The voluminous succession of tholeiitic basalts, calc-alkaline andesites and minor high-K basalts that form the Late Miocene Altos de Jalisco mafic province of the western Trans-Mexican Volcanic Belt is interpreted as the magmatic manifestation of a lithospheric dripping event, which removed mantle lithosphere and lower crustal lithologies beneath the study area. During this process, the release of fluids from the foundering materials, coupled with mantle upwelling around the sinking mass, promoted abundant melting of a spinel peridotite and the production of large volumes of tholeiitic magma with low La/Yb and Gd/Yb ratios. Negative correlations of these ratios with MgO contents, Nd isotopes and Rb/Nd ratios indicate that the parental basalts subsequently experienced high-pressure fractional crystallization and contamination with a newly exposed felsic continental crust, thus producing the more evolved calc-alkaline compositions. Stronger garnet signatures and marked enrichments in highly incompatible elements in the high-K suite support derivation from a garnet- and phlogopite-bearing pyroxenitic source, presumably formed by reaction of mantle peridotites with hydrous silicic melts derived from the foundering lithologies. This new petrogenetic model for the Altos de Jalisco volcanic district suggests that the loss of mafic lower crust during lithospheric dripping might be balanced by production of abundant flood basalts within continents, and thus indicates that additional mechanisms may be required for the stabilization of andesitic crust on Earth.

The Madre de Dios Metamorphic Complex (MDMC) in southern Chile is a fossil frontal accretionary prism, which is mainly composed of metapsammopelitic rocks, intercalations of oceanic rocks (greenstone and metachert) and platform carbonate. We concentrated on the metabasite to decipher the metamorphic evolution. This rock type contains assemblages of the pumpellyite–actinolite facies: pumpellyite ± actinolite–chlorite ± grandite ± phengite ± epidote–albite–quartz–titanite ± K-feldspar ± calcite. The metamorphic phases mainly grew by prograde hydration reactions during various episodes of restricted fluid influx. Fundamental phase relations of the pumpellyite–actinolite facies and adjacent facies were reproduced by pseudosections calculated for the system K₂O–Na₂O–CaO–FeO–O₂–MgO–Al₂O₃–TiO₂–SiO₂–H₂O–CO₂ at 200–400°C and 1–9 kbar. The calculated stability fields of the metamorphic assemblages as realized in the MDMC metabasite indicate highest metamorphic conditions restricted to 290–310°C, 4–6 kbar for the MDMC, presumably as a result of the main fluid influx at these conditions. Nevertheless, earlier local equilibria are still preserved as a result of strongly kinetically controlled mineral reactions and a lack of recrystallization and compositional homogenization at thin-section scale. Hence, thermodynamic calculations of local multivariant mineral equilibria using the entire compositional variation of minerals in the MDMC show that the prograde PT path evolved from 4 ± 1 kbar, 200–220°C to 5 ± 1 kbar, 290–330°C. The prograde PT path reflects nearly horizontal particle paths after reaching the maximum depth typical for frontal accretionary prisms. Long residence at maximum depth resulted in thermal re-equilibration. ⁴⁰Ar/³⁹Ar spot ages were measured by in situ UV laser ablation of local phengite concentrations in a deformed metapelite at 233·2 ± 1·8 Ma and in an undeformed metabasite at 200·8 ± 2·4 Ma. Whereas the first age represents an age of accretion, the latter age can be attributed to mineral growth either during a younger stage of accretion or during a retrograde stage. ⁴⁰Ar/³⁹Ar isotopic analyses of two further metabasite samples reflect a prominent resetting of ages at 152·0 ± 2·2 Ma and white mica growth during external fluid access triggered by either a local intrusion or a late Jurassic extensional episode.

