Introduction

The genus Arvicola Lacépède, 1799 is represented in the extant European fauna by two species: Arvicola terrestris (Linnaeus, 1758), the northern water vole; and Arvicola sapidus Miller, 1908, the southern water vole, which is larger and which is limited to the Iberian Peninsula and southern France (Díaz de la Guardia and Pretel 1978; Rekovets 1990; Abbassi and Desclaux 1996; Abbassi et al. 1998; Desclaux et al. 2000; Maul et al. 2000; Cubo et al. 2006; Román Sancho 2007; Centeno-Cuadros et al. 2009a, b). The morphological characters of the limbs of water voles are linked with their fossorial/aquatic adaptations; the aquatic or semiaquatic character probably represent the ancestral condition, as seen in A. sapidus (Cubo et al. 2006; Roman Sancho 2007). The related species, Mimomys savini Hinton, 1910, was probably a semiaquatic rodent that spread from Eastern to Western and Southern Europe along the wet plains of coast and fluvial environments (Cuenca-Bescós et al. 2001: fig. 2).

During the Pleistocene several species of the genus Arvicola have been recognized: Arvicola mosbachensis (Schmidtgen, 1911); Arvicola cantianus (Hinton, 1910); Arvicola bactonensis Hinton, 1926; Arvicola grenii Hinton, 1926; Arvicola praeceptor Hinton, 1926; Arvicola sapidus aupsensis Râdulescu, 1989; Arvicola chosaricus Alexandrova, 1976; Arvicola kalmankensis Zazhigin, 1980. These taxa were discussed in a number of papers, so we shall not touch upon them here (see Koby 1965; Râdulescu 1989; Rekovets 1990; Gromov and Polyakov 1992; Abbassi et al. 1998; Maul et al. 2000; Kalthoff et al. 2007; Markova 2007).

The large voles of the species Mimomys savini and its putative descendant, Arvicola, provide evidence for European biostratigraphy (Heinrich 1978; Kolfschoten 1993), not always coincident with the chronostratigraphic limits. More precisely, the Mimomys—Arvicola transition has been proposed as the biostratigraphic Early—Middle Pleistocene limit in Europe. The last appearance datum (LAD) of Mimomys and first appearance datum (FAD) of Arvicola is also used as the marker of the Cromer II—Cromer III boundary in England, Thuringia and the Rhine basin (Koenigswald and Kolfschoten 1996), dated around 600 Ka, while the chronostratigraphic limit of the Early—Middle Pleistocene is fixed in the Matuyama-Brunhes paleomagnetic reversal, at 780 Ka (Koci and Sibrava 1976; Kuiper et al. 2008).

Institutional abbreviations.—MPZ, Museo Paleontológico de Zaragoza, Zaragoza, Spain.

Other abbreviations.—a, anteroconid complex length; ACC, anteroconid complex of m1; acd, anteroconid; AL, anterior loop; ATA, Atapuerca; BRA, buccal re-entrant angle; LRA, lingual re-entrant angle; m1, m2, m3, the lower molars; M1, M2, M3, the upper molars; p, posterior lobe length; PL, posterior loop; ROI, region of interest; SDQ, Schmelzband Differenzierung s Quotient (enamel thickness differentiation quotient); T1…T5, the dentine fields or triangles. The re-entrant angles and the triangles are numbered consecutively from PL towards ACC of the m1; TE, Sima del Elefante site; TE-LRU, Lower Red Unit of TE; Z, depth.

Evolutionary characters for the study of fossil arvicolines.—In arvicoline molars, characters such as the cementum in the re-entrant angles, the enamel-free areas in occlusal view (which form the occlusal intersection with the dentine tracts in lateral or medial views), and the enamel differentiation between the trailing and leading edges, as well as the arrangement of this differentiation, are unique among rodent molars. These characters are linked to hypsodonty, or the increase in height of the molar crowns, that resulted from adaptation to an herbivorous, distinctly abrasive diet entailing substantial dental wear. As an adaptation to abrasive grasses the hypsodont molars became hypselodont in some arvicoline lineages, acquiring the character of continuous growth. The biological advantage of this molar type and its continuous growth is the ability to process very abrasive food, because the tooth tissue lost through wear is continuously replaced (Kesner 1980; Koenigswald et al. 1994). The loss of roots in the molars of some arvicoline lineages during the Pleistocene can be described in terms of a heterochronic process of paedomorphosis, or neoteny (Agustí et al. 1993).

