Geochemistry of Jurassic Oceanic Crust beneath Gran Canaria (Canary Islands):
Implications for Crustal Recycling and Assimilation
KAJ HOERNLE ∗
DEPARTMENT OF VOLCANOLOGY AND PETROLOGY, GEOMAR, WISCHHOFSTR. 1–3, D24148 KIEL, GERMANY
RECEIVED MAY 2, 1997; REVISED TYPESCRIPT ACCEPTED JANUARY 9, 1998
Trace elements and Sr–Nd–Pb isotopes have been analyzed on INTRODUCTION
sedimentary and igneous (metabasalt, metadiorite and metagabbro) Overwhelming evidence that the mantle is heterogeneous samples from the Jurassic oceanic crust beneath Gran Canaria on both small (mm) and large (global) scales has come (Canary Islands). The igneous crust exhibits extreme heterogeneity from isotopic studies of ocean island basalts (OIB), fresh in87Sr/86Sr (0·7029–0·7052),206Pb/204Pb (18·2–20·8) and mid-ocean ridge basalt ( MORB) glasses, mantle xenoliths
208Pb/204Pb (38·1–41·3). Leaching experiments indicate that and ultramafic massifs (e.g. Zindler & Hart, 1986). Re- seawater alteration has elevated the 87Sr/86Sr ratio but has not cycling of oceanic crust through subduction zones is appreciably affected 143Nd/144Nd (0·51295–0·51306). An generally accepted as one of the major processes, if Sm–Nd isochron gives an age of 178± 17 Ma, which agrees not the major process, causing heterogeneity within the with the age predicted from paleomagnetic data. Hydrothermal mantle (e.g. Chase, 1981; Hofmann & White, 1982).
alteration near the ridge axis has increased207Pb/204Pb (15·59– Studies of Pb isotopes in OIB and MORB have been 15·73),208Pb/204Pb (as well asD7/4Pb andD8/4Pb),238U/ used to constrain the time necessary to recycle oceanic
204Pb (l) and Ce/Pb but has not appreciably changed 206Pb/ crust through subduction zones and have come up with
204Pb. The large range in 206Pb/204Pb and 208Pb/204Pb reflects widely differing estimates ranging from 300 to 2000 Ma radiogenic ingrowth withl being as high as 107. Portions of the (e.g. Sun, 1980; Silver et al., 1988; Hanan & Graham, Jurassic ocean crust have trace element and isotope characteristics 1996; Thirlwall, 1997). Estimates of recycling ages are within the range found at St Helena, the Atlantic type locality for critically dependent on the effects of hydrothermal al- the HIMU (highl) mantle end-member. Evaluation of the published teration near the ridge axis and the subduction process isotopic data for Gran Canaria volcanic rocks indicates that the on the Pb isotope systematics of ocean crust.
isotopic composition of these melts primarily represents the com- Ocean island basalts have long been considered to be position of their mantle sources rather than crustal assimilation. one of the major means for evaluating mantle hetero- geneity, based on the belief that ocean crust, in contrast to continental crust, does not significantly contaminate ascending plume magmas. Indeed, several factors indicate that ocean crust is less likely to contaminate plume- derived magmas than continental crust. These factors include the smaller thickness, higher density and lower
KEY WORDS:Canary Islands; oceanic crustal recycling; seafloor hydro- abundance of most incompatible elements in oceanic thermal alteration–metamorphism; HIMU; crustal assimilation in ocean crust. In addition, oceanic crust is more similar in isotopic composition to mantle melts than to typical continental island basalt
Fig. 1.Map of Canary Islands located in the eastern North Atlantic Ocean offthe coast of northwest Africa. Approximate locations of the M25 (156 Ma) and S1 (175 Ma) paleomagnetic anomalies are also shown (Roeser, 1982; Klitgord & Schouten, 1986; Roest et al., 1992).
crust. Several recent studies, using thorium, oxygen and GENERAL GEOLOGY
osmium isotope systems, however, suggest that crustal The Canary Islands, located in the Atlantic Ocean off contamination can affect the chemistry of ocean island
the coast of NW Africa (Fig. 1), sit on Jurassic ocean magmas (e.g. Nicholson et al., 1991; Marcantonio et al.,
crust, some of the oldest in situ crust in the ocean basins.
1995; Widom & Shirey, 1996; Eiler et al., 1996a, 1996b;
The age of the crust beneath the Canary Islands can be Thirlwall et al., 1997). To more fully evaluate the role of
bracketed by the 175 Ma S1 magnetic anomaly between crustal contamination on ocean island magmas, it is
the easternmost Canary Islands and northwest Africa necessary to improve our understanding of the range in
and the 156 Ma M25 magnetic anomaly between the composition of altered and aged oceanic crust.
westernmost islands of La Palma and Hierro (Roeser, In this study, an extensive trace element and Sr–Nd–Pb
1982; Klitgord & Schouten, 1986; Roest et al., 1992).
isotope data set is presented for the Jurassic oceanic crust
Gran Canaria, the third easternmost island, lies on ocean beneath Gran Canaria. As the Canary Islands are located
crust in the Jurassic magnetic quiet zone between these on some of the oldest oceanic crust, this crust serves as
two anomalies. The thickness of the oceanic crust to an end-member for evaluating the effects of (1) alteration
the north and northeast of Gran Canaria is ~10 km processes, (2) radiogenic ingrowth and (3) contamination
(Schmincke et al., 1997; Ye et al., 1998). Layer 1 consists of plume-derived, ocean-island melts. Furthermore, be-
of ~6 km of mid-Miocene to Jurassic sediments and cause of the proximity of Gran Canaria to the African
possibly a Jurassic carbonate platform. Layer 2 (pillows continental margin (Fig. 1) and its old age, Gran Canaria
and dikes or sills) is ~2 km thick, whereas crustal layer 3 is situated on one of the thickest packages of sediments
(intrusives, primarily gabbros) has a thickness of ~2·5 km.
