Journal of Petrology Advance Access originally published online on September 24, 2004
Journal of Petrology 2005 46(1):109-133; doi:10.1093/petrology/egh064
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology vol. 46 issue 1 © Oxford University Press 2004; all rights reserved
The Cocos and Carnegie Aseismic Ridges: a Trace Element Record of Long-term Plume–Spreading Center Interaction
1 DEPARTMENT OF GEOLOGY, COLGATE UNIVERSITY, 13 OAK DRIVE, HAMILTON, NY 13346, USA
2 DYNAMICS OF THE OCEAN CRUST, IfM-GEOMAR, WISCHHOFSTR. 1–3, 24148 KIEL, GERMANY
3 TETHYS GEOCONSULTING GmbH, WISCHHOFSTR. 1–3, 24148 KIEL, GERMANY
RECEIVED JULY 10, 2003; ACCEPTED JULY 29, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The aseismic Cocos and Carnegie Ridges, two prominent bathymetric features in the eastern Pacific, record
20 Myr of interaction between the Galápagos hotspot and the adjacent Galápagos Spreading Center. Trace element data determined by inductively coupled plasma-mass spectrometry in >90 dredged seamount lavas are used to estimate melt generation conditions and mantle source compositions along the ridges. Lavas from seamount provinces on the Cocos Ridge are alkalic and more enriched in incompatible trace elements than any in the Galápagos archipelago today. The seamount lavas are effectively modeled as small degree melts of a Galápagos plume source. Their eruption immediately follows the failure of a rift zone at each seamount province's location. Thus the anomalously young alkalic lavas of the Cocos Ridge, including Cocos Island, are probably caused by post-abandonment volcanism following either a ridge jump or rift failure, and not the direct activity of the Galápagos plume. The seamounts have plume-like signatures because they tap underlying mantle previously infused with Galápagos plume material. Whereas plume heterogeneities appear to be long-lived, tectonic rearrangements of the ridge plate boundary may be the dominant factor in controlling regional eruptive behavior and compositional variations. KEY WORDS: mantle plume; mid-ocean ridge; Galápagos; abandoned rift; partial melting of the mantle
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Most aseismic ridges are considered to represent the long-term history of hotspot activity as a tectonic plate passes over a mantle plume (e.g. Morgan, 1971
). The Cocos and Carnegie Ridges, located in the eastern Pacific Ocean, have likewise been attributed to the Galápagos plume (Fig. 1; Holden & Dietz, 1972). In this case, however, the system has been complicated by the proximity of the hotspot to the Galápagos Spreading Center (GSC). Because the ridge has been migrating northeastward for more than the past 5 Myr (Wilson & Hey, 1995) and migrated southward before that (Meschede & Barckhausen, 2000), volcanic material from the plume has been erupted variably onto both the Nazca and Cocos plates. Consequently, the accumulated lavas of the Cocos and Carnegie Ridges constitute a record of interaction between the Galápagos plume and the Galápagos Spreading Center for close to 20 Myr (e.g. Werner et al., 1999).
|
The present-day Galápagos archipelago exhibits an anomalously wide variety of geochemical compositions, from enriched, hotspot-like signatures in the west and south to mid-ocean ridge basalt (MORB)-like lavas in the central and northern regions. The spatial zonation has been attributed to a heterogeneous plume and to extensive interaction between the plume and the asthenosphere (e.g. White & Hofmann, 1978
; Geist et al., 1988; White et al., 1993; Harpp & White, 2001). One of the controversial questions about the Galápagos system is whether the geochemical zonation in the present-day archipelago is a long-term phenomenon related to inherent plume heterogeneity (Hoernle et al., 2000) or only the recent result of upper-mantle contamination.
