Grotta Romanelli and its chronological and palaeoenvironmental framework.
New fieldwork at Grotta Romanelli allowed a litho-, morpho- and chronostratigraphical assessment of the successions inside and outside the cave. This achievement is of paramount importance since the former chronostratigraphical setting of Grotta Romanelli, placed the beginning of the sedimentation inside the cave during MISs 5e and became a paradigm for Quaternary archaeology, palaeoanthropology, palaeontology, stratigraphy and sea-level history of Italy and the Mediterranean (20). In Grotta Romanelli, the sedimentary evidence points to the deposition within an active humid karst setting (6). Moreover, the position of the cave relative to the sea increases its importance concerning the geodynamics and the eustatic history of this sector of the Mediterranean (44). In fact, the position of the water table in the coastal zone is linked to eustatic sea-level changes (45) and the succession within Grotta Romanelli is one of the older in-cave marine sedimentary records not ‘blown-out’ by high rates of water flow. After the works of Blanc (20, 21) (Suppl.Mat.1) the PSL indicators within the cave (mn1 and mn2 in Fig. 2 and Fig. 3) were considered coeval with and associated to the coarse-grained basal part of the sequence, which included both ISU1 and ISU2, and were attributed to the Last Interglacial. This morpho-stratigraphical assumption was based on the observation that these were the only PSL higher than the present-day sea level along the cliff of Grotta Romanelli. Our research brought new evidence based on litho-stratigraphical, morpho-stratigraphical and geochronological data depicting a different history of Grotta Romanelli during the Quaternary. ISU1 is the only sedimentary RSL marker within Grotta Romanelli, deposited in a coastal marine environment where the finer-grained sediments were preserved only in small conservative features, such as potholes carved within the bedrock, whereas gravels, pebbles and cobbles entered the cave during high-energy events. Moreover, it is found as a lag resting over mn2, the lower marine notch carved on the bedrock. Outside of the cave, above and below the cave entrance, two others sedimentary RSL markers and one PSL (mn3) are for the first time found and described: OSU1 located at ca. 18 m a.s.l. (Fig. 4c), OSU3 found at ca. 3.8 m a.s.l. (Fig. 4d, Fig. 4e) and mn3 at 5.5 m a.s.l (Fig. 4a.2). Therefore, RSL and PSL markers, sediments and marine notches found within Grotta Romanelli cannot be uniquely associated with the Last Interglacial. ISU2 provides the older geochronological constraint for the whole succession. In fact, two U/Th dates, made on the flowstones embedded within ISU2 in the N sector (Fig. 5), provided ages of 360 ± 87 ka and 218.8 ± 34 ka (Table S1 in Suppl.Mat.) respectively from the lower-mid part and the top of the Unit. Furthermore, the flowstone formed on top of the bedrock is dated to 325 ± 39 ka in sect. N buried under ISU2 (Fig. 5). This chronology indicates that the marine deposition of ISU1 should be dated back to the Middle Pleistocene. ISU2 is made of autochthonous debris derived from the ceiling and walls of the cave or from short-distance mass-deposition, re-worked within the cave according to the floor morphology. Partial transport by water run-off is also suggested by the rough trough crossbedding and by the rare presence of re-worked rounded marine pebbles, cannibalised from the underlying ISU1. Periods of non-deposition are indicated by the precipitation of carbonates forming the wavy irregular flowstone on top of the beds (Fig. 3) and by the local presence at the bottom of phytoclastic beds, possibly related to local small ponds with standing carbonate-rich water. The origin of the coarse-grained angular debris of ISU2 is associated with rock degradation and stress release from the cave ceilings and walls, lacking any other palaeoenvironmental markers such as freeze-thaw features or frost slabs. Therefore, the presence and distribution of sediments belonging to ISU2 does not provide specific palaeoclimatic nor palaeoenvironmental indications being mainly associated with the morphology of the cave (6, 46). The deposition of ISU2, according to the U/Th dates, occurred across a long timespan, suggesting that during this part of the Middle Pleistocene the cave was almost empty and that it underwent scarce, mainly autochthonous depositional events, also incorporating the older evidence of human and animals’ frequentation of the cave, including fire features, limestone lithic tools and burnt and unburnt bones (20, 21).