(~6 km) of all ocean islands, enhancing the possibility
The 15 my subaerial volcanic history of Gran Canaria of sediment assimilation during magma ascent. Finally,
(Schmincke, 1982; Hoernle & Schmincke, 1993a, 1993b) volcanic and plutonic rocks from Gran Canaria have been
can be divided into two major cycles of volcanic activity:
extensively studied for trace element and Sr–Nd–Pb–O
(1) Miocene or shield cycle (~10–15 Ma) and (2) post- isotopic compositions (e.g. Sun, 1980; Schmincke, 1982;
Miocene rejuvenated (post-erosional) cycle(s) (~0–5·5 Crisp & Spera, 1987; Cousens et al., 1990, 1993; Hoernle
Ma). The tholeiitic to alkalic shield basalts (~14–15 Ma) et al., 1991; Hoernle & Schmincke, 1993a, 1993b; Freundt
form the oldest and most voluminous subaerially exposed
& Schmincke, 1995; Thirlwall et al., 1997; Wiechert et
unit. At 14 Ma, a composite peralkaline rhyolite–basalt al., 1997). To date, evidence for crustal assimilation has
ignimbrite (P1) was erupted, resulting in caldera for- only been noticed in the Miocene volcanic rocks (Freundt
mation. Peralkaline rhyolites and trachytes, and rare
& Schmincke, 1995; Thirlwall et al., 1997). Two major
alkali basalts and intermediate rocks, were erupted for goals of this paper are (1) to constrain the age of ocean
the next ~1 my. Between ~10 and 13 Ma, volcanic crust recycling and the origin of the HIMU component
rocks became more SiO2 undersaturated, ranging from in OIB, and (2) to evaluate the role of crustal assimilation
on the chemistry of OIB. trachytes to phonolites, including rare basanites and
nephelinites. In summary, volcanic and plutonic rocks Within sample B9117, the degree of hydrothermal al- teration, as indicated by the abundance of secondary became more evolved and SiO2 undersaturated with
decreasing age during the Miocene cycle of volcanism. phases such as chlorite, Fe-actinolite and epidote, in- creases from B9117.1 (relatively fresh) to B9117.3 (the Eruption and magma production rates also generally
decreased during the Miocene. most altered). Igneous samples are mafic (7·4–10·8%
MgO, 78–206 ppm Ni and 79–516 ppm Cr) and have After an ~4·5 my hiatus in volcanism, the post-Miocene
cycle began ~5·5 my ago with eruption of small volumes tholeiitic compositions (48–53% SiO2).
of highly undersaturated nephelinitic and basanitic lavas.
As eruption rates and magma production rates increased with decreasing age, alkali basalt became dominant. A
ANALYTICAL METHODS single interval of tholeiitic pahoehoe flows (representing
Chips from the interior of samples (to minimize alteration several km3 in volume) was erupted during this time
after subaerial exposure) were ground to a flour in an interval. The peak of activity was reached ~4 my ago.
agate mill. Plagioclase separates were prepared by picking Between 3·5 and 4·0 Ma, complete suites of alkali basalt
under a binocular microscope and then rinsed with cold to trachyte and basanite to phonolite were erupted, with
3 N HNO3. Trace elements on the samples were analyzed more evolved compositions dominating. Although the
with a VG Plasmaquad PQ1 inductively coupled plasma- degree of differentiation generally increases up-section,
mass spectrometer at the Geological Institute, Christian basalts and evolved volcanic rocks are intercalated
Albrechts University in Kiel, using the method of Garbe- throughout this time interval. Between 0 and 3·0 Ma,
Scho¨nberg (1993). The trace element data, replicate eruptives consisted primarily of basanite, nephelinite and
analyses and standards (except for house standards) run melilite-bearing nephelinite.
during the study are presented in Table 1.
The Sr, Nd and Pb isotopic compositions of whole-rock and plagioclase samples were determined by thermal- SAMPLE DESCRIPTION ionization mass spectrometry ( TIMS) at the Universities of California in Santa Barbara (on a Finnigan MAT261 The carbonate and silicic sediment samples (layer 1)
mass spectrometer) and Santa Cruz (on a VG Sector 54- occur as xenoliths within late Pliocene and Quaternary
30 mass spectrometer). Analytical procedures are the mafic lava flows and come from Deep Sea Drilling Project
same as those outlined by Hoernle & Tilton (1991). A (DSDP) Site 397, ~100 km southeast of Gran Canaria.
second split of powder for samples 303903, 303905 and The igneous crustal samples (layers 2 and 3) occur as
303906 (denoted by a.w.—acid washed) were boiled in cobbles within a Miocene (~14 Ma) fanglomerate exposed
a mixture of 50% 6 N HCl and 50% 8 N HNO3for 1 at the bottom of Barranco (Canyon) de Balos in the SE
h. The residues were analyzed for their Sr and Nd portion of Gran Canaria, and are interpreted to have been
isotopic composition. The measured87Sr/86Sr ratio was brought to the surface in phreatic eruptions (Schmincke et
normalized within-run to 86Sr/88Sr= 0·1194 and then al., 1998). These samples consist primarily of greenschist
to an 87Sr/86Sr value of 0·710235 for NBS987. The facies metabasalts, metadiorites and metagabbros, but
143Nd/144Nd ratio was normalized within-run to146Nd/
also include some relatively fresh coarse-grained gabbros
144Nd = 0·7219 and then to a 143Nd/144Nd value of with 2 cm long plagioclase. Some metabasalt clasts appear
0·51185 for the La Jolla standard. 143Nd/144Nd ratios to represent fragments of pillows or dikes, as is evident
were measured in dynamic mode. The Pb isotope data from their fine-grained margins.
