Journal of Petrology Advance Access originally published online on October 14, 2004
Journal of Petrology 2005 46(1):79-107; doi:10.1093/petrology/egh062
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology vol. 46 issue 1 © Oxford University Press 2004; all rights reserved
Rapid Change of Lava Composition from 1998 to 2002 at Piton de la Fournaise (Réunion) Inferred from Pb Isotopes and Trace Elements: Evidence for Variable Crustal Contamination
1 MAISON DES GÉOSCIENCES, LABORATOIRE DE GÉODYNAMIQUE DES CHAÎNES ALPINES, UMR 5025, BP 53, 38041 GRENOBLE CEDEX 9, FRANCE
2 OBSERVATOIRE VOLCANOLOGIQUE DU PITON DE LA FOURNAISE, INSTITUT DE PHYSIQUE DU GLOBE DE PARIS, 14 RN3, LE 27°KM, 97418 LA PLAINE DES CAFRES, LA RÉUNION, FRANCE
3 INSTITUT DE PHYSIQUE DU GLOBE DE PARIS, 4 PLACE JUSSIEU, 75005 PARIS, FRANCE
RECEIVED JUNE 23, 2003; ACCEPTED JULY 23, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
After an unusually long quiet period of nearly 6 years, in 1998 the Piton de la Fournaise volcano started a new cycle of intense volcanic activity. We report geochemical data on the first nine events (53 samples), from the long-lived initial eruption (six and a half months) of 1998 to the high-flux picritic eruption of January 2002. Pb isotopes and trace elements display systematic, coupled variations, which are mostly confined to the beginning and the end of the period. Two well-defined binary mixing trends are shown by Pb–Pb and Pb–trace element relationships. These trends indicate a change of end-member components between March and June 2001 that coincides with the transition from steady-state basalts to picrites. A three-component mixing model involving a homogeneous plume and two contaminants successfully explains the data. The Pb–Pb relationship requires that two mixing processes occur successively: plume-derived magma interacts first with altered oceanic crust, and the resulting hybrid product then interacts at shallower level with the old lavas constituting the base of the volcanic edifice. Assimilation of edifice material decreased continuously from 1998 to 2002, whereas assimilation of oceanic crust drastically increased during the late-stage picritic eruption. These results suggest that picrites may have resided for an unusually long time at an oceanic crustal level before ascending rapidly through the volcanic edifice with little interaction with channel walls.
KEY WORDS: assimilation; lead isotopes; picrites; Piton de la Fournaise; trace elements
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Some active volcanoes are quiet for thousands of years, some show episodic and catastrophic activity, and a few show continuous activity. The last, which are commonly located on oceanic islands, are extremely useful guides to how volcanic systems work. The compositional evolution of lavas erupting at a single edifice has been investigated at different time-scales. On a scale of several thousand years, Albarède et al. (1997)
inferred changes of petrogenetic processes at Piton de la Fournaise, whereas Lassiter et al. (1996) and Schiano et al. (2001) found evidence of the tapping of different mantle components during the various stages of building of Mauna Kea and Mount Etna, respectively. Rapid compositional changes, such as those occurring during a single eruption or between consecutive eruptions, have been poorly studied until recently. This is simply due to the lack of dense sampling during ancient eruptions. Recent work on Hawaiian (Pietruszka & Garcia, 1999) or Réunion (Albarède & Tamagnan, 1988) active volcanoes has revealed that short-term (50–200 years) variations in lava chemistry cover a large fraction of long-term compositional variability, suggesting that similar processes occur on various time-scales. The continuous sampling now achieved by permanent observatories allows the study of yearly (Garcia et al., 1996, 2000; Semet et al., 2003) to weekly (Rhodes, 1988) variations, allowing new constraints to be placed on the origin of the geochemical signal. In particular, the possibility of developing an understanding of geochemical trends through the observational and physical data collected during eruptions makes the short-term approach very promising.
With 200 eruptions reported during the last 300 years, Piton de la Fournaise (Réunion) is one of the most regularly active volcanoes in the world. Continuous volcanic monitoring started in 1931, but became really detailed in 1980 with the installation of a permanent observatory by the Institut de Physique du Globe de Paris. The volcano recently started a cycle of high eruptive activity (Staudacher et al., 2001) after a quiet period of nearly 6 years, with nine eruptions occurring between March 1998 and January 2002. During this period, volcanologists from the Piton de la Fournaise Observatory systematically sampled all events. The main motivation of this work is to use this dense and well-documented sampling to resolve, if possible, any rapid and systematic variations of lava composition that could be linked to eruption processes. Particular emphasis is placed on potential relationships between changes in lava composition and observational and physical data pertaining to the rate and style of eruption.
Among long-lived radiogenic tracers, Pb isotopes are the most suitable for this study because they display by far the largest variations during the histories of Piton des Neiges (Oversby, 1972) and Piton de la Fournaise (Bosch et al., 1999). In addition, the multi-dimensional Pb isotope system should help to determine the number and nature of components in the Piton de la Fournaise magmas and to show how these components interact during the course of eruptions and between successive eruptions. Trace element concentrations are also presented and used to discuss the nature of the processes.
|
GEOLOGICAL SETTING AND HISTORY OF PITON DE LA FOURNAISE |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Réunion (60 km x 40 km x 3 km) is located in the Indian Ocean at 21°10'S, 55°30'E (Fig. 1a). It is the emergent part of a basaltic cone of 240 km x 200 km x 7 km. The island is inferred to be the present location of the hotspot that created the Deccan Traps, the Chagos–Maldive–Laccadive Ridge, the Mascarene Plateau and Mauritius. The composition of rocks recovered from the South Tethyan suture zone of Pakistan suggests a pre-Deccan phase of activity
73 Myr ago (Mahoney et al., 2002). The age of initiation of the Réunion edifice can be estimated at 7 Ma assuming an average effusion rate of 0·34 m3/s (Lénat & Bachèlery, 1988). The edifice formed on a NW–SE rift (today aseismic) that has controlled the main volcano-tectonic features of the island and its slightly elongated shape. The island is at present made of two main volcanoes, the extinct Piton des Neiges in the NW and the active Piton de la Fournaise in the SE (Fig. 1b). According to common evolution models (Gillot & Nativel, 1989; Gillot et al., 1990), Piton de la Fournaise started to rise 530 kyr ago on the SE flank of the Piton des Neiges. Both volcanoes were active simultaneously until the final eruption of Piton des Neiges 12 kyr ago (Deniel et al., 1992). Based on subaerial gravity and magnetic data, Lénat & Gibert-Malengreau (2001) proposed a different scenario involving one and possibly two other volcanic centres (Les Alizés and Takamaka) during the early stage of island building. Those researchers speculated that Les Alizés volcano was active at the eastern end of the Island and that Piton de la Fournaise grew on its flank.
|
Three major events, either collapses or landslides, marked the growth history of Piton de la Fournaise at 250, 35 and 4·7 ka (Gillot et al., 1994
). The result is three concentric calderas or depressions, the most recent of which (the Enclos Fouqué) is U-shaped (8 km x 13 km) and open to the sea on its eastern side (Fig. 1c). A 400 m high cone with two coalescent summit craters (Bory and Dolomieu) rises inside the Enclos Fouqué. Three centuries of historical volcanic activity indicate that 95% of the eruptions occur inside the Enclos Fouqué, either on the central cone, or, less often, on the plain or seaward slopes. A precise quantification of the Piton de la Fournaise eruptive regime (eruption frequencies, durations and volumes) is available only since 1931. Major eruptions (>50 Mm3) occurred in 1931, 1961, 1977 and most recently in 1998. These events possibly define 15–30 year volcanic cycles (Stieltjes & Moutou, 1989). Between these major events, periods of low (1967–1972, 1987–1992, 1992–1998) and high eruptive activity (1963–1967, 1985–1987) alternate. |
PREVIOUS GEOCHEMICAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The geochemistry of Piton de la Fournaise lavas has been extensively studied, mostly at long time-scales. The most striking feature is the quasi-absence of He, Sr and Nd isotopic variations (Fisk et al., 1988
; Graham et al., 1990; Staudacher et al., 1990; Albarède et al., 1997; Fretzdorff & Haase, 2002) that make the Réunion hotspot source famous for its homogeneity. In detail, picrites have slightly more elevated 87Sr/86Sr (+0·0001), possibly reflecting an interaction with altered cumulates (Albarède et al., 1997), whereas 
Nd displays a subtle, progressive increase (+1·5) during the growth history of the volcano (Luais, 2004). In Sr–Nd isotope space, the Piton de la Fournaise signature (87Sr/86Sr 0·7041; Nd +4) lies midway between mantle end-members (Fisk et al., 1988). Pb isotopes, on the other hand, are more variable (18·79 < 206Pb/204Pb < 19·01), suggesting the involvement of multiple components in the volcano source (Bosch et al., 1999). The general decrease of 20xPb/204Pb isotopic ratios (x = 6, 7, 8) over the last 530 kyr (Bosch et al., 1999) has been ascribed to a progressive increase of the lithospheric contribution relative to that of the plume.