Continuous growth generally gives rise to crowns in the form of high columns or dentine prisms, with an osteodentine core and bordered by enamel (Hinton 1926). On the occlusal surface, the prisms form a certain number of folds that produce reentrant and salient angles of enamel on the buccal and lingual sides. These dentine fields or prisms bordered by enamel form a pattern of alternating polygons that may either be closed or confluent and are termed triangles by authors (Meulen 1973; Rabeder 1981; Martin 1987; Repenning 1992). The dentine fields are varyingly confluent with one another, and with the anteroconid complex (ACC). The confluence pattern is one of the most useful diagnostic characters in arvicolines (Meulen 1973; Rabeder 1981, Repenning 1992). Closed or confluent fields in combination with the ACC shape in the wearing surface of cheek teeth make identification possible even when fossil remains are limited or incomplete.

Enamel differentiation and Schmelzband Differenzierungs Quotient (SDQ).—There are arvicolines with almost uniformly thick enamel bordering the triangles (undifferentiated type), and with a differentiation in enamel thickness between the leading and trailing edges (Koenigswald et al. 1994). Thickness differentiation is observed in the clade Mimomys Forsyth-Major, 1902, where rooted molars have a thick trailing edge. This character evolves into a thin trailing edge in the extant lineages derived from Mimomys: the genus Microtus Schrank, 1798 and the species Arvicola terrestris (Linnaeus, 1758). The Microtus-type represents the apomorphic character state, while in the plesiomorphic stage the enamel band is thicker in the trailing edge: this is the Mimomys-type (Heinrich 1978; 1987; Koenigswald et al. 1994; Koenigswald and Kolf-schoten 1996). In spite of that, in earlier representatives of the Microtus-type, in Allophaiomys Kormos, 1932, the enamel thickness is variable; it may be of the Microtus-type (Markova 2005), undifferentiated (Alcalde et al. 1981) or of the Mimomys-type (Meulen 1974). The enamel is undifferentiated or slightly Mimomys-type in the first representatives of the genus Allophaiomys (see Meulen 1974, Garapich and Nadachowski 1996; Tesakov 1998b). The examination of enamel differentiation is one of the parameters measured in the evolution and ontogeny of arvicolines (Heinrich 1978; Röttger 1987; Ventura and López Fuster 1989; Kalthoff et al. 2007).

Morphology of the molars.—Species of the genus Mimomys have a lower first molar (m1) with a simple and primitive occlusal morphology (Hinton 1926): from distal to mesial, the occlusal surface of the m1 comprises the posterior lobe, three distal triangles, and an anterior complex formed by two open triangles convergent with one another and with the anterior lobe or anteroconid complex (Meulen 1973; Repenning 1992). The three distal triangles are completely closed in the species of Arvicola, but in the species of Mimomys this character is more variable. The anteroconid complex of m1 also displays unique structures such as the Mimomys-kante, or ridge, and the enamel islet. The third upper molar (M3) is simple, with a single, closed triangle and a posterior complex with two salient angles, one buccal and one lingual. The posterior lobe varies from being rounded to triangular and is broadly connected to the two salient angles. In some species M3 has a simpler morphology, in which the triangles are broadly confluent both with one another (there is no closed triangle) and with the anterior and posterior lobes (Garapich and Nadachowski 1996). The more primitive species exhibit enamel islets in m1 and in M3 (Rabeder 1981).

Morphology of the mandible.—The mandible, which is scarcely represented in the fossil record, plays an important part in the systematics of arvicoline rodents (Hinton 1926; Repenning 1968; Kesner 1980). The diagnostic characters are: the ascending ramus and the double crest for the insertion of the masseter, which is divided into the upper or dorsal crest and the lower or ventral crest, the masseteric crests (upper and lower) originating at m1, the arvicoline groove, the deep and long internal temporal fossa separating the second and third lower molars (m2, m3) from the ascending ramus, the symphyseal eminence for the insertion of the digastric muscle. The eminence of the alveolus of the lower incisor is large and is visible on the lateral side of the mandible (capsule for base of incisor in Repenning 1968); in medial view it reaches up to the mandibular foramen in the ascending ramus, which has a high position. In this paper we also note the strong genial tubercle on the medial side of the water vole mandible. This is considerably larger in Arvicola than in Terricola or Microtus.

Fig. 1.

A. General situation of the Atapuerca localities in the Iberian Peninsula. B. A map depitcing the main karstic system of the Atapuerca Hill, and the main two sets of localities: the Trinchera del Ferrocarril or La Trinchera localities, where the Sima del Elefante belongs; and the Cueva Mayor-Sima de los Huesos system. Note that Sima del Elefante could be an ancient opening to the Cueva Mayor-Galería Baja karstic tunnel.

