Neoarchean Plume and Arc Volcanism and Their Interaction, in Allochthonous Terranes: A Synthesis
Neoarchean greenstone terranes of the Superior, Dharwar, and Yilgarn cratons record contemporaneous plume- and arc-related magmatism. The voluminous komatiite-tholeiitic basalt association represents oceanic plateaus erupted from mantle plumes. Komatiites record Nb/Th of 8-20, relative to a primitive mantle value of 8, signifying processing of ocean lithosphere through a subduction zone, and its recycling into the mantle source of the plumes; high ratios are consistent with significant growth of continental lithosphere before 2.7 Ga. Basalts record Nb/Th 4-12 where low values imply subduction-erosion of arc lithosphere into the asthenosphere. Together, the two magma series reflect radially zoned plumes. Tholeiitic, low-K, arc-related basalts are identified by the conjunction of LREE-enrichment with negative anomalies at Nb, Ta, P, Ti, often associated with a compositional spectrum of increasing LREE to MKS- and HKS-basalts, with rare arc shoshonites and leucitites. Boninites are associated with low-Ti tholeiites, likely from fore-arcs. An association, originally defined from Cenozoic intraoceanic arcs, of tholeiitic LKS- to MKS- and HKS-basalts, with adakites, Mg-andesites, and Nb-enriched basalts, has now been documented from many greenstone terranes. The implication is for flat subduction of hot ocean lithosphere <20 Ma old, with slab dehydration- wedge melting evolving to slab melting. If this association is plotted in Th/Yb vs. Nb/Yb co-ordinate space, the trend is close to the crustal contamination vector but does not have that significance. Given that arcs migrate, and komatiites are imbricated with arc sequences, greenstone terranes cannot be autochthonous. Cratonization involves coupling of refractory continental mantle lithosphere (CLM) to crust. CLM is dominated by the refractory residue of melting in plumes, whereas crust is imbricated plateau and arc crust.
Characteristics of Plate Tectonics with a Warmer Mantle
The mantle was probably warmer in the past than it is today. Continental plates underlain by such a warmer mantle would have experienced less subsidence than modern ones following extension, because extension would have led to widespread melting of the underlying mantle and the generation of large volumes of mafic rock. A 200° increase in mantle temperature leads to the production of more than 12 km of melt beneath a continental plate extended by a factor of 2, and the resulting thinned plate rides with its upper surface little below sea level. The thick, submarine, mafic-to-ultramafic volcanic successions on continental crust that characterize many Archean regions could therefore have resulted from extension of continental plates above warm mantle. A warmer mantle in the past would have led to greater degrees of melting, and melting to greater depths in the mantle rising at oceanic spreading centres. Both the oceanic crust and the underlying layer of depleted mantle would therefore have been thicker than at present. This is widely thought to have led to more buoyant oceanic lithosphere, which might have prevented subduction, or at least have resulted in widespread shallow subduction. Although greater melting in a column of rising asthenosphere does indeed reduce the density of oceanic plates to less than that of undepleted and undifferentiated asthenosphere, the negative buoyancy drive for subduction still remains, because cooled differentiated asthenosphere is replaced at the surface by uncooled but still differentiated asthenosphere. Once subduction is established, the dynamics of subduction zones are governed largely by slab-pull forces. Such forces would have been significantly lower with a warmer mantle, because the oceanic lithosphere would have cooled less than the modern one before attaining thermal equilibrium. For the same reason however, the flexural rigidity of the oceanic lithosphere would have been much lower. It is not clear, therefore, that shallow subduction would have been more prevalent with a warmer mantle. Calculated temperatures at the base of the thick crust beneath a warmer mantle in the past are well above the granitoid solidus for oceanic lithosphere of young ages. Oceanic crust may therefore have had a considerably more complex petrological structure than is believed to be the case today, with complex interactions between granitoid melts and earlier-crystallized mafic rocks.
