Geoarchaeological and microstratigraphic view of a Neanderthal settlement at Rambla de Ahíllas in Iberian Range: Abrigo de la Quebrada (Chelva, Valencia, Spain)

The Abrigo de la Quebrada is a Middle Palaeolithic rockshelter located in the Rambla de Ahíllas in the Iberian Range (Valencia, Spain). Archaeological work began in 2007 and was completed in 2015, reaching the rockshelter substratum and uncovering a record that spans from MIS 5 to MIS 4/3. The data from the geoarchaeological and micromorphological study of the site allow us to deduce that it was formed by alluvial contributions from the ephemeral stream, in different subenvironments varying from channel/bar to floodplain facies. These alternate with debris from different displacement processes depending on the unit, such as solifluction–gelifluction, mass displacement, and diffuse runoff. In addition, collapse episodes of the overhanging rockshelter roof influenced the pedological evolution of the record, with implications for the archaeological levels, especially in Unit G (Level IV). From a paleoenvironmental point of view, a more contrasted variability is reflected in the upper units of the site (MIS 4/3), especially in Unit G (Level IV), which, based on data, suggests temperate conditions, and in Unit H (Levels III and II) indicate cold conditions. In contrast, the lower units (MIS 5) are generally temperate, with the exception of Unit C (Level VIIIa), which reflects a colder phase.

It is in the latter corridor where most of the sites are concentrated, among which the Cova del Bolomor (Fernández-Peris, 2007), Cova Beneito (Doménech, 2005), Cova Negra (Eixea, Oltra, et al., 2020), Abric del Pastor , and El Salt (Galván et al., 2014) stand out for their archaeological records. In the central zone, which is the area of interest in our study, there are records such as Rambla de los Morenos, San Luís, and Abric del Barranc del Carcalín (Eixea, Chacón, et al., 2020;Fernández-Peris & Martínez-Valle, 1989), although these sites are not as large as the previous sites mentioned.
The site of the Abrigo de la Quebrada, located in the Turia river corridor, specifically in the Rambla de Ahíllas (Figure 1b,c), is an important enclave for information on the Neanderthal populations in this area since it is chronostratigraphically included in MIS 5 and possibly reaches MIS 3 (Carrión-Marco et al., 2019;Eixea, Chacón, et al., 2020;Real et al., 2020). It also has a geogenic and anthropogenic sedimentary record at the base of the rockshelter resting directly on limestone bedrock.
The record offered by the Abrigo de la Quebrada within the regional context is especially important because there are few sites analyzed from a geoarchaeological point of view in alluvial deposits such as the ephemeral streams corresponding to the Middle Palaeolithic, except for the records located in the Altiplano de Yecla in Murcia, Spain (López-Campuzano & Conesa, 2008;López-Campuzano, et al., 2007). These deposits, in general, represent a rare type of record due to the erosive action of the ephemeral streams characterized by steep slopes and a changing morphology.

| Location and geological context
The Abrigo de la Quebrada (39°48ʹ25ʹʹN, 01°00ʹ49ʹʹW) is located in Chelva, 69 km northwest of Valencia (see Figure 1b) and 728 m a.s.l. (Badal et al., 2012;Carrión-Marco et al., 2019). Discovered in 2004, and although a test pit was used to evaluate the possibilities of the site, it was not until 2007 that systematic archaeological excavation began (Eixea et al., 2011). This work finished in 2015 after reaching the bedrock of the rockshelter.
The current dimensions of the rockshelter are 38 m wide and between 2 and 9 m deep from the dripline to the back wall, which would have originally extended 11 m. The surface is slightly uneven, with a slight slope in a N-S direction, front to back.