Because of the complex tectonic relationship between the Galápagos Spreading Center and the Galápagos hotspot system, volcanic activity along the Cocos and Carnegie Ridges provides a means of addressing important questions about plume heterogeneity and longevity, as well as plume–ridge interaction dynamics. Seamounts on the Cocos and Carnegie aseismic ridges, as well as the related Malpelo and Coiba ridges were systematically sampled for the first time during dredging operations of the 1999 PAGANINI expedition, R.V. Sonne cruise SO 144-3 (Fig. 1). Two of the fundamental goals of the cruise were to document the temporal variations in Galápagos plume composition and eruptive activity over the past >20 Myr, and to determine the extent of plume interaction with the migrating GSC. Here, we present results from over 90 new inductively coupled plasma-mass spectrometry (ICP-MS) trace element analyses of lavas from seamounts along the aseismic ridges and their implications for the dynamic history of the Galápagos plume. Major element and radiogenic isotopic analyses (Sr, Nd, Pb) of the recovered lavas have been reported, along with a discussion of the magnetic profiles collected during the cruise, by Werner et al. (2003); hafnium isotopic analyses of selected lavas have been presented by Geldmacher et al. (2003). Additional trace element and isotopic data (Sr, Nd, Pb) from the Cocos Ridge and Costa Rica Seamount Province have been reported by Hoernle et al. (2000).
|
BACKGROUND INFORMATION |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Tectonic history
The Galápagos Islands are located on the Equator,
1000 km west of the South American coast. They emerge from an extensive, shallow submarine volcanic platform that marks the western terminus of the Carnegie Ridge (Fig. 1). The Galápagos Spreading Center lies directly north of the archipelago and marks the boundary between the Nazca and Cocos plates.
The tectonic history of the plume-spreading center system has been summarized by Wilson & Hey (1995), Meschede & Barckhausen (2000), Barckhausen et al. (2001) and Werner et al. (2003). On the basis of sea-floor magnetic anomaly analysis, Barckhausen et al. (2001) proposed that spreading initiated on the GSC at 22·7 Ma, causing the breakup of the Farallon plate. At 19·5 Ma, the GSC began a major southward jump, shifting strike by 22° clockwise, where it remained for another 5 Myr. The ridge underwent a second jump south at 14·5 Ma, taking on its current orientation (Meschede & Barckhausen, 2000). Between 14·5 and 9·5 Ma, the Malpelo Ridge was rifted away from the Carnegie Ridge toward its present location. The 14·5 Ma jump was the first in a series of smaller, more frequent southward jumps that continue today (Wilson & Hey, 1995; Barckhausen et al., 2001). The GSC overlay the Galápagos plume between 5 and 9 Ma, but the interactions have been complicated by the multiple southward ridge segment jumps and the formation of the 91°W transform fault at 3 Ma (Wilson & Hey, 1995). Currently, the GSC is 150–250 km north of the hotspot center, presumed to be located slightly west of Fernandina Island (e.g. Kurz & Geist, 1999). Consequently, the relationship between the plume and the spreading center has varied from a ridge-centered plume to an off-axis configuration over the past 20 Myr (Wilson & Hey, 1995).
Description of study area
Werner et al. (2003) have described the morphology of the study area in detail; only the general features are reviewed here.
The Cocos Plate
The Cocos Ridge, long believed to represent the track of the Cocos Plate over the Galápagos hotspot (e.g. Hey, 1977; Lonsdale & Klitgord, 1978), extends over 1000 km between the 91°W transform fault on the GSC and the west coast of Costa Rica, parallel to the current direction of plate motion (35·5°; Gripp & Gordon, 1990; Fig. 1). The Cocos Ridge reaches elevations of less than 1000 m below sea level, making it the most prominent positive bathymetric feature in the eastern Pacific (e.g. Hey, 1977; Walther, 2003). The NE end of the Cocos Ridge and the adjacent seamounts have been dated at 13·0–14·5 Ma (Werner et al., 1999). The ridge narrows slightly to the SW, decreasing progressively in volume toward its intersection with the GSC. Magnetic anomaly studies indicate that the lithosphere underlying the NE end of the Cocos Ridge ranges in age from 15 Ma in the south to 19 Ma in the north (Barckhausen et al., 2001).
