Population Genetic Analysis of Pinus Koraiensis in China Inferred Using EST-SSR Markers


 Background Pinus koraiensis (commonly known as Korean pine), is a well-known conifer species in China with high economic, ornamental and ecological values. More than 50% of the P. koraiensis forests in the world are distributed in Northeast China, a region with abundant germplasm resources. However, these natural P. koraiensis populations are in danger of genetic erosion caused by continuous climate changes and frequent human activity. Little work has been conducted on the population genetic structure and genetic differentiation of P. koraiensis in China. Here, representative individuals from 16 natural P. koraiensis populations were sampled and genotyped, and polymorphic expressed sequence tag-simple sequence repeat (EST-SSR) markers were used to comprehensively evaluate genetic diversity, population structure and differentiation of P. koraiensis populations in China.ResultsA total of 480 samples from 16 populations were collected in the natural distribution area of P. koraiensis. Analysis of molecular variance (AMOVA) of the EST-SSR marker data showed that 33% of the total genetic variation was among populations and 67% was within populations. A high level of genetic diversity was found across P. koraiensis populations (average Na=10.33, Ne=2.514, He=0.521), and the highest levels of genetic diversity were found in Heihe (He=0.449), Zhanhe (He=0.413), Liangshui (He=0.370) and Tieli (He=0.414) populations. Moreover, pairwise Fst values reveled significant genetic differentiation among populations (mean Fst=0.177). Structure and Neighbor-joining (NJ) tree analyses and principal component analysis (PCA) revealed two genetic clusters: cluster 1 from Xiaoxinganling Mountains and cluster 2 from Changbaishan Mountains, which were consistent with the geographical distributions of the natural populations. ConclusionsThe findings provide new genetic information for future genome-wide association studies (GWAS), marker-assisted selection (MAS) and genomic selection (GS) in natural P. koraiensis breeding programs and can aid the development of conservation strategies for this valuable conifer species.

Unlike morphological and biochemical markers, many DNA molecular markers are codominant and highly polymorphic, and many have been identi ed in the genome and transcriptome [23][24][25]. Because of their low cost, easy detection by Polymerase Chain Reaction (PCR), high polymorphism, and codominance, simple sequence repeats (SSRs) are considered powerful and advantageous molecular tools for genetic diversity analysis, genome-wide association analysis, core collections and genetic linkage map construction in many plants and animals [26][27][28]. Furthermore, multiple EST-SSR markers can easily be developed from microsatellite loci of public transcriptome data. At present, there are few reports of analyses of genetic diversity in P. koraiensis based on DNA molecular markers. Studies to date have employed random ampli ed polymorphic DNA (RAPD) analysis [29], single primer ampli cation reaction (SPAR) [30], intersimple sequence repeat (ISSR) analysis [31][32] and expressed sequence tag-simple sequence repeat (EST-SSR) analysis [33]. All of these studies have identi ed high levels of genetic diversity in P. koraiensis, with the greatest levels of genetic differentiation occurring within populations. However, a systematic and comprehensive population genetic study is lacking because of the limitations of the species' geographical distribution, the number of available molecular markers and population size. Therefore, widespread germplasm collection and abundant polymorphic markers are necessary to study the genetic relationships and diversity of P. koraiensis, which can be facilitated by abundant molecular markers developed from high-throughput sequencing, especially for some endangered and non-model plant species.
In this study, we collected germplasm resources from 480 individuals of 16 natural populations of P. koraiensis within the species main distribution area in northeastern China, and we analyzed genetic diversity using 15 EST-SSRs. We aimed to (1) investigate genetic variation using polymorphic EST-SSRs, (2) evaluate the genetic diversity and structure of natural populations, (3) conduct a comprehensive, range-wide genetic diversity study of P. koraiensis in China, and (4) propose a protection strategy. The results provide insight into the conservation of this species and lay a foundation for further studies of marker-assisted selection (MAS) and genomic selection (GS) in P. koraiensis for genetic improvement.