At Mt. Moncuni (Lanzo Massif, Western Alps) plagioclase peridotites and early mid-ocean ridge basalt (MORB) gabbroic dykes are deformed by shear zones containing cataclastic bands and both fault-vein and injection-vein pseudotachylytes, which are crosscut by late MORB porphyritic dykes. Fault-vein pseudotachylytes have thicknesses of the order of 1 mm; injection-vein pseudotachylytes have a typical thickness of 1–10 cm and contain spinifex textures. Structural, petrological and geochemical data show that the pseudotachylytes formed by near-complete melting of the host peridotite, at ambient temperature–pressure conditions (T = 600 ± 100°C, P < 0·5 GPa) close to the brittle–ductile transition of ultramafic rocks, during exhumation of the lithospheric mantle in the early stages of formation of the Ligurian Tethys oceanic basin. Estimates of the average volume fraction of unmelted clasts and of the ambient and liquidus temperature, together with thermophysical parameters, allow the determination of the melting energy per unit volume. Coseismic displacement is not observable at Mt. Moncuni, and consequently the dynamic shear resistance cannot be inferred. We show that commonly proposed relations between fault-vein thickness and displacement are of limited value, given the difficulty in identifying ‘single-event’ pseudotachylytes and the mobility of the melt. However, we also show that dynamic shear resistance can be predicted to decrease sharply if the melt coats the whole fault plane, partly as a consequence of the nonlinear viscosity of silicate melts at high strain rates. The Mt. Moncuni pseudotachylytes are the result of upper mantle seismicity at shallow depth (z < 20 km) over a time period of at most 5 Myr. Estimation of the total seismic energy release and moment (caused by an unspecified number of small to moderate earthquakes) requires an assessment of the total pseudotachylyte volume. This is highly uncertain, with a probable qualitative error margin of ±1 order of magnitude. The inferred values of cumulative seismic energy release and moment are of the order of 10^{15 ± 1} J and 10^{19 ± 1} N m, respectively, resulting in a seismic energy release rate of approximately 10^{8 ± 1} J/a. This value is compatible with present-day seismic rates at extensional plate margins.

In the Twin Sisters ultramafic massif, NW Washington, an ~0·5 cm thick, isolated ultramafic ultramylonitic shear zone displaces orthopyroxenite and clinopyroxenite dikes, by a minimum of 21 cm. The shear zone exists only adjacent to the orthopyroxenite and clinopyroxenite dikes, with deformation distributed along strike into the wall-rock less than 10 cm from the dikes in the outcrop face. Microstructurally, the shear zone contains domains of different grain sizes and phase proportions. A marginal domain of almost pure olivine displays a mean grain size of ~30–100 µm and an olivine lattice preferred orientation (LPO) indicating that glide occurred on (010)[100] and (010)[001]. A central domain of mixed olivine, pyroxene and amphibole displays a finer grain size, ranging down to ~5 µm. Aligned grain and phase boundaries and weak olivine and pyroxene LPOs indicate that this zone deformed by a grain-size-sensitive deformation mechanism (e.g. grain boundary sliding accompanied by diffusion creep). Geothermometry indicates that shearing occurred at temperatures in the range 650–750°C. We interpret the formation of this ultramylonitic shear zone as a shear instability caused by the presence of compositional heterogeneity. Localization was promoted by a deformation mechanism switch from dislocation creep to grain-size-sensitive creep as a result of syn-deformational grain-size reduction. Mineral compositions indicate that this grain-size reduction was associated with reaction. The fine grain size was stabilized by the presence of multiple phases, particularly pyroxene, within the central shear zone domain. The shear zone did not propagate beyond the vicinity of the pyroxenite dikes because the fine grain sizes necessary for the deformation mechanism switch to occur could not be maintained in the monophase olivine forming the surrounding dunite.

A combined petrological, geochemical, and geochronological (Rb–Sr and Sm–Nd whole-rock, U–Pb and Lu–Hf zircon, and Ar–Ar hornblende) study on a section of pre-Witwatersrand basement drilled at the northwestern margin of the Witwatersrand Basin has revealed new insights into the nature and tectonic setting of the likely source area for some of the Mesoarchaean auriferous Witwatersrand sediments. The protoliths of intersected altered granite and hornblende metagabbro are of indistinguishable age (3062 ± 5 Ma) and have very similar geochemical signatures. Trace element characteristics typical of calc-alkaline magmatism and evidence of variable contamination with older crust (subchondritic _Nd and _Hf in zircon) point to an active continental margin setting. The Ar–Ar hornblende ages are within error of the magmatic crystallization age or slightly older. Alteration of presumably primary magmatic hornblende to magnesio-hornblende immediately after gabbro emplacement during late magmatic autometasomatism is suggested. The presence of hydrous melts (>4 wt % H₂O), comparable with fertile Au-bearing magmatic–hydrothermal mineralizing systems in Phanerozoic volcanic arcs, is inferred. Thus, a kind of hinterland is proposed for the Witwatersrand that compares favourably with the tectonic domains that are known to host the majority of post-Archaean gold deposits. Later retrograde hydrothermal alteration at c. 2720 and 2630 Ma led to variable Pb loss in zircon and the resetting of the whole-rock Rb–Sr isotope system whereas the Ar–Ar and Lu–Hf isotope systems in the hornblende and zircon grains, respectively, were not significantly affected. Comparison with published data suggests that these alteration events are the same as those that affected the Witwatersrand Basin fill associated with major early Ventersdorp flood basalt volcanism and possibly a pre-Transvaal thrusting event in response to contractional deformation in the Limpopo Belt.

