Introduction
Walnut (Juglans spp., Juglandaceae) comprises an economically important group of dual-purpose nut and timber trees worldwide. Nuts are included as one of the important elements of healthy dietary guidelines because they are considered a good dietary source of nutrients including unsaturated fatty acids. The walnut production of world was 3,662,507 tons in 2018, while the production of China reached 1,586,367 tons (FAOSTAT, 2012). Eight walnut species are found in China, two of which, common walnut: Juglans regia and iron walnut: Juglans sigillata, are widely cultivated species available as walnut breeding resources. J. regia is distributed globally, whereas J. sigillata is restricted to southwestern China (Wang et al., Reference Wang, Hao, Wang and Pei2007; Feng et al., Reference Feng, Yuan, Sun, Hu, Zulfiqar, Ouyang, Dang, Zhou, Woeste and Zhao2018; Yuan et al., Reference Yuan, Sun, Bai, Dang, Feng, Zulfiqar and Zhao2018). J. sigillata is cultivated in Yunnan Province, where the native species is economically important.
Yunnan Province, a distinct alpine region with various climatic and geographical features harbouring rich plant diversity, is a centre of origin and endemism of J. sigillata (Xi and Zhang, Reference Xi and Zhang1996). J. sigillata can be found in all 127 counties of the province at elevations of 850–2900 m (Lu, Reference Lu2009). The unique geography and climate of Yunnan are responsible for superior agronomic properties and valuable traits, such as high oil and protein contents, kernel quality and strong disease like cold and drought resistances (Xi, Reference Xi1987), that have significant potential value for variety improvement.
Genetic diversity results from the long-term existence, evolution and eco-adaption of a species; consequently, assessment of the genetic diversity of a species greatly aids its utilization and conservation. Most walnut trees are open-pollinated, which would lead to accumulate abundant genetic diversity. As a result of long-term environmental pressures and increasing human activities during the past decades, however, walnut resources have been depleted, thereby reducing genetic diversity. Although analysis on characteristics and economic quality of walnut nut in Yunnan has been done (Lu, Reference Lu2009) and molecular-based genetic analysis on some germplasm resource was finished (Ning et al., Reference Ning, Ma, Zhang, Wang, Lu and Pei2011; Xiao et al., Reference Xiao, Chen, Ning, Wu and He2018), the overall evaluation on the genetic diversity has not been systematically evaluated.
Several molecular marker techniques have been applied to examine the genetic diversity of walnut. Examples include methods based on randomly amplified polymorphic DNA (Wu et al., Reference Wu, Pei, Xi and Li2000; Nazeer et al., Reference Nazeer, Mir, Mir, Nazir, Rizwan, Shabir, Wajida, Hidayatullah and Sheikh2012), restriction fragment-length polymorphism (Fjellstrom and Parfitt, Reference Fjellstrom and Parfitt1994), amplified fragment-length polymorphism (Bayazit et al., Reference Bayazit, Kazan, Gulbitti, Evik, Ayanoglu and Ergul2007; Chen et al., Reference Chen, Hu, Zhang and Li2008; Wang et al., Reference Wang, Zhao, Gao, Zhang and Xuan2011), inter-simple sequence repeats (SSRs) (Potter et al., Reference Potter, Gao, Aliello, Leslie and McGranahan2002; Li et al., Reference Li, Ai, Zhang, Wei and Liu2011; Liu et al., Reference Liu, Liu, Wei and Gu2017), and SSRs (Wang et al., Reference Wang, Hao, Wang and Pei2007, Reference Wang, Pei, Gu and Wang2008; Shah et al., Reference Shah, Mir, Ahmed and Fazili2018; Shamlu et al., Reference Shamlu, Rezaei, Lawson, Ebrahimi, Biabani and Khan-Ahmadi2018). Lack of systematic study of the genetic diversity of a germplasm resource seriously restricts its genetic improvement and further utilization. In this paper, we therefore, drew on the work of the previous resource survey and analysed the SSR-based genetic diversity and genetic structure of J. sigillata germplasm sampled from 14 prefectures (cities) of Yunnan. The results should serve as a valuable reference for the conservation and breeding programmes.