Genetic diversity at different loci among populations
The genetic diversity analysis was performed on 480 individuals from 16 natural P. koraiensis populations using 15 EST-SSRs markers ( Table 2). The allele size ranged from 151 bp at locus NEPK-65 to 301 bp at loci NEPK-168 and NEPK-184. In total, 155 alleles across all 15 loci were detected in the sampled individuals; the number of alleles per locus ranged from 4 (NEPK-67) to 21 (NEPK-145), with a mean value of 10.33. We identi ed 58 alleles as private alleles, accounting for 37.42% of the alleles. The effective number of alleles (Ne) ranged from 1.170 at locus NEPK-40 to 6.605 at locus NEPK-145, with an average of 2.514 per locus. Observed and expected heterozygosity ranged from 0.008 to 0.984 and from 0.145 to 0.849, respectively, with mean values of 0.374 and 0.521, respectively. Polymorphism information content (PIC) varied from 0.142 (NEPK-40) to 0.833 (NEPK-145), with a mean value of 0.461. Four loci exhibited high polymorphism (PIC > 0.5), and 8 exhibited moderate polymorphism (0.2 < PIC < 0.5). In addition, across the 480 samples, all of the loci conformed to Hardy-Weinberg equilibrium. F-statistics were calculated to detect genetic subdivision and revealed moderate inbreeding, and the mean value of Fst was 0.347, indicating moderate genetic variation. The gene ow (Nm) value ranged from 0.080 to 17.691 among the populations, with an average of 2.667.     Table 4. The AMOVA results indicated that 67% of the total genetic variation existed within populations, indicating high genetic diversity within populations. AMOVA of the two genetic clusters identi ed by the STRUCTURE analysis indicated that 63.79% of the total variation was attributable to differences within populations, and the overall Fst was 0.362 (Fst > 0.25), indicating high genetic differentiation between the 2 clusters. In addition, the AMOVA of two groups classi ed according to geographical location indicated low genetic variation among populations within each group (2.77%). All of these results indicated high genetic differentiation within populations and groups. Note: a The analysis included all collected populations as one hierarchical group. b The analysis included two geographical groups (G1 and G2). c The analysis included two genetic clusters (Clusters 1 and Clusters 2).
The Nei's genetic distance and pairwise Fst values are shown in Table 5. The Fst-value was considered the main genetic parameter for evaluating genetic differentiation among populations. In this study, the pairwise Fst values ranged from 0.014 to 0.348, and most of the P. koraiensis population pairs exhibited high values (Fst > 0.15), indicating high levels of genetic diversity. The greatest level of differentiation was observed between populations Helong and Liangshui, and the lowest was observed between Jiaohe and Hulin. The highest genetic distance was observed between populations Helong and Liangshui (0.813), consistent with the pairwise Fst values and indicating pronounced differentiation between these two populations. The relative migration network among the sixteen P. koraiensis populations was constructed using relative migration rate with the divMigrate function in R software. Analysis of gene ow between populations suggested a biased geographic distribution, and gene ow was not uniform among all populations (Fig. 2). A high degree of gene ow was observed among three populations located close to each other (Muleng, Maoershan and Fangzheng), consistent with the principal coordinate analysis and dendrogram analysis. In addition, one genetically isolated population (Boli) displayed high levels of gene ow with the three nearby populations Muleng, Maoershan and Fangzheng. Moreover, a moderate level of gene ow was found among three admixed populations, and two genetically distinct populations (Zhanhe and Wangqing) exhibited distant segregation from the other populations. Analysis of population structure of 16 natural P. koraiensis populations was performed based on a Bayesian approach using STRUCTURE software. The number of clusters within the range of 1 to 10 was evaluated for 10 repetitions in each run. In the structure plot, the maximum delta K value appeared at K = 2, with a noticeable peak apparent at this value. Thus, this value was considered the optimal genetic cluster number for all of the EST-SSR markers (Fig. 3b, c). To further analyze clustering patterns, principal component analysis (PCA) based on the pairwise genetic distance matrix of fteen EST-SSRs was performed; and the results are shown in Fig. 5a. The 480 individuals from the sixteen populations were roughly divided into two clusters according to the rst two axes in the PCA plot. Principal axes 1 and 2 accounted for 22.99% and 12.46%, respectively, of the total genetic variation among the individuals, together accounting for 35.45% of the total genetic variation (Fig. 4a). Five populations (Heihe, Liangshui, Zhanhe, Tieli and Hegang) were grouped into cluster 1, and the remaining populations were grouped into cluster 2. The same clustering was obtained in the STRUCTURE analysis using the same dataset, indicating marked genetic differentiation. Furthermore, the Neighbor-joining (NJ) dendrogram based on Nei's genetic distance clustered the 480 P. koraiensis individuals from the 16 populations into two clusters, consistent with the above results (Fig. 4b, Fig. 5).
In this study, the geographic distance and genetic distance values ranged from 37.72 km to 825.45 km and from 0.02 to 0.83, respectively. We examined whether the genetic distance estimated based on molecular markers may be related to the distribution of the species under study and the geographic distance between individuals or populations. To investigate the correlations between genetic distance and geographic distance, the Mantel test was carried out. The results showed that genetic distance was not signi cantly correlated with the geographic distance among P. koraiensis populations (p = 0.26, R 2 = 0.01), indicating a lack of association between geographical distance and the genetic differentiation of P. koraiensis (Fig. 6). Liangzihe and Hegang exhibited the lowest geographic distance and were not grouped in the same cluster. Therefore, there was not apparent isolation by genetic and geographical distance among the sampled populations.

