Fuzzy petrology in the origin of carbonatitic/pseudocarbonatitic Ca-rich ultrabasic magma at Polino (central Italy)

The small upper Pleistocene diatreme of Polino (central Italy) is known in literature as one of the few monticellite alvikites (volcanic Ca-carbonatite) worldwide. This outcrop belongs to the Umbria-Latium Ultra-alkaline District (ULUD), an area characterized by scattered and small-volume strongly SiO2-undersaturated ultrabasic igneous rocks located in the axial sector of the Apennine Mts. in central Italy. Petrographic and mineralogical evidences indicate that Polino olivine and phlogopite are liquidus phases rather than mantle xenocrysts as instead reported in literature. The presence of monticellite as rim of olivine phenocrysts and as groundmass phase indicates its late appearance in magma chambers at shallow depths, as demonstrated by experimental studies too. The absence of plagioclase and clinopyroxene along with the extremely MgO-rich composition of olivine (Fo92–94) and phlogopite (average Mg# ~93) suggest for Polino magmas an origin from a carbonated H2O-bearing mantle source at depths at least of 90–100 km, in the magnesite stability field. In contrast with what reported in literature, the ultimate strongly ultrabasic Ca-rich whole-rock composition (~15–25 wt% SiO2, ~31–40 wt% CaO) and the abundant modal groundmass calcite are not pristine features of Polino magma. We propose that the observed mineral assemblage and whole-rock compositions result mostly from the assimilation of limestones by an ultrabasic melt at a depth of ~5 km. A reaction involving liquidus olivine + limestone producing monticellite + CO2 vapour + calcite is at the base of the origin of the Polino pseudocarbonatitic igneous rocks.

We propose that the observed mineral assemblage and whole-rock compositions result mostly from the assimilation of limestones by an ultrabasic melt at a depth of ~5 km. A reaction involving liquidus olivine + limestone producing monticellite + Co 2 vapour + calcite is at the base of the origin of the polino pseudocarbonatitic igneous rocks.
The identification of Ca-carbonatitic volcanic activity has been considered a key feature to unravel the geodynamic setting of the Plio-Quaternary volcanism of central Italy. In particular, the presence of carbonatitic magmas has been proposed to deny the existence of active or fossil subduction tectonic settings along peninsular Italy, proposing the existence of a deep seated variably asymmetric (eastward-pushing) mantle plume or hot spot 1,[11][12][13] . From a general point of view, "a strong evidence for the link between the carbonatite genesis and the locations of deep-mantle plumes" has been recently proposed 14 .
The mantle source of the calcite-, melilite-and kalsilite-bearing rocks is assumed to be the same of the potassic and ultrapotassic volcanic rocks of central-southern Italy (Latium-Campania administrative regions 11 ). On this

Carbonatites
Carbonatites are igneous rocks composed of >50% modal primary carbonate and with <20 wt% SiO 2 (ref. 5 ). Carbon must be juvenile -magma-derived -and not from wall rocks or external hydrothermal fluids. Calcite-carbonatite (also defined as calcio-carbonatite) magmas are by far the most abundant lithologies of this extremely rare rock group. Carbonatites can be either alkali-rich or alkali-poor, and are commonly associated with silicate rocks characterized by alkaline ultrabasic compositions such as nephelinite/nephelinolite, melilitite/melilitolite, essexite, aillikite, melteigite and kimberlite, but can also be genetically related to evolved and SiO 2 -richer lithologies such as trachyte/syenite, phonolite/nepheline syenite 25 . Albeit rare 26 , carbonatitic inclusions have been reported in mantle xenoliths found in mafic sodic magmas (e.g., refs 27,28 ). On the other hand, mantle xenoliths have been never 29 or rarely 30 recorded in carbonatitic magmas because of their preferential settling in low viscosity and low density magmas.
