Journal of Petrology | Volume 38 | Number 2 | Pages 231-275 | 1997
© Oxford University Press 1997
Generation and Polybaric Differentiation of East Greenland Early Tertiary Flood Basalts
Department of Geology, University of California Davis, CA 95616, USA
Received July 24, 1995; Revised typescript accepted September 13, 1996
 |
ABSTRACT |
|---|
 TOP ABSTRACT IntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
Major element, trace element, isotope, and petrographic studies of the basal volcanics in the Kangerdlugssuaq Fjord region of the East Greenland margin provide constraints on the generation and subsequent differentiation and contamination histories of magmas during the early stage of continental rifting creating the North Atlantic ocean basin. Most primitive lavas of the succession are shown to have accumulated olivine. Corrections for crystal accumulation yield estimates for erupted primitive liquids with MgO contents between 9 and 13 wt %. The occurrence of suspended Fo88 olivine phenocrysts further requires the existence of parental magmas with MgO contents up to 17 wt % and FeO contents up to 14 wt %. Quantitative modeling of fractionation indicates that Kangerdlugssuaq magmas primarily differentiated at moderate pressures (
8 kbar) to form dunitic and wehrlitic cumulates at depths of 25 km. Evolved East Greenland flood basalts from the Scoresby Sund region record a more complex fractionation history involving olivine ± clinopyroxene fractionation in the lower crust, as found for the Lower Basalts, followed by olivine + plagioclase (± clinopyroxene) fractionation at shallow crustal levels. Correlations between SiO2, incompatible elements, and Sr–Nd isotopic ratios for the Lower Basalts are consistent with bulk assimilation of up to 15–20% Archean leucogneiss with unradiogenic Sr and Nd isotopic ratios characteristic of Lewisian-type lower crust. Comparisons between inferred magma types for Lower Basalts and estimated primary liquids from elsewhere in the North Atlantic show complex relationships among FeO, MgO, TiO2 contents and CaO/Al2O3 ratios that relate to regional and temporal changes in mantle temperatures, lithospheric thickness, and involvement of at least two end-member mantle sources. We postulate that the Lower Basalts of East Greenland were derived from iron-poor mantle, depleted by prior melt extraction (and distinct from MORB source mantle), that was enriched in TiO2 and LREEs by a metasomatic event shortly before Early Tertiary magmatism. We associate this ‘depleted’ mantle source with the ancestral Iceland plume anomaly. KEY WORDS: continental flood basalts; East Greenland; geochemistry; magma differentiation; mantle melting
|
Introduction |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
The distinctive geochemical features of continental flood basalts (CFBs) are strongly controlled by the composition and structure of the continental lithosphere and underlying mantle. Significant controversy remains concerning the balance of contributions from the lithosphere and the mantle, and the mechanisms by which the various source components contribute to basalt petrogenesis. How are basalt compositions influenced by generation in anomalous regions in the deep mantle (plumes), by interaction between rising magmas and the continental lithospheric mantle, and by contamination with continental crust? How are the differentiation, contamination, magma mixing, and storage histories of CFBs controlled by the structure of the continental lithosphere? And how do the thermal and chemical features of the mantle and the extensional history of the continental lithosphere affect mantle melting and compositions of primary liquids?
These questions are best addressed in the context of a geologically and geochemically well-constrained CFB province, such as the North Atlantic Tertiary Province. Separation of East Greenland from Northern Europe in the Early Tertiary at 60 Myr ago was accompanied by voluminous volcanic activity. The Blosseville Coast of East Greenland, extending from Kangerdlugssuaq Fjord to Scoresby Sund (Fig. 1), contains spectacularly exposed sections of lavas. These sections begin at the basal contact of the lavas with underlying Cretaceous–Tertiary sediments and Archean basement, and continue to the transition between continental and oceanic volcanism. A major tectonic boundary in the continental lithosphere, the Caledonide Front, crosses the coast between Kangerdlugssuaq and Scoresby Sund, thus providing a unique opportunity to examine contemporaneous, closely related magmatism in distinct structural settings. In addition, the trail of the modern Iceland plume anomaly extends along the Greenland–Iceland–Faeroes ridge, intersecting the East Greenland margin at Kangerdlugssuaq (Fig. 1). This feature of the province permits characterization of the thermal and compositional nature of the mantle and the relationship between East Greenland margin volcanism and the Iceland anomaly.
|
The volcanic succession in the Scoresby Sund region has been extensively studied by Larsen et al. (1989)
. The volcanic succession in the Kangerdlugssuaq area has been examined by Wager (1947), Brooks et al. (1976), Soper et al. (1976), Nielsen et al. (1981), Brooks & Nielsen (1982a, b), Holm (1988), Gill et al. (1988), and Hansen (1990), who established the geologic relations and developed a preliminary overview of the geochemistry of the lavas. In this study we present new major, trace, and rare earth element, Sr and Nd isotopic, and mineral composition data for lavas representing the initial volcanic rift products (lowermost 1.5 km of volcanic statigraphy) of the East Greenland CFBs exposed in the Kangerdlugssuaq region. We use correlations among geochemical data to isolate the nature of interactions between magmas and the continental crust. We further constrain the differentiation histories of the lavas using phase equilibria constraints and calculated liquid lines of descent. We establish differences between the fractionation and contamination histories of magmas erupted in the Scoresby Sund and Kangerdlugssuaq areas, and attribute these differences to variations in the crustal architecture along the East Greenland margin. We investigate these issues in turn, giving careful attention to the interdependence of processes and underlying assumptions in modeling. Finally, we look beyond crustal differentiation processes and evaluate the constraints on primary mantle melt compositions for the East Greenland basalts provided by major elements. These results, coupled with trace element and isotopic data, are discussed in the light of current models of mantle source and melting conditions for basalts of the North Atlantic as a whole.
|
Geology and Stratigraphic Relations |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
Early Tertiary volcanic rocks in the Kangerdlugssuaq region erupted through continental basement largely composed in outcrop of Archean quartzo-feldspathic gneisses and a thin veneer of Cretaceous–Tertiary clastic sediments (Wager, 1947
; Leeman et al., 1976; Soper et al., 1976; Nielsen et al., 1981; Brooks & Nielsen, 1982a, b). The volcanic succession begins with the Lower Basalts, a sequence of lavas, tuffs, and volcaniclastic deposits totaling 1.5 km in thickness. The Lower Basalts are subdivided into the Vandfaldsdalen and Miki Formations (Fig. 2). The Lower Basalts are overlain by the Hængefjeldet Formation, a massive tuff unit, and by the Plateau Basalts, a sequence of lavas compositionally similar to the Scoresby Sund flood basalts (Wager, 1947; Nielsen et al., 1981). Total thickness of the succession is estimated to be 7 km (Nielsen & Brooks, 1981). The volcanic succession at Scoresby Sund is estimated to be 3 km thick and the lowermost units are possibly correlated with the Lower Basalts of the Kangerdlugssuaq region (Larsen et al., 1989).
|
The Kangerdlugssuaq region lies to the southwest of the main zone of deformation associated with the mid-Paleozoic Caledonian Orogeny (Escher & Watt, 1976
; Henriksen & Higgins, 1976) (Fig. 1) and was a relatively stable block from late Proterozoic to the time of continental break-up. Soper et al. (1976) and Nielsen et al. (1981) described occurrences of Cretaceous to Paleocene marine sediments and volcaniclastic deposits locally reaching thicknesses of 300 m infilling embayments in the basement terrain. However, these sedimentary successions are small compared with the thick accumulation of late Paleozoic–Mesozoic sediments (16–18 km) found in the Jameson Land basin northeast of Scoresby Sund (Larsen & Marcussen, 1992) (Fig. 1). Seismic reflection data for the Jameson Land basin indicate anomalously thin (6–9 km) continental crust below the basin fill and the presence of mid-crustal dike and sill complexes within the basin, inferred to be Early Tertiary in age and correlative with the main phase of flood basalt volcanism. Larsen & Marcussen (1992) suggested that extensional structures associated with the Jameson Land basin may have helped to guide early rift magmas into mid-crustal level chambers within the basin and laterally to eruption sites at the basin margins. This model is supported by the work of Larsen et al. (1989) showing that the bulk of the lavas erupted in the Scoresby Sund region at the southern edge of the Jameson Land basin are evolved Fe–Ti basalts with compositions characteristic of fractionation and mixing in upper- to mid-crustal magma chambers. Much less is known about the role of crustal structure in determining pathways for magma ascent along the southern portions of the Blosseville Coast where large pre-breakup extensional basins are absent.
The present study is based on mapping and sampling of the Lower Basalts in the vicinity of Miki Fjord (Fig. 2). A total of six individual stratigraphic profiles were measured through the Lower Basalts corresponding to a composite section of 1.2 km (Fig. 2). Detailed descriptions of the petrology of the Lower Basalts have been given by Wager (1947), Nielsen et al. (1981), Brooks & Nielsen (1982a), Hansen (1990), and Fram (1994). Salient petrographic features are described below.
The Vandfaldsdalen Formation contains lavas ranging in composition from picrite to basaltic andesite. A majority of the lavas are aphyric, although some units contain sparse microphenocrysts of olivine and, less commonly, of clinopyroxene, plagioclase, or orthopyroxene. Several prominent phenocryst-rich units are present and classified as picrites and ankaramites based on petrography.
The Miki Formation is relatively homogeneous, consisting almost entirely of pahoehoe flow units of olivine tholeiite to picrite lavas (Fig. 2). Lavas contain between 0 and 30 vol. % phenocrysts with the assemblages olivine or olivine + clinopyroxene. Plagioclase phenocrysts have not been observed in Miki Formation lavas. The internal structure of a representative Miki Formation pahoehoe flow is described in the Appendix.
Hydrothermal alteration in Lower Basalts is extensive and a function of proximity to Tertiary intrusions, the vesicle structure of the flow, brecciation, and the depth of burial (Bird et al., 1985). Within individual flows the massive interiors are generally the least altered and were sampled where possible. In the Appendix we report the results of a number of critical tests to assess the mobilities of components during hydrothermal alteration of the Lower Basalts.
The Scoresby Sund basalts are characterized by plagioclase-bearing phenocryst assemblages, although many lavas are aphyric. Olivine, olivine + plagioclase, and finally olivine + plagioclase + clinopyroxene phenocryst assemblages predominate with increasing degree of fractionation of the lavas (Larsen et al., 1989).
|
Analytical Methods |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
Forty-two samples of massive flow units of the Lower Basalts were chosen from the composite section for major, trace, and rare earth element analysis, electron microprobe analysis of phenocryst phases, and petrographic examination. Compositions of an additional 11 samples provided by T. F. D. Nielsen and C. K. Brooks from collections housed at the Greenland Geological Survey and University of Copenhagen were also determined in the study.
Olivine, spinel, and clinopyroxene compositions were obtained by wavelength-dispersive electron microprobe analysis using either the Cameca Camebax/Micro microprobe at the Lamont–Doherty Earth Observatory (LDEO) or the SX50 Cameca microprobe at the University of California, Davis (UCD). The microprobes are operated at 15 kV accelerating voltage, 1 µm beam width, and 25 nA (LDEO) or 10 nA (UCD) sample current. All elements were calibrated against mineral standards. Olivine, spinel, and clinopyroxene compositions were characterized by 3–10 spot analyses on 1–4 mineral grains per sample. Zoning profiles with 5–20 µm point spacing were made across clinopyroxene megacrysts of several samples. Olivine compositions are reported in Table 1 and clinopyroxene in Table 2.
|
|
Major and trace elements were analyzed in whole-rock powders using the Spectrometrics SMI III direct current plasma emission spectrometer (DCP) at LDEO. The procedures followed those outlined by Klein et al. (1991)
for major and trace elements and Miller et al. (1992) for REE, except that the solutions were spiked with Ge and Co, and Be, respectively, for short-term drift corrections. Analyses are accurate to within << 1% to 2% for all major elements, and to better than 5% for most trace elements, based on an analysis of standard NBS-688 as an unknown (Table 3). Ba, Zn, Zr, and Y analyses are slightly less accurate, but the Zr/Y ratio is still within 10% of the accepted value. Inter-run precision is 1% for major elements and 1–4% for trace elements, based on replicate analyses of sample CL89-23b (Table 3). Major and trace element analyses of Lower Basalts samples are reported in Table 4 and REE analyses in Table 5. Table 4 also summarizes the phenocryst assemblage found in each sample.
|
|
|
A subset of the samples analyzed by DCP were chosen for Sr and Nd isotopic analysis. Clinopyroxene phenocrysts from three Vandfaldsdalen and nine Miki Formation lavas were separated magnetically by a Franz Isodynamic magnetic separator. All samples were then hand-picked immersed in methanol or alcohol under a binocular microscope. The clinopyroxenes were lightly crushed in an agate mortar and leached in 5% HF and 6 N HCl before analysis. Catastrophically leached whole rocks were analyzed for samples without separable clinopyroxene. Whole rocks were leached in multiple aliquots of 6 N HCl for
150 h. The samples lost 50–75% of their original mass by dissolution and slight loss of fines during decanting. Separates of hydrothermal albite and epidote from two Miki Formation samples were also made by magnetic separation followed by hand-picking. A chlorite separate was made by chiseling out tube vesicles from one rock. DCP analysis of these vesicle fillings confirms that they are almost entirely chlorite (see Table A2).
|
After leaching, the clinopyroxenes (
20–25 mg) or powders (50 mg) were transferred to clean beakers and spiked with a multi-element spike (LaMES III) containing 150Nd, 84Sr, 149Sm, and 87Rb. The samples were dissolved in 8:1 HF:HClO4 in sealed beakers at 125°C for 24 h. The procedures for isolation of Rb, Sr, Nd, and Sm followed the standard procedures in the LDEO clean laboratory (Richard et al., 1976; Hart & Brooks, 1977; Zindler et al., 1984). Blanks were insignificant compared with the samples (<0.1 ng Sr, <20 pg Rb, <0.04 ng Sm or Nd). Following chemical separation, samples were analyzed on the VG 54–30 mass spectrometer at LDEO (results are reprorted in Table 6).
|
Based on examination of variability in chemical composition of samples from a single Miki Formation flow (see the Appendix) and comparison with a similar study of Keweenawan lavas (Schmidt, 1990
), we screened out samples and elements in our Lower Basalts dataset that have been noticeably disturbed by post-magmatic alteration. Briefly, the results are as follows. Concentrations of TiO2, P2O5, Zr, Y, Al2O3, Sc, and all but the lightest REE, and 143Nd/144Nd isotope ratios are not noticeably disturbed by hydrothermal alteration (see the Appendix). Their concentrations are representative of the original values even in highly altered samples. CaO, SiO2, and LREE are only noticeably disturbed in rocks in which the plagioclase is completely pseudomorphed by alteration phases. Samples with completely altered plagioclase were not considered in this study (and are not included in Table 4). Na2O, Sr, V, and Zn were determined to have been mobilized in all the samples (see the Appendix), but only to a limited degree in the samples included in the screened dataset. Data for these elements will be used with discretion and only to illustrate general points. 87Sr/86Sr isotope ratios of clinopyroxene mineral separates are considered representative of magmatic values, whereas those of leached whole rocks may be displaced from magmatic values. Thus Sr isotopic data are used with caution. K2O, Ba, and Cu are significantly remobilized in all of the Lower Basalts samples and will not be considered further. |
Geochemical and Petrographic Results |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
The compositions of Lower Basalts whole rocks and phenocryst phases determined in this study corroborate the general compositional features reported by Brooks et al. (1976)
, Nielsen et al. (1981), Brooks & Nielsen (1982a, b), Holm (1988), and Gill et al. (1988). Two distinct populations of olivine phenocrysts are observed (Table 1). The most common is Fo83–84 in composition, and is abundant in Miki Formation lavas. Less common are Fo88–90 olivines, which are often larger than the others and generally restricted to Vandfaldsdalen Formation picrites. Resorption of magnesian olivine was reported by Nielsen et al. (1981), but this is not a universal feature. Most olivine grains, however, are replaced by chlorite, serpentine, and magnetite.