Northern Norrbotten, Sweden is a key part of Baltic Shield and provides a record of magmatic, tectonic and related, superimposed, Fe oxide–apatite and iron oxide–copper–gold (IOCG) mineralization, during the Svecokarelian orogeny. Titanite and allanite from a range of mineral deposits in the area have been analysed for U–Pb isotope systematics and trace element chemistry using laser ablation quadrupole inductively coupled plasma-mass spectrometry (LA-ICP-MS). Analyses of a single sample from the regional scapolite–albite alteration give an age of 1903 ± 8 Ma (2) and may be contemporaneous with the early stages of Fe mineralization (1890–1870 Ma). Analyses of titanite and allanite from undeformed IOCG deposits indicate initial alteration at 1862 ± 16 Ma. In many deposits subsequent metamorphic effects reset titanite isotope systematics from 1790 to 1800 Ma, resulting in a spread of U–Pb isotope analyses along concordia. In some instances core regions may record evidence of early thermal events at around 2050 Ma. Titanite and allanite from deformed IOCG deposits on major shear zones record ages from 1785 ± 21 Ma to 1777 ± 20 Ma, corresponding to deformation, metamorphism and secondary hydrothermal alteration as a result of late orogenic movements. The lack of intracrystalline variations in titanite and allanite trace element chemistry suggests that hydrothermal fluid chemistry and metal source were the main controls on mineral trace element chemistry. Titanite from undeformed Fe oxide–apatite and IOCG deposits is typically light rare earth element (LREE) enriched, and shows low U/Th ratios and low Ni in both intermediate to acid and basic volcanic-hosted deposits. This is consistent with a granitic source for metals. Minor variations in trace element patterns are consistent with the influence of aqueous complex formation on relative REE solubility. Deposits related to the Nautanen Deformation Zone have relatively heavy REE (HREE)-enriched titanite, and LREE-depleted allanite, with high U/Th ratios and elevated Ni contents, consistent with leaching of metals from the local basic volcanic rocks. All hydrothermal titanites are high field strength element enriched (Nb, Ta, Zr) indicating their transport as a result of either high salinities or high F contents, or both. The data overall support models of IOCG-type mineralization as a result of regional circulation of saline hydrothermal fluids in association with major crustal structures, with at least some metallic components derived from the granitoid rocks of the area. All the deposits here show evidence of subsequent metamorphism, although penetrative fabrics are restricted to regional-scale deformation zones.

The Cardigan Pluton, located in southwestern New Hampshire, USA, is a strongly peraluminous granodiorite that contains distinctive, meter-sized pods consisting of 50–70 modal % garnet (+sillimanite + biotite + plagioclase + quartz). The presence of fibrolitic mats and flat, unzoned major and trace-element garnet profiles provide evidence for prograde metamorphism and single-stage garnet growth from biotite dehydration melting. Melt-depleted, bulk-rock compositions indicate that the garnetites are either fragments of restite or melt-depleted xenoliths. Comparison of the Nd and Sr isotopic compositions of the garnetites and the Cardigan granitic rocks indicates an equilibrium relationship as required in the restite model. Additionally, plagioclase and garnet compositions are the same in the garnetites and the most mafic host rocks, also permissive of a restite origin for the garnetites. Temperature and pressure calculated using garnet–biotite thermometry and garnet–aluminum silicate–quartz–plagioclase (GASP) barometry yield estimates of 800°C and 6–7 kbar. These temperatures are probably lower than the peak melting temperature because major element modeling subtracting a granodiorite composition from an average calc-pelite metasediment requires greater than 45% melting to generate garnetite sample CP-23G. Such high melting percentages require temperatures of ~900°C, near the biotite-out curve. We infer that the heat required for such high amounts of melting was obtained from asthenospheric upwelling and basaltic underplating. Inherited zircons with 600 Ma U–Pb ages suggest that the Cardigan Pluton does not have a Laurentian source, consistent with thermobarometric calculations that place the depth of melting below the décollement between the basement and the Central Maine Trough metasediments. A peri-Gondwanan basement source is inferred. Calculated ascent rates of >1000 km/yr allowed preservation of restite in the Cardigan Pluton whereas slower rates in other peraluminous plutons could account for the paucity of restite in most peraluminous plutons.

