), genotypic diversity (measured by the Shannon-Wiener index [H] and corrected Simpson’s index [corrected λ]), genotypic evenness (measured by E.5), and gene diversity (measured by expected heterozygosity [H_E] and mean allelic richness) (Table 3) support that high variation occurred in the P. capsici population of Sichuan, Chongqing and Guizhou provinces. Correspondingly, previous research reported that P. capsici was considered to be a “plastic” pathogen with high variation in zoospore progeny (34), especially for the A2 mating type (29). Recent research (29) has shown that dynamic extreme aneuploidy plays an important role in the adaptation and evolution of P. capsici and ultimately leads to a high degree of genetic diversity even in clonal lineages. In our study, we also found that some isolates had three or more alleles in some SSR loci, which was similar to previously reported dynamic extreme aneuploidy (20, 29). Thus, dynamic extreme aneuploidy might be a common phenomenon in P. capsici. The specific molecular mechanism behind aneuploidy inheritance in P. capsici will help us to unravel the variation mechanism of P. capsici in the future.

In terms of the geographical characteristics of P. capsici, we found that diverse clonal lineages of P. capsici existed across most of the counties in Sichuan, Chongqing, and Guizhou provinces although both the A1 and A2 mating types simultaneously existed in some areas. However, sexual reproductions only existed in Jiulongpo county of Chongqing, which was authenticated by the mating type composition, linkage disequilibrium test, and oospore detection. In previous studies, P. capsici from Yunnan (35) and Gansu (8), two other provinces of China, were also found to have both the A1 and A2 mating types isolates in the fields. Phytophthora capsici in Yunnan provinces were presumed to have had sexual reproduction (35). On the contrary, long-lived genetically diverse clonal lineages were found to exist across Gansu provinces (8). Associated with genetic diversity results (Table 3), extensive variation also occurred in Sichuan and Guizhou within the context of long-lived clonal lineages, which was similar to the findings regarding P. capsici populations in Gansu province of China as previously reported (8). The opposite inferences between linkage disequilibrium and the mating type composition in Dafang, Weiling, and Pengzhou counties might be caused by the high variation in clonal lineages as a previous report (8). The absence of sexual reproduction in most fields in these three provinces indicates that raised beds, plastic mulch, and drip irrigation can still work well in disease control of pepper blight due to the lack of oospores in the soil. The population of P. capsici in Chongqing was different from that in Sichuan and Guizhou in that the former had higher genetic diversity with a limited amount of sexual reproduction. In terms of geography and climate, Chongqing has more valleys, a lower altitude, and higher temperature and humidity than the two other provinces. The different linkage disequilibrium and obvious differentiations among the different counties in Chongqing also corresponded with P. capsici being an atypical soilborne pathogen, which cannot easily outcross between long-distance regions (31). The population characteristics of long-lived diverse clonal lineages in most locations in Southwest China indicate that a reasonable layout of cultivars, which have different resistance genes, is also beneficial for the prevention of pepper blight. Considering that there was no obvious population differentiation between the isolates from Sichuan and Guizhou, with two identical genotypes even simultaneously appearing in these two provinces separated by a large geographical distance, we should pay more attention to the contamination of P. capsici in the market circulation of pepper fruits and seeds. Indeed, many peppers are imported from Guizhou to meet the consumption demands in Sichuan province according to the market information.

In terms of the temporal characteristics of P. capsici, the population of P. capsici in our study areas was relatively stable during the two consecutive years (2019 and 2020) as the populations had similar levels of genetic diversity and similar structure assignment, and none of them had population differentiation. The result of the AMOVA showed that the major variance in the different regions was within populations, which was also in line with the finding that the population of P. capsici in our study areas was relatively stable during those two consecutive years.

Finally, the genetic characteristics of P. capsici from the two different types of pepper (hot peppers and bell peppers) were also compared. The small but significant population differentiation between hot peppers and bell peppers implied that cultivars put selective pressure on P. capsici populations. This conclusion was also supported by a previous report stating that P. capsici varies significantly among different host species (11). Thus, we also suggested that the adaption and virulence variation in P. capsici in different cultivars should also be considered in the process of developing effective disease-resistant varieties. Variations in the virulence factors (such as effectors) of P. capsici populations from hot peppers and bell peppers will help us uncover the molecular mechanisms behind the adaption of P. capsici to different types of peppers in the future.

Taken together, we proposed that the P. capsici populations in Sichuan, Chongqing, and Guizhou provinces of Southwest China have high and complex genetic diversity with uneven geographic variation, limited sexual reproduction, relative stability between the two consecutive years, differentiation between the isolates from hot peppers and bell peppers, and a distant relationship with the two referenced isolates from American. The population dynamics in these three provinces of China were speculated to be driven by differences in geographical environment, different cultivars, two reproductive patterns, and potentially artificial long-distance transfers.

