Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://vertebrata.bio.msu.ru/Alticola_zsc_.pdf Äàòà èçìåíåíèÿ: Mon Feb 8 19:40:44 2010 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 19:26:37 2012 Êîäèðîâêà:

Molecular phylogeny of the genus Alticola (Cricetidae, Rodentia) as inferred from the sequence of the cytochrome b gene
Blackwell Publishing Ltd

V

LADIMIR

S. L

EBEDEV,

A

NNA

A. B

ANNIKOVA,

A

LEXEY

S. T

ESAKOV

&N

ATALIA

I. A

BRAMSON

Submitted: 29 January 2007 Accepted: 5 August 2007 doi:10.1111/j.1463-6409.2007.00300.x

Lebedev, V. S., Bannikova, A. A., Tesakov, A. S. & Abramson, N. I. (2007) Molecular phylogeny of the genus Alticola (Cricetidae, Rodentia) as inferred from the sequence of the cytochrome b gene. -- Zoologica Scripta, 36, 547­563. Central Asian mountain voles Alticola is one of the least known groups of voles both in evolution and life history. This genus includes three subgenera Alticola s.str., Aschizomys and Platycranius, and belongs to the tribe Clethrionomyini comprising also red-backed voles Clethrionomys and oriental voles Eothenomys. In order to elucidate the phylogenetic relationships within Alticola and to examine its position within the tribe, mitochondrial cytochrome b (cyt b) gene variation was estimated, and the results were compared with morphological and palaeontological data. Maximum likelihood (ML), neighbor-joining (NJ), maximum parsimony (MP) and Bayesian phylogenetic analyses show that the genus Alticola does not appear to be a monophyletic group since the representatives of Aschizomys branch within Clethrionomys, whereas two other subgenera (Alticola and Platycranius) form a separate monophyletic clade. Flat-headed vole Alticola (Platycranius) strelzowi is nested within the nominative subgenus showing close association with A. (Alticola) semicanus. Surprisingly, the two species of Aschizomys do not form a monophyletic group. The results of the relaxed-clock analysis suggest that the Alticola clade splits from the Clethrionomys stem in early Middle Pliocene while basal cladogenetic events within Alticola s.str. dates back to the late Middle to early Late Pliocene. A scenario of evolution in Clethrionomyini is put forward implying rapid parallel morphological changes in different lineages leading to the formation of Alticola-like biomorphs adapted to mountain and arid petrophilous habitats. Corresponding author: Vladimir S. Lebedev, Zoological Museum, Moscow State University, B. Nikitskaya 6, 125009 Moscow, Russia. E-mail: wslebedev@hotmail.com Anna A. Bannikova, Lomonosov Moscow State University, Vorobievy Gory, 119992 Moscow, Russia. E-mail: hylomys@mail.ru Alexey S. Tesakov, Geological Institute RAS, Pyzhevsky 7, 119017 Moscow, Russia. E-mail: tesak@ginras.ru Natalia I. Abramson, Zoological Institute RAS, Universitetskaya nab. 1, 199034 St Petersburg, Russia. E-mail: lemmus@zin.ru

Introduction
Previous data on phylogeny and taxonomy of Alticola The Central Asian high mountain or rock voles Alticola Blanford, 1881 is still one of the least known groups of voles both in evolution and life history. These elegant rodents featured by silver-grey pelage, long vibrissae, rootless molars and angular skull shape are well adapted to life in high altitude stony habitats. They occur in mountain areas of Central and North Asia from Chukotka Peninsula in the north-east to Kugitang Range in the west, and to the Tibet and Himalayas in the south (Musser & Carleton 2005; Shenbrot & Krasnov 2005; Fig. 1). The genus is attributed to the tribe Clethrionomyini

(Arvicolinae, Cricetidae, Rodentia), which also includes red-backed voles Clethrionomys and oriental voles Eothenomys. Alticola is currently thought to comprise 12 species (Pavlinov 2003; Musser & Carleton 2005) grouped into three subgenera: Alticola s.str., Platycranius Kastschenko, 1899 and Aschizomys Miller, 1899. Alticola s.str. is the most widespread and speciose group within the genus [see Rossolimo & Pavlinov (1992) for revision]. Monotypic Platycranius is remarkable for its unusual flattened skull. Aschizomys is the least known form and certain controversies exist concerning its content and taxonomic position. In general, this subgenus is believed to contain two allopatric species: Alticola (Aschizomys) lemminus
547

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Fig. 1 Distribution of subgenera of the genus Alticola [after Shenbrot & Krasnov (2005) with modifications] and collection localities for specimens used in the study. Solid circles, open circles and triangles correspond to Aschizomys, Platycranius and Alticola s.str., respectively. Numbers refer to the collection sites as given in Table 1. Asterisks denote the sites for A. (Aschizomys) lemminus (Cook et al. 2004).

and A. (Aschizomys) macrotis (Musser & Carleton 2005), although sometimes the former is regarded as a subspecies of the latter (Gromov & Polyakov 1977). In contrast to that, some authors (Ognev 1950) recognized them as separate species considering A. lemminus as the sole representative of Aschizomys and attributing A. macrotis to the subgenus Alticola s.str. Until now the phylogeny of the genus is unclear and its position within the tribe Clethrionomyini remains problematic. The available morphologic evidence concerning relationships among Aschizomys, Alticola s.str and Clethrionomys is contradictory and permits to question the monophyly of the genus Alticola (Gromov & Polyakov 1977). While some authors suggested affinity of Aschizomys to Clethrionomys (Hinton 1926) others rejected this hypothesis and supposed that the former taxon is rather close to Alticola s.str. (Vinogradov 1927). Affinities between Aschizomys and Eothenomys or Phaulomys were discussed as well ( Jameson 1961; Gromov & Polyakov 1977; Gromov & Erbajeva 1995). The issue is further complicated by the possibility that, in its turn, the genus Clethrionomys may be non-monophyletic in respect to Alticola as follows from several genetic studies that provide evidence of isolated position of the subgenus
548

Craseomys (comprising Clethrionomys rufocanus and related species). Thus, according to DNA­DNA hybridization studies (Gileva et al. 1990), both Clethrionomys and Alticola appeared to be polyphyletic with A. lemminus joining Craseomys, and A. argentatus being closer to Clethrionomys s.str. In contrast to that, G-banding data support the alliance of all subgenera of Alticola and Clethrionomys s.str. to the exclusion of Craseomys (Agadjanian & Yatsenko 1984). An important potential synapomorphy shared by the former two taxa is the presence of 1/9 reciprocal translocation contrasting the primitive condition of these chromosomes found in C. rufocanus and C. andersoni (Modi & Gamperl 1989). A similar phylogenetic pattern is suggested by the results of allozyme analysis (Mezhzherin & Serbenyuk 1992; Mezhzherin et al. 1995) indicating that Clethrionomys s.str. and Craseomys are more distant from each other than the former is from A. argentatus. The inferred position of A. lemminus as a sistergroup to A. argentatus was considered by the latter authors as preliminary due to small number of loci studied in this species. Thus, the results of previous genetic studies suggest nonmonophyly of Alticola and/or Clethrionomys s.l.; however, they contradict each other in details whereas the levels of support

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

for conflicting patterns remain undeterminable. Despite the considerable success of molecular studies towards understanding the evolutionary relationships of species and genera within Clethrionomyini (Cook et al. 2004; Luo et al. 2004), there is still a complete lack of data on the molecular phylogenetics of the morphologically polymorphic genus Alticola. Only one species, A. lemminus (called A. macrotis) has been included in previous molecular studies (Cook et al. 2004). Its results are consistent with the hypothesis of Clethrionomys s.l. paraphyly.
Palaeontological background Palaeontological record of the entire tribe Clethrionomyini is quite scarce and is mostly represented by isolated teeth. Most clethrionomyines have dentitions with rooted molars and crown cement, and some lineages show the development of more advanced rootless cheek teeth. Generally, in evolution of arvicolines, the primitive forms with brachyodont rooted molars adapted to granivory showed evolution towards more hypsodont and finally rootless forms adapted to feeding on grasses (Hinton 1926; Kretzoi 1969; Gromov 1972). In recent fauna, clethrionomyines with rooted molars (Clethrionomys s.l.) are associated with wooded habitats, whereas forms with rootless teeth (Alticola s.l., Eothenomys) inhabit mountainous biotopes. The modern diversity of the tribe is distinctly focused in the eastern regions of Palaearctic implying a possible centre of the tribe's origin. The earliest fossil records of the group is, however, so far known in Europe and represents a small form of Clethrionomys dating back to the boundary of Middle and Late Pliocene (c. 2.6 My) (Tesakov 1996). Almost equally, old Clethrionomys is listed from southern West Siberia (Zykin et al. 1989). The earliest fossil record of eothenomyine voles is known since Late Pliocene and Early Pleistocene of south China (Zheng & Li 1990; Musser & Carleton 2005). The most morphologically primitive clethrionomyine showing some archaic `protomimomys' characters, unknown in other early forms, was recently described from Late Pliocene of central China as Villanyia (Zhang 2004; Zhang et al. 2006). The earliest smaller Clethrionomys in Europe (Chaline & Michaux 1974; Tesakov 1996; Popov 2000) and West Siberia (Smirnov et al. 1986; Zykin et al. 1989) are morphologically close to lineages of recent C. glareolus and C. rutilus, respectively. Thus, the split between these lineages based on geological position of their early representatives is estimated at no less than 2.5 My. The palaeontological data on Craseomys provide little relevant information since the earliest record of the lineage of C. rufocanus is as late as Early Pleistocene of southern West Siberia represented by a notably more primitive form described as C. major (Borodin 1988). Other members of this group, Japanese Phaulomys, are still younger and this lineage

