A Transitional Gundi (Rodentia: Ctenodactylidae) from the Miocene of Israel

We describe a new species of gundi (Rodentia: Ctenodactylidae: Ctenodactylinae), Sayimys negevensis, on the basis of cheek teeth from the Early Miocene of the Rotem Basin, southern Israel. The Rotem ctenodactylid differs from all known ctenodactylid species, including Sayimys intermedius, which was first described from the Middle Miocene of Saudi Arabia. Instead, it most resembles Sayimys baskini from the Early Miocene of Pakistan in characters of the m1-2 (e.g., the mesoflexid shorter than the metaflexid, the obliquely orientated hypolophid, and the presence of a strong posterolabial ledge) and the upper molars (e.g., the paraflexus that is longer than the metaflexus). However, morphological (e.g., presence of a well-developed paraflexus on unworn upper molars) and dimensional (regarding, in particular, the DP4 and M1 or M2) differences between the Rotem gundi and Sayimys baskini distinguish them and testify to the novelty and endemicity of the former. In its dental morphology, Sayimys negevensis sp. nov. shows a combination of both the ultimate apparition of key-characters and incipient features that would be maintained and strengthened in latter ctenodactylines. Thus, it is a pivotal species that bridges the gap between an array of primitive ctenodactylines and the most derived, Early Miocene and later, gundis.


Introduction
Neev [1] indicated the discovery of large mammals in the Neogene of the Rotem Basin (Israel) based on unpublished data of other geologists.Savage and Tchernov [2] mentioned again the discovery of macromammals (mostly proboscideans) in the Gidron member of the Hatzeva Formation of the Rotem and Yeroham basins and provided provisional identifications.They suggested that these fossils could be Early Burdigalian in age (MN3 equivalent 18Ma) or even older.Goldsmith et al. [3] first signaled the presence of rodent remains (such as incisors) in the Rotem Basin (see Fig 1), which they identified as: Megapedetes sp., Metasayimys sp., and Cricetodon sp.
They also mentioned the presence of a variety of other mammals as well including birds, reptiles, and fishes ( [3]: fig.2).A few of these taxa, but no rodents, were also said to be present in the Yeroham Basin ( [3]: fig.1).The fossils were found in a site called Anthracothere Hill, which is situated on an East-facing escarpment of a North-South oriented ridge, about 100 m above the wadi in the north Rotem Basin.The fossil bearing area is situated about 60 m above the ridge base.Rodents were recovered at the lowest levels (Goldsmith et al., unpublished data).Goldsmith et al. [3] estimated the age of these fossils at 16-17 Ma (MN4 equivalent).A more detailed report was published by Tchernov et al. [4].These authors identified the rodents from the Rotem Basin as Megapedetes cf.pentadactylus, Metasayimys sp., and "a probable representative of the Bathyergoidea and (?) Cricetodontidae" ( [4]: 301).They suggested that the Negev fauna correlated with lower MN3.Goldsmith et al. [5] discussed the age of the mammals from the Hatzeva Formation.They considered the fauna a correlate of the upper MN3 and based also on radiometric dating of Rusinga [6], whose fauna is comparable, concluded that it is 17.5-17.0Ma.Savage [7] discussed the age of the Negev fauna and found that it can hardly be differentiated from Gebel Zelten, which he correlated with uppermost MN3.Goldsmith et al. [8] radiometrically-dated oysters from a level a few meters above the vertebrates at 17.9 Ma.Wood and Goldsmith [9] re-examined the micromammal fauna and suggested the presence of a "new species of primitive pedetid, provisionally referred to Megapedetes", Metasayimys cf.intermedius, (?) Bathyergoides sp., and an indeterminate species that was either a cricetid or a sciurid.As part of a study of the Miocene fauna in Israel, we re-examined the rodent material from the Rotem Basin housed at the National Natural History collections of the Hebrew University of Jerusalem.Preliminary results hinted at the presence of only a pedetid and a ctenodactylid in the early Miocene of the Rotem Basin [10].The aim of the present paper is to describe in detail this ctenodactylid and analyse the significance of this novel taxon in its full extent.
The new specimens have been described and compared with all the valid species of Ctenodactylinae as recognized by López-Antoñanzas and Knoll [22] and a few further species erected subsequently [23,24].However, the detailed comparisons reported in full below are only those carried out with the species considered to be the closest relatives to the new Israeli taxon (Prosayimys spp.and "Sayimys" spp.).
First, second, and third lower molars are designated as m1, m2, and m3, respectively, and first, second, and third upper molars as M1, M2, and M3, respectively.Lower and upper permanent premolars are designated as p4 and P4, respectively, and lower and upper deciduous premolars as dp4 and DP4, respectively.The terminology used in the tooth descriptions follows the works of Baskin [16] and López-Antoñanzas and Knoll [22].
The occlusal measurements (greatest length and greatest width; Table 1) of the teeth of Sayimys negevensis sp.nov.from Israel have been obtained with a Nikon digital counter CM-6S measuring device.
The evolutionary history of ctenodactyline rodents have been clarified recently [22,25].The analysis of the phylogenetic relationships of the Rotem gundi that we present in this work builds upon the character/taxon matrix from López-Antoñanzas et al. [25].Karakoromys Matthew and Granger, 1923 [26] and Tataromys Matthew and Granger, 1923 [26], basal ctenodactylid genera [24,27], were selected as the outgroup.The ingroup included all the valid species of Ctenodactylinae known to date except for those having over 50% of missing data [25].The informative dental characters used in this work are listed in S2 File.
The data matrix (S3 File) was processed with TNT [28] with the "traditional search" option (using TBR).

