Karydomys is a rare and little diversified democricetodontine, of which only six species are currently recognized. This group of rodents is first recorded in the early Miocene (MN3) in China and spread quickly thereafter to Kazakhstan and Greece (MN4). Karydomys reached south‐western and central Europe by early middle Miocene times (MN5), from where it became extinct shortly thereafter (MN6). A new species of Karydomys is here described from the Miocene Keramia Formation of Chios Island (north‐eastern Aegean Sea). Karydomys strati sp. nov. is characterized by the presence of a labial spur on the anterolophule, the lack of anterior protolophule on the M1 and by the presence of a double metalophule on the M2. A cladistics analysis involving all the species of Karydomys and some closely related species of Democricetodon and Cricetodon show that Karydomys split early into two different lineages, an ‘eastern stock’, which includes the central Asian (K. debruijni and K. dzerzhinskii) and Greek (K. symeonidisi, K. boskosi and K. strati) species, and a ‘western lineage’ comprising the western and central European species (K. wigharti and K. zapfei). The evolutionary stage of Karydomys strati suggests that the lowermost part of the Keramia Formation, usually attributed to the middle Miocene, is older than previously thought.
Asia minor has been an important area for the exchange of faunas between Europe, Asia and Africa, especially during the early Miocene (Rögl 1999; Koufos et al. 2005). At that time, major dispersals from the east occurred through that area, transforming the mammalian composition of Europe (van der Made 1999; Agustí et al. 2001). The faunal changes were particularly important for rodents, since the entrance of ‘modern’ cricetids into Europe at the end of the early Miocene disrupted the structure of communities, which had until then been dominated by eomyids and glirids (Daams & van der Meulen 1984; van der Meulen & Daams 1992; Agustí et al. 2001). The appearance of cricetids in Europe was not the result of a single dispersal event but instead occurred in several successive waves, as shown by the timing of the arrival of the three main groups of cricetids: the Democricetodontinae, Megacricetodontinae and Cricetodontinae (de Bruijn et al. 1992; Theocharopoulos 2000; van den Hoek Ostende et al. 2015; Oliver & Peláez‐Campomanes 2016). In order to fully understand such faunal interchanges, it is important to improve our knowledge of the fossil record from key areas and develop phylogenetic hypotheses about the taxa involved in a robust chronological framework. For this reason, the fossil record of small mammals from the early and middle Miocene of the north‐eastern Aegean island of Chios, at the crossroad between Anatolia and south‐eastern Europe, is of critical importance for the establishment of a more precise biogeographical framework.
In the summer of 2010, two of us (RLA and FK) prospected in Chios Island. After sampling various levels in the Michalos’ pit and its surroundings, we located two layers with Miocene vertebrates (López‐Antoñanzas & Knoll 2011). The higher fossiliferous horizon is an arenaceous stratum that yielded only a cranial fragment of a large indeterminate mammal. However, lower in the Keramia Formation a thin argillaceous lens produced a diverse vertebrate fauna after screen‐washing and sorting, which motivated a return to the site in the summer of 2012 and the exploitation of the lens to near exhaustion. Altogether, over two tonnes of rock were processed, resulting in the recovery of nearly 200 teeth of small mammals. In fact, even though remains of large mammals (comprising a phalange of an indeterminate ruminant, a possible tooth fragment of Cainotherium and another of Sanitherium) and reptilian jaw and osteoderm fragments were retrieved from this new spot, rodents and, to a lesser extent, lagomorphs and insectivores are best represented by far. Preliminary identification of the lagomorphs suggests the presence of Albertona. The insectivore record consists of galericines. Rodents include a wide range of taxa belonging to cricetids, glirids, sciurids and ctenodactylids. Among them, cricetids are the most abundant, in particular species belonging to genera usually present in the eastern Mediterranean region during the early Miocene, such as Cricetodon, Megacricetodon and Democricetodon, but remains of the unusual democricetodontine genus Karydomys were also found.
