Fingerprinting 146 Chinese chestnut (Castanea mollissima Blume)accessions and selecting a core collection using SSR markers

NIE Xing-hua,WANG Ze-hua,LIU Ning-wei,SONG Li,YAN Bo-qian,XING Yu,ZHANG Qing,FANG Ke-feng,ZHAO Yong-lian,CHEN Xin,WANG Guang-peng,QIN Ling,CAO Qing-qin

1 Beijing Advanced Innovation Center for Tree Breeding by Molecular Design,Beijing University of Agriculture,Beijing 102206,P.R.China

2 College of Plant Science and Technology,Beijing Key Laboratory for Agricultural Application and New Technique,Beijing University of Agriculture,Beijing 102206,P.R.China

3 College of Landscape Architecture,Beijing University of Agriculture,Beijing 102206,P.R.China

4 Chestnut Technology Experiment and Extension Station of Huairou District,Beijing 102206,P.R.China

5 Shandong Institute of Pomology,Shandong Academy of Agricultural Sciences,Tai’an 271000,P.R.China

6 Changli Institute of Pomology,Hebei Academy of Agriculture and Forestry Sciences,Changli 066600,P.R.China

Abstract Chinese chestnut is an important nut tree around the world.Although the types of Chinese chestnut resources are abundant,resource utilization and protection of chestnut accessions are still very limited.Here,we fingerprinted and determined the genetic relationships and core collections of Chinese chestnuts using 18 fluorescently labeled SSR markers generated from 146 chestnut accessions.Our analyses showed that these markers from the tested accessions are highly polymorphic,with an average allele number (Na) and polymorphic information content (PIC) of 8.100 and 0.622 per locus,respectively.Using these strongly distinguishing markers,we successfully constructed unique fingerprints for 146 chestnut accessions and selected seven of the SSR markers as core markers to rapidly distinguish different accessions.Our exploration of the genetic relationships among the five cultivar groups indicated that Chinese chestnut accessions are divided into three regional type groups:group I (North China (NC) and Northwest China (NWC) cultivar groups),group II (middle and lower reaches of the Yangtze River (MLY) cultivar group) and group III (Southeast China (SEC) and Southwest China (SWC)cultivar groups).Finally,we selected 45 core collection members which represent the most genetic diversity of Chinese chestnut accessions.This study provides valuable information for identifying chestnut accessions and understanding the phylogenetic relationships among cultivar groups,which can serve as the basis for efficient breeding in the future.

Keywords:Castanea mollissima,SSR markers,fingerprinting,phylogeny,core collections

1.Introduction

Chinese chestnut is one of the earliest nut trees domesticated in China,and it has been cultivated in China for more than 6 000 years (Zhanget al.2005;Hao and Zhang 2014).The nut yield of Chinese chestnut is ranked first worldwide at approximately 1.6 million tons,and production of the Chinese chestnut accounted for 83.3% of the total annual production(http://www.fao.org/home/en/).Due to long-term artificial selection,Chinese chestnut has significant edible qualities and strong resistance,with important genetic resources for the improvement of edible chestnut in various countries around the world (Zhanget al.2005;Langet al.2006).

Today,more than 350 Chinese chestnut cultivars(accessions) have been bred through seedling selection and cross-breeding by farmers and breeders.To preserve Chinese chestnut accessions for the future,a national chestnut germplasm repository was established in Tai’an City,Shandong Province,and a provincial chestnut germplasm repository is located in Changli,Hebei Province.In addition,some local Chinese chestnut banks have also been established (Yanget al.2017).These preserved chestnut accessions are crucial for the modern chestnut industry because of their excellent traits,such as good quality,suitability,dwarf characteristics and high yield.However,because grafting scions are widely spread without labels among various producing regions,there are numerous synonyms or homonyms for some accessions and widespread mislabeling of germplasm,which hinder cultivar identification,exploitation,evaluation and utilization (Xuet al.2016).Therefore,an effective strategy for identifying accessions is urgently needed to ensure the purity of the breeds during propagation and the uniqueness of new accessions,consequently providing a basis for protecting the rights and interests of breeders,farmers and merchants.