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new name contained herein is available under that Code from the electronic edition of this article.This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN.The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/".The LSID for this publication is: urn:lsid:zoobank.org:pub:E4252CEA-BB80-4994-832A-97F70627EF80.The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Systematic Paleontology
Order RODENTIA Bowdich, 1821 [29] Family ctenodactylidae Gervais, 1853 [30] Genus SAYIMYS Wood, 1937   Differential Diagnosis: Distinct from Prosayimys flynni in the absence of a distinct metalophulid II separated from the metalophulid I on the lower molars, in having the mesoflexid shorter and deeper than the metaflexid and having a well-developed posterolabial cingulum on the lower molars.Differing from Sayimys obliquidens Bohlin, 1946 [32] in the absence of a distinct metalophulid II on the lower molars.Distinct from Sayimys baskini López-Antoñanzas and Sen, 2003 [33] and Sayimys sivalensis in having a well-developed paraflexus that is longer than the metaflexus, and from the latter also in having a metalophulid II and an open V-shaped mesoflexid shorter than the metaflexid.Differing from Sayimys obliquidens, Sayimys assarrarensis, and Sayimys giganteus in the wear pattern of the upper molars (the paraflexus is more affected by wear than the metaflexus) and from the latter also in having the M3 more posteriorly reduced.Different from Sayimys intermedius in having the mesoflexid shorter than the metaflexid and having a well-developed posterolabial edge on the lower molars.Sayimys negevensis sp.nov.differs from Pireddamys De Bruijn and Rümke, 1974 [34] and Sardomys De Bruijn and Rümke, 1974 [34] in the absence of a distinct independent metalophulid II on the lower molars.Additionally, it differs from Sardomys in lacking cement infilling the valleys on the lower molars.Sayimys negevensis sp.nov.greatly differs from that of Irhoudia Jaeger, 1971 [35] and Pellegrinia De Gregorio, 1887 [36] in the morphology of its cheek-teeth.For instance, it is much less hypsodont, lacks cement infilling in the valleys of the molars, and has four-lobed upper molars.Sayimys negevensis sp.nov. is also distinct from Metasayimys Lavocat, 1961 [21] in the absence of cement filling the valleys of the molars.Finally, the new Israeli species differs from all species of Africanomys Lavocat, 1961 [21] in having the M3 less reduced posteriorly.

Lower incisors
The lower incisors are rounded and usually ungrooved.

Lower molars
Unfortunately, only two lower molars have been found, a m1 and a m1 or m2 (AH2211 and AH1938, Fig 2K and 2L-2N, respectively).The outline of these teeth is subrectangular in occlusal view.The mesoflexid is wide, open V-shaped and clearly shorter and deeper than the metaflexid.These teeth show a remnant of metalophulid II, which is combined with the metalophulid I in a single loph (see discussion below).The hypolophid is roughly transverse and does not oppose exactly the hypoflexid.The protoconid is larger and extends more labially than the hypoconid.The posterolophid does not constrict before reaching the triangular wear surface of the hypoconid.These teeth show a low and well-developed cingulum on their posterolabial side.