Karydomys is a rare and little diversified democricetodontine, which shows a discontinuous record, geographically as well as chronologically. Its oldest record (Karydomys debruijni) comes from the early Miocene (MN3) of northern Xinjian (China) (Maridet et al. 2011). The genus may have rapidly expanded its range westward, as its presence is attested shortly later in the early Miocene (MN4) of eastern Kazakhstan (Karydomys dzerzhinskii) (Kordikova & de Bruijn, 2001) and Greece (Karydomys boskosi and Karydomys symeonidisi) (Theocharopoulos, 2000). Some undescribed findings of Karydomys from the early Miocene (MN4) of western Anatolia (Ünay & Göktaş 1999; Kaya et al. 2007) indicate that this area may well have been on a route between central Asia and Greece. Karydomys may have reached south‐western (Karydomys zapfei) (Mein & Freudenthal 1981) and central Europe (Karydomys wigharti) (Mörs & Kalthoff 2004; Prieto 2012, 2013) by middle Miocene times, from where it most probably became extinct at about 13.8 Ma (a single M2 from the fissure filling Petersbuch 6; Prieto 2012).
Because of its geographical position between Asia and Europe as well as its age, somewhat older than the first European occurrences known so far, the presence of Karydomys in the Keramia Formation (‘Keramaria’ in Besenecker 1973) of Chios is not a trivial record. The aim of the present work is to provide a description of the teeth of the new species of Karydomys discovered in the Keramia Formation of Chios Island and shed light on the evolutionary history of this group of rodents by deciphering their phylogenetic relationships with other Democricetodontinae and representative Cricetodontinae, touching upon the possible dispersal routes of this hamster.Geological setting
Chios Island is situated in the north‐eastern part of the Aegean Sea. The continental Cenozoic sediments of Chios are located mainly in the south‐eastern part of the island, unconformably overlying Mesozoic strata (Fig. 1). The Neogene deposits were subdivided by Besenecker (1973) into four lithostratigraphical units, which are from bottom to top: the Thymiana Formation (early Miocene), Zyfia Formation (early–middle Miocene), Keramia Formation (middle Miocene) and Nenita Formation (middle Miocene – Pliocene).Figure 1 Open in figure viewerPowerPoint Simplified geological map of Chios Island (redrawn from Kondopoulou et al. 1993) showing the location of the fossiliferous site, Thy 0. Colour online.
The Keramia Formation is about 120 m thick (Kondopoulou et al. 2011). It consists mainly of layers of green‐reddish clay and silt intercalated with less important layers of green sandstone (Besenecker 1973). The green and reddish colours are characteristic of this formation and make it easily recognizable in the field (Besenecker 1973). Besides, the sediments of the Keramia Formation are clearly delimited by a light ‘Tuff‐horizon’ below, and by the overlying white limestones of the Nenita Formation (Besenecker 1973).
Kondopoulou et al. (1993) conducted a magnetostratigraphical study in a 46 m‐thick section of the Keramia Formation in the Michalos’ pit area that included the three fossiliferous localities THA, THB and THC. The long reversal dominating the section was tentatively correlated with either chron C5Br or chron C5Cr based on the biochronological age then suggested by the mammals. Later, Kondopoulou et al. (2011) provided supplementary magnetostratigraphical data resulting from the sampling of the almost complete Keramia Formation in the Michalos section (about 120 m in thickness) also including approximately 10 m of the Zayfa Formation at the bottom. Thus, they presented information about the lower and uppermost parts of the section that was not considered in Kondopoulou et al. (1993). While the detailed magnetostratigraphical study is still pending publication, the preliminary results of Kondopoulou et al. (2011) seem to be in agreement with a correlation of the section with the interval C5Dr to C5Bn (see Hilgen et al. 2012, fig. 29.9).