Typical Miki Formation lavas contain olivine and clinopyroxene phenocrysts. The clinopyroxene phenocrysts commonly display strong compositional zoning and contain spinel inclusions that are chrome rich [Cr/(Cr+Al) 0.68] (Nielsen et al., 1981; Fram, 1994). Analyzed core compositions vary over a small range of mg-number (0.83–0.85) and are Cr rich (1 wt % Cr2O3) (Table 2). Zoning patterns are clearly defined by Al2O3 contents. Grains tend to have uniform Al2O3-rich cores (2–3.5 wt %), surrounded by uniform Al2O3-rich mantles (3.5–4 wt %), and finally by progressively less aluminous rims (Table 2). The most magnesium-rich and aluminum-poor clinopyroxene are found in the ankaramitic lavas of the Vandfaldsdalen Formation.
Major element compositions of the Lower Basalts lavas analyzed in this study are shown in Fig. 3. As originally observed by Brooks et al. (1976), the Lower Basalts are dominated by primitive whole-rock compositions (MgO > 8 wt %), whereas the Scoresby Sund succession contains few lavas with MgO >7 wt % (Larsen et al., 1989). Based on incompatible element abundances, particularly TiO2 (Fig. 3b), Brooks & Nielsen (1982a) and Gill et al. (1988) divided the Lower Basalts into three geochemical groups. The picrite–ankaramite series (PAS) includes lavas with high TiO2 abundances and MgO contents up to 20 wt %. PAS lavas include the picritic and ankaramitic lavas of the Vandfaldsdalen Formation and their differentiates and are most common in the early stages of magmatism. The tholeiite series (TS) lavas are less enriched in incompatible elements, rarely have >10% MgO, and define the lower TiO2 portion of the data array in Fig. 3b. TS lavas are present through the whole Lower Basalt sequence and are by far the dominant type found in the Scoresby Sund section (Brooks & Nielsen, 1982a; Gill et al., 1988; Larsen et al., 1989). The third group, Miki-type, dominates the Miki Formation and has TiO2 contents intermediate between those in PAS and TS lavas (Fig. 3b). Brooks and Nielsen (1982a) suggested that these lavas represent mixtures of the other two types. A few alkaline lavas occur in the Miki Formation at the top of the measured section (Fig. 2). These lavas are nepheline normative by virtue of their low silica rather than by high alkali contents. They can be distinguished in Fig. 3 by their low SiO2, and high CaO, TiO2, and FeO contents.
|
Lavas of the Lower Basalts with >11 wt % MgO show increasing SiO2, Al2O3, CaO, and Na2O contents as MgO decreases (Fig. 3), suggesting control by addition or subtraction of olivine. Below
11 wt % MgO, CaO declines and Al2O3 rises, indicative of clinopyroxene fractionation. In contrast, Scoresby Sund lavas generally have higher CaO and lower SiO2 contents than the Lower Basalts at comparable MgO content (Fig. 3a, d). Moreover, the steeply increasing FeO and TiO2 contents of Scoresby Sund lavas with decreasing MgO contents (Fig. 3b, c) are consistent with fractionation of plagioclase and mafic phases (Larsen et al., 1989). As will be shown below, these differences result from the contrasting fractionation histories of these lavas.
In addition to complex internal variability, the Lower Basalts display significantly higher TiO2 and FeO and lower CaO and Al2O3 contents at a given MgO content than do modern North Atlantic basalts. It is interesting to note that Na2O contents of the East Greenland basalts have a much smaller range than do TiO2 contents (Fig. 3a, f), suggesting that these two elements behave differently in this magmatic system. We have previously linked the gross compositional differences between the earliest rift products and those erupted from the subsequently established North Atlantic spreading system to the presence of continental lithosphere restricting mantle upwelling during the initial phase of continental break-up (Fram & Lesher, 1993). On a regional scale, the compositions of the most primitive lavas from the Scoresby Sund region fall mostly within the range of Lower Basalt compositions (see Fig. 3), although relatively primitive depleted compositions (e.g. low TiO2) approaching primitive Icelandic basalts are also found in the Scoresby Sund succession (Larsen et al., 1989).
The relative importance of olivine and clinopyroxene fractionation in the Lower Basalts and of plagioclase in the Scoresby Sund basalts can be further appreciated by examining variations of Ni, Sc, and Sr (Fig. 4a, b, c). Ni, Sc, and Sr are compatible in olivine, clinopyroxene, and plagioclase, respectively. Sr contents of the Lower Basalts increase from <100 p.p.m. to >500 p.p.m. as MgO decreases from 20 to 4 wt % (Fig. 4b). There is also a wide range of Sr at a given MgO content. The general increase in Sr with decreasing MgO indicates no significant fractionation of plagioclase, consistent with the petrographic observations. In contrast, the Scoresby Sund lavas, many of which do contain plagioclase phenocrysts as noted above, generally have Sr contents below the trend defined by the Lower Basalts. Relatively constant Sr contents of the Scoresby Sund lavas (Fig. 4b) are consistent with plagioclase (and olivine ± clinopyroxene) fractionation (Larsen et al., 1989).
|
Sc contents also show a range of values at a given MgO content in the Vandfaldsdalen and Miki Formation lavas (e.g. 30–36 p.p.m. at 12% MgO) (Fig. 4c). Because Sc is moderately compatible in clinopyroxene (D = 2–5; Gallahan & Nielsen, 1992
), moderately incompatible in olivine (D = 0.1–0.3; Beattie et al., 1991), and highly incompatible in plagioclase, its concentration in evolved magma is most sensitive to the proportion of clinopyroxene in the fractionating assemblage. The gradual increase in Sc with decreasing MgO found for the Lower Basalts between 22 and 13 wt % MgO suggests olivine control owing to either dilution from accumulated olivine in liquids with as low as 11 wt % MgO or olivine fractionation from primitive picritic liquids. In either case, maximum Sc concentrations (36 p.p.m.) are found in lavas with 10–13 wt % MgO and concentrations drop markedly with decreasing MgO contents (Fig. 4c). The maximum at 11 wt % MgO is interpreted as marking the appearance of clinopyroxene in the fractionating assemblage for the Lower Basalts. In contrast, Sc concentrations vary unsystematically between 24 and 44 p.p.m. in Scoresby Sund basalts containing 5–7 wt % MgO. These relationships for the evolved Scoresby Sund compositions are consistent with the petrographic observations indicating important roles for olivine and plagioclase fractionation, and to a lesser extent, clinopyroxene. On the other hand, the similarity in the Sc contents of the most primitive Scoresby Sund lavas and moderately evolved Lower Basalts (11 wt % MgO) is difficult to explain without the early involvement of clinopyroxene fractionation in both suites if they are indeed related to common parental magmas.
Zr/Y ratios in primitive Lower Basalts range from 4 to 10, whereas those for relatively primitive Scoresby Sund lavas range from 2 to 6 (Fig. 4d). Zr concentrations show a similarly large degree of variation (Fig. 4e). Bernstein (1994) observed increasing Zr/Y ratios with decreasing MgO contents in the Faeroe Islands Lower Series basalts and attributed it to crystal fractionation involving garnet at the base of the continental crust (40–50 km in depth). Such a correlation is not well displayed for the East Greenland succession. Rather large ranges in Zr/Y ratios and Zr concentrations are displayed by the most primitive lavas of the Lower Basalt and Scorseby Sund suites, suggesting that variations in these incompatible elements are linked more directly to mantle melting systematics and source composition.
Figure 5 presents representative REE patterns for the Lower Basalts. In general, the lavas are enriched in LREE relative to HREE, but significant differences are noted. Most Vandfaldsdalen Formation lavas exhibit steep, straight patterns, resembling those determined by Gill et al. (1988) for dikes from the Kangerdlugssuaq and Fladø regions. (Ce/Sm)N varies from 1 to 2 and (Dy/Yb)N from 1.2 to 1.6, resulting in crossing REE patterns (Fig. 5a). The majority of the Miki Formation lavas display parallel, distinctly curved REE patterns with (Ce/Yb)N = 3.0–3.8 (Fig. 5b). The LREEs internally display a relatively flat pattern with (Ce/Sm)N = 1.0–1.3, and HREE fractionation in these lavas spans a limited range of (Dy/Yb)N = 1.4–1.5. The lavas display no or small positive Eu anomalies (Eu/Eu* = 1.0–1.1) but the magnitude of the Eu anomaly is not correlated with degree of fractionation of the sample. These curved patterns were not observed in the population of dikes analyzed by Gill et al. (1988), although they are characteristic of the Lower and Middle Series lavas on the Faeroe Islands (Gariépy et al., 1983). Intercalated with the typical Miki Formation lavas are flows that have slightly steeper LREE slopes [(Ce/Sm)N = 1.3–1.5] than those of the main group, but parallel HREE slopes (Fig. 5c). The two alkaline lavas, MF91-57b and 57c (Table 5), at the top of the section have REE patterns steeper than all others in the Lower Basalts.
|
Figure 6 summarizes the stratigraphic variations in Sr and Nd isotopic ratios for East Greenland basalts. In the Vandfaldsdalen Formation initial 87Sr/86Sr ratios vary from 0.7036 to 0.7054 and initial 143Nd/144Nd ratios vary from 0.5122 to 0.5130. Lavas of the basal 150 m all have relatively radiogenic Sr and unradiogenic Nd. Most of the flows in the Miki Formation have a limited range of values (87Sr/86Sr = 0.7032–0.7034 and 143Nd/144Nd = 0.5128–0.5130) that approach values observed in modern Icelandic basalts (Fig. 6). These lavas are intercalated with flows with distinctly more radiogenic 87Sr/86Sr and less radiogenic 143Nd/144Nd.
|
87Sr/86Sr and 143Nd/144Nd ratios in lavas from Scoresby Sund and along the Blosseville Coast between Kangerdlugssuaq and Scoresby Sund (Carter et al., 1979
; Holm, 1988; Larsen et al., 1989) cluster at the high 143Nd/144Nd, low 87Sr/86Sr end of the Lower Basalts array (Fig. 6). A few samples from near the base of the Scoresby Sund succession have markedly higher 87Sr/86Sr. Larsen et al. (1989) modeled contamination of these lavas by bulk assimilation of up to 20% local Archean basement gneiss. |
Data Analysis and Discussion |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
Crustal contamination
Before we can investigate the range of primary liquid compositions for the East Greenland basalts, we must determine what portion of the observed variation in lava compositions is due to interaction of magmas with continental lithospheric mantle and crust.
Previous Sr–Nd–Pb isotopic investigations of the Lower Basalts reached the conclusion that the observed isotopic variation reflected interaction between the ancestral Iceland plume, asthenospheric mantle, and 3-Byr-old lithospheric mantle (Holm, 1988). Holm (1988) suggested that crustal contamination was not a significant factor in producing isotopic variation in the lavas because of the poor correlations he observed between incompatible element ratios in contaminated lavas and ratios in the local basement rocks.
Our examination of a broader set of geochemical data for the Lower Basalts leads us to conclude that crustal contamination played a large role in modifying the compositions of Lower Basalt magmas. Figure 7 shows strong correlations between the SiO2 contents of East Greenland lavas and their 87Sr/86Sr and 143Nd/144Nd ratios. Only samples with mg-number [MgO/(FeOtot + MgO)] < 0.65 are plotted because calculated liquid lines of descent for Lower Basalt liquids (discussed below) indicate that the SiO2 content changes little during fractionation of liquids with mg-number < 0.65. At higher mg-number, SiO2 increases with decreasing MgO as expected for fractionation of olivine alone. Simple mixing calculations indicate that the range observed in SiO2 in the Lower Basalts (47.5–52.5 wt %; Fig. 7) can be accounted for by mixing of basalt having 47 wt % SiO2 with up to 15% crustal component with 65–70 wt % SiO2.
|
Correlations between isotopic ratios and trace elements provide for a more comprehensive assessment of the impact of crustal contamination on magma compositions. Incompatible element contents of representative Vandfaldsdalen and Miki Formation lavas are shown in Fig. 8a. The lavas chosen for this diagram encompass the full range of Sr and Nd isotopic ratios observed in the Lower Basalt suite. The data are shown in a double normalized format. The data for each sample are first normalized to the incompatible element contents of the sample (MF91-55d) with the lowest 87Sr/86Sr and highest 143Nd/144Nd. The data are then normalized to the TiO2 content of this reference sample. This second normalization eliminates differences in concentrations of incompatible elements between samples owing to crystal fractionation or accumulation. Thus, the reference sample plots as a horizontal line at unity in Fig. 8a and deviations from unity represent enrichments not readily explained by fractionation. It is noteworthy that enrichments in Sr, LREE, and Zr correlate with higher 87Sr/86Sr and lower 143Nd/144Nd, and are reasonably linked to crustal contamination. Phosphorus, Y, and HREEs have values close to unity and thus are not enriched relative to TiO2. It is also evident from a comparison of major element and isotopic composition that TiO2 is itself not enriched by contamination. For example, at a given MgO content, lavas with the highest 87Sr/86Sr and lowest 143Nd/144Nd ratios tend to have the lowest TiO2 contents, a difference attributable to dilution by a felsic crustal component. This correspondence is consistent with the elevated SiO2 contents of the most contaminated lavas. A similar relationship has been reported for crustally contaminated lavas from the Asuk and Kuganguak members of the Vaigat Formation on Disko Island, West Greenland (Pedersen, 1985
).
|
Both Vandfaldsdalen and Miki Formation lavas show similar patterns of incompatible element enrichment coupled with increasing 87Sr/86Sr and decreasing 143Nd/144Nd (Fig. 8a). Several samples show anomalous behavior, however. For example, sample CL89-8e exhibits a large depletion in Sr compared with the other samples, which is probably due to leaching of Sr during post-eruptive alteration (see the Appendix). Normalized LREE, Zr, and P2O5 contents are also lower in sample CL89-8e than expected for the isotopic composition of the sample. This may be due to elevated TiO2. When compared at a similar stage of differentiation, CL89-8e has TiO2 and FeO contents higher than the main group of primitive Lower Basalts (Table 4, Fig. 3b).
