O/C Isotopic and EPR Signature of Marble from the Apuan Alps (Italy): A Critical Review

Abstract

A critical review of studies concerning the attribution of the provenance of marble from the Apuan Alps (Italy) (AAM) used for historical–monumental buildings and artefacts is proposed based on its O/C isotopic and EPR signature. First, a summary of the geological origin of AAM and its geo-structural evolution and setting is presented. A review of the exploitation history of AAM is then discussed. This geological and historical information is used as categorical information to better constrain the literature multimethodic database, containing numerous data, including O/C isotopic and EPR spectroscopic parameters. A robust multivariate statistical analysis of the combination of all these data is performed. The results point to the fact that the O/C isotopic and EPR signature can help in attributing an analysed AAM sample to a marble extraction district, and to a certain extent also to a site, whereas the discrimination of the individual quarry appears to not yet be achievable.

1. Introduction

The use of the worldwide-famous Apuan Alps marble (AAM) as a dimensional and ornamental stone for outstanding sculptures, artefacts and cladding of buildings started with the Romans in the I century B.C. For two centuries, the use of marble from Luni (Marmor Lunensis), the Roman harbour from which the marble was shipped to the whole Empire, overrode all other marble across the Mediterranean area [1].

In the II century A.D., its use decreased, and it ceased in the V century with the fall of the Western Empire. In the XI century, the use of AAM started again and reached its fame in the Renaissance with Michelangelo’s outstanding artefacts [1,2,3,4,5,6].

In archaeology and art-history, it is very important to define the type of marble, and the relative provenance, of a certain artefact or building for both knowledge and conservation purposes. Three different approaches can be used in determining the type of stone material:

-

Documental analysis carried out on commercial and historical text and archives that report the orders, origin and types of stone in the materials used.

-

Technical analysis including petrographic, mineralogical, geochemical and isotopic analyses allowing the attribution to a lithotype based on existing databases (DBs).

-

Expert evaluation carried out by professionals, based on their knowledge and regarding lithology, stone grain size, texture and warp.

An attribution is reliable if more than one of these approaches point towards the same attribution. Where there is a lack of historical documents, defining the provenance of historical marble through only skilled expert judgment is not always an easy task; therefore, reliable technical analysis can help in the attribution.

In the last several decades, the C and O isotopic ratio has emerged as a discriminating technical analysis suitable for the correct attribution of the provenance of marble [7,8,9,10,11,12]. This analysis discriminates on the basis of the C/O isotopic ratio, a parameter that is established during the sedimentary process, is a function of the climate of that time and can be altered by metamorphism [11,12].

As reported by the abovementioned authors, the factors in play are as follows:

-

The origin of the carbonate: chemical precipitate or organic mud.

-

The composition of water present during diagenesis and its later history.

-

The temperature of metamorphism and its thermal gradient.

-

Fractionation with pore waters and other mineral phases during metamorphism.

A uniform isotopic composition can be obtained if the following criteria are met:

-

The protolith was deposited and underwent diagenesis in a uniform environment.

-

Isotopic equilibrium was attained and maintained during sedimentation or metamorphism.

-

The marble unit is homogeneous, preferably almost pure carbonate, and thick enough.

-

The metamorphic gradient was not too steep.

The principal variations can be related to the following features:

-

The exchange between the carbonates and accessory silicates or oxides within the marble or near its contact with other formations.

-

A steep metamorphic gradient; a thermal gradient of 15 °C/km or less is not a problem, while a gradient towards 30 °C/km results into a great range in δ¹⁸O values.

-

Significant weathering of the analysed sample.

From these points, high constraints result for the correct attribution of a marble sample based solely on isotopic analysis; this implies the need for a deep knowledge of the geological history and context of the analysed materials before proposing reliable attribution.

In the last decades, Electron Paramagnetic Resonance (EPR) investigations, exploiting the weak paramagnetic properties of marble, have also been profitably added to the pool of techniques able to aid in the determination of marble’s origin [13,14,15]. The EPR determination proceeds through the comparison of features arising from (1) hyperfine interactions, (2) fine interactions and (3) the line width of the spectrum of Mn(II) replacing Ca in calcite with those of a set of reference rocks. These latter are chosen as unweathered representatives of a specific locality and a specific lithogenesis. The reliability of both of these two types of analysis depends on an extended DB of certified measurements of the several possible sites of provenance of the many types of historical marble used during the centuries.

A recent review paper [11] warns about the limits of these analyses and fosters the importance of multiple convergences among technical data, expert evaluation and documentation.

At present, for AAMs, namely marble from the Carrara, Massa, and Seravezza districts, the existing DB [10,15,16] numbers 111, 1 and 57 analysed samples, respectively. For all references, isotopic and EPR data and the quarry from which the sample was collected are registered, together with a number of additional parameters and other information.

In the present study, this DB has been implemented with additional categorical information regarding the geological context; namely the marble type and the connected sedimentary facies, and the structural levels which present slight differences in tectono-metamorphic parameters (temperature and pressure).

