The Departament de Cultura of the Generalitat de Catalunya, Diputació de Barcelona, Ajuntament de Capellades, and Arts Gràfiques Romanyà-Valls S.A provided support for the Cinglera del Capelló excavations. This research was partly sponsored by Ministerio de Educación y Ciencia (HUM2007-62915). We would like to thank Florent Rivals for correcting the French abstract and captions. We also want to thank Brian Edwards, Benjamin Schwartz and two anonymous reviewers for their very helpful comments.
1The relationship between environmental and cultural changes is a primary topic in paleoanthropological research. The impact of climatic oscillations on the density, availability and predictability of resources can modify human adaptive strategies and, as a consequence, the archeological record. The Late Pleistocene and Early Holocene are characterized by accelerated archeological change defined by two main cultural transitions: 1) from the Middle to the Upper Paleolithic and 2) from the Upper Paleolithic to the Mesolithic. Regarding the former, several studies have established a relationship between the Middle/Upper Paleolithic divide and the environmental instability of the Marine Isotope Stage (MIS) 3 (Mellars, 1998; d’Errico & Sánchez Goñi, 2003; Sepulchre et al., 2007; Carto et al., 2009; Golovanova et al., 2010; Müller et al., 2011). Regarding the latter, the passage from the Upper Paleolithic to the Mesolithic has been associated with the environmental changes characterizing the Pleistocene/Holocene boundary (Straus, 1995; Dolukhanov, 1997; Aura et al., 1998; Tankersley & Kuzmin, 1998; Jones, 2007). Moreover, paleoenvironmental arguments have been used to explain the cultural transformations within each of these cultural periods (Rolland & Dibble, 1990; Stiner & Kuhn, 1992; Blades, 2001; Giles et al., 2003; Straus, 2005).
2Paleoenvironmental changes can have profound consequences on another factor related to the reconstruction of cultural change: the preservation of the archeological evidence. Erosive events or sedimentary hiatuses are climatically induced factors that may condition the formation and conservation of archeological assemblages. Some cultural horizons may be underrepresented at a regional level, giving a punctuated appearance to the cultural evolution. These temporal hiatuses have been noted at different regions of the Iberian Peninsula (Villaverde & Roman, 2004; Aubry et al., 2011), affecting the Upper Paleolithic sequence in particular.
3To test the consequences of climatic changes on cultural dynamics, it is important to achieve a good correlation between the paleoenvironmental and archeological data. One of the main barriers to this goal is the difference between the resolutions of the paleoclimatic marine and ice core and terrestrial archeological records. Late Pleistocene and Holocene environmental changes are known by high-resolution records that allow the identification of millennial-scale climatic events (Dansgaard et al., 1984; Bond et al., 1993; Dansgaard et al., 1993; Sánchez Goñi et al., 2000). However, the cultural changes are based on coarse-grained continental sequences that can only be correlated with large paleoclimatic periods. The archeological sequences covering most of the last glacial period with millennial-scale resolution are rare. We present in this paper one of these rare sequences, that of the Cinglera del Capelló.
4We present a synthesis of the stratigraphical, paleoenvironmental, radiometric and archeological data from the Capelló sites, attempting to correlate this sequence with the millennial-scale events identified at both ice and marine cores. The synthesis allows for the construction of a general framework of cultural and environmental evolution for the whole Capelló, spanning the second half of the Late Pleistocene and the Early Holocene (from ca. 70 to ca. 7 ka).
5The Cinglera del Capelló (Capelló cliff) is a 1-km-long tufa-draped cliff located above the west bank of the Anoia River (Capellades, Barcelona, Spain) (fig. 1). It is the border of a Pleistocene lacustrine tufa, formed over the slates of the Paleozoic basement. These tufas are nearly 50 m thick and cover a surface of 5 km2. Their origin is linked to the springs of the regional aquifer, and their base has yielded U-series dates older than 350 ka (Bischoff et al., 1988). The Capelló presents many rock-shelters (abrics), some of them containing Late Pleistocene and Early Holocene deposits with Middle Paleolithic (MP), Upper Paleolithic (UP) and Mesolithic remains. A research project underway since 1983 has focused on the excavations at the Romaní rock-shelter (Carbonell et al., 1994; Vaquero et al., 2001); however, within the last several years, excavation has been extended to other sites. This study is based on a multiproxy analysis of four rock-shelters: Romaní, Pinyons, Can Manel and Agut (fig. 2). Another Cinglera site, the Consagració rock-shelter, will not be presented here because the most current data has already been published (Giralt & Julià, 1996).
6The Romaní rock-shelter was discovered and excavated by Amador Romaní at the beginning of the 20th century. A second excavation was carried out by Eduard Ripoll between 1956 and 1961 (de Lumley & Ripoll, 1962). The present project began in 1983. Fourteen archeological levels (from A to O) have been excavated so far. The discovery and first excavations in the Agut rock-shelter were also carried out by Amador Romaní from 1910 to 1914. Leslie Freeman and colleagues carried out additional excavation in 1976 (González Echegaray & Freeman, 1998). However, all data presented here correspond to excavations carried out between 1999 and 2001 (Vaquero et al., 2002). Pinyons was also discovered and first excavated by Amador Romaní, although only the uppermost archeological level was affected. In 2000 and 2001, a 3 x 2 m sounding pit was excavated. Can Manel is the only site in this paper that was not discovered by Amador Romaní. Four excavation seasons were carried out between 2003 and 2006, during which a 20 m2 area was excavated.
7The multiproxy approach to these sequences incorporates a) paleoenvironmental reconstruction through lithostratigraphic, pollen and charcoal analyses, b) radiometric dating, mainly by U-series but also by 14C AMS and TL, and c) the cultural adscription of the archeological layers. The geoarcheological study is based on the lithostratigraphic description of profiles and the identification of sedimentary units. Environmental conditions are inferred from the formation processes associated with the main lithofacies.