Most of the Piton de la Fournaise lavas have major element compositions transitional between alkali and tholeiitic basalts (Upton & Wadsworth, 1972; Ludden, 1978). Albarède et al. (1997) noted that the lavas evolved from ‘mildly alkalic’ to ‘mildly tholeiitic’ during the growth history of the volcano. Based on the absence of a correlation between alkalinity and 87Sr/86Sr, they suggested that this transition does not reflect variation of the mantle source composition. Instead, following the original idea of O'Hara (1968), they proposed that massive removal of clinopyroxene during magma chamber processes has increased the alkalinity of the liquids. Because clinopyroxene crystallizes before olivine at high pressure (>9 kbar)—and the opposite at low pressure—the decrease in alkalinity could have resulted from crystallization becoming shallower with time. Albarède et al. (1997) also noted that whereas crystallization seems to occur at decreasing depth with time, the extent of differentiation is nearly constant for the steady-state basalts (non-picritic). They suggested that the buffering of compatible elements is best explained if the lavas equilibrate in a low-porosity medium before eruption. This observation, together with the model of damping of geochemical fluctuation of Albarède (1993), calls into question the existence of a large magma chamber. The size of the magma reservoir would not exceed 1 km3 and elements would reside within it for 10–30 years (Albarède, 1993). 230Th–226Ra radioactive disequilibria depend on the dynamics of magma segregation from the mantle and ascent to the surface. They have been used to estimate the size of a storage reservoir (a few km3) and liquid residence time (a few hundred years) (Sigmarsson et al., 2001).
The study of the 1931–1986 eruptions by Albarède & Tamagnan (1988) revealed a negative correlation between Ce/Yb and Ca/Al in volcanic cycles spanning 17 years. This relation suggests that the lower the degree of melting, the greater the degree of crystal fractionation. However, no significant chemical variation has been resolved within individual eruptive events. The most voluminous eruptions (>50 Mm3, in 1931, 1961 and 1977) produced picritic lavas. The abundant olivine crystals (up to 50%) in these lavas have a xenocrystic origin, and are ascribed to magma pulses that swept clean the magma conduits (Albarède & Tamagnan, 1988).
|
1998–2002: A CYCLE OF HIGH ERUPTIVE ACTIVITY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
In 1998, after an unusually long quiet period of nearly 6 years, Piton de La Fournaise started to erupt regularly and continuously. This paper reports on the nine events that occurred between March 1998 and January 2002. The principal characteristics of these eruptions are given in Table 1. Locations of vents and lava flows are shown in Fig. 1c. A detailed description is being prepared by Staudacher et al. and only a short summary is given here. The long-lived 1998 eruption (six and a half months) initiated this period. An initial main vent (Piton Kapor) was active between 9 March and 21 September and produced about 60 Mm3 of lava, a volume similar to that of the three other major eruptions of the century. Two other vents opened on 12 March (Hudson Crater) and 9 August (outside l'Enclos) and produced modest volumes. The three vents of 1998 occur along a line, suggesting that they are related to a single, long fissure. After 10 months of inactivity, five eruptions occurred at regular intervals (every 2–3·5 months) between 19 July 1999 and 13 November 2000, and produced increasing volumes of lava (from 1·5 to 9 Mm3) at increasing fluxes (from 0·7 to 3·3 m3/s). The volume of the following eruption, four and a half months later (27 March 2001) dropped to 4·8 Mm3, but the flux was higher than for the previous eruptions (6·9 m3/s). Shortly after, in June 2001, an eruption produced 9·5 Mm3 at a rate that stayed relatively high (4·2 m3/s). Six months passed before the next eruption, in January 2002, which produced 13 Mm3 of lava at the remarkably high rate of 12·5 m3/s. The two last events produced olivine-rich and picritic lavas, respectively.
|
Based on eruption volumes and flux, four periods can be distinguished during the 1998–2002 eruptive cycle: (I) the voluminous eruption of 1998; (II) the low-flux eruptions (<3 m3/s) that occurred between July 1999 and June 2000 (n = 4); (III) the following three events (October 2000 to June 2001), which produced lavas at higher rates (>3 m3/s); (IV) the high-flux picritic eruption of January 2002. Symbols in the figures are keyed to these groups.
|
SAMPLES AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
From the 174 samples previously analysed for their major element composition (the full dataset is given in Electronic Appendix 1, which may be downloaded from the Journal of Petrology website at http://www.petrology.oupjournals.org), 53 samples covering the life spans of the nine eruptions were selected for Pb isotope and trace element analysis (Tables 2 and 3). Two eruptions were studied in more detail: the March 1998 eruption (10 samples) because of its long duration, and the January 2002 picritic eruption (10 samples) because of its different Pb isotopic signature. One sample (986-115) from a separate vent (Hudson Crater) that was briefly active during the 1998 eruption was also analysed. No samples from the eruption that took place outside l'Enclos in August 1998 were analysed. Most of the samples were collected during the course of an eruption, or soon after cooling. Samples include quenched lava, spatter and Pelé's hair. Two groups can be distinguished: (1) basalts transitional between the tholeiitic and the alkaline series with a matrix containing a small amount of clinopyroxene, plagioclase, and olivine; (2) olivine-rich and picritic lavas containing various amounts of olivine phenocrysts and xenocrysts (up to 40%). These lavas were produced during the late stages of the 2000 and 2001 eruptions, and during the high-flux eruption of January 2002. A complete description of the petrology of the samples will be given elsewhere (Semet et al., in preparation).
|
|
Trace element concentrations were measured in Grenoble by inductively coupled plasma mass spectrometry (ICP-MS) using a Fison PlasmaQuad system and a thulium-doping technique. Following Barrat et al. (1996)
, thulium was used to monitor within-run machine drift. An external correction was also applied by measuring the BHVO-1 standard every two samples. The external precision of the method (2
error), estimated from repeated analyses of standards (BIR and BHVO-1) and sample duplicate analyses (n = 20), was <5% for rare earth elements (REE) and <10% for other elements (Table 2). Because errors correlate during ICP-MS measurements, errors on ratios did not exceed errors on individual elements. To estimate the accuracy of the ICP-MS measurements, all the powders were independently analysed by instrumental neutron activation analysis (INAA) for U, Th, Hf, Ba, Sr, Rb, La, Ce, Sm, Eu, Tb and Yb (at CEA by J. L. Joron). The differences obtained were found to be within the uncertainties given above.
For lead isotope analyses, millimetre-sized chips (about 400 mg) were hand picked and leached with 6N HCl for 1 h at room temperature to remove any possible traces of contaminants at the surface. It was estimated that 30–70% of natural lead is removed during this step. Such high efficiency could be due to the vesicular texture of most samples. Samples were subsequently repeatedly washed with de-ionized water and dissolved in an HNO3–HF mixture. Lead was separated and purified using an HNO3–HBr procedure modified from Lugmair & Galer (1992). During the study, the total procedural blank ranged from 18 to 70 pg (n = 5), which is negligible compared with the amount of Pb extracted. Isotopic compositions were determined by multicollector (MC)-ICP-MS (VG Plasma 54) at ENS Lyon using a Tl 205–203 spike to monitor mass fractionation. White et al. (2000) observed different mass fractionation behaviours for Pb and Tl, and suggested using a fractionation coefficient for Pb derived from that of Tl. However, we found that this approach did not improve the precision of our data. We thus simply corrected the data to 205Tl/203Tl using an exponential law. As discussed extensively since 1998, Tl-corrected data may not be accurate (Rehkämper & Halliday, 1998; Thirlwall, 2002). An external correction was thus applied by measuring the NBS981 standard every two samples. Each pair of sample measurements was corrected using the mean value of the two bracketing NBS981 standards. The external reproducibility of the technique was estimated by measuring repeatedly (n = 20) an in-house standard (a natural sample with isotopic composition similar to those of Réunion samples). The result was 350 ppm (2) for ratios involving 204Pb and 250 ppm (2) or less for other ratios. The accuracy was verified by measuring the same solution by thermal ionization mass spectrometry (TIMS) using a triple-spike technique (at MPI in Mainz). The difference obtained was found to be within the uncertainty of the techniques (Table 3).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Trace elements
When normalized to the average composition of depleted mid-ocean ridge basalt (MORB) (Fig. 2a), trace element distributions show a factor of 10 enrichment in incompatible elements, and slight Th, Pb and Sr positive anomalies. Normalizing to the average composition of 1998–2001 non-picritic lavas reveals fine variations in the trace element pattern (Fig. 2b). Samples from the long-lived eruption of 1998 display slightly depleted patterns, whereas the lava from the separate vent (Hudson) is enriched. July 1999 to June 2000 samples display patterns similar to the average composition used as reference, but some samples have negative Sr–Rb and positive Pb anomalies. A significant depletion of the less incompatible elements (Tb to Lu) characterizes the 2000–2001 non-picritic lavas. Olivine-rich and picritic samples display flat, uniformly depleted patterns, with slight positive Sr and negative Th spikes in the most depleted picrite. When plotted versus time, trace element abundances and ratios display systematic variations (Figs 3 and 4). The concentrations of incompatible elements such as Ce increase during the 1998 eruption. They decrease slightly between 1999 and 2001, and drop to low values in the picritic lavas produced in January 2002. In contrast, less incompatible elements such as Yb seem to display an overall decrease. Ce/Yb progressively increases during the 1998 eruption (from 18·6 to 21·3), and remains elevated (20·9–22·6) during the following events. The short-lived separate vent (Hudson Crater), which formed in 1998, produced the most enriched lavas of the sample set (Ce/Yb = 23). As expected, the abundance of olivine crystals in picrites does not significantly affect most trace element ratios. However, we note that the January 2002 picrites have more elevated Sr/Th, Rb/Th and, to a lesser extent, Ba/Th ratios compared with normal, non-picritic lavas. Ratios involving elements of similar incompatibility (Th/U, Nb/U, Nb/Ta) display small variations that are only slightly larger than analytical uncertainty. Th/U (4·07 ± 0·21 2
) is similar to that estimated for the Bulk Silicate Earth (Allègre et al., 1986) but significantly higher than the value (3·0) reported for the Hawaiian plume (Hofmann & Jochum, 1996). Piton de la Fournaise lavas are also characterized by a remarkable correspondence between measured Th/U and time-integrated Th/U inferred from 208Pb/206Pb (
* = 4·030 ± 0·004 2). Nb/U (38·64–44·83) and Ce/Pb (18·71–29·77) are, within error, within the ranges reported by Hofmann et al. (1986) (47 ± 10 and 25 ± 5, respectively). The lowest Ce/Pb values (below 25) of our dataset are due to higher Pb concentrations, and could, at first glance, be attributed to anthropogenic contamination. These high Pb concentrations (1·8 ppm < Pb < 2·6 ppm) are found, however, to correspond to specific time periods: namely, April–June 1998, the last sample of the July 1999 eruption and the first sample of the following September eruption. Because contamination is expected to affect the sample set randomly, these high Pb contents are thought to reflect the natural compositional variability of the lavas. This idea is also supported by the absence of Pb isotopic anomalies in these samples. Only samples from 2 June 2001 appear to have been markedly contaminated (as confirmed by isotopic measurement; see below).
