Mitochondrial genetic diversity , phylogeny and population structure of Hydropotes inermis in South Korea

Baek-Jun Kim†, Yun-Sun Lee†, Yong-Su Park, Kyung Seok Kim, Mi-Sook Min, Sang-Don Lee and Hang Lee* Department of Climate and Ecology, National Institute of Ecology, 1210, Geumgang-ro, Maseo-myeon, Seocheon-gun, Chungcheongnam-do 325-813, Korea Conservation Genome Resource Bank for Korean Wildlife and Research Institute for Veterinary Science, College of Veterinary Medicine, Seoul National University, San 56-1, 1 Gwanak-ro, Gwanak-gu, Seoul 151-742, Korea Department of Ecological Conservation, National Institute of Ecology, 1210, Geumgang-ro, Maseo-myeon, Seocheon-gun, Chungcheongnam-do 325-813, Korea Department of Environmental Science and Engineering, College of Engineering, Ewha Womans University, 52, Ewhayeodae-gil, Seodaemun-gu, Seoul 120-750, Korea

because of illegal hunting and habitat loss.Therefore, a small number of water deer now live only in four fragmented areas (Sheng, 1992;Wang, 1998;Xu et al., 1998;Zhang and Guo, 2000;Guo andZhang, 2002, 2005).Recently, the subspecies was classified as a vulnerable species in the China Red Data Book of Endangered Animals (Wang, 1998) and the IUCN Red List (Harris and Duckworth, 2008), and its total population is about 10,000 (Butzler, 1990).The Korean subspecies was previously found along the ranges of Mt.Taebaek and Mt.Nangrim, distributed extensively throughout the Korean peninsula (Won and Smith, 1999).Like the Chinese subspecies, its abundance and distribution declined severely due to excessive harvesting and habitat destruction in the 1950s.Although the current population seems to be abundant and widely distributed in local habitats (Woo et al., 1990, Kim et al., 2011a), the total population size of the Korean subspecies has not yet been surveyed.
To date, a few genetic studies have been conducted on water deer using mitochondrial DNA markers (e.g., cytochrome b and the control region) (Hu et al., 2006;Koh et al., 2009) and nuclear microsatellite markers (Hu et al., 2007).Hu et al. (2006) ascertained that the overall Chinese subspecies displays no clear phylogenetic relationship between different regions of origin, even though a relatively higher genetic diversity in the mitochondrial D-loop region was found compared to that in other rare Cervidae.However, an analysis of molecular variance (AMOVA) showed a significant differentiation between the Zhoushan archipelago population and the mainland population, which indicated the existence of a geographic barrier to gene flow between the two populations (Hu et al., 2006).Furthermore, Hu et al. (2007) found that the Chinese water deer has a relatively high level of genetic diversity among microsatellite markers (n = 7) compared to other rare Cervidae, and strong differentiation between different regions of origin (mainland and Zhoushan archipelago).More recently, Koh et al. (2009) reported that the Chinese and Korean subspecies seem to be very closely related, displaying low levels of genetic distance for cytochrome b (1.3%) and the control region (2.1%), which contrasts with previous expectations that the two subspecies would have distinct characteristics based on the pelage color of the deer.This previous study only focused on the taxonomic status of the two water deer subspecies (Koh et al., 2009).Detailed genetic information about the Korean subspecies, such as the genetic diversity and population structure among various regions or localities of Korea, is scarce.Therefore, further studies are essential to improve our understanding of the genetic characteristics of the Korean subspecies.
The focus of this study was to examine in detail the mitochondrial genetic characteristics of the Korean subspecies.To this end, the genetic diversity, phylogeny and degree of genetic differentiation of the Korean subspecies were investigated for Korean water deer sampled across geographic localities from various regions in South Korea.Anthropogenic interference and climate change as possible causal effects to explain the present status of mitochondrial diversity of water deer in Korea are also discussed.

