From the outset, the study of stalagmite breakage and speleofact structures in the cave of Saint-Marcel has been anchored in the spatial understanding of the features that can be seen in the cave. Our study includes identifying the relationships between the stalagmites that have been broken and those that have been integrated into the human-made structures, studying the organisation of the structures, and understanding the relationship between the components of the structures and the relative chronology of various human activities and natural processes (natural breakage, calcite deposition). This entire set of information was then used to identify the sectors for absolute dating (see Section 3).
For our study of the geography of the observed cave features, we created a high-resolution map using two different methods: laser scanning was used for entire study sectors and photogrammetry was used for the anthropic structures. In the three study sectors, surveys were carried out using a phase-shift terrestrial scanner (Faro Focus 3D). Several hundred scenes, each with 7 million points (quality of x3), were acquired. They were consolidated using the sphere method and the point clouds were cleaned using 3DReshaper. The floor was separated from the vaults and ceilings and rasterised using CloudCompare (grid size of 1 cm). The raster images were produced from the Digital Terrain Model (DTM) and shaded or slope maps (Jaillet et al., 2017) were generated using QGIS 3.16. In the Colonnes sector, a complementary series of photogrammetric data were collected after the removal of a thick coating of clay. The 231 photos acquired with a Nikon D810 APN were processed in Photoscan to produce a three-dimensional model with 162 million points. This very high-resolution model (sub-millimetre) was used to conduct a detailed analysis of this key sector, where over 90 speleofacts are arranged on the floor. The analysis was conducted after the speleofacts have been cleaned meticulously; the speleofacts were covered by a thick layer of clay that has been deposited by visitor and speleologist traffic due to the presence of clay soils on both sides of the study sectors. The clean-up uncovered broken stalagmites that were hidden beneath the clay; it enabled us to clearly identify connections between broken stalagmites and the relationship between broken stalagmites and the underlying soil.
We used the DTM for archaeo-geomorphological mapping and identified the various natural processes and human activities that were responsible for the current architecture of the Colonnes area. While this mapping exercise is based on the principle and experience of geomorphological mapping that have been acquired during the mapping of the floors of Chauvet and Garma caves (Delannoy and Geneste, 2020; Arias and Ontañon, 2020) it has been enriched with new cartographic terms (e.g., superposition, interlocking, presence of wedges, sealing by stalagmite deposits) to facilitate detailed descriptions of the relationships between various structural elements and those between the elements and the evolution of the cave (Fig. 4). Our aim is to track all the natural processes and human activities that were responsible for the creation of the landscape that is currently visible in the Colonnes area and to highlight (in)coherences in morphologies (Fig. 5).
We mapped all the visible objects that are on the floor, associated each object with a specific process, examined the various links between mapped objects, and established a relative chronology (i.e., object A is older than object B and younger than object C). As a result, a status was conferred to even the minutest element, including stalagmite regrowth at the centimetre scale and micro-impacts from strikes. This cartographic absolutism required us to consider all elements involved in the creation of the landscape and, above all, examine the presence or absence of specific objects, their status, and their history. This approach enabled us to identify the following elements (Fig. 6):
1- In the northern sector of the Salle des Colonnes, there is a major structure consisting of an alignment of 69 broken stalagmites and 23 limestone blocks on the ground. The broken stalagmites that make up this structure are relatively large in size and diameter (the largest ones are 60-100 cm in length and 18-38 cm in diameter) (Fig. 7). Regrowth that is several centimetres thick was found on several recumbent stalagmites. Similar regrowth can be seen on various stalagmite stumps in the vicinity of the structure. Stumps that are blackened by a thick, hardened soot deposit can also be seen in the same area.
2- At the southern end of this first structure, recumbent stalagmites were found on the outer rim of a gour that predates the structure. In addition, a piece of broken column (160 cm long with average diameter of 34 cm) lies on the same outer rim of the gour (generation 3); it is wedged in by smaller broken stalagmites and slightly sealed by a calcite crust (generation 5). This first piece of the column lies perpendicularly to a second piece of the same column (170 cm long with average diameter of 39 cm); the second piece rests against the inner rim of the gour and shows no traces of impact that could be associated with the fall of the column. The end of the column that is on the top of the gour is partially sealed by a calcite deposit that overlays the gour rim (generation 4). (This feature is hereafter referred to as Feature A.)
3- Continuing along the direction of the first structure into the central area of the room, which is rich in speleothems (median sector, Fig. 6), there are numerous recumbent stalagmites (136); they are sealed by a flowstone pavement that is fed by active stalagmites and flowstones (generation 4). In the same central area, there are many smaller broken speleothems (30) at the bottom of the immersed gours; they probably came from the numerous broken stalagmites that are in this area, as well as from more recent extractions of calcite crystals. Numerous stalagmite stumps (on the ground or on the edges of stalagmite edifices) are sealed by recent regrowth (generation 4).
4- At the edge of the central gour, two paths go around the towering stalagmite column; along both paths, there are stalagmite stumps and recumbent stalagmites sealed by calcite crust. The path on the right (west) features an arrangement that forms a “staircase” that crosses the inner rim of the gour. On the outer rim, there is a similar arrangement of recumbent stalagmites.