is known since Middle Pleistocene (Kawamura 1988). The fossil record of Alticola proper is extremely scarce and reliably starts in late Early Pleistocene of peri-Baikalian Russia (Alexeeva 1998) and Middle Pleistocene of China (Young 1934; Pei 1936). The Chinese records were later attributed to recent species A. stoliczkanus and A. roylei (Zheng & Li 1990). In caves of Russian Altai, the fossil record of A. strelzowi and A. macrotis starts at least from the beginning of Late Pleistocene (Agadjanian & Serdyuk 2005). All Middle­Late Pleistocene forms, like all extant species of Alticola, have rootless cheek teeth. It means that most Alticola forms fully developed rootless cheek dentition as the adaptation to the abrasive herbivory by Middle Pleistocene. The preceding stage of evolution is obscured by the formal practice of attribution of all fossil clethrionomyines with rooted teeth to Clethrionomys s.l. Late Pliocene forms that may be representatives of some proto-Alticola lineage were described as Villanyia klochnevi from Transbaikalian Russia (Erbajeva 1998) as aberrant lagurines. These poorly known voles have rooted molars without cement (primitive feature) and show distinct alticoline dental morphology. A slightly younger form, referable to ancestral Alticola, was described from Early Pleistocene fauna in south-eastern Kazakhstan (Kozhamkulova et al. 1987; Tjutkova & Kaipova 1996) as Clethrionomys mirus Savinov et Tjutkova, 1987. This hypsodont rooted from piedmonts of Tien Shan Mountains displays a combination of characters which makes us believe its phyletic link with extant Alticola ex gr. argentatus. In the Altai Mountains, an ancestral form of the Alticola s.str. with very hypsodont rooted molars is probably present as early as at the level of 2 My, that is, at the beginning of Early Pleistocene (Serdyuk & Tesakov 2006). As evident from the reviews given above, the available data on both extant and fossil clethrionomyines does not provide a clear picture of the tribe's phylogeny leaving serious gaps unfilled and controversies unresolved. In order to elucidate the phylogenetic relationships within the genus Alticola and to examine its position within the tribe Clethrionomyini, we estimated the mitochondrial cytochrome b (cyt b) gene variation and compared the results with morphological and palaeontological data. For the first time, the study of the relationships of the genus Alticola contains 6 of 12 recognized species representing all three known subgenera. Most of representatives of the tribe Clethrionomyini were included in the analysis.

Materials and methods
Nomenclature accepted In our study, we follow the classification system presented in the third edition of Mammal Species of the World (Musser & Carleton 2005) implying that Eothenomys includes Anteliomys,
549

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Table 1 List of specimens analysed.
Cytochrome b sequences Museum catalogue number, tissue or field code -- 179263 ZMMU 179045 ZMMU 179593 ZMMU 179584 ZMMU 179592 ZMMU ZIN RAS t.c.39 90986 ZIN RAS 90980 ZIN RAS Length (bp) 1140 1140 1140 1140 1140 1140 1140 900 900 1140 780 1140 GenBank accession no. DQ845186 DQ845187 DQ845192 DQ845193 DQ845188 DQ845194 DQ845190 DQ845191 DQ845189 DQ845196 DQ845195 DQ845185

Specimens 1 2 3 4 5 6 7 8 9 10 11 12

Collecting locality Turkmenia, Kugitang Mongolia, Hara-Nur, 48°19, 96°07 Mongolia, Harhorin, Orhon left bank, 47°06, 102°46 Mongolia, Ulan-Bator 75S, 47°17, 106°39 Mongolia, Tsogt, 45°23, 96°45 Mongolia, Baian-Under, 44°44, 98°46 Altai, Ulaganskii area, Bashkaus River, middle reaches Altai, Ulaganskii area, Kuraiskii Range, Aktash, Yarlyamry, 50°19, 87°46 Eastern Kazakhstan, Zyryanovskii area, origin of the Khamir River, 50°11, 84°35 Altai, Ust`Koksinskii area, Kholzun Range 50°09, 84°49E Altai, Ulaganskii area, Kuraiskii Range, Aktash, Yarlyamry, 50°19, 87°46 Kazakhstan, Alma-Ata region, Kaskelen

A. (Alticola) argentatus A. (Alticola) semicanus A. (Alticola) semicanus A. (Alticola) semicanus A. (Alticola) barakshin A. (Alticola) barakshin A. (Platycranius) strelzowi A. (Platycranius) strelzowi A. (Aschizomys) macrotis vinogradovi A. (Aschizomys) macrotis vinogradovi A. (Aschizomys) macrotis vinogradovi Clethrionomys centralis

f.c.85 f.c.7 f.c.147 f.c.33 f.c.109 f.c.31 f.c.25

90981 ZIN RAS f.c.26 90985 ZIN RAS f.c.30 ZMMU f.c.77/00 24320

and that Clethrionomys contains Craseomys (C. rufocanus, C. rex) and Phaulomys ( C. andersoni and C. smithii), which are otherwise treated as subgenera (Pavlinov 2003) or full genera (Musser & Carleton 1993 for Phaulomys; and Mezhzherin & Serbenyuk 1992; Mezhzherin et al. 1995 for Craseomys). We do not follow the substitution of the name Clethrionomys with the name Myodes (Carleton et al. 2003; Musser & Carleton 2005; Pavlinov 2006), but believe that the name Clethrionomys overwhelmingly used in the literature for about a century should be conserved to enforce the stability of the zoological nomenclature.
Specimens examined Tissue samples have been taken from specimens of A. (Alticola) argentatus, A. (Aschizomys) macrotis, A. (Platycranius) strelzowi, A. (Alticola) barakshin, A. (Alticola) semicanus and C. centralis. Species identifications for this study were based on morphological criteria and were performed prior to the molecular analysis. All voucher specimens are deposited in the Zoological Museum of Moscow (Lomonosov) State University (ZMMU) or in the Zoological Institute RAS (ZIN RAS) in St Petersburg. The list of species, collecting sites, museum catalogue numbers and GenBank accession numbers are given in Table 1. For cyt b gene phylogenetic analysis, 47 sequences were retrieved from the GenBank (Table 2). Dicrostonyx torquatus and six species of Microtus were included in the phylogenetic analysis as outgroups. DNA isolation, PCR amplification and sequencing Genomic DNA was isolated from ethanol-fixed liver, kidney or muscles by proteinase K digestion, phenol­chloroform
550

deproteinization and isopropanol precipitation (Sambrook et al. 1989). The complete cyt b gene (1140 bp) was sequenced in nine voles including one specimen of C. centralis (Table 1). A gene region that included the whole of the mitochondrial cyt b gene was amplified by PCR with the forward/reverse primer combination: L14734/H15985 (Ohdachi et al. 2001) or L14728/ H15906arvic. In two specimens, we sequenced a portion of the gene (~900 bp) with the use of UCBU and UCBL primers and for one specimen 780 bp were sequenced (Tables 1 and 3). Primers H15906arvic, UCBU, UCBL and most of internal primers were designed for this study and presented in Table 3. Double-stranded polymerase chain reaction (PCR) usually entailed 30­35 thermal cycles as follows: 30 s denaturation at 94 °C, 1 min annealing at 55 °C or 62 °C and 1 min extension at 72 °C. All PCR experiments included negative controls. PCR products were visualized on 1.5% agarose gel and then purified using DEAE Watman and Qiagen QIAquick kit. Approximately 10­30 ng of the purified PCR product was used for sequencing with each primer by autosequencing system 3730 DNA Analyzer (Applied Biosystems, Foster City, CA, USA) using ABI PRISM® BigDyeTM. Terminator v. 3.1.
Phylogenetic analysis The final alignment of the mitochondrial regions included 1140 bp. The sequence length for A. (Platycranius) strelzowi DQ845191 and A. (Aschizomys) macrotis vinogradovi DQ845189 were shorter (900 bp), nevertheless, they were included in all analyses. Only 780 bp were obtained for A. (Aschizomys) macrotis vinogradovi from Kuraiskii Range DQ845195, so its phylogenetic position was assessed only in parsimony trees.