MATERIALS AND METHODS
Sampling sites and DNA samples A total of 107 individual samples of water deer were studied in the wilds of the Korean peninsula (Table 1).Three different habitats were targeted for tissue and hair sampling: the midwestern region (Gyeonggi Province: 12 locations), the mid-eastern region (Gangwon Province: 5 locations), and the southwestern region (Jeolla Province: 9 locations) (Fig. 1).All tissue samples were collected in 2008 and 2009, and were frozen at −30°C until they were used for DNA extraction.
DNA extraction, PCR amplification and sequencing Genomic DNA was extracted from tissue samples (e.g., muscle, tongue, skin, liver, kidney and spleen) using a LaboPass Tissue kit (Cosmo Genetech, Seoul, Korea).Additionally, a chelex method was used for DNA extraction from hair samples (Walsh et al., 1991).Two sequential PCR assays were carried out for the mitochondrial D-loop region in this study.One pair of primers, L15775 and H651, was first used to amplify a complete D-loop fragment (Kocher et al., 1989).A second pair of newly designed primers, BJWDDLOOP_F (5′-GCT CCA TAA AAA CCA AGA AC-3′) and BJWDDLOOP_R (5′-TTG AGT ACA AAA AAT GGC GC-3′), was then used to amplify a partial fragment from the first PCR products.PCR was carried out in a 20-μl reaction that contained 1 μl of DNA template, 1X PCR buffer (iNtRON Biotechnology, Seongnam, Korea), 2 mM MgCl 2 (iNtRON), 0.2 mM of each dNTP (iNtRON), 0.1 μM of each primer, 2.5 μg of BSA (Promega, Fitchburg, WI, USA) and 1.25 U of i-Star Taq polymerase (iNtRON).PCR was performed in a PTC-100 PCR Thermal Cycler (MJ Research, Waltham, MA, USA) under the following conditions: initial denaturation for 4 min at 94°C followed by 40 cycles of [94°C for 60 s, 50°C or 57°C (for the first and second PCR, respectively) for 45 s, and 72°C for 60 s], with a final extension for 5 min at 72°C.The PCR products were resolved by electrophoresis on a 2% agarose gel, stained with ethidium bromide, and visualized under a UV transilluminator.All the PCR products were purified from the gel using a Zymoclean Gel DNA Recovery kit (Zymo Research, Orange, CA, USA).The purified PCR products were bidirectionally sequenced using forward (BJWDDLOOP_F) and reverse (BJWDDLOOP_R) primers with an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
Data analysis Partial sequences of the mitochondrial D-loop region were aligned using the AlignIR program, version 2.1 (LI-COR, Lincoln, NE, USA).These sequences were manually edited at least five times on the alignment program.Any sequence that varied by at least one nucleotide was considered to be a different haplotype.The number of polymorphic sites, parsimonious informative sites, singletons and haplotypes (A), and the amount of haplotype diversity (h) and nucleotide diversity (π), were calculated using DNASP version 4.10.9(Rozas et al., 2003).Sequence divergences (p s ) at each region and all regions, and pairwise sequence divergences among regions and clades were calculated using the Kimura 2 parameter (K2P; Kimura, 1980) with MEGA version 4.0 (Kumar et al., 2004).
The phylogenetic relationships between the haplotypes of the Korean and Chinese subspecies (accession no.EU315254 and NC011821) were reconstructed using the neighbor joining (NJ) method under the K2P model and the maximum parsimony (MP) method (Saitou and Nei, 1987).The maximum likelihood (ML) phylogenetic tree was also constructed.To select the best-fit model, ML analysis was subjected to the Akaike Information Criterion (Akaike, 1974) using jModelTest version 0.1 (Posada and Crandall, 1998).The HKY+I model was determined to be most appropriate for our DNA sequences with the following parameters.Both MP and ML heuristic searches were conducted using a procedure with a tree-bisectionreconnection branch swapping algorithm with 100 random addition sequence replications.The mitochondrial control region sequence of the Siberian roe deer (Capreolus pygargus) was chosen as an outgroup to root the phylogenetic trees, and retrieved from GenBank (accession no.Z70317) (Douzery and Randi, 1997).Confidence in the estimated relationship was determined using the bootstrap approach and obtained through 1,000 (for NJ) or 100 (for MP and ML) replications incorporating the same model as detailed above (Felsenstein, 1985).Bootstrap analysis and phylogeny reconstruction were both conducted using PAUP version 4b10 (Swofford, 2001).Both NJ and MP were used, but the ML tree was not used due to its poor resolution with relatively low bootstrap values (< 50%).In addition, a median-joining network was created by combining the haplotypes of the water deer using the program Network version 4.0 (Bandelt et al., 1999).
Fisher's exact test was performed with GENEPOP DOS version 4.0.10 (Raymond and Rousset, 1995).In addition, an AMOVA was conducted to investigate population differentiation (both F ST and Φ ST ), using the program ARLEQUIN version 3.0 (Excoffier et al., 2005, Weir andCockerham, 1984).Pairwise Φ ST values were calculated in the analysis with 1,000 permutations of haplotypes between the three regions, which consisted of 26 locations under the hypothesis of panmixia.The software was further used to examine the demographic histories of the populations for regions and lineages.The distribution of the pairwise differences (mismatch distribution) was constructed to distinguish stable populations over time with regard to recent expansion or reduction.Tajima's D and Fu's F statistics were also computed.
Phylogeny and demographic history NJ and MP trees were reconstructed with 17 Korean water deer haplotypes (accession no.KC581917-KC581933), one Chinese water deer haplotype (two reference sequences: accession no.EU315254 and NC_011821) and one Siberian roe deer haplotype (Z70317) (Fig. 2).Both the NJ tree and the MP tree showed identical topologies, with two major clades of Korean populations.Between the  two clades (lineage A and B), one subclade (A1) was grouped with the other subclade (A2) that consisted of two Chinese water deer (Fig. 2).There was no clear locationspecific distribution pattern in the Korean populations (Fig. 2 and Supplementary Fig. S1).The median-joining network also showed the same pattern as that of the phylogenetic trees (Fig. 3).Tajima's D and Fu's F test indicated that historical demography has had an important effect on the pattern of genetic variation of mitochondrial DNA and genetic diversity for lineage B (P < 0.05 in Fu's F) (Supplementary Table S1).In addition, the mismatch distribution of the mitochondrial DNA lineage exhibited a clear, smooth, and unimodal shape (or bell shape) for lineage B (Supplementary Fig. S2), which suggests that a certain level of population expansion occurred for lineage B, although the raggedness index (P RAG ) was not supportive of the population expansion of the lineage (Supplementary Table S1).

