Revision of TD1 and TD2 stratigraphic sequence of Gran Dolina cave (Sierra de Atapuerca, Spain)

Gran Dolina is a cavity infilled by at least 25 m of Pleistocene sediments divided into 12 lithostratigraphic units and 19 sedimentary facies. These sedimentary facies have been divided into allochthonous facies, defined as sediment inputs from the outside, and autochthonous facies, defined as sediments generated within the karst; but this division has been challenged in recent works. In this study, TD1 and TD2 units of Gran Dolina have been detailed studied and the use of autochthonous facies has been assessed. For that purpose, we have studied the stratigraphic excavation profile, combining field observation with laboratory sedimentary analysis (sieving, laser diffraction, and XRD) to characterize the texture and structure of the sediments. Based on these studies, a total of 8 sedimentary facies have been identified. Consequently, TD1 unit has been separated into two sub-units and 13 layers, while the TD2 unit has been divided into three sub-units. The facies associations indicate a succession of phreatic and vadose phases that would define together epiphreatic conditions inside the cave, related to the transition between Arlanzón valley terraces T3 and T4. Interior facies (and entrance facies for allochthonous facies) is proposed to define Gran Dolina's sediments since the facies analyses indicate transport by underground flows.


Introduction
Caves are natural traps that can preserve sediments and archaeological remains from outside erosion. Sedimentary infills have been well studied in many caves since the end of the XIX century until now (Arriolabengoa et al., 2015;Bonifay, 1956;Bull, 1981;Cooke, 1938;Farrand et al., 2001;Kadlec et al., 2008;Kampolis et al., 2022;Martini et al., 2021;Ossowski, 1882;Shaw, 1992, among others), providing data about the processes of formation and the relationship between the caves and their environment. Classification of cave deposits traditionally divides cave sedimentary facies into three main groups: allochthonous, autochthonous, and chemical deposits (Ford & Williams, 2007;White, 2007).
Allochthonous sediments are sediment inputs from the outside. These sediments enter the cave from near entrances and can accumulate through different sedimentary processes such as gravity (e.g. debris fall, debris flow, mudflow), fluvial (floodplain and channels), or aeolian agents. In contrast, autochthonous sediments are generated within the karst, resulting from cave processes such as weathering or breakdown. These kinds of sediment are usually produced when the cave is closed to the outside, preventing autochthonous sediment accumulation. Although, at some times, autochthonous facies can produce in open caves or rock shelters, as breakdowns (Martini et al., 2018). Chemical deposits are mineral growths that are produced inside the caves. The most known are calcite speleothem, but also, they can be evaporites, phosphates, or even ice. These usually are very slow growth deposits that need a closed environment.
During the excavation of 2013, a vertical test pit in the NW section was started at the base of the Gran Dolina site (Fig. 3) to reach the cave's floor. It was reached at seven meters depth, showing a total of nine meters of the sedimentary sequence of sterile sediments mainly composed of sand, silt, and clay. After, a borehole was drilled in the Railway Trench next to the Gran Dolina site revealing 10-m thick sediment fill composed of silts and clays (Bermejo et al., 2020). These sediments were included in the TD1 and TD2 units of Gran Dolina, defined as autochthonous facies  following the definition of White (2007) and Ford and Williams (2007), and previous works in Sierra de Atapuerca sites (Pérez-González et al., 1995. The assignment of these sediments to autochthonous facies was due to comparison with the clastic sediments of Gran Dolina, defined as allochthonous facies, and because the Sierra de Atapuerca karst was less understood. The use of this terminology has been challenged since a possible fluvial origin for TD1 and TD2 has been previously postulated by different studies based on sedimentology (Campaña,  1 3 2018), particle shape analyses (Campaña et al., 2016), and geochronology (Duval et al., 2022;Parés et al., 2018). Although TD1 and TD2 are sterile lithostratigraphic units, their study is necessary to understand the evolution of the karst system, providing key information about the formation and sedimentation processes in the cavities during the persistence of water table conditions related to fluvial phreatic levels. The formation of the middle level of the karst, where Gran Dolina cave belongs, was related to the terrace T3 of the Arlanzón River , which was dated by ESR to 1.14 ± 0.13 Ma (Moreno et al., 2012). Recently, TD1 and TD2 have been dated by Electron Spin Resonance (ESR) and single-grain thermally transferred optically stimulated luminescence (SG TT-OSL) methods applied to quartz grains, in combination with magnetostratigraphy, constraining these units to between ~ 1.4 Ma and ~ 0.9 Ma (Duval et al., 2022;Parés et al., 2018). This chronology means that the sedimentary deposit of Gran Dolina cavity started quickly after its formation. Therefore the study of these sediments can indicate the evolution of the environment during the Early Pleistocene, at a time when other caves in the karstic system already had sediments from outside, such as Sima del Elefante , and the oldest hominins from West Europe were living in the Sierra de Atapuerca .
This work aims to understand the cave dynamic and hydric regime within Gran Dolina cave during the period of sedimentation of TD1 and TD2, as well as its relationship with the rest of the karst and the hydrographic basin. To reach this aim, the first detailed description and interpretation of the TD1 and TD2 stratigraphic units is needed, with particular emphasis on sedimentary processes and sediment provenance.