The Upper Basin Member rhyolites are the oldest known post-collapse rhyolites of the third Yellowstone caldera. They were erupted near the caldera's resurgent domes between 516 ± 7 and 473 ± 9 ka and at 257 ± 13 ka. An unusual characteristic is their low ¹⁸O signature. Few data are available on their mineralogy and glass geochemistry, and this study fills an important gap in understanding their petrogenesis. We report new mineralogical observations and plagioclase, whole-rock and glass compositional data. Based on our observations, we describe a new lava flow for which we propose the name East Biscuit Basin flow. This unit is a quartz- and sanidine-free low-silica rhyolite (71–72% SiO₂ in the whole-rock) in which the dominant mineral, plagioclase, comprises two populations: (1) small fresh euhedral crystals of An_20–48 (average of An₃₁) composition, commonly part of aggregates with pyroxenes and Fe–Ti oxides; (2) large sieve-textured isolated crystals, which are slightly more sodic in composition (An_19–34, average of An₂₇). Plagioclase compositions in most other Upper Basin Member rhyolites are similar. The range of compositions for trace elements such as Rb, Th, Y and the rare earth elements is small (e.g. 158–189 ppm Rb, 20–25 ppm Th, 52–63 ppm Y, 60–82 ppm La in the whole-rock), and there is no systematic variation of these elements as a function of SiO₂ content or mineralogy. Certain trace element signatures and ratios are specific to each of these rhyolites, allowing us to propose that the Upper Basin Member rhyolites originate from six independent magma batches. The coexistence of the two types of plagioclase and their progressive disappearance in the more evolved rhyolites suggest the following petrogenetic model for each magma batch. A low-¹⁸O rhyolitic protolith is heated by replenishing magmas, which initiate melting, forming a crystal mush. Replenishment by buoyant silicic magma may enhance melting and cause mixing with the mush material. As a consequence, crystal-poor eruptible magma batches are formed, which contain small, more calcic plagioclase crystals (aggregates) formed during cooling and mixing of the replenishing silicic melt, and larger, lower temperature crystals exhibiting dissolution features inherited from the protolith.

Experiments have been conducted in the range 3–15 GPa and 850–1800°C to investigate the P–T stability field of OH-apatite in an average mid-ocean ridge basalt (MORB) and a model Mg-basalt, to study the compositional evolution of apatite and its breakdown products and the partitioning of P between phosphates and silicates. In the bulk compositions investigated OH-apatite is stable to <7·5 GPa at 950°C in a typical eclogite assemblage garnet + omphacite + SiO₂ + TiO₂. This is ~5 GPa below the breakdown P of pure OH-apatite. The high-P breakdown product is tuite [-Ca₃(PO₄)₂]. Both apatite and tuite are stable in a wide range of subduction zone T regimes but not along an average mantle adiabat. This precludes apatite or tuite stability in the asthenospheric mantle. Apatite may be stable in cold continental lithosphere (40 mW/m²) but is restricted to P < ~4–5 GPa. The apatite breakdown reaction is an important limit for the crust–mantle transport of Cl in subduction zones and can contribute to the Cl depletion of subducted cust. Both apatite and tuite are important storage sites for large ion lithophile elements (LILE) and rare earth elements (REE), therefore apatite breakdown does not greatly affect LILE or REE transport in subduction zones. In an eclogite assemblage only garnet can accommodate significant P. In the presence of apatite or tuite, P₂O₅ contents in garnet range from ~0·2 to 0·6 wt % between 3 and 11 GPa and increase to ~0·8 wt % at 15 GPa in the absence of a detectable phosphate phase. The P-storage capacity of clinopyroxene is limited to ~250 ppm. Because of the extreme preference of P for the garnet structure, virtually the entire P budget of subducted MORB will be locked up in garnet well into the lower mantle provided fO₂ is high enough to prevent the stability of a metal phase.