|
|
|
The 1998–2002 trace element systematics are compared with those of the 1931–1986 time period (Albarède & Tamagnan, 1988
) in Figs 3 and 4 (no data are available for the 1992 eruption). This comparison is based on the premise that no analytical bias exists between the two sample sets, which is difficult to assess given that no reference standard was given by Albarède & Tamagnan (1988).
Restricting the comparison first to non-picritic lavas, the trace element concentrations of the recent volcanic episode correspond to the lowest values previously reported, which generally occur just before or after picritic eruptions. For instance, the Th concentration does not exceed 2·6 ppm between 1998 and 2002 (except for Hudson Crater) although it frequently exceeded this value earlier in the 20th century. An exception is Zr, whose concentration matches the highest values of the 1931–1986 period. The Ce/Yb range is similar in both sample sets, although the Hudson Crater sample displays the highest Ce/Yb ratio since 1931. In contrast, ratios involving highly incompatible elements (such as Ba/Th) show much less variation during the recent volcanic episode than during the whole century. The only notable discontinuity between 1986 and 1998 concerns Zr/Th. This ratio clearly decreases from 90 at the beginning of the 20th century to 75 in 1986, and increases between 1986 and 1998 to reach a value of 85–90. This results from a decrease of Th concentration between 1986 and 1998 (Zr concentration remaining the same). Zr/Th is slightly increasing at present (from 85 in 1998 to >90 in January 2002).
The most recent picritic lavas show degrees of trace element depletion similar to those of picrites from earlier in the century (Fig. 3). Incompatible element ratios display some anomalies in picrites, but these are not systematic. For instance, between 1931 and 1986, low Zr/Th ratios are observed either during picritic events (1977, 1961), or shortly after (1931) or before (1948). This results in regularly spaced, negative spikes along the secular trend of decreasing Zr/Th (Fig. 4). Surprisingly, the January 2002 lavas do not show such a negative Zr/Th spike, but instead have high values compared with previous events. Contrasting behaviours between January 2002 lavas and previous picritic lavas are also observed for Rb/Th and Sr/Th.
Pb isotopes
Despite the strong leaching step for the isotopic analyses, the samples from 2 July 2001 with unusually high Pb contents also have abnormal and highly variable isotopic compositions. This suggests that some contamination occurred before cooling. These samples will not be considered further.
As expected, the variations shown by Pb isotopes are small: 18·870 < 206Pb/204Pb < 18·904; 15·586 < 207Pb/204Pb < 15·602; 38·959 < 208Pb/204Pb < 38·998. These values plot within the field of historical eruptions of Piton de la Fournaise [these data, from Bosch et al. (in preparation), have been acquired on the same instrument and using the same correction technique as in this study]. At such small variations, the apparent range of isotopic ratios versus analytical precision must be carefully examined and is expressed as range/precision ratios (Table 4). The 207Pb/204Pb and 208Pb/204Pb variations, about 1000 ppm, represent only three times the external precision of the technique (2 error: 350 ppm). For this reason, we consider the variations of 207Pb/204Pb and 208Pb/204Pb as insignificant, except for the extreme values. In contrast, the 206Pb/204Pb, 208Pb/207Pb, 207Pb/206Pb and 208Pb/206Pb variations are more clearly outside analytical error. When plotted versus time (Fig. 5), these ratios display clear and systematic variations: during the eruption of March 1998, 206Pb/204Pb decreases continuously whereas 207Pb/206Pb and 208Pb/206Pb increase. The lavas from the short-lived separate vent (Hudson Crater) have higher 208Pb/207Pb than the lavas produced at the same time at the main vent, whereas other ratios do not markedly differ. During the 1999–2001 period, Pb isotopic variations are modest but nevertheless appear systematic when mean values for each event are considered (bold line in Fig. 5). 206Pb/204Pb follows the decreasing 1998 trend but in a more subtle way. In contrast, 207Pb/206Pb and 208Pb/207Pb are constant from 1999 to 2000, then increase and decrease, respectively, in 2001. The picritic lavas from January 2002 have distinctively low 206Pb/204Pb and 208Pb/207Pb, and high 207Pb/206Pb and 208Pb/206Pb. With the exception of the long-lived 1998 eruption there is no resolvable, systematic Pb isotopic evolution during each event. However, we note that the first lavas produced in January 2002 have the most extreme composition (low 206Pb/204Pb and 208Pb/207Pb, high 208Pb/206Pb and 207Pb/206Pb) and that Pb isotopes evolved towards more conventional values during the course of the eruption.
|
|
Data are plotted in 208Pb/206Pb vs 207Pb/206Pb space (Fig. 6a), which is less dependent on analytical error than spaces involving 204Pb in the denominator (higher range/precision ratios). It should be noted that binary mixing translates into straight lines in Pb isotope space as long as normalization is made to the same isotope. Data points from 1998 to 2001 plot on a single array, with the exception of the Hudson Crater sample, which plots well above the line. Data from the January 2002 picrite eruption plot on a sub-parallel trend, which is clearly shifted towards higher 207Pb/206Pb. Because the isotopic variations within individual eruptions from 1999 to January 2002 are close to analytical precision, average values will be used for these events (Fig. 6b). This plot gives a slightly different view on the data and reveals two well-defined mixing trends: the first accounts for the 1998–2000 period, and the second appears to link the Hudson Crater lava with the lavas from October 2000 to January 2002 (high-flux eruptions). The October 2000 and March 2001 samples plot at the intersection of the two trends. These trends, later referred to as trends A and B, imply that at least three distinct components are involved in the genesis of the lavas.
|
Trace elements–Pb isotopes relationship
As with the Pb isotopes, mean values have been plotted for eruptions displaying chemical and isotopic variations close to analytical uncertainty (all except the 1998 eruption). Pb isotopes show clear correlations with the enrichment of incompatible elements in liquids expressed by La/Yb (Fig. 7). This relationship is particularly well defined for the 1998 eruption, during which 206Pb/204Pb decreases as La/Yb increases. This is the first time that such an evolution of trace element and Pb isotopes has been resolved within a single volcanic event at Piton de la Fournaise. The 1999–2001 data plot on the high-La/Yb extension of the 1998 correlation. In contrast, lavas from the separate vent of 1998 (Hudson Crater) and from the January 2002 eruption plot on a different trend, which is orthogonal to the main 1998–2001 trend. The October 2000 to June 2001 eruptions plot at the intersection of the two trends. These two orthogonal trends correspond to the trends A and B previously identified in 208Pb/206Pb vs 207Pb/206Pb space. Although trend B is defined by fewer samples than trend A, its existence is strengthened by the consistency between Pb–Pb isotope and Pb isotope–La/Yb plots.
|
Trend A is characterized by an unexpected, negative correlation between 206Pb/204Pb and La/Yb. It is well established that the relative incompatibility of the elements during the melting process controls the abundances of the elements within the mantle. Given that U and La are more incompatible than Pb and Yb, mixing enriched and depleted mantle components is expected to result in a positive correlation between 206Pb/204Pb and La/Yb. The observed negative correlation requires that the process responsible for La/Yb variations is recent and has not been recorded by Pb isotopes. Thus, trend A cannot result from mixing mantle components and necessarily reflects a superficial, recent process, such as variations in the degree of partial melting or crustal contamination. In contrast, trend B shows a positive correlation between 206Pb/204Pb and La/Yb, which could simply result from mixing sufficiently old enriched (plume type) and depleted (MORB type) components.
The relationships between trace elements and Pb isotopes described above contrast with those observed for historical lavas from Kilauea volcano (Pietruszka & Garcia, 1999). At Kilauea, 206Pb/204Pb correlates positively with La/Yb but negatively with long-term source enrichment (thought to be best recorded by Sr and Nd isotopes), suggesting that both 206Pb/204Pb and La/Yb variations reflect melting or more superficial processes.