We can further constrain the nature of the contami- nant by simple mixing models. Unlike the example of interaction between macrodike magmas and their host gneiss investigated by Blichert-Toft et al. (1992), there is no clear field evidence for interaction between Lower Basalt lavas and any specific crustal rock. For this reason, we can only estimate the crustal end-member in the mixing calculations. We used three plausible contaminants: a leucogneiss from Kangerdlugssuaq (Blichert-Toft et al., 1992); a leucogneiss from the Lewisian of Scotland (Thompson et al., 1982; Dickin et al., 1987); and a granophyric melt of gneiss from Kangerdlugssuaq (Blichert-Toft et al., 1992). The two leucogneiss samples have similar trace element abundances and Nd and Pb isotopic ratios, but Kangerdlugssuaq gneiss has high 87Sr/86Sr whereas Lewisian gneiss has low 87Sr/86Sr (Fig. 9). The two Kangerdlugssuaq samples have similar isotopic ratios (Fig. 9), but the granophyre has lower Sr, and higher HREE and Y concentrations.
|
Figure 8b shows the results of simple mixing calculations involving the Lower Basalt reference liquid and the three prospective crustal contaminants. The reference liquid is derived from the composition of MF91-55d (Table 4) by removal of olivine from the whole-rock composition to generate a calculated composition with mg-number = 0.65, and concomitant increase in concentrations of incompatible elements calculated with the Rayleigh fractionation equation. Mixtures composed of 85% basalt and 15% crustal contaminant are shown in Fig. 8b. Comparison of the shapes of the incompatible element patterns in the three mixtures (Fig. 8b) with the patterns of the contaminated suite (Fig. 8a) indicates that mixtures of leucogneiss and basalt more closely resemble the composition of contaminated lavas than does the mixture of granophyre and basalt. In particular, the granophyre–basalt mixture contains higher concentrations of HREE and Y compared with the lavas.
Sr and Nd isotopic arrays for mixing between the reference liquid and the three prospective contaminants are shown in Fig. 9. Two of the three models yield good correspondence. The least suitable is the model assuming bulk assimilation of the local basement. Contamination by Sr-depleted partial melts of this local gneiss is plausible based on Sr–Nd isotope systematics, but as shown in Fig. 8b this contaminant gives a poor fit to the trace element inventory. Thus, considering both the elemental and isotopic data, depleted Lewisian leucogneiss appears to provide the most consistent end-member contaminant for the Lower Basalts. Incorporation of up to 20% of this crustal lithology is required to account for the isotopic composition of the most contaminated lavas.
In view of the scarcity of basement lithologies with Lewisian-type isotopic compositions in the Kangerdlugssuaq area (Leeman & Dasch, 1978; Kays et al., 1989; Blichert-Toft et al., 1992), the modeling suggests that contamination of the Lower Basalts occurred at depth in the crust. This conjecture contrasts with the work of Larsen et al. (1989) on the Scoresby Sund lavas indicating involvement of a felsic contaminant with a high abundance of radiogenic Sr. The occurrence of suitable crustal lithologies immediately underlying the volcanic succession in the Scorseby Sund area led those workers to suggest that contamination occurred at shallow crustal levels, as similarly documented for the Skaergaard and macrodike intrusions (White et al., 1989; Stewart & DePaolo, 1990; Blichert-Toft et al., 1992; Geist & White, 1994). These inferences concerning the crustal level for contamination based on the characteristics of an inferred contaminant can be strengthened by considering the polybaric fractionation histories for East Greenland lavas.
We conclude from these observations that magmas of the Miki and Vandfaldsdalen Formations examined thus far interacted with crustal material compositionally similar to Lewisian basement. The stratigraphic relations further show that there is a temporal change in the frequency of contaminated lavas. Crustally contaminated lavas are ubiquitous in the basal 150 m of the Vandfaldsdalen Formation, common in the remainder of the Vandfaldsdalen, and infrequent in the Miki Formation (Fig. 6). Although further work is necessary to establish in detail the connection between these time–stratigraphic relationships and the development of the magmatic conduit system, the gross variations suggest that crustal conduits to the surface matured rapidly after volcanism began. There is no evidence for the eruption of silicic magmas or a hiatus in volcanic activity as recently discovered in the continental succession during ODP drilling off SE Greenland (Larsen et al., 1994).
The temporal variation in contamination histories involving lower-crustal gneisses displayed by the Lower Basalts is also well documented in the British Tertiary Province (Moorbath & Thompson, 1980; Dickin, 1981; Thompson et al., 1982; Thirlwall & Jones, 1983; Dickin et al., 1987). However, it should also be noted that the basal lavas, which are only found locally, also show an additional contamination signal from crustal lithologies found at higher levels of the crust (Morrison et al., 1985; Thompson et al., 1986). A similar basal sequence is not recognized in the suite of Lower Basalts investigated in this study, but may be represented by lavas of the lower Vandfaldsdalen Formation under study by H. Hansen (personal communication, 1996).
Finally, East Greenland basalts display none of the geochemical features indicative of significant involvement of the continental lithospheric mantle, contrary to the conclusion reached by Holm (1988). Arndt & Christensen (1992) suggested that magma–lithosphere interaction will fractionate Nb and Ta from the LREE, U, Th, Rb, Ba, and Sr. This interaction is then responsible for producing the large negative Nb and Ta anomalies seen in incompatible element patterns of lavas from CFB provinces such as the Karoo and Tasmanian provinces (Cox et al., 1984; Hergt et al., 1991). It is noteworthy that none of the lavas and dikes analyzed by Gill et al. (1988) from the Kangerdlugssuaq area show large negative Nb and Ta anomalies. Nb and Ta depletions in British Tertiary Province lavas (e.g. Thompson et al., 1986) are greater than can be accounted for by crustal contamination, but it is also apparent that most other compositional features of the lavas are well explained by crustal contamination. If the continental lithospheric mantle is responsible for excess Nb and Ta depletions in the British Tertiary lavas, then the differences between East Greenland and British Tertiary lavas may reflect regional variations in the composition of the continental lithospheric mantle and/or the efficiency of melt-wallrock interaction during magma ascent. Cox (1988) and Ellam & Cox (1991) have suggested that ascending CFB magmas scavenge low-degree partial melts (lamproites) from the continental lithospheric mantle that have high contents of incompatible elements, particularly K2O, Ba, Rb, and LREE. This behavior is well documented in picrites from the Karoo Province that contain up to 4 wt % K2O. Although we cannot completely dismiss this potential lithospheric contaminant, contaminated Lower Basalts do not show enrichments of K2O, Ba, Rb and LREEs in excess of that expected from the degree of crustal contamination inferred from enrichment of silica in the lavas [Table 4 and Nielsen et al. (1981)].
Estimating parental magma compositions
Quantitative constraints on the conditions for fractionation of East Greenland basalts can be established by combining the petrographic observations, geochemical results, and available experimental data on the liquid line of descent (LLD) of basaltic magmas. Two issues are of central importance for quantitative modeling of the LLD. The first is to establish the real range of eruptive liquid compositions by critically evaluating the extent of crystal accumulation, particularly for phyric lavas. Second, estimates of primary liquid compositions must be made assuming a justifiable fractionation scheme. It is obvious that these two issues are not independent and for the most evolved lavas, reflecting the most complex history, would result in circular (or unconstrained) reasoning. However, as previously shown, many Lower Basalts are relatively primitive with MgO contents of 9–20 wt %. Nearly all of these high-magnesium samples contain olivine phenocrysts, so the question remains whether any of these MgO-rich lavas represent liquid compositions. The importance of olivine accumulation in Lower Basalts was first recognized by Brooks & Nielsen (1982a). They noted that accumulation of Fo85 olivine could account for much of the major element variation found in primitive PAS lavas. The procedure described below establishes the permissible compositions of Lower Basalt liquids, correcting for the effects of olivine accumulation. Below we also restrict the analysis to lavas least affected by contamination. These include lavas with 143Nd/144Nd > 0.5126, and La/Zr < 0.07 and Zr/Ti < 80. These last two criteria were used for screening samples lacking Nd isotope data.
Most of the olivine phenocrysts analyzed from Miki Formation lavas have compositions close to Fo84 (Table 1; Nielsen et al., 1981), and as can be seen in Fig. 10 many of the high-magnesium Lower Basalts could be interpreted as related by accumulations of 0–30% Fo84 olivine in liquids with 10 wt % MgO. It is considered significant that this array for accumulation of Fo84 olivine intersects the locus of compositions in Fe–Mg exchange equilibrium with Fo84 olivine at approximately the composition of several aphyric and nearly aphyric (<3% phenocrysts) lavas (CL89-7c, -7a; GM 40641; Table 4) (Fig. 10). Whole-rock samples plotting along the Fo84 control line all contain olivine phenocrysts.
|
. The arrow extends from a point on the Fo84 equilibrium line with Highly olivine–phyric lavas are found in the basal portions of a number of other CFB provinces. In particular, the picrites of West Greenland and Baffin Island (e.g. Clarke, 1970
; Pedersen, 1985) contain olivine phenocrysts with forsterite contents up to Fo92.4. Pedersen (1985) estimated that these olivine phenocrysts were in equilibrium with melt with 19 wt % MgO, and postulated that such highly magnesian liquids may have erupted during the early Tertiary in West Greenland. Similarly, olivine phenocrysts with forsterite contents up to Fo88 and Fo90 have been reported in picrites from the Deccan and Karoo CFB provinces, respectively (Cox & Jamieson, 1974; Krishnamurthy & Cox, 1977). Parental melts for these lavas are also inferred to be picritic with 16–18 wt % MgO (e.g. Krishnamurthy & Cox, 1977; Cox, 1988), although individual lavas with these compositions generally erupted as suspensions of olivine ± clinopyroxene as we have found in East Greenland.
Estimates of parental magma compositions for the Lower Basalts are made first by correcting for accumulation of Fo84 olivine to give a spectrum of liquids with mg-number = 0.64 and then simulating olivine fractionation in reverse by incrementally adding 0.5% by weight of the equilibrium olivine until liquids obtain an mg-number of 0.70. We choose this final mg-number for the liquid because it corresponds to a liquid in equilibrium with Fo88 olivine, which is among the most magnesian olivine reported in the Lower Basalts. This fractionation procedure requires only that the iron–magnesium exchange distribution coefficient for olivine 
be specified, and in these calculations is taken as 0.31 (e.g. Longhi, 1991). In performing this fractionation correction we considered only whole-rock compositions with >9 wt % MgO and ignored the accumulation correction step for lavas plotting to the high-FeO side of the Fo84 control line in Fig. 10 for which olivine compositional data are lacking.
The range of hypothetical parental liquids at mg-number = 0.70 (i.e. in equilibrium with Fo88 olivine) are shown in Fig. 10. Table 7 lists the compositions for three parental liquids along this array. Composition A is the estimated parental melts for the alkaline lavas, whereas compositions B and C represent prevalent parental liquids for the tholeiitic lavas. No estimate is given for the parental magma of the ankaramitic lavas because of the difficulty in correcting for possible olivine and clinopyroxene accumulation.
|
Phase equilibria
We employ two approaches to quantify the fractionation histories of East Greenland basalts. The first is a graphical approach comparing experimental phase relations for basaltic systems over a range of pressures with the lava and calculated parental magma compositions. The second approach involves forward modeling of the LLDs using hypothetical parental magma compositions B and C (Table 7).
Figure 11a shows the compositional relations in terms of olivine–clinopyroxene–plagioclase components projected from silica, using the projection scheme of Tormey et al. (1987). Use of this scheme does not require that the projected compositions be saturated with the projection phase, in this case silica, because the projection is onto the olivine–clinopyroxene–plagioclase plane, not the intersection with the silica saturation surface. Silica was chosen as the projection component because its use minimizes the effects of alteration and contamination on the projected compositions. The positions of the olivine–clinopyroxene, olivine–plagioclase, and plagioclase–clinopyroxene cotectics at 1 bar, 8 kbar, and 12 kbar shown in Fig. 11a are based on experimental data for MORB-type compositions (Bender et al., 1978; Walker et al., 1979; Fujii & Bougault, 1983; Elthon & Scarfe, 1984; Baker & Eggler, 1987; Tormey et al., 1987; Kinzler & Grove, 1992) and on calculated LLDs for Lower Basalt-type compositions (Longhi, 1991).
|
The vast majority of Scoresby Sund lavas lie in a restricted field close to the low-pressure olivine–plagioclase–clinopyroxene and olivine–plagioclase cotectics (Fig. 11a) as expected from the work of Larsen et al. (1989)
. Most of the Lower Basalts plot within the olivine primary phase volume at low pressure. The difference in projected positions of contaminated and uncontaminated Lower Basalts is minor because silica is the projection component.
Increasing pressure causes the clinopyroxene phase volume to expand, shifting the olivine–clinopyroxene cotectic away from the clinopyroxene apex towards lower clinopyroxene component. The available experimental data suggest that rotation of the boundary during this shift is minimal in the pressure range of interest. Aphyric samples (Fig. 11b) project between the 1 bar olivine–plagioclase and olivine–plagioclase–clinopyroxene cotectics (Vandfaldsdalen Formation samples only) and 12 kbar olivine–clinopyroxene cotectic. Other samples projecting into the olivine phase volume have olivine and clinopyroxene phenocrysts, consistent with crystallization along an olivine–clinopyroxene cotectic, whereas still others form an array radial to the olivine component reflecting olivine accumulation as discussed above (Fig. 11b). The calculated liquids in equilibrium with Fo84 olivine (see Fig. 10) project in a region clustering about the 8 kbar olivine + clinopyroxene cotectic (Fig. 11b), whereas the calculated parental liquids A, B and C plot well within the olivine phase volume at this pressure. Moreover, whole-rock compositions corrected for olivine accumulation to an mg-number in equilibrium with cores of clinopyroxene phenocrysts also project close to the 8 kbar olivine + clinopyroxene cotectic. This inferred high-pressure origin for the lavas is supported by the high Al2O3 contents (up to 3.75 wt %) in clinopyroxene (Table 2) and more quantitatively by application of the clinopyroxene–liquid geobarometer of Putirka et al. (1996) based on Al2O3 and Na2O exchange, which yields equilibration pressures of 7–13 kbar.