This implemented DB has been analysed through a robust statistical approach to verify the capability of the dataset to discriminate the provenance of AAMs within the different exploited districts.

2. Material and Methods

2.1. Geological Setting

Geological studies carried out in recent decades in the Apuan Alps (Northern Apennines, Italy) have clarified the genesis and technical characteristics and cultivation history of the marble present there [6,17,18,19,20,21,22,23,24,25,26]. The main features related to the AAMs’ origin and setting are here outlined.

2.1.1. Sedimentary Protolith

The AAMs derive from the tectono-metamorphic deformation of carbonate-platform sediments deposited in the Hettangian about 201–199 Ma ago (absolute ages are from [27,28], in a marginal environment: as today is the area of the Persian Gulf. This platform was located on the western side of the Tethyan Gulf and covered a large area now dismembered by the Apennine orogeny into the several tectonic units constituting the Alpine–Apennine–Maghreb orogenic belt (in the Northern Apennines: Apuan Alps unit, Tuscan Nappe, and Umbrian–Marche units) [29,30].

This type of carbonate platform presents various morphological and sedimentary environments, corresponding to different sedimentary facies; from the outside to the inside, they consist of the following:

-

External ramp with carbonate muds deposited under the wave base.

-

High-energy zone with oolithic and coarse well-sorted and -washed sediments deposited in high-energy shores and tidal channels.

-

Internal lagoon with fine and pure carbonate mud sedimentation.

-

Restricted lagoon with fine and pure carbonate mud sedimentation containing residues of abundant organic and oxide material.

-

Tidal zone with carbonate-dolomitic deposits.

-

Emerged and karst areas with karst features and residual soils.

The evolution of the AAM platform lasted about 2 My and evolved vertically and laterally in a same geological environment. General warming from the Hettangian to the Sinemurian occurred, as shown by the variation in the oxygen isotope, with δ¹⁸O carb ‱PDB grading from −1.9 to −1.4, and other features, while the carbon isotope δ¹³C carb ‱PDB grades ranged from 0.1 to 2.1 [28]; these data are common for the whole Tethyan Gulf carbonate platform. The Calcare Massiccio of the Tuscan Nappe, which contiguously deposited in the same context, represents the protolith of the AAMs. The authors of [21,22], on the basis of a field map of the whole Apuan Alps area at the scale 1:5000, studied the AAMs’ lithofacies and distribution and defined a regular sequence of facies grading from south to north from external high-energy facies to low-energy inner lagoon to intertidal flat.

The AAMs lie above a previous carbonate-dolomitic platform (namely Grezzoni) developed in the Upper Triassic (about 230–201 Ma ago). The lower portion of the AAMs consists of dolomitic marble referable to an intertidal flat and is followed upward by a fully developed carbonate platform. In the Late Hettangian–Sinemurian, this carbonate platform underwent faulting and sinking, grading to breccia and mega-breccia complexes and/or shallow water marly limestone (Ammonitic Rosso) and successively to pelagic cherty limestone (Calcare Selcifero) [17,31].

2.1.2. Tectono-Metamorphic Deformation

During the Tertiary period, the whole Apuan Alps sedimentary sequence underwent tectono-metamorphic deformation, which resulted into superimposed deformation events (see updated reviews in [26,32,33]). The tectono-metamorphic deformation lasted about 15 My (27 to 12 Ma) and gave rise to a complex tectonic assemblage resulting in a series of first-order ductile (NE-E verging) nappe-folds. Namely, they are the Carrara Syncline, Vinca Anticline, Orto di Donna–Mt. Corchia Syncline, Tambura Anticline, and Arnetola Syncline, with the crest of the late uplifting dome along the alignment of Boana–Mt. Corchia [17,18,31] (Figure 1).

The tectono-metamorphic deformation occurred at low-grade metamorphic conditions (general condition with T at 300 to 450 °C and P up to 0.6–0.8 GPa), which resulted in green-schist facies (pirofillite + phengite-muscovite + quartz + albite + biotite + chlorite ± epidote ± chloritoid) for pelites. These conditions correspond to a depth of about 12 km [18,31,34].

Deformation rates ranged from 0.8 to 2.5 mm/y (2.5 to 7.9 × 10⁻¹⁵ cm/s) [34,35,36], with finite strain ratios varying from 4.0/1 to 12.5/1 with average values of 7.0/1. The authors of [33,37] argue for a deformation path reaching 30 °C/km of geothermal gradient, but below 35 °C/km according to [38].

The authors of [39,40] envisaged that in the early tectono-metamorphic deformation stage, the uppermost (west side) AAMs were located in the deepest positions at higher equilibrium temperature (430 °C) with respect to the lowermost (east side) AAMs that were in a higher position at a lower temperature (380 °C). This situation reversed during the late tectono-metamorphic deformation stage when the folds-stack was formed with a difference in T from the top (west) at 340 °C to the bottom (east) at 380 °C.