Materials and methods
Plant material
A total of 1122 iron walnut individuals sampled from 14 prefectures of Yunnan were used for the analysis. All of these individuals were derived from wild plants growing naturally in the mountains or in villages, except for samples from KM (the capital of Yunnan). To better reflect geographic characteristics of districts and the effects of human activities on germplasm, we considered individuals from the same administrative district to be one population; consequently, 14 populations (namely, 14 administrative districts) were analysed (Fig. 1). In each population, 18–221 naturally growing, adult trees of seedling origin (>60 cm diameter at breast height) were sampled; to reduce the probability of sampling from family clusters, individuals separated from each other by a distance of more than 50 m were selected. Fresh mature leaves were randomly sampled from each individual, stored in plastic bags, and taken to the laboratory for SSR analysis. Locations and sampling sizes of the 14 populations in the study are listed in Table 1.
Fig. 1. Geographic locations of sampling sites of the 14 iron populations in this study. Note: The full names of abbreviated locations are given in Table 1.
Table 1. Locations and sampling sizes of the 14 iron walnut populations in this study
DNA extraction
Genomic DNA was extracted from stored leaves using an improved cetyltrimethylammonium bromide method (Yang et al., Reference Yang, Chen, Zhang, Fang and Xi2005). DNA quality and concentration were determined by gel electrophoresis and ultraviolet–visible photometry (Thermo Scientific, USA). Purified DNA was diluted with Tris-EDTA (Ethylene diamine tetraacetic acid) buffer to ~20 ng μl−1 and preserved in a freezer at −20°C for use in SSR PCR (Polymerase chain reaction) amplification.
SSR-PCR amplification
To amplify SSRs of all samples, 20 primer pairs developed for J. sigillata (Chen et al., Reference Chen, Ning, Wu, Xiao, Lin, Zhu, He and Pan2017) were synthesized by Aoke Biological Technology Co. (Beijing, China). Information on the 20 primers is listed in the online Supplementary Table S1. PCR amplifications were performed in 20-μl reaction volumes containing 10.0 μl of 2 × ES TaqMaster mix, 0.4 μl each of forward and reverse primers, 1.0 μl template DNA and 8.2 μl RNase-free water. Touchdown PCR amplifications were carried out as follows: initial denaturation for 5 min at 94°C, followed by 35 cycles of 30 s at 94°C, 1 min at an annealing temperature of 53–60°C (based on the specific primer used, see online Supplementary Table S1) and 1 min at 72°C, with a final extension of 10 min at 72°C. The amplification products were detected by capillary electrophoresis (3730xl DNA analyser) (Chen et al., Reference Chen, Ma, Chen, Wang and Pei2014).
Data statistics and genetic diversity analysis
Diversity of SSR loci
The sizes of detected amplification fragments were input into Excel, and GenAIEx 6.5 was used to calculate the number of alleles per SSR locus (N) and the following parameters: observed heterozygosity (H o), expected heterozygosity (H e), Shannon's information index (I), Nei's genetic distance (D) and gene flow (N m) (Peakall and Smouse, Reference Peakall and Smouse2012).
Population genetic diversity and genetic differentiation
The number of private alleles, mean allelic richness (AR), mean private allelic richness (PAR), H e and I were used to measure the diversity of the populations. The number of private alleles, AR and PAR were calculated with HP-RARE version 1.0 (Kalinowski, Reference Kalinowski2005), which uses rarefaction and hierarchical sampling to adjust for uneven sample sizes. Eight individuals (16 allele) were sampled at random from each population to match the smallest LC regional sample size. GenAIex 6.5 was used to calculate H e and I of the populations and it was also used to measure population genetic differentiation by calculating the within-population inbreeding coefficient (F IS) and the among-population genetic differentiation coefficient (F ST). Among-and within-population genetic differences were further estimated by analysis of molecular variance (AMOVA) of population genetic variation in Arlequin v3.1 (Testolin et al., Reference Testolin, Marrazzo, Cipriani, Quarta, Verde, Dettori, Pancaldi and Sansavini2000).