Discussion
To understand the genetic differentiation of forest tree populations and contribute to the development of effective breeding strategies, comprehensive evaluations of natural germplasm resources of individual species are essential. Such evaluations can accelerate breeding-strategy and industrial development [34][35]. Naturally, P. koraiensis mainly grows in the cold temperate zone, especially in northeast China, and natural forests of this species have been shown to be sensitive to climate factors. Thus, to conserve genetic resources of this species, it is important to obtained data on its genetic diversity and population structure. In present study, we conducted a population genetic analysis using codominant molecular markers, representing the rst such analysis in P. koraiensis. The results can help guide the genetic improvement and resource conservation of this important conifer species.
Genetic diversity Genetic diversity has been increasingly evaluated in species lacking a reference genome, including some conifers [36], endemic species [37] and endangered plants [38]. Studies of genetic diversity can provide insight into speciation and genetic variation within and among populations and can aid the development of conservation strategies. P. koraiensis is a valued woody plant that provides high-quality timber and pine nuts and possesses high economic, ecological and ornamental value. However, transcriptome data and molecular markers remain lacking for P. koraiensis. The available genetic data provide few markers suitable for the study of population genetics in this species. Evaluating the germplasm resources of this species represents the rst step towards understand the genetics of natural P. koraiensis populations.
In this study, we evaluated the genetic diversity of natural P. koraiensis resources in China using 15 polymorphic EST-SSR markers. High levels of genetic diversity in natural P. koraiensis populations were detected, with mean values of 10.33 and 0.521 for Na and He, respectively. High genetic diversity was observed in the Heihe population in the northern Xiaoxinganling Mountains, possibly due to less human disturbance in this region than in other areas. According to a previous study, a PIC value equal to or more than 0.5 indicates high genetic information for genetic markers. In the present study, the PIC values obtained for the multiallelic EST-SSR markers ranged from 0.142 to 0.833, with a mean value of 0.461, indicating high level of genetic information among the 480 P. koraiensis individuals from the 16 natural populations. The genetic diversity of P. koraiensis obtained in the present study is higher than that reported in Pinus bungeana (Na = 3.70, He = 0.36) [39], Pinus dabeshanensis (He = 0.36) [40] and Pinus yunnanensis (Na = 4.10, He = 0.43) [41] but lower than that reported in Pinus tabulaeformis (Na = 6.52, He = 0.68) [42].
The genetic diversity of a species may vary with characteristics such as adaptability, pollination mechanism and population size [43][44][45]. The observed genetic diversity in the present study might be attributed to the genetic background, life history, and population dynamics of P. koraiensis. Previous reports have found that P. koraiensis has a large population size, long life cycle, strong adaptability, long pollination distance and large genome size, and it has a complex genetic background, which allows it to generate high genetic diversity [46][47][48]. Over a long period of natural selection and succession, the natural distribution range and population number of P. koraiensis have greatly decreased, while those of man-made forests have increased in recent years. Furthermore, under the increasing in uences of human activity and climate change, seed yield, plant resistance to stresses and other traits may be reduced and may hinder the maintenance of high levels of genetic diversity. Natural selection under the changing environment conditions is likely to lead to differences in genotype frequency among populations. In addition, previous studies have suggested that evaluations of genetic diversity are limited by low number of populations and molecular markers [49][50].
Studies on genetic diversity of natural P. koraiensis populations have been conducted using a variety of molecular-marker techniques. Kim et al. [51] analyzed allozyme loci variation and found a moderate level of genetic diversity among natural P. koraiensis populations in Korea. The genetic diversity of natural P. koraiensis populations detected by allozyme markers (He = 0.18) in Russia [52] was much lower than that identi ed using EST-SSRs markers in this study. These ndings indicate that P. koraiensis maintains high genetic diversity worldwide. The level of genetic diversity detected in this study is similar to that detected based on nine EST-SSR markers in seven natural populations of P. koraiensis in northeast China (He = 0.610) [53]. Xiaoxinganling Mountain of China was considered the distribution center for P. koraiensis, possessing abundant germplasm resources and ancient founding stocks and maintaining considerable numbers of individuals. In addition, the genetic diversity of P. koraiensis populations from Xiaoxinganling Mountains was higher than that of the Changbaishan Mountains populations, with high expected heterozygosity and abundant private alleles found for the former populations (Fig. 5). All of these results indicate that Xiaoxinganling Mountains may be the center of genetic diversity of P. koraiensis in China.

