| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 2 | Pages 279-304 | 2003
© Oxford University Press 2003
The Prinsen af Wales Bjerge Formation Lavas, East Greenland: the Transition from Tholeiitic to Alkalic Magmatism during Palaeogene Continental Break-up
1DANISH LITHOSPHERE CENTRE, ØSTER VOLDGADE 10-L, DK-1350 COPENHAGEN K, DENMARK
2LABORATOIRE DES SCIENCES DE LA TERRE, ÉCOLE NORMALE SUPÉRIEURE DE LYON, 46 ALLÉE D’ITALIE, 69364 LYON CEDEX 7, FRANCE
3GEOSCIENCES RESEARCH DIVISION, SCRIPPS INSTITUTION OF OCEANOGRAPHY, LA JOLLA, CA 92093-0220, USA
4GEOLOGICAL INSTITUTE, UNIVERSITY OF COPENHAGEN, ØSTER VOLDGADE 10-L, DK-1350 COPENHAGEN K, DENMARK
5GEOLOGICAL MUSEUM, ØSTER VOLDGADE 5, DK-1350 COPENHAGEN K, DENMARK
6COLLEGE OF OCEANIC AND ATMOSPHERIC SCIENCES, OREGON STATE UNIVERSITY, CORVALLIS, OR 97331, USA
RECEIVED November 1, 2001; ACCEPTED August 8, 2002
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL BACKGROUND TO THE...RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We present elemental and isotopic (Sr–Nd–Pb–Hf–Os–He) data on primitive alkalic lavas from the Prinsen af Wales Bjerge, East Greenland. Stratigraphical, compositional and 40Ar–39Ar data indicate that this inland alkalic activity was contemporaneous with the upper parts of the main tholeiitic plateau basalts and also post-dated them. The alkalic rocks show a marked crustal influence, indicating establishment of new magmatic plumbing systems distinct from the long-lived coastal systems that fed the relatively uncontaminated plateau basalts. The least contaminated lavas have high 3He/4He isotope ratios (R/RA 12·4–18·5), sub-chondritic 187Os/188Osi (0·120–0·126), low
Ndi (
+4) and Hfi (+6) that plot below the ‘Nd–Hf mantle array’, and trace element characteristics similar to HIMU ocean island basalt (OIB). The uncontaminated magma is inferred to have more radiogenic 206Pb/204Pb values (>19·2) than the plateau basalts and Icelandic basalts, and thus represents a possible ‘enriched’ component to explain the compositional variations within the plateau basalts. One model to explain these compositional features is preferential melting of recycled material within the plume upwelling beneath the thick lithospheric cap, with 3He contributed from volatile-rich fluids from elsewhere in the Icelandic plume. The exact nature of the recycled component is not yet resolved, although Hf isotope compositions rule out any significant role for recycled pelagic sediment, and the low 187Os/188Os limits the participation of recycled basaltic material and argues instead for a contribution from the mantle section of the recycled slab. KEY WORDS: alkalic lavas; flood basalts; high 3He/4He; East Greenland; recycled lithosphere; Iceland plume
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUND TO THE...RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The extensive Early Palaeogene tholeiitic magmatism on the margins of Greenland and NW Europe, which comprise the North Atlantic Igneous Province, is generally attributed to mantle decompression melting, driven both by the arrival of the Iceland mantle plume and the subsequent rifting that led to continental break-up and the opening of the North Atlantic Ocean (e.g. Saunders et al., 1997). Volumetrically minor alkalic magmatism is found along the East Greenland coast (66–74°N), especially as highly silica-undersaturated rocks that occur mainly as intrusive complexes, but with some lavas and tephra (Nielsen, 1987). The possible origins of such alkalic rocks associated with tholeiitic flood basalts include low-degree melting of plume mantle beneath a thick lithospheric cap (e.g. Brown et al., 1996), preferential melting of enriched components within the plume such as recycled oceanic crust (e.g. Bernstein et al., 2000), and melting of enriched material within the continental mantle lithosphere by conductive heating from the plume (e.g. Carlson et al., 1996).
The most extensive sequence of alkalic lavas is in the inland Prinsen af Wales Bjerge (Fig. 1), where they form a series of small volcanic centres overlying the main tholeiitic plateau basalts (Wager, 1947; Brown et al., 1996; Hansen et al., 2002). Four stratigraphic profiles were sampled through lava sequences in the northern Prinsen af Wales Bjerge by parties from the Danish Lithosphere Centre. We present new elemental and isotopic (Sr–Nd–Pb–Hf–Os–He) data on these samples to investigate the relationship between the alkalic lavas and the underlying tholeiitic flood basalts. In addition, we address the nature of the mantle sources involved during the waning stages of magmatism that accompanied continental break-up in the North Atlantic region.
|
|
GEOLOGICAL BACKGROUND TO THE PRINSEN AF WALES BJERGE LAVAS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUND TO THE...RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
East Greenland flood basalt succession
The main Palaeogene lava sequences in East Greenland are in the Blosseville Kyst region (Fig. 1a). The lava stratigraphy of the coastal region is well established, and most of these tholeiitic lavas were erupted in two distinct phases (Storey et al., 1996) comparable with the main magmatic episodes recognized throughout the North Atlantic Igneous Province (Saunders et al., 1997): (1) the earliest volcanic rocks that include the Lower Basalts (Nielsen et al., 1981) at 61–57 Ma, followed by (2) the eruption at 56–54 Ma of the voluminous plateau basalts. The main phase of flood volcanism is divided into four stratigraphic formations (from base to top: Milne Land, Geikie Plateau, Rømer Fjord, and Skrænterne), based on field appearance, petrography, and composition (Larsen et al., 1989; Pedersen et al., 1997). The term ‘plateau basalt(s)’ is used here as a collective term for these four formations (after Hansen et al., 2002). A younger magmatic phase (50–47 Ma) is represented by volumetrically minor Igtertivâ Formation lavas that cap the plateau basalts in the NE of the Blosseville Kyst region (Larsen et al., 1989), by some mafic intrusions along the coast (Tegner et al., 1998a), and by some offshore lavas (Tegner & Duncan, 1999). A tectonic model to account for these discrete mantle melting episodes has been proposed by Saunders et al. (1997) and Tegner et al. (1998a). The initial phase is inferred to represent the initial impact and rapid dispersal of Iceland plume-head material over a broad area in the shallow upper mantle, resulting in synchronous initiation of magmatism from Baffin Island to Scotland. The driving force behind the second phase of activity is likely to have been decompression melting accompanying the initial plate break-up and opening of the North Atlantic Ocean. Tegner et al. (1998a) interpreted the third phase as resulting from the passage of the Iceland plume stem beneath the East Greenland rifted margin, although hotspot track reconstructions (Lawver & Müller, 1994; Torsvik et al., 2001) suggest that the plume axis was beneath the thick Greenland craton,
300 km inland at this time.
