Journal of Petrology Advance Access published online on January 22, 2009
Journal of Petrology, doi:10.1093/petrology/egn081
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Geochemical Stratigraphy of Submarine Lavas (3–5 Ma) from the Flamengos Valley, Santiago, Southern Cape Verde Islands
1Department of Earth Sciences, Uppsala University, Villavägen 16, SE 752 36, Uppsala, Sweden
2Department of Geography and Geology, University of Copenhagen, Øster Voldgade 10L, Copenhagen 1350 K, Denmark
3Department of Geoscience, University of Iowa, 121 Trowbridge Hall, Iowa City, IA 52242, USA
4School of Earth Sciences, Victoria University of Wellington, PO BOX 600, Wellington, New Zealand
Received December 20, 2007; Revised typescript accepted December 24, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
New high-precision Pb–Sr–Nd isotope, major and trace element and mineral chemistry data are presented for the submarine stage of ocean island volcanism on Santiago, one of the southern islands of the Cape Verde archipelago. Pillow basalts and hyaloclastites in the Flamengos Valley are divided into three petrographic and compositional groups; the Flamengos Formation lavas (
4·6 Ma) dominate the sequence, with the younger Low Si and Coastal groups (2·8 Ma) found near the shoreline. Olivine and clinopyroxene compositions and isotopic data for minerals and their host melts indicate disequilibrium between some crystals and the melt. Intra-sample disequilibrium suggests homogenisation of liquids but eruption before complete equilibration between crystals and melt preserves the heterogeneity. Pressures of crystallization for clinopyroxene (0·4–1·1 GPa) indicate stalling and crystallization of the magmas over a range of depths in the lithosphere. Major element compositions indicate melting of a carbonated eclogite source. Sr–Nd–Pb isotope data suggest the involvement of FOZO-like and EM1-like components in the mantle source, which are simultaneously available at all depths in the melting column. The Flamengos Valley lavas display large compositional variations, often between stratigraphically adjacent flows; these frequent abrupt changes of magma composition suggest stalling and crystallization of discrete magma batches on transport through the lithosphere. KEY WORDS: Cape Verde; crystal–melt disequilibrium; submarine volcanism; source heterogeneity; Pb–Sr–Nd isotopes
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The Cape Verde Islands provide an excellent opportunity to sample the early submarine stage of ocean island growth that is otherwise often deeply buried in ocean islands or is accessible only through drilling (e.g. Hawaii; Rhodes, 1996
). Extensive valley erosion has led to excellent exposure of continuous stratigraphic sections on several of the Cape Verde islands (e.g. Santo Antão, Holm et al., 2006; Santiago, this study). Their study allows investigation of detailed temporal geochemical variations that might result from changing conditions and sources of melt generation and from variations in the shallow-level plumbing systems.
The Cape Verde Islands are located 500 km west of the coast of Senegal, West Africa (Fig. 1). There are nine main islands forming a horseshoe-shaped archipelago opening to the west. The Cape Verde Islands are located on top of the Cape Verde Rise, a plateau 500 km in diameter with 2 km of elevation above the adjacent oceanic floor (Dash et al., 1976). The archipelago is situated close to the pole of rotation for the African continent, resulting in negligible plate movements of 3 mm/a towards the SSE (Pollitz, 1991), and thus it can be considered stationary over the lifetime of the hotspot. There is anomalous heat flow and an associated geoid anomaly that are consistent with the presence of a mantle plume (Courtney & White, 1986; Pim et al., 2008). Seismic tomographic studies suggest that a plume-like anomaly extends to depths greater than 1900 km beneath the islands (Montelli et al., 2004). Although located close to the edge of the African continent, the continental margin lies to the east of the Cape Verde Rise and therefore the islands are not underlain by continental crust, but are built on 130–135 Ma oceanic crust and late Jurassic sediments (Klerkx et al., 1974; Dosso et al., 1993; Ali et al., 2003).
Volcanism in the Cape Verdes is thought to have initiated at c. 24–22 Ma (Holm et al., 2008). There is a subtle age progression within the southern islands from the currently active volcanism on the western island of Fogo (
350 ka), to the significantly eroded island of Maio (22–7 Ma) to the east (Christensen et al., 2001; Holm et al., 2008). We have chosen to focus on Santiago as its moderate erosion means that the volcanic record remains relatively coherent and erosion has served to maximize exposure of the older submarine sequences. Previous research by Davies et al. (1989) showed that lavas on Santiago range from basanites to more silica-undersaturated compositions. The stratigraphic relations are complex, leading Serralheiro (1976) to suggest that the submarine Flamengos Formation was preceded by the extensive ‘Complex Antigo’. However, 40Ar–39Ar dating indicates that many outcrops associated with the ‘Complex Antigo’ are in fact much younger (Holm et al., 2008). The Pico da Antónia Formation represents the subsequent shield-building stage of the island, followed by the more recent post-erosional flows of the Assomada Formation and contemporaneous scoria cones. This paper focuses on the petrogenetic evolution of a stratigraphic sequence of lavas exposed in the Flamengos Valley, considered representative of the 2·8–4·6 Ma submarine stage of magmatism on Santiago (Holm et al., 2008). The compositional variations throughout the history of volcanism on Santiago will be considered elsewhere (Barker et al., in preparation).
|
Field characteristics and stratigraphy
The deeply incised Flamengos Valley reveals a mostly continuous exposure of lavas that extends inland for 14 km (Fig. 1); 49 samples were collected along a profile from inland (sample 301) to the coast (sample 371). The submarine pillow basalts and hyaloclastites of the Flamengos Formation, as mapped by Serralheiro (1976
), extend stratigraphically towards the coast dipping at 20–30°, and as the river valley is perpendicular to the strike, the profile generally samples progressively younger lavas towards the coast (Fig. 1e). The coastal 0·6–1 km were mapped as younger submarine units associated with the Pico da Antónia Formation, consistent with their younger Ar–Ar ages (Holm et al., 2008). The majority of the profile is unconformably overlain by an erosional beach deposit, which truncates the pillow lavas (Fig. 1e). Younger subaerial lavas of the Pico da Antónia Formation cover the beach deposits high in the valley walls, dipping a few degrees coastward, and they were not sampled as part of this study. We have divided the Flamengos Valley samples into three groups based on petrography, geochemistry and stratigraphy: (1) the Flamengos Formation; (2) the Low Si group; (3) the Coastal group. The Flamengos Formation accounts for 38 of the 49 samples (301–329 and 351–359), and forms the bulk of the sequence (8·5 km of the profile). 40Ar–39Ar dating of two samples that span the stratigraphic extent of the Flamengos Formation gives similar weighted plateau ages of 4·57 ± 0·31 (2
) Ma and 4·59 ± 0·09 (2) Ma (Holm et al., 2008). There is a significant hiatus before the eruption of the Low Si and Coastal group samples at about 2·8 Ma. The Low Si group consists of five exposed flows at 13·2–14 km along the profile with an age of 2·87 ± 0·31 (2) Ma. These are overlain by the Coastal group 14·3–14·6 km along the profile with an age of 2·83 ± 0·08 (2) Ma (Holm et al., 2008). The 5 km of the profile between the youngest dated sample of the Flamengos Formation and the Low Si group lavas lacks good outcrop, but the unconformable contact with the beach deposits above continues to truncate both units. The only sample (359) collected from this poorly exposed section has rather anomalous compositional characteristics (see below) and it could have been erupted at any time during the 1·8 Myr interval between 4·6 and 2·8 Ma. The significant age difference between the Flamengos Formation and the Low Si and Coastal groups, despite their similar field characteristics, confirms that they are not necessarily genetically related. |
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Olivine, clinopyroxene and groundmass plagioclase compositions in 13 representative samples were analysed by wavelength-dispersive spectrometry using a JOEL JXA-8200 Superprobe at the University of Copenhagen (e.g. Andreasen et al., 2004
). Whole-rock samples were jaw-crushed and altered surfaces removed prior to milling in an agate mill. Major element compositions were analysed on fused glass discs using a Philips PW 1606 X-ray fluorescence spectrometer at the Geological Survey of Denmark and Greenland using the methods of Kystol & Larsen (1999). Trace element analyses were measured by Philips PW 1400 X-ray fluorescence spectrometer at the geological Institute, University of Copenhagen, Denmark following the general procedures of Norrish & Chappel, (1977). Rare earth elements in selected samples were analysed on unleached powders by isotope dilution and multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) at the Danish Lithosphere Centre (DLC) using the methods of Baker et al. (2002).