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

Table 2 Cytochrome b sequences of voles retrieved from GenBank (Cook et al. 2004; Luo et al. 2004).
Specimens GenBank accession no. AY309411 AF119273 AY309421 AY309420 AY309419 AY309429 AY309434 AF272639 AF272636 AY309423 AY309422 AF272631 AF272632 AF272638 AY309424 AY309427 AY309428 AB031582 AB031574 AB031576 AF272640 AY309412 AY309418 AB037311 AB104508 AB104503 AB037302 AY426691 AY426690 AY426687 AY426677 AY426676 AY426686 AY426683 AY426679 AY426678 AY426680 AY426675 AY426682 AY426682 AY513830 AY513814 AY513808 AF163893 AY167150 AF163900 AF119275

Table 3 Primers used for PCR amplification and sequencing of the cytochrome b gene in Alticola and Clethrionomys.
Primer L14734 H15985 L14729 H15906arvic H15195MK H15576MO CBU UCBL Sequence (5­3) AAAAACCATCGTTGTTATTCAACT TAGAATGTCAGCTTTGGGTGCT GACATGAAAAATCATCGTTGTTATT ACTGGTTTACAAGACCAGTGTAAT GCCGATGTATGGGATAGCTGATA GATCGTAGGATGGCGTAGG CCATCAAACATCTCATCCTGATGAAA TCAACTGGTTGGCCGCCAATTCATGT Reference Ohdachi et al. (2001) Ohdachi et al. (2001) This study This study This study Brunhoff et al. 2003 This study This study

Alticola (Aschizomys) lemminus A. (Aschizomys) lemminus Clethrionomys glareolus C. glareolus C. glareolus C. gapperi C. gapperi C. gapperi C. gapperi C. californicus C. californicus C. rutilus C. rutilus C. rutilus C. rutilus C. rutilus C. rutilus C. rex C. rufocanus C. rufocanus C. rufocanus C. rufocanus C. rufocanus C. smithii C. smithii C. andersoni C. andersoni Eothenomys proditor E. olitor E. olitor E. custos E. custos E. miletus E. miletus E. eleusis E. eleusis E. fidelis E. cachinus E. melanogaster E. melanogaster Microtus socialis M. majori M. juldaschi M. chrotorrhinus M. agrestis M. montebelli Dicrostonyx torquatus

To reveal saturation for different substitution classes/codon positions uncorrected pairwise differences (P-distances) were plotted against maximum likelihood (ML) distances based on total data set.
Tree reconstruction In order to control for potential biasing effects of base compositional heterogeneity and saturation in 3rd position transitions each phylogenetic reconstruction was performed with two data sets. The first one contained the 3rd codon positions recoded as R or Y (purines or pyrimidines) in order to exclude 3rd position transitions from consideration (ti3­ analysis) and the second retained information on all substitutions (ti3+ analysis). Phylogenetic neighbor-joining (NJ), ML and maximum parsimony (MP) analyses were performed using PAUP (Swofford 2003). Unweighted MP analysis was performed using heuristic search starting with stepwise addition trees (random addition sequence, 100 replicates) and employing TBR branch-swapping. To reconstruct the ML tree, the appropriate model of sequence evolution was chosen on the basis of hierarchical likelihood ratio tests (hLRT) as implemented in MODELTEST v. 3.7 (Posada & Crandall 1998). Subsequent heuristic search with fixed model parameters used the NJ tree as the initial topology which was subjected to SPR swapping. NJ trees were derived from ML distances obtained using the same substitution models and parameter values as in the ML analysis. Bootstrap analysis with 1000 pseudoreplicates was used to measure support of the resulting MP and NJ topologies. The MULTREES option was not in effect for the MP bootstrap analysis of the ti3­ data set. The ML bootstrap analysis was based on just 100 replicates (NJ tree as the start and NNI swapping) due to prohibitively long computation times. Bayesian phylogenetic analysis (Huelsenbeck et al. 2001) was conducted using the program MRBAYES 3.04 (Ronquist & Huelsenbeck 2003) with either HKY + I + model for the total data (ti3­ variant) or separate models for each codon
551

To test the significance of compositional heterogeneity at 3rd and 1st codon positions, 2 tests were performed in PAUP*version 4.0b10 (Swofford 2003) and STATIO (Rzhetsky & Nei 1995). Principal component analysis (PCA) as implemented in STATISTICA v. 6.0 was employed to explore the variation of nucleotide frequencies.

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

positions (ti3+ variant). In the latter case, a combination of codon-specific models was obtained with MODELTEST comprising GTR + I + models for both 1st and 3rd positions, and HKY + I + for 2nd position. All model parameters were unlinked across partitions. Two independent analyses of four chains each were run for 2.5 or 2.0 million generations for ti3+ and ti3­ variants, respectively. Trees and parameters were sampled every 1000 generations; the heating parameter was set to its default value of 0.2. To check for convergence the trace of log-likelihoods against generation number was visualized in the program TRACER v.1.3 (Rambaut & Drummond 2004). The first 1.5 (ti3+) or 1.0 (ti3­) million generations were discarded as the `burn-in'.
Hypothesis testing Three a priori formulated hypotheses, based upon currently accepted system of the genus Alticola, were tested. They include: (i) the monophyly of Alticola s.l.; (ii) the monophyly of the subgenus Alticola s.str.; and (iii) the monophyly of the subgenus Aschizomys. A version of SOWH test (Hillis et al. 1996; Swofford et al. 1996; Goldman et al. 2000) adapted to the parsimony criterion was employed. First, the best tree constrained to contain the monophyletic group in question was obtained. Since parsimony analysis never resulted in a single tree, the majority-rule consensus was used as the optimal topology. If polytomies existed, they were resolved in accordance with the ML tree. Next, most appropriate parameters of the GTR + I + model as well as optimal branch lengths were estimated on the basis of the best constrained topology in PAUP. Using the resulting suite of parameters, 500 data sets were simulated in SEQGEN v. 1.3.2 (Rambaut & Grassly 1997) and subjected to both constrained and unconstrained parsimony analysis. H0 is rejected if the difference in length between constrained and overall optimal trees for the original data falls out of the 95% range for the distribution of this test-statistic generated from simulated data sets. In the case of the ti3­ analysis, two separate sets of model parameters and branch lengths were employed for simulating evolution of 380 bp at 3rd position, and 760 bp at combined 1st and 2nd positions. After the 3rd position data sets were recoded into purines and pyrimidines they were concatenated with the sets of 1st + 2nd positions and processed as described above. Global and local clock The assumption of rate constancy was tested via LRT based on likelihood values for the ML topology obtained with and without global clock constraint (Felsenstein 1981). Rate heterogeneity among taxa was determined with the use of relative-rate tests as implemented in RRTree (Robinson et al. 1998; Robinson-Rechavi & Huchon 2000). Likelihoods of trees assuming a limited number of different rates (local clock
552

approach) were calculated using BASEML program (PAML 3.14 package) (Yang 1997) on the basis of ML topology and GTR + model.
Relaxed-clock analysis Bayesian estimation of divergence times under the relaxedclock model was performed using MULTIDIVTIME package (Thorne et al. 1998; Kishino et al. 2001; Thorne & Kishino 2002). Separate F84 + models were employed for each of the codon positions and their parameters were estimated in PAML (T able 6). The choice of this model is explained by the limitations of the estbranches_dna software responsible for estimation of the variance­covariance matrix of branch lengths, which is a necessary step in the algorithm implemented in multidivtime. However, the analysis of saturation plots (not shown) demonstrates that the relationship between GTR + I + G and F84 distances remains nearly linear for most of their ranges, indicating that the bias due to the application of a simpler model is negligible here. Rather than assuming a priori calibration dates, we preferred to obtain a set of relative divergence times which are independent of any assumptions about rates and dates, but can be easily transformed into absolute times afterwards. Thus, the prior for expectation of the age of the root was set to be equal to one arbitrary time unit. Following the recommendations of the program manual the expected rate for the root node was chosen to be close to the median root-to-tip distance (0.25 for ti3+ and 0.07 for ti3­ analyses). The prior for this parameter needs to be flat in order to accommodate high variation among the three codon positions, so the variance of the root rate was set at 2.0 and 1.0, respectively. Default values were used as the priors for the parameter of rate variation (mean = 1.0, SD = 1.0) and the parameter specifying prior probability of internal node depth MinAb (= 1.00). Several independent runs were conducted in order to assess the length of chain sufficient for convergence, mixing rate and sensitivity to choice of priors. Since a large proportion of variance of raw time estimates reflect nuisance variation of the time of the root node, they need to be standardized prior to statistical analysis. Instead of dividing node times in each sampled generation by its corresponding root time as it is outputted by multidivtime, we applied a method of standardization, which takes into account variation not of a single arbitrary point but of the entire ensemble of node times. The procedure is as follows: (i) median estimate Tmi is computed for each node time i across all sampled generations; (ii) a scaling factor Sj is obtained for each generation as the median of Tij / Tmi, where Tij is the time estimate of node i in generation j; and (iii) standardized relative time estimates (ij) are computed as Tij /Sj, and their distribution is subjected to basic statistic analysis. The choice of the median-based method is justified by the necessity to account for possible effects of node times