Depositional environment and dynamics suddenly changed after ISU2, which is paraconformably buried by ISU3, macroscopically characterised by reddish colour (“Terre Rosse”). ISU3 is made of thinly layered planarly or cross-bedded silts, clays and sands indicating aqueous deposition through low-energy runoff. The lack of any evidence of in situ clay illuviation or other long-term soil-formation related features suggests that the iron-enrichment and the carbonate leaching of the groundmass is derived from the erosion of leached, argillic limestone soils in the overlying plateau and surrounding slopes. In fact, the sandy and silty mineral fraction of ISU3 is made primarily of quartz suggesting its provenance from the erosion of aeolian deposits such as coastal dunes or other older deposits originally present on top of the plateaux (i.e. OSU1) or from colluvial sediments or soils covering the surrounding landscape. The sediments therefore have been washed into the cave system and transported by run-off and/or karstic waters, indicating the beginning of landscape destabilisation (47) within a warm and humid climatic context as suggested also by the presence of pollen spectra with a consistent amount of olive tree (48). Once arrived inside the cave the sediments were redistributed over short distances by bi-directional water flows both toward the internal and the external part of the cave, by means of predominant sheet-flows (plane parallel bedding) and occasionally within small channels (local cut and fill), possibly related to events of increased water availability (i.e. strong rainfall) alternating with periods of standing water (fining-upward trends and laminated clays) (Fig. S3d in Suppl.Mat.). Contemporaneous physical weathering of the bedrock within the cave provided the coarse-grained autochthonous carbonate fragments. Although the lack of evidence of dense vegetation, phases of non-deposition are highlighted by the common presence of biological voids, including chambers and channels (Fig.S3c in Suppl.Mat.), burrowing features, faecal pellets and locally small-sized coprolites. Rare elements of anthropogenic origin (i.e. burnt bone fragments) and fire-related origin (charcoals and charred plant tissues) (Fig. S3a, Fig. 3b in Suppl.Mat) also suggest a short-distance transport of anthropogenic deposits within the cave system. In the N and S sector of the cave, ISU3 is buried under a roof spall (ISU4 in Fig. 3) covered by a flowstone dated at 112.5 ± 1 ka (N sector, Fig. 5, Table S1 in Suppl.Mat.) and 74 ± 6 ka (S sector, Fig. 5, Table S1 in Suppl.Mat). Assuming that the age of the speleothem might reflect a long period of non-deposition within the cave and dipping vadose waters with precipitation of calcite, these ages post-date the deposition of ISU3 at least at the beginning of the Late Pleistocene. Due to the “colluvial” nature of the sediments inherited by the erosion of Interglacial leached soils ISU3 can be correlated to the early phases of climate deterioration that followed the Last Interglacial (i.e., MISs 5d-b). ISU3 and ISU4 are in turn buried by ISU5, the so-called “Terre Brune '' (due to their brownish colours) of Blanc (20) (Fig. 5). The sedimentary characteristics are very similar to those of ISU3 and point to deposition in a low-energy sheet-wash and runoff environment alternating with phases of standing water within local ponds formed by consecutive events (Fig. S3e, Fig.S3f in Suppl.Mat.), although the lack of major discontinuities within the succession suggests that it was probably laid down at a regular pace. The random distribution of cross bedding dip suggests a direction of sediments transport according to local irregular morphology, i.e. the top of ISU4 that dammed the inner part of the cave, where finer grained sedimentation prevailed, and concentrated the coarser grained deposition in the southernmost sector of the cave (Sect. S in Fig. 3) where a broad and shallow channel was filled by coarser-grained sediments including bedrock boulders and cobbles. Similarly, to ISU3, there is no evidence of soil formation features and the composition suggests the re-working of both ISU4 (limestone gravelly fraction) and ISU3 (colluviated reddish clays, Fig.S3g in Suppl.Mat). The overall brownish colour and the lack of iron sesquioxide’s precipitation, together with the abundance of plant tissues, also suggests a sediments’ source from the erosion of brown soils covering the slopes surrounding the cave washed within the cave. Moreover, the abundance of anthropogenic components, such as charcoal fragments, bones (Fig.S3h in Suppl.Mat) locally with evidence of thermal impact, charcoals and charred plant remains, indicate the reworking of anthropic deposits from other sectors of the cave (35). However, most of the anthropogenic components are reduced in size and with seldom evidence of weathering or secondary impregnation by phosphate (Fig. S3g in Suppl. Mat) and their origin should be related to short-distance re-working within the cave, lacking any features that could be linked to trampling or crushing. ISU5 almost completely filled the cave up to the roof and its chronology can be constrained by the age of the speleothem found on top of ISU3 in the W sector, which provided a U/Th age of 43.3 ± 8 ka (Fig. 5, Table S1 in Suppl. Mat.). Moreover, radiocarbon dates of the same Unit in the S, W and SW sectors of the cave, provided ages for its deposition between 13,6 cal ka BP and 11,4 cal ka BP (31, 49) indicating high sedimentation rates at the Pleistocene-Holocene transition. Such rapid sedimentation occurred by both erosion of the surrounding landscape and related sediments and soil covers and partial re-working of the older succession within the cave, including the autochthonous coarse-grained gravels and cobbles of bedrock produced by the degradation of the cave walls and ceiling.