During these tectono-metamorphic deformation phases, strong transposition events occurred in the axial-plane high-strain zones. The events resulted into the formation of sheets of “fault-rock”, the layering of which especially marks the synform–syncline axial plane.

Deformation occurred with no evidence of hydrothermal and exotic fluids flowing through the AAM body [41]. Post 8 Ma, during uplifting and exhumation, as core complex, the AAMs were cut in the ductile–brittle transition stage by crack veins filled by spathic calcite in the core of the marble body, and by mixed fluids near the contacts with the other embedding lithologies at 370–380 °C and 0.30–0.325 GPa [37,38]. Later, brittle sub-vertical fractures occurred; no faults are presents in the body of the Apuan Alps metamorphic core complex.

The rock composition, consisting almost totally of calcite (>97%) in most of the AAMs, prevents assessment of whether the dynamic microfabrics overprint an early annealed fabric or if they are related to the earlier stages of deformation surviving the late static recrystallisation and grain growth [18,40].

Calcite flow laws suggest that, under P/T conditions, the AAMs deformed by grain-size-sensitive super-plastic deformation mechanisms [42,43,44]. This means that the AAM grain sizes resemble those of the original carbonate-platform facies.

The main elements of the mesoscopic tectonic fabric recognisable in the field [19,20,45] in the AAM body is the main tectono-metamorphic schistosity (S1) called verso by quarriers. The verso has the morphology of a continuous cleavage type 1, marked by the white and grey texture of the AAMs. The verso draws the main structures, anticlines and synclines and has the AAM types laying along it.

Detailed field work, at the scale 1:5000 [23], led researchers to recognise and map the several types of AAMs, namely Statuario, Ordinario, Venato, Venato forte, Venatino, Breccia, Bardiglio, Calacatta, Arabescato, Fior di Pesco, Paonazzo, Zebrino and Cremo, whose origin is related to both original sedimentary features and tectonic transposition. Their setting is not casual but follows the fold-schistosity assemblages.

These works also envisaged that the great structures of the Apuan Alps dip gently towards the north; therefore, in the north of the Apuan Alps the highest structural levels outcrop, and in the south the lowermost ones. The late uplift has its dome in the Boana area, where the deepest structural level is exposed. In total, a thickness of about 10 km of vertical fold-pile stack, exposed from Boana to the Carrara Syncline reverse limb, has been calculated.

2.1.3. Characteristic of the AAMs

Under transmission optical microscope (TOM) analysis, AAMs show a homeoblastic grano-xenoblastic texture, grain size ranging from 0.1 to 0.8 mm, sutured and indented grain boundaries, relics of twins and glide surfaces and sub-grain partitioning in the larger grains (>0.6 mm). New crystals free of strain have linear sharp boundaries and polygonal shapes [18].

All these features are differently distributed in the AAM lithotypes:

-

Marmo dolomitico: Dolomite is abundant, has a rhombohedral habit, ranges in size from 0.2 to 0.4 mm of diameter and shows sharp boundaries. In the marble layers, calcite has an average grain size of 0.1 mm; in this lithotype, there are also regions with larger calcite crystals (0.3 to 0.8 mm) containing twins and gliding surfaces. Calcite grain size decreases to micro-crystalline (<0.05 mm) where there are dolomite or second-phase particles. Boundaries are indented or suturated towards the Venato Forte and become progressively curved or gently curved toward the Venato.

-

Venato forte: Venato appears strongly heterogeneous in crystal size, ranging from <0.3 to 0.5 mm; some regions, aligned in the verso, are characterised by the presence of twinned crystals whose size is larger than average (0.2 to 0.3 mm). The grey bands aligned in the verso number particles of additional phases and the presence of regions of calcite micro-crystals (<0.01 mm) comprising totally, or in part, the grey veins. Dolomite is present as rhombohedral crystals close to phyllosilicate layers, ranges from 0.1 to 0.2 mm and in some cases possesses a core and rim structure. Phyllosilicates (white mica and chlorite) are widespread in the calcite matrix or concentrated in sub-millimetric layers within which there are abundant particles of accessory minerals with a 0.01 mm average diameter. Phyllosilicates are dimensionally elongated in the main schistosity and behave as strong fibres driving the shape of calcite, whose boundaries in this case are always very sharp and straight.

-

Venato: Calcite crystals range from 0.1 to 0.4 mm with boundaries gently concavely curved outwards or sharp and polygonal in the smaller ones. Laying parallel with the main schistosity, regions of larger calcite crystals (0.4 to 0.8 mm) occur, where calcite presents twins and gliding surfaces and indented or suturated boundaries. Veins result, enriched in dolomite and additional minerals. Dolomite crystals are generally larger than calcite ones and isolated from each other. Their boundaries are sharp, and rhombohedral gliding planes can be present.

-

Venatino: Calcite crystals have the same features as the Venato, but sizes are clustered at 0.2–0.4 mm and relics of twinning and gliding surfaces become rare. Dolomite crystals are rare, but always with a rhombohedral habit.