In situ electron microprobe monazite dating and mineral equilibria modelling of amphibolite–granulite-facies metapelites from the southern Prince Charles Mountains, East Antarctica has been carried out to unravel the P–T conditions, spatial extent and structural style of two overprinting orogenic records. This study shows that: (1) rocks of the northern Palaeoproterozoic Lambert Complex were pervasively reworked at peak conditions (6·5–7·1 kbar and 790–810°C) during the Early Neoproterozoic Rayner orogenic event; (2) rocks of the southern Lambert Complex experienced pervasive deformation and metamorphism at peak conditions (5·8–6·1 kbar and 625–635°C) during Early Palaeozoic Prydz orogenic activity; (3) in regions of the Lambert Complex reworked during the Rayner orogenic event, Prydz-aged orogenesis was highly localized. The distribution of orogenic activity pertaining to the Rayner and Prydz orogenic events in the southern Prince Charles Mountains can be attributed to (1) the development of a southward directed (current coordinates) orogenic front that propagated from an Early Neoproterozoic collision between India and Antarctica, and (2) rock fertility (i.e. availability of free fluid) during Early Palaeozoic intraplate orogenesis that was driven by far-field stresses generated by a collision of India–Antarctica with the Mawson Craton.

We conducted a microstructural study of a high-strain mantle shear zone from the Josephine Peridotite, SW Oregon, USA. The goal of this study is to understand how microstructural evolution at large strains leads to transitions in rheological behavior. The shear zone we investigated exhibits higher strain and greater localization than previously studied shear zones in the Josephine Peridotite. The margins of the shear zone have a homogeneous microstructure, characterized by moderately strong olivine fabrics, fairly weak orthopyroxene fabrics, and grain sizes of 2–3 mm. The highly deformed samples from the center of the shear zone display two distinct microstructural domains—a relatively coarse-grained domain (~550 µm) that contains only olivine and a finer-grained domain (~250 µm) that contains both olivine and orthopyroxene. The coarse-grained domain has a strong E-type olivine lattice-preferred orientation (LPO). Within the fine-grained domain the olivine LPO is also E-type, but significantly weaker. The E-type fabrics are rotated slightly past the shear plane, providing the first field-based confirmation of similar experimental observations. The presence of E-type fabrics, which form in the presence of moderate quantities of water, also highlights the potential importance of water to shear zone evolution. The orthopyroxene in the fine-grained domains has no LPO, suggesting that a transition to grain-size sensitive deformation occurred. The microstructural transition in orthopyroxene may have resulted in a marked weakening of the rock, suggesting that orthopyroxene plays a critical role in shear localization. These samples provide a crucial microstructural link between moderately localized shear zones and highly deformed ultramylonites.

Microstructural and in situ mineral chemistry studies on mantle peridotite xenoliths from the Late Neogene alkaline volcanic center of Cabezo Tallante (SE Spain) reveal an exceptional record of a multi-stage history of deformation, recrystallization, melt–rock interaction and melt intrusion tracking the progressive exhumation of this lithospheric mantle sector. Xenoliths include porphyroclastic to equigranular spinel peridotites, impregnated plagioclase peridotites, and composite xenoliths made up of peridotites intruded first by gabbronorite veins and later by amphibole-bearing pyroxenites. The earliest stage involved subsolidus re-equilibration from garnet- to spinel-facies conditions, represented by rounded opx + spinel ± cpx clusters indicative of precursor garnet. The spinel-facies equilibration was followed by development of a porphyroclastic fabric, accentuated in many xenoliths by spinel trails, in response to shear deformation that may be related to the early stages of Neogene extension. Porphyroclastic spinel peridotites subsequently underwent multiple episodes of reactive porous melt percolation documented by crystallization of undeformed olivine replacing pyroxene porphyroclasts, and of undeformed poikilitic orthopyroxene at the expense of both pyroxene porphyroclasts and newly crystallized olivines. The porphyroclastic and melt–rock reaction textures are progressively obliterated by an equigranular structure developed as the result of static, possibly melt-assisted, annealing recrystallization. Clinopyroxenes in equigranular peridotites (i.e. the most equilibrated with the percolating melts) display slight light rare earth element (REE) depletion and almost flat middle to heavy REE spectra (La_N/Yb_N = 0·37–0·62; Sm_N/Yb_N = 0·89–1·23). Computed equilibrium liquids have an enriched tholeiitic affinity, consistent with the sub-alkaline magmatism of the Alboran Domain. Overall, the tectonic and magmatic stages recorded in spinel peridotites from Tallante are remarkably consistent with the evolution documented in the Ronda peridotites of the western Betics. Reactive porous flow and annealing recrystallization were followed by an impregnation event, documented by crystallization of interstitial (plag ± opx ± ol) aggregates in porphyroclastic and equigranular xenoliths; this indicates further exhumation to shallower depths. Diffuse melt percolation was followed by intrusion of melts with distinct chemical affinity. The first event is documented by thin gabbronoritic–noritic veins, showing opx reaction rims against the host peridotite. Comparable gabbronorites were previously ascribed to slab-derived melts. The norite veins are crosscut by centimeter-thick dikelets of amphibole pyroxenite. Geobarometric estimates and the observed crystallization order (ol–cpx–amph–plag) point to 0·7–0·9 GPa for pyroxenite intrusion. Computed melts in equilibrium with clinopyroxene show alkaline affinity, similar to the host Tallante alkali basalts. Textural and geochemical features in the xenoliths thus indicate that the progressive uplift of the Tallante lithospheric mantle was accompanied by interaction with melts of different sources, reflecting the magmatic evolution of the Alboran Domain in response to lithosphere extension and thinning leading to the formation of the Betic–Rif arc.