Erosional processes on the talus outside the dripline explain the truncation of the record, which would have affected the upper levels of the site. The excavated area was 30 m 2 for the upper levels and about 27 m 2 for the lower levels, with a total thickness of 4 m (Real et al., 2020). The Abrigo de la Quebrada is located on the left bank of Rambla de Ahíllas, around 8-10 m above the current river bed. This ephemeral stream, which descends from the Sierra de Javalambre (Pérez-Cueva, 1985), runs through mountainous and steep relief, forming a canyon. In the final stretch, it flows into a plain together with other ephemeral streams, and these feed the river Tuéjar-Chelva, a tributary of the river Turia. At present, Ahíllas is an episodic flood channel, and its water supply is mainly rainwater. Its topographic course generally runs from north to south, although in the study area, it runs from southeast to northwest.
The modeling of this area of Rambla de Ahíllas forms a narrow corridor with an irregular longitudinal profile. Its slopes were formed by steep drops between the tabular limestone strata that shaped the relief of the site, exposing numerous rockshelters (Figure 2a). These are preferentially located in the limestone stratification planes, areas of weakness that favor erosion, as well as by processes of mechanical alteration and dissolution that generated inlets along the limestone in this sector, as is the case of Abrigo de la Quebrada itself. At the foot of the slopes there are mainly boulders and occasional debris.
Thirty meters from the site in a NW direction on the left bank of the ephemeral stream, there are 20-50 cm of alluvial deposits formed by subrounded gravels of mainly micritic limestone with almost no fine-matrix ( Figure 2b). Longitudinally furrowed potholes are located in the bedrock and silty-clayey fillings are retained in some sectors.
Vegetation colonizing some sectors of the channel (see Figure 2a) is consistent with a river system that alternates flash floods with long periods of water inactivity (Mateo et al., 1988).
The NW orientation of the rockshelter means that direct sunlight is scarce, which is also a consequence of the narrowness of the ravine with its canyon-like characteristics and the steepness of its slopes.
Significant drip zones can be observed on the walls of the rockshelter (see Figure 1c), and in some zones, there is an incipient formation of carbonate concretions.
The stratigraphic and sedimentary record of Abrigo de la Quebrada has been obtained mainly from the archaeological profiles shown in

| Archaeological assemblage
The Abrigo de la Quebrada comprises nine archaeological levels.
Human presence corresponding to the Middle Palaeolithic is recorded in throughout the site, except Levels I and VI. Level I corresponds mostly to the use of the rockshelter as a stable in historical times and will not be dealt with in this article. For Level VI, two optically stimulated luminescence dates on the feldspar fraction are 80,000 ± 4700 a and 83,200 ± 5400 a (Carrión-Marco et al., 2019;Real et al., 2020).
From an archaeological point of view, the levels ranging from II to V contain the most evidence of occupation and have several radiocarbon dates. Levels III-V were dated by radiocarbon MERCÈ BERGADÀ ET AL. | 681 accelerator mass spectrometry: III at 40,500 ± 530 14C a B.P. and IV at 43,930 ± 750 14C a B.P. (Real et al., 2020;Villaverde et al., 2008), although Levels III and IV were also dated by acid-base oxidation, giving results >50,800 14C a B.P. and >51,600 14C a B.P. (Eixea et al., 2011(Eixea et al., -2012 and V > 47,100 14C a B.P. (Real et al., 2020). The lithic industry includes discoid and centripetal Levallois technologies, and the tools are simple sidescrapers and Mousterian points. The abundance of cortical fragments, cores, and flakes of different sizes suggests that the knapping was carried out in situ (Carrión-Marco et al., 2019). Flint, mainly local, was used as raw material, although other raw materials are also used like limestone and quartzite (Eixea, Chacón, et al., 2020 (Eixea, Chacón, et al., 2020;Villaverde et al., 2017). In the lithic industry, discoid knapping dominates and the raw material is mainly flint. Regarding the fauna, specifically in VII, leporids are the best represented group and are likely the result of nonanthropic consumption activities. The leporids are followed by ungulates, with the Equidae families standing out, of which part of the sample could correspond to the donkey (Equus hydruntinus), followed by Cervidae and Bovidae-Caprinae, most of which seem to correspond to activities of Neanderthal groups (Real et al., 2019)  dominate, although leaves and wood/bark of woody plants are also found (Esteban et al., 2017).