Population structure and cluster analysis
The program structure, version 2.3.2 (Pritchard et al., Reference Pritchard, Stephens, Rosenberg and Donelly2000) was used to detect the number of populations within germplasm collection of 1122 genotypes. The membership of each genotype was tested from K = 1 to K = 6 with admixture model, along with a burn-in period of 100,00 and a run length of 100,000. Based on the results, the ΔK parameter was estimated to identify the optimal number of clusters, as described in Evanno et al. (Reference Evanno, Regnaut and Goudet2005). A value of K was selected when the estimate of ΔK peaked in the range of 1–6 and inferred ancestry estimates of individuals (membership coefficients) were derived for the selected populations (Pritchard et al., Reference Pritchard, Stephens, Rosenberg and Donelly2000).
The unweighted pair-group method with arithmetic (UPGMA) means was used to construct a dendrogram based on Nei's distances in NTsys 2.10 (Rohlf, Reference Rohlf2000). Correlation of population genetic distance with geographic distance was estimated by a Mantel test (Mantel, Reference Mantel1967).
Results
SSR-PCR amplification and diversity of SSR loci
As an example, amplification products detected by capillary electrophoresis were shown for the case of primer CUJRA123 amplified from samples W092–W095 in online Supplementary Figure S1.
A total of 351 alleles were detected from 1122 samples using 20 pairs of primers, all of which were polymorphic with lengths ranging from 96 to 250 bp (online Supplementary Table S1). The number of alleles detected by each primer varied from 7 (JSI73) to 27 (CUJRB307), with an average of 17.55 (N, online Supplementary Table S2). Genetic diversity parameters for each locus were listed in online Supplementary Table S2. For all 20 loci, observed heterozygosity (H o) and expected heterozygosity (H e) ranged from 0.18 to 0.75 and 0.26 to 0.78, respectively, with corresponding averages of 0.46 and 0.57. Shannon's information index (I) ranged from 0.55 to 1.86, with an average of 1.19. Polymorphism information content (PIC) varied from 0.22 to 0.79, with an average of 0.57. With regard to specific loci, CUJRB307 had the highest values of H e (0.78), I (1.861) and PIC (0.79), while JUG13 exhibited the lowest values (H e = 0.26, I = 0.55 and PIC = 0.22).
Population genetic diversity
According to estimates of genetic diversity of the 14 iron walnut populations in Yunnan (Table 2), AR and PAR ranged from 3.40 to 4.62 and 0.11 to 0.36, with an average of 3.93 and 0.30, respectively. H e, I, and PIC ranged from 0.49 to 0.65, 0.84 to 1.40 and 0.41 to 0.60, with corresponding averages of 0.58, 1.19 and 0.53, respectively. These parameters differed among the 14 populations, with the DQ population having the highest values of AR, H e, I and PIC (4.62, 0.65, 1.40 and 0.60, respectively) and the LC population having the lowest values (AR = 3.40, H e = 0.49, I = 0.84, PIC = 0.41). As demonstrated by these results, the genetic diversity of J. sigillata in the 14 populations was highest in the DQ population and lowest in the LC population.
Table 2. Genetic diversity of 14 iron walnut populations in Yunnan
AR, mean allelic richness; PAR, mean private allelic richness; H o, obseved heterozygosity; H e, expected heterozygosity; I, Shannon's information index; PIC, polymorphic information content; F IT, total population inbreeding coefficient.
