In a study published in Earth Science Informatics, Huanbao Zhang of the Hunan Institute of Technology and colleagues, working with collaborators at the University of South China, compiled an extraordinary dataset: 13,397 granite samples drawn from the global GEOROC database, each characterized by its major and trace element composition. Rather than relying on a few curated elements, the team let the full chemical spectrum speak. Every sample was assigned to one of five tectonic-setting classes: intraplate volcanics, convergent margins, Archean cratons, continental flood basalt provinces, and rift volcanics. This taxonomy spans the dramatic range of environments in which granitic magmas are generated, from the crushing pressures above sinking ocean slabs to the hot, stable roots of the oldest continental nuclei on Earth.

The scale of the compilation matters. Traditional discrimination diagrams, such as the influential trace element schemes developed by Julian Pearce and colleagues in the 1980s, were built on comparatively small reference suites and depend on subjective empirical criteria for drawing field boundaries. A single sample plotted near a dividing line can be interpreted one way or another depending on the eye of the beholder. By contrast, a dataset of more than thirteen thousand samples allows an algorithm to learn the statistical texture of each tectonic environment, including the overlapping zones where settings genuinely blur into one another. It is the difference between memorizing a few textbook examples and having read the entire library.

Before any learning could happen, the team invested heavily in data preprocessing and feature engineering, the craft of transforming raw measurements into inputs that algorithms can use effectively. Geochemical data are notoriously messy: detection limits vary between laboratories, elements are reported over many orders of magnitude, and some measurements correlate so strongly with others that they carry redundant information. Feature engineering addresses these problems by creating derived variables, such as element ratios, that often encode petrological meaning more directly than raw concentrations. The care taken at this stage is frequently what separates a machine learning result that works from one that merely memorizes noise.

The heart of the method is a stacking ensemble, an architecture in which multiple different learners are trained on the same problem and a higher-level model learns how best to combine their votes. The team integrated three complementary algorithms: XGBoost and LightGBM, two highly efficient implementations of gradient-boosted decision trees, and a multilayer perceptron, a classic neural network capable of capturing smooth nonlinear relationships. Gradient boosting excels at tabular data like geochemical compositions, building sequences of decision trees that each correct the errors of the last, while the neural network contributes a different inductive bias. Stacked generalization, a technique pioneered by David Wolpert in 1992, lets a meta-learner discover which base model to trust under which circumstances, often squeezing out performance that no single model could achieve alone.

The numbers are striking. After optimization, the ensemble achieved an accuracy of 0.91, a balanced accuracy of 0.90, and a macro-F1 score of 0.89 across the five tectonic classes. Balanced accuracy and macro-F1 are particularly important here because they prevent a model from looking good simply by favoring the most common classes; they require the algorithm to perform well even on rare tectonic settings. In practical terms, roughly nine out of ten granites can be assigned to their correct tectonic environment from chemistry alone, a level of reliability that approaches, and in some respects exceeds, what expert-driven diagram-based classification typically delivers, especially for ambiguous samples.

But raw accuracy is only half the story, and arguably the less interesting half. The real advance lies in the team’s use of SHAP analysis, a technique from explainable artificial intelligence developed by Scott Lundberg and Su-In Lee that attributes each prediction to the contributions of individual input features. SHAP values, grounded in cooperative game theory, reveal not just which elements matter most overall but how each element pushes a particular sample toward or away from a particular classification. For a discipline that has long depended on interpretable, if crude, graphical tools, this transparency is what makes machine learning a genuine scientific partner rather than a black-box oracle.

The SHAP results are geochemically telling. The most predictive features were associated with holmium, yttrium, the heavy rare earth elements, thorium, neodymium, and dysprosium. That lineup is not arbitrary. Yttrium and the heavy rare earths are sensitive to the presence or absence of garnet and other minerals in the source region or in the residue left behind during melting, and their behavior therefore tracks the depth and temperature conditions of magma generation. Thorium enrichments speak to crustal contributions and the degree of differentiation, while neodymium participates in the well-known geochemical systems used to trace mantle versus crustal sources. In other words, the algorithm independently rediscovered, through pure pattern recognition, element groups that petrologists consider diagnostic, and it did so across all five tectonic settings with distinct contribution patterns for each.

That last finding carries a conceptual punch. The analysis emphasized the multivariate nature of granite tectonic discrimination: no single element or pair of elements cleanly separates the settings, and the discriminative information is distributed across a web of interacting concentrations. This helps explain why conventional diagrams, which by necessity flatten high-dimensional chemistry into two or three axes, so often produce ambiguous results. The machine learning framework does not replace those diagrams so much as reveal their limits, offering an interpretable complement that can quantify how much each chemical variable contributes to a decision and expose where the boundaries between tectonic environments genuinely overlap in nature.

