The Cretaceous of the Cordillera Oriental in Bolivia is primarily represented by the Puca Group (Kimmeridgian to Paleocene) which unconformably overlies Palaeozoic sedimentary rocks (Sempere et al. 1997). The Puca Group within the Potosi Basin is deposited in an under-filled, back-arc foreland basin formed by subduction along the western margin of the South American plate (Sempere et al. 1997; Tasistro-Hart et al. 2019). Today, however, Cretaceous rocks are found only as relicts in small, overthrust nappes or synclines. One notable structure is the Maragua syncline, a large bowl-shaped structure with a NW-SE trending axis that forms part of the Potosi Basin, exactly in the Betanzos-Ravelos syncline. The Puca Group is deposited across an angular unconformity that formed over the Jurassic La Puerta Formation, and deposition commences within the siliciclastic Chaunaca Formation. The overlying El Molino Formation can be divided into three members: Lower, Middle, and Upper, respectively (Sempere et al. 1997 and Fig. 1).

The Lower Member of the El Molino Formation has been dated between ~73 Ma to 67 Ma (Sempere et al. 1997). Sempere et al. (1997) placed the Cretaceous-Paleogene (K/Pg) boundary between the Middle and Upper Member, a lithostratigraphic boundary that was also regarded as coinciding with the K/Pg boundary by Tasistro-Hart et al. (2019) based on cyclostratigraphy and radiometric ages.

The depositional environment of the El Molino Formation has long been debated. Based on the presence of selachians and (estuarine) actinopterygian fishes, Gayet et al. (1991, 2001) and Sempere et al. (1997) favour fully marine conditions; however, Rouchy et al. (1993), based on stable isotopes and palynomorphs, prefer an almost continuous lacustrine system in Bolivia during the Late Cretaceous. Furthermore, the El Molino Formation contains ostracods, gastropods, turtles, and catfishes, which support the ephemeral lacustrine basin model (e.g., Meyer et al. 2001; Meyer et al. 2019).

In the Maragua syncline, several dinosaur tracksites have been reported and which broadly occupy the same stratigraphic package in the uppermost part of the Lower Member of the El Molino Formation (Lockley et al. 2002; Meyer et al. 2019; Riguetti et al. 2021). Lockley et al. (2002) thoroughly documented the Humaca site (Fig. 1A) which provides evidence of a herd of eleven (subadult) titanosaurids (av. 45 cm pes length) and small theropods (12–30 cm pes length) in the uppermost part of the Lower Member the El Molino Formation (contraLockley et al. 2002 and Meyer et al. 2019). The Niñu Mayu site (Fig. 1A) has yielded tracks of cf. Tetrapodosaurus attributed to small nodosaurid ankylosaurs based on manus morphology (Riguetti et al. 2021), a large theropod trackway, and tracks of large titanosaurids (ca. 80 cm pes length; larger than at Humaca and our Mullurina Mayu site). Several other sites are known but currently remain undocumented in the area.

Herein, we report on a track surface within this area, which is located at Mullurina Mayu (Quechua for “arc-shaped river”) at 3450 m.a.s.l., 5.6 km WNW of the small village of Maragua, 23 km west of Sucre (19°03'46.96"S, 65°28'51.61"W, Distrito 8; Departamiento de Chuquisaca), Bolivia (Fig. 1). It is situated 450 m east of the Niñu Mayu locality (19°03'58.88"S, 65°28'43.97"W; 3434 m.a.s.l.), reported by Riguetti et al. (2021), and 2.25 km northeast of the Humaca locality (19°04'40.81"S, 65°27'58.97"W 3179 m.a.s.l.) described by Lockley et al. (2002).