Quantitative liquid lines of descent
Figure 12a shows the variation of CaO/Al2O3 with MgO for the East Greenland lavas. This diagram is particularly useful given its sensitivity to the proportions of olivine, clinopyroxene, and plagioclase making up the fractionating assemblage. The lines in the upper left corner show the effect of fractionation of 5% of each of the three phases from an initial liquid with 10% MgO and CaO/Al2O3 = 0.9 (the lines have been transported to the upper left for clarity). Because clinopyroxene has a high CaO/Al2O3 and MgO content (see Table 2), fractional crystallization of clinopyroxene rapidly drives the residual liquids to lower CaO/Al2O3 and MgO. Conversely, plagioclase contains essentially no MgO and has CaO/Al2O3 of 0.4–0.5 so removal of plagioclase from magmas increases both MgO and CaO/Al2O3. Olivine fractionation decreases MgO rapidly but leaves CaO/Al2O3 unchanged. Co-crystallization of plagioclase, clinopyroxene, and olivine in low-pressure cotectic phase proportions results in little change in CaO/Al2O3 and a slow decrease in MgO.
|
Lower Basalts display a wide range of CaO/Al2O3 ratios and MgO contents (Fig. 12a). At high MgO contents, the dominant trend is horizontal, consistent with olivine control. All of the samples with >11% MgO contain olivine phenocrysts and, as discussed above, many represent accumulations of olivine in liquid. The close compositional correspondence between high-magnesium samples with and without clinopyroxene phenocrysts suggests that although clinopyroxene crystallized, it was not efficiently separated from the magmas. Given that the olivines in these lavas are generally larger and slightly denser than the clinopyroxene, Stokes settling velocities of the olivine phenocrysts are estimated to be 10–100 times faster than those of the clinopyroxene phenocrysts. As many of the magmas erupted contained suspended olivine it is unlikely that clinopyroxene would have preferentially settled out. Krishnamurthy & Cox (1977)
noted that geochemical variations in picritic basalts from the Deccan CFB Province are also largely controlled by removal and addition of olivine, although clinopyroxene phenocrysts are also present. They suggested a more complex crystal settling model in which both clinopyroxene and olivine settle together, but clinopyroxene does not appear to fractionate because crystals settling from any given parcel of magma in a vertical column are balanced by crystals accumulated from above (Krishnamurthy & Cox, 1977). Some of the range in CaO/Al2O3 in the more magnesian samples may reflect differences in primary liquid compositions. The alkaline and ankaramitic lavas have the highest CaO/Al2O3 (Fig. 12a). The absence of clinopyroxene phenocrysts in the alkaline lavas indicates that their high CaO/Al2O3 is a primary feature of these liquids rather than an artifact of clinopyroxene accumulation.
An important constraint on the fractionating assemblage during differentiation of Lower Basalt magmas is the sharp decline in CaO/Al2O3 ratio below MgO contents of 11 wt % (Fig. 12a). This is most easily explained by involvement of clinopyroxene in the fractionating assemblage when magmas reached 11 wt % MgO. Evidence for clinopyroxene fractionation in the Lower Basalt suite was noted above in discussing the Sc–MgO relationships (Fig. 4c). In contrast, Scoresby Sund lavas display a restricted range in CaO/Al2O3 and MgO content. A few evolved Vandfaldsdalen Formation lavas show comparatively high CaO/Al2O3 at low MgO content overlapping the Scoresby Sund field (Fig. 12a).
Quantitative modeling of the liquid line of descent over a range of pressure was undertaken to reproduce the general features shown in Fig. 12a. We used the fractional crystallization program MAGFOX of Longhi (1991) to compute the LLDs for inferred parental liquids and performed calculations under both isobaric and polybaric conditions. The program MAGFOX is conceptually based on portrayal of crystallization as a path followed by a fractionating liquid on a series of pseudoternary liquidus phase diagrams. The positions of phase boundaries in multicomponent space are fitted with empirical equations that are based on experimental data (e.g. Longhi & Pan, 1987; Longhi, 1991). The parameterization accounts for the shifts in the positions of phase boundaries owing to (1) the addition of components (i.e. FeO, TiO2, alkalis) and (2) increasing pressure relative to their positions in the CaO–MgO–Al2O3–SiO2 simple system (Presnall et al., 1978; Longhi, 1987). Crystallization is simulated by solving the equations for the phase boundaries to determine on which boundary the liquid lies, removing the appropriate mineral phases, and repeating this procedure over small increments of fraction of liquid crystallized. MAGFOX includes phase boundaries involving all the major phases encountered in basaltic systems (olivine, plagioclase, augite, pigeonite, orthopyroxene, spinel, and ilmenite), utilizes experimental data at pressures from 1 bar to 40 kbar, and includes the effects of minor elements such as Al2O3, Cr2O3, Na2O, MnO, K2O, and P2O5 on mineral compositions and phase relations. The model is set up to run in FORTRAN on a Macintosh computer and results are easily translated into graphical representations. Longhi (1991) demonstrated the accuracy of the program for a range of basaltic compositions, even those not used explicitly in the parameterizations, by testing predicted positions of phase boundaries against experimental data.
We use the estimated primary liquids (C-I and B-I) from Table 7 as starting compositions in the calculations, and focus our attention on the tholeiitic lavas, as the alkaline and ankaramitic lavas represent only a very small portion of the Lower Basalts. Furthermore, as crustal contamination apparently affects CaO/Al2O3, the calculated LLDs in Fig. 12b are only compared with uncontaminated samples.
Results of isobaric calculations of the LLDs at 1 bar and 8 kbar are shown in Fig. 12b. The model predicts the crystallization sequence olivine 
clinopyroxene plagioclase for all pressures, with the most important difference being the stage at which clinopyroxene begins crystallizing. At low pressure for primary liquid C clinopyroxene appears at 8 wt % MgO, followed by plagioclase at 6 wt % MgO. For primary liquid B clinopyroxene crystallization is followed immediately by plagioclase. For both starting compositions it is seen that with increasing pressure clinopyroxene joins olivine crystallization earlier in the evolutionary scheme. This reflects a shift of the olivine–clinopyroxene cotectic to higher MgO contents (and temperatures) by 0.3 wt % MgO/kbar for the model of Longhi (1991). Thus, at 8 kbar, clinopyroxene saturation occurs at 11 wt % MgO for both compositions (Fig. 12a). In the present context, the importance of this change in clinopyroxene stability is that it results in a marked change in the trajectory of the LLD in terms of the MgO content and CaO/Al2O3. When compared with the natural data, it is obvious that the decrease in CaO/Al2O3 ratios for Lower Basalts with MgO contents less than 11 wt % is consistent with fractionation at high pressures and would be difficult to reconcile with low-pressure evolution. These results are consistent with the projected phase relations shown in Fig. 11 and clinopyroxene compositions discussed above. The low-pressure model fractionation paths for both parental liquids B and C intersect the compositions of the Scoresby Sund lavas (Fig. 12b). However, the isobaric calculations do not correctly predict the phenocryst assemblages found for Scoresby Sund lavas, specifically the plagioclase–phyric nature of the lavas. This enigma posed by the petrographic observations and results from simple LLD calculation on East Greenland parental magmas suggests that a more complex fractionation history was involved.
The effects of polybaric fractionation can be illustrated by modifying the program MAGFOX (Longhi, 1991) to simulate successive stages of isobaric crystallization and decompression. The results presented in Fig. 12b show plausible and representative scenarios involving an early stage of high-pressure polybaric fractionation followed by decompression, storage, and fractionation at upper-crustal conditions. Although there is considerable leeway in prescribing a decompression history in terms of the amount of crystallization per kilobar, and no attempt has been made to link these calculations directly to the physics of magma ascent, there are certain restrictions on the polybaric path that will yield erupted lavas with the unique compositions and phenocryst assemblages typical of Scoresby Sund lavas.
The model paths shown in Fig. 12b are for primary liquids B and C. In each case the calculations assume a starting pressure of 15 kbar. The system is permitted to crystallize under fractional conditions at a rate of 0.5% (by mole fraction) per kbar of decompression from 15 to 8 kbar, followed by storage until 30% of the original liquid has crystallized. Liquids are then permitted to continue decompressing, again crystallizing by 0.5% per kbar until they reach 2 kbar. Fractionation calculations are continued until 70% of the original liquid has crystallized.
Olivine is the liquidus phase for both compositions and is the only phase to fractionate in the first 20% of crystallization (initial flat portion of paths in Fig. 12b). The portion of the LLD path with decreasing CaO/Al2O3 beginning at 11 wt % MgO (Fig. 12b) corresponds to fractionation of olivine and clinopyroxene at high pressure as discussed previously for the isobaric case. The model also predicts a small amount (<1% by volume) of pigeonite crystallization. When the remaining liquid is further decompressed crystallization returns to olivine only (flat portion of paths between 9 and 7 wt % MgO). The portion of the LLD path displaying an increase in CaO/Al2O3 reflects the appearance of plagioclase in the fractionating assemblage. Saturation with clinopyroxene occurs for the second time at low pressure when the liquid reaches 5 wt % MgO, and the CaO/Al2O3 ratio of the liquid again decreases with decreasing MgO, albeit more gradually than at high pressures. In summary, the hypothetical polybaric fractionation histories for primary liquids B and C yield the fractionation sequence olivine and olivine + clinopyroxene at high crustal pressures, followed by olivine, olivine + plagioclase, and olivine + plagioclase + clinopyroxene upon decompression and evolution at low pressure. It is proposed that the major element compositions of the Scoresby Sund lavas reflect primarily the latter stages of polybaric evolution (below 8 wt % MgO) portrayed in Fig. 12b. However, we emphasize that an earlier stage of high-pressure fractionation, similar to that inferred for the Lower Basalts, is necessary to yield plagioclase (± olivine) phyric lavas at low pressure. This model of polybaric fractionation of East Greenland magmas is similar to that proposed by Cox (1980) to explain the evolution of Karoo flood basalts and the Rooi Rand dolerites.
The role of high-pressure fractionation has also been shown to be important for magma evolution in the British Tertiary Province. Morrison et al. (1985) and Thompson et al. (1986) compared lavas from Skye and Mull with the positions of 1 bar and 9 kbar cotectics in the normative ternary system olivine–hypersthene-diopside. They showed that the earliest lavas, represented by the Staffa Magma Type on the island of Mull and the Torvaig lavas on Skye, project close to the 1 bar cotectic, as recognized for the Scoresby Sund plateau basalts. Lavas of the main part of the volcanic succession, the Skye Main Lava Series and the Mull Plateau Group, preserve a signature of fractionation at 9 kbar, similar to the result we have presented for the Lower Basalts. Bernstein (1994) has concluded, based on trace element modeling for the Lower Series of the Faeroe Islands, that garnet became an important fractionating phase at pressures of 15 kbar. This prediction is higher than pressures indicated in our analysis of the Lower Basalts, and by Morrison et al. (1985) and Thompson et al. (1986) for the British Tertiary Province. However, based on the work of Waagstein (1988), the Lower Series of the Faeroe Islands are not correlative with the Lower Basalts of East Greenland and may be related to an earlier phase of rifting associated with thicker continental crust.
The differentiation histories of the Lower Basalts and Scoresby Sund lavas inferred here provide a context for understanding the crustal contamination histories of the lavas discussed above, the temporal evolution of magma conduit systems, and the contrast in conduit system development across the Caledonide Front. Figure 13 shows a schematic representation of the differentiation and contamination histories of the East Greenland basalts. In summary, lavas of the Vandfaldsdalen Formation preserve compositional relations reflecting differentiation at pressures ranging from 1 bar to >8 kbar, although the proportion of lavas fractionating at low pressure is small. Vandfaldsdalen Formation lavas also exhibit a large range in degree of fractionation and have a high frequency of contamination, particularly in the earlier stages. This suggests that magmas ascended along tortuous paths with extensive contact with fresh crustal material and subject to arrest at various depths for variable amounts of time (Fig. 13). In contrast, lavas of the Miki Formation record more uniform differentiation pressures, and exhibit a more uniform degree of differentiation and a lower frequency of contamination. This implies a more mature conduit system (Fig. 13). During eruption of Miki Formation lavas, magmas had apparently easy access to the surface from the deep crustal ponding site and ascended without significant fractionation and contamination. The dominant differentiation pressure recorded by Lower Basalt lavas corresponds to differentiation at depths of 20–25 km. We further surmise, based on the major element, trace element and isotopic relationships among contaminated and uncontaminated lavas, that crustal contamination occurred largely after the high-pressure stage of differentiation, and thus during ascent to the surface. In contrast, basalts in Scoresby Sund experienced extensive fractionation at depths <10 km that overprinted the earlier higher-pressure differentiation history.
|
Magmatic additions to the continental crust
The contrasting differentiation histories of the Kangerdlugssuaq Lower Basalts and the Scoresby Sund lavas have implications for the nature of magmatic additions to the crust along the East Greenland margin. Seismic surveys of the North Atlantic continental margins reveal the existence of sections of new igneous crust more than twice the thickness of normal oceanic crust (White et al., 1987
; Larsen & Jakobsdóttir, 1988; Mutter & Zehnder, 1988; Mutter et al., 1988; Zehnder et al., 1990). The upper 6 km typically comprise seaward-dipping reflectors and correspond to extrusive lavas. The thick, high seismic velocity (>7.3 km/s) lower-crustal sections have been interpreted as underplated basaltic material (e.g. Mutter & Zehnder, 1988; White & McKenzie, 1989; White, 1992).
Comparison between lava compositions and the polybaric fractionation models presented in Fig. 12b can be used to estimate the composition, proportion, and distribution of underplated or intraplated cumulate rocks. The two histograms in Fig. 13 give the modal composition of the cumulates predicted at depths of 20–25 km (8 kbar) and 4–6 km (2 kbar) for polybaric fractionation of primary liquid C. Liquid compositions containing 9–13 wt % MgO, comparable to the Kangerdlugssuaq Lower Basalts, are produced after 10–25% crystallization of primary liquid C involving olivine and olivine + clinopyroxene fractionation. Similarly, the production of evolved Fe–Ti basalts typical of the Scoresby Sund succession (5.50 wt % MgO) can be accounted for by 30% fractionation of olivine and olivine + clinopyroxene at high pressure, followed by an additional 20% crystallization of olivine, plagioclase, and clinopyroxene on decompression.
A corollary to these predictions is the addition of dunitic and wehrlitic cumulates to the lower crust, and troctolitic and gabbroic cumulates to the upper crust during development of the East Greenland volcanic margin (Fig. 13). Evidence for the latter is supported by the seismic properties of the sills imaged by Larsen & Marcussen (1992) in the Jameson Land basin, which compare favorably with troctolitic and gabbroic rocks at low confining pressure (Christensen, 1982). High-resolution crustal seismic studies now under way (SIGMA project) should help to establish if the predicted ultramafic cumulates reside within the crust or formed at the crust–mantle boundary.
Primary melts of North Atlantic basalts
The preceding discussions provided insights into the fractionation and contamination histories of East Greenland flood basalts, as well as the compositions of parental magmas (Table 7). Below we consider the deeper issues related to the conditions for mantle melt generation during continental breakup and proposed hotspot activity. Conceptually at least, one can invert primary liquid compositions to extract information about the composition of the mantle source region and the systematics of mantle melting. Practically however, such an exercise requires knowledge of the composition of primary melts, the melting behavior of mantle peridotite, and the physics of melt segregation. Recognizing obvious deficiencies in this understanding, it remains possible to relate many first-order characteristics of mantle-derived melts to the conditions for their formation (e.g. Klein & Langmuir, 1987; McKenzie & Bickle, 1988; Kinzler & Grove, 1992).