Whereas there is a relatively even distribution of seamounts along the Cocos Ridge crest, the greatest concentration of seamounts is on the NW flank (Fig. 1). More than 30 volcanic structures that reach over 1000 m bathymetrically dominate this region, alongside many additional, smaller seamounts. Werner et al. (2003) divided the seamount population along the Cocos Ridge into three regions (Fig. 1): (1) the Costa Rica Seamount Province, the largest with 18 major volcanic structures arranged in three >300 km long chains of seamounts that run parallel to the Cocos Ridge crest; (2) the Cocos Island Province, a cluster of seven major seamounts, Cocos Island, and additional minor structures that extend 200 km to the west and SW of Cocos Island; (3) the Southwest Seamount Province, six major seamounts scattered along the northern flank of the SW Cocos Ridge.
The Nazca Plate
The Galápagos hotspot track on the Nazca Plate is manifested as the 600 km long Carnegie Ridge, which is up to 300 km wide at its eastern end (Fig. 1). The Carnegie Ridge is oriented nearly parallel to the direction of Nazca Plate motion (91°; Gripp & Gordon, 1990). Meschede & Barckhausen (2001) estimated that the most ancient parts of the Carnegie Ridge were formed at around 20 Ma. In general, seamounts along the Carnegie Ridge are less abundant than on the Cocos Ridge and rarely reach more than several hundred meters above base level (Werner et al., 2003); they are more abundant and evenly distributed across the eastern part of the ridge and on the Galápagos Platform (Christie et al., 1992), but decrease in number in the bathymetric saddle centered at 86°W. Dredging operations were less successful along the Carnegie Ridge, often yielding no rocks or heavily manganese-encrusted samples.
The Malpelo Ridge is c. 300 km long and 100 km wide, trending in a northeasterly direction (Fig. 1; Werner et al., 2003). This structure is believed to be an older section of the Carnegie Ridge (e.g. Hey, 1977; Lonsdale & Klitgord, 1978; Werner et al., 2003), whereas Meschede et al. (1998) proposed that it originated as part of the Cocos Ridge. Hoernle et al. (2002) reported 40Ar/39Ar ages from Malpelo Island ranging from 15·8 to 17·3 Ma.
On the NE Nazca Plate, the Coiba Ridge is a gently sloping, sediment-covered plateau 150 km long by 100 km wide (Fig. 1; Werner et al., 2003). Hoernle et al. (2002) proposed that it is a continuation of the Cocos Ridge that was offset along a paleo-transform fault. Coiba Ridge sediments include fossils that date back to 15 Ma (van Andel et al., 1973).
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
During R.V. Sonne cruise SO 144-3 (9 November–18 December 1999), we sampled >80 seamounts on the Cocos, Carnegie, Malpelo, and Coiba Ridges (Fig. 1). Most rocks were collected by dredge (DR samples), although a few were acquired with a remotely controlled grab sampler with real-time video (TVG samples). Samples were crushed, sieved (125 µm), and sonicated in de-ionized water to remove fine particles and surface contamination. The chips were examined under a binocular microscope to collect c. 10 g of the freshest, most pristine lava fragments. In many instances, the rocks were severely altered. Chips were powdered in agate and split for isotopic, major element, and trace element analysis. For dredges with more than one distinct lithology, several representative samples were processed.
Approximately 250 mg of rock powder were digested in a closed PFA Teflon container with 20 ml HNO3 and 5 ml HF sample for 40 h (e.g. Harpp, 1995). The solution was then evaporated to dryness. The solid residue was redissolved in 5 ml 50% HNO3 and diluted 1000-fold in 1% HNO3. All chemical procedures were conducted in HEPA-filtered Plexiglas clean boxes and all reagents were purified.
