DISCUSSION

In response to changing herbivorous feeding habits, cricetid rodents developed an enormous variability of cheek teeth patterns during the Miocene. The molars of the primitive cricetid rodents and arvicolids considered here represent only limited insight into the immense diversity of molar patterns and offer no more than a small window into the dental history of this extremely successful group of rodents. Based on the comments on the considered taxa, the following conclusions can be drawn.

Cricetini with brachyodont-bunodont molars and a primitive dental pattern (tooth group A) appeared in the late Early Miocene (Late Orleanian, MN 4) of Europe, among them Democricetodon and Megacricetodon (Mein 2003). In Anatolia, Democricetodon appeared apparently earlier (Agenian -MN 1) and Megacricetodon during the Early Orleanian (MN 3) (Ünay et al. 2003). However, the validty of these occurrences remains to be confirmed. Both genera retained low-crowned molars and disappeared from the European fossil record during the Early Vallesian (MN 9) (Megacricetodon) and Late Vallesian (MN 10) (Democricetodon), respectively (Mein 2003; Sesé 2006). Kowalskia appeared in the Early Vallesian (MN 9) and survived until the Early Villanyian (MN 16), and is fairly specialised in its own direction but no tendencies toward hypsodonty are seen in its European representatives. Collimys (MN 7 – MN 11), the first European member of the Cricetini that increased the height of the tooth crowns, was followed by Cricetulodon (MN 9 – MN 10) and Rotundomys bressanus (MN 10) (tooth group B).

The oldest Eurasian microtoid cricetid is Microtocricetus from the Vallesian (MN 9 – MN 10) of Central Europe. It is characterised by mesodont molars that display an initial prismatic dental pattern (tooth group C).

Microtoid cricetids with mesodont-hypsodont prismatic molars that have opposing or alternating triangles (tooth group D) appeared first with Paramicrotoscoptes in the Early Hemphillian of North America considered to be equivalent of Early to Middle Turolian (MN 11, MN 12) of Europe. Eurasian records of Microtoscoptes have only been reported from the Late Turolian (MN 13) to Early Ruscinian (MN 14) so far.

Mesodont-hypsodont prismatic molars characterised by opposing triangles and trilophodont upper (M1) and lower molars (m1, m2) are traits of Trilophomys (tooth group E), a microtoid rodent, which was distributed in Europe during the Ruscinian (MN 14 – MN 16).

The first microtoid cricetid having mesodont molars and a true prismatic dental pattern with opposing and alternating triangles is Pannonicola (= Ischymomys), known from the Late Vallesian (MN 10) and the Turolian (MN 11 - MN 12) of Eurasia (tooth group F). Members of tooth group G are characterised by mesodont molars with alternating triangles. They were reported from the Late Turolian (MN 13) to Early Ruscinian (MN 14) (e.g., Anatolomys, Microtodon, Promimomys, Celadensia) of Eurasia, the Late Hemphillian of North America (Promimomys, or Prosomys of some authors) as well as from the Ruscinian (MN 14–MN 15) (e.g., Bjornkurtenia ) and Ruscinian (MN 14–MN 15) to Villanyian (MN 16) (e.g., Baranomys,) of Eurasia. The recent proposal that Microtodon and Anatolomys are synonyms of Baranomys (Fahlbusch and Moser 2004) appears to be unjustified, since the occlusal patterns and overall shape can be rather clearly distinguished (Figure 1, descriptions above).

Hypsodont prismatic molars with alternating triangles (tooth group H) are first known from Aratomys in the Early Ruscinian (MN 14) of Asia. Further early true arvicolids included in this tooth group are, for instance, Mimomys vandermeuleni (early Late Ruscinian, MN 15a) and Dolomys adroveri (Late Ruscinian, MN 15), Propliomys hungaricus (late Early Ruscinian (MN 14 b to Early Villanyian, MN 16) of Eurasia as well as Protopliophenacomys parkeri (Late Hemphillian), Ophiomys mcknighti (Blancan I), and Cosomys taylori (Middle Blancan) of North America; all of which are characterised by increasing height of crowns, enlargement of the anteroconid in m1, enlargement of the posterior portion of M3, and the development of dentine tracts.

