Insectivores are a group of mammals whose fossil record has assumed a special relevance in the Iberian Peninsula in the last few years (e.g. [10], [11], [21] and [45]). To this group belongs Beremendia fissidens (Mammalia, Soricidae), a kind of red-toothed shrew which inhabited Eurasia from the Early Pliocene to the beginnings of the Middle Pleistocene [30] and [62].

Although trustworthily described by Rzebik-Kowalska [61] and Reumer [52] in central Europe, the biostratigraphy, biogeography and morphological peculiarities of Beremendia are poorly known in the Iberian Peninsula. Besides, its general phylogenetic affinities urgently need a revision at the light of new challenging discoveries [30], [31] and [57]. As if it were not enough, its precise palaeoecological preferences have remained obscure and highly speculative.

In Spain, in particular, the records of B. fissidens up to now are scanty and all come from the Late Pliocene [1], [21], [37] and [44]. Pleistocene records correspond to previous misidentifications, and, with the exception of the unpublished Ph.D. dissertation of Furió [21], all said references are limited to the inclusion of B. fissidens in general faunal lists. In this paper we present evidence of the first bona fide examples of this large-sized soricid in the Iberian Peninsula during the Early Pleistocene (as chronologically stated by Aguirre and Pasini [3], based on the Vrica section in Calabria, Italy), which comes from the site of Sima del Elefante in Atapuerca (Burgos, Spain). We also take advantage of the opportunity to carry out a comprehensive review and updating of the current knowledge we have on the species, throwing light on such aspects as those mentioned before.

Sima del Elefante (TE labelled in stratigraphic, paleontological and archaeological samples) is a major cave infill located in the railroad cutting (Trinchera del Ferrocarril) of the Sierra de Atapuerca (Fig. 1). It may correspond 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 of which were exposed during the construction of the railway. This section comprises 22 levels, the lowermost of which, grouped and known as the Lower Red Unit (TE-LRU), are Early Pleistocene in age [5], [8], [10], [38], [58] and [59]. The sample of fossil shrews analysed here comes from the Lower Red Unit of the Sima del Elefante site (Fig. 2).

TE-LRU comprises levels TE8 to TE14 and is extremely rich in faunal remains, most of which are very well preserved; some of them were even found in anatomical connection [58] and [59]. The latter include (updated from Cuenca-Bescós and García [8]): (1) Eulipotyphla (Insectivores) – Asoriculus gibberodon, B. fissidens, Crocidura kornfeldi, Crocidura sp., Sorex cf. minutus, Sorex cf. araneus, Erinaceus cf. praeglacialis, Galemys cf. kormosi, Talpa cf. europaea; (2) Carnivora – Canis cf. arnensis/mosbachensis, Vulpes cf. alopecoides, Ursus cf. dolinensis, cf. Baranogale antiqua, Mustela cf. palerminea/praenivalis, Pannonictis nestii, Panthera gombaszoegensis, Lynx cf. issiodorensis; (3) Perissodactyla – Stephanorhinus etruscus, and Equidae indet.; (4) Artiodactyla – Hippopotamus sp., Eucladoceros giulii, Megaloceros savini, Cervidae indet., Bison sp., Bovidae indet, Suidae indet.; (5) Rodentia – Sciurus cf. wharthae, Eliomys quercinus, Castor fiber, Ungaromys nanus, Pliomys cf. simplicior, Allophaiomys lavocati, A. burgondiae, A. nutiensis, Arvicolinae gen. et sp. nov., Castillomys rivas, Apodemus sp.; (6) Lagomorpha – Lepus sp. and Oryctolagus sp.; (7) Aves – Carduelis chloris, Perdix paleoperdix, Coturnix coturnix, Columba livia, Haliaetus albicilla, Circaetus gallicus, Falco tinunculus, Falco sp., Turdus sp., Alaudidae indet., Anas sp.; (8) Fishes, amphibians and reptiles – Salmonidae indet., Bufo bufo, B. calamita, and Chelonia indet.

Small mammals proved to be crucial in dating the TE-LRU as the oldest one in Atapuerca (i.e. more than 1 Myr old [8], [10], [38] and [56]). The arvicolid and insectivore associations, in particular, showed that this unit may be Early Pleistocene, pre-Jaramillo in age, at least as old as the localities of Fuente Nueva 3 and Barranco León in the Granada basin (ca. 1.2–1.5 Myr [55]). This is consistent with the reverse polarity of the TE9 to TE16 levels, interpreted as Matuyama Magnetochron by Parés et al. [46], and with the presence of the large-sized mustelid Pannonictis nestii, referable to the Early Pleistocene [22]. Recent cosmogenic nuclide analysis dated the TE9 level to ∼1.2 Ma [5], thus confirming the biostratigraphically-inferred chronology.

