The Ontong Java Plateau, mainly emplaced during the Early Cretaceous on the Pacific Plate, is the most voluminous extant oceanic plateau. The exceptional volume and flux of magma that formed this plateau is widely thought to result from melting induced by a hot, buoyant mantle plume head; however, a purely thermal plume predicts uplift to above sea level, inconsistent with the plateau's mostly submarine emplacement. An alternative formation mechanism is rapid seafloor spreading inducing decompression melting of a mantle portion with a relatively high proportion of dense fusible component. Here we use thermodynamic models simulating decompression melting of heterogeneous mantle sources to constrain the mantle potential temperatures and dense fusible mantle proportions required to form the Ontong Java Plateau under mantle plume and seafloor spreading scenarios. We show that the seafloor spreading model requires an unreasonably high mantle potential temperature or dense fusible pyroxenite proportion. By contrast, a thermochemical plume, with a temperature 135-200 degrees C higher than ambient mantle and up to 13% dense fusible pyroxenite, can explain not only the spatial variations in crustal thickness and lava compositions of the plateau, but also its mostly submarine emplacement. Thus, we propose that the formation of this plateau is best explained by a thermochemical mantle plume.

The fate of continental lithosphere during rifting is central to the process of continental extension. The continental lithospheric mantle comprises both depleted and enriched domains that may contribute to magma generation during extension. The East African Rift System is the archetypal example of a magma-rich continental rift, with the Turkana Depression containing the most extensive temporal record of mafic magmatism. There is debate as to the contribution of continental lithosphere to this mafic magmatism, with suggestions that HIMU-like isotopic signatures, often attributed to the continental lithosphere, are derived instead from a heterogeneous mantle plume. We focus on Miocene lavas that are characterized by radiogenic Pb-206/Pb-204 > 19.3, requiring a contribution from an HIMU-like endmember in their origin. We present a novel two-stage chromatographic metasomatism model that demonstrates that a HIMU-like endmember can be generated through time-integrated evolution within the continental lithospheric mantle. The first model stage uses an initial composition for the metasomatizing agent equivalent to a subduction magma to generate metasomes within the continental lithosphere during the Pan-African stabilization of the regional lithosphere (similar to 700 Ma). During Mesozoic rifting, the second model stage simulates destabilization and melting of these initial metasomes, re-enriching the surrounding lithosphere to generate new Mesozoic metasomes. Melts of these metasomes, when combined with melts of the regional asthenosphere, are consistent with the observed trace element and isotopic signatures of Turkana Miocene lavas. These findings suggest an important role for the continental lithospheric mantle during rifting and obviate the need for a complex, heterogenous plume. Plain Language Summary The Turkana Depression of the East African Rift System contains an extensive record of rift-related magmatism associated with a mantle plume. While most Turkana magmatism is attributed to melts of a plume-influenced upper mantle, some lava geochemistry suggests melting of an enriched source. Previous work has interpreted these enriched geochemical signatures as melts of a heterogenous plume. However, melting of a metasomatized continental lithospheric mantle may also generate enriched lavas. This study presents new major, trace element, and isotopic data of enriched Turkana lavas from the Miocene. We performed simulations of metasome formation and melting to assess the potential contribution of a metasomatized lithospheric mantle to Turkana lavas. We also apply isotopic constraints to this model to infer the age of metasomatic enrichment events. We find that Turkana Miocene lava geochemistry can be explained by melting enriched metasomes in the lithosphere and mixing these melts with upper mantle melts.