The available paleoecological and technological data suggest that the Neanderthal communities that inhabited the site exploited the immediate environment by hunting ibex, horse, and deer in a pine forest environment. Most of the lithic raw material corresponds to local flint and comes from an area within 2-8 km (Eixea et al., 2011).

| METHODS
The study procedure that we used consisted of stratigraphic-sedimentary field description (see Table 1) and the application of soil micromorphology applied to the entire stratigraphic record, except for Unit B, in which we were only able to sample the sandy laminations. The study was based on the following profiles of the site's grid: A-3, C-3, E/3-4, E-4, E-5, E-7, G-5, I-4, I-5, I/4-5, and K-3 (Figure 6a-f). The analysis of some combustion areas, specifically in profiles E-3 and I-5, and grid square G-3 is also included in the study.
The protocol that we followed for the extraction of intact block samples, a total of 20, consisted of introducing the sediment into boxes covered with plaster, which allowed us to obtain blocks without altering the structure and arrangement of the sediments.
Samples collected by M. Mercè Bergadà were processed at the Micromorphology and Image Analysis Service in the Department of Environmental and Soil Sciences at the University of Lleida. The blocks were air-dried and then oven-dried at 40°C for 48 h to minimize precipitation of soluble salts during sample drying. They were then impregnated with an unsaturated colorless orthophthalic polyester resin and diluted with styrene at a ratio of three parts resin to one part styrene with a Methyl Ethyl Ketone Peroxide catalyst and activator (cobalt octoate). The embedded blocks were cut into smaller blocks a few centimeters thick and two thin sections were made for each block, generally 13.5 × 5.5 cm in size, although some are 7.7 × 5.5 cm and all are 30 µm thick. Thin sections (in total 43) were observed under a polarizing stereomicroscope and a petrographic microscope at magnifications between ×6.3 and ×400, using planepolarized light (PPL), crossed-polarized light (XPL), and oblique incident light (OIL). Thin-section description and interpretations follow the guidelines proposed by Bullock et al. (1985), Stoops (2021), Stoops et al. (2010), and Nicosia and Stoops (2017).
When presenting the microstratigraphic study of the stratigraphic units, we use the term microunit (MU) to distinguish Composed of some thin intercalations of angular gravels, in the form of platelets <6 cm with slight concretions toward the lower part, with some laminations of fine and medium brown sands. In profile I/6, there is a net contact with respect to the previous unit.

Clasts and boulders:
Slopes toward the north northeast. It consists of a deposit of platy clasts <15 cm with grayish-brown sandy silts with practically no carbonate concretions, except for profile E-6. In profile E-4, the platy clasts, 12-15 cm, appear in a vertical arrangement together with some blocks. The clasts deposit is located in sector E-G-I/3-5, totaling 9 m 2 , and the blocks make up the rest of the excavated area. There are combustion features, specifically the one located in the profile E-3, about 12 cm in length with a thickness of 5 cm, that could be described and is distinguished by a whitish calcitic accumulation (7.5YR6/1) next to charcoal.
Middle Palaeolithic E/4-7 G-I/4 D (VII) 43 cm 7.5YR6/4 Silty clays with cobbles and some boulders: Composed of brown silty clays with subangular to subrounded limestone cobbles, and some are carbonate fragments. Toward the base of the unit, there are blocks with traces of carbonation and toward the top, they decrease in size and the presence of cobbles increases. The E-G-I/4 sector is where its geometry is best represented, with a tabular tendency and a slight northeast slope. Toward the top, there are traces of biological activity (roots and soil fauna activity). Diffuse contact in sectors close to the wall of the rockshelter and abrupt toward Rambla de Ahillas.
