The material recovered from Artazu VII consists of microfaunal (amphibians, reptiles, birds and small mammals) and macrofaunal (ungulates and carnivores) remains collected from the sedimentary fill of the fissure. The site was located in a zone that was to be destroyed because it was inside a working quarry, so all the sediment remaining inside was removed. In total, seventeen 7–15 cm thick arbitrary spits were excavated (Fig. 1C). These spits were designated in descending order from top to bottom, each one being allocated a depth, a letter or number. Thus, in the Upper Ledge they were ordered by depth (0–15), in the Lower Ledge alphabetically (from A to L) and in the Chamber numerically (from 1 to 5) (Fig. 1C). Most of the macrofaunal remains were disturbed, with species and individuals totally mixed, both horizontally and vertically; even with different anatomical elements corresponding to a single individual were distributed in more than one spit. Therefore, no squares and sectors were established. Also, we also studied all the material of macrofauna without stratigraphic context, so as to compose a list of all taxa present in Artazu VII.

While the macrofauna remains were being collected, all the sediment inside the fissure was removed to extract the microfauna. Two samples were taken from the Upper Ledge (total sediment volume: 2.5 L), twelve in the Lower Ledge (total sediment volume: 156 L) and five in the Chamber (total sediment volume: 82 L), amounting to an overall sediment volume of 240.5 L. The present study analyses part of Samples A, C, G, K and L from the Lower Ledge and Layers 1, 3 and 5 from the Chamber, representing a total sediment volume of 55.25 L. These samples were chosen because they were on the two sides of the cave, near its top and bottom, and therefore represent the whole stratigraphic column. The sediment samples were washed and sieved using an upper screen with a 2 mm mesh and a lower 0.5 mm mesh. The fossil remains were selected and grouped according to anatomical criteria. The dental elements were studied first because they are more diagnostic for taxonomical identification. The specimens were photographed and measured with a Nikon SMZ 1500 stereomicroscope at 10 × magnification in the Stratigraphy and Palaeontology Department at the University of the Basque Country (UPV-EHU). All recovered material will be storage in Gordailua [Cultural Heritage Center of Gipuzkoa, Irun (Spain)].

A bone sample was selected for an Accelerator Mass Spectrometry (AMS) radiocarbon determination to establish the exact age of the deposit. The bone chosen was a Rupricapra pyrenaica metapodial, because of its consistency and good state of preservation. It came from the Upper Ledge (Fig. 1C), which, a priori, is the most recent part of the deposit. The sample was sent to Beta Analytic (Florida) and it yielded the minimum weight of 1 to 2 g collagen. The sample proved to be outside the maximum age limit for Beta Analytic (43.5 ka), showing it was at least not referable to the later part of the Late Pleistocene.

As it was also impossible to date the deposit with the U/Th technique owing to the absence of speleothems within the deposit, the method of amino acid racemization (AAR) was applied in the Artazu VII site, the only dating method that can be used in these chronologies, as in Las Callejuelas (Domingo et al., 2015). The dating of palaeontological and archaeological sites using AAR analysis has become a reliable method, despite certain difficulties. The amount of amino acids in organisms with carbonate shells depends on diverse factors related to the depositional environment, taphonomical processes (mechanical fragmentation or bioerosion) and chemical dissolution (Carroll et al., 2003, Davies et al., 1989, Kidwell, 1998, Kidwell et al., 2005 and Meldahl et al., 1997). Moreover, different results for the same chronology can be explained by the absorption of solar radiation according to the orientation of the entrance of the site (Ortiz et al., 2015a). In addition, several studies have reported variations in the ratio of measured amino acids depending on the part of the organisms with carbonate shells from which the sample is recovered (Haugen and Sejrup, 1992, Ortiz et al., 2015b and Wehmiller, 1980). Furthermore, racemization analysis varies depending on the genus and species examined (Murray-Wallace, 1995), so it should be preferable to analyze samples of the same species, or at least, of the same genus.