The Prinsen af Wales Bjerge region and sampled profiles
The inland lavas north of Kangerlussuaq (Fig. 1) have rarely been visited. Previous expeditions to the Prinsen af Wales Bjerge region in 1935–1936 (Wager, 1947; Anwar, 1955; Fawcett et al., 1982; Noble et al., 1988; Brown et al., 1996, 2000) and 1982 (Hogg, 1985; Hogg et al., 1989; Hansen et al., 2002) found alkalic lavas with variable dips, overlying a series of flat-lying tholeiitic lavas. This was interpreted as a series of late-stage alkalic shield volcanoes that developed as the main plateau basalt eruptions came to an end. Published K–Ar ages for Prinsen af Wales Bjerge lavas lie in the range 51–60 Ma (Noble et al., 1988). Hansen et al. (2002) developed a regional stratigraphy for this inland area, based on photogrammetry, new sampled profiles (Fig. 1b), and a compilation of existing and new compositional data. Parts of these sequences can be correlated with the existing coastal stratigraphy (Pedersen et al., 1997), but some compositional units are not found at the coast. In the Urbjerget profiles, there are a few lavas, lying on basement gneisses, that are correlated stratigraphically with the coastal Lower Basalts, and dated by 40Ar–39Ar at 61 Ma (Hansen et al., 2002). These are overlain by a younger sequence of high-Ti picrites, unknown elsewhere along the Blosseville coast, and then by tholeiitic basalts inferred to be part of the Milne Land Formation, the lowermost part of the plateau basalts (Hansen et al., 2002). In the northern Prinsen af Wales Bjerge profiles, tholeiitic basalts are interbedded with, and overlain by, a series of alkalic lavas. These alkalic lavas have generally been referred to as the Prinsen af Wales basalts, but Hansen et al. (2002) formalized them as the Prinsen af Wales Bjerge Formation. Compositional data on the lower parts of this inland lava sequence as well as alkalic lavas from 1982 Nunatak (Fig. 1) have been given by Hansen et al. (2002). Here, we focus on the upper parts, to evaluate the origin of the alkalic basalts that make up the Prinsen af Wales Bjerge Formation, and their relationship to the underlying tholeiitic plateau basalts.
Multi-model photogrammetric analysis using stereo photographs obtained in 1994 (Pedersen et al., 1997; Hansen et al., 2002) was used to select four stratigraphic profiles for detailed flow-by-flow sampling during expeditions in 1995 and 2000. Profile locations and stratigraphic details are shown in Figs 1b and 2, respectively. Reconnaissance sampling in 1995 of ‘Lindsay Nunatak’ to the NW of the Prinsen af Wales Bjerge (Fig. 1) found younger, tholeiitic picrites above the alkalic lavas. We also present major element data on dykes from the southern part of the Prinsen af Wales Bjerge and from areas farther south to demonstrate their compositional similarity to the Prinsen af Wales Bjerge alkalic lavas, attesting to the widespread distribution of this compositional type: one dyke from the Urbjerget 2 section (Fig. 1b: Hansen et al., 2002), and five dykes collected in 1975 that cross-cut the 445 Ma Batbjerg intrusion (Fig. 1a: Brooks et al., 1976; Fawcett et al., 1982).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUND TO THE...RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Sample classification and petrography
Using the IUGS classification scheme (Le Bas, 2000), most samples classify as picrites, basalts, trachybasalts and basanites. The total alkalis vs silica diagram (Fig. 3) shows the clear division of the analysed samples into alkalic and tholeiitic magmas, using the alkalic–tholeiitic boundary from Hawaii (MacDonald & Katsura, 1964). The tholeiitic lavas are found at the base of the sampled profiles, and overlap with the field for the coastal plateau basalts. The uppermost, tholeiitic flows at ‘Lindsay Nunatak’ are an exception as they overlie alkalic lavas that are younger than the plateau basalts based on new 40Ar–39Ar dating (see below). The alkalic lavas are mildly alkalic, with CIPW norm calculations indicating that just over half are nepheline normative. The lowermost flow at ‘Lindsay Nunatak’ (429289) is nepheline normative but different from the other alkalic rocks as it has lower SiO2 and thus classifies as a nephelinite.
|
Detailed petrographic descriptions of the Prinsen af Wales Bjerge lavas have been given by Anwar (1955), Hogg (1985) and Brown et al. (1996, 2000). The tholeiitic lavas are virtually aphyric (<5% plagioclase phenocrysts), except for two samples (429246 and 429250) that have up to 15% plagioclase phenocrysts, and the uppermost flow at ‘Lindsay Nunatak’ that is olivine-phyric (429288: 30% 1–2 mm euhedral olivine phenocrysts). Most alkalic samples are porphyritic, with up to 40% sub- to euhedral phenocrysts (2–10 mm) of olivine and/or clinopyroxene, although a few are virtually aphyric. Although our samples are mainly picritic (olivine-dominated, with <5% clinopyroxene phenocrysts), Brown et al. (1996, 2000) documented some ankaramitic (clinopyroxene-dominated) flows from other localities.