Upper incisors
The upper incisors are characterized by having a longitudinal groove located close to the labial side of the tooth.

DP4
The only available DP4 is very badly worn (AH2143, Fig 2D -2F).Due to the large degree of dental wear of this specimen, the hypoflexus seems to be very short and the hypostria very shallow.This tooth has three roots.

M1-2
The M1 is likely smaller than the M2.The occlusal outline of the M1 and the M2 is sub-quadrate.The protoloph connects to the medial or posterior margin of the protocone.The anteroloph and the protoloph and the metaloph and the posteroloph are fused in an early stage of wear.Therefore, the paraflexus and the metaflexus have disappeared in all specimens except for the less worn of them (AH2051 and AH1874, see Fig 2A-2C and 2H-2J respectively), in which they are well-developed.The upper molars of this new taxon seem to have a wear pattern in which the paraflexus is more affected by the increase in wear than the metaflexus and, therefore, it experiences a more rapid shortening than the metaflexus.In case of heavy wear, the paraflexus is obliterated, but even then the metaflexus may persist as an indentation.In the less worn specimens (AH2051, Fig 2A -2C and AH 1874, Fig 2H -2J), the mesoflexus extends across the tooth slightly beyond half of it and flexes posteriorly at its internal termination.The hypoflexus is shorter and much deeper than the mesoflexus and flexes anteriorly.These teeth have three roots (two labial and a lingual one).

M3
The morphology of this tooth (AH2051) soundly recalls that of the M1 and M2 with its posterior side somewhat reduced and, therefore, with the hypocone smaller than the protocone (Fig 2A -2C).