Geologic context
The Gran Dolina site is situated in the south of the Sierra de Atapuerca in central Spain (N42° 21′ 05.29ʺ; W3° 30′ 39.20ʺ; WGS84) ( Fig. 1). The Sierra de Atapuerca is a NNW-SSE trending anticlinal ridge belonging to the most north-western outcrop of the Iberian Chain and is situated in the NE Neogene Duero Basin . This anticlinal ridge was folded during the Oligocene-Lower Miocene, and it is mainly formed by late Cretaceous limestones and dolostones. This Vertical test pit at the base of Gran Dolina cave, where the TD1 and TD2 section was excavated. A Image of TD1 and TD2 section. B Top view of 3D model of TD1 and TD2 made by photogrammetry deformation caused the deposit of syn-orogenic conglomerates, sandstones, and mudstones.
In the Late Miocene-Pliocene, the Duero Basin was opened to the Atlantic Ocean, finalizing the endorheic phase and starting the incision fluvial network of the sediments. During the Quaternary, the main fluvial courses around the Sierra de Atapuerca is the Arlanzón River, and its tributaries the Vena and Pico rivers. The studies of the evolution of these rivers identified a terrace sequence of 14 levels Benito-Calvo et al., 2008;Zazo et al., 1987) that was dated from the Early Pleistocene to the Holocene . Terraces T1 (+ 92-97 m) to T4 (+ 60-65 m) have been constrained to the Early Pleistocene, while T5 (+ 50-58 m) to T11 (+ 12-14 m) and T12 (+ 10-11 m) to T13 (+ 5 m) have been correlated to the Middle and Late Pleistocene, respectively. Finally, T14 (+ 2-3 m) was tentatively attributed to the Holocene (Benito-Calvo & Pérez-González, 2015; Moreno et al., 2012). The chronologies suggest a putative correlation between terrace deposits and cold Marine Isotope Stages and have allowed reconstruction of the incision rate evolution of the main valley , 2018.
In the Early and Middle Pleistocene, a multilevel endokarst system developed in the Late Cretaceous limestones and dolostones. This karst system consists of 4.7 km of explored passages (Martín-Merino et al., 1981), composed mainly of three levels of subhorizontal water-table controlled galleries (Ortega, 2009;Ortega et al., 2013). These levels appear spatially and chronologically related to some of the Arlanzón fluvial terraces Benito-Calvo et al., 2018;Ortega et al., 2013Ortega et al., , 2014Parés et al., 2015). The karst passages developed during the relatively short period corresponding to terrace aggradation, while vadose entrenchments are related to prolonged incision phases of the Arlanzón River . During the formation of the karst, the Arlanzón waters entered the galleries; they passed through the interior of the karst and left at springs in the Pico Valley headwaters .
The upper level developed in a position similar to the base levels associated with the Lower Páramo limestones and the fluvial terrace T2 (+ 82-86 m) , which occur at a similar elevation near the Sierra de Atapuerca . It is more than 600 m long with ceilings at 1015-1022 m in altitude, with some vertical chimneys reaching 1030 m in altitude . This level is composed by Galería del Sílex, El Portalón, Salón del Coro and Galería de las Estatuas (Fig. 1). The Salón del Coro is the biggest cavity of the Sierra de Atapuerca karst since it includes the three levels. The sediments in this level are mostly composed of clay and silt (Ortega, 2009).
The middle level, where the Gran Dolina cavity is located, is a sinuous subhorizontal phreatic passage about 700 m long and about 1000-1005 m in altitude, associated with the period of stability represented by terrace T3 (+ 70-78 m) , during the Early Pleistocene (Benito-Calvo et al., 2008Moreno et al., 2012). The vadose regime in this level was marked by incision of the Arlanzón River between T3 and T4 (+ 60-65 m) Ortega et al., 2013). Currently, this level has a subterranean zone formed by Sala de Los Cíclopes, Galería del Silo and Galería Baja (Fig. 1), in addition to other sediment-filled galleries recently discovered (Bermejo, 2021;Bermejo et al., 2017), and another aerial zone cut by a Railway Trench that revealed numerous sedimentary fills and cavities, including three important archaeological sites: Sima del Elefante cave, Galeria-Tres Simas cave and Gran Dolina cave . These three caves are sediment-fill caves that preserve important archaeological and paleontological remains (Rodríguez et al., 2011).
The Galería Complex is situated in the south of Gran Dolina and is composed of three sub-sections: Covacha de los Zarpazos to the north, Galería in the centre, and Tres Simas in the south (Ortega, 2009). Covacha de los Zarpazos is about a 15 m length cavity with direction northeast that continues to Galería at the south. Galería is the main subhorizontal conduit of the site. Tres Simas is three vertical shafts connected in their lower part with Galería. The stratigraphic sequence of Galería Complex has 13 m of thickness in the current section, which continues with at least 17 m under the current path (Bermejo et al., 2020); thus the cave has more than 30 m thick of sediments. The Galería sediments have been separated into five litho-stratigraphic units, from the bottom to the top, GI to GV units (Pérez-González et al., 1995, 2001. Matuyama/Brunhes paleomagnetism boundary is observed near the top of the GI unit , which is formed by similar facies of TD1 and TD2. Meanwhile, the luminescence chronology of GIV shows results of 0.185 ± 0.026 Myr (Berger et al., 2008) and 0.255 ± 0.021 Myr (Demuro et al., 2014).
The Sima del Elefante is a 25 m thick cave sedimentary infill divided into 16 litho-stratigraphic units, called TE7 to TE21, from bottom to top (Huguet et al., 2017;Rosas et al., 2001Rosas et al., , 2006. These units were grouped into three sedimentary phases: lower, middle, and upper phases. Sediments composing the Sima del Elefante site are all entrance facies formed by mudflows and debris flows. The lower phase has been dated by cosmogenic to 1.13 ± 0.18 Myr for TE7, and 1.22 ± 0.16 Myr for TE9 . In this site, the Matuyama/Brunhes paleomagnetism boundary is described between TE16 and TE17 units  according to the faunal associations, TL-IRSL, and cosmogenic dating . The lower cave level has been correlated to T4 (+ 60-65 m)/T5 (+ 50-54 m) . Their ceilings are situated at 990 m.a.s.l with vertical chimneys reaching 994 m in altitude . This level is located further west than the upper and middle levels and is formed by Cueva del Silo, Cueva Peluda, Cueva del Compresor and Sala del Caos (Fig. 1). These conducts have a phreatic morphology and are smaller than the cave of the other levels .
In the lower level's south, a small karst sub-level was developed. To this sub-level belongs Sima de los Huesos and the lowest area of Cueva del Silo, which has ceilings below 985 m.a.s.l (Ortega, 2009).