40Ar–39Ar geochronology
Two alkalic lava samples were selected for dating by the step-heating 40Ar–39Ar method. The data are presented in Table 1 [see table legend and Storey et al. (1998) for analytical details]. Sample 436223 is a lava from the middle of the ‘Lindsay Nunatak’ profile (Fig. 2), and a whole-rock sample gave a plateau age of 52·5 ± 0·3 Ma (Table 1). The isochron age for this sample is concordant with the plateau age and the initial 40Ar/36Ar ratio is within error of the atmospheric value. Sample 436231 is an evolved, pegmatitic sample from a vent site in the Spring Journey Nunataks (vent 1, Fig. 1). A plagioclase separate gave a plateau age of 54·9 ± 1·0 Ma and a concordant isochron age of 55·1 ± 1·1 Ma with an atmospheric initial 40Ar/36Ar ratio. The age for sample 436223 confirms inferences from field relations that parts of the Prinsen af Wales Bjerge Formation magmatism postdate the eruption of the coastal plateau basalts (56–54 Ma: Storey et al., 1996). However, the older age for sample 436231 provides the first indication that some of this alkalic magmatism was contemporaneous with the plateau basalt eruptions.
|
Major element variations
Major element data on 77 samples, analysed by X-ray fluorescence spectrometry, are given in Table 2. The variations of selected major elements with MgO, an index of crystal fractionation, are shown in Fig. 4. Some samples have elevated ‘volatiles’ values (see table legend) consistent with petrographic evidence for post-eruptive alteration and weathering. However, these make up only a small proportion of the analysed samples (only 12 out of 77 analysed samples have ‘volatiles’ values >2·5 wt % and most of these are dykes). Although alteration affects most samples to some extent (e.g. the greater data scatter in Fig. 4 of K2O, a fluid-mobile element, compared with TiO2), we believe that our overall petrogenetic conclusions are robust as they are based in general on variations of the more immobile elements in the freshest-looking samples.
|
|
The main tholeiitic lavas have a restricted compositional range (MgO 6·0–7·2 wt %) and lower TiO2, relative to the alkalic lavas (MgO 4–25 wt %). The youngest tholeiitic flow (429288) at ‘Lindsay Nunatak’ has high MgO (20 wt %), and some other highly magnesian tholeiites have been reported from Trekantnunatakker and ‘Tjeldebjerge’ (Brown et al., 1996, 2000). The main tholeiitic lavas show a significant range in TiO2 for a given MgO content, indicating that the samples cannot simply be related to a common parental magma by different extents of fractional crystallization. Two distinct subgroups can be recognized: Group 1 have TiO2 <3·2 wt % and SiO2 >48·2 wt %, whereas Group 2 have TiO2 >3·2 wt % and SiO2 <48·2 wt %. Two samples (429246 and 429250) have anomalously high Al2O3 indicative of plagioclase accumulation, which is consistent with their plagioclase-phyric petrography. The Group 1 flows dominate the tholeiitic lavas in the sampled profiles, with the Group 2 flows comprising just one or two relatively thin flows in each profile (Fig. 2).
Most of the alkalic lavas have between 7 and 20 wt % MgO, and they can be divided into two subgroups: a main group with low SiO2 (41–45 wt %) and high TiO2 (4·0–5·5 wt %), and an Si-enriched group with higher SiO2 (46–51 wt %) and lower TiO2 (2·5–4·7 wt %). The most mafic sample (429235; 25 wt % MgO) is also considered to belong to the high-Si group. The ‘Lindsay Nunatak’ nephelinite is distinct from the other alkalic lavas, in having very low SiO2 (39 wt %), low MgO (7 wt %), and high TiO2 (8 wt %), although broadly similar compositions are found in dykes from the nearby Gardiner Intrusion (Fig. 1: Nielsen, 1994) and in dykes traversing the Kangerlussuaq region to the south (Brooks & Rucklidge, 1974).
For the high-MgO samples (>10 wt % MgO), major element variations are controlled by olivine fractionation or accumulation. Clinopyroxene joins the fractionating assemblage at 10 wt % MgO, as indicated by the inflection on the CaO vs MgO diagram (Fig. 4e). Despite the high MgO contents of the alkalic lavas, their highly porphyritic nature suggests that they are unlikely to represent liquid compositions. However, Brown et al. (1996) found some very magnesian olivines (up to Fo91·3) in one alkalic picrite, and commented that such olivines would be in equilibrium with a liquid similar in composition to the host bulk-rock sample, and therefore that such high-MgO (18 wt %) magmas might have existed.
Trace element variations
Trace element data, analysed by inductively coupled plasma mass spectrometry (ICP-MS) on selected samples, are presented in Table 3. Primitive-mantle-normalized trace element patterns of representative unaltered samples are shown in Fig. 5a. The two tholeiitic compositional sub-groups, recognized from major element differences, are also distinct in terms of trace elements. The high-TiO2 Group 2 lavas have higher incompatible element abundances and more light rare earth element (LREE) enriched patterns compared with the main low-TiO2 Group 1 lavas (La/YbN 3·9–4·2 vs 3·0–3·5: where subscript N indicates chondrite normalized). The upper tholeiitic lava at ‘Lindsay Nunatak’ (429288) has a similar trace element pattern to the Group 2 tholeiites (La/YbN 4·9), except for higher Pb, Ba, Rb, K and Sr, whereas the other ‘Lindsay Nunatak’ tholeiite (429291) has an enriched pattern with La/YbN of 11, similar to some of the high-Si alkalic lavas (neither are illustrated in Fig. 5a). Similarly enriched tholeiitic lavas have been found at 1982 Nunatak (Hogg et al., 1989).
|
|
The alkalic lavas have higher incompatible trace element abundances and strongly LREE-enriched patterns relative to the tholeiitic lavas (La/YbN 12–28 vs 3–5). The main group show relatively smooth patterns, with primitive-mantle-normalized abundances reaching a maximum at Nb and Ta. The Si-enriched group tend to have lower extents of heavy REE (HREE) fractionation, i.e. lower Dy/YbN, than the main group (Fig. 6: Dy/YbN 1·9–2·3 vs 2·3–2·7). The Si-enriched example plotted in Fig. 5a shows a broadly similar trace element pattern to the main group, but with enrichments in Rb, Ba, K, and especially Pb and Sr.
|
The ‘Lindsay Nunatak’ nephelinite has very different trace element characteristics from the other lavas, with much higher incompatible trace element contents, a strongly LREE-enriched pattern (La/YbN 49), and marked relative depletions in K, Pb, Sr and P, and low Cs, Rb and Ba (Fig. 5a). These features are also seen in nephelinites from the Gardiner Intrusion (T. F. D. Nielsen & C. K. Brooks, unpublished data, 1997) that is of broadly similar age (56–54 Ma: Waight et al., 2002b).