Population genetic differentiation
The inbreeding coefficient for the total population (F IT) per locus varied from −0.42 to 0.54, with an average of 0.19. The within-population inbreeding coefficient (F IS) ranged from −0.41 to 0.59 and the average value was 0.20. The among-population genetic differentiation coefficient (F ST) ranged from 0.01 to 0.09, with an average of 0.05 (online Supplementary Table S2), thus indicating low differentiation among populations (i.e. only 5.00% of the total genetic variability being inter-populational). The level of gene flow (N m) was estimated to be 6.70 (online Supplementary Table S2), which implied a relatively high level of gene flow among populations and thus prevention of gene differentiation among populations. As shown in online Supplementary Table S2, values of F IT and F IS were positive for most loci, thus indicating a deficiency of heterozygosity at these loci.
The among-population P-value returned by the AMOVA was <0.001, which indicated the difference was extremely significant (online Supplementary Table S3). Most genetic diversity was within-population, accounting for 94.85% of the total variance, with only 5.15% existing among populations. The results of the AMOVA regarding population differentiation were consistent with the above-estimated F-statistics.
Population structure and cluster analysis
STRUCTURE program was used to analyse the genetic structure of the germplasm accessions. As showed in Fig. 2(a), ΔK parameter reached the sharp peak when K = 2, suggesting that all accessions were divided into two genetic clusters. In Fig. 2(b), showing from the results of the STRUCTURE analysis, each accession was represented by a vertical histogram with two colours segments that represent the individual's membership fraction in two clusters. The results indicated that all accessions contained a certain membership fraction in two clusters and no obvious differentiation among the 14 populations.
Fig. 2. Prediction of sub-populations (K = 2) for collection of 1122 iron walnut genotypes using Evanno method in (a) STRUCTURE analysis and (b) STRUCTURE plot depicting sub-populations. Note: In (b), each individual is represented by a vertical line and populations are separated by a vertical black line. Different colours in the same line indicate the individual's estimated membership percentage in K clusters.
To reveal genetic relationships between populations and further clarify among-population genetic differentiation, Nei's unbiased genetic identity coefficients were analysed (Table 3). Genetic identities varied from 0.85 to 0.98, with an average of 0.94, which corresponded to a relatively high genetic similarity among populations. However, populations of J. sigillata germplasm in Yunnan had a relatively low level of among-population genetic differentiation. The highest identity coefficient was between populations DL and PE (genetic identity = 0.98), indicating their close genetic relationship, while the lowest identity coefficient was that between populations DQ and LC (genetic identity = 0.85). A UPGMA dendrogram based on genetic distance, constructed to further illustrate population relationships, clustered the 14 populations into two major groups under the genetic distance of 0.075 (Fig. 3). The first group (I) contained three populations (ZT, NJ and DQ) distributed mainly in northwestern and southeastern Yunnan. The second group (II) included other 11 populations. The group III was divided into two sub-groups under the genetic distance of 0.065, LC as one sub-group, and the second consisting of other ten populations.
Fig. 3. UPGMA dendrogram for 14 iron walnut populations in Yunnan.
Table 3. Nei's unbiased genetic identity (above diagonal) and genetic distance (below diagonal) among 14 iron walnut populations
Mantel testing of correlations between pairwise linearized genetic differentiation values and the natural logarithm of between-site geographic distances revealed that geographic distance was significantly correlated with genetic distance at P < 0.05. We found a significant correlation between geographic distance and genetic distance (r = 0.33, P = 0.04), which implied that the geographic distance between populations affects population differentiation to some extent.