Fig. 2: (A) Topographic map of the Cinglera del Capello showing the location of the four sites discussed in the text. (B) General view of the Cinglera: 1/ Romaní, 2/ Pinyons, 3/ Agut, 4/ Can Manel.
8The sediments were treated using the method proposed by Goeury & de Beaulieu (1979), slightly modified by the elimination of acetolysis according to Girard & Renault-Miskovsky (1969), and the protocol developed by Burjachs et al. (2003). The sediments were reduced by HCl, humic acids washed in OHNa, floated in dense Thoulet liquor, filtered with glass fibre filter and destroyed the filter with 70 % HF. All steps were followed by washes in H2O distilled water and centrifuging. The final residue of each sample was mounted in preparation for biological microscopy diluted with glycerin.
9Pollen concentration (PC), calculated as pollen grains per gram of dried sediment (volumetric method; Loublier, 1978), was generally low (ca. 100 grains/g). For this reason, 40 g of sediment were taken for pollen chemical treatment. The PC reached a maximum of about 3000 grains/g at the top and bottom of the Capelló sequence. The number of counted pollens varies between 152 and 482 per sample. The pollen diagram and calculations for CONISS zonation has been developed with Tilia software (Grimm, 1987, 1991-2011).
Fig. 1: Location and geomorphic setting of the Cinglera del Capelló.
1/ Plutonic Rock, 2/ Paleozoic, 3/ Mesozoic, 4/ Cenozoic, 5/ Quaternary tufa, 6/ Quaternary, 7/ Anticline, 8/ Syncline, 9/ Thrust fault, 10/ Inverse fault, 11/ Fault
10Charcoal fragments were picked by hand during excavation. For charcoal identification, the remains were fragmented to obtain the three wood anatomy sections and observed with a reflected light microscope using dark and light fields. The samples were classified by comparison with reference collections and a wood anatomy atlas (Schweingruber, 1990). The charcoal assemblages were quantified by counting the number of fragments or noting the presence/absence of the different taxa.
11Tufas provide good material for U-series dating. Carbonate-rich karstic-discharge waters from the Capellades springs have flowed across the plateau and over the cliff throughout the past 100 ka. The waters are highly charged with CO2 and elevated, but trace, amounts of uranium (ca. 1 ppb). Isotopic disequilibrium between 230Th and 234U has been widely applied to the dating of solids precipitated from aqueous solutions under a variety of conditions. Samples of the highest purity were processed using standard techniques (Bischoff & Fitzpatrick, 1991; Bischoff et al., 1994, 2005). The samples were dissolved in strong HNO3 solutions, after which artificial spikes were added and U and Th were isolated and purified using anion exchange resins. In earlier stages of this project, U and Th isotopes were analyzed by alpha spectrometry (Bischoff et al., 1988), thermal ionization mass spectrometry (TIMS, Bischoff et al., 1994) and ion-microprobe mass spectrometry (SHRIMP RG) at the U.S. Geological Survey (Bischoff et al., 2005). In later years, they were analyzed by inductively coupled plasma mass-spectrometry (ICPMS) by Ross Williams of the Lawrence Livermore National Laboratory. The coeval sample-to-sample reproducibility was approximately ± 3 percent, indicating a general “geological” integrity of the samples in which a small amount of internal post-depositional migration of uranium has likely taken place. Over 60 tufa samples from the Romaní sequence have been dated. U-series dating has also been performed at the other Cinglera sites: Pinyons (5 samples), Can Manel (6 samples) and Agut (7 samples).
12Twenty-four organic samples have been dated by 14C AMS: 12 from Romaní, 6 from Can Manel, 5 from Agut and 1 from Pinyons. Except for one dating from Can Manel, all 14C AMS dating has been made on charcoal. In general, these samples were recovered during surface excavations. Only samples from levels A and B of Romaní were taken from remnant stratigraphic profiles. AMS dating was carried out at four different laboratories: Arizona Accelerator Mass Spectrometry Laboratory (Romaní), Beta Analytic (Can Manel), Oxford Radiocarbon Accelerator Unit (Agut and Pinyons), and Rafter Radiocarbon Laboratory (Romaní). Lastly, three luminescence dates were obtained from Romaní and Pinyons. TL and OSL dating was carried out at the Laboratorio de Datación y Radioquímica de la Universidad Autónoma de Madrid.
13Finally, the archeological horizons have been culturally characterized through the technological and typological analyses of lithic assemblages.
14All deposits found in the different rock-shelters show similar formation processes. four main lithofacies can be distinguished: a) moss-generated tufa and algal mats (stromatolites) (viles & goudie, 1990); b) gravels and fallen blocks that are found in poorly stratified discontinuous beds; c) conglomerates, carbonatic sands and silts appearing in graded and horizontal discontinuous beds; and d) red silts and sands of aeolian origin that form irregular and massive stratified beds. these lithofacies provide some clues to the climatic context. lithofacies a) and c) are related to dripping water and are therefore indicative of wet conditions. in contrast, lithofacies d) corresponds to arid and semiarid periods in which the capellades springs were inactive or diverted.
15Two main lithological units characterize the stratigraphy. The uppermost unit (Sequence 0), composed of 3 m of red silts and sands, is an aeolian deposit and was largely removed by early excavations. The basal unit, with a minimum thickness of 17 m, is mainly composed of tufas, tufa gravels, carbonatic sands and silts (Bischoff et al., 1988; Carbonell et al., 1994; Giralt & Julià, 1996). Only the uppermost 9 m of this unit will be described in this paper because the lower 8 m are known only by a narrow hand-dug shaft (the Pou Romaní) excavated at the beginning of the 20th century. There are some differences between the outcrop used in this paper (Coveta Nord outcrop) and that previously published (e.g., Carbonell et al., 1994; Giralt & Julià, 1996), which corresponded to Pou Romaní. As the modern excavations have proceeded, large exposures have been available, allowing a more detailed stratigraphic study. The Coveta Nord outcrop corresponds to one of these exposures. Four units have been distinguished from bottom to top (fig. 3):
-
Unit 4 (at least 2 m thick). Close to the wall, gravel wedges alternate with tufa domes. Below the cornice and beyond the dripline, conglomerates and sands alternate with gravels. Archeological level L is included in this sequence.
