The Great Wall of China: a physical barrier to gene flow?


One population from each of six plant species along both sides of the Juyong-guan Great Wall, together with one population from each of five species along both sides of a path on a mountain top near Juyong-guan, were selected to study the effect of the Great Wall as a barrier on genetic differentiation between two subpopulations using RAPD markers. Significant genetic differentiation was found between the subpopulations on both sides of the Great Wall. A wind-pollinated woody species, Ulmus pumila, showed less genetic differentiation than four insect-pollinated species: Prunus armeniaca, Ziziphus jujuba, Vitex negundo, and Heteropappus hispidus. Cleistogenes caespitosa, a wind-pollinated perennial herb, displayed more genetic differentiation between subpopulations than the insect-pollinated species because of its propagation strategy. Although AMOVA analysis showed that subpopulations divided by a mountain path had diverged genetically, the variance component between the subpopulations on both sides of the Great Wall was significantly larger than that between the subpopulations at the control site. Therefore, it is reasonable to deduce that the Juyong-guan Great Wall has served as a physical barrier to gene flow between subpopulations separated for more than 600 years.


Geographic isolation is expected to have significant effects on the genetic structure of populations (Smith, 1999). A population may be physically separated when its original habitat becomes divided by a natural barrier (eg, a river, shoreline, mountain range or glacier) or even an artificial barrier (eg, a man-made canal or highway). These barriers can restrict or prevent gene flow and result in the genetic differentiation of isolated subpopulations (Hartl, 1980; Corre et al, 1997; Bauert et al, 1998; Nesbo et al, 1998). Furthermore, the mating system, pollination biology and life history of a species can also have a critical role in the genetic differentiation of its populations (Turner et al, 1982; Wolff et al, 1997; Gauer and Cavalli-Molina, 2000).

The Great Wall was built by the ancient Chinese for the purpose of defence. The earliest record of the Great Wall dates back to 656 BC, and many Emperors in successive dynasties had the Great Wall renovated or extended. The most recent large-scale renovation was in the Ming Dynasty (1368–1644) and was the biggest renovation project in its history. The most famous sites, such as Bada-ling, Jiayu-guan, Juyong-guan, were built during the Ming Dynasty (Luo and Liu, 1992). As an artificial physical barrier, the Great Wall could be an excellent model for studying its effect on the genetic differentiation of plant populations around it.

In this study, six plant species with different habits, pollination modes or reproduction systems along both sides of the Juyong-guan Great Wall at two sites were collected and analyzed. Five of the same or closely related species along both sides of a mountain path without the Great Wall were taken as a control. RAPD markers were used as molecular markers to reveal the genetic variation, since they have been widely used in the studies of plant genetic diversity and population genetic structure (Dawson et al, 1993; Huff et al, 1993; Martin et al, 1997; Buso et al, 1998; Hsiao and Lee, 1999; Zhao et al, 2000).

Materials and methods

Sampling sites and strategy

The Juyong-guan Great Wall (40°13′20″N, 116°04′72″E) was recorded as built in 1368 (Luo and Liu, 1992), and located approximately 60–70 km north of Beijing. Plant materials were collected in September 1999 and April 2000 at two sites (sites A and B) along both sides of the Juyong-guan Great Wall, and one site (site C) along both sides of a path, as control (Figure 1). Sites A and B were about 2 km apart from the control site and separated by an express way. At site A the wall runs from east to west with north and south aspects and at site B the wall runs from south to north with east and west aspects (Figure 1). The Great Wall is 6 m high and 5.8 m wide (on average) at the sampling sites, and the mountain path is approximately 1.5 m wide with north and south aspects. Sampling within populations was carried out randomly in an area approximately 100–150 m long and 7 m wide along both sides of the Wall and the path. Six species were collected in the Juyong-guan Great Wall area, five of them at site A and one species, Prunus armeniaca, at site B (Table 1a). Heteropappus hispidus, or its related species, were not found in the control area; therefore, only five species were collected along the path and four of them were of the same species as in the Great Wall area (Table 1b). The fifth one, Ulmus macrocarpa, replaced U. pumila which was only found in the Great Wall area. For each species, approximately 40 individuals were collected per population, with approximately 20 individuals on each side of the Great Wall or path, respectively. In total, 11 populations and 416 individual plants were collected (Table 1a, b).

Figure 1

Map of sampling sites.