Discussion
Shannon's information index (I) and expected heterozygosity (H e) were used to evaluate genetic diversity (Shannon and Weave, Reference Shannon and Weave1959; Nei, Reference Nei1973). I can effectively reflect population genetic diversity (Huang, Reference Huang2013), whereas H e can be applied not only to evaluate genetic diversity, but also to indicate the richness and evenness of an allele in populations (Wang et al., Reference Wang, Hao, Wang and Pei2007). As it is less influenced by sample size, H e is more suitable for estimating population genetic diversity, with a higher H e corresponding to richer genetic diversity (Frankham et al., Reference Frankham, Ballou and Briscoe2002). In the present study, species-level H e and I were 0.578 and 1.190, respectively. Compared with other Juglans species assessed using the same molecular markers, the parameters estimated in this study were higher than those of J. regia and cultivars in Xinjiang (H e = 0.282, I = 0.4225; Liu et al., Reference Liu, Chen, Zhang, Chen, He, Zhang and Zhang2008) and J. regia and J. sigillata in Shanxi, Sichuan, Chongqing and Yunnan (H e = 0.525; Wang et al., Reference Wang, Pei, Gu and Wang2008), similar to those of J. regia and J. sigillata in Xizang (H e = 0.586; Wang et al., Reference Wang, Pan, Ma, Zhang and Pei2015), and lower than those of J. regia and J. sigillata in Shanxi, Henan, Sichuan and Yunnan (H e = 0.617; Wang et al., Reference Wang, Hao, Wang and Pei2007), J. regia in the Daba mountains of Sichuan (H e = 0.657, I = 1.324; Xu et al., Reference Xu, Han, Wang, Chen, Ma and Pei2016), J. nigra in the central hardwood region of the United States (H e = 0.793; Victory et al., Reference Victory, Glaubitz, Rhodes and Woeste2006) and J. mandshurica in central and southwestern China (H e = 0.806; Wang et al., Reference Wang, Pei, Gu and Wang2008). PIC is another important index of genetic diversity. According to Botstein et al. (Reference Botstein, White, Skolnick and Davis1980), a locus is highly polymorphic if PIC > 0.5 and moderately polymorphic if 0.25 < PIC < 0.5, with a PIC < 0.25 corresponding to low polymorphism. In the present study, the average PIC of 20 loci was 0.570, with 13 loci having PICs > 0.5. The average PIC of the 14 populations was 0.5265. The PICs of 12 populations were near or >0.5; the exceptions were LC (0.4104) and PE (0.4576). AR is a better measure of diversity and informative when the population size vary significantly. The mean AR for 14 populations was 3.93, with the highest value of 4.62 (DQ) and the lowest of 3.40 (LC). According to our comprehensive analysis, the provincial-wide genetic diversity of J. sigillata germplasm resources in Yunnan is moderately high, which implies that great potential exists for their further exploitation and development.
Internal factors (e.g. breeding systems, genetic drift, gene mutation and gene flow), as well as external ones (e.g. population isolation and habitat fragmentation resulting from environmental changes and human activities), will interactively influence species genetic diversity. Hamrick and Loveless (Reference Hamrick, Loveless, Bock and Linhart1989a, Reference Hamrick and Loveless1992) have stated that the genetic diversity level of a species is related to its life history characteristics and ecological traits. Rare and endemic species and species of narrow distribution have low levels of genetic diversity, whereas those characterized by a broad distribution, outcrossing and animal-dispersed seeds should possess high genetic diversity (Hamrick and Godt, Reference Hamrick and Godt1996). Although J. sigillata is only distributed in southwestern China (mainly in Yunnan, Guizhou, western Sichuan and southern Tibet) (Wang et al., Reference Wang, Yan, Zhang and Pei2009), this distribution area includes varied ecological and geographic environments with different climatic features, a large range of growing elevations (from 800 to 3200 m) and complicated topographic and geomorphic conditions. The ability to adapt to various environments results in relatively high genetic diversity (Wen et al., Reference Wen, Han and Wu2010). Another key factor leading to relatively high genetic diversity is breeding system. Iron walnut is wind-pollinated, a situation in which pollen is spread more easily compared with insect pollination. Furthermore, walnut is a dichogamous, monoecious plant with unisexual flowers, which promotes outcrossing and discourages selfing, thereby helping to maintain relatively high levels of genetic diversity. Finally, two other characteristics, overlapping generations and a long reproductive life span, should contribute to the maintenance of relatively high genetic diversity.