The reconstruction of sea-level history and palaeo-climatic implications.
Relative sea level (RSL) markers at Grotta Romanelli are of two types: deposits and PSL indicators, mainly marine notches. PSL are found within the cave, from the higher to the lower, namely mn1 at 9.2 m a.s.l., mn2 at 7 m a.s.l., ISU1, as well as the RSL marker represented by ISU1 (Fig. 3, Fig. 4). A PSL marker is also located along the cliff, below the cave entrance (mn3 at 5.5 m a.s.l. in Fig. 4a,Fig. 4d, Fig. 4e). On the same cliff, above and below the cave entrance two RSL markers are respectively located at 18 m (OSU1 in Fig. 4c), and between 3.8 m and 4.8 m a.s.l. (OSU3 in Fig. 4a,Fig. 4d, Fig. 4e). The highest RSL marker (Fig. 4c), OSU1, indicates a high-energy littoral deposition over a marine abrasion platform modelled during a marine highstand. The intermediate RSL, ISU1, found inside the cave is associated with the tidal notches mn1 (9 m a.s.l.) and mn2 (7 m a.s.l.) with Lithophaga burrows. Assuming mn1 and mn2 as belonging to the same transgressive phase (43) and dated before 320 ka BP (the average lower age of the overlaying ISU2) should be associated to the highstand of the MIS 9 (Fig. 6, Fig. 7), as already debated by Mastronuzzi et al., (9) and Antonioli et al., (11) based on correlation with the elevation of other marine notches along the tectonically stable Apulian and Italian coasts. This attribution dismantles the Grotta Romanelli paradigm (20) of the Last Interglacial age for the marine features and deposits found within the cave. A further consequence of this chronological setting is the possible attribution of the older marine record represented by OSU1 to the highstand of the MIS 11 (Fig. 6, Fig. 7), although currently, a standardised review of pre-MIS 5 Mediterranean sea-level proxies is not available (50). This Middle Pleistocene marine high-stand deposit was followed by an erosional regressional phase (MIS 10) marked by typical low-stand and cooler climate indicator debris-slope unit (OSU2) (Fig. 4a, Fig. 4b). Moreover, the finding of the marine terrace located just below the entrance of Grotta Romanelli (Fig. 4d) brings new light to the Last Interglacial sea-level of this sector of the Mediterranean. In fact, the inner tip of the top of the deposit characteristic of high-energy littoral environment reaches 4.8 m a.s.l. and rests over a marine abrasion platform carved over the bedrock at an elevation of ca. 3.8 m a.s.l. The terrace is linked to the mn3 located at 5.5 m a.s.l. in front of the cave entrance (Fig. 4, Fig. 6, Fig. 7) that can be correlated to the MISs 5e, with an elevation within the variability for the same highstand in the same area of the Mediterranean (9, 11, 50). After the Last Interglacial, the sea level lowstand and climate deterioration were followed by the emplacement of the thick debris units (OSU4, 5 and 6 in Fig. 4, Fig. 6, Fig. 7) that partially buried the cave entrance and were dismantled during the history of the excavations. OSU4 accumulated on top of OSU3, the MISs 5e terrace, and the chaotic setting as well as the presence of boulders made of cemented stratified breccias and the occasional presence of rounded pebbles indicates that it originated by rockfalls originated by OSU1 and OSU2 from the overhanging cliff (Fig. 4). The subsequent OSU5 shows a thick layer dipping up to 35 degrees according to the slope, burying both OSU3 and OSU4 and dip below the present-day sea level (Fig. 4), marking its deposition during a well-established lowstand (MIS 4 − 3). Finally, OSU6, made of finer-grained flake-shaped limestone clasts, suggests the onset of cooler conditions along the slopes and the action of freeze-thaw on bedrock.