Middle Palaeolithic E/3-7 C-3 G-I/4 I/3-7 E (VI) 90-100 cm 7.5YR5/6 Sands: Sands with brown silty-clay laminations with some scattered gravels, especially in C/E-3, the inner sector of the rockshelter, with some CaCO 3 nodules. The contact is erosive and has a wavy bedding, especially in profile G-I/4. Silty clays: Formed by silty clays with brown-colored sands and with subangular cobbles and gravels that diminish toward the outside of the rockshelter. Cementation can be seen in some sectors, as well as an increase in organic matter. Archaeological components and combustion areas abound. Very diffuse contact and the geometry tends to be tabular.
A/3-5 E-5 G-I-K/5 H (III-II) 37 cm 7.5YR6/4-7.5YR5/4 Gravels with sandy silt: Characterized by angular and subangular gravels giving rise to 1-4 cm platy clasts limestone that tend to decrease toward the G-I-K/5 sector and with a subrounded morphology. Toward the top, they appear with CaCO 3 concretions, especially in grid square A-3, and the sandy silt matrix is brown. Erosive contact in some sectors, especially in those between A/3-5, and its geometry tends to be irregular. There are combustion features, specifically a flat combustion area in profile I-5, where the following sequence from bottom to top was distinguished: 2. Thickness <1 cm consisting of a dark brown silty-clay fraction (7.5YR5/4) with charcoal residues.

| RESULTS
In this section, we present the most important micromorphological aspects of each unit and MU in stratigraphic order. These are shown in more detail in Table 3. From now on, when we refer to archaeological levels, they will be referred to with the acronym AL.

| Microstratigraphic description
The units and MU analyzed are as follows (see Table 3
In general, fine to medium sands ( Figure 12b)-MU.7, 9, 10, 11, 12, and 14-mainly composed of subangular to subrounded quartz and, to a lesser extent, micritic limestone, some clay aggregates, especially in MU.14, and some chert. They can appear, as in MU.10 and MU.12, with laminations of very fine sands (Figure 12c). Coarse to very coarse sands are also found in MU.8 and MU.13 (Figure 12d), composed of subrounded to rounded micritic limestone, quartz, muscovite, and aggregates of red clays.
Despite the particle size of the different MU, it is worth highlighting their composition, in which, apart from quartz and mica and in some cases chert, micritic limestones also appear repeatedly in the current bed of the ephemeral stream (rambla), as well as in the alluvial profiles. The subrounded aggregates of silty clays (terra rossa) from the limestone formations that can currently be seen in the bed of the rambla are also noteworthy. (e) Detail of (d), view of the hearth: upper, calcitic ashes, slightly phosphatized, and lower, thermoaltered silty clays (TH), in PPL. (f) Detail of (e), brown to brownish gray bone (calcined) and bubble-shaped pores due to high combustion, in PPL. MU, microunit; PPL, plane-polarized light; XPL, crossed-polarized light. Verrecchia et al., 1995). We propose that its origin in Abrigo de la Quebrada in some sectors was favored by the presence of boulders in sectors E-G-I/7-10 that did not facilitate drainage of thaw water and so this bacterial activity developed. This also reveals that Unit C was exposed to the atmosphere for a certain period of time.
Another interesting feature in the same sector is the formation of pendants on the clasts. Their genesis is related to processes of dissolution and precipitation of carbonates, which is accentuated in the lower part of the detrital material as more humidity has been retained (Dorronsoro et al., 2020;Ducloux & Laouina, 1989;Durand et al., 2010). These accumulations appear in the form of layers; the ones closest to the clasts are older and are formed of micritic calcium carbonate and the most recent ones are formed by an increase in detrital material and appear darker. According to Durand et al. (2010), the carbonate-rich laminations form in more arid phases with less biological activity, the latter more typical of a more humid environment; therefore, it also demonstrates certain environmental changes in the same unit that would correlate with the above processes. Finally, we identified illuviation of fine material in voids and gravels that is likely due to liquefaction processes on the surface and/or rapid percolation of surface runoff after thawing (Van Vliet-Lanoë, 2010).
The paleoenvironmental conditions of this Unit C were initially cold and arid, but gradually became wetter as conditions moved toward a less harsh environment.
