The specimens of Asiatic short-tailed shrews were collected during the surveys of small mammals conducted by the Russian Academy of Sciences and the Chinese Academy of Sciences in Gansu and Sichuan provinces of China. Voucher specimens are deposited in the Zoological Museum of Lomonosov Moscow State University (ZMMU). These specimens were compared with the Asiatic short-tailed shrews kept in the collection of the Natural History Museum, London, UK (NHMUK) and the Zoological Institute of the Russian Academy of Sciences, Saint Petersburg, Russia (ZIN). Among them, the holotype of B. griselda (NHMUK 1912.8.5.23) was genetically studied for the first time. On the whole, we obtained 23 new sequences from 11 specimens of Blarinella and two specimens of Chodsigoa hypsibia (Table 1). For the phylogenetic analysis, 47 mitochondrial and nuclear sequences of Blarinella, Blarina, Chodsigoa, Chimarrogale, Neomys, Anourosorex, and Sorex from our previous studies (Abramov et al. 2017b; Bannikova and Lebedev 2010; Bannikova et al. 2015, 2017, 2018) and additional 116 sequences of different genera of Soricinae as well as Crocidurina (Crocidura fuliginosa) from GenBank were used (see Suppl. material 1: Table S1).

Figure 1. 

Sample localities of specimens used for molecular analyses 1 China, Gansu, Lingtan County (NHMUK 12.8.5.23, holotype of Blarinella griselda Thomas, 1912) 2 China, Gansu, Taizishan NR (ZMMU S-195179) 3 China, Sichuan, Ruoergai (Zoige) (ZMMU G17-87) 4 China, Sichuan, Songpan (ZMMU G18-252) 5 China, Shaanxi, Mt.Qinling (after He et al. 2018) 6 China, Gansu, Huixian (ZMMU G17-12) 7 China, Sichuan, Baoxing (type locality of Blarinella quadraticauda (Milne-Edwards, 1872)). Unnumbered localities based on the GenBank data.

Genomic DNA from ethanol-preserved tissues of the recent specimens was extracted using a standard protocol of proteinase K digestion, phenol-chloroform deproteinisation and isopropanol precipitation (Sambrook et al. 1989). We sequenced the complete mitochondrial cytochrome b (cytb) gene and fragments of four nuclear loci: apolipoprotein B (ApoB), exon 11 of the breast cancer type 1 susceptibility protein (BRCA1), recombination activating gene 2 (RAG2), and the interphotoreceptor retinoid binding photoreceptor (IRBP). Primers and polymerase chain reaction protocols for nuclear loci (ApoB, BRCA1, IRBP and RAG2) are described in Abramov et al. (2017a) and Bannikova et al. (2018). New primers were designed specifically for amplification and sequencing the complete cytb gene and its short fragments from the historical museum specimen, the holotype of B. griselda (Suppl. material 1: Table S2). General methods for the amplification and sequencing of cytb from recent samples are described in Bannikova et al. (2010). PCR products were sequenced on the autosequencing system ABI 3100-Avant using ABI PRISM BigDyeTM Terminator v. 3.1 (Applied Biosystems, Foster City, CA, USA).

Molecular analysis (DNA extraction and PCR preparation) of the holotype of B. griselda NHMUK1912.8.5.23 was performed in the laboratories of Department of Bioinformatics and Genetics of the Swedish Museum of Natural History in the special laboratory for historical museum samples. DNA was extracted from a ~1.5 mm × 1.5 mm skin sample that was washed in ethanol prior to initiating the extraction procedure.

Extraction of DNA was performed using Qiagen QIAamp DNA Micro Kit following the protocol “Isolation of Genomic DNA from Tissues” with some changes: (1) additional 5 µl of Proteinase K after overnight lysis and incubation at room temperature for 30 min; (2) two steps of elution with AE buffer, each with 20 µl of buffer and 5 min of incubation at room temperature. Amplification of cytb fragments was performed in 25 µl reaction volume containing 1–2 µl DNA, 2.5 µl 10× buffer, 1 µl of each primer (10 pmol/µl), 0.5 µl of dNTP mix, 1 µl MgCl2, 2.5 µl BSA, and 0.4 µl Taq polymerase. Extraction was performed twice, the second time with blank as a negative control. Double-stranded polymerase chain reaction entailed 50-thermal cycles and was performed as follows: 95 °C for 10 min, (94 °C for 30 sec, 52 °C for 30 sec, 72 °C for 30 sec) ×50 cycles, 72 °C for 7 min, 12 °C for forever. Negative controls were used both for DNA and PCR mix. PCR products were verified on 1% agarose gels stained with Gel Green. Primer pairs which resulted in bands on the gel and empty negative controls: L467x – H601x, L240a – H400a, L62x – H190x, L170ax – H330x, L580x – H670x. Amplicons were sequenced directly by Sanger sequencing on Applied Biosystems 3130xl Genetic Analyzer. Each fragment was sequenced several times to ensure the authenticity of the sequence.