Origin of hypselodont taxa derived from Mimomys.—The genus Mimomys was widely distributed across the Northern Hemisphere during the Pliocene (Chaline et al. 1999, Repenning 2003), but at the beginning of the Pleistocene its specific diversity diminished considerably. This was accompanied by an important evolutionary change: various lines became more hypsodont, eventually leading to hypselodont genera such as Allophaiomys (Kormos 1932) and Arvicola (Agustí 1991; Agustí et al. 1993; Maul et al. 1998; Maul 2001). The first genus with hypselodont molars, presumed to be a descendant from a species of Mimomys, was Allophaiomys, small in size and with enamel that was undifferentiated or of the Microtus-type, which first appeared at the beginning of the Early Pleistocene, circa 1.7 Ma (Repenning 1992; Garapich and Nadachowski 1996). Less well-known is the appearance of hypselodont forms such as Mimomys oswaldoreigi Agustí, Castillo, and Galobart, 1993 in Spain (Agustí et al. 1993), likewise at the beginning of the Early Pleistocene. Much later on, at the beginning of the Middle Pleistocene (around 600 Ka), other groups of medium-sized to large arvicolines that lose their roots are Arvicola mosbachensis, Arvicola cantianus, and Arvicola terrestris in central Europe and the Pannonian Basin (Heinrich 1987; Kolfschoten 1993; Maul et al. 2000).

Origin of Arvicola derived from Mimomys.—Arvicola is alleged to be a descendant of Mimomys because of the great similarity in the morphology of the occlusal surface of the molars (Koenigswald and Kolfschoten 1996), the loosening of roots and the complexity of the third upper molar (M3) being the evolutionary differences. The extant species of Arvicola share the rootless molars, and the M3 more complex, although they differ in their enamel differentiation: it is of the Mimomys-type in Arvicola sapidus, while of the Microtus-type in Arvicola terrestris. The SDQ values of the Mimomys-type were also present in the species Arvicola cantianus and Arvicola mosbachensis of the Middle Pleistocene.

While the origin of Arvicola terrestris as a gradual evolution of the phyletic lineage Mimomys savini—Arvicola cantianus (or Arvicola mosbachensis, see Maul et al. 2000)— A. terrestris, seems to have some consensus (Rekovets 1990; Kolfschoten 1993), the origin of A. sapidus is more doubtful. The first appearance of the most primitive representative of the lineage of A. terrestris seems to be well-established as having occurred during the second half of the “Cromerian Complex” (Kolfschoten 1993). Transitional forms are known from several sites (Râdulescu and Samson 1993; Kolfschoten 1990; Koenigswald and Kolfschoten 1996; Heinrich 1987).

For the origin of A. sapidus and its relationship to A. terrestris as well as to M. savini different models have been discussed. Rabeder (1981) postulates two lineages: one leading to A. sapidus and the other to A. terrestris; according to this model, the lineages became separated long before the beginning of the middle Pleistocene. Molecular genetic studies have dated the split of A. terrestris and A. sapidus around 250 Ka (Centeno-Cuadros et al. 2009b) and the genetic distance between A. terrestris and A. sapidus confirms their specific independence (Diaz de la Guardia and Pretel 1978; Rekovets 1990).

The aim of this paper is to present a new species of a large vole of early Pleistocene age, related to Mimomys and to Arvicola, in terms of size, morphology and enamel differentiation. So far it has only been found in the lower levels of the Sima del Elefante site from the Lower Pleistocene of the karstic complex of the Sierra de Atapuerca in Burgos (Spain).

Fig. 2.

Geologie profile and lithostratigraphic units (TE7 to TE21) of the cave deposits of the Sima del Elefante locality, in the Trinchera del Ferrocarril of Sierra de Atapuerca Hill. Below the trench (Trinchera) level there are nearly 6 m of unearthed sediments, though levels TE8 and TE7 were sampled in the north trench perforation during the field campaign of 1996. The floor of the cave was identified by the presence of a 40-cm thick limestone layer in the north drill. Underneath the limestone layer, a distinct set of sediments was found, which was interpreted to represent the infill of a cavity from a lower karst floor. Two strata have been distinguished: the uppermost, composed of clay, small clasts (1–2 cm), and manganese oxide; the lowermost, with rounded quartzite pebbles, is interpreted as part of a terrace of the Arlanzón River left into the cave. Modified from Huguet Pàmies 2007.

