Phylogenetic relations and systematics of Allactaginae
Intergeneric relations
Our data produced a robust phylogeny for Allactaginae above species level and thereby firmly proved that Allactaga s.l. (as recognised by Holden and Musser17) is paraphyletic to both Pygeretmus and Allactodipus. Both of the latter taxa are morphologically distinct from Allactaga by a number of unique apomorphies: a unique molar pattern and glans penis morphology in Allactodipus as well as high-crowned terraced molars, reduction of the premolar, and particular glans penis morphology in Pygeretmus. At the same time, the morphology of all other five-toed jerboas is relatively monotonous with variation only in terms of body size, relative molar crown height, size of auditory bullae, m1 morphotype frequency, and the rate of M3 reduction1,45. Such level of differences never allowed recognition of more than one genus.
Thus, allactagines represent a case when descendant lineages with derived morphology are nested within a group with overall conserved morphology. This can be compared to paraphyly of white-toothed shrews Crocidura relative to Diplomesodon46, rorquals (Balaenoptera) relative to humpback whales (Megaptera)47, or tits (Parus s.l.) relative to morphologically aberrant ground tit (Pseudopodoces humilis)48. In such cases, the taxonomy should be changed in accordance with the monophyly principle, which is achieved by combining genera (as done in whales) or splitting the genus in question into new taxa (as done in tits). Unfortunately, any decision in this context is arbitrary as it is based on subjective weighting of morphological differences. For Allactaginae, the splitting approach was implemented18, which resulted in the elevation of Scarturus and Orientallactaga to the generic rank2, despite the fact that a synapomorphy-based morphological diagnosis of Scarturus can hardly be formulated.
As an alternative to the morphology-based approach, temporal banding—a method which uses node age as a measure of rank49—was suggested as a standardised method for taxonomic ranking. In the present study, the age of divergence of major Allactaginae lineages was dated to the Pliocene. However, in other groups of Myodonta, Pliocene divergences were found both among genera (as in voles50 or hamsters51) and among congeneric species (as in Sicista52). Thus, the ambiguity remains unresolved; we see no better option than to retain the generic classification established by Michaux & Shenbrot2 (Table S10). However, it should be noted that the inferred age of divergence between S. tetradactylus + S. hotsoni and the VECE clades (3.9–4.1 Mya) is comparable or even larger than the divergence time of Allactodipus from Allactaga. If the temporal criterion (sensu Avise, Johns49) is accepted, one should consider elevating the VECE clade at least to subgeneric rank, with Scarturus proper including only two species. The diagnosis of the new taxon should be polythetic (medium to small jerboas with five-toes, bullae not enlarged, glans penis with longitudinal fold, molar low-to medium crowned, M3 not reduced). Although the name Paralactaga is traditionally used as a subgeneric for the S. euphraticus group and therefore may have been applied to the whole VECE clade, we believe that this is incorrect. The type species of Paralactaga—P. anderssoni Young, 1927—was described from the Late Miocene of China, which is inconsistent with the estimated time of origin of the VECE clade. Apparently all similarities between S. euphraticus group and Paralactaga proper are because of plesiomorphy. Therefore, we suggest that Paralactaga should be attributed to fossil taxa only.
Species groups within Scarturus
In the present study, we analysed in detail the phylogenetic reconstructions and divergence times estimations for the species and species groups of the genus Scarturus. Our study is the first to examine the phylogenetic position of the enigmatic taxon described from Afghanistan and which is currently termed Scarturus williamsi caprimulga. The mitochondrial data provided clear evidence that this taxon is not closely related to any member of the S. euphraticus species group including S. williamsi. Instead, it belongs to a separate divergent lineage of Scarturus, which should be considered a separate species, Scarturus caprimulga. It also includes the jerboa from Kopet Dag provisionally classified by Hamidi et al25 as Paralactaga cf. williamsi. The mitochondrial difference between specimens from Afghanistan and those from Kopet Dag suggested a potential subspecies rank of the latter form, which is provisionally referred to as S. aff. caprimulga. More research on the distribution and genetic structure of this species is needed for further clarification. Our study has added more representative genetic data on the poorly known S. vinogradovi and confirmed it as a separate divergent branch within Scarturus s.l. and likely a distant sister group of S. caprimulga.
Previous phylogenetic reconstructions of the S. euphraticus species group based on mtDNA data recovered a divergent branch within S. euphraticus53, which was subsequently classified as S. aulacotis2. With further addition of comprehensive nuclear data, the full species rank of this taxon is now completely supported. The relationships among the three species in the S. euphraticus group correspond to a hard trichotomy dated to the late Early Pleistocene.