High-precision Pb isotope compositions were measured on hand-picked leached rock chips by MC-ICP-MS at DLC, using the double-spike method of Baker et al. (2004) to correct for mass bias. Rock chips were leached in 6N HCl for 1 h at 100°C. A leaching experiment was undertaken on sample 327 to evaluate the effects of different leaching protocols (unleached sample; 30 min 2N HCl cold leach; 15 min, 1 h and 2 h 6N HCl hot leaches) on the Pb isotope composition. The SRM 981 Pb standard gave an average of 206Pb/204Pb = 16·9425 ± 11, 207Pb/204Pb = 15·4995 ± 14 and 208Pb/204Pb = 36·7269 ± 38 (2 s.d., n = 15). Groundmass–clinopyroxene separates were analysed for U/Pb by isotope dilution and high-precision Pb isotope compositions for sample 366 in an attempt to determine a U/Pb eruption age. Analytical Pb blanks were negligible (< 50 pg). Sr and Nd isotopes were measured on the eluted matrix from the Pb chemistry, followed by separation with standard ion-exchange procedures (e.g. Luais et al., 1997). Sr isotope analyses were performed by VG 54-30 thermal ionization mass spectrometer at the Geological Institute, University of Copenhagen. Sr isotope data were corrected for mass fractionation using 86Sr/88Sr = 0·1194, and the SRM987 standard gave 87Sr/86Sr = 0·710231 ± 21 (2 s.d.). Nd isotopes were analysed by MC-ICP-MS at DLC, with the isotope ratios corrected for mass fractionation using 146Nd/145Nd = 2·071943 as both masses are free from Sm isobaric interferences. The data were acquired during three sessions with the AMES standard of 143Nd/144Nd = 0·512133 ± 11 (2 s.d., n = 11), 143Nd/144Nd = 0·512164 ± 13 (2 s.d., n = 10) and 143Nd/144Nd = 0·512147 ± 16 (2 s.d., n = 6), and all Nd isotope data were normalized to a 143Nd/144Nd value of 0·51213 for this AMES standard, equivalent to La Jolla 143Nd/144Nd = 0·51185. Analytical blanks for Nd were < 67 pg.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Petrography and mineral chemistry
Coastal group (2·8 Ma)
These samples are porphyritic (
10–15% phenocrysts; 0·3–1·6 mm) with subhedral phenocrysts of olivine (2–5%) and zoned clinopyroxene (5–10%) containing spinel inclusions. The groundmass is dominated by clusters of clinopyroxene with Fe–Ti oxides, glass, feldspar needles in most samples and occasional olivine, amphibole, and zeolite. The samples exhibit a distinctive appearance in thin section from the contrast of the clinopyroxene clusters with the black glass (Fig. 2a). Alteration of phenocrysts is shown by amphibole growth along rims and cracks. Olivines are Fo86–88 composition and the clinopyroxenes have diopside and salite compositions with Mg-number 73–83 (Fig. 3).
|
|
Low Si group (2·9 Ma)
These samples are sparsely porphyritic (<3% phenocrysts: 1–2% olivine, 1–2% spinel, and 1–2% clinopyroxene), vesicular and brown in appearance. They are characterized by a fine groundmass of euhedral Fe–Ti oxides, olivine, rare clinopyroxene, amphibole and zeolites, and a conspicuous absence of feldspar (Fig. 2b). Alteration is shown by amphibole replacement of earlier minerals, dissolution of clinopyroxene, and zeolite development in vesicles. The olivines have Fo contents of 81–83, some of which have clinopyroxene overgrowths, and the clinopyroxenes are salites with Mg-number 71–76 (Fig. 3).
Flamengos Formation (4·6 Ma)
These samples fall into three petrographic types (Table 1): (1) aphyric (< 2% phenocrysts); (2) olivine and/or clinopyroxene phyric (5–20% phenocrysts); (3) very porphyritic (15–40% phenocrysts, 0·1–3·8 mm) with olivine ± clinopyroxene, and minor spinel (Fig. 2d). Olivines are sub- to euhedral, and sometimes have clinopyroxene rims and spinel inclusions. Clinopyroxenes show a variety of textures from radiating clusters, glomerocrysts, small tabular phenocrysts to large sub- to euhedral phenocrysts, often with magnetite as inclusions or rims. Groundmasses are holocrystalline to melanocrystalline composed of glass, prehnite, needles or clusters of clinopyroxene, Fe–Ti oxides, feldspar needles (An74–92), hornblende and rare olivine. Some of the samples contain dark enclaves (< 0·5 cm; Fig. 2c), visible only in thin section, that display the entire range of phenocryst and groundmass mineralogy found in the enclosing lavas. In some cases, these enclaves are sheared out, together with vesicles, and mixed to different extents with the groundmass. Table 1 summarizes the stratigraphic variation of several distinctive petrographic features within the Flamengos Formation. Most of the aphyric samples occur inland less than 2·13 km along the profile; samples with dark enclaves are also more prevalent in the youngest part of the sequence. Samples with similar clinopyroxene clusters to the Coastal group make an early sporadic appearance, are absent in the middle of the sequence, and then reappear irregularly towards the upper part of the sequence. Samples lacking feldspar (reminiscent of the Low Si group) occur late in the sequence.
|
Olivines are Fo79–90 in composition. Clinopyroxenes are diopside and salite (plus rare augite) with Mg-number 67–88 (Fig. 3). Clinopyroxene rims associated with olivine cores have low Mg-number (70–77). Many samples contain normally zoned clinopyroxenes (Fig. 3b), which occasionally show reverse zonation at the rim. A profile across a clinopyroxene phenocryst in sample 306 (Fig. 3c) shows a core of Mg-number 80, with an increase in Mg-number to 82, and is generally normally zoned towards the rim although there are oscillations to higher Mg-number recording up to four episodes of crystallization of higher Mg-number from core to rim. On a plot of Na2O vs MgO (Fig. 3d), clinopyroxenes form two broad groups; one with higher Na2O at a given MgO content than the other; samples often contain pyroxenes from both groups, even within single zoned phenocrysts (Fig. 3c and d).
Whole-rock major and trace elements
Major and trace element data are reported in Tables 2 and 3. The total alkalis vs silica diagram classifies the six Coastal group samples as basanites and the five Low Si group samples, given their mineralogy and silica–alkali content, as mela-nephelinites (Fig. 4; Le Bas, 1989). The Flamengos Formation samples include 16 basanites and 11 picrobasalts: six of the picrobasalts have > 12 wt % MgO and classify as picrites (Le Bas, 2000). There are also four basalts and two alkali basalts. The aphyric samples have < 7·7 wt % MgO and all but two of the samples with < 6·2 wt % MgO are aphyric (Fig. 5).
|
|
|
|
The Flamengos Formation lavas show a wide range in MgO contents (6–16 wt %, Fig. 5) at relatively constant SiO2 (42–46 wt %) and Fe2O3(t) contents (11–14 wt %). In contrast, the Low Si and Coastal groups both have restricted major element compositions. The sparsely porphyritic low Si group have MgO contents of 9·5–10·2 wt % and low SiO2 (38–40 wt %), together with higher Fe2O3(t) (16–17 wt %), TiO2 (4·8 wt %), Na2O, P2O5 and Nb than the Flamengos Formation. Sample 359, from the poorly exposed section between the Flamengos Formation and Low Si group outcrops, has compositional features, such as elevated Na2O, P2O5 and Nb, and low SiO2, compared with the main Flamengos Formation samples, that are more similar to the Low Si group lavas, although it has much lower MgO (
7 wt %). The coastal group samples have 12·3–13·2 wt % MgO, and are compositionally more similar to the Flamengos Formation, except for higher Al2O3, Na2O, P2O5 and Nb, and lower SiO2. Mantle-normalized trace element patterns for all the Flamengos Valley lavas show broadly similar patterns, with marked light REE (LREE) enrichment, and a maximum at Nb (Fig. 6). The Low Si group shows variability in Rb, Ba and K, which are susceptible to alteration, but with a very narrow homogeneous range for the middle REE (MREE) to heavy REE (HREE). The Coastal group patterns show very little variability, with similar enrichment and depletion patterns to the Flamengos Formation and Low Si group, offset to lower normalized values than the latter because of their more primitive nature.
