INTRODUCTION
Thanks to their vast extent, the fossil content, and historical significance, “the formations of Tuscany” became the cradle of the Pliocene Epoch (Brocchi, 1814; “Older Pliocene” in Lyell, 1833, p. 54; see Rudwick, 2008, p. 272-273), capturing the interest of researchers for its exceptionally rich marine fossil assemblages to our days (Mortillet, 1863; Pecchioli, 1864; D’Ancona, 1871; Seguenza, 1873; Lawley, 1874; Dainelli & Videssot, 1930; Bianucci et al., 1998; Bianucci & Landini, 2005; Ragaini & Bernieri, 2007; Dominici et al., 2018).
A variety of shallow and deep water litho- and biofacies allows for the recognition of small-and medium-scale unconformity-bounded units (Benvenuti et al., 2007, 2014; Nalin et al., 2016) and enables the quantitative tracking of palaeoenvironmental and palaeobiological gradients in time (Tomašových et al., 2014; Dominici et al., 2018; Dominici & Danise, 2023). The multiplicity of facies in Tuscany positively correlates with Pliocene biological diversity, but the influence of facies on fossil content notoriously complicates the use of calcareous plankton for broad-scale biostratigraphic correlation and leads to different opinions on the exact age of strata (see Martini et al., 2016).
The Pliocene of Tuscany has also been the object of an ongoing dispute on which tectonic regime-controlled basin development and large-scale depositional dynamics (Benvenuti et al., 2014; Mirabella et al., 2022; Poneti et al., 2024). Formed in an overall extensional tectonic regime related to the opening of the Tyrrhenian Sea (Jolivet et al., 2008), Tuscan basins have been interpreted either as bowl-shaped basins evolving into graben bounded by normal faults (Pascucci et al., 2007; Brogi, 2020) or as thrust-top basins (broken foreland basins: Poneti et al., 2024). Given the wider interest, high-resolution stratigraphy achieved by integrating multiple tools emerges as a key approach for resolving controversies. In this paper we present a first step in this direction through the integrated study of a ca. 120 m thick succession, exposed at the Belvedere waste facility near Legoli, in the municipality of Peccioli (Pisa: Fig. 1). The succession is part of the Valdera-Volterra basin (VVB), one of the three largest Pliocene basins of Tuscany (Dominici et al., 2018; Dominici & Danise, 2023), where Pliocene deposits reach a thickness up to about 1,600 m (Poneti et al., 2024).
- Schematic Geological map of the Valdera-Volterra basin showing the major structures and the location of the seismic line used in this study (green line). Black star, studied locality; VFS, Villamagna fault system.
At Legoli we integrate the study of sedimentary facies, quantitative analysis of mollusk shell beds and calcareous plankton (foraminifera and calcareous nannofossils), and magnetostratigraphy. We interpret the results within the broader geological framework established in Tuscany through geological mapping and seismic stratigraphy. This integrative approach allows us to better constrain available chronostratigraphic data on the Pliocene of the VVB, define the depositional dynamics during a specific phase of basin infill, and improve the understanding of the general tectonic and climatic regime affecting the region around the Zanclean-Piacenzian boundary.
GEOLOGIC SETTING
The Neogene-Quaternary Valdera-Volterra basin (VVB) is oriented in a NNW-SSE direction. According to Poneti et al. (2024) the two bounding structures are two major E–NE-verging, thrust-related anticlines: the Mid-Tuscan Ridge (MTR) to the northeast and the Perityrrhenian Ridge (PTR) to southwest (Fig. 1). These structures controlled its evolution, leading to the accumulation of an upper Serravallian–Pleistocene sedimentary succession up to about 2,500 m thick (Bossio et al., 1996; Costantini et al., 2000; Lazzarotto et al., 2002; Martini et al., 2001; Pascucci et al., 2007). A Miocene-Pliocene unconformity and a Zanclean-Piacenzian unconformity were clearly identified along all seismic lines in a recent analysis that integrated 2D stacked seismic reflection profiles, deep exploration wells, gravity data, and geological and biostratigraphic literature. The Miocene-Pliocene unconformity is marked by reflectors that clearly onlap onto underlying truncated units, a feature generated by thrust propagation at the end of the Messinian compressive phase, which created accommodation space for Zanclean units (Poneti et al., 2024 and references therein). According to this model, the Neogene development of the VVB is framed within an alternation of crustal recompression and relaxing, respectively accommodated by activation (or reactivation) of thrust faults and basin-margin normal faults. In other words, the basin developed, similarly to other hinterland basins of Tuscany, within a complex tectonic setting as expected in a still evolving collisional orogen. The alternative scenario justifies increased accommodation in a purely extensional geodynamic regime (e.g., Costantini et al., 2000, and references therein; see also Milaneschi et al., 2024 for the neighbouring Valdelsa Basin).
The Zanclean has been further subdivided into four smaller seismo-stratigraphic units (Z1-Z4 in Fig. 2) deposited during the final stage of compressive deformation and forming a gentle syncline (Poneti et al., 2024). Z1 and Z2 are present in the southern sector of the basin, each bounded below by surfaces of strong subaereal erosion. Z3, which is thicker in the axial part of the basin between Villamagna and Laiatico (Fig. 1), wedges out towards the basin shoulders, suggesting differential uplift. In contrast, Z4 does not show significant thickness variations, suggesting the deactivation of thrust-related anticlines (Poneti et al., 2024). In the northern sector, the Zanclean succession comprises only Z3 and Z4 (Fig. 3), which are gently inclined westward in the study area. The Piacenzian deposits (P1-P3 in Fig. 2) onlap older units and mark the end of the compressive phase, when thrust faults gave way to normal faults, creating new accommodation space and inducing the tilting of underlying units. In the northern sector of the basin, two second-order unconformities have been identified marked by onlap geometries, which subdivide the Piacenzian into three smaller units. These units thicken towards the eastern basin shoulder, in correspondence with the Villamagna Fault System (VFS), where the Piacenzian deposits reach a maximum thickness of 900 m (Figs. 1, 3). Southward, only P1 and P2 are recognized in the subsurface at Lajatico and Villamagna, while at Volterra only P1 is present (Poneti et al., 2024; Figs. 9–10).