One of the major problems in identifying true carbonatitic rocks is to distinguish magmatic textures from later replacement and secondary textures (e.g., ref. 42 ). Indeed, subsolidus plastic flow, deuteric alteration and solution-precipitation recrystallization of magmatic calcite can lead to a difficult distinction from hydrothermal and metasedimentary carbonate rocks 29 . Also the presence of minerals typical of carbonatite rocks cannot be considered a proof for their mantle origin, being these phases also found in calc-skarns 38,43 . We report petrographic and mineralogical evidences according which the carbonate-rich fraction of Polino rocks cannot be considered a feature of primary-i.e., in equilibrium with mantle rocks -conditions.
Carbonatites from central Italy are associated to small monogenetic centres (diatreme structures) pyroclastic rocks and minor lavas with ultrapotassic composition located close to the Apennine Chain axis (Fig. 1). Overall, the trace element content and Sr-Nd-Pb isotopic ratios of Italian carbonatites match those of the Roman Province rocks 47 , but their origin remains highly debated. Stoppa et al. (ref. 56 ) interpret the carbonate-rich lithologies associated with the ultrapotassic rocks as true Ca-carbonatitic melts, while Peccerillo (ref. 57 ) relates them to interaction between silicate melt and sedimentary limestone country rocks.
According to several authors, the volcanic rocks of Polino, classified as monticellite calcio-carbonatites 7-10,45,58 , are characterized by abundant mantle xenoliths and xenocrysts (forsterite and phlogopite). Here, we demonstrate that the petrographic features, the mineralogy and the chemical composition of Polino ultrabasic rocks are more compatible with shallow depth assimilation of sedimentary carbonate wall rocks by SiO 2 -poor Mg-rich carbonated melts.

polino
Despite the limited outcrop (<100 m 2 ), the Polino volcanic rocks occur in two facies. The first (pyroclastic facies) is characterized by the presence of mm-to cm-sized droplets of silicate melt agglutinated by sparry calcite cement, while the second show a more massive aspect (Fig. 2). Both facies are characterized by abundant euhedral to subhedral forsteritic olivine (Fo 92-94 ) followed by euhedral phlogopite (Mg# 90-94) and monticellite, coexisting with accessory phases as Ca-Ti/Ca-Si perovskite, schorlomitic garnet and Fe-Ti oxides, all associated with variable amounts of sparry, sedimentary or ameboid calcite (e.g., ref. 59 ; Fig. 2). All the minerals are SiO 2 -poor to SiO 2 -free phases, i.e., to complete occupancy of tetrahedral sites in the schorlomitic garnet, all the Al and half of Fe must be assigned to tetrahedral coordination 59   www.nature.com/scientificreports www.nature.com/scientificreports/ pyroxenes, with the exception of very rare (<0.1% vol) occurrences of sanidine and diopside xenocrysts from older phonolitic pyroclastic cover pierced by the diatreme 8 . The Polino diatreme crosses the Calcare Massiccio Formation, a Lower Jurassic peritidal carbonate succession cropping out in the Apennines originally deposited along the passive margin of the Adria micro-plate (e.g., ref. 60 ). During two campaigns in 2016 and 2017, twelve samples were collected and analysed for petrographic investigation using the polarizing microscope, scanning electron microscopy (SEM) and electron microprobe (EMP), while three of them have been analyzed for major oxides (ICP-AES) and trace elements (ICP-MS) at the Activation Laboratories (Ontario, Canada). Only massive facies samples have been chosen for analyses, excluding pyroclastic facies rocks rich in sparry calcite cementing silicate globules. About 0.2 g sample have been thermally decomposed in a resistance furnace in a pure nitrogen environment at 1000 °C, directly releasing CO 2 . H 2 O is removed in a moisture trap prior to the detection of CO 2 in the infra-red cell. Full details in http://www.actlabs.com. The EMP details are reported in Lustrino et al. (ref. 61 ).