-

Breccia: The matrix appears to be formed by very small calcite crystals (<0.05 mm), accessory mineral particles and rarer dolomite with a rhombohedral habit (size 0.2–0.3 mm). Clasts consist of calcite crystals, with sizes ranging from 0.2 to 0.4 mm, which have the same features as in the Venato.

-

Ordinario and Statuario: They consist exclusively of calcite crystals, whose sizes are clustered around the mean (from 0.3 to 0.8 mm). Relics of twins and glide surfaces are common. Calcite crystals often show bands of light undulated extinction; the boundaries are fairly sharp but indented.

-

Bardiglio: Calcite crystals have sizes from 0.05 to 0.1 mm in the dark grey background, whereas they range from 0.2 to 0.4 mm in the white stains; quartz is present even if in very small amounts.

The authors of [21,22] analysed several lithological sections across the AAMs and related the main AAM types to the carbonate-platform facies on the basis of grain sizes and texture (

Table 1. AAM types, environmental attribution, feature and grain sizes (elaboration, M. Coli).

AAM Type	Facies	Features	Grain Size
Zebrino	External ramp (Ammonitico Rosso, marly calcareous facies)	Thick alternating of white and grey layers following the verso	White 0.2 to 0.4 mm Grey < 0.1 mm
Cremo	External ramp (Ammonitico Rosso, calcareous facies)	Pale pink	0.2 to 0.4 mm
Bardiglio	Restricted lagoon	Dark grey with white stains elongated in the verso	Grey < 0.1 mm White 0.2 to 0.4 mm
Statuario	Beach, tidal channels	Ivory white	0.6 to 0.8 mm
Ordinario	Inner lagoon	White, pale white, with pale grey stains	0.3 to 0.6 mm
Breccia	Fault scarp, sinking fractures	White clasts, grey matrix	Matrix < 0.05 mm Clasts 0.2 to 0.4 mm
Paonazzo, Calacatta	Karst and emerged areas	Pale white with pale grey and pale pink flames	Pale white 0.4 to 0.6 mm Flames 0.05 to 0.1 mm
Venato forte	External ramp	Thick alternating strikes of white and grey along the verso	White 0.2 to 0.4 mm Grey < 0.1 mm
Venatino	Inner lagoon shores	White with rare pale grey strikes along the verso	White 0.2 to 0.4 mm Grey < 0.1 mm
Venato	External lagoon	White with pale grey strikes along the verso	White 0.1 to 0.4 mm Grey < 0.1 mm
Marmo dolomitico	Upper intertidal flat	Alternating layers of pale grey marble and white dolomite marking the verso	Dolomite 0.2 to 0.4 mm Calcite 0.3 to 0.8 mm

).

Back-folding the Apuan Alps tectonic edifice [21,22] obtained a thickness for the AAMs of about 200 m, as for the Calcare Massiccio protolith of the overlying Tuscan Nappe. The restored paleogeography of the AAM Hettangian carbonate platform [21,22] led to an extension of about 25 km (NW-SE) for 45 km (SW-NE). This resumes a graduation from external facies to the south, to inner platform facies to the north, as for the Calcare Massiccio of the Tuscan Nappe [30,46], at that time in lateral continuity with the AAM platform. This similarity fosters the method and the attribution of the AAM types to the different carbonate-platform environments.

During the tectono-metamorphic deformation, strong transposition occurred in the axial-plane high-strain and -shear zones. That resulted in the formation of sheets of “fault-rock”, the layering of which especially marks the synform–syncline axial plane. In the AAMs, these sheets are “tectonic-origin marble” with the main synclines marked by these types of marble, namely the Nuvolato, rich in quartz due to the transposition into the marble of the overlying cherty limestone, and Arabescato and Fior di Pesco in which even the Late Cretaceous Varicoloured Schist is involved.

Indeed, the settings of AAM commercial types derive [18] from both tectono-metamorphic folding and refolding of the Hettangian carbonate platform (AAMs from sedimentary origin), as well as by transposition processes (AAMs of tectonic origin). The distribution in the AAM body of these different types of AAMs is a function of the carbonate-platform facies assemblages and how they were set in the folding by the tectono-metamorphic deformation for the AAMs of sedimentary origin, whereas the distribution of the main high-strain and -shear zones generated the AAMs of tectonic origin.

2.2. Synthesis

From the above short review of the geological knowledge about the genesis of the AAMs, these principal highlights follow:

-

The AAMs and the Calcare Massiccio formed contiguously in a wide Hettangian carbonate platform of the western side of the Tethyan Gulf.

-

This carbonate platform was organised into continuous environments grading from the external ramp, in the south, to the inner tidal flat, in the north; each one of these environments corresponds to a different facies and grain size that had some correspondence with the different types of AAMs.

-

During the Hettangian, the δ¹⁸O and δ¹³C isotope ‱PDB increased due to general warming, whereas it decreased in the Sinemurian.

