Sample collection

A total of 47 red deer samples (dried skins, hairs, or bloods) from eight populations were collected (Fig. 1 and Table 1). No specimens from the subspecies macneilli (see Fig. 1) were available. Dried skins of two Mongolian red deer were offered from Dr. Sumiya Ganzorig of Hokkaido University. Hairs of one Hungarian and one Polish red deer were supplied from Mr. Geza Voros of Director of Oreglak Game Processing Plant of Hungary. Total DNAs from those tissues were extracted using the phenol/proteinase K/sodium dodecyl sulphate method of Sambrook et al. (1989) with slight modifications (Masuda and Yoshida, 1994) or using Chelex 100 (Bio-Rad) (Walsh et al., 1991). The following sequences were obtained from GenBank (Table 1): European and North American C. elaphus, accession Nos.  AF016976,  AF016970,  AF016957,  AF016966,  AF016972,  AF016973; Cervus nippon,  AF016974 (Polziehn and Strobeck, 1998).

Table 1

Sample profiles ofCervus elaphus in the present study

PCR and direct sequencing of mtDNA control region fragments

Primers (LD5 and HD8) for polymerase chain reaction (PCR) amplification and sequencing (Nagata et al., 1998) were used as shown in Fig. 2. An aliquot (10 μl) of the DNA extract was subjected to PCR amplification in a reaction mixture of 50 μl, which included 10 mM Tris (pH 8.3), 50 mM KCl, and 1.5 mM MgCl₂, each dNTP at 0.2 mM, 1.25U of Taq DNA polymerase (Takara), and each primer at 0.5 μM. Step programs for PCR were as follows: 94°C for 1 min; 50°C for 1 min and 72°C for 2 min. The cycle was repeated 35 times followed by a completion reaction at 72°C for 10 min. PCR products were purified with the QIAquick PCR purification kit (Qiagen). Then, 5 μl of purified PCR products were sequenced using the ABI Prism™ 377 DNA sequencing system (Perkin-Elmer).

Fig. 2

Schematic diagram of the Cervus elaphus mtDNA control region. Numbers 1–6 indicate tandem repeats of 38 to 43-bp motifs. Arrows show positions of primers used for amplification and sequencing. Shaded motifs 3 and 4 were absent in both Tarim and European populations.

Sequence analysis

Sequence alignment was performed using a computer software GeneWorks (Intelligenetics). A phylogenetic tree was reconstructed using the neighbor-joining method (Saitou and Nei, 1987) in a computer software MEGA (Kumar et al., 1993). Numbers of nucleotide substitutions per site were estimated for multiple substitutions using the Kimura's (1980) two-parameter method in MEGA. Insertion or deletion (indel) sites were eliminated for sequence analysis. Bootstrap (Felsenstein, 1985) values were derived from 1,000 replications. Parsimonious relationships between haplotypes were presented by hand-written networks. The C. nippon sequence was used as outgroup to root the neighbor-joining phylogenetic tree.

Tandem repeats and polymorphisms in mtDNA control region sequences

Nucleotide sequencing in the present study yielded 402-bp fragments of the C. elaphus mtDNA control region. From the 47 individuals, we identified 23 haplotypes (CE-05 to -27)(Table 1). Haplotypes CE-01 to -04 and CE-28 and -29 were cited from previous reports of Polziehn and Strobeck (1998) as mentioned in Materials and Methods. A sequence alignment showed that 57 of 304 sites (19%: 57/304 sites) excluding indels were polymorphic (see Appendix).

This fragment included tandem repeats (four or six copies) of 38 to 43-bp motifs: four copies in populations of Tarim (CE-21 to -26) and Europe (CE-27 to -29), while six copies in populations of Alashan (CE-17 and -18), Gansu (CE-19), Tibetan (CE-20), northeastern China (CE-07 to -09), Tianshan (CE-10 to -14), Altai (CE-15 and -16), Mongolia (CE-05 and -06), and North America (CE-01 to -06) (Table 1 and Appendix). Consequently, two copies (third and fourth copies) of the motifs were absent in populations of both Tarim and Europe, compared with those of the other areas (Fig. 2 and Appendix). In addition, CE-25 and CE-26 from the Tarim population shared a 19-bp deletion within the first copy of the motif (Table 1 and Appendix).

Phylogenetic relationships among populations

The neighbor-joining tree showed that mtDNA haplo-types from Asia, Europe, and North America were classified into two major clusters, named the Eastern (consisting of CE-01 to -20) and Western (consisting of CE-21 to -29) lineages with 91% and 76% bootstrap values, respectively (Fig. 3). The separation of the two lineages was in congruence with haplotype classification by copy numbers of tandem repeats: six copies for the Eastern lineage and four copies for the Western lineage (Table 1).