Concentrations of the trace elements were determined by ICP-MS using a Hewlett–Packard HP4500 system at Colgate University (Table 1). Measurements were made using an online internal standard correction consisting of a 1:20 dilution of a 1 ppm 115In, 133Cs, and 182W solution. Raw data were corrected to the closest internal standard mass (e.g. Doherty, 1989; Eggins et al., 1997). Contributions from polybaric oxide and doubly charged interferences were consistently below 1%. At least three replicate analyses of each solution were performed; precision and accuracy are represented by multiple analyses of the USGS standard W-2 run as an unknown (Table 2).
|
|
Samples selected for geochemistry were first crushed to small pieces, then washed in deionized water and carefully handpicked under a binocular microscope. Rock powders were dried in a furnace at 110°C for 12 h to drive off moisture. Water and CO2 were analyzed in an IR photometer (Rosemount CSA 5003). Major elements of whole-rock samples were determined on fused beads using a Phillips X'Unique PW1480 X-ray fluorescence spectrometer (XRF) equipped with a Rh-tube at IFM-GEOMAR. Accuracy of international reference standards JB-2, JB-3 and JA-2, measured with the samples, is better than 3% (SiO2, TiO2, Al2O3, MgO, CaO, K2O, 4% (Fe2O3), P2O5), and 10% (Na2O).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Lavas collected from the Cocos, Carnegie, Malpelo, and Coiba Ridges reveal a wide spectrum of lithologies, ranging from predominant basaltic pillow and sheet flow fragments to volcaniclastic and sedimentary rocks, as well as plutonic samples from the SE Cocos Ridge near the Panama fracture zone (Fig. 1). Major element, isotopic, and petrographic data have been reported by Werner et al. (2003)
. The complete major and trace element dataset is included as Electronic Appendix 1, which may be downloaded from the Journal of Petrology website at http://www.petrology.oupjournals.org/.
Severely altered samples
Many of the dredged samples are severely altered. The following criteria were applied to determine which samples should be removed from the dataset prior to interpretation.
(1) Ten lavas with P2O5 >1·5 wt % and/or MnO >0·3 wt % and with extensive visible alteration (i.e. zeolite-filled vesicles, major iron staining or olivine alteration, several millimeter-thick manganese oxide crusts or chlorite coatings) have been eliminated from consideration, predominantly from the Costa Rica Seamount Province (DR-5-1, 5-11, 39-2, 48-1, 58a-1, 64a-1, 73-1, 75-1, 77-1, 82-1).
(2) Sample DR-17-1 has extreme levels of Ba and Sr, and its Ba/Th and Sr/Zr are much higher than any lava from the islands or aseismic ridges (White et al., 1993; Harpp & White, 2001). We believe its composition is strongly altered by seawater; this sample is not considered in subsequent discussion either.
Trace element results by region
The spatial and geochemical complexity of our dataset requires that we consider the new trace element results in groups, on the basis of their location, common morphological characteristics, and apparent petrogenetic relationships (Figs 2–5; Table 3).
|
|
|
|
, Proposed mantle end-member compositions of Harpp & White (2001)
|
The Carnegie Ridge (dredges 8a, 11a-13, 17–19, 26, 28, and 29a)
Lavas dredged from the Carnegie Ridge are predominantly tholeiitic basalts, have trace element concentrations characteristic of both MORB and tholeiitic ocean island basalt (OIB) (Fig. 2; Table 3), and a two-fold variation in MgO content (5·11–10·01 wt %; Werner et al., 2003
). In general, the lavas follow coherent differentiation trends (e.g. MgO vs TiO2), but their relatively wide range in radiogenic isotope ratios (
Nd 5·97–9·77; Werner et al., 2003) precludes a relationship among the sampled lavas by fractional crystallization or partial melting of a single, homogeneous source.
In general, the more depleted lavas are located on the flanks of the ridge, whereas the enriched lavas are on the main crest (Figs 3–5). Thus, the most depleted lavas (dredges 8a, 28, 29a) may represent normal oceanic crust (Werner et al., 2003). In contrast, 26-TVG-1 displays a relatively flat rare earth element (REE) pattern, similar to those from samples collected to its south on the PLUME02 cruise (Harpp & White, 2001), and is consequently considered as part of the Carnegie Ridge despite its location on the northern flanks. In general, there is a broad trend of increasing enrichment in incompatible trace elements (ITE) and radiogenic isotope signatures eastward along the ridge.