- Stratigraphic log of the Neogene infill of the Valdera-Volterra basin (modified after Poneti et al. 2024). SSDS: small-scale depositional sequence; LSDS: large-scale depositional sequence.
- Uninterpreted (A) and interpreted seismic reflection profile (B) passing through Legoli (green line in Fig. 1). LF, Laiatico fault; VMF: Villamagna fault. Note that the studied succession falls in areas of poorer quality and constraints (modified from Poneti et al., 2024).
On outcrops, four main lithostratigraphic units have been identified: sabbie di San Vivaldo (“San Vivaldo sands”, SVV - Zanclean), Argille Azzurre (“Blue Clays”, FAA - Zanclean- Piacenzian), calcari di Volterra (“Volterra limestone”, VTR - a Piacenzian calcarenite) and formazione di Villamagna (“Villamagna sands”, VLM - Piacenzian) (Servizio Geologico d’Italia, 2000). The Argille Azzurre, primarily composed of mud-rich deposits, constitute the thickest part of the succession and include gravelly and sandy members in the southernmost sector of VVB (Costantini et al., 2000). Biostratigraphic data collected between Volterra and Mazzolla (Fig. 1) indicate that the Argille Azzurre have a thickness of 800-1,000 m and span from the lowermost Zanclean to the Piacenzian, up to the lower part of the biozone Globorotalia aemiliana (Bossio et al., 1994; Costantini et al., 2000; Riforgiato et al., 2005). The latter zone occurs tens of meters below the contact with the formazione di Villamagna and the calcari di Volterra. The latter two formations are assigned to the Globorotalia crassaformis crassaformis subzone of planktonic foraminifera and to nannofossil zones NN16-18 (following the scheme of Martini, 1971; Giannelli et al., 1981). However, no analytical data on marker distributions were published in these and similar studies carried out in the VVB (Bossio et al., 1996; Sandrelli et al., 2004), nor in the adjacent Valdelsa Basin (Capezzuoli et al., 2005; Milaneschi et al., 2024).
Analytical data on foraminifera and nannofossils are available via the web portal of Regione Toscana (Geoportal, 2024; see also Poneti et al., 2024). Additional data from the uppermost 100 m of the Argille Azzurre near Volterra suggest that these and the overlying sandy interval correspond to the top of the Piacenzian (Bianucci et al., 1998). This finding contradicts seismic data, which instead indicate to the absence of P2 and P3 at Volterra (Poneti et al., 2024). An integrated study of the Pliocene cropping out near Parlascio, in the northwestern sector of VVB (Fig. 1), which includes biostratigraphy and magnetostratigraphy, attributes the calcari di Volterra to the middle Piacenzian (Nalin et al., 2016). This unit is correlated with biocalcarenites from other Tuscany basins, supporting its significance in relation to global palaeoclimatic events (Benvenuti et al., 2014; Nalin et al., 2016; Dominici et al., 2018; Dominici & Danise, 2023; Milaneschi et al., 2024).
The Pliocene deposits of the adjacent Valdelsa Basin have been subdivided into six synthems (S1-S6), each bounded by major unconformities (Benvenuti et al., 2014), a subdivision tentatively extended to the VVB and other basins of Tuscany (Dominici et al., 2018; Dominici & Forli, 2021; Dominici & Danise, 2023). However, the correspondence between synthems identified in outcrops and seismic profiles and their correlation with the geological time scale remains to be verified (Fig. 2).
The fossil macrofauna of the VVB includes abundant mollusk shell beds (Ragaini & Bernieri, 2007; Dominici & Danise, 2023), studied at Legoli since the 1870s (Seguenza, 1873, 1875), and a diverse marine vertebrate assemblage (Bianucci et al., 1998; Bianucci & Landini, 2005; Dominici et al., 2018). Furthermore, a fragmented proboscidean remain attributed to Archidiskodon gromovi (= Mammuthus meridionalis Nesti) was reported in 1921 from an unspecified locality at “Legoli” (Bianucci & Landini, 2005, p. 11).
The area is influenced by the northernmost expression of the VFS, a NNW-SSE-aligned structure identified in both outcrops and seismic profiles (Figs 1, 3). This fault system acted as a normal fault during the Piacenzian, generating new accommodation space and ultimately placing P3 in contact with Z4 and P1 west of Legoli (Costantini et al., 2000; Poneti et al., 2024). At what particular moment this movement interested sedimentary dynamics at Legoli and in general in the VVB is among the objects of the present study.
MATERIALS AND METHODS
We conducted a survey of the area around the Belvedere waste disposal plant and mapped the main lithologic units (Fig. 4). The succession exposed on the flanks of the waste plant (Fig. 5) was measured and sedimentary facies were described (Figs. 6-7) following criteria tested in other Pliocene sedimentary basins of Tuscany (Benvenuti et al., 2007). The succession was sampled to collect quantitative data on calcareous plankton and benthic mollusks. Oriented blocks and drilled cores were collected for palaeomagnetic analyses from mudstones of units B and D, while sandstones were not compacted enough to be sampled for a palaeomagnetic study.