Sedimentological studies
The study and classification of the different facies and sedimentary environments of Gran Dolina required a detailed description of the available stratigraphic excavation profiles. Fieldwork has been combined with laboratory analyses to describe the mineralogy and texture of the sediments. For this, eight samples were taken from non-cemented layers in the stratigraphic section.
Particle size sieving and laser diffraction techniques have been used. For sieving techniques, φ size sieves, ranging from − 1φ to 4φ, were used (Geology Laboratory, CENIEH). Larger sizes have not been analyzed because a large amount of sample would have been required to obtain a representative analysis. A Beckman Coulter LS13 320 laser diffraction particle size analyzer was used to measure the silt and clay fraction particle size. Particle size has been classified following Blott and Pye (2012).
The survey of the stratigraphic profiles and facies maps of the sections was performed using 3D laser scanning techniques (Leica C10), total stations, and photogrammetry (Digital Mapping and 3D Analysis Laboratory, CENIEH), which have served to obtain a 3D model of textures with an RGB image of Gran Dolina and the test pit. Ortho-image of TD1 and TD2 was extracted from the 3D model to study the stratigraphic section.

Mineralogical and chemical analyses
Mineralogical and chemical composition of the bulk samples were obtained by combining two techniques at the Archaeometry laboratory of the CENIEH. The mineralogical phase composition was determined by bulk mineralogy powder X-ray diffractogram (XRD) using a PANalytical X'Pert PRO instrument equipped with a Cu target and a secondary monochromator. The operating conditions for XRD were 45 kV/40 mA in a continuous scan mode performed in the range of 2θ from 3° to 70° with an increment of 0.02°. Semiquantitative analysis was carried out according to the Chung method (Chung, 1975) by using High Score Plus software, based on the Reference Intensity Ratios (RIR) of the existing phases.
Chemical compositions of the major elements were obtained by wavelength dispersive X-ray fluorescence method (XRF) using a PANalytical Axios instrument. For each sample, 0.5 g of bulk grounded material was homogeneously mixed with 5 g of 66:34 mixture of Li 2 B 4 O 7 and LiBO 2 flux with LiBr as bead releasing agent. The mixture was then fused to a glass bead in a Pt-Au crucible with a PANalytical Perl'X3 automatic fusion machine. The following elements were measured: SiO 2 , Al 2 O 3 , Fe 2 O 3 total, MnO, MgO, CaO, Na 2 O, K 2 O, TiO 2 , P 2 O 5 , and SO 3 . Loss on ignition (LOI) was calculated too. . TD1 is the lowermost stratigraphic unit lying on the cave floor and topped by a speleothem marking a sedimentary discontinuity with the upper TD2 unit (Fig. 4). The upper discontinuity of TD2 is another speleothem growth, separating this unit from the red clastic sediment of TD4 unit. Based on fieldwork description and textural analyses, TD1 may be divided into two sub-units, comprising thirteen layers, while TD2 consists of three sub-units (Tables 2 and 3

Textural analyses
Textural analyses were carried out on uncemented layers of TD1 and TD2 (Fig. 5). In general, the layers of TD1 and TD2 are composed of millimetric laminations that could not be distinguished during bulk sampling. Therefore, particle size analyses result from both kinds of laminates, a darker laminate and a lighter laminate.

