To investigate the phylogenetic relationships between the New World Sciurus and the Old World Sciurus and their biogeographic history, the partial mitochondrial cytochrome b gene sequences (1,040 base pairs) were analyzed on six Sciurus species: S. aberti, S. carolinensis, S. lis, S. niger, S. stramineus, and S. vulgaris. Phylogenetic trees (maximum parsimony, neighbor-joining, and maximum likelihood methods) commonly showed two groups with high bootstrap values (73–100%): one consisting of the New World Sciurus and the other consisting of the Old World Sciurus. Genetic distances among the New World Sciurus species were remarkably larger than that between two Sciurus species of the Old World, suggesting the earlier radiation of the New World Sciurus than the Old World Sciurus.
The genus Sciurus includes 27 extant squirrel species: 24 species of the New Continent and three species of the Old World (Corbet and Hill, 1991). These animals adapted themselves to the temperate forests, and are widely distributed in the northern parts of Eurasia, North America, Central America, and the northern to central parts of South America.
Yet despite the large amount of ecological information on Sciurus (e.g., Gurnell, 1987; Moncrief et al., 1993; Koprowski, 1996; Steele et al., 1998; Tamura, 1998; Lee and Fukuda, 1999), the phylogenetic relationships within this genus remain uncertain. From the paleontological study, Black (1972) described that the ancestral Sciurus could have been present already during the Miocene in European and North American Continents, and that some Sciurus species were also found in the early Pliocene in Spain and in Germany. Nadler and Sutton (1967) reported that the chromosomal constitutions are closely related within the North American species of Sciurus: S. carolinensis, S. niger, and S. aberti. In addition, the immunological analysis of serum albumin (Ellis and Maxson, 1980) and the study of protein variation (Hafner et al., 1994) strongly supported the close phylogenetic relationship between S. carolinensis and S. niger. Meanwhile, in the Eurasian species, the close phylogenetic relationship between S. lis and S. vulgaris was inferred from the 12S ribosomal RNA (rRNA) gene sequences (Oshida et al., 1996) and the chromosomal characteristics (Oshida and Yoshida, 1997). However, very few are known on the phylogenetic relationships among worldwide species of Sciurus.
In the present study, the phylogenetic relationships among six species of Sciurus from Asia and North and South America were examined based on the mitochondrial cytochrome b gene sequences. We here discuss the biogeographic history of Sciurus from the Eurasian Continent and the New Continent.
MATERIALS AND METHODS
Profile of squirrels examined in the present study is shown in Table 1. One specimen of S. stramineus was commercially obtained from a pet store in Japan. DNA sequence data of three North American species (S. aberti, S. carolinensis, and S. niger) previously reported by Thomas and Martin (1993) and Wettstein et al. (1995) were included for the present phylogenetic analysis (Table 1). Tamiasciurus hudsonicus was used for an out-group.
Squirrel species used in the present study
DNA preparation and sequencing
Total DNAs of S. lis, S. vulgaris, S. stramineus, and T. hudsonicus were extracted from muscle tissues with the phenol/proteinase K/sodium dodecyl sulfate method of Sambrook et al. (1989). A partial region (1,040 base pairs, bp) of the mitochondrial cytochrome b gene was amplified with polymerase chain reaction (PCR), using a set of primers described by Oshida et al. (2000): L14724 5′-GATATGAAAAACCATCGTTG-3′ and H15910 5′-GATTTTTGGTTTACAAGACCGAG-3′. Primer names correspond to the light (L) or heavy (H) strand and the 3'end-position of the primers in the human mitochondrial DNA (mtDNA) sequences (Anderson et al., 1981). The 50 μl of reaction mixture contained 100 ng of genomic DNA, 25 picomoles of each primer, 200 μM dNTPs, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, and 2.5 units of rTaq DNA polymerase (Takara). Amplification was carried out for 35 cycles and the program was 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min, and then the extention reaction was performed at 72°C for 10 min. PCR products were purified with the Qia-quick PCR purification kit (QIAGEN) and directly sequenced using an automated DNA sequencer (SQ5500L, Hitachi).