The East African Rift System (EARS) provides an opportunity to constrain the relationship between magmatism and plate thinning. During continental rifting, magmatism is often considered a derivative of strain accommodation-as the continental plate thins, decompression melting of the upper mantle occurs. The Turkana Depression preserves among the most extensive Cenozoic magmatic record in the rift. This magmatic record, which comprises distinct basaltic pulses followed by periods of relative magmatic quiescence, is perplexing given the lack of evidence for temporal heterogeneity in the thermo-chemical state of the upper mantle, the nonexistence of lithospheric delamination related fast-wave speed anomalies in the upper mantle, and the absence of evidence for sudden, accelerated divergence of Nubia and Somalia. We focus on the Pliocene Gombe Stratoid Series and show how lithospheric thinning may result in pulsed magma generation from a plume-influenced mantle. By solving the 1D advection-diffusion equation using rates of plate thinning broadly equivalent to those measured geodetically today we show that despite elevated mantle potential temperature, melt generation may not occur and thereby result in extended intervals of quiescence. By contrast, an increase in the rate of plate thinning can generate magma volumes that are on the order of that estimated for the parental magma of the Gombe Stratoid Series. The coincidence of large-volume stratiform basalt events within the East African Rift shortly before the development of axial zones of tectonic-magmatic activity suggests that the plate thinning needed to form these stratiform basalts may herald the onset of the localization of strain. The magmatic record in the Turkana Depression-part of the East African Rift System-is characterized by pulses of basaltic activity that are followed by long periods of relative magmatic quiescence. This is a puzzling observation assuming that these magmas are generated by decompression melting of the upper mantle; there is no obvious changes in the rate of plate motion between Nubia and Somalia. This study presents new geochemical data on the final pulse of basaltic volcanism (during the Pliocene) and interprets these data in the context of a mantle melting model. We find that pulses of basaltic volcanism and intervening periods of quiescence could be simulated using different rates of thinning of the plate. We examine the consequences of a period of enhanced plate thinning in context of melt generation both below and within the plate. Pulses of basaltic volcanism and intervening periods of quiescence could be simulated using different rates of thinning of the plate

As continents break apart, the dominant mechanism of extension transitions from faulting and lithospheric stretching to magma intrusion and oceanic crust formation in a new ocean basin. A common feature of this evolution preserved at magmatic rifted margins worldwide are voluminous lava flows that erupted close to sea level during the final stages of development of the continent-ocean transition (COT). The mechanisms responsible for the generation of the melts that contribute to these voluminous flows, the so-called seaward dipping reflectors (SDR), and their significance in the context of COT development, are relatively poorly understood; they lie deep below post-rift strata along submarine rifted margins where they cannot be studied directly. Extensive coring of the Afar Stratoid Series - an areally-extensive sequence of Pliocene-aged basalts and intercalated sediments that lie atop the developing COT in the sub-aerial Afar Depression, northern Ethiopia - offers fresh scope to address this issue. We present a numerical model simulating the formation of enriched metasomes within the continental lithospheric mantle by the passage of magmas resembling modern axial basalts. Thermal destabilization of the metasome, caused by plate stretching, initiates melt formation within the metasome. These melts, when mixed with a depleted lithospheric mantle component, closely match the range of compositions of the Afar Stratoid Series lavas in this study. Metasomatic re-enrichment and subsequent melting of the lithospheric mantle during the COT may contribute to further plate thinning. These results demonstrate a novel mechanism by which large-volume flows may be erupted during the COT.