It is known that DNA undergoes degradation over time (Hofreiter et al. 2001), such that mtDNA sequences from even relatively recent museum specimens can exhibit sequencing artifacts (Sefc et al. 2007). Most errors involves C→T changes, presumably due to the deamination of cytosine bases in the template (Hofreiter et al. 2001). Here, neither clear double C/T peaks nor an excess of C→T transitions were observed in the cytb sequences obtained from the historical Blarinella sample relative to modern samples, suggesting that the holotype sequences are authentic.

The sequences obtained in this study can be accessed via GenBank (accession numbers: MN199101 to MN199123, Suppl. material 1: Table S1).

All sequences were aligned by eye using Bioedit v. 7.0.9.0 (Hall 1999). Heterozygous positions in nuclear genes were coded using the IUB ambiguity codes and sequences were used as unphased genotypes. The ModelFinder routine (Kalyaanamoorthy et al. 2017) as implemented in IQTREE v. 1.6 (Nguyen et al. 2015) was used to determine the optimum partitioning scheme and the best-fit substitution models for each subset under the BIC criterion.

Phylogenetic reconstructions were performed with each nuclear gene separately and all nuclear genes combined. Phylogenetic trees were reconstructed from nuclear concatenation under Maximum Likelihood (ML) and Bayesian criteria. Maximum likelihood reconstructions were conducted in IQTREE v. 1.6 (Nguyen et al. 2015). Clade stability was tested using Ultrafast Bootstrap (Minh et al. 2013) with 10 000 replicates. Bayesian tree reconstructions were performed in MrBayes v. 3.2 (Ronquist et al. 2012). Models with either two or six rate matrix parameters were selected for each subset using ModelFinder. For most parameters, default priors were used. Compound Dirichlet priors for branch lengths combined with gamma prior on the tree length were invoked. All parameters except branch lengths were unlinked across partitions. The analysis included two independent runs of four chains with the default heating scheme. The chain length was set at 20 million generations with the sampling of every 10 000 generation. Tracer v. 1.6 software (Rambaut and Drummond 2003) was used to check for convergence and to determine the necessary burn-in fraction, which was 10% of the chain length. The effective sample size exceeded 200 for all estimated parameters.

The mitochondrial phylogeny of Blarinella was generated from the cytb alignment containing 46 sequences of Asiatic short-tailed shrews including the partial sequence of the holotype of Blarinella griselda. The ML and Bayesian trees were reconstructed as described above and rooted using Blarina and Cryptotis as the outgroups. In addition, we performed the analyses of the extended set of taxa of Soricinae with the aim to examine more thoroughly the relationships among taxa of the Blarinini-Blarinellini clade and other soricine genera. The p-distances were calculated in PAUP* v. 4.0b10 (Swofford 2003).

Karyotypes of two Asiatic short-tailed shrews, a male ZMMU S-195179 (ID Chi111) from southern Gansu, Taizishan and a female G17-12 from southern Gansu, Huixian were examined. Mitotic chromosome preparations were made in the field from both bone marrow and spleen after colchicine treatment in vivo following Ford and Hamerton (1956) with some modifications (Bulatova et al. 2009).

In case of the male, preparations were made from spleen using a simple technique without centrifugation proposed by Krysanov et al. (2009). Briefly, after colchicine treatment in vivo an incised spleen incubates with 5 mL of KCl hypotonic solution (0.07 M) for 20–30 min at room temperature, and fix with 5 mL of freshly prepared glacial acetic-methanol (1:3) for 5 min twice. Such samples can be stored at −10 °C up to 6 months. To prepare air-dried slides a fixed tissue incubates with 200 µl of 50% glacial acetic acid for 3–4 min, resuspends, and then drops suspension onto a hot slide (30 min at 90 °C). After drying, a slide incubates with pure methanol for 5 min, and then dried again. Air-dried chromosome spreads of both specimens were stained conventionally with 4% Giemsa for 8 min.

CBG-banding was performed using the standard technique (Sumner 1972) to determine C-heterochromatin blocks (for the female karyotype only).

The total matrix used in the cytb analysis (74 sequences, 1140 bp) contained 46 specimens of Blarinella and 28 specimens of other soricids. Models for the cytb gene estimated by IQTREE and employed in the Maximum likelihood analysis were as follows: 1^st codon position TIM2e+I+G4, 2^nd codon position TIM3+F+I+G4, and 3^rd codon position TIM2+F+I+G4.