Table 1 Species and subpopulations studied from the (a) Juyong-guan Great Wall and (b) control site

For the convenience of discussion, the two parts of a population divided by the Great Wall or the path were defined as subpopulations. The abbreviation for each subpopulation was designated as follows, the first capitalized letter is the first letter of the genus name, the following two letters are the first two letters of specific epithet; the fourth letter is either J or C representing the Juyong-guan or control site, respectively; the last letter represents the sampling position. For example, UpuJN means the subpopulation of U. pumila collected on the northern side of the Juyong-guan Great Wall.

DNA preparation and RAPD analysis

Young leaves from an individual plant were collected and stored in silica gel at room temperature. Total DNA was extracted from the dried leaves according to the CTAB method of Qu et al (2000). A total of 100 primers (S41–60, S201–220, S401–420, S481–500, S2121–2140, Sangon Biotech Company) were used for a two-step primer screening. In the first step, one plant sample per population was randomly selected to identify primers that produced at least three clear bands. In the second step, another three individuals from each population were randomly selected for RAPD analysis with primers selected in the first step. Following the two-step primer screening, 10–11 primers with clear and reproducible bands were chosen for each species (Table 2). Amplification was carried out in 25 μl with 10 mM Tris-HCl, 50 mM KCl, 0.1% Triton-100, 2 mM MgCl2, 200 μM each of dATP, dGTP, dCTP and dTTP, 30 ng of primer, 2 units Taq DNA polymerase and 5–40 ng template DNA. The thermal cycles started with five cycles of 94°C for 1 min, 35°C for 2 min, 72°C for 2 min, then 40 cycles of 94°C for 30 s, 35°C for 2 min, 72°C for 2 min and ended with 72°C for 10 min. The amplification products were separated on 1.4% agarose gel in 1 × TAE buffer with 0.5 μg/ml ethidium bromide at 100–120 V for 2 h, and documented using the AmpGene Gel 100 system.

Table 2 Primers, number of total RAPD bands and polymorphic bands detected in subpopulations

Statistical analysis

RAPD bands were scored as present (1) or absent (0) for each DNA sample. The genetic diversity within subpopulations was measured by the percentage of polymorphic bands, which was calculated by dividing the number of polymorphic bands by the total number of bands surveyed for the corresponding subpopulations (Table 2). Both Nei and Li's and Euclidean genetic distances within populations were calculated employing RAPDistance version 1.04 (Armstrong et al, 1994). Based on Nei and Li's genetic distance matrix, a dendrogram of individuals within a population was constructed using the neighbour-joining method. The variance components between subpopulations and among individuals within subpopulations were calculated with WINAMOVA version 1.55 (Excoffier, 1993) based on Euclidean genetic distances. The significance of each variance component was tested by a random permutation test. The significance of the difference in variance component between the Great Wall subpopulations and the control subpopulations was tested using a paired t-test.


Genetic variation and genetic distance among individuals within subpopulation

For each subpopulation, the total number of RAPD bands ranged from 102 to 151, and the number of polymorphic bands ranged from 50 to 101 (Table 2). Vitex negundo on the northern side (VneCN) of the path had the highest percentage of polymorphism (85.5%), and U. pumila in the same location (UpuCN) had the lowest percentage of polymorphism (43.9%).

The results also showed that among 416 individuals studied, 409 had a unique RAPD banding pattern. Two individuals of Ziziphus jujuba in subpopulations ZjuCN and ZjuCS had the same banding pattern as did three individuals of U. macrocarpa in the southern subpopulation (UmaCS) along the path (Figure 2).

Figure 2

RAPD banding patterns of subpopulation UmaSC amplified using primer S419. M: λDNA/EcoRI+HindIII markers. 1–17: individuals from subpopulation.

The result of neighbour-joining analyses on the individuals of each population showed that the individuals from the same subpopulation along the Juyong-guan Great Wall tended to cluster together, while the individuals of two subpopulations from one population along the path were always mixed. A typical example of neighbour-joining results for P. armeniaca is shown in Figure 3.

Figure 3

(a) Neighbour-joining dendrogram of 40 P. armeniaca individuals from the Great Wall site. (b) Neighbour-joining dendrogram of 40 P. armeniaca individuals from the control site.

Genetic differentiation between the two subpopulations within a population

The results of AMOVA analysis indicated that for the total genetic variation of a population, 71.1–97.1% of the variance was distributed among individuals within subpopulations and 2.9–28.9% of variance was distributed between subpopulations (Table 3). In the Great Wall area, the perennial herb, Cleistogenes caespitosa, had the highest variance component within subpopulations (4.7), and U. pumila and P. armeniaca had the lowest variance component within subpopulations (2.3 and 2.3, respectively). The subpopulations of all species were significantly (P<0.001) genetically differentiated. In the control area, the subpopulations of each species were also significantly differentiated (P<0.001 or <0.05). However, when the degree of genetic differentiation between subpopulations of the Juyong-guan Great Wall sites and that of the control site was compared by the paired t-test, the differentiation between subpopulations of the Juyong-guan Great Wall was significantly greater than that of the control site (P<0.05).

