The Buena Pinta Cave lies some 80 km north of Madrid in the upper reaches of the Lozoya Valley, a depression of tectonic origin in the Sierra de Guadarrama, a mountain range forming part of the Spanish Central System (Fig. 1). To the west of the valley lies the Peñalara Massif, which reaches an altitude of 2428 m – the highest in the Sierra de Guadarrama. The valley itself lies between the altitudes of 1000 and 1100 m. With at least 41 mammal species it has a high biological diversity (Prieto Cana and de Lucio, 1995), a consequence of the many habitats resulting from its steep altitude gradient and varied geomorphology, climate, soil and hydrology. Indeed, the upper reaches of the valley were designated a National Park in 2013.

The cave itself, of reduced dimensions, is developed in Cretaceous carbonatic rocks, and was discovered in 2003 during surveys performed as part of a project on human occupation of the Peninsular centre during the Pleistocene (Arsuaga et al., 2011, Baquedano et al., 2014 and Pérez-González et al., 2010). The cave has an external chamber where excavation work has been focused since 2003 (Fig. 2). The limestone roof in this area has been eroded and has fallen away. To the east of this area a narrow gallery (about 1.5 m wide) opens that continues some 10 m into the Cretaceous limestones and dolostones in a NNE direction. The sediments in the external chamber and this gallery show stratigraphic continuity. In grid squares M49 and M50 (Fig. 2) at the mouth of the gallery, excavation work has revealed a stratigraphic section with five distinguishable levels (Pérez-González et al., 2010) (Fig. 3). Level 1 is formed from Holocene colluvium some 1.80 m thick. It has a clay sand texture and contains grey-brown and dark grey carbonate clasts. Lying below is a 1.70 m thick layer of yellow, silt-clay sands containing disperse carbonate clasts measuring up to 40 cm along their longest axis. This layer can be divided into four levels (numbered 2–5 from top to bottom) distinguishable by the colour and texture of their sediments. Levels 2–5 are rich in macro- and microvertebrate remains, that have been dated to the middle of the Late Pleistocene (see Section 4.3). These levels lie discordantly over a much eroded speleothem. Beneath this speleothem lie conglomerates and carbonate breccias of limestone and dolomite boulders that make up one of the first stages of infilling of the cave; they also crop out on the northwestern wall forming the boundary of the external chamber. This unit shows continuity to the northeast and completely fills another gallery, the roof of which has disappeared through erosion. A survey of this gallery (known as the Galería Pit) returned a wide range of microvertebrates. The presence and evolutionary stages of the detected species Microtus vaufreyi and Microtus brecciensis allow the materials of this gallery to be dated to the end of the Middle Pleistocene (Laplana et al., 2013a).

The remains of the cold-climate rodents examined in the present work all come from levels 2–5 of the external chamber of the cave.

Since its discovery, the Buena Pinta Cave has been the object of annual excavation campaigns. All the archaeological and palaeontological remains found were spatially referenced on an excavation grid. Samples of sediment were taken at 10 cm depth intervals in each 1 × 1 m grid square. For the detection of microvertebrate remains, the collected samples were washed through a series of fine-mesh sieves (down to 0.5 mm). The vertical and horizontal variations in the microvertebrate assemblages therefore became appreciable as work progressed.

The nomenclature used to describe the arvicoline dental elements recovered is that proposed by Van der Meulen (1973) and Nadachowski (1991) (Fig. 4a). The systematic taxonomy followed is that proposed by Musser and Carleton (2005). Measurements of the first lower molars (m1 onwards) were taken following the procedure shown in Fig. 4b. The measurements L (maximal length of the occlusal surface), W1 (width of the posterior part of the anteroconid complex), a (length of the anteroconid complex), b (shortest distance between BRA3 and LRA4), and c (shortest distance between BRA3 and LRA3) are those proposed by Van der Meulen (1973). W2 is the width across T2 and T3. The variables La1 and La2 are similar to those proposed by Cuenca-Bescós et al. (1995). The morphometric ratios A/L, B/W1 and C/W1 are those proposed by Van der Meulen (1973). The ratio L/W was also calculated, as were two others reflecting transversal asymmetry: La1/W1 and La2/W2. The morphotypes identified in the description of the variability of the first lower molars of Chionomys nivalis are those proposed by Nadachowski (1991), and in this work they are used with the same meaning.