Population differentiation
The null hypothesis that "the haplotype distribution is identical across three regions (Gyeonggi, Gangwon and Jeolla Provinces)" was rejected for the three regions comprising 26 locations in South Korea (Fisher's exact test: P < 0.001; data not shown).Moreover, the result of AMOVA showed a significant genetic barrier among the three regions (F ST = 0.28416, P < 0.00001; Φ ST = 0.19239, P < 0.00001; Table 4).
In particular, population differentiation was observed between Jeolla and Gyeonggi (Φ ST = 0.31) and between Jeolla and Gangwon Provinces (Φ ST = 0.23) (Table 5).Tajima's D and Fu's F test for three regions was not supportive of population fluctuation (Supplementary Table S1).

DISCUSSION
Of all the small ruminant species in East Asia, China and Korea, the water deer is one of the rarest, only inhabiting two countries.Despite this biogeographical status, the Korean subspecies has not been protected due to its high population density in South Korea, while the Chinese subspecies is protected in China.Recently, a restoration project for the Chinese subspecies began in Shanghai and the subspecies has been continuously studied in China.However, studies on water deer in Korea have rarely been conducted, regarding its ecology (e.g., population size, distribution and density) and genetics (e.g., genetic diversity, taxonomic status and population differentiation) (Kim et al., 2011a).In particular, the present status of the genetic diversity, phylogeny and population structure of the Korean subspecies is an important issue that needs to be examined to obtain a better understanding of its genetic characteristics and management.
Several hypotheses can be suggested to explain the relatively low genetic diversity of the Korean population.First, this low genetic diversity (number of haplotypes, sequence divergence, nucleotide diversity and haplotype diversity) may result from historical anthropogenic effects, such as excessive harvesting, poaching, and political upheavals such as war and famine.Historically, for Phylogeny and population structure of water deer more than a thousand years, the distribution of water deer in the Korean peninsula seems to have been quite different from their current distribution.Currently, water deer are found in all South Korean regions, including high-altitude mountainous areas (Woo et al., 1990).However, the historical range of water deer in the Korean peninsula was restricted mainly to the southwestern, lowland region of the peninsula.This was probably because water deer prefer lowland habitats near water sources, such as riverside wetlands.A high density of predator (tigers, leopards and wolves) and competing ungulate species (sika deer and roe deer) might also have prevented the expansion of their range further into the eastern mountainous areas of the peninsula.However, the situation changed during the last hundred years: most major predator species along with the major competing species became extinct in South Korea, and consequently water deer became the most abundant ungulate species in the whole of the country.Thus, the historical distribution of water deer had been restricted to the southwestern part of the Korean peninsula, which coincides with the most intensively developed agricultural area and the highest density of human population of the peninsula.Therefore, the historically localized water deer population may have been highly vulnerable to human activities like hunting, land reclamation and war.The South Korean water deer population has become more abundant during the last two decades, but more time will be required to recover from their lowered genetic diversity.