Radiogenic isotopes
Sr, Nd, Pb, Hf and Os isotope data on selected samples are presented in Tables 4 and 5. Additional Sr–Nd–Pb isotope data on lavas from the Prinsen af Wales Bjerge, Trekantnunatakker, and ‘Lindsay Nunatak’ have been given by Brown et al. (1996, 2000) and Ellam & Stuart (2000). The Sr–Nd–Pb isotope data are illustrated in Fig. 7, which includes fields for data from the plateau basalts, Lower Basalts, offshore SE Greenland basalts and Iceland for comparison. The data from Brown et al. (1996, 2000) show slightly elevated 87Sr/86Sri compared with the data in this study (Fig. 7a), which is probably due to their samples not being leached before analysis.
|
|
|
The tholeiitic lavas have 87Sr/86Sri of 0·7032–0·7048, Ndi of +4·7 to +7·6, and 206Pb/204Pb of 17·1–18·3, consistent with existing data on the coastal plateau basalts (Brown et al., 1996; Andreasen et al., in preparation; Peate & Stecher, in preparation). The young tholeiitic flow from ‘Lindsay Nunatak’ has high 87Sr/86Sri of 0·7063, low Ndi of -2·0, and very unradiogenic 206Pb/204Pb of 15. The alkalic lavas, in general, have higher 87Sr/86Sri (0·7037–0·7061) and lower Ndi (-1·6 to +6·2) than the tholeiitic lavas. They show a wide range in Pb isotope composition, with 206Pb/204Pb varying from 14·7 to 18·5, but appear to be offset to lower 207Pb/204Pb relative to the tholeiitic lavas and coastal plateau basalts. This is not an analytical effect, as alkalic and tholeiitic samples were run interspersed in the same analytical session. The extreme isotopic variations, particularly in Pb isotopic compositions, seen in the alkalic lavas and some of the tholeiitic lavas are probably due to crustal assimilation with an unradiogenic Pb component (see discussion below). In terms of Hf and Nd isotopes, the Prinsen af Wales Bjerge lavas all lie below the ‘mantle array’ (Fig. 8a: Vervoort et al., 1999), with low Ndi and Hfi relative to published data on Icelandic lavas and crustally uncontaminated samples related to the North Atlantic break-up (Kempton et al., 2000). The ‘Lindsay Nunatak’ nephelinite has a similar Sr, Nd and Hf isotope composition to the least contaminated Prinsen af Wales Bjerge alkalic rocks, but it has a very different Pb isotope composition, with an extremely radiogenic 206Pb/204Pb of 23 (not shown in Fig. 7).
|
Six Prinsen af Wales Bjerge alkalic lavas were analysed for Os concentrations and isotope compositions. All samples have relatively high Os abundances from 0·17 to 1·17 ppb, and they have a range in 187Os/188Osi from 0·120 to 0·132. It should be noted that the sample with the lowest 187Os/188Osi (429244: 187Os/188Osi 0·120), also has the lowest Os content and highest Re/Os of all the samples, and its low initial Os isotope ratio might just be the result of an erroneously large age-correction owing to late-stage Re addition. The observed range in initial Os isotope compositions overlaps with the known range of Icelandic lavas (187Os/188Os 0·127–0·137: Skovgaard et al., 2001; Smit et al., 2001), although extending to sub-chondritic values lower than any known Iceland sample (Fig. 8b). 187Os/188Osi values as low as 0·122 have been measured in 61 Ma lavas from Baffin Island (A. J. R. Kent, unpublished data, 2001), albeit with significantly higher Ndi than the Prinsen af Wales Bjerge samples (Fig. 8b).
Helium isotopes
In general, Icelandic basalts are characterized by 3He/4He values higher than those found in mid-ocean ridge basalt (MORB) (R/RA = 8 ± 1: Farley & Neroda, 1998), indicating a source component derived from a region with higher time-integrated 3He/(U + Th) than the convecting upper mantle, and usually inferred to be the lower mantle (e.g. Condomines et al., 1983; Kurz et al., 1985; Poreda et al., 1986; Hilton et al., 1999; Breddam et al., 2000). Thus, He isotopes can potentially be used as a tracer of the contribution of Icelandic plume mantle to the Palaeogene North Atlantic magmatism. 3He/4He values greater than those of MORB have been found in early phase lavas in West Greenland (R/RA = 31; Graham et al., 1998), NE Greenland (R/RA = 21; Marty et al., 1998) and Scotland (R/RA = 22; Stuart et al., 2000). No studies have yet been carried out on East Greenland lavas of either the initial or main magmatic episodes. Single 3He/4He values of 10·8 (Lilloise intrusion: Bernstein et al., 1998b) and 10·5 (Prinsen af Wales Bjerge: Ellam & Stuart, 2000) have been measured in olivines from the post break-up magmatic episode, but these values are barely resolvable within error from the canonical MORB range (Farley & Neroda, 1998).
