R. N. Thompson

Subcontinental mantle plumes, hotspots and pre-existing thinspots

The magmatism occurring when a hot convective mantle plume is sited beneath a lithospheric plate may be more complex if the latter is continental, rather than oceanic, because of the characteristic local physical inhomogeneity of continents. Thus, the surface volcanic expression of the plume (hotspot) may be displaced from immediately above its rising stem, if the continent has been previously locally thinned nearby. The magmatism of the British Tertiary Igneous Province and Parana basin, South America, appears to fit this model, and it may also explain why the Miocene Columbia River basalts of the NW United States overlie an Early Tertiary sedimentary basin.

High-Ti and low-Ti mafic potassic magmas: Key to plume-lithosphere interactions and continental flood-basalt genesis

The role of mantle plumes in the genesis of continental flood-basalts (CFB) remains controversial, primarily due to our limited knowledge of the composition of the subcontinental lithospheric mantle (SCLM). In this study we use the widespread Cretaceous mafic potassic magmatic rocks, emplaced around the margins of the Paranasedimentary basin, to probe large-scale compositional variations of the SCLM beneath southern Brazil and Paraguay. On the basis of Ti contents, together with major-, trace-element and isotopic ratios, these mafic potassic rocks may be subdivided into high-Ti and low-Ti groups. The former have relatively high average TiO2 (4.64), CaO/Al2O3 (1.74) and 143Nd/144Ndi (0.51232), together with low La/Nb (1.1) and 87Sr/86Sri (0.7050). The latter are characterised by much lower average TiO2 (1.77), CaO/Al2O3 (0.72) and 143Nd/144Ndi (0.51182), together with higher La/Nb (2.01) and 87Sr86Sri (0.7068). These high-Ti and low-Ti groups are spatially separate and their distribution correlates with tectonic setting; the low-Ti magmas are associated with cratonic regions, whereas the high-Ti magmas are in Proterozoic mobile belts. The distribution of the subgroups of mafic potassic magmatic rocks correlates closely with the geochemical provinciality of the Early Cretaceous high-Ti and low-Ti Paranaflood-basalts. This is the first reported occurrence of extensive low-Ti mafic potassic magmatism associated both spatially and temporally with the low-Ti region of a major CFB province. Our study further reveals similar relationships between tectonic setting and the geochemical provinciality of mafic potassic magmas and continental flood-basalts across Gondwana. We use the bulk-rock compositions and radiogenic isotopic ratios of both the high-Ti and low-Ti mafic potassic magmatic rocks as end members in models of CFB genesis. Mixing calculations involving Sr and Nd isotopic ratios indicate that the flood-basalts may contain up to 50% of mafic potassic lithosphere-derived melts. Overall, the results of our geochemical modelling agree with geophysical arguments that the convecting asthenosphere is the predominant source of CFB magmas.

Ferropicrites: geochemical evidence for Fe-rich streaks in upwelling mantle plumes

A comparison of high-MgO magmas from both oceanic and continental settings reveals that they exhibit wide variations in their bulk-rock contents of FeO* (9–16 wt% at 15 wt% MgO). The high-FeO* picrites (ferropicrites) range from Archean to recent in age but are relatively rare at the Earth’s surface, typically forming thin isolated flows near the base of thick lava piles in large igneous provinces. They are characterised by high contents of compatible trace elements (e.g. Cr=400–1650 parts per million (ppm) and Ni=250–1050 ppm) and unaltered samples (Parana–Etendeka and Madagascar) have relatively smooth, normalised multi-element patterns that lack significant relative enrichments in strongly incompatible elements (e.g. [Ba/La]n=0.5–1.0) and [La/Nb]n=1.2–1.4). The ferropicrites are distinguished from other picritic rocks (e.g. Deccan, Hawaii, West Greenland) by their relatively low abundances of Al2O3 (∼10 wt%) and heavy rare-earth elements (HREEs, e.g. Lu=<10×chondrite). They have 87Sr/86Sri ratios of ∼0.7048 and ϵNd values of ∼+4 that are comparable to those of ocean-island basalts. Modelling calculations of combined Sr, Nd and Pb isotopic ratios indicate that some of the ferropicrites may have assimilated upper or lower crust but this appears to have had little effect on major element abundances. The high-FeO* contents of world-wide ferropicrites, relative to ‘normal’ picrites, cannot simply be attributed to variations in degrees of partial melting and/or depth of melt segregation of an anhydrous lherzolite mantle source. Quantitative partition modelling suggests that the contributing parental melts of the ferropicrites were derived by adiabatic decompression melting of a mantle source that was similar in composition to experimentally studied Fe-rich peridotite PHN1611. The parental melts of the Parana–Etendeka ferropicrites appear to have been generated by ∼10% partial melting, at high pressures (45–35 kbar) and high mantle potential temperatures (Tp=1550°C). The relatively low volume of world-wide ferropicrites and their association with igneous rocks of ‘normal’ FeO* contents in mantle plume-related igneous provinces suggest that the former may be derived from Fe-rich streaks in mantle plume starting-heads.