The sauropod trackways consist of manus-pes couplets, with the preserved pes preserved as an elongated depression visible posterior to the manus impression. Although the overall dimension of the pes is unknown, similar specimens from the higher level, 1.5 m above (see Fig. 2) average 56 cm in length and 49 cm in width. The manus is half-moon shaped, averaging 41 cm in width and 20 cm in length. Soft sediment deformation features are associated with manus impressions near the depression described below (Figs 4, 5). While digit impressions on the current surface are poorly defined, many well-preserved tracks with digit impressions are visible on the higher surface (Fig. 3, heteropody index = 0.3). Data from these trackways indicate an average stride of 1.7 m and a gleno-acetabular distance of ~1.8 m. One measurement from the level with the depression indicates a slightly higher gleno-acetabular distance (~ 2–2.2 m) while stride varies between 1.8 and 2 m.

Towards the western part of the surface, a single, relatively deep, bowl-shaped depression is cut into the palaeosurface. The length along the dip of the strata is 4.2 m, the width along the strike is 3.2 m, and the maximum depth is ~60 cm. The complete surface area is 12 m² (Figs 2, 3). There is a relatively sharp rim (break in slope) and a deeper inner “basin” which shows two longitudinal trough-like structures at the western end (Fig. 4D–G). The margin is frayed, irregular in shape, and has scalloped edges (Fig. 5). Several areas show groups of parallel, linear grooves at the margin of the depression, which appear to be bound to what seem to be semi-circular impressions (Figs 4, 5). The bundles of linear grooves (1 cm wide) are confined to these impressions (width 23–28 cm) and have an average interspace (between scratches) of 1.8 cm (Fig. 6). There are six to seven parallel grooves per bundle (impression) with a maximum length of 30 cm in transverse section; they are hummocky-shaped in section. Furthermore, there are isolated broad single scratches that seem to be claw marks (length 5–10 cm, width 2 cm; Fig. 6D). Soft sediment deformation features (sediment bulges, sediment collapse) are associated with the impressions and scratch bundles.

The irregular Mullurina Mayu bowl-like structure could be explained as a feature, such as a pothole or a large scour, within a fluvial setting. A scour would form on the surface due to high-velocity, sediment-laden flow over the semi-cohesive palaeosurface, producing erosional sole marks (which include scours and gutter casts; Miall 2006). These features are characterised by U-, V-, or rectangular cross-sections and are commonly elongate and asymmetric in plane view. Diagnostic criteria would include smooth or locally polished walls, tool marks, and a geometry that indicates unidirectional flow. In our case, the bowl-like morphology in planform, rim, and syn-sedimentary internal structures is inconsistent with flow-parallel, elongate structures typical of hydraulic scouring. We observe no scouring, tool marks, or polished surfaces on any part of the palaeosurface.

Additionally, the structure can be compared to a weathering pit (panhole) that forms during water pooling and repeated physical or chemical weathering (Udomsak et al. 2021; Timms 2021). In this case, we would expect to see textures consistent with weathering and no internal bedding disturbance. Typically, panhole pits occur as multiples on the same surface and within a similar size range (Timms 2021), which does not align with the singular bowl nor its overall morphology, which displays syn-sedimentary features rather than secondary weathering. Furthermore, no comparable pits have been identified on the same surface, and the structure’s morphology falls outside the size range typical of panhole associations (e. g. 2 m by 1 m).

A biogenic process of formation for the Mullurina Mayu depression is supported by the presence of grouped, parallel grooves (several confined to defined areas; Figs 5, 6) along the basin walls, together with isolated, single scratch-like marks on the basin floor, which are interpreted as ungual traces. Associated titanosaurid trackways and the size and morphology of the scratch traces make the most parsimonious interpretation that a sauropod was involved in the formation and/or modification of this basin-like structure (Fig. 6). Additionally, sauropod tracks (manus impressions) can be seen near the basin margin (Figs 2, 4, 6).