We report new helium isotope data on separated olivines from four Prinsen af Wales Bjerge alkalic lavas in Table 5. Analyses were carried out by crushing in vacuo so as to liberate helium trapped in inclusions within the olivine crystals. This technique gives the best estimates for the magmatic helium isotope signature, although the results probably represent minimum values (e.g. Hilton et al., 1999). Three samples show 3He/4He isotope ratios clearly elevated with respect to MORB (R/RA 12·4, 16·0, 18·5) and the fourth sample has a low 3He/4He value (R/RA 4·9). The new He and Pb isotope data on the Prinsen af Wales Bjerge lavas combined with those of Ellam & Stuart (2000) fall on a broad hyperbolic trend with the low 3He/4He samples (R/RA 1·8–9·7) having unradiogenic 206Pb/204Pb (14·7–17·6) consistent with assimilation of radiogenic 4He-bearing, low 206Pb/204Pb crustal material (Fig. 9). These data represent the first clear evidence for a high 3He/4He component present in Palaeogene lavas from the Kangerlussuaq region, which lies on the hotspot track of the Iceland plume axis (Lawver & Müller, 1994; Torsvik et al., 2001).
|
Olivine compositions
We carried out a reconnaissance survey of olivine compositions in the Prinsen af Wales Bjerge Formation alkalic lavas to look for possible xenocrystic lithospheric mantle olivines. The four analysed samples show a similar range of core olivine compositions (Fig. 10: Fo85·0–89·4; n = 66: sample 429264 has two olivines with Fo78). Crystal zoning appears to be limited, and in all cases is normal. However, two samples contain rare olivines with more magnesian compositions: 429235 (Fo91·0 and Fo91·6) and 429244 (Fo93·2). The question is whether these Mg-rich olivines crystallized from melts or represent mantle olivine xenocrysts. Mantle olivines have low CaO and Cr2O3 contents relative to magmatic olivines (e.g. Gurenko et al., 1996; Larsen & Pedersen, 2000). Analyses of local lithospheric mantle olivines are available from xenoliths in 40 Ma dykes from Wiedemann Fjord (Fig. 1: Bernstein et al., 1998a). The Prinsen af Wales Bjerge and xenolith data are plotted in Fig. 10 together with a reference suite of magmatic olivines (West Greenland picrites: Larsen & Pedersen, 2000). Only one analysed olivine (the one with the highest Fo content) has suitably low Cr2O3 to make it a possible xenocrystic lithospheric mantle olivine, although its CaO content, while low, is still higher than that of most mantle olivines.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUND TO THE...RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Stratigraphic relationship of the tholeiitic lavas to the coastal plateau basalts
The 40Ar–39Ar age for the ‘Lindsay Nunatak’ alkalic lava is younger than the plateau basalts (56–54 Ma: Storey et al., 1996; Heister et al., 2001; Hansen et al., 2002). However, alkalic lavas are found interbedded with tholeiitic lavas at Midway Nunatak (Fig. 2) and Tjeldebjerge (Fig. 1: Anwar 1955), although lavas from these sections have yet to be dated. If these tholeiitic lavas can be correlated with the coastal plateau basalt stratigraphy, this would place a maximum age on the initiation of alkalic magmatism. The 55 Ma age for the vent sample already suggests that some alkalic magmatism was contemporaneous with the plateau basalts. Hansen et al. (2002) showed from compositional criteria and photogrammetric studies that most of the tholeiitic lavas in the Urbjerget region are equivalent to the lowermost plateau basalts (Milne Land Formation: Larsen et al., 1989; Pedersen et al., 1997). The tholeiitic lavas in the northern Prinsen af Wales Bjerge should lie stratigraphically above those at Urbjerget, given the general northward dip of the plateau basalts in this area, and are thus most likely to represent either the upper part of the Milne Land Formation or the overlying Geikie Plateau Formation.
It might be possible to use distinctive compositional features to correlate the lavas with a particular plateau basalt formation. For example, using samples from a 6 km composite section through the coastal plateau basalts at Sortebræ (Fig. 1), Tegner et al. (1998b) demonstrated a contrasting behaviour in REE systematics between the lower and upper parts. In the lower part (Milne Land and Geikie Plateau Formations), there is a negative correlation of Dy/YbN with La/SmN, whereas in the upper part (Rømer Fjord and Skrænterne Formations), these ratios show a positive correlation (Fig. 11a). Furthermore, within the lower part, there are systematic variations with stratigraphic height, with La/SmN increasing and Dy/YbN decreasing up through the Milne Land and Geikie Plateau Formations.
|
The main Group 1 tholeiites have higher La/SmN and lower Dy/YbN than the stratigraphically lower Urbjerget tholeiites that are clearly equivalent to the lower Milne Land Formation (Fig. 11a: Hansen et al., 2002). This stratigraphic compositional transition in the inland area from the Urbjerget tholeiites to the Group 1 tholeiites appears to be equivalent to the temporal variations within the lower part of the plateau basalts, suggesting that the Group 1 tholeiites are equivalent to the upper Milne Land or Geikie Plateau Formations. The relatively low La/SmN of the Group 1 tholeiites rules out any correlation with the Rømer Fjord Formation lavas. There is an overlap in REE compositions with some lavas from the Skrænterne Formation, but in terms of major elements the Group 1 tholeiites tend to have higher SiO2 at a given TiO2/FeO(t) value than Skrænterne Formation lavas (Fig. 11b: Hansen et al., 2002). The volumetrically minor Group 2 flows have higher La/SmN than most Milne Land Formation lavas and higher Dy/YbN than the Geikie Plateau Formation lavas. Although they have similar La/SmN and Dy/YbN to the Rømer Fjord Formation lavas, they can be distinguished from these lavas by their higher SiO2 at a given TiO2/FeO(t) and higher Zr/Y. It is possible that the Group 2 flows represent a minor compositional variant with the Milne Land or Geikie Plateau Formation that is not seen in the Sortebræ profile.
Thus, it appears that alkalic magmatism in the Prinsen af Wales Bjerge region initiated during the eruption of the early–middle part of the plateau basalts (upper Milne Land or Geikie Plateau Formations). This is consistent with the 55 Ma age obtained from the vent 1 sample. This activity lasted until after the plateau basalt eruptions had ceased (53 Ma age from ‘Lindsay Nunatak’), a total duration of at least 2 Myr. Additional evidence for alkalic magmatism contemporaneous with the upper two formations of the tholeiitic plateau basalts comes from nephelinitic tuffs found interbedded within the Rømer Fjord Formation lavas (Larsen et al., 1989), and a phonolitic tuff within the Skrænterne Formation lavas that is inferred to have been erupted from the Gardiner complex at 54 Ma (Fig. 1: Heister et al., 2001). Furthermore, whereas the full stratigraphic sequence of the plateau basalts is preserved in the inland area near Gronau West (Fig. 1: Heister et al., 2001), only lavas of the lower formations appear to have reached the Prinsen af Wales Bjerge region to the east.