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

Table 4 Base composition at 1st, 2nd and 3rd positions of codons for the analysed taxa.
Base frequencies (%) mean (min­max) Codon position 1st 2nd 3rd A 29.6 (26.7­31.1) 20.8 (20.1­21.1) 39.3 (36.6 ­ 41.1) C 24.1 (22.6 ­25.5) 24.5 (23.9 ­26.6) 41.1 (36.6 ­ 44.5) G 23.4 (22.4 ­24.8) 12.7 (12.4 ­13.4) 3.4 (1.6 ­ 6.1) T 22.9 (21.3 ­24.5) 42.0 (40.3 ­ 42.5) 16.2 (13.2­20.0) R :Y 53.0 : 47.0 (51.3 : 48.7­53.7 : 46.3) 33.5 : 66.5 (32.8 : 67.2­33.9 : 66.1) 42.7 : 57.3 (44.7 : 55.3 ­ 41.1 : 58.9)

with disproportionately large errors. For convenience of description an additional scaling factor is introduced so that the mean time for the split between lineages leading to C. rutilus and C. glareolus is equal to 2.5 My. The rational for this calibration is given in the Introduction section. Alternative fossil-based calibration points such as the one suggested by Luo et al. (2004) based on hypothesis of Kawamura (1988) are incompatible with early diversification of C. rutilus and C. glareolus lineages and, hence, result in underestimation of divergence times. Two sets of standard deviations and confidence intervals were calculated: the first corresponds to variation of original relative estimates and is independent of the error of rate estimate while the second takes into account covariation with the time estimate for the calibration node and its variance and, hence depends on the uncertainty of rate estimation (but not on the error of the calibration date itself). The latter set was inferred on the basis of the observed distribution of (ij /rgj) where rgj is the relative age for C. glareolus­ C. rutilus split in the jth generation. Since some nodes in the Clethrionomyini tree could not be resolved with confidence, we used two tree topologies differing in branching pattern within Clethrionomys: while the first topology was equivalent to the ML tree (ti3+ analysis), the second differed from it in containing [C. rutilus + (C. gapperi + C. californicus)] branching. Despite the deficit in support of Alticola s.str. + Clethrionomys s.str. grouping, we preferred our trees to retain this clade since this pattern is corroborated by previous cytogenetic and allozyme data. Another problem that we tried to address was the extent to which our time estimates were sensitive to the influence of ancestral polymorphism. (For a complete review of potential sources of bias and random error in molecular date estimation see Arbogast et al. 2002.) Since the date estimates refer to splits between haplotype lineages rather than between species, the former are expected to predate the latter by a certain amount of time that is dependent on the ancestral effective population size (Nef). Based on the output of multidivtime, we modelled the effect of non-zero ancestral coalescent times assuming that the ancestral Nef is a random exponentially distributed variable with the expectation of 105. The latter value is inferred from the levels of variation within contemporary populations of Clethrionomys ( ~ 0.005, see Runck & Cook 2005; Deffontaine et al. 2005). We employed the divergence

rate of ~0.05 substitutions per My which roughly corresponds to the median value for Clethrionomyini as follows from the results of our relaxed-clock analysis.

Results
Base composition and its variation All obtained cyt b sequences follow the pattern of compositional biases expected in mammals (Table 4) (e.g., Irwin et al. 1991). The test for base composition heterogeneity as realized in PAUP did not reject the null hypothesis of constant base frequencies in any of the codon positions. In contrast, the test implemented in STATIO (Rzhetsky & Nei 1995), which takes into account possible phylogenetic correlations, reveals significant heterogeneity in composition of the third (2 = 105, d.f. = 78, P = 0.02) but not the first (2 = 83, d.f. = 78, P = 0.32) codon position. This test was applied to the alignment including only 27 haplotypes (one to two haplotypes per species) due to memory limitations of the program. According to the results of PCA for 3rd codon positions, the first principal component (PCI), which accounts for 67% of total variance, is highly correlated with variation of base frequencies within pyrimidines (Pearson r between and C­T = 0.99). PCII (27.5%) and PCIII (5.5%) are associated mainly with differences in A­G (r = 0.92) and R­Y (r = 0.91), respectively. The most deviant base composition in the analysed data set is found in Eothenomys custos AY426677 (C: 37%, T: 20%) and A. (Aschizomys) macrotis (C: 38%, T: 19%) contrasting to AY426686 E. miletus (C: 44%, T: 13%). As concerns the R /Y proportion the most dissimilar are E. melanogaster AY426682 (R: 45%) and AY309411 A. (Aschizomys) lemminus (R: 41%); however, the difference between them is not significant (2 = 1.05, d.f. = 1, P = 0.30). Hence, one can conclude that the stationarity condition is not violated for 3rd position transversions. Saturation As it is evident from the saturation plots (Fig. 2), the 3rd position transitions are saturated when the level of divergence reaches ~10%­20%. As long as this level corresponds to some of the splits of interest, phylogenetic reconstructions were performed with both the original alignment (ti3+) as well as the data set in which the 3rd codon positions were R/Y recoded (ti3­).
553

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Fig. 2 Saturation plot: uncorrected (P) pairwise distances for different substitution types and codon positions are plotted vs. ML distances (GTR + I + ) based on all substitution types in all codon positions.

Table 5 Models and parameters for reconstruction of ML trees and relaxed-clock analyses.
Stationary base frequencies Data ti3+ ti3­ 1st position 2nd position 3rd position 3rd position (transversions only) Model Gtr + I + Hky + I + F84 + F84 + F84 + CF + R-Matrix parameters rAC rAG 0.4261 9.8372 Tratio = 3.5623 = 8.69929 = 4.93087 = 12.34628 rAT 0.9339 rCG 0.4963 rCT 9.4793 rGT 1 A 0.3350 0.2589 0.3086 0.2080 0.3806 R 0.43819 C 0.3648 0.2596 0.24898 0.25735 0.40004 G 0.0892 0.1733 0.2083 0.1261 0.04944 Y 0.56174 T 0.211 0.3082 0.2341 0.4085 0.1670 Shape parameter of distribution 1.2126 0.6785 0.10693 0.00921 1.38405 0.35308 Proportion of invariant sites 0.5743 0.6896 -- -- -- --

Model choice The model chosen for the ti3+ data set by GTR + I + that is, the most parameter-rich PAUP. In contrast to that, the preferable model excluding 3rd position transitions were less HKY + I + (Table 5).

hLRT was available in for the data complex --

Phylogenetic results All phylogenetic methods employed in our study displayed trees with similar topology (Figs 3 and 4). Although the trees inferred on the basis of the two data sets and different optimality criteria were not completely identical, no strongly supported (> 80% bootstrap) incongruent clades were revealed. The ti3+ and ti3­ analyses produce concordant results differing mainly in values of support for several clades. Expectedly,
554

resolution was somewhat lower for the second data set. Thus, despite a significant departure from base composition stationarity in 3rd positions, the compositional bias does not appear to have any serious effect on the tree structure. None of the trees that we constructed is consistent with the monophyly of Alticola s.l. which appears as a polyphyletic assemblage of three lineages. The subgenera Alticola s.str. and Platycranius form a well-supported monophyletic group; however, the latter is nested within the nominative subgenus showing close association with A. (Alticola) semicanus. The level of divergence (GTR + I + distance = 6.7%) between these two species is similar to that between such closely related species as C. (Phaulomys) andersoni and C. (Phaulomys) smithii. Alticola barakshin, A. (Alticola) argentatus, and the clade comprising A. (Alticola) semicanus and A. (Platycranius) strelzowi

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

Fig. 3 The ML tree for 1140 nucleotide positions generated using GTR + I + model based on all substitutions in all codon positions. Bayesian posterior probabilities (BPP) and bootstrap values ( 50%) obtained from 100 replications for ML analysis and 1000 replications for NJ and MP analyses are given in the following order ((BPP/(ML/NJ))/MP). Dicrostonyx torquatus is used as an outgroup. The sequences obtained in the present study are marked by asterisks. A. (Aschizomys) macrotis vinogradovi DQ845195 (its branch is given as dotted line) was used only in parsimony analysis.

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563

555


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Fig. 4 The ML tree based on all substitutions except of transitions in 3rd codon positions generated using HKY + I + model. The designations are as in Fig. 2.