are corrected for mass fractionation using a factor of Petrography, mineral chemistry, and major and trace
0·11% per atomic mass unit for the Santa Cruz data element data from X-ray fluorescence (XRF ) have been
(samples 303903, 303905 and 303906) and 0·125% for presented by Schmincke et al. (1998) and therefore are
the Santa Barbara data (all other samples).
only briefly summarized here. Primary igneous minerals are calcic plagioclase, augitic clinopyroxene ± ortho- pyroxene and Fe–Ti oxides. The primary mineralogy has been modified to variable extents at upper greenschist
TRACE ELEMENT AND ISOTOPE facies conditions, as reflected by secondary actinolite or
ferroactinolite, Fe–Mg chlorite, epidote, sphene, ilmenite DATA
±hematite, Ab-rich plagioclase, muscovite, biotite and The trace element data for Jurassic ocean crust samples iron sulfide. Most fine-grained metabasalts (e.g. sample from Gran Canaria are presented in Table 1. Immobile 303905) consist almost entirely of the greenschist facies incompatible elements, such as Th, Nb, Ta and the rare phase assemblages; however, many of the coarser-grained earth elements (REE), of igneous samples, show good samples contain igneous plagioclase and pyroxene with correlations with each other (Fig. 2). Th and U also correlate well (Fig. 3). Incompatible element abundances optically fresh cores (e.g. samples B914, B9112, 412938).
Table1:Traceelementconcentrations(inppm)fromICP-MSforsamplesfromoceancrustlayer1(sediments),layer2(pillowlavas,sheetflowsand dikes)andlayer3(gabbros)beneathGranCanaria;alsoshownaremeansforanalysesofstandardsBHVO-1andBIR-2andblanks,aswellas1r standarddeviationsofBHVO-1 Sample:917c91249303905303906B9117.1B9117.2B9117.3303903B913B914B914B914B9112B9112412938B9114BHV0-1SDBIR-2Blank Crustal layer:1122222333333333 SedimentSedimentTholeiiticTholeiiticTholeiiticTholeiiticTholeiiticTholeiiticTholeiiticTholeiiticPlagioclasePlagioclaseTholeiiticTholeiiticTholeiiticTholeiiticn=6 xenolithxenolithbasaltbasaltbasaltbasaltbasaltgabbrogabbrogabbroseparateseparategabbrogabbrogabbrogabbro Li14·511·87·6813·00·020·020·570·6119·60·9502·150·240·881·031·545·250·063·320·000040 Sc4·925·2842·227·444·442·948·438·926·340·63·633·2631·231·435·747·734·50·145·70·000130 Cu65·124·947·672·615·714·916·746·882·340·96·856·7340·039·754·158·915031310·000070 Zn28·455·459·984·044·844·750·668·210142·415·014·055·254·425·9121123275·90·000070 Ga6·175·7812·715·515·715·315·613·313·512·619·019·012·012·210·014·423·40·616·00·000010 Rb31·735·51·055·920·520·440·710·311·402·010·820·831·281·360·171·399·680·290·230·000020 Sr11392·915812312912112960·236·491·216216172·974·069·496·5422141170·000190 Y8·578·9525·722·723·122·423·122·016·28·800·360·42114·414·45·3012·824·62·416·50·000010 Zr25·621·931·045·534·331·635·019·029·77·100·710·5918·117·03·6011·6170514·90·000010 Nb6·505·164·7111·710·710·49·943·489·080·510·100·093·593·640·211·2618·10·90·560·000000 Ba13113944·668·131·229·936·454·013·338·035·235·623·525·913·324·513637·200·000240 La14·015·13·7210·56·696·438·122·606·840·851·091·082·632·550·391·1615·80·30·630·000080 Ce28·029·88·4321·714·113·716·16·3513·412·031·701·716·986·940·962·8437·70·62·020·000020 Pr3·233·401·292·771·951·942·040·981·720·300·190·190·850·840·150·465·310·160·400·000000 Nd12·012·46·6212·18·658·458·795·187·181·650·770·754·204·210·922·4424·00·62·470·000020 Sm2·492·682·293·082·412·432·501·901·950·680·170·151·371·420·410·935·990·221·160·000010 Eu0·530·570·820·910·840·840·860·710·720·420·550·540·540·540·230·522·000·090·530·000010 Gd2·262·563·183·492·922·993·012·702·330·950·160·171·901·790·591·366·110·141·860·000020 Tb0·310·340·620·630·530·540·530·520·420·210·020·020·350·350·120·270·920·030·380·000010 Dy1·611·704·303·893·553·593·583·632·771·410·120·142·402·390·861·925·150·232·610·000010 Ho0·280·300·920·810·890·760·760·800·580·320·020·030·530·520·200·420·950·030·580·000010 Er0·780·802·792·382·442·212·262·401·720·950·060·061·571·550·581·262·460·081·740·000010 Tm0·110·110·400·340·460·320·320·340·250·140·010·010·230·220·080·190·330·010·250·000010 Yb0·640·682·682·142·272·102·092·271·640·950·070·061·521·490·561·301·970·061·670·000020 Lu0·090·100·400·310·340·300·310·330·250·140·010·010·220·220·090·190·270·010·250·000010 Hf0·660·621·211·531·131·111·030·680·990·260·020·010·620·590·120·384·450·230·600·000020 Ta0·530·430·280·700·590·560·190·530·040·040·040·220·210·010·071·100·060·040·000000 Pb3·8926·90·313·560·670·360·130·541·940·410·140·160·320·290·250·832·080·083·280·000020 Th3·623·220·321·260·870·710·720·240·620·120·060·060·230·220·030·151·200·070·030·000000 U0·520·550·070·330·230·140·130·050·160·020·000·000·050·040·010·040·420·030·010·000000
[normalized to primitive mantle after Hofmann (1988)] 18·2–20·8, 207Pb/204Pb = 15·59–15·73, 208Pb/204Pb = 38·1–41·3 (Fig. 7). Surprisingly, analyses of three different for representative crustal samples are compared with
Neogene tholeiites from Gran Canaria and MORB in portions of a single basalt sample (B9117) yielded almost the entire observed range (Table 2b). The 206Pb/204Pb Fig. 4a. Of the crustal samples, the gabbros have lower
incompatible element abundances than the basalts. All and 208Pb/204Pb isotope ratios for these three samples correlate positively with the degree of hydrothermal crustal samples have lower abundances of highly and
moderately incompatible immobile elements than the alteration. In the Pb isotope diagrams, almost all samples plot above the field for Atlantic N-MORB and the Neogene tholeiites and have nearly flat patterns of the
Northern Hemisphere Reference Line [NHRL, from mildly incompatible elements, in contrast to the steep
Hart (1984)]. Sample B9117.3 with the highest 206Pb/
patterns of the Neogene tholeiites, which cross over the
204Pb plots slightly below the NHRL in the 207Pb/204Pb crustal patterns for Y and the HREE from Tb to Yb.
