The control region phylogeny

There were 42 variable sites out of 266 bp within brown bears (Table 1). All substitutions were transitions except one transversion. Percentage sequence differences are shown in Table 2. Thirteen haplotypes of the mtDNA control region (Table 1) were compared to reconstruct a phylogenetic tree among them (Fig. 1a). The four major clades emerging in the phylogeny were named Groups I-IV in the present study, although Groups I and II were supported with only 67% and 54% bootstrap values, respectively.

Table 1

Positions of the nucleotide substitution in the control region partial sequence (284 bp).

Table 2

Percentage of pairwise differences of mtDNA control region (266 bp)

Fig. 1

Molecular phylogenetic trees (a, 266 bp of the control region; b, 700 bp of Cyt. b) of Alaskan, European, and Hokkaido brown bears using the neighbor-joining method. Gray, open, and closed circles show the haplotypes from Europe, Alaska or eastern Siberia, and Hokkaido, respectively. Numbers on the branches are bootstrap values (%) derived from 1,000 replications, while values under 50% were not shown.

European brown bears fell into two different clades: Group I (the western European lineage comprising Pyr and Gre) and Group II (the eastern European lineage comprising Ro1 and Rus1) as reported by Taberlet and Bouvet (1994). We confirmed that the two lineages of Europe arose even when all the haplotypes reported by Taberlet and Bouvet (1994) and by Kohn et al. (1995) were applied to the phylogenetic tree reconstruction (data not shown). Therefore, in the present study, the four brown bears (Pyr, Gre, Ro1, and Rus1) were selected as representatives of the two lineages of the European brown bears. Meanwhile, the brown bears of Hokkaido were divided into three lineages: HK1-D and HK2-D with a 72% bootstrap value in Group II, HK3-D and HK4-D with a 94% bootstrap value in Group III, and HK5-D and HK6-D with a 90% bootstrap value in Group IV, which correspond to Clusters A, B, and C reported by Matsuhashi et al. (1999), respectively, and these three lineages were distributed in central, eastern, and southern Hokkaido, respectively. Two individuals were selected as representatives from each of the three lineages of the Hokkaido brown bears. In the Group II, the brown bears of Europe were not intermixed with the Hokkaido brown bear group (Fig. 1a). The Kodiak brown bear (KD1) was included in the Group II, standing near the eastern European brown bear (Rus1, Fig. 1a). Thus, the Group II consisted of the brown bears of eastern Europe, Kodiak, and central Hokkaido, though the bootstrap value is low (54%). The Group III comprised only eastern Hokkaido brown bears. The Tibetan brown bear was clustered with the southern Hokkaido group (Group IV) with a 91% bootstrap value. The Gobi brown bear (Gobi-D) and the western European group (Group I) formed a cluster with a 50% bootstrap value, which was distantly related to the other brown bears (Fig. 1a).

The Cyt. b phylogeny

Twelve haplotypes of Cyt. b (700 bp) obtained from 15 brown bears (six bears from Alaska, six from Hokkaido, one from Gobi, two from Tibet) and one polar bear Ursus maritimus (Table 3) were compared by constructing the neighbor-joining phylogenetic tree (Fig. 1b). A total of 54 variable sites (including two transversions) at all codon positions of Cyt. b (700 bp) and 37 variable sites at the third codon positions (235 positions) among 12 haplotypes were identified (Table 3). Percentage sequence differences at the third codon positions are shown in Table 4.

Table 3

Substitution sites of the third codon positions in the partial portion of Cyt. b (700 bp).

Table 4

Percentage of pairwise nucleotide differences at the third codon positions in Cyt. b(235 positions/700 bp). Above the diagonal: Both transitions and transversions were counted for calculation of divergence time using the method of Irwin et al. (1991); below diagnal: only transitions were counted for estimation of divergence time using the method of Talbot & Shields (1996).