High-Mg Magmatism Through Time: Implications for the Thermal and Compositional Structure of the Earth's Mantle
There was an abrupt change in the maximum MgO content of mantle-derived magmas at the end of the Archean. The hottest magmas identified in the Archean are high-Mg komatiites with up to 32-34% MgO that erupted at ~1610-1630C (e.g., Abitibi, Barberton, Belingwe, Norseman-Wiluna), those in the Proterozoic are low-Mg komatiites with up to 22% MgO that erupted at ~1480C (e.g., Thompson), and those in the Phanerozoic are low-Mg komatiites/high-Mg picrites with 19-23% MgO that erupted at ~1430-1500C (e.g., Baffin, Gorgona). It has been suggested that some if not all high-Mg komatiites were produced by hydrous melting. Although some komatiites contain vesicles and/or igneous amphibole, they occur only locally, even within individual volcanic-stratigraphic units, which suggests that water was derived locally during emplacement. Critically, most komatiites are moderately-strongly depleted in all highly-incompatible lithophile elements (HILE), indicating that melting was not facilitated by addition of water and that the komatiite source may have contained as little as 20 ppm H2O, in agreement with the oxidation state of Fe in komatiites (Berry et al. 2008 Nature). Thus, the 9% MgO change in the composition of mantle-derived magmas corresponds to a 115C change in eruption temperature and a 150C change in the potential temperature of the sampled mantle source region. Most estimates of upper mantle temperatures have used the compositions of basalts in Archean greenstone belts, which likely represent the heads of plumes rather than equivalents of MORB, but the temperatures do not appear to have been significantly hotter. Thus, the higher plume temperatures, along with the plethora of other fundamental changes in geological processes that occurred at the end of the Archean, suggests that there was a fundamental and irreversible change in the thermal and/or compositional structure of the mantle at that time from one that could produce hotter plumes but that involved only slightly hotter upper mantle. The three most widely discussed scenarios are: 1) generation of a thermal/compositional boundary layer in the lowermost mantle, 2) a change from whole-mantle to 2-layered convection, and 3) a change from 2-layered to whole-mantle convection. Having no (or fewer) thermal boundary layer(s) at the core-mantle boundary in the Archean would allow hotter mantle to accumulate along the boundary and might also explain the PGE depletion in early Archean komatiites if there was any partitioning of PGE into the core, but Nd isotopic data suggest that D" formed much earlier, komatiites appear to be sourced from depleted not enriched mantle, and the change in PGE abundances appears to have occurred between 3.2 and 2.7 Ga, not at 2.5-2.6 Ga. Deriving plumes primarily from the core-mantle boundary in the Archean and primarily from the 670 km boundary layer after the Archean might account for the change in maximum MgO content, but isotopic data for noble gases and HILE do not permit whole-mantle depletion and seismic tomography studies suggest that the present 670 km phase transition does not impede mass transfer. Sourcing komatiites from an intermediate boundary layer (e.g., majorite-rich cumulates left over from the magma ocean disrupted by the major 2.7 Ga global komatiite event) would also explain the greater abundance of Al-Sc-HREE- and PGE-depleted komatiites in the early Archean if that layer retained the signature of the very early core-mantle differentiation event. In order to better evaluate these possibilities we need more information about the potential temperature of Archean upper mantle and the abundances of PGE in a wider range of early Archean komatiites.
