2.1. General context on Provence dinosaur eggs and nests

As mentioned in Introduction, the Provence in southern France is famous for dinosaur eggs and nests with first record from Matheron (1869) and more extensive observations—notably in the Sainte-Victoire mountain surroundings—from de Lapparent (1947). Since then, many more nests have been discovered in the Réserve Naturelle de Sainte-Victoire (RNSV), most of those housed in the Museum of Natural History of Aix-en-Provence (MHNAix). Beyond these “nests”, many more isolated eggshells were also collected making the Provence eggshell collection particularly rich. Most of those eggs—classified as Megaloolithidae in the literature (Vianey-Liaud, Mallan, et al., 1994)—were traditionally attributed to titanosaurian dinosaurs (Chiappe et al., 2003; Vila et al., 2010) but recent Romanian discoveries challenges this unique combination in associating it with basal hadrosaurids (Grigorescu et al., 2010). So, it is more parsimonious to envisage this oofamily associated with herbivorous dinosaurs, especially since Rhabdodon (Ornithopoda) was found as the most abundant dinosaur in local layers rich in megaloolithid eggs (despite the lack of non direct in embryo nor in utero discoveries).

In the Sainte-Victoire mountain foothills, the continental deposits were not directly dated but inferred from magnetostratigraphic studies combined with sedimentological analysis (Cojan and Moreau, 2006) and paleontological markers such as charophytes, ostracods, pollen, and gastropods surrounding the interval rich in dinosaur eggshells (Babinot et al., 1983; Tortosa, 2014). However, thanks to the richness in eggshells, a local stratigraphy based on oospecies was proposed, increasing the precision of the scale but remaining still poorly anchored to the Geological Time Scale (Garcia and Vianey-Liaud, 2001) and highly dependent on the reliability and precision of determinations (Sellés, 2012; Dhiman et al., 2019; van der Linden et al., 2024).

2.2. Sampling sites

The dinosaur eggshells were collected in several places in Provence (Bouches-du-Rhône and Var departments), southern France (Figure 1).

2.2.1. Beaurecueil—Grands Creux 2 (BGC2)

Grands Creux 2 locality lies within the core restricted area of the Réserve Naturelle Sainte-Victoire (Beaurecueil, Bouches-du-Rhône department), at the foothill of the Montagne Sainte-Victoire, about 7 km east of Aix-en-Provence (Figure 1). This historical site, also known as Roques-Hautes, was discovered by de Lapparent (1947) and subsequently received considerable attention during the following decades for studies on dinosaur eggs (see Tortosa and Vianey-Liaud, 2021). It is located in the central part of the Aix-en-Provence Basin and belongs to the continental deposits historically referred to as the lower “Argiles et Grès à Reptiles” or lower “Argiles Rutilantes” Formation, dated to the upper Campanian (lower Rognacian local facies; Figure 2). In this area, the eggs—organized in nests—occur in floodplain fluvial deposits consisting of reddish siltstones to fine sandstones forming badland morphologies.

3.1. Sample collection

A total of 10 eggshell samples were examined in this study. These specimens are curated in two public institutional collections: the Bouches-du-Rhône Departmental Collection (CD13) and the Muséum d’Histoire Naturelle d’Aix-en-Provence (MHNAix), France. Five eggshell samples originated from Beaurecueil—Grands Creux 2 (BGC2; Figure 1). Those eggshells were collected in 2017 during planned excavations conducted by the RNSV and MHNAix teams. The BGC2 series is housed in the CD13 collections, those also have a national collection number: BGC2.1A.2017.16 (collection number CD13-PV.2025.01.001), BGC2.1A.2017.32 (CD13-PV.2025.01.002), BGC2.1A.2017.49 (CD13-PV.2025.01.003), BGC2.1A.2017.51 (CD13-PV.2025.01.004) and BGC2.1A.2017.52 (CD13-PV.2025.01.005). The remaining specimens are curated in the MHNAix collections: one eggshell named hereafter CRA from Châteauneuf-le-Rouge—La Crau Autoroute (CRA in Figure 1) was collected from a complete egg (collection number MHNAIX.PV.2025.1.2). One eggshell named hereafter GAL is from Châteauneuf-le-Rouge—La Galinière (GAL in Figure 1), sampled from a complete egg (MHNAIX.PV.2025.10.1). One eggshell sample hereafter named FAM (collection number MHNA-PV.2005.24.18) is a subsample of eggs collected during excavation campaigns conducted by the MHNAix between 1997 and 2001 at Fox-Amphoux—Métisson (FAM; Figure 1). One eggshell named hereafter PSS (collection number MHNAIX.PV.2025.9.1) was collected from a sandstone slab from Puyloubier—Saint-Ser (PSS in Figure 1). Finally, the eggshell sample named hereafter VLP (collection number MHNA-PV.1999.13) is from Vitrolles—La Plaine (VLP in Figure 1) and was collected in 1999.