Megacricetodon and Democrocricetodon display dental features indicating that these Miocene cricetids might have given rise to microtoid cricetids and extant cricetids as well. The origin of Megacricetodon and Democricetodon is far from clear, but Shamalina from the Early Miocene of Saudi Arabia (Whybrow et al. 1982) may have given rise to Megacricetodon (Lindsay 1994) and Spanocricetodon might be ancestral to Democricetodon (Mein 2003). Megacricetodon and Democrocricetodon appeared as Asian immigrants in Europe already in the late Early Miocene (late Orleanian, MN 4) for the first time and reached their highest point of evolutionary development and maximum abundance during the late Middle Miocene (Late Astaracian, MN 8) and early Late Miocene (Early Vallesian, MN 9). Kowalskia, a descendant of Democricetodon, made its first appearance in Europe during Late Miocene (Early Vallesian, MN 9) and survived until the Late Pliocene (Villanyian, MN 16).

During the Vallesian (MN 9 to 10) and possibly again at the Pliocene transition (MN 13 to 14) drastic ecological changes can be observed with the spread of more dry open habitats supporting grasslands, which is concluded from the increase of large and small mammal taxa with hypsodont teeth (Fortelius et al. 2002, 2006). At this time also the spread of prairies in North America with an increase of C4 plants has been recorded (Retallack 1997). This is also a period of many other changes in micromammals, like the MN 9 decline of glirids, the radiation of Murinae, the MN 10 appearance of Hystrix, Pliopetaurista, Apodemus etc., and the extinction of seven cricetine genera. Hypsodont forms were not only deveolped in cricetines, but also in murines (e.g., Microtia).

During the Late Miocene some cricetids with brachyodont-bunodont molars developed a more complex tooth crown pattern by the acquisition of more or less transverse ridges (lophs). Moreover, the molars became gradually high-crowned. A typical representative of this stage of cricetid evolution is Rotundomys bressanus known from the Late Vallesian (MN 10) in Western Europe. It seems that the basic lophodont tooth crown pattern of Rotundomys bressanus must have been quite similar to that of the forerunner of early arvicolid forms like Microtodon that is still to be discovered.

During the Late Vallesian (MN 10) Pannonicola (= Ischymomys) appeared in Eurasia for the first time and survived until the mid-Turolian (MN 12). The molar pattern of Pannonicola (= Ischymomys) is distinctly more derived than that of Rotundomys because its particular molar specialization resulted in the development of prismatic cheek teeth with opposing and alternating triangles. Pannonicola (= Ischymomys) belongs to the so-called Ischymomyini (Topachevskij and Nesin 1992), which were mostly considered an aberrant side branch of "micotoid cicetids" that became extinct during the Turolian without descendants. Therefore, closer relationship between Pannonicola (= Ischymomys) and true arvicolids met with disapproval (e.g., Nesin and Topachevskij 1991). However, this rejection is not justified because the relatively modern molar pattern of Pannonicola (= Ischymomys) (e.g., mesodont prismatic cheek teeth, the deep BRA 3 syncline, overall pattern of chewing surface) indicates closer relationships to Dolomys (Repenning et al. 1990) and possibly to Dicrostonyx. If so, Pannonicola (= Ischymomys) must be considered as one of the most important starting points of arvicolid evolution leading to the Ondatrinae (Dolomys, Propliomys) and probably to the Dicrostonychini (Predicrostonyx, Dicrostonyx). In various phylogenetic studies based on mitrochondrial and nuclear DNA (e.g., Galewski et al. 2006, Buzan et al. 2008; Robovský et al. 2008; Abramson et al. 2009; Triant and Dewoody 2009) ondatrins and lemmings are sister taxa to all other recent arvicolid genera that might confirm this pattern.