The Skaergaard intrusion is one of the most thoroughly studied layered mafic intrusions on Earth and an exceptional example of (near) closed-system magmatic differentiation. We report new Fe isotope data for whole rocks, and magnetite and ilmenite separates through the layered series (LS) and upper border series (UBS) of the intrusion. d56Fe values for gabbroic rocks range from 0.033 parts per thousand to 0.151 parts per thousand with an abrupt step increase at the base of Lower Zone c (LZc) within LS with the appearance of cumulus magnetite and subsequent decline accompanying FeTi oxide fractionation. The lowest d56Fe values are found near the Upper Zone b (UZb) to c (UZc) boundary followed by a sharp rise across UZc approaching the Sandwich Horizon. Magnetiteilmenite separates straddle bulk rock compositions with fractionation factors (? 56Fe) of 0.081 parts per thousand to 0.239 parts per thousand, consistent with subsolidus equilibration. Granophyric rocks occurring as pods, sheets and wispy layers from the upper zone and UBS equivalents and having unradiogenic Sr similar to gabbroic rocks of Skaergaard, are isotopically heavier than their host ferrodiorites (? 56Fe?=?0.1 parts per thousand) reaching a maximum d56Fe of 0.217 parts per thousand in UBS. A fused xenolith from UBS has d56Fe ?=?0.372 parts per thousand This range in d56Fe spans much of that reported for terrestrial igneous rocks, and like the global dataset, shows a pronounced increase in d56Fe with inferred silica content of modeled Skaergaard liquids. Forward modeling of closed system fractional solidification was undertaken to account for Fe isotope systematics, first by testing published liquid lines of descent (LLD), and then by exploring improvements and considering the impacts of liquid immiscibility, crustal contamination, fluid exsolution and diffusional processes. Our modeling relies on published Fe+2 and Fe+3 force constants for magmatic minerals and silicate glasses, and the most reliable estimates of the average bulk composition and mass proportions of the well-defined subzones of the intrusion. We show that the increase in d56Fe across the LZbLZc boundary is readily explained by the increased incorporation of Fe+3 into the crystallizing solid including magnetite. We further demonstrate that the classic Fenner LLD, involving strong Fe enrichment at nearly constant silica, does not lead to a rise in d56Fe toward the end stages of evolution, while a Bowen-like LLD, with little Fe enrichment and strong Si enrichment, also underestimates enrichment in heavy Fe isotopes in the ferrodiorites of UZc. A LLD following an intermediate path involving modest Fe and Si enrichment, followed by Fe depletion best explains the observations. We predict similar to 3.5% (by mass) residual liquid after crystallization of UZc having a composition similar to felsic segregations in pegmatitic bodies found in the intrusion. While liquid immiscibility may have been encountered within fractionating mush at the margins of the intrusion, the Fe isotope systematics do not support liquid phase separation of the bulk magma. Crustal contamination, fluid exsolution, hydrothermal alteration and thermal diffusion are also shown to have no resolvable effect on the Fe isotope composition of the gabbroic and granophyric rocks. We conclude that the Fe isotope systematics documented in the Skaergaard intrusion reflect the dominant role of fractionating Fe-rich minerals from gabbroic through ferrodioritic to rhyolitic liquids. The success of our model to account for the observed Fe isotope systematics for Skaergaard demonstrates the utility of Fe+2 and Fe+3 force constants determined at ambient conditions to model magmatic conditions and gives critical insights into plutonic processes fractionating Fe isotopes complementary to the volcanic record.

The recycling of lithospheric material into the mantle by subduction, and the subsequent melting of this recycled material beneath mid-ocean ridges and ocean islands is a fundamental process in the ongoing differentiation of the Earth.Much of our understanding of lithosphere recycling comes from studies that apply mantle melting models to constrain mantle potential temperature and the proportion of recycled material required to match oceanic basalt composition and crustal thickness [1][2][3][4].While this approach has been fruitful for constraining the relative abundances of fertile recycled basaltic crust (present as pyroxenite) and lherzolite, constraining the amount of highly depleted peridotite (harzburgite) is more difficult because it contributes little, if anything, to melt production.Despite this challenge, recent studies [2,3] have estimated harzburgite abundances by applying a combination of three constraints: crustal thickness, proportion of pyroxenite-derived melt comprising the crust (Xpx; derived from basalt compositions), and either olivine crystallization temperatures or source buoyancy to mantle melting models.These studies conclude that >40% harzburgite is present in the sources for MORBs and OIBs, including Iceland basalts.This contrasts with work showing that basalt compositions and crustal thicknesses at Iceland do not require harzburgite in the mantle source [4].These diametrically opposed views are rooted in model assumptions that require closer examination.To explore the effects of model assumptions, we have extended our mantle melting code, REEBOX PRO [5], to encompass the range of assumptions made in previous models, including the incorporation of olivine crystallization temperatures, batch polybaric melt productivity, and source compaction accompanying melt extraction.Combining REEBOX PRO with Markov chain Monte Carlo sampling, we provide a rigorous assessment of the sensitivity of inferred harzburgite abundances to model assumptions and further illustrate the value of using basalt compositions, instead of Xpx, as a model constraint.