Unit D (AL VII) is caused by several processes: (a) an episode of rockshelter roof collapse that is particularly reflected in grid squares E/4-5, not as important as in the previous episode, although in E-3, there are still gravels and cobbles from the rockshelter wall, but not as abundant as in the previous episode, and (b) a mass displacement as the coarse material is embedded in a very fine matrix and tends to have an orientation depending on the slope of the deposit (Karkanas & Goldberg, 2019). Some laminations of very fine material are also observed in the matrix as a result of this displacement.
One aspect to consider is that there are carbonate concretions in the cobbles and blocks (see Figure 5a) that did not originate in situ, but are instead from the initial roof or wall of the rockshelter. This feature indicates that between Units C and D, there was a hiatus in which conditions were favorable for the precipitation of carbonates, due to a warm environment and availability of water.
After the formation of Unit D, conditions of stability were established. This is indicated by localized root activity in the form of calcification-decalcification features associated with impregnation of root tissues known as queras (Durand et al., 2010), silica phytoliths, and some spores. There is also evidence of biological activity, namely, faunal passages, biogenic calcite granules excreted by earthworms (biospheroids) (Canti, 2017;Durand et al., 2010), and the abundant presence of terrestrial gastropods. All these features suggest that Unit D was subaerially exposed for a period of time and that pedogenesis began.
Unit E (AL VI) is due to the alluvial dynamics of the ephemeral stream, in which different episodes can be distinguished with a sequence generally decreasing in grain size. It can also be seen by the erosive contact in some sectors, as well as by its wavy bedding, which is indicative of this dynamic (see Figure 5b). We located several episodes, among which we distinguish from bottom to top: 1. Sands with angular gravels from the dismantling of the previous unit, which could correspond to episodes of higher energy, perhaps of the flash flood type.
2. Well-sorted mainly fine to medium sands denote a low-traction energy environment perhaps in the water recess or could correspond to overflows of suspended load. There are also some episodes of medium to coarse sands without gravel or pebbles, a fact that makes us believe that these correspond to episodes of higher energy.

| Anthropogenic activity in the sedimentary record
In Abrigo de la Quebrada, Neanderthal activity is evident in practically the entire record from a microstratigraphic point of view, except for Units A (AL IX) and E (AL VI), which have geogenic characteristics not very conducive to the rockshelter's habitability.
However, the identification of anthropogenic activity manifests itself mainly in hearths and especially in Unit G (AL IV), where the density of components is considerable.

| Hearths
The hearths analyzed were located in Unit C (AL VIIIa), Unit G, (AL IV), and, finally, Unit H (AL III). They are characterized by their flat or slightly concave morphology, in which it has been possible to identify a single combustion event. The hearths in general consist of rubified sediment overlain by ash and a few associated lithic and bone remains.
In Unit C, a hearth was located in profile E-3, composed of woody tissues and ashes characterized by calcitic accumulation together with some prismatic pseudomorphs of calcium carbonate representative of bark and wood ash (Brochier, 1996). The support shows traces of thermal alteration slightly affected by processes of illuviation of fine material.
In Unit G, a hearth characterized by silty clays with CaCO 3 was found in MU.23, specifically in grid square G-3. We can distinguish cemented detrital material with marked traces of thermoalteration. In its upper part, calcitic ashes with calcined bone remains are evidence of a high-intensity fire (Mentzer, 2014).
In Unit H, profile I-5, there is also a hearth in which the fire is established on a detrital layer with moderate thermal transformations together with partially carbonized plant remains and phytoliths that seem to correspond to a previous anthropic activity in Unit G. In the upper part, there are calcitic ashes with a high content of silica phytoliths and calcined bone remains, all denoting high-intensity combustion (Bergadà, 1998;Villagran et al., 2017), but probably below 750-850°C, as the phytoliths do not show signs of melting (Canti, 2003). This hearth appears with the freeze and thaw traces typical of Unit H. It also subsequently underwent dissolution and reprecipitation processes, forming microsparitic to sparitic CaCO 3 infillings in the ashes as well as in the fissures of the bones, which, in some cases, led to their fragmentation. Its formation is due to local dissolution of ashes, followed by precipitation. Slow crystallization is evidenced by crystal sizes of >50 µm, indicating a very cool environment and stable conditions (Durand et al., 2010).