All of the vertebrate specimens recovered so far from the Miocene layers of Chios (at least in modern history) come from the exposure of the Keramia Formation in or close to the brickyard quarry known as ‘Michalos’, which is situated SSE of the village of Thymiana, about 8 km from Chios city. The first find of vertebrate remains in Michalos’ pit was made in 1924 by the geologist Georgalas, but Paraskevaidis (1940, 1955; see also Paraskevaidis 1977) was the first to publish on the Miocene vertebrate fauna from Chios (macromammals, Sanitherium and others, but also evidence of tortoises). Additional, but more cursory, observations on similar vertebrate remains (macromammal, particularly of sanitheriid and bovid affinities, but also chelonian fragments) from the Neogene of the vicinity of Thymiana were published by Kreatsas (1963). Palaeontological fieldwork in the Michalos area conducted in 1967 and 1968 in the framework of a Germano‐Hellenic project significantly improved our understanding of the mammalian fauna from the Keramia Formation (see Melentis & Tobien 1967, 1968; Tobien 1968, 1969, 1977, 1980; Rothausen 1977; Lehmann & Tobien 1995). In total, nine fossiliferous layers were recognized by the end of 1960s (Thy 1–9) and described by Rothausen (1977). Most of the material collected then was fragmentary remains of bovids and giraffids, but a nearly complete skull of Gomphotherium angustidens was also discovered (and is still remembered locally!). Micromammals were also found, but were never described in detail and their present whereabouts is unknown. The same holds true for the lizards, tortoises and fish remains that were retrieved. Complementary studies on the vertebrate faunas from Chios arose from Helleno‐French field campaigns of 1991 and 1993 (Kondopoulou et al. 1993; Koufos et al. 1995; de Bonis et al. 1997a, b, 1998; López‐Antoñanzas et al. 2005). In addition to the nine fossiliferous levels described by Rothausen (1977), three new levels were found on this occasion: ‘Thymiana A’ (THA), ‘Thymiana B’ (THB) and ‘Thymiana C’ (THC) (Kondopoulou et al. 1993). Besides a few interesting macromammal remains (mostly of sanitheriid and bovid affinities, some of them well preserved) this fieldwork resulted in the accumulation of a rich micromammal collection, which is yet to be studied in its entirety.
Our main fossiliferous layer is situated on the coast, about 340 m south‐south‐west from the principal brick building. It does not correspond to any of the localities previously sampled. The stratigraphical position of the new layer suggests that it may be coeval with the most basal fossiliferous layers (Thy 1 and Thy 5), which are older than the others. We name it informally Thy 0.
The mammal fauna of the Keramia Formation is considered to belong to the biozone MN5 (Mein 1989; Steininger et al. 1989; de Bruijn et al. 1992; de Bonis et al. 1997a, b, 1998; de Bonis & Koufos 1999; van der Made 1999; Koufos 2006). This age is based both on the macromammals and the evolutionary stage and composition of the micromammal fauna, notably the cricetids. However, the macromammal remains from this area are fragmentary and the micromammals have never been properly described. In fact, only Tobien (1968) and de Bonis et al. (1997a) listed the micromammals that they recognized and only a single group of rodents, the ctenodactylines, has been the object of detailed study (López‐Antoñanzas et al. 2005). Kondopoulou et al. (1993) showed that the fossiliferous levels THA, THB and THC were included in a long reverse episode that, considering the MN5 age provided by the mammalian fauna, could be correlated with chron C5Br, which in turn provided an age of about 15.5 Ma (Koufos 2006). For instance, the micromammals from THA and THC were supposed to be younger than those of Aliveri (MN4) but quite similar to the Komotini (MN5) assemblage (de Bonis et al. 1997a; de Bonis & Koufos 1999). Considering that our fossiliferous level Thy 0 is at the base of the Keramia Formation, it may be correlated with chron C5Cr, which would correspond to an early Miocene age of 16.7–17.23 Ma.Material and method