- Geological map of the Belvedere waste plan to the SE of Legoli showing the distribution of sandstone and mudstone units considered in this study and the position of the studied succession.
- The Pliocene succession freshly-exposed at the Belvedere waste plant. A) Overview of the main Belvedere outcrop, from W to E; the dashed lines indicate tracts of the studied section. B) Detail of stratal terminations of sandstone C on top of mudstone B. C) Interpretation of the geometric relationships between sedimentary units (for the relative position see inset in A). D) Detail of the angular unconformity separating units B and C. E) Interpretation of the geometric relationships between sedimentary units (for the relative position see inset in A).
- Sedimentary characters and macrofossils of sandstone units cropping out in the study area. A) Bioturbated calcarenite at Legoli (35 cm-long hammer for scale). B) Disarticulated and transported bivalve shells in sandstone C (10 cm-long folding knife for scale). C) Recurring facies in sandstone B outlining partitioned beds referred to density-stratified flows (see text; 100 cm-long scale). D) Erosional contact between sandstone C and underlying mudstone B (outcrop about 10 m-thick). E) Detail of the facies stacking in the basal portion of sandstone in C. To notice the large mud clasts in the basal facies a (35 cm-long hammer for scale). F) Water- escape structures in sandstone C.
- Sedimentary characters and macrofossils of mudstone units cropping out in the study area. A) Parallel bedding in the upper part of mudstone C (35 cm-long hammer for scale). B) Costellamusiopecten cristatum and Anadara diluvii in life position (mudstone C) (18 cm- long hammer handle for scale). C) Undetermined tellinid in life position (mudstone C) (25 mm- coin for scale). D) Centrocardita rudista in life position (mudstone C) (22 mm-coin for scale). E) Stratal surface in the upper part of mudstone C, showing small bioconstructions on the ancient seafloor by cemented vermetid gastropods (35 cm-long hammer for scale). F) Pelecyora brocchii in life position (mudstone C).
Mollusk palaeoecological analyses
Mollusk shell beds were sampled for quantitative palaeoecological analyses. Due to the hardened muddy matrix, mollusk shell beds were sampled by picking shells from the surface. To make meaningful quantitative comparisons between assemblages, all shells were collected consistently at each major shell bed either by one person over approximately 30 minutes or by two persons for 15 minutes. Collected shells were washed through a 1 mm mesh, dried, and counted following standard procedures (Dominici et al., 2018). Species-level abundance data from seven such samples were standardized and interpreted by comparison with a large dataset available for the Mediterranean Pliocene, which includes 638 species from 132 samples across six different basins. This comparison allowed for the integration of Legoli palaeobathymetries within a broader Mediterranean-wide gradient (Dominici & Danise, 2023).
Differences in mollusk composition between samples were assessed via ordination analysis on a Bray-Curtis matrix. The latter was obtained after the exclusion of singletons and square-root transformation of proportional abundance data, carried out to dampen the contribution of the most abundant species in samples. The structure of macrobenthic communities was explored through non-metric multidimensional scaling (NMDS). Multivariate statistics were carried out using PRIMER® software (Clarke et al., 2014).
Shell bed diversity was expressed in terms of per-sample number of species (S) and palaeoecology was interpreted by comparison with known species- or genus-level ecological data available for the Mediterranean (Dominici & Danise, 2023).
Calcareous plankton analyses
A total of 27 samples were analyzed for calcareous plankton. Sample resolution ranges from an average spacing of 2-3 meters in the upper part of Unit B, and from 3 to 9 meters in Unit C and D.
Slides for calcareous nannofossil analysis were prepared following the technique of Flores & Sierro (1997) and analyzed under a polarized light microscope at 1000x magnification. Semi- quantitative analyses were conducted on the whole assemblage by scanning an area of approximately 1 mm2. Quantitative analyses focused specifically on all Discoaster species, scanning an area of about 4 mm2. Abundance patterns of taxa were plotted as number of specimens/mm2, providing a more accurate representation of stratigraphic distribution patterns (Rio et al., 1990). Biozonal identification follows the scheme of Rio et al. (1990).
Samples for planktonic foraminifer analysis were dried and washed through 63 and 125μm sieves. Residues>125μm were split until a representative aliquot containing about 300 specimens was obtained. Quantitative analyses were carried out for key biostratigraphic species such as Globorotalia puncticulata and Globorotalia crassaformis. Abundances were determined as percentages on the total number of planktonic foraminifer specimens in each aliquot. The planktonic foraminifer biozonation and biochronology proposed by Lirer et al. (2019) has been here adopted.
Palaeomagnetism
For the palaeomagnetic study, we sampled the Belvedere section at various depths using either oriented blocks or palaeomagnetic drilled cores. At each depth, we collected between 3 and 10 samples, resulting in a total of 344 samples from 58 depths. Blocks were oriented using a Brunton magnetic compass, while palaeomagnetic cores (~2.5 cm in diameter) were drilled using a portable rock drill cooled with water. The orientation of the cores was determined using both magnetic and sun compasses.
The blocks were cut with a diamagnetic saw into standard palaeomagnetic samples (8 cm3 of volume) for analyses of anisotropy of magnetic susceptibility (AMS), and alternating field (AF) demagnetization. Additionally, selected fragments were used for rock magnetic analyses, specifically for susceptibility vs temperature curves.
At the INGV palaeomagnetic laboratory, we conducted AF demagnetization protocols using the 2G cryogenic Magnetometer hosted in a magnetically shielded room. A total of 15 demagnetization steps (ranging from 0 to 80 mT) were applied to isolate the Characteristic Remanent Magnetization (ChRM) from the Natural Remanent Magnetization (NRM). Results were analyzed using the PuffinPlot software (Lurcock & Florindo, 2019), applying Fisher statistics and Principal Component Analyses (PCA).
