A new Lower Pleistocene archeological site in Europe (Vallparadís, Barcelona, Spain)

Abstract

Here we report the discovery of a new late Lower Pleistocene site named Vallparadís (Barcelona, Spain) that produced a rich archeological and paleontological sequence dated from the upper boundary of the Jaramillo subchron to the early Middle Pleistocene. This deposit contained a main archeological layer with numerous artifacts and a rich macromammalian assemblage, some of which bore cut marks, that could indicate that hominins had access to carcasses. Paleomagnetic analysis, electron spin resonance-uranium series (ESR-US), and the biostratigraphic chronological position of the macro- and micromammal and lithic assemblages of this layer reinforce the proposal that hominins inhabited Europe during the Lower Pleistocene. The archeological sequence provides key information on the successful adaptation of European hominins that preceded the well-known fossil population from Atapuerca and succeeded the finds from Orce basin. Hence, this discovery enables us to close a major chronological gap in the early prehistory of Iberia. According to the information in this paper and the available data from these other sites, we propose that Mediterranean Western Europe was repeatedly and perhaps continuously occupied during the late Matuyama chron.

During the early Pleistocene, hominins moved into new environments outside of Africa to the Euroasian continent in one or several migratory waves (1). By developing strategies and technologies to adapt to new environments, hominins succeeded in the next several hundreds of millennia in occupying diverse habitats from eastern to western Eurasia. The pace and causes of this early colonization are still the source of intense scientific debate (2–6). According to some authorities, the first human colonization of Europe would have been episodic or intermittent, at least until the early Middle Pleistocene, and of little demographic importance (7). However, there is increasingly solid data from Spain (8–13) that clearly indicate that Europe was occupied for the first time over a million years ago, and which thus supports the “Mature Europe” hypothesis (14). In addition, paleoanthropological interpretations of the first hominins’ morphotypes suggest that there is sufficient evidence for biological continuity within the Eurasian fossil record (8, 15, 16).

Vallparadís is a recently discovered Lower Pleistocene site in the Mediterranean Iberian Peninsula, which provides sound information concerning both the ecological adaptation and the lithic technology of the hominins, and which suggests that this region has been continually settled for at least 1 million years.

Geomorphology and Sedimentology.

The site is located in a sequence of intermittent fluvial and alluvial sediments on the banks of Vallparadís River at the location where it flows through the city of Terrassa (Fig. 1B). Large Quaternary fossiliferous Pleistocene deposits were discovered in this river valley during excavations in 1997 at the neighboring site of Cal Guardiola (17). From 2005 to 2007, an ≈20-m thick deposit was exposed in this area during the building of a new train station. The deposits were part of the alluvial fan of a tributary in the town of Terrassa. The sedimentary sequence was divided into two parts, separated by an erosional surface (unit 5), and is formed by a series of interspersed layers of fluvial and alluvial origin (Fig. 1A). The fluvial layers are lutites deposited within the flood plain, weak developed river bars, and travertine formations; the alluvial layers are lutites and alluvial/colluvial conglomerates. There is also a lake-marsh layer rich in macromammals and fossil wood remains (at the bottom of unit 4). Archeological and/or paleontological remains have been found in practically all of the stratigraphic units (Fig. S1. The archeological excavation has been limited to the richest unit of the sequence (unit 7).

Fig. 1.

Geographic setting and geological context. (A) Summary of the lithostratigraphy and chronology of the Quaternary sedimentation at the Vallparadís site. The synthetic column shows the different stratigraphic units to the Right and the magnetostratigraphy column to the Left. (1) archeological remains (lithic and fauna), (2) root marks, (3) Upper Pleistocene terrace, (4) clays and muds with gastropods, (5) unit 5, (6) red clays and muds, (7) unit 7 (layers 10 and 10c), (8) brown clays and muds, (9) conglomerates, and (10) paleo-floor. (B) Geographic setting of Vallparadís site (Terrassa), near Barcelona, Spain. EVT, acronym of Vallparadís site.