556

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

form a poorly resolved trichotomy with the average level of divergence between the three lineages being slightly larger than those observed among species of Clethrionomys s.str. (14%­15% vs. 11.3%­12.7%). The within-tribe position of the (Alticola s.str. + Platycranius) clade as a member of the group also comprising Clethrionomys and Craseomys is well supported by all analyses. Moreover, in some cases moderate support is found for closer association of Alticola and Clethrionomys (parsimony and NJ analyses with ti3+ data set) with Craseomys splitting basally to them. However, ti3­ analysis results in a trichotomy among Alticola, Clethrionomys s.l. and Craseomys or suggests Craseomys + Alticola grouping, though never supported by bootstrap. We believe that the latter result should be regarded rather as an effect of insufficient information content in the ti3­ data set. The representatives of the subgenus Aschizomys do not form a monophyletic group. The most unexpected result concerns a close association between haplotypes found in A. macrotis from the Altai Mountains, and those of C. centralis and C. glareolus. The distance among these three lineages is surprisingly small (5.0%­6.3%). In contrast to that, the type species of Aschizomys -- A. (Aschizomys) lemminus tends to represent the basal branch within the poorly resolved complex of lineages corresponding to C. rutilus, C. californicus, C. glareolus + C. centralis + A. (Aschizomys) macrotis and two forms of C. gapperi. The support for the basal position of A. (Aschizomys) lemminus is low for the complete data set but increases slightly (BPP/ML/NJ/MP: 71/57/55/53) when the 3rd position transitions are excluded.
Phylogenetic hypotheses testing We tested three a priori hypotheses postulating the monophyly of Alticola s.l. and its two polytypic subgenera (Aschizomys and Alticola s.str.) using the variant of the SOWH test adapted to parsimony. In neither case the difference between the lengths of the overall and constrained optimal trees for the original data ( steps) was exceeded by the test-statistic for any of the 500 simulated data sets. So, all a priori hypotheses are rejected at a high level of significance. The same outcome was obtained when the hypothesis of Alticola monophyly was tested on the basis of the ti3­ data set. Cytochrome b clock tests The assumption of rate constancy was rejected by the LRT for both the original and ti3­ data sets (2 = 95.27; P < 0.001 and 2 = 84.36; P < 0.01, respectively). Relative-rate test has demonstrated that one of the reasons for clock violation is the acceleration of the rate of synonymous (but not nonsynonymous) substitutions in Microtus in comparison to Clethrionomyini (P < 0.03 for Ks, P > 0.3 for Ka). However, the local clock approach assuming just two rates of evolution (one for Microtus and the other for the rest of taxa) produced

the tree that was still significantly less fit than the rateunconstrained tree (2 = 92.98, P < 0.001). This result warranted the application of relaxed-clock analysis for date estimation.
Relaxed-clock analysis The preliminary runs demonstrated that the chain length of 500 000 generations was sufficient for convergence of relative date estimates. Alterations of root time and rate priors did not affect the resulting dates significantly. For both tree topologies the two arrays of dates generated from ti3+ and ti3­ data sets are highly correlated (Pearson r = 0.98) with the relationship between them being linear. Taking into account 35%­50% lower variation for the node dates resulting from the ti3+ analysis we base our conclusions exclusively on the latter. The dates produced with the two different topologies are congruent as well (r = 0.99 for ti3+ analyses based on 24 deeper nodes); in absolute values the times for corresponding internal nodes deviate from each other by 130 000 years on average (with the maximum of 0.4 My for the split between east and west lineages of C. gapperi). So, we present the results for just one of the trees (Fig. 5; Table 6), the node dates for which are slightly younger than for the second one (3% on average). According to the results produced by multidivtime the rate of divergence of cyt b sequence ranges from 4.4% to 5.9% within Clethrionomyini but is ~1.25 times higher in Microtus (6.0%­7.4%). The latter value might be an underestimate since according to the two-rate local clock analysis (PAML) the clock ticks ~1.7 times faster in Microtus compared to Clethrionomyini. As follows from our simulations, the expected bias due to coalescence times will not exceed ­160 000 generations (years) for the root node and +100 000 generations for the youngest nodes. Although standard deviations of node dates increased by 200 000 years on average, the lower limit of the confidence range decreased by just 90 000 years on average (to the maximum of 200 000). Thus, provided the simple model that we employ is applicable, the obtained divergence date estimates deviate from expected values just moderately. It should be pointed out that if the effective population size in the ancestor of Clethrionomys was substantially larger than average then all node times are underestimated. However, using a single gene and a single calibration point one can not expect the time estimates to be free from error and, hence, we regard the above results as preliminary.

Discussion
Phylogenetic relationships within Clethrionomyini One of the main tasks of the study was to clarify the relationships between different branches attributed to Alticola and
557

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Fig. 5 The chronogram of radiation in Clethrionomyini. The divergence times correspond to the mean posterior estimate of their age in My. The bars represent the 90% HPD interval for the divergence time estimates. Two confidence intervals are presented for each node: the first takes into account covariation with the time estimate for the calibration node and its variance (empty bars), the second corresponds to original relative estimates (grey bars). Nodes 2­25 are numbered.

558

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

Table 6 The estimates of divergence times. Two sets of SEs and confidence intervals (90%) are presented for each node: the first take into account covariation with the time estimate for the calibration node and its variance, the second (in parentheses) correspond to original relative estimates (see text for explanation).
Cytochrome b divergence (%) estimated from ML ultrametric tree (GTR + I + ) 21.8 15.9 11.6 9.2 7.7 7.0 3.4 6.8 6.3 3.2 2.5 5.7 5.7 4.9 2.8 5.5 4.9 4.8 1.9 1.9 1.7 0.7 12.1 10.4 6.8

Node 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Description of node

Divergence time ± SE (My) 6.6 5.2 4.05 3.39 2.89 2.63 1.27 2.72 2.50 1.25 0.96 2.29 2.00 1.78 0.97 2.3 2.05 1.96 1.01 0.72 0.77 0.29 4.25 3.53 2.80 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 1.46 0.95 0.63 0.42 0.47 0.47 0.38 0.19 0.00 0.28 0.26 0.13 0.23 0.25 0.28 0.53 0.49 0.49 0.32 0.26 0.21 0.13 0.88 0.83 0.70 (±1.10) (±0.62) (±0.42) (±0.33) (±0.37) (±0.38) (±0.35) (±0.30) (±0.28) (±0.28) (±0.26) (±0.28) (±0.29) (±0.29) (±0.28) (±0.39) (±0.37) (±0.38) (±0.27) (±0.23) (±0.22) (±0.13) (±0.61) (±0.64) (±0.55)

Confidence interval of divergence time (My) 4.50 ­9.25 3.83 ­ 6.92 3.16 ­5.17 2.78 ­ 4.13 2.18 ­3.70 1.91­3.45 0.71­1.96 2.51­3.09 2.50 ­2.50 0.82­1.75 0.57­1.42 2.04 ­2.48 1.60 ­2.35 1.34 ­2.18 0.54 ­1.46 1.60 ­3.31 1.35 ­2.94 1.24 ­2.83 0.58 ­1.59 0.38 ­1.21 0.45 ­1.15 0.10 ­ 0.53 2.99 ­5.80 2.31­ 4.98 1.79 ­ 4.03 (4.97­8.62) (4.25 ­ 6.28) (3.39 ­ 4.77) (2.86­3.95) (2.27­3.51) (2.00 ­3.26) (0.74­1.89) (2.25 ­3.24) (2.04 ­2.98) (0.82­1.75) (0.56 ­1.41) (1.83 ­2.78) (1.51­2.49) (1.30 ­2.27) (0.54 ­1.45) (1.75 ­3.03) (1.47­2.69) (1.36 ­2.64) (0.62­1.51) (0.40 ­1.15) (0.45 ­1.16) (0.09 ­ 0.53) (3.26 ­5.28) (2.49 ­ 4.62) (1.94 ­3.75)

Microtus/Clethrionomys Eothenomys/Clethrionomys s.l. + Alticola Craseomys/Clethrionomys s.str. + Alticola Alticola/Clethrionomys s.str. Alticola/ basal A. argentata/A. semicanus P. strelzowi/A. semicanus Aschizomys lemminus/Clethrionomys s.str. C. glareolus/C. rutilus C. glareolus/C. centralis A. macrotis/C. centralis C. rutilus/C. gapperi C. gapperi/C. californicus C. gapperi west/C. gapperi east C. gapperi west/C. gapperi central Phaulomys/Craseomys C. rufocanus/C. rex C. andersoni/C. smithii C. andersoni/basal C. rufocanus/ basal C. rutilus/ basal C. glareolus/ basal Anteliomys/Eothenomys E. custos/E. olitor + E. proditor E. melanogaster/E. fidelis + E. cachinus + E. miletus + E. eleusis

Clethrionomys. The suggested model of phylogenetic relationships within the tribe is illustrated in Fig. 6. Our results provide further evidence to support the association of Alticola and Clethrionomys s.l. and suggest that the latter is paraphyletic due to basal position of Craseomys. This pattern is fully compatible with previous cytogenetic (Agadjanian & Yatsenko 1984; Modi & Gamperl 1989) as well as allozyme data (Mezhzherin et al. 1995). Another long-standing issue addressed by the current analysis involves the monophyly of the genus Alticola which our mtDNA data clearly reject. This conclusion agrees with the previous DNA­DNA hybridization study (Gileva et al. 1990); however, the inferred branching patterns differ in position of Aschizomys. The apparent discrepancy might be accounted for by the obvious internal conflict in DNA­DNA hybridization data with respect to the affinities of the latter taxon. In comparison to previous cyt b analyses of Clethrionomyini, the inclusion of additional species of Alticola in our study did not change the branching pattern in Clethrionomys s.str., Eothenomys-Anteliomys and Craseomys-Phaulomys clades obtained earlier (Cook et al. 2004; Luo et al. 2004).