The present-day geomagnetic field expected at this location (43.571389 N°; 10.795° E) has a declination (angle between the magnetic vector component and true north) of 3.84°, and an inclination (angle between the component of the vector and the horizontal plane) of 59.9°, calculated using the WMM2020 model (World Magnetic Model, 2020 edition; https://www.ngdc.noaa.gov/geomag/calculators/magcalc.shtml?useFullSite=true#igrfwmm)
To determine the main magnetic carriers in the samples, we measured their rock magnetic properties. Using a KLY5- Kappabridge, we analyzed the bulk susceptibility in 270 samples from 48 depths and the Anisotropy of the Magnetic Susceptibility (AMS) in 78 samples from 17 depths. Data were processed using the AGICO software Anisoft. (https://www.agico.com/text/software/anisoft/anisoft.php). Finally, we conducted four susceptibility-temperature curve measurements using the furnace unit on the KLY5-Kappabridge.
THE BELVEDERE SUCCESSION
Sedimentary facies analysis
The Belvedere succession under study is about 120 m thick. It is attributed to the Argille Azzure (FAA in Costantini et al., 2000) and probably includes an equivalent of the S3 synthem established in the Valdelsa Basin (Benvenuti et al., 2014).
Following a sedimentary facies analysis, the succession was subdivided into four lithological units, labelled A-D (figs 4–6). Additionally, a fifth unit (Unit E) crops out around the village of Legoli (Fig. 6A). Units D and E consist of partially cemented, often completely bioturbated sands (Fig. 6A) which correspond to the formazione di Villamagna and the calcari di Volterra described in previous studies (VLM and VTR, respectively, in Costantini et al., 2000). These units are here equated to S4 synthem of Benvenuti et al. (2014). However, Units D and E were not here studied in detail.
Unit A is partially covered by vegetation and was sampled only for the macrofauna. In contrast, Units B (upper part) and C, due to their fresh outcrop conditions (Fig. 5), were sampled for macro- and microfossil investigation, as well as for paleomagnetic analysis. Ideally, each unit consists of a lower sandstone interval and an upper mudstone interval. In this study area, the sandstone interval of unit A does not crop out, and the lower part of unit B could be only partially described in the field.
The thickest part of the succession consists of blue-grey mudstones with variable clay-silt mixtures (Fig. 7). These mudstones are predominantly massive, with only occasional stratification, displaying crude bedding (Fig. 7A). Gastropods, bivalves and scaphopods are typically found scattered or in life position (Fig. 7B–F). In some areas, particularly in unit A and in the uppermost part of unit D, the bioclastic fabric appears loosely packed (Fig. 7D–E). Additionally, occasional carbonized plant remains are present.
Greyish-yellowish, fine- to medium-grained sandstones define the base of unit B and, less frequently, its upper part, where they appear interbedded with crudely bedded mudstones.
At the base of Unit B, a meters-thick, massive sandstone bed transitions upward into an alternation of mudstone and cm- to dm-thick, massive to graded sandstone beds, which are then overlain by massive mudstones. Higher up, cm- to dm-thick sandy beds indicate a renewed sand supply, followed once again by predominant mudstones, interspersed with m- to dm-thick marly beds (Fig. 8).
- Non-metric multidimensional scaling (NMDS) from abundance data applied to a large dataset of Pliocene assemblages (132 samples, 614 species: Dominici & Danise, 2023), including seven samples collected at Legoli (larger and outlined symbols).
The thick sandstone of unit C, well exposed in the excavation, exhibits an overall lobate geometry (Fig. 5A), resting unconformably on unit B (Fig. 5D–E). It is characterized by a complex internal architecture (Fig. 6C–F). Amalgamated, meter-thick beds of coarse-to fine-grained yellowish sands alternate with discontinuous, deformed cm- to dm-thick mudstone beds. Sandstones show a graded structure, with a clear partition between a lower coarser portion, rich in mud clasts and shell fragments, overlain by a finer-grained upper portion (respectively facies a and b in Fig. 6C–E) massive or with low-angle (sub-facies b1 in Fig. 6D) or horizontal lamination (sub-facies b2 in Fig. 6D), topped by a mudstone bed (facies c). Mudstone partings eventually coalesce, forming wedges that disrupt the primary sandstone structures (Fig. 6F). The lobate geometry of unit C is further defined by downlap terminations of the sandstone beds, which are laterally replaced by marly beds (Fig. 5A–C). The upper portion of the section, represented by unit D, once again consists of blue-greyish massive mudstones with dispersed shells of gastropod and bivalve shells.
Interpretation
The studied succession largely records deposition in a shelf setting below fair-weather wave base, interrupted by a phase of shallowing marked by bioturbated sands (lower part of unit B). A pulsatory arrival of sands from hyperpycnal flows fed by riverine floods or by rejuvenation of the depositional profile due to tectonism (unit C: Fig. 5D–E), does not necessarily imply a shallowing. Sandy density flows spreading onto the bottom, forming lobate bodies, can be invoked. The bed partition suggests traction in density-stratified flows where the lower portion congested with sediment, including mollusk remains eroded from coeval nearshore sandy bottoms and transported offshore (Fig. 7B), behaved as a sediment gravity flow, whereas the upper portion was characterized by traction in transcritical condition, as attested by low-angle to planar lamination (Fig. 7C–E). Deposition took place during sudden events, as testified by water-escape structures and other features disrupting the primary sedimentary structures when sands were still loaded with water (Fig. 7E-F). Regarding the lowermost sandy interval, the unsorted nature of the mollusk assemblage attests to a shoreface setting. The background muddy deposition that formed the largest part of the Belvedere succession (unit A, upper part of B, D) resulted from diluted density flows and suspension settling. Biogenic shell beds (Fig. 7) attest to low-energy conditions below storm-weather wave base. Marly beds may record periods of starvation in the flux of clastic supply to the basin (Fig. 7A), but the lack of laterally continuous densely-packed biogenic shell beds testifies to an average high sedimentation rate.