Remains of micromammals were obtained by screen washing almost 2 tonnes of sediment with a mesh of 0.5 mm. First, second and third lower molars are designated as m1, m2 and m3, and first, second and third upper molars as M1, M2 and M3, respectively. The terminology used in the tooth descriptions follows those of Freudenthal et al. (1994) and López‐Antoñanzas et al. (2010). Measurements of the occlusal surface of the teeth (L, maximum length; W, maximum width) were obtained with a Nikon digital counter CM‐6S measuring device, following the method of van de Weerd (1976) (Table 1). The teeth of the species of Karydomys described in this work were compared with all the known species belonging to this genus. Scans and SEMs of the teeth of the specimens were taken with a Nikon XT H‐160 μ‐CT‐scanner and an FEI QUANTA 200 scanning electron microscope.Table 1. Measurements of the teeth of Karydomys strati sp. nov. from the Island of Chios (Greece) Length (mm) Width (mm) Karydomys strati Thy 0–39 m1 1.79 1.224 Karydomys strati Thy 0–2 m2 1.667 1.385 Karydomys strati Thy 0–5 m2 1.865 1.513 Karydomys strati Thy 0–145 m3 (broken) – – Karydomys strati Thy 0–9 M1 2.398 1.477 Karydomys strati Thy 0–30 M2 1.857 1.372 Karydomys strati Thy 0–48 M2 (eroded) 1.527 1.316 Karydomys strati Thy 0–54 M2 (broken) – –
For our phylogenetic analysis, all the species of Karydomys known to date were included in the ingroup. These are: K. debruijni, K. dzerzhinskii, K. symeonidisi, K. boskosi, K. wigharti, K. zapfei and K. strati. In addition, a selection of closely related non‐congeneric species were added to the ingroup to test the monophyly of the genus Karydomys and to clear up its phylogenetic relationships with species currently attributed to Democricetodon, from which Karydomys may have evolved, and Cricetodon. These are Democricetodon anatolicus, D. doukasi, D. gracilis, D. franconicus, D. mutilus, Cricetodon versteegi, C. tobieni, C. kasapligili, C. aliveriensis, C. trallesensis, C. orientalis, C. meini, C. engesseri, C. hungaricus, C. bolligeri. Eucricetodon wangae, a primitive cricetid rodent, was selected as outgroup. A total of 32 phylogenetically informative characters (mainly of dental morphology) have been coded (see López‐Antoñanzas et al. 2018). Of these, 19 characters are binary and 13 are multistate. The data matrix (López‐Antoñanzas et al. 2018) was built using Mesquite 3.04 (Maddison & Maddison 2009) and the analysis was run in TNT (Goloboff et al. 2008) with the ‘traditional search’ option (using TBR). Owing to the lack of a priori information, all characters were unordered and equally weighted (Fitch optimality criterion). As some species are known so far from only a few specimens, the influence of intraspecific variation in the scoring of the characters could not be assessed. Branch support was estimated through two complementary indices: Bremer Support (Bremer 1994) and Relative Bremer Support (Goloboff & Farris 2001).Systematic palaeontology
Karydomys symeonidisi Theocharopoulos, 2000.
From stratus, Latinized form of stratos (Greek: Στράτος), genitive. In honour of Stratos Asmanis in recognition of his warm‐hearted hospitality during our work in the Michalos pit.
UM Thy 0–30, a left M2. This and the paratype specimens are housed in the palaeontological collections of the University of Montpellier (Montpellier, France).
UM Thy 0–9, left M1; UM Thy 0–39, right m1; UM Thy 0–2, right m2; UM Thy 0–5, right m2; UM Thy 0–145, right m3 (broken); UM Thy 0–48 left M2; UM Thy 0–54, left M2 (broken).
Thymiana 0. Lowermost part of the Keramia Formation, SSE of the village of Thymiana, about 8 km from Chios city (Greece).
Late early Miocene (MN4).
Species of Karydomys characterized by having a backward paracone spur and long and narrow mesolophs on the M1 and M2, having labial spur of the anterolophule and lacking anterior protolophule and metalophule on the M1 but having double protolophules and metalophules on the M2; lower molars with ectomesolophid and long and narrow mesolophids.