Comparisons
The new species from Israel is compared below with all valid species of Prosayimys and "Sayimys" (see discussion) known to date [22].
The lower molars of Prosayimys flynni are easily distinguished from those of Sayimys negevensis sp.nov.by the presence of a distinct metalophulid II that is not fused with the metalophulid I. Furthermore, the lower molars of Sayimys sp.nov.have the mesoflexid shorter and deeper than the metaflexid.In contrast, those of Prosayimys flynni show a mesoflexid and a metaflexid nearly equal in length as well as in depth [16].The lower molars of Prosayimys flynni have a poorly developed posterolabial cingulum, whereas it is well-developed in those of Sayimys negevensis sp.nov.
The M1-3 of Prosayimys flynni and Sayimys negevensis sp.nov show also some differences.In the upper molars of Sayimys sp.nov from Israel, the paraflexus and metaflexus almost disappear in an early stage of wear, whereas in Prosayimys flynni both persist until after very advanced wear.This is probably due to the weaker hypsodonty of the teeth of Prosayimys, which have deep labial flexus and lingual flexids in relation to the tooth crown height.In Sayimys sp.nov., the labial flexus and lingual flexids are not so deep with regard to the crown height and, therefore, are more prone to wear.
Finally, the posterior side of the M3 of Sayimys negevensis sp.nov. is somewhat reduced, whereas that of the M3 of Prosayimys flynni is not.
Comparison with Sayimys obliquidens Bohlin, 1946 [32] The holotype (T.b. 268 b) of this species ( [32]:111, fig.30b, 30b', 30b''), a left lower jaw with p4-m3, comes from an horizon of the Tiejianggou Formation in the Tabenbuluk region (Gansu, China) that is possibly Middle Miocene in age [38].The m1-2s of Sayimys negevensis sp.nov.have a metalophulid II that fused with the metalophulid I early in wear, whereas those of Sayimys obliquidens have a distinct metalophulid II.Furthermore, the posterolophid is not constricted before reaching the triangular wear surface of the hypoconid in AH1938, whereas this constriction is observed on the m1s and m2s of Sayimys obliquidens.On the lower molars of Sayimys negevensis sp.nov., as in Sayimys obliquidens, the mesoflexid is much shorter than the metaflexid and the hypolophid is anteriorly directed.The comparison between the upper molars from the Rotem Basin and those belonging to Sayimys obliquidens are based on the single tooth row described and attributed by Bohlin ([32]: fig.30a‴) to this species.It is necessary to note here that, as proposed implicitly by Stehlin and Schaub ([39]: fig.182) and Schaub ([40]: fig.212) and explicitly by Wang ([27]: 63), all of the specimens from Tabenbuluk described by Bohlin [32] possibly belong to Sayimys obliquidens.On the M1-2s of Sayimys obliquidens, the metaflexus seems to be more affected than the paraflexus by the increase in wear.In fact, in the M1 of specimen 279a ( [32]: fig.30a‴), the metaflexus is nearly obliterated, whereas the paraflexus is still preserved.In contrast, as seen above, the paraflexus on the upper molars of Sayimys negevensis sp.nov. is more affected by wear than the metaflexus.The upper molars described by Bohlin [32] have a mesoflexus that is much narrower than in Sayimys negevensis sp.nov.Both Sayimys negevensis sp.nov.and Sayimys obliquidens have the M2 larger than the M1, but all the teeth of Sayimys obliquidens are well over the size range of those belonging to Sayimys negevensis sp.nov.(Fig 3).
Comparison with Sayimys baskini López-Antoñanzas and Sen, 2003 [33] The holotype of this species, a right P4 (GSP Y747/48125), has been recorded from the Early Miocene (MN4) locality Y747 of the Kamlial Formation (Potwar Plateau, Pakistan) [33].The morphology of the m1-2 of this taxon soundly recalls Sayimys negevensis sp.nov.However, the lower molars of this taxon are larger than those belonging to Sayimys baskini (Fig 3) and show a strong posterolabial ledge that is lacking in the latter species.The lower molars of the Rotem gundi and those of Sayimys baskini (Y747/48136) have a wide, open V-shaped mesoflexid shorter than the metaflexid that allows inferring the former presence of a metalophulid II (see discussion).The hypolophid of AH1938 is slightly oblique, does not oppose exactly the hypoflexid, and is partially lined up with the posterior arm of the protoconid.According to Baskin [16], the hypolophid of the m1 of GSP 48136 is more or less transverse, opposite the hypoflexid, and not aligned with the posterior arm of the protoconid.In the m2 of the same specimen, the hypolophid is more oblique and partially aligned with the posterior arm of the protoconid.As in Y747/48136, the hypoflexid is much larger and deeper than the mesoflexid in AH1938.The only DP4 of Sayimys negevensis sp.nov.available is badly worn.However, it is over the size range given by Baskin [16] for the DP4 of Sayimys baskini.The upper molars of Sayimys negevensis sp.nov.differ from those of Sayimys baskini in having a well-developed paraflexus that is longer than the metaflexus in the less worn specimens.Of the 19 M1 or M2s from the Kamlial Formation (localities Y721 and 747) described by Baskin [16], only one (GSP 36353) has a paraflexus that is longer and deeper than the metaflexus.The paraflexus is absent in the remaining teeth.Baskin [16] argued that the absence of a paraflexus is, at least in some specimens, real and not the result of wear because there are little worn specimens in which the anteroloph and/or protoloph taper(s) as it/they extend(s) labially.Furthermore, all the upper molars of the new taxon from Israel are well over the size range provided by Baskin [16,33] for Sayimys baskini (Fig 3).
The cheek teeth of Sayimys negevensis sp.nov.are smaller than those of Sayimys giganteus (Fig 3).Moderately worn teeth of the former taxon show a pattern of three lophs due to the fusion of the anteroloph and the protoloph at an early stage of wear.Even after moderate wear, the upper molars of Sayimys giganteus have four very distinct lophs.In addition, the M3 of Sayimys negevensis sp.nov. is more reduced posteriorly than those of Sayimys giganteus.
Comparison with Sayimys assarrarensis López-Antoñanzas and Sen, 2004 [17] The holotype of this species (AS21-1023), a fragmentary left maxilla with P4-M2, comes from the Lower Miocene locality of As-Sarrar, Saudi Arabia [17].In Sayimys assarrarensis, the paraflexus is present whatever the degree of wear of its upper molars (even when the metaflexus becomes obliterated).In contrast, the upper molars of Sayimys negevensis sp.nov.has the paraflexus more affected by wear than the metaflexus.In addition, the protocone and hypocone are connected by a straight endoloph, whereas it is oblique in Sayimys negevensis sp.nov.
The lower molars of Sayimys negevensis sp.nov.have a wide, open V-shaped mesoflexid that is shorter than the metaflexid and a well-developed posterolabial edge.In Sayimys intermedius, the mesoflexid and the metaflexid are equal in length and there is a weak or absent posterolabial ledge.With respect to the upper molars, those of Sayimys negevensis sp.nov.have the anteroloph and the protoloph fusing after moderate wear.In contrast, the upper molars of Sayimys intermedius show four distinct lophs even at a more advanced wear stage.In addition, the M3 of Sayimys negevensis sp.nov.are more reduced posteriorly than those belonging to Sayimys intermedius.
Comparison with Sayimys sivalensis (Hinton, 1933) [20] The holotype of Sayimys sivalensis (GSI D284) is a left dentary fragment with m2 and m3 from the Middle Miocene Chinji Formation, Pakistan [20].On the m1-2s from the Israeli species, the mesoflexid is much shorter than the metaflexid and does not reach the longitudinal axis.This condition can also be observed in GSI D284, as figured by Black [41].The specimens described as Sayimys sivalensis by De Bruijn et al. [42] and Baskin [16] also have a mesoflexid that is shorter than the metaflexid, but only slightly so and, in contrast with the condition in the Israeli m1-2s, it extends up to the longitudinal axis.According to Wessels et al. [43], the mesoflexid and the metaflexid of Sayimys sivalensis extend equally far labially in the m1, whereas the mesoflexid extends farther labially in the m2.In the specimens figured by Munthe ([44]: fig.8C-D), the mesoflexid and metaflexid are approximately equal in length.In the lower molars of Sayimys negevensis sp.nov., the hypolophid is less oblique than usually observed in Sayimys sivalensis.In AH2299 and AH1938, the anterior arm of the hypoconid is constricted at its connection with the posterior arm of the protoconid as in the specimens of Sayimys sivalensis.The m1-m2 of Sayimys negevensis sp.nov.have a well-developed posterolabial ledge.In Sayimys sivalensis, this ledge is usually distinct, but it can be weak or even missing.The upper molars of Sayimys sivalensis have a very short paraflexus.In contrast, the M1-M3 of the less worn specimen from Israel (AH2051) have a well-developed paraflexus.The dental wear pattern of both taxa is characterized by having the paraflexus more prone to obliteration with wear than the metaflexus.