MATERIALS AND METHODS

Sources and single-zoospore isolation of P. capsici.

A total of 239 isolates of P. capsici were collected, including 235 isolates from Southwest China (Chongqing municipality, Guizhou province, and Sichuan province), 1 isolate from Northwest China (Qinghai province), 1 isolate from East China (Shandong province) and 2 isolates from America (Table S1 in Supplemental File 2). The sampling was carried out at as many sites as possible, and only one sample with obvious disease symptoms was collected in a 200-m² plot of pepper. Only one pathogen strain was isolated per sample. To isolate pathogens, plant patches with healthy and diseased tissue were cut from infected plants, surface-disinfested in 1% NaClO solution for 30 s, rinsed with sterile water three times, and cultured on 10% clarified V8-juice agar medium with rifampicin (20 mg/L), ampicillin (100 mg/L) and pentachloronitrobenzene (67 mg/L) for 3 to 4 days at 22°C. Zoospores of P. capsici were prepared by placing P. capsici discs (5 mm diameter) of 4-day-old cultures grown on clarified V8 agar into a Petri dish containing a mixture of sterile distilled water with a soil-leaching solution and incubated for 3 days to promote zoospore production. Every isolate of P. capsici was purified by single-zoospore isolation by spreading zoospores on a V8 agar plate and picking one colony on a new V8 agar plate as a purified single zoospore isolate. All the isolates were confirmed by PCR with specific primers (PC-F, 5′-GTCTTGTACCCTATCATGGCG-3′, and PC-R, 5′-CGCCACAGCAGGAAAAGCATT-3′) as previously described (27).

Detection of mating type.

To determine the mating type, a 5 mm plug of all the isolates was picked from the 5-day-old culture to pair with the known A1 and A2 mating type isolates, which were provided by Yuee Tian (Henan University of Science and Technology), on V8 agar plates. Each purified isolate was also paired cultured with itself to determine whether it could self-fertilize. After incubation at 22°C for 5 to 10 days, oospores were examined by microscope. The Chi-squared (χ²) calculation was used to test whether the ratio of the A1/A2 mating types was significantly different from a ratio of 1:1 in the same location.

DNA extraction.

Each P. capsici strain was grown on 10% clarified V8-juice agar medium at 22°C in the dark for 4 days. Disks (~1 cm in diameter) from the cultured V8 plates were transferred to 30 mL of 5% (vol/vol) V8 broth in a 90-mm Petri dish and grown at 23°C in the dark for 3 days. Then, the mycelia from these disks, which were grown from one isolate, were pooled for DNA extraction. DNA from each P. capsici isolate was extracted with the Rapid Fungi Genomic DNA Isolation kit (Sangon Biotech, Shanghai, China) and stored at −20°C.

Preparation of genomic library and sequencing.

Genomic DNA was extracted from mycelia using TruSeq DNA Sample Prep kit (Illumina, San Diego, CA, USA) and used to construct paired-end sequencing libraries with the TruSeq DNA Sample Prep kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. Cluster generation, template hybridization, isothermal amplification, linearization, blocking, denaturation, and hybridization of the sequencing primers were also performed according to the manufacturer's instructions. The base sequence of the template DNA fragment was obtained by counting the fluorescence signal with the SolexaPipeline. All these raw data were deposited in the National Genomics Data Center (NGDC) Genome Sequence Archive (GSA) under the accession number CRA007265.

Processing of Illumina data and screening of polymorphic SSRs.

For the raw sequencing reads, adapters and low-quality bases (base quality <20) from either the start or the end of the reads were cut off using Trimmomatic v0.36. We mapped all the generated clean data against the updated reference genome (21) of P. capsici using bowtie2 (36). We removed reads with identical external coordinates and insert lengths using MarkDuplicates in Picard (http://broadinstitute.github.io/picard). The multi-isolate indels were identified using the HaplotypeCaller and GenotypeGVCFs modules in the GATK pipeline (version 4.1.2.0) (37). Indels were subjected to quality control and filtered if they met the following criteria: (i) GATK hard filtering: quality by depth <2.0, mapping quality <40.0, read position rank sum test <−8.0, Fisher strand >60.0, haplotype score >13.0, mapping quality rank sum test <−12.5; (ii) quality scores (GQ) <20 and depth values over all the isolates less than 10; (iii) missing rate >80% and minor allele frequency <0.05; and (iv) genotyped in, at most, 40% of the isolates.