Some of the major oxides of the three Polino volcanic rocks analyzed in this study (full dataset in Table A1) are reported in Figs 3 and 4, together with the Polino literature data, the composition of the ULUD, Mt. Vulture rocks and other intraplate circum-Mediterranean ultrabasic rocks with SiO 2 content <40 wt%. The LOI of the Polino rocks is very high (19.8-25.2 wt%), as expected for calcite-rich compositions.
The rocks analyzed in this study show low SiO 2 contents (~18.8-23.8 wt%), within the composition reported in literature of ~15. 1-24.8 wt% (refs 7,8,45 ). The range in silica content is surprisingly wide considering the outcrop extension (<100 m 2 ) and is here explained as due to the difficulty to completely separate the sparry cement from the silicatic portion in the pyroclastic facies rocks. The CaO content ranges from 31.6 to 39.7 wt% and shows negative correlation with SiO 2 (R 2 = 0.89). With the exception of two literature analyses, CaO is correlated with CO 2 (14.5-26.0 wt%; R 2 = 0.81; Fig. 4). These correlations highlight two components, one relatively SiO 2 -rich but CaO-and CO 2 -poor, and the other SiO 2 -poor (to SiO 2 -free) and Ca-CO 2 -rich. The second term has been considered the most representative Italian carbonatite end-member by Martin   www.nature.com/scientificreports www.nature.com/scientificreports/ <0.92 wt%, a feature in contrast with the coeval and nearby potassic to ultrapotassic composition of the Italian Pleistocene volcanic rocks and the other ULUD products 4,20,21,42 .
Compared with the average Ca-carbonatite compositions 65

Whole Rock and Mineral Chemical Constraints for the origin of polino Rocks
The relationship between olivine and monticellite is the key aspect to unravel the origin of the Polino magma. Polino monticellite has been interpreted as the result of the interaction between a Ca-carbonatitic mantle melt with xenocrystic mantle forsterite. According to several authors (refs 7-9,58 ), the Ca-carbonatitic melt should be generated at mantle depths by immiscibility from a carbonatite-melilitite liquid. Then, it partially reacted with portions of mantle wall rocks during its ascent to the surface, with the chemical interaction between liquid carbonatite and mantle olivine being expressed in the form 9 : This reaction would result in a dilution of the original CaO content of the carbonatitic magma coupled with an increase of its SiO 2 and MgO content. According to Bell and Kjarsgaard (ref. 67 ; p. 86) the presence of mantle xenoliths at San Venanzo and Polino "is a testament to the rapid ascent of the kamafugitic and carbonatitic magmas", rendering "highly problematic" an open system interaction between a silicatic (kamafugitic) magma and local limestone. On the other hand, Barker (ref. 29 ; p. 46) underlines that ultramafic mantle xenoliths have never been reported in carbonatite lavas or intrusions, due to "the low viscosity and low density of carbonatitic liquid or to the separation of an immiscible liquid from mantle-derived silicate magma within a crustal reservoir. In both cases, ultramafic xenoliths would settle out".
Fort Portal carbonatite contains a variable cargo of silicatic rocks as gneissic, amphibolitic, and gabbroic xenoliths. In order to calculate the relatively pristine composition of the carbonatitic magma, Eby et al. (ref. 68 ), following the approach of Barker and Nixon (ref. 69   www.nature.com/scientificreports www.nature.com/scientificreports/ A microstructural and compositional study of the main mineral phases found in Polino rocks allows us to exclude a xenocrystic origin for forsterite and phlogopite. These two minerals, indeed, are characterized by euhedral to subhedral shape, with phlogopite mostly represented by tiny elongated euhedral laths (Figs 2 and 5; see also pictures in Table A2). Interestingly, the age of Polino rocks (~246 ka) has been estimated via 40 Ar/ 39 Ar on phlogopite separates, indicating that this is a phase crystallized from a melt rather than a mantle xenocryst.