Palaeoecology
Unit A
Two samples were collected in a laterally-continuous shell bed in Unit A, cropping out on the flanks of the road that leads to the waste plant (RD1, RD2: Fig. 4, Tab. 1), yielding a total species richness (S) = 37. The overall assemblage is dominated by encrusting, cemented or byssate suspension feeders, including the vermetid gastropod Petaloconchus glomeratus (about 45%) and the bivalves Neopycnodonte cochlear, Talochlamys multistriata, Chama gryphoides and Barbatia barbata (totalling 13-27%). Shallow infaunal suspension feeders include the gastropod Helminthia tornata and the bivalves Corbula gibba and Centrocadita rudista (making 13% to 27% of the total abundance). Carnivores are well represented by Tritia prysmatica, Cochlis raropunctata and Ocinebrina imbricata (2-5%).
The association suggests an open shelf palaeoenvironment and a low sedimentation rate, favoring the formation of a vermetid construction that hosted a well-diversified mollusk fauna. Petaloconchus reefs are known in the Mediterranean from the Upper Miocene to the Pleistocene, having been functionally replaced in the Holocene by Dendropoma reefs. Depth ranges from inner shelf (30–50 m) to slope depths (Vescogni et al., 2008). The abundance of Neopycnodonte cochlear (= N. navicularis) suggests a mid- or outer shelf setting, where this species is an important bioconstructor (Cardone et al., 2020).
Unit B
The two associations found in the samples collected in the lower part of the unit characterize the two different lithologies with which they are associated. Unsuitable conditions (hardened mudstone matrix, vertical outcrop) prevented the collection of a sufficient number of fossils for a quantitative comparison in the mudstones overlying sample UF2.
The lowermost sample UF1 is dominated by the shallow-burrowing suspension-feeding bivalves Chamelea gallina and Glycymeris nummaria, which together total 39% of abundance. These species and the carnivore gastropod Neverita olla (3,7%) are currently found in shallow-water sandy bottoms of the shoreface, where they typically occur at depths of 5-10 m (Legac & Hrs-Brenko, 1999; Huelsken et al., 2008; Cardone et al., 2020; Grazioli et al., 2022).
Overlying sample UF2 has a completely different association, largely dominated by the five gastropod species Oligodia spirata, Aporrhais uttingeriana, Tritia semistriata, Gemmula contigua and Comitas dimidiata, together accounting for 87% of total sample abundance. All of them are extinct, and the only closely related species in the modern Mediterranean is Tritia ovoidea, a dominant species in muddy offshore bottoms at mid- to outer-shelf depths (Moya-Urbano et al., 2019; see also Dominici & Danise, 2023).
In the remaining part of unit B, fossils are scattered or aligned along shell pavement. The matrix is hardened and we could not collect samples for a quantitative analysis. Recurring species, such as Oligodia spirata, Aporrhais uttingeriana, Tritia semistriata, Comitas dimidiata, Ringicula buccinea, Venus nux and Corbula gibba, are also recorded in samples UF2, CC2 and CC7 (see below). Other species found along the succession include bivalves of the species Limopsis aurita and spatangoid echinoderms.
Unit C
Shells abound in the lowermost sandstone of unit C, but are always fragmented and too fragile to be collected. Sturdier fragments allow for the recognition of species such as Ostrea edulis, Pecten jacobaeus (Fig. 6B), Glycymeris nummaria and Chamelea gallina, confirming that these bioclasts were transported basinward from sandy bottoms of the shoreface, where such species commonly lived.
Mollusks of the mudstones of unit C are scattered, but more frequent than in mudstones of unit B. Bivalves, including Lembulus pella, Glossus humanus, Myrtina meneghinii, Costellamussipecten cristatum, Anadara diluvii and an undetermined tellinid (Fig. 7B,C), are almost invariably in life position. The serpulid Ditrupa arietina is common in the lowermost part of the unit and gastropods such as Cochlis raropunctata and Aporrahais uttingeriana are common throughout the succession.
Two samples were collected in the upper part of the unit (CC2, CC7: Fig. 4, Tab. 1), corresponding to two horizons where shells are more frequent and exhibit a loosely-packed biofabric. Species in life position include shallow burrowers such as bivalves Centrocardita rudista and Pelecyora islandicoides (Fig. 7D,F) and incrusting gastropod Thylacodes arenarius, forming small constructions (Fig. 7E).
The analysis of the two samples reveals a recurring association of gastropods Cochlis raropunctata, Tritia semistriata, Oligodia spirata and Ringicula buccinea, bivalves Anadara diluvii, Costellamussipecten cristatum, Venus nux and Myrtina meneghinii and scaphopod Dentalium rectum, totalling about 60% of the total abundance in each sample. Gastropod Comitas dimidiata is instead present only in CC2 (15%) and scaphopod Antalis fossile only in CC7 (25%).
This association is richer, but functionally similar to that found in sample UF2 and higher in unit B. As in unit B, the fauna of unit C is dominated by extinct species, but Venus nux and Tritia ovoidea, a close living relative of Tritia semistriata, indicate modern muddy offshore bottoms at mid- to outer shelf depths.