Torsion experiments on partially molten aggregates of olivine + chromite + 4 vol. % mid-ocean ridge basalt provide new insights into the interactions between deformation and melt segregation. When samples are sheared, melt segregates into distinct melt-rich bands oriented ~20° antithetic to the macroscopic shear plane. In one series of experiments, samples were deformed at similar shear strain rates (or stresses) to a range of finite shear strains to explore the evolution of melt-rich bands. In another series of experiments, samples were deformed to similar finite shear strains at a range of strain rates to explore the effect of strain rate (or stress) on band spacing and microstructure. We relate variations in strain rate to the compaction length and show that band spacing increases with increasing compaction length. These experiments provide new information on the evolution of melt distribution, the partitioning and localization of strain, and the scaling of experimental results to the Earth's mantle.

Major, trace, and rare earth element compositions of both tonalite–trondhjemite–granodiorite (TTG) and modern adakite-like magmas are typically used in conjunction with batch melting experiments and models to infer source rock composition, depth of melting and tectonic setting. However, batch melting does not capture the impact of melt segregation processes on magma geochemistry. We have used melting experiments in conjunction with numerical modelling to investigate the impact of melt segregation on TTG arc crust formation. Our melt segregation equilibrium (MSE) experiments are designed to reproduce the local changes in bulk composition that are predicted by the numerical model to occur as buoyant melt migrates upwards along grain boundaries and accumulates to form a magma that leaves the source region. The MSE experimental results show distinct differences in the melt and solid phase compositions and solid phase stability when compared with direct partial melting experiments. They yield a significant reduction in hornblende and plagioclase modal proportions at lower temperatures and partial melt compositions that are lower in An and have higher Mg-numbers. These results suggest that dynamic melt segregation and equilibrium processes may have a significant impact on modes, melt compositions and geochemical indicators such as Mg-numbers. Mantle wedge interaction may not be necessary to generate varying Mg-numbers in TTG and adakite magmas. Moreover, the use of batch melting models or experiments to interpret these geochemical signatures may not be appropriate.

Carbonatites commonly occur together with alkaline silicate rocks. It is frequently argued in the literature that the varied ⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd and Pb isotopic ratios in the alkaline silicate rocks reflect source heterogeneity, and that in many cases bear no genetic significance to the coexistence of silicates and carbonatites. Such a hypothesis may not be universal as it fails to explain the observations that in numerous carbonatite–alkaline complexes both rock types are contemporaneous, they exhibit complementary trace element patterns, and most importantly their initial radiogenic isotopic ratios overlap, all of which when considered together suggest a common parentage. In addition, the alkaline silicate rocks have more variable and higher ⁸⁷Sr/⁸⁶Sr, and lower ¹⁴³Nd/¹⁴⁴Nd and Pb isotopic ratios compared with those of the carbonatites and show hyperbolic trends in isotopic ratio vs concentration plots, hinting at a possible contamination of their parental magma by crustal or lithospheric material. Using a mathematical model that quantifies the isotopic ratios and concentrations of an element during concurrent assimilation and fractional crystallization of silicate rocks combined with immiscible separation of carbonate melt (AFCLI), I propose that the isotopic ratio variations in most carbonatite–alkaline silicate complexes can be explained by assimilation of crustal material by parental carbonated silicate parental magmas. A highly plausible scenario that emerges from this exercise is that not only are the carbonate and associated silicate magmas derived from a single parental magma, but also that the lower crust plays an important role in their diversification.