Nuclear data strongly support deep structuring within the S. elater species group, as previously demonstrated using mtDNA19,22,54, and confirmed the species status of S. indicus and S. heptneri. The divergence between S. elater and S. indicus estimated based on the nuclear loci was dated to approximately 1.5 Mya, which was slightly older than the 1.26 Mya inferred from mtDNA by Bannikova et al.22. Both S. indicus and S. elater included allopatric lineages that have separated 600–800 kya (i.e. dzungariae and strandi within elater, and aralychensis within indicus). Their formal taxonomic rank appears controversial: the level of divergence apparently conforms to species rank, whereas genetic data indicates potential gene flow between them. Thus, the mtDNA haplotypes of Scarturus specimens from the Zaisan depression (S. e. zaisanicus) form a subclade within S. elater s.str., whereas nuclear data suggest that S. e. zaisanicus is relatively close to S. e. dzungariae. This pattern suggests that the Zaisan population, while being a derivative of the Dzungar form, experienced mtDNA capture as a result of a past hybridisation event with S. elater. Gene flow between S. strandi and S. elater proper was indicated by the occurrence of elater mtDNA haplotypes in certain populations of strandi from north-western Kyzylkum22. All these taxa require additional research to produce a more accurate evaluation of gene flow intensity. Nevertheless, we suggest that dzungariae, strandi, and aralychensis should be considered semispecies or species in statu nascendi. Taxonomically, we regard them as parts of elater and indicus superspecies and refer to them as S. (elater) dzungariae, S. (elater) strandi, and S. (indicus) aralychensis, respectively.
Phylogenetic relations within Orientallactaga
Within Orientallactaga, O. bullata and O. balikunica were supported as sister taxa based on nuclear data, which is consistent with their common morphology (enlarged bullae). However, mtDNA suggested that O. bullata is a sister taxon to O. sibirica, and the reason for this discrepancy is unclear, with ancient mtDNA introgression being the most obvious explanation. The crown age of Orientallactaga was dated to the early Early Pleistocene (Gelasian). Neither O. bullata nor O. balikunica show substantial intraspecific variation.
In contrast, O. sibirica consists of several genetic lineages, which partly correspond to recognised subspecies. The mtDNA data tentatively supported subdivision of O. sibirica into western and eastern groups separated by the Tianshan–Altay zoogeographic boundary. The structure of variation in the eastern portion of the range (Mongolia, China) is well-studied23; however, the genetic data on the western portion are still fragmentary. Available mtDNA data provisionally support recognition of western subspecies such as O. s. ognevi (north-eastern to central Kazakhstan), O. s. dementjevi (Issyk-Kul region), and O. s. altorum (central Tianshan). The latter two forms are distributed in high-altitude areas of Tianshan, thus indicating that, in contrast to most other jerboa species, mountain areas might serve as foci of diversification in O. sibirica.
The westernmost part of the range (western Kazakhstan, Qyzylkum) was assumed to be inhabited by a single O. s. suschkini subspecies after morphological revision1. However, three divergent mtDNA lineages were recovered based on the preliminary analysis of mtDNA data retrieved from museum specimens from the area, which suggests that the diversity of western populations is likely underestimated and in need of further examination.
The crown age of O. sibirica was estimated at 500–600 kya, which was substantially younger than 2.2–3.2 Mya as inferred by Cheng et al.23; this discrepancy, however, can be explained by mtDNA saturation effects and usage of inaccurate secondary calibrations in their study.
Variation within Allactaga and Pygeretmus
Considering the phylogenetic position of Pygeretmus, our data firmly corroborated its separate phylogenetic position and rejected any affinity with Orientallactaga bullata as reconstructed by Wu et al.55. The latter result should be attributed to identification error. In our study, all three species of Pygeretmus were analysed to confirm phylogenetic proximity of P. shitkovi and P. platyurus relative to P. pumilio. Thus, the subgeneric status of Alactagulus containing the latter species was not contradicted; however, the split age between Pygeretmus s.str. and Alactagulus is relatively young, dated as Pliocene/Pleistocene boundary, indicating that morphological and life history traits of the former (e.g. slower locomotion) have evolved rather recently.
A further taxon demonstrating a complex structure is Allactaga major. Our mtDNA data indicated that A. major consisted of several genetic lineages partly corresponding to morphological subspecies (A. m. spiculum, A. m. djetysuensis). A high level of divergence was observed between specimens from the northern Caucasus and Kazakhstan. A specimen of morphologically distinct A. m. spiculum (north-eastern Kazakhstan, western Siberia) was placed as a sister species to all other A. major with a divergence level compatible with species status.
Several other species included unexpected genetic lineages that were apparently divergent at subspecies level (e.g. a southern Uzbekistan lineage of A. severtzovi and an Ili lineage of P. shitkovi). However, the resolving power of the employed set of 15 nuclear genes is insufficient for clarifying relationships within species. Therefore, these cases should be studied using larger samples and further nuclear loci.
Divergence time estimates within Allactaginae
Our estimated divergence times were generally more recent than those produced by most previous studies. The root node of crown Allactaginae was dated to 7.7 (5.4–9.9) Mya by Wu et al.55, 8.1 (4.2–12.7) Mya by Zhang et al.56, or 8.87 (8.3–9.85) Mya by Pisano et al.4. The results by Wu et al.55 may be affected by a node density effect as their re-analysis with reduced taxon sampling of Allactaginae and Dipodinae produced younger dating at 5.8 (3.1–8.6) Mya. The latter two studies used only one to four nuclear loci and calibrated their analysis using non-Dipodidae calibration points. In both cases, the Early Miocene age of Sicista primus was used to calibrate crown Sicista, which lacks proper justification and may result in upward bias, as argued by Rusin et al.57.
