A step back to move forward: a geological re‐evaluation of the El Castillo Cave Middle Palaeolithic lithostratigraphic units (Cantabria, northern Iberia)

El Castillo Cave is one of the most important sites for understanding the Middle and Upper Palaeolithic in Europe. Despite its importance, the absence of a widely used stratigraphic section with detailed lithostratigraphic descriptions and correlations between the different geological and archaeological interpretations has led to confusion in the correct identification of lithostratigraphic units in the lowermost, Middle Palaeolithic sequence. This study establishes a new lithostratigraphic framework for the site, which can be accurately correlated to previous geological and archaeological studies and generates a solid working basis for framing the Mousterian of El Castillo Cave in the Cantabrian region and southwestern Europe. The geological re‐evaluation of Unit XX (‘Mousterian Alpha’) has expanded its chronology, now ranging from 49 130–43 260 cal bp to 70 400 ± 9600 bp. Unit XXII (‘Mousterian Bet’) would consequently yield an age older than 70 400 ± 9600 bp and younger than the underlying speleothem (Unit XXIIIb), dated to 89 000 +11 000/‐10 000 bp.


Introduction
El Castillo Cave is one of the most important sites for understanding the Middle and Upper Palaeolithic in Europe, yielding 25 lithostratigraphic units with archaeological assemblages ranging from the Early Middle Palaeolithic to the Azilian (Cabrera Valdés, 1984;Cabrera Valdés et al., 2006) and an assortment of rock art (Alcalde del Río, 1906;Valladas et al., 1992;Pike et al., 2012). The cave is located in Puente Viesgo village in Cantabria (northern Iberia; Fig. 1).
The site was discovered in 1903 by Hermilio Alcalde del Río (Alcalde del Río, 1906) and was excavated by him alongside Hugo Obermaier and Henri Breuil between 1910and 1914(Breuil and Obermaier, 19121913;1914;Obermaier, 1924). After years scattered throughout several institutions, nowadays the majority of the archaeological material is hosted at the National Archaeological Museum (MAN) and at the Prehistory and Archaeology Museum of Cantabria (MUPAC). Despite the undoubted importance of these assemblages, the archaeological materials were studied decades later (Cabrera Valdés, 1979Valdés, , 1984Castaños, 2018).
The stratigraphy of the site, initially described by Breuil and Obermaier (Obermaier, 1914;1924), was expanded by Cabrera Valdés (1984) using Obermaier's excavation notes, and a numerical unit designation (1 being the youngest to 25 being the oldest) for the site was introduced. Archaeological and palaeontological remains hosted in the museums are organised and separated according to these designations.
Independently, Butzer (1981) carried out detailed sedimentological analyses of the El Castillo sequence, and similarly described and numerically ordered the stratigraphic levels (1 being the oldest unit, 19 being the youngest). Although Butzer (1981) produced a stratigraphic section of the Castillo sequence, it has not been used in archaeological studies because not all lithostratigraphic units could be precisely correlated to Obermaier's units and archaeological levels (Cabrera Valdés, 1984).
Excavations at El Castillo were resumed, led by Victoria Cabrera Valdés and Federico Bernaldo de Quirós, from 1980to 2017. Cabrera Valdés (1984, with careful inspection of Obermaier's field notes, was able to confidently relate her Unit 18 to Obermaier's 'Aurignacian Delta', with the visual aid of several large boulders and its stratigraphic position between two sterile units (Units 17 and 19; Cabrera . Since then, studies have been centred around the Middle-Upper Palaeolithic transition and the Mousterian (Cabrera Valdés and Bischoff, 1989;2001;2006;Bernaldo de Quirós and Maíllo-Fernández, 2009;Pike-Tay et al., 1999;Rink et al., 1997;Garralda et al., 2019;Wood et al., 2018, Jones et al., 2019, with less attention paid to the lowermost, Middle Palaeolithic sequence. Despite the success of Cabrera  in correctly identifying Unit 18, the absence of a widely used stratigraphic section, with detailed descriptions of levels and correlations between the different geological and archaeological interpretations, has led to some confusion in the correct identification of stratigraphic units in the lowermost Castillo sequence, as will be discussed below. Assigning archaeological assemblages to incorrect lithostratigraphic units  can not only foster an incorrect chronology, but can also be incorrectly related to certain sedimentary environments or processes which can deem the assemblage as not maintaining its in-locus properties. Additionally, if new materials are retrieved in recent excavations from incorrectly identified lithostratigraphic units and added to the historical collection (organised and separated according to Obermaier's levels), the resulting association would be an amalgamation of archaeological and palaeontological material belonging to completely different assemblages.
For this reason, the main objectives of this study are: 1) to produce a detailed stratigraphic section of the lowermost, Middle Palaeolithic sequence; 2) describe exhaustively each level, in order to facilitate their correct identification in future studies; 3) using this geological information, correlate the different archaeological and geological levels historically used in the Middle Palaeolithic sequence; and 4) to create a lithostratigraphic framework, taking into account all these data, to provide a reference for future archaeological, palaeontological and geological studies carried out in this important Palaeolithic site.