Unit 7 is made up of two archeological layers: the upper layer 10 and the lower layer 10c. Layer 10 has a maximum thickness of 2 m. The sediment is formed by gray-green clays and mud with angular and subangular clasts of quartz, lydite, and metamorphic rock, ranging in size from 1 mm to 1 cm in a matrix supported fabric. Layer 10c is made up of gravel and conglomerates dominated by large-scale rounded blocks (up to 40 cm thick) in a matrix supported fabric. In the NE sector of the excavation, a carbonate layer, formed of oncolithics (oncolithic rudstone), covers layer 10c and descends at a slope of 15°; the seasonal circulation of water would have created this carbonated platform through the shoaling of conglomerates. Layer 10c was probably formed by the accumulation of transported massive debris flow and to a lesser extent by sheet flows with erosive capabilities. Layer 10 was formed on layer 10c when clay was deposited as a result of dense mudflows moving in a NE-SW direction. These mudflows were caused by successive seasonal river level rises and were rapidly colonized by vegetation.

Chronological Data.

Paleomagnetic analysis and electron spin resonance-uranium series (ESR-US) readings were used to date. The Vallparadís sequence shows three magnetozones: (i) an initial normal magnetozone N1 (units 8 and below), (ii) a reverse magnetozone R1 (units 4–7, inclusive), and (iii) a second normal magnetozone N2 (unit 3 and above). Stone tools have been recorded in the N2 and R1 magnetozones, the latter of which contains layers 10 and 10c (unit 7). The rodent assemblage of these layers (see below) shows that they are below the Matuyama-Brunhes boundary of 0.78 Ma. Furthermore, this rodent association is younger than that of Sima del Elefante in Atapuerca (8), which indicates that the paleomagnetic sequence as a whole is younger than the Olduvai subchron. Between units 3/4 we identify the paleomagnetic transition between the Lower and the Middle Pleistocene. Consequently, the normal magnetozone at the base of the Vallparadís section should correspond to the Jaramillo subchron, which means that the age of the layers 10 and 10c can thus be set at ca. 0.98 Ma.

The combined method of ESR/U-series was applied on two horse teeth unearthed from layer 10 (SI Text). For such an old period (>800 ka), the description of the U-uptake in dental tissues is often very difficult as illustrated by one of the two samples presented here. However, the recorded U-series data allow us to calculate the age of one sample (EVT0601). The high ²³⁰Th/²³⁴U ratio values measured in the dentine and cement of the EVT0602 sample (Table S1) prevent a combined ESR-US calculation. In this case, the age is considered as an Early-uptake model, which corresponds to the modelized maximum dose-rate associated with dental tissues (Table S2), thus providing a minimum age (Table 1). We conclude that the ESR-US age of EVT0601 is about 0.83 ± 0.07 Ma. This determination is therefore in agreement with the results obtained from the paleomagnetic observations and supports the age of late Lower Pleistocene of the site.

Table 1.

ESR-US combined ages and associated data obtained on teeth samples from Vallparadís

				p factor
Sample	Unit	Total dose rate (μGy/a)	DE (Gy)	Enamel	Dentine	Cement	ESR-US age (Ma)
EVT0601	10	2943 ± 244	2432 ± 65	−0.54	−0.47	−0.61	0.83 ± 0.07
EVT0602	9/10	5843 ± 488	3151 ± 69	NC	NC	NC	NC (EU age 0.54 ± 0.05)

The ESR-US combined age cannot be calculated for EVT0602 (NC). An ESR-EU age is mentioned as a minimum age.

The rodent assemblage from layer 10 includes Mimomys savini, Iberomys huescarensis, Ungaromys sp., Eliomys quercinus, and Sciurus sp. The most characteristic element in biochronological terms is I. huescarensis, which is also present at other sites such as Huéscar 1 (the Guadix-Baza basin), Sima del Elefante, and Gran Dolina (the Atapuerca karstic complex) (18, 19). The metric values of the rodents’ teeth are comparable to those in Gran Dolina TD3, TD4, and TD4B, especially when the L and A/L values (sensu Meulen, 18) are compared. However, the lower molars of the Vallparadís I. huescarensis have a less derived morphology than those from the lower layers at Gran Dolina (less prominent BSA3, sensu Meulen, 20), and are thus closer to those from Huéscar 1 and Sima del Elefante. Moreover, the absence of advanced microtine species such as Stenocranius gregaloides and Terricola arvalidens also differentiates Vallparadís from Gran Dolina. Together, these data indicate that Vallparadís is older than TD3 and TD4, and closer to Sima del Elefante and Huéscar 1 (Fig. S2).

The Archeological Layer.

