Peng Peng

The mechanisms responsible for the formation and compositional evolution of continental crust remain central questions in Earth science. Post‐collisional magmatism is widely preserved in the continental record; however, its contribution to crustal growth may be underestimated because enriched isotopic signatures obscure juvenile mantle input. Here, we present integrated whole‐rock geochemical, Sm–Nd–O isotopic, and zircon U–Pb–Hf–O isotopic data for ∼2.49–2.45 Ga post‐collisional mafic and felsic rocks from the Daqingshan Terrane, western North China Craton. The Daqingshan mafic rocks lack diagnostic arc‐like signatures, show no pronounced depletion in high‐field‐strength elements, and have relatively radiogenic Nd–Hf isotopic compositions, indicating derivation from a depleted but variably hybridized mantle source. Systematic mafic–felsic isotopic correspondences, together with large isotopic variations (whole‐rock ε _Nd ( t ) = −3.1 to +4.5, zircon ε _Hf ( t ) = −11.2 to +6.4, zircon δ ¹⁸ O = 2.9‰ to 9.2‰), indicate that the felsic batholiths were genetically linked to the mafic magmas and formed through variable combinations of mafic magma differentiation, magma hybridization, and crustal anatexis. Hf–O isotope mixing and mass‐balance estimates suggest that juvenile mantle‐derived material accounts for at least ∼36% of the exposed felsic–intermediate assemblage and may approach ∼50% when unexposed mafic underplates are considered. Comparable magmatic timescales and mafic compositional trajectories in other collisional belts suggest that similar processes may have operated more widely. These results identify post‐collisional mafic magmatism as a cryptic mechanism for continental crust preservation, growth, and reworking since the late Archean.

The geochemical heterogeneity of the mantle, recorded by mantle‐derived basalts, offers crucial insights into the evolution of mantle sources. However, recognizing mantle end‐members of basalts is challenging, as traditional trace element proxies yield ambiguous overlapping results in older basalts. Here we employ machine learning techniques to classify basalts, by their major and trace element characteristics, derived from three primary mantle sources: Depleted Mantle, Enriched Mantle, and Hydrated Mantle. By training on Cenozoic basalts, our machine learning achieves a high accuracy of global test sets spanning Phanerozoic to late Proterozoic, proving both its spatial and temporal effectiveness. Additionally, comparing our source predictions with full‐plate paleogeography since 1 billion years ago and testing the method in complex tectono–magmatic settings further assures spatiotemporal robustness and general applicability. We thus provide an effective machine learning classifier for accurately tracing mantle sources of basalts globally farther back in geologic time.

The African large low shearwave-velocity province (LLSVP) advected plumes manifest as large igneous provinces (LIPs) during Pangea breakup. The existence of such basal mantle structures during earlier supercontinents is unknown. Here we analyze samples from China and Brazil to demonstrate with literature data a series of seven LIPs of supercontinent Rodinia from 940–720 million years ago. Multi-proxy reconstruction results in a coherent ring of LIPs comparable with the size of Rodinia that resembles the product of an ancient LLSVP whose edges advected LIP-producing plumes. Oceanic lithosphere recycling following supercontinental assembly is evident in Nd isotopes as evolving LIP-sources. There is a same secular variational pattern of LIP sources with a replenished depleted component advected by plumes as well as a similar manner yet contrasting paleogeographic configurations corresponding to supercontinental breakup between Rodinian and Pangean LLSVPs. This evolving pattern of LLSVPs suggests their partial regeneration (replenishment and migration) between supercontinents.