Apatites crystallized from different types of igneous rocks show significant variations in the abundances of some minor and trace elements. In this study, electron probe microanalysis and laser ablation inductively coupled plasma mass spectrometry were used to determine the concentrations of 25 minor and trace elements in apatite separated from three principal rock types of the Transhimalayan igneous plutonic suite: S-type granites, the I-type Gangdese batholith and post-collisional adakites. F, Mn, Sr and rare earth elements (REE) in apatite vary systematically with the composition of the host magma and thus have high potential as petrogenetic tracers. More specifically, the F and Mn contents of apatite can be used as an indicator of magma aluminosity or differentiation index. Combined with Sr and REE data, which show significant variations in apatite from different rock types, these elements are useful for constructing ‘discrimination diagrams’. This study also reveals that apatite has the capacity to retain geochemical information about the host magma through the course of magmatic evolution. Systematic variations of Sr and REE in apatite with bulk-rock aluminosity are the results of partition competition with pre-existing and coexisting major and accessory minerals in silicate melts, and thus are useful for more detailed investigations of petrogenetic processes such as fractional crystallization and magma mixing, which is signaled by inconsistent Eu anomalies, Sr abundances and REE patterns relative to bulk-rock compositions.

It is now well established that lithospheric thinning of the eastern North China Craton (NCC) occurred during the Phanerozoic. The cause and extent of the thinning, however, remains highly debated. In this study, mantle xenoliths from the Paleozoic Mengyin kimberlites, along with xenoliths from the Cenozoic Penglai and Shanwang basalts in Shandong Province, are investigated via traditional petrographic and elemental, Sr–Nd–Hf–Os isotopic and platinum-group element (PGE) analyses. Late Archean Os model ages of c. 2·5 Ga for the Mengyin peridotites provide confirmation that refractory, Archean lithospheric mantle existed beneath the easternmost portion of the NCC during the Paleozoic. Some Paleoproterozoic lithospheric mantle fragments may also be present in the Mengyin xenolith suite. In contrast, the spatially associated Penglai and Shanwang peridotite xenoliths are more fertile, and have Sr–Nd isotopic compositions similar to the depleted mantle. They differ dramatically from the Archean peridotites sampled by the Mengyin kimberlites. Osmium model ages for single samples range from mid-Proterozoic to modern, similar to variations observed in Phanerozoic convecting upper mantle as sampled by modern abyssal peridotites. This suggests that the present lithospheric mantle beneath the eastern NCC formed in the Phanerozoic, despite the fact that Os model ages extend back to the Proterozoic. Some samples from the Penglai suite yield Proterozoic Lu–Hf clinopyroxene mineral isochron ages that are consistent with the Os model ages, suggesting that Hf isotopes in modern convective upper mantle can preserve evidence for ancient melt depletion, similar to Os isotopes. Our results are consistent with thinning of the eastern NCC as a result of foundering of the deep crust and lithospheric mantle, but are inconsistent with stretching or refertilization models, as remnants of Archean mantle are expected to be present in these scenarios. The match between ¹⁸⁷Os/¹⁸⁸Os in convective upper mantle and the eastern China lithopspheric mantle also precludes models that seek to explain the present lithosphere as being due to lateral translation of Proterozoic lithosphere.

The subduction-related Michoacán–Guanajuato Volcanic Field (MGVF) in central Mexico contains ~900 cinder cones and numerous larger shield volcanoes of Late Pliocene to Holocene age. We present data for major, trace and volatile (H₂O, CO₂, S, Cl) elements in olivine-hosted melt inclusions from eight calc-alkaline cinder cones with primitive magma characteristics and one more evolved alkali basalt tuff ring. The samples span a region extending from the volcanic front to ~175 km behind the front. Relationships between H₂O and incompatible trace elements are used to estimate magmatic H₂O contents for 269 additional volcanic centers across the MGVF and central Mexico. The results show that magmatic H₂O remains high (3–5·75 wt %) for large distances (~150 km) behind the front. Chlorine and S concentrations are strongly correlated with melt H₂O and are also high across most of the arc (700–1350 ppm Cl, 1500–2000 ppm S). The alkali basalt, located far behind the front (~175 km), has much lower volatile contents (<1·5 wt % H₂O, 200 ppm Cl, 500 ppm S), and is compositionally similar to other melts erupted in this region. Oxygen isotope ratios of olivine phenocrysts (5·6–6) from the calc-alkaline samples are higher than for typical mantle-derived magmas but do not vary systematically across the arc. Calc-alkaline samples have high large ion lithophile element concentrations relative to Nb and Ta, as is typical of subduction-related magmas, but alkali basalt samples far behind the front have high Nb and Ta and lack enrichments in fluid-mobile elements. Modeling based on volatiles and trace elements suggests that the calc-alkaline magmas were generated by 6–15% partial melting of a variably depleted mantle wedge that was fluxed with H₂O-rich components from the subducted slab. In contrast, the alkali basalts formed by small degrees of decompression melting of an ocean island basalt source that had not been fluxed by slab-derived components. Based on high ¹⁸O_olivine values and trace element characteristics, the H₂O-rich subduction components added to the mantle wedge beneath the MGVF are likely to be mixtures of oceanic crust derived fluids and sediment melts. Integrating these results with new 2-D thermo-mechanical models of the subduction zone beneath the MGVF, we demonstrate that the present-day plate configuration beneath the MGVF causes fluids to be released beneath the forearc and volcanic front, and that sediment melts can be produced beneath the volcanic front by the waning stages of fluid released from the oceanic crust percolating through already dehydrated sediments. Down-dragging of serpentine- and chlorite-bearing peridotite in the lowermost mantle wedge probably plays a role in fluid transport from the forearc to beneath the arc. H₂O-rich magmas located more than ~50 km behind the volcanic front can be explained by mantle hydration related to a shallower slab geometry that existed at ~3 Ma. Rollback of the slab over the last ~2 Myr has resulted in strong mantle advection that forms low-H₂O, high-Nb alkali basaltic magmas by decompression melting far behind the present-day volcanic front.