Phylogenetic Analysis
A few attempts to decipher phylogenetic relationships in extant [45,46] and extinct [17,18] ctenodactylines had been carried out before López-Antoñanzas & Knoll [22] provided a comprehensive cladistics analysis of the subfamily.The subsequent discovery of a new genus of ctenodactyline from the Late Miocene of Lebanon [25] also contributed to improve our understanding of the evolutionary history of Ctenodactylinae.
The cladistic analysis including all valid species of Ctenodactylinae with less than 50% of missing data [25] has produced a single most parsimonious tree that confirms that the genus Sayimys is not monophyletic, as suggested by López-Antoñanzas and Knoll [22].Vianey-Liaud et al. ([24]: fig.15) in their interesting study of a new ctenodactylid from the Oligocene of China, Helanshania deserta, recovered a monophyletic Sayimyini.However, as their aim was not to resolve the relationships within the genus Sayimys but rather to determine the phylogenetic position of Helanshania, they did not include in their analysis all the species of Sayimys (e.g., Sayimys obliquidens appears in the matrix but not in the cladogram, Sayimys assarrarensis and Sayimys giganteus are lacking altogether).
The cladistic analysis including all valid species of Ctenodactylinae [25] resulted in a single most parsimonious tree (Fig 4).It shows that the Rotem gundi is more derived than the clade (Sayimys assarrarensis + Sayimys intermedius), but more primitive than Sayimys baskini.Sayimys negevensis sp.nov. shares some important synapomorphies (characters 31:1, 34:1) with the most derived taxa within Ctenodactylinae, which evidence some clear evolutionary trends inside this subfamily.
Primitive ctenodactylinae are characterized by having the paraflexus and the metaflexus on the upper molars well developed (29:0).The paraflexus is long in the most primitive ctenodactylines (Prosayimys flynni, Sardomys dawsonae, Pireddamys rayi, Sayimys giganteus, Sayimys intermedius, and Sayimys assarrarensis), whereas it obliterates very early with wear in more derived ctenodactylines (Sayimys baskini, Sayimys sivalensis, Proafricanomys libanensis, Africanomys cf.solignaci, Africanomys spp.) due to the fusion of the anteroloph and the paracone into a single loph.In fact, the most derived ctenodactylines (Irhoudia spp., Pellegrinia panormensis, and the living ctenodactylines) have completely lost it.Sayimys negevensis sp.nov. is the most derived ctenodactyline with a well-developed paraflexus, but this structure is lost after moderate wear.Therefore, this species can be seen as illustrating the transition within the evolution of Ctenodactylinae where the trend to lose the paraflexus is initiated.
The well-developed metaflexus (31:0) shown in primitive ctenodactylines disappears at earlier stages of wear in the course of ctenodactyline evolution until its complete disappearance in the crown group (Pectinator spekei + more derived ctenodactylines).Sayimys negevensis sp.nov. is the most plesiomorphic ctenodactyline that has clearly reduced the metaflexus on upper molars.Interestingly, this evolutionary trend is evidenced in the DP4 only higher in the tree (Africanomys major, Africanomys pulcher, Africanomys minor + more derived Ctenodactylinae).
The reduction of the posterior side of the M3 (34:1) characterizes all Ctenodactylinae more derived than the clade (Sayimys assarrarensis + Sayimys intermedius) and less derived than (Irhoudia spp.+ more derived ctenodactylines) except for Metasayimys curvidens.Once again, Sayimys negevensis sp.nov. is the most primitive species within ctenodactylines to show this reduction.On a side note, as mentioned above, the tree confirms that the genus Sayimys is not monophyletic.The taxa "Sayimys" obliquidens, "Sayimys" giganteus, "Sayimys" assarrarensis, and "Sayimys" intermedius are not closely related to Sayimys sivalensis, the senior synonym of the type species, Sayimys perplexus [22].In addition, Sayimys sivalensis has numerous apomorphies that are not shared by any of the above mentioned species (characters: 27:2, 28:2, 29:1, 31:1, and 34:1).Thus, "Sayimys" obliquidens, "Sayimys" giganteus, "Sayimys" assarrarensis, and "Sayimys" intermedius cannot be considered any longer as belonging to the genus Sayimys.Sayimys negevensis sp.nov.records the appearance, at least in inchoate form, of the synapomorphies that characterize the more derived ctenodactylines.Sayimys negevensis sp.nov. is very close to Sayimys baskini, which is, in turn, very close to Sayimys sivalensis.The unique combination of plesiomorphic and apomorphic characters of this taxon suggests that the erection of a new genus would be warranted.However, we refrain from doing so in view of the missing data due to the current lack of premolars.The same holds true for Sayimys baskini, which is located phylogenetically between Sayimys negevensis sp.nov.and Sayimys sivalensis.The acquisition of new data about these species of gundis may result in a shift of their phylogenetic position.