Lunar mare basalts illuminate the nature of the Moon’s mantle, the lunar compositional asymmetry and the early lunar magma ocean (LMO)^1,2,3. However, the characteristics of the mantle beneath the vast South Pole–Aitken (SPA) basin on the lunar farside remain a mystery. Here we present the petrology and geochemistry of basalt fragments from Chang’e-6 (CE6), the first returned lunar farside samples from the SPA basin^4,5,6,7. These 2.8-billion-year-old CE6 basalts⁸ share similar major element compositions with the most evolved Apollo 12 ilmenite basalts. They exhibit extreme Sr–Nd depletion, with initial ⁸⁷Sr/⁸⁶Sr ratios of 0.699237 to 0.699329 and ε_Nd(t) values (a measure of the neodymium isotopic composition) of 15.80 to 16.13. These characteristics indicate an ultra-depleted mantle, resulting from LMO crystallization and/or later depletion by melt extraction. The former scenario implies that the nearside and farside may possess an isotopically analogous depleted mantle endmember. The latter is probably related to the SPA impact, indicating that post-accretion massive impacts could have potentially triggered large-scale melt extraction of the underlying mantle. Either way, originating during the LMO or later melt extraction, the ultra-depleted mantle beneath the SPA basin offers a deep observational window into early lunar crust–mantle differentiation.

During the mid‐Proterozoic, nearly all of Earth's continents collided to form supercontinent Columbia. However, the exact timing of its formation is still debated, with estimates ranging from 1.8 to 1.6 Ga. This period, known as the Statherian, is also debated for its affinity to either the Paleoproterozoic or Mesoproterozoic eras. To address these spatiotemporal uncertainties, here we present a reliable 1.70–1.68 Ga paleomagnetic pole from mafic dyke swarms in the North China Craton, enabling global paleogeographic comparison during this critical interval. Compared with paleopoles from the likely neighbor of the North Australian Craton, although these two cratons achieved proximity as early as 1.78 Ga, they underwent slight but significant adjustments in their relative positions between 1.73 and 1.65 Ga, consistent with geological evidence. Similar reconfigurations are observed between Laurentia and North Australia, suggesting that Columbia was gradually becoming more stable in configuration during the Statherian Period.

Large low shear-wave velocity provinces (LLSVPs) in the lowermost mantle are the largest geological structures on Earth, but their origin and age remain highly enigmatic. Geological constraints suggest the stability of the LLSVPs since at least 200 million years ago. Here, we conduct numerical modeling of mantle convection with plate-like behavior that yields a Pacific-like girdle of mantle downwelling which successfully forms two antipodal basal mantle structures similar to the LLSVPs. Our parameterized results optimized to reflect LLSVP features exhibit velocities for the basal mantle structures that are ~ 4 times slower than the ambient mantle if they are thermochemical, while the velocity is similar to the ambient mantle if purely thermal. The sluggish motion of the thermochemical basal mantle structures in our models permits the notion that geological data from hundreds of millions of years ago are related to modern LLSVPs as they are essentially stationary over such time scales.

Precambrian paleomagnetic studies are critical for testing paleogeographic reconstructions in deep time but rely on the fidelity of the assumption of the geocentric axial dipole (GAD) hypothesis. With high‐reliability data from mafic dykes and volcanic rocks, the scatter of individual virtual geomagnetic poles (VGPs) can be used to test simple GAD models. In order to conduct such a test, the VGPs must be adequate in number and in spatial coverage of the sampling sites. In this study, we targeted the 2.1 Ga Indin dyke swarm of the Slave craton. Building on previous sampling of the Indin dyke swarm in the western and central parts of southern Slave craton, we report results from 9 additional sites in the central and eastern parts of the craton, sites that significantly expand the width of the dyke swarm across the entire craton. The VGPs obtained from 7 of 9 newly identified Indin dykes are broadly similar to previously reported directions, expanding the total of VGPs for individual Indin dykes to n = 28, which is sufficient for a test of the GAD‐based statistical models using VGP scatter. The high VGP scatter of the Indin swarm can be attributed to the relatively high paleolatitude of 56° ± 6° for the Slave craton at the time of dyke emplacement. The Indin data have VGP scatter that is consistent with field models associated with the GAD hypothesis for the indicated paleolatitude, thus confirming the fidelity of the GAD field at ca. 2.1 Ga.