Materials and methods
The southern and western profiles, from the base of Cabrera Valdés (1984) Unit 18, were exposed and freshly scraped (Fig. 2). Lithostratigraphic units were described, measured, logged using scaled drawings and photographed. Level limits were measured and recorded with a Total Station. For clarity in distinguishing Cabrera Valdés (1984) units (using Arabic numerals) with the units here described, Roman numerals will be used.
Representative samples of each unit were collected for granulometrical analyses from freshly scraped exposures. Sieves sized -1-4 φ were used for measuring grain size distribution. Sediment types have been classified based on particle size distributions, following Blott and Pye (2012).
For detailed mineralogical analyses of the units, exposures were freshly scraped and~5 cm-thick unaltered samples were collected. Fifteen grams of each sample were homogenised by quartering and ground manually with an agate mortar until they could be sieved completely through a 53 µm metallic mesh. Bulk mineralogy powder X-ray diffraction (XRD) spectra were produced using a Philipps Analytical PW 1752 Cu Kα radiation X-ray diffractometer (graphite monochromator radiation Kα1 = 1540.6 nm). Diffraction spectra were recorded continuously at 2θ angles from 2°to 68°with 0.02 stepping intervals and 1 s per step. A semi-quantitative analysis was carried out with EVA software, following Chung (1975).