The traditional methods of cultivar identification obviously have great limitations for a large number of resources.While traditional plant cultivar identification is based on morphological characteristics and pedigree information,morphology can be influenced by environmental factors(Alcaraz and Hormaza 2007).In the last two decades,molecular markers have been developed and extensively applied in genetic diversity,genetic map and genomic analyses in plants,such as restriction fragment length polymorphism (RFLP),randomly amplified polymorphic DNA(RAPD),amplified fragment length polymorphism (AFLP),simple sequence repeat (SSR) and single nucleotide polymorphism (SNP) identification strategies for plant accessions (Kubisiaket al.1997,2013;Casasoliet al.2001,2004;Siscoet al.2005;Ariaset al.2016;Cuestaset al.2017;Olssonet al.2017).Of these,SSRs (microsatellites)are sequences of 1 to 6 bases that are repeated in tandem and widely distributed in the genome.The mutation rate of SSR tandem repeats is very high,resulting in many alleles and forming a highly polymorphic SSR (Gur-Arieet al.2000),so using a few markers combinations can increase detection efficiency.In addition,SSR markers are codominant and developed from genome and transcriptome sequencing of a given species.Based on these advantages,the International Union for the Protection of New Varieties of Plants (UPOV)recommends using SSR and SNP markers together as the preferred method for building DNA fingerprints (UPOV 2007).

In recent studies ofCastaneaplants,SSRs have mainly been applied to genetic diversity studies,marker association and the construction of genetic maps.Cross-species amplifications using 17 SSR markers were successful amongC.mollissima,C.crenata,andC.sativa;additionally,65 samples among three species exhibited considerably high genetic diversity (Inoueet al.2009).Jianget al.(2017)conducted an SSR-based analysis of 95 accessions from ten provinces in China to evaluate the genetic diversity,patterns of genetic structuring,and linkage disequilibrium(LD) of Chinese chestnut.Mülleret al.(2018) reported that genotyping 272 American chestnut individuals with 24 SSR markers revealed five outlier loci significantly associated with environmental variables of the population sites.However,DNA fingerprinting using well-characterized SSR primer sets has not yet been carried out.Researchers are eager to know how many SSR markers are required to distinguish Chinese chestnut accessions.The fingerprinting power of SSR markers can be estimated based on their allele frequencies,although new accessions are emerging continuously.

In the 1980s,Frankel (1984) and Brown (1989)proposed and improved on the concept of core collections.Because the guiding principle of core collections is that the minimum sampling quantity represents the maximum genetic diversity,the difficulties and pressures associated with large germplasm collection are effectively alleviated.A core collection is the vital subset for in-depth resource research,fine gene mining and new technology application,which can improve the effective utilization rate of germplasm resources.It is necessary to establish a core collection of Chinese chestnut,since it is a perennial woody plant,and the management of large germplasm collections is often costly,time consuming and labor-intensive.

This study focused on the following principal objectives:(1) to screen a core set of markers for fingerprinting a collection of Chinese chestnut accessions;(2) to clarify the genetic background,differentiation characteristics and genetic relationships of five cultivar groups;and (3) to construct a rational core collection to be used as a genetic diversity library of germplasm resources,which will facilitate the breeding of new accessions with great agronomic traits.

2.Materials and methods

2.1.Plant materials and genomic DNA extraction

A total of 146 Chinese chestnut (Castanea mollissimaBlume) accessions were collected from the National Chinese Chestnut Germplasm Repository located in Tai’an City,Shandong Province,China,and from the Chestnut Technology Experiment and Extension Station of Huairou District,Beijing,China.Accessions were locally selected from 14 different provinces.According to the main economic characteristics of nuts combined with the distribution of producing areas,the 146 accessions were divided into six cultivar groups (Zhanget al.2005):the North China(NC),middle and lower reaches of the Yangtze River(MLY),Northwest China (NWC),Southeast China (SEC),Southwest China (SWC) and Northeast China cultivar groups.Of these,the Northeast China cultivar group belongs to Japanese chestnut,so it was not included in this study.Detailed information for each accession,including breeding methods,germplasm type,cultivation region and accession name,is presented (Appendix A).

