The abandoned quarry of inestimable value
The seed from which this research grew was a simple question: what do carbonate-rich metasediments heated to more than 1070 °C/GPa in the lower continental crust of orogens record? To answer this question, we collected samples during a broader campaign through the exposed section of Variscan lower crust in the Serre Massif of Calabria, southern Italy. Different sites were spot out, among which an old abandoned quarry near San Vito sullo Ionio. Here, marble appears weathered, and it did not seem particularly promising. Then we removed the surface alteration to collect a fresh sample. The rock underneath immediately revealed a texture and mineralogical complexity that had been completely hidden by weathering. We knew that we had something surprising in our hands, although at that point we had no idea how surprising it would become. Back in the laboratory, the promise of that fresh surface only grew. Under the microscope, olivine grains appeared surrounded by dolomite-rich pockets and delicate reaction coronas, while thin films of calcite occupied grain boundaries, as if liquid had moved between the crystals. These features raised a simple question: did the marble melt?
The problem with a magma that does not crystallise in situ
Carbonatites are unusual igneous rocks composed for more than 50% by carbonate minerals. Most known examples are linked to mantle-derived magmas, but high-grade metacarbonate rocks can also melt. Proving this in nature has always been difficult because carbonatitic liquids have extremely low viscosity: they can rapidly leave their source, migrate and react with other rocks. The magma may accumulate far away, leaving its source looking like an ordinary recrystallised marble.
Therefore, we usually study the crystallised melt and never the residue from which it escaped, simply because the latter had not already been characterised. Outcrops from San Vito sullo Ionio in Calabria offered the opposite perspective: a marble depleted by melt extraction, preserving the negative image of a vanished magma.
Building a case from ambiguous clues
One of the most memorable moments came during a scanning electron microscope session. In a backscattered-electron image, we encountered a microstructure unlike anything else in the sample: extremely small, intensely bright droplets embedded in a complex, finely intergrown matrix of calcite and dolomite. Because phases containing heavier elements appear brighter in these images, the droplets immediately caught our attention. When we analysed them, our initial surprise became genuine shock: they consisted of about 98 wt.% of pure lead.
The timing made the discovery even more striking. That same morning, Fabrizio had shown students an analogous droplet microstructure - although with a completely different composition - while explaining liquid immiscibility, the separation of one liquid into two distinct liquids. Only hours later, a remarkably similar geometry appeared on our own microscope screen. Our reaction was one of pure surprise. The lead-rich droplets did not prove the entire melting history by themselves, but they became one of its most unforgettable clues.
The hardest part of the work was resisting the temptation to treat one striking texture as definitive proof. Reaction coronas and carbonate pockets can form during melting, but similar structures may also develop through solid-state metamorphic reactions. Each clue, considered alone, was open to more than one interpretation. Our task was therefore to test whether a completely subsolidus history - one involving no melt at all - could explain the whole rock.
We combined detailed petrography with mineral chemistry of major, minor and trace-element. The solid-state explanation could reproduce parts of the story, but not all of them together. FTIR analyses showed us that olivine contains water and fluorine, two components expected to have been transported by an early volatile-rich carbonatitic liquid. The simplest explanation became a sequence of fluid-present melting, melt extraction and fractionation, and later reaction between the remaining melt and the solid residue.
Following what was missing
The rare earth elements provided the decisive mass-balance test. Despite their name, these elements are useful here not because of their abundance, but because they act as sensitive tracers of mineral reactions and melt movement. Comparing the rare earth element budgets of the reactant and product minerals, we estimated that at least 35% of the original budget was removed from the rock. The extracted component was strongly biased towards the light rare earth elements, with a light-to-heavy rare earth element ratio of about 9:1.
This chemical loss matched the microstructural evidence. We interpret the more mobile, calcium-rich carbonatitic melt as the fraction that escaped, carrying rare earth elements with it. Melt extraction and fractionation resulted in magnesium-richer residual liquid remained close to the newly formed olivine and eventually crystallised as dolomite-rich pockets. As the system cooled, this residual melt reacted with the surrounding minerals, producing the intricate coronas now visible under the microscope. What initially looked like a collection of unrelated textures became successive frames in a single process.
Why the residue matters
Our study connects two parts of the carbonatite story that are usually found separately: a crustal carbonate source and the melt extracted from it. It shows that granulite-facies marbles are not necessarily passive layers heated and recrystallised during mountain building. Under suitable conditions, they can generate highly mobile carbonatitic melts and redistribute trace elements through the lower crust.
Next challenges
The next challenge is to follow these melts beyond their source. Residual marbles like the one at San Vito sullo Ionio provide a search image for recognising crustal carbonatite generation elsewhere, while isotopic and trace-element fingerprints may link depleted sources to the bodies where the liquids accumulated. Such source-product pairs can reveal how far carbonatitic melts travel and how strongly they modify the crust.
Take-home message
A magma can be recorded not only by the rock it crystallises into, but also by the source it leaves depleted. In the San Vito marble, microstructures, volatiles, modelling and rare earth elements converge on the same conclusion: part of the rock melted, a calcium-rich carbonatitic liquid escaped, and the residue preserved the history of that loss.