The entire area (781 m²) of layer 10 was excavated. The archeological material consists of more than 57,600 elements (24,854 faunal remains, the remainder being stone tools and debris) (Tables S3, S4, and S5) and is distributed along the horizontal plane without forming any significant concentrations. Polyspecific accumulations of fossil remains were only found lying directly on the conglomerates at the bottom of layer 10. In these polyspecific accumulations, the most numerous and complete remains were of hippopotamus, the bones of which, in some cases, were still anatomically connected or semiconnected. Vertically, the projection of the archeological remains shows three main individual accumulations of fossil material separated by sterile sediments. The main accumulation of archeological material is located in the middle of the thickness of layer 10. This accumulation is of interest due to the high concentration of hyena coprolites, as well as a large number of Dama antlers, which are mostly complete. These antlers are still attached to parts of the cranium, often bearing gnawing marks, thus ruling out the possibility that they were shed. The second densest accumulation is located directly above the conglomerates, at the bottom of layer 10. Most of the mammalian skeletons appear dismembered, which would rule out the hypothesis that the animals had been surprised by mud floods and buried (21).

Three refittings carried out in the field demonstrate that the lithic assemblage was accumulated as a result of in situ reduction processes. In addition, there were objects in secondary position, especially in layer 10c, this layer yielding larger objects, a greater technical diversity, and variety of local raw materials.

The Lithic Assemblage.

The technical characteristics of the artifacts are strongly conditioned by the following factors: the small size of the raw material clasts, of which more than 70% were quartz; the simplicity of the reduction and retouching processes; short chaîne opératoires; and the bipolar detachment of flakes. Other raw materials were also employed, such as lydite, flint, quartzite, and metamorphic rock. All these small clasts were transported by natural processes to the site. It seems likely that the small clasts were selected intentionally and that this choice was as culturally determined and not conditioned by the kinds of raw material available (22, 23). This observation is reinforced by the fact that the conglomerates in the immediate vicinity of the site were not used to provide large blocks of raw material. All of the elements of the chaîne opératoire are represented and we believe that this is because the mudflows have moved the archeological record in its entirety regardless of the size and density of the lithic elements. This in turn means that the integrity of the archeological record is ensured (Fig. 2).

Fig. 2.

Lithic tools and faunal remains with cut marks. (1) Chopper of metamorphic rock from layer 10c, (2) cubical core of flint knapped using the bipolar on anvil method, (3) denticulate on a small cobble of lydita, (4) bec of quartz, (5), bec of flint, (6) anvil stone of flint, (7) notched of quartz, (8) denticulate of flint, (9) denticulate of flint, (10) denticulate of quartz, (11) mandible of Rhinocerotidae with several isolated short and deep cut marks in the tongue side, and (12) dorsal vertebra with two parallel short and deep cut marks in the apophysis and carnivore marks beside them.

The flakes show the systematic use of orthogonal core reduction, mainly bipolar on anvil. This method is common at other European Lower and Middle Pleistocene sites (24–27). This knapping technique results in flakes with two opposing butts. The flakes often bear cortical surfaces, both on the butts and on the dorsal face. The bipolar flakes also tend to show trimming, or marginal retouches with a denticulate delineation on the distal extremity produced accidentally during knapping when the core was struck on the anvil with a hammerstone.

When flake detachment was extended to a new knapping plan, polyhedrical and cubical forms were generated. There are a few examples of centripetal cores and débordant flakes. Low quantities of these centripetal elements are also present in the other Lower Pleistocene sites (24–27). The recovered anvils are flint, quartzite, and metamorphic rock. They are between 10 and 15 cm long and are flat rolled boulders upon which percussion traces and depressions can be seen. Large-size objects on cobbles (e.g., a chopper in a metamorphic rock) were found in layer 10c.

Retouched tools such as denticulates and notches were abundant. There are flakes that combine a denticulate retouch on a long side and a notch on the transversal distal end, which in many cases forms the trihedral shape of a bec. This repeated retouching is the same habit that has been recorded at other sites (24, 28, 29) and is a recurrent morphotype among the assemblages produced by the first hominin populations outside of Africa.

The characteristics of the lithic record are comparable to those of other mode 1 technology (Oldowan-like) sites (1). These characteristics can be summarized as (i) the use of local raw materials in knapping processes; (ii) the mastery of orthogonal methods of core exploitation and, at this site, bipolar on anvil; (iii) the small size of the objects; (iv) short and not very intensive knapping processe; (v) the presence of large objects on cobbles (choppers); and (vi) the presence of denticulates, notches, and becs as morphotypes of retouched tools.