Grotta Romanelli is one of the few localities in this area of the Mediterranean where 3 RSL markers and PSL indicators are stacked along the same cliff at decreasing elevation from 18 m down to 5,5 m a.s.l. (MIS 11, MIS 9, MISs 5e). This suggests that this part of the Apulian region, although under an overall long-term tectonic stability, underwent some vertical uplift in the last 350 ka. Further chronological constraints for the uppermost and lowermost RSLs would bring new data for the assessment of a more reliable vertical displacement rate of the Apulian coast. The newly assessed chronology of the sedimentary succession filling the cave indicates an early frequentation of the cave under conditions of relative environmental stability, dominated by debris accumulation at very low sedimentation rates (ca. 1 m in 150 ka). The oldest human and faunal frequentation occurred between MIS 9 and MIS 7, earlier than previously supposed (20). The environmental conditions suddenly changed after the Last Interglacial when soil erosion processes on the surrounding slopes brought fine-grained sediments in the cave. A long-lasting hiatus in the sedimentation is marked by the occurrence of several flowstones and speleothems resting on top of ISU3 and ISU 4 spanning from 112 to 43 ka. The complete and final filling of the cave occurred mostly during the MIS 2 − 1 transition, recording human and animal frequentation once again. Due to the intense excavation activities of the past and the consequent lack of sediment, we assume that different sedimentary environments and conditions may have characterised the cave sector closer to the entrance, probably providing more suitable conditions for human settling, as indicated by the evidence of fire, food and frequentation previously reported. In fact, in the inner part of the cave the elements of anthropogenic origin are anyhow present (see also, 35) although at a minor extent and as minor components, as evidenced also by micromorphological observations. The new chronostratigraphic assessment of Grotta Romanelli infilling deposit allows to redefine several important bioevents for European mammal palaeocommunities and add important information for human presence in the Mediterranean area:
- the occurrence of Stephanorhinus hundsheimensis (Toula, 1902) in ISU1 and ISU2 (29) represents the last occurrence of this taxon in Europe, previously attested at about 600 ka (Isernia Faunal Unit). This effectively suggests that a large revision of the Middle Pleistocene rhinoceros is needed to clarify the evolution of S. hundsheimensis;
- the fireplaces reported by Blanc (20, 21) on top of ISU1 should be referred to Middle Pleistocene hominins (H. neanderthalensis or Homo heidelbergensis Schoetensack, 1908) representing some of oldest fireplaces in Europe Mediterranean area (52, 53);
- referring ISU3 to the Last Interglacial (MISs 5e) or to the oldest substages of MIS 5 implies that the mandible of the historical collection of Blanc ascribed to the Eurasian River otter, Lutra lutra (Linnaeus, 1758) (20, 21, 34) is the oldest record of this species in Europe;
- fossils of Palaeoloxodon antiquus (Falconer & Cautley, 1847) and Hippopotamus amphibius Linnaeus, 1758 from ISU3 were considered among the last occurrences of these taxa in Europe, allowing to hypothesise their survival up to MIS 4 − 3 (22). The new chronology of ISU3 suggests that the straight-tusked elephant and the hippopotamus went extinct after MIS 5 (54);
- the lithic industry of the Middle Pleistocene on top of ISU1 (layer K sensu 20, 21, n = 2) and ISU2 (layer I sensu 20; 21, n = 6), and those from the oldest Late Pleistocene ISU3 (layer G sensu 20; 21; n = more than 1100 artefacts) are attributed to the Mousterian (30, 55). Following the new chronology for Grotta Romanelli, these are the oldest evidence of human technology in the Salentine Peninsula (56).