The Microtoscoptinae include two North American genera (Goniodontomys, Paramicrotoscoptes) and one Eurasian genus (Microtoscoptes) and represent an aberrant lineage of advanced microtoid cricetids known from the Late Turolian (MN 13) and Early Ruscinian (MN 14) of Eurasia and the Early Hemphillian of North America.

The molars look rather modern because of the prismatic structure, however, some buccal and lingual triangles and reentrants are still opposing and not strictly alternating, as in true arvicolids. In these chacteristics Paramicrotoscoptes from the Early Hemphillian of North America is more derived than Microtoscoptes from the Late Turolian (MN 13) and Early Ruscinian (MN 14) of Eurasia. Nesin and Topachevskij (1992) suggest closer relationships between Microtoscoptes and Pannonicola (= Ischymomys), however, this suggestion needs further study.

Because of this unique feature we consider the genera Microtoscoptes, Paramicrotoscoptes, Goniodontomys, and Pannonicola as representing primitive microtoid cricetid branches of the first independent hypsodont-prismatic dental adaptation; the first three in Eurasia and North America and the less hypsodont, more primitively rooted but more alternating-triangled and younger Pannonicola only in Eurasia.

Along with Microtocricetus, the three genera of the subfamily Microtoscoptinae belong to the earliest Late Miocene prismatic cricetid in the Northern Hemisphere; the records of Goniodontomys and Paramicrotoscoptes in the Early Hemphillian of Idaho, Nevada, Wyoming, Oregon, and Nebraska predate the Old World sites with Microtoscoptes. Consequently, the Mongolian, Siberian, and Russian occurrences of Microtoscoptes seem to represent North American immigrants that crossed the Bering Land Bridge. This immigration does not preclude the possibility that the unknown ancestor of Paramicrotoscoptes and Goniodontomys came to North America from Eurasia.

Microtoscoptinae are known earlier in North America than in other parts of the Northern Hemisphere and must have dispersed to Asia during the faunal exchange that took place in the Late Miocene (Repenning 1987). Neither Paramicrotoscoptes nor Goniodontomys are known from Late Hemphillian faunas of North America (Repenning 1987).

Baranomys, which ranged in Europe from the Early Ruscinian (MN 14) to Early Villanyian (MN 16) and Anatolomys from the Late Turolian (MN 13) and Early Ruscinian (MN 14) of Asia share some dental similarities with Eurasian Microtodon (MN 13, MN 14). However, in contrast to Microtodon and other arvicolid taxa, the Baranomyinae never developed additional prisms (triangles) in the anterior portion of the m1. Therefore, the Baranomyinae are to be considered as an aberrant side branch, like the Trilophomyini, which were widely distributed in Europe during the Ruscinian (MN 14, MN 15) and Early Villanyian (MN 16).

The most successful (in number of taxa and geographic distribution) evolutionary line of arvicolid evolution can be traced through Microtodon, Promimomys, and Mimomys that gave rise first to Microtus and later to Arvicola. The ancestor of Microtodon (MN 13, MN 14) is still unknown. Special attention needs to be given to Promimomys sp. from the Late Turolian (MN 13) of eastern Europe and western Asia (Zazhighin and Zykin 1984) because it displays a molar pattern apparently in between Microtodon and Promimomys. Mimomys evolved from Promimomys. Among the oldest and larger species are Mimomys antiquus from the late Early Ruscinian (MN 14b) of Siberia (Zazhigin 1980), Mimomys vandermeuleni, and Mimomys davakosi from the early Late Ruscinian (MN 15a) of the Iberian Peninsula and southern Europe, respectively, followed by Mimomys hassiacus known primarily from the Late Ruscinian (MN 15b) and Early Villanyian (MN 16) of central Europe.