In Artazu VII, to reduce taxonomically controlled variability in measured values, AAR was applied to 16 gastropods belonging to two genera (eight to Clausilia sp. and eight to Xerosecta sp.) and two Panthera pardus dental elements. Analyzed gastropods were extracted from Upper Ledge (spits A, C, G, K and L) and from Chamber (layers 1, 3 and 5) (Fig. 1), collecting the same two genera for each spit. In this way, it was possible to contrast two dating for each level. The first sample of panther was an upper left canine from Lower Ledge Level L (LEB-14069) and the second was a lower left first molar from Layer K (LEB-14070) in the Lower Ledge. All samples were sent to the Biomolecular Stratigraphy Laboratory (LEB) in the Higher School of Mining Engineers in Madrid (E.T.S.I de Minas). They were prepared according to the LEB protocol (Torres et al., 2014) and analysed in an Agilent HPLC-1100 high performance liquid chromatograph with a fluorescence detector, taking between 17 and 36.3 mg of each sample. The elements being analysed were subjected to hydrolysis and derivatization processes, using hydrochloric acid hydrolysis and an automatic injector with three mobile phases for the derivatization. The AAR results were obtained based on the D/L relationships for each amino acid identified and the total contents (pmol/mg) of each enantiomer (Torres et al., 2014).

The most of the taxa identified at the site still exist today, which means that the correlation between species and habitats is quite reliable. Four habitats have been defined in accordance with the ecological preferences of each species, after the studies of Berto (2013), Cuenca-Bescós et al. (2009), Pemán (1985), Pokines (1998) and Sesé (2005). These habitats are: woodland (scrubland and edges of forests with moderate cover), grassland (open environments with certain humidity, also with dense pastures and high plant cover), aquatic environments (any kind of surface water environment, either a watercourse or ponded water: streams, lakes, ponds, bogs and marshes) and crags (open areas with a rocky substrate, generally above woodland, but not necessarily a montane environment). However, this classification occasionally lacks clear boundaries as the transitions between them can be gradual.

The partial destruction of the site prevents the reconstruction of the morphology of the cave. However, the preliminary evidence suggests that Artazu VII was a karstic cavity that acted as a natural trap. All remains preserved in the site display no signs of biological action due to carnivore activity. Additionally, the absence of human activity evidences, such as cut marks by artefacts or intentional breakage features discarded the possibility of anthropogenic origin. Besides, many macrofaunal remains were even found in anatomical connection.

In contrast, normally, the microfauna is accumulated as part of the food of birds of prey. Accumulations of microfauna in caves tend to be the result of the action of predators, either from owl pellets or the scats of small carnivores (Andrews, 1990). This action causes physical (tooth marks) and/or chemical traces (corrosion by gastric juices) on the bones to a greater or lesser extent (Andrews, 1990). In Artazu VII, the microfaunal bones hardly display any evidence of digestion, so nocturnal owls were probably the potential agents producing the accumulation of small vertebrates in the sequence. Nevertheless, we cannot rule out the possible accidental incorporation of the microvertebrates to the site.

In addition, some of the remains exhibit post-depositional fractures, so that after the accumulation has formed, the remains are thought to have been disseminated by some kind of flow (water, mud…) that deposited them in the final part of the cave.

Artazu VII has yielded a large number of remains of great taxonomical diversity, as regards both the macrofauna and the microfauna, and consequently the detailed biometric and morphological study is still in progress. Because most of the macrofaunal remains were found whole or with post-depositional fractures, with many of them preserving anatomical connections, this excellent preservation allows the palaeobiological interpretation and general palaeoenvironmental reconstruction in the present study.

The microvertebrate assemblage consists of herpetofauna, avifauna and small mammals. Within the herpetofauna, amphibians are represented by five taxa (Salamandridae indet., Alytes obstetricans, Bufo bufo, Hyla arborea and Rana temporaria-iberica) and the reptiles by four (Lacertidae indet., Anguis fragilis, Coronella girondica and Vipera seoanei) (Fig. 2, Table 1). Among the amphibians, the most common specie is R. temporaria-iberica, while among the reptiles, the most numerous taxon is Lacertidae indet. R. temporaria-iberica and the Lacertidae indet. belong to families with species that nowadays live to the north of the Atlantic-Mediterranean watershed in the Basque Country, and inhabit areas with certain plant cover, mainly broadleaf and mixed forests, near almost permanent sources of water. The avifauna consists of seven species (Perdix perdix, Coturnix coturnix, Lyrurus tetrix, Bubo Bubo, Pica pica, Pyrrhocorax pyrrhocorax and P. graculus) (Table 1). The most abundant taxa remains are the two species of the genus Pyrrhocorax.