To understand the significance of the compositional range of East Greenland parental magmas in relation to plausible primary mantle melts, it is instructive to first compare these estimates with compositions inferred for basalts elsewhere in the North Atlantic region. Although the approach in this paper focuses on major element data, it is conceptually consistent with the trace element approach of Fram & Lesher (1993). Our treatment is based on published major element analyses for basalts from East Greenland (Table 4; Brooks et al., 1976; Nielsen et al., 1981; Brooks & Nielsen, 1982b; Gill et al., 1988; Larsen et al., 1989), Northeast Greenland (Thirlwall et al., 1994); the Faeroe Islands (Rasmussen & Noe-Nygaard, 1969; Waagstein & Hald, 1984), the British Tertiary (Thompson et al., 1972, 1980, 1986; Beckinsale et al., 1978; Scarrow & Cox, 1995), Southeast Greenland (Fitton et al., 1997; Larsen et al., 1997), and the modern Reykjanes, Iceland, Kolbeinsey, Mohns, and Knipovich segments of the Mid-Atlantic Ridge (Jakobsson et al., 1978; Wood et al., 1979; Schilling et al., 1983; Neumann & Schilling, 1984; Meyer et al., 1985; Hemond et al., 1993) (see Fig. 1 for locations). To compare magma types from these regions, the database is screened for primitive basalt compositions (defined by MgO > 8.5 wt %). This restriction to lavas with >8.5 wt % MgO identifies compositions that reasonably can be related to primary mantle melts by addition or subtraction of olivine.
The approach used in this analysis is analogous to the correction procedure we used above to identify the compositions of parental liquids for the Lower Basalts. We were careful in our previous discussions not to equate these estimates of parental magma compositions with ‘primary’ mantle melts. For example, if one assumes that the olivine–liquid 
is constant and fractionation-corrected compositions are compared at the same mg-number and considered representative of primary melts, we have then prescribed the composition of residual olivine in the mantle source. For example, in the studies of Albarède (1992), Fram & Lesher (1993), and Francis (1995) the composition of olivine in the mantle source was assumed to be Fo91, Fo89, and either Fo90 or Fo86, respectively. An alternative approach taken by Scarrow & Cox (1995) for basalts from the Isle of Skye is to fractionation-correct primitive lava compositions to a constant MgO content (i.e. 15 wt %). In this case, the composition of residual mantle olivine is not fixed, and variations in iron contents of corrected compositions could be attributed to differences in iron content of the mantle source composition. However, such variations may also reflect differences in the overall extent of melting and/or pressures for decompression melting. The correction to a constant MgO content is similar to the approach of Klein & Langmuir (1987), who referenced MORB compositions to 8 wt % MgO for comparative purposes. However, in the latter treatment compositions so defined were not assumed to be primary liquid compositions, but the major element composition of the MORB source was assumed to be uniform. In the present study, we cannot a priori assume that source mantle compositions for CFBs and ocean ridge magmas in the North Atlantic were similar, and therefore any fractionation correction must be scrutinized in an effort to separate source compositional effects from melting systematics. For this reason, we begin by directly comparing results obtained for both the MgO and mg-number normalization procedures.
Figure 14a gives a comparison of hypothetical primary melt compositions for East Greenland, modern Iceland, and the Reykjanes Ridge basalts for the assumptions of equivalent MgO content and mg-number. The thin continuous curves show the range of MgO and FeO contents for compositions A, B and C given in Table 7 with addition and subtraction of equilibrium olivine. As expected, variations in FeO contents are small compared with MgO. The filled circles on these curves represent primary liquid compositions in equilibrium with Fo91 [mg-number(liq) = 0.758], in equilibrium with Fo88 [mg-number(liq) = 0.70], and with 16 wt % MgO. For the Lower Basalts the difference between primary liquids compared at mg-number = 0.70 and those normalized to MgO = 16 wt % is small, whereas for many Icelandic and normal MORB compositions the discrepancy is much greater (Fig. 14a; Table 7). In fact, employing a correction to 16 wt % MgO for many Icelandic primitive basalts will grossly overestimate the mg-number of the mantle source. Similarly, for characteristically high-FeO primitive flood basalts this correction procedure could underestimate the mg-number of the source, and lead one to conclude that the source region for flood basalts is iron rich. Thus to avoid circularity, we choose not to consider the MgO normalization procedure further in this analysis. Instead, we compare primary liquids at fixed mg-number, but because mantle composition is one of the variables we are trying to investigate we further consider the effect of varying this parameter.
|
The trends defined by normalization of the full set of primitive Lower Basalts assuming mg-number = 0.758 and 0.70 are shown as stippled bands in Fig. 14a. The compositional range for Icelandic and Reykjanes Ridge primary melts are shown as filled and open bands, respectively. Primary melts estimated for Iceland generally have lower iron contents than estimated primary melts for the Lower Basalts, although primary melts for these regions, as well as the Reykjanes Ridge, overlap between 11 and 13 wt % FeO.
Figure 14b shows the range in FeO and MgO contents for estimated primary liquids for each of the regions included in our North Atlantic database at mg-numbers of 0.758 and 0.70. FeO contents vary from 7 to 14 wt % and MgO contents from 8 to 24 wt %. In general, basalts from the British Tertiary, Faeroe Islands, NE Greenland, Scoresby Sund, and Kangerdlugssuaq Lower Basalt successions have primary liquids with higher FeO contents than modern basalts.
Figure 15a presents the covariations of TiO2 contents and CaO/Al2O3 ratios for these same suites. For ease of portrayal we only include compositions corrected to mg-number = 0.70. For higher values of mg-number, the relative positions of the fields and absolute values of the CaO/Al2O3 ratios remain the same, but the TiO2 concentrations are shifted to lower values owing to dilution by olivine. TiO2 contents for the whole database show a factor of 8.8 variation (from 0.38 to 3.4 wt %), and CaO/Al2O3 ratios vary from 0.55 to 1.20 (Fig. 15a). For comparison, if the mg-number of primary liquids is assumed to be 0.76, the resulting range in TiO2 contents would be 0.35–3.0 wt %, or a factor of 8.5 variation. It is noteworthy here that estimated primary liquids for most of the Early Tertiary basalts, including the Kangerdlugssuaq Lower Basalts, define a positive correlation between TiO2 and CaO/Al2O3 in Fig. 15a, whereas modern basalts show the opposite trend.
|
Mantle melting systematics—principles
To relate estimated primary liquid compositions in Figs 14a, b and 15a to mantle melting systematics and composition, it is necessary to consider the effects of pressure, extent of melting, and residual mantle mineralogy on the composition of partial melts. It has long been recognized that the MgO and FeO contents of liquids derived by melting of peridotite increase as the pressure of melting increases (e.g. Jaques & Green, 1980
; Langmuir & Hanson, 1980). Figure 14c shows FeO and MgO contents of liquids produced in recent experiments using the diamond aggregrate segregation technique (Hirose & Kushiro, 1993; Baker & Stolper, 1994; Baker et al., 1995) and the near-liquidus, multiphase saturation approach (Kinzler & Grove, 1992; Longhi, 1995). We have not included experiments using the sandwich technique.
The experimental data presented in Fig. 14c are divided into groups according to pressure. The thin continuous lines contour near-solidus compositions at 10, 15, 20, and 30 kbar. Although there is considerable uncertainty in the exact position of these contours, owing to the paucity of experimental data at low extents of melting (Baker et al., 1995), MgO and FeO contents of near-solidus mantle melts are clearly shown to increase with pressure. The thick continuous lines in Fig. 14c contour the composition of olivine in equilibrium with experimental liquids. It should be noted that these contours are steeper than lines of constant mg-number for the liquid, which is consistent with an expected increase in olivine–liquid with pressure (e.g. Takahashi & Kushiro, 1983). Thus the line of mg-number = 0.70 intersects the contour for liquids in equilibrium with Fo88 olivine at 7.5 wt % FeO, whereas it intersects the Fo87 contour at 12 wt % FeO. This difference corresponds to an increase in from 0.31 to 0.35.
The utility of Fig. 14c is that contours of constant olivine composition can be viewed as proxies for mantle composition given that the mg-number of the source is nearly equivalent to the mg-number for the olivine. Thus, as expected intuitively, source mantle containing forsteritic olivine produces partial melts that, in general, have lower FeO and higher MgO contents than does mantle with more iron-rich olivine at comparable pressure and extent of melting. Moreover, Fig. 14c provides limits on the minimum MgO and FeO contents of primary melts generated with increasing pressure. However, absolute values for mantle mg-number and mean pressure of melting for a given suite of estimated primary liquids cannot be read directly from this diagram. Mantle melting is a fractional rather than an equilibrium process, and Langmuir et al. (1992) showed that an accumulated fractional melting model yields estimates for the mean pressure of melting based on FeO contents of partial melts that can be significantly higher than mean pressures calculated assuming equilibrium melting over the same melting interval.
To complement the regional comparison shown in Fig. 15a, we also attempted to systematize the available experimental data in terms of CaO/Al2O3 and TiO2. This proved difficult largely because of the wide range of TiO2 contents of experimental starting materials. Nevertheless, the results for individual experimental studies reveal systematic relationships between CaO/Al2O3, TiO2 and mantle mineralogy (Fig. 15b). For example, at 10 kbar Baker & Stolper (1994) and Baker et al. (1995) showed that over a significant range of melt fractions CaO/Al2O3 increases and TiO2 decreases during partial melting of spinel peridotite. At very low melt fractions TiO2 may behave compatibly; however, the highest CaO/Al2O3 ratio and lowest TiO2 content are found in liquids in equilibrium with a harzburgite residue (see Fig. 15b). In contrast, Longhi (1995) showed that liquids produced during partial melting of garnet peridotite at 28 kbar display the opposite behavior, with CaO/Al2O3 ratios dropping along with TiO2 as the extent of melting increases. This behavior at high pressure is attributed to the compatibility of Al2O3 in garnet and the high proportion of garnet (compared with spinel at lower pressures) entering the melt (e.g. Kinzler, 1992; Kinzler & Grove, 1992; Longhi, 1995). Thus melting of garnet peridotite results in the continual addition of Al2O3 to the liquid that apparently outpaces the addition of CaO by melting of clinopyroxene. If garnet is exhausted before clinopyroxene, the CaO/Al2O3 ratio is expected to then rise upon further melting, producing the V-shaped melting trajectory as shown in Fig. 15b. Although melting scenarios based on the isobaric experiments presented in Figs 14c and 15b cannot be directly applied to mantle melting occurring under fractional fusion and polybaric conditions, the gross effects of pressure, extent of melting, and source mineralogy on bulk composition of primary melts can be used in relating the compositions of North Atlantic primary melts to relative differences in mantle melting conditions from the early Tertiary to the present.
Mantle melting systematics—application to North Atlantic
Fram & Lesher (1993) examined the trace element compositions of North Atlantic primary liquids and noted that normalized TiO2 contents and HREE ratios generally decreased with time from the Early Tertiary CFBs, through the seaward-dipping reflector series (SDRS) basalts, to the modern North Atlantic MORBs. We rationalized this observation by considering mantle melting beneath a progressively thinning continental lithosphere. Decompression melting of the mantle beneath a thick lithosphere was confined to a restricted melting zone at greater depth where contributions from melting garnet lherzolite were not highly diluted by melting at lower pressures. The resulting pooled melts have high TiO2 contents and (Dy/Yb)N ratios. As the lithosphere thinned, mantle upwelling proceeded further, resulting in a greater mean extent (
) of melting at lower mean pressure (
) producing pooled liquids with lower TiO2 contents and (Dy/Yb)N ratios.
To a first order, this ‘lid-effect’ is apparent in the general decrease in FeO and MgO contents (and thus by inference ) from the CFB to the modern MORB estimated primary liquids shown in Fig. 14b. However, in detail, there are apparent contradictions in the data. For example, the FeO contents of primary melts of the Reykjanes Ridge and East Greenland overlap (Fig. 14a, b), suggesting that the two groups formed by melting over the same pressure interval. However, (Dy/Yb)N ratios in the Lower Basalts are consistently higher than those in the Reykjanes Ridge basalts, indicating higher for the Tertiary Lower Basalts, as expected for melt formation beneath a thick continental lithosphere (Fram & Lesher, 1993). This contradiction can be reconciled if primary liquids for the East Greenland flood basalts were derived at higher from mantle with higher mg-number than those for the Reykjanes Ridge.
Figure 16 summarizes schematically the effect of melting two different mantle sources influenced by a ‘lid-effect’ and a variation in mantle temperature. Melting of peridotite with higher mg-number beneath a progressively thinning lithospheric lid yields primary liquids with progressively decreasing FeO and MgO contents (melts 1–2–3–4; Fig. 16). Melting of peridotite with a lower mg-number beneath the same thinning lithosphere also results in primary liquids with progressively decreasing FeO and MgO, but the absolute values are displaced to higher FeO content (compare melts 3–4 with melts 9–8 in Fig. 16). Coupling the systematics portrayed for MgO-FeO in Fig. 16 with the relationships shown for North Atlantic primary melts in Fig. 14b suggests that primary melts for most basalts from Kangerdlugssuaq, Faeroe Islands, Scoresby Sund, and NE Greenland were on average derived from a mantle source with higher mg-number than elsewhere in the North Atlantic.
|
It is noteworthy that Francis (1995)
and Scarrow & Cox (1995) reached the opposite conclusion concerning the composition of the mantle feeding CFBs (i.e. the mantle source was iron rich). Scarrow & Cox (1995) showed that FeO and SiO2 contents of normalized compositions (constant MgO = 15 wt %) of British Tertiary basalts overlapped the compositions of liquids produced by melting iron-rich peridotite similar to HK-66 (mg-number = 0.855), whereas they were notably different from melts derived from more magnesium-rich peridotite such as KLB-1 (mg-number = 0.897; Hirose & Kushiro, 1993). We have already discussed the problems inherent in comparing normalized primary liquid compositions directly with experimental liquids. It is also possible that the normalized compositions have high SiO2 rather than high FeO compared with the KLB-1 liquids, which could be related to mantle melting under hydrous conditions (Kushiro, 1972, 1990; Gallagher & Hawkesworth, 1992). Granted, there is no a priori reason to conclude that the mantle source involved in the anomalous melting event at the initiation of spreading in the Early Tertiary North Atlantic was hydrous or anhydrous.
The FeO and MgO contents of primary mantle melts are also directly linked to mantle temperature, which controls the depth at which upwelling mantle intersects its solidus. The trend defined by compositions 5–6–7–8 in Fig. 16 shows schematically the effect of increasing mantle temperature on primary melt composition. Klein & Langmuir (1987) showed that the compositions of basalts from ridge segments of the North Atlantic were consistent with an increase in mantle temperature, and mean pressure of melting, approaching Iceland. We can see this trend in Fig. 14b for estimated primary liquids from the Knipovich, Mohns and Reykjanes ridges. The Kolbeinsey segment is not well behaved in this regard, whereas Iceland covers the full range shown for the modern ridge segments.