Mineralogy and chemical analyses
The layers of TD1 and TD2 show a very similar composition, either mineralogy or elemental composition (Tables 4  and 5). The values of the mineralogy are semiquantitative. Quartz, phyllosilicates, and calcite mainly dominate their composition. Quartz seems to be the main mineral of the samples, except in TD1.1.4, which is primarily composed of calcite. This can be due to the high proportion of gravels, either limestone clasts or cemented nodules, in this specific layer.
The elemental analysis shows a clear relation with the mineralogy, with Si as the principal element, except in TD1.1.4, where Ca is the majority due to its richness in calcite. TD1.1.8 has higher values of Al, Fe, and K than the other layers, which can explain by its clay content, which is significantly higher than in the rest of the layers (Fig. 5).

Sedimentary facies
The study of the stratigraphic section of TD1 and TD2 has allowed making a new classification that extends the sedimentary facies classification of Campaña et al., 2017 ( Table 1). The layers of TD1 and TD2 have been classified into sedimentary facies based on field observation, particle size analyses, and previous cave sediment classifications Ford & Williams, 2007;White, 2007). Eight facies have been differentiated and grouped into five main sedimentary facies (Table 6, Figs. 4 and 6).

Facies A
This facies is found at the top of TD1 and TD2, and the bottom of TD1. It consists of millimetric laminated yellowish brown sand and silt with rip-up clasts of silt and clay. The lamination is horizontal, although it is found slightly folded. It is mainly composed of well-sorted quartz, with a few phyllosilicates. No vertical particle size variation is observed in these facies. This facies presents sharp base and top boundaries. Rip-up clasts of other lithology, like clay and silt, are observed, in addition to some layers partially cemented. Depending on its particle size and cementation, three sub-facies have been separated.

Facies A1
Millimetric laminated 10YR 6/4 light yellowish-brown silty sand alternating dark and light laminates. It is composed of more than 50% sand and about 25% silt. This facies is observed in TD2.2 and TD1.2.4, although the last shows a darker Hue.

Facies A2
Millimetric laminated 10YR 6/4 light yellowish brown sandy silt alternating dark and light laminates. It is composed of more than 45% of silt and about 20% of sand. This facies is observed in TD1.1.2 and TD1.2.2.

Facies C2
Massive 10YR 5/6 yellowish-brown silty sands in TD1.2.3. This layer wedges through the West, resting its east side on the wall of the host rock. No internal sedimentary structures are observed.

Facies E
Breakdown facies are characterized by large and angular boulders breaking off caves' ceiling and walls. This facies appears in the TD2.3 layer, where boulders larger than 50 cm are observed (Fig. 6E). The rock falls have caused sediment deformation in the underlying layers, as in the TD1.1 subunit, where laminated layers are folded because of the TD2.3 breakdown.

Soft-sediment deformation structures
The soft-sediment deformation structures are defined as "deformations that occur in still unlithified sediments or in sedimentary rocks that had not yet undergone lithification before the deformation structures started to be formed." (Van Loon, 2009). In TD1 and TD2, two kinds of soft-sediment deformation have been described (Campaña et al., 2015). On the one hand, the layers of the section show different dips (Fig. 4) through the south-west, TD1.2 layers have a dip of 30°, TD1.1 layers have a dip of 12°, that to the south-west increased to 33°, and finally TD2 are sub-horizontal with a dip of 15° to the south-west (Campaña et al., 2015). On the other hand, the breakdown of TD2.3 caused the folding of the underlying layers as TD1.1.2 and TD1.1.3. This deformation is formed by rounded, low-amplitude anticlines and synclines that conform to the shape of the breakdown.

Comparison with previous works
In Campaña et al. (2017), seven autochthonous facies were differentiated in Gran Dolina (Table 1). Five were described Autochthonous fluvial facies were defined by Campaña et al. (2017) as "sediments deposited by the action of a more or less constant stream of groundwater that selects and concentrates particles of similar size. It is formed by well-sorted sands, which display a unimodal particle size distribution and usually include a very high percentage of quartz grains". These fluvial facies were differentiated into three sub-facies: facies A (FLA) as laminated sandy silt with soft nodules, facies B (FLB) as laminated clayey silt with cemented layers, and facies C (FLC) as sands and clays. The facies include in the current work include these facies. Thus, FLA is separated in facies A1 and facies A2, according to their particle size. FLB is the same that facies A3 since they have the same textural and fabric description. Finally, FLC includes sands and clays that in the current work have been separated in two different facies: sands are included in facies A1, while clays have been distinguished as facies B, defined as millimetric laminated clayed silt after particle size analysis.