The implications reach well beyond academic taxonomy. Reconstructing ancient tectonic settings is foundational to understanding crustal evolution and geodynamic processes, and it feeds directly into mineral exploration, since many ore deposits are intimately tied to specific magmatic environments. A tool that can reliably fingerprint a granite’s birthplace from a routine whole-rock analysis could sharpen exploration targeting, refine reconstructions of vanished supercontinents, and help test competing models of how the earliest continental crust, including the Archean cratons included in this study, was assembled. The work also joins a growing wave of machine learning applications across the geosciences, from basalt discrimination to tracking crustal thickness variations, that suggest the era of big data in Earth science is moving from promise to practice. For a rock that has been silent for billions of years, granite is suddenly talking, and algorithms are learning to listen.

Subject of Research: Machine learning classification of granite tectonic settings using global geochemical data and SHAP interpretation

Article Title: Machine learning-based discrimination of granite tectonic settings: Integrating global geochemical datasets, feature engineering and SHAP interpretation

Article References: Zhang, H., He, H., Shi, Y., He, Y., Zeng, T., Tao, Y., Wang, Z., Xie, Y., Huang, X., & Li, X. (2026). Machine learning-based discrimination of granite tectonic settings: Integrating global geochemical datasets, feature engineering and SHAP interpretation. Earth Science Informatics, 19(11), Article 199. https://doi.org/10.1007/s12145-026-02256-x

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02256-x

Keywords: granite, tectonic settings, machine learning, geochemistry, SHAP, XGBoost, LightGBM, stacking ensemble, rare earth elements, crustal evolution, GEOROC, petrology

Xenoliths are among the most direct windows geologists have into the otherwise inaccessible sub-continental lithospheric mantle. When basaltic magma ascends rapidly from depth, it can rip fragments of the wall rock it passes through and transport them fast enough that the fragments survive the journey. At Gangassao, the xenoliths are impressive specimens, measuring 4 to 8 centimeters across. They are mostly ovoid to sub-ovoid in shape, a form the authors attribute to long transport within a turbulent magma, and they display sharp contacts with their basaltic host lava, with no reaction rims—another sign that the ascent was too quick for the fragments to be chemically digested by the melt around them.

Under the microscope, the Gangassao xenoliths reveal a classic four-phase mantle assemblage of olivine, orthopyroxene, clinopyroxene and spinel, with no hydrous minerals such as phlogopite, amphibole or apatite. Modal proportions classify them firmly as spinel lherzolites. Two textural types are present: protogranular, characterized by large olivine crystals 5 to 8 millimeters across, sometimes rimmed by aggregates of small, strain-free polygonal olivine; and porphyroclastic, marked by elongated, kinked olivine crystals, some with distinctive holly-leaf shapes. According to the classification scheme of Mercier and Nicolas, these textures record recrystallization during partial melting and syn- to post-tectonic deformation at elevated temperatures—a history consistent with the tectonically active setting of the Adamawa plateau.

The chemical fingerprints of the minerals are equally telling. Olivine crystals are highly magnesian, with forsterite contents between 87.6 and 89.1, and mostly low calcium oxide contents of 0.07 to 0.18 weight percent, a combination characteristic of mantle-derived olivine rather than crystals grown in a magma. Interestingly, some olivine grains carry elevated CaO values of 0.35 to 0.47 weight percent, approaching magmatic compositions. The researchers interpret this as evidence of reheating after the xenoliths were sampled by the ascending basalt, with calcium diffusing within olivine or from adjacent clinopyroxene. Orthopyroxene is enstatite and clinopyroxene is chromium-rich augite, while the spinel phase is an aluminum spinel with very low TiO2, below 0.02 weight percent, and low chromium numbers between 11.0 and 11.8. In the olivine–spinel mantle array diagram, the compositions plot squarely within the mantle field.

To pin down the conditions under which these rocks last equilibrated, the team applied two-pyroxene geothermometry and geobarometry, using the formulations of Putirka alongside independent thermometers from Taylor, Wells, and Brey and Köhler. The results are remarkably congruent. Temperatures cluster mostly between 1100 and 1150 degrees Celsius, with pressures of 1.13 to 2.00 gigapascals, implying depths of 37 to 66 kilometers and a geothermal gradient close to 30 degrees Celsius per kilometer. Notably, the range of values calculated for one sample may trace the physical journey of the xenolith itself: from an initial position at about 2.00 gigapascals and 1247 degrees Celsius, around 66 kilometers deep, to re-equilibration with the host magma at roughly 1.20 gigapascals and 1100 degrees Celsius, about 40 kilometers down, just below the Moho. The final dash through the crust, the authors conclude, was too rapid for any further thermal or pressure adjustment.