Population genetic differentiation
Detection of genetic differentiation is a key process in the genetic improvement of forest trees. Regarding the estimation of genetic differentiation, previous studies have considered an Fst value higher than 0.15 but lower than 0.25 to indicate signi cant divergence [54][55][56]. In the present study, the genetic differentiation assessed by EST-SSRs among P. koraiensis populations ranged from 0.014 to 0.348, with a mean value of 0.177, indicating signi cant differentiation among populations in China. However, previous studies reported low genetic differentiation among populations assessed by allozyme loci variation in Korea Far East (Fst = 0.06) [51] and Russia (Fst = 0.015) [52] and by EST-SSRs in China (Fst = 0.02) [53]. In addition, Kim et al. [29] studied the genetic variation of P. koraiensis in Korea, Russia and China using allozymes and RAPDs and detected small differences among the three regions. Different degrees of genetic differentiation were observed in natural P. koraiensis populations in these countries, with low Fst values. The main reason for these differences is that only limited numbers of natural populations and molecular markers were analyzed.
The genetic differentiation index (Fst) is correlated with gene ow (Nm). Generally, the greater the degree of differentiation, the weaker the gene ow, i.e., a lower gene migration rate among populations [57][58][59]. Thus, the genetic differentiation is facilitating gene ow. Natural P. koraiensis forest originated in Siberia in Northeast Asia and has undergone regeneration, succession and migration over millions of years [60][61]. After the Quaternary glaciation, many species died out, but the P. koraiensis forests persisted into the present and underwent a range of changes and varying degrees of differentiation. In natural P. koraiensis populations, low levels of genetic differentiation have been observed in Korea [29], whereas high genetic differentiation has occurred in Northeast China, which may have contributed to the rich P. koraiensis germplasm resources (representing approximately 60% of the world's total) and broad distribution area (more than 3000 hectares) in this country.
The mean He (0.521) across all loci was greater than Ho (0.374), indicating a high heterozygosity among the sampled populations of P. koraiensis. This high heterozygosity is attributable to the fact that Pinus species exhibit cross-pollination and wind pollination. Furthermore, the AMOVA suggested that most of the genetic variation (more than 60%) in P. koraiensis exists within populations, with a small proportion occurring among populations; suggesting that genetic differentiation between populations exists.
Population structure and gene ow Analyses of population structure can provide insight into population size, breeding system, extent of isolation and population migration or gene ow [62][63]. Furthermore, such analyses can help reveal the relationships between genetic variation and environmental stresses and enhance our understanding of evolution. Evaluating population structure is a key component of genome-wide association analysis (GWAS) and marker-assisted selection (MAS) [64]. P. koraiensis is mainly distributed in Xiaoxinganling Mountains and Changbaishan Mountains in northeast China, areas with a humid climate. Due to the environmental conditions, the germplasm resources of P. koraiensis from different locations display high phenotypic and genetic variation. The STRUCTURE analysis of population structure identi ed two groups (optimal K = 2) from the sixteen natural populations, with ve populations in one group and the remainder in another group. Similar results were obtained in the PCA and dendrogram (neighbor-joining tree) analysis, indicating genetic differentiation of P. koraiensis in China.
Interestingly, individuals from Xiaoxinganling Mountains were clustered into one group, occupying a northern area, which makes them more like an ancestral group. Furthermore, the samples from Changbaishan Mountains and adjacent ridge region were clustered into the other group; the populations corresponding to these samples are distributed in a southern area and exhibit different degrees of genetic differentiation and gene ow. However, some of the individuals from Xiaoxinganling Mountains were clustered into cluster 2, although the majority were clustered into cluster 1 (Fig. 5a, Fig. 5b). The main potential reasons for this nding are as follows: (1) these two mountain regions are close to each other, and some hybridization events may occur; (2) for populations separated by a short spatial distance, the probability of gene ow is high, which will affect population genetic structure; and (3) pollen and seed dispersal occurs over long distances in this species, which promote gene ow. The genetic structure of the natural P. koraiensis populations in China determined in this study is consistent with the current geographical distribution of these populations. Furthermore, the ndings are consistent with previous studies showing that populations in similar geographical locations or environments tend to cluster into the same group [65][66].
Gene ow among populations is closely related to geographical distance and effective population size and can generate new genetic combinations, potentially enhancing species resilience and persistence [67][68][69]. In plants, migration or gene ow is achieved via seeds, pollen and other propagules, and in uences the genetic diversity and differentiation among independent evolutionary units [70][71]. We found that two genetically distinct populations (Zhanhe and Wangqing) exhibited segregation from other populations, which may be related to their geographic distance from those other populations (approximately 565 km), limiting the level of gene ow between them. These independent units play an important role in maintaining the genetic diversity of this species. This interpretation is consistent with previous studies demonstrating that isolated populations of plants with long-distance pollination may have higher levels of genetic diversity than large contiguous populations [72][73][74].
Moreover, high levels of gene ow were found among Helong, Maoershan and Fangzheng populations. High levels of gene ow can reduce the effects of arti cial selection or genetic drift and promote the maintenance of genetic information. Similar results were obtained for Oryza sativa accessions [61]. Extensive gene ow can alter the gene frequencies in populations to affect genetic diversity and structure. In our study, although a strong correlation between gene ow and geographic distance between populations was observed, some degree gene ow was also evident between geographically distant populations. In addition, geographic distance was not correlated with genetic distance in the natural P. koraiensis populations in this study, suggesting that geographic distribution may not be a determinant factor for the genetic structure of populations.