The Ruscinian was the starting point of two important "directions" of Mimomys evolution in Europe (the taxa cited below are not in all cases in ascendant-descendant relation): (1) Mimomys with Mimomys vandermeuleni (early MN 15a), Mimomys davakosi (late MN 15a), Mimomys hassiacus (MN 15b/Mn 16a), Mimomys polonicus (MN 16b), Mimomys pliocaenicus (MN 17), Mimomys savini (Biharian), Arvicola mosbachensis (early Toringian), Arvicola terrestris (late Toringian); and (2) Mimomys with Mimomys gracilis (MN 15), Mimomys stehlini (MN 16), Mimomys reidi (MN 17), Mimomys pusillus (Biharian). Later on Pusillomimus, Pliomys, Clethrionomys, Borsodia, Villanyia etc. appeared. Further investigations are expected to solve the question whether these sequences represent true lineages or merely Stufenreihen with members that invaded successively from Asia.

It has been suggested (Kotlia and von Koenigswald 1992; Kotlia 1994) that Mimomys (Cseria) dispersed to southern Asia where it evolved to Kilarcola prior to 2.5 Ma. The immigration of Mimomys (Cseria) was correlated with the megafaunal turnover in southern Asia that took place about 2.5 Ma (Kotlia and von Koenigswald 1992). Kilarcola that existed in geographic isolation south of the Himalaya and gave rise to the following lineage (Kotlia 1994): Kilarcola indicus – Kilarcola indicus sahni – Kilarcola kashmiriensis. Note that Kilarcola kashmiriensis was referred to Mimomys (Aratomys) kashmiriensis by Repenning (2003, p. 484). Prior to these discoveries, Mimomys (Cseria) cf. gracilis and Mimomys sp. were reported from Hadji Rona (Sarobi Basin) in Afghanistan (Sen et al. 1979), likely of Ruscinian age.

The center of origin of Promimomys was located in the northern parts of Asia. According to Repenning (1987), a Late Hemphillian dispersal event introduced Promimomys, which was first described as Prosomys mimus (Shotwell 1963) to North America. The North American Promimomys mimus seems to be slightly more primitve than Promimomys insuliferus from the Early Ruscinian of Europe (Repenning et al. 1990). Also Nesin and Nadachowski (2006) consider evolutionary centers of early arvicolids in both Eurasia and N-America.

It has been suggested that Promimomys became extinct in North America, but gave rise to Mimomys in Asia (Repenning 1980, 1987). A dispersal framework established by Repenning (1980, 1987, 2003; see also Repenning et al. 1990) proposes that Mimomys entered North America via the Bering land bridge during Blancan I for the first time, followed by further immigration waves from the Eurasian continent. This model was challenged by von Koenigswald and Martin (1984) who argued, based on their studies of Schmelzmuster, that early Mimomys never immigrated to North America. According to these authors it is more likely that the North American Cosomys, Ophiomys, and Ogmodontomys (considered as subgenera of Mimomys by Repenning 1987) and the Eurasian Mimomys developed in parallel from the Promimomys basic stock (von Koenigswald and Martin 1984). In contrast to von Koenigswald and Martin (1984), Martin (2003b) has recently suggested that Mimomys occurred in North America but was limited to the early Pleistocene (early Irvingtonian) species Mimomys virginanus and Mimomys dakotaensis. We agree that Promimomys gave rise to Mimomys and its long history with extended diversification of several subgenera and many species in Eurasia. Nevertheless, we think that the fossil record in North America may suggest a more complicated iterative dispersal pattern of immigration and endemic evolution followed by diversification, and that the solution of the Mimomys question is a topic requiring more material and extensive further study.

The fossil record of microtoid cricetids and early arvicolids in Eurasia is remarkably complex. The discussed fossil record and the complete lack of hypsodont species of the American Copemys indicate that the place of origin and the primary evolutionary centre for arvicolids were situated in the northern parts of Asia. The present large arid/semiarid areas indicate extensive grass plains – the suitable habitat for both microtoid and arvicolid ancestors.