-

Tectono-metamorphic deformation occurred at T at 300 to 450 °C and P up to 0.6–0.8 GPa, with a geothermal gradient of about 30–35 °C/km, with a slight difference between the lowermost and uppermost structural levels.

-

No fluid was added to the AAM core body during the tectono-metamorphic deformation, and fluids from the embedding formations were incorporated during the last ductile–brittle deformation stage in the boundary portions of the AAMs.

-

Some AAMs derive directly from the different carbonate platform facies, others from the Ammonitic Rosso, others from the ductile strong inter-fingering of marble and others from lithologies in high-strain shear zones.

-

For the O/C isotope analysis, the main features usually taken into account are the fabric, crystal boundary shape and Maximum Grain Size (MGS). Referring to the AAM features, as outlined above, discrimination by means of these features is difficult, and, in particular, the MGS as obtained for those analyses appears to have a very limited significance, even for attributing the sample to a specific AAM type, because the discriminating grain-size feature is the modal distribution and not the maximum.

Accordingly, the only sedimentary reason for differentiating O/C isotopic and EPR signatures for AAMs all across the whole Apuan Alps complex could be the facies. Slight differences can be found in the tectono-metamorphic P/T path in the tectono-geometric stack of the folds pile (Figure 2), which resembles a sequence of structural levels, here numbered from #1, the lowermost, to #9, the topmost.

Regarding the data in the isotope and EPR DB [10], the Carrara District is at the top and comprises more structural levels, which namely from top to bottom are the quarry sites of Boccanaglia (#9), followed in turn by Torano (#8), Miseglia (#7), and Colonnata (#6) at the bottom; but all these settings are above the Vinca Anticline. The Massa District comprises the quarry sites of Caglieglie (#6) and Frigido (#4); the Seravezza District comprises the quarry sites of Seravezza (#6) and, below the Vinca Anticline, the Altissimo, and Cervaiole sites (#4), in the hinge zone of the Orto di Donna Syncline. No analysis is available for the structural levels #1 (Boana), #2 (Arnetola), #3 (Focolaccia) and #5 (Orto di Donna).

Concerning the AAM types, for the Carrara District, samples from Bardiglio, Calacatta, Ordinario, Statuario, and Venato were analysed; for the Massa District, Ordinario; and for the Seravezza District, Ordinario, Statuario, and Venato. Moreover, Nuvolato at Carrara and Arabescato at Seravezza were analysed. Both of these last two are marble of tectonic origin. The list of types of marble and of the structural levels of the samples can be devised from

Table 2. List of the quarries in the Apuan Alps marble complex. Active exploitation during the Roman and Renaissance periods (A = active; NA = not active at that time). Categorical HISTORY parameter. Type of marble quarried, together with district, quarry site and structural level. In bold are the quarries exhibiting samples included in the original DB (elaboration, M. Coli).

Quarry	Roman	Renaissance	History	Type of Marble	District	Quarry Site	Structural Level
Artana	A	NA	1	Venato	Carrara	Colonnata	8
Bacchiotto	A	NA	1	Venato	Carrara	Colonnata	9
Battaglino	NA	A	2	Statuario-Ordinario	Carrara	Torano	8
Boccanaglia	NA	A	2	Paonazzo-Venato	Carrara	Boccanaglia	9
Calagio	A	NA	1	Nuvolato-Venato	Carrara	Colonnata	7
Canalgrande	A	NA	1	Venato	Carrara	Miseglia	7
Cervaiole	NA	A	2	Venato	Seravezza	Altissimo	4
Collestretto	A	A	3	Statuario	Carrara	Torano	8
Grotta Colombara	NA	A	2	Ordinario	Carrara	Torano	8
Corvaia-Ceragiola	A	A	3	Ordinario	Seravezza	Seravezza	6
Crestola	A	A	3	Statuario	Carrara	Boccanaglia	9
Crocifisso	NA	A	2	Venato	Carrara	Boccanaglia	9
Fossa del Cecchino	NA	A	2	Statuario-Paonazzo	Carrara	Boccanaglia	9
Facciata	A	A	3	Ordinario	Carrara	Torano	8
Fantiscritti	A	A	3	Ordinario	Carrara	Miseglia	7
Finestra	A	NA	1	Venato	Carrara	Miseglia	7
Finocchioso	NA	A	2	Venato	Carrara	Boccanaglia	9
Fondone	NA	A	2	Ordinario	Massa	Frigido	4
Fossacava	A	NA	1	Nuvolato	Carrara	Colonnata	7
Fusarolo	NA	A	2	Bardiglio	Carrara	Boccanaglia	9
Gioia	A	NA	1	Venato	Carrara	Colonnata	6
La Mossa	NA	A	2	Arabescato	Seravezza	Altissimo	4
La Para	A	A	3	Venato	Carrara	Boccanaglia	9
La Tagliata	A	NA	1	Ordinario	Carrara	Miseglia	7
Mandria	A	A	3	Cremo-Statuario	Carrara	Boccanaglia	9
Mortarola	NA	A	2	Venato	Carrara	Boccanaglia	9
Pianello	NA	A	2	Ordinario	Carrara	Torano	8
Piastraio	NA	A	2	Breccia-Bardiglio	Carrara	Seravezza	6
Poggio Dovizio	A	A	3	Statuario	Carrara	Torano	8
Polvaccio	A	A	3	Statuario-Ordinario	Carrara	Torano	8
Porcinacchia	NA	A	2	Statuario-Paonazzo	Carrara	Torano	8
Ravaccione	A	NA	1	Ordinario	Carrara	Torano	8
Ruggeta	NA	A	2	Calacatta-Bardiglio	Carrara	Boccanaglia	9
Scalocchiella	A	A	3	Ordinario	Carrara	Colonnata	6
Scaloni	A	NA	1	Ordinario	Carrara	Miseglia	7
Sponda	A	A	3	Statuario	Carrara	Boccanaglia	9
Strinato	A	NA	1	Ordinario	Carrara	Miseglia	7
Tacca Bianca	NA	A	2	Statuario	Seravezza	Altissimo	4
Tarnone	A	NA	1	Nuvolato	Carrara	Colonnata	8
Trambiserra	NA	A	2	Ordinario	Seravezza	Stazzema	6
Trugiano	A	NA	1	Ordinario	Carrara	Colonnata	7
Vara	A	NA	1	Venato	Carrara	Boccanaglia	9
Zampone	A	A	3	Statuario	Carrara	Boccanaglia	9