-
Unit 3 (2 m). The basal beds onlap toward the shelter wall and consist of a moss-generated tufa interbedded with thin gravel lenses. A disconformity in the upper third part of this unit represents erosion. Archeological level K.
-
Unit 2 (2.5 m). The bottom consists of two beds of carbonatic sands and moss-generated tufa interbedded with gravels and fallen blocks. The upper part fines upward at the dripline and dips towards the shelter wall. The uppermost layer shows the first occurrence of loess. Archeological levels E, F-G, H, I and Ja-Jb are included in this package.
-
Unit 1 (3 m). This unit consists of fallen blocks capped by massive domal laminated tufas toward the shelter wall. Two channel deposits contain beds of pisoids and oncolites, carbonatic sands and silts. The upper half of this unit consists of stratified and cemented gravels, carbonatic sands and tufa. It is caped by a second layer of loess (Unit 0). Archeological levels A, B, C and D.
Fig. 3: Lithostratigraphy of Romaní, Pinyons, Can Manel and Agut sections, showing the stratigraphic units and the location of the archeological layers.
silts and sands; f/ stalagmitic dome; g/ bounding surfaces; h/ sequence limits.
Lithology: a/ tufa, conglomerates and carbonatic sands; b/ briophyte bioconstruction; c/ travertinic and angular gravels; d/ laminated algal tufas; e/ red
16The sequence has a minimum thickness of 6 m and is composed of laminated algal and moss-generated tufa, oncolites, carbonatic sands and red silty sands. Tufas are dominant at the bottom and top of the sequence, but red aeolian silty sands are the dominant facies in the middle part of the stratigraphy. From bottom to top, 7 stratigraphic units have been defined:
-
Unit 7 (2 m). Dome-shaped layers of moss-generated tufas interbedded with sands and gravels.
-
Unit 6 (0.5 m). Stratified red silty sands. Archeological level C.
-
Unit 5 (0.8 m). Crudely stratified tufas, weathered gravels and sands.
-
Unit 4 (0.8 m). Stratified red silty sands, carbonatic sands, gravels and blocks.
-
Unit 3 (1.2 m). Crudely stratified red silty sands, carbonatic sands and weathered gravels. Archeological level B.
-
Unit 2 (0.4 m). Aggregated-organic rich bed of red silty sands and fallen blocks. Archeological level A.
-
Unit 1 (1.7 m). Laminated and thin-bedded tufas.
17The sequence exposed at Can Manel is 9 m thick with 7 lithostratigraphic units:
-
Unit 7. Crudely stratified red silty sands above a tufa dome. Archeological level F.
-
Unit 6 (1.5 m). Microstratified carbonatic sands, stem tufas with fallen blocks and red silty sands. Archeological level E.
-
Unit 5 (0.5 to 1 m). Stratified carbonatic sands with gravels and blocks, stem tufas with fallen blocks.
-
Unit 4 (0.4 m). Stratified carbonatic sands and gravels with red silty sands. Archeological level D.
-
Unit 3 (2.5 m). Poorly stratified coarse gravel, carbonatic sands and fallen blocks and well-stratified graded gravels, granules, carbonatic sands and tufa beds and blocks. The upper part contains a microstratified bed of red sands with blocks. Archeological levels C-B.
-
Unit 2 (2.5 m). Domic coset of thin bedded tufa of mosses, stems and stalks.
-
Unit 1 (1.5 m). Fining upward coset of carbonatic sands interrupted by a microaggregated organic rich paleosol and upper set of fine bedded tufas bounded by domic small-scale laminated tufas. Archeological level A.
18The stratigraphic sequence at Agut is near 3.7 m thick and subdivided into three units from bottom to top:
-
Unit 3 (approx. 1.3 m). Crudely stratified gravel, carbonatic sands and blocks above basal fallen blocks.
-
Unit 2 (1 m). Crudely stratified and inclined carbonatic sands and gravels with fallen blocks at the top.
-
Unit 1 (3 m). Tufa, gravel, carbonatic sands and organic-rich lutites with an upper domic tufa. All the archeological layers (4.5 and 4.7) correspond to this unit.
19The oldest dates correspond to the Romaní sequence, which has yielded a series of U/Th dates between 70 and 40 ka cal (tab. 1). The plot of U-series dates against depth is relatively smooth and indicates a rather constant sedimentation rate for the tufa of a very rapid 570 cm/ka (fig. 4). A series of 14C AMS dates from the upper archeological levels (A to J) has also been obtained (tab. 2), ranging from 23,160 ± 490 (level A) to 47,100 ± 2,100 14C years BP (level J). With the exception of some of the 14C dates from the uppermost levels, there is a good agreement between the U-series and radiocarbon dating. The loess at the top of the Romaní stratigraphy (unit 0) has been dated by OSL at 29,500 ± 1,900 years (tab. 3).
20U/Th dating shows that the Pinyons sequence spans from 52,700 ± 1,900 years (unit 7) to 8,800 ± 600 years (unit 1) (tab. 4). The absence of inverted dates, as well as the high uranium content (0.6 to 1.7 ppm) and low detrital contamination (230Th/232Th ratios ranging from 35 to 228, except for the sample 00-258), indicate that the results are highly reliable. The accuracy of these dates is confirmed by the AMS radiocarbon date from level C (tab. 2). Furthermore, two luminescence dates from units 3 and 4 are the in correct order relative to the U/Th and 14C dates (tab. 3).