Suggestions for conservation
Evaluations of germplasm resources are needed to maintain abundant genetic variation and high levels of genetic diversity of some species of interest and establish sound conservation strategies. Our population genetic analysis revealed that the populations distributed in the Xiaoxinganling Mountains area (Zhanhe, Heihe, Liangshui, Tieli and Hegang) exhibit high levels of genetic diversity and moderate levels of gene ow (Fig. 4). These populations represent the core populations and have stronger environmental adaptability and evolutionary potential than the other populations, and they can be considered independent genetic units. Hence, measures such as in situ conservation should be implemented for conserving natural P. koraiensis resources.
In addition, the marginal populations represent special germplasm resources; and they are characterized by low genetic diversity but have high levels of genetic differentiation relative to the other natural populations. Habitat fragmentation can reduce gene ow among populations, leading to a loss of genetic diversity. In this study, the Helong population, which occurs in a marginal area, should be targeted for conservation measures, such as ex situ measures. In addition, the greatest level of differentiation observed between populations was between Helong and Liangshui, indicating that these populations can be considered independent units. Therefore, regulations and management strategies must be established to protect the natural habitat of this species and prohibit harvest. More importantly, a national-level core germplasm resources library of P. koraiensis should be established by the government, with the objectives of maintaining genetic variation, improving plant adaptability to environmental changes, and developing new breeding materials. Under these measures, the existing natural P. koraiensis populations in China can be protected and be better used as a source of material for genetic improvement.

Conclusion
This study investigated the genetic diversity and population structure of natural P. koraiensis populations in northeast China, which is the rst comprehensive report of the genetic diversity of natural P. koraiensis populations in China. We found that the existing P. koraiensis populations in China maintain high levels of genetic diversity, which provide a foundation for germplasm innovation and genetic improvement of P. koraiensis. The population genetic analysis in this study identi ed two independent genetic units (Liangshui and Helong) that exhibit high degrees of genetic differentiation. The populations distributed in the Xiaoxinganling Mountains area are highly genetically diverse and may represent the central population of natural P. koraiensis in China. Our ndings provide genetic information useful for future genome-wide association studies (GWAS) and marker-assisted selection (MAS) and genomic selection (GS) studies. Furthermore, the genetic structure of P. koraiensis populations identi ed in this study is consistent with the geographical distribution of these populations in China. These results have signi cant implication for the protection of natural P. koraiensis germplasm resources in China. Thus, we suggest that appropriate in situ and ex situ conservation measures should be taken to preserve the germplasm resources.