Shatsky Rise is a large oceanic plateau formed at a spreading ridge triple junction, but its origin mechanism is unclear. Voluminous magmatism at Shatsky Rise has been regarded either as the product of decompression melting of fertile recycled material in a divergent plate boundary setting (plate model) or a hot mantle plume head (plume model). Here we use thermodynamic models to simulate decompression melting of heterogeneous mantle sources. We provide constraints on both mantle potential temperature and fertile source abundance under seafloor spreading and plume–ridge interaction scenarios. As constrained by crustal thickness and lava chemical compositions, a seafloor spreading origin requires mantle potential temperatures of 1,490–1,585 °C, which are unreasonably high for mid-ocean ridge systems far from hotspots. In contrast, the plume–ridge interaction model can explain observations with mantle potential temperatures of 1,470–1,515 °C for the buoyant plume and upwelling rates up to six times the rate of plate separation during the formation of the main massifs. Fertile recycled material comprises less than 7% of the mantle source, and predicted melt fractions are consistent with independent estimates. We thus suggest that Shatsky Rise magmatism is best explained by plume–ridge interaction—a combination of the plume and plate models.

To examine the effect of upper mantle compositional variations on seismic velocities, we exploit the Perple_X thermodynamic solver using the Holland and Powell database, and the mineral physics parameters of Schutt and Lesher (2010), to estimate seismic wave speeds through a wide range of mantle xenoliths.Our database includes peridotitic and eclogitic xenoliths from abyssal, forearc, cratonic, off-cratonic continental, passive margin, ophiolite, and massif settings.Initial results show that, although large compositionally modulated velocity variations occur in each setting, the mean velocities in these geologic regions are almost the same.Thus, we conclude that, on average, different settings have almost no compositional effect of seismic velocity.This finding suggests that seismic velocities vary compositionally over small scales, but that the mean seismic velocity observed on a regional scale is nearly independent of composition.Then, using these measured velocities, we estimate temperatures at the base of the crust in the U.S., China, and Australia by mapping Pn velocities to temperature.These temperatures are compared to xenolith-derived geotherms to determine spatio-temporal thermal variations in the mantle lithosphere.Our lithospheric temperatures in the U.S. (updated from Schutt et al., 2018) show patterns consistent with recent tectonism, whereas temperatures in Australia and China are more complex and show large variations.Areas that are predicted to be anomalously hot using our Pn conversion are generally consistent with Cenozoic-Recent xenolith locales where geothermometry indicates high temperatures, but these large variations also may be due to instability in derived Pn velocities.

Heavy rare earth elements (HREEs) in mafic and ultramafic volcanic rocks are useful recorders of mantle source processes because their ratios are not easily modified by differentiation. Here we utilize REEBOX PRO, a simulator of adiabatic decompression melting of the mantle, to study the behavior of HREEs in the formation of continental flood basalt (CFB) parental magmas in the mantle. We simulate partial melting of depleted peridotite, pyrolitic peridotite, pyroxenite, and peridotite-pyroxenite mixtures at mantle potential temperatures of 1350-1650 degrees C and lithospheric thicknesses of 50-150 km, and compare the results to natural data. Many large igneous provinces are typified by low-Ti and high-Ti CFBs with contrasting HREE patterns. Our results show that low-Ti CFBs originate mainly from peridotitic sources. Flat mid-ocean ridge basalt-like HREE patterns typical of low-Ti CFBs can be generated beneath thick lithosphere (similar to 100 km), given that mantle potential temperatures are high (>1500 degrees C) and garnet is completely consumed from the source. We thus challenge the common interpretation that flat HREE patterns always indicate shallow sources for CFB parental magmas. High-Ti CFBs require pyroxenite-bearing sources (>= 10%). Contrary to a common view, their steep oceanic island basalt-like HREE patterns can be generated beneath quite a thin lithosphere (similar to 50 km), which is due to increased garnet stability in pyroxenite sources. When applied to CFBs of the Karoo large igneous province, the results are compatible with a model where a mantle plume penetrates a progressively thinning Gondwana lithosphere.