21The bottom of the Can Manel sequence is a tufaceous surface dated at 39,388 ± 264 years, while three 238U/230Th dates from units 1 and 2 suggest that these uppermost layers were formed during the Late Glacial (tab. 4). The uppermost age at 12,700 ± 160 years is the most reliable, being relatively free of detrital contamination (230Th/232Th ratio = 40). The two underlying dates have significant detrital contamination (230Th/232Th ratios of 3 and 4), rendering their corrected dates of 11,900 ± 1600 and 13,600 ± 700 years as only semiquantitative. The 14C AMS dating of the archeological levels can be seen in table 2. Dates are lacking for the lowermost archeological unit (level F), but a shell from the immediately overlying layer yielded a date of 29,420 ± 250 14C years BP.
22The isotopic composition of the Agut tufas is very similar to that from Romaní, with uranium contents between 1 and 2 ppm and 234U/238U ratios of 2 to 3 (Bischoff et al., 1988; Vaquero et al., 2002). The U/Th dates range from 7,731 ± 370 years (level 4.4) to 12,672 1200 years (level 4.8) (tab. 5). The thorium ratio (230Th/232Th) is greater than 10, indicating that most samples are relatively free of detrital contamination. The 14C AMS dates (tab. 2) are internally coherent and consistent with the U-series dates. The overall sedimentation rate is 273 cm/ka.
Fig. 4: Plot of U/Th dates of Romaní against stratigraphic depth.
Tab. 1: Analyses of U-series isotopes and derived dates on tufa samples from Abric Romani.
Sample numbers with superscripts a (alpha spectrometry) and i (ICPMS) are previously unpublished dates. Sample numbers 90-AR3 and AR4 are isochron dates (averaged) from multiple analyses of co-eval samples reported in Bischoff et al. (1994). All other samples are analyzed by alpha spectrometry and reported in Bischoff et al. (1988).
Tab. 2: 14C (AMS) dates from the Cinglera del Capelló sites. 2σ calibration has been achieved using IntCal09 (Reimer et al., 2009).
Tab. 3: Unpublished datings obtained by optical stimulated luminescence (OSL) and thermoluminescence (TL) on silty-sand sediments from Pinyons and Romaní.
In both methods, the accumulated dose and annual dose were determined by standard techniques. The resulting ages fit in stratigraphic order with U-series ages from bracketing units (see tab. 4).
Tab. 4: Unpublished analyses of U-series isotopes and derived ages from tufa samples of Can Manel and Pinyons rock-shelters
Sample numbers with superscripts a, i and m were respectively analyzed by alpha spectrometry, ICPMS and ion-microprobe.
Tab. 5: Analyses of U-series isotopes and derived ages from tufa samples of Abric Agut after Vaquero et al. (2002).
Sample numbers with superscripts a and t were respectively analyzed by alpha spectrometry and TIMS.
23Romaní and Agut provided the most reliable pollen results, whereas many Pinyons units lacked significant pollen (fig. 5). According to the pollen data, two main units can be recognized in the Cinglera sequence. Unit A (70-40 ka) corresponds to the Romaní sequence and the bottom of Pinyons. It is subdivided in two large subunits (A1 and A2) according to the cluster analysis (CONISS). The Romaní analysis was presented elsewhere (Burjachs & Julià, 1994, 1996; Burjachs et al., 1996), and we will only summarize the main results here. The high resolution of this deposit allowed the recognition of five pollen zones between 70 and 40 ka. Zone A1a1 (70-65 ka) shows two warm phases with thermophilous taxa. Conifers (Pinus and cf. Juniperus) dominate, followed by Mediterranean taxa (Quercus evergreen, Olea) and Poaceae. Zone A1a2 (65-57 ka) is characterized by the dominance of Poaceae, Artemisia and Pinus, indicating a cold phase. Zone A1a3 (57-49 ka) shows short and abrupt oscillations. This phase is characterized by dominance of Pinus, followed by abrupt advances and setbacks of mesic taxa. Artemisia decreases and gramineous increase, while Asteraceae begins to become important. Cold and arid conditions characterize pollen zone A1b (49-45 ka); the low percentages of trees and the dominance of Asteraceae, Poaceae and Artemisia suggest steppe vegetation. Pollen zone A2 (45-40 ka) is characterized by warming, humid conditions. The landscape was formed by open pinegroves that were gradually colonized by cf. Juniperus. Mesophilous trees and heaths (cf. Erica, Cistus) increased, and the typical thermophilous communities of the Mediterranean appeared.
24Romaní hearths yielded large amounts of charcoal (fig. 6). From level O to D, Pinus sylvestris dominated, reflecting open woodland during a cold and dry climate. Levels O and N differ slightly, with the appearance of cf. Juniperus, Prunus and Rhamnus cathartica/saxatilis. In level D, a few fragments of deciduous species such as Acer, deciduous Quercus and other undetermined angiosperm reflect a change that corresponds to the climatic improvement also suggested by pollen zone A2.
25Unit B corresponds to the Pinyons and Agut sequences. Pollen data from Pinyons show a clear difference between the Upper Pleistocene (stratigraphic units 7 to 2) and Holocene (unit 1). The Pleistocene landscape was dominated by pinewoods, among which grasslands of gramineous, Artemisia, Asteraceae and Chenopodiaceae dominate, representing cold conditions. Refugia of mesic and Mediterranean taxa (evergreen Quercus, deciduous Quercus, Olea) are also observed. Unit 6 (pollen zone B1a1) shows a cold landscape of semi-open pinewoods (cf. Pinus sylvestris) that allow the proliferation of seasonal gramineous and Asteraceae grasslands. During unit 5, conditions improved, as evidenced by the presence of the meso-termophylous Quercus. In units 3 and 4 (pollen zone B1a2), the arboreal cover is reduced, even at the level of conifers, and there is a significant presence of Betula. The beginning of the Holocene (unit 1) corresponds to the expansion of the thermo-mesophilous taxa, with important shrub diversity (Erica, Buxus, Coriaria, Cistus, Phillyrea, Vitis) and a decrease in pinewoods (pollen zone B2).