Within the small mammals, 13 species have been identified, belonging to the Orders Rodentia (Arvicola sapidus, A. amphibius, Microtus (Microtus) agrestis, M. (Microtus) arvalis, M. (Terricola) sp., Pliomys lenki and Apodemus sylvaticus-flavicollis) Eulipotyphla (Sorex [Sorex] araneus-coronatus, S. [Sorex] minutus, Neomys sp., Talpa sp. and Erinaceus europaeus) and Chiroptera (indeterminate Chiroptera) (Fig. 2, Table 1). The most abundant Orders are Rodentia and Eulipothypla, with the species A. sylvaticus-flavicollis, S. araneus-coronatus and S. minutus. The first appearance of P. lenki in the Iberian Peninsula was in the Early Pleistocene and it persists until the Early Holocene (Cuenca-Bescós et al., 2009). Some authors associate this rodent with a woodland environment (Pokines, 1998, Rofes et al., 2014 and Sesé, 2005). The group A. sylvaticus-flavicollis have also lived in Iberia since the Early Pleistocene (Maul, 1990), and are indicative of tree cover and characteristic of the temperate climate in northern Spain, as clearly shown by their Holocene presence at the sites of Santimamiñe (Rofes et al., 2014) and Peña Larga (Rofes et al., 2013) and the Late Pleistocene site of Antoliña Koba (Rofes et al., 2015). Finally, the group S. araneus-coronatus and the species S. minutus live in continental climates and are associated with very humid environments, preferably temperate, with dense plant cover (Cuenca-Bescós et al., 2009 and Rofes, 2009). It should be noted that no species associated with cold climates (e.g. Microtus [Alexandromys] oeconomus) have been found (Cuenca-Bescós et al., 2009) in Artazu VII.

Ungulates and carnivores form the large mammal assemblage: three ungulate species (Bison priscus, Rupicapra pyrenaica and Cervus elaphus), and eight carnivore taxa (Panthera spelaea, P. pardus, Lynx sp., Felis silvestris, Canis lupus, Vulpes vulpes, Mustela lutreola and Martes sp.) (Fig. 3, Table 1). The best represented species is Rupicapra pyrenaica, followed by P. pardus and C. elaphus. The taxonomical association of the large mammals suggests a complex environment consisting of a landscape with different habitats. The presence of a main sample of R. pyrenaica indicates the existence of areas of rocky mountains next to the site. The existence of forest near the site is justified by the presence of species such as the P. pardus, P. spelaea, Lynx sp. and F. silvestris. Besides, the presence of members of a steppe bison (B. priscus) and Panthera genera are consistent with some grassland in the surroundings, next to the forest assemblage. This landscape is complemented by the existence of freshwater currents nearby, as attested by the presence in the sample of species such as M. lutreola.

Given the material fell outside the range of radiocarbon dating, 18 AAR dates were obtained for other large mammal and gastropods remains. The results provided by the gastropods racemization are inconsistent for several reasons. On the one hand, in the same level, the gastropods AAR had a difference of about 30 ka. On the other hand, numerous datings are inverted and show non-consistent results, without any order. As was mentioned before, AAR in carbonate shells depends on many factors, and although it all necessary precautions were taken, the results are inconclusive. A similar observation has also been made in other works with AAR, like Penkman et al. (2008) and Ortiz et al. (2013). Ortiz et al. (2015a) noticed that ostracods racemization speed varies depending on the marine influence (pH, salinity, …). They also noted that the concentration of amino acid present in Patella Vulgate shells, varies within the same archaeological level (Ortiz et al., 2015a). The AAR results obtained from the panther samples are more consistent with AMS results. The one from Lower Ledge Layer L (LEB-14069) gave an age of 88,500 years, while the other, from Lower Ledge Layer K (LEB-14070) was dated to 98,400 years (Table 2). This chronological inversion has been interpreted as being influenced by atmospheric temperature inside the palaeontological site. Although these results do not help to know the exact age, they are valid to give an approximate age. Roughly, the mean value between both AAR datings is 93,000 years, so Lower Ledge (Fig. 1C) can be dated to the early Late Pleistocene.

Deep ice cores from central Greenland provide a long climate record going back to 122 ka, differentiating alternating cold and warm periods. The AAR dates and general environmental data from Artazu VII have been correlated with the δ ¹⁸O curve obtained by the North Greenland Ice Core Project, 2004 (NGRIP). According to ARR datings, the deposit of Artazu VII corresponds to the previous interglacial MIS 5 (Helmens, 2014). Whereas MIS 5b represents a period of cooling with an open environment and sub-Arctic and sub-Alpine steppe species (Helmens, 2014), MIS 5c is interpreted as a warmer period, in which broadleaf forests spread in a temperate climate. The transition from MIS 5d to MIS 5c is marked by the disappearance of the deciduous woodland and the spread of coniferous forests, indicative of cold, dry climates (Helmens, 2014). Artazu VII shows a relatively temperate climate, so it would correspond with substages 5c and/or 5b. Therefore, it is tentatively correlated to MIS 5c because of the large development of woodland.