Young leaves for all accessions collected were immediately flash frozen in liquid nitrogen and then stored at−80°C until processing.Total genomic DNA was extracted from the leaves using an improved cetyltrimethylammonium bromide (CTAB) method (Caoet al.2007).The quality and quantity of DNA were tested using 1% agarose gel electrophoresis and the NanoDrop One spectrophotometer(Thermo Fisher Scientific,MA,USA),respectively.Subsequently,the concentration of each DNA sample was diluted to 20–50 ng μL–1for PCR amplification.

2.2.SSR marker selection and allele reading

SSR markers were screened from aC.mollissimagenetic map that contained 1 401 markers involved in 330 SSRs and 1 071 SNPs (Kubisiaket al.2013).First,twelve chestnut accessions distributed in 11 provinces were used as primary screening materials,and their allele bands were analyzed on 8% silver-stained polyacrylamide gels.We selected a total of 18 pairs of polymorphic primers scattered throughout the nine linkage groups of theC.mollissimagenome (Appendix B).Afterwards,PCR amplification was performed in 20 μL reaction mixtures containing 1 μL of template DNA (20–50 ng μL–1),10 μL of 2× PremixTaqTM(TaKaRa,Dalian,China),0.1 μL of forward primer (10 μmol L–1),0.3 μL of reverse primer (10 μmol L–1),0.2 μL of M13 primer (10 μmol L–1)with a fluorescent label (6-FAM (6-Carboxyfluorescein),HEX(Hexachlorofluorescein),ROX (Carboxy-X-Rhodamine) or TAMRA (Carboxytetramethylrhodamine)) (Tsingke,Tianjing,China),and 9.6 μL of ddH2O.The thermal cycling program consisted of predenaturation at 94°C for 3 min,35 cycles at 94°C for 30 s,56°C for 30 s,and 72°C for 30 s,and a final extension at 72°C for 10 min.Finally,PCR-amplified products from three or four pairs of primers were pooled together.Then,mixtures were analyzedviacapillary electrophoresis using an ABI 3730 XL DNA Sequencer(Applied Biosystems,Foster City,CA,USA),and alleles were read using GeneMarker v 2.2.0 Software package(SoftGenetics LLC,State College,PA,USA).

2.3.Data analysis

The key genetic diversity statistics of the SSR markers,including major allele frequency (MAF),number of alleles(Na),gene diversity (GD),heterozygosity (Ho),polymorphic information content (PIC),the frequency of alleles in different cultivar groups and genotype distribution of each marker and Nei’s genetic distance (Neiet al.1983),were calculated by PowerMarker v3.25 (Liu and Muse 2005).

To evaluate the fingerprinting ability of the SSR markers,the probability of identity (PI) of each marker and their combinations (PIsibs) were estimated by GenAlEx 6.51(Peakall and Smouse 2006;Smouse and Peakall 2012).Based on these results,to generate an appropriate core marker set,several criteria were established (Patzaket al.2007;Tanet al.2015;Liuet al.2017),including (1) the amplified products were unambiguous without nonspecific amplification;(2) the marker showed a high degree of polymorphism and (3) the PI<0.3,so that a combination of some markers can provide sufficient discriminatory power.

A phylogenetic tree of cultivar groups was constructed based on Nei’s genetic distances and the neighbor-joining method and created in MEGA 7.0 (Kumaret al.2016).There were six cultivar groups (five chestnut cultivar groups and an external group):the NC,MLY,NWC,SEC,and SWC cultivar groups and an external group consisting of nineC.henryiaccessions (Ch).

The genetic differentiation fixation index (FST) with two levels involving each locus and between two groups,Shannon-Wiener index (I),gene flow (Nm) and analysis of molecular variance (AMOVA) were calculated by PowerMarker v3.25 and GenAlEx 6.51.