26The Pinyons site yielded several poorly preserved charcoals. Taxa include Pinus sylvestris type, Rhamnus cathartica/saxatilis and other conifers and angiosperms. This finding indicates the same environment as in Romaní, where pine is dominant. The charcoal record from Can Manel shows the presence of Pinus sylvestris type. Other taxa, including Acer, Prunus, Juniperus, Pomoidea and Salix/Populus, are also present. At the base of the sequence, Pinus is more important, whereas mesophilous taxa spread out on the top. This upper sequence shows the transitional vegetation between Romaní, dominated by pine, and Agut, dominated by mesophilous taxa and junipers.
27The Agut succession spans the Last Glacial and beginning of the Holocene (pollen zones B1a3 to B2), with pinewoods constituting the majority of the flora. Archeological complex 4.7 (unit Agut 1) coincides with a thermal (Mediterranean and mesic taxa) and humid (riverside arboreal taxa) phase. The shrubs (cf. Erica, Buxus, Rhamnus) increase, as do Asteraceae and Chenopodiaceae. The tufa above level 4.7 indicates the deterioration of the deciduous forests in favor of conifers. Level 4.5 represents a cold phase. Mediterranean taxa are scarce, and human activity is observed again in terms of the presence of Asteraceae.
28The charcoal remains correspond to levels 4.5 and 4.7 only. Pinewood is absent, and the presence of Juniperus, Acer, Rhamnus, Prunus, Maloideae and Hedera reflects a shift to more humid, milder conditions. This assemblage is characterized by colonizer species, which are the pioneer vegetal cover that evolve into the Holocene deciduous forest. During this period, spring water was more abundant and affected the pine forests in terms of the slope.
29
Fig. 5: Pollen diagram from the Cinglera del Capelló.
The lower part of the sequence (70 to 40 ka) corresponds to Romani, the central part (40 to 15 ka) to Pinyons and the upper part (15 to 7 ka) to Agut.
30The Capelló archeological sequence spans the Middle Paleolithic, the Upper Paleolithic and the Mesolithic. The Middle Paleolithic is mainly represented by the Romaní levels, although it could also be present at the bottom of Pinyons. Flint was the primary raw material for toolmaking, although quartz and limestone were also used. Discoidal and levallois methods are dominant in reduction sequences, and denticulates are the best-represented artifacts. Horse and red deer are the dominant fauna, although bovid and rhinoceros bones are found in some layers. Artifacts were scarce in the lowermost level (C) of Pinyons, but the technical characteristics and absence of blade technologies, support MP attribution. According to the 14C AMS date of 37,900 ± 500 14C years BP, it could be correlated with the uppermost MP of Romaní.
31Upper Paleolithic assemblages are represented at Romaní, Pinyons and Can Manel; however, they are generally sparse and culturally non-diagnostic. The most significant Early Upper Paleolithic assemblage was found in level A of Romaní. Although it was entirely removed by early excavators, Dufour bladelets indicate that this level corresponds to the Archaic Aurignacian (Laplace, 1962). According to dates, Aurignacian culture may also be present in level F of Can Manel. Typical Aurignacian tools have yet to be recovered, but some bladelets indicate that the assemblage is UP. Level E of Can Manel should also be attributed, according to the obtained dates, to the EUP, although no diagnostic artifact has been found.
32The Late Upper Paleolithic has been documented at Pinyons and Can Manel. Level B of Pinyons was found in the uppermost silts of unit 2. A precise cultural attribution is not possible due to the absence of diagnostic artifacts, but some blades indicate UP. Level A, described by Romaní under the tufas that sealed the deposit (unit 1), was removed by clandestine excavators and only some isolated pieces have been recovered. Lithics from the upper levels of Can Manel (A to D) are attributed to the Late UP (most likely Magdalenian). The artifacts from level B are numerous, including some typical UP tools, such as endscrapers and burins.
33Finally, the archeological horizons from Agut are Mesolithic. The technical and typological characteristics represent a return to a flake-based technology. Discoidal knapping is dominant, and there is no evidence of blade techniques. Denticulates are dominant, and a few worked cobbles have been found. This sequence should be attributed to the Mesolithic of macrolithic facies. This cultural stage is well defined in Mediterranean Spain and is a clear departure from the previous UP technical traditions. The fauna are dominated by rabbits but also includes other species, such as red deer, goats, turtles and wolves in very low percentages.
Fig. 6: Charcoal analysis diagram showing percentages based on determined fragments.
Some sample percentages are inflated due to the low number of fragments.
34Correlation between sedimentary and vegetation data allows nine successive stages to be identified in the formation of the Cinglera del Capelló sequence (tab. 6). The oldest stages (Capelló IX and VIII) correspond to the bottom of the Romaní sequence and are known only through pollen data. Capelló IX (70-65 ka) corresponds with pollen zone A1a1 and shows two strong warm pulses. The following stage, Capelló VIII (65-57 ka), corresponds to pollen zone A1a2 and shows an open, park-type landscape, where pine and prairie/steppe dominated.