Table 3 Analysis of molecular variation (AMOVA) for 11 subpopulations comparison


The RAPD analysis showed that the degree of polymorphism in six populations along both sides of the Great Wall and five populations along the path fits into the range of genetic polymorphism (45–100%) of out-crossing plants (Huff et al, 1993; Liu, 1994). Except for seven individuals in three subpopulations at the control site, all plant samples studied had unique RAPD banding patterns, giving sufficient polymorphic loci for genetic differentiation analysis.

The variance component analysis indicated that over 71% of genetic variation was attributed to the individual differences within subpopulations (Table 3). Although less genetic variation was detected between corresponding subpopulations, the genetic differentiation between the subpopulations separated by the Great Wall was significant (P<0.001; Table 3). The individual dendrograms, based on neighbour-joining analysis, revealed that all populations in the Great Wall area had two main clusters, and that almost all individuals in a subpopulation were clustered together, except for one to four individuals, indicating that the gene flow between subpopulations separated by the Great Wall was disrupted to a certain extent. Ulmus pumila, Z. jujuba, C. caespitosa and H. hispidus have one or two individuals clustered with the individuals from the other subpopulation in both directions, respectively (Figure 4a–d). Vitex negundo has only two southern individuals clustered with the northern subpopulation (Figure 4e). Only one western individual of P. armeniaca was clustered with the eastern subpopulation (Figure 4f). Ulmus pumila is wind-pollinated and has wind-dispersible samara, while Z. jujuba and P. armeniaca are insect-pollinated and have edible fruits (animal- or human-dispersed). The other three plant species have either insect-pollinated flowers/wind-dispersed fruit or wind-pollinated flowers/insect-dispersed fruit. It is not clear, based on these data, whether the limited gene flow between subpopulations is because of wind, animal, or human activities.

Figure 4

Samples collected along two sides of the Juyong-guan Wall: (a) U. pumila, (b) Z. jujuba, (c) C. caespitosa, (d) H. hispidus, (e) V. negundo, and (f) P. armeniaca. The scales of Figure 4b–e are the same as that in Figure 4a.

The fact that the subpopulations of the wind-pollinated species, U. pumila, along the Juyong-guan Great Wall, has less genetic differentiation between subpopulations than four insect-pollinated species ( Table 1 and Table 3) follows the general rule that wind-pollinated plants have less genetic differentiation than insect-pollinated plants (Govindaraju, 1988). Ulmus pumila flowers in early April and sets wind-dispersed samaras in the same month or early May during the season with strong winds in the Beijing area. In the four insect-pollinated species, the woody species P. armeniaca had less genetic variation between subpopulations than the herbaceous H. hispidus. The latter had the smallest population size because the individuals were only distributed in an area of 80 × 1 m2 along the northern side (14 individuals collected) and of 10 × 1 m2 along the southern side (nine individuals collected) of the Great Wall (Figure 4d). In such a small population, genetic drift may play an important role in genetic variation between subpopulations.

Plant population genetic structure is also strongly influenced by life cycle and reproductive strategy (Hamrick et al, 1979; Schaal et al, 1998; Gauer and Cavalli-Molina, 2000). Cleistogenes caespitosa, a perennial herb, is a wind-pollinated species and has a few individuals from both subpopulations clustered together in the Great Wall area. However, it has the greatest genetic differentiation between subpopulations (28.9%) among the six species. Even at the control site, it has the greatest genetic differentiation between subpopulations among the five species. This phenomenon could have resulted from its reproductive strategy. It is the only species in this study with both sexual and asexual propagation. Therefore, although gene flow between the two sides of the Great Wall or path could be aided by the wind, as indicated by cluster analysis, the individuals at one side may be the progeny of a few clones. Vegetative reproduction will reduce the genetic variation within the subpopulations of C. caespitosa, and increase the genetic differentiation between subpopulations.