The production of the magma volumes necessary for the formation of continental flood basalts is thought to result from extension-related decompression melting of thermo-chemically anomalous mantle. The relative importance of elevated mantle temperature or decompression associated with lithospheric thinning remains difficult to constrain because the temporal relationship between the initiation of extension and continental flood basalt is often ambiguous. The -1.1 Ga Midcontinent Rift (MCR) in North America provides an opportunity to probe these magma generation factors because the mantle thermo-chemical anomaly (Keweenaw plume) thought responsible for the continental flood basalts of the Keweenaw Large Igneous Province (LIP) intersected an existing rift. The role of elevated mantle temperatures in Keweenaw LIP magmatism remains ambiguous because a temporal decrease in MgO content in primitive lavas may reflect decreasing melting depths with time due to progressive lithospheric thinning, a temporal decrease in mantle temperature, or both. Furthermore, paleomagnetic studies show that Laurentia was moving rapidly during Keweenaw LIP magmatism, raising the question of how the plume remained connected to the MCR during 25 Myr of volcanism. Using samples from the most complete volcanic section preserved at Mamainse Point, Ontario, Canada, we constrain mantle potential temperature and lithospheric thickness by combining three major element thermobarometry models with forward modeling of rare earth element compositions. We find that the earliest stage of Keweenaw LIP magmatism (Early phase; 1109-1104 Ma) was governed by initially deeper melting (>85-100 km thick lithosphere) and high mantle potential temperatures (>1480-1630 degrees C) related to the plume. Potential temperatures decreased to near-ambient conditions (-1400-1445 degrees C) during the latter portion of the Early phase, with these lower temperatures persisting throughout the Latent (1104-1098 Ma) and Main (1098-1090 Ma) phases. Major element-derived pressures of melt equilibration place broad constraints on lithospheric thickness, which decreased from -60-110 km at the onset of magmatism to 45-65 km by the Main phase. The decrease in primitive MgO contents thus reflects thinning lithosphere and decreasing mantle potential temperatures. We argue that the temporal decrease in mantle potential temperature reflects the severing of the connection between the MCR and Keweenaw plume after the Early phase of magmatism by the rapid plate motion. Our constraints on mantle potential temperature resolve the relationship between the MCR, the plume, Keweenaw LIP magmatism, and rapid plate motion, and indicate that an alternative mechanism, such as enhanced small-scale convection due to rapid plate motion and plume-lithosphere interaction, was required to generate the voluminous Main magmatic phase.(c) 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Continental flood basalt provinces (CFBPs) are large igneous features formed by the extrusion of massive amounts of lavas that require significant evolution within the lithosphere. Although sequential lava flows are effective probes of magmatic systems, CFBPs are typically poorly preserved. We focus on lava flows from the well-preserved 1.1 Ga Keweenawan CFBP that erupted within the Midcontinent Rift System. We present a new geochemical, petrographic, and stratigraphic synthesis from the Main stage Portage Lake Volcanics (PLV). Flow-by-flow analysis of the PLV reveals that major element behaviour is decoupled from trace element behaviour; MgO exhibits limited variability, while compatible and incompatible trace elements deviate from high to low concentrations throughout the sequence. The concentrations of incompatible trace elements slightly decrease from the base of the sequence to the top. We investigate these observations by applying a recharge, evacuation, assimilation and fractional crystallization model to geochemical and petrographic data. Our modelling demonstrates a magmatic system experiencing increased evacuation rates while fractionation and assimilation rates decrease, indicating an increase in magmatic flux. The outcome of this modelling is a progressively more efficient magma system within the PLV. This study highlights the utility of joint petrographic and geochemical interpretation in constraining CFBP magma evolution.