|
The Flamengos Formation lavas show a wide variation in Zr/Y that is seen across the full range of MgO contents (Fig. 7b); for example, the aphyric lavas (MgO
6–7 wt %) have Zr/Y of 9–14. These variations in Zr/Y broadly correlate with Zr/Nb (3·0–4·3) (Fig. 7c). The Coastal group samples plot in the middle of this trend, whereas the Low Si group samples plot to even higher Zr/Y (15·5) and Zr/Nb (4·5). In terms of REE, the Flamengos Valley lavas have a restricted range in Dy/Yb (2·8–3·3) that is significantly higher than mid-ocean ridge basalt (MORB) values, but slightly lower than most samples from the northern Cape Verde Islands (Fig. 7d). The Flamengos Formation samples show a cluster with La/Yb of 30–42, with one sample at lower La/Yb of 23. A Coastal group sample falls in the middle of this cluster with La/Yb of 38, whereas the Low Si group samples have higher La/Yb (50–58) than the Flamengos Formation.
|
Radiogenic isotopes (Sr–Nd–Pb)
High-precision Pb isotope data for the Flamengos Formation samples form a broad array from 206Pb/204Pb of 19·07–19·49 (Fig. 8a and b, Table 4). The array follows the Northern Hemisphere Reference Line (NHRL) in 207Pb/204Pb (
7/4 –0·6 to + 0·9 c.f. ± 0·2 2se: 7/4 and 8/4 are the vertical deviation from the NHRL; Hart, 1984), but is higher than and oblique to the NHRL in 208Pb/204Pb. The Low Si group and Coastal group have restricted variation at relatively unradiogenic Pb: 206Pb/204Pb of 19·04–19·08 and 19·09–19·13, respectively. These new data are consistent with the positive 8/4 of the southern islands found in previous studies of the Cape Verde Islands (Gerlach et al., 1988; Davies et al., 1989: Doucelance et al., 2003), although extending the field of the southern islands to higher 206Pb/204Pb values.
|
|
The Flamengos Formation samples show a wide range in
Nd from 1·5 to 4·0, across a wide range in Pb isotopes from 206Pb/204Pb of 19·15–19·5 (Fig. 8c). The Low Si group has low Nd of 0·9–1·5 at low 206Pb/204Pb of 19·05, and the Coastal group follows stratigraphically with intermediate Nd of 2·2–2·7 at low 206Pb/204Pb of 19·1. It is notable that at 206Pb/204Pb (19·0–19·1), the Flamengos Valley lavas display a wide range in Nd of 0·8–3·7 (Fig. 8c). For Sr–Nd isotopes the Flamengos Formation samples form a broad array of decreasing Nd from 4·0 to 1·5 as 87Sr/86Sr increases from 0·7031 to 0·7035 (Fig. 8d). The Low Si group shows a shift to lower Nd of 0·9–1·5 and higher 87Sr/86Sr > 0·7036. The Coastal group marks a change back to lower 87Sr/86Sr of 0·7034–0·7035, at high Nd of 2·2–2·7 for a given 87Sr/86Sr comparable with the highest 87Sr/86Sr of the Flamengos Formation. These data are consistent with previous data from the southern islands (Gerlach et al., 1988; Davies et al., 1989; Doucelance et al., 2003) extending to slightly lower 87Sr/86Sr and lower Nd at low 87Sr/86Sr. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Alteration
The compositions of lavas erupted in submarine environments have the potential to be modified by seawater. The presence of prehnite, zeolites (mainly in vesicles), and amphibole alteration rims on phenocrysts provides petrographic evidence that the Flamengos Valley lavas have all been influenced by seawater alteration. One way to assess the effects of such alteration on lava compositions is to look at groups of related lavas with a restricted compositional range. For example, most elements in the Low Si group vary by < 5%, and elements expected to be immobile such as Ti, Nb, Zr, Y and REE vary by < 1·5%. However, several elements (Rb, Sr, Ba, K) show significantly greater variability (10–100%) indicating mobility during the alteration process. Other groups of related samples within the Flamengos Formation, such as (1) 307, 308 and 309 and (2) 325, 326 and 327, show limited variability of large-ion lithophile elements (LILE), suggesting that alteration is largely insignificant for samples with LOI <
2·5%. Therefore, samples with LOI > 2·5% are excluded from further discussion. Although the anomalous sample (359) shows greater petrographic evidence of alteration (and LOI > 3%), its high Nb and Y indicate that it does have a distinct composition that cannot be explained by alteration.
Sr–Nd–Pb isotopes were all measured on samples that had been leached in an effort to remove the effects of low-T alteration. Despite the clear evidence for mobility of LILE in the Low Si group lavas, Sr–Nd–Pb isotope data show a very narrow compositional range that is inconsistent with significant and variable alteration by seawater: 206Pb/204Pb = 19·05 ± 1470 ppm (2 s.d.), 87Sr/86Sr = 0·70367 ± 100 ppm (2 s.d.), and 143Nd/144Nd = 0·51270 ± 43 ppm (2 s.d.). Sample replicates and compositionally similar groups such as the Flamengos Formation samples 325, 326 and 327 show remarkable reproducibility: a leaching experiment on five replicates of sample 327 gave 206Pb/204Pb of 19·210–19·214, and additionally the total of eight analyses for samples 325, 326 and 327 are within this range (Table 4). Although the presence of Pb-rich Fe–Mn nodules can potentially modify the Pb isotope composition of submarine lavas, analyses of Fe–Mn crusts from near the Cape Verdes (Abouchami et al., 1999) have different compositions from the Flamengos Valley lavas. The Fe–Mn crusts overlap the 206Pb/204Pb range of the lavas (18·7–19·0 vs 19·0–19·5), but they have significantly higher 207Pb/204Pb (15·68–15·73 vs 15·56–15·61), and therefore clearly had no influence on controlling the observed Pb isotope variations in the Flamengos Valley lavas.
Fractional crystallization and crystal accumulation
The major and trace element trends (Figs 5 and 7a) of the Flamengos Valley samples indicate that fractional crystallization and crystal accumulation had a significant influence on the compositional evolution of these magmas.
The Flamengos Formation lavas have a wide range of MgO contents, from 4 to 16 wt %, and increasing Al2O3, Na2O, K2O and TiO2 and generally decreasing CaO with decreasing MgO content (Fig. 5). However, samples with MgO > 9 wt % are variably porphyritic, with up to 50% olivine and clinopyroxene phenocrysts, and so their compositions might not reflect liquid compositions but instead might be affected by crystal accumulation. If we consider only the aphyric samples, then their compositional range (7·6 wt % MgO for sample 314 to 5·8 wt % MgO for sample 327) can be modelled by 24% fractional crystallization of an assemblage of clinopyroxene and plagioclase in proportions of 83:17 (Fig. 5). However, there are significant ranges in ratios of highly incompatible elements in the aphyric samples, such as Nb/Zr (0·24–0·33) and La/Nb (0·6–1·1), which indicate that this trend cannot represent a single liquid line of descent but must reflect a range of parental magmas generated through variations in melting conditions and source compositions. Trace element data are consistent with the fractional crystallization of olivine (decreasing Ni) and clinopyroxene (decreasing Sc).