Geological setting

The Sima del Elefante site (labelled TE in stratigraphic, palaeontological and archaeological samples) is a major cave infill located in an ancient railroad cut (Trinchera del Ferrocarril) of the Sierra de Atapuerca (Figs. 1–3). It corresponds to an ancient opening to Galería Baja in the Cueva Mayor Karst subsystem, and is completely full of sediments of allochthonous origin. The stratigraphic section of Sima del Elefante, including Lower and Middle Pleistocene deposits, is 25 m thick, 19 m of which were exposed during the construction of the railway (Figs. 2, 3). This section comprises 22 levels, the lowermost of these, known as the Sima del Elefante Lower Red Unit (TE-LRU), are Early Pleistocene in age (Cuenca-Bescós et al. 2001; Rosas et al. 2004; Cuenca-Bescós and García 2007; Carbonell et al. 2008). The sample of fossil voles analysed here comes from the Lower Red Unit of the Sima del Elefante Atapuerca site (Fig. 3).

TE-LRU comprises levels TE7 to TE 14 and is extremely rich in faunal remains, most of which are very well preserved. The TE-LRU represents the Atapuerca Faunal Unit 1 of Cuenca-Bescós and García 2007 and Cuenca-Bescós et al. 2010). The faunal assemblage of TE-LRU is given in Table 1. Small mammals proved to be crucial in dating the Lower Red Unit of the Sima del Elefante site as the oldest one in Atapuerca (i.e., pre-Jaramillo faunas; Laplana and Cuenca-Bescós 2000; Rofes and Cuenca-Bescós 2006, 2009, in press; Cuenca-Bescós and García 2007). Moreover, the Lower Red Unit of the Sima del Elefante site is below a normal polarised zone in the stratigraphic sequence as indicated in Fig. 3. The arvicoline and insectivore associations, in particular, showed that this unit indicates an Early Pleistocene age, similar to the localities of Fuente Nueva 3 and Barranco León in the Granada basin (ca. 1.2–1.5 Ma, Cuenca-Bescós and García 2007; Carbonell et al. 2008; Cuenca-Bescós et al. 2010). This is consistent with the presence of the large-sized mustelid Pannonictis nestii, referable to the Early Pleistocene (García et al. 2008). Recent cosmogenic nuclide analysis dated the Sima del Elefante TE9 level to ∼1.2 Ma (Carbonell et al. 2008), thus confirming the biostratigraphically-inferred chronology (Table 1).

Fig. 3.

Synthetic column showing the lithostratigraphic units of the north section of the Sima del Elefante site, labelled TE (levels TE7 to TE19). Levels that correspond to the Lower Red Unit, TE-LRU are TE7 to TE14. Units TE20 to TE22 do not outcrop here, but in the southern section (see Fig. 2). Black arrow points to level TE9 from which the recently discovered human remains were retrieved, and where the burial dating, based on the radioactive decay of cosmogenics, has been calculated (Carbonell et al. 2008). The small arvicoline mandible points to the level (TE13) where the holotype of Arvicola jacobueus sp. nov. was found. The figure has been modified from Carbonell et al. (2008).

Table 1.

Early Pleistocene vertebrates of the lower levels or Lower Red Unit of the Sima del Elefante site, Atapuerca, Burgos, Spain.

Noteworthy is that the Lower Red Unit from Sima del Elefante, from where the new fossil vole was retrieved, yielded, during the field campaign of 2007, a hominin mandible that, together with Mode 1 lithic tools and faunal remains bearing traces of hominin processing, represents the oldest and most accurately dated record of human presence in Europe (Carbonell et al. 2008).

Material, methods, and terminology

A total of 121 dental and mandibular elements have been analysed. Some of the teeth, like those of the holotype, are still associated with the dentary (Figs. 4, 5, 6), although the majority consist of isolated teeth (Appendix 1, Fig. 7). The specimens were obtained by concentrating the sediment after a process of washing and sieving, with 0.5 mm mesh-size, the sediments excavated from Sima del Elefante during the course of the Atapuerca campaigns between 1999 and 2007, plus two previous sample pits in 1995–96. The fossil remains of small vertebrates are then packed in plastic sacs, labelled with an abbreviation indicating the year of the field campaign, the site, stratigraphic level, the excavation grid unit, and the depth (Z) in cm where appropriate, (e.g., ATA 04 TE-13 H-31 Z: 1150–1160), enabling us to maintain stratigraphic control throughout. Afterwards, in the Laboratory of Palaeontology of the University of Zaragoza, the small fossil remains are picked and sorted using a hand-held or binocular magnifying glass, and a preliminary classification of the fossils is recorded in the data base of the Atapuerca small vertebrates sampling record. The fossils are stored in the Palaeontology section of the Department of Earth Sciences of the University of Zaragoza, Spain. For the measurements and study in detail of the molars and mandible we use a binocular microscope with 7×, 16×, 20×, and 40× of magnification. The details and photographs of Figs. 4, 5, 7, were made with a Scanning Electron Microscope. The specimens studied for this paper are kept in the MPZ.