Continental flood basalt lavas often contain deeply-sourced, thermo-chemically anomalous material that can provide a potential probe of inaccessible reservoirs. However, continental flood basalts interact with geochemically diverse domains within the continental lithosphere, which may complicate interpretations of deep mantle signatures. We examine the role of continental lithospheric mantle in continental flood basalts erupted as part of the 1.1 Ga Keweenawan large igneous province, centered on the Lake Superior region of North America. We show that flood basalts at Mamainse Point exhibit a range of εHf 1100 from −14.1 to +6, plotting along the global εHf—εNd mantle array. Lithospheric mantle melts represented by alkaline rocks from the Coldwell and Seabrook Lake Complexes yield positive εNd 1100 (+0.7 to +4.3) and εHf 1100 from −6.9 to +2.4, placing them below the mantle array. Mamainse Point lavas are interpreted to be variably crustally contaminated melts of the Keweenawan plume and ambient upper mantle; there is no clear evidence for contributions from an enriched lithospheric mantle.

Basalts are generated by adiabatic decompression melting of the upper mantle, and thus provide spatial and temporal records of the thermal, compositional, and dynamical conditions of their source regions. Uniquely constraining these factors through the lens of melting is challenging given the complexity of the melting process. To limit the a priori assumptions typically required for forward modeling of mantle melting, and to assess the robustness of the modeling results, we combine a Markov chain Monte Carlo sampling method with the forward melting model REEBOX PRO [1] simulating adiabatic decompression melting of lithologically heterogeneous mantle. Using this method, we invert for mantle potential temperature (Tp), lithologic trace element and isotopic composition and abundance, and melt productivity together with a robust evaluation of the uncertainty in these system properties. We have applied this new methodology to constrain melting beneath the Reykjanes Peninsula (RP) of Iceland [2] and here extend the approach to Iceland’s Northern Volcanic Zone (NVZ). We consider melting of a heterogeneous mantle source involving depleted peridotite and pyroxenite lithologies, e.g., KG1, MIX1G and G2 pyroxenites. Best-fit model sources for Iceland basalts contain more than 90% depleted peridotite and less than 10% pyroxenite with Tp ~125-200 °C above ambient mantle. The trace element and Pb and Nd isotope composition of the depleted source beneath the Reykjanes Peninsula is similar to DMM [3], whereas depleted mantle for the NVZ is isotopically distinct and more trace element enriched. Conversely, inverted pyroxenite trace element compositions are similar for RP and NVZ and are more enriched than previously inferred, despite marked differences in their Pb and Nd isotope composition. We use these new constraints on the Iceland source to investigate their relative importance in basalt genesis along the adjoining Reykjanes and Kolbeinsey Ridges. We find that the proportion of pyroxenite diminishes southward along Reykjanes Ridge and is seemingly absent to the north along the Kolbeinsey Ridge. Moreover, abundances of inverted RP and NVZ depleted mantle also diminish away from Iceland and give way to a common depleted source for the North Atlantic. These findings further illuminate the along-strike variability in source composition along the North Atlantic ridge system influenced by the Iceland melting anomaly, while reconciling geochemical, geophysical and petrologic constraints required to rigorously test plume vs. non-plume models. [1] Brown & Lesher (2016); G^3, v. 17, p. 3929-2968 [2] Brown et al. (2020); EPSL, v. 532, 116007 [3] Workman and Hart (2005); EPSL, v.231, p. 53-72

The ophiolitic peridotite and gabbro of Moncuni (Southern Lanzo Massif, Western Alps) retain pre-subduction mantle-to-oceanic, high-temperature (>700C∘) ductile fabrics. These fabrics are overprinted by seismic fracturing and faulting associated with pseudotachylytes. Within the gabbro, the pseudotachylytes preserve dry glass and pristine microlites. The occurrence of rare, minute garnet and the static development of eclogite-facies assemblages in local hydrated domains indicate that pseudotachylytes experienced subduction conditions of 600C∘ and 2.1 GPa. The exceptional survival of glass and the absence of post-oceanic ductile deformation demonstrate prevailing dry conditions during the entire Alpine subduction and exhumation path. Dry conditions inhibited reaction kinetics and viscous flow. In contrast, the majority of the Alpine ophiolites, derived from the upper hydrated portions of the oceanic lithosphere, show pervasive fluid-assisted metamorphism and ductile deformation. The Moncuni body can, therefore, be regarded as representative for the rheological behaviour during subduction of seismic, dry, deeper oceanic lithosphere that is rarely exhumed to the Earth's surface. In Moncuni, the brittle-ductile transition of dry oceanic rocks is constrained to be between 600 and 750C∘. This temperature range corresponds to the observed cut-off of intermediate-depth seismicity within subducting slabs. We infer that the base of the seismic layer corresponds to the brittle-ductile transition of a dry slab rather than the locus of antigorite breakdown triggering earthquakes by dehydration embrittlement.

