In the current study, we, for the first time, sequenced and de novo assembled the complete mitochondrial genome of a rare rodent endemic to Kazakhstan, the Desert Dormouse Selevinia betpakdalaensis, for which no genetic data has been obtained so far. In addition, we assembled complete mitochondrial genomes of three representatives of Gliridae (Myomimus roachi and Glirulus japonicas) obtained from museum specimens, and Graphiurus murinus from SRA data published previously).
The gene order and organisation of the mitochondrial genome of S. betpakdalaensis, is similar to those of other Gliridae representatives, and typical for other vertebrates18. The Mitochondrial genome of S. betpakdalaensis is characterised with average GC value (37.4%) compared with other Gliridae mitochondrial genomes (where it was about 34.6–38.6%).
The results obtained clearly demonstrate the strong influence of saturation on the topology, moreover, the topology may be erroneous, but well supported (Fig. 2A). Saturation tests (Table S3) visualise the effect of saturation on the phylogeny obtained (Fig. 2). As a result of the BI analysis with the complete exclusion of the 3rd codon position (Fig. 2C), the trichotomy, with an uncertain position of the cluster of garden and forest dormice is observed. Meanwhile, ML analysis in both variants (with the exception of transitions in 3rd codon position and the exclusion of the entire 3rd position) shows the same result, identical to both, ML and BI, phylogenies obtained on a complete set of genes and species (Fig. 3). These results are well consistent with modern classification1. At the same time, ML analysis showed it to be more resistant to saturation.
Phylogenetic reconstruction of 17 Gliridae species inferred from the 13 PCGs (3rd codon position excluded due to saturation) and 12S concatenated alignment revealed the tree topology in general similar to the previously published by Bover et al.2 – three subfamilies: Glirinae, Graphiurinae and Leithiinae were identified. Nevertheless, our result allowed us to resolve some complex nodes and obtain a supported topology. Clustering of Glirinae and Graphiurinae was only shown by Bover et al.2, however, support of this cluster was low (bpp = 0.49). In the same paper, the monophyly of the Leithiinae was shown, but the branching order within it was not resolved. Our results resolved the topology within Leithiinae – the cluster of Myomimus and Selevinia turns out to be early derivative, and Muscardinus is sister to the group of Eliomys and Dryomys. In the previous molecular study2, the genus Muscardinus was basal for the subfamily Leithiinae, probably as a result of saturation, see upper section. On the other hand both the basal position of Muscardinus and the lack of support for Myomimus in their study may be related to incomplete taxa sampling, namely absence of Selevinia. It should be mentioned here that Montgelard et al.15 also showed the basal position of Muscardinus within Leithiinae with good support in the study involving 12s and some nuclear genes.
Trying to combine morphological and molecular data for living and extinct Gliridae, Dalmasso et al.19 performed Bayesian divergence dating including fossil species (as tip dates) alongside their living relatives in the tree-building process (the fossilised birth–death models). As a result, it turned out that Dryomys (Eliomys was not analysed) turned out to be phylogenetically closer to Glirulus than to Myomimus, which fits our most saturated results (Fig. 2A,B,D).
As for the primary goal of our study, the first molecular data obtained for Selevinia betpakdalaensis strongly support its position as sister to Myomimus roachi, as it was showed earlier in the studies based on the middle ear features, cranial and mandibular morphology11,12,20.
The divergence dates based on 13 PCGs and 12S rRNA as a whole does not differ much from the estimates made in the previous studies (Table 1), except that we manage to significantly reduce the confidence intervals.
Table 1
Divergence ages of the main nodes reported in this and previous studies. All ages are in millions of years ago.
Node | Current study | Bover et al., 2020 | Mouton et al., 2017 | Mouton et al., 2012 | Nunome et al., 2007 | Montgelard et al., 2003 |
Gliridae | 34.6 (28.03–41.04) | 38.5 (26.91–50.08) | n/a | n/a | 55 | 50 |
Leithiinae | 31.24 (25.04–37.27) | 31.0 (20.6–41.4) | n/a | n/a | n/a | 40.8 (37.0-44.6) |
Selevinia — Myomimus | 23.98 (17.9-30.44) | n/a | n/a | n/a | n/a | n/a |
Eliomys — Dryomys | 24.44 (18.86–30.09) | 23.15 (14.47–31.84) | 18.46 (13.08–24.4) | 6.96 (4.87–8.88) | 14.5 (12.1–16.9) | 28.5 (25.7–31.3) |
Eliomys, Dryomys — Muscardinus | 28.41 (22.66–34.31) | n/a | n/a | n/a | 22.3 (19.5–25.1) | n/a |
Glirinae | 28.34 (21.56–35.05) | 28.7 (16.1–41.3) | n/a | n/a | 27.0 (24.1–29.9) | 27.7 (24.7–30.7) |
Graphiurinae | 16.64 (11.43–21.97) | 17.02 (9.00-25.04) | n/a | n/a | n/a | 8.7 (7.7–9.7) |
Thus, the age of the Gliridae family was estimated as 34.6 (28.03–41.04), which is slightly less than the estimate of Bover et al.2, based on the same root calibration, and equal to 38.5 (26.91–50.08) million years. In the work of Dalmasso et al.19, the average age of the node uniting modern representatives of Glirulus, Dryomys and Myomimus (representatives of Graphiurus were not used in that study) was estimated at about 30 Mya.
Our estimate for the MRCA of subfamily Leithiinae is 31.24 (25.04–37.27) Mya, that agrees well with the average estimate 31.0 (20.6–41.4) made by Bover et al.2 based on the 1330 bp fragment (CYTB and 12S).
The age of the subfamily Glirinae, estimated in our study as 28.34 (21.56–35.05) Mya, turns out to be very close to the results of all previous studies (Table 1), both based on nuclear16 and mt genes2, and their combination15.
The mean age of the subfamily Graphiurinae estimated as 16.64 (11.43–21.97) Mya is also similar to the previous estimate of Bover et al.2 at about 17 Mya.
A fundamentally new result obtained in our study is the determination of the divergence time of Selevinia from Mouse-tailed Dormouse with an average of 23.98 (17.9–30.44) Mya, that is, approximately the Oligocene-Miocene boundary.