In all, the microstratigraphic study shows that there are practically no differences between the hearths analyzed, despite their time difference. These combustions reached high temperatures, which is probably why there are practically no charcoal fragments. It is difficult to hypothesize the function of these hearths as there is a lack of micro-scale studies on the extent of these combustion areas.
Finally, the good preservation of the ashes, with no evidence of dispersion especially in the hearths of Units C and H, could indicate a lack of reiterated occupations in the rockshelter. However, these considerations must be interpreted with caution, as very few hearths have been analyzed from a microstratigraphic point of view at this site.

| Unit G (AL IV)
This unit is associated with a high density of anthropogenic and biogenic components that distinguish it from the rest of the record.
Two of the most important are bones and teeth, some with traces of thermoalteration and to a lesser extent fat-derived char.
Some bones have in situ breakage that resulted in articulated and slightly displaced fragments, most likely due to trampling (Miller et al., 2009;Villagran et al., 2017). These data reinforce evidence for human activity in the unit that, in some cases, is related to combustion. One of the aspects worth highlighting about these components is their diagenetic evolution according to the localized MU. In the silty clay MU.22 and MU.24 with high organic contents, bone remains are more abundant. The bone remains are generally reddish-brown, and in some cases, have stains that appear reddishorange in PPL. These features are due to microbial activity leading to the precipitation of iron and manganese compounds (Bergadà et al., 2013;Polo-Díaz & Fernández-Eraso, 2010;Villagran et al., 2017) and also due to the precipitation of humic acids from the soil (Villagran et al., 2017). This would accentuate the dissolution features, giving rise to an increase in porosity and more subrounded contours at the bone edges due to the acidic environment (Karkanas & Goldberg, 2019). Other bone fragments have calcified filament infillings of fungal origin (Durand et al., 2010).
In contrast, in the CaCO 3 silty clay MU.21 and MU.23, microbial activity in the bone fragments is considerably reduced, which is also an indication that carbonate precipitation was rapid, immediately after the formation of the MU. One of the features affecting bone remains is their enrichment in carbonate in their pores or at their edges, which, in some cases, causes their fragmentation. This process is due to carbonate-rich solutions derived from the surroundings reacting with the phosphate material in the bones. It is these MU that we believe best preserve the hearths such as the one located in grid square G-3 discussed in the previous section, since cementation prevents the proliferation of biological activity in the soil.
The components of plant origin are both uncharred and charred.
Some are woody and associated with calcium oxalate crystals, and their presence may be due to combustion taking place at a low temperature (Milek, 2014) or due to poor oxidizing conditions (Courty et al., 1989), data that, when associated with the main traces of thermoalteration observed in some bones, also reflect moderate temperatures. A smaller proportion of woody ash remains are reworked mainly by biological activity.
It is worth mentioning that some coprolites of animals with carnivorous and herbivorous diets have been found in different MU but penecontemporaneous to Neanderthal activity. All these features, as well as the pedogenic activity that developed, allow us to infer that the rate of sedimentation was slow, diffuse, and irregular, and therefore, the occupation surfaces were subaerially exposed, favoring the formation of cumulative palimpsests (Bailey, 2007).
In this unit, occupations were more recurrent in the settlement, but at the same time intermittent and/or seasonal, as there are also features that indicate that it was frequented by animals.

| Chronostratigraphic and paleoenvironmental evolution
Some challenges arise in establishing the chronostratigraphic and paleoenvironmental evolution of the site (  -Gallego, 1987), which is part of the drainage basin of the Rambla de Ahíllas and the lower Turia, level III of 5-10 m (Carmona, 1982;Pérez-Cueva, 1985).