Collection of plant materials and genomic DNA extraction
The plant materials used in this study were obtained from the wild and permission was obtained to collect samples. The collection of plant materials also complied with institutional, national, or international guidelines. The formal identi cation of the samples used in this study was performed by Xi-Yang Zhao. Voucher specimens were deposited in the herbarium of Northeast Forestry University. We investigated 16 populations of P. koraiensis from Jilin Province (J) and Heilongjiang Province (H) in the current study. A total of 480 samples were collected throughout the natural distribution areas in northeastern China (Table 1 The fresh needle samples of P. koraiensis with no signs of pests or disease were immediately frozen in liquid nitrogen and stored at -80 °C for subsequent genomic DNA extraction and PCR ampli cation. In addition, nucleic acids were extracted from needles using the improved cetyltrimethyl Ammonium Bromide (CTAB) method described by Li et al. [75]. WDNA quality and concentration were evaluated using 1.0% agarose gel electrophoresis and the K5500 Plus microspectrophotometer (KAIAO Technology Development Co., Ltd., Beijing, China), respectively.

PCR ampli cation and SSR analysis
To detect polymorphisms in the 16 sampled P. koraiensis populations, 15 highly polymorphic and reproducible EST-SSR markers of P. koraiensis developed in our laboratory were selected in this study. The primers of P. koraiensis was developed as described by Li et at. [22].

Data analysis
GeneMapper was used to obtain the microsatellite allele data, and the Microsatellite toolkit v 3.1.14 was used to convert the data into the necessary format for analysis. The genetic diversity analysis was conducted using GENALEX software version 6.50 [76] with the following parameters: number of alleles (Na), effective number of alleles (Ne), observed (Ho) and expected (He) heterozygosity, number of rare alleles (NRA), Shannon diversity index (I), Hardy-Weinberg equilibrium (HWE), F-statistics (Fis, Fit and Fst) and Nei's genetic distance. The TBtools software [77] was used to plot the map of heatmap of expected heterozygosity (He). In addition, we calculated the polymorphism information content (PIC) values of each SSR primer using the PICcalc program [78]. Gene ow (Nm) was calculated as Nm = (1-Fst)/4 × Fst and used to measure the degree of gene exchange among or within the 16 populations. ALREQUIN software (version 3.5) [79] was used to analyze the level and sources of molecular genetic variation via AMOVA based on the evolutionary distances among and within the sampled populations and the observed genetic clusters. The total genetic variation was divided into three components: among groups, among populations within groups and within populations.
To evaluate the population genetic structure of P. koraiensis, a Bayesian clustering algorithm was performed in STRUCTURE software (version 2.3) [80] with the following settings: K-values from 1 to 10, with ten runs per K value and a burn-in period and number of Markov chain Monte Carlo (MCMC) reps after burn-in of 100,000 iterations and 100,000, respectively. The optimal K value for the number of populations was based on the delta-K values calculated by the Evanno method [81], using an algorithm of the online tool of STRUCTURE HARVESTER [82]. A clear peak was observed in the plot of delta K. In addition, principal component analysis (PCA) was performed to evaluate the genetic relationships among different populations using GENALEX software version 6.50. Based on the Nei's genetic distance (1983), a Neighbor-joining (NJ) phylogenetic tree of the populations was constructed using PowerMarker software (version 3.25) [83] and annotated and visualized using the online tool interactive Tree Of Life (iTOL) [84]. Geographic distance among populations was calculated as described in Li et al.'s study [75]. Finally, to detect the gene ow among the sixteen populations, a relative migration network was constructed using the 'diveRsity' [85] package of R software (version 3.5.0) [86]. The funding agency has no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

List Of Abbreviations
Authors' contributions: XZ and XL conceived the study, developed the experimental design and provided the suggestions and comments for this manuscript. XL, MZ and YX carried out eldwork. XL and YL performed the data analysis. XL wrote the original draft and all authors contributed in preparing the manuscript in its nal form. MT revised the original manuscript. All authors have read and approved the submission of the manuscript.