Orthopyroxene chemistry preserves evidence for hydrous komatiite melts at Commondale, South Africa
The Commondale Ultramafic Suite of South Africa contains a unique assemblage of komatiites. First described by Wilson (2003) and Wilson et al. (2003), the komatiite units contain spinifex zones of orthopyroxene that overlay cumulate zones of olivine. The units also preserve clearly defined chill margins between komatiite units, indicating cooling unit thicknesses that are greater than 1 m thick. Although the majority of the samples found on the surface show variable extents of metamorphism to a talc-tremolite dominated assemblage, there are still decimeter scale regions of the samples that preserve greater than 30% of the original igneous minerals, including most of the coarse orthopyroxene spinifex. The preserved orthopyroxene spinifex grains exhibit primitive high-Mg# cores with normal zoning of the Mg# to the perimeters. Anhydrous experiments at 1-atm were performed to investigate crystallization under equilibrium and cooling conditions on a representative Commondale komatiite liquid (31.9 wt% MgO). The initial pyroxene that crystallizes in equilibrium conditions is a protoenstatite. The chemistry of this protoenstatite is very distinct (0.005 wt% CaO, and Mg# 0.95), and different than the natural orthopyroxenes (Mg# 0.92 and 0.75 wt% CaO). When the natural pyroxenes are compared to the pyroxenes in cooling rate experiments of 100°C/hr and 10°C/hr, the minor element variations, Ca, Cr, and Al are qualitatively duplicated, however the Mg# of the pyroxenes is too high to match the natural pyroxenes. The chemical effects observed in these undercooling experiments is also expected for more moderate undercooling of the komatiite melt, and rules out the possibility that the pyroxene spinifex was formed by rapid quenching. Because the anhydrous experiments fail to reproduce the observed orthopyroxene chemistry, hydrous experiments were performed to investigate the affect of H2O on the komatiite phase equilibria. The addition of H2O serves to suppress the appearance of the initial pyroxene by a greater amount than it suppresses the olivine liquidus, allowing the liquid to evolve to a lower Mg# before pyroxene begins to crystallize. This serves to lower the Mg# of the initial orthopyroxene, bringing it closer to the observed natural pyroxenes. By combining the effects of high-H2O contents, ∼4 wt%, and the effects of a small undercooling, the natural orthopyroxene compositions can be successfully modeled. This indicates emplacement of hydrous komatiite melt in the ultramafic suite of Commondale, South Africa similar to the hydrous komatiite melts of the Barberton Mountainland, South Africa.
The Transition From Archaean TTGs and Enderbites to Proterozoic Anorthosites; From Cannibal Basaltic Plateaux to Plate Tectonics
The applicability of plate tectonics to pre-Phanerozoic time is controversial. Archean cratons are dominated by tonalite-trondhjemite-granodiorite (TTG) and greenstone belt terranes which form dome-and-keel structures with no modern equivalents. The chemical signatures of TTGs indicate ultimate formation by anatexis of metabasites in the presence of rutile and garnet. However, although commonly identified as root-zones of continental arcs, the dominant older TT components are not calc-alkaline, are not distributed as linear belts, and the observed volumes could not have formed by slab melting in the timespan available. Nd-isotopic signatures and zircon xenocrysts indicate multiple remelting events after their formation. NE Superior TTGs contain abundant enderbite (pyroxene-tonalites, ca 25%) with liquidus temperatures near 1000oC, too hot for arc slab melts, but consistent with a major basaltic underplating event that would have remelted most pre- existing TTGs at ca 2.73Ga, preserving fossil geochemical signatures. Calc-alkaline lavas in greenstone belts tend to be sandwiched between tholeiitic flood basalt sequences, are subordinate, and are best interpreted as localized remelts of the basalts; not as coeval plume-arc products. In contrast, many Proterozoic (Ptz) sequences show well-developed passive margin sequences, and linear orogenic (deformation and metamorphism) belts similar in many ways to Phanerozoic ones. The principal difference is the superabundance of anorthosite-mangerite-charnockite-gabbro-granite massifs (AMCG), which are restricted to Ptz time. AMCGs locally dominate Ptz crust and form massifs up to 300km in diameter. Although frequently interpreted to be anorogenic, recent progress in understanding Ptz geology implies they are post-orogenic. Trace element inversion models suggest that the anorthosite cumulates formed from remelts of thickened arc crust with residual garnet, while the gabbroic facies formed from melts having more of a mantle component. The trace element signatures of the mantle melts suggest these were tholeiitic basalts extracted from a source affected by slab-derived melts and fluids. The absence of AMCGs from Archaean terranes suggests that pre- Ptz crust was too soft to be thickened tectonically. This is in accord with the appearance of major crustal dyke swarms in the Ptz. Their absence from post-Ptz time could be due to secular cooling of the Earth.
