INTRODUCTION

The mammalian species observed on the earth have derived from the primitive shrew-like small beasts. Among their descendants there are four groups that are believed to share a close relationship each other among eutherians. They are the Insectivores, the Macroscelideans, the Scandentians, and the Chiropterans. Moreover, the Dermoptera and the Primates are said to be of insectivore origin (Colbert and Morales, 1991). However, it is very difficult to resolve their phylogenetic relationships only from the morphological characteristics and/or the fossil evidences, because of poor fossil record. In adition, although the insectivoran order is generally divided into three suborder, the relationships among these suborders remain uncertain due to inconsistencies in morphological traits.

In order to investigate the intraordinal relationships of Insectivora among primitive placental groups, we sequenced the cytochrome c oxidase subunit II (COII) genes of mtDNA at first, since this gene has been used extensively in the investigation for systematic relationships within and among mammalian orders. Although the nucleotide sequence of animal mtDNA is useful to investigate relationships among closely related species because of its high evolutionary rate, the amino acid sequence encoded by mtDNA can be also useful to probe deep branchings because of its slower rate than the rate at a nucleotide level. The other advantage of using mtDNA in a phylogenetic work is that we are relatively free from the danger of comparing paralogous genes, as may sometimes be the case in nuclear genes (Cao et al., 1994).

In our study we used three insectivoran species, musk shrew (Suncus murinus), belonging to the family Soricidae, and shrew mole (Urotrichus talpoides) and mole (Mogera wogura), belonging to the family Talpidae. Mole and shrew mole are specialized for underground life and they have the shovel-like limbs for digging soil and their visual organs are covered with the skin. On the other hand, musk shrew has slender limbs with five toes and small normal visual organs. Then we compared new amino acid sequences from these three insectivora with those of the other mammalian orders, with special attention to the other insectivoran species, the order Chiroptera and the order Primates.

MATERIALS AND METHOD

Data analysis

Raw sequence data were analyzed with the software DNASISMac (Hitachi Software Engineering). We chose the human sequence (Horai et al., 1992) as a reference sequence. Amino acid sequences encoded by mtDNA are particularly useful to probe deep branches, because selection (mainly negative selection in the framework of the neural theory; Kimura, 1983) is likely to be operating stronger on an amino acid site than on a nucleotide site (Kocher et al., 1989). Therefore, we used amino acid sequence of the COII gene for analysis, together with the published data (Table 1).

Table 1

COII gene sequences included in this study

In analyzing the COII amino acid sequence data of protein, we used both neighbour-joining method (Saitou and Nei, 1987) and maximum likelihood method (Felsenstein, 1981). As for the case of the NJ method, we used the MEGA program (Kumar et al., 1993) with a gamma distance a = 0.7 (Nei et al., 1976). For maximum likelihood analysis, we used the PROTML program in MOLPHY package, ver 2.3 (Adachi and Hasegawa, 1996).

For the PROML analyses, the mt REV-F models for mtDNA-encoded protein (Adachi and Hasegawa, 1996) were used for the these data. Since it seems very likely that each of Primates, Ferungulata [Artiodactyla, Cetacea and Carnivora (Cao et al., 1994)], Chiroptera, Insectivora and Rodentia forms a monophyletic group, we add arumadillo into these 5 groups and examined all of the 945 possible trees among these 6 major lineage of eutheria. We estimated the bootstrap probability by the RELL (resampling of the estimated loglikelihood) method by Kishino et al. (1990). Although the sequence data of goldenmole and hedgehog, which are assumed to have been also the member of Insectivora, were available in a public data-base, both of them were excluded in our analyses from the following reasons. Goldenmole is suspected in a recent molecular study (Springer et al., 1997) to be in a clade that contains members of presumed African origin and its phylogenetic position in Insectivora is also uncertain with morphological uncertainties (Van Valen, 1967; Eisenberg, 1981). On the other hand, hedgehog shows an higher levels of compositional bias in nucleotide bases of the overall mtDNA genes.

The branching orders within each of major lineage of eutherian and within the outgroup were assumed as the consensus of previous works (Cao et al., 1994; D'Erchita et al., 1996; Bulmer et al., 1991; Li et al., 1990; Szalay, 1977; McKenna, 1975; Van Valen, 1967; Mindell et al., 1991; Adkins and Honeycutt, 1991). The species and their assumed within-group relationships are Primates: (human, orangutan, siamang, rhesus monkey, green monkey, mantled howler, Humboldt's wooly monkey, ring-tailed lemur, fat-tailed mouse lemur, slow loris, bushbaby); Insectivora: ((short-tailed shrew, musk shrew), (shrew mole, Japanese mole)); Ferungulata: ((harbor seal, grey seal) dog, cat) ((goat, (cow, blue whale)), horse); Rodentia: (rat, mouse); Chiroptera: (Leschenault's rousette, greater spear-nosed, horse shoe); and outgroup: ((wallaroo, opossum), platypus).