Fig. 6 The generalized scheme of phylogenetic relationships of the Clethrionomyini. All names correspond to valid genus-level taxa to the exception of Platycranius and `Aschizomys macrotis' (given in smaller font).

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563

559


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Relationships within Alticola s.str. Two main hypotheses were proposed regarding the phylogenetic relationships between species of Alticola s.str. (Rossolimo et al. 1992, 1994; Rossolimo & Pavlinov 1992). According to the first one (Rossolimo & Pavlinov 1992), the stem species of Alticola s.str., which was close to the modern A. roylei, gave rise to A. argentatus followed by A. tuvinicus, A. semicanus and finally A. barakshin as the terminal stage in the series. The second hypothesis (Rossolimo et al. 1994) postulates an early separate origin of A. barakshin and A. stoliczkanus from a stem species independent of all other species of Alticola s.str. The phylogenetic analysis of allozyme data by Hille & Stubbe (1996) just marginally supported affinities of A. argentatus, A. barakshin and A. semicanus as proposed by Rossolimo & Pavlinov (1992). Although we studied a restricted number of species, it should be mentioned that our results are congruent with the second hypothesis (Rossolimo et al. 1994) but not with the first one (Rossolimo & Pavlinov 1992). Evolutionary scenario of the tribe Clethrionomyini The reconstructed chronological sequence of phylogenetic events within the tribe Clethrionomyini (Fig. 5) is subject to cross-checking with the available data set. The two earliest successive splits on the tribe's tree are represented by the Chinese mountain voles and Craseomys group (Craseomys + Phaulomys). Our estimates based on cyt b sequences suggest that both these cladogenetic events occurred in Early Pliocene, though the first fossil record of craseomyines does not predate Late Pliocene and the first fossil craseomyines date back to Early Pleistocene. A later branching point corresponds to the split between Clethrionomys s.l. and Alticola s.l. We hypothesize that the common ancestor of these groups could have been ecologically and morphologically close to contemporary red-backed voles in its rather primitive dentition, broader dietary spectrum and affinity to woodland habitats. According to our date estimates, it must have existed not later than Middle Pliocene. As was shown above, there is so far no fossil record of the tribe for this time interval. Alticola demonstrates a whole complex of characteristics suggesting that it is a derivative of a Clethrionomys-like ancestral form that emerged as a result of adaptation to high mountain open habitats. This habitat shift could have been a result of the orogenesis activation in Central Asia in Middle­Late Pliocene and Early Pleistocene when the uplift of some areas reached several kilometres (Dodonov 2003). The same events should inevitably entail range fragmentation and promote splitting processes. This scenario is in line with the results of the cyt b relaxed-clock analysis suggesting that isolation of at least three lineages of Alticola s.str. dates back to Late Pliocene or Middle­Late Pliocene boundary and with first fossil records of protoAlticola forms in Late Pliocene of Central Asia.
560

Molecular results demonstrate that the radiation of most of species of Clethrionomys s.str was rather rapid that is consistent with the existence of a single ancestral species with transpalaearctic distribution. A presumed expansion of its range in the very beginning of Late Pliocene can be correlated with the expansion of the boreal biome through Palaearctic as a consequence of the strong global cooling cycle at c. 2.6 My. (Grichuk 1989; Repenning & Brouwers 1992.) However, this pattern is in poor correspondence with relatively late arrival of Clethrionomys in North America. According to the available fossil record, the first dispersal of Clethrionomys to the North America dates back to Irvingtonian II, c. 0.85 ­ 0.78 My (Bell et al. 2004). So far no details of the local endemic evolution that produced modern C. gapperi and C. californicus are available. If our calibration is correct, to reconcile the chronogram with fossil record one should suggest a substantial period of Palaearctic development of this group before its eastward migration via the Bering Bridge. Otherwise, the observed discrepancy can be explained by the lack of palaeontological record on earlier stages of evolution which could take place in high latitudes of North America and subsequent southward range shift. Our molecular-clock results suggest that C. centralis separated ~1 My ago that is concordant with fossil data according to which a small form of Clethrionomys is present in Central Asia (Tajikistan and Kyrgyzstan) for at least 1 My since late Early Pleistocene (Zazhigin 1988), persisting through early Middle Pleistocene (Dodonov 2003) and late Middle Pleistocene (Markova 1992). Morphologically, it is likely a form ancestral to modern C. centralis. We suggest that A. (Aschizomys) lemminus and A. (Aschizomys) macrotis originated from Clethrionomys s.str., perhaps as a result of independent adaptation to mountain habitats. Thus, they acquired the rock vole complex of petrophylic adaptations (rootless hypsodont molars, flattened angular-shaped cranium and specific external characters) at a later moment in time than Alticola s.str. and independently from each other. On one hand, the apparent similarity between the two Aschizomys species is partly accounted for by plesiomorphies indicating that they both represent an earlier stage of specialization retaining a number of characters common with their Clethrionomys-like ancestors such as thicker enamel. On the other hand, this pattern can be interpreted as an example of remarkable parallelism in morphologic evolution of clethrionomyine voles as a response to habitat shift. Life in suboptimal fluctuating mountain steppe or tundra habitats favours graminivory, thus promoting rapid morphologic transformations. Other groups of Clethrionomyini also demonstrate a tendency for rapid acquisition of rootless molars albeit in a different ecological background. This can be exemplified by `Microtus-like' trends observed among forms of the C. rufocanus­ C. rex group.

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

The endemic Japanese group of Phaulomys was shown to have the ancestor with rooted dentition, C. japonicus, in Middle Pleistocene (Kowalski & Hasegawa 1976; Kawamura 1988). Kawamura (1988) accurately documented the transition to complete rootless condition from C. japonicus to Phaulomys in the interval from 150 to 20 ka. It is clear from the above scenario that the predicted splitting times for branches of Clethrionomyini notably predate their occurrences in fossil record. Most likely it is a realistic picture that is influenced by the combination of cryptic evolution periods before broad geographical spread of a certain lineage and an insufficiently dense fossil record. However, it should be stressed once again that results based on a single genetic marker whatever informative, should be regarded as a hypothesis requiring corroboration from independent genetic data. An important potential source of distortion of both the retrieved branching pattern and date estimates is hybridization resulting in introgression of alien mtDNA (e.g., Patterson et al. 2006). Taking into account evidence on past hybridization between C. glareolus and C. rutilus (TegelstrÆm 1987), we cannot discard such a possibility as improbable.
Taxonomic implications Our results provide unequivocal support for the phylogenetic grouping comprising all species contemporarily referred to Alticola s.l. and Clethrionomys s.l. Within this wider group three lineages corresponding to Alticola s.str., C. rufocanus + C. smithii/C. andersoni and Clethrionomys s.str. + Aschizomys are easily recognized. However, both molecular and cytogenetic data favour a somewhat closer association between Clethrionomys s.str. and Alticola to the exclusion of C. rufocanus­C. smithii/ C. andersoni group, thus rendering Clethrionomys s.l. paraphyletic. Following the principles of phylogenetic systematics, we should either lump all the above-mentioned taxa into a single genus or reinstate the full genus rank for Craseomys Miller, 1900 comprising C. rufocanus with related forms, C. rex, and all species of C. andersoni/C. smithii complex (subgenus Phaulomys). Taking into account high levels of morphologic and genetic divergence among the three main phylogenetic lineages, we prefer the second solution. Similar interpretations were suggested earlier (Mezhzherin & Serbenyuk 1992; Mezhzherin et al. 1995). A simpler concept accepted in the latest edition of Mammal Species of the World (Musser & Carleton 2005) and recognizing all-inclusive genera Clethrionomys and Alticola should be abandoned as contradicting phylogenetic data. In light of our mtDNA data, the validity of the subgenus Platycranius appears questionable. Current results suggest that the flat-skulled vole is just a morphologically deviant offshoot within Alticola s.str. sharing relatively recent common ancestor with A. semicanus. The latter association is partly supported by the observation that in comparison to other

Alticola species the skull of A. semicanus is flattened to a larger extent that might be interpreted as an earlier stage of the striking transformation found in A. strelzowi. Taxonomic interpretation of phylogenetic position of species included in Aschizomys is less straightforward. First, the data obtained is definitely incompatible with the status of Aschizomys as a subgenus within Alticola since the type species of Aschizomys (A. lemminus) forms a well-supported association with Clethrionomys. If its basal position within this clade receives additional support from other data, Aschizomys can be regarded as a valid subgenus within Clethrionomys (or even a separate genus). Next, if the inferred mtDNA haplotype tree correctly reflects relationships among species, A. macrotis and A. lemminus do not form a monophyletic group and, hence, the former should be excluded from Aschizomys. The taxonomic status of `Aschizomys' macrotis seems most controversial. High level of morphological differentiation from typical Clethrionomys prompts its recognition as a representative of a separate generic-level taxon. However, if such a standpoint is accepted and provided our phylogenetic hypothesis is correct, the genus Clethrionomys would inevitably become a paraphyletic assemblage. No adequate resolution of such a conflict between morphology-based evolutionary approach and limitations imposed by principles of phylogenetic taxonomy is evident thus far.