In summary, trace elements and isotopes behave consistently along trend B (Hudson lava–2002 picrite) whereas a different process, probably superficial, is required to explain trend A (1998–2001 lavas). The four time periods that have been distinguished based on volume and flux characteristics (see Table 1) also show distinctive chemical and isotopic systematics. (1) Progressive and systematic chemical and isotopic variations occurred during the long-lived eruption of 1998. During this eruption, the Hudson separate vent produced lavas with unusual chemical and isotopic signature. (2) The July 1999–June 2000 eruptions (n = 4) followed the isotopic evolution started in 1998. (3) The October 2000–June 2001 eruptions (n = 3) displayed a change of magma components. (4) The picritic eruption of January 2002 showed an extreme isotopic signature.
Relation with magma flux and volume
The 1998–2002 eruptive cycle is characterized by large variations of volume (from 1 to 60 Mm3) and flux (from 0·6 to 12·5 m3/s) (Table 1), which offer the opportunity to test the relationship between these physical parameters and magma composition (Fig. 8). The concentrations of incompatible elements (such as La) correlate negatively with magma flux. This relationship results from variable abundances of olivine crystals, which are controlled predominantly by magma flux during the time period investigated. Lanthanum also correlates with volume but only for the 1999–2001 eruptions. In contrast to historical Kilauea eruptions (Pietruszka & Garcia, 1999), flux and volume do not correlate with a parameter sensitive to partial melting degree such as La/Yb. Excluding the voluminous eruption of 1998, 206Pb/204Pb decreases with increasing volume and flux. The relationship between magma composition and physical eruption parameters is a new observation that might be very helpful for understanding the eruption mechanisms. Whether the chemical–physical relationship is actively controlled by the composition of deep magma, or alternatively, passively controlled by magma chamber process can be discussed.
|
Inter- vs intra-eruption processes
Processes occurring during eruptions have to be distinguished from those occurring between successive eruptions. Intra-eruption processes, which control how a magmatic reservoir empties, are investigated using both physical (magma volume and flux, seismicity) and geochemical data measured during eruptions. Inter-eruption processes include magma reservoir refilling, magma mixing and the assimilation of country rocks, a combination of which can trigger an eruption. These processes are investigated using seismic and deformation data. Another new approach consists of comparing the compositional variations occurring within eruptions with those occurring between successive eruptions. Such an approach provides an estimate of the relative contribution of the two types of mechanism in controlling magma compositional variability. It has been applied to the 1998–2002 eruption series. Because no clear pattern emerges when raw data are plotted versus time (Figs 3–5), a quantitative approach has been undertaken. Among the numerous tests that can be made, we compared the ranges (Fig. 9a) and the variations (Fig. 9b) of the inter- vs intra-eruption compositions. The range is defined as the difference between the highest and the lowest value of a given time period, and is necessarily positive. It should be noted that inter-eruption ranges are necessarily based on two data points, whereas intra-eruption ranges are based on more than two data points, which could yield overestimated intra-eruption ranges. The variation is the difference between the last value and the first value of a time period, and could be either positive or negative. Inspection of Fig. 9a reveals that the ranges are generally greater during eruptions than between eruptions, except for the eruption of January 2002, in which the first lava significantly differs in composition (La/Yb and 206Pb/204Pb) from the last lava of the preceding event (June 2001). Inter- and intra-eruption variations do not show any systematic trends for trace elements. In contrast, from 1998 to 2001, 206Pb/204Pb tends to increase between eruptions and decrease during the course of an eruption, whereas the opposite behaviour was observed in 2002.
|
(x) = (x highest value) – (x lowest value), and the variation as 
(x) = (last value of x) – (first value of x). (a) Comparison of ranges. (b) Comparison of variations. (See comments in text.) |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Although the chemical and isotopic variations occurring between 1998 and 2002 at Piton de la Fournaise are not of high amplitude, they appear to correlate with the eruption processes. The particularly short time-scale of these variations raises the question of their origin. This issue will be addressed before discussing the nature of the components involved and the mixing relationships.
Origin of the geochemical variations
In a simple scenario, the rapid chemical and isotopic variations of Piton de la Fournaise lavas reflect compositional changes in the plume that melted to form the magmas. However, this scenario can be ruled out for the following reason: at such a short time-scale, the plume rises only a few centimetres. Any centimetre-scale compositional zonation of the plume, if any, has little chance to be preserved in the liquids rising from melting depth to the surface. Another possibility, initially introduced by Sleep (1984), is that the plume material contains small-scale heterogeneities that are preferentially sampled by low melt degrees. Although this explanation cannot be definitively ruled out, it is not supported by the relationship between trace elements and Pb isotopes. Mantle heterogeneities are characterized by variable enrichments in incompatible elements, which have been preserved for a sufficiently long time to be recorded by long-lived radiogenic isotopes. An illustration is the consistent behaviour of trace elements and isotopes observed in oceanic basalts. Thus, sampling old mantle heterogeneities is expected to yield a positive correlation between La/Yb and 206Pb/204Pb. As previously noted, most of the 1998–2002 samples (trend A samples) do not meet that requirement. Only trend B samples (Hudson lavas–picrites) show a positive relationship between La/Yb and 206Pb/204Pb, which, however, probably results from crustal processes as suggested by the relationship between 206Pb/204Pb and eruption flux and volume (Fig. 8). The remaining possibility calls upon superficial processes such as magma chamber processes or interaction of the magmas with the material they ascend through. These shallow-level processes have been shown to play a first-order role in controlling the composition of ocean island basalts (Thirlwall et al., 1997; Garcia et al., 1998; Klügel et al., 2000). Supporting this idea, the composition of melt inclusions trapped in olivine crystals suggests that interaction between magma and channel rocks is common at Piton de la Fournaise (Bureau et al., 1998). Before reaching Piton de la Fournaise summit, the liquids must have ascended through potential contaminants, including the 60 Ma Indian oceanic crust, the old lavas from Piton des Neiges that constitute the base of the volcano and the early products of Piton de la Fournaise. All have variable Pb isotopic compositions (Oversby, 1972; Bosch et al., 1999).
Several observations support a scenario in which the 1998–2002 Pb isotopic variations result from various types of interaction between a relatively homogeneous liquid and crustal material.
(1) Between 1998 and 2002, Pb isotopes (a) display variations that are small (about 20%) compared with the known range of Piton de la Fournaise (Bosch et al., 1999) and (b) have values similar to the less radiogenic values reported for historical lavas. Thus, the 1998–2002 values differ significantly from the most radiogenic values reported for the oldest lavas, which are thought to be the most representative of the plume material (Bosch et al., 1999). Considering as a first approximation only 206Pb/204Pb, and assuming that the Pb concentration of the lavas has not changed during Piton de la Fournaise eruptive history, the 1998–2002 variations could result from assimilation of 0–20% of this old material.
(2) During the eruption of 1998, lavas produced at Hudson Crater have different isotopic compositions (higher 208Pb/207Pb) from lavas produced simultaneously at the main vent (Piton Kapor). Because both vents are fed by the same deep reservoir, assimilation is expected to occur predominantly along one of the two melt channels.
(3) Except for the high-flux eruption of January 2002, chemical and isotopic variations principally occur during, and not between the eruptions (Fig. 9a). In addition, Pb isotopes evolve systematically in opposed directions between and during eruptions. These observations are best explained if liquids interact with material of different composition before erupting.
(4) Pb isotopes correlate both with magma flux and volume (Fig. 8). In particular, a remarkable negative correlation between, on one hand, 206Pb/204Pb and, on the other hand, magma flux and volume is observed for the 1999–2002 time period. Eruption flux varies by a factor of 10 between 1998 and 2002 whereas the cumulative volume plotted versus time (Fig. 10) indicates that magma supply from the mantle at depth has been constant since the beginning of the eruptive cycle (0·63 m3/s). Therefore, the correlation between Pb isotopes and flux is thought to result from crustal processes. How magma flux and magma composition are linked remains to be explained. In a simple scenario, the pressure excess in a magma reservoir that controls magma flux could enhance the rate of erosion and assimilation of surrounding old lavas. On the other hand, rapid ascent to the surface results in limited interaction with channel walls. In the last section (origin of picrites), after having identified mixing components and relationships, it will be shown that both deep magma storage and rapid ascent to the surface are required to explain the Pb isotopic signature of high-flux lavas.
|
Despite the Pb isotopic spread shown by the erupted lavas, the observations listed above suggest that primary melts from the Réunion plume have maintained a uniform composition during the 1998–2002 eruptive cycle. At such a short time-scale, chemical and isotopic variations appear to result from shallow processes, which are discussed below.
Components, mixing relationship and processes
The characteristics of the components involved, as well as the mixing relationships, are best illustrated in 208Pb/206Pb vs 207Pb/206Pb space when mean values are used for events showing isotopic variations close to analytical uncertainty (all except the eruption of 1998) (Figs 6b and 11). This plot reveals two well-defined binary mixing trends. As mentioned above, the first is defined by samples from 1998 to July 2000 (trend A), whereas the second is defined by samples from Hudson Crater (March 1998), as well as the June 2001 and January 2002 eruptions (trend B). The signature of lavas produced in October 2000 and March 2001 plots at the intersection of the two trends. It is worth noting that the two trends are also observed in trace element–Pb isotope space (Fig. 7) as well as other Pb–Pb isotopic spaces with 207Pb and 208Pb as denominator isotope (not shown), which reinforces the observations made in 208Pb/206Pb vs 207Pb/206Pb space (Fig. 6b).
|
Pb isotope relationships suggest that (1) the composition of the end-member components changed between March and June 2001, and (2) variations of mixing proportions occurred during the 1998–2001 and 2001–2002 time periods. To quantify the mixing process, we need to know the number and nature of the end-members. The first constraint comes from the observation that the high 208Pb/206Pb–207Pb/206Pb end-member of trend A is not pure, but results from mixing along trend B. This has the following consequences: (1) three-component mixing can explain the two binary trends; (2) along trend B, the mixing proportions were constant between 1998 and 2001, namely at the hybrid end-member component of trend A; (3) the mixing process B should occur before mixing process A, and thus probably takes place deeper along the magma path. In addition, the observation that historical samples (grey field in Fig. 11; data from Bosch et al., in preparation) plot along the two trends A and B, but not above trend B, indicates that this mixing configuration is stable at the century scale.