35During stage Capelló VII (pollen zone A1a3), between 57 and 49 ka, the environment was characterized by abrupt changes from Mediterranean-type temperate groves to arid steppe. However, sedimentary data indicate a dominance of interstadial conditions, given that tufa deposits are dominant (Henning et al., 1983). This stage corresponds to Romaní units 4 and 3, as well as Pinyons unit 7. Tufa from the base of Romaní unit 3 suggests an interstadial stage characterized by a low seasonality in dripping activity. In the top of unit 3, the increase of clastic strata indicates a higher seasonality in rainfall. The small mammal assemblages from these layers of Romaní are dominated by taxa related to an open forest environment and species requiring high humidity (Burjachs et al., 2012). The travertine layers at the bottom of the Consagració sequence (units 5 to 3 in Giralt & Julià, 1996), dated between 47,000 ± 2,700 and 54,700 ± 3,200 years, also indicate that interstadial conditions were dominant during this period. The oldest MP levels so far excavated in the Abric Romaní correspond to this time span.
36The following stage, Capelló VI (49-45 ka), is basically correlated to Romaní unit 2. Pollen subzone A1b indicates that it corresponds to a particularly cold period. The landscape became a steppe of Asteraceae, Artemisia and gramineous, with scarce pines. However, sedimentary facies indicate interstadial conditions for most of this unit. The onlap pattern in the basal part of unit 2 exhibits decarbonation in its upper surface, suggesting an interstadial stage (Vallverdú, 2002). Coarser upward thickening of slab gravels against the shelter’s wall, interbedded with tufa domal deposits at the dripline, suggests dripping activity in the upper part of this unit. Harsh conditions are only evident at the top of unit 2, where an irregular bed of red sands suggests aeolian activity and the absence of dripping. The abundance of the natterjack toad (Bufo calamita) in level E also highlights these dry conditions (Burjachs et al., 2012). This aeolian sedimentation is capped by an abrupt lithologic change, marked by roof collapse and biochemical deposits (base of unit 1). Romaní archeological levels J, I, H, F-G and E correspond to this time span.
37The stage Capelló V (45-40 ka) is an interstadial phase that corresponds to Romaní unit 1 and pollen subzone A2. The thick tufas and speleothems in the base of this sequence also indicate a warm, wet period. This abrupt lithological change suggests a correlation with the abrupt warming at the end of the Bond cycles (Bond et al., 1993). The end of this warm period is also represented by the tufas at the bottom of the Can Manel stratigraphy, which have been dated to 39,388 ± 264 years. At Pinyons, this stage coincides with a major temporal hiatus between units 6 and 7. This discontinuity occurs in domic tufas, suggesting non-deposition. The final MP levels of Romaní (B, C and D) correspond to this period.
38The end of this warm phase and the start of Capelló IV stage (40-23 ka) mark the first great discontinuity in the Capelló sequence. It is the beginning of the pollen subzone B1a1 that corresponds to Pinyons units 5 and 6, Romaní unit 0 and Can Manel units 7 to 5. Pollen and charcoal are scarce, but the data indicate the continuation of pine groves, with few mesic taxa and a steppe domain (Artemisia, Asteraceae and Poaceae). Pteridophyta are practically absent, symptomatic of the aridity of this time. This change is clear in sedimentary dynamics. From this moment and until the end of the Pleistocene, tufas become less important and the formation of aeolian deposits dominates. Unit 2 of Consagració, consisting of silts and red sands (Giralt & Julià, 1996), also shows this change in sedimentary conditions. Stratigraphic discontinuities and temporal gaps are more important, such as that recorded in the Pinyons units 5/6 boundary between 43,185 - 41,745 and 32,800 ± 1,000 cal. years BP. Archeological information is also scarce. However, we note some inconsistencies between Pinyons and other sites. According to the date of unit 6 of Pinyons, the formation of aeolian deposits in this site started in approximately 43,185 - 41,745 cal. years BP, whereas in Romaní and Can Manel tufas have been dated to ca. 39-40 ka. Although the lithic assemblage from unit 6 is small, its technical features suggest a MP attribution, which reinforces the correlation with the top of the Romaní MP sequence. This difference concerning the beginning of the aeolian sedimentation may be explained by local factors affecting each rock-shelter.
39This climatic and sedimentary change is associated with the MP/UP transition. Aurignacian is represented by the Archaic Aurignacian from Romaní level A and level F from Can Manel. Level A of Romaní shows an average AMS date of 36,800 14C years BP and is located in the boundary between units 1 and 0. The chronology of the purported Aurignacian assemblage of Can Manel is uncertain because the date of 29,420 ± 250 14C years BP (34,655-33,380 cal. years BP) corresponds to a sterile layer above the archeological horizon.
40Above the Aurignacian from Romaní and Can Manel, the only EUP layer in the Cinglera is level E of Can Manel, located below the tufa dated at 33,742 ± 260 years. This dating would suggest a Late Aurignacian adscription, but the 14C AMS dates from the archeological horizon – between 31,485 and 29,805 cal. years BP - suggest a Gravettian attribution instead. Lithics are of little help to resolve this question because the assemblage is small and non-diagnostic.
41Approximately at 32-34 ka, dripping sedimentation is again documented in Pinyons and Can Manel, corresponding to the last tufas formed in the Cinglera before the Tardiglacial, which suggests that an extraordinarily harsh environment was dominant between ca. 32 and 13 ka. The onset of these drier conditions at ca. 30 ka coincides with the longest sedimentary hiatuses, although their duration varies by site. At Pinyons, there is a ca. 9 ka hiatus between units 5 and 4 - dated by TL to 23,702 ± 1591 years, whereas the temporal gap between units 5 and 4 at Can Manel is even larger (ca. 15 ka). The discontinuity in the boundary of isotope stages 3/2 suggests condensed sedimentary successions (Dabrio & Hernando, 2003) at Can Manel unit 4 and Pinyons units 4 to 2.