High-MgO (>12 wt%) magmas represent some of the most primary and high temperature melts from mantle plumes. The compositional diversity of high-MgO magmas gained by fractional melting within inhomogeneous mantle sources, is often overprinted by wall rock assimilation, magma mixing and fractional crystallisation within magma chambers at various depths within the crust. The deepest intrusions, at the base of the crust, are the first stop for magmas as they leave the mantle and such localities offer vital insights into the crustal processes that modify high-MgO melts. The Selland Igneous Province (SIP) represents a rare exposure of a deep crustal magma conduit system that transported large volumes of mantle-derived melts through the lower continental crust. In this work, trace element compositions of clinopyroxenes in dunite, wehrlite and olivine dinopyroxenite samples from the Reinfjord intrusion were measured using laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). The melts calculated to be in equilibrium with the clinopyroxenes represent derivatives of mantle-derived magmas and have steep rare earth element (REE) profiles with La/Yb-PM of 8.4-14., Sm/Yb-PM of 3.5-4.8, and negative anomalies in Nd, Zr and Hf. Assimilation, recharge and fractional crystallisation modelling shows that the ultramafic cumulates in the Reinfjord intrusion formed through concurrent fractional crystallisation and repetitive recharge (and mixing) of new primitive magmas from the mantle with very limited input from crustal sources. The recharge and mixing model is strongly supported by field and petrographic data. Two end-member ultramafic magmas are identified; the first end-member melt composition (MELTCPXO) is similar to the melt compositions that are in equilibrium with the most trace element depleted clinopyroxenes from Reinfjord. The second end-member melt composition (MELTPICO) is similar to a number of picrite dykes found throughout the SIP, which have less steep LREE/HREE and lack strong negative anomalies in Zr and Hf. The REEBOX PRO melting model (Brown and Lesher, 2016) was used to forward model adiabatic decompression melting of lithologically homogeneous and heterogeneous sources containing anhydrous/hydrous peridotite +/- pyroxenite. It was found that MELTPICO formed by melting of a peridotite source with up to 10% pyroxenite component and mantle potential temperatures of up to 1450 degrees C. MELTCPXO. however, formed by melting of a carbonate- metasomatised peridotite with potential temperatures of up to 1630 degrees C. These results indicate a compositionally and thermally inhomogeneous mantle plume beneath the SIP. (C) 2020 The Authors. Published by Elsevier B.V.

Mantle melting dynamics of the Central Atlantic Magmatic Province (CAMP) is constrained from new platinum group element (PGE), gold (Au), rare earth element (REE), and high field strength element (HFSE) data and geochemical modelling of flood basalts in Morocco. The PGE are enriched similarly to flood basalts of other large igneous provinces. The magmas did not experience sulphide saturation during fractionation and were therefore fertile. The CAMP is thus prospective for PGE and gold mineralization. The Pt/Pd ratio of the Moroccan lavas indicates that they originated by partial melting of the asthenospheric mantle, not the subcontinental lithospheric mantle. Mantle melting modelling of PGE, REE and HFSE suggests the following: (1) the mantle source for all the lavas was dominated by primitive mantle and invariably included a small proportion of recycled continental crust (<8%); (2) the mantle potential temperature was moderately elevated (c. 1430 degrees C) relative to ambient mantle; (3) intra-lava unit compositional variations are probably a combined result of variable amounts of crust in the mantle source (heterogeneous source) and fractional crystallization; (4) mantle melting initially took place at depths between c. 110 and c. 55 km and became shallower with time (c. 110 to c. 32 km depth); (5) the melting region appears to have changed from triangular to columnar with time. These results are best explained by melting of asthenospheric mantle that was mixed with continental sediments during the assembly of Pangaea, then heated and further mixed by convection while insulated under the Pangaea supercontinent, and subsequently melted in multiple continental rift systems associated with the breakup of Pangaea. Most probably the CAMP volcanism was triggered by the arrival of a mantle plume, although plume material apparently was not contributing directly (chemically) to the magmas in Morocco, nor to many other areas of CAMP.