Noteworthy is that the Lower Red Unit from Sima del Elefante (TE9) represents the oldest and most accurately dated record of human occupation in Europe. This takes the form of a hominin mandible, 32 Mode 1 stone tools, and many large-mammal bones with clear evidence of human processing [5].

We introduce here in advance the taxonomic authorities for the main genera and species of soricids mentioned in the text with the purpose of facilitating further reading. Those taxa are, in alphabetic order, namely as follows: Beremendia Kormos, 1934 [33]; Beremendia fissidens (Petényi, 1864) [48]; Beremendia minor Rzebik-Kowalska, 1976 [61]; Beremendia pohaiensis (Kowalski and Li, 1963) [35]; Beremendia ucrainica Pidoplitschka, 1956 [49]; “Dolinasorex glyphodon” Rofes and Cuenca-Bescós, in press [57]; Lunanosorex Jin and Kawamura, 1996b [31]; Lunanosorex lii Jin and Kawamura, 1996b [31]; Nectogalinia altaica Gureev, 1979 [25]; Neomys sinensis Zdansky, 1928 [71]; Peisorex Kowalski and Li, 1963 [35]; Peisorex pliocaenicus Flynn and Wu, 1994 [20]; Peisorex pohaiensis Kowalski and Li, 1963 [35].

The specimens were obtained by concentrating the sediment after a process of washing and sieving the sedimentary materials acquired from the excavations of Sima del Elefante during the course of the Atapuerca campaigns between 1999 and 2007, plus two previous sample pits in 1995–1996 (roughly 12–15 tons of sediment each year out of 25 processed for all the Atapuerca locations). The product was a concentrate consisting fundamentally of fossil remains of small vertebrates (packed in small plastic boxes). This was duly labelled with an abbreviation indicating in order the site and stratigraphic level from which it came, the excavation grid unit, the depth (Z) in centimeter where appropriate, and the year in which the remains were recovered (e.g. TE-9B/K-29/Z: 600-620/ATA 02), enabling us to maintain rigorous stratigraphic control throughout. Subsequently, the small fossil remains were separated from these concentrates using a hand-held or binocular magnifying glass. Afterwards, a preliminary classification of the fossils was undertaken, and these were recorded in the database and provisionally stored in the Department of Palaeontology of Zaragoza University. The specimens studied for this paper are kept in the Palaeontology Museum of Zaragoza University, abbreviated as MPZ.

The 171 specimens analysed in the present study have been drawn from the enormous volume of remains gathered in the course of 11 years of continuous excavations. The sample is composed of mandibles, maxillae and teeth that are either loose or in situ. Here we follow Reumer's [52] anatomical nomenclature, above all for comparative purposes given its nearly standardized use among scholars dealing with fossil shrews. For a slightly different approach, particularly with reference to the so-called “unicuspid” teeth, see Hutterer [29].

Metric values were recorded using a stereo-microscope connected to a video camera, which sent the images to a computer programme (Matrox Inspector) that made it possible to visualize and measure the specimens with great accuracy. We also follow Reumer [52] in the choice of measurements to be taken and the PE-index, with a few additions from Rabeder [50] and Rofes and Cuenca-Bescós [56]. The PE-index quantifies the degree of posterior emargination of the large upper teeth. This has been calculated for P4, M1 and M2 by means of the formula: LL + BL/2 PE – 1. For the morphometric analysis we used the PAST v1.77 statistical programme [26].

There is no doubt regarding the assignation of the remains from Sima del Elefante to the Soricinae subfamily, on account of the characteristic red pigment of the teeth and the presence of a conspicuous posterolingual basin in the fourth lower premolar. Even though it lacks a formal diagnosis, the genus Beremendia, defined by Kormos in 1934, does have a detailed description. Repenning [51] and Reumer [52] selected a series of characters from this description as the diagnostics of the group, to which Jin and Kawamura [30] added further observations. The result of this is the version presented in the following section.

Class MAMMALIA Linnaeus, 1758 [39]

Order EULIPOTYPHLA Waddell, Okada, and Hasegawa, 1999 [69]

Family SORICIDAE Fisher von Waldheim, 1817 [19]

Subfamily SORICINAE Fisher von Waldheim, 1817 [19]

Genus Beremendia Kormos, 1934 [33]

Beremendia fissidens (Petényi, 1864) [48]

Fig. 3 and Fig. 4.