There is a group of porphyritic samples with MgO between 7·5 and 10·5 wt % that have lower CaO than the group of most primitive aphyric samples with 8 wt % MgO. These form a trend with the more evolved aphyric samples (with 6 wt % MgO), highlighted by the solid black line in Fig. 5d; together these samples have a restricted range of 10·2–12·4 for Zr/Y, compared with 9–14 for the total Flamengos Formation. This trend is consistent with accumulation of clinopyroxene in a melt compositionally similar to the more evolved aphyric samples.
Samples with MgO > 9 wt % show broadly constant CaO, FeO and Sc, which is consistent with fractionation of only olivine, but there is considerable scatter in the trends (e.g. at a given MgO value, CaO varies by almost 3 wt %: Fig. 5f). Given the highly porphyritic nature of these samples, it is likely that crystal accumulation has influenced their compositions, producing some of this scatter. The magnitude of the effects of crystal accumulation can be modelled from the observed phenocryst compositions and their modal abundance in a sample (Table 1). This is illustrated in Fig. 5 for phenocryst accumulation from one of the most primitive aphyric samples (sample 308, 7·9 wt % MgO) for the range of olivine (10–30%) and clinopyroxene (10–40%) phenocrysts apparent in samples with MgO > 9 wt %. The range of accumulation trends (black dashed lines in Fig. 5), shows increasing Al2O3, Na2O and TiO2 and constant SiO2 with decreasing MgO content. Samples with a high proportion of olivine (10–30%) in comparison with clinopyroxene (< 10%) phenocrysts plot to higher TiO2 and Fe2O3 wt % at a given MgO than samples with higher clinopyroxene to olivine ratios (0–20% vs 10–30%, respectively).
The sparsely porphyritic Low Si group and the porphyritic Coastal group both have restricted major element ranges and minor compositional variations indicating minimal differences in fractional crystallization between samples. Although the major element compositional variations within the Flamengos Formation are broadly consistent with fractional crystallization and/or accumulation processes, it is apparent that the three groups cannot be related to each other by differentiation. The lower MgO and SiO2 of the silica-undersaturated Low Si group relative to the coastal group cannot be a result of fractional crystallization as differentiation will not produce a melt with lower MgO and SiO2 or vice versa (Figs 4 and 5). Likewise the Flamengos Formation has higher SiO2 than the Low Si group and shows a trend of fractional crystallization that is distinct from the compositions of both the Low Si group and the Coastal group, notably the lower Na2O and P2O5 for a given MgO value.
Evidence for crystal–melt disequilibrium
Olivine and clinopyroxene phenocrysts show a large compositional range within many single samples (Fig. 3), which means that many of the crystals cannot be in equilibrium with their host lava. Equilibrium can be evaluated from the partitioning of Fe/Mg between crystal and melt based on the known distribution coefficients: Fe/Mg Kd between olivine and melt is 0·32 ± 0·04 (for P in the range 0·5–2·0 GPa; Roeder & Emslie, 1979; Putirka, 2005a), and Fe/Mg Kd between clinopyroxene and melt is 0·275 ± 0·067 (1 s.d.; Putirka et al., 2003). For these calculations, an estimate of the Fe3+/Fetotal ratio of the melt is needed. Lavas from Santo Antão and São Nicolau in the northern Cape Verde Islands have minimum measured Fe3+/Fetotal values of 0·26 and 0·19, respectively (Holm et al., 2006; Duprat et al., 2007). The oxygen fugacity measured on Cape Verde xenoliths (Ryabchikov et al., 1995) is similar to estimates for lavas from La Palma, Canary Islands (compositionally similar to the Santiago lavas), which are equivalent to a Fe3+/Fetotal value of 0·21 (Klügel et al., 2000). These Fe3+/Fetotal ratios are more oxidized than in typical MORB, consistent with a recent observation that ocean island basalts (OIB) seem to fall into two groups of Fe3+/Fetotal 0·1 and 0·15–0·25 (Herzberg & Asimow, 2008). Based on these studies, the lavas from Santiago are expected to have a relatively oxidized Fe3+/Fetotal and we assume a value of 0·21 for the Fe3+/Fetotal ratio in the melt.
The compositions of olivine and clinopyroxene phenocrysts are shown in Fig. 9, in comparison with trends for the expected Fo content of olivines and Mg-number of clinopyroxenes in equilibrium with melts of a given Mg-number. The liquid compositions are represented by the whole-rock analysis of the samples. For the porphyritic samples, corrected liquid compositions are also shown based on subtraction of the modal proportions of olivine and clinopyroxene phenocrysts (Table 1) from the bulk-rock compositions, in order better to approximate the liquid composition (e.g. Mordick & Glazner, 2006). Although some olivine and clinopyroxene phenocrysts are in equilibrium with the melt in most samples, there is a large range of phenocryst compositions that exhibit disequilibrium. Most of the virtually aphyric, low-MgO Flamengos Formation samples (e.g. 307, 310, 318) and the sparsely porphyritic Low Si group contain equilibrium olivines. The olivines have core compositions that are in equilibrium with the melt but are paired with disequilibrium clinopyroxene rims. Most samples contain a small proportion of clinopyroxenes in equilibrium with the host melt, but most of the measured clinopyroxene compositions scatter to much lower Mg-number (Fig. 9b). The clinopyroxenes have cores that range from equilibrium to disequilibrium compositions with the rims showing disequilibrium. Holm et al. (2006) showed that disequilibrium between melt and both olivine and clinopyroxene phenocrysts is also common in lavas from Santo Antão, one of the northern Cape Verde islands.
|
The Fe/Mg Kd between mineral and melt is not very sensitive to pressure (Yang et al., 1996
), and so it is unlikely that variations in pressure can account for the formation of such a wide range in phenocryst compositions. There is little correlation between Fo content or Mg-number and phenocryst CaO wt %, suggesting that the dominant control is not fractional crystallization (Putirka, 1997), and confirming the disequilibrium between olivine and melt (Norman & Garcia, 1999). Olivines have relatively high CaO contents (0·18–0·51 wt %) consistent with a magmatic origin, suggesting that disequilibrium olivines are inherited from cumulates or through magma mixing rather than from disaggregated lithospheric mantle rocks (Cape Verde xenoliths have mean CaO of 0·17 ± 0·1 wt % (2 s.d.); Ryabchikov et al., 1995; Bonadiman et al., 2005). The preservation of zoned phenocrysts is consistent with the mineral chemistry indications of disequilibrium.
There is an indication from high-precision Pb isotope data that some isotopic heterogeneity is preserved between phenocrysts and melt (Table 4). Figure 10 shows isotopic differences between whole-rock, light and dark clinopyroxenes, and groundmass fractions. Figure 10c and d illustrates these differences plotted as 
6/4, 7/4 and 8/4 (deviations in 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb of clinopyroxene and groundmass from the whole-rock; Bryce & DePaolo, 2004), suggesting isotopic heterogeneity between clinopyroxene and melt. The differences extend beyond generous uncertainties shown by the box (5 se; ± 262 ppm 206Pb/204Pb, ± 321 ppm 207Pb/204Pb, ± 256 ppm 208Pb/204Pb; blanks 20–50 pg). Duplicate Pb isotope analyses of whole-rock chips also suggest within-sample heterogeneity. For some aphyric samples, the duplicates are within error (310, 325, 327, Table 4), confirming the precision of the method. However, there are other aphyric and porphyritic samples that do not duplicate within error (Coastal group sample 360 and aphyric sample 314; Fig. 10). The inference from these data is that there is isotopic heterogeneity within the sample groundmasses, with different sample aliquots having sampled different mixtures of isotopic components within the sample. These preliminary results suggest that additional work is needed to extensively investigate within-sample heterogeneity more fully.
|
Mineral thermobarometry
Mineral compositions can potentially be used to calculate the conditions (temperature and pressure) under which phenocrysts formed, thus providing important information about crystallization processes and the mechanisms by which the crystals became incorporated into the final erupted magma. Although it is possible to determine pressures from only clinopyroxene compositions using the geobarometers of Nimis (1999
), which could potentially be applied to clinopyroxenes out of equilibrium with the host melt, such geobarometers are not calibrated for the silica-undersaturated compositions of the Cape Verde lavas (P. Nimis, personal communication, 2007). Consequently, equilibrium between phenocryst and host melt is a prerequisite for geothermometry and geobarometry, and this can be evaluated for olivine and clinopyroxene using Fe/Mg partitioning (see above). We have used the geothermometers and geobarometers of Putirka et al. (2003, 2005b), which are calibrated for silica-undersaturated melt compositions similar to those of the Cape Verde lavas (K. D. Putirka, personal communication, 2008), to calculate the temperature and pressures of crystallization of equilibrium olivine and clinopyroxene phenocrysts (Table 5, Fig. 11).
|
|
There is a wide range in estimated crystallization temperatures (1125–1260°C) and pressures (0·4–1·1 GPa); these results are summarized in Fig. 11 with the samples shown in their stratigraphic order. Single samples crystallize over ranges of 30–80°C and
0·3 GPa and crystallization commences at higher pressures > 0·45 km along the profile. Adjacent samples have commonly crystallized over similar pressure ranges. This suggests that adjacent compositionally contrasting magmas have stalled, cooled and crystallized at similar depths within the lithosphere.