5- Below this central space, numerous recumbent stalagmites can be seen along the outer rim of the gour and below the central stalagmite edifice (southern margin, Fig. 6). Approximately 80 speleofacts are thus arranged along the left wall of the passage that leads southwards to the main gallery beyond the central stalagmite edifice. These speleofacts are partially sealed by a thick deposit of moonmilk that is present at the bottom of the gours that flow from north to south through the gallery.
Over the entire mapped area, more than 350 recumbent stalagmites were identified (Table 1). This number only takes into account visible elements; elements that are covered by moonmilk and with size and morphology that resemble broken speleothems were not taken into account. Over 172 stalagmite stumps were found in the same geographical area, including 68 with calcite regrowth. It is interesting to note that the diameters of the stumps near the first speleofact structure (northern sector) are generally well below the diameters of the speleofacts that are integrated into the structure. In the same sector, the number of stumps is also far below that of the speleofacts in the structure.
Mapping data reveal a difference between ancient and recent stalagmite breakages on the eastern margin of the Salle des Colonnes. Ancient breakages are always located at a certain height from the ground (5-10 cm), whereas more recent ones are systematically flush with the ground. The presence or absence of regrowths and patinas are also used in the differentiation between ancient and recent breakages. This difference in breakage type is even more striking in the Colonnades sector and suggests that different actions were taken to achieve different goals. In this section of the gallery, people have also extracted crystals by breaking off the edges of gours and flowstones; debris from this looting can be seen at the bottom of the gours and in the hollows dug into the flowstone.
Taken together, these observations provide a clear indication of the anthropic origin of the structures of broken stalagmites. The structures are characterised by speleofacts that were aligned with each other and sometimes wedged in place with smaller elements. An aerial view of the study sector indicates an arrangement that extends over the entire length of the mapped area. In the southern margin and the northern and median sectors, structures are sealed by stalagmite regrowth or crust, flowstone, or moonmilk; this indicates that the structures are linked to a history of past human occupation. The appearance of the speleothems and the actions that led to their breakage differ from those associated with more recent breakage and looting. This difference can also be seen in the numbers of stalagmite stumps and recumbent stalagmites that are visible (Table 1). In the case of recent breakages, there are hardly any stalagmite tips on the ground, indicating people’s desires to take the speleothems with them as trophies. In the case of ancient breakages, almost 40% of the stalagmite tips are visible, on the ground, and integrated into the structures. There is certainly a difference in the intentions behind recent and ancient breakages. The question of the origin of the broken stalagmites still remains: did the stalagmites break naturally or were they broken by humans? Given the large number of broken speleothems in the cave, the possibility of breakages caused by seismicity immediately comes to mind. However, the diversity in the diameter of the broken stalagmites and location of the breakages (top, bottom, or outgrowths of flowstones or large stalagmite edifices) does not correspond to seismic breakage, even if such a hypothesis cannot be totally ruled out for certain breakages, given the proximity of Saint-Marcel cave to active faults with recent activity (Teil earthquake, 11 November, 2019) (Gilli, 2000, Lacave et al., 1999, 2004, 2011; Pons-Branchu et al., 2004). Extraction of the clay substratum could have also caused the stalagmites to break. Several examples can be seen in the cave (in Galerie du Lac and Trou de l'Enfer in the main gallery); however, the breakage mechanism in these sectors is different (e.g., tilting, stress cracks distributed throughout the columns) from that observed in the study sectors, where the bedrock is relatively close. Finally, we note that as soon as we cross the former terminus of the main gallery (beyond the chatière De Joly that was widened in 1947), the number of broken speleothems is much lower. Stalactites that have fallen from the ceiling could be related to earthquakes. Broken stalagmites and columns are generally associated with clay extraction or recent breakage. Without totally ruling out natural causes, careful examination of old stalagmite stumps indicates that the stalagmites have been broken by intentional strikes, although interpretation is often delicate where there is stalagmite regrowth or patina. In the case of the most “recent” breakages, strike impacts are particularly visible.
Regardless of the cause of breakage, broken stalagmites have been moved, transported, and arranged on the ground by humans to build structures. It is interesting to note that in the Colonnes sector, particularly for the speleofact structure in the northern sector, a large number of the speleofacts did not originate from the immediate surrounding area. The number of stumps (20) in the vicinity of the structure (92 elements including 20 stalagmite tips) is small, and the diameters of the stumps are well below those of the recumbent speleofacts. Stump diameters range from 6 to 15 cm, with 60% around 10-12 cm; speleofact diameters are between 25 and 12 cm, with 48% over 20 cm. By matching the diameters of speleothems across the study sector, we found that most of the speleofacts in the structure in the northern sector originated from the central area of the room, and perhaps from edifices that are further north.
All the data from the mapping and field surveys highlight deliberate human modification of the underground space. Large numbers of broken speleothems have been used to build structures and the time at which these activities took place remains to be determined. The breakages of the pieces that form these structures differ from “more recent” ones associated with looting. While close observation of the structures enables us to identify a relative chronology between human activity and natural processes (calcite deposition), only absolute dating can place these events within the history of the cave and human use.