Conclusion
The Rotem ctenodactyline, which was assigned originally to "Metasayimys" cf.intermedius [9], can be distinguished from Sayimys intermedius and assigned to a new species, Sayimys negevensis.Compared to Sayimys intermedius, the lower molars of Sayimys negevensis have the mesoflexid shorter than the metaflexid and have a well-developed posterolabial edge.Furthermore, the anteroloph and the protoloph on the upper molars of the new taxon fuse at a moderate stage of wear and the M3 is reduced posteriorly, whereas in Sayimys intermedius the upper molars show four distinct lophs and the M3 is not reduced posteriorly.The taxon from Israel resembles Sayimys baskini from the Early Miocene of Pakistan.However, its less worn upper molars (Fig 2A and 2B) have a well-developed paraflexus, which is absent in Sayimys baskini.Thus, the morphological and dimensional features of the Rotem gundi show that it represent a new, endemic species.Sayimys negevensis nov.sp.places phylogenetically between (Sayimys intermedius, Sayimys assarrarensis) and Sayimys baskini.Sayimys negevensis sp.nov.adds a new facet to our currently poor knowledge of Middle Eastern Miocene rodents.This holds especially true for the Levant, which is an area of particular paleogeographical significance that, since Early Miocene time, has acted as a corridor of dispersal between Eurasia and Africa and where no other site of this age yielding rodents has so far been discovered.

Fig 3 .
Fig 3. Length/width scatter diagrams of the upper molars and first and second lower molars of the species included into Sayimys.Red circles indicates the size of Sayimys negevensis sp.nov.doi:10.1371/journal.pone.0151804.g003

Fig 4 .
Fig 4. Single most parsimonious tree generated by the cladistic analysis of all valid species of the Ctenodactylinae with less than 50% of missing data [25] (matrix in S3 File).The phylogenetic position of Sayimys negevensis sp.nov. is highlighted in red.doi:10.1371/journal.pone.0151804.g004