3.2. Eggshells preparation

To perform microscopic observations and geochemical analysis by LA-ICP-MS, the samples were mounted in radial section in a 1-inch resin epoxy plug—araldite 2020 produced by ESCIL. Once prepared, the radial section was polished in order to abrade any residual resin on the eggshell. The resin used is depleted in Pb and U, thus preventing any contamination of the sample.

3.3. Eggshells optical microscopy, cathodoluminescence and numerical microscopy

Eggshells were observed in optical microscopy with a binocular lens Leica M125 with a magnification from ×8 up to ×100 scale of observation commonly used for the parataxonomic assignment of dinosaur eggshells (Mikhailov, 1997). However, this scale and methodology was not adequate to fully explore the degree of preservation of the eggshell structures.

All samples were studied with a cathodoluminescence installed on an optical microscope with ×4 and ×10 objectives and ×10 oculars. The microscope used was an Olympus BH-2, the cathodoluminescence a NewTec Scientific Cathodyne connected with an iDS camera. Depending on the luminescence of the eggshells, the electric potential was set between 12 kV and 18 kV, the intensity between 100 μA and 200 μA and 3 s photo acquisition time by default. Set of pictures after LA-ICP-MS analyses were performed using a Keyence VHX-7000 with magnification from ×30 up to ×300 under coaxial light.

3.4. Eggshells SEM and EDS

Before the preparation in plugs, BGC2.1A.2017.32, BGC2.1A.2017.51, CRA, GAL and FAM were observed with a SEM Hitachi S-3000N. Few uncalibrated EDS (Bruker Nano 129 eV XFlash Detector 5010) analyses were also performed on those samples with a range of spot analysis about few micrometers. These devices were used to identify minute structures and to observe possible chemical contamination or traces of diagenetic overprints. Sample VLP SEM images were taken after the U–Pb analyses.

3.5. U–Pb analyses

U–Pb analyses were carried out at CEREGE, Envitop Analytical facility, Aix-en-Provence, France, using an ESI 193 nm excimer laser ablation system coupled to an Element XR SF-ICP-MS (more detailed methodology and the U–Pb dataset are presented in supplementary material). At least 30 spots of 150 μm diameter were placed manually in zones showing the best preservation on cathodoluminescence images and having the highest U/Pb using quick laser ablation prescreening of Pb, U and Th signals. During prescreening, an empirical cut-off value for the ²³⁸U/²⁰⁶Pb of 5 is applied (Table 1). This value, although not very radiogenic for samples <100 Ma, allowed to quickly preselect zones with the best chance of providing spread in the ²³⁸U/²⁰⁶Pb ratios and hence U–Pb ages after data reduction.

After data acquisition, processing was done using Iolite (Paton et al., 2011) and an in-house excel spreadsheet to derive ages from Tera-Wasserburg diagrams using IsoplotR software (Vermeesch, 2018). WC1 was used as a primary standard to correct ²³⁸U/²⁰⁶Pb fractionation (Roberts et al., 2017) and AUG-B6 as a secondary standard to check for accuracy, yielding an age of 43 ± 2 Ma in agreement with Pagel et al. (2018). Ages are quoted at a 2s, including the propagation of systematic uncertainty of the standards following (Horstwood et al., 2016).

For LA-ICP-MS elemental mapping, a square spot size of 100 μm, raster speed, and MS cycling time were synchronized to provide low-resolution images with 100 μm² pixels, sufficient to correlate these mappings to other imagery techniques used in this study. A JCp^∗-1-NP pellet from myStandards GmbH was used as a reference.