Crustal assimilation during turbulent magma ascent (ATA); new isotopic evidence from the Mull Tertiary lava succession, N. W. Scotland

Assimilation of crustal rocks with concomitant fractional crystallisation (AFC) is a well documented phenomenon in many igneous suites, but geochemical evidence from the Tertiary Mull lava succession suggests that in these magmas crustal contamination occurred by a distinctly different mechanism. Lavas from the lower half of the Mull Plateau group (MPG) can be divided into two broad sub-types; high (>8%) MgO basalts with elevated Ba and K; and lower MgO (<8%) basaltic-hawaiites with lower Ba and K. The lower crust and most of the upper crust beneath Mull is probably of Lewisian age. The Sr-, Nd-and Pb-isotope compositions of local Lewisian crustal samples yield the following ranges: 87Sr/86Sr=0.71002–0.72348, 143Nd/144Nd=0.51045–0.51058 and 206Pb/204Pb=14.0–14.6. Ten lavas have also been analysed and yield the following ranges: 87Sr/86Sr=0.7028–0.7042, 143Nd/144Nd=0.51214–0.51230 and 206Pb/204Pb=15.1–17.9. However, within this range, it is predominantly the more primitive mafic compositions, with elevated Mg, Ba and K, that show the lowest Nd- and Pb-, and the highest Sr-isotope values. Modelling of these isotopic results, in conjunction with major and trace element data, show that: (1) contamination by Lewisian lower crustal material does occur; (2) that the process involved was not one of assimilation with concomitant fractional crystallisation (AFC). The proposed contamination process is one whereby the hottest (most MgO rich) magmas have assimilated acidic partial melts of Lewisian lower crust during turbulent ascent (ATA) through thin, poorly connected dyke- and sill-like magma chambers. The chemical composition of the contaminated lavas can be modelled successfully through addition of ∼5% acidic Lewisian crust to an uncontaminated lava. In contrast, the more evolved magmas — which probably fractionated at sub-crustal levels — were either not hot enough to molt significant amounts of crust, or did not ascend turbulently because of their higher viscosity, and so are less contaminated with crust.

The heterogeneous Iceland plume: new insights from the alkaline basalts of the Snaefell volcanic centre

Abstract It has been accepted for some time that mixing between MORB-source mantle and the Iceland plume (geochemically relatively ‘reriched’ mantle) occurs along the Reykjanes Ridge. The composition of the basalts becomes progressively more enriched in lithophile elements towards Iceland. Within Iceland itself, there is significant variation in the composition of the mafic volcanic rocks. Although variations in degree of melting of a single source could have produced the elemental compositions of the Icelandic picritic, tholeiitic and alkalic basalts, their isotope systematics are not consistent with this model and require differences in source chemistry. Therefore, either MORB-source mantle has been mixed into the plume or the plume itself is heterogeneous. The clearest indications come from Pb isotopic data, which suggest that MORB-source mantle was excluded from the generation of Icelandic volcanic rocks.

Erratum to “High-Ti and low-Ti mafic potassic magmas: Key to plume—lithosphere interactions and continental flood-basalt genesis” [Earth Planet. Sci. Lett. 136 (1995) 149–165]