The origin of Large Igneous Provinces (LIPs) associated with continental breakup and the reconstruction of continents older than ca. 320 million years (pre‐Pangea) are contentious research problems. Here we study the petrology of a 615–590 Ma dolerite dyke complex that intruded rift basins of the magma‐rich margin of Baltica and now is exposed in the Scandinavian Caledonides. These dykes are part of the Central Iapetus Magmatic Province (CIMP), a LIP emplaced in Baltica and Laurentia during opening of the Caledonian Wilson Cycle. The >1,000‐km‐long dyke complex displays lateral geochemical zonation from enriched to depleted basaltic compositions from south to north. Geochemical modelling of major and trace elements shows these compositions are best explained by melting hot mantle 75–250 °C above ambient mantle. Although the trace element modelling solutions are nonunique, the best explanation involves melting a laterally zoned mantle plume with enriched and depleted peridotite lithologies, similar to present‐day Iceland and to the North Atlantic Igneous Province. The origin of CIMP appears to have involved several mantle plumes. This is best explained if rifting and breakup magmatism coincided with plume generation zones at the margins of a Large Low Shear‐wave Velocity Province (LLSVP) at the core mantle boundary. If the LLSVPs are quasi‐stationary back in time as suggested in recent geodynamic models, the CIMP provides a guide for reconstructing the paleogeography of Baltica and Laurentia 615 million years ago to the LLSVP now positioned under the Pacific Ocean. Our results provide a stimulus for using LIPs as piercing points for plate reconstructions.

The compositions and volumes of basalt generated by partial melting of the Earth's mantle provide fundamental constraints on the thermo-chemical conditions of the upper mantle. However, using melting products to interpret uniquely these conditions is challenging given the complexity of the melting and melt aggregation processes. Forward models simulating melting of lithologically heterogeneous mantle sources can account for this complexity, but require assumptions about key model input parameters, and the quality of the model fits to the observations are rarely, if at all, considered. Alternatively, inverse melting models can provide estimates of the quality of model fits to the observations, but as of yet, do not account for the presence of lithologic heterogeneity in the mantle source. To overcome these limitations, we present an inverse method coupling a Markov chain Monte Carlo (MCMC) sampling method with the REEBOX PRO forward mantle melting model. We use this tool to constrain mantle potential temperature, melt volumes, and the trace element and isotopic compositions of mantle source lithologies beneath the Reykjanes Peninsula of Iceland. We consider a range of plausible pyroxenite compositions (KG1, G2, and MIX1G) that span much of the range of natural pyroxenite compositions, and constrain the mantle potential temperature between 1455 and 1480°C and pyroxenite abundance between 6.5 and 8.5%. These results are independent of the choice of pyroxenite composition and indicate that elevated potential temperatures and modest pyroxenite abundances are robust features of the Reykjanes Peninsula mantle source. The permitted ranges of pyroxenite trace element compositions vary as a function of pyroxenite fertility and mineralogy, but differ from the compositions of subduction-modified recycled oceanic crust typically used in previous models, indicating a more complex petrogenetic origin for the pyroxenite source than previously considered. As all of the pyroxenites employed here yield equally good fits to the geochemical and geophysical observations along the Reykjanes Peninsula, forward models should not be used to constrain the major element character of pyroxenite present in mantle source regions based solely on the trace element/isotopic compositions (and volumes) of basalts. Given the range of lithologies included in REEBOX PRO and the flexibility of MCMC inversion, this method may be applied to constrain thermal and compositional source characteristics in a wide variety of basalt source regions.