Most parts of the Great Wall were built on the top of steep mountains, or even on top of cliffs. Naturally, environmental factors, such as light, temperature and moisture, along the two sides could differ slightly, even without the Wall. In order to have an objective evaluation of the effect of the Great Wall on the genetic differentiation of the separated subpopulations, a control site with similar geographic conditions, but without the Wall, on top of a mountain near the Juyong-guan Great Wall was selected. Ideally, the subpopulations at the control site should be separated by the same distance as those at the Wall area, but we failed to find such a site. Therefore, the plant individuals from two subpopulations were collected at the control site in such a way that they were either at least 5.8 m apart, such as V. negundo and C. caespitosa, or less than 5.8 m apart since no individual could be found at the ‘standard’ distance. AMOVA analysis showed that the subpopulations of all species at the control site had significant genetic differentiation (P<0.001 or <0.05, Table 3). Population genetic differentiation caused by microenvironmental differences has been well documented for plant species, such as Yushania niitakayamensis (Hsiao and Lee, 1999), wild emmer wheat (Nevo et al, 1988) and others (Hamrick and Allard, 1972; Schaal, 1975; Nevo, 1988; Nevo et al, 1981,1994). However, it is unexpected to find that the microgeographic difference between two sides, separated only by a 2-m wide path, could cause significant genetic differentiation between subpopulations. Although AMOVA analysis showed that the subpopulations within populations at the control site were significantly differentiated genetically, all individuals in two subpopulations were clustered together when neighbour-joining analysis was applied, that is, there was no clear cluster of subpopulations. Nevertheless, the variance component between subpopulations at the control site was about half that between corresponding subpopulations at the Juyong-guan Great Wall, and the variance component at the Great Wall sites was significantly greater than that at the control site. Therefore, based on this study, it is reasonable to deduce that the Juyong-guan Great Wall has served as a physical barrier to gene flow of separated plant subpopulations. It needs to be pointed out that more sampling sites along the Great Wall as well as relevant control areas should be selected in order to evaluate the effect of the Great Wall to the gene flow in general.


  1. Armstrong J, Gibbs A, Peakall R, Weiller G (1994). Randomly Amplified DNA Analysis. A computer program distributed by the Australian National University, Canberra, Australia.

    Google Scholar 

  2. Bauert MR, Kälin M, Baltisberger M, Edwards PJ (1998). No genetic variation detected within isolated relict populations of Saxifraga cernua in the Alps using RAPD markers. Mol Ecol 7: 1519–1527.

    CAS  Article  Google Scholar 

  3. Buso GSC, Rangel PH, Ferrira ME (1998). Analysis of genetic variability of South American wild rice populations (Oryza glumaepatula) with isozymes and RAPD markers. Mol Ecol 7: 107–117.

    CAS  Article  Google Scholar 

  4. Corre VLe, Dumolin-Lapègue S, Kremer A (1997). Genetic variation at allozyme and RAPD loci in sessile oak Quercus petraea (Matt.) Liebl.: the role of history and geography. Mol Ecol 6: 519–529.

    Article  Google Scholar 

  5. Dawson IK, Chalmers KJ, Waugh R, Powell W (1993). Detection and analysis of genetic variation in Hordeum spontaneum populations from Israel using RAPD markers. Mol Ecol 2: 151–159.

    CAS  Article  Google Scholar 

  6. Excoffier L (1993). WINAMOVA1.55 (Analysis of Molecular Variance). A computer program distributed by the Department of Anthropology and Ecology, University of Geneva, Carouge, Switzerland.

    Google Scholar 

  7. Gauer L, Cavalli-Molina S (2000). Genetic variation in natural populations of maté (Ilex paraguariensis A. St-Hil., Aquifoliaceae) using RAPD markers. Heredity 84: 647–656.

    CAS  Article  Google Scholar 

  8. Govindaraju DR (1988). Relationship between dispersal ability and levels of gene flow in plants. Oikos 52: 31–35.

    Article  Google Scholar 

  9. Hamrick JL, Allard RW (1972). Microgeographical variation in allozyme frequencies in Avena barbata. Proc Natl Acad Sci USA 69: 2100–2104.

    CAS  Article  Google Scholar 

  10. Hamrick JL, Linhart YB, Mitton JB (1979). Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Ann Rev Ecol Syst 10: 173–200.

    Article  Google Scholar 

  11. Hartl DL (1980). Principles of Population Genetics. Sinauer Associates: Sunderland, MA.

    Google Scholar 

  12. Hsiao JY, Lee SM (1999). Genetic diversity and microgeographic differentiation of Yushan cane (Yushania niitakayamensis; Poaceae) in Taiwan. Mol Ecol 8: 263–270.

    Article  Google Scholar 

  13. Huff DR, Peakall R, Smouse PE (1993). RAPD variation within and among a natural population of outcrossing buffalo grass (Buchloë dactyloides (Nutt.) Engelm.). Theor Appl Genet 86: 927–934.