4.1. Egg shells oospecies identification, microstructure, preservation and contamination

Typical structures of eggshells were observed in optical microscopy based on radial sections and complementary information were made based on the sculpture of the outer surface. Shell units were discretispherulitic (or tubospherulitic) with ovoid mammilla and radial-tabular columns (Mikhailov, 1997, Figure 3A). The spaces between the mammila were filled with sediments with a different color and texture than the shell itself. The discretispherulitic structure confirmed the attribution to Megaloolithidae oofamily. Pore canals were not easily observed in most samples, with the exception of FAM and PSS which had tubocanaliculate pore canals (Figure 3B). In many samples—when visible—the outer face of the egg was covered by nodes. The proximal face was smooth—when not encrusted. More precisely, BGC2.1A.2017.16 (Supplementary Figure 1), BGC2.1A.2017.49, BGC2.1A.2017.51 (Figure 3A) and BGC2.1A.2017.52 (Supplementary Figure 2) were attributed to the oospecies Megaloolithus petralta (Vianey-Liaud, Mallan, et al., 1994) based on the node diameter (0.4–0.6 mm), its slightly arched growth lines, and partially fused, narrow fan-shaped shell units. The sample BGC2.1A.2017.32 was tentatively attributed to Megaloolithus sp. Although historically attributed to Megaloolithus aureliensis (ibid.) and later to Megaloolithus cf. aureliensis (Vianey-Liaud and Garcia, 1999), it is now regarded as a distinct oospecies with a smaller general size, thinner shell thickness, arched growth lines, and partially fused fan-shaped shell units. So far, Megaloolithus sp. was exclusively reported from BGC2 locality. The sample GAL (Supplementary Figure 3) from Châteauneuf-le-Rouge—La Galinière and the sample VLP from Vitrolles—La Plaine were attributed to the oospecies Megaloolithus mamillare (Vianey-Liaud, Mallan, et al., 1994) based on arched growth lines following the surface and unfused short and broadly fan-shaped shell units. The sample PSS (Supplementary Figure 4) from Puyloubier—Saint-Ser was attributed to the oospecies Megaloolithus siruguei (ibid., Figure 3B), with strongly arched growth lines, unfused fan-shaped shell units, and a reticulate canal pore system. Finally, the samples CRA (Châteauneuf-le-Rouge—La Crau Autoroute) and FAM (from Fox-Amphoux—Métisson; Supplementary Figure 5) were attributed to the oospecies Cairanoolithus dughii (ibid.) based on horizontal growth lines in the interlocking shell units and more arched single units as well as the presence of very smooth nodes on the external surface. The Supplementary Figure 6 summarizes the oospecies identifications.

SEM pictures of sample CRA C. dughii, sample GAL M. mamillare, sample FAM C. dughii, sample BGC2.1A.2017.32 Megaloolithus sp. and sample BGC2.1A.2017.51 M. petralta showed that the prismatic structure of the thick CaCO₃ layer was preserved (Figure 4 and Supplementary Figures 7–10). The cathodoluminescence showed that recrystallisation and secondary crust were luminating (Figures 5, 6 and Supplementary Figures 7–10). Those were concentrated in the inner and outer faces of eggshells, probably due to the contribution of clays. In most case, luminescence was observed between two shell units, within radial thin structures. If some of those structures may correspond to pore canals (Supplementary Figure 4), in most case those seemed too narrow to correspond to tubocanals and thus may correspond to fractures and/or interface between two shell units (Figure 5 and Supplementary Figure 1). In PSS (Supplementary Figure 4), the pore canals displayed circularly banded luminescence interpreted as recrystallisation within the pores.

Most of the prismatic calcite forming the majority of the eggshell was depleted in transition metals (Figure 4 and Supplementary Figures 7–10) and did not present any luminescence (Figures 5, 6 and Supplementary Figures 1, 2, 5). To conclude, the faces, thin fractures, transversal canal pores, and longitudinal pores were encrusted with clays and/or CaCO₃ enriched in transition metals testifying secondary crystallization. The rest of the prismatic layer showed comparatively a better preservation and was targeted for laser ablation analyses.