Hokkaido brown bears were divided into three lineages: (1) HK1-C and HK2-C (both from central Hokkaido; their sequences were identical in Group Y); (2) HK3-C and HK4-C (both from eastern Hokkaido; sequences were identical in Group Z); (3) HK5-C and HK6-C (both from southern Hokkaido; sequences were identical in Group W), indicating the same phylogenetic relationships among groups as for the control region. Brown bears from Alaska and eastern Siberia were divided into three groups: (1) GB01 and GB04, both from ABC Islands of Alaska (Group X); (2) GB19 from Kodiak and GB27 from eastern Siberia (Group Y); (3) GB09 and GB10, both from eastern Alaska (Group Z). Haplotypes (GB01 and GB04) in Group X were closer to U. maritimus with a 95% bootstrap value. Both Groups Y and Z embraced subpopulations from Alaska and Hokkaido: bears of central Hokkaido (HK1 and 2-C) formed Group Y with Kodiak bears of western Alaska (GB19) and eastern Siberia (GB27) with a 95% bootstrap value; bears of eastern Hokkaido (HK3 and 4-C) were included in Group Z together with those of eastern Alaska (GB09 and GB10) with a 89% bootstrap value. Group W (HK5 and 6-C: bears of southern Hokkaido) split from Groups Y and Z. Two bears of Tibet (Tibet-1 and 2) had identical sequences (Tibet-C). Brown bears of Gobi (Gobi-C) and Tibet (Tibet-C) were remote from both Hokkaido and Alaskan populations.

The sequence data of the control region and Cyt. b determined in the present study will appear in the DDBJ, EMBL, and GenBank nucleotide sequence databases with the following accession numbers: the control region sequence of KD1 ( AB041258); Cyt. b sequences of Tibet-C ( AB041259) and Gobi-C ( AB041260).

Relationships between phylogeny and geographic origins

Each population from Europe, Alaska, and Hokkaido was clearly split into two or three mtDNA lineages (Taberlet and Bouvet, 1994; Kohn et al., 1995; Talbot and Shields, 1996b; Waits et al., 1998; Matsuhashi et al., 1999). As shown by Taberlet and Bouvet (1994) and by Kohn et al. (1995), brown bears of western Europe were genetically distant from those of eastern Europe. Brown bears of Alaska were genetically separated in three regions: ABC Islands, western Alaska including Kodiak Island, and eastern Alaska, while the western Alaskan population is very close to the eastern Siberian population (Talbot and Shields, 1996b). In Japan, there were three lineages distributed in eastern, southern, and central Hokkaido (Matsuhashi et al., 1999). The present study suggested that some groups located in distant areas were phylogenetically closer to each other than to geographically neighboring groups. For instance, bears of eastern Europe, Kodiak Island, and central Hokkaido were classified into Group II (Fig. 1a) in the control region phylogenetic tree, though the bootstrap value (54%) is low. On the other hand, bears of western Alaska including Kodiak Island, eastern Siberia, and central Hokkaido were classified into Group Y (Fig. 1b) in the Cyt. b tree. The Cyt. b data of eastern European brown bears were not included because they were unavailable in the present study. These results suggest that Group II (control region tree, Fig. 1a) and Group Y (Cyt. b tree, Fig. 1b) form a common lineage. Brown bears included in this lineage could have been widespread in eastern Europe, Siberia, central Hokkaido, and western Alaska before the final formation of the Bering Strait and the Soya Strait (located between Hokkaido and Sakhalin).

Brown bears of both eastern Alaska and eastern Hokkaido were classified into Group Z (Cyt. b tree, Fig. 1b), suggesting that they are also descendent from a common ancestral lineage although they currently inhabit separate and distant areas (Fig. 2).

Fig. 2

Worldwide distribution of eight groups (I-IV, X-W) in brown bear populations.

The control region tree (Fig. 1a) showed that the Tibetan brown bear (Tibet-D) is genetically closer to southern Hokkaido brown bears (Group IV), whereas the Cyt b tree (Fig. 1b) did not support it (Tibet-C - Group W). The incongruity between the two phylogenetic trees (Fig. 1a and 1b) might be attributed to the shortage of the length of the control region sequences. Another possibility is that the substitution rates of the control region and Cyt. b may be different from one another. For example, at the control region, the sequence difference (excluding indels) between HK1-D and HK6-D was 3.76% and that between HK6-D and Selenarctos thibetanus (outgroup) was 12.03% (Table 2). Meanwhile, at the Cyt. b gene (transitions at the third codon positions), the pairwise difference between HK1 and 2-C and HK5 and 6-C was 2.56%, but that between HK5 and 6-C and S. thibetanus was 24.79%.