All sequences were aligned using a computer software Gene Works (Intelligenetics). The phylogenetic trees were constructed with the maximum parsimony (MP) method using the branch and bound search algorithm (Hendy and Penny, 1982) with the 50% majority-rule consensus in PAUP (Swofford, 1993), with the neighbor-joining (NJ) method (Saitou and Nei, 1987) in Clustal W (Thompson et al., 1994), and with the maximum likelihood (ML) method using DNAML in PHYLIP package program (Felsenstein, 1993). In NJ and ML methods, numbers of nucleotide substitutions per site were estimated for multiple substitutions by the Kimura's (Kimura, 1980) two-parameter method. MP tree was produced by unweighted parsimony. To assess the branching confidence, bootstrap values (Felsenstein, 1985) were derived from 1,000 replications in MP and NJ methods and 100 replications in ML method.
To estimate the divergence time between Sciurus species, the transversional substitution rate (0.5% / million years, Myr) at the third codon positions of mammalian cytochrome b gene (Irwin et al., 1991) was employed.
The partial sequences (1,040 bp) of the cytochrome b gene were successfully determined for two S. lis, three S. vulgaris, one S. stramineus, and one T. hudsonicus. Percentage differences corrected by Kimura's two-parameter model and numbers of transitions and transversions obtained from pairwise comparison are shown in Table 2. The transversional substitutions at the third codon positions (Table 3) were used for estimation of the divergence time.
Pairwise comparison of cytochrome b nucleotide sequences (1,040 bp) between ten squirrel specimens
Pairwise comparison of transversional substitutions at the third codon positions of cytochrome b gene between nine specimens of Sciurus
Maximum parsimony (MP), neighbor-joining (NJ), and maximum likelihood (ML) analyses yielded similar branching in trees, all of which contained the same two major groups with high bootstrap values: the New World groups consisting of S. aberti, S. carolinensis, S. niger, and S. stramineus (84% in MP tree of Fig. 1a, 80% in NJ tree of Fig. 1b, and 73% in ML tree of Fig. 1c); the Old World groups consisting of S. lis and S. vulgaris (100% all in MP, NJ, and ML trees, Fig. 1). In MP analysis, only one most-parsimonious phylogenetic tree was obtained by unweighted parsimony, with a consistency index of 0.688.
Although the groups consisting of four Sciurus species from the New World were supported with high bootstrap values (73–84%), the phylogenetic relationships within the New World Sciurus analyzed in the present study were unclear because of the polychotomy found in MP tree (Fig. 1a) and the low bootstrap values on NJ tree (51–77% in NJ tree, Fig. 1b) and ML tree (32–40% in ML tree, Fig. 1c). Genetic distances were 14.7 –19.8% among the New World Sciurus (Table 2). On the other hand, the monophyly of two Old World Sciurus species was clearly shown with very high bootstrap values (100% all in MP, NJ, and ML trees, Fig. 1) and the genetic distances between two species were 4.9–6.8% (Table 2). In addition, genetic distances between the New World Sciurus and the Old World Sciurus were 18.0–21.3% (Table 2).
Phylogeny of the New World Sciurus
Sciurus aberti, S. carolinensis, and S. niger are currently distributed in the North American Continent, while S. stramineus occurs in the South American Continent. Although the distribution area of S. stramineus is separated from those of the former three species, the phylogenetic position of S. stramineus was included in their group (Fig. 1) and did not reflect the geographically expected position. Ellis and Maxson (1980) and Hafner et al. (1994) described the close relationship between S. carolinensis and S. niger, based on the serum albumin and the protein variation, respectively. Our results agreed with their opinions. Moreover, genetic distances among species in NJ and ML trees and the polycotomy in MP tree suggest that the radiation of Sciurus might have explosively occurred in the New World (Fig. 1) Based on the fossil records, Black (1972) reported that the ancestral Sciurus had already been distributed in North America during the Miocene. Kurtén and Anderson (1980) described that four species of Sciurus (S. alleni, S. arizonensis, S. carolinensis, and S. niger) were identified in Pleistocene deposits of the New World. From the transversional substitutions at the third codon, the divergence time among the New World Sciurus was estimated to be approximately 9.8–14.4 Myr ago. Therefore, the radiation of the New World Sciurus might have occurred during the Miocene.