A strategy was chosen before the core collection was constructed,namely,a simulated annealing algorithm based on the maximizing genetic diversity (SAGD) was used(Wanget al.2012;Xuet al.2016;Wuet al.2019).Groups including 5,10,15,20,25,30,35,40,45,50,55,60,65,70,75,80,85,90,95 and 100 samples were built.According to five-fold averages,a relationship between the sample capacity and the number of alleles retained in the core collection was identified to determine the sample capacity in PowerMarker v3.25 Software.Afterwards,we selected accessions with special traits,such as belonging to earlymaturing,brachyblast,pendulous branch,male abortion and red bur types,to serve as the required germplasm in the core collection.After determining core collection capacity and required accessions,a total of 1 000 independent runs were conducted using the PowerMarker v3.25 Software,and the final core collection was determined according to the frequencies of occurrence per 1 000 times in each accession.

SPSS v20.0 (SPSS,Chicago,IL,USA) was used to evaluate the final core collection by performing Levene’s test and t-tests for MAF,Na,Ho,and PIC values between the core and the original collection.

3.Results

3.1.Polymorphism of SSR markers

Eighteen SSR markers were detected to analyze the alleles and locus genotypes of 146 Chinese chestnut accessions distributed among five cultivar groups in China,and a total of 133 alleles and 305 locus genotypes for 18 loci were obtained (Appendices C and D).To explore the genetic properties of these markers,we subsequently calculated key genetic statistics,including MAF,Na,GD,Ho,and PIC(Table 1).The number of alleles varied from four alleles at loci CmSI0742 and CmSI0809 to 16 alleles at CmSI0930 with a mean of 8.10 alleles per locus.The MAF ranged from 0.216 at locus CmSI0922 to 0.640 at locus CmSI0933 with an average of 0.420.TheHovaried from 0.479 to 0.767 with an average of 0.622.Furthermore,as an important genetic diversity estimator,the PIC value certified that the 18 SSR markers have high polymorphism,with an average of 0.652.Of the screened markers,16 exhibited highly polymorphic trends (PIC>0.5),with PIC values varying from 0.513 at locus CmSI0396 to 0.813 at locus CmSI0922.Two markers were moderately polymorphic (0.25

3.2.Diversity of Chinese chestnut cultivar groups with unique alleles

We detected a total of 133 alleles among the 146 diploid accessions,and no locus was missing.Afterwards,the frequency and distribution of alleles of each marker in the different cultivar groups were also counted,and the five cultivar groups had 119,110,77,93 and 80 alleles (Appendix C).Among the five cultivar groups,the NC cultivar group had the most alleles,accounting for 89.5% of all alleles;the MLY cultivar group accounted for 82.7% of all alleles;and the lowest number of alleles was found in the NWC cultivar group,which accounted for 57.9% of all alleles.A number of common alleles were discovered,but the frequencies of the common alleles had distinct differences between groups.For instance,the frequencies of the 114 bp allele at locus CmSI0809 were 0.274,0.308,0.500,0.083 and 0.214 in the MLY,SEC,NC,NWC and SWC cultivar groups,respectively.In addition,many alleles were unique to one of the cultivar groups,such as the 175 bp allele of CmSI0871,the 281 bp allele of CmSI0930,the 276 bp allele of CmSI0853 and the 193 bp allele of CmSI0933,which occurred in only the NCcultivar group,MLY cultivar group,SWC cultivar group and NWC cultivar group,respectively.

Table 1 Key genetic statistics of the 18 SSR markers1)

3.3.The power of fingerprinting

A total of 305 locus genotypes were amplified by the 18 SSR markers,and they consisted of polymorphic and repeatable alleles.Based on the genotype data,we carried out a multilocus match analysis in GenAlEx 6.51,and no match was detected,showing that each of the 146 chestnut accessions had a unique SSR multilocus combination.To assess the fingerprint identification ability of the 18 markers,two crucial statistics,PI and PIsibs,were calculated(Table 1).The PI value of each locus ranged from 0.049 to 0.289,with an average of 0.144.Assuming that all marker loci segregate independently,the probability of identifying two random individuals with the same multilocus locus genotypes at all 18 loci was reckoned to be 7.398×10–17.PIsibs,which was ascertained as the upper limit of PI(Waitset al.2001),ranged from 0.345 to 0.560 with an average of 0.437 for each locus,and the combined PIsibs was 2.880×10–7.