4.2. U–Pb–Th concentrations

Elemental mapping (LA-ICP-MS) of sample VLP showed distinct spatial distributions of Mn (not shown), Pb, Th, U. Concentrations ranged from 0 to 150 ppm, 50 ppm, 1.5 ppm, 10 ppm respectively (Figure 7). The highest values for all elements occurred near sutures and in the inner and outer layers, which were interpreted as zones enriched in clays and/or affected by diagenetic processes. In contrast, the prismatic calcite outside these areas exhibited significantly lower concentrations (Th ≪ 0.5 ppm, Pb ≪ 1 ppm, U ≪ 1 ppm), suggesting minimal contamination and secondary alteration—consistent with observations from cathodoluminescence imaging. ²³⁸U/²⁰⁶Pb varied from 0 to ∼25 with the most radiogenic values (highest ratios) found in areas with the lowest Pb and U concentrations. Age calculation from the elemental mapping could not be applied due the overall low Pb content in the best-preserved zones that prevented the use of ²⁰⁷Pb measurements, as the signal on the 207 mass was too low to be reliable.

4.3. U–Pb ages

Only two samples, CRA and VLP, yielded reliable U–Pb isochrons (Figure 8). Some other samples were tested (see Table 1), but did not yield reliable isochrons mainly due to unfavorable ²³⁸U/²⁰⁶Pb ratios—below the cut-off value of 5—and excluded to the exception of sample PSS presented in Supplementary Figure 11. The age of sample VLP attributed to Megaloolithus mamillare was calculated to 67.4 ± 4.4 Ma (2s, n = 60) (Figure 8A). The age of sample CRA attributed to Cairanoolithus dughii was calculated to 69.5 ± 9.9 (2s, n = 50) (Figure 8B). Considering the uncertainties, sample VLP was dated within the Maastrichtian-Danian (Paleocene) time interval and sample CRA within the Campanian-Selandian (Paleocene) time interval. Since these eggshells coming from dinosaur species cannot be younger than the Cretaceous–Paleogene (K–Pg) boundary, VLP age’s was considered Maastrichtian and CRA Campanian to Maastrichtian. For both samples, the ²³⁸U/²⁰⁶Pb were relatively low (<35) compared to the ²³⁸U/²⁰⁶Pb radiogenic concordia intercept of ∼100 for ages ∼68 Ma, leading to moderate uncertainties (6.5% and 15%, respectively). Based on LA-ICP-MS, CRA had lower Pb concentrations and lower U concentrations (<100 ppb in average for both Pb and U) compared to VLP (∼500 ppb in average for both Pb and U) which explain its greater uncertainty despite showing a similar ²³⁸U/²⁰⁶Pb range. For each sample, the initial ²⁰⁷Pb/²⁰⁶Pb ∼0.84 was consistent with the expected value of less than 100 Ma formations (Stacey and Kramers, 1975). The ²³²Th concentrations were well below 0.5 ppm (Figure 7) limiting the possible contamination from secondary phases (see previous section).

5.1. U–Pb system in eggshells and diagenesis

Previous U–Th analyses of Quaternary ostrich eggs showed that the prismatic calcite forming the shell behaved as an open U-series system, requiring advection–diffusion models to account for U mobility during diagenesis and to derive ²³⁰Th–U ages estimates (Sharp et al., 2019). On U–Pb timescales, it can be expected that the U-series system was subject to several post-depositional processes, including contamination from soil-derived detrital particles filling in the intercrystalline microporosity of the eggshell (Loewy et al., 2020) and/or mobility of U and Pb, potentially driven by meteoric or sediment derived fluids circulation along pores or microfractures of the shells (Tucker et al., 2025). The time scale for which the U-series system in prismatic calcite remains open is still an open question and hence U–Pb ages must be considered as minimum ages. Tucker et al. (ibid.) demonstrated that U–Pb ages obtained from dinosaur eggshells in the Western Interior Basin of the USA—where preservation was rigorously controlled through both optical and geochemical analyses—agree within a 5% margin with Maximum Deposition Ages derived from detrital zircons. In the samples studied here, elemental mapping (Figure 7) defined two zones with distinct spatial distribution of U and ²³⁸U/²⁰⁶Pb. The first one, corresponding to the inner and outer layers as well as the fractures and pores, was characterized by variable and high U and Th contents, along with high Mn and Pb and low ²³⁸U/²⁰⁶Pb. This zone showed evidence of contamination with detrital material in both optical and cathodoluminescence images. The second zone, located in the core of the prismatic calcite, exhibited lower U (<1 ppm), Th, Mn and Pb and relatively high and variable ²³⁸U/²⁰⁶Pb ratios. Optical and cathodoluminescence imaging indicated that this zone was well preserved.