In the combined analyses of four nuclear genes, the final alignment consisted of 2914 nucleotide positions, including 472 bp of ApoB, 840 bp of BRCA1, 741 bp of RAG2, and 861 bp of IRBP. In total, the nuclear dataset contained 34 specimens, including 19 outgroups. We also performed a separate analysis of the extended ApoB data (44 sequences, 472 bp) because this nuclear gene is represented by the largest number of Blarinella sequences in GenBank. The best-fit substitution models employed for each of the five partitions found by IQTREE are given in Suppl. material 1: Table S3.

As a result of the genetic analysis of the type specimen of B. griselda, we obtained sequences of three fragments of cytb: 90, 160 and 120 bp. The analysis of these fragments showed that the holotype of griselda is very close to our specimen from southern Gansu, two other specimens from northern Sichuan and three specimens from Shaanxi named as Pantherina in He et al. (2018) (Fig. 2; Suppl. material 1: Fig. S1). Together all seven specimens form a clade which appears distinctly separate from all the species of Blarinella (p-distance ~19%).

Figure 2. 

The phylogenetic relationships in Blarinella as reconstructed in MrBayes based on cytb data. Numbers above or below branches correspond to Bayesian posterior probabilities and ML bootstrap values (>50%) generated using fast bootstrap algorithm in IQTREE. The genera Blarina and Cryptotis are used as outgroups.

Overall, three clades of the species of Blarinella may be recognized in the cytb tree: (I) the first clade consists of B. wardi (described from Myanmar and found also in western Yunnan); (II) the second one corresponds to B. quadraticauda: these are specimens previously identified as B. griselda from different localities in China and northern Vietnam and B. quadraticauda from Sichuan Province, including specimens from Baoxing (the type locality of B. quadraticauda); this clade stands as a sister branch to B. wardi; (III) the third clade contains the holotype of griselda, one specimen from southern Gansu, two specimens from northern Sichuan and three specimens from Shaanxi; it is highly divergent from clades I and II. Based on these data combined with morphological and nuclear results presented below, we consider this third clade rather as a separate genus, hereinafter referred to as Parablarinella. A detailed justification of this decision and the description of the new taxon is given in the Discussion.

Phylogenetic analysis of the relationships of the species of Blarinella based on nuclear genes (Fig. 3; Suppl. material 1: Fig. S2) supported the separate position of the specimens that were close to the holotype of B. griselda on the mitochondrial tree (Fig. 2). This clade occupied the sister position to all the remaining species of Blarinella (0.99/94). This pattern is consistent with the previous results by Bannikova et al. (2017) and He et al. (2018). The genetic distance between the B. griselda clade (Parablarinella) and the B. quadraticauda clade is higher than that between the genera Blarina and Cryptotis (p-distance ~0.5% and 0.038% accordingly). The specimens of B. quadraticauda formed a single clade sister to B. wardi (1.0/100).

Figure 3. 

MrBayes tree of Soricinae genera as inferred from the concatenation of four nuclear genes. Numbers above or below branches correspond to Bayesian posterior probabilities and ML bootstrap values (>50%) generated using fast bootstrap algorithm in IQTREE. Crocidura fuliginosa is used as outgroup.

The relatively close relationship of the Blarina / Cryptotis clade to the Blarinella / Parablarinella clade is clearly shown in Figure 3, demonstrating that these taxa form a clade separate from the other soricid outgroups. Divergence times as inferred from nuclear concatenated data using BEAST (Table 2) are nearly half of those obtained in He et al. (2018) based on mitogenomic data, which is likely explained by a bias due to saturation in the fast evolving mitochondrial DNA. The estimated divergence time between Parablarinella griselda and Blarinella proper is 9.61 My (95% HPD = 6.87–12.74), that is ~1.5 times higher than the most recent common ancestor (tmrca) of Blarinini. The tmrca of B. quadraticauda and B. wardi was estimated at 1.68 My (95% HPD = 0.71–3.25).

The three species are very similar in external appearance; the tail is approximately half the length of the head and body (48–63% in B. quadraticauda, 50–61% in B. wardi and 47–51% in B. griselda). The eyes are small, the ears small and almost completely concealed in the pelage, the claws on all feet are moderately large, and a gland is indicated on the mid-ventral surface of males.

Despite the marked genetic divergence between “B. griselda” (Parablarinella) and the other two species, differences in cranial and dental morphology are comparatively limited and not as great as might be expected to distinguish genera, and for some characters there is equal variation between B. quadraticauda and B. wardi as between either one of these species and Parablarinella. In their description of the new genus, He et al. (2018) presented characters to separate B. griselda from the other two species. Here, based on historical material and the recently acquired specimens available to us, we elaborate on these characters and provide additional ones. The craniodental characters that in combination serve to distinguish the three taxa, and those that separate the two genera, are shown in Table 3 and Figures 4–6.

Figure 4. 