The euhedral shape strongly differs from olivine crystals found in mantle xenoliths (but also in Alpine-type ophiolitic massifs or dredged peridotites), always characterized by anhedral habit. The interpretation of Polino olivines as mantle xenocrysts seems, therefore, highly improbable (e.g., ref. 70 ). In addition, the absence of any deformation texture in the olivine crystals in Polino olivines is a further proof at odd with a mantle xenocrystic origin hypothesis, supporting the derivation as a liquidus phase. The lack of peridotite mantle xenoliths or mantle xenocrysts would be sufficient to confute the reaction (1).
Polino olivines are characterized by relatively high Ni (ranging from ~550 to ~5700 ppm; average = ~3300 ppm, st. dev. ±~1180 ppm; Table A2). According to Ammannati et al. (ref. 71 ), the Italian plagio-leucititic rocks have olivine poor in Ni (~1100-1960 ppm; average ~1330 ppm) because generated from a peridotitic mantle metasomatized by CaCO 3 -rich agents. On the other hand, SiO 2 -richer lamproitic rocks are characterized by much higher Ni (~1960-4950 ppm; average ~3140 ppm). The reason of Ni enrichment in olivine from lamproitic melts would be the derivation of these liquids from olivine-poor to olivine-free mantle source. The www.nature.com/scientificreports www.nature.com/scientificreports/ depletion in olivine would result after the reaction of mantle forsterite with silica-rich agents released from recycled terrigenous sediments, forming enstatite-rich metasomes 72,73 . On the other hand, the low-Ni (and high-Ca; ~2400-4000 ppm) content of olivines in the plagio-leucitites would reflect "the effect of the reaction between melts of carbonate-rich sediments with peridotite, stabilizing newly formed olivine and clinopyroxene at the expense of orthopyroxene" (ref. 71 ; p. 72). Polino olivines do not have the low Ni content of the melts generated from a carbonate-metasomatized sources, but share with them the relatively Ca-rich (~430-6220 ppm; average ~2720 ppm) content. This questions the validity of the general rule proposed by Ammannati et al. (ref. 71 ).
Forsterite is never observed in the groundmass, but as phenocryst only, characterized by a variably thick monticellite rim. The chemical composition of Polino olivine is, also in the largest crystals, homogeneous (Table A2), suggesting that the melt from which the euhedral olivine crystallized was chemically homogeneous and did not change over time. An abrupt compositional variation occurs in the outer rims only, when CaO content sharply increases from values usually <0.8 wt% to >31 wt%, typical of monticellite, coupled with MgO decrease from >50 wt% to <20 wt%, only few microns across ( Fig. 8; Table A2). The boundary between olivine and monticellite  www.nature.com/scientificreports www.nature.com/scientificreports/ rim appears always abrupt and corroded, with signs of reaction and irregular contacts. This can be interpreted as a chemical shock in the form of a change of composition of the crystallizing melt.
We interpret this feature as the indicator of the external input of CaO-rich lithologies in the silicatic magma when it started to pond at shallow crustal depths within sedimentary carbonate magma chamber at a pressure of ~200 MPa. During shallow depth ponding, the hydrous (because phlogopite-bearing) silicatic magma transferred latent heat of crystallization to the country rock, allowing partial dissolution of sedimentary calcite, strongly enriching the melt with CaO and CO 2 formed by calcite breakdown. Worth noting, the upwelling magma at Polino had to pierce a Jurassic dolostone and limestone succession ~5 km thick 76 , and it is hard to believe that such a small volume of magma can have escaped significant interaction with wall rocks.
A peculiarity of Polino rocks is the common presence of phlogopite despite the low whole-rock K 2 O content (average 0.63 wt% ±0.21 wt%; Table A1). The high SiO 2 content of sanidine (~65 wt%) prevents its formation in strongly SiO 2 -undersaturated melts such as those of Polino. Alternatively, anhydrous strongly ultrabasic K 2 O-rich magma could stabilize kalsilite and/or leucite as the most stable K-bearing minerals. The absence of these two minerals is related to the overall low amount of K 2 O in the magma, at odds with the high K 2 O content of leucite (~22 wt%) and kalsilite (~30 wt%). The presence of dissolved H 2 O and high MgO contents in the magma allowed the formation of phlogopite (~8.8-10.5 wt% K 2 O) as the most stable K-bearing phase. The presence of this mica is, therefore, not to be considered as an anomalous feature, due to the specific conditions of the Mg-rich parental magma (before the interaction with limestones at shallow depths) and the availability of water in the magma.