We propose that the end-member components include the plume material and two contaminants. In contrast to contaminants, the plume material is expected to be a major component in all lavas, which implies that its signature plots relatively close to all data points (probably within the frame of Fig. 6). It should also plot along trend B because: (1) the plume material is expected to be involved in the first, deepest mixing process, and (2) trend B cannot result from mixing of contaminants, otherwise the March 2001 to January 2002 lavas would contain no plume material. It is possible that the plume component plots at the intersection of the two trends. However, we view this as less likely because four components (plume + three contaminants) would be required. Thus, the signature of the plume material is best represented either by Hudson Crater lavas or by 2002 picritic lavas. On the one hand, picritic lavas are commonly believed to be little modified by magma-reservoir processes (Clague et al., 1991). On the other hand, interaction of picritic liquids with genetically unrelated, deep cumulates has been demonstrated at Piton de la Fournaise (Albarède & Tamagnan, 1988). This apparent disagreement results from the fact that the former study considered high-MgO glasses, and thus near-primitive melts, whereas the second considered lavas rich in xenocrystic olivine having interacted with cumulates. The January 2002 lavas are of the second type, and thus probably do not represent the plume component. Supporting this idea, the less radiogenic Pb isotopic signature of the 2002 picrites compared with normal lavas suggests a reduced plume influence. In contrast, the particularly high enrichment in incompatible elements (high La/Yb) of Hudson Crater lavas identifies these liquids as a likely candidate. In support of this idea, the primitive signature, the unusual crystallization depth (below the crust–mantle boundary), and the liquid state of the Hudson lavas, together with other petrological observations (Bureau et al., 1999; Semet et al., 2003) suggest that these lavas rose unusually rapidly to the surface, without significant interaction with wall rocks, thus preserving the deep signature of the plume. The plume component would therefore represent the low 208Pb/206Pb–207Pb/206Pb end-member of mixing trend B.
Potential contaminants include: (1) the altered Indian oceanic crust; (2) the old lavas from Piton des Neiges and Piton de la Fournaise that constitute the deep levels of the edifice; (3) the oceanic sediments trapped in the volcanic edifice. Because the Indian MORB array is sub-parallel to trend A (Fig. 11), Indian crust cannot be involved directly in mixing trend A. In contrast, old lavas (1–2 Ma) from Piton des Neiges and the early products of Piton de la Fournaise have Pb signatures that plot close to the low 208Pb/206Pb–207Pb/206Pb extension of trend A, indicating a possible interaction between 1998–2001 liquids and these old lava flows.
The nature of the high 208Pb/206Pb–207Pb/206Pb end-member of trend B is not obvious. We first note that the picritic lavas of 1931 plot right on the extension of trend B and, thus, these lavas appear to have undergone the same process (to an even greater extent) as the January 2002 lavas. Some observations suggest that trend B reflects an interaction with the underlying oceanic crust: (1) despite its very large spread, the Indian MORB field has the required isotopic composition (high 208Pb/206Pb–207Pb/206Pb); (2) as required by the configuration of Pb–Pb mixing trends, mixing B should occur before mixing A. As mixing A is inferred to take place within the volcano edifice, mixing B should take place below, probably deeper in the oceanic crust. On the other hand, other observations suggest an interaction with seawater-derived material, as follows. (1) Among the different pre-anthropogenic (0·2–2 Ma) Pb signatures of Indian Ocean seawater (Vlastélic et al., 2001), only that of Madagascar basin plots on trend B extension (Fig. 11). This basin lies precisely between Réunion and the adjacent spreading ridges, namely the Central Indian Ridge and the Southwest Indian Ridge. A 20 Myr record inferred from an Fe–Mn crust (109D-C) located in the south of Madagascar basin (O'Nions et al., 1998) indicates that the Pb isotopic signature of seawater has varied little at this location. The signature of crust 109D-C plots, however, between trend B and the MORB field, possibly reflecting its proximity to the ridge axis during its growth history. (2) Albarède et al. (1997) noted that picritic lavas have slightly higher 87Sr/86Sr at a given 143Nd/144Nd, a characteristic that is indicative of seawater contamination. (3) Ratios involving elements concentrated in seawater (Sr, Rb, Ba) over an element with very low concentration in seawater (Th) are elevated in January 2002 picrites. The two types of observations may be reconciled by involving interaction with hydrothermally altered oceanic crust. Far from continents, the Pb isotopic composition of seawater is dominantly controlled by mantle inputs, which could explain why both MORB and seawater signatures plot along trend B. Alternatively, an interaction with oceanic sediments trapped in the volcanic edifice (Gallard et al., 1999) cannot be excluded because the isotopic signature of these sediments is expected to be similar to that of seawater.
From these considerations, it is suggested that trend B results from mixing plume material with altered oceanic crust, whereas trend A results from mixing material from the volcano edifice with products derived from deeper mixing (the B process).
The contribution of the various components through time has been quantified using the values given in Table 5, and results are shown in Fig. 12. The observed Pb isotopic spread can be explained by varying the fractions of components from the plume, old lavas flows, and the oceanic crust from 95 to 80%, 15 to 0% and 10 to 5%, respectively. The calculated contribution of the oceanic crust is roughly constant between 1998 and 2001 (5%) and doubles in January 2002. In contrast, assimilation of the old lavas that make the deep layers of the volcanic edifice shows a general decreasing trend. The resulting plume fraction increases markedly during the 1998 eruption (by about 10%) and between June 2000 and October 2000, before decreasing in 2001–2002.
|
|
The computed plume fraction broadly correlates with La/Yb variations (Fig. 12), which would be consistent with a plume component having higher La/Yb than the contaminants. Whereas this condition is fulfilled for the oceanic crust, the old lavas of Piton de la Fournaise and the underlying Piton des Neiges have La/Yb ratios similar to present-day lavas (Fisk et al., 1988
; Albarède et al., 1997). Thus, we cannot explain La/Yb variations during the period 1998–2002 only by mixing of the components inferred from the Pb model. This suggests that part of the La/Yb variations observed between 1998 and 2002 results from changes in the extent of melting: the lower the melting degree (high La/Yb), the greater the plume fraction. According to trace element–isotope systematics (Fig. 7), trend B is dominated by mixing between sources with different La/Yb ratios (plume + oceanic crust), whereas trend A (mixing B product + edifice old lavas) results from co-variations between the mixing proportions and the degree of partial melting. Because the proportion of olivine crystals shows large variations in 1998–2002 lavas, MgO can be used as a proxy for olivine abundance. The relationship between crust fraction and MgO content (Fig. 13a) indicates that the lavas with high olivine abundances assimilated a high amount of crust component. For high-flux eruptions (>3 m3/s) the calculated crust fraction increases with magma flux (this is even clearer if the March and June 2001 eruptions are considered as a single event; see Fig. 13b). These observations are based, however, on few events, and only continuous monitoring of future eruptions will tell us if these relationships can be generalized.
|
The progressive upward migration of earthquake hypocentres from 5 km below sea level to the surface that preceded the eruption of 1998 (Staudacher et al., 1998
), together with the systematic geochemical variations seen at the beginning of the eruptive cycle, suggest that a magma batch ascended from melting depth and interacted with the different layers it met on its way (Fig. 14). The assimilation of material constituting the deep levels of the edifice, important at the beginning of the eruptive cycle, decreased continuously between 1998 and 2001, possibly reflecting the progressive formation of dykes that facilitate the passage of magma to the surface, or, alternatively, the progressive emptying of a zoned magma chamber. The systematic decrease of 206Pb/204Pb within individual events from 1998 to 2001 (Fig. 9b) can also be ascribed to a decrease of assimilation of edifice material during the course of the eruptions. Such a systematic behaviour of Pb isotopes supports the existence of a zoned magma reservoir: the beginning of an eruption would sample the upper, most contaminated liquids, which would have resided the longest in the reservoir, whereas the end of an eruption would sample the lower, more primitive and less contaminated liquids. In support of such a scenario, the lavas became more primitive (Semet et al., in preparation) as contamination decreased during the course of the 1998 eruption. In contrast, the late-stage picritic lavas of January 2002 are products of an unusual process, which is discussed below.
|
Origin of Piton de la Fournaise picritic lavas
The origin of Piton de la Fournaise picritic lavas was first ascribed to the accumulation of olivine phenocrysts (Kieffer et al., 1977
; Ludden, 1978; Clocchiatti et al., 1979). The discovery, in picritic olivines, of common deformation features and evidence of disequilibrium with the host magma led Albarède & Tamagnan (1988) to propose a xenocrystic origin of the olivine. They suggested that these crystals originate from cumulates in the magma chamber that were disrupted by, and entrained in, high-flux magmas. Stieltjes & Moutou (1989) proposed the existence of a large, deep (20–30 km) magma reservoir that fed a smaller, shallower (5–10 km) reservoir. The shallow reservoir buffers magma pulses during normal eruptions, and only during picritic eruptions, when high magma supply exceeds the capacity of the shallow chamber, does the magma rise directly from the deep reservoir. Interaction with deep cumulates, and rapid ascent to the surface, has also been proposed as an explanation for olivine-rich lavas from the Hawaiian plume (Garcia, 1996). A contentious issue is the nature of contaminants, if any, and the processes involved during picritic eruptions (Norman & Garcia, 1999; Norman et al., 2002). Whereas it is generally thought that these cumulates originate from previous melting events that fed these volcanoes (Albarède & Tamagnan, 1988; Clague et al., 1995; Garcia, 1996), the contribution of genetically unrelated components, such as underlying oceanic crust, is also advocated (Baker et al., 1996; Eiler et al., 1996; Hauri et al., 1996; Garcia et al., 1998). The relatively unradiogenic Sr, Nd and Pb signature of some picrites is most easily explained by assimilation of lithosphere (Borisova et al., 2002) or altered oceanic crust (Kerr et al., 1996). The January 2002 eruption at Piton de la Fournaise offers an opportunity to investigate the origin of picrites. Lavas from this eruption display physical, petrological and geochemical characteristics that differ from those of most previous events, as follows.