The Cocos Ridge
Cocos Ridge axis (dredges 35, 37, 39, 43–45, 49–55). Lavas collected from the Cocos Ridge axis are all tholeiitic basalts with MgO ranging from 2·90 to 11·61 wt % (Werner et al., 2003). Limited radiogenic isotope analyses from six samples exhibit less variation than is observed on the Carnegie Ridge (Nd 6·01–7·74; Werner et al., 2003). Dredge 29a sampled the ocean floor at the 91°W transform fault on the GSC, revealing a tholeiitic basalt (MgO 8·52 wt %) that is slightly more depleted (Nd 7·56; Werner et al., 2003) than lavas erupted along the GSC axis both to the east and west of the fracture zone (Schilling et al., 1982; Detrick et al., 2002).
Cocos Ridge axial samples exhibit a similar range in trace element contents to those from the Carnegie Ridge, but extend to more enriched ITE ratios (Figs 3–5). Lavas with the steepest REE patterns also have the most pronounced negative Eu anomalies (e.g. dredges 51, 53; Fig. 3), corresponding to the greatest modal abundance of plagioclase phenocrysts (25–30%). The depleted lavas are not typical N-MORB, but broadly resemble those of the southern Carnegie Ridge flanks (Harpp & White, 2001) and Genovesa Island (Harpp et al., 2002), with concave-down REE patterns (Fig. 3). Similar to the Carnegie Ridge, Cocos Ridge lavas become increasingly enriched in ITE and radiogenic isotope ratios eastward along the axis.
Costa Rica Seamount Province (dredges 46–48, 56, 58–63, 73–87). Lavas from the Costa Rica Seamount Province are both tholeiitic and alkalic, with a large range of differentiation (Table 3; MgO 1·93–8·04 wt %). Radiogenic isotope signatures are slightly more enriched than those along the Cocos Ridge axis (Nd 3·90–6·63; Werner et al., 2003). All the Costa Rica Province lavas are highly enriched in ITE, displaying steeply dipping REE patterns (Fig. 3; Hoernle et al., 2000).
Seamounts in the center of the Costa Rica Seamount Province have elevated (La/Sm)n ratios compared with those on the outskirts, and lavas at the northern periphery are the most depleted of the region. The lavas with the shallower heavy REE slopes are found on the western and eastern edges of the province.
Cocos Island Province (dredges 38, 40–42, 64–72, Cocos Island). Cocos Province lavas are alkalic, with a wide range of MgO contents (0·48–8·83 wt %; Werner et al., 2003). The lavas define coherent trends in major element diagrams, consistent with clinopyroxene fractionation (i.e. positive Sc/Y and CaO/Al2O3 vs MgO slopes). Isotopic compositions (Nd 5·77–7·41; Werner et al., 2003) suggest that the Cocos Province lavas are derived from a relatively enriched mantle source; Cocos Island lavas have a limited range in isotopic ratios (Nd 6·34–6·83; Werner et al., 2003), as first noted by Castillo et al. (1988).
Cocos Island and the nearby seamounts are highly enriched in incompatible trace elements (Figs 3–5). Some lavas exhibit a distinctive concave-up REE pattern, with steep slopes in the light REE (LREE) but nearly horizontal slopes in the heavy REE (HREE), causing some of the patterns to cross (Fig. 3), reminiscent of Floreana Island and SW Galápagos seamount lavas (White et al., 1993; Harpp & White, 2001); the spoon-shaped patterns are observed most strongly in the submarine lavas.
Interestingly, lavas from dredge 38, located over 150 km east of Cocos Island, display the same enriched compositions as the Cocos Island Province. Sample 39-DR-2, however, collected from a seamount approximately midway between dredge 38 and Cocos Island (Fig. 1), has a nearly flat REE pattern (Fig. 3) and bears little resemblance to the Cocos Province lavas.