Fig. 3

Phylogenetic relationships among haplotypes of the Cervus elaphus mtDNA control region, constructed by the neighbor-joining method. Haplotype names refer to those of Appendix. Scale bars represent genetic distances calculated by the Kimura's two-parameter method, exclusive of indel sites. E1-8 and W1-2 were clusters with sampling locality names. Numbers near internal branches are bootstrap values (%) derived from 1,000 replications.

Since haplotypes from geographically closer localities (see Fig. 1) were genetically closer to each other and formed their own clusters with 47–100% bootstrap values, we named clusters E1-8 and W1-2 (Fig. 3). Parsimonious networks supported the existence of the Western and Eastern lineages and the haplotype clusters found in the neighbor-joining tree (Fig. 4).

Fig. 4

Parsimonious networks between haplotypes of the Cervus elaphus mtDNA control region. Haplotype names in open circles correspond to those of Fig. 3. Closed circles indicate predicted haplotypes. One slash means a nucleotide substitution, excluding indel sites. The networks basically support the separation between the Western and Eastern lineages, the ten clusters (E1-8 and W1-2), and the haplotype relationships which were formed in the neighbor-joining tree (Fig. 3).

In the Western lineage, haplotypes (CE-21 to -26) from the Tarim populations of southern Xinjiang formed their own cluster (W1) with a 92% bootstrap value, while haplotypes (CE-27 to -29) from Europe (W2) were clustered with a 100% bootstrap value (Fig. 3). CE-25 and CE-26, both which shared the 19-bp deletion (Table 1), formed a sub-cluster within the Tarim cluster (W1) with a 97% bootstrap value.

Meanwhile, in the Eastern lineage, CE-19 from Gansu (E7) and CE-20 from Tibet (E8) were closer to each other with a 88% bootstrap value (Fig. 3). They were closely related to the Alanshan population (Cluster E6; CE-17 and CE-18). Haplotypes CE-15 and CE-16 from Altai Mountains of northern Xinjiang (Cluster E5; 99% bootstrap value) and haplotypes (CE-10 to -14) from Tianshan Mountains of northern Xinjiang (E4; 47% bootstrap value) formed one larger cluster with CE-07 to -09 from northeastern China (E3; 56% bootstrap value), CE-05 and CE-06 from Mongolia (E2; 100% bootstrap value), and CE-01 to -04 from North America (E1; 61% bootstrap value), showing a 80% bootstrap value (Fig. 3). Fig. 3 showed that North American populations were relatively closer to the Mongolian population (Cluster E2). In addition, Tianshan (E4) and Altai (E5) populations from northern Xinjiang were included in the Eastern lineage, whereas the Tarim population (W1) from southern Xinjiang was grouped into the Western lineage (Fig. 3). Thus, mtDNA lineages of southern and northern Xinjiang were different from each other. The boundary of the two lineages could have occurred between Tianshan Mountains and Tarim River in Tarim Basin. Parsimonious networks (Fig. 4) also showed that the relationships among haplo-types were the same as those of Fig. 3.

Genetic distances were 6.3% within the Western lineage (W1 and W2), 1.8–5.4% within the Eastern lineage (E1-8), and 6.9-9.9% between the Western and Eastern lineages (Table 2).

Table 2

Genetic distances (%) between clusters, calculated using Kimura's two-parameter method. Clusters E1–8 are from the Eastern lineage, while W1 and W2 are from the Western lineage (see Fig. 3).

Phylogeography of Cervus elaphus in the Holarctic region

The present study demonstrated that there are two major mtDNA lineages in C. elaphus of Eurasia: the Western and Eastern lineages (Table 1; Figs. 3 and 4). It was supported by geographic distribution of four (Western lineage) or six (Eastern lineage) copies of motifs in tandem repeats identified in the present study, in addition to the molecular phylogenetic relationships between haplotypes. The similar tandem repeats of motifs in the mtDNA control region were found in the Japanese sika deer Cervus nippon, whose repeat numbers were more variable (four to seven copies) and population-specific within Japanese islands (Nagata et al., 1998, 1999). Compared with the data of C. nippon, it is noticed that C. elaphus from much wider ranges (Eurasia and North America) shared only two kinds (four or six) of copy numbers.

Each cluster of the trees was almost congruent with their geographical distribution as well as subspecies classification (Figs. 3 and 4). This means that genetic characterization in the present study was consistent with morphological differentiation in local populations of C. elaphus.