Skulls from left to right of the holotype of Parablarinella griselda NHMUK 1912.8.5.23; Blarinella quadraticauda NHMUK 1911.2.1.59; the holotype of Blarinella wardi NHMUK 1915.2.1.3 (please note that the number written incorrectly as 12.2.1.3 on the skull of this species should read 15.2.1.3). Top row: dorsal view; middle row: ventral view; lower row: left lateral view.

Figure 5. 

Variation in morphology of the talonid of the lower molars A lingual view of left mandibular ramus of Parablarinella griselda ZMMU G18-252 B lingual view of left mandibular ramus of Blarinella quadraticauda ZMMU G17-12 C lingual view of right mandibular ramus of holotype of Blarinella wardi NHMUK 1915.2.1.3 D comparison of right lower molars to show variation in development of the entoconid and entoconid crest on m1 and m2 and the talonid of m3. Above holotype of Parablarinella griselda NHMUK 1912.8.5.23, middle Blarinella quadraticauda NHMUK 1911.2.1.57, below holotype of Blarinella wardi NHMUK 1915.2.1.3. The arrows indicate the entoconid and entoconid crest on m2. Scale bar: 1 mm.

Figure 6. 

Comparison of lingual view of posterior region of right mandible to show variation in mandibular and ramal foramina. Mandibular foramen: horizontal arrow; ramal foramen: vertical arrow. From left to right: holotype of Parablarinella griselda NHMUK 1912.8.5.23; Blarinella quadraticauda NHMUK 1911.2.1.59; holotype of Blarinella wardi NHMUK 1915.2.3. Scale bar: 1 mm.

The karyotyped specimens were assigned to B. quadraticauda and “B. griselda” (Parablarinella) based on the combination of molecular and morphological traits.

Blarinella quadraticauda. The diploid chromosome number of the studied female (G17-12) was 2n = 49, and the fundamental autosome number (NFa) was 62 (Fig. 7A). The autosomal complement was represented by the largest polymorphic metacentric pair (# 1), three pairs of large submetacentrics (# 2–4), one medium-sized (#5) submetacentric pair, three medium-sized (#6–8) metacentric pairs, and 15 pairs of medium-sized to small acrocentrics (#9–23). The X chromosomes were medium-sized submetacentrics.

Figure 7. 

The female karyotype of Blarinella quadraticauda (G17-12) with 2n = 49, NFa = 62: routine Giemsa staining (A) and CBG-banding (B).

C-heterochromatic blocks (Fig. 7B) were revealed in the pericentric regions of all acrocentric autosomes (#9–23) as well as in four pairs of bi-armed autosomes (#4 and 6–8). Several C-blocks found in one submetacentric pair (#3) were localized interstitially and at telomeric region of the short arm. The polymorphic pair (#1) shown C-positive blocks in pericentric regions of both acrocentrics while the homologous metacentric was C-negative. Two pairs of medium-sized submetacentrics (#2 and 5) were C-negative. The X chromosomes had C-positive short arms.

Parablarinella griselda. A short description of chromosome set of this male (2n = 49; NFa = 50; ZMMU S-195179) was previously reported in Sheftel et al. (2018) under the name Blarinella cf. griselda. Here we describe this karyotype in more details and present a karyogram of this specimen for the first time (Fig. 8). The studied male (ZMMU S-195179) had 2n = 49; NFa = 50. The autosomal part of the karyotype consisted of one largest metacentric pair (#1), one large polymorphic submetacentric pair (#2) and 21 pairs of medium-sized to small acrocentrics (#3–23). The X chromosome was the medium-sized submetacentric and the Y was a small acrocentric. Only conventional Giemsa staining was applied for this specimen because of a poor quality of chromosome suspension.

Figure 8. 

The male karyotype of Parablarinella griselda (ZMMU S-195179) with 2n = 49; NFa = 50.

There is no doubt from the results of the current study that the clade comprising the holotype of Blarinella griselda and a few other specimens from southern Gansu, northern Sichuan and Shaanxi is the true griselda clade, which is highly divergent from the B. quadraticauda + B. wardi clade. We believe that griselda should be attributed to a separate genus because the age of divergence corresponds to that among recognized genera in the Soricinae (Dubey et al. 2007; He et al. 2010; our data). Recently, He et al. (2018) suggested the same taxonomic decision. However, without including comparisons with the holotype or topotypical material they had insufficient evidence for their attribution of the new name to B. griselda Thomas, 1912, and unfortunately the name Pantherina that they chose is already preoccupied by the same name for a subgenus of African beetles (Agrilus (Pantherina) Curletti, 1998). Thus, we provide a new replacement name for this genus: Parablarinella nom. nov. While some of the specific morphological traits of the new genus are highlighted in the description by He et al. (2018), a more detailed account is warranted and an enhanced diagnosis of the new genus is provided.