.

2.3. Exploitation History and Uses

The exploitation of the AAMs was occasional before the Romans [47] and systematically “proto-industrial” with the Romans [1]. AAM exploitation stopped with the fall of the Western Roman Empire, but started again in the XIII century and increased during the Renaissance [48], through the centuries up today [1,2,3,4,6]. From the XVI century, marble was also exploited from the area of Seravezza [6,48]; in the XIX century, the quarries were also opened on the eastern side of the Apuan Alps [49]. Recent studies [6,50] have allowed us to recognise the active quarries in Roman times and in the Renaissance at Michelangelo’s time and the types of AAMs quarried there (Table 2).

This framework must be taken into account in attributing a determinate artefact to a quarry that must have been active at the time of its realisation. For instance, it is well known that Michelangelo took its marble from the Carrara District, mainly from the Polvaccio quarry. Michelangelo also worked in the Seravezza area but, according to the literature, only collecting five columns, one of which alone reached Firenze [51]; according to this author, the first marble from the famous Tacca Bianca quarry at the Mt. Altissimo was extracted in 1569 for Giambologna.

On the basis of the available information on the historical period of exploitation of a single quarry, the list of all quarries presented in Table 2 has also been implemented with an additional categorical parameter. The historical information was thus categorised by three possibilities: active exploitation in Roman times or during the Renaissance, or both.

2.4. Database Definition

From the above cited review of the existing literature, we decided to operate an integration of an existing wide DB of isotopic and spectroscopic EPR information of several AAMs discriminated by sampling point (with specific reference to the district and the quarry) with the geological and historical information available on the same sampling points. In particular, the widest known DB on AAM results assembled by Attanasio and co-workers [10]. This DB consists of 8 categorical parameters to produce information on sample provenance and labelling, 19 numerical parameters linked to the quantification of EPR features, 2 isotopic numerical parameters (δ¹³C and δ¹⁸O), 3 organoleptic numerical parameters and 1 petrographical numerical parameter, for a total of 33 parameters. The entries of this DB, dedicated to the characterisation of the marble most commonly used in the classical age, number 1346. Among them, 169 pertain to AAM. The reader is referred to the original study [10] for further information concerning the experimental settings of the performed instrumental determinations.

We maintained several parameters out of the original setting, including the following: 3 categorical parameters (sampled quarry, district and marble group), 2 isotopic parameters, 4 EPR parameters (WAV, SPLI, SPREAD and DOLOM), and 1 petrographic parameter. To this set, 3 additional categorical parameters were added: TM, representing the main type of marble extracted in the sampling quarry; LS, representing the structural level of the sampling quarry; and HISTORY, relating to the information about the historical exploitation period. In particular, following the information reported in the above paragraph, TM can assume the entries V (Venato), O (Ordinario), A (Arabescato), S (Statuario) and N (Nuvolato).

LS can assume the values #4, #6, #7, #8 and #9 (with reference to the structural levels illustrated in Figure 2). HISTORY can assume the values 1 (Roman period), 2 (Renaissance period) and 3 (both periods). The reason for reducing from the original 6 EPR parameters coded by [52] to the 4 proposed here is mainly linked to the difficulty in achieving reproducibility in absolute quantitative EPR investigations, as described in, e.g., [53].

Analyses over the occurrence of spurious correlations in the DB were operated through the use of the software R-4.4.2 [54,55]. Principal component analysis of the final DB was operated using the Orange software R 3.3.38.1 [56].