The genetic distances (6.9%−9.9%, Table 2) between the Western and Eastern lineages corresponded to 0.3–0.4 million years ago for the divergence time, based on the mutation rate (0.118×10⁻⁶ substitutions/site/year) of the human mtDNA control region reported by Stoneking et al. (1992). After the divergence between the two lineages, some populations with the Western lineage could have immigrated into Europe. Fossil data showed that C. elaphus appeared in Europe in Middle Pleistocene (approximately 0.7 to 0.55 million years ago) (Kurten, 1968; Nilsson, 1983; Lister, 1984, 1990, 1993). Divergence times estimated by us supported that C. elaphus occurred in Europe in Middle Pleistocene, although the estimated divergence times were somewhat younger than the fossil data.

The Eastern lineage included North American populations which were closer to Mongolian and Northeastern Chinese populations (Figs. 3 and 4). This fact indicates that the ancestral populations of C. elaphus immigrated from northeastern Eurasia to North America via Beringia which was formed during the last glacial period, as considered from fossil data (Geist, 1998). From the genetic distances (2.0%, Table 2) between North American and the Mongolian populations, the divergence time was estimated to be 80,000 years ago. Polziehn and Strobeck (1998) estimated that the divergence time between Siberian and North American populations of C. elaphus was 52,000 years ago based on the mtDNA control region data. Geist (1971, 1983, 1998) described from fossil data that at the Riss (Illinoian) age the C. elaphus remained in Alaska, moved into Alberta approximately 40,000 years ago, and entered southern parts of North America. Our results were not in discordance with the divergence time estimated by previous DNA analysis (Polziehn and Strobeck, 1998) and fossil data (Geist, 1971, 1983, 1998). Recent molecular phylogenetic studies on intercontinental populations of brown bears Ursus arctos showed that some parts of Eurasian populations are very close to those of Alaska and that ancestral populations immigrated from northeastern Eurasia to North America, via Beringia of Late Pleistocene (Leonard et al., 2000; Matsuhashi et al., 2001). Thus, large mammals such red deer and brown bears, which have adapted themselves to relatively cold climates, likely have a similar history on intercontinental migration in Pleistocene.

Migration history within Eurasia

Cervus elaphus is considered to have evolved from the sika deer-like ancestor when they immigrated from the Himalayan foothills to the Chinese continent. In addition, C. elaphus expanded their distribution to Europe through Middle Asia, then, in connection with glacial retreats and glacial cup disappearance in Middle Pleistocene, went back to northern parts of the Chinese continent through Xinjiang, and finally immigrated to Siberia and North America (Geist, 1971, 1983; Groves and Grubb, 1987; Ohtaishi, 1992, 1995). The present study, however, demonstrated that the European populations did not belong to the Eastern lineage but to the Western lineage with populations of Tarim Basin (southern Xinjiang). This result indicates that the European populations were not a direct ancestor of current populations in Northeastern Asia and North America. Figure 5 summarizes a conceivable migration history of C. elaphus using clusters defined in the present study. Combining our data with the previous reports on C. elaphus evolution (Geist, 1971, 1983; Groves and Grubb, 1987; Ohtaishi, 1992, 1995), it is likely that the Eastern lineage (Altai and Tianshan Mountains in northern Xinjiang, northeastern China, Mongolia, and North America) evolved from ancestral populations around the Tibetan plateau.

Fig. 5

Migration history of Cervus elaphus in Eurasia and North America. Clusters E1-8 and W1-2 were the same as in Fig. 3. ‘Origin’ means an assumed ancestor of C. elaphus.

Occurrence of the mtDNA lineage boundary within Xinjiang

Our results showed that the boundary between the Western and the Eastern lineages occurred between Tianshan Mountains and Tarim River within Tarim Basin of Xinjiang (Table 1; Figs. 1, 3, and 4). Green belts along Tarim River in Tarim Basin are current habitats of C. elaphus (Fig. 1). Desert areas lying north of Tarim River could have been geographic barriers between the Tianshan and Tarim populations, and genetically differentiated them. Genetic characteristics in Xinjiang populations could reflect their migration history in Middle Asia. When ancestral populations of C. elaphus immigrated into Tarim Basin from the south in Middle Pleistocene and adapted themselves to the habitat, Tianshan Mountains had been already covered with glacial sheets (Wen, 1994). Therefore, they might not be able to reach northern Xinjiang across glacial sheets, and might have stayed in Tarim Basin. Meanwhile, formation of deserts had started from eastern Tarim since the end of Early Pleistocene (Wen, 1994). Even after the glacial retreat in Holocene (Wen, 1994), Tarim populations could not cross Tianshan Mountains by geographic barriers of deserts formed between the Tarim River and Tianshan Mountains, and then they have been isolated in green belts developing along the river. As shown in Fig. 5, only populations having the Eastern lineage of mtDNA could have colonized northern Xingjian (Altai and Tianshan Mountains) from East Asia.