A second cause of this low genetic diversity may have been a founder effect during a glacial period, as historical evidence of climate change in the Korean peninsula suggests the existence of glaciers in high mountainous northern regions during the Pleistocene period (Kim, 1994;Kong and Watts, 1993).In contrast, no glaciers existed in the central or southern parts of the Korean peninsula despite the extreme cold.However, since southern coastal areas are presumed to have been comparatively warmer (Kong and Watts, 1993), a refuge may have formed somewhere along these warmer coastlines.The existence of a small founder population in such a refuge could be the cause of the comparatively lower genetic diversity.This hypothesis would concur with the population expansion in lineage B based on mismatch distribution.Phylogeny Water deer were previously considered to be a species consisting of two subspecies, H. inermis argyropus and H. inermis inermis (Cooke and Farrell, 1998;Geist, 1998).These two subspecies have very similar phenotypes, except for their coat colors (Tate, 1947).The Korean subspecies has a darker coat color and a much more reddish-colored head than the Chinese subspecies (Koh et al., 2009).Koh et al. (2009) reported two sympatric phylogroups and low levels of mitochondrial genetic distance between the two subspecies (Korean and Chinese) in both the cytochrome b gene (1.3%) and the Dloop region (2.1%).Therefore, they suggested that the validity of the subspecies designation amongst water deer needs to be re-evaluated by performing detailed statistical analyses of morphological characters, including coat color, as well as genetic characteristics.
Likewise, we found that there are two sympatric clades (A and B) in water deer (Fig. 2).The Chinese D-loop sequences (EU315254 and NC_011821) are grouped with one (A) of the two clades.The genetic distance between lineage A and lineage B was 1.332 ± 0.340% (data not shown).The standard equation d = 2Tu (where u is the substitution rate and T is time (Saitou and Nei, 1987)) and a conservative fossil record calibration of 25 million years ago (MYA) for the most recent common ancestor of Cervidae and Bovids (Vrba and Schaller, 2000) produced a substitution rate of approximately 0.39% (u = 0.00393) per site per million years (Ludt et al., 2004).These estimates concordantly suggest that the A and B lineages (2.1-1.3MYA) separated at the beginning of the Pleistocene period.The genetic distances within lineage A and lineage B were 0.525 ± 0.167% and 0.264 ± 0.113%, respectively.This indicates that both lineages have a very close phylogenetic relationship in terms of mitochondrial D-loop haplotypes.The median-joining network showed that of the two location-nonspecific lineages, one sub-lineage (A1) was grouped with the Chinese sublineage (A2) (Fig. 3).This result also supports the possibility that the two water deer subspecies are from the same taxonomic group.
Moreover, the results of the previous research in China and Korea and ours indicate that both lineages A and B appeared in Korea while only lineage A appeared in China (Hu et al., 2006;Koh et al., 2009).This suggests that lineage B originated in Korea.Furthermore, the two lineages were separated from each other in the early Pleistocene period when climatic fluctuations greatly affected the distribution of species.At the beginning of the Pleistocene, South Korea was connected with China, as a fresh-water lake or river existed where the Yellow Sea currently resides (Dobson, 1994;Keigwin and Gorbarenko, 1992;Kim, 1994;Nelson, 1993).The water deer may therefore have been distributed widely in both Korea and China without a specific barrier.At the start of the Ice Age (about 1.6 MYA), two refuges may have formed in comparatively warm areas.Their populations gradually became lineages A and B. Lineage B may have inhabited the southern coastline of Korea, and then population expansion toward the north and west during interglacial periods, and/or exchanges between the two lineage populations in China and Korea during relatively mild glacial periods, may have occurred.As some groups of lineage A migrated eastward, a complete separation of Korea and China, about 10,000 years ago, may have been the cause of the two dependent lineages existing in Korea (Kong and Watts, 1993).Although the two lineages also existed in China, the survival of only one lineage may have been due to a decrease in population because of human intervention (Koh et al., 2009).However, more detailed analyses, including DNA sequencing using fossils in China, would be required to test this hypothesis.