Petrogenesis of Sanukitoids : Example of the Bulai Pluton, Central Limpopo Belt, South Africa
The word sanukitoid refers to Archaean high-Mg dioritic, granodioritic and monzogranitic plutons. These magmatic rocks display geochemical features of both Archaean TTG (fractionated REE patterns, strong HFSE- depletion) and modern BADR series (calc-alkaline differenciation trend, high LILE contents). Their transitional characteristics are not only compositional, but also temporal; indeed sanukitoids were reported in almost all Late Archaean terranes, where they emplaced at Archaean-Proterozoic boundary (2.5 Ga). Consequently, sanukitoids are considered as the result of the petrogenetic and geodynamic changes that took place at the end of the Archaean. In NE South Africa, the 2.6 Ga old Bulai pluton is a calc-alkaline granodioritic pluton intruded into the granulitic Limpopo Belt. It shares several petrographic characteristics with sanukitoids : it is a hornblende- and biotite- bearing porphyritic granodiorite, associated with monzodioritic enclaves and both mafic and granitic dykes. Whole-rock major- and trace-element analysis show high LILE contents (K2O > 3 wt. % ; Rb 60-200 ppm ; Ba 600-2600 ppm) and fractionated REE patterns (25 < (La/Yb)N < 80; with La > 80 ppm). However it slightly differs from typical sanukitoids by lower Ni and Cr contents, Mg# and higher SiO2 contents and K2O/Na2O. In Harker's plots, data draw linear trends for both major and trace elements, over a wide range of compositions, from 47 up to 74 wt. % SiO2. In case of fractional crystallization or of partial melting, this would imply that the composition of the cumulate or of the residue remained constant throughout the whole process, which is unlikely over a so large range of SiO2. Consequently, such long linear is better explained by mixing processes. The calculated poles of the mixing are : i) an acid one represented by a granitic dyke (SiO2 = 74 wt.%) and ii) a mafic one, whose composition is close to that of monzodioritic enclaves (SiO2 = 47 wt.%). The SiO2-rich pole is assumed as generated by melting of the surrounding TTGs. On the other hand, the mafic pole would represent the primitive sanukitoid magma : monzodioritic enclaves and basic dykes are the richer, not only in Fe, Mg, V, Ni and Cr but also in both REE and incompatible trace elements thus precluding a cumulative origin. Modeling shows that this pole can have been produced by partial melting of a mantle peridotitic source, which was previously enriched in LREE and LILE. In addition, the Nb/Y ratios of these magmas points to a metasomatic agent that would be a TTG-like slab melt rather than a fluid. These results are in good agreement with previous studies focusing on the petrogenesis of sanukitoids which identified a peridotitic source metasomatized by slab-derived melts. They also enhance the role of mixing with crust-derived magmas, which seems to be a key process in the genesis of many sanukitoid suites. We interpret the genesis of the Bulai sanukitoids in terms of decreasing TTG production in course of time. This interpretation is based on the effective melt/rock ratio of Rapp et al. (1999): during Archaean times, due to high geothermal gradients, great volumes of TTG magmas were produced such they were not all consumed in reaction with mantle peridotite. At the Archaean-Proterozoic boundary, due to cooling of the Earth, TTG magmas were produced in smaller amounts and totally consumed by reaction with peridotite, which subsequent melting gave rise to sanukitoids.
New Geodynamic Inferences from the Trace element Systematics of Eo- to Neoarchean Volcanic Rocks, SW Greenland
Archean greenstone belts in SW Greenland contain mafic to ultramafic volcanic rocks ranging in age from 3800 to 2800 Ma. Volcanic rocks in the Eoarchean (3800-3700 Ma) Isua greenstone belt are composed of boninites, picrites, and tholeiitic pillow basalts. The Mesoarchean (3075 Ma) Ivisaartoq belt contains tholeiitic pillow basalts, and minor picrites and boninites. Greenstones associated with the Neoarchean (2900-2800 Ma) Fiskenaesset Anorthosite Complex are characterized predominantly by tholeiitic basalts. On primitive mantle- normalized diagrams, these volcanic rocks display variably depleted to enriched LREE patterns, near-flat HREE (except boninites), and pronounced negative Nb anomalies. On the Nb/Yb versus Th/Yb and projection, SW Greenland volcanic rocks, plot mainly within the field of tholeiitic intra-oceanic subduction-derived magmas, as for Phanerozoic ophiolites. Greenland Archean volcanic rocks are characterized by variably positive initial epsilon-Nd values (+0.5 to +5), consistent with long-term depleted mantle sources. There is no field evidence that these Archean greenstone belts were deposited on older continental crust, suggesting that the negative Nb anomalies and variable initial Nd isotopic compositions reflect the enrichment of their mantle sources by subduction-derived fluids and/or melts. On recently developed new petrogenetic discrimination diagrams, based on the log-transformed ratios of immobile elements (e.g., La/Th, Sm/Th, Yb/Th , Nb/Th), Eo- to Neoarchean volcanic rocks display a trend projecting from mid-ocean ridge basalt (MORB) to island arc basalt (IAB) field. This trend is interpreted as reflecting the entrainment by induced convection of Archean depleted upper mantle (i.e., the source of Archean MORB) into a subarc mantle wedge following the initiation of intra-oceanic subduction and arc migration.