According to the available data, this unit corresponds to a torrential flood under arid climatic conditions with a decrease in vegetation cover and an increase in erosion and consequently an increase in sediment production. It is suggested that aggradation probably occurred during the cold stadials of MIS 5. A period of aggradation of the fluvial systems in the Iberian Range and in other Mediterranean localities has been identified during stadial 5d (Soria-Jáuregui et al., 2016), perhaps also the case of Abrigo de la Quebrada.
Subsequently, calcium carbonate cementation processes occurred, giving rise to a petrocalcic horizon, which is a common advanced-phase calcic horizon in Pleistocene alluvial records from Iberia (López-Campuzano et al., 2007;Roquero et al., 1999), usually found in arid or semi-arid environments (Soil Survey Staff, 2015).
After a sedimentary interruption, Unit B (AL VIIIc and b) was formed by debris contributions, as well as by episodes of alluvium from the ephemeral stream. The energy is less in these episodes relative to those before. The layout of the deposit suggests that the rockshelter was not part of the main channel of the ephemeral stream. This is an episode, unlike the previous one, where more evidence of human activity is found. This is followed by Unit C (AL VIIIa). As we have pointed out in the previous section, its formation and evolution are complex, as Note: The dashed lines show proposed episodes not recorded in the sedimentary fill. The MIS age have been based on LR04 stack of marine benthic foraminiferal δ18O data (Railsback et al., 2015).
thermoluminescence at 45 and 46 ka and attributed to MIS 3 (Fumanal, 1996). This indicates that around this time, there were phases of stability that allowed the development of soils in the Pleistocene records. To all this, we must add that in Unit G, there are bone remains of the Mediterranean tortoise (Testuto hermanni).
There are also some indications of their presence in MIS 4/3, although they were mostly present in the Iberian Peninsula during the Last Interglacial (Morales Pérez & Serra, 2009;Real et al., 2021).
From Unit G onwards, an erosional contact is observed in some sectors and the environmental conditions worsen considerably, as determined in Unit H, dated at 40.5k 14C a B.P. and >50.8k 14C a B.P. (AL III). There are episodes of frost shattered clasts and gelifluction that would be associated with cold oscillations. The dating, as in Unit G, does not offer a good resolution; however, it could also be placed in MIS 4/3. In any case, in the absence of a set of dates, we believe that the inferred paleoenvironmental data are insufficient to make any chronostratigraphic approximation for Units G and H. We are further limited by few specific records for MIS 4 in Iberia that can offer a view of this period . It should also be added that the geoarchaeological study carried out at the nearby Abric del Pastor (Alcoi, Spain), located in MIS 4 and the beginning of MIS 3, does not reflect any paleoenvironmental differences either, with evidence of generally cold conditions . In any case, what we can suggest is that these units correspond to contrasting paleoenvironmental conditions: Unit G (AL IV) corresponding to a more temperate, interstadial phase and Unit H (AL III-II) corresponding to a colder, stadial phase. These units may perhaps already be located at the beginning of MIS 3, as character- Ahíllas ravine and, in short, of the Turia river.
A relevant aspect of the Rambla de Ahíllas ravine evolution is the incision of the ephemeral stream, between Unit E (AL VI) and F (AL V). This in fact preserved the site from later erosive actions and contributed, from our point of view, to the fact that human occupations were more recurrent at the site, since it would have made the ravine a much more accessible route.
The site can be described as an alluvial terrace in which a water flow dynamic has been detected, with intermittent floods alternating with debris from around the rockshelter. In addition to these processes, the retreat of the rockshelter roof along the sequence favored postdepositional dynamics and soil formation. It is during these biostasis phases, with a tendency toward stability, that human occupations increase, while at the same time, they have a palimpsestic nature due to the slow burial of the surfaces of the occupation, as is the case of Unit G (AL IV). In some MU of this same unit, the formation of carbonate laminae has preserved anthropogenic features in the sedimentary fill.
Discontinuities are reflected throughout the stratigraphic record.
There is a more contrasting paleoenvironmental variability from a | 709