Population differentiation
The AMOVA test suggested the existence of distinct populations with slightly different genetic structures (F ST = 0.28416, P < 0.00001; Φ ST = 0.19239, P < 0.00001; Table 4).This indicates that there has been a barrier between the populations to prevent random gene flow among them.In particular, population isolation was observed between the Jeolla and Gyeonggi (pairwise Φ ST = 0.31246, P < 0.00001) or Gangwon (pairwise Φ ST = 0.23187, P < 0.00001) regions.Hu et al. (2006) also showed that a distinctive genetic barrier existed between the Zhoushan archipelago and mainland China (F ST = 0.088, P < 0.001; Φ ST = 0.075, P = 0.043).According to Hu et al. (2006), a geographical separation of island and continent caused population differentiation.However, population differentiation could also be derived from a different dispersal ability in males and females.Territory and dispersal of male water deer are generally wider than those of females (Kim et al., 2011b).The result of analysis of mtDNA, which is maternally inherited, indicated a slight population differentiation in this study.Further study is needed using other markers, such as a nuclear microsatellite markers, to show more precise regional differentiation.
This study confirmed the low genetic diversity and controversial phylogenetic status of the Korean water deer, and showed that there may be a certain level of differentiation among the populations at the regional scale.However, only water deer from restricted regions (n = 3) were analyzed.Furthermore, we did not analyze the genetic differentiation at finer scales, such as between water deer from the southern and northern parts of Gyeonggi Province (or the Han River), between the northern areas of Gyeonggi Province and western areas of Gangwon Province, and between the western and eastern sites of Gangwon Province (or the Taebaek Mountains).In addition, we did not include the other provinces in South Korea (e.g., Chungcheong and Gyeongsang Provinces), due to the insufficient number of samples collected from these regions.Future studies using more samples collected from various localities in China as well as Korea will provide better resolution of the genetic structure of water deer at regional and local scales.Our analyses were based only on matrilineal mitochondrial Dloop markers, and therefore do not reflect any patrilineal influence.Therefore, further use of other genetic markers (e.g., nuclear microsatellite DNA or Y-linked DNA markers) would help us to find unbiased and comprehensive information related to the genetic diversity, phylogeny and population differentiation of water deer in East Asia.

Fig. 2 .
Fig. 2. Phylogenetic tree of mitochondrial control region haplotypes of the Korean subspecies of water deer.The tree was rooted using a Siberian roe deer sequence as the outgroup.The numbers of changes over the whole sequence are indicated by the scale.H: haplotype; A, B: lineage or clade; A1 and A2: sublineage or subclade.The first and second numbers on branches are bootstrap values obtained from NJ and MP analyses, respectively.Only values over 50% are indicated; those under 50% are shown as '-'.

Fig. 3 .
Fig. 3. Median-joining network of 18 control region haplotypes of water deer including Chinese subspecies.Branch lengths are scaled to the number of nucleotide substitutions.The size of the circles is proportional to the haplotype frequencies.H: haplotype; A and B: lineage or clade; A1 and A2: sublineage or subclade; CWDH: Chinese water deer haplotype.

Table 1 .
Geographical region, sample size and type, and GenBank accession number for each sample of water deer used in this study

Table 2 .
Summary of haplotype definition of all obtained mitochondrial control region sequences of water deer individuals showing the polymorphic sites only.The numbers in the columns refer to variable nucleotide positions in the 822-bp DNA sequences

Table 3
. Summary of sample size (N), number of haplotypes (A), haplotype diversity (h), mean sequence divergence (p s ) and nucleotide diversity (π) for the Korean water deer populations (Gyeonggi, Gangwon, Jeolla and total)

Table 4 .
Hierarchical analysis of molecular variance based on partial mitochondrial control region sequences.Percentage of variation explained by different levels of population structure, variance components, and F-and Φ-statistics *P < 0.001, ***P < 0.00001.