Karydomys strati differs from K. debruijni in lacking the posterior metalophulid on the m1, in having a well‐developed ectomesolophid on some m2, a strong posterior spur of the paracone on the M1 and M2 and a double metalophule on the M2 (whereas it is either anteriorly connected or absent in K. debruijni). It differs from K. dzerzhinskii in lacking the anterior protolophule on the M1, in having the anterior and posterior protolophule unequally developed and in showing a double metalophule on the M2 (in K. dzerzhinskii the single metalophule is usually anteriorly located). It is distinguished from K. symeonidisi in lacking the anterior protolophule on the M1, which is present in most of the M1 of K. symeonidisi, and in having a double metalophule on the M2. It is unlike K. boskosi in being larger, in lacking a posterior ridge of the metaconid and in having a strong metalophulid and longer and narrower mesolophid on the m1. It differs from K. wigharti in being noticeably smaller (Fig. 4), in lacking the anterior protolophule on the M1, and in having a backward paracone spur on the M1 and M2 and a double protolophule on the M2. It is different from K. zapfei in being significantly smaller (Fig. 4), in having ectomesolophid on the m1, a long mesoloph on the M1, a distinct backward paracone spur on the M1 and M2, and a markedly bent double metalophule on the M2.Figure 4 Open in figure viewerPowerPoint Length/width scatter diagrams of the first two upper and lower molars of the different species belonging to the genus Karydomys. Colour online.
The m1 has a small and single oval anteroconid. Both labial and lingual anterolophids are distinct. The labial anterolophid is of medium length and it does not reach the base of the protoconid, whereas the lingual one is short and encloses a rounded anterosinusid. The anterior metalophulid that connects the metaconid with the anteroconid is well developed, whereas the posterior one is absent. The long and narrow mesolophid reaches the lingual side of the tooth, where it forms a mesostylid. The ectomesolophid is long and narrow. The entoconid is located anteriorly to the hypoconid and the hypolophid is close to the mesolophid. The posterolophid is long and reaches the posterior side of the entoconid lingually, closing the posterosinusid. The roots of this tooth are not preserved.
Two lower second molars, of which one is very well preserved and the other is damaged, have been found. The occlusal outline of these teeth is rectangular. Both show well‐developed labial and lingual anterolophids. The labial anterolophid is longer than the lingual one. The metalophulid connects to the anteroconid. The anterolophulid is very short. The mesolophid is nearly transverse, either long and narrow (UM Thy 0–5) or short (UM Thy 0–2). It can reach the lingual border of the tooth, where it finishes in a mesostylid (UM Thy 0–5). The metaconid and the mesolophid are connected by a longitudinal, thin crest. The ectomesolophid is long (UM Thy 0–5) or absent (in the damaged specimen UM Thy 0–2). The entoconid is anteriorly located with respect to the hypoconid. The strong hypolophid, very close to the mesolophid, bends anteriorly. The long posterolophid joins with the posterolingual side of the entoconid. It encloses a deep and large posterosinusid. The roots of these teeth are not preserved.
The most anterior part of the single m3 is missing. Its occlusal outline is sub‐triangular, its posterior part being reduced and rounded. The metalophulid is connected to the anterior arm of the protoconid and the mesolophid is absent. The well‐developed hypolophid is slightly oblique and encloses the much reduced entoconid. The hypoconid is large although somewhat smaller than the protoconid. The posterolophid is long and joins with the posterolabial side of the entoconid, enclosing a large and circular posterosinusid. The roots of this tooth are not preserved.
The anterocone of the single first upper molar found is slightly divided. The lingual anteroloph is a very thin crest that runs to meet the anterior side of the protocone, enclosing the protosinus. The strong anterolophule connects the protocone to the middle part of the anterocone. This tooth shows a distinct labial spur on the anterolophule. The anterosinus is labially closed by a low ridge, which runs from the labial side of the anterocone to the anterior part of the paracone. This tooth has no anterior protolophule (protolophule I) and the posterior protolophule (protolophule II) connects to the entoloph well‐posterior to the posterior of the protocone. The protocone and the entoloph are firmly connected. The tooth shows a short but very distinct backward paracone spur. The mesoloph is long and narrow, reaching the labial border of the tooth. The sinus is closed by a ridge that connects the anterolingual side of the hypocone to the base of the protocone. The metalophule is short and joins with the posteroloph. This latter connects to the metacone, enclosing a small posterosinus. The roots of this tooth are not preserved.