Based on the data obtained,the identification ability of the 18 SSR marker combinations to distinguish each Chinese chestnut cultivar was calculated.The results showed that all of the chestnut accessions could be identified independently when a minimum of seven SSR markers were combined(Fig.1).According to the principle of core marker selection and the genotype data,seven high-quality SSR markers were screened as the core marker set for fingerprinting the 146 Chinese chestnut accessions.First,with a high PIC value and low PI value,CmSI0922 was selected,and the 146 accessions were divided independently into 28 groups;but only six accessions were separated from all other accessions.When a second marker (CmSI0614) was added for further analysis,79 of 146 accessions could be uniquely identified,accounting for 54.10% of all accessions.When the third (CmSI0561),fourth (CmSI0509),fifth (CmSI0933)and sixth (CmSI0396) markers were successively added to the analysis,the number of unique marker combinations increased to 102 (69.86%),107 (73.28%),125 (85.62%)and 138 (94.52%).Finally,the addition of a seventh marker(CmSI0800) was able to completely distinguish all 146 accessions,and the combined PI and PIsibs of the seven core markers were 7.146×10–8and 3.122×10–3,respectively.Therefore,we constructed fingerprints of all 146 Chinese chestnut accessions using the seven determined core markers (Appendix E),and the allele sizes and frequencies of these core markers are summarized (Appendix F).

3.4.Phylogenetic tree analysis of cultivar groups

To accurately determine the relationships among the five Chinese chestnut cultivar groups,we added aCastanea henryigroup consisting of nine accessions as an outgroup reference (Appendix A).We constructed a phylogenetic tree of the six individual cultivar groups based on Nei’s genetic distance and the neighbor-joining method in MEGA 7.0 (Fig.2).Cluster analyses clearly placed the NWC and NC cultivar groups into a subclade,and the SWC and SEC cultivar groups clustered into a second subclade.It was speculated that the MLY cultivar group was independent of these two subclades.

Fig.1 Evaluation for the fingerprinting power of 18 SSR markers in Chinese chestnut accessions.PI,the average probability of two random individuals having the same genotype.

3.5.Genetic diversity analysis of different cultivar groups

The genetic differentiationFSTwith two levels,including each locus and population differentiation,was estimated.TheFSTof each locus ranged from 0.012 to 0.099 with an average of 0.042 (FST<0.05) (Table 1).A matrix ofFSTpairwise groups among the five Chinese chestnut cultivar groups was also established (Appendix G);theFSTranged from 0.013 to 0.051,with an average of 0.026 (FST<0.05).This suggested that genetic differentiation rarely existed among the groups.Moreover,Shannon’s Information Index showed differences in genetic diversity among the five cultivar groups,with the diversity decreasing in the order MLY>NC>SEC>SWC>NWC (Appendix H).Finally,an AMOVA was carried out to determine the total amount of variation and the distribution among or within groups in five Chinese chestnut cultivar groups.AMOVA revealed that the genetic variation in Chinese chestnut accessions mainly existed within the groups,accounting for 98% of the total variation,and that 2% of the genetic variation occurred among groups (P<0.01) (Appendix I).This result was consistent with that obtained from theFSTvalues.In addition,theNmaverage value was 7.434 (Table 1),which indicated that gene exchanges occurred frequently within the groups.

Fig.2 Neighbor-joining (NJ) phylogenetic tree for six cultivar groups using the Nei’s genetic distances matrix.Ch is C.henryi cultivar group,MLY,SWC,SEC,NWC and NC represent the middle and lower reaches of the Yangtze River cultivar group,Southwest China cultivar group,Southeast China cultivar group,Northwest China cultivar group,and North China cultivar group of Chinese chestnut,respectively.In the figure,0.02 shows the standard ruler for distance cluster of each cultivar group.