Pre-Neoarchean Rocks Differ Greatly From Modern Rocks Because They Record Different Tectonic Processes
Archean through Mesoproterozoic rocks and assemblages differ strikingly (individually, collectively, geologically, and geophysically) from the island arcs and other modern plate-tectonic features with which analogies are conventionally forced. Archean crust consists largely of tonalite, trondhjemite, and granodiorite (TTG). These are mostly richer in Si, Na, and K and poorer in Mg and Ca, and have steeper REE patterns, than modern arc rocks with the same names, and are vastly more felsic in bulk than island arcs. TTG are the only basement rocks seen beneath Archean supracrustals, sections of which commonly begin with basement- derived clastic strata, continue with mafic and ultramafic lavas (quite different from both modern arc and seafloor rocks) and then with sharply bimodal mafic-and-felsic volcanic sections and quartzose clastic strata. (Island arcs lack such strata, and have unimodal igneous rocks.) Widespread olivine and clinopyroxene komatiites and mafic ferroan andesite have no modern equivalents; modern MORB and OIB lack Archean twins. Stratal sections are subregional, not in narrow belts, where mapping and dating are good and subsequent disruption minimal. Submarine plains, followed by hills and basins consequent on granite doming, are indicated. Severe lateral deformation of upper crust that effectively floated on mobile deeper TTG, unlike arcs, commonly accompanied doming. Sutures, accretionary wedges, ophiolites, and ensimatic rocks are unknown. Geologic, petrologic, geochemical, and isotopic data from Archean crustal and mantle assemblages accord with fractionation, ca. 4.5 Ga, of thick melabasaltic protocrust (geochemistry's "missing enriched reservoir," complementary to upper-mantle harzburgite) which became the primary source for tonalites. Misnamed "mantle separation ages" date, at best, derivation of TTG from protocrust with chondritic Sm/Nd. Incremental delamination and sinking of restitic garnet-rich protocrust enabled rise of harzburgite, 200-300 degrees C hotter than modern asthenosphere, producing more TTG from remaining protocrust and recycling and refining early TTG. Paleoproterozoic orogens have sedimentary and ensialic igneous (mostly mafic and ultramafic) sections that onlap far onto Archean rocks. Where zircon spot dates are abundant, reworked Archean TTG commonly is shown to be present also in tracts beneath the orogens, and to be reworked within them. Felsic members of strongly bimodal Paleoproterozoic volcanic and plutonic rocks are moderately to richly potassic, utterly unlike modern island arcs. Aluminous and quartzose sediments, rare in island arcs, are abundant. Sutures, accretionary wedges, ophiolites, and ensimatic rocks are unknown. A general origin in thickly sedimented and largely ensialic extensional basins is indicated, with heat provided for melting of basement and basin rocks by then-high U-K-Th in blanketed materials, by continuing delamination of remnant protocrust, and by mantle melts (mostly of previously sunken protocrust?) enabled by tensional depressurization. Because valid petrologic and geologic analogies with modern plate settings cannot be made, plate proponents base wished-for assignments on selected trace-element ratios. These ratios poorly discriminate modern rocks of known settings, and there is no basis for the common assumption that they somehow define ancient settings in the absence of any corroborative data. Nor is there any basis for the common conjecture that multiple imaginary plate and "plume" settings can be mixed in whatever combination might account numerically for ratios that misfit hoped-for settings.