Three M2 teeth, two of them complete (UM Thy 0–30 and UM Thy 0–48) and the other one damaged (UM Thy 0–54), have been found. The teeth show well‐developed labial and lingual anterolophs. The labial anteroloph is connected to the base of the paracone, enclosing a narrow anterosinus, whereas the lingual one reaches the base of the protocone enclosing the protosinus. The protolophule is double, but its degree of development varies from one specimen to another. Specimen UM Thy 0–30 shows the anterior protolophule slightly interrupted and weaker than the posterior one. UM Thy 0–48 has the anterior protolophule strong, whereas the posterior one is weaker and interrupted. UM Thy 0–54 has both anterior and posterior protolophules complete. All specimens have a long and narrow mesoloph that ends with a mesostyle at the labial border of the teeth. The backward paracone spur is short but strong and reaches the mesostyle in two out of the four specimens available. The teeth have a double metalophule with the posterior one stronger than the anterior one. Both the anterior and the posterior metalophules are nearly longitudinal and they can even form an additional longitudinal crest on their own (specimen UM Thy 0–30). The posteroloph is short and connects to the posterior metalophule. It nearly reaches the metacone, enclosing a small posterosinus. The roots of these teeth are not preserved.Phylogenetic results
The cladistic analysis including all species of Karydomys as well as relevant species currently placed in the genera Democricetodon and Cricetodon yielded a single most parsimonious tree with a length of 86 and a relatively high degree of homoplasy (CI = 0.512; RI = 0.806). Our results (Fig. 5) provide evidence of two groups separated at the most basal node. The first one corresponds to all the species of Cricetodon included in this analysis, whereas the second encompasses the various species of Democricetodon taken into account in this work and all known species of Karydomys. However, while Karydomys forms a clade within the latter lineage, Democricetodon constitutes an array of successive sister‐species to Karydomys. Furthermore, two lineages diverge at the basal node of Karydomys: one constituted by the western Eurasian species (K. wigharti + K. zapfei) and the other by the eastern Eurasian ones (K. debruijni + more derived species).Figure 5 Open in figure viewerPowerPoint Cladogram showing the relationships among all known species of Karydomys as well as selected Democricetodon and Cricetodon and the phylogenetic position of Karydomys strati sp. nov. (for data matrix see López‐Antoñanzas et al. 2018). Nodes show Bremer and Relative Bremer indices. Discussion
The examination of the new rodent remains from Chios has revealed eight molars (Figs 2, 3) belonging to a medium‐sized Karydomys (Fig. 4) but different from the molars of any species of this genus known so far. Karydomys strati is morphologically close to the two known species from Karydia. However, it shows some important differences, particularly concerning the protolophule and the metalophule on the M1 and M2 and the degree of development of the mesolophids, which prevent it being considered as belonging to either of them. The presence of a double metalophule on the M2 (Fig. 6) may suggest that the new species from Chios could be, in fact, slightly older than K. symeonidisi and K. boskosi, in which the anterior metalophule is lost not only on the M1 but also on the M2. The absence, on the M1 of K. strati, of the anterior protolophule, which is present in K. symeonidisi, and the presence of a spur on the anterolophule on the M1, a primitive character absent in K. boskosi, is consistent with this hypothesis. The presence of a posterior metalophule on the M2 of K. strati suggests that it is more evolved than K. dzerzhinskii and K. debruijni (Fig. 6), which are the oldest species of Karydomys. In addition, Karydomys from Chios is distinct from middle Miocene species such as K. wigharti and K. zapfei in being noticeably smaller, in lacking the anterior protolophule on the M1 (present in K. wigharti), in having a well‐developed posterior paracone spur on the M1 and M2 as well as in the presence of the ectomesolophid on the m1 (Fig. 6).Figure 6 Open in figure viewerPowerPoint Evolutionary stage of various species of Karydomys with some important structures signaled. A–C, Karydomys dzerzhinskii (taken from Kordikova & de Bruijn 2001, with permission). D–F, Karydomys strati sp. nov. G–I, Karydomys wigharti (taken from Mörs & Kalthoff, 2004, with permission). Morphological comparison only; not to scale.