The excavation work performed over the last 12 years at the Buena Pinta Cave has led to the recovery of tens of thousands of small mammal teeth. For instance, grid units M49 and M50, which represent only a small part of the entire site (Fig. 2) have provided 2744 rodent first lower molars (Table 1). Indeed, the site is home to one of the most species-rich assemblages for the entire Iberian Late Pleistocene [see provisional list in Baquedano et al. (2010)]. From levels 2–5, 18 rodent species were identified, mostly arvicolines (Table 1). Their abundance was estimated from the total number of first lower molars recovered for each.

During the coldest moments of the Pleistocene, ice sheets spread across the North of Europe (Ehlers and Gibbard, 2004). The expansion of steppe-tundra ecosystems towards the south and west led to important modifications in the distributions of the animals and plants that inhabited the continent. The large mammals that became displaced south and westward are together known as the Mammuthus-Coelodonta faunal complex (Kahlke, 1999). They reached their most southerly and westerly distribution during the Late Pleistocene (Kahlke, 2014); indeed, many large species penetrated the peninsulas of the Mediterranean, including the Iberian Peninsula (Álvarez-Lao and García, 2010, Álvarez-Lao and García, 2011 and García and Arsuaga, 2003) where remains of Mammuthus primigenius, Coelodonta antiquitatis, Rangifer tarandus, Gulo gulo, Alopex lagopus, Ovibos moschatus and Saiga tatarica have all been discovered. Some of these species, such as the woolly mammoth, reached as far as Granada in southern Spain (Aguirre et al., 1973 and Álvarez-Lao et al., 2009). These large animals were accompanied in their displacement by a wide range of smaller ones, including some rodents and lagomorphs. Among them, M. oeconomus, M. gregalis, S. betulina, Sp. superciliosus, Sp. major, L. timidus and O. pusilla have all been recorded for the Late Pleistocene of the Iberian Peninsula (Arribas, 2004, Laplana et al., 2009, López-García, 2011, López-García et al., 2010 and Rofes et al., 2012). Currently, however, they are not found there, but much further to the north and east of Europe, where the climate is colder. Most of the palaeontological sites where these have been found are in the northern third of the Peninsula, especially in the Cantabrian Mountains and Catalonia (Álvarez-Lao and García, 2011). Only a few pieces of evidence exist that suggest they reached any more southerly latitudes; the Buena Pinta Cave is one of these. The record for M. oeconomus at this site is one of the most southerly for the species in Europe. Indeed, it is known from only one more southerly location, the Abrigo del Monte site (Sevilla et al., 2009) some 25 km to the south of the Buena Pinta Cave, where it appears in a somewhat more modern Magdalenian setting (MIS2). A citing for M. aff. ratticeps (a synonym of M. oeconomus) from further south at the Las Yedras site in Granada (Ruiz Bustos, 1978) may be the result of an erroneous identification. An inspection of these materials revealed many to show a partial confluence of T4 and T5, which does not occur in M. oeconomus. The materials thus identified may in fact belong to Microtus huescarensis (see Figs. 8 and 9 in Plate 1 of Ruiz Bustos, 1978) and perhaps to Microtus chalinei (see Figs. 10 and 11 in Plate 1 of Ruiz Bustos, 1978).

The record for M. gregalis at the cave is the southernmost for Europe. Until now, the most southerly citings were those for Erralla (Pemán, 1985), El Mirón (Cuenca-Bescós et al., 2008), Sopeña (Pinto-Llona et al., 2012) and perhaps the Laminak II (Pemán, 1994) sites, all in the Cantabrian region to the north.

The chronology of the sediments at the cave is important in understanding the biogeographical significance of M. oeconomus and M. gregalis. Levels 2–5 have been dated to an interval ranging from MIS4 to the beginning of MIS3. Although some northern European rodent species colonised the Iberian Peninsula during the beginning of the Late Pleistocene, as shown by the presence of S. betulina in MIS5 levels at the Lezetxiki II site (Rofes et al., 2012), most northern rodent records are for sites corresponding to MIS3 and (especially) MIS2, which includes the Last Glacial Maximum (LGM) (López-García, 2011 and Sesé, 2005). Until now, the oldest records for M. oeconomus inhabiting the centre of the Peninsula were from the Magdalenian (MIS2) levels of the Peña de Estebanvela site, some 30 km to the north of the Buena Pinta Cave (Sesé, 2006 and Sesé, 2013), and the Abrigo del Monte site, some 20 km to the south (Sevilla et al., 2009). The finding of M. oeconomus and M. gregalis remains in the Buena Pinta Cave confirms these species arrived well before MIS2.