The Moho beneath the Cape Verde archipelago has recently been seismically determined to be at 18 ± 1 km (Lodge & Helffrich, 2006), which is equivalent to 0·49 GPa. The Flamengos Formation magmas stall and differentiate at higher pressures than the Moho, comparable with compositionally similar magmas from Madeira and the Canary Islands (Klügel et al., 2000; Nikogosian et al., 2002; Schwarz et al., 2004; Longpré et al., 2008; Stroncik et al., 2009), suggesting that the Moho does not lead to stalling of magmas on these ocean islands, which are all built on relatively old, thick oceanic lithosphere. Ali et al. (2003) presented a best-fit elastic plate thickness of 29 km (0·8 GPa; elastic thickness is the brittle upper part of the lithosphere above a depth corresponding to around 400°C) based on seismic reflection data for the Cape Verde region; however, they reported that gravity and topographic data allow for an elastic thickness range of 20–29 km (0·5–0·8 GPa). Clinopyroxenes from six samples indicate pressures in this range. Thus the stalling of some magmas may be influenced by the brittle–ductile transition, as suggested for Hawaiian magmas (Putirka, 1997). The pressures of crystallization for five samples, like those for many of the Madeira and Canary Island samples (Klügel et al., 2000; Nikogosian et al., 2002; Schwarz et al., 2004; Longpré et al., 2008; Stroncik et al., 2009), indicate that crystallization occurred deeper within the oceanic mantle lithosphere, below the elastic thickness and above the 90 km deep seismic Low Velocity Zone (LVZ; Lodge & Helffrich, 2006).
Melting and source variations
In this section we consider melting systematics and investigate the influence of source heterogeneity and composition on the Flamengos Valley lavas. Mantle source heterogeneity is evident from the observed range in radiogenic isotope compositions (Fig. 8). Isotopic variability within the southern islands as a whole has been attributed to variable contributions from two isotopically enriched sources: an EM1-like source (low 206Pb/204Pb < 18·8, low Nd < –1) and a FOZO-like source (higher 206Pb/204Pb 19·5–19·9; low Nd +5; Gerlach et al., 1988; Barker et al., in preparation). The Flamengos Formation exhibits a wide range of Pb isotope compositions (206Pb/204Pb 19·07–19·49), representing large variations in the relative contributions of the FOZO and EM1 components (Fig. 8). The Low Si group has the lowest 206Pb/204Pb and Nd, implying the greatest contribution from the EM1-like source amongst the Flamengos Valley lavas. The Coastal group samples have slightly higher 206Pb/204Pb and Nd than the Low Si group and lower 206Pb/204Pb than the majority of the Flamengos Formation, suggesting a greater influence from an EM1-like source than in the Flamengos Formation but less than in the Low Si group. Variations between trace elements and isotopes are not simple, as the latter give an indication of the relative proportions of the various source components involved in mixing, whereas the degree and pressure of melting also influence the trace element characteristics. The Low Si group shows subtle trace element and isotopic differences from the Flamengos Formation and Coastal group, indicating small variations in source compositions. In contrast, the major element differences, especially high Fe2O3(t) and low SiO2, are much more pronounced, implying significant differences in melting conditions or source mineralogy.
The Cape Verde Islands are constructed on thick, old (132–135 Ma; Ali et al., 2003) oceanic lithosphere, which acts as a lid on the melt column and restricts melting in the underlying asthenosphere to low extents at relatively high pressures. The marked MREE–HREE fractionation seen in all the Flamengos Valley lavas (e.g. Dy/Yb 2·8–3·3) compared with average MORB indicates the presence of garnet in the source (Fig. 7d). Prytulak & Elliott (2007) have shown from a careful consideration of Ti partition coefficients in mantle minerals that it is not possible to generate lavas with high TiO2 contents by melting a fertile peridotite source. This is especially true for the Flamengos Valley lavas, which have TiO2 contents of 3–5 wt % at MgO contents of 10 wt %, corresponding to some of the highest TiO2 contents observed in OIB globally. An anhydrous garnet peridotite source would also produce melts that are too high in SiO2 and low in CaO to be the parental magmas for the Flamengos Valley lavas (Fig. 5; Hirose & Kushiro, 1993). Likewise a garnet pyroxenite source produces melts that have lower TiO2 and CaO than required to produce the primitive counterparts to the Flamengos Valley samples (Hirschmann et al., 2003; Prytulak & Elliott, 2007). However, carbonated peridotite and/or eclogite melting in the presence of CO2 has the potential to generate melts with appropriate SiO2 and CaO contents to produce the Flamengos Formation, Coastal and Low Si group lavas (Dasgupta et al., 2006, 2007). This is consistent with the presence of broadly contemporaneous carbonatite magmatism in the Cape Verde Islands, including on Santiago (Hoernle et al., 2002; Jørgensen, 2003).
The Low Si group samples have more extreme compositions compared with the Flamengos Formation and Coastal group samples, with higher Fe2O3(t) (15 wt %) and TiO2 (4·5–5 wt %), and lower SiO2 (38–40 wt %). The Low Si group also has the highest La/Yb and Zr/Y (Fig. 7), implying a more enriched source or lower degrees of melting. Partial melting of carbonated garnet peridotite at higher pressures is likely to produce low SiO2 melts that are enriched in Fe (Dasgupta et al., 2007). Therefore, one hypothesis to explain the Low Si group might be the development of residual mantle at shallow depths as melting proceeds, depressing melting depths to where more fertile material is encountered beneath the refractory root (e.g. Holm et al., 2006; Lodge & Helffrich, 2006). However, this is unlikely to be the cause of the high melting pressures of the Low Si group, as it is difficult to explain why there would be such an abrupt temporal change to the subsequent Coastal group samples and other intermediate and young volcanic rocks from Santiago, as well as the temporal range of volcanic rocks from Santo Antão, northern Cape Verde (Holm et al., 2006; Barker et al., in preparation), which revert to similar major and trace element compositions, and hence melt fraction and pressure, to the Flamengos Formation. The Low Si group appears to be a unique and localized feature of the magmatic record on Santiago (Barker et al., in preparation). Based on melting experiments (Dasgupta et al., 2007) that show a negative correlation between SiO2 and CO2 in partial melts of carbonated peridotite, the unusually low SiO2 of the Low Si group is more likely to be a function of higher CO2 than higher pressures. Although higher CO2 might be expected to increase the CaO content of the Low Si group, which is not observed relative to the Flamengos Formation, at low degrees of melting the dominant control over the behaviour of CaO is the extent of melting rather than the proportion of CO2 (Dasgupta et al., 2007). The higher Fe2O3(t) and TiO2 contents of the Low Si group suggest that the mantle source might be dominated by eclogitic rather than peridotitic lithologies (Dasgupta et al., 2007). The low Ni contents of the Low Si group lavas (85 ppm Ni in sparsely porphyritic lavas with 10% MgO) are also consistent with derivation from an eclogite source (Huang & Frey, 2005).