Carbonate-silicate Magma Interaction styles
Based on textural and chemical evidences, we propose a two-stage process to explain the origin of olivine and monticellite. The first step includes the growth of compositionally homogeneous euhedral forsterite crystals as liquidus phases in equilibrium with an H 2 O-and CO 2 -bearing ultrabasic melt, followed by crystallization of phlogopite at water-saturated conditions. The second step is associated to an abrupt change of composition of the melt manifested by the appearance of monticellite at the expenses of forsterite around olivine phenocrysts and as groundmass phase not associated to any contemporary olivine crystallization. Once monticellite appeared, olivine stopped to grow. Groundmass monticellite is closely associated with precipitation of microcrystalline calcite after assimilation of limestone wall rocks.
An important contribution over the last ten years has come from experimental studies on the melt composition after assimilation of carbonates. All the previous studies have used "basaltic" compositions (i.e., plagioclase-bearing melts) as starting materials to reproduce assimilation paths as function of temperature, pressure and fO 2 . These experiments agree in the occurrence of olivine with relatively high Mg content (Mg# ~95) followed by the crystallization of clinopyroxene as result of the magma interaction with limestones according to the reactions: 3(limestone) 2(melt) (melt) 2 6(cpx) 2(vapour)    75 ) and a Vesuvius shoshonite. Following these experimental evidences, some Alban Hills volcanic rocks have been interpreted as related to silicatic magma experiencing extensive carbonate assimilation (both limestone and dolostone) on the basis of the common presence of skarns and magmatic calcite 51,54,77 .
Other reactions include the increase of larnite component in olivine, but not the formation of pure monticellite 81 : Also in this case, the major impact of carbonate (calcite/limestone) digestion on the mineralogy of hybrid magma is the increase of modal abundance of clinopyroxene and its increase in kushiroite (CaAl 2 SiO 6 ), esseneite (CaFe 3+ AlSiO 6 ) and grossmanite (CaTiAlSiO 6 ) components in clinopyroxene. In the case of Polino, the absence of clinopyroxene and plagioclase, as well as the presence of a monticellite rim around liquidus Mg-Fe olivine and as groundmass phase indicate strongly SiO 2 undersaturated compositions of the pristine (pre-carbonate contamination) mantle melt, suggesting its ultrabasic nature. The reaction we envisage can be simplified as follows: 3(limestone) 2 4(olivine) 4(monticellite) ( melt) 2(vapour) Equation (9) was proposed by Walter (ref. 82 ) and can explain the reaction observed in the rim of olivine phenocrysts in contact with monticellite. Another possible reaction could be represented by Bowen (ref. 83 ): 3(limestone) 2 4(olivine) 2 6(melt) 4(monticellite) 2(vapour) In their experiment (run 486) carried out in the CaO-MgO-SiO 2 -CO 2 -H 2 O system at 1038 °C and 2 kb, Otto and Wyllie (ref. 84 ; p. 357) report "clusters of subhedral crystals of monticellite with enclosed euhedral crystals of forsterite" associated with subhedral mostly angular calcite crystals and ~20 wt% SiO 2 in the experimental run. This paragenesis and the SiO 2 content closely resemble that of Polino rocks, and is, therefore, compatible with a process of monticellite crystallization at shallow depths rather than invoking equilibrium with forsterite at P > 1 GPa, as instead proposed by literature models. In the Otto and Wyllie (ref. 84

Carbonatite Melt-silicate solid vs. silicate Melt-Carbonatite solid Constraints
The data presented above and those from previous experimental studies suggest a different origin for the CaO enrichment in Polino rocks, i.e. a shallow depth provenance rather than an upper mantle source. The strongly radiogenic 87 47,85,86 ), questioning the interpretation of Polino calcite as a mantle carbonatitic component. The 87 Sr/ 86 Sr of Polino calcite and whole-rock carbonatite is also higher than that of local Calcare Massiccio limestone (<0.7075; ref. 85 ), but this cannot be interpreted against a sedimentary origin of the Polino carbonate fraction. Indeed, in a hypothetical limestone-silicatic magma interaction, the Sr-rich ULUD magma (Sr up to 4000 ppm) strongly controls the variation of 87 Sr/ 86 Sr ratios in the contaminated melts. In this case, minimum amounts of magma can easily modify the original 87 Sr/ 86 Sr of the carbonate-rich www.nature.com/scientificreports www.nature.com/scientificreports/ fraction of the contaminated magma, being the sedimentary fraction characterized by more than one order of magnitude less Sr.