3. Discussion

3.1. O/C Isotope Discrimination by TM, LS and HISTORY

The results of the O/C isotope show most of the data falling out of the range of the δ¹⁸O and δ¹³C reported for both the Hettangian and the Sinemurian by the Global Time Scale (Figure 3); moreover, there is a large overlap among the data from diverse quarries and districts. The data distribution has a close range for δ¹³C (1.06 ÷ 2.83) but a large range for δ¹⁸O (−3.01 ÷ −0.46). All of this suggests a strong alteration in isotope ratio due to diagenetic and or tectono-metamorphic features.

Despite no correspondence of the isotope data with the Global Time Scale being revealed, scholars have tried hard to find differences among the Carrara District quarrying areas: Boccanaglia, Torano, Miseglia, and Colonnata. From a geological point of view, any difference can be hypothesised in the carbonate platform from which these AAMs derive, apart the environmental facies, which grade from external to internal from south to north; this analysis, too, does not envisage significant differences among the AAMs (Figure 4).

Structurally, the Carrara District quarry sites of Colonnata, Miseglia, and Torano belong to the normal limb of the Vinca Anticline, whereas the strips of AAMs outcropping west of the Carrara Syncline (Boccanaglia) belong to its reverse limb and represent the structurally uppermost outcropping AAMs, at structural level #9 (Figure 2).

In the normal limb of the Vinca Anticline, more parasitic ductile folds are present in the marble body, with the syncline axial plane marked by the Nuvolato (cherty limestone “digested” into the marble during tectono-metamorphic transposition processes). The only difference is their diverse structural level, where the Colonnata quarry site is the lowermost (#6), followed upward by the Miseglia (#7) and Torano (#8) on top.

For the Massa District, which in the DB is included into the Carrara one, only one sample is in the DB and refers to structural level #4; in the DB, the Seravezza District has data from the quarry site of Stazzema (#6), and from the Altissimo and Cervaiole one (#4). Structural level #4 represents the structurally lowermost sampled level, because no samples were collected at the structural levels #1, #2, #3 and #5.

Also, considering these data, no discriminating difference arises for the O/C isotope distribution in the AAMs (Figure 5).

3.2. EPR Discrimination by TM, LS and HISTORY

Considering the three numerical EPR parameters SPLIT, SPREAD and WAV, one can easily observe a certain degree of internal dependence (Figure 6).

This phenomenon has already been suggested in the literature (e.g., [50]), and it can be attributed to the definition of the three parameters. They refer, in fact, to spectral features that arise from combinations among the fine and hyperfine interactions with the line width. However, the occurrence of internal dependence among the dataset asks for a reduction in its dimension to purely independent parameters. Here, this task was accomplished through the Compositional Data Analysis approach (CODA).

In this procedure, the sum of the three parameters is normalised, and the reduction in the three- to a two-dimensional parameter space was accomplished by the selection of two isometric coordinate balances [50]. These balances allow us to sort out the directions of maximal variability in the original 3D space. The balances resulting in the best expression of the original parameter variability and correlation resulted in the following:

$$\begin{array}{l}(1)\hspace{1em}irl1=\sqrt{{\displaystyle \raisebox{1ex}{$2$}\!\left/ \!\raisebox{-1ex}{$3$}\right.}}ln{\displaystyle \frac{SPREAD}{\sqrt{SPLI·WAV}}}\\ (2)\hspace{1em}irl2=\sqrt{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}ln{\displaystyle \frac{SPLI}{WAV}}\end{array}$$

The resulting projected 2D space exhibits a very clear linear correlation (Figure 7). When the cases are discriminated by the TM, LS and HISTORY categories, the discrimination results are again not complete. As an example, in Figure 7, the cases are plotted with reference to the LS category. Apparently, a certain degree of discrimination can be observed, e.g., between the cases of the four and seven stratigraphic levels. However, one cannot fully discriminate among all cases considering all of the categories.

3.3. Overall DB Analysis by TM, LS and HISTORY

From the above discussed information achievable considering the C/O isotope data or EPR parameter values only, a reliable discrimination of the three additional categories is not achievable. Accordingly, we decided to test the whole six-dimensional DB, including the petrographic MGS, the isotopic δ¹³C and δ¹⁸O and the reduced EPR irl1 and irl2 and experimental DOLOM variables through robust multivariate approaches.

A dendrogram evaluated over the whole DB has been obtained after having standardised (zero mean and unitary σ²) the variables and evaluated the Euclidean distances among cases (Figure 8). The chosen linkage is the Ward one. In the obtained dendrogram, the cases appear not to be fully discriminated by district. Namely, one can observe that some of the groups considered at the sixth level of branching (i.e., those highlighted in colour in Figure 8) still exhibit some cases not pertaining to the district representative of the group.