diagram, within the field for St Helena. The entire whole- The crustal samples, however, have similar incompatible
rock data set, in particular the data for B9117 samples, element abundances to MORB, which also have flat
falls within error of a line with a slope (0·049) equivalent HREE patterns. The shapes of the incompatible element
to an age of 170 Ma, in agreement with the ages patterns for some basalt samples (e.g. 303906) are re-
determined from Sm–Nd and Ar–Ar isotope systematics.
markably similar to those for basalts from HIMU ocean
In the thorogenic Pb isotope diagram, the data for sample islands such as St Helena (Fig. 4b).
B9117 form a line with a slope equivalent to Th/U of Sr–Nd–Pb isotope data are presented in Table 2 for
3·8 (j=3·9).
ocean crust samples from Gran Canaria and a tholeiitic
The initial Pb isotope data for whole-rock samples fall gabbro xenolith in the 1949 eruption on La Palma.
between or within the fields for Atlantic N-MORB and Samples from Gran Canaria and La Palma form an
sediments (Fig. 7b). Whole-rock sample B913 and the elongated, horizontal field in the Sr vs Nd isotope cor-
fresh plagioclase separate from B914 are within error of relation diagram (Fig. 5), which overlaps the field for
Jurassic Atlantic N-MORB. The whole-rock sample for fresh samples of Atlantic MORB. In contrast to the143Nd/
B914, which contains secondary phases (such as actinolite,
144Nd ratio (0·51294–0·512306), which falls completely
chlorite and sulfide) indicative of hydrothermal alteration, within the range for Atlantic N-MORB, the 87Sr/86Sr
has similar initial 206Pb/204Pb but significantly higher ratio (0·7029–0·7052) exhibits a relatively large range
initial 207Pb/204Pb and 208Pb/204Pb, and is displaced to- extending from the N-MORB field to values that are
wards the sediment field.
significantly more radiogenic. The Sr and Nd isotopic composition of the Gran Canaria samples is similar to the composition of altered (120 Ma) upper oceanic crust in the western Atlantic (Fig. 5; Staudigel et al., 1995). A
DISCUSSION fresh plagioclase mineral separate from sample B914 has
Effect of post-exhumation alteration on U a value of 0·70288, which plots within the field for
and Pb concentrations Atlantic N-MORB. The87Sr/86Sr isotope ratio correlates
positively with H2O content [data from Schmincke et al. Alteration occurring after exhumation of the crustal (1998)]. Leaching experiments were carried out on three samples on Gran Canaria (e.g. alteration related to the samples (303903, 303905 and 303906) with high 87Sr/ formation of the fanglomerate containing the crustal
86Sr and high H2O. After acid-washing, the residues samples and post-depositional circulation of groundwater yielded significantly lower Sr isotope ratios but Nd isotope or hydrothermal fluids associated with later magmatic ratios within analytical error (see Table 2 and Fig. 5). activity at Gran Canaria) has affected the concentration The Sm–Nd data define an isochron with an age of of some elements, as is best illustrated by sample B9117.
178±17 Ma (1r) with a mean square weighted deviate The initial Pb isotope data for the three portions of ( MSWD) of 2·52 (Fig. 6) and an initial143Nd/144Nd ratio B9117 show relatively large differences (e.g. 206Pb/204Pb of 0·51277. The 143Nd/144Nd ratio does not correlate ranges from 17·9 to 19·0). Several observations suggest with 1/Sm, ruling out the possibility of two-component that thel and j ratios, calculated from the measured mixing to explain the variation in Nd isotope ratio. The U, Th and Pb concentrations, of samples B9117.2 and Sm–Nd age data, as well as40Ar/39Ar single- and multiple- B9117.3 do not reflect the long-term, time-integrated crystal laser dates of plagioclase (164 ± 3 Ma) and ratios. First, the initial Sr and Nd isotope data for the hornblende (173 ± 1 Ma) (Schmincke et al., 1998), three portions of B9117 are within error of each other.
provide the strongest evidence that these samples are Second, the Pb isotope data for these samples fall within from the Jurassic oceanic crust and not the Canary error of a 170 Ma reference isochron (Fig. 7a). Third, plume. the excellent linear correlation in the206Pb/204Pb vs208Pb/
The Pb isotope ratios for whole-rock samples from 204Pb diagram indicates that the time-integrated j for these samples was 3·9 ( Th/U = 3·8). Only sample Gran Canaria exhibit extreme variation:206Pb/204Pb =
Fig. 2. Correlations between immobile incompatible trace elements for igneous samples from the Gran Canaria oceanic crust. The estimated U concentrations (Μ; Table 3) for samples B9117.2 and B9117.3 before they were brought to the surface at ~14 Ma are connected to the measured values (Α; Table 1).