    CAS  Article  Google Scholar 

  14. Liu Z (1994). Genetic relationships and variation among ecotypes of seashore paspalum (Paspalum vaginatum) determined by random amplified polymorphic DNA markers. Genome 37: 1011–1017.

    CAS  Article  Google Scholar 

  15. Luo ZW, Liu WY (1992). The Great Wall: A Miracle in the World (in Chinese). Culture Relic Press: Beijing.

    Google Scholar 

  16. Martin C, Gonzalez-Bentio ME, Iriondo JM (1997). Genetic diversity within and among populations of a threatened species: Erodium paularense Fern. Gonz & Izco. Mol Ecol 6: 813–820.

    Article  Google Scholar 

  17. Nesbo CL, Magnhagen C, Jakobsen KS (1998). Genetic differentiation among stationary and anadromous perch (Perch fluviatilis) in the Baltic Sea. Hereditas-Lund 129: 241–249.

    Article  Google Scholar 

  18. Nevo E, Brown AHD, Zohary D, Storch N, Beiles A (1981). Microgeographic edaphic differentiation of allozyme polymorphisms in wild barley. Plant Syst Evol 138: 287–292.

    Article  Google Scholar 

  19. Nevo E, Beiles A, Krugman T (1988). Natural selection of allozyme polymorphisms: a microgeographic climatic differentiation in wild emmer wheat (Triticum dicoccoides). Theor Appl Genet 75: 529–638.

    Article  Google Scholar 

  20. Nevo E (1988). Genetic diversity in nature: patterns and theory. Evol Biol 23: 217–246.

    Article  Google Scholar 

  21. Nevo E, Krugman T, Beiles A (1994). Edaphic natural selection of allozyme polymorphisms in Aegilops peregrina at a Galilee microsite in Israel. Heredity 72: 109–112.

    CAS  Article  Google Scholar 

  22. Qu L-J, Li D, Zhang Y, Liu M, Gu H, Chen Z (2000). Cloning and expression of a cDNA encoding ribosomal protein S4 from rice (Oryza sativa). Chin Sci Bull 45: 168–173.

    CAS  Article  Google Scholar 

  23. Schaal BA (1975). Population structure and local differentiation in Liatris cylindricea. Am Nat 108: 511–528.

    Article  Google Scholar 

  24. Schaal BA, Hayworth DA, Olsen KM, Rauscher JT, Smith WA (1998). Phylogeographic studies in plants: problems and prospects. Mol Ecol 7: 464–474.

    Article  Google Scholar 

  25. Smith JM (1999). Evolutionary Genetics. Oxford University Press: Oxford.

    Google Scholar 

  26. Turner ME, Stephen JC, Anderson WW (1982). Homozygosity and patch structure in plant populations as a result of nearest neighbor pollinations. Proc Natl Acad Sci USA 79: 203–207.

    CAS  Article  Google Scholar 

  27. Wolff K, El-Akkad S, Abbott RJ (1997). Population substructure in Alkanna orientalis (Boraginaceae) in the Sinai Desert, in relation to its pollinator behaviour. Mol Ecol 6: 365–372.

    Article  Google Scholar 

  28. Zhao GF, Felber F, Kuepfer P (2000). Subpopulation differentiation of Anthoxanthum alpinum (Poaceae) along an altitudinal gradient detected by random amplified polymorphic DNA. Acta Phytotaxon Sinica 38: 64–70.

    Google Scholar 

Download references


This study is financially supported by a State Key Basic Research and Development Plan (G2000046800). We thank Drs Martin Lascoux (Uppsala University, Sweden), Mei Sun (Hongkong University), Dik Hagenbeek (Hongkong Science and Technology University) and Guangyuan Rao (Peking University) for their critical reading of the manuscript, valuable suggestions and comments. The authors also thank Prof M Liu for her guidance on experimental techniques, Mr J Chen for his kind help in statistical analysis, and S Chang, X Chen, Zh Fan, B Li, D Liu, Y Liu, L Ming and X Shi for their assistance in collecting plant samples. The authors are grateful to an anonymous editor and two reviewers for their important suggestions and help in English language. Special thanks go to Dr Peter Raven (Missouri Botanical Garden, USA) for his inspiration and encouragement that finally led to this project.

Author information



Corresponding author

Correspondence to H Gu.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Su, H., Qu, L., He, K. et al. The Great Wall of China: a physical barrier to gene flow?. Heredity 90, 212–219 (2003).

Download citation


  • natural population
  • genetic differentiation
  • gene flow
  • RAPD

Further reading