The genus Karydomys was erected by Theocharopoulos (2000) on the basis of characters such as the presence of a small posterosinus and well‐developed ectolophs on the first two upper molars (in some of the taxa), the reduction of the size of the third molars and the presence of a small, blade‐like anteroconid situated very close to the protoconid and the metaconid on the first lower molar. However, as pointed out by Maridet et al. (2011), some of these characters, such as the presence of a small posterosinus, resulting from a posterior position of the metalophule, as well as the development of the ectolophs (posterior paracone spur in the present work), are not present in all the species of the genus. Therefore, Maridet et al. (2011) suggested that the diagnosis of the genus would have to be revised when more material became available. Even though we concur with Maridet et al. (2011) that there are a number of issues related to the diagnosis of the genus Karydomys, we are inclined to maintain this taxon in its current composition as the species of Karydomys form a monophyletic grouping in our analysis. Indeed, the clade Karydomys shares one exclusive synapomorphy: a slightly divided or ‘crest‐like’ anterocone on the M1 (3 (0→1)). In fact, the M1 of Karydomys are characterized by a wider and flater anterocone more independent from the anteroloph than in Democricetodon, in which it is a well‐defined cusp that may fuse with the anteroloph with wear. The presence of a short lingual anterolophid (23 (0→1)) on the m1 also supports the clade, but this occurs in some species of Democricetodon as well (e.g. D. anatolicus of our ingroup). A very interesting synapomorphy is the connection of the anterior metalophulid on the m1. This crest joins a distinct anterolophulid in Democricetodon (24(0→1)), but directly connects to the anteroconid or to a tiny anterolophulid in Karydomys (24 (1→2)). This latter feature is shared by the members of the genus Cricetodon.
According to our results, Karydomys split early into two discrete lineages. One constituted by the central Asian (K. debruijni and K. dzerzhinskii) and Greek (K. symeonidisi, K. boskosi and K. strati) species, which we term here the ‘eastern lineage’, and the other formed by the species from Germany and France (K. wigharti and K. zapfei), the ‘western lineage’. Maridet et al. (2011) pointed out morphological differences between the species of Karydomys, those from western and central Europe (France and Germany) and south‐eastern Europe (Greece) on the one hand and those from central Asia (Kazakhstan and north‐western China) on the other. The latter were considered to be the least derived, particularly on the basis of the morphology of the M2. Our results place the two central Asian species of Karydomys (K. debruijni and K. dzerzhinskii) in a basal position with respect to the Greek ones, but on a separate lineage from the western European species K. wigharti and K. zapfei. The cladogram shows sister group relationships between K. dzerzhinskii from Kazaksthan and the clade composed by the three Greek species (K. symeonidisi (K. strati + K. boskosi)). These results are consistent with the suggestion of Kordikova & de Bruijn (2001) about a close relationship between K. dzerzhinskii and K. symeonidisi, with the former being a plausible ancestor of the latter. Our results also identify K. wigharti as sister species of K. zapfei. Mörs & Kalthoff (2004) emphasized the morphological affinities between them. Some synapomorphies of the species of Karydomys for central Asia and south‐eastern Europe differentiate them from the western and central European taxa. In particular, the presence of a short but distinct backward paracone spur on the M1 (8 (0→1), a derived character state, and the presence of a well‐developed ectomesolophid on the m1 (27 (0→1)), another derived character state, are features that are usually lacking on the two western and central European species K. wigharti and K. zapfei. However, these are not clear‐cut discriminators as several M1 of K. debruijini show a backward paracone spur that varies from more or less short to being absent altogether, and the ectomesolophid is equally variable in its degree of development on the m1 in K. wigharti. Besides morphology, the western and central European species are also larger than the other species of Karydomys and, indeed, Democricetodon. The grouping of the Greek species of Karydomys is supported by stout cusps (2 (0→1)), which is a parallelism with the western and central European species. The Greek species also show a well‐developed backward paracone spur on the M2 (15 ((0→1)), which is a character state incipient in K. dzerzhinskii.