In the Cyt. b tree (Fig. 1b) in the present study, the position of Ursus maritimus was within the brown bear cluster, forming a sister clade to the population of ABC Islands (Group X), as reported by Talbot and Shields (1996a, b) and Shields and Kocher (1991). Fig. 2 shows the worldwide distribution of groups identified from mtDNA lineages.

Divergence time of the brown bear populations

To reconstruct migration history of brown bear populations, the divergence time was calculated from control region sequence differences (Table 3) by using the sequence divergence rate of the homologous human control region sequence (approximately 8.4% difference per million years, Myr) (Vigilant et al., 1989). Divergence times were as follows: 0.8 Myr between Gobi + Group I and the others; 0.7 Myr between Gobi and Group I (western Europe); 0.5 Myr between Group IV + Tibetan bear and Groups II + III; 0.3 Myr between Groups II and III; and 1.3 Myr between Ursus arctos and Selenarctos thibetanus.

The divergence time was also calculated from the Cyt. b data. Irwin et al. (1991) reported a rate of approximately 10%/Myr for nucleotide substitutions at the third codon positions of Cyt. b. Talbot and Shields (1996a) calculated nucleotide substitution rate with fossil records and indicated that transitions at the third codon positions of the ursid Cyt b gene occur at an approximately 6% divergence per Myr. Using these two divergence rates and the Cyt b data obtained in the present study (Table 4), divergence times were estimated as follows (when using the rates of 10%/6% per Myr): 0.8/1.2 Myr between Gobi and others; 0.6/0.9 Myr between Group X + Tibet and Hokkaido + the other Alaskan brown bears; 0.3/0.5 Myr between Group W and Groups Y + Z; 0.2/0.3 Myr between Groups Y and Z; 0.2/0.3 Myr between Group X and U. maritimus; and 2.4/4.0 Myr between U. arctos and S. thibetanus.

Mazza and Rustioni (1994) suspected that some brown bears in Asia immigrated into Europe at the end of the Early Pleistocene. Taberlet and Bouvet (1994) suggested that the western and eastern Europe lineages diverged 0.85 Myr ago. Waits et al. (1998) estimated 0.11–0.14% divergence (Kimura's two-parameter distance) of the control region per 0.01 Myr, from the data of Talbot and Shields (1996b). Using this divergence rate, Waits et al. (1998) estimated the divergence time as follows: 0.146–0.185 Myr between U. maritimus and the ABC Island brown bear (the same population as Group X in the present study); 0.245–0.31 Myr between western (Group Y) and eastern Alaskan bears (Group Z); 0.28–0.356 Myr between western Alaskan (Group Y) and Rocky Mountain brown bears; and 0.404–0.515 Myr between eastern Alaskan (Group Z) and Rocky Mountain brown bears. These values were relatively close to our estimates, although we did not examine any sequences of Rocky Mountain brown bears.

From the estimated divergence time, each of the four lineages (Gobi, Group I, Group X, and other groups) classified in the present study are likely to have diverged from the others approximately 0.6–0.9 Myr ago. These ages may relate to the early stage of the evolution of the brown bear. The ancestral lineage of Group II/Group Y could have been distributed in the extensive area ranging from eastern Europe to eastern Siberia, and some populations from this lineage could have crossed the land bridge (Beringia) between Siberia and Alaska and immigrated into North America. Other populations from this lineage could have migrated through Sakhalin to Hokkaido. This presumption does not conflict with Kurten's (1968) report on the basis of the fossil record that the brown bear could have existed in North America at least 0.5 Myr ago. Thus, our results indicate that some minor (probably relict) lineages of mtDNA likely occur in western Europe, ABC Islands of Alaska, Gobi of Mongolia, Tibet, and Hokkaido (Fig. 2), which are considered to be refugia of brown bears during the last glacial age of Pleistocene. Further studies involving more data of the same DNA sequence regions on worldwide brown bear populations will more fully illustrate their origin and migration history.