Multivariate analysis
The multivariate Q-mode comparison of species-level abundance distributions among the Legoli samples and 164 other quantitative samples previously collected with analogous methodology in the Pliocene of the Mediterranean area (Dominici & Danise, 2023) confirms the palaeoenvironmental interpretations based on the autoecology of the dominant species found at Legoli. In the MDS multivariate plot, axis 1 represents palaeodepths, and axis 2 represents vegetation cover (at least in shallow depths). In particular, sample UF1 represents the shallowest palaeocommunity, aligned with other shoreface assemblages along axis 1, but with fewer species indicating bottom vegetation (see Dominici & Danise, 2023). UF2, CC2 and CC7 record the deepest palaeosettings, at outer shelf depths. RD1 and RD2 occupy an intermediate position along axis 1, indicating mid-shelf depths (Fig. 8).
Biostratigraphy
The abundance of the calcareous nannofossil assemblage ranges from rare to common, with greater abundances in Unit B. The most significant taxa are represented by Calcidiscus leptoporus, C. macintyrei, Coccolithus pelagicus, small Gephyrocapsa, Helicosphaera carteri, H. sellii, Pseudoemiliania lacunosa, Reticulofenestra spp., Rhabdosphaera spp., Syracosphaera spp., Umbilicosphaera spp. and discoasterids. The latter are never abundant, and become rare in the upper part of Unit D. The genus is mainly represented by Discoaster tamalis, D. asymmetricus, rare D. surculus, and very rare, scattered D. pentaradiatus and D. brouweri. Their quantitative distribution (Fig. 10) provides more detailed information on their abundance through the studied interval.
Based on the occurrence of D. tamalis, P. lacunosa and H. sellii, and the absence of R. pseudoumbilicus and Sphenolithus spp., the succession is assigned to biozone MNN16a (Rio et al., 1990). The very rare, discontinuous occurrences of D. pentaradiatus, ranging between 0 and < 1 specimens/mm2 (Fig. 10), suggest that the studied interval falls within the paracme interval of D. pentaradiatus (Driever, 1981; Di Stefano, 1998). This evidence supports the inference that the succession falls within the lower part of Zone MNN16a and specifically above the Last Occurrence (LO) of R. pseudoumbilicus (dated at 3.84 Ma; Raffi et al., 2006) and below the end of the paracme of D. pentaradiatus (dated at 3.56 Ma; Di Stefano, 1998), at the Zanclean/Piacenzian transition. The identified interval correlates with the Subzone MNN15a of Di Stefano et al. (2023).
Planktonic foraminifera, in the lower part of Unit B, show varying states of preservation and low diversity; three samples were poorly preserved or barren. The assemblages are well preserved and diversified in Unit C and Unit D.
Among the investigated key species, Globorotalia puncticulata shows the highest abundances and a more continuous record within Unit B (Fig. 10). The taxon has the highest abundances in the lowermost portion of the section, but it strongly decreases from about 15% to about 2.5 % of the total assemblages, in the uppermost part of Unit B (Fig. 10). Rare occurrences of G. puncticulata are also found within Unit D. Sample CC1 (at 56 meters in the section) marks the LO of G. puncticulata. Starting from the uppermost part of Unit B, Globorotalia crassaformis occurs within the assemblages, although it becomes scattered along Unit D. The First Occurrence of G. crassaformis is found in sample LL86 at 25.7 meters in the section.
The distribution of G. puncticulata and G. crassaformis supports referring the studied interval to the upper part of MPL4a and lower part of MPL4b subzones (Lirer et al., 2019). The boundary between the subzones is marked by the LO of Globorotalia puncticulata dated at 3.57 Ma, in the Mediterranean (Lourens et al., 2004; Lirer et al., 2019), which closely approximates the Zanclean- Piacenzian boundary and the magnetic polarity transition from the reversed Gilbert (C2Ar) to the normal lower Gauss dated at 3.60 Ma (C2An.3n: Lirer et al., 2019).
Magnetostratigraphy
The average NRM intensity is 14.56 x 10-6A/m (ranging from 0.74 to 213 x 10-6A/m; (Supplementary Information: Table 1 and Fig. 9e), and the bulk susceptibility is 140 x 10-6 SI (ranging from 13 to 262 x 10-6 SI; Supplementary Information: Table 1 and Fig. 9f). These values co-vary in the lower part of the section, with relatively higher values in the first 4 meters (at the base) and at 16 meters, but they do not appear to co-vary in the upper part of the section.
- A) stratigraphic units; B) Declination and C) inclination calculated with PCA Analyses where black circles are the data for every sample and blue circles are the results averaged by depth. D) VGP, Virtual Geomagnetic Pole, Latitudes calculated for each depth with positive/negative values indicating normal/reverse polarity. E) NRM, Natural Remanent Magnetization and F) susceptibility for each measured sample.
In Unit B, AMS results show a typical sedimentary fabric, with a k3 oriented sub-vertically, while k1 and k2 forming a girdle around the sub-horizontal plane (Fig. 9g and Supplementary Information: Fig. S2). The AMS tensor in unit B is gently tilted of about 20-25° towards the WNW. Similarly, in Unit C, the mean AMS tensor is still typical of a sedimentary fabric and appears tilted towards WNW of ca. 10-20° (Fig. 5d,e and Fig. 9g). The orientation of Unit B and C AMS tensors is consistent with the orientation of the base of Unit C, suggesting a gentle post-depositional tectonic effect on the entire section (with a shallowing effect towards the top).