In light of our suggestion that the mantle beneath the North Atlantic is heterogeneous with respect to the forsterite content of olivine, we must further consider the possibility that it is also heterogeneous with respect to the contents of TiO2 and other incompatible elements. TiO2 contents of North Atlantic estimated primary liquids vary by a factor of 8 (Fig. 15a). Such a large range is difficult to achieve without appealing to very small extents of melting (<1%) for the highest TiO2 compositions. Indeed, the melting model of Fram & Lesher (1993), which used a single TiO2 content for the starting mantle, was unable to reproduce the full range of TiO2 contents observed in the North Atlantic database. To reduce the range in implied by the TiO2 contents of the estimated primary liquids, one must conclude that the Early Tertiary CFBs (i.e. most of the Lower Basalts, Scoresby Sund, Faeroe Islands, NE Greenland) were derived from a TiO2-rich source compared with the mantle source of modern North Atlantic MORBs. In this respect, although Knipovich Ridge primary liquids have lower TiO2 contents than Lower Basalt primary liquids (Fig. 15a), they need not represent higher degrees of melting.
In the melting scenarios outlined in Fig. 16, is proportional to the length of the melting zone, whereas TiO2 is dependent on both the TiO2 content of the source mantle and the length of the melting zone. A large degree of melting of a high TiO2 mantle (melt 4; Fig. 16) will yield a liquid with approximately the same TiO2 content as a smaller degree of melting of a low TiO2 mantle (melt 6; Fig. 16). The relative decrease in TiO2 contents upon increasing owing to thinning lithosphere can be seen in the pattern of TiO2 contents of melts 1–2–3–4 in Fig. 16. This pattern resembles the relationship between the TiO2 contents of primary liquids of the alkaline (high TiO2) and tholeiitic (low TiO2) Lower Basalts and several of the other Early Tertiary sequences (Fig. 15a). The relative decrease in TiO2 contents upon increasing owing to increased mantle temperature can be seen in the pattern of TiO2 contents of melts 5–6–7–8 in Fig. 16. This pattern resembles the relationship between the TiO2 contents of primary liquids of the Knipovich, Mohns, Reykjanes, and Kolbeinsey Ridge basalts (Fig. 15a).
Recalling Fig. 15b, the CaO/Al2O3–TiO2 relations shown by North Atlantic primary melts suggest that the general systematics may be related to the contributions of melts formed at high pressure and within the stability field of garnet lherzolite. Garnet, rather than spinel, is the aluminous phase present in peridotite at pressures >20–30 kbar (Takahashi & Kushiro, 1983; Takahashi, 1986; Falloon & Green, 1988; Kinzler, 1992). As shown in Fig. 15b, melting at pressures high enough (and extents of melting low enough) for garnet stability yields liquids with decreasing CaO/Al2O3 ratio as increases (i.e. TiO2 decreases). We have assumed, however, that the melting effect on CaO/Al2O3 ratio is larger than any reasonable bulk compositional effect.
Thus by combining the three variables controlling mantle melting, i.e. the thickness of the lithosphere, the mantle temperature, and the mantle composition, we can construct an internally consistent explanation for the first-order systematics displayed for primary liquids for North Atlantic basalts in Figs 14a, b and 15a. We propose that primary liquids for the Kangerdlugssuaq Lower Basalts, NE Greenland, and the main portion of the Faeroe Island and Scoresby Sund sequences formed by melting of a high-TiO2, high-mg-number mantle, with the melting systematics governed primarily by the lid-effect (schematically portrayed by the sequence 1–2–3 in Fig. 16). Primary liquids for the Scoresby Sund and Faeroe Islands basalts that lie on the opposite limb of the V-shaped array in Fig. 15a correspond to the MORB-type basalts of the Rømer Fjord Formation of Scoresby Sund and the Upper Series of the Faeroes. We suggest that these liquids record introduction of MORB-type mantle into the melting zone. Schilling & Noe-Nygaard (1974) and Gariépy et al. (1983) similarly proposed involvement of a second mantle component to explain the abrupt transition in the REE patterns of Faeroe Islands basalts between the Middle and Upper Series.
As previously suggested, primary liquids for the Knipovich, Mohns, Reykjanes, and Kolbeinsey Ridge basalts formed by melting of the low-TiO2, low-mg-number mantle, with the melting systematics primarily governed by mantle temperature variations (melts 5–6–7–8; Fig. 16). By virtue of being the MORB source mantle, this low-TiO2, low-mg-number mantle end-member is inferred to have mg-number = 0.890 (e.g. Langmuir et al., 1992).
Primary liquid compositions for the SE Greenland basalts have both low CaO/Al2O3 ratios and low TiO2 contents relative to the primary liquids for the continental Early Tertiary sequences (Fig. 15a). At first glance, this suggests relatively high extents of melting of the same mantle source as for the other Early Tertiary sequences (i.e. melt 3; Fig. 16). However, TiO2 contents for the SE Greenland primary liquids extend to values as low as 0.75 wt %, which is a factor of 4.5 lower than TiO2 contents of the primary liquids for alkaline lavas in the other Early Tertiary successions. If the alkaline primary liquids represent mean melt fractions of 0.02–0.03 then the primary liquids with TiO2 = 0.75 wt % must represent an extent of melting of at least 0.10 and perhaps as great as 0.16, depending on assumptions for partition coefficients for TiO2 between mantle phases and the liquid. At these relatively large extents of melting, garnet may be completely exhausted owing to melting, and thus, the CaO/Al2O3 ratio would be expected to be high (i.e. melt 4; Fig. 16).
Alternatively, we can explain the range of TiO2 contents at low CaO/Al2O3 ratios shown by the SE Greenland suite by also postulating involvement of a low-TiO2 mantle source. Melting of both the low- and high-TiO2 mantles together under similar conditions of lithospheric thickness and mantle temperature (i.e. melts 3 and 9; Fig. 16) could produce the observed variation. This conclusion is consistent with the findings of Fram et al. (1997) showing that incompatible trace element ratios and concentrations in primary liquids for basalts from the ODP Leg 152 Site 917 Upper Series and from Sites 915 and 918 on the SE Greenland margin imply derivation of the basalts from two mantle sources, one depleted in incompatible elements compared with the other, undergoing decompression melting under similar conditions of lithospheric thickness and mantle temperature. It is interesting to note that volcanism in SE Greenland begins by tapping the more incompatible element depleted mantle (i.e. lower La/Sm and Ba/Zr), whereas the less incompatible element depleted mantle becomes important late in the rifting history (Fitton et al., 1997; Fram et al., 1997).
Compositions of primary liquids for British Tertiary basalts are difficult to explain using the scenario outlined in Fig. 16. Based on inferences from FeO–MgO relations (Fig. 14c) and HREE fractionations (Fram & Lesher, 1993), the CaO/Al2O3–TiO2 relations should more closely resemble those of the Faeroe Islands primary liquids. At present, we have no satisfactory explanation for this discrepancy.
The ancestral Iceland plume?
In this final section we return to the Lower Basalts and their connection to the ancestral Iceland plume. Based on major element data we postulate that two end-member mantle compositions were involved in North Atlantic basalt generation: a relatively high-TiO2, high-mg-number mantle and a relatively low-TiO2, low-mg-number mantle (the MORB source).
Modern Icelandic volcanism clearly taps a mantle source region with heterogeneous Sm/Nd ratios (e.g. Schilling & Noe-Nygaard, 1974; Zindler et al., 1979; Schilling et al., 1982; Hemond et al., 1993; Fig. 17). To determine whether Early Tertiary volcanism at Kangerdlugssuaq tapped a similarly heterogeneous source, we must first assess the consequences of both mantle composition and the extent of mantle melting on the Sm/Nd ratios of basalts. Figure 17 shows Nd isotopic data for modern Icelandic lavas with mg-number > 0.35 and 
18O > +5 from Hemond et al. (1993). To compare the modern Icelandic lavas with the Kangerdlugssuaq Lower Basalts, 143Nd/144Nd ratios in the Icelandic lavas are adjusted for decay of 147Sm in the mantle source over the last 55 Myr (Class et al., 1993). The source region will have Sm/Nd greater than or equal to the Sm/Nd of the lavas owing to the higher compatibility of Sm during partial melting.
|
The shaded field in Fig. 17 represents the compositions Icelandic lavas would have had if they had erupted 55 Myr ago. 143Nd/144Nd at 55 Myr ago are calculated assuming that measured Sm/Nd ratios of the lavas represent minimum possible values for the Sm/Nd of the source mantle. Lower Basalt samples with the highest 143Nd/144Nd fall within the range expected for the hypothetical ancestral Iceland mantle. This link between the Nd (and Sr) isotopic ratios of Early Tertiary East Greenland and modern Icelandic lavas was first recognized by Carter et al. (1979)
, and later noted by Holm et al. (1992). The bands at the top of Fig. 17 give the range in Sm/Nd ratios for 0–25% partial melting of mantle with initial Sm/Nd ratios of 0.322 and 0.440. An Sm/Nd ratio of 0.322 corresponds to a flat chondrite-normalized LREE pattern, whereas an Sm/Nd ratio of 0.440 corresponds to an LREE depleted chondrite-normalized pattern. To produce the full range of Sm/Nd observed in Icelandic lavas, mantle with Sm/Nd ranging from 0.25 to 0.44 is required.
Based on modeling of HREE ratios and TiO2 contents of liquids during partial melting, Fram & Lesher (1993) inferred extents of melting of 8–14% for the Kangerdlugssuaq Lower Basalts. Melting of mantle with Sm/Nd = 0.322 produces liquids with Sm/Nd ratios equal to those of the Lower Basalts (Sm/Nd = 0.3) at 10% melting (Fig. 17). Melting of mantle with Sm/Nd greater than 0.35 could not produce liquids with Sm/Nd ratios appropriate for the Lower Basalts, except by <1% partial melting, which is inconsistent with the major element compositions. The range in Sm/Nd of the Lower Basalts with high 143Nd/144Nd suggests mantle having Sm/Nd ratios <0.322 was also involved. Contamination by continental crust accounts for the low Sm/Nd ratios of lavas with 143Nd/144Nd below 0.5128. The range in Sm/Nd of the Lower Basalts with 143Nd/144Nd falling in the ‘55 Myr Iceland’ field implies that the Early Tertiary volcanism in Kangerdlugssuaq tapped mantle with Sm/Nd ratios ranging from a maximum of 0.35 to a minimum of 0.25. These Sm/Nd ratios correspond to rare earth patterns ranging from slightly LREE depleted to slightly LREE enriched relative to chondrites. There is no indication of involvement of mantle as LREE depleted (Sm/Nd = 0.35–0.44) as the mantle tapped for modern Icelandic picrites (Zindler et al., 1979; Hemond et al., 1993).
The relationships shown in Fig. 17 suggest that the relatively high-mg-number, high-TiO2 end-member mantle composition has lower Sm/Nd and slightly less long-term LREE depletion than the relatively low-mg-number, low-TiO2 end-member mantle composition. These compositional features have interesting implications for the prior history of the mantle, particularly of the high-mg-number, high-TiO2 end-member. High mg-number implies that the mantle was previously depleted of FeO and other fusable components by melt extraction. The long-term LREE depletion reflected in Nd isotopes is consistent with this being an ancient melt extraction event, yet the Sm/Nd ratio inferred for this mantle component from the basalts is nearly chondritic (Fig. 17). This implies that the mantle underwent a relatively recent (metasomatic) enrichment event that preferentially added LREEs, lowering the Sm/Nd ratio. Interestingly, this decrease in Sm/Nd ratio is correlated with an increase in TiO2 content, suggesting that LREE and TiO2 may have been added by the same metasomatic agent. TiO2-rich oxide phases are indeed found in some metasomatized mantle xenoliths (e.g. Roden & Murthy, 1985; Nixon, 1987).
The two end-member mantle components for North Atlantic basalts identified in the preceding analysis both represent long-term ‘depleted’ mantle sources with inferred low Rb/Sr and high Sm/Nd ratios giving rise to their low 87Sr/86Sr and high 143Nd/144Nd ratios relative to primitive mantle and Bulk Earth. We have already related one end-member to the source mantle for MORBs (e.g. Zindler & Hart, 1986). This end-member has a lower mg-number, lower TiO2, and a higher Sm/Nd ratio than the second source component, which we tentatively associated with the ancestral Iceland plume. Regardless of the ultimate origin of this latter mantle component, it has geochemical characteristics inferred for recycled lithosphere and is the dominant source for the East Greenland flood basalts.
Implications for plume structure
The spatial and temporal distribution of basalts derived from the two end-member mantle components has implications for the structure of the Iceland anomaly and its role in Early Tertiary volcanism in the North Atlantic region. As stated above, we correlate the high-mg-number, high-TiO2, low-Sm/Nd component with the Iceland compositional anomaly, and the low-mg-number, low-TiO2, high-Sm/Nd component with the MORB source mantle. Early Tertiary regions relatively close to the ends of the Greenland–Iceland–Faeroes Ridge (Kangerdlugssuaq Lower Basalts, Scoresby Sund, Faeroe Islands) are dominated by the ‘Iceland’ component. This is consistent with predictions of plume models that suggest the source region for volcanism near the center of the plume axis at the initiation of rifting is dominantly plume mantle material (Richards et al., 1989; Campbell & Griffiths, 1990).
Our analysis and that of Thirlwall et al. (1994) suggest that the source region for basalts in NE Greenland was also compositionally similar to ‘Iceland plume’ mantle. This implies extensive lateral distribution of the ‘Iceland’ component in the Early Tertiary. It has been suggested that the plume axis was located (1) beneath Kangerdlugssuaq at the time of volcanism (e.g. White & McKenzie, 1989), or (2) 400–600 km to the northwest or west under Greenland (Duncan & Richards, 1991; Lawver & Muller, 1994). In either case, NE Greenland eruption sites were remote from the presumed plume axis. The Early Tertiary sequence in SE Greenland, far to the south of the plume axis, is dominated by the ‘MORB’ component at the initiation of volcanism, with the ‘Iceland’ component providing a greater contribution once seafloor spreading initiated.