The role of crustal assimilation
Magmas traversing the continental crust have the potential to modify their compositions by assimilating crustal material. The effects of crustal contamination on lava compositions in East and SE Greenland are most readily apparent in Pb isotope compositions because the local basement has a distinctive, very unradiogenic 206Pb/204Pb isotope composition (Leeman et al., 1976; Taylor et al., 1992; Kalsbeek et al., 1993). Many of the Lower Basalts (Holm, 1988; Fram & Lesher, 1997; Hansen & Nielsen, 1999) and the offshore lavas drilled by the Ocean Drilling Program off SE Greenland (Fitton et al., 1998, 2000; Saunders et al., 1999) show clear evidence for significant contamination by such crustal material (Fig. 7). In detail, two compositionally distinct crustal contaminants affecting lavas in both regions can be recognized, distinguished primarily by differences in Sr isotopes (amphibolitic gneiss with relatively high 87Sr/86Sr, and granulitic gneiss with relatively low 87Sr/86Sr) but both with relatively unradiogenic 206Pb/204Pb and low Ndi. From the Pb isotope variations (Fig. 7), it is clear that the extent of crustal assimilation is, in general, lower in the plateau basalts than in the earlier Lower Basalts, and this limited crustal influence is also seen in the main group of tholeiitic lavas from the Prinsen af Wales Bjerge profiles. In contrast, the younger ‘Lindsay Nunatak’ tholeiites appear to be highly contaminated, as is evident from their unradiogenic 206Pb/204Pb of 14·9–16·2.
The Sr–Nd–Pb isotope data indicate a strong crustal influence in the Prinsen af Wales Bjerge alkalic lavas and also that the crustal assimilant is broadly similar in composition to that involved in the plateau basalts, Lower Basalts, and the early phase lavas drilled offshore further south. The relatively high 87Sr/86Sr of the inferred contaminant suggests dominance of amphibolitic rather than granulitic crustal material. The effects of crustal assimilation are also clear from the major and trace element data: the Si-enriched group have lower Ndi and 206Pb/204Pb than the main group, and tend to have high Ba/La, K/Nb and U/Pb ratios, and lower Ce/Pb and Dy/YbN ratios, indicative of a higher proportion of assimilated Si-rich crust. The highest 187Os/188Os values (0·130–0·132) are found in the samples with the lowest Ndi and 206Pb/204Pb. This is consistent with addition of ancient crustal material with time-integrated elevated Re/Os and low Sm/Nd and U/Pb ratios, as is illustrated by one example of a plausible mixing curve (curve 1 in Fig. 8b). The question, though, is whether the main group are themselves slightly contaminated. Mixing arrays on Pb–Pb isotope diagrams are linear, and it is interesting that extrapolation of the alkalic lava trend on the 207Pb/204Pb vs 206Pb/204Pb (Fig. 7b) to higher 206Pb/204Pb (i.e. potentially less-contaminated compositions) intersects the trend of uncontaminated plateau basalt lavas and Icelandic basalts at 206Pb/204Pb values of 19·2–19·8. This is a plausible source composition for the Prinsen af Wales Bjerge lavas, and also explains their low 207Pb/204Pb relative to the plateau basalt samples. If we assume that this intersection point represents the Pb isotope composition of the uncontaminated magma, estimates for the amounts of contamination required to explain the lava data can be made using plausible compositions for the assimilated crustal material. These amounts depend on the choice of elemental and isotopic compositions assumed for the crustal end-member. If we take reasonable values as used in other studies of East Greenland magmatism (e.g. Blichert-Toft et al., 1992; Hansen & Nielsen, 1999; Saunders et al., 1999), then the main group of alkalic lavas can be explained by 4% assimilation whereas the most contaminated samples require 15–20% assimilation. The differences in SiO2 between the main group and the high-Si group are also consistent with such amounts of assimilation.
Thus, rather surprisingly, there is clear evidence for greater extents of crustal assimilation in the alkalic lavas, despite their elevated trace element contents, than in the associated underlying tholeiitic lavas. This marked crustal influence in the Prinsen af Wales Bjerge alkalic lavas, together with the absence of mantle xenoliths, is suggestive of small magma batches that stalled in the crust en route to the surface. The high levels of contamination compared with the relatively uncontaminated tholeiitic plateau basalts provide further evidence that the Prinsen af Wales Bjerge alkalic lavas, and the younger tholeiites on ‘Lindsay Nunatak’, are locally erupted through newly established magma conduits, whereas the voluminous plateau basalt lavas probably flowed into the region fed from long-lived plumbing systems nearer the coast or from the east (Pedersen et al., 1997). Possible eruption vent sites for the Prinsen af Wales Bjerge alkalic lavas have been identified in the ‘1982 Nunatak’, ‘Tjeldebjerge’, Trillingerne, and Spring Journey Nunataks regions (Fig. 1: Wager, 1947; Brown et al., 2000; Hansen et al., 2002). The scatter of the data on the various isotope plots (Figs 7 and 8) rules out a single contamination trend, consistent with frequent small magma batches traversing the crust and stalling at different levels at different times.
Mantle source for ‘uncontaminated’ Prinsen af Wales Bjerge alkalic lavas
As discussed above, none of the Prinsen af Wales Bjerge Formation samples represent pristine uncontaminated mantle-derived magmas. From the three least contaminated samples (429244, 429265, 429272) and the contamination model calculations, an ‘uncontaminated’ magma is inferred to have the following distinctive isotopic characteristics: 87Sr/86Sri 0·7035, Ndi +5 to +6, Hfi +8 to +9, 206Pb/204Pbi 19·2–19·8, 187Os/188Osi 0·120–0·126, and 3He/4He >18 R/RA. These Sr–Nd–Pb isotope characteristics are broadly similar to the ‘high 206Pb plume component’ that contributed to the 56–54 Ma tholeiitic dykes on the Faeroes (Holm et al., 2001), and the ‘206Pb-rich, p component’ that Hanan & Schilling (1997) inferred to be representative of the main Icelandic plume mantle.