Results and discussion
El Castillo Cave stratigraphic section and profiles The exposed southern and western profiles (Fig. 2) below Cabrera Valdés (1984) Unit 18 revealed a~3.6 m sequence, containing a total of 16 lithostratigraphic levels. A detailed stratigraphic section of this lowermost, Middle Palaeolithic sequence (Units XXV to XIX) is shown in Fig. 3. Detailed stratigraphic profiles are shown in Figs 4-7 and discussed below.   El Castillo Cave stratigraphic units: descriptions, interpretations and correlations Unit XXV: The lowermost level of the sequence comprises massive, slightly clayey, sandy silts (Table 1; Fig. 8), with alternating~5-10 cm-thick beige (light beige/pinkish white 7.5YR 8/2 to pinkish grey 7.5YR 7/2) and brown (10YR 3/2) levels (Figs 3 and 4). Two <2 cm-thick,~2 m-wide, very dark grey (10YR 3/1) to very dark greyish brown (10YR 3/2) levels can be observed towards the right of the profile shown in Fig. 4. Mineralogically, quartz (41-53%) and clay minerals (21-27%) are most abundant, followed by significant quantities of calcite, plagioclase and K feldspar (Table 2). These deposits are interpreted as backswamp deposits (Gillieson, 1996;Bosch and White, 2018), deposited during, at least, five very low-energy events (Units XXVa-e; Fig. 4). These levels, most likely parallel to the underlying depositional surface (yet to be uncovered), with thicknesses ranging from 50 to 200 mm, show no internal sedimentary structures, suggesting that they were decanted from a homogeneous suspension that was steadily renewed with very low velocities (Ford and Williams, 2013). Towards the top of Unit XXVa, the abundance of limestone blocks, interpreted as éboulis, could reflect a major cold episode (Courty, 1989;Laville et al., 1980;Macphail and Goldberg, 1999).
Unit XXIV: Units XXV is overlain disconformably by Unit XXIV (Figs 3, 4 and 5). The lowermost~5 cm are composed of a gastropod shell conglomerate with a massive, sandy clayey silt matrix (Table 1; Fig. 8). The silt, yellowish brown (10YR 5/ 4) to dark yellowish brown (10YR 4/4), extends over the rest of the level (~70 cm). Quartz (37-50%) and clay minerals (22-34%) are the most abundant minerals, followed by significant quantities of calcite, aragonite, apatite, plagioclase and K feldspar (Table 2). Limestone blocks of 5-10 cm are abundant in the lowermost 20-30 cm of the level, with some scarce 15-20 cm blocks, all interpreted as éboulis. Towards the left of the profile, two large quartz blocks (70 × 40 cm and 60 × 30 cm) are found. Two more levels of the shell conglomerate can be observed in the lower half of the level, pinching out laterally. Abundant faunal remains are found embedded in this unit, including red deer (Cervus elaphus), large bovids (Bos/Bison) and cave bear (Ursus spelaeus; Castaños, 2018). Archaeological remains found in Unit XXIV correspond to the Early Middle Palaeolithic, and can be found in two archaeological levels. The uppermost section of the unit (30-40 cm), which is more clayey and where charcoal fragments are common, contains the upper archaeological level, which is characterised by discoid cores and Levallois type blanks with type 0 cleavers and bifaces (Solano-Megías, 2016), and can be correlated to Obermaier (1924) and Cabrera (1984) 'Acheulean' level. The lower section of Unit XXIV contains an archaeological assemblage similar to the upper level, but with a lower abundance of Levallois types (Bilbao-Malavé, 2016).
The massive silts comprising Unit XXIV are interpreted as backswamp facies (Bosch and White, 2018), which include three more energetic events which deposited the gastropod shell conglomerate levels. As in Unit XXVa, the presence of abundant éboulis is indicative of cold conditions during the deposition of the level (Courty, 1989;Laville et al., 1980;Macphail and Goldberg, 1999).
Unit XXIV can be confidently related to Butzer's (1981) Level 4 (Table 3; Figs. 9 and 10), described as an exceptionally  148.00 (Continued ) poorly sorted brown (7.5YR) loam with coarse-to block-grade rubble and archaic bifaces. Unit XXIV can also be correlated to Obermaier's level 'b) Acheulean' (Obermaier, 1924) and Units 24 and 25, described as brown silts and clays with charcoal fragments (Cabrera Valdés, 1984; Fig. 10). Unit XXIII: Unit XXIII is subdivided into three units (Figs 3 and 5). Unit XXIIIc is a~10 cm-thick unit of a dark yellowish brown (10YR 4/4) slightly clayey sandy silt (Table 1; Fig. 8) with abundant fossil remains and some scarce limestone blocks, with a channel-like incision into Unit XXIV in the centre of the southern profile, and overlies it conformably throughout the rest of the site (Fig. 5). The incised section contains 10-15 cm blocks. Although the cross-section is channel-like in the incised section, textural characteristics (massive deposit, matrix-supported, chaotic, unbedded and unsorted), point to a diamicton cave deposit (debris flow; Bosch and White, 2018). Mineralogically, Unit XXIII is very similar to Units XXIV and XXV, with quartz (43%) and clay minerals (27%) being the most abundant minerals, with significant quantities of calcite, aragonite, apatite, plagioclase and K feldspar ( Table 2).
Unit XXII contains both archaeological (Mousterian) and palaeontological remains. Discoid knapping methods on quartzite and flint are common in this level, which includes abundant sidescrapers, denticulates and notches (Cabrera Valdés, 1983, 1984. The faunal assemblage is dominated by Equus caballus, followed by Cervus elaphus and Bos/Bison remains (Castaños, 2018).
Unit XXI: Unit XXI is a sterile, very pure yellow (10YR 8/6) sandy clayey silt marl (Table 1; Fig. 8), overlying Unit XXII conformably with a very discrete contact (Fig. 5) contains some phosphate precipitation. The upper half of the unit contains very few 5-10 cm blocks, and abundant 2-4 cm blocks (mostly limestone, but some quartz; Figs 5 and 6).
All Unit XX deposits contain abundant archaeological (Mousterian) and palaeontological remains. Archaeological remains consist of sidescrapers, denticulates, notches and cleavers, the majority on quartzite (Cabrera Valdés et al.,

Conclusions
The new, detailed stratigraphic section, alongside the detailed descriptions and correlations of the stratigraphical levels, will hopefully facilitate their correct identification in future studies, avoiding any possible confusion. The proposed level correlations will allow future studies to incorporate and combine data from the different archaeological and geological levels historically used in the Middle Palaeolithic sequence of El Castillo Cave. This study therefore allows accurate correlation of new archaeological finds with those stored in the museums, belonging to the collections from the Obermaier excavations in the early twentieth century, and those carried out later, in the 1980s, by Victoria Cabrera Valdés and Federico Bernaldo de Quirós.
Finally, from an archaeological point of view, it is noteworthy that this study entails an important change in the archaeological conclusions of the Mousterian since the work of Rink et al. (1996). Previous studies analysing Units 20, 21 and 22 all refer, after our study, to different subunits of Unit XX. Although this does not substantially change archaeological interpretations of previous Mousterian techno-typological studies, it does change their stratigraphic position and overall geological interpretation.
In conclusion, this study establishes a new lithostratigraphic framework for the site, which can be accurately correlated to previous geological and archaeological studies and generates a solid working basis for framing the Mousterian of El Castillo Cave in the Cantabrian region and southwestern Europe.