Dolostones are present in minor amount in the carbonate succession pierced by the magma and could be potentially responsible for the high Fo content of Polino olivines. According to Iacono Marziano et al. (ref. 75 ), the magnesite end-member of dolomite reacts with SiO 2(melt) via: Dolomite digestion often results in crystallization of MgO-rich olivine together with clinopyroxene, this latter being the only crystallizing phase in calcite-doped experiments 75 . The very low NiO of MgO-rich and CaO-rich composition of Alban Hills volcano olivines is considered not a mantle origin feature, but rather the effect of digestion of dolostone 43 , a conclusion confirmed by experimental data 75 . North Baikal olivines formed after digestion of dolostones by mafic magma are characterized by similar low NiO (<0.25 wt%; average 0.12 wt%) 87 . Polino olivines are characterized by high NiO (~0.06-0.95 wt%; average ~0.36 wt%, ~3300 ppm Ni), without any correlation between NiO and MgO (R 2 = 0.06; Table A2). These and the previous considerations lead us to hypothesize for the Fo-rich composition of Polino olivines a mantle origin, rather than a process of interaction with dolostones. This conclusion is corroborated by the very homogeneous composition of the largest olivines (Table A2), indicating the absence of variation from the core to the near rim sectors. Only the outermost olivine rims show MgO increase, associated to the CaO increase, as consequence of interaction with limestone. This process leads Ca entering the M2 olivine site, leaving Mg free entering the M1 site, not preferred by Fe (ref. 81 ).
As concerns the stable isotopes, the high δ 18 O (~+24‰; ref. 9 ) of the calcite component cannot be used to infer a mantle origin for the carbonate component in Polino rocks. The low δ 13 C of Polino calcite (from −8.0 to −13.1‰; ref. 9 ) is within the δ 13 C values of Alban Hills skarns (from +3 to −12‰; ref. 53 ), Alban Hills lavas (from +5 to −19‰; ref. 53 ) and the Morron de Villamayor sedimentary calcite entrapped in olivine (from −11 to −12‰; ref. 88 ). Natural Alban Hills clinopyroxenes show good correlation between δ 18 O and IV Al with increasing CaO content 51 . Experimental modelling of AFC process adding carbonate component to a silicatic magma shows the same compositional evolution of clinopyroxenes, with the highest kushiroitic component recorded in the experimental runs with higher amounts of added carbonates (e.g., refs 78,81 ). If these stable isotope data cannot be used to definitively assume a sedimentary origin for the carbonate fraction of Polino rocks, they are at least compatible with such an origin.
As monticellite is known to be unstable at pressures >1 GPa (ref. 89 ), the hypothesis of monticellite formation after the reaction of mantle minerals with Ca-carbonatitic melt below the Moho appears improbable, and implies its secondary origin. In other words, our view overturns the classically accepted interpretation 8,9 of monticellite as the reaction product between a carbonatitic magma infiltrating mantle matrix, then scraped off by the same carbonatitic liquid. We believe, instead, that the interaction occurred between an ultrabasic melt and the sedimentary carbonate wall rocks en route to the surface.