Plotted on a temperature–pressure diagram, the Gangassao lherzolites fall within the field of incipient melting, close to dehydration solidus boundaries, and span the transition from the spinel stability field toward the plagioclase field. This distribution points to decompression during mantle upwelling, or exhumation from deeper levels followed by incomplete re-equilibration during ascent. The team proposes that the mantle beneath Gangassao has undergone repeated melt extraction, a process supported by rounded outlines of some olivine, clinopyroxene and orthopyroxene crystals and by suspected droplets of silicate glass around them. They identify filter pressing—the squeezing of melt out of a crystal mush—as the most plausible mechanism for separating the silicate liquid generated by decompression of the sub-continental lithospheric mantle, with the Tertiary uplift of the entire Adamawa plateau providing the driving force.

The authors are careful to note the limits of their evidence. The filter-pressing interpretation would ideally be backed by geochemical data showing enrichment of incompatible elements in extracted melts relative to their residues, which the present study does not provide. They also raise alternative mechanisms that could accomplish melt extraction without filter pressing, including fracture-assisted melt escape driven by melt overpressure and buoyancy-driven segregation arising from density contrasts between melt and residue. Degrees of melting estimated from spinel chromium numbers range from about 2.0 to 2.65 percent, though the team cautions that this formula was calibrated for mid-ocean-ridge peridotites and its applicability to continental rift-related lherzolites is debated. An independent estimate based on aluminum and magnesium contents of coexisting pyroxenes suggests the xenoliths experienced less than 15 percent melting, consistent with the low-degree melting inferred for the source of the Adamawa basalts themselves.

The second process acting on the Gangassao mantle is silicate metasomatism—the infiltration of melt into the solid rock, subtly altering its chemistry. Noticeable variations in the major-element compositions of coexisting minerals within and between the xenoliths point to this heterogeneous modification. Together, melt extraction and melt infiltration emerge as the main processes sculpting the lithospheric mantle beneath the plateau. The broader regional context reinforces the picture: across the Adamawa plateau, ultramafic xenoliths include nine lherzolite occurrences, three harzburgites, two wehrlites and one olivine websterite, each recording different temperatures, pressures and depths. At Youkou, just 40 kilometers southwest of Gangassao, a refertilization process was invoked; at Dibi, 50 kilometers to the south, mantle upwelling is recorded by the transformation of spinel into plagioclase and a Moho at only about 20 kilometers depth.

The geological setting ties all of this together. The Adamawa plateau is a dome-shaped tectono-volcanic horst within the Pan-African mobile belt, cut by a network of ancient strike-slip faults trending in multiple directions that likely penetrate down to the mantle. Geophysical data show the crust thinning from about 35 kilometers in the south to as little as 14 to 20 kilometers north of Ngaoundéré, while the asthenosphere–lithosphere boundary has risen from around 120 kilometers to roughly 80 kilometers beneath the plateau. The authors argue that the history of the Adamawa lithospheric mantle reflects the Tertiary rejuvenation of this Pan-African fault system, which channeled magmas upward and allowed them to sample the mantle at a range of depths. The Gangassao lherzolites, equilibrated at 37 to 66 kilometers, fit neatly alongside lherzolites from Ngao Sey and complement the shallower harzburgites and deeper websterites found elsewhere on the plateau.

What makes the discovery compelling is how much information a handful of fist-sized rock fragments can carry. Each xenolith is a small time capsule, preserving textures and mineral chemistry that record partial melting, deformation, metasomatism and rapid volcanic transport. Together they demonstrate that the mantle beneath the Adamawa plateau is not a single homogeneous reservoir but a patchwork, stitched together by repeated melt extraction and infiltration over hundreds of millions of years. As volcanic activity continues to expose fresh fragments across the Cameroon Volcanic Line, each new occurrence refines the map of this hidden layer, bringing scientists closer to understanding how continental mantle evolves when ancient faults, rising plumes and uplifting plateaus collide in one of Africa’s most geologically dynamic regions.

Subject of Research: Petrology and geothermobarometry of spinel lherzolite xenoliths revealing lithospheric mantle heterogeneity beneath the Adamawa plateau, Cameroon

Article Title: Lithospheric mantle heterogeneity under Adamawa plateau (Cameroon, Central Africa) evidenced from Gangassao lherzolite xenoliths

Article References: Feukam Feuto, J. B., Bardintzeff, J. M., Okomo Atouba, L. C., Fagny Mefire, A., Adama, H., Njankouo Ndassa, Z. N., Nikara Voundou, D., Nkouandou, O. F., & Bonin, B. (2026). Lithospheric mantle heterogeneity under Adamawa plateau (Cameroon, Central Africa) evidenced from Gangassao lherzolite xenoliths. Discover Geoscience, 4(1), Article 364. https://doi.org/10.1007/s44288-026-00728-7

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00728-7

Keywords: lithospheric mantle, lherzolite, xenoliths, Adamawa plateau, Cameroon, spinel lherzolite, geothermobarometry, melt extraction, silicate metasomatism, Pan-African faults, Ngaoundéré, petrology