Interpretation of the facies and depositional environments of TD1 and TD2
Facies A, defined as millimetric laminated yellowish brown sand and silt with rip-up clasts of silt and clay, shows similar features to other cave facies described in the bibliography, such as the channel facies described in the Mammoth Cave by Bosch and White (2004), facies Sh in Lapa Doce cave system (Laureano et al., 2016) or facies Laminated Sand in the caves of the Cheat River (Springer & Kite, 1997a, 1997b. These facies are composed of sand and silty sands with horizontal lamination and are interpreted as the result of a fluvial flow. The rip-up clasts indicate some energy in the hydric flow that was capable of eroding previous sedimentary deposits since these rip-up clasts are formed by silt and clay, and the energy needed to erode these cohesive sediments is similar to the required to erode and transport gravels (Herman et al., 2012;Sundborg, 1956). The lack of gravels in the deposit can be due to the lack of them in the source area or that the fissures were too small to allow the entry of them. This facies has three sub-facies. The facies A1 and A2 are very similar, differing only in that the main particle size component in facies A1 is sand, meanwhile in facies A2 is silt. This suggests that facies A2 represents a period with less energy than facies A1. Facies A3 shows cemented layers intercalated between the silty sand layers. These cemented layers, with some speleothem growth, indicate cycling cease of the hydric flow. Although some of these cemented layers can be post-depositional, the speleothem growths, that only appears in sub-aerial conditions, suggest that the drop of the water table caused the cementation. This major cycle contains other minor cycles that formed the lamination of the layers. Therefore, this facies indicates a period with constant changes in the hydric flow, with cycling variations in the energy and numerous hiatuses. Red mud facies are very common in caves (Bull, 1981;Springer & Kite, 1997a, 1997bAuler et al., 2009;Martini, 2011;Iacoviello & Martini, 2012;Laureano et al.,;Martín-Perea et al., 2022). Facies B, defined as reddish brown millimetric laminated clay and silt, can be considered like these facies since it shows similar sedimentological features as lamination, color and particle size. These deposits have been explained as slackwater deposits that have been interpreted as the result of sequential flooding events that can build up very fine-scale layering (Bull, 1981;Herman et al., 2012), in some cases related to underground lakes (Martini, 2011).
The facies C1 do not show any internal structure or lamination, but the rip-up clasts observed in this facies indicates some energy in the flow and, therefore, some transport of the sediment. As in facies A, the rip-up clasts are formed by silt and clay that surely were eroded from older layers from inside the karst. The lack of sedimentary structures can indicate that this facies could be deposited in a single event.
As discussed above, facies similar to facies A, B, and C1 have been described by other authors as deposits from hydric flows associated with phreatic conditions (Bull, 1981;Springer & Kite, 1997a, 1997bLauretano et al., 2016;Kampolis et al., 2022). These facies are indicators of a more or less constant hydric regime inside the cave, which can be related to hydric processes at a regional level (Hunt et al., 2010;Wagner et al., 2011). This suggests that, at the time of the deposition of these facies, Gran Dolina was partially flooded below the water table (Springer & Kite, 1997a, 1997b. In this sense, facies A, B, and C1 would be considered as phreatic facies.
On the other side, the features of facies C2, a wall-supported massive silty sand deposit, suggest that its sedimentation could be related to runoff water from the cave wall that gradually accumulates it by accretion. It can be similar to backswamp facies described by Bosch and White (2004) that consists of "weathering residue of the bedrock and infiltrate material filtering into the conduit system from overlying soils with little or no lateral transport" (Bosch & White, 2004) and the FO facies in La Cala cave and Oscurusciuto rock shelter described as sandy silt without sedimentary structure that was deposited due to infiltration processes from overlying soils and sediments (Martini et al., 2018(Martini et al., , 2021. In this case, the small amount of sand in the residue of the bedrock  indicates that this deposit is the result of the soil washdown, and, therefore, a slow sedimentation rate would be expected. In addition, the dip and geometry of the layer, with a wedge shape, could suggest that there was not any water table at the moment of the sedimentation. The formation of speleothems (facies D) implies subaerial conditions, being an indicator of vadose conditions (Springer & Kite, 1997a, 1997b. During the growth of a speleothem, other sedimentary deposits should not occur; it means that a speleothem indicates a hiatus in the stratigraphic section (Gillieson, 1986). The speleothems of the section of TD1 and TD2 are found near the East wall. Vertically, they are located in the same line, suggesting a drip point in this site that was active during most TD1 and TD2 sedimentation.
Breakdown deposits (facies E) are very common deposits in caves (e.g. Gillieson, 1996;Martini et al., 2018;White, 2007). Causes of collapse can be because of undermining of the underlying support, loss of hydrostatic pressure caused by the transition from phreatic to vadose conditions, cryoclastism, secondary mineral wedging, or earthquakes (Sasowsky, 2007). The deformation of the underlying layers in TD1 suggests the fall of the rocks from a high height, perhaps the ceiling of the cave.