The primitive-mantle-normalized diagram (Fig. 5a) shows relatively smooth trace element patterns, except for slight negative anomalies at K and P, with normalized abundances reaching maxima at Nb and Ta. The strongly fractionated HREE patterns (Dy/YbN 2·1–2·7) indicate melting in the presence of residual garnet. These magmas share many trace element characteristics with HIMU ocean island basalt (OIB) magmas (Fig. 5b: e.g. Sun & McDonough, 1989; Weaver, 1991; Hofmann, 1997), namely low La/Nb (0·7–0·8), K/Nb (<180), Th/Nb (<0·08) and Ba/La (5–8), but without the extreme 206Pb/204Pb (>20) values found in typical HIMU OIB such as St. Helena. Thirlwall (1997) refers to other OIB magmas with similar characteristics as being derived from ‘young HIMU’ mantle; i.e. mantle modified to give similar trace element features to the extreme HIMU source, but sufficiently recently so as not to have allowed in-growth of radiogenic 206Pb. The positive primitive-mantle-normalized anomaly of Nb relative to La, Th, Ba and K in the HIMU OIB source is usually attributed to a contribution from recycled subduction-processed residual ocean slab material.
The ‘Lindsay Nunatak’ nephelinite is compositionally distinct from the Prinsen af Wales Bjerge Formation lavas (i.e. contrasting primitive-mantle-normalized trace element patterns in Fig. 5, and markedly different Pb isotope compositions). These differences are indicative of a different petrogenetic origin. It has trace element features consistent with small-degree melting in the presence of minor residual phases that hold back elements such as K, Sr, Pb and P. The Pb isotope data and U–Th–Pb abundances for this sample are consistent with being derived from a source broadly similar to that of the Palaeogene tholeiitic and alkalic lavas but that had elevated U/Pb and Th/Pb ratios for the last 400–800 Myr. Thus it is plausible that this nephelinite, as well as the contemporaneous Gardiner intrusion, represents melts of local lithospheric mantle that was metasomatized during the Caledonian orogeny.
Recycled components in Iceland plume-related magmatism
Recycled components are increasingly invoked to explain much of the isotopic variations seen within OIB globally (e.g. Weaver, 1991; Hofmann, 1997). It has been proposed that the compositional variations in Icelandic and Hawaiian lavas can be explained by melting of different components that make up a complete section of recycled subducted slab (e.g. Lassiter & Hauri, 1998; Chauvel & Hémond, 2000). The low-Nd component (alkalic basalts, Iceland; ‘Koolau’ component, Hawaii) is interpreted as the upper part of recycled oceanic crust (basalt ± pelagic sediment), whereas the high-Nd component (depleted picritic lavas, Iceland; ‘Kea’ component, Hawaii) is interpreted as the gabbroic part of the recycled slab. The role for DMM (depleted MORB mantle) as a magma source at both places appears to be minimal (e.g. Lassiter & Hauri, 1998; Chauvel & Hémond, 2000), and the high 3He/4He signal is inferred to be primarily from the plume mantle hosting the recycled slabs (e.g. Eiler et al., 1998; Hilton et al., 1999).
In this paper, we are concerned primarily with the nature of the ‘enriched’ low-Nd component(s). It is not clear if more than one ‘enriched’ component is required to explain the compositional variety of trace element enriched, low-Nd magmas found throughout the North Atlantic region, and also how the Prinsen af Wales Bjerge alkalic lavas compare compositionally with these other magmatic suites. Examples include alkalic Iceland lavas (e.g. Hards et al., 1995; Chauvel & Hémond, 2000; Prestvik et al., 2001), Jan Mayen lavas (Schilling et al., 1999; Trønnes et al., 1999), Vestbrona nephelinites, offshore Norway (56 Ma: Prestvik et al., 1999) and Palaeogene melilitites, NE Greenland (Bernstein et al., 2000). Trønnes et al. (1999) have argued, for example, that the compositional features of Jan Mayen alkaline lavas can be explained by contributions from Nb-rich melts derived from a recycled oceanic crust component.
The highly alkaline ultrabasic lavas found in the Nunatak Region of NE Greenland are the most compositionally extreme of all the Palaeogene alkaline rocks in East Greenland (Bernstein et al., 2000), and they are similar in composition to the meymechites associated with the Siberian flood basalts (Horan et al., 1995). Bernstein et al. (2000, 2001) suggested that these rocks represent low-degree melting at high pressures (>5 GPa) of subducted oceanic lithosphere material within the Iceland plume and also that variable mixing of such a component with a depleted MORB-like melt can explain the main compositional features of the plateau basalts. The primitive-mantle-normalized trace element pattern of such a melilitite is broadly similar to that of the main group of Prinsen af Wales Bjerge alkalic lavas, although the melilitite has higher incompatible trace element abundances and higher La/Yb and Dy/Yb (Fig. 5b). The Nunatak Region lavas have lower SiO2 and Al2O3, and higher Fe2O3 and TiO2 (Fig. 4), which, together with their higher Dy/Yb, suggests low-degree melting at higher pressures than for the Prinsen af Wales Bjerge alkalic lavas, perhaps beneath a thicker lithospheric cap. Isotope data are needed to test the possible similarity of mantle sources between the Nunatak Region melilitites and the Prinsen af Wales Bjerge alkalic lavas, although the few Sr and Nd isotope data of Bernstein et al. (2001) are broadly consistent with such a model. Furthermore, the 206Pb/204Pb value (>19·2) inferred for an uncontaminated Prinsen af Wales Bjerge magma is sufficiently radiogenic to represent the ‘enriched’ component that is required to explain the isotopic variations within the plateau basalts (Peate & Stecher, 2002).