U–Pb spot analyses in the well-preserved prismatic calcite of samples VLP and CRA define Tera-Wasserburg diagrams with robust linear alignment and consistent intercept ages (Figure 8). These results demonstrate that the U–Pb system remained closed for approximately 68 Ma in both samples. Notably, both samples exhibit limited radiogenic ingrowth, as evidenced by ²³⁸U/²⁰⁶Pb below 30—significantly lower than the expected radiogenic intercept of around 100 for 68 Ma Tera-Wasseburg isochron (Supplementary Figure 12). Similar ²³⁸U/²⁰⁶Pb ranges were reported by Chen et al. (2025) and Tucker et al. (2025) for samples aged between 95 and 66 Ma from diverse continental settings. These consistencies in ²³⁸U/²⁰⁶Pb are interpreted as indicative of a shared mechanism of U incorporation during early diagenesis in the prismatic calcite of dinosaur eggshells.

U sensitivity to redox conditions influences its geochemical behavior and mobility during diagenesis (Langmuir, 1978), in contrast to Pb. Modern avian eggs incorporate negligible U during calcification (U <1 ppt) (Loewy et al., 2020), and arguably this can be speculated to apply to non-avian dinosaur eggs. Post-depositional incorporation of U in CaCO₃ (as $\mathrm{UO}_{2}^{2+}$) would require calcite dissolution-precipitation (Reeder et al., 2001), which is inconsistent with the pristine mineralogical preservation observed in optical, SEM, and cathodoluminescence images of the well-preserved zones. Avian eggshells also contain significant amounts of organic matter, such as proteins, that impregnate the calcite matrix (D. L. Smith and Hayward, 2010). Given that tetravalent U(IV) has a known affinity for organic matter (Bone et al., 2017), the presence and spatial distribution of the organic phase—though not analyzed during this study—could explain both the observed heterogeneity in U distribution and the long-term stability of the U–Pb system. If U uptake within the eggshell is controlled by the early diagenesis of organic matter intertwined with calcite, U uptake would be inherently linked to the bio-mineralogical composition of the shell and its interaction with external factors such as meteoritic or sediment-driven fluid circulation. U uptake would not need dissolution–reprecipitation to take place. This mechanism could greatly expand the applicability of U–Pb dating of dinosaur eggshells in other continental deposits, as also demonstrated by the successful results published by Chen et al. (2025) and Tucker et al. (2025).

5.2. Consistency of the U–Pb ages with stratigraphy of the sampling sites

The two U–Pb ages are consistent with each other, with a combined age of 67.7 ± 4.2 Ma (2s, n = 110) (see also Supplementary Figure 13), and both align with the stratigraphic age derived from independent markers (Figure 9). However, while these U–Pb ages are broadly consistent with the expected ca. 70 Ma range, their uncertainties, reaching up to 15% due to relatively low ²³⁸U/²⁰⁶Pb and low U and Pb contents, remain too large to resolve detailed stratigraphic relationships. The U–Pb ages obtained are less precise than the local stratigraphy. Although stratigraphy in continental deposits are often less precise than in marine deposits, the Provence region gathered almost 80 years of litho-, bio- and magnetostratigraphy literature limiting the input of calcite U–Pb ages. Together with the limited range in ²³⁸U/²⁰⁶Pb and the low success rate discussed in the previous section, these results highlight both the promise and current limitations of the method. In cases where dinosaur eggshells are the only available biostratigraphic markers, U–Pb analyses could provide a valuable complementary approach to microstructural studies and possible age anchors in the regional stratigraphy. Nevertheless, such analyses require thorough assessment of sample preservation before conducting in situ U–Pb measurements, similarly to results of Tucker et al. (ibid.).