Whereas the latter species no longer live in the Iberian Peninsula, C. nivalis still does, although only in mountainous regions, especially at altitudes of between 1000 and 2600 m (depending on the mountains in question) (Luque-Larena and Gosálbez, 2002). Although temperatures are usually low at this altitude, the species is not as well adapted to the cold as M. oeconomus or M. gregalis (Nappi, 2002 and Pérez-Aranda Serrano, 2009). Its link with mountainous areas lies in that these produce the petricolic environments that provide the species its microhabitat. Chionomys nivalis lives in the spaces between rocks (normally of periglacial origin) in open areas, where the conditions of thermal stability upon which the species depend are found (Krystufek and Kovacic, 1989 and Luque-Larena et al., 2002). Indeed, C. nivalis still lives not too far from the Buena Pinta Cave, at altitudes of over 1800 m (Pérez-Aranda Serrano, 2009). The presence of its remains at the site, which is at 1105 m, indicates that during the Late Pleistocene the conditions it requires were available at lower altitude, i.e., there must have been periglacial scree available and open spaces. For this to have been the case, the temperature must have been lower than it is today. The presence of C. nivalis at the cave is therefore an indication of a colder climate having reigned during the Late Pleistocene.

Along with the remains of M. gregalis, M. oeconomus and C. nivalis, the single remain (a jaw fragment) belonging to O. pusilla found in level 3 (Laplana et al., 2009) is also indicative of a colder climate during the Late Pleistocene. This lagomorph is no longer found in the Iberian Peninsula, but on the steppes of central Asia. Like M. gregalis and M. oeconomus the abundance of O. pusilla is under 1%. Its identification was in part due to the concerted sampling effort undertaken and the richness of the site in terms of microvertebrate remains. Less intense sampling at other Central System sites of similar age may be the reason why they have not been noticed there – which may have affected the palaeoclimatic interpretations made for these sites. Nonetheless, the small number of remains recovered does indicate a low density of individuals for these species in the area during the Late Pleistocene. The Buena Pinta Cave now marks the second most southerly known Pleistocene site for M. oeconomus, and the southwesternmost limit for both M. gregalis and O. pusilla, and it is well known that population density tends to decrease away from the centre of a species’ distribution (Brown, 1984). For C. nivalis, the site does not mark its known limit of distribution; it has been cited for the Carihuela Cave in Granada, 400 km to the south (Ruiz Bustos, 2000 and Ruiz Bustos and García Sánchez, 1977), in levels chronologically similar to those of the Buena Pinta Cave. This may explain why it is more common than the other cold-climate indicators, although it never surpasses an abundance of 1% in any level.

The Buena Pinta Cave provides the southernmost European record of M. gregalis, and one of the most southerly (along with those of the Abrigo del Monte [Sevilla et al., 2009] and Peña de Estebanvela [Sesé, 2006 and Sesé, 2013] sites) for M. oeconomus. Thermoluminescence datings of levels 2–5, where these species were found, place them within a time interval ranging from MIS4 to the beginning of MIS3. This confirms that these species were present in the centre of the Iberian Peninsula before the LGM. Neither is currently found in the Peninsula, but farther north and east in Europe and Asia (and indeed in North America in the case of M. oeconomus). They may therefore be considered indicators that the climate was cold when the site formed.

C. nivalis, which still survives in the Peninsula, is also a marker of cold conditions during the Late Pleistocene at the site; it now lives at altitudes of over 1800 m (Pérez-Aranda Serrano, 2009). Since the site is located at 1100 m, the periglacial phenomena that produce the scree among which this animal lives must have been active at this lower altitude during the Late Pleistocene. Such a cold-climate would have produced the more open environments suitable to the species.

The relative and absolute rarity of M. oeconomus and M. gregalis at the site may be the result of their lying far from the centres of their distributions – indeed, at the southernmost edge of their ranges. This may explain why they have not been cited from sites of similar age and geographical location; they may not have been common enough for their remains to have been detected.

The presence at the cave of these three rodent species indicates that the climatic conditions of the surrounding region during the middle of the Late Pleistocene were colder than they are today. This agrees with conclusions drawn from the types of pollen found at the site (Ruiz-Zapata et al., 2008 and Ruiz-Zapata et al., 2012).