The nomenclature and measurements for the various parts of the occlusal surface of the arvicoline molars follows the proposal put forward by Meulen (1973), with certain modifications (Fig. 4), derived mainly from the works of Rabeder (1981), Repenning (1992), and our own studies. We here apply a new way of measuring the occlusal surface of the molars of arvicolines by the use of an image-capturing programme that allows one to measure different parameters in objects inscribed within a particular region of interest (ROI). Length, width, anteroconid length “a” and posterior complex “p” were measured in the rectangle (ROI) in which the outline of the tooth is inscribed (Figs. 4C, 7M). The parameter “p” is the length of the posterior complex in M2 and M3 measured from the mesial limit of T3 to the distal edge of the molar (Fig. 7). This parameter makes it possible to estimate the posterior elongation of both molars.

Fig. 4.

Dental terminology and measurements taken on the occlusal surface of the m1 of arvicoline molars from Heinrich (1978) and Meulen (1973), modified. A. Leading and trailing edges of a dentine field or triangle. B. The occlusal surface of a lower first molar of Arvicola jacobaeus sp. nov., with the triangles (T) labelled from distal to mesial (T1–T5). C. Measurements taken on the occlusal surface of the molar. Abbreviations: a, anteroconid length (AC-2+T4-T5); AC-2, anteroconid complex; b, width of the anteroconid complex; BRA 1–3, buccal re-entrant angles; c, opening or separation of the T4 and T5 triangles; EFA, enamel free areas; L, total length; LRA 1–4, lingual re-entrant angles; W, total width.

The enamel thickness is measured in accordance with the method devised by Heinrich (1978; 1987) and Koenigswald and Kolfshoten (1996), where the SDQ (enamel differentiation quotient) is the average quotient obtained by dividing the thickness of the trailing edge by that of the leading (see the meaning of trailing-leading edges in Kesner 1980) edge of the enamel fields of the first lower molar (m1). This method was modified by Kolfschoten (1990), applying the measurements to all the molars, both upper and lower (Fig. 4A). We calculate the SDQ with the formula proposed by Heinrich (1987), and Koenigswald and Kolfschoten (1996) slightly modified:

Where teet is the trailing edge enamel thickness of the dentine fields and leet is the leading edge enamel thickness of the dentine fields. The sum of the resulting index of each dentine field is divided by N, the number of total calculated indices of each tooth.

The study of the anatomical characters of the mandible is based on the work of Repenning (1968) and Kesner (1980).

Systematic palaeontology

Fig. 5.

Occlusal view of the right mandible with m1, m2, m3 of the cricetid rodent Arvicola jucobaeus sp. nov. from Sima del Elefante level TE 13. Holotype (ATA04 TE13 H31 1150–1160, MPZ 2008/380).

Fig. 6.

Medial view of the right mandible with il, m1, m2, m3 of the cricetid rodent Arvicola jacobaeus sp. nov. from Sima del Elefante level TE 13. Holotype (ATA04 TE13 H31 1150–1160, MPZ 2008/380). The photograph was made during the field campaign of 2005 by Javier Trueba.

Fig. 7.

Occlusal views (except some root or basal views) of isolated teeth of the cricetid rodent Arvicola jacobaeus sp. nov. from Sima del Elefante level TE9. A. Right m1 (TE9 1 1997 MPZ 2008/343). B. Left m1 (TE9 1 1998 MPZ 2008/344). C. Left m1 (TE9 s/n 1998 MPZ 2008/372). D. Right lower m2 (tooth in mandible with the m1 of A) E. Left lower m2 (tooth in mandible with the m1 of B). F. Right lower m3 (TE9 1 1998 MPZ 2008/345). G. Left M1 (TE9 s/n 1998 MPZ 2008/374) H. Left M1 (TE9 s/n 1998 MPZ 2008/376). I. Root view of left M1 (TE9 s/n 1998 MPZ 2008/375). J. Left M2 (TE9 s/n 1998 MPZ 2008/377). K. Root view of right M3 (TE9 1 1998 MPZ 2008/346). L. Detail of the incipient root development in the M3 in K. M. Left M3 (TE9 1 1997 MPZ 2008/342). Abbreviations: L, total length; W, total width; P, posterior length; 2,3, triangles 2,3. The arrow in L shows the thicker enamel and the closening of the crown in basal view.