Macromammalian Fauna, Taphonomy, and Paleoecological Reconstruction.

The taxonomic identification of the mammalian species indicates that they belong to the Epivillafranchian biochron (30, 31). These are: Pseudodama vallonetensis, Sus cf. scrofa, Equus altidens, Premegaceros verticornis, Bison sp., Stephanorhinus hundsheimensis, Hippopotamus antiquus, Elephas antiquus, Lynx sp., Canis mosbachensis, Ursus deningeri, and Panthera gombasgoezensis. In layer 10c, an almost complete adult cranium and other anatomically semiconnected appendicular remains of Pachycrocuta brevirostris were recovered. Likewise, dental and maxillary remains of Macaca sylvanus cf. florentina have also been recorded (32). This faunal association indicates that ecologically the site was located on an ecotone, near a river with an abundant and regular water flow.

The faunal record is dominated by cranial remains, isolated dental elements, and appendicular long bones. The skeletons are dismembered and there is an overlaying of large animal remains, which could be attributed to the differential preservation of the taphocenosis after the intense activity of scavengers. Rolling and polishing caused by water abrasion affected only 12.1% of the bones analyzed (334 bones) and always at the lowest stages of alteration. This rules out the selection of the bones according to density and the impact of water flows and tractive transport on the composition of the record. Scavenger activity has been widely identified at the site in almost 50% of the bones analyzed (334 bones). This takes the form of bites, gnawing, and regurgitated fragments. Medium-sized individuals and equids are mainly represented by cranial remains, isolated dental elements, and distal ends of the appendicular skeleton (metapode, calcaneus, and astragal). The absence of the rest of the skeleton probably means that the other elements were destroyed by hyenas (33, 34).

The cranial and appendicular skeleton is also better represented among the larger species. The entire sequence of scavengers’ activities can be seen in the long bones, which were abandoned in the form of diaphyseal cylinders. Coprolites are abundantly represented in the deposit and the final composition of the bone collection was determined by the scavengers’ activities. It is especially important to note that no hyena infant remains have been found. These would be common in a hyena den given the ethology of this species (34, 35).

The sample also shows the impact of humans on animal skeletons. Twelve elements (i.e., 5% of the sample) bear cut marks, including dismembering and defleshing, and five elements show fractures of anthropic origin (Table S6). The diversity of taxa, the skeletal elements, and the age of the individuals consumed by the hominins are consistent with a general predator strategy. The percentage of cut marks in Vallparadís is significantly higher than in other sites of the same age such as Orce, Dmanisi, and ‘Ubeidiya (36, 37). The low percentage of cut marks in these sites contrasts with remains from cave sites such as TD6, where cut marks reach 25%. This high percentage in TD6 is directly related to a low incidence of scavenger activity (38).

Half of the bones with cut marks also show bite marks, which would indicate competition between hominins and large predators for the same prey (2, 39). No overlaid marks have yet been found that would enable us to determine the sequence of these actions. However, in a diaphyseal cylinder (a rhinoceros tibia), a large incision covers the whole length of the long bone. We deduced that hominins defleshed the tibia before the hyenas could consume the epiphysis and turn it into a cylinder (40). Also interesting is the low representation of bones with anthropic fractures. The possible lack of interest in bone marrow has also been observed at other sites in Eurasia and Africa and was interpreted as evidence that hominins did not act as scavengers (37).

To determine the environment of layer 10, a combination of pollen records and amphibian remains was used. At the Vallparadís site, pollen was only found at the bottom of unit 4, a lake-marsh deposit (chronologically on the Matuyama-Brunhes boundary). These data suggest that during this period, the landscape was forested with more than 60% arborean pollen (AP), above all Pinus, Quercus ilex-coccifera, and Corylus. However, to gain a better insight into the environment of layer 10, we had to use a palynological analysis of the richest fossil layer (D2) at the neighboring Cal Guardiola site, which is located in the same sedimentary deposit as Vallparadís and is separated from it by less than 100 m. The Cal Guardiola site is geochronologically and biostratigraphically dated to the lower boundary of the Jaramillo subchron (41). Here, the analysis shows an open landscape with only 12% AP, including 1% of riverbank tree species. The key characteristic of this landscape is that it experienced seasonal rainfall, which would have enabled the development of megaforb meadows during the rainy seasons. This flora suggests a temperate, subarid, Mediterranean climate, with a seasonal distribution of rainfall (Fig. 3). In contrast, layer 10 has no similar pollen record; however, it does have amphibian and squamate reptile fossil remains such as the western spadefoot toad (Pelobates cf. cultripes), one indeterminate anuran, one indeterminate lacertid, and two colubrine snakes (Natrix natrix and cf. Rhinechis sp.). The overall association may characterize a Mediterranean landscape (meso and thermo-Mediterranean bioclimatic levels) with sunny open-dry areas with few or no vegetation and rocky soils and laterally more humid areas with presence of water and open Mediterranean woodlands.