3.6.Selection of the core collection

First,to determine the most suitable size of the core collection,20 different sample grab volumes from all samples were designed using a simulated annealing algorithm (Fig.3).As described in Fig.3,the allele retention approached the number of total alleles.When the core selection capacity reached 45,the allele number was approximately 122,so it captured close to 92% (122/133)of the total alleles.Then,12 accessions were selected with distinctive traits as required materials,which were important for use in genetic research and breeding resources based on the unique features of early maturity,dwarf branching and bending,bur color and commercial importance in the Chinese chestnut industry (Appendix J).Ultimately,we constructed a core collection for Chinese chestnut that retained 30.8% (45/146) of the original collection using a simulated annealing allele-first strategy (Appendix K).The retention rates of MAF,Na,Hoand PIC were 104.2%,88.7%,99.0% and 98.8%,respectively,between the core collection and the entire set of accessions.Significant differences were not found in the genetic diversity index of the core collection constructed through Levene’s test andt-test rooted in the mean values of MAF,Na,Ho,and PIC (Table 2).

4.Discussion

Fig.3 Allele numbers shown as a function of the sample size based on the simulated annealing algorithm to evaluate the capacity of the core collection.

Table 2 Genetic diversity comparison,Levene’s test and t-test between the original collection,and the core collection constructed by simulated annealing algorithm1)

SSR molecular markers are characterized by stability,repeatability and the polymorphism of the genome;and they are also applied in various fields,such as resource identification,genetic diversity and the selection of core collections (Tautz and Renz 1984;Sardaret al.2016).For example,Guet al.(2019) studied genetic diversity,core collections and population structure ofCapsicumgermplasm accessions using 29 SSR markers;Wanget al.(2019) selected 17 pairs of SSR primers from genome to identify 11 commercial cultivars ofAgaricus bisporus.However,the effectiveness and success of these studies greatly rely on the quality of markers and the accuracy of genotype data (Selkoe and Toonen 2006;Liuet al.2017).In this study,18 pairs of primers with high polymorphism were selected from 330 SSR markers.We acquired 133 allele loci and 305 locus genotypes with an average of 7.38 alleles among 146 chestnut accessions,which was nearly three times as large as the result of Jianget al.(2017) and consistent with the amount reported by Tianet al.(2009).The mean value ofHowas calculated to be 0.622 using 18 SSR markers,showing high heterozygosity,a level which was highly consistent with the 0.87% heterozygosity level of the entire Chinese chestnut genome (Xinget al.2019).The highly heterozygous characteristic of Chinese chestnut is inseparable from its being a cross-pollination plant (MoKay 1942;Payneet al.1983;Shi and Stösser 2005).The average PIC was 0.652 (>0.500),and the maximum was 0.813,representing high polymorphism.The reason was likely because more ecological accessions were collected in the current study than in previous studies,and they spanned five major Chinese cultivar groups.In addition,capillary electrophoresis was utilized to read and record the data,which not only tremendously improved the accuracy of loci detection but also greatly overcame the problem of the low resolution of polyacryamide gel electrophoresis (PAGE).These relatively high polymorphism markers will contribute greatly to research on genetic diversity,the identification of resources and the determination of fingerprints.In addition,considering the locus analysis of distinct cultivar groups,we observed that each group has its own unique allele loci and frequency differences of shared loci,which might result from the selection of both distinctive geographical environment conditions and seedling selection.

After genotyping,PI and PIsibs were determined,which are essential indexes for fingerprinting identification(Donnelly 1995;Waitset al.2001;Tanet al.2015;Liuet al.2017).These values were extremely low (7.398×10–17and 2.880×10–7,respectively) when 18 SSR markers were used,suggesting that the identification capacity of these fingerprints should be excellent.In fact,the population substructure is tightly related to selection by the environment and human intervention,so that some alleles occur very infrequently.In addition,the assumption of independent segregation among loci is not credible.In conclusion,the PI value tends to be overestimated (Waitset al.2001).In addition,the PIsibs value is taken as the upper limit of the PI value.Waitset al.(2001) thought that when PI is approximately 1×10−4–1×10−2,most natural populations can be identified.According to the results of multilocus matches analysis,we ensured that when the minimum number of core markers needed was seven,the 146 accessions could be completely distinguished (Fig.2).Furthermore,the PI and PIsibs of these core markers were 2.429×10–5and 3.122×10–3,respectively,which was just within the bounds of Waits’ hypothesis.