All of the Flamengos Valley lavas have low K/Nb values (100–200) compared with MORB and typical OIB (250: Sun & McDonough, 1989). As the alkalis (Ba, Rb, K) and Ti behave compatibly in amphibole and phlogopite (LaTourette et al., 1995), low K/Nb can potentially be generated by residual amphibole or phlogopite during melting, which would require a source within the oceanic lithosphere (Class et al., 1998). Although the K and Rb contents of the Flamengos Formation samples can potentially have been affected by alteration, the majority of the samples plot at similar K/Nb and Rb/La values. Importantly, K/Nb and Rb/La do not correlate with La/Sm, La/Nb or other indications of the degree of melting, which implies that the dominant control is not residual amphibole or phlogopite. Furthermore, the TiO2 content of the Flamengos Valley samples, especially the Low Si group, is high, which requires a low Ti partition coefficient in the source (Prytulak & Elliott, 2007), in contrast to the high Ti partition coefficient of melting in the presence of residual amphibole or phlogopite (La Tourette et al., 1995). Therefore, the low K/Nb charateristics of these lavas are probably an inherited source feature rather than an indicator of lithospheric melting.
Temporal variations in trace element and isotope ratios
The Flamengos Valley samples show wide variability in incompatible trace element ratios and isotope compositions, which indicates that crystal fractionation and accumulation cannot be the only processes causing compositional variations in these lavas. In this section we investigate whether there are any consistent stratigraphic variations in composition within the 4·6 Ma Flamengos Formation that forms the majority of this lava sequence. We focus on two immobile incompatible element ratios that display a relatively wide range of values (Zr/Y 9–14; Zr/Nb 3·0–4·5; Fig. 12). Although Zr/Y will potentially increase with decreasing MgO content as fractional crystallization involving clinopyroxene proceeds, Fig. 7 demonstrates that the full range in Zr/Y values is found in samples across the range in MgO contents, and so the variations in Zr/Y are not exclusively controlled by fractional crystallization and phenocryst accumulation. Poor correlations of Zr/Y and Zr/Nb with radiogenic isotopes (e.g. 206Pb/204Pb; Fig. 13a) indicate that the variations in trace element ratios are dominated by the melting process and mixing of melts rather than source heterogeneity effects. For example, a subset of Flamengos Formation samples shows a wide range in Zr/Y (8·9–14·1; Fig. 13a) and Zr/Nb (3·0–4·2) at constant 206Pb/204Pb values (19·2–19·3), suggesting a range in the degree of partial melting required to generate the parental magmas to the Flamengos Formation lavas.
|
|
Stratigraphic variations show an average Zr/Y of 11–12·5 and Zr/Nb of 3·4–3·9 along the sequence (Fig. 12a and b); however, in addition distinct groups of samples are found: a group at high Zr/Nb (4·2) and high Zr/Y (14) less than 0·5 km along the sequence; a group at low Zr/Nb (3) and low Zr/Y (10) between 4 and 5 km; and a group of porphyritic samples at the end of the sequence with high Zr/Nb (3·7–4·0) with contrasting high and low Zr/Y (14–9·8; Fig. 12a and b). Successive flows often show large compositional differences, with groups of only 3–4 samples maintaining similar compositions over distances of 140–400 m. Similar stratigraphic variability between successive flows or groups of flows is not uncommon on other ocean islands; for example, it is seen in lava sequences from Santo Antão, one of the northern Cape Verde islands (Holm et al., 2006
), and on Hawaii, where the continuous stratigraphy of the Hawaiian Scientific Drilling Project (HSDP) shows variations that are gradual to abrupt, never maintaining constant Zr/Y for more than a few adjacent flows (Rhodes, 1996). The Flamengos Formation stratigraphic sequence also preserves evidence of source variations despite the complexity added by melting, transport and magma chamber processes. Correlations between Zr/Y and 206Pb/204Pb can be seen in Fig. 13a, with low 206Pb/204Pb extending to high Zr/Y and high 206Pb/204Pb extending to low Zr/Y (samples 354, 355), corresponding to EM1-like (low 206Pb/204Pb) and FOZO-like (high 206Pb/204Pb) sources, respectively. The samples with the highest and lowest 206Pb/204Pb ratios show greatest source control (i.e. FOZO-like and EM1-like components, respectively), whereas samples that show constant relative contributions from EM1-like and FOZO-like sources (i.e. intermediate 206Pb/204Pb) preserve a wide range of Zr/Y that reflects selective tapping of melts generated at different levels within the melt column. High, intermediate and low Zr/Y samples are juxtaposed with no particular stratigraphic relationship, indicating that melts segregated at different depths are not mixed prior to eruption, potentially following different pathways during transport through the entire lithosphere. Single samples must result from mixing of melts or sources at similar melting depths within the melting zone; that is, shallow melts do not seem to have interacted with deeper melts, which would average the Zr/Y, but sources have hybridized at discrete depths. As crystallization seems to have taken place predominantly within the lithospheric mantle, then olivine-hosted melt inclusions might provide a more complete sampling of the compositional diversity of melts generated within the melt column rather than being dominated by shallow crustal level processes.
Younger subaerial lavas from Santiago (Barker et al., in preparation) and lavas from Santo Antão, a northern Cape Verde island, also show a wide range in Zr/Y from 8 to 16 (for samples with 7–19 wt % MgO; Fig. 13b and c; Holm et al., 2006), comparable with the range seen in the Flamengos Formation lavas. There is a wide range with relatively constant 206Pb/204Pb and correlated increases in Zr/Y and Zr/Nb suggesting that the melting systematics do not change significantly either temporally or spatially within the archipelago.
Implications for the transport and interaction of melts
All samples represent mixtures of melts derived from EM1-like and FOZO-like mantle sources. Stratigraphically adjacent samples are compositionally varied and the stratigraphy is characterized by abrupt compositional differences, implying that the supply of source components to the melt zone is not constant. The presence of adjacent lavas with contrasting compositions suggests that melts hybridize discrete proportions of EM1-like and FOZO-like components to form single magma batches whose diversity is represented by the variable supply of the end-members. Simultaneous compositional evidence for mantle source contribution and melting systematics suggests that the melts were hybridized prior to aggregation, stalling and crystallization, and were probably transported in proximal plumbing conduits, preventing interaction between compositionally distinct melts.
The mineral chemistry and within-sample isotopic heterogeneity indicate within-sample disequilibrium between phenocrysts and melt and between groundmass separates. This implies that the duration of juxtaposition of magmas prior to eruption was enough to hybridize or partially mix the liquids but not long enough for re-equilibration of phenocrysts (Bryce & DePaolo, 2004). Magmas leave the melting and mixing region in small batches and grow phenocrysts in the lithospheric mantle. The magmas crystallize as they traverse the oceanic mantle lithosphere and subsequently single magma batches aggregate their phenocryst load and mingle with other magmas (Davidson et al., 2005; Stroncik et al., 2009); the liquids hybridize almost completely and eruption follows maintaining the disequilibrium between phenocrysts and melt. This preserves the sample-scale heterogeneity, as eruption occurs prior to complete hybridization of the liquids. The crystals and melt are derived from different magma batches, which may represent temporal and spatial differences between melts segregated from the source. Stroncik et al. (2009) have developed a similar model to explain the petrographic and compositional features in submarine lavas from El Hiero in the Canary Islands. They have argued that small, intermittent and isolated magma chambers that each have distinct fractionation and magma-mixing histories might be a common feature of oceanic intraplate volcanoes, such as in the Canary Islands and the Cape Verde Islands, where there is a low flux of magma from the mantle. Further investigation of the crystal population in the Flamengos Valley lavas by methods such as crystal size distribution, trace element diffusion profiles and melt inclusions, plus detailed study of enclaves hosted by the lavas, should provide more information about the timing and origin of phenocrysts, heterogeneity and mixing processes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The Flamengos Valley on Santiago, one of the southern Cape Verde islands, reveals a semi-continous stratigraphic sequence of submarine pillow lavas and hyaloclastites. We have divided the sequence into three stratigraphic groups: the Flamengos Formation (4·6 Ma), which forms the bulk of the sequence, followed by the Low Si and Coastal groups (2·8 Ma). The younger Low Si and Coastal lavas show little internal compositional variation and are geochemically distinct from each other and the older Flamengos Formation lavas, which show a broad range in major and trace elements and Pb–Sr–Nd isotope composition.