Compositional similarities between alkalic magmas from Jan Mayen, Iceland and Vestbrona (e.g. Fig. 12) led Trønnes et al. (1999) to suggest a common ‘enriched’, low-Nd mantle component with low 3He/4He (Schilling et al., 1999) that was present within the ancestral Iceland plume and that is now widely dispersed throughout the North Atlantic upper mantle. However, the Prinsen af Wales Bjerge alkalic lavas and the Nunatak Region lavas are markedly different in composition from this component, both having lower Ba/La and higher Zr/Nb (Fig. 12). This implies that the ancestral Iceland plume contained more than one ‘enriched’, low-Nd component, as suggested by Hanan & Schilling (1997).
|
At the time of the Prinsen af Wales Bjerge alkalic magmatism, the Iceland plume stem is inferred to have been beneath thick Greenland cratonic lithosphere, within 200–300 km of the Prinsen af Wales Bjerge region (Lawver & Müller, 1994; Torsvik et al., 2001). Melting beneath the inland regions was largely restricted to the enriched recycled slab components (see also Bernstein et al., 2000, 2001) because the thick lithospheric cap prevented the plume mantle from upwelling to depths shallow enough for melting to occur above its dry solidus (115 km: Ito et al., 1999). At deeper levels, perhaps up to 180–210 km (Ito et al., 1999), incipient ‘wet’ melting caused by dehydration and defluidization of the upwelling plume mantle will have produced fluids that were probably highly mobile and capable of transporting significant amounts of helium plus other rare gases and volatiles (e.g. Schilling et al., 1999; Breddam et al., 2000). Thus, one model to explain the high 3He/4He ratios measured in the Prinsen af Wales Bjerge Formation lavas is addition of Iceland plume-derived helium from such deep He-rich fluids to melts of the enriched recycled slab components.
Hf and Os isotope constraints on the nature of the recycled component(s)
Isotope data on the least contaminated Prinsen af Wales Bjerge alkalic lavas can provide some constraints on the nature of the recycled low-Nd component. For example, ancient pelagic sediments have unradiogenic Nd and Hf isotope compositions that plot above the ‘mantle array’ in Fig. 8a (i.e. they have more radiogenic Hf than normal OIB for a given Nd value). The shallow slope of the Hawaiian data trend and relatively high Hf of the low-Nd ‘Koolau’ component indicate the presence of old recycled pelagic sediment in this mantle source (Blichert-Toft et al., 1999). This contrasts with the steep slope of the Icelandic data trend, relative to the ‘mantle array’. The inferred composition of the uncontaminated Prinsen af Wales Bjerge magma lies on an extension of this data trend below the ‘mantle array’, limiting the role of pelagic sedimentary material in this component. Nd and Hf isotope data on HIMU OIB lavas plot in a similar position to the Prinsen af Wales Bjerge lavas, below the ‘mantle array’, and Salters & White (1998) demonstrated that this is consistent with derivation primarily from an old recycled oceanic crust component. Blichert-Toft & Albarède (1997) explained such characteristics by remelting ancient residues produced by garnet-absent melting.
The compatibility of Os and moderate incompatibility of Re during melting produces high Re/Os and, with time, high 187Os/188Os in mafic melts, whereas the peridotitic mantle residue will have low Re/Os, eventually leading to low 187Os/188Os. Thus, ancient recycled mafic components in OIB can be identified from their elevated 187Os/188Os values as seen, for example in HIMU OIB lavas (187Os/188Os 0·155: Hauri & Hart, 1993; Shirey & Walker, 1998). However, the Prinsen af Wales Bjerge alkalic lavas have sub-chondritic Os isotope compositions (187Os/188Osi 0·120–0·126; 
Osi -7 to -2), indicating a peridotitic rather than pyroxenitic source. There are several possibilities to consider:
- melting of a DMM source, which has suitably low 187Os/188Os ratios of 0·123–0·128 (defined by abyssal peridotites and the least radiogenic MORB samples: Shirey & Walker, 1998). However, this can be ruled out as it has significantly higher Nd than the Prinsen af Wales Bjerge lavas (+10 vs +4).
- Another possible source is sub-continental lithospheric mantle. This is often characterized by sub-chondritic Os isotope values reflecting ancient melt-related Re depletion (average 187Os/188Os = 0·113: Shirey & Walker, 1998), and data on the Wiedemann Fjord xenoliths show that such depleted lithospheric mantle occurs beneath East Greenland (187Os/188Osm 0·101–0·126: Hanghøj et al., 2001). The very low 187Os/188Osi (0·112–0·121: Carlson et al., 1996) of some mafic potassic magmas in central Brazil demonstrates that alkalic magmas can result from low-degree melting of lithospheric mantle, rather than the underlying asthenospheric or plume mantle. However, it is difficult to reconcile the elevated 3He/4He ratios of the Prinsen af Wales Bjerge alkaline lavas with an origin purely from melting of continental lithospheric mantle, as such mantle is generally characterized by 3He/4He ratios similar to, or lower than, MORB (e.g. Dunai & Baur, 1995).
- Assimilation of lithospheric mantle material by an ‘enriched’ plume-derived melt, as suggested for high-MgO meymechites from Siberia (Horan et al., 1995), would rapidly decrease the Os isotope composition of the melt with little effect on other radiogenic isotope systems (Sr–Nd–Pb–Hf) owing to the marked contrast between the Os and Sr–Nd–Pb–Hf contents in the two end-members. The ‘enriched’ plume component seen in early phase magmatism in West Greenland and Baffin Island (Schaefer et al., 2000; A. J. R. Kent, unpublished data, 2001) provides a suitable end-member to test this mixing model for the Prinsen af Wales Bjerge alkalic lavas. It has similar Ndi (+5) to the ‘uncontaminated’ Prinsen af Wales Bjerge magma, but with higher Osi (+5 vs -5). The Prinsen af Wales Bjerge data can be modelled by addition of 15–25% of a lithospheric mantle component (curve 2 in Fig. 8b). Although such a model cannot be excluded, several lines of evidence suggest that this might not be the origin of the observed sub-chondritic Os isotope compositions. First, there is no clear evidence for any xenocrystic lithospheric mantle olivines in the samples, and, second, analyses of Holocene alkalic basalts from Iceland (Smit et al., 2001) show that material with relatively low Ndi (