Cultivated chestnuts are mainly derived from seedling selection and are divided into five primary cultivar groups based on their economic characteristics and ecological type:the NC,MLY,NWC,SEC,and SWC cultivar groups (Zhanget al.2005).To deeply explore the genetic diversity and relationships of chestnut groups,a phylogenetic tree was built with aC.henryigroup as an external reference.This tree revealed that the NC cultivar group was clearly relatively close to the NWC cultivar group,forming an independent subclade,which agrees with the research results of Bao and Huang (2002).Additionally,this phylogenetic tree also revealed that the SWC cultivar group and the SEC cultivar group made up another subclade,having a fairly large genetic distance from the MLY cultivar group.Subsequently,we calculated the Shannon-Wiener index of each cultivar group and found that the MLY cultivar group had the most genetic diversity,which was consistent with the results of Qinet al.(2002) and Zhou (2005).

Then,to evaluate the differentiation of the five chestnut cultivar groups,theFSTvalue was evaluated first.The averageFSTvalues were 0.042 and 0.026 (<0.05) at the marker level and the group level,respectively,which showed that different cultivar groups present a low degree of differentiation.To determine the reason for this low differentiation,we carried outNmanalysis and AMOVA of the five cultivar groups.The averageNmwas nearly 7.434,showing a high frequency of gene exchange.Wright (1931) suggested that when theNmof populations is over one,gene flow can play an efficient role in preventing genetic differentiation among various district groups.Meanwhile,genetic variation in Chinese chestnut was primarily rooted within the group,accounting for 98% of the AMOVA.Thus,we obtained a reasonable explanation for the low differentiation:Chinese chestnut,as a typical cross-pollination plant,had highly frequent gene exchange under natural conditions,and only this strong gene exchange prevented group differentiation.In addition,Chinese chestnut accessions were mostly bred by seedling selection,and varietal cultivation was launched late.For these reasons,a complex genetic background may preserve the polymorphism that existed in the early stages of population evolution.

As a cross-pollinated plant,the selection and breeding of Chinese chestnut accessions mainly involve excellent accessions,which leads to a complex genetic relationship.Therefore,we drew lessons from the nongroup strategy that was used to construct the core collection of jujube and torreya (Xuet al.2016;Wuet al.2019).A nongroup strategy using a simulated annealing algorithm based on the maximizing genetic diversity (SAGD) was determined.For the sampling proportion of the core collection,most researchers believe that sampling proportions between 5 and 30% can represent the genetic diversity of the entire collection (Brown 1989;Shashidharaet al.2003;Singhet al.2014).To avoid the loss of some rare resources due to the randomness of selection by PowerMarker Software,12 chestnut accessions with special traits such as early maturation,brachyblast,pendulous branch,and red bur,were used as the required materials.According to dynamic grab results,we discovered that when the samples collected randomly reached their maximum,nearly 30.8% (45/146) of allele loci obtained accounted for 122 alleles,accounting for approximately 92% of all allele sites.Notably,the retention value of Chinese chestnut core collection genetic diversity was slightly lower than the allele retention values of 95.74,100 and 99.5% in pear,grape and jujube,with sampling ratios reaching 24.2,19.4 and 15.6%,respectively (Songet al.2014;Huet al.2015;Xuet al.2016).The most likely reason was that the breeding of a majority of Chinese chestnut accessions were still from seedling selection,and lacked extensive gene exchanges from cross-breeding,which led to an increase in the ratio of the core collection.

5.Conclusion

In this study,the fingerprints of 146 Chinese chestnut accessions were successfully constructed with seven core markers screened from 330 reported markers.Subsequently,the relationships among five Chinese chestnut cultivar groups were determined,providing significant information to determine the phylogenetic relationships,group differentiation and genetic origins of the cultivar groups.Finally,we selected 45 samples for a core collection representing the largest genetic diversity of Chinese chestnut accessions,and this core collection will provide the basis for efficient breeding in the future.

Acknowledgements

This work was supported by grants from the Project of Construction of Innovative Teams and Teacher Career Development for Universities and Colleges under Beijing Municipality,China (IDHT20180509);the National Key Research &Development Program of China(2018YFD1000605) and the Opening Project of Beijing Key Laboratory of New Technology in Agricultural Application,China (kf2018024).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available on http://www.ChinaAgriSci.com/V2/En/appendix.htm