According to the previously described features of the sedimentary facies, it can be considered that TD1 and TD2 represent a period of repetitive variations in the water table level, with fluctuations between phreatic conditions, indicated by the deposits of facies A, B, and C1; and vadose conditions, indicated by cementation and facies C2, D and E. These fluctuations can indicate that Gran Dolina was in the epiphreatic zone, but on the one hand, the formation of speleothem needs stable vadose conditions during long time (Ford & Williams, 2007), and in the other hand, the continuous lamination of facies A and facies B in sub-units with more than one meter of thick suggests stable phreatic conditions (Gillieson, 1996;Springer & Kite, 1997a, 1997b. Therefore, the epiphreatic conditions have been used only in deposits where the changes in the conditions were clearly fast, as putatively in the facies A3 (Fig. 3), being not possible to differentiate precisely the phreatic and vadose conditions. These deposits can generally be considered as the result of an underground fluvial flow that transports sediments from farther sources than entrance facies. Furthermore, the fact that in certain cementations, a speleothem develops, as in layers TD1.1.6 and TD1.1.8, suggests that the cementations were produced by sedimentary stops and aerial exposure. Therefore, the cementations found in the TD1 and TD2 sequence would indicate vadose conditions (Springer & Kite, 1997a, 1997b. Based on the last dating study of TD1, which inferred an age of about 1.4 Ma for the base of this deposit (Duval et al., 2022), the sedimentation of Gran Dolina cave began shortly after its vadose incision. TD1.2 represents the first registered sedimentation of Gran Dolina cave; it starts with the layer TD1.2.4 composed of laminated silty sand (facies A1) dipping towards the West. At the top of this layer, the lamination is truncated by an erosive surface, above which TD1.2.3 is deposited. This layer is accretion sediment that, with the erosive surface, suggests a stop in the phreatic sedimentation in the sequence that would indicate the first registered vadose conditions in Gran Dolina cave after the vadose incision of the cave (Fig. 7). The sedimentation is reactivated with laminated sandy silt (facies A2) of TD1.2.2 which suggests the return of phreatic conditions in the cave, but with lesser energy than in TD1.2.4. Finally, at the top of TD1.2, the cementation of TD1.2.1 would indicate vadose conditions and another hiatus in the sequence.
The TD1.1 sub-unit shows an important change in the dynamic of the sedimentation. From now on, the sedimentation is dominated by centimeter-thick or decimeter-thick cycles of laminated silty sand, in some cases cemented (facies A1 and A3); and laminated clayed silt (facies B), within each of which a millimetric cyclical lamination can be observed (Figs. 4 and 7). In addition, there are at least four speleothems and a level of facies C1. All these levels indicate a moment with successive hydric changes from decreased energy when the clays of facies B are deposited; to increased energy, when the silts of facies A are deposited. The end of some cycles with speleothem would indicate the drop of the water table and vadose conditions (Springer & Kite, 1997a, 1997b. It is especially remarkable in TD1.1.4 where two layers of speleothem are observed (Fig. 4). The layer TD1.1.7, massive silty sands with rip-up clasts (facies C1), indicates a break in these cycles, with the reworking of previous deposits of the cave. This can be due to local changes inside the karst system, such as variations in the course of the water flow and reworking of near deposits.
These cycles thicken towards the top until TD1.1.4 and TD1.1.3, that has about 1.5 m thick. This increase in thickness may be related to temporal increases in the environmental cycles that regulate this sedimentation and more stable conditions than in the underlying cycles.
Furthermore, TD1.1.2 shows an alternating of facies A2 and facies B that suggests a relative water table rise and stable phreatic conditions in Gran Dolina cave, although with a difference in the energy flow. This layer presents partial cementation, but as they are only observed in facies B, namely, in clayey silt layers, and no speleothem growths are observed in these layers, it is preliminary interpreted as by post-depositional cementation of the most clayey layers. Finally, the water level falls again, allowing the speleothem growth at the top (TD1.1.1).
The TD2 unit starts with a breakdown (facies E) of more than 1-m diameter boulders that caused deformation in the layers of TD1.1 (Fig. 6E). These boulders collapsed from the walls and the ceiling of the cave. Later, TD2.2 is deposited with a similar pattern to TD1.1.2, an alternating of facies A1 and facies B. This indicates the last phreatic condition in the cave (Fig. 7), with substantial energy flow fluctuations. The particle size of facies A1 suggests more energy in TD2.2 than in TD1.1.2. Finally, the TD2 unit ends sealed by more than 40 cm thick speleothem which indicates a relatively long time before the entry of the entrance facies of TD3 and TD4 . The variations of the TD1 and TD2 sediments indicate that, in a broad temporal sense, Gran Dolina would be in the epiphreatic zone of the hydric regime, with the base level oscillating about the altitude of the cavity, but never completely flooding it. These oscillations would reflect the changes in the hydrogeological basin of the Arlanzón River since karst systems are more sensitive to changes in the water table: oscillations of centimeters outside can imply metric rises or falls inside karst (Ford & Williams, 2007). These variations should be related to the transition between the terraces T3 and T4, which is related to the vadose incision in Gran Dolina cave . The variations between vadose and phreatic conditions observed in the TD1 and TD2 stratigraphic sequence suggest that this transition had, at least, ten stages (Fig. 7). The TD2.1 speleothem would mark the final phreatic influence and deposition of interior facies in Gran Dolina cave related to the T3-T4 transition, which would finish with the final construction of the terrace T4 in the hydrogeological basin of the Arlanzón River. This terrace has an estimated age of about 0.85 Ma Moreno et al., 2012) which is consistent with the proposed dates for the end of TD2 (about 0. 89 Ma; Duval et al., 2022). This stratigraphic section mainly matches with the Railway Trench borehole (BH2 in Bermejo et al., 2020), composed of fine clays and fine sands. The limitation of the mechanical drilling is that it avoids observing the sedimentary details in the borehole and prevents a direct correlation between it and the main section. Nonetheless, this borehole and the GPR survey demonstrate that Gran Dolina cave continues to the west (Bermejo et al., 2020), and, therefore, the flow direction was from east to west.