The diagnostic characters provided by Theocharopoulos (2000) to differentiate Karydomys from Democricetodon are not valid for all species. The presence of a small anteroconid, which was used to discriminate the two genera, is a character also seen in various species of Democricetodon. Furthermore, the alleged presence of better developed ectolophs in Karydomys than in Democricetodon is actually lacking in the two western and central European members of Karydomys (K. wigharti and K. zapfei) but present in some species of Democricetodon (e.g. D. anatolicus, whose ectolophs are well developed). As mentioned above, only one exclusive synapomophy supports the clade Karydomys, namely a slightly divided or ‘crest‐like’ anterocone on the M1. We have reservations about the adequacy of such a unique character to justify the legitimacy of the genus Karydomys. However, the differences between Democricetodon and Karydomys regarding the nature of the connection of the anterior metalophulid should also be taken into account. The species currently attributed to Karydomys may as well be seen as pertaining to Democricetodon, as derived members of the genus. At this point, however, we refrain for formally synonymizing Karydomys with Democricetodon given the incomplete nature of our ingroup. Should a future, more comprehensive, phylogenetic analysis of the species of Democricetodon and Karydomys yield a pattern of results in line with our own (i.e. Democricetodon forming a paraphyletic array of species relative to a Karydomys clade) then the merits of the recognition of the genus Karydomys will have to be seriously questioned. In any case, the close relationships between the genera Karydomys and Democricetodon suspected by Theocharopoulos (2000) are confirmed.
The teeth of the taxa belonging to Cricetodon are morphologically quite different to those of Democricetodon and Karydomys. Our cladogram clearly supports the monophyly of Cricetodon. The species of Cricetodon included in the ingroup share some exclusive synapomorphies. They have the anterocone strongly divided (3 (0→1)) and a short mesoloph (9 (0→1)) on the M1. Their m1 are characterized by having a large anteroconid (22 (0→1)) and by usually lacking the lingual anterolophid (23 (0→1)). With respect to the m2, the less derived members of the group have a weak lingual anterolophid (28 (0→1)), which completely disappear in the more evolved representatives of the clade (28 (1→2)). Their m3 show a more or less circular protosinusid (31 (0→1)) instead of having it retracted and directed forward as is the case in Democricetodon and Karydomys. However, to confirm or refute the monophyly of this clade, a more comprehensive cladistic analysis that includes all the species belonging to the subfamily Cricetodontinae is necessary.Conclusions
The new species Karydomys strati has been erected in this work to accommodate remains of Karydomys recently recovered from the Miocene of Chios. According to our phylogenetic analysis, the genus Karydomys split at an early stage into two different lineages, an ‘eastern stock’ that includes the central Asian and Greek species and a ‘western lineage’, which is constituted by the western and central European species of the genus. K. strati belongs to the eastern lineage. This relatively large‐sized democricetodontine rodent is characterized by a labial spur on the anterolophule and the lack of the anterior protolophule on the M1 as well as by the presence of a double metalophule on the M2. This latter character suggests that K. strati is more evolved than K. dzerzhinskii and K. debruijni, which are both devoid of the posterior metalophule, but less evolved than the Greek species K. symeonidisi and K. boskosi, which both lack the anterior metalophule.
The presence of Karydomys is of interest from a biostratigraphical point of view. The Keramia Formation has usually been assigned a middle Miocene age. However, the evolutionary stage of the new species of Karydomys that it yields suggests that its stratigraphical level could in fact be slightly older than those from which K. symeonidisi and K. boskosi were collected (MN4). The lowermost part of the Keramia Formation may, therefore, be late early Miocene in age (MN4), which would be in accordance with the latest published magnetostratigraphical data of the area.Acknowledgements
We sincerely thank I. Antoñanzas‐Asso, E. López‐Antoñanzas and J. A. Molina‐Anadón for their help during the field work and/or the sorting of sediment. Permission to conduct field research in the Island of Chios carried out by RL‐A and FK was granted by the Greek Institute of Geology & Mineral Exploration (Acharnae). Permission to prospect the private Michalos brickyard quarry was ensured by Stratos Asmanis, who made our sojourns in Chios most enjoyable. T. Mörs (Swedish Museum of Natural History, Stockholm, Sweden), S. Thomas (The Palaeontological Association, UK) and an anonymous reviewer enhanced this work through careful, critical reading. M. Furió, A. García, C. Paradela, and L. Tormo (Museo Nacional de Ciencias Naturales‐CSIC, Madrid) were kind enough to take the μ‐CT scan images and SEMs.
This work is dedicated to the memory of my beloved father, Manuel López Gálvez, who will live forever in our hearts.