Discussion

Compared with the rest of the arvicolines of the Lower Red Unit of the Sima del Elefante (Atapuerca) site, Arvicola jacobaeus is the largest. The occlusal morphology of the m1 is simpler than that in Allophaiomys lavocati, with an elongated anteroconid and Microtus-type enamel differenciation (Laplana and Cuenca-Bescós 2000). The species Allophaiomys nutiensis and Allophaiomys burgondiae are also smaller and have the Microtus-type of enamel (Cuenca-Bescós et al. 2010).

Compared with other medium to large voles of the Early Pleistocene, Arvicola jacobaeus is similar in the occlusal surface of its molars to Mimomys savini, with a variable confluence between T3, T4, and the posterior lobe. The T4 is less developed than in Mimomys savini. The new species of Sima del Elefante lacks the enamel islet (which occurs mainly in rather young specimens in M. savini samples), the Mimomys fold, and the roots. In basal view, one M3 has a thick wall and a slight curvature which may be the stage prior to the formation of true roots (i.e., the sample of Mosbach 2, Maul et al. 2000) as in the M2 of Mimomys intermedius described by Râdulescu and Samson (1993). Note that M. intermedius is considered a synonymy of M. savini by most students (see i.e., Rekovets 1990; Gromov and Poliakov 1992; Maul et al. 2000).

During the Middle and Late Pleistocene appeared Arvicola cantianus, Arvicola sapidus, and Arvicola terrestris, as recorded in the introductory part. The three species are somewhat larger than Arvicola jacobaeus, both in length and in width, while Arvicola mosbachensis from Mosbach and their common ancestor (Maul et al. 2000), Mimomys savini, are somewhat smaller than A. jacobaeus (Fig. 8).

Arvicola jacobaeus has abundant cementum in the re-entrant angles, unlike the species of the genus Borsodia Jánossy and Meulen, 1975, a small- to medium-sized lagurine vole, with an occlusal surface superficially similar to that of Mimomys, which lacks cementum (Jánossy and Meulen 1975; Popov 2001).

Fig. 8.

Length—width scatter diagram of large arvicolinae rodents: Mimomys savini from the Lower Pleistocene levels of Gran Dolina, Atapuerca; Arvicola terrestris from the Upper Pleistocene levels of E1 Mirón, Cantabria; Arvicola jacobaeus sp. nov. from the Lower Pleistocene, Sima del Elefante levels TE9–TE13; Arvicola mosbachensis from Mosbach 2. Except the data from Mosbach 2, taken from Maul et al. (2000), the measurements were made by us.

The genus Kislangia Kretzoi, 1954 has the m1 larger than 4 mm, and confluent dentine fields, especially in Kislangia gusii from the Late Pliocene of Almenara-Casablanca 1 and Galera 2 in Spain (Agustí et al. 1993). In Arvicola jacobaeus the dentine fields are closed or exhibit just slight confluence, in particular the three distal ones, T1–T3 (Kretzoi 1954; Agustí et al. 1993; Minwer-Barakat et al. 2004).

The species Allophaiomys vandermeuleni Agusti, 1991 is slightly smaller, the enamel is of the undifferentiated type (Agustí 1991), and lacks roots.

The species Tibericola deucalion (Kretzoi, 1969) is also smaller, the SDQ lower, the enamel of the Mimomys-type, and the roots are lacking (Meulen 1974; Garapich and Nadachowski 1996; Tesakov 1998a). Ruiz Bustos and Sesé (1985) proposed the re-assignation of Allophaiomys deucalion to Arvicola deucalion, and at the same time as the first representative of the A. deucalion—Allophaiomys chalinei— Arvicola mosbachensis lineage. We here want to remark that (i) the molars of the species A. deucalion has the Mimomys-type. of differentiation of the enamel-thickness and lack roots, therefore it represents the most primitive species of the lineage Allophaiomys—Microtus; (ii) it cannot be the ancestor of Arvicola mosbachensis because the molars of this species exhibit incipient roots to different extent (Maul et al. 2000).

The molars of the species Mimomys oswaldoreigi Agustí, Castillo, and Galobart, 1993 are smaller, though similar to Arvicola jacobaeus in the derived occlusal characters such as the lack of the enamel islet and Mimomys-ridge in the m1 (Agustí et al. 1993). The lack of roots, except in the M3, is also mutually shared.