(1) The relatively large volume of lava (13 Mm3) produced in a short time (12 days) requires a high eruption rate (12·5 m3/s), which is about 20 times that calculated for the average melt production rate (0·63 m3/s) between 1998 and 2002. However, the volume is none the less small compared with other picritic eruptions (usually >50 Mm3). For example, the picritic eruption of 1931 produced 130 Mm3 at a mean rate of 7 m3/s. On the other hand, the 1998 event shows that a large volume (60 Mm3) of normal lava could be produced at relatively low rate (3·5 m3/s). From these observations, a distinctive feature of picritic eruptions appears to be high effusion rates, rather than large magma volumes.
(2) The Pb isotopic signature of the picrites indicates interaction with a genetically unrelated component. Olivine contains nearly no Pb (Dolivine–melt = 0·0076; Green, 1994) except in melt inclusions, which are, however, commonly observed. Unless these melt inclusions have very different Pb isotopic composition compared with the common liquids of Piton de la Fournaise, which is unlikely, the contribution of Pb from melt inclusions is expected to be negligible. This is at least suggested by the composition of the olivine-free sample (Pelé's hair, 0201-09-5), which is one of the less radiogenic. Thus, picritic liquids have probably mixed with, or assimilated, unrelated material. According to the three-component mixing model, picrites have undergone pronounced interaction with altered oceanic crust (10% assimilation) but little or no interaction with the overlying volcanic edifice. To estimate how robust these results are, we have used the mixing proportions inferred from Pb isotopes to estimate how assimilation of altered oceanic crust would affect Sr and Nd isotopic ratios. Because seawater contains a significant amount of Sr but very little Nd, the altered oceanic crust is expected to have a Sr isotopic composition overprinted by seawater but a Nd composition similar to that of fresh MORB. Consequently, the composition estimated for the 60 Ma Indian oceanic crust is: [Sr] = 115 ppm, [Nd] = 11 ppm, 87Sr/86Sr = 0·7078 (Hess et al., 1986), Nd = +7·4 (averaged Indian MORB). The composition used for the plume component is that of historical steady-state basalts (Albarède et al., 1997; Luais, 2004): [Sr] = 350 ppm, [Nd] = 26 ppm, 87Sr/86Sr = 0·70405 and Nd = +4·50. For picrites, we found that an increase of oceanic crust fraction from 5 to 10% (Fig. 12) increases 87Sr/86Sr from 0·70411 to 0·70418 and Nd from +4·56 to +4·63. These shifts are similar to those observed in historical picrites (Luais, 2004), suggesting that a 5% increase of assimilation of altered oceanic crust is realistic for picrites.
Flux and isotopic data can be reconciled given that (1) the observed high flux at the surface can explain the absence of extensive interaction with melt channel walls within the edifice, and (2) the process that produces high magma flux at the surface is the same as that causing assimilation at crustal depths. As mentioned above, surface flux is highly variable between 1998 and 2002 whereas the rate of magma supply at the base of the edifice appears to be constant (Fig. 10). This is best explained if the magma reservoir is continuously fed from below, but empties by upward magma migration only when some critical pressure is reached. Therefore, magma storage appears as a likely process for controlling, and thus linking magma flux and assimilation. The remaining, unsolved question concerns the unusually deep and prolonged magma storage suspected for the picritic liquids and its relationship with high olivine content (up to 40% of the mass according to incompatible element depletion in whole-rock analyses). A distinctive feature of the olivine-rich lavas (June 2001 and January 2002) is the low elevation of their eruptive sites (Table 1). Based on the 1985–1992 eruptions, Aki & Ferrazzini (2000) found that low-elevation (<2100 m), flank eruptions were systematically preceded by a longer seismic crisis than summit eruptions. This observation, together with the absence of long-period signals in the precursory crisis of summit eruptions, led those workers to suggest the existence of two distinct paths for the two types of eruption, namely the ‘summit path’ and the ‘rift zone path’. Vent locations and elevations suggest that the magma followed the ‘summit path’ at the beginning of the eruptive cycle (1998–2000) and the ‘rift zone path’ during the late-stage olivine-rich and picritic eruptions. The reason why magma changed path is not obvious. It should be first noted that the January 2002 eruption occurred after 3 years of intense volcanic activity, during which 98 Mm3 of lavas were produced. It can be estimated that the volume of olivine accumulated during this period represents about half of the volume of lava erupted. Melt and fluid inclusion studies (Bureau et al., 1998, 1999) indicate that crystallization and accumulation of olivine occurred over a wide depth range, down to below the crust–mantle boundary. Because melts preferentially pond at a depth where there is a density barrier, the recently formed, dense and voluminous olivine-rich cumulates may have played a role in deepening and increasing magma storage during the late-stage picritic eruption. Contrasting with previous events of the cycle, 206Pb/204Pb increased during the course of the 2002 eruption (Fig. 9b), which can be ascribed to a decrease of assimilation of altered oceanic crust. As for the 1998–2001 events, such an evolution suggests the presence of a zoned magma reservoir, the most contaminated melts being present at the top of it and erupting first. In a simple scenario (Fig. 14), the picrites could result from growing magma batches that pond below olivine-rich cumulates. Once critical pressure is reached, the lavas disrupt the olivine cumulates and ascend rapidly and directly to the surface through the rift zone path. Although this model remains speculative, it successfully accounts for the relationship between magma flux, olivine content, and Pb isotopic composition of the January 2002 picrites.
In summary, flux and geochemical data suggest that the January 2002 picrites have resided longer at greater depth than common lavas. The model proposed agrees with previous views in the sense that it suggests a deeper origin for picrites compared with other lavas (Stieltjes & Moutou, 1989), a rapid ascent to the surface (Albarède & Tamagnan, 1988) and contribution of a material having interacted with seawater (Albarède et al., 1997). However, our short time-scale study allows us to refine the ‘magma pulse’ model of Albarède & Tamagnan (1988). Indeed, if the 1998–2002 period is considered as a single, long eruption, it effectively could be considered as resulting from a deep magma pulse that ultimately produces picrites. However, picrites appear during late-stage, high-flux eruptions, suggesting that they result from passive crustal processes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Previous studies on Hawaiian volcanoes have constrained the time-scale of compositional change at a single edifice to between 1 month (Rhodes, 1988
) and 1 year (Garcia et al., 2000). The cycle of high eruptive activity that started in 1998 at Piton de la Fournaise offers the opportunity to study rapid compositional change at a comparable hotspot. A dense, well-documented sample collection covering nine events from March 1998 to January 2002 has been analysed for Pb isotopes and trace elements, and the following results were obtained. (1) Pb isotopic variations are small, but nevertheless outside analytical error for 206Pb/204Pb, 208Pb/206Pb, 207Pb/206Pb and 208Pb/207Pb.
(2) Systematic co-variations of Pb isotopes and trace elements occur between 1998 and 2002. Variations are essentially confined to the 6 month eruption that initiated the cycle and to the late-stage eruptions of 2001 and 2002.
(3) Two well-defined mixing trends are shown by Pb–Pb isotope and Pb isotope–trace element relationships. The first trend is defined by samples from 1998 to July 2000 (trend A), and the second is defined by samples from 1998 separate vent (Hudson Crater), June 2001 and January 2002 eruptions (trend B). The signature of lavas produced in October 2000 and March 2001 plots at the intersection of the two trends. These observations suggest a change of end-member components between March and June 2001, which precisely corresponds to the appearance of olivine-rich and picritic lavas.
(4) A three-component mixing model involving a homogeneous plume and two contaminants successfully explains the data. Pb–Pb isotope relationships indicate that two types of binary mixing occur successively: plume-derived magmas interact first with altered oceanic crust (mixing B), and the resulting hybrid product interacts at shallower level with old lavas at the base of the edifice (mixing A). Assimilation of edifice material has decreased continuously since the beginning of the cycle, reflecting either the progressive formation of melt channel or the emptying of a zoned magma reservoir. In contrast, the amount of assimilated oceanic crust is relatively constant from 1998 to 2001, and drastically increases between 2001 and January 2002.
(5) The Pb isotope signature of the January 2002 picrites probably results from an unusually long residence time at oceanic crust level. Olivine-rich cumulates formed during previous eruptions of the cycle may have played a role in deepening and increasing magma storage. After rupture of the magma reservoir, the lavas ascended rapidly through the volcanic edifice with little interaction with channel walls.