Major element compositions rule out melting of an anhydrous peridotite source and instead point towards the derivation of the parental magmas from a carbonated peridotite and/or eclogite source. The low SiO2 content of the Low Si group compared with the other Flamengos Valley lavas suggests melting in the presence of higher CO2 concentrations rather than higher pressures, and the elevated Fe2O3(t) suggests a carbonated eclogite rather than a carbonated peridotite source.
There is petrographic, mineral chemistry and Pb isotopic evidence for disequilibrium between some phenocrysts (olivine and clinopyroxene) and their host melts. Preservation of heterogeneity requires mixing of compositionally different melts and their crystal cargoes shortly prior to eruption to allow homogenization of liquids without significant re-equilibration between phenocrysts and host melt.
Clinopyroxene geobarometry indicates that significant crystallization occurred within the oceanic lithosphere below the Moho (0·4–1·1 GPa). Stratigraphically adjacent but compositionally distinct samples stalled, cooled and crystallized at broadly similar depths in the lithosphere. During transport through the lithosphere the magmas underwent fractional crystallization; interaction between discrete magma batches led to within-sample heterogeneity by aggregation of multiple phenocryst populations hosted by hybridized melts.
FOZO-like and EM1-like melts segregate and mix at specific depths in the mantle and are probably transported as small magma batches in proximal plumbing conduits.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
Claude Herzberg and Thomas Kokfelt are thanked for discussions regarding mantle sources, and Paolo Nimis and Keith Putirka are thanked for discussions concerning clinopyroxene geobarometry. Godfrey Fitton, Christian Tegner, Jörg Geldmacher, Dennis Geist and Karsten Haase are thanked for their constructive comments and thorough reviews, which helped us to improve and clarify several aspects of the manuscript. We would like to thank Toby Leeper and Berit Wenzell for analytical assistance. We received funding from the Danish National Research Council to facilitate this research at the former Danish Lithosphere Centre.
*Corresponding author. Telephone: +46 18 471 25 85. E-mail: Abigail.Barker{at}dunelm.org.uk
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Abouchami W, Galer S. JR, Koschinsky A. Pb and Nd isotopes in NE Atlantic Fe–Mn crusts: proxies for trace metal paleosources and paleocean circulation. Geochimica et Cosmochimica Acta (1999) 63:1489–1505.[CrossRef][Web of Science]
Ali MY, Watts AB, Hill I. A seismic reflection profile study of lithospheric flexure in the vicinity of the Cape Verde Islands. Journal of Geophysical Research (2003) 108(B5):2239–2263.[CrossRef]
Andreasen R, Peate DW, Brooks CK. Magma plumbing systems in large igneous provinces: Inferences from cyclical variations in Palaeogene East Greenland basalts. Contributions to Mineralogy and Petrology (2004) 147:438–452.[CrossRef][Web of Science]
Baker J, Waight T, Ulfbeck D. Rapid and highly reproducible analysis of rare earth elements by multiple collector inductively coupled mass spectrometry. Geochimica et Cosmochimica Acta (2002) 66(20):3635–3646.[CrossRef][Web of Science]
Baker J, Peate DW, Waight T, Meyzen C. Pb isotopic analysis of standards and samples using a 207Pb–204Pb double spike and thallium to correct for mass bias with a double focusing MC-ICP-MS. Chemical Geology (2004) 211:275–303.[CrossRef][Web of Science]
Barker AK, Holm PM, Peate DW, Baker JA. A five million year record of compositional variations in mantle sources to magmatism on Santiago, southern Cape Verde archipelago. in preparation.
Bonadiman C, Beccaluva L, Coltorti M, Siens F. Kimberlite-like metasomatism and ‘garnet signature’ in spinel-peridotite xenoliths from Sal, Cape Verde archipelago: Relics of a subcontinental mantle domain within the Atlantic Ocean lithosphere? Journal of Petrology (2005) 46:2465–2493.
Bryce JG, DePaolo DJ. Pb isotope heterogeneity in basaltic phenocrysts. Geochimica et Cosmochimica Acta (2004) 68(21):4453–4468.[CrossRef][Web of Science]
Christensen BP, Holm PM, Jambon A, Wilson JR. Helium, argon and lead isotope composition of volcanics from Santo Antão and Fogo, Cape Verde Islands. Chemical Geology (2001) 178:127–142.[CrossRef][Web of Science]
Class C, Goldstein SL, Altherr R, Bachèlery P. The process of plume–lithosphere interactions in the ocean basins—the case of Grande Comore. Journal of Petrology (1998) 39:881–903.[CrossRef][Web of Science]
Courtney RC, White RS. Anomalous heat flow and geoid across the Cape Verde Rise: Evidence for dynamic support from a thermal plume in the mantle. Geophysical Journal of the Royal Astronomical Society (1986) 87:815–867.[Web of Science]
Dasgupta R, Hirschmann MM, Stalker K. Immiscible transition from carbonate-rich to silicate-rich melts in the 3 GPa melting interval of ecologite + CO2 and genesis of silica-undersaturated ocean island lavas. Journal of Petrology (2006) 47:647–671.
Dasgupta R, Hirschmann MM, Smith ND. Partial melting experiments of peridotite + CO2 and genesis of alkalic ocean island basalts. Journal of Petrology (2007) 48:2093–2124.
Dash P, Ball MM, King GA, Butler LW, Rona PA. Geophysical investigation of the Cape Verde Archipelago. Journal of Geophysical Research (1976) 81(29):5249–5259.
Davidson JP, Hora JM, Garrison JM, Dungan MA. Crustal forensics in arc magmas. Journal of Volcanology and Geothermal Research (2005) 140:157–170.[CrossRef][Web of Science]
Davies GR, Norry MJ, Gerlach DC, Cliff RA. A combined chemical and Pb–Sr–Nd isotope study of the Azores and Cape Verde hot-spots: the geodynamic implications. In:. Magmatism in the Ocean Basins. Geological Society, London, Special Publications—Saunders AD, Norry MJ, eds. (1989) 42:231–255.
Dosso L, Bougault H, Joron JL. Geochemical morphology of the North Mid-Atlantic Ridge, 10°–24°N: Trace element-isotope complementarity. Earth and Planetary Science Letters (1993) 120:443–462.[CrossRef][Web of Science]
Doucelance R, Escrig S, Moreira, M. Gariepy C, Kurz M. Pb–Sr–He isotope and trace element geochemistry of the Cape Verde Archipelago. Geochimica et Cosmochimica Acta (2003) 67(19):3717–3733.[CrossRef][Web of Science]
Duprat HI, Friis J, Holm PM, Grandvuinet T, Sørensen RV. The volcanic and geochemical development of São Nicolau, Cape Verde Islands: Constraints from field and 40Ar/39Ar evidence. Journal of Volcanology and Geothermal Research (2007) 162:1–19.[CrossRef][Web of Science]
Gerlach DC, Cliff RA, Davies GR, Norry M, Hodgson N. Magma sources of the Cape Verdes archipelago: Isotopic and trace element constraints. Geochimica et Cosmochimica Acta (1988) 52:2979–2992.[CrossRef][Web of Science]
Hart SR. A large-scale isotopic anomaly in the southern hemisphere mantle. Nature (1984) 309:753–757.[CrossRef]
Herzberg C, Asimow PD. Petrology of some oceanic island basalts: PRIMELTS2.XLS software for primary magma calculation. Geochemistry, Geophysics, Geosystems (2008) 9. Q09011, doi:10.1029/2008GC002057.
Hirose K, Kushiro I. Partial melting of dry peridotites at high pressure; Determination of composition of melts segregated from peridotite using aggregates of diamond. Earth and Planetary Science Letters (1993) 114:447–489.