B9117.1 has a jof 3·9; samples B9117.2 and B9117.3 time-integrated Th/U ratio of 3·8. On plots of U vs immobile elements, such as Nb, Ta and REE (e.g. Fig. 2), have substantially higher j values of 5·2 and 5·9, re-
spectively. These observations suggest that the initial Pb the estimated U concentrations (Table 3) for samples B9117.2 and B9117.3 plot closer to the trends formed isotope ratios were similar and that recent mobilization
of U, Th and/or Pb has occurred. by the other samples than the measured U concentrations.
A plot of Th vs U for whole-rock samples suggests that Under oxidizing conditions U can have a valence of
6+, forming the uranyl ion (UO22+). As the uranyl ion all crustal samples may originally have had a Th/U ratio of ~3·8 (Fig. 3).
forms compounds that are soluble in water, it becomes
highly mobile under oxidizing conditions. Th, on the Using these U concentrations, we see that the initial Pb isotope ratios (using an age of 170 Ma) are much other hand, only occurs in the tetravalent state and thus
forms compounds that are generally insoluble in water closer, but they are still outside error of each other. Recent hydrothermal alteration, associated with volcanism on (Faure, 1986). Assuming that Th has not been mobilized
during alteration, the time-integrated U concentration Gran Canaria, may also have affected the Pb con- centrations of samples B9117.2 and B9117.3. If we es- can be estimated using the measured Th and the inferred
Fig. 3.Th vs U for ocean crust samples. Symbols same as in Fig. 2.
timate Pb concentrations for samples B9117.2 and did not change the 143Nd/144Nd isotope ratio outside error (Fig. 5). Third, the good correlation between147Sm/
B9117.3 by assuming that these samples have the same
initial208Pb/204Pb isotope composition as sample B9117.1, 144Nd and143Nd/144Nd has a slope equivalent to an age of 178±17 Ma (Fig. 6), similar to that expected from then the206Pb/204Pb and207Pb/204Pb isotope ratios for all
three samples are within error (Table 3). In conclusion, seafloor paleomagnetic data in the region and obtained by40Ar/39Ar age dating. The 147Sm/144Nd ratios range post-exhumation alteration appears to have removed U
and added Pb, resulting in an increase injand a decrease from 0·15 to 0·27 in the whole rocks, resulting in an increase of up to 0·0003 in the 143Nd/144Nd ratio over inlin some but not all samples. Pre-exhumationland
Ce/Pb ratios were as high as 107 and 134, respectively, as 178 my.
The87Sr/86Sr isotope ratios for all three B9117 samples is illustrated by samples B9117.2 and B9117.3 (Table 3).
and sample B914 fall within the field for Atlantic N- MORB (Fig. 5 and Table 2). Seawater alteration, how- ever, has increased the Sr isotopic composition of the Effect of seafloor alteration–metamorphism other samples (0·7037–0·7052), reflecting high water/
and aging on trace element and isotopic rock ratios. Not only do the residues of all three leached compositions of oceanic crust samples have significantly lower 87Sr/86Sr than the un- Both hydrothermal alteration–metamorphism and low- leached residues (Fig. 5), but the87Sr/86Sr isotope ratio temperature alteration occur within the igneous portion of also correlates positively with H2O content. In addition, the oceanic crust. In normal oceanic crust, hydrothermal the lack of correlation of Rb and Sr with immobile alteration will occur as long as the newly formed crust elements reflects the high mobility of both of these is near the ridge axis. An age of 10 Ma is probably elements. The measured 87Rb/86Sr ratios range from a reasonable upper limit for hydrothermal alteration 0·01 to 0·13, correlating with a maximum increase of (Staudigel et al., 1981), in the absence of later seamount 0·0003 in87Sr/86Sr. The fresh samples have the lowest or ocean island volcanism. In contrast, low-temperature 87Rb/86Sr ratios, whereas the altered samples have the alteration will continue, although at low rates, as long as highest ratios.
there is sufficient permeability within the crust. Both hydrothermal alteration and radiogenic ingrowth A number of observations indicate that neither meta- have played a major role in the evolution of the Pb morphism nor later low-temperature alteration affects isotope systematics. In Pb isotope diagrams, most of the the Nd isotopic system. First, the Sm and Nd abundances crustal data plot above the field for Jurassic Atlantic N- and the Sm/Nd and 143Nd/144Nd ratios correlate with MORB (Fig. 7b). Although it cannot be ruled out that immobile elements (e.g. Fig. 2). Second, leaching of three the source for the Jurassic ocean crust beneath Gran Canaria had higher 207Pb/204Pb and 208Pb/204Pb than greenschist facies samples with an aqua regia solution
Fig. 4. Incompatible multi-element diagrams normalized to primitive mantle (Hofmann, 1988). (a) Immobile incompatible elements of tholeiitic basalts and gabbros from the ocean crust beneath Gran Canaria [data from Table 1 and Schmincke et al. (1998) for Zr and Ti] are compared with Miocene and Pliocene plume-derived tholeiites from Gran Canaria (Hoernle & Schmincke, 1993a) and with average enriched (E)-MORB and normal (N)-MORB from Sun & McDonough (1989). (b) As illustrated by tholeiitic basalt sample 303906, some Gran Canaria ocean crust samples have similar incompatible element characteristics to HIMU basanites from St Helena [sample 68 from Chaffey et al. (1989)].