(6) The geochemical variations occurring within individual eruptions are generally small and close to analytical error (except for the 1998 eruption). However, a quantitative approach reveals that the first liquids are systematically the most contaminated, supporting the existence of zoned magma reservoirs.
These results show that high-precision Pb isotope data can place constraints on shallow-level contamination processes, a topic more commonly studied with O or Os isotopes. In addition, and as previously emphasized by some recent studies (Thirlwall et al., 1997; Garcia et al., 1998) they suggest some caution when discussing mantle heterogeneity from isotopic systematics of island lavas.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
Samples were collected with the help of the Piton de la Fournaise Observatory staff. The authors are grateful to A. Pietruszka, A. Klügel and F. Hauff for their thorough and constructive reviews, and to C. W. Devey for handling the manuscript. D. Bosch is thanked for sharing her unpublished data. We also thank F. Albarède for his advice, N. Arndt for comments on the manuscript, and M. Collombet for discussion. P. Télouk, A. Agranier and F. Keller provided some help with sample analyses.
* Corresponding author. Telephone: (33) 4 76 63 59 08. Fax: (33) 4 76 51 40 58. E-mail: Ivan.Vlastelic{at}ujf-grenoble.fr
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND HISTORY...PREVIOUS GEOCHEMICAL RESULTS1998-2002: A CYCLE OF...SAMPLES AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Aki, K. & Ferrazzini, V. (2000). Seismic monitoring and modeling of an active volcano for prediction. Journal of Geophysical Research 105, 16617–16640.[CrossRef]
Albarède, F. (1993). Residence time analysis of geochemical fluctuations in volcanic series. Geochimica et Cosmochimica Acta 57, 615–621.[CrossRef][Web of Science]
Albarède, F. & Tamagnan, V. (1988). Modelling the recent geochemical evolution of the Piton de la Fournaise volcano, Réunion island, 1931–1986. Journal of Petrology 29, 997–1030.
Albarède, F., Luais, B., Fitton, G., Semet, M., Kaminski, E., Upton, B. G., Bachèlery, P. & Cheminée, J.-L. (1997). The geochemical regimes of Piton de la Fournaise volcano (Réunion) during the last 530 000 years. Journal of Petrology 38, 171–201.[CrossRef][Web of Science]
Allègre, C. J., Dupré, B. & Lewin, E. (1986). Thorium/uranium ratio of the Earth. Chemical Geology 56, 219–227.[CrossRef][Web of Science]
Baker, M. B., Alves, S. & Stolper, E. M. (1996). Petrography and petrology of the Hawaii scientific drilling project lavas: inferences from olivine phenocryst abundances and compositions. Journal of Geophysical Research 101, 11715–11727.[CrossRef]
Barrat, J. A., Keller, F., Amossé, J., Taylor, R. N., Nesbitt, R. W. & Hirata, T. (1996). Determination of rare earth elements in sixteen silicate reference samples by ICP-MS after Tm addition and ion exchange separation. Geostandards Newsletters 20, 133–139.
Borisova, A. Y., Nikogosian, I. K., Scoates, J. S., Weis, D., Damasceno, D., Shimizu, N. & Touret, J. L. R. (2002). Melt, fluid and crystal inclusions in olivine phenocrysts from Kerguelen plume-derived picritic basalts: evidence for interaction with the Kerguelen Plateau lithosphere. Chemical Geology 183, 195–220.[CrossRef][Web of Science]
Bosch, D., Albarède, F. & Telouk, P. (1999). The Piton de la Fournaise volcano (Réunion island, Indian Ocean): temporal evolution from high resolution Pb isotopes. Journal of Conference Abstracts 4, 345.
Bureau, H., Métrich, N., Pineau, F. & Semet, M. P. (1998). Magma conduit interaction at Piton de la Fournaise volcano (Réunion Island): a melt and fluid inclusion study. Journal of Volcanology and Geothermal Research 84, 39–60.[CrossRef][Web of Science]
Bureau, H., Métrich, N., Semet, M. P. & Staudacher, T. (1999). Fluid–magma decoupling in a hot-spot volcano. Geophysical Research Letters 26, 3501–3504.[CrossRef][Web of Science]
Clague, D. A., Weber, W. S. & Dixon, J. E. (1991). Picritic glasses from Hawaii. Nature 353, 553–556.[CrossRef]
Clague, D. A., Moore, J. G., Dixon, J. E. & Friesen, W. B. (1995). Petrology of submarine lavas from Kilauea's Puna Ridge, Hawaii. Journal of Petrology 36, 299–349.
Clocchiatti, R., Havette, A. & Nativel, P. (1979). Relations pétrogénétiques entre les basaltes transitionnels et les océanites du Piton de la Fournaise (île de la Réunion, Océan Indien) à partir de la composition chimique des intrusions vitreuses des olivines et des spinelles. Bulletin de Minéralogie 102, 511–525.
Deniel, C., Kieffer, G. & Lecointre, J. (1992). New 230Th–238U and 14C age determinations from Piton des Neiges volcano, Réunion. A revised chronology for the Differentiated Series. Journal of Volcanology and Geothermal Research 51, 253–267.[CrossRef][Web of Science]
Eiler, J. M., Valley, J. W. & Stolper, E. M. (1996). Oxygen isotope ratios in olivine from the Hawaii scientific drilling project. Journal of Geophysical Research 101, 11807–11813.[CrossRef]
Fisk, M. R., Upton, B. G. J., Ford, C. E. & White, W. M. (1988). Geochemical and experimental study of the genesis of magmas of Reunion Island, Indian Ocean. Journal of Geophysical Research 93, 4933–4950.
Fretzdorff, S. & Haase, K. M. (2002). Geochemistry and petrology of lavas from the submarine flanks of Réunion Island (western Indian Ocean): implications for magma genesis and mantle source. Mineralogy and Petrology 75, 153–184.[CrossRef][Web of Science]
Gallart, J., Driad, L., Charvis, P., Sapin, M., Hirn, A., Diaz, J., de Voogd, B. & Sachpazi, M. (1999). Perturbation to the lithosphere along the hotspot track of La Réunion from an offshore–onshore seismic transect. Journal of Geophysical Research 104, 2895–2908.[CrossRef]
Garcia, M. O. (1996). Petrography and olivine and glass chemistry of lavas from Hawaii Scientific Drilling Project. Journal of Geophysical Research 101, 11701–11713.[CrossRef]
Garcia, M. O., Rhodes, J. M., Trusdell, F. A. & Pietruszka, A. J. (1996). Petrology of lavas from the Puu Oo eruption of Kilauea volcano: III. The Kupaianaha episode (1986–1992). Bulletin of Volcanology 58, 359–379.[CrossRef][Web of Science]
Garcia, M. O., Ito, E., Eiler, J. M. & Pietruszka, A. J. (1998). Crustal contamination of Kilauea volcano magmas revealed by oxygen isotope analyses of glass and olivine from Puu Oo eruption lavas. Journal of Petrology 39, 803–817.[CrossRef][Web of Science]
Garcia, M. O., Pietruszka, A. J., Rhodes, J. M. & Swanson, K. (2000). Magmatic processes during the prolonged Pu'u ’O’o eruption of Kilauea volcano, Hawaii. Journal of Petrology 41, 967–990.
Gillot, P.-Y. & Nativel, P.-E. (1989). Eruptive history of the Piton de la Fournaise volcano, Réunion Island, Indian Ocean. Journal of Volcanology and Geothermal Research 36, 53–65.[CrossRef][Web of Science]
Gillot, P.-Y., Nativel, P.-E. & Condomines, M. (1990). Evolution volcano-structurale du Piton de la Fournaise depuis 0·53 Ma. In: Lénat, J.-F. (ed.) Le Volcanisme de la Réunion. Clermont-Ferrand: Centre de Recherches Volcanologiques, pp. 243–256.
Gillot, P.-Y., Lefèvre, J.-C. & Nativel, P.-E. (1994). Model for the structural evolution of the volcanoes of Réunion Island. Earth and Planetary Science Letters 122, 291–302.[CrossRef][Web of Science]
Graham, D., Lupton, J., Albarède, F. & Condomines, M. (1990). Extreme temporal homogeneity of helium isotopes at Piton de la Fournaise, Réunion Island. Nature 347, 545–548.[CrossRef]
Green, T. H. (1994). Experimental studies of trace-element partitioning applicable to igneous petrogenesis: Sedona 16 years later. Chemical Geology 117, 1–36.[CrossRef][Web of Science]
Hart, S. R., Blusztajn, J., Dick, H. J. B., Meyer, P. S. & Muehlenbachs, K. (1999). The fingerprint of seawater circulation in a 500-meter section of ocean crust gabbros. Geochimica et Cosmochimica Acta 63, 4059–4080.[CrossRef][Web of Science]
Hauri, E. H., Lassiter, J. C. & DePaolo, D. J. (1996). Osmium isotope systematics of drilled lavas from Mauna Loa, Hawaii. Journal of Geophysical Research 101, 11793–11806.[CrossRef]
Hess, J., Bender, M. L. & Schilling, J. G. (1986). Evolution of the ratio of strontium-87 to strontium-86 in seawater from the Cretaceous to present. Science 231, 979–984.