With respect to spatial distribution, the genetic diversities of the 14 populations were different. The DQ population had the highest level of genetic diversity (AR = 4.62; H e = 0.648; I = 1.402), whereas that of LC was the lowest (AR = 3.40; H e = 0.489; I = 0.838). When arranged from highest to lowest genetic diversity based on AR, H e and I, the relative order of populations was as follows: DQ, KM, NJ, ZT, QJ, HH, YX, WS, DL, LJ, BS, CX, PE and LC. This order may be mostly due to geographic location, germplasm type, population size and the influence of human activities. DQ, NJ and ZT are located in northern and northwestern Yunnan, a region characterized by complicated geographic and climatic conditions, and, most importantly, they adjoin Tibet, Guizhou and Sichuan, where J. sigillata is reportedly native (Gunn et al., Reference Gunn, Aradhya, Salick, Miller, Yang, Liu and Hai2010). In addition, these three populations include a greater number of naturally growing trees in high mountains and valleys that are less influenced from human activities. Kunming, the capital of Yunnan, had a higher level of genetic diversity; this is because it serves as a collection centre, with germplasm from throughout the province concentrated to some extent. The low genetic diversity of LC may be a consequence of small sampling size.
Private alleles, which are generally low-frequency alleles often reflects the selective advantage of populations. Thus, estimating PAR would be useful to reveal the genetic distinctiveness of populations. The mean PAR of 14 populations was 0.30, and ZT had the highest value of 0.36, then DQ (0.34) and BS (0.30). CX showed the lowest PAR of 0.11. That implied that ZT, DQ and BS contained more private alleles and should take more attention in the purposed genetic selection.
F ST is the main index reflecting population differentiation (Templeton, Reference Templeton2010). In the present study, the F ST of the 14 populations was 0.05 (online Supplementary Table S2), thus indicating that only 5% of the total variation existed among populations. This finding was confirmed by the results of AMOVA showing that the genetic variation in walnut germplasm in the area was mainly inter-populational (within-population) and also confirmed by STRUCTURE analysis showing that high genetic diversity among individuals not among populations. The genetic differentiation uncovered in this study was relatively low compared with that found in other studies of related walnut populations, such as J. nigra (F ST = 0.017, Victory et al., Reference Victory, Glaubitz, Rhodes and Woeste2006), J. regia (F ST = 0.103) and J. sigillata in Tibet (F ST = 0.111; Wang et al., Reference Wang, Pan, Ma, Zhang and Pei2015), eight J. regia populations in China (F ST = 0.196; Wang et al., Reference Wang, Hao, Wang and Pei2007) and eight populations in northern China (G ST = 0.4934; Ji et al., Reference Ji, Wang, Wu, Wu, Yang and Wang2014). Evaluation of gene flow (N m), a very important factor influencing within and among-population genetic variation (Slatkin and Barton, Reference Slatkin and Barton1989), led to the same conclusion. Generally speaking, species with a high N m have low among-population genetic variation. When N m > 1, according to Hamrick et al. (Reference Hamrick, Godt, Sherman-Broyles, Hoch and Stephnon1995), gene flow can overcome genetic drift and prevent differentiation among populations. Conversely, when N m < 1 genetic drift would lead to the among-population differentiation. The average N m in this study was 6.70, which was >1, thus indicating the existence of relatively high gene flow among populations. Also, that would prevent population genetic differentiation. This result is consistent with the findings of Chen et al. (Reference Chen, Hu, Zhang and Li2008), who investigated the genetic diversity of walnut germplasm in Sichuan Province.
Bussell (Reference Bussell1999) summarized F ST values of 35 species and concluded that the genetic differentiation of inbreeding species is obviously higher than that of outbreeding ones. Hamrick et al. (Reference Hamrick, Godt and Brown1989b) statistically analysed numerous seed plants and found that the genetic differentiation of long-lived, wind-pollinated, and outcrossing plants was generally the lowest. In our study, the relatively low genetic differentiation of J. sigillata in Yunnan can be primarily attributed to the long lifespan and wind-pollination and outcrossing traits of this species. Germplasm exchange and seed spread is another important reason for the low differentiation. Yunnan is a mountainous province where traditional agriculture is practiced. Iron walnut, an economically important local crop, has been grown and cultivated for hundreds of years in this area because of its economic value and cultural appeal. During the course of walnut cultivation, human-mediated selection, germplasm exchange and seed spread have frequently occurred. Given these facts, relatively strong human interference may be another primary reason for the low genetic differentiation.