We isolated the ChRMs in 172 out of 344 total samples from 41 depth levels (black dots in Fig. 9b and c; Supplementary Information: Table 1). However, only depths with three or more ChRMs allow for the calculation of the Fisherian mean for the given depths, which are considered robust if α95 is less than approximately 20°. In total, palaeomagnetic results from 27 depths are considered robust (blue dots in Fig. 9b,c and d; Supplementary Information: Table 2). From these data, we calculated the VGP (Virtual Geomagnetic Pole; Supplementary Information: Table 2) using pmag.dia_vgp.py (PmagPy, 2025) and interpreted negative VGP latitudes as reversed polarity and positive VGP latitudes as normal polarity. Thus, we identified a magnetic polarity transition between 58 and 64 meters in the section.
By integrating bio- and magnetostratigraphic results, it is possible to infer that the identified magnetic polarity transition corresponds to the reversed Gilbert Chron (C2Ar) transitioning to the Gauss normal Chron (C2An.3n), dated at 3.6 Ma (Fig. 10)
- Sedimentary log, bio- and magnetostratigraphy at the Belvedere waste plant (Legoli).
DISCUSSION
The integrated stratigraphic analysis of the Belvedere section resulted in a detailed resolution of the portion of the Pliocene infill of the northern sector of the VVB that may be compared with adjacent areas of VVB, and with adjacent basins (Fig. 11).
- Stratigraphic scheme for the infills of the VVB and adjacent Tuscan basins. From the left: stack of benthic δ18O records (Lisiecki and Raymo, 2005, strong glacials are marked in light blue, in red the middle Piacenzian Warm Climate (mPWP: Dominici & Danise, 2023); magnetostratigraphy and Mediterranean foraminifera biostratigraphy are from Lirer et al. (2019), calcareous nannofossils from Rio et al. (1990). VE: Valdelsa Basin (adapted from Benvenuti et al., 2014; Nalin et al., 2016, and references therein); SVS: sabbie di San Vivaldo; FAA: Argille Azzurre; VLS: formazione di Villamagna; VL: calcari di Volterra (Costantini et al., 2000); PF: formazione di Pianosa (Foresi et al., 2008); PRL: biocalcareniti di Parlascio (Mazzanti, 2016); FAAs: sabbie di Mazzolla; SRZ: conglomerati di Serazzano (Lazzarotto et al. 2002); PCG: conglomerati di Strolla (Bossio et al., 1991; Abbazzi et al., 2008).
In the southern sector of VVB planktonic foraminifera biostratigraphy from samples collected ESE of Volterra (Fig. 1; Giannelli et al., 1981), suggested to place the Zanclean–Piacenzian boundary in the upper part of a mudstone succession, up to 600 m-thick (Argille Azzurre), around the MPL4a-MPL4b boundary. The mudstones were overlain, through a sharp boundary, by a sandy mudstone unit that we can attribute to Zone MPL4b (as defined by Lirer et al., 2019), or to MPL4c, should the presence of G. bononiensis be common (no analytical data available from the literature). This is in turn overlain by a thick calcarenite, also Piacenzian in age (calcari di Volterra: Giannelli et al., 1981; Costantini et al., 2000). Planktonic foraminiferal data collected E of Volterra (Monte Voltraio section; Fig. 1) can similarly be attributed to Zone MPL4b, according to the zonal scheme of Lirer et al. (2019), based on the co-occurrence of G. crassaformis crassaformis, G. aemiliana, and G. bononiensis, albeit at low frequency. According to the published analytical data (Fig. 2 in Bianucci et al., 1998), the presence of Zone MPL4c in the upper part of the clay sequence cannot be ruled out, where G. bononiensis becomes more frequent.
In the NW sector of the VVB, onto the western shoulder of the basin, an analogous biocalcirudite unit crops out at Parlascio, near Casciana Terme (Fig. 1). According to Nalin et al. (2016), the carbonates of the Casciana Terme unit were deposited in an interval between the FCO of G. bononiensis and the upper limit of the magnetochron C2An.2n. Thus, in zonal terms, from the MPL4c Zone to the lower part of the MPL5a Zone (zonal scheme of Lirer et al., 2019). The calcareous nannofossils implied the biozone MNN16, above the highest occurrence of Sphenolithus spp. The Legoli section is located about 20 km to the W of Parlascio, on the opposite shoulder of the VVB and in a symmetrical position with respect to the basin depocentre (Figs. 1 and 3). The stratigraphic position on top of the Argille Azzurre suggests a stratigraphic correlation between units D and E and the Casciana Terme Unit. The Legoli succession and available data on the VVB therefore indicate that the unit, including the calcari di Volterra, rests on an erosional surface and that it correlates with other similar bodies encountered from NW to SE. Calcarenites rest on top of the MPL4b–MPL4c boundary at Casciana Terme, within MPL4c at Legoli, Volterra and Voltraio, the latter on the eastern shoulder of the VVB and in proximity of the Villamagna fault (Fig. 1). In nearby basins, calcarenites may be coeval with Parlascio (calcarenite di Buonriposo in the Valdelsa Basin, near San Gimignano: Capezzuoli et al., 2005; see also Milaneschi et al., 2024). However, in structural highs they occur also in the lower part of MPL4b or in MPL4a, like at Pianosa, in the Northern Tyrrhenian, SW of Elba Island. Eastward, in the Valdelsa Basin at Montegabbro, calcarenites are referable to the lower part of MPL5a of the scheme of Lirer et al. (2019: Milaneschi et al., 2024), confirming that calcarenites are time-transgressive on a larger scale (Fig. 11).