At this point, mention should also be made of the affinities of Arvicola jacobaeus with Tibericola jordanica Haas, 1966, from Ubeidiya (Koenigswald et al. 1992) and Yesher Benot Ya'aqov, Israel (Goren-Inbar et al. 2000). T. jordanica was initially classified within the genus Arvicola and described as a medium-sized to large species, lacking roots, with abundant cementum in the re-entrant angles and with T4–T5 separated to slightly confluent. Some specimens of T. jordanica display morphologies resembling Chionomys and not developed in Arvicola jacobaeus. The M3 is complex, with three well formed triangles. Another difference from A. jacobaeus is that the enamel thickness of T. jordanica is undifferentiated.

Table 2.

Measurements of the upper molars of Arvicola jacobaeus sp. nov. from Sima del Elefante, Atapuerca, Burgos, Spain, levels TE9–TE13.

A less derived species of Tibericola, Tibericola vandermeuleni, was described by Agusti (1991) from the Guadix-Baza basin (Granada, southern Spain), and originally assigned to the genus Allophaiomys. Tibericola vandermeuleni shares with Arvicola jacobaeus the ever-growing molars and a relatively simple and short anteroconid complex. However, the species are clearly different in many other respects. Arvicola jacobaeus is larger than T. vandermeuleni and shows Mimomys-type differentiated enamel, while the latter species has undifferentiated or slightly Mimomys-type. enamel. Moreover, A. jacobaeus has larger B values and much larger C values (sensu Meulen 1973; Agustí 1991). As in other derived hypselodont microtine species, the M3 of T. vandermeuleni is more complicated than this of A. jacobaeus, which in this way is more comparable to M3 of Mimomys.

Arvicola jacobaeus also differs from some large species of Allophaiomys, such as Allophaiomys chalinei Alcalde, Agustí, and Villalta, 1981. Again, A. jacobaeus shares with the latter species the presence of hypselodont molars and its relatively short, simple anteroconid complex. However, the species from Sima del Elefante is considerably larger and presents much larger B values (Tables 2, 3). While in A. jacobaeus the enamel is always of the Mimomys-type differentiated, in Allophaiomys chalinei it varies, being even of the Microtus-type differentiated in the more recent populations (such as Gran Dolina levels TD5 and TD6; Cuenca-Bescós et al. 1995, 1999b, 2001). As with Tibericola, the occlusal morphology of the M3 of Allophaiomys chalinei is more complicated than this of A. jacobaeus and in basal view it lacks any trace of roots.

Table 3.

Measurements of the lower molars of Arvicola jacubaeus sp. nov. from Sima del Elefante, Atapuerca, Burgos, Spain, levels TE9–TE13.

Comparison of the enamel thickness of Arvicola jacobaeus with that in other large voles of the Pleistocene of Europe.—The separation between the species of the genus Mimomys and those of the genus Arvicola is based exclusively on the presence or absence of roots. For this reason it is particularly important to analyse the evolution of enamel thickness (SDQ) in the European large voles (Fig. 9, Table 4). The enamel differentiation of A. jacobaeus is of the Mimomys-type; see this work and descriptions in Röttger (1987) and Ventura and López-Fuster (1989).

In 1978 and subsequently, Heinrich observed that the enamel of Arvicola evolved from a differentiated enamel of the Mimomys-type to a differentiated enamel of the Microtus-type, thicker in the leading edge (mesial in the lower molars and distal in the upper molars) (Heinrich 1978). Arvicola jacobaeus retains the Mimomys conditions, both in terms of enamel thickness and the form of the anterior lobe of m1. However, it lacks (through presumed loss) three of the features that characterise Mimomys, the fold, the islet, and the roots. On the other hand, the enamel differentiation of A. jacobaeus shows great resemblance to that of Mimomys savini, Arvicola cantianus, and Arvicola sapidus (Fig. 9). What is more, the enamel of A. jacobaeus tends to be uniformly thicker throughout the enamel wall (Fig. 7A–H, J) in comparison to Mimomys, Allophaiomys, and other Arvicola species.

The origin of the southern water vole Arvicola sapidus

In agreement with the above discussions we can conclude that both extant species of Arvicola share a common ancestor, with some authors implicating Mimomys savini (Kolfschoten 1990, 1993; Râdulescu and Samson 1993; Koenigswald and Kolfschoten 1996; Heinrich 1978; Maul 2001). However, some authors, such as Rabeder (1981), postulate two different lineages leading to the extant Arvicola sapidus and Arvicola terrestris, which became separated before the beginning of the Pleistocene. Ruiz Bustos and Sesé (1985) also postulated a different origin (from the species Allophaiomys deucalion) for the lineage Arvicola mosbachensis—Arvicola sapidus.