On the basis of genetic similarity, 14 populations were divided into two major groups under the genetic distance of 0.075 and genetic distance was found to be significantly correlated with geographic distance. In other words, most populations that were geographically close to one another generally clustered in the same group, such as DQ and NJ (northwestern Yunnan) in one group, and CX, DL and PE (central Yunnan) in another. Geographic location, environment, resource state and genetic diversity were related to group compositions. For example, members of the first group, consisting of DQ, NJ and ZT, were collected from the border of the province, which is characterized by complex landforms of high mountains, valleys and rivers and various climates. Given the underdeveloped status of this area, samples obtained for our investigation were from natural growing populations rather than products of human-mediated selection. The three populations constituting the first group also had the highest genetic diversities in 14 populations. In contrast, the remaining nine populations (except for LC and KM) comprising the second group were located in the provincial interior, which is characterized by flatter land and a better-developed economy. Although some samples in this group were also natural growing in villages, they might be more or less interfered by human activities, e.g. they might be planted by previous generations of hundreds of years ago, which would be expected to have low levels of genetic diversity and higher genetic identities. The LC population, which was adjacent to PE, CX and DL, had the smallest number of samples and the lowest genetic diversity, but the genetic difference of the samples in this group were relatively high. More samples should be taken from LC for further study. The KM population represented resources gathered from throughout the province; it thus displayed high levels of genetic diversity and constituted its own sub-group. When taking 1122 individuals as a total or with in population, the average Nei’ genetic distance between 1122 genotypes was 0.5578, the minimum and maximum values were 0.0149 and 0.9984, respectively.
Improvement programmes of iron walnut have been carried out through natural selection and crossbreeding during the past 20 years, and superior cultivars and strains have been officially registered (Fang et al., Reference Fang, Yang, Huang, Zhang, Fan and Qi1998; Fan et al., Reference Fan, Xi, Fang, Zhang and Zhao2005; Zhao et al., Reference Zhao, Fang, Fan, Xi and Zhang2007). Recently, phenotypic traits and diversity of nuts, genetic diversity of resources in some area of Yunnan have been evaluated (Xiao et al., Reference Xiao, Chen, Ning, Wu and He2018; Chen et al., Reference Chen, Wu and Xiao2019). However, the present research is the first time comprehensive characterization of germplasm in Yunnan has been carried out, which may help future genetic improvement.
In conclusion, genetic diversity is the foundation of germplasm exploitation, conservation and utilization. In the present study, a moderately high level of genetic diversity in J. sigillata germplasm in Yunnan was uncovered which suggests that great potential exists for the further exploitation and development of these germplasm resources, especially those from populations containing richer genetic diversity, such as DQ, NJ and ZT. Further systematic study needs to be carried out on these resources, and conservation measures, such as in situ conservation or establishment of a resource collection nursery, are even required. Given that the genetic variation in the iron walnut germplasm in the study area is mainly within populations, future breeding and conservation efforts should focus on the selection of individuals and families within populations, especially those within DQ, ZT and BS populations of high PAR, in addition to the selection of populations with high genetic diversity.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262120000441.
Acknowledgements
The authors thank the staff members of the local Forestry Bureau and Forest Research Institutes of various prefectures and cities for providing information on walnut distribution and assisting with the collection of experimental materials. This work was financially supported by the National Natural Science Foundation of China (31660214), the Forestry Innovation Project in Yunnan ([2014]CX01) and the Yunnan Provincial Science and Technology Major Project (2018ZG002-1).