Hirschmann MM, Kogiso T, Baker MB, Stolper EM. Alkalic magmas generated by partial melting of garnet pyroxenite. Geology (2003) 31:481–484.
Hoernle K, Tilton G, Le Bas MJ, Duggan S, Garbe-Schönberg D. Geochemistry of oceanic carbonatites compared with continental carbonatite: mantle recycling of oceanic crustal carbonate. Contributions to Mineralogy and Petrology (2002) 142:520–542.[Web of Science]
Holm PM, Wilson JR, Christensen BP, Hansen L, Hansen SL, Hein KH, Mortensen AK, Pedersen R, Plesner S, Runge MK. Sampling the Cape Verde mantle plume: Evolution of melt compositions on Santo Antão, Cape Verde Islands. Journal of Petrology (2006) 47:145–189.
Holm PM, Grandvuinet T, Friis J, Wilson JR, Barker AK, Plesner S. An 40Ar–39Ar study of the Cape Verde hotspot: Temporal evolution in a semi-stationary plate environment. Journal of Geophysical Research (2008) 113. B08201, 2007JB005339.
Huang S, Frey F. Recycled oceanic crust in the Hawaiian Plume: evidence from temporal geochemical variations within the Koolau Shield. Contributions to Mineralogy and Petrology (2005) 149:556–575.[CrossRef][Web of Science]
Jørgensen J. Ø. A geochemical study of carbonatites from the Cape Verde Islands, Basaltic rocks from São Vicente and fenites from Brava. In: Unpublished PhD Thesis (2003) University of Copenhagen: Geological Institute. 1–135.
Klerkx J, Deutsch S, De Paepe P. Rubidium, strontium content and strontium isotopic composition of strongly alkalic basaltic rocks from the Cape Verde Islands. Contributions to Mineralogy and Petrology (1974) 45:107–118.[Web of Science]
Klügel A, Hoernle KA, Schminke H.-U, White J. DL. The chemically zoned 1949 eruption on La Palma (Canary Islands): Petrologic evolution and magma supply dynamics of a rift zone eruption. Journal of Geophysical Research (2000) 105:5997–6016.[CrossRef]
Kystol J, Larsen LM. Analytical procedures in the rock geochemical laboratory of the Geological Survey of Denmark and Greenland. Geology of Greenland Survey Bulletin (1999) 184:59–62.
LaTourette T, Hervig RL, Holloway JR. Trace element partitioning between amphibole, phlogopite and basanite melt. Earth and Planetary Science Letters (1995) 135:13–30.[CrossRef][Web of Science]
Le Bas MJ. Nephelinitic and basanitic rocks. Journal of Petrology (1989) 30:1299–1312.
Le Bas MJ. IUGS reclassification of the high-Mg and picritic volcanic rocks. Journal of Petrology (2000) 41(10):1467–1470.
Lodge A, Helffrich G. Depleted swell root beneath the Cape Verde islands. Geology (2006) 34:449–452.
Longpré M-A, Troll VR, Hansteen TH. Upper mantle magma storage and transport under a Canarian shield-volcano, Teno, Tenerife (Spain). Journal of Geophysical Research (2008) 113. B08203, doi:10.1029/2007JB005422.
Luais B, Telouk P, Albarède F. Precise and accurate neodymium isotopic measurements by plasma-source mass spectrometry. Geochimica et Cosmochimica Acta (1997) 61:4847–4854.[CrossRef][Web of Science]
Montelli R, Nolet G, Dahlen FA, Masters G, Engdahl ER, Hung S.-H. Finite-Frequency Tomography reveals a variety of plumes in the mantle. Science (2004) 30:338–343.
Mordick BE, Glazner AF. Clinopyroxene thermobarometry of basalts from the Coso and Big Pine volcanic fields, California. Contributions to Mineralogy and Petrology (2006) 152:111–124.[CrossRef][Web of Science]
Nikogosian IK, Elliott T, Touret J. LR. Melt evolution beneath thick lithosphere: a magmatic inclusion study of La Palma, Canary Islands. Chemical Geology (2002) 183:169–193.[CrossRef][Web of Science]
Nimis P. Clinopyroxene geobarometry of magmatic rocks. Part 2. Structural geobarometers for basic to acid, tholeiitic and mildly alkalic magmatic systems. Contributions to Mineralogy and Petrology (1999) 135:62–74.[CrossRef][Web of Science]
Norman MD, Garcia MO. Primitive magmas and source characteristic of the Hawaiian plume: petrology and geochemistry of shield picrites. Earth and Planetary Science Letters (1999) 168:27–44.[CrossRef][Web of Science]
Norrish K, Chappell BW, eds. X-ray Fluorescence Spectrometry. Physical Methods in Determinative Mineralogy (1977) 2nd edn. London: Academic Press.
Pim J, Peirce C, Watts AB, Grevemeyer L, Krabbenhoeft A. Crustal structure and origin of the Cape Verde Rise. Earth and Planetary Science Letters (2008) 272:422–428.[CrossRef][Web of Science]
Pollitz FF. Two-stage model of African absolute motion during the last 30 million years. Tectonophysics (1991) 194:91–106.[CrossRef][Web of Science]
Prytulak J, Elliott T. TiO2 enrichment in ocean island basalts. Earth and Planetary Science Letters (2007) 263:388–403.[CrossRef][Web of Science]
Putirka K. Magma transport at Hawaii: Inferences based on igneous thermobarometry. Geology (1997) 25(1):69–72.
Putirka KD. Mantle potential temperatures at Hawaii, Iceland, and the mid-ocean ridge system, as inferred from olivine phenocrysts: evidence for thermally-driven mantle plumes. Geochemistry, Geophysics, Geosystems (2005a) 6(5). Q05L08, 2005GC000915.
Putirka KD. Igneous thermometers and barometers based on plagioclase and liquid equilibria; tests of some existing models and new calibrations. American Mineralogist (2005b) 90:336–346.
Putirka K, Mikaelian H, Ryerson FJ, Shaw H. New clinopyroxene–liquid thermobarometers for mafic, evolved and volatile-bearing lava compositions, with applications to lavas from Tibet and the Snake River Plain, ID. American Mineralogist (2003) 88:1542–1554.
Rhodes JM. Geochemical stratigraphy of lava flows samples by the Hawaii Scientific Drilling Project. Journal of Geophysical Research (1996) 101:11729–11746.[CrossRef]
Roeder PL, Emslie RR. Olivine–liquid equilibrium. Contributions to Mineralogy and Petrology (1979) 29:275–289.
Ryabchikov ID, Ntaflos T, Kurat G, Kogarko LN. Glass-bearing xenoliths from Cape Verde: evidence for a hot rising mantle jet. Mineralogy and Petrology (1995) 55:217–237.[CrossRef][Web of Science]
Schwarz S, Klügel A, Wohlgemuth-Ueberwasser C. Melt extraction pathways and stagnation depths beneath the Madeira and Desertas rift zones (NE Atlantic) inferred from barometric studies. Contributions to Mineralogy and Petrology (2004) 147:228–240.[CrossRef][Web of Science]
Serralheiro A. A geologia da Ihla de Santiago (Cabo Verde). Boletim do museu e laboratório mineralógico e geológico da Faculdade de Ciências (1976) 14:157–369.
Stroncik NA, Klügel A, Hansteen TH. The magmatic plumbing system beneath El Hierro (Canary Islands): constraints from phenocrysts and naturally quenched basaltic glasses in submarine rocks. Contributions to Mineralogy and Petrology (2009) in press.
Sun S.-s, McDonough WF. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In:. Magmatism in the Ocean Basins. Geological Society, London, Special Publications—Saunders AD, Norry MJ, eds. (1989) 42:313–345.
Yang H.-J, Kinzler RJ, Grove TL. Experiments and models of anhydrous basaltic olivine–plagioclase–augite saturated melts from 0·001 to 10 kbar. Contributions to Mineralogy and Petrology (1996) 124:1–18.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||