42The Capelló III stage (23-17 ka) is particularly poorly known. The pollen information is based on only one sample from Pinyons (pollen zone B1a2). The Late Glacial Maximum (LGM) is mainly documented at units 3 and 4 of Pinyons. Red silts of Pinyons suggest partial sedimentation and erosion, characterized by massive and microstratified beds that are related to semiarid to arid conditions. Reworking is indicated by weathered clasts in Pinyons 3 and 2 units. Human presence has been attested at unit 3 of Pinyons, dated to 19,962 ± 1402 years, but lithic remains are also scarce and culturally undiagnostic. The archeological layer at Can Manel unit 4, dated to 18,645-18,050 cal. years BP, would also correspond to cold and dry conditions.
43The information from the following stage, Capelló II (17-12 ka), is also sparse, although at the end the presence of mesic taxa and Pterydophyta suggests a warming (pollen subzones B1a3 and B1b). Wet conditions after LGM periods are also documented by the sedimentary processes, as observed in the tabular beds of gravels and sands in Can Manel unit 3. The end of the Pleistocene represents the second great turnover in the sedimentary processes at the Cinglera. As in the period previous to Henrich Event 4 (HE 4), thick tufa deposits return, as observed in Can Manel, Pinyons and Agut, although dating indicates that this growth began somewhat later in Pinyons, in the Early Holocene. These sedimentary patterns began at the end of the Pleistocene and continued throughout the Early Holocene. In Can Manel, the bottom of unit 2 is dated at 13,600 ± 700 years and marks the beginning of this sedimentary change. Archeological level C of Can Manel, dated to 14,880 - 13,940 cal. years BP, would correspond to this period. Although their cultural characteristics are yet to be defined, the abundance of artifacts suggests a human impact greater than the earlier UP levels. The number of sites in NE Spain increases sharply for the final stages of the UP and especially in the Upper Magdalenian (13-12 ka onwards), clearly indicating an expansion of human settlement.
44The last stage in the Cinglera sequence, Capelló I (12-7 ka), corresponds to Agut unit 1 and Pinyons unit 1 and is characterized by the expansion of Mediterranean trees (pollen zone B2). Sedimentary processes are characterized by wet regimes in unit 1 of Pinyons and unit 1 of Agut. Domic tufas and tabular beds of calcarenites were dated to roughly the first Holocene altithermal. This was also the case at Consagració, where travertine layers were dated to 9600 ± 1100 years (Giralt & Julià, 1996). The beginning of the Holocene is associated with the second great cultural change, which is defined by the disappearance of UP traditions and a return to a flake-based industry, as observed in Agut.
45The well-constrained, absolute chronology of the Cinglera sequence allows the correlation of paleoenvironmental changes with the Dansgaard-Oeschger (D-O) variability evident in ice core and marine records (Blunier & Brook, 2001; Cacho et al., 1999; Dansgaard et al., 1993; Genty et al., 2003; Grootes et al., 1993) (fig. 7). In some cases where the correlation with specific events is doubtful, the sequence of events may nevertheless be related to the framework of longer-term cooling cycles (“Bond cycles”) (Bond et al., 1993). This is the case for the bottom of the Capelló sequence, which can be correlated with the end of MIS 5 (Capelló IX) and MIS 4 (Capelló VIII).
46The climatic pulsations that characterize the Capelló VII stage can be correlated with the D-O oscillations 17 to 14 (Dansgaard et al., 1984; Dansgaard et al., 1993; Hinnov et al., 2002). The cold conditions of stage VI are associated with HE 5. Dry conditions are particularly evident at the end of this stage, at the top of Romaní unit 2 (archeological level E), whose chronology (ca. 45 ka) is very close to that of the HE 5 (Blunier & Brook, 2001; Sánchez Goñi & Harrison, 2010). Both pollen and sedimentary analyses indicate that HE 5 was particularly rigorous in the Cinglera. It was colder than the former MIS 4 and represents the first loessic layer identified in the Cinglera so far, which indicates previously unknown dry conditions. Other Mediterranean records also reveal the extreme impact of HE 5, which was characterized by a strong decline in rainfall (d’Errico & Sánchez Goñi, 2003; Müller et al., 2011).
47The climate improvement of stage Capelló V correlates to the warming event D-O 12-9, which is one of the best defined episodes in Mediterranean records. It is the most pronounced warm event in the Villars speleothems (Genty et al., 2003, 2010) and appears as a marked phase of forest development in high-resolution pollen records such as MD95-2043 (Sánchez Goñi et al., 2002; Fletcher & Sánchez-Goñi, 2008), Tenaghi Philippon (Müller et al., 2011) and Monticchio Lake (Brauer et al., 2007).
48The cold and dry period at the beginning of the stage Capelló IV that marks the first great break in sedimentary dynamics can be correlated with HE 4. The short temporal range of disconformities in the Romaní stratigraphy suggests that irregular beds of red silts at the top of level A correlate to HE 4; however, the shift to colder and drier conditions is already evident at the bottom (Courty & Vallverdú, 2001). According to the marine record from the Alboran Sea (Cacho et al., 1999; Sánchez Goñi et al., 2002), the climatic conditions during HE 4 were particularly severe in the Mediterranean region, indicating that it was colder and drier that other HE.
49From this great sedimentary change, correlation with millennial-scale events becomes more difficult. However, some layers of tufa from Can Manel and Pinyons allow a tentative correlation with D-O interstadials. Tufas from unit 5 of Pinyons (dated to 32,800 ± 1000 years) and units 5 and 6 of Can Manel (33,742 ± 260 and 35,954 ± 239 years, respectively) suggest wetter events that may be correlated to D-O 7-5. This finding agrees with the data from the Villars stalagmites, whose growth stopped at ca. 30 ka (Genty et al., 2010). High-resolution pollen records also indicate the establishment of a much drier climate from ca. 30 ka, coinciding with the HE 3 event (Brauer et al., 2007; Fletcher & Sánchez-Goñi, 2008; Fletcher et al., 2010).