Southwest Seamount Province (dredges 30–33)
The Southwest Seamount Province consists of tholeiitic basalts with a large range in MgO content (4·62–10·17 wt % for four samples; Werner et al., 2003) but little variation in isotopic composition (Nd 5·95 and 6·28; Table 3; Werner et al., 2003). The Southwest Province lavas vary from LREE-enriched to LREE-depleted, but never achieve the enrichment of the Costa Rica or Cocos Island Province lavas (Fig. 3).
The Coiba Ridge (dredges 88, 90)
Both of the Coiba Ridge lavas are tholeiitic basalts (MgO 5·19–5·56 wt %; Table 3) enriched in ITE (Fig. 2), like the more depleted of the Costa Rica seamounts and the most enriched lavas from the main Cocos Ridge axis (Figs 3 and 4).
The Malpelo Ridge (dredges 1–7)
The Malpelo Ridge lavas (dredges 3–7) are all plagioclase-phyric, tholeiitic basalts, except for 6-DR-1, which is alkalic. The major element compositions (MgO 4·68–7·75 wt %) resemble those of the more primitive Costa Rica Seamount Province lavas, with moderately well-correlated major oxide variations and a narrow isotopic range (Nd 5·62–6·34; Werner et al., 2003). Malpelo Ridge lavas exhibit uniformly ITE-enriched compositions, similar to those observed on the Coiba Ridge and the Cocos Ridge axis, but with gentler slopes in the HREE and lower absolute concentrations on average.
Lavas dredged from an abandoned spreading center north of the Malpelo Ridge (dredges 1 and 2) are tholeiitic (MgO 5·70–8·72 wt %) and distinctly more depleted (Nd 10·05; Figs 3–5; Werner et al., 2003) than those from the main Malpelo Ridge. Their location and composition suggest that these lavas are probably normal oceanic crust rather than the products of hotspot activity.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Heterogeneous mantle sources
In a statistical analysis of regional Galápagos compositions, Harpp & White (2001)
proposed that the complex isotopic and trace element variations observed throughout the archipelago are the result of interaction between four mantle source reservoirs with distinct radiogenic isotope ratio signatures (Table 4): (1) a mantle plume with relatively enriched compositions (PLUME); (2) the depleted upper mantle (DUM); (3) an incompatible element-enriched, possibly metasomatically altered reservoir (FLO), which may or may not be part of the Galápagos plume (e.g. Kurz & Geist, 1999; Blichert-Toft & White, 2001); (4) a fourth component distinctive primarily in its lead isotope ratios (WD; elevated 207Pb/204Pb and 208Pb/204Pb for a given 206Pb/204Pb).
|
Radiogenic isotope ratios
The compositions of the four mantle end-members interacting in the Galápagos emerged from principal component analysis (PCA) of radiogenic isotope ratios in Galápagos lavas (Harpp & White, 2001
). Principal component analysis is a multivariate statistical method designed to reduce the complexity of a dataset by transforming the original, observed variables (in this case, isotopic ratios and later trace element ratios) into a new set of variables called principal components. The new variables are vectors that consist of different linear combinations of the original geochemical data. The first principal component vector accounts for as much of the variance in the dataset as possible; the second does the same for the remaining variability, and so forth. The extent to which the principal component represents the original dataset is indicated by the eigenvalue, or variance proportion. Effectively, PCA provides a method for reducing a multivariate system to a more manageable number of dimensions, without making any assumptions about the original data. The distribution of the variance across the principal component vectors indicates the number of variables necessary to describe the dataset efficiently. For example, when the majority of the variance is taken into account by the first principal component vector, the data can be described in one-dimensional space by two end-member compositions without any significant loss of information. Similarly, if two eigenvectors share the bulk of the variance, three end-member compositions adequately represent the original dataset's variability.