The C–D boundary at Legoli — correlative with the sharp surface separating the Argille Azzurre from the formazione di Villamagna at Volterra — is therefore here considered as part of a major unconformity formed around 3.3 Ma (MPL4b–MPL4c boundary), in coincidence with Marine Isotope Stages M2 (De Schepper et al., 2014) and with a major global sea-level fall (Dumitru et al., 2019; see also Nalin et al., 2016). This unconformity can be correlated with the S3–S4 boundary recognized in the Valdelsa Basin (Benvenuti et al., 2014; Dominici & Danise, 2023; not at the base of the Piacenzian, as indicated by Mirabella et al., 2022), marking an anomalously cold stage described as “a failed attempt at Northern Hemisphere Glaciation” (Vega et al., 2020) (Fig. 11).
In parallelism with the Valdelsa Basin infill, the Pliocene succession underlying the S3-S4 unconformity was deposited before 3.3 Ma, thence before the onset of glacial-interglacial cyclicity paced by eccentricity-modulated orbital precession (Grant et al., 2019) and was controlled instead by regional tectonism (Benvenuti et al., 2014; Dominici & Danise, 2023). A tectonic signal is evidenced at Legoli by a general dip of the section of about 10-20° towards the WNW (Fig. 5). This observation is corroborated by the AMS mean tensors of Unit B and C (Fig. 9g and Fig. S2) which suggest a similar tilting. The sandy lobe at the base of Unit C shows an angular unconformity and is of limited extent and with shells transported from the shoreface (Fig. 6). This was deposited in an offshore setting at the same depth, as documented by the palaeoecological analysis in the mudstone below (samples LG, CC2 and CC7 in Fig. 8) and above (sample UF2 in Fig. 8) the angular unconformity. As reported in Poneti et al. (2024), the Tortonian-Piacenzian tectono-sedimentary evolution of the VBB was characterized by an acme of compressive deformation during the Messinian that progressively attenuated during the Zanclean. A collapse of the basin shoulders occurred during the Piacenzian, driven by normal faults such as the Villamagna fault. The angular contact between units B and C in the Belvedere section, clearly visible thanks to the excellent outcrop conditions (Fig. 5), may attest to an early fault activity around the Zanclean-Piacenzian boundary.
A broader significance is assigned to the sandy interval at the base of unit B, as this is characterized by a biogenic shell bed from a shoreface palaeoenvironment (UF1 in Fig. 8). This body is laterally continuous, cropping out also eastward of the Belvedere waste plant (Mazzanti, 1961; Dominici et al., 1995), interposed between two mudstone units that rest on top of the sabbie di San Vivaldo (Zanclean: Costantini et al., 2000). The mudstone underlying the sandy unit, attributed to MPL4a (Costantini et al., 2000), corresponds to unit A, the lowermost cropping out at Legoli (Figs. 4,9,11). The base of sandstone of unit B is therefore interpreted as the base of a depositional sequence and tentatively correlated with the lower boundary of synthem S3 recognised in Valdelsa (Benvenuti et al., 2014). Its age remains unconstrained, here as well as in the Valdelsa Basin where, at Borro Strolla, Argille Azzurre are referable to MPL1-2 (Abbazzi et al., 2008; Synthem 1 in Benvenuti et al., 2014). West of Volterra Argille Azzurre are up to 1,000 m-thick, and South of Volterra, near Mazzolla, they rest on top of the Messinian and include Zones MPL1-MPL3. Here marine clays are intercalated with conglomerates in the lower part (Serrazzano conglomerates) and to sand units in the upper part (Mazzolla sands: Lazzarotto et al., 2002) (Fig. 11).
The results of our study provide a reasonable attempt to link the surface geology with the buried seismo-stratigraphic architecture of part of the Pliocene basin fill described in Poneti et al. (2024). The lower sequence boundary of sequence Z4, recognized in seismic lines in proximity of Legoli (Fig. 3), correlates with the lower boundary of S3 as described in Valdelsa (Fig. 2). The boundary separating Z4–P1 is the minor unconformity temporarily interrupting the general transgressive interval of S3 (that is, the angular unconformity at the base of sandstone C), a tract covering the Zanclean–Piacenzian boundary. Finally, the boundary between P1–P2 seismic sequences corresponds to the boundary between S3–S4 synthems of Benvenuti et al. (2014) (Fig. 2).
CONCLUSIONS
Facies analysis and quantitative mollusk palaeoecology provided reconstruction of the sedimentary dynamics controlling the deposition of the Pliocene succession freshly exposed at the Belvedere waste plant, near Legoli (PI), in the northeastern sector of the Valdera-Volterra basin. The succession is thus subdivided into allostratigraphic units formed by sandstone-mudstone couplets, m or tens of m-thick. By integrating biostratigraphic data based on the analysis of calcareous plankton (foraminifera and nannofossils) with magnetostratigraphic data, we identified magnetic polarity transition from the reversed Gilbert Chron (C2Ar) to the lower Gauss normal Chron (C2An.3n), dated at 3.6 Ma and thus the Zanclean- Piacenzian boundary. Thanks to chronostratigraphic constraints recognized at Legoli and at other sections of the Valdera-Volterra basin, we recognize here synthems S3-S4 previously defined in the sedimentary infill of the nearby Valdelsa basin. These two allostratigraphic units were deposited during the latter part of the Zanclean and in the early-middle Piacenzian and display evidence of sedimentary dynamics during a crucial interval in Earth’s history at the onset of the first signs of a Northern Hemisphere glaciation. The study shows a tectonic tilting of the section and confirms that in Tuscany this interval is marked by a depositional sequences controlled by increased sediment accommodation due to basin subsidence, possibly characteristic of a thrust top basin with an adaptive accommodation space towards a middle and upper Piacenzian basin infill.
ELECTRONIC SUPPLEMENTARY MATERIAL
This article contains electronic supplementary material which is available https://doi.org/10.3301/IJG.2026.02.