50The following period (Capelló III) is the least known of the Cinglera sequence, although dates from Pinyons indicate that it can be correlated with the Last Glacial Maximum. The warming event at the end of Capelló II can be related to the Bölling Complex interstadials of the Last Glacial (D-O 1). At Can Manel, wet conditions formed tufas and tabular beds of calcarenites with an upper age close to the beginning of the Younger Dryas. In Agut, the discontinuity between units 2 and 1 most likely resulted from erosion during the Younger Dryas. Finally, Capelló I marks the beginning of the Holocene.
51It is important to stress that HE tend to be associated with important boundaries in the sedimentary sequence. HE 5 represents the first appearance of loessic deposits in the Cinglera sequence at the top of stage Capelló VI; HE 4 marks the beginning of the dominance of eolian sedimentation; HE 3 is associated with the end of tufa formation ca. 30 ka; and HE 1 coincides with the end of the loessic sequence before the resumption of tufa growth in the Cinglera. According to the data from core MD95-2043 (Cacho et al., 1999; Sánchez Goñi et al., 2002), HE were particularly outstanding in the Mediterranean region because they were colder and drier than other stadial periods. Dryness would have especially profound consequences in the sedimentary dynamics of the Cinglera due to the primary role played by tufa formation. The Cinglera system would be therefore particularly sensitive to major decreases in rainfall values.
Tab. 6: Formation stages of the Capelló sequence and correlation between pollen zones, sedimentary units and archeological layers.
Fig. 7: Correlation of mineralogical, pollen and archeological sequences with the GISP2 oxygen isotope curve.
Pollen curves shown for temperate and mesic taxa. 1/ Capello stages, 2/ Pollen zones, 3/ Stratigraphic units and lithofacies, 4/ Archeological levels and cultural periods. AR/ Abric Romani, PI/ Pinyons, CM/ Can Manel, AG/ Agut.
52The Cinglera del Capelló provides a unique environmental and archeological record spanning the Late Pleistocene and Early Holocene of NE Iberia, between 70 and 7 ka. This sequence is characterized by a high temporal resolution within a well-calibrated chronology, especially compared with other archeological deposits. The well-constrained chronology and investigation at high resolution allows the observation of the rapid climatic oscillations and correlation with millennial-scale paleoenvironmental records.
53The Capelló sequence indicates that the temporal resolution strongly depends on the climatic conditions and the resulting sedimentary dynamics. On the one hand, wet and mild phases, associated with tufas and carbonatic sediments, are represented by thick deposits and show the highest temporal resolution, as observed in Romaní and Agut. These wet periods tend to be overrepresented in the Capelló due to the role played by the Capellades springs in the formation of the deposits. On the other hand, dry events, characterized by the absence of tufa, are associated with sedimentary hiatuses or aeolian layers, in which the temporal resolution is lower. The richest and best preserved archeological assemblages generally correspond to stratigraphic units dominated by tufas and carbonatic sediments, whereas the periods of aeolian sedimentation – except level E of Romaní - are characterized by low-quality, badly preserved assemblages.
54The Cinglera displays the environmental evolution from the last events of MIS 5 to the beginning of the Holocene. Nine climatic-sedimentary stages have been identified. The first stages (Capelló IX to V) span from the end of MIS 5 to ca. 40 ka and correspond to the Romaní stratigraphy – except the uppermost unit 0 - and the bottom of Pinyons. Although different sedimentary facies have been recognized, tufa formation played a primary role over this period. The end of MIS 5 included mild conditions. MIS 4 was characterized by a climate that was not particularly cold or dry, unlike the MIS 4 documented at the rest of Europe. MIS 3 was mild and wet, characterized by the D-O oscillations in the beginning of this interstadial, and a cold and dry phase – including the HE 5 ca. 45 ka - prior to the mild and wet interstadial.
55At ca. 40 ka, coinciding with HE 4, a phase (Capelló IV to II) begins, represented at the top of Romaní, the main part of Can Manel and Pinyons sequences, and the bottom of Agut. Tufas are less common, and the formation of aeolian deposits of red silty sands increases. Conditions were extraordinarily dry during this period, and significant temporal hiatuses are recognized in Pinyons and Can Manel, especially between 33 and 18 ka. Some wet events, correlated to D-O 7-5, are still recorded at the end of MIS 3. After this period, MIS 2 was cold and dry, marked by the dominance of steppe taxa (Artemisia, Asteraceae) and the absence of tufas. There was a gradual recovery of the thermal and pluviometric conditions from ca. 13 ka to the Early Holocene (Capelló I). Dripping water from the Capellades springs was again the main formation process, and thick tufa layers grew once more at Can Manel, Pinyons and Agut.
56In general terms, these climatic-sedimentary phases match the three main cultural periods identified at the Capelló sequence (MP, UP, and Mesolithic), indicating a correlation between environmental conditions and cultural changes. The most intense and continuous human occupation is associated with periods in which tufas dominate. The MP and the Mesolithic, well represented in Romaní and Agut, correspond to such periods. In contrast, the long-term periods in which tufa formation was largely interrupted are characterized by a decrease in human presence. This was the case during the UP, which seems to be characterized by particularly dry conditions. In addition, this depositional context seems less suitable for the preservation of the archeological evidence because temporal hiatuses are longer and more common. The MP/UP boundary would be associated in NE Iberia with a marked environmental turnover. From the viewpoint of formation processes, this change was the most noticeable of the Capelló sequence and can be compared to the environmental change associated with the Pleistocene/Holocene boundary. It seems, therefore, that environmental, cultural and settlement changes were correlated during the Upper Pleistocene and Early Holocene of NE Iberia.