Phylogeny of the Old World Sciurus
Sciurus vulgaris is widespread throughout the northern parts of the Eurasian Continent and this species is divided into several subspecies (Sidorowicz, 1971; Wilson and Reeder, 1993). However, the phylogenetic relationships among sub-species have not ever been studied. In three specimens of S. vulgaris examined here, one individual from Russia (SVU3) seemed to be closer to S. v. koreae from Korea (SVU2) than to S. v. orientis, which is endemic to Hokkaido in Japanese islands (SVU1), with high bootstrap values (89% in MP tree, Fig. 1a; 84% in NJ tree, Fig. 1b; 91% in ML tree, Fig. 1c). Ohshima (1990, 1991) suggested that Hokkaido was separated from the Eurasian Continent and Sakhalin during the late Pleistocene via the formation of the straits. Sciurus v. orientis may be a population which has been geographically isolated in Hokkaido since then. However, so as to consolidate this hypothesis, further analysis of geographic variation of S. vulgaris is required.
The phylogenetic relationship and classification between S. vulgaris and S. lis are controversial. S. lis is endemic to Honshu, Shikoku, and Kyushu islands of Japan (Corbet and Hill, 1991). Imaizumi (1960) classified S. lis as an independent species on the basis of differences in tail hair color, body size, and cranial characteristics, while Oshida et al. (1996) pointed out that the sequence difference of mitochondrial 12S rRNA gene between S. lis and S. vulgaris corresponded to intraspecific differences of the genera Petaurista and Tamias. In addition, Oshida and Yoshida (1997) reported the karyotypic similarity between S. lis and S. vulgaris. In the present study, genetic distance between two species was 4.9–6.8% (Table 2), and referred to interspecific differences of other squirrel genera, Glaucomys (Arbogast, 1999) and Petaurista (Oshida et al., 1999). Since intraspecific cytochrome b differences of other squirrels reported heretofore are <3.0% (Wettstein et al., 1995; Arbogast, 1999; Oshida et al., 2000), our results of cytochrome b are not discordant with that S. lis and S. vulgaris are regarded as an independent species. In general, the substitution rate of cytochrome b gene is more rapid than that of 12S rRNA gene (Irwin et al., 1991). The substitution rate differences might be responsible for the conflict between the 12S rRNA phylogeny (Oshida et al., 1996) and the cytochrome b phylogeny of the present study. Our results also indicate that cytochrome b is a more suitable marker to perceive the phylogenetic relationships between closely related squirrel species.
Based on fossil records, Kawamura (1988, 1990) and Kawamura et al. (1989) considered that an ancestral S. vulgaris immigrated from Hokkaido to Honshu through the land bridge formed during the middle Pleistocene, and then S. vulgaris had diverged to S. lis. However, in the present study, divergence time between the two species estimated using the transversional substitution rate at the third codon positions of mammalian cytochrome b gene was approximately 4.0–5.2 Myr ago. Accordingly, the divergence period between the two species may be earlier than the estimation of Kawamura (1988, 1990) and Kawamura et al. (1989).
Phylogenetic relationships between the New World and the Old World Sciurus
The present study revealed the phylogenetic differentiation of Sciurus between the New World and the Old World (Fig. 1). Fossils of the ancestral Sciurus are found in the deposits of the Miocene in Europe and in North America (Black, 1972). In addition, Sciurus species are seen in the early Pliocene in Europe (Mein, 1970; Black, 1972). Although the origin of the genus Sciurus is currently ambiguous, judging from the information about fossil remains, it is obvious that ancestral Sciurus were widespread in the New World and in the Old World during the Miocene. At present, the range of the genus Sciurus are absolutely separated by the Bering Strait. Recently, based on the paleontological records, Marincovich Jr. and Gladencov (1999) suggested that the first opening of the Bering Strait might have occurred between 4.8 and 7.3–7.4 Myr ago. The divergence time (approximately 22.4–26.6 Myr ago) between the two Sciurus groups estimated in the present study was older than the first opening time of the Bering Strait reported by Marincovich Jr. and Gladencov (1999). Therefore, it may be reasonable that the geographic isolation by the opening of the Bering Strait was not the main cause of the divergence between the two Sciurus groups. However, it is likely that the absolute geographic isolation by the opening of the Bering Strait during the Pliocene or the Miocene influenced secondarily the independent evolution of the two Sciurus groups.
We would like to thank Mr. F. Sekiyama (Iwate Prefectual Museum), Dr. H. Yanagawa (Obihiro University of Agriculture and Veterinary Medicine), Dr. K. Murata (Kobe Municipal Oji Zoo), Dr. A. V. Abramov (Zoological Institute, Russian Academy of Sciences), Awaji Farm Park, and Noboribetsu Bear Park for supplying specimens. This study was partly supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture, Japan.