The most directly comparable example for the scratch-like parallel grooves comes from Late Cretaceous titanosaur track soft tissue traces (Nemegt Formation, Mongolia; Bell et al. 2025), in which both the manus and pes show evidence of parallel striae (scale scratch lines; Bell et al. 2025; Fig. 7C–F) created by keratinised scales during foot movement in the sediment. The width and interspacing of scale scratch lines vary alongside variations in scale size, morphology and geometry. In the manus of the Nemegt trackmaker, the diameter of the scales scratch lines is ~10–12 mm, with interstitial area width of 2–7 mm. The low relief smooth/hummocky, large scales grade into pyramidal or spur-like scales at the palmar margins (Bell et al. 2025). Similar bundling of parallel grooves (scale scratch lines) within the walls and floor of a 3D sauropod manus impression cast (e.g., Late Jurassic, Milàn et al. 2005) provides evidence of manus movement. In other latest Maastrichtian titanosaurian skin traces (southern Pyrenees), sub-rounded pentagonal and hexagonal polygons (11–16 mm in diameter) are arranged in rosettes in the manus (Fondevilla et al. 2017) when impressed.

In the Nemegt pes (Fig. 7C, D), Bell et al. (2025) measure scratch line diameters between ~2–3 mm at the margin of digit I, which increase to ~10 mm in digits IV and V, with the posteromedial area of the heel preserving the widest scratch lines at ~20 mm ending in polygonal skin impressions. The Mullurina Mayu parallel striae traces are ~10 mm wide, with an interspacing distance of ~18 mm. These are comparable to the scale scratch lines from the Nemegt titanosaurid traces (~20 mm around the heel, 2–10 mm around digits; Bell et al. 2025) in the mid-foot to distal digital scales. In addition to the presence of scratch lines, the Mullurina Mayu traces occur in semi-circular depressions and exhibit two main orientations. They are inconsistent with a single direction of movement. The multidirectional striae bundles are interpreted as produced by repeated foot movements around the margin of the bowl depression. These differ from the previously outlined skin impressions associated with locomotion. The traces in the depression show evidence of scraping or scratch digging. Bell et al. (2025) discuss the Mongolian titanosaurid pedal ungual traces, and estimated them to estimated to represent only 10–13% of the ungual protruding from the fleshy cuticle, they have hypothesised that scratch digging is possible for pedal digit functionality. Bonnan (2005) was one of the first to come up with this possibility. This is supported by Cretaceous nesting structures (see below; Chiappe et al. 2004; Vila et al. 2010).

We hypothesise that the Mullurina Mayu bowl-like depression may represent the following, in order from most to least likely:

A dinoturbation basin like a trampled watering hole. In contemporary semi-arid settings, topographic low points retain standing water and serve as congregation points for animal visitation (Jarman 1972). Continuous trampling compacts and deforms the substrate, while scraping and slipping behaviour introduces grooves and irregular surface textures. Periodic quiet-water inundation deposit fine-grained laminated sediment between trampling events, recording episodic activity. This model predicts a substrate with compaction structures, scratch or slip grooves, and interbedded fine-grained laminae reflecting alternating wet and dry phases. However, at Mullurina Mayu, both the underlying substrate and infill lack lamination (structureless, no interbedding), which contradicts the hypothesis of repeated trampling cycles. Pedogenesis indicators in the overlying bed suggest prolonged subaerial exposure, consistent with a single low-energy event rather than episodic ponding. Moreover, sauropod tracks do not show a congregation point or trampling near the basin.

Geophagy: A dinoturbation basin for mineral mining or water. Geophagy, the ingestion of substrate for minerals, by large herbivores such as elephants (Jarman 1972) is one possible explanation for the creation of the depression. The present substrate does not contain any traces of minerals (except quartz and mica), therefore we can likely exclude geophagy. Mammals are also known to excavate holes to reach the water table and for purifying drinking water (Ramey et al. 2013). This could be another avenue for the creation of the structure; however, it is hard to prove or disprove either of these assumptions.