Abstract The role of mantle plumes in the genesis of continental flood-basalts (CFB) remains controversial, primarily due to our limited knowledge of the composition of the subcontinental lithospheric mantle (SCLM). In this study we use the widespread Cretaceous mafic potassic magmatic rocks, emplaced around the margins of the Parana sedimentary basin, to probe large-scale compositional variations of the SCLM beneath southern Brazil and Paraguay. On the basis of Ti contents, together with major-, trace-element and isotopic ratios, these mafic potassic rocks may be subdivided into high-Ti and low-Ti groups. The former have relatively high average TiO2 (4.64), CaO Al 2 O 3 (1.74) and 143 Nd 144 Nd i (0.51232), together with low La Nb (1.1) and 87 Sr 86 Sr i (0.7050). The latter are characterised by much lower average TiO2 (1.77), CaO Al 2 O 3 (0.72) and 143 Nd 144 Nd i (0.51182), together with higher La Nb (2.01) and 87 Sr 86 Sr i (0.7068). These high-Ti and low-Ti groups are spatially separate and their distribution correlates with tectonic setting; the low-Ti magmas are associated with cratonic regions, whereas the high-Ti magmas are in Proterozoic mobile belts. The distribution of the subgroups of mafic potassic magmatic rocks correlates closely with the geochemical provinciality of the Early Cretaceous high-Ti and low-Ti Parana flood-basalts. This is the first reported occurrence of extensive low-Ti mafic potassic magmatism associated both spatially and temporally with the low-Ti region of a major CFB province. Our study further reveals similar relationships between tectonic setting and the geochemical provinciality of mafic potassic magmas and continental flood-basalts across Gondwana. We use the bulk-rock compositions and radiogenic isotopic ratios of both the high-Ti and low-Ti mafic potassic magmatic rocks as end members in models of CFB genesis. Mixing calculations involving Sr and Nd isotopic ratios indicate that the flood-basalts may contain up to 50% of mafic potassic lithosphere-derived melts. Overall, the results of our geochemical modelling agree with geophysical arguments that the convecting asthenosphere is the predominant source of CFB magmas.

Quaternary volcanism in northwestern Colorado: Implications for the roles of asthenosphere and lithosphere in the genesis of continental basalts

Abstract Quaternary volcanic rocks were erupted at four locations in NW Colorado; Dotsero (4150 y.B.P.), Willow Peak (undated), McCoy (0.64 m.y. B.P.), and Triangle Peak (1.98-1.87 m.y.B.P.). At Triangle Peak, there are at least eleven lava flows, but eruptions at the other locations were monogenetic. Dotsero was the only hydrovolcanic eruption. The volcanic rocks are alkali basalts, containing the phenocryst assemblage: olivine, Fe-Ti oxide, ± clinopyroxene, ± plagioclase. The basalts are chemically similar to OIB, as would be expected from their intraplate crustal setting. Nevertheless, they have La/Ta, K/Ta, Ba/Ta and K/La ratios which are significantly higher than those of oceanic OIB. These differences cannot be explained by contamination during uprise of OIB-like basalt by continental crust of reasonable composition. It is, therefore, logical to assume that the Quaternary magmas contained a component of partial melt of subcontinental lithospheric mantle. This conclusion is in accord with the low 143 Nd/ 144 Nd ratios of the basalts. The geochemistry of the Quaternary basalts can be explained by mixing between three separate mafic magma end-member groups that were erupted in the same area during the Miocene. Group 1 magmas were OIB, representing partial melts of OIB-source asthenosphere. Group 2 magmas were minettes, with low 143 Nd/ 144 Nd ratios, regarded as partial melts of sub-Colorado lithospheric mantle. Group 3 magmas had high La/Ta ratios, and, generally, low LIL/HFS ratios. During the Miocene, the latter group of magmas are interpreted to have been derived by partial melting of asthenosphere that had been modified by subduction of oceanic lithosphere below the North America plate. The presence of a component of Group 3 magma in the Quaternary basalts indicates that the mantle source of this group was trapped for 8 m.y. in an uppermost asthenospheric layer which experienced very sluggish flow. We propose that this layer is equivalent to the thermal boundary layer — situated between the rigid part of the lithospheric mantle (above), and the convecting asthenosphere (below) — originally identified by calculations of the thermal histories of lithospheric plates.

Asthenosphere-derived magmatism in the Rio Grande rift, western USA: implications for continental break-up

Abstract Magmas that are generated at continental rift zones provide an insight into the processes operating during the early stages of continental break-up. Our detailed study of mafic volcanism along the axis of the Rio Grande rift shows that, throughout both phases (30–17 and < 13 Ma) of its evolution, magmas with compositions interpreted as melts from the asthenospheric mantle have reached the surface. This recognition of early phase (26 Ma) magmas with incompatible trace element concentrations and radiogenic isotope ratios resembling those normally associated with ocean-island basalts and small seamounts (OIB) is significant because: (1) magmas dominated by the composition of asthenosphere-derived melts are not usually thought to be characteristic of early-phase continental rifting; (2) Tertiary mafic magmatism of an age greater than late Miocene in Colorado and New Mexico was hitherto regarded as subduction-related. Previous studies have shown that the final erupted composition of asthenosphere-derived melts is determined by the potential temperature of the convecting mantle, the amount and rate of lithosphere extension, fractional crystallization and crustal contamination. However, in the Rio Grande rift and elsewhere, such as the Basin and Range province, Eifel, NW Sardinia and the Cameroon Line, the final composition of these melts is also significantly influenced by earlier magmatic episodes. During the initial stages of asthenosphere melt generation the magma batches that first penetrate may heat a previously undisturbed segment of lithosphere and mix with strongly potassic, low temperature melt fractions. When these segments have been subsequently temporarily purged of such fusible potassic fractions the asthenosphere-derived melts can rise unimpeded through the sub-continental lithosphere.