We report on microstructural data obtained by optical microscopy and transmission electron microscopy concerning the crystallographic relationships of serpentine minerals with their host olivine in two contrasting situations. In the first case, mesh-textured lizardite (liz) is developed in a standard 60% serpentinized oceanic harzburgite from the Oman ophiolite where olivine converts to columnar lizardite. The joined columns are perpendicular to the basal plane (001)_liz, corresponding to the pseudofibres observed optically. The plane (001)_liz is locally parallel to the narrow boundary ol–liz; thus column orientations register the interface of serpentinization. The ol–liz relationships are not strictly topotactic, but reflect preferred cracking orientations in olivine, parallel to (010)_ol. In the second case, antigorite (atg) develops in a rare sample of antigorite schist in a kimberlite from Moses Rock (Colorado Plateau), representative of a suprasubduction-zone mantle wedge. High-resolution transmission electron microscopy (HRTEM) images along [010]_atg show domains of very regular modulation with a 43·5 Å wavelength (m = 17, where m is the number of silicate tetrahedra along the wave), with few defects, indicative of HP–HT antigorite, and also heavily kinked regions as fingerprints of strong tectonic shear. TEM imaging and electron diffraction patterns reveal two topotactic relationships between antigorite and olivine: [100]_atg//[010]_ol and <100>_atg//<100>_ol; the planes in contact are (001)_atg//(100)_ol and (001)_atg//(010)_ol, respectively. The [010]_atg//[001]_ol and antigorite lamellae are parallel to the forsterite b-axis. In both cases, the topography of olivine–serpentine interfaces is controlled by open fluid pathways along microcracks oriented according to the anisotropy of the olivine aggregate. In the cases studied, the serpentine aggregate exhibits a preferred orientation inherited from that of the peridotite. These results have some relevance to the seismic anisotropy of serpentinized mantle. Anisotropy of propagation of seismic waves as a result of the olivine fabric is maintained and reinforced with the development of lizardite. Conversely, the development of antigorite produces a trench-parallel fast S-wave polarization and an anisotropy that is lowered at low degrees of serpentinization and then increased with increasing serpentinization.

High-strain torsion experiments were performed on a series of samples composed of anorthite plus <1 to 12% melt to investigate the formation of melt-rich bands produced by stress-driven melt segregation. Fine-grained (3–4 µm) samples were deformed in the diffusion creep regime at a temperature of 1450 K and a confining pressure of 300 MPa at shear strain rates of 1 x 10^–4 to 16 x 10^–4 s^–1 and shear stresses of 15–150 MPa to shear strains between = 1·9 and 6·6. The dependence of viscosity, , on melt fraction, , for these partially molten aggregates can be expressed as = 2·6 x 10¹² exp (–24 ) Pa s. In each sample, melt-rich bands develop by a shear strain of = 1, forming a population of bands at an angle of 5–15° to the shear plane and 40–30° to the applied maximum principal stress. The spacing between and width of the melt-rich bands increases as melt fraction increases from <0·01 to 0·06, then roughly levels off as melt fraction increases to 0·12. This band spacing, _s, increases linearly with increasing compaction length, _c, according to the relation _s = 0·07 _c when the bulk viscosity is assumed to be twice the shear viscosity. In the Earth, spontaneous stress-driven segregation of fluids is an important mechanism for localizing deformation into shear zones.