Sedimentation rate and paleoclimate
TD1 and TD2 units are proposed to develop during about 400 kyr, according to TT-OSL, ESR, and palaeomagnetic dating results (Duval et al., 2022). In TD1, three magnetic excursions have been described, tentatively identified as Cobb Mountain (1221 ± 11 ka, Channel et al., 2020), Punaruu (1126± 2 ka, Channel et al., 2020, and Jaramillo (from 1070 ± 3 ka to 990 ± 4 ka, Channel et al., 2020) from bottom to top. Cobb Mountain and Punaruu are found near the bottom and the top of the TD1.2.2 layer, respectively ( Fig. 7) (Duval et al., 2022). This is a 1.8 m level that shows no apparent sedimentary discontinuity (Table 2). Therefore, an approximate sedimentation rate of 1.89 cm/kyr can be calculated for this layer. If we assume that this sedimentation rate was constant from TD1.2.2 to the bottom of the sequence, an age of 1.31 Ma can be tentatively proposed for the beginning of the sequence, which agrees with the numerical TT-OSL age estimate of 1.30 ± 0.14 Ma presented by Duval et al. (2022), near to the bottom of the infill, but is younger than the 1.44 Ma estimated by these authors. In the first instance, one may note that the existing climate oscillations between Cobb Mountain and Punaruu magnetic events are apparently not registered in this sedimentary record, although future studies specifically focused on the millimetric laminations of this layer could find them.
Between the magnetic excursion Punaruu and the base of Jaramillo 56 kyr passed, and there is about a 3 m thick TD1 sequence that corresponds from TD1.2.1 to the middle of TD1.1.3. Then, an approximate sedimentation rate of 5.35 cm/kyr can be calculated for these layers. This sedimentation rate is significantly higher than that calculated for TD1.2.2, even more so if we take into consideration that TD1.2 is a more continuous record than TD1.1, where some hiatuses are identified (Fig. 7). If the magnetic excursion Punaruu is not considerate and the sedimentation rate is calculated from Cobb Mountain, an approximate sedimentation rate of 3.04 cm/kyr is obtained. If these estimations are correct, it would mean that sedimentation rates in TD1 would be inconsistent between layers, and that age estimation by sedimentation rate must be viewed with caution. The time between Punaruu and the base of Jaramillo is related to the Marine Isotope Stage 32. Then, the changes observed in TD1.1 cannot associate with the Marine Isotope Stages. The alternation of facies B and facies A of TD1.1 may be due to thousand-year environmental cycles.