most depleted mantle, comparison of the Pb isotopic falls within the field for Jurassic N-MORB in both Pb isotope diagrams, whereas the altered whole-rock sample composition of the fresh plagioclase separate with the
altered whole rock for B914 suggests that hydrothermal has similar 206Pb/204Pb but significantly higher 207Pb/
204Pb and208Pb/204Pb. This relationship is consistent with alteration played a role in elevating these ratios. The
initial Pb isotopic composition for the plagioclase separate hydrothermal alteration adding a crustal Pb component
Table2:IsotopicdataforGranCanariaoceancrustsamplesandonegabbroxenolithinthe1949eruptiononLaPalma (a)SrandNdisotopicdata 87Sr/86Sr87Rb/86Sr(87Sr/86Sr)T143Nd/144NdeNd147Sm/144Nd(143Nd/144Nd)TeNd(T) GranCanaria Sedimentlayer1 917cwhole-rock0·709643(19)0·810·707680·511985(3)−12·70·120·51185−11·2 91249whole-rock0·715426(13)1·110·712750·511978(5)−12·90·130·51183−11·4 DSDP397-60-4whole-rock0·710291(12)1·210·707370·511965(7)−13·10·110·51184−11·3 DSDP397-101-1whole-rock0·723619(13)0·090·723400·512080(7)−10·90·110·51196−9·03 DSDP397-30-1whole-rock0·709362(23)1·480·705790·511664(12)−19·00·110·51154−17·2 DSDP397-40-2whole-rock0·709363(16)0·120·709080·512131(6)−9·890·110·51201−8·08 DSDP397-49-1whole-rock0·709288(25)0·110·709030·512157(8)−9·380·110·51203−7·52 Basaltlayer2 303905whole-rock0·705243(10)0·020·705200·513004(8)7·140·210·512776·89 303905a.w.acid-washed0·704716(10)0·513032(18) 303906whole-rock0·704004(8)0·140·703670·512949(9)6·070·150·512787·00 303906a.w.acid-washed0·703822(14)0·512935(10) B9117.1whole-rock0·702862(13)0·010·702830·512965(4)6·380·170·512787·01 B9117.2whole-rock0·702859(11)0·010·702830·512962(11)6·320·170·512776·83 B9117.3whole-rock0·702923(17)0·020·702880·512954(25)6·160·170·512766·72 Gabbrolayer3 303903whole-rock0·703716(14)0·010·703680·513042(11)7·880·220·512807·36 303903a.w.acid-washed0·703146(21)0·513026(18) B913whole-rock0·704399(14)0·110·704130·512958(10)6·240·160·512786·96 B914whole-rock0·702992(28)0·060·702840·513061(8)8·260·250·512797·15 B914plagioclase0·702879(21)0·010·702840·512938(15)5·850·130·512807·36 LaPalma(gabbro) 241921Bwhole-rock0·703272(13)0·080·703100·512980(8)6·670·170·512816·67
Table2:IsotopicdataforGranCanariaoceancrustsamplesandonegabbroxenolithinthe1949eruptiononLaPalma (b)Pbisotopicdata 206Pb/204Pb207Pb/204Pb208Pb/204Pblj(206Pb/204Pb)T(207Pb/204Pb)T(208Pb/204Pb)T GranCanaria Sedimentlayer1 917c19·05915·69939·2998·677·2018·8315·6938·77 9124918·92215·70039·1771·316·0918·8915·7039·11 DSDP397-60-419·01815·74539·2016·574·4018·8415·7438·96 DSDP397-101-118·90315·69939·09221·41·6818·3315·6738·79 DSDP397-30-118·97515·72639·01510·43·4018·7015·7138·72 DSDP397-40-218·94915·70139·18317·62·1618·4815·6838·86 DSDP397-49-118·95115·69539·18215·32·6118·5415·6838·85 Basaltlayer2 30390518·15515·60538·07814·24·7217·7715·5937·51 30390618·38515·59738·2465·873·9518·2315·5938·05 B9117.118·44315·59638·45621·63·9417·8715·5737·74 B9117.118·44715·58038·401 B9117.219·35715·66339·55325·45·2318·6815·6338·43 B9117.320·75315·72841·30365·55·8719·0015·6438·06 Gabbrolayer3 30390318·22115·59138·0975·834·9618·0615·5837·85 B91319·06715·62038·8415·304·0018·9315·6138·66 B914w.r.18·66415·63838·4863·116·2018·5815·6338·32 B914plag.18·54515·57638·2550·0018·5415·5838·03 Pliocenebasalts RNB6018·99415·52539·039 RN124918·93115·50938·98824·94·6318·9115·5138·96 Analysesofresiduesafterboilingfor1hinamixtureof50%6NHCland50%8NHNO3aredenotedasacid-washed(a.w.).Theerrorsinparenthesesafterthe measuredisotoperatiosare2randpertaintothelastdigits.Srblanksmeasuredduringthisstudywere<300pg;PbandNdblankswereΖ100pg.Agecorrections weremadeusinganage(T)of170Maforallsamples,exceptfortheLaPalmagabbro(T=156Ma).InitialSr–Nd–Pbisotopicdataforsedimentsamplesfrom DSDPSite397(usingtheageofdeposition)andSrandNdisotopedataforRN1249andRNB60samples,aswellasPb,UandThconcentrationsforRN1249, havebeenpresentedbyHoernleetal.(1991).