High-pressure fractionation in rift-related basaltic magmatism: Faeroe plateau basalts

The lower 3000 m basalt stratigraphy on the Faeroe Islands shows strong positive correlation between degree of fractionation and Zr/Y ratio. This is ascribed to the presence of garnet in the cumulus assemblage during fractionation at the base of the continental crust or in the upper lithospheric mantle at 40-50 km depth. Later basaltic magmas fractionated at low pressures, which is a typical shift in fractionation regime among continental flood basalts. Seismic velocity ( V p ) for the garnet-bearing high-pressure fractionates is in excess of 7 km/s. This is comparable to velocities observed in high- V p bodies in the basal crust of Precambrian shields. High-pressure fractional crystallization of basaltic magmas may thus be an important process in generating the chemical complexity of continental flood basalts and has implications for crustal underplating.

Diverse mantle and crustal components in lavas of the NW Cerros del Rio volcanic field, Rio Grande Rift, New Mexico

Products of Pliocene (2–4 Ma) mafic to intermediate volcanism in the northwestern Cerros del Rio, a dominantly mafic volcanic field in the Española Basin of the Rio Grande Rift (RGR), range from 49% to 63% SiO2 and exhibit diversity in silica saturation, trace-element patterns, and isotopic compositions. Tholeiites, which are largely confined to west of the Rio Grande, have trace-element abundances that resemble those of oceanic basalts, but with mild depletions in Nb and Ta, and high 87Sr/86Sr, low 143Nd/144Nd, and high δ18O compared to typical OIB. They are regarded as asthenospherically-derived magmas contaminated with continental crust. Alkali basalts and hawaiites erupted from vents east of the Rio Grande are geochemically distinct, having generally higher overall incompatible-element abundances, but with pronounced depletions in K, Rb, Nb and Ta with respect to Th and LREE. Spatially-associated benmoreites, mugearites and latites (collectively termed “evolved” lavas) have similar trace-element characteristics to the mafic mildly-alkaline compositions, but are typically not as depleted in K. Hawaiites and evolved lavas exhibit a good negative correlation of 143Nd/144Nd with SiO2, due to interaction with lower continental crust. The most silicic “evolved” lavas carry the highest proportions of crustal material, and consequently have higher K/Th than the related hawaiites. Several (mostly mafic) lavas contain abundant crustally-derived resorbed quartz xenocrysts in O-isotope disequilibrium with the host magma. The δ18O values of xenocrystic quartz range over 4‰, indicating a variety of quartz-bearing crustal contaminants beneath the Española Basin. The hawaiites, with their unusual combination of trace-element enrichments and depletions, cannot be generated by any process of fractionation or crustal contamination superposed on a common mantle source type (oceanic or arc-source). It is a regional mantle source type, inasmuch as it was also present beneath NW Colorado during the mid-late Cenozoic. We argue that the hawaiite source must have originally existed as arc-source mantle enriched in LILE, generated during Mesozoic to early Cenozoic subduction at the western margin of North America. This arc-source mantle lost K, Rb and Ba, but not Th or LREE, prior to magmagenesis. Selective element loss may have occurred during lithospheric thinning and uprise of hydrated phlogopitebearing peridotite-possibly as a thermal boundary layer between lithosphere and asthenosphere — to shallow mantle depths, with consequent conversion of phlogopite to amphibole (an inferior host for K, Rb and Ba). We suggest that this occurred during the early extensional phase of the northern RGR. Further extension was accompanied by partial melting and release of magma from this source and the underlying asthenosphere, which by the Pliocene was of oceanic type. The hawaiite source mantle is the product of a long history of subduction succeeded by lithospheric extension of the formerly overriding plate. Similar chemical signatures may have developed in the mantle beneath other regions with comparable histories.