Soft-sediment deformation structure of the stratigraphic sequence
The soft-sediment deformation structure inside a cave, where tectonics is not contemplated, can be assigned to postdepositional processes such as the collapse of the cave floor, erosion of the lower levels, or loss of volume of the underlying sediments due to desiccation or compaction (Marean et al., 2004;Van Loon, 2009). The eastern section, close to the cave wall, does not show this deformation. This may be due to the successive speleothems and cementations found in this area of the section, could suggest that the compaction of the sediments formed the deformation due to the weight of the upper lithostratigraphic units or that undermining processes of lower layers occurred to the west. This eastern area could preserve undeformed by two factors: on the one hand, the speleothems served as support and, on the other hand, the absence of large masses of sediment vertically in the eastern area, since part of the cave wall closes in that zone avoiding the sedimentary accumulation of entrance facies on it .

Source of the sediment ¿Autochthonous sediments?
The autochthonous facies have been defined as the materials that are derived locally within the cave (White, 2007). Different sediments have been included in autochthonous facies, such as weathering detritus, speleothem, breakdown, infiltrates, or guano (Goldberg & Sherwood, 2006;White, 2007). One of the possible origins that have been postulated for the sediments found inside the karst of the Sierra de Atapuerca, as TD1 and TD2 units, is the weathering detritus Ortega et al., 2013;Pérez-González et al., 1999). It is the sediment derived from insoluble residues in the bedrock and is found as sand, silt, and clay deposits with evidence of little or no stream transport. As the process for accumulating this type of sediment is relatively slow, it is usually deposited in areas without other types of sedimentation, i.e., in isolated areas or far from inputs of sediments from outside. For this reason, it usually appears inside the endokarst or at the base of the sedimentary cave infills, where this sediment is deposited before the cave's opening to the outside. Although we cannot ignore a certain influence of this residue on the sediment, the low percentage of residue in the Sierra de Atapuerca limestone, less than 0.1 g per kg , suggests that there is another source for the nine meters of sedimentary fill found at the base of Gran Dolina. Therefore, of the five main facies observed in TD1 and TD2, only facies D (Speleothem) and facies E (Breakdown) can be included in this definition of autochthonous facies, since Facies A, B, and C shows characteristics that indicate a certain transport, as discussed above, such as the horizontal lamination, the predominance of quartz in the mineralogy, the presence of rip-up clasts and the particle size dominate of sand in some layers.
The mineralogy of the sediments is very similar in all the stratigraphic section (Table 4), indicating that the source of the sediments is the same, and, therefore, that the changes in the granulometry are due to changes in the hydric regime. The richness of sand particle size observed in some layers in TD1 and TD2 (Fig. 5), e.g. TD2.2 or TD1.1.5, contrasts with the little sand found in the residue of the Sierra de Atapuerca limestone . In addition, the ESR and TT-OSL dates suggest the lack of heritage sand from the limestone (Duval et al., 2022). This evidence suggests a source of the sediment outside of the karstic system, being the sediment added by a fluvial stream that selected the sediment and entry into the karstic system from a relatively far entrance. The location of this entry is unknown since the intermediate level of the karst is not entirely known, although the southern contact between the Sierra de Atapuerca and the valley of the Arlanzón River has been postulated as a possible entry of these sediments (Ortega & Martín-Merino, 2019).
Therefore, the suitability of the term 'autochthonous' to describe TD1 and TD2 facies should be reconsidered. Kite (1997a, 1997b) separated the cave sediments into three categories: phreatic, vadose, and residual. As mentioned before, facies A, B, and C can be included in the phreatic category, but facies D and E are vadose facies since speleothems and gravity falls are formed in vadose conditions.
Other authors have used interior facies to define this kind of sediments (Kukla & Lozek, 1958;Creer & Kopper, 1976;Sroubek et al., 2007). This terminology refers to the distance at which an entry is located; thus, the interior facies is defined as the sediments found deeper in the cave, beyond the reach of surface weathering. In this environment, clays, silts, and sands predominate and are described as accumulation by fluvial flows (Creer & Kopper, 1976;Springer & Kite, 1997a, 1997bBosch & White, 2004;Sroubek et al., 2007). Since the sedimentological study suggests that facies A, B, and C can be interpreted as transported sediment by fluvial flows, in agreement with particle shape analyses (Campaña et al., 2016), the use of interior facies (and entrance facies for allochthonous facies), seems to be more suitable to the conditions observed in Gran Dolina than the previous classification of Allochthonous and Autochthonous   (Table 1).

Conclusion
TD1 and TD2 units are 9 m thick of interior facies in the base of Gran Dolina cave. This sedimentation began shortly after its vadose incision in the Early Pleistocene and was directed by the fluctuations in the level of the water table that caused the succession of phreatic and vadose phases, which would indicate general epiphreatic conditions inside the cave. Flooding processes deposited millimetric laminated sand and silts during phreatic conditions, while during vadose conditions, speleothem growths and cementations occurred.
These deposits have been defined as interior facies since they were deposits the inside of the cave when there was no entrance nearby. This terminology replaces the previous autochthonous term because it is more exact with the features observed in TD1 and TD2 units. The interior facies are divided by hydrological conditions. The phreatic facies (A, B, and C1) resulted from an underground fluvial flow that transported sediments from the outside to the inside of the cave. These sediments entered the cave from an entry situated in the contact between the Sierra de Atapuerca and the Arlanzon valley. The vadose facies (C2, D, and E) were deposited when the water table was below the sedimentation level, and they were formed by accretion, dripping, or falls.
At least ten stages of phreatic-vadose conditions have been identified, but indeed there are more. These stages should be related to the evolution of Arlanzón catchment during the transition between terrace T3 and terrace T4.
The sedimentation rate calculated by available geochronological data suggest great variation in sedimentation. The bottom of the sequence (from unit TD1.2.2 to TD1.2.4) has a sedimentation rate of 1.89 cm/kyr, while the TD1.1 deposits have an approximate sedimentation rate of 5.35 cm/kyr.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s41513-022-00200-8. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.