To obtain a deeper insight into the DB structure, we explored the internal variability of the DB through a principal component analysis (PCA) approach, developing a three-component model, always operating on the standardised data. The chosen model accounts for 79.0% of the overall variability of the DB. The results of the PCA are shown in Figure 9, where the first two components (PC1 and PC2, explained variability of 36.9% and 23.2%, respectively) are plotted and the cases discriminated by colour according to the chosen category. Indeed, in each single diagram, the colour code with which the cases are represented takes into account the different grouping hypotheses investigated: district, site, type of marble, stratigraphic level and use in history.

The diagrams in Figure 9a still confirm (and reinforce) the considerations emerging from the hierarchical clustering analysis of the dendrogram in Figure 8 concerning discrimination by district. Namely, Carrara appears almost well separated (even if not completely) from the group of Massa and Seravezza. The latter, in turn, are practically not discriminated. It is worth highlighting that a region of certain sample attribution to Massa/Seravezza can be highlighted at high PC1 and PC2 values; on the other hand, the Carrara region still contains cases pertaining to the other districts.

By focusing onto the quarry sites in Carrara, the Altissimo, Boccanaglia, Colonnata, Frigido, Miseglia, Seravezza, and Torano cases are not discriminated from each other (Figure 9b). A similar consideration can be proposed also concerning the AAM type (TM category). No significant discrimination is observed in fact among venato, ordinario, arabescato, statuario, and nuvolato AAMs (Figure 9c). This consideration suggests that the method for provenance identification could be independent of the type of marble. One can recall, in fact, that the authors of [57] applied the same DB for the discrimination of different types of marble, such as Bigio Antico.

Concerning the structural levels (LS category, Figure 9d—for quarry sites refer to Table 2), two clusters almost distinct from each other are identified: those labelled #4 (reverse limb of the Vinca Anticline–Orto di Donna Syncline) and those labelled from #6 to #9 = uppermost AAM levels.

Since the difference envisaged in Figure 9d is mainly due to EPR parameters, the levels #4 and #6-#9 have registered the slight difference in the tectono-metamorphic deformation path that can be detected by this technique. This can confirm the relation between the P/T path history and the EPR signature.

Considering the HISTORY category, the AAM quarries exploited during Roman and Renaissance times seem almost well discriminated (Figure 9e—for quarry sites refer to Table 2): the Roman quarries were only in the Carrara District (#6 → #9), whereas in the Renaissance, with Medici, many blocks came from the Seravezza District (#4 → #6); the presence of the LS category #6 in both times explains the partial overlap.

4. Conclusions

From the general data on the Early Jurassic—carbonate O/C isotope data—the only difference that can be found derives from diverse age, but this is not the case for AAM isotope analysis.

Table 3. Comparison between constraining features for isotope analysis attribution and the AAM situation (elaboration by the authors).

Constraining Features	AAM Situation	Verified
Origin of the carbonate: chemical precipitate or organic mud	Organic mud	Yes
Composition of water associated during diagenesis and its later history	Warming with increasing of the δ18O and δ13C values	No
Marble unit is homogeneous and preferably almost pure carbonate and thick	Most contain 97% calcite	Yes
Protolith was deposited and underwent diagenesis in a uniform environment	Carbonate platform with several environments and sedimentary facies	No
Temperature of metamorphism and its thermal gradient—thermal metamorphic gradient of 15 °C/km or less is not a problem, while a gradient towards 30 °C/km resulted into a great range of δ18O values	Temperature about 380–400 °C Thermal gradient up to 30–35 °C/km	No
Fractionation with pore waters and other mineral phases during metamorphism	No, almost pure calcite	No
Isotopic equilibrium was attained and maintained during sedimentation or metamorphism	Exchange between the marble and the embedding lithologies occurred near the contacts	No everywhere

reports a comparison between the main constraining features for detecting marble provenance on the basis of isotopic analysis and the situation for AAMs.

Therefore, the chance to discriminate the quarry of provenance for the AAMs solely on the basis of the O/C isotope appears unrealistic. The authors of [11] made the same considerations: “the positive identification of the quarries of ancient marble artefacts may be obtained only by a multimethod approach”.

In the approach fostered in the present study, the combination of isotopic data, petrographic parameterisation of the fabric and EPR spectroscopic parameterisation, this latter modified to avoid intrinsic self-correlation trends, seem able to identify, at least partially, some subsets among the geological, geographical and historical corpus of the AAMs. In particular, significant, but not complete, discrimination by EPR analysis has been identified in the AAM DB of [10] concerning the structural levels above (#6 → #9) and below (#4) the Vinca Anticline, which it also resembles into the historical exploitation period (Roman age versus Renaissance).

Accordingly, when using this procedure to contribute to the attribution of a sure origin to AAM samples collected as part of an artefact, there is a certain chance of referring to the fully discriminated regions of the PCA 3D space; however, this method cannot be considered fully efficient in the task. Once again, the convergence of organoleptic attribution, technical laboratory analysis and documentation data appears to be the only reliable way to attain a sure attribution of an artefact to a specific AAM district or quarry site.