Fig. 5. 87Sr/86Sr (measured) vs143Nd/144Nd (measured) isotope correlation diagram for Balos ocean crust samples from Gran Canaria (triangles), a tholeiitic gabbro xenolith from the 1949 eruption on La Palma (circle), and the field for Lanzarote tholeiitic gabbro xenoliths (Vance et al., 1989; E.-R. Neumann & M. Whitehouse, unpublished data), showing that oceanic crust is present beneath the entire Canary Island chain. After boiling samples from Gran Canaria for 1 h in a mixture of 50% 6 N HCl and 50% 8 N HNO3, residues (Α) have lower Sr but similar Nd isotopic compositions, illustrating the effects of seawater alteration. Also shown are fields for (1) Atlantic N-MORB between 30°N and 6°S, excluding area influenced by Sierra Leone hotspot (1–3°N) (Ito et al., 1987; Dosso et al., 1991; Schilling et al., 1994), (2) St Helena (Chaffey et al., 1989), and (3) composites of the upper ~800 m of ~120 Ma ocean crust from DSDP–ODP Sites 417 and 418 in the western Atlantic (Staudigel et al., 1995).
Fig. 6.In the147Sm/144Nd vs143Nd/144Nd diagram, the Balos whole-rock and plagioclase samples define an isochron with an age of 178±17 Ma (1r) with a mean square weighted deviate ( MSWD) of 2·52 and an initial143Nd/144Nd ratio of 0·51277. An age of 178±17 Ma agrees well with that inferred from paleomagnetic data (see Fig. 1).
Fig. 7.(a) Measured Pb isotope ratios for the oceanic crust beneath Gran Canaria (from Table 2). In the206Pb/204Pb and207Pb/204Pb diagram, the whole-rock data fall within error of a 170 Ma reference isochron; whereas in the206Pb/204Pb vs208Pb/204Pb isotope diagram, the samples form an array with Th/U of ~3·8. In the uranogenic Pb diagram, all samples except B9117.3 lie above the field for Atlantic N-MORB and the Northern Hemisphere Reference Line (NHRL; Hart, 1984). Sample B9117.3 has206Pb/204Pb and207Pb/204Pb ratios similar to those for basalts from St Helena, the Atlantic type locality for the HIMU component in ocean islands. The208Pb/204Pb ratio, however, is significantly higher. As shown in the thorogenic diagram (solid cross), a208Pb/204Pb ratio similar to that for St Helena samples could have been generated if the Th/U ratio had been 2·5 instead of 3·8 over the last 170 my. Compositions (open cross) similar to those of lavas from Mangaia, the most extreme example of HIMU in ocean islands, could evolve if portions of the crust similar to sample B9117.3 were allowed to age another 50 my and if the crust had Th/U=2·2 over its entire ~220 my life-span (i.e. since formation at a mid-ocean ridge). Fields for the Gran Canaria Early and Late Miocene volcanic rocks are shown in the thorogenic Pb isotope diagram. (b) Initial Pb isotope data. In both Pb isotope diagrams, the data corrected for in situ decay fall between the fields estimated for Jurassic Atlantic N-MORB (assumingl=5 andj=2·5 for the depleted mantle, i.e. MORB source; see White, 1993) and Jurassic Atlantic oceanic sediments (using thelandjmeasured in each sample).
Magmatic plagioclase from sample B914 falls within the field for N-MORB in both Pb isotope diagrams, whereas the hydrothermally altered whole-rock has higher207Pb/204Pb and208Pb/204Pb ratios, which plot within or just below the field for oceanic sediments. The high207Pb/204Pb and208Pb/204Pb (and highD7/4Pb andD8/4Pb) in the whole-rock samples are consistent with the addition of crustal Pb from oceanic sediments to the Jurassic oceanic crust by hydrothermal alteration. The initial Pb isotope ratios for samples B9117.2 and B9117.3 are from Table 3. Data sources not given in Fig. 5 are as follows: N-MORB and sediments (Sun, 1980; Ben Othman et al., 1989; Hoernle et al., 1991), Mangaia (Nakamura & Tatsumoto, 1988), Tubuai (Vidal et al., 1984) and Cameroon Line (Halliday et al., 1988, 1990).