Although we have grouped all of the Early Tertiary volcanism into one time period for ease of discussion, in detail, volcanism in SE Greenland probably preceded volcanism in NE Greenland by a few million years (Upton et al., 1995; Sinton & Duncan, 1997). This suggests a time delay between the initiation of volcanism and the flow of mantle material from axial to distal locations. If mantle flow was radially symmetric from the plume axis, then the SE Greenland volcanism may have captured an earlier stage in the evolution of the plume structure than recorded in NE Greenland volcanism. This underscores the importance of linking the history of continental rifting and mantle flow in the starting plume to the timing of volcanism as emphasized by Hill (1991). Nevertheless, the widespread compositional anomaly observed in the Early Tertiary basalts is apparently a transient feature in view of the appearance of basalt with MORB characteristics high in the volcanic successions on the Faeroe Islands and in the Scoresby Sund region.
|
|
|
|
|
errors. If no error bar is shown, then it is smaller than the size of the symbol. All Nd and Sr isotopic ratios are corrected for decay of 147Sm and 87Rb in the 55 My since the lavas erupted. The eight whole-rock samples and vesicle-free separates of rock matrix from CL89-7eI, -7eV, and -7eVII were analyzed. Separates were made by breaking 1–2 cm rock chunks into 2–4 mm chips with a small chisel to isolate vesicle-free portions of the rock matrix. Approximately 30 g of chips were cleaned and powdered by the procedures described above. In addition, tube vesicles from CL89-7eI were chiseled out and analyzed. Major, trace, and rare earth element analyses for these samples are presented in |
Acknowledgements |
|---|
Troels Nielsen of the Geological Survey of Greenland and Kent Brooks of the University of Copenhagen kindly provided additional samples of the Lower Basalts. Discussions over the years with John Longhi, Dan Miller, Ro Kinzler, Jeff Rosenbaum, Henriette Hansen, Lotte Larsen, Troels Nielsen, Louise Kellogg, and Peter Thy have helped shape the ideas presented in this paper. We are grateful for their interest and absolve them of any responsibility for deficiencies that remain. We thank Keith Cox, Godfrey Fitton, Andy Saunders, and Marge Wilson for their thorough and insightful reviews, and for challenging us to integrate our work on the generation and differentiation of North Atlantic magmas. Support for this research has been provided by NSF Grants EAR91-17239 and EAR94-19382 to C.E.L. Additional support for field-work was provided in 1989 by an NSF Graduate Fellowship to M.S.F., and in 1991 by the Department of Geological Sciences, Columbia University, and the Danish Natural Science Research Council (to Kent Brooks and Ella Hoch, University of Copenhagen). M.S.F. received support from a University of California Presidential Postdoctoral Fellowship during preparation of this manuscript. Support from the Danish Lithosphere Center during completion of this study is also gratefully acknowledged by C.E.L.
* Corresponding author. Telephone: 916-752-0350. Fax: 916-752-0951. e-mail: fram{at}geology.ucdavis.edu
|
References |
|---|
|
TOPABSTRACTIntroductionGeology and Stratigraphic...Analytical MethodsGeochemical and Petrographic...Data Analysis and DiscussionReferences |
|---|
Albarède F. How deep do common basaltic magmas form and differentiate? Journal of Geophysical Research (1992) 97:10997–11009.
Arndt N. T., Christensen U. The role of lithospheric mantle in continental flood magmatism: thermal and geochemical constraints. Journal of Geophysical Research (1992) 97:10967–10982.
Baker D. R., Eggler D. H. Compositions of anhydrous and hydrous melts coexisting with plagioclase, augite, and olivine or low-Ca pyroxene from 1 atm to 8 kbar: application to the Aleutian volcanic center of Atka. American Mineralogist (1987) 72:12–28.[Abstract]
Baker M. B., Stolper E. M. Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta (1994) 58:2811–2827.[Web of Science]
Baker M. B., Hirschmann M. M., Ghiorso M. S., Stolper E. M. Compositions of near-solidus peridotite melts from experiments and thermodynamic calculations. Nature (1995) 375:308–311.
Beattie P., Ford C., Russell D. Partition coefficients for olivine–melt and orthopyroxene-melt systems. Contributions to Mineralogy and Petrology (1991) 109:212–224.[Web of Science]
Beckinsale R. D., Pankhurst R. J., Skelhorn R. R., Walsh J. N. Geochemistry and petrogenesis of the early Tertiary lava pile of the Isle of Mull, Scotland. Contributions to Mineralogy and Petrology (1978) 66:415–427.[Web of Science]
Bender J. F., Hodges F. N., Bence A. E. Petrogenesis of basalts from the project FAMOUS area: experimental study from 0 to 15 kbars. Earth and Planetary Science Letters (1978) 41:277–302.[Web of Science]
Bernstein S. High-pressure fractionation in rift-related basaltic magmatism: Faeroe plateau basalts. Geology (1994) 22:815–818.
Bird D. K., Rosing M. T., Manning C. E., Rose N. M. Geologic field studies of the Miki Fjord area, East Greenland. Bulletin of the Geological Society of Denmark (1985) 34:219–236.
Blichert-Toft J., Lesher C. E., Rosing M. T. Selectively contaminated magmas of the Tertiary East Greenland macrodike complex. Contributions to Mineralogy and Petrology (1992) 110:154–172.[Web of Science]
Brooks C. K., Nielsen T. F. D. The E.Greenland continental margin: a transition between oceanic and continental magmatism. Journal of the Geological Society, London (1982a) 139:265–275.
Brooks C. K., Nielsen T. F. D. The Phanerozoic development of the Kangerdlugssuaq area, East Greenland. Meddelelser om Grønland, Geoscience (1982b) 9:30.
Brooks C. K., Nielsen T. F. D., Petersen T. S. The Blosseville Coast basalts of East Greenland: their occurrence, composition and temporal variations. Contributions to Mineralogy and Petrology (1976) 58:279–292.[Web of Science]
Campbell I. H., Griffiths R. W. Implications of mantle plume structure for the evolution of flood basalts. Earth and Planetary Science Letters (1990) 99:79–93.[Web of Science]
Carter S. R., Evensen N. M., Hamilton P. J., O'Nions R. K. Basalt magma sources during the opening of the North Atlantic. Nature (1979) 281:28–30.[Web of Science]
Christensen N. I. Seismic velocities. In: Handbook of Physical Properties of Rocks, Vol. II—Carmichael R. S., ed. (1982) Boca Raton, FL: CRC Press. 1–228.
Christie D. M., Carmichael I. S. E., Langmuir C. H. Oxidation states of mid-ocean ridge basalt glasses. Earth and Planetary Science Letters (1986) 79:397–411.[Web of Science]
Clarke D. B. Tertiary basalts of Baffin Bay: possible primary magmas from the mantle. Contributions to Mineralogy and Petrology (1970) 25:203–224.
Class C., Goldstein S. L., Galer S. J. G., Weis D. Young formation age of a mantle plume source. Nature (1993) 362:715–721.
Cox K. G. A model for flood basalt vulcanism. Journal of Petrology (1980) 21:629–650.
Cox K. G. The Karoo Province. In: Continental Flood Basalts—MacDougall J. D., ed. (1988) Boston, MA: Kluwer Academic. 239–271.
Cox K. G., Jamieson B. G. The olivine–rich lavas of Nuanetsi: a study of polybaric magmatic evolution. Journal of Petrology (1974) 15:269–301.
Cox K. G., Duncan A. R., Bristow J. W., Taylor S. R., Erlank A. J. Petrogenesis of the basic rocks of the Lebombo. In: Petrogenesis of the Volcanic Rocks of the Karoo Province. Geological Society of South Africa, Special Publication—Erlank A. J., ed. (1984) 13:149–169.
Dickin A. P. Isotope geochemistry of Tertiary igneous rocks from the Isle of Skye, N.W. Scotland. Journal of Petrology (1981) 22:155–189.
Dickin A. P., Jones N. W., Thirwall M. F., Thompson R. N. A Ce/Nd isotope study of crustal contamination processes affecting Palaeocene magmas in Skye, northwest Scotland. Contributions to Mineralogy and Petrology (1987) 96:455–464.[Web of Science]
Duncan R. A., Richards M. A. Hotspots, mantle plumes, flood basalts, and true polar wander. Reviews of Geophysics (1991) 29:31–50.[Web of Science]
Ellam R. M., Cox K. G. An interpretation of Karoo picrite basalts in terms of interaction between asthenospheric magmas and the mantle lithosphere. Earth and Planetary Science Letters (1991) 105:330–342.[Web of Science]
Elliott T. R., Hawkesworth C. J., Grönvold K. Dynamic melting of the Iceland plume. Nature (1991) 351:201–206.
Elthon D., Scarfe C. M. High-pressure phase equilibria of a high-magnesia basalt and the genesis of primary oceanic basalts. American Mineralogist (1984) 69:1–15.[Abstract]
Escher A., Watt W. S., eds. Geology of Greenland (1976) Copenhagen: Geological Survey of Greenland.
Falloon T. J., Green D. H. Anhydrous partial melting of peridotite from 8 to 35 kbar and the petrogenesis of MORB. Journal of Petrology, Special Lithosphere Issue (1988) 379–414.
Fitton J. G., Saunders A. D., Larsen L. M., Hardarson B. S., Norry M. J. Volcanic rocks from the East Greenland margin at 63°N: composition, petrogenesis and mantle sources. In: Proceedings of the Ocean Drilling Program, Scientific Results, 152—Larsen H. C., Saunders A. D., Wise S., eds. (1997) College Station, TX: Ocean Drilling Program. in press.
Fram M. S. Petrogenesis of the Early Tertiary flood basalts of the East Greenland continental margin. (1994) New York: Columbia University. Ph.D. Thesis.
Fram M. S., Lesher C. E. Geochemical constraints on mantle melting during creation of the North Atlantic basin. Nature (1993) 363:712–715.
Fram M. S., Lesher C. E., Volpe A. M. Mantle melting systematics: the transition from continental to oceanic volcanism on the southeast Greenland margin. In: Proceedings of the Ocean Drilling Program, Scientific Results, 152—Larsen H. C., Saunders A. D., Wise S., eds. (1997) College Station, TX: Ocean Drilling Program. in press.
Francis D. The implications of picritic lavas for the mantle sources of terrestrial volcanism. Lithos (1995) 34:89–105.[Web of Science]
Fujii T., Bougault H. Melting relations of a magnesian abyssal tholeiite and the origin of MORBs. Earth and Planetary Science Letters (1983) 62:283–295.[Web of Science]
Gallagher K., Hawkesworth C. Dehydration melting and the generation of continental flood basalts. Nature (1992) 358:57–59.
Gallahan W. E., Nielsen R. L. The partitioning of Sc, Y, and the rare earth elements between high-Ca pyroxene and natural mafic to intermediate lavas at 1 atmosphere. Geochimica et Cosmochimica Acta (1992) 56:2387–2404.[Web of Science]
Gariépy C., Ludden J., Brooks C. Isotopic and trace element constraints on the genesis of the Faeroe lava pile. Earth and Planetary Science Letters (1983) 63:257–272.[Web of Science]
Geist D., White C. Assimilation and fractionation in adjacent parts of the same magma chamber: Vandfaldsdalen macrodike, East Greenland. Contributions to Mineralogy and Petrology (1994) 116:92–107.[Web of Science]
GGU. Geological map of Greenland, Sheet 13, Kangerdlugssuaq (1988) Copenhagen: Geodætisk Institut.
Gill R. C. O., Nielsen T. F. D., Brooks C. K., Ingram G. A. Tertiary volcanism in the Kangerdlugssuaq region, E. Greenland: trace-element geochemistry of the Lower Basalts and tholeiitic dyke swarms. In: Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society Special Publication—Morton A. C., Parson L. M., eds. (1988) 39:161–179.
Hansen H. The lowermost basalts and volcanogenic sediments. In: Kangerdlugssuaq Studies: Processes at a Rifted Continental Margin—Brooks C. K., ed. (1990) Copenhagen: Geologisk Centralinstitut. 35–37.
Hanson G. N. Rare earth elements in petrogenetic studies of igneous systems. Annual Review of Earth and Planetary Science (1980) 8:371–406.
Hart S., Brooks C. The geochemistry and evolution of early Precambrian mantle. Contributions to Mineralogy and Petrology (1977) 61:109–128.[Web of Science]
Hemond C., Arndt N. T., Lichtenstein U., Hofmann A. W. The heterogeneous Iceland plume: Nd–Sr–O isotopes and trace element constraints. Journal of Geophysical Research (1993) 98:15833–15850.
Henriksen N., Higgins A. K. East Greenland Caledonian fold belt. In: Geology of Greenland—Escher A., Watt W. S., eds. (1976) Copenhagen: Geological Survey of Greenland. 182–247.
Hergt J. M., Peate D. W., Hawkesworth C. J. The petrogenesis of Mesozoic Gondwana low-Ti flood basalts. Earth and Planetary Science Letters (1991) 105:134–148.[Web of Science]
Hess P. C. Phase equilibria constraints on the origin of ocean floor basalts. In: Mantle Flow and Melt Generation at Mid-Ocean Ridges. Geophysical Monograph, American Geophysical Union—Morgan J. P., Blackmun D. K., Sinton J. M., eds. (1992) 71:67–102.
Hill R. I. Starting plumes and continental break-up. Earth and Planetary Science Letters (1991) 104:398–416.[Web of Science]
Hirose K., Kushiro I. Partial melting of dry peridotites at high pressures: determination of compositions of melts segregated from peridotite using aggregates of diamond. Earth and Planetary Science Letters (1993) 114:477–489.[Web of Science]
Holm P. M. Nd, Sr and Pb isotope geochemistry of the Lower Lavas, E. Greenland Tertiary Igneous Province. In: Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society Special Publication—Morton A. C., Parson L. M., eds. (1988) 39:181–195.
Holm P. M., Hald N., Nielsen T. F. D. Contrasts in composition and evolution of Tertiary CFBs between West and East Greenland and their relations to the establishment of the Icelandic mantle plume. In: Magmatism and the Causes of Continental Break-up. Geological Society Special Publication—Storey B. C., Alabaster T., Pankhurst R. J., eds. (1992) 68:349–362.
Jakobsson S. P., Jónsson J., Shido F. Petrology of the western Reykjanes Peninsula, Iceland. Journal of Petrology (1978) 19:669–705.
Jaques A. L., Green D. H. Anhydrous melting of peridotite at 0–15 kb pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology (1980) 73:287–310.[Web of Science]
Kays M. A., McBirney A. R., Goles G. G. Xenoliths of gneisses and the conformable, clot-like granophyres in the Marginal Border Group, Skaergaard intrusion, East Greenland. Contributions to Mineralogy and Petrology (1981) 76:265–284.[Web of Science]
Kays M. A., Goles G. G., Grover T. W. Precambrian sequence bordering the Skaergaard intrusion. Journal of Petrology (1989) 30:321–361.
Kinzler R. J. Mantle melting processes at the spinel–garnet transition (17–21 kb) (abstract). EOS, Transactions, American Geophysical Union (1992) 73:615.
Kinzler R. J., Grove T. L. Primary magmas of mid-ocean ridge basalts 1. Experiments and methods. Journal of Geophysical Research (1992) 97:6885–6906.
Klein E. M., Langmuir C. H. Global correlations of ocean ridge basalt chemistry with axial depth and crustal thickness. Journal of Geophysical Research (1987) 92:8089–8115.
Klein E. M., Langmuir C. H., Staudigel H. Geochemistry of basalts from the Southeast Indian Ridge, 115°E-138°E. Journal of Geophysical Research (1991) 96:2089–2107.
Krishnamurthy P., Cox K. G. Picrite basalts and related lavas from the Deccan Traps of western India. Contributions to Mineralogy and Petrology (1977) 62:53–75.[Web of Science]
Kushiro I. Effect of water on the composition of magmas formed at high pressures. Journal of Petrology (1972) 13:311–334.