Acknowledgements
The authors thank William Modi who read the first draft of the manuscript and made valuable comments. We are grateful to Igor Pavlinov for discussion of the results. We are indebted to Alexander Stekolnikov (ZIN RAS, St Petersburg), Yury Litvinov (Novosibirsk), Alexey Surov (Severtsov Institute RAS, Moscow) and Alexey Bogdanov (Koltzov Institute RAS, Moscow) for tissue samples. We also thank Tatyana Petrova (St Petersburg University) for technical assistance in the laboratory. This work was supported by the Russian Foundation for Basic Research, projects, 06-04-49294a, 05-04-49240a and 06-05-64049a.

References
Agadjanian, A. K. & Serdyuk, N. V. (2005). History of mammalian communities and paleogeography of the Altai Mountains in the Paleolithic. Paleontological Journal, 39 (Suppl. 6), 645 ­ 812. Agadjanian, A. K. & Yatsenko, V. N. (1984). Phylogenetic interrelationships in voles of northern Eurasia. Proceedings of Zoological Museum of Moscow State University, 22, 135 ­190 [in Russian]. Alexeeva, N. V. (1998). Paleoecology of Transbaikal Allophaiomys faunas. Paludicola, 2 (1), 16 ­19. Arbogast, B. S., Edwards, S. V., Wakeley, J., Beerli, P. & Slowinski, J. B. (2002). Estimating divergence times from molecular data on phylogenetic and population genetic timescales. Annual Review of Ecology and Systematics, 33, 707­740. Bell, C. J., Lundelius Jr, E. L., Barnosky, A., Graham, R. W., Lindsay, E. H., Ruez Jr, D. R., Semken, Jr. H. A., Webb, S. D. &

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563

561


Molecular phylogeny of the genus Alticola · V. S. Lebedev et al.

Zakrzewski, R. J. (2004). The Blancan, Irvingtonian, and Rancholabrean mammal ages. In M. O. Woodburne (Ed.) Late Cretaceous and Cenozoic Mammals of North America: Biostratgraphy and Geochronology (pp. 232­314). New York: Columbia University Press. Borodin, A. V. (1988). History of voles of the genus Clethrionomys in the West Siberia Lowland. In A. V. Borodin (Ed.) Moder n Status and Histor y of Animals from the West Siberia Lowland (pp. 23­31). Sverdlovsk: USSR Academy of Sciences, Uralian Branch [in Russian]. Brunhoff, C., Galbreath, K. E., Fedorov, V. B., Cook, J. & Jaarola, M. A. (2003). Holarctic phylogeography of the root vole (Microtus oeconomus): implications for late Quaternary biogeography of high latitudes. Molecular Ecology, 12, 957­968. Carleton, M. D., Musser, G. G. & Pavlinov, I. Ya. (2003). Myodes Pallas, 1811, is the valid name for the genus of red-backed voles. In: A. O. Averianov & N. I. Abramson (Eds) Systematics, Phylogeny and Paleontology of Small Mammals (pp. 96 ­98). St Petersburg: Zoological Institute of Russian Academy of Sciences. Chaline, J. & Michaux, J. (1974). Les micromammifÕres et la biostratigraphie de la bordure occidentale du remplissage bressan. Bulletin Scientifique de Bourgogne, 29, 55 ­ 83. Cook, J. A., Runck, A. M. & Conroy, C. J. (2004). Historical biogeography at the crossroads of the northern continents: molecular phylogenetics of red-backed voles (Rodentia: Arvicolinae). Molecular Phylogenetics and Evolution, 30, 767­777. Deffontaine, V., Libois, R., Kotlik, P., Sommer, R., Nieberding, C., Paradis, E., Searle, J. B. & Michaux, J. (2005). Beyond the Mediterranean peninsulas: evidence of central European glacial refugia for a temperate forest mammal species, the bank vole (Clethrionomys glareolus). Molecular Ecology, 14, 1727­1739. Dodonov, A. E. (2003). Quaternary of Middle Asia. Stratigraphy, Correlation, PaleoGeography. Moscow: Geos [in Russian]. Erbajeva, M. A. (1998). Late Pliocene Itantsinian faunas in Western Transbaikalia. Mededelingen Nederlands Instituut Voor Toegepaste Geowetenschappen TNO, 60, 419 ­ 430. Felsenstein, J. (1981). Evolutionary trees from DNA sequences: a maximum likelihood approach. Journal of Molecular Evolution, 17 (6), 368 ­376. Gileva, E. A., Rybnikov, D. E. & Miroshnichenko, G. P. (1990). DNA­DNA hybridization and phylogenetic relationships in two genera of voles, Alticola, and Clethrionomys (Microtinae, Rodentia). Doklady Akademii Nauk SSSR, 311, 477­ 480 [in Russian]. Goldman, N., Anderson, J. P. & Rodrigo, A. G. (2000). Likelihoodbased tests of topologies in phylogenetics. Systematic Biology, 49, 652­ 670. Grichuk, V. P. (1989). History of Flora and Vegetation of the Russian Plain in Pleistocene. Moscow: Nauka [in Russian]. Gromov, I. M. (1972). Systematic groups above species level in the vole subfamily (Microtinae) and their possible affinities. Proceedings of Zoological Museum of Moscow State University, 13, 8­33 [in Russian]. Gromov, I. M. & Erbajeva, M. A. (1995). The Mammals of Russia and Adjacent Territories Lagomorphs and Rodents. St Petersburg: Zoological Institute of Russian Academy of Sciences. Gromov, I. M. & Polyakov, I. Ya. (1977). Voles (Microtinae). Fauna of the USSR Mammals, 3 (8), 1­504. Leningrad: Nauka [in Russian]. Hille, A. & Stubbe, A. (1996). Biochemical systematics of four taxa of Asian highmountain voles, Alticola (Rodentia, Arvicolinae). Folia Zoologica, 45 (4), 289 ­299.

Hillis, D. M., Mable, B. K. & Moritz, C. (1996). Applications of molecular systematics: the state of the field and a look to the future. In D. M. Hillis, C. Moritz & B. K. Mable (Eds) Molecular Systematics, 2nd edn (pp. 515 ­543). Sunderland, MA: Sinauer. Hinton, M. A. C. (1926). Monograph of the Voles and Lemmings (Microtinae) Living and Extinct, 1. London: British Museum (Natural History). Huelsenbeck, J. P., Ronquist, F., Nielsen, R. & Bollback, J. P. (2001). Bayesian inference of phylogeny and its impact on evolutionary biology. Science, 294, 2310 ­2314. Irwin, D. M., Kocher, T. D. & Wilson, A. C. (1991). Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution, 32, 128 ­144. Jameson, E. W. (1961). Relationships of the red-backed voles of Japan. Pacific Science, 15 (4), 594 ­ 604. Kawamura, Y. (1988). Quaternary rodent faunas in the Japanese Islands (Parts I and II). Memoirs of the Faculty of Science, Kyoto University, Series Geology and Mineralogy, 53, 1­348. Kishino, H., Thorne, J. L. & Bruno, W. J. (2001). Performance of a divergence time estimation method under a probabilistic model of rate evolution. Molecular Biology and Evolution, 18, 352­361. Kowalski, K. & Hasegawa, Y. (1976). Quaternary rodents from Japan. Bulletin of the National Science Museum. Series C (Geology and Paleontology), 2 (1), 31­ 66. Kozhamkulova, B. S., Savinov, P. F., Tjutkova, L. A. & Pak, T. K. (1987). Pliocene mammals from the Aktogay locality. Materials on the History of Fauna and Flora of Kazakhstan, 9, 82­120. Kretzoi, M. (1969). Skizze einer Arvicoliden Phylogenie-Stand. Vertebrata Hungarica, 11 (1­2), 155 ­192. Luo, J., Yang, D., Suzuki, H., Wang, Y., Chen, W.-J., Campbell, K. L. & Zhang, Y. (2004). Molecular phylogeny and biogeography of Oriental voles: genus Eothenomys (Muridae, Mammalia). Molecular Phylogenetics and Evolution, 33, 349 ­362. Markova, A. K. (1992). Fossil rodents (Rodentia, Mammalia) from the Sel-Ungur Acheulian cave site (Kirghizstan). Acta Zoologica Cracoviensia, 35, 217­239. Mezhzherin, S. V. & Serbenyuk, M. S. (1992). Biochemic variability and genetic divergence of Palearctic Arvicolidae. The genus Clethrionomys Thilesius, 1850. Genetika, 28 (2), 143­153 [in Russian, English (Abstract)]. Mezhzherin, S. V., Morozov-Leonov, S. Yu & Kuznetsova, I. A. (1995). Biochemical variation and genetic divergence in Palearctic voles (Arvicolidae: Subgenus Terricola, true lemmings Lemmus Link 1975, pied lemmings Dicrostonyx Gloger 1841, steppe lemmings Lagurus Gloger 1842, mole voles Ellobius Fischer von Waldheim 1814. Genetika, 31 (6), 788 ­797 [in Russian, English (Abstract)]. Modi, W. S. & Gamperl, R. (1989). Chromosomal banding comparisons among American and European red-backed mice, genus Clethrionomys. Zeitschrift fÝr SÄugetierkunde, 54, 141­152. Musser, G. G. & Carleton, M. D. (1993). Subfamily Arvicolinae. In D. E. Wilson & D. M. Reeder (Eds) Mammal Species of the World, 2nd edn (pp. 501­535). Washington, London: Smithsonian Institute Press. Musser, G. G. & Carleton, M. D. (2005). Subfamily Arvicolinae. In D. E. Wilson & D. M. Reeder (Eds) Mammal Species of the World, 3nd edn (2, pp. 956­1039). Baltimore: Johns Hopkins University Press. Ognev, S. I. (1950). Mammals of the USS Royal and Adjacent Countries. Mammals of Eastern Europe and Northern Asia, 7. Moscow-Leningrad: USSR Academy of Sciences [in Russian].