The species Arvicola sapidus is morphologically more primitive than the species Arvicola terrestris (Cubo et al. 2006, this work), and also in its karyotype (Díaz de la Guardia and Pretel 1978). The new species Arvicola jacobaeus displays a mosaic of primitive, mimomyan and derived microtian character states. This is not unusual in the Iberian Peninsular, where other species of Arvicolinae exhibiting mixed characteristics have been described, such as Mimomys oswaldoreigi of the Early Pleistocene and Tibericola vandermeuleni of the middle Early Pleistocene. A morphological affinity with the species Mimomys savini is indicated by the size and the morphology of the occlusal surface in Arvicola jacobaeus, with enamel fields that are slightly confluent in some specimens, the general thickness of the enamel, the type of differentiation and the loss of roots in conjunction with the retention of the capacity to form them. This same morphology, the enamel differentiation and the loss of roots, also indicate an affinity with the species A. sapidus, which is nonetheless distinct from A. terrestris, in particular as regards the SDQ values (Fig. 9, Table 4) and the more advanced character of the occlusal morphology in m1 and M3.

Fig. 9.

Comparison of the SDQ values of large arvicoline rodents from selected literature (see text) as well as extant Arvicola sapidus from Spain (localities and ages listed in the same order as they appear in Table 4).

Table 4.

Enamel thickness (SDQ) values of large arvicoline rodents of Europe. Measurements: Gran Dolina, levels TD3–TD6, Cuenca-Bescós et al. 1995; Gura Dobrogei and Casian caves, Radulescu and Samson 1993; several Holsteinian, Eemian, Saalian, Weichselian, and Weimar-Ehringsdorf localities in the Pannonian basin and Central Europe in Heinrich 1982; Miessehheim 1, Kolfschoten 1990; two Recent populations of Arvicola sapidus in Röttger 1987; Untermassfeld, Voigtstedt in Maul et al. 1998. Extant A. sapidas, and the Pleistocene A. sapidus from L'Abric Romani together with A. aff. sapidus from Galeria, Atapuerca, and A. terrestris from El Mirón, were measured for this work.

Table 5.

Table of characters of the first lower molars (m1) and upper third molars (M3) of Arvicola jacobaeus sp. nov. from Sima del Elefante, Atapuerca, Burgos, Spain, levels TE9–TE13, and related arvicoline species used in the cladistic analysis.

The characterisation of the chromosomal complement, in number of chromosomes, shows that A. terrestris is a derived form (with more advanced chromosome number, 2n = 36), which split off the main lineage at an unknown time in the Pleistocene. There is an opinion that A. sapidus, with an ancestral karyotype (2n = 40) was present in most Eurasian faunas prior to the Late Pleistocene, whereas the chromosome speciation and a wide distribution of A. terrestris occurred in Late Pleistocene (Díaz de la Guardia and Pretel 1978; Kalthoff et al. 2007). Our own studies in Middle Pleistocene localities of Atapuerca such as Galería (Cuenca-Bescós et al. 1999a), as well as those by other authors based on materials from other sites in southern Europe (López Martínez and Ruiz Bustos 1977; López Martínez 1980; Râdulescu 1989), demonstrate that an ancestral form of A. sapidus, generally classified in the mentioned works as A. cf. sapidus, was already present during the Middle Pleistocene. Synaptic sex chromosomes, the ancestral condition of arvicolines, are present in A. sapidus (Megías-Nogales et al. 2003), reinforcing the idea of the primitive character of this species. Heterochronic evolution of the species A. sapidus also proves its primitive condition (Cubo et al. 2006).

Phylogenetic analysis

Morphologic affinity among Arvicola sapidus and Arvicola terrestris suggests a common ancestor; probably Mimomys savini as mentioned above. However, we know very little about the relationships among the extant species and their purported relatives of early, middle, and late Pleistocene age. Consequently we perform here a preliminary initial data exploration in a phylogenetic approach by analysing the characters discussed throughout the text in the m1 and the M3 (Table 5) using PAST (Hammer et al. 2001). Being well aware of its tentative quality the resulting cladistic analysis reveal that M. savini is probably the sister group of the Arvicola species, and that A. sapidus and A. terrestris are more closely related to each other than to other Arvicola species (Fig. 10) including the new species described here.

Fig. 10.

Cladogram of fossil and extant arvicoline species related with Arvicola jacobaeus sp. nov. from the Lower Pleistocene, Sima del Elefante levels TE9–TE13. The cladistic analysis has been performed using PAST; to find the shortest tree we use the branch and bound algorithm and the Fitch character optimisation criteria, and 1000 bootstrap replicates (Hammer et al. 2001). The number of Most Parsimonious Tree is one, the tree length 22, the Consistency Index 0.7273, and the Retention Index 0.6. The Bremer support or Decay Index has not been analysed.