Hofmann, A. W. (1988). Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust. Earth and Planetary Science Letters 90, 297–314.[CrossRef][Web of Science]
Hofmann, A. W. & Jochum, K. P. (1996). Source characteristics derived from very incompatible trace elements in Mauna Loa and Mauna Kea basalts, Hawaii scientific drilling project. Journal of Geophysical Research 101, 11831–11839.[CrossRef]
Hofmann, A. W., Jochum, K. P., Seufer, M. & White, W. M. (1986). Nb and Pb in oceanic basalts: new constraints on mantle evolution. Earth and Planetary Science Letters 79, 33–45.[CrossRef][Web of Science]
Holm, P. M. (2002). Sr, Nd and Pb isotopic composition of in situ lower crust at the Southwest Indian Ridge: result from ODP Leg 176. Chemical Geology 184, 195–216.[CrossRef][Web of Science]
Kerr, A. C., Tarney, J., Marriner, G. F., Klaver, G. T., Saunders, A. D. & Thirlwall, M. F. (1996). The geochemistry and petrogenesis of the late-Cretaceous picrites and basalts of Curaçao, Netherlands Antilles: a remnant of an oceanic plateau. Contributions to Mineralogy and Petrology 124, 29–43.[CrossRef][Web of Science]
Kieffer, G., Tricot, B. & Vincent, P. M. (1977). Une éruption inhabituelle (avril 1977) du Piton de la Fournaise (Ile de la Réunion): ses enseignements volcanologiques et structuraux. Comptes Rendus des Séances de l'Académie des Sciences 285, 957–960.
Klügel, A., Hoernle, K. A., Schmincke, H.-U. & White, J. D. L. (2000). 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 105, 5997–6016.[CrossRef]
Lassiter, J. C., DePaolo, D. J. & Tatsumoto, M. (1996). Isotopic evolution of Mauna Kea volcano: results from the initial phase of the Hawaii scientific drilling project. Journal of Geophysical Research 101, 11769–11780.[CrossRef][Web of Science]
Lénat, J.-F. & Bachèlery, P. (1988). Dynamics of magma transferts at Piton de la Fournaise volcano (Réunion Island, Indian Ocean). In: King, C.-Y. & Scarpa, R. (eds) Modeling of Volcanic Processes. Wiesbaden, pp. 57–72.
Lénat, J.-F. & Gibert-Malengreau, B. (2001). A new model for the evolution of the volcanic island of Réunion (Indian Ocean). Journal of Geophysical Research 106, 8645–8663.[CrossRef]
Luais, B. (2004). Temporal changes in Nd isotopic composition of Piton de la Fournaise magmatism (Réunion Island, Indian Ocean). Geochemistry, Geophysics, Geosystems 5, Q01008, doi:10.1029/2002GC000502.
Ludden, J. N. (1978). Magmatic evolution of the basaltic shield volcanoes of Réunion Island. Journal of Volcanology and Geothermal Research 4, 171–198.[CrossRef][Web of Science]
Lugmair, G. W. & Galer, S. J. G. (1992). Age and isotopic relationships among the angrites Lewis Cliff 86010 and Angra dos Reis. Geochimica et Cosmochimica Acta 56, 1673–1694.[CrossRef][Web of Science]
Mahoney, J. J., Frei, R., Tejada, M. L. G., Mo, X. X., Leat, P. T. & Nägler, T. F. (1998). Tracing the Indian Ocean mantle domain through time: isotopic results from old West Indian, East Tethyan, and South Pacific seafloor. Journal of Petrology 39, 1285–1306.[CrossRef][Web of Science]
Mahoney, J. J., Duncan, R. A., Khan, W., Gnos, E. & McCormick, G. R. (2002). Cretaceous volcanic rocks of the South Tethyan suture zone, Pakistan: implications for the Réunion hotspot and Deccan Traps. Earth and Planetary Science Letters 203, 295–310.[CrossRef][Web of Science]
Norman, M. D. & Garcia, M. O. (1999). Primitive magmas and source characteristics of the Hawaiian plume: petrology and geochemistry of shield picrites. Earth and Planetary Science Letters 168, 27–44.[CrossRef][Web of Science]
Norman, M. D., Garcia, M. O., Kamenetsky, V. S. & Nielsen, R. L. (2002). Olivine-hosted melt inclusions in Hawaiian picrites: equilibration, melting, and plume source characteristics. Chemical Geology 183, 143–168.[CrossRef][Web of Science]
O'Hara, M. J. (1968). The bearing of phase equilibria studies on synthetic and natural systems on the origin and evolution of basic and ultrabasic rocks. Earth-Science Reviews 4, 69–133.
O'Nions, R. K., Frank, M., von Blanckenburg, F. & Ling, H.-F. (1998). Secular variation of Nd and Pb isotopes in ferromanganese crusts from Atlantic, Indian and Pacific Oceans. Earth and Planetary Science Letters 155, 15–28.[CrossRef][Web of Science]
Oversby, V. M. (1972). Genetic relations among the volcanic rocks of Réunion: chemical and lead isotopic evidence. Geochimica et Cosmochimica Acta 36, 1167–1179.[CrossRef][Web of Science]
Pietruszka, A. J. & Garcia, M. O. (1999). A rapid fluctuation in the mantle source and melting history of Kilauea volcano inferred from the geochemistry of its historical summit lavas (1790–1982). Journal of Petrology 40, 1321–1342.[CrossRef][Web of Science]
Rehkämper, M. & Halliday, A. N. (1998). Accuracy and long-term reproducibility of lead isotopic measurements by multiple-collector inductively coupled plasma mass spectrometry using an external method for correction of mass discrimination. International Journal of Mass Spectrometry 181, 123–133.[CrossRef]
Rhodes, J. M. (1988). Geochemistry of the 1984 Mauna Loa eruption: implications for magma storage and supply. Journal of Geophysical Research 93, 4453–4466.
Schiano, P., Clocchiatti, R., Ottolini, L. & Busà, T. (2001). Transition of Mount Etna lavas from a mantle-plume to an island-arc magmatic source. Nature 412, 900–904.[CrossRef][Medline]
Semet, M. P., Joron, J.-L. & Staudacher, T. (2003). The 1998–2002 activity of Piton de la Fournaise, Réunion Island: lessons in magma supply and transfers. Geophysical Research Abstracts 5, 10736.
Sigmarsson, O., Condomines, M. & Bachèlery, P. (2001). Magma dynamics beneath Piton de la Fournaise volcano, Réunion island, as inferred from U-series disequilibria. Journal of Conference Abstracts 6, 807.
Sleep, N. (1984). Tapping of magma from ubiquitous mantle heterogeneities: an alternative to mantle plumes? Journal of Geophysical Research 89, 10029–10041.
Staudacher, T., Sarda, P. & Allègre, C. J. (1990). Noble gas systematics of Réunion island, Indian Ocean. Chemical Geology 89, 1–17.[CrossRef][Web of Science]
Staudacher, T., Bachélery, P., Semet, M. P. & Cheminée, J. L. (1998). Piton de la Fournaise. Bulletin of Global Volcanism Network of the Smithsonian Institution 23, 2–4.
Staudacher, T., Aki, K., Bachèlery, P., Catherine, P., Ferrazzini, V., Hochard, D., Kowalski, P., Ricard, L. P., Villeneuve, N. & Cheminée, J. L. (2001). Piton de la Fournaise volcano, Réunion island, starts a new cycle of high eruptive activity. Journal of Conference Abstracts 6, 825.
Stieltjes, L. & Moutou, P. (1989). A statistical and probabilistic study of the historic activity of Piton de la Fournaise, Réunion Island, Indian Ocean. Journal of Volcanology and Geothermal Research 36, 67–86.[CrossRef][Web of Science]
Thirlwall, M. F. (2002). Multicollector ICP-MS analysis of Pb isotopes using a 207Pb–204Pb double spike demonstrates up to 400 ppm/amu systematic errors in Tl-normalization. Chemical Geology 184, 255–279.[CrossRef][Web of Science]
Thirlwall, M. F., Jenkins, C., Vroon, P. Z. & Mattey, D. P. (1997). Crustal interaction during construction of ocean islands: Pb–Sr–Nd–O isotope geochemistry of the shield basalts of Gran Canaria, Canary Islands. Chemical Geology 135, 233–262.[CrossRef][Web of Science]
Upton, B. G. J. & Wadsworth, W. J. (1972). Aspects of magmatic evolution on Réunion island. Philosophical Transactions of the Royal Society of London, Series A 271, 105–130.
Vlastélic, I., Abouchami, W., Galer, S. J. G. & Hofmann, A. W. (2001). Geographic control on Pb isotope distribution and sources in Indian Ocean Fe–Mn deposits. Geochimica et Cosmochimica Acta 65, 4303–4319.[CrossRef][Web of Science]
White, W. M., Albarède, F. & Télouk, P. (2000). High-precision analysis of Pb isotopic ratios by multi-collector ICP-MS. Chemical Geology 167, 257–270.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. J. Pietruszka, E. H. Hauri, and J. Blichert-Toft Crustal Contamination of Mantle-derived Magmas within Piton de la Fournaise Volcano, Reunion Island J. Petrology, April 20, 2009; (2009) egp016v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. P. Marske, M. O. Garcia, A. J. Pietruszka, J. M. Rhodes, and M. D. Norman Geochemical Variations during Kilauea's Pu'u 'O'o Eruption Reveal a Fine-scale Mixture of Mantle Heterogeneities within the Hawaiian Plume J. Petrology, July 1, 2008; 49(7): 1297 - 1318. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||