562

Zoologica Scripta, 36, 6, November 2007, pp547­563 · © 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters


V. S. Lebedev et al. · Molecular phylogeny of the genus Alticola

Ohdachi, S. D., Dokuchaev, N. E., Hasegawa, M. & Masuda, R. (2001). Intraspecific phylogeny and geographical variation of six species of northeastern Asiatic Sorex shrews based on the mitochondrial cytochrome b sequences. Molecular Ecology, 10, 2199 ­2213. Patterson, N., Richter, D. J., Gnerre, S., Lander, E. S. & Reich, D. (2006). Genetic evidence for complex speciation of humans and chimpanzees. Nature, 441, 1103 ­1108. Pavlinov, I. Ya. (2003). Systematics of modern mammals. Moscow: Moscow State University [in Russian]. Pavlinov, I. Ya. (2006). Myodes Pallas, 1811, is the valid name for the genus of redbacked voles (Cricetidae). Zoologicheskii Zhurnae, 85 (5), 667­ 669. Pei, W. C. (1936). On the mammalian remains from locality 3. Choukoutien. Paleontologica Sinica, C, 7 (5), 70 ­75. Popov, V. V. (2000). Late Pliocene voles (Mammalia: Arvicolidae) from Varshets (North Bulgaria). Acta Zoologica Cracoviensia, 44, 143 ­172. Posada, D. & Crandall, K. A. (1998). MODELTEST: testing the model of DNA substitution. Bioinformatics, 14, 817­818. Rambaut, A. & Drummond, A. (2004). TRACER, Version 1.3 Computer Program Distributed by the Authors. UK: Department of Zoology, University of Oxford. Available via . Rambaut, A. & Grassly, N. C. (1997). Seq-General: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Computers and Applied Biosciences, 13, 303 ­306. Repenning, Ch & Brouwers, E. M. (1992). Late Pliocene­Early Pleistocene ecologic changes in the Arctic Ocean Borderland. US Geological Survey Bulletin, 2036, 1­37. Robinson, M., Gouy, M., Gautier, C. & Mouchiroud, D. (1998). Sensitivity of the relativerate test to taxonomic sampling. Molecular Biology and Evolution, 15, 1091­1098. Robinson-Rechavi, M. & Huchon, D. (2000). RRTree: relativerate tests between groups of sequences on a phylogenetic tree. Bioinformatics, 16, 296­297. Ronquist, F. & Huelsenbeck, J. P. (2003). MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19, 1572­1574. Rossolimo, O. L. & Pavlinov, I. J. (1992). Species and subspecies of Alticola s.str. (Rodentia: Arvicolidae). In I. Horacek & V. Vohralik (Eds) Prague Studies in Mammalogy (pp. 149 ­176). Prague: Charles University Press. Rossolimo, O. L., Pavlinov, I. J. & Shenbrot, G. I. (1992). Taxonomic status of Alticola tuvinicus Ognev. (Mammalia: Arvicolinae). Biologicheskie Nauki, Moscow, 6, 58­ 62 [in Russian, English (Abstract)]. Rossolimo, O. L., Pavlinov, I. Ya. & Hoffmann, R. S. (1994). Systematic distribution of the rock voles of the subgenus Alticola s.str. in the People's Republic of China (Rodentia, Arvicolinae). Acta Theriologica Sinica, 14 (2), 86 ­99. Runck, A. M. & Cook, J. A. (2005). Postglacial expansion of the southern red-backed vole (Clethrionomys gapperi) in North America. Molecular Ecology, 14, 1445 ­1456. Rzhetsky, A. & Nei, M. (1995). Tests of applicability of several substitution models for DNA sequence data. Molecular Biology and Evolution, 12, 131­151. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. Serdyuk, N. V. & Tesakov, A. S. (2006). New form of rhizodont voles (Rodentia, Arvicolinae, Clethrionomyini) from Pleistocene of Central Altai (Russia). Russian Journal of Theriology, 5 (2), 79 ­ 83.

Shenbrot, G. I. & Krasnov, B. R. (2005). An Atlas of the Geographic Distribution of the Arvicoline Rodents of the World (Rodentia, Muridae: Arvicolinae), 45. Sofia-Moscow: Pensoft, Series Faunistica. Smirnov, N. G., Bolshakov, V. N. & Borodin, A. V. (1986). Pleistocene Rodents from the North of Western Siberia. Moscow: Nauka [in Russian]. Swofford, D. L. (2003). PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods), Version 4. Sunderland, Massachusetts: Sinauer Associates. Swofford, D. L., Olsen, G. J., Waddell, P. J. & Hillis, D. M. (1996). Phylogenetic inference. In D. M. Hillis, C. Moritz & B. K. Mable (Eds) Molecular Systematics (pp. 407­514). Sunderland, MA: Sinauer. TegelstrÆm, H. (1987). Transfer of mitochondrial DNA from the northern red-backed vole (Clethrionomys rutilus) to the bank vole (C. glareolus). Journal of Molecular Evolution, 24 (3), 218­227. Tesakov, A. S. (1996). Evolution of the bank voles (Clethrionomys, Arvicolidae). Acta Zoologica Cracoviensia, 39, 541­547. Thorne, J. L. & Kishino, H. (2002). Divergence time and evolutionary rate estimation with multilocus data. Systematic Biology, 51, 689­702. Thorne, J. L., Kishino, H. & Painter, I. S. (1998). Estimating the rate of evolution of the rate of molecular evolution. Molecular Biology and Evolution, 15, 1647­1657. Tjutkova, L. A. & Kaipova, G. O. (1996). Late Pliocene and Eopleistocene micromammal faunas of southeastern Kazakhstan. Acta Zoologica Cracoviensia, 39, 549­557. Vinogradov, B. S. (1927). Notes on Some Mammals of Yakutia 1 LemmingVoles (Genus Aschizomys), 17, 1­14. Leningrad: Commission for Exploration of Yakut Autonomous Soviet Republic [in Russian, English (Abstract)]. Yang, Z. (1997). PAML: a Program Package for Phylogenetic Analysis by Maximum Likelihood. CABIOS, 13, 555 ­556. Young, C. C. (1934). On the Insectivora, Chiroptera, Rodentia and Primates other than Sinanthropus from locality 1 of the Choukoutien. Paleontologica Sinica, 8 (3), 1­160. Zazhigin, V. S. (1988). Paleontological descriptions: Insectivora, Lagomorpha, Rodentia. In E. A. Vangengeim & K. V. Nikiforova (Eds) Biostratigraphy of Late Pliocene­Early Pleistocene of Tadjikistan (Mammalian Fauna) (pp. 18­24). Moscow: Nauka [in Russian]. Zhang, Y. Q. (2004). Morphological variation of molars of Villanyia from Renzidong, Anhui. Proceedings of the Ninth Annual Meeting of the Chinese Society of Vertebrate Paleontology (pp. 93 ­100). Beijing: Ocean Press, [in Chinese, English abstract]. Zhang, Y. Q., Kawamura, Y. & Jin, C. Z. (2006). A new species of the extinct vole Villanyia from the Early Pleistocene of Central China, with comments on the relationship to the Transbaikalian species. In N. V. Alexeeva, M. A. Erbajeva & N. G. Mironov (Eds) Stratigraphy, Paleontology and Paleoenvironment of Pliocene­Pleistocene of Transbaikalia and Interregional Correlations. Abstracts (p. 99). Ulan-Ude: Buryatian Scientific Center of the Siberian Division of RAS. Zheng, S. H. & Li, C. K. (1990). Comments on fossil arvicolids of China. In O. Fejfar & W.-D. Heinrich (Eds) International Symposium Evolution Phylogeny and Biostratigraphy of Arvicolids (pp. 431­ 442). Prague: Geological Survey. Zykin, V. S., Zazhigin, V. S. & Prisyazhnyuk, V. A. (1989). Pliocene Stratigraphy of the South of West Siberian Plain. In V. S. Zykina (Ed.) Cenozoic of Siberia and North-east of the USSR (pp. 9 ­18). Novosibirsk: Nauka [in Russian].

© 2007 The Authors. Journal compilation © 2007 The Norwegian Academy of Science and Letters · Zoologica Scripta, 36, 6, November 2007, pp547­563

563