Historical demographyThe population

Historical demography
The populations from Region 2 showed the genetic imprints of a demographic expansion as inferred by a negative Fs test, which was also supported by the mismatch distribution analysis and Rog- ers test of sudden population expansion (Rogers, 1995; Table 1). The expansion time was estimated to be between 1.043 and 2.087 mya. However, the rise of the Tibet–Qinghai Plateau, which is believed to have a profound influence on the organisms on the Tibetan plateau itself and its neighboring areas including Hainan and Guangdong (Harrison et al., 1992; Molnar et al., 1993; Shi et al., 1998), occurred 0.60–1.10 million years ago (Zheng, 1981; Shi et al., 1998), later than the expansion time. Glaciation and deglaciation and the accompanying lowering and rising of sea lev- els during the Pleistocene is known to have greatly affected land mass configurations and plant and animal distributions in South- east Asia (Voris, 2000). Given that these lizards are associated with sand at the seaside, sea-level changes are more important for these lizards than the rise of the Tibet–Qinghai Plateau. Depending on the size of a land mass and its connectivity to other land masses, sea level fluctuations could have led to local extinctions of some populations or cause populations to be isolated reproductively. Subsequent dispersal and accumulation of genetic differences over time are possible evolutionary mechanisms that account for the genetic divergence observed in this study.
4.3. Genetic diversity and dispersal
Low genetic diversity could be due to a number of factors, one of which is overexploitation (Kochzius and Nuryanto, 2008). Over- exploitation of L. reevesii as a dietary item and for medicinal use is the main cause of population decline over the past three decades (Zhao, 1998). In Clade A [individuals from the two neighboring localities of Lingshui (LS) and Sanya (SY)], for example, high genetic diversity is found in LS (a less disturbed locality by human) and low genetic diversity is observed in disturbed and exploited popu- lations in SY, a developed travel industry site where hunting pres- sure is extremely high.
The median-joining network shows that H85 on one tip of the network is a young haplotype (Fig. 2), shared by all individuals from Beihai (BH), Shiwan (SW) and Wujia (WJ), and four (25%) individuals from Fucheng (FC). This result provides evidence show- ing that BH, SW and WJ are most recently colonized by L. reevesii in Region 3. Clade C includes haplotypes from Region 3 that exhibit a trend of northwardly decline in nucleotide diversity with the value of 0 inferred from cytochrome b in the BH, WJ and SW populations (Fig. 2), showing a northward dispersal route. It is known that ancestral populations often lose their genetic diversities during their dispersal or expansion and contraction phases (e.g. Yang et al., 2004). Low genetic diversity in these populations presum- ably reflects the consequence of a founder effect. Genetic diversity is lower in Kelu (KL) than in Nanxing (NX) but higher in KL than in Fucheng (FC). This geographic trend highlights the importance of dispersal in shaping the spatial pattern of genetic diversities at that site.
Given that haplotypes from Vietnam is relatively ‘‘old” (ancient) and that haplotypes from Guangdong and Guangxi (Region 3) is relatively ‘‘new” (recent) (Fig. 2), one plausible scenario to explain our genetic data is a historical dispersion of L. reevesii as proceed- ing from Vietnam (Region 4) to Hainan (Region 1 and Region 2), followed by a second wave of dispersal from Hainan to Guangdong and Guangxi (Region 3). However, another equally plausible sce- nario is a historically widespread population that has been struc- tured by vicariant factors such as the mountains in Hainan and sea level fluctuations. Isolation caused by the orogenesis and sub- sequent genetic drift, together with adaptation to local environ- ments led to genetic differentiation and further speciation of Leiolepis (Macey et al., 2000). In this study, the phylogenetic tree indicates a three-way polytomy among the three major Clades (A, B and C), thus provides no evidence that any one clade is older than the other two (as would be expected in a dispersal scenario). The three-way polytomy might be the natural result of vicariant events subdividing widespread populations (though undoubtedly Clade A diverged from Clades B and C much earlier than B from C). AMOVA showed that more than half of the total genetic varia- tion occurred among the regions and among population within the regions, indicating high levels of geographical structuring and restricted gene flow. This population genetic structure could be lar- gely due to the fact that the lizard has a limited ability to disperse. Herptiles including L. reevesii that are less likely to disperse are more prone to suffering substantial losses in genetic diversity resulting from habitat loss and fragmentation (Taylor et al., 1994; Bouzat et al., 1998; Wisely et al., 2002; Lin et al., 2007). This was a factor to be considered when evaluating dispersal versus vicariance scenarios. At this time, there is no overwhelming evi- dence supporting one scenario over another. Thus, it may have the benefit that alternative scenarios are evaluated in the future with additional genetic data (microsatellites might be very useful here) or by comparing similar taxa for genetically congruent patterns.
Acknowledgments
The Provincial Forestry Departments of Hainan, Guangdong, Guangxi, Thanh Hoa, Nghe An and Binh Dinh provided permits for collecting lizards. This work was supported by grants from Nat- ural Science Foundation of China (Project No. 30860046) and Nan- jing Normal University Innovative Team Project to Ji’s group. Thanks are given to Xin-Yang Chen, Yong Guo, Guang-Wu Huang, Hua-Song Lei, Qing-Bo Qiu, Zu-Chun Wang, Jia-Yong Zhang and Qun-Li Zhang for their assistance in collecting animals, and to Lu-Xi Mao, Yan-Fu Qu and Qing-Qing Chen for their laboratory assistance.