RESULTS AND DISCUSSION

The nucleotide sequences of cytochrome c oxidase subunit II gene and the amino acid sequences are shown in Fig. 1 and in Fig. 2, respectively, for three insectivoran species and human.

Fig. 1

Nucleotide sequence of the cytochrome c oxidase subunit II (COII) gene from the three insectivoran species along with the human sequence (Anderson et al., 1981). Numbering is according to the human sequence. The predicted amino acid sequence of the human protein is shown above by one-letter code. Nucleotides identical to those of human are indicated by dashes. human = Homo sapiens, musk shrew = Suncus murinus, shrew mole = Urotrichus talpoides, mole = Mogera wogura.

Fig. 2

Amino acid sequence of the cytochrome c oxidase subunit II gene along with the human sequence (Anderson et al., 1981). human = Homo sapiens, musk shrew = Suncus murinus, shrew mole = Urotrichus talpoides, mole = Mogera wogura.

The phylogenetic tree based on NJ analysis is shown in Fig. 3 and the result is generally consistent with the morphological and molecular phylogenetic trees already reported (Honeycutt et al., 1995; Johnson et al., 1994), but shows some interesting differences.

Fig. 3

The neighbor-joining tree of amino acid sequence of COII gene. Bootstrap values 40% or greater (derived from 1000 replicates) are shown along branches. The horizontal length of each branch is proportional to the estimated number of amino acid substitutions.

As is expected, the close relationships between mole and shrew mole and between musk shrew and short-tailed shrew are observed with high bootstrap probabilities (97% in both cases) and the results are in accord with the traditional relationship by obtained morphological data (Butler, 1988; Szalay, 1977; McKenna, 1975; Van Valen, 1967). The NJ tree, however, have complicated the issue of fixing Insectivoran position among major orders, with Chiroptera as many as three independently evolving lineages observed (Fig. 3). The bootstrap probabilities of [(short-tailed shrew, musk shrew), greater spear-nosed bat] clad and {horse shoe bat, [Leshenult's rousette,(shrew mole, Japanese mole)} clade are only 27% and 20%, respectively, which suggest the resolving power of the present data would not be enogh to distinguish markedly these groups. This, in fact, may suggest that the relationship between Insectivora and Chiroptera may be very close. The close relationship of armadillo to Primates is also not in accord with the evidence from the complete mtDNA data of armadillo (Arnason et al., 1997).

Despite the predominantly divergent nature of most molecular evolution, parallel and convergent evolution sometimes occur in the specific genes (Stewart et al., 1987), and such a discrepancy could occur in 5% of the cases without any particular reason if the divergence among groups in question occurred within a short period.

Furthermore, the suborder Anthropoidea seems to be too distant from the other eutherians in Fig. 3. This may be probably due to the higher level of variation and the higer rate of amino acid replacement in this lineage.

One of our purpose is to see the degree of interrelationship among the extant eutherian orders and to seek the most close relative of Insectivora. Therefore, we examined the data by maximum-likelihood (ML) method (Kishino et al., 1990; Adachi and Hasegawa, 1992) for protein sequences with known consensus relationship of eutherians. The branching orders within each of the Primates, Ferungulata (Artiodactyla, Cetacea, Perissodactyla), Insectivora, Chiroptera, Rodent, and outgroup are assumed as the consensus of the several morphological and molecular studies. Figure 4 shows the phylogenetic relationship among the seven major eutherian groups constructed from COII data alone by the PROTML. Although the bootstrap probability is not high, the ML tree with the highest boot strap probability (45.7%) locates Chiroptera closer to Insectivora than to the other eutherian orders.

Fig. 4

The maximum likelihood tree of amino acid sequence of COII gene. The Prot ML program with the mtREV-F model was applied. The horizontal length of each branch is proportional to the estimated number of amino acid substitutions.

From both NJ and ML phylogenetic analyses, the following interesting results were obtained. Firstly, the most closest relative of Insectivora seems to be the order Chiroptera. Secondly, there would be a sister-group relationship between Ferungulata and (Insectivora + Chiroptera).

Novacek (1986, 1989), from the analyses on skull characters, suggested the possibility that Insectivora may be part of a clade that includes Tublidentates and Carnivorans. Miyamoto and Goodman (1986), from the analyses on amino acid sequence changes in several proteins, suggested that Insectivora may be most closely related to Carnivorans and Pholidotans. But these evidences supporting each association are admittedly weak. The insectivoran position has been not clear in eutherian cladgram yet. The COII gene trees obtained by the present study do not support either Novacek or Miyamoto and Goodman's hypothesis, and radically contradict the traditional view, which says that Chiroptera would be closely associated with Primates, flying lemur and tree shrew in the superorder Archonta (Novacek et al., 1988 and Szalay, 1977).

We consider it appropriate to use molecular data bacause it seems that there are no recognizably non-primitive traits of Insectivora to distinguish them from other eutherians. Since analyses of individual genes could lead to an erroneous tree with an apparently significant confidence level, to avoid such possibility, we will carry out further analyses based on as many different genes as possible and synthesize the results.