Kushiro I. Partial melting of mantle wedge and evolution of island arc crust. Journal of Geophysical Research (1990) 95:15929–15939.
Langmuir C. H., Hanson G. N. An evaluation of major element heterogeneity in the mantle sources of basalts. Philosophical Transactions of the Royal Society of London, Series A (1980) 297:383–407.
Langmuir C. H., Klein E. M., Plank T. Petrological systematics of mid-ocean ridge basalts: constraints on melt generation beneath ocean ridges. In: Mantle Flow and Melt Generation at Mid-Ocean Ridges. Geophysical Monograph, American Geophysical Union—Morgan J. P., Blackmun D. K., Sinton J. M., eds. (1992) 71:183–280.
Larsen H. C., Jakobsdóttir S. Distribution, crustal properties and significance of seawards-dipping sub-basement reflectors off E. Greenland. In: Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society Special Publication—Morton A. C., Parson L. M., eds. (1988) 39:95–114.
Larsen H. C., Marcussen C. Sill-intrusion, flood basalt emplacement and deep crustal structure of the Scoresby Sund Region, East Greenland. In: Magmatism and the Causes of Continental Break-up. Geological Society Special Publication—Storey B. C., Alabaster T., Pankhurst R. J., eds. (1992) 68:365–386.
Larsen H. C., Saunders A. D., Clift P. D., et al. Proceedings of the Ocean Drilling Program, Initial Reports, 152 (1994) College Station, TX: Ocean Drilling Program.
Larsen L. M., Watt W. S., Watt M. Geology and petrology of the lower Tertiary plateau basalts of the Scoresby Sund region, East Greenland. Grønlands Geologiske Undersøgelse Bulletin (1989) 157:164.
Larsen L. M., Fitton J. G., Fram M. S. An overview and comparison of basaltic magma compositions in the North Atlantic Tertiary. In: Proceedings of the Ocean Drilling Program, Scientific Results, 152—Larsen H. C., Saunders A. D., Wise S., eds. (1997) College Station, TX: Ocean Drilling Program. in press.
Lawver L. A., Muller R. D. Iceland hotspot track. Geology (1994) 22:311–314.
Leeman W. P., Dasch E. J. Strontium, lead and oxygen isotopic investigation of the Skaergaard intrusion, East Greenland. Earth and Planetary Science Letters (1978) 41:47–59.[Web of Science]
Leeman W. P., Dasch E. J., Kays M. A. 207Pb/206Pb whole-rock age of gneisses from the Kangerdlugssuaq area, eastern Greenland. Nature (1976) 263:469–471.
Lindsley D. H., Andersen D. J. A two-pyroxene thermometer. In: Proceedings of the 13th Lunar and Planetary Science Conference. Journal of Geophysical Research (1983) 88:A887–A906. Supplement.
Longhi J. Liquidus equilibria and solid solution in the system Anorthite–Forsterite–Wollastonite–Silica at low pressure. American Journal of Science (1987) 287:265–331.
Longhi J. Comparative liquidus equilibria of hypersthene-normative basalts at low pressure. American Mineralogist (1991) 76:785–800.[Abstract]
Longhi J. Liquidus equilibria of some primary lunar and terrestrial melts in the garnet stability field. Geochimica et Cosmochimica Acta (1995) 59:2375–2386.[Web of Science]
Longhi J., Pan V. A reconnaisance study of phase boundaries in low-alkali basaltic liquids. Journal of Petrology (1988) 29:115–147.
McKenzie D., Bickle M. J. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology (1988) 29:625–679.
Mertz D. F., Devey C. W., Todt W., Stoffers P., Hofmann A. W. Sr–Nd–Pb isotope evidence against plume–asthenosphere mixing north of Iceland. Earth and Planetary Science Letters (1991) 107:243–255.[Web of Science]
Meyer P. S., Sigurdsson H., Schilling J.-G. Petrological and geochemical variations along Iceland's neovolcanic zones. Journal of Geophysical Research (1985) 90:10043–10072.
Miller D. M., Langmuir C. H., Goldstein S. L., Franks A. L. The importance of parental magma composition to calc-alkaline and tholeiitic evolution: evidence form Umnak Island in the Aleutians. Journal of Geophysical Research (1992) 97:321–343.
Moorbath S., Thompson R. N. Strontium isotope geochemistry and petrogenesis of the early Tertiary lava pile of the Isle of Skye, Scotland, and other basic rocks of the British Tertiary Province: an example of magma–crust interaction. Journal of Petrology (1980) 21:295–321.
Morrison M. A., Thompson R. N., Dickin A. P. Geochemical evidence for complex magmatic plumbing during development of a continental volcanic center. Geology (1985) 13:581–584.
Mutter J. C., Zehnder C. M. Deep crustal structure and magmatic processes: the inception of seafloor spreading in the Norwegian–Greenland Sea. In: Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society Special Publication—Morton A. C., Parson L. M., eds. (1988) 39:35–48.
Mutter J. C., Buck W. R., Zehnder C. M. Convective partial melting 1. A model for the formation of thick basaltic sequences during the initiation of spreading. Journal of Geophysical Research (1988) 93:1031–1048.
Neumann E.-R., Schilling J.-G. Petrology of basalts from the Mohns–Knipovich Ridge: the Norwegian–Greenland Sea. Contributions to Mineralogy and Petrology (1984) 85:209–223.[Web of Science]
Nielsen T. F. D., Brooks C. K. The E. Greenland rifted continental margin: an examination of the coastal flexure. Journal of the Geological Society, London (1981) 138:559–568.
Nielsen T. F. D., Soper N. J., Brooks C. K., Faller A. M., Higgins A. C., Matthews D. W. The pre-basaltic sediments and the Lower Basalts at Kangerdlugssuaq, East Greenland: their stratigraphy, lithology, palaeomagnetism and petrology. Meddelelser om Grønland, Geoscience (1981) 6:25.
Nixon P. H., ed. Mantle Xenoliths (1987) Chichester, UK: John Wiley.
Pedersen A. K. Reaction between picrite magma and continental crust: early Tertiary silicic basalts and magnesian andesites from Disko, West Greenland. Grønlands Geologiske Undersøgelse Bulletin (1985) 152:126.
Presnall D. C., Dixon S. A., Dixon J. A., O'Donnell T. H., Brenner N. L., Schrock R. L., Dycus D. W. Liquidus relations on the join diopside–forsterite–anorthite from 1 atm to 20 kbar: their bearing on the generation of and crystallization of basaltic magma. Contributions to Mineralogy and Petrology (1978) 66:203–220.[Web of Science]
Putirka K., Johnson M., Kinzler R., Longhi J., Walker D. Thermobarometry of mafic igneous rocks based on clinopyroxene–liquid equilibria, 0–30 kbar. Contributions to Mineralogy and Petrology (1996) 123:92–108.[Web of Science]
Rasmussen J., Noe-Nygaard A. Geology of the Faeroe Islands (in Danish). Bulletin of the Geological Society of Denmark (1969) 24:370.
Richard P., Shimizu N., Allègre C. J. 143Nd/144Nd, a natural tracer: and application to oceanic basalts. Earth and Planetary Science Letters (1976) 31:269–278.[Web of Science]
Richards M. A., Duncan R. A., Courtillot V. E. Flood basalts and hot-spot tracks: plume heads and tails. Science (1989) 246:103–107.
Roden M. F., Murthy V. R. Mantle metasomatism. Annual Review of Earth and Planetary Sciences (1985) 13:269–296.[Web of Science]
Scarrow J. H., Cox K. G. Basalts generated by decompressive adiabatic melting of a mantle plume: a case study from the Isle of Skye, NW Scotland. Journal of Petrology (1995) 36:3–22.
Schilling J.- G., Noe-Nygaard A. Faeroe–Iceland plume: rare-earth evidence. Earth and Planetary Science Letters (1974) 24:1–14.[Web of Science]
Schilling J. -G., Meyer P. S., Kingsley R. H. Evolution of the Iceland hotspot. Nature (1982) 296:313–320.
Schilling J. -G., Zajac M., Evans R., Johnston T., White W., Devine J. D., Kingsley R. Petrologic and geochemical variations along the Mid-Atlantic Ridge from 29°N to 73°N. American Journal of Science (1983) 283:510–586.
Schmidt S. T. Alteration under conditions of burial metamorphism in the North Shore Volcanic Group, Minnesota—mineralogical and geochemical zonation. Heidelberger Geowissenschaftliche Abhandlungen (1990) 41:309.
Schmidt S. T. Regional and local patterns of low-grade metamorphism in the North Shore Volcanic Group, Minnesota, USA. Journal of Metamorphic Geology (1993) 11:401–414.[Web of Science]
Sinton C. W., Duncan R. A. 40Ar–39Ar ages of lavas from the southeast Greenland margin. ODP Leg 152 and the Rockall Plateau, DSDP Leg 81. In: Proceedings of the Ocean Drilling Program, Scientific Results, 152—Larsen H. C., Saunders A. D., Wise S., eds. (1997) College Station, TX: Ocean Drilling Program. in press.
Soper N. J., Higgins A. C., Downie C., Matthews D. W., Brown P. E. Late Cretaceous–early Tertiary stratigraphy of the Kangerdlugssuaq area, East Greenland, and the age of opening of the north-east Atlantic. Journal of the Geological Society, London (1976) 132:85–104.
Stewart B. W., DePaolo D. J. Isotopic studies of processes in mafic magma chambers: II. the Skaergaard intrusion, East Greenland. Contributions to Mineralogy and Petrology (1990) 104:125–141.[Web of Science]
Takahashi E. Melting of a dry peridotite KLB-1 up to 14 GPa: implications on the origin of peridotitic upper mantle. Journal of Geophysical Research (1986) 91:9367–9382.
Takahashi E., Kushiro I. Melting of a dry peridotite at high pressures and basalt magma genesis. American Mineralogist (1983) 68:859–879.[Abstract]
Thirlwall M. F., Jones N. W. Isotope geochemistry and contamination mechanisms of Tertiary lavas from Skye, northwest Scotland. In: Continental Flood Basalts and Mantle Xenoliths—Hawkesworth C. J., Norry M. J., eds. (1983) Nantwich, UK: Shiva. 186–208.
Thirlwall M. F., Upton B. G. J., Jenkins C. Interaction between continental lithosphere and the Iceland plume: Sr–Nd–Pb isotope geochemistry of Tertiary basalts, NE Greenland. Journal of Petrology (1994) 35:839–879.
Thompson R. N., Esson J., Dunham A. C. Major element chemical variation in the Eocene lavas of the Isle of Skye, Scotland. Journal of Petrology (1972) 13:219–253.
Thompson R. N., Dickin A. P., Gibson I. L., Morrison M. A. Elemental fingerprints of isotopic contamination of Hebridean Palaeocene mantle-derived magmas by Archaean sial. Contributions to Mineralogy and Petrology (1982) 79:159–168.[Web of Science]
Thompson R. N., Gibson I. L., Marriner G. F., Mattey D. P., Morrison M. A. Trace-element evidence of multistage mantle fusion and polybaric fractional crystallization in the Palaeocene lavas of Skye, NW Scotland. Journal of Petrology (1980) 21:265–293.
Thompson R. N., Morrison M. A., Dickin A. P., Gibson I. L., Harmon R. S. Two contrasting styles of interaction between basic magma and continental crust in the British Tertiary Volcanic Province. Journal of Geophysical Research (1986) 91:5985–5997.
Tormey D. R., Grove T. L., Bryan W. B. Experimental petrology of normal MORB near the Kane Fracture Zone: 22°-25°N, mid-Atlantic ridge. Contributions to Mineralogy and Petrology (1987) 96:121–139.[Web of Science]
Upton B. G. J., Emeleus C. H., Rex D. C., Thirlwall M. F. Early Tertiary magmatism in NE Greenland. Journal of the Geological Society, London (1995) 152:959–964.
Waagstein R. Structure, composition and age of the Faeroe basalt plateau. In: Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society Special Publication—Morton A. C., Parson L. M., eds. (1988) 39:225–238.
Waagstein R., Hald N. Structure and petrography of a 660 m lava sequence from the Vestmanna-1 drill hole, Lower and Middle basalt series, Faeroe Islands. In: The Deep Drilling Project 1980–81 in the Faeroe Islands—Berthelsen O., Noe-Nygaard A., Rasmussen J., eds. (1984) Tórshavn: Føroya Fródskaparfelag. 39–70.
Wager L. R. Geological investigations in East Greenland. Meddelelser om Grønland (1947) 134:64.
Walker D., Shibata T., DeLong S. E. Abyssal tholeiites from the Oceanographer Fracture Zone. Contributions to Mineralogy and Petrology (1979) 70:111–125.[Web of Science]
White C. M., Geist D. J., Frost C. D., Verwoerd W. J. Petrology of the Vandfaldsdalen macrodike, Skaergaard region, East Greenland. Journal of Petrology (1989) 30:271–298.
White R., McKenzie D. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research (1989) 94:7685–7729.
White R. S. Crustal structure and magmatism of North Atlantic continental margins. Journal of the Geological Society, London (1992) 149:841–854.
White R. S., Spence G. D., Fowler S. R., McKenzie D. P., Westbrook G. K., Bowen A. N. Magmatism at rifted continental margins. Nature (1987) 330:439–444.
Wood D. A., Joron J. -L., Treuil M., Norry M., Tarney J. Elemental and Sr isotope variations in basic lavas from Iceland and surrounding ocean floor. Contributions to Mineralogy and Petrology (1979) 70:319–339.[Web of Science]
Zehnder C. M., Mutter J. C., Buhl P. Deep seismic and geochemical constraints on the nature of rift-induced magmatism during breakup of the North Atlantic. Tectonophysics (1990) 173:545–565.[Web of Science]
Zindler A., Hart S. Chemical geodynamics. Annual Review of Earth and Planetary Science (1986) 14:493–571.
Zindler A., Hart S. R., Frey F. A., Jakobsson S. P. Nd and Sr isotope ratios and rare earth element abundances in Reykjanes Peninsula basalts: evidence for mantle heterogeneity beneath Iceland. Earth and Planetary Science Letters (1979) 45:249–262.[Web of Science]
Zindler A., Staudigel H., Batiza R. Isotope and trace element geochemistry of young Pacific seamounts: implications for the scale of upper mantle heterogeneity. Earth and Planetary Science Letters (1984) 70:175–195.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. K. Barker, J. A. Baker, and D. W. Peate Interaction of the rifting East Greenland margin with a zoned ancestral Iceland plume Geology, June 1, 2006; 34(6): 481 - 484. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Bernstein, C. K. Brooks, and O. Stecher Enriched component of the proto-Icelandic mantle plume revealed in alkaline Tertiary lavas from East Greenland Geology, September 1, 2001; 29(9): 859 - 862. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. Brandriss and D. K. Bird Effects of H2O on Phase Relations during Crystallization of Gabbros in the Kap Edvard Holm Complex, East Greenland J. Petrology, June 1, 1999; 40(6): 1037 - 1064. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||