Table 3: Pre-exhumation (>14 Ma) Pb (ppm), U (ppm),l, initial206Pb/204Pb and initial207Pb/204Pb for ocean crust basalt samples B9117.2 and B9117.3
Pb U Nb/U Ce/Pb 206Pb/204Pb 207Pb/204Pb l (206Pb/204Pb)T (207Pb/204Pb)T
B9117.1 0·67 0·23 47 21 18·44 15·60 22 17·87 15·57
B9117.2 0·22 0·19 56 62 19·36 15·66 55 17·90 15·59
B9117.3 0·12 0·19 52 133 20·75 15·73 107 17·89 15·59
Values were estimated using an age (T) of 170 Ma and assuming that samples B9117.2 and B9117.3 have the samej(3·94) and initial208Pb/204Pb (37·74) as sample B9117.1. A value of 3·94 forjwas also inferred for all three B9117 samples from the slope of these data on the thorogenic Pb isotope diagram (Fig. 7a). Measured Nb and Ce concentrations (Table 1) were used to calculate pre-exhumation Nb/U and Ce/Pb ratios. Measured206Pb/204Pb amd207Pb/204Pb are from Table 2(b)
(derived from oceanic sediments) to the igneous ocean systematics of ocean islands with the most radiogenic Pb crust (Hanan & Graham, 1996). or HIMU-type compositions. Recycling times for end- Hydrothermal alteration can also affect the U/Pb, Th/ member HIMU ocean islands are generally believed to Pb and Ce/Pb ratios but does not appear to significantly be ~2 gy, although recently Hanan & Graham (1996), fractionate Th/U or Nb/U. Neither Nb or Ce appears to with a model similar to the one presented in this study, be significantly affected by seafloor metamorphism, as is proposed that it may be possible to generate St Helena evident from the good correlation between these elements type HIMU within 300 my. It is clear from this study and other immobile elements (e.g. Fig. 2; also Staudigel et that old altered ocean crust can have HIMU-type trace al., 1995). If Nb/U is 47±10 and Ce/Pb is 25±5 in element characteristics (Fig. 4b) and Sr–Nd–Pb isotopic fresh MORB as proposed by Hofmann et al. (1986), then compositions even before subduction. For example, we can estimate the original U/Pb (andl) ratios for these sample B9117.3 has 87Sr/86Sr (0·7029), 143Nd/144Nd samples. The estimatedlratios (18–26, excluding coarse- (0·51295),206Pb/204Pb (20·8) and207Pb/204Pb (15·73) sim- grained gabbros) show only minor variation, falling within ilar to present-day samples from St Helena (87Sr/86Sr= the upper range for MORB (2–25; White, 1993). As is 0·7028–0·7032,143Nd/144Nd=0·51282–0·51297,206Pb/
illustrated by sample B9117 (Table 3), hydrothermal al- 204Pb = 20·4–20·9 and 207Pb/204Pb = 15·71–15·81;
teration near the ridge axis can substantially raiseland Chaffey et al., 1989). Sample B9117.3, however, is distinct Ce/Pb ratios (to values as high as 107 and 133, respectively) from HIMU-type OIB in having even more radiogenic but does not appear to have significantly affected Nb/U 208Pb/204Pb (41·3) than St Helena (39·7–40·2) and all (47–56) or Th/U (3·8), suggesting that the highland Ce/ other end-member HIMU ocean islands (see Fig. 7a).
Pb ratios in the altered samples primarily reflect Pb loss. The higher208Pb/204Pb ratio for B9117.3 reflects a high The high Th/U ratio and the enriched abundances of time-integrated Th/U of 3·8, which falls at the upper immobile incompatible elements in basaltic samples, as end of Th/U found in MORB (e.g. ~1·2–4·3; White, well as the HIMU-type incompatible element char- 1993). If sample B9117.3 had a Th/U of 2·5 (an average acteristics in some samples, may reflect the presence of value for MORB) instead of 3·8, then it would at present young (<0·5 Ga) recycled oceanic crust in the source of have a208Pb/204Pb ratio of 40·1, within the range for St the Jurassic oceanic crust (Figs 3 and 4b). Helena (solid cross in Fig. 7a), instead of the measured
After hydrothermal alteration near the mid-ocean value of 41·3.
ridge, it is unlikely that the Pb isotopic composition will Even though the crust beneath the Canary Islands is be significantly affected by subsequent low-temperature about the oldest in situ oceanic crust that exists at present, seawater alteration, because of the low concentration of it will probably be significantly older before it is sub- Pb in cold seawater (e.g. Faure, 1986). During this ducted, because of its location on a passive margin. If second evolutionary stage of ocean crust (Ζ10 Ma until other portions of the crust in the vicinity of the Canary subduction), the Pb isotopic composition will primarily Islands have the same Pb isotopic composition andlas change as a result of radiogenic ingrowth, especially in sample B9117 andjof 2·4, they will evolve Pb isotope portions of the crust with extremely highlvalues of 55 ratios similar to those for Mangaia (see Fig. 7a), with the (B9117.2) or 107 (B9117.3). Radiogenic ingrowth will most extreme Pb isotopic or HIMU-like composition of not only result in the isotopic composition of the crust all ocean islands, in another 53 my (or when the age of becoming more radiogenic but will also serve to widen
the crust is ~223 Ma). If the ocean crust along the NW the isotopic range of the crust (increase its heterogeneity),
African and Iberian margins ages beyond 223 Ma before especially in older (≥100 my) ocean crust. it is subducted, even more extreme Pb isotopic com-
In summary, the ocean crust beneath Gran Canaria
positions could be generated than those for Mangaia.
had an Sr, Nd and Pb isotopic composition ~170 my
Although intermediate compositions are often ex- ago similar to that expected for the Jurassic MORB
plained through mixing between end-members, the Cam- source (depleted mantle). Hydrothermal alteration near
eroon Line end-member cannot be derived through the ridge axis, subsequent low-temperature alteration and
mixing of HIMU (defined by either St Helena or Man- radiogenic ingrowth, however, have significantly changed
gaia) and MORB in the thorogenic Pb diagram (Fig. 7a).
the Sr and Pb isotopic compositions and U/Pb, Th/Pb
However, assuming an initial Pb isotope composition and Ce/Pb ratios in the oceanic crust beneath Gran
similar to whole-rock sample B914 (Table 2) and a l Canaria from that of present-day MORB.
andjsimilar to those estimated for B9117.3 (Table 3), the Cameroon Line samples with the most radiogenic Constraints on the age of recycled ocean Pb could be generated within 115 my.
crust and the HIMU component in ocean Even though the most extreme HIMU can be generated islands within 223 my, these estimates reflect minimum ages, as the most extreme l was used. Maximum ages can be Age constraints for the recycling of oceanic crust through
a subduction zone are primarily based on the Pb isotope calculated using the bulk composition of the crust (see next