For the Galápagos archipelago dataset, we normalized all of the isotopic ratios to their mean prior to calculating the PCA results, because their values differ by orders of magnitude; in this way, all geochemical observations are afforded equal weight in the analysis. When only the isotopic ratios are considered for all subaerial and submarine data available in the present-day Galápagos archipelago, the first two eigenvectors (e1 + e2) account for 94·3% of the variance and the first three explain 97·3% (Table 5a; Harpp & White, 2001). Based on these results, Harpp & White (2001) concluded that three or four compositionally distinct mantle end-members must be interacting in the region to explain the observed isotopic variations.
|
When the isotopic ratios from the Cocos and Carnegie Ridges are included in the calculations with the Harpp & White (2001)
dataset (i.e. all available submarine and subaerial isotopic analyses from the Galápagos archipelago), the PCA result does not change significantly (Table 5b; e1 + e2 = 93·5%; e1 + e2 + e3 = 97·2%). The eigenvectors are not dominated by any single isotopic ratio, consistent with the results from the archipelago lavas alone. These statistical calculations confirm the conclusion of Hoernle et al. (2000) and Werner et al. (2003) that the lavas of the Cocos, Carnegie, Malpelo, and Coiba Ridges are derived from the same mantle source materials as <3 Ma Galápagos archipelago lavas. Thus, the Galápagos plume has probably maintained its distinctive compositional heterogeneity for the past 20 Myr.
Incompatible trace element ratios
To date, conclusions about the longevity of the chemical zonation in the Galápagos plume have been based exclusively on radiogenic isotope analyses. Because the geochemical variations are probably the result of mixing between different mantle reservoirs (White et al., 1993; Blichert-Toft & White, 2001; Harpp & White, 2001), the compositional heterogeneity should be reflected in elemental concentrations as well, particularly ratios of incompatible trace elements that can serve as tracers of mantle sources.
Furthermore, Harpp & White (2001) found that ITE ratios were consistent with their model involving variable contributions of four mantle sources for the present-day Galápagos plume (Table 4). The aseismic ridge samples fall within the limits defined by the four Galápagos archipelago mantle components (Harpp & White, 2001; Fig. 5). To test this conclusion statistically, we added aseismic ridge data for Ba/La, La/Ce, La/Nb, Th/U, Y/Gd, and Sm/Zr to the isotopic dataset tested in the previous PCA. The elements used in the ratios were chosen on the basis of similar distributions in mantle reservoirs (Sun & McDonough, 1989), as well as broadly similar partition coefficients for clinopyroxene (GERM website).
Inclusion of the ITE ratios from the aseismic ridges with the archipelago isotopic dataset has a profound effect on the PCA results (Table 5c). The first two eigenvectors account for only 77·8% of the variance, and the first three explain nearly 10% less of the variance than they do in the equivalent dataset for young Galápagos lavas (e1 + e2 + e3 = 87·9%).
Whereas radiogenic isotope ratios are unaffected by melting and crystallization processes, making them ideal tracers of mantle source compositions, even the carefully chosen ITE ratios used in the PCA can be affected by differences in melt generation conditions and, to a lesser extent, crustal contamination. Taken together, the statistical results suggest the following fundamental conclusions: first, the heterogeneous mantle source(s) responsible for the variation observed in the Galápagos archipelago is the same as that for the seamounts along the aseismic ridges, as predicted by Hoernle et al. (2000) and supported by Werner et al. (2003). Second, PCA results that indicate decoupling between the isotopic and ITE ratios imply that the conditions responsible for melt generation in the aseismic ridge lavas differ, at least in part, from those affecting the present-day Galápagos archipelago.
Both the Galápagos archipelago and the aseismic ridges exhibit similar ranges in isotopic and ITE ratios. The lavas with the most extreme isotopic enrichment in the archipelago (Floreana and the SW Galápagos Platform) also possess the most extreme ITE ratios, and are thought to be the result of low-degree melts generated in the spinel peridotite facies (Fig. 5; White et al., 1993; Harpp & White, 2001). In contrast, aseismic ridge lavas from this dataset showing the greatest degree of isotopic enrichment (Costa Rica Seamount Province, Carnegie, and Malpelo Ridges) are not those with the most extreme ITE ratios (Fig. 5; Cocos Island Province). This observation suggests that segments of the aseismic ridges have experienced different melting conditions from those affecting the Galápagos archipelago, although nevertheless sharing the same compositionally heterogeneous mantle source(s).