Synonyms: Beremendia ucrainica Pidoplitschka, 1956 [49]

Nectogalinia altaica Gureev, 1979 [25]

Type species: Crossopus fissidens Petényi, 1864 [48]

Type locality: Beremend 1 (Hungary).

Studied locality: Sima del Elefante (Lower Red Unit) in the Sierra de Atapuerca (Burgos, Spain).

Selected other localities: See reviews by Rzebik-Kowalska [62], Storch et al. [67], and Furió [21].

Geographic and stratigraphic distribution: Eurasia, from the Early Pliocene (MN 14) to the late Early Pleistocene; early Middle Pleistocene?

Diagnosis: As stated by Jin and Kawamura [30]: dental formula 1-4-3/1-2-3; teeth stained red to almost black; first upper incisor markedly bifid; upper antemolars decreasing in size from A1 to A4; A4 reduced or lacking, and invisible from the buccal side of the skull; posterior emargination in P4 and upper molars moderate; parastyle of M1 normal and not strongly developed; coronoid process leaning anteriorly to varying degrees; upper pterygoid fossa markedly or slightly depressed; upper facet narrowly elliptic to oval; interarticular area very broad, and with lingual emargination; lower facet placed greatly anteriorly, and thus invisible in buccal view; internal temporal fossa deeply pocketed; first lower incisor without cusplets on its cutting edge; its apex markedly bent upward; p4 with two cusps and a posterolingual basin; entoconid crest with various degrees of development in lower molars.

Measurements: See Table 1a and Table 1b and the morphometric analysis below.

On the basis of a comparison between diverse fossil associations, Reumer [52] proposed that B. fissidens was an ecological opportunist, a ubiquitous species which apparently did not have narrow preferences. Nonetheless, the high metabolic rate of the soricids in general and of the red-toothed shrews in particular [7] would have made the consumption of vast quantities of food essential for these enormous shrews (i.e. with a body mass of 40–45 g [11]), whose development would not have been feasible but for a favourable ecological environment. We consider, therefore, that the abundance of Beremendia in the Lower Pleistocene levels of Sima del Elefante (TE8 to TE14) was linked with a temperate, humid and relatively stable climate, which is in accordance with the panorama inferred on the basis of previous studies of faunal associations [10], [22], [56], [58] and [59].

A tentative palaeoenvironmental reconstruction of the TE-LRU, based on the distribution of insectivores throughout the stratigraphic sequence [10], reflects a succession of three phases: warm–cold–warm (w–c–w), correlated with the Waalian pollen episode of northern-central Europe [70], the only one that contains the same phases w–c–w, and the chronology of which (1.25–1.5 Myr) concurs with the period of inverse magnetic polarity prior to the positive Jaramillo Subchron pinpointed for this period [5].

The best modern candidate for an actualistic approach to Beremendia could be the North American short-tailed shrew Blarina brevicauda, which is the typical opportunist: it lives in forests, grasslands, marshes and brushy areas and is not restricted to any habitat [4]. This moderately large species of soricine also possesses another singular characteristic: its saliva is one of the most toxic forms described for any mammal to date [32], even though it does not have such a specialized injection mechanism (i.e. grooved lower incisors) as the Caribbean solenodon, a very close relative of the shrew [17]. A recent study [11] shows that B. fissidens not only possessed incisors with a medial channel in the manner of the solenodons, but also a buccal architecture specially designed to increase the power of its bite upon possible prey.

The typical diet of soricids includes insects, worms, molluscs and other invertebrates [7]. However, the venomous saliva of B. brevicauda (and a few other shrew species) enables the animal to deal with relatively large prey, which is immobilized by the fluid, thereby reducing its struggling [14]. This also enables Blarina to store food that is living but comatose, paralyzed by the toxin [17]. We assume similar behaviour in Beremendia. It is possible that the great size reached by B. fissidens was an adaptation to the hunting and handling of increasingly large prey (e.g. fish, amphibians, reptiles, small birds and mammals, including other shrews). The need to immobilize such prey would have led to the selection and subsequent development of a sophisticated mechanism, the first of its kind among soricids, for injecting them with venomous saliva in a quick and efficient way, details of which are laid out in Cuenca-Bescós and Rofes [11].

Beremendia is a Plio-Pleistocene genus with a broad European distribution and a moderate presence in the Asiatic continent [62] and [67]. Following the biogeographical approach of Jin and Kawamura [30], we can construct a scenario in which Beremendia first appeared in the Early Pliocene of Europe and then expanded eastward to East Asia by the Middle Pliocene, becoming extinct there by the end of the Early Pleistocene while purportedly surviving until the early Middle Pleistocene in Europe.