An attempt at a nest scrape (Chiappe et al. 2004). Trackmakers are known to excavate a shallow to moderately deep pit-like structure for egg-laying (Varicchio et al. 1997; Grellet-Tinner et al. 2006; Chiappe et al. 2004; Dhiman et al. 2023). Chiappe et al. (2004) has documented pit-like structures used for egg-laying by sauropods. Morphologically, these show some similarity with the Mullurina Mayu depression. Chiappe et al. (2004) suggest 4 criteria for recognising such a structure: (a) presence of eggs/eggshell/juvenile body fossils in the depression; (b) depression truncates primary stratification; (c) elevated rim of displaced sediment around the depression, and (d) contrasting infill. Vila et al. (2009, 2010) reported dinosaur nests from the Late Cretaceous of Spain and concluded that the structures were excavated by the hind feet of the sauropods. The Mullurina Mayu depression shows two of these features (truncation and infill differs from host substrate). There are no repeated basin-like or nest-like structures (i.e., no evidence of communal nesting behaviours).

A less likely hypothesis is a mal-formed, lek-like display scrape area (Lockley et al. 2016; Buntin et al. 2025). In this case, the structure would show repeated, long, scratch-like grooves that create a depression within the sediment, which can be ascribed to the ichnogenus Ostendichnus. Ostendichnus is characterised by an oval-shaped depression separated by a central ridge, bisecting the structure into two parallel troughs with internally parallel and longitudinal grooves (scratch marks). This trace has been linked to theropod dinosaurs, indicating areas for nesting or display behaviours (courtship) (Lockley et al. 2016; Buntin et al. 2025). The Mullurina Mayu bowl-like structure doesn’t show any of the characteristics of Ostendichnus and therefore is unlikely to represent a lek-type trace.

Of the hypotheses evaluated, abiotic formation is considered least consistent with the syn-sedimentary nature, scratch-like traces, and bowl morphology of the structure. Among biogenic interpretations, the dinoturbation basin and nest scrape remain the most morphologically plausible, though neither can be confirmed without more detailed study and the presence of more concrete examples. The scratch traces are comparable to skin-slip impressions documented for both the manus and the pes of sauropods (e.g., Milàn et al. 2005; Fondevilla et al. 2017; Paik et al. 2017; Bell et al. 2025; Fig. 7).

Sauropod nesting and reproductive biology is poorly understood from the fossil record (Chiappe et al. 2004), and the unknown ontogenetic stage of the Mullurina Mayu trackmakers makes interpretations complicated. We have assumed, based on size, that these tracks represent subadults, while acknowledging that track size alone is not a direct proxy for ontogenetic stage. However, adjacent sites within the same sedimentary package show large pes tracks (Niñu Mayu, pes length: >80 cm). Nonetheless, if assuming abandoned nesting for our depression, subadult titanosaurid sauropods could be capable of reproductive behaviour because sexual maturity is reached before maximum body size (Curry Rogers and Erickson 2005).

It is also possible that the Mullurina Mayu depression represents the activity of another animal. For instance, longirostrine crocodyliforms are known from the Bolivian Late Cretaceous (e.g., Jouve et al. 2021) and Late Cretaceous notosuchian crocodyliforms from Brazil are known to produce egg clutches in nests, preserved in situ, in floodplain mudstones (Paixão et al. 2026) while eggshell microstructure, i.e., oospecies Bauruoolithus, indicate burial-style nesting (Oliveira et al. 2011) consistent with crocodyliforms. No crocodyliform tracks are currently known in Bolivia, and in the absence of other evidence (in situ egg clutches/eggshells), a crocodyliform origin is considered less likely. Although the overall dimension of the Mullurina Mayu structure does show some geometric overlaps with modern crocodile hole nests (Table 1), it is considerably larger than those of known crocodyliform nest chambers (modern hole-nests; Modha 1967; and fossil clutch size; Paixão et al. 2026). Moreover, the size and geometry of the scratch trace bundles, confined within oval depressions, along the basin walls, are more consistent with the geometry and sizing of sauropod pes movement than with crocodyliform limb actions.