The earliest Allactaginae appeared in the Early Miocene and, in the Middle Miocene, the members of the primitive genus Protalactaga Young, 1927 became a common element of the Asian fauna3. During the Late Miocene, the diversity of allactagines persisted, and new genera emerged including Paralactaga Young, 1927 which is morphologically similar to Allactaga and is often considered its subgenus3,45. However, as can be derived from our results, all but one of the Middle and Late Miocene lineages went extinct without leaving any recent descendants, and all current diversity is a product of the Pliocene–Pleistocene evolution. This diversification pattern is unlike that observed in a different jerboa subfamily, Dipodinae, which includes lineages that had diverged in the Middle and early Late Miocene (Paradipus and Dipus, respectively)4,58.
As estimated here, the onset of radiation among crown Allactaginae occurred in the latest Messinian and thus was nearly coincident with the Messinian crisis. However, it remains unclear how (or whether at all) climatic perturbations at the Miocene /Pliocene boundary affected the evolution of Allactaginae. The results of the diversification analysis suggested that, throughout the Pliocene and Pleistocene, the rate and mode of speciation in five-toed jerboas remained constant, indicating high tolerance of this group towards the climatic changes of this period.
The minimum age of split observed between sympatric species was approximately 1 Mya as demonstrated by heptneri versus elater s.str. (and strandi). This was the estimate for the minimum time necessary for formation of effective reproductive barriers in allactagines (post- or pre-zygotic). Other phylogenetically close sympatric species pairs were S. elater/S. indicus (1.5 My), O. bullata/O. balikunica (1.5 My), and A. major/A. severtzovi (2.0 My).
Geography of speciation
Of 17 analysed episodes of speciation in Pliocene–Pleistocene, the patterns of range fragmentation in 10 episodes matched well to the classical vicariance scenario and those of six episodes matched to the founder-event speciation scenario; in one episode, both scenarios were equally probable. As the location of arising isolation barriers within the ancestor range seemed incidental, only in three cases the ancestors’ range was subdivided into two parts which were more or less equal in size: first, into East and West Central Asia; second, into Turan and Iran; third, into Anatolia with trans-Caucasus and northern Zagros and Levant with northern Mesopotamia and southern Zagros. In all other cases, the ancestors’ range was subdivided into the main part and relatively small peripheral isolates. As can be expected from the modern patterns of species diversity of Allactaginae, the discovered speciation events were unequally distributed: one episode in North Africa, one in the eastern part of Central Asia, three in the Middle East, four in the Iranian highland, four in Turan, and five in Kazakhstan. In most cases, range fragmentation coincided with extreme climate conditions within the analysed time periods: warmest and wettest (decrease of the area of arid lands: nodes 2–3, 5, 10, 12, and 14–15) or coldest and driest (closing narrow mountain passages due to mountain glaciation: nodes 4, 6–9, 13, and 16–18). In one case (node 11), fragmentation of the range coincided with moderate climate conditions.
Successful modelling of fragmentation of geographic ranges as a base of speciation events seemed to agree with the hypothesis of Peterson et al.15, which states that ecological niches evolve little at or around the time of speciation events, whereas niche differences accumulate later. This hypothesis was supported by Peterson’s analysis59 of data published between 1999 and 2008 which demonstrated that niche conservatism was found in more than 70% of comparisons within species and between sister species, but in less than 50% of comparisons among closely-related (but not sister) species and across monophyletic lineages of species. Moreover, analysis of habitat niche evolution of arvicoline rodents16 demonstrated that closely related species with allopatric or parapatric distribution demonstrated small niche differences, whereas they were larger in species with sympatric distribution. This is a clear indication that interspecific competition forces natural selection to increase niche differences resulting in species co-occurrence. It was demonstrated that niche divergence/conservatism can be differently expressed between different niche/resource axes60. In voles, which have a highly specialised folivorous diet, habitat segregation seems to be the only type of niche differentiation. Closely related Allactaginae species are similar in diet and typically occur in allopatric or parapatric distribution patterns1, which may indicate their niche conservatism. The only exception to a pattern where species with similar diets show widely overlapping geographic distributions are Scatrurus elater and S. heptneri (these two species are similar in both, macro- and micro-habitat niches, and it is unclear which mechanisms allow them to co-occur22). Distantly related sympatric species typically show similarities regarding macro-habitat niches but marked differences in terms of micro-habitat niches (Allactaga major and Orientallactaga sibirica; O. sibirica and O. bullata; O. sibirica and O. balikunica; Pygerethmus pumilio and P. platyurus; P. pumilio and P. shitkovi; personal observations) and diet (Allactaga and Allactodipus; Allactaga and Scarturus; Allactaga and Pygeretmus; Orientallactaga and Pygeretmus; Scarturus and Pygeretmus1,61). Thus, macro-habitat niche conservatism may be expected even in sympatric species.
Source: Ecology - nature.com