The oldest records of B. fissidens, in particular, come from Hungary and Poland in Europe (MN 14 [62]) and from West Siberia in Asia (MN 15 [67]). In Europe, remains of the species have been recovered from more than a 100 sites, distributed over 17 countries, including Russia, the Ukraine, Hungary, Romania, Bulgaria, Poland, Germany, the Netherlands, Slovakia, Croatia, the Czech Republic, Switzerland, Austria, France, Greece, Italy, Spain and the United Kingdom [6], [34], [54], [62] and [63], and covering a time period that ranges from the Early Pliocene to the beginnings of the Middle Pleistocene. However, the remains of B. fissidens from this latter time are very scarce and restricted to just a few specimens from Hungary, Romania, France and the United Kingdom [62]. As it will be shown below, the reliability of said references should be reconsidered at the light of new challenging discoveries [57].

In Asia there are remains of B. fissidens in western Siberia and Mongolia dating from the middle to the end of the Pliocene (i.e. MN 15–17 [67]), with the sole exception of the fragmentary mandible described as “Nectogalinia altaica” by Gureev [25] and recognized as B. fissidens by Storch et al. [67], which comes from an Early Pleistocene site (i.e. Razdolye in western Siberia).

B. minor, whose presence is restricted to the Pliocene, was a species with a very limited distribution both in Europe and in Asia [62] and [67], while B. pohaiensis was an endemic species that inhabited the North of China during the Early Pleistocene [30]. It is possible that both species derived from B. fissidens, albeit in distinct circumstances and contexts: following a sympatric speciation model in the centre of Europe in the case of B. minor, and possibly an allopatric model in the case of B. pohaiensis, due to the geographical isolation of populations in the Far East of Asia. Another possibility, pointed out by Rofes and Cuenca-Bescós [57], is that the difference in size between B. fissidens and B. minor, groups that are otherwise identical [61], is a reflection not of specific diversity but merely sexual dimorphism.

In the Iberian Peninsular all the specimens of Beremendia found to date correspond to the species B. fissidens and all the references are from the Late Pliocene. We thus have Valdeganga 2 in Albacete (MN 16 [44]), Orrios 3 and Barranco del Monte 1 in Teruel (MN 16 [1] and [37]) and Almenara-Casablanca 1 in Castellón (MN 17 [21]). As Beremendia sp. it is cited by De Bruijin [15] in Moreda (Granada, MN 16).

There is also a series of references for the Early Pleistocene [2], [9], [12], [18], [23], [24], [40] and [65], all of which refer without exception to specimens identified as B. fissidens in levels TD4-6 of Gran Dolina, the site adjacent to Sima del Elefante in the Sierra de Atapuerca. However, Rofes and Cuenca-Bescós [57] have demonstrated that this material in fact corresponds to Dolinasorex glyphodon, a new species of larger soricine with certain features similar to B. fissidens, but phylogenetically closer to Asiatic forms (i.e. Lunanosorex lii). If we take into consideration the previous ignorance on this new Beremendia-like species, potential misidentifications throughout Europe included, together with some degree of stratigraphic uncertainty, particularly in the Middle Pleistocene records of the species (e.g. Kozi Grzbiet and Betfia [61] and [63]), we cannot be sure if B. fissindens really made it beyond the Early Pleistocene.

The Sima del Elefante specimens constitute by far the most recent population of B. fissidens recorded in the Iberian Peninsular to date, and the only one from after the Pliocene in this geographical area, which, on account of its particular environmental features, played the role of a “last redoubt” for a significant number of species in addition to the one that concerns us here. Examples of this from the group of insectivores include Asoriculus gibberodon, Crocidura kornfeldi, Dolinasorex glyphodon, Galemys kormosi and Erinaceus praeglacialis [21], [56] and [57].

The 171 specimens analysed from the Lower Red Unit (levels TE8–TE14) of the Sima del Elefante site would correspond, without any doubt, both from a morphological and a morphometric point of view, to the species Beremendia fissidens. It constitutes by far the most recent population of B. fissidens recorded in the Iberian Peninsular to date, and the only one from the second third of the Early Pleistocene in this geographical area. Slight variations in size, in comparison to other populations from central and southern Europe, may be due to particular environmental conditions in the Iberian Peninsula. The abundance of Beremendia in the Lower Pleistocene levels of Sima del Elefante was undoubtly linked with a temperate, humid and relatively stable climate, which is in accordance with the panorama inferred on the basis of previous studies of faunal associations.