To excavate polymorphic SSRs, MISA was used for the prediction of SSRs in the whole updated reference genome as previously described (38). Second, the cleaned paired reads were assembled de novo using MEGAHIT (version 1.2.9) (39) with the default settings. Subsequently, the polymorphism SSRs were calculated using the CandiSSR pipeline (40) with the settings (-l 150 -s 95 -c 95 -t 50). Finally, the polymorphism SSRs were improved using the following stepwise protocol: (i) no indels in the 150 bp flanking sequence and (ii) read depth of the same SSR allele in each isolate greater than 5.

SSR amplicons detection.

Six frequently used SSRs (18, 20) and four novel polymorphism SSRs, which were calculated as previously described, were used to detect the genotyping of all the isolates. For the population genetics study, each primer pair was labeled with FAM or HEX fluorescent labeling, as in previous reports (8). PCR with high-fidelity DNA polymerase (2× High-Fidelity Master Mix, NEB, USA), was performed with a 25-μL volume, using the following conditions: 94°C for 4 min; 35 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 25 s; and a final extension of 72°C for 5 min. Amplicons were visualized using agarose gel electrophoresis and then detected by an ABI3730XL (ThermoFisher Scientific, America) in Zhongke Yu Tong Biotechnology Co., Ltd. (Shaanxi). To confirm the sequence of the PCR products, amplicons of different sizes, produced by general primer, were linked into pBM16A (pBM16A-TOPO Clone Smart kit, China) and sequenced using the Sanger method in Sangon Biotech Co. Ltd. (Shanghai).

Population analysis of P. capsici using SSR markers.

Considering that P. capsici is a diploid organism, we assumed that there were two alleles at each locus, as in previous studies (8). Thus, isolates with more than two alleles were filtered for further SSR data analysis here. In total, 163 isolates with no more than two alleles were used for the subsequent study. The genotype accumulation curve was calculated to assess the ability of SSR loci to distinguish distinct individuals in our study. Genetic diversity was estimated by both genotypic diversity and gene diversity. The population genotypic diversity was described by the number of isolates, the observed multilocus genotype (MLG), the expected MLGs (eMLG) in the smallest sample size of ≥ 10 based on rarefaction (41), the Shannon-Wiener Index (H) of MLG diversity (42), the corrected Simpson index (corrected λ) (43) and genotypic evenness (E.5) (44). All these indices, genotype distributions, and “clone-corrected data sets,” in which each genotype in each subpopulation includes only one representative isolate, were calculated using the R package poppr (45, 46). On the other hand, the mean number of observed alleles (A_O), the observed heterozygosity (H_O), and the expected heterozygosity (H_E) were calculated using POPGENE v. 1.32 (47) with the “clone-corrected data.” The fixation index (48) was calculated as F_IS = 1 − H_O/H_E. Gene (allelic) richness was estimated using rarefaction with the ADZE 1.0 program (49). Both H_E and Gene (allelic) richness were used to describe the gene diversity in our populations. To determine whether the population structure had regular sexual reproduction, we also tested the SSR loci for linkage disequilibrium with the R package poppr (45, 46).

Analysis of molecular variance (AMOVA), the Weir and Cockerham coefficient of differentiation θ (equivalent to Wright’s F_ST), discriminant analysis of principal components (DAPC), and model-based Bayesian clustering were used to estimate the population subdivision or structure. AMOVA was calculated using GenAlex v.6.5 (50) with the “clone-corrected data.” The hypothesis of nondifferentiation among populations was tested by comparing the observed θ (equivalent to Wright’s F_ST) value with the value calculated for the “clone-corrected data” in which alleles were resampled without replacement (1,000 randomizations) using the MULTILOCUS v. 1.3b program (32). Population subdivision was also examined by model-free discriminant analysis of principal components (DAPC), which was implemented by the R package adegenet (51). The population structures of our P. capsici isolates were investigated using model-based Bayesian clustering with STRUCTURE 2.3.4 (52, 53). The program was run with the admixture model, and cluster numbers (K) were evaluated from K = 1 to K = 9 using 100,000 iterations after a burn-in period of 100,000 iterations. To evaluate the stability of the results across repeated runs, seven independent runs were conducted. The runs for each value of K were evaluated based on the second-order rate of change of the likelihood function with respect to K using the online STRUCTURE HARVESTER v.0.6.94 program (54).

To reveal the genetic relationships among all the detected P. capsici genotypes, a minimum spanning network (MSN) was produced from the distance matrix using the R packages magrittr and poppr. On the other hand, an unrooted unweighted pair group method based on shared allele distance was also constructed with all the clone-corrected data using PowerMarker v3.25 (55), and the phylogenetic trees were visualized using iTOL (https://itol.embl.de/).