The presence of phlogopite, forsterite, monticellite, perovskite and periclase (the first four phases identified in Polino rocks) is not a proof to discriminate a carbonate-rich rock as carbonatite, being these minerals present in impure marbles too 38,50,90 . Also, the presence of schorlomitic garnet in Polino rocks 59 cannot be considered a proof for a carbonatitic origin of the melt, being this mineral a typical phase of impure limestones, marbles and skarns (e.g., refs [90][91][92] ). According to several authors (refs 29,38,43,51 ), no decisive chemical and mineralogical criterions by which to establish a magmatic heritage for a carbonate-rich rock do exist, considering that also magmatic carbonatite can undergo drastic compositional changes after emplacement.
Recognition of primary carbonate components must be based on the association with igneous rocks, the presence of fenites, euhedral crystals of carbonates evidencing flow structures, the mineral paragenesis, the incompatible element composition and the radiogenic and stable isotopic ratios of the carbonate-rich rocks 27 . None of these evidences can be found at Polino: (1) no silicate pair rocks are associated to the "carbonatitic" diatreme; (2) no fenitization is recorded on country rocks; (3) no euhedral tabular or rhombic crystals of calcite have been found; (4) no flow structures of carbonate materials have been identified; (5) monticellite stability field is constrained to shallow P (<1 GPa), rendering impossible the formation of this mineral at mantle depths; (6) monticellite appearance is contemporaneous with cessation of olivine growth and crystallization; (7) the incompatible and compatible element budget of Polino "carbonatite" is different from average worldwide Ca-carbonatites 62,63,93 and Fort Portal monticellite carbonatite 68 ; (8) Radiogenic isotopic ratios indicate interaction with crustal lithologies, while stable isotopic ratios are compatible with C-O exchanges at relatively low-T conditions.
On the basis of petrographic evidences (euhedral shape of olivine and phlogopite crystals, presence of monticellite as groundmass phase and as rim around olivine), mineral chemical compositions (homogeneous Fo-rich and modestly CaO-enriched olivine, high Mg# phlogopite), experimental constraints (stability field of monticellite limited to <1 GPa), isotopic data (strongly radiogenic 87 Sr/ 86 Sr, high δ 18 O, low δ 13 C) and geological-volcanological considerations (small volume of the magma, limited diameter of the diatreme, over-thickened sedimentary carbonate country-rocks pierced by the upwelling magma), we propose that Polino rocks cannot be considered as true carbonatites (monticellite alvikite), as instead reported in literature. The origin of these rocks is, rather, more compatible with interaction between an ultrabasic melt likely generated by a magnesite-bearing residual peridotite and local sedimentary carbonate rocks. (2019) 9:9212 | https://doi.org/10.1038/s41598-019-45471-x www.nature.com/scientificreports www.nature.com/scientificreports/ The classification of Polino volcanic rocks remains debated. It cannot be defined as alvikite (too low primary carbonate content), basalt (no plagioclase), kamafugite (no kalsilite), melilitite (no melilite), foidite (no foids), dunite (not a plutonic/metamorphic rock) or picrite (SiO 2 <30 wt%). It represents an alkali-poor strongly ultrabasic melt whose original composition was modified by the digestion of sedimentary carbonates, and can consequentially be classified as pseudocarbonatite 30 . The final message is that the carbonate component of Polino rocks is not mantle-derived, but rather is the result of partial digestion of shallow limestones.
On the other hand, the original Polino magma (i.e., the composition before interaction with sedimentary carbonates) is considered to derive from a carbonated (magnesite-bearing) peridotitic source, to explain the ultrabasic compositions (absence of feldspars and clinopyroxene) and the Mg-rich olivine and phlogopite phenocrysts. In other words, the petrogenetic model proposed is a fuzzy one, based not on a carbonatite-yes or carbonatite-no choice. At the same time, the Polino rocks are not true carbonatites, but they derive from a carbonated mantle source.