Fig. 3.

Composite pollen diagram from the Cal Guardiola and Vallparadís sites (Terrassa, Spain). Sample D2 is an average of 27 samples from Cal Guardiola layer D2, whereas the two samples from unit 4 belong to Vallparadís. Samples from both sites show vegetation and climate changes between Lower/Middle Pleistocene. During layer D2 (Lower Pleistocene) the climate was arid (Chenopodiaceae, Lygeum, Tamarix) and warm (e.g., Quercus type ilex-coccifera, Olea-Phillyrea). Immediate surroundings were dominated by Poaceae grasslands, with fragmented patches of savannah-type trees. The mountain areas had mixed forest of Holm Oaks and deciduous trees, and in the highest areas there would have been red pine, fir, and birch trees. Of particular note is the presence of exotic taxa (e.g., Pinus type haploxylon, Taxodiaceae, Carya), characteristics of the Pliocene that confirm the importance of the Iberian Peninsula as a refuge area. In this landscape, shrubs would be well represented. However, these taxa are palynologically underestimated. The lake-marsh environment is well represented by a high diversity of hygro-hydrophytes and Pterydophyta taxa and riverside trees such as Salix, Ulmus, Tamarix, and Populus. Later, at the beginning of Middle Pleistocene, the climate became cool (cool and humid), with a landscape dominated by coniferous (Pinus) and deciduous (Quercus type deciduous) forest. Thermophylous elements (e.g., Olea-Phillyrea) and steppic and cold elements (e.g., Ephedra) coexisted in this forest. The landscape lost diversity and Mediterranean taxa (e.g., Pistacia, Cistus, Chamaerops-type). At this point, all of the Pliocene exotic taxa disappear.

Conclusions

The sedimentary sequence of Vallparadís dates from the Jaramillo subchron to the early Brunhes. Lithic remains and cut marks on animal bones show that hominins were present in the area. Therefore, we hypothesize that the region was repeatedly and perhaps continuously occupied between the late Lower and early Middle Pleistocene periods. Consequently, the first European hominins were capable of reproducing and surviving for generations in the Mediterranean regions of the Iberian Peninsula. This situation continued at Vallparadís until the early Brunhes period (unit 4), when changes in the archeological record show the arrival of an arboreal landscape, the presence of a fully Galerian species such as Cervus elaphus, and larger lithic artifacts than those which are found in the preceding layers.

The first hominins began to expand from their original gallery forests in East Africa into other habitats, and their diet became increasingly carnivorous. This migration into other ecosystems indicates that behaviorally and technologically these early hominins were not limited to a specific distribution of resources or climatic conditions. At Vallparadís, paleoenvironmental analyses show that the local environmental conditions varied, especially in terms of rainfall. This evidence is supported by the climate models obtained from the marine, continental, and palynological records (42, 43). Nevertheless, the hominin populations of Vallparadís faced these changing climatic conditions and their effect on the prehistoric landscapes and fauna (44). The most decisive factor regarding the expansion and adaptation of hominins outside Africa was probably their carnivorous diet rather than any cultural, ecological, and even physical features (36). In the Mediterranean temperate ecosystems, hominins must have depended on animal resources for subsistence at certain times of the year. This would have led to direct competition between hominins and other large European predators and scavengers of the time (45). The hominin groups that occupied Vallparadís exploited a rich biological environment and adapted in a way that shows they were not at all selective in terms of either the prey they consumed or the raw material they used in their knapping processes. This broad adaptive strategy meant that the hominins would not have required great technological skills to exploit the resources available. Therefore, we propose that these first European hominins would have been a cohesive group of general predators that would have taken advantage of the best parts of the carcasses that they found in the areas along river banks. We also hypothesize that they had primary access to prey and, along with large predators, occupied a position near and possibly at the top of the food chain.