Spatial trends in melt generation across the aseismic ridges
Harpp & White (2001) used isotopic ratios (Sr, Nd, and Pb) to define relative contributions of the four mantle end-members to every Galápagos lava. The resulting source mixture was then melted using a polybaric melt model of trace element ratios (La/Sm and Sm/Yb; Fig. 6). We have applied the results of this model to the REE of the aseismic ridge lavas in regional groups to determine plausible ranges for end-member contributions and approximate melting conditions (Figs 5 and 6; Table 3).
|
The Carnegie and Malpelo Ridges (dredges 1–8, 11a—13, 17–19, 26, 28, 29a)
Lavas from the main Carnegie Ridge (dredges 11–13) exhibit dominantly PLUME-like compositions, with some contribution from DUM, and are consistent with >1–5% melting of their mantle source, at least partially in the garnet stability field (Harpp & White, 2001
). These lavas are nearly indistinguishable from those erupted more recently in the central Galápagos archipelago (White et al., 1993; Harpp & White, 2001). Lavas from the Malpelo Ridge exhibit similar characteristics (Fig. 6a).
Lavas from the ridges' flanks (dredges 1, 2, 8a, 26, 28, 29a) appear to be derived from >10% melting of a more depleted mantle source. The southern ridge flank lavas (dredges 17–19) resemble those dredged during the PLUME02 cruise on the southern periphery of the Galápagos platform (Christie et al., 1992; Harpp & White, 2001), and are derived from a more PLUME-enriched source than those from the north flank, as indicated by the isotopic data (Werner et al., 2003). The southern Carnegie Ridge lavas may incorporate a minor contribution from the FLO end-member, manifested in the isotopic ratio–ITE ratio plot as an array offset toward the enriched FLO composition (Fig. 5).
The Cocos Ridge axis and Coiba Ridge (dredges 35, 37, 39, 43–45, 49–55, 88, 90)
Compositions of Cocos axial lavas and those from the Coiba Ridge substantially overlap those of the Carnegie Ridge (Fig. 6b). The melt model indicates that lavas from the axis of the Cocos Ridge are primarily the result of 2–10% melting of a PLUME source, with the exception of the samples that appear to be normal MORB (e.g. dredges 37, 44). Most of the Cocos axis samples resemble lavas erupted in the western Galápagos archipelago, in that their sources are PLUME-dominated but diluted by depleted mantle (Fig. 5). Coiba Ridge lavas are produced from lower degrees of melting (Fig. 6b) of a more pristine PLUME source (1–2%).
Costa Rica Seamount Province (dredges 46–48, 56, 58–63, 73–87)
Most of the lavas from the Costa Rica Seamount Province (Fig. 6b) appear to be derived from a source that is predominantly PLUME, with small, variable contributions from FLO and some WD. The lavas with the most prominent WD signatures are located at the western edge of the province (dredges 74 and 78; determined from Pb isotopic ratios, not shown here). Most of the Costa Rica seamounts are compositionally similar to lavas from the SW corner of the Galápagos archipelago, near Cerro Azul on Isabela Island (White et al., 1993; Harpp & White, 2001). An important feature distinguishes the bulk of the Costa Rica seamounts from the SW Galápagos Platform lavas; the Costa Rica seamount lavas have elevated (Sm/Yb)n ratios in addition to high (La/Sm)n. These REE systematics suggest that the Costa Rica Province is the product of low degrees of melting (<2%) of a PLUME + FLO source, mostly in the garnet stability field. No lavas dominantly derived from PLUME in the present Galápagos exhibit comparable REE systematics. Lavas from the seamounts at the outer edges of the province are products of greater extents of melting (5–10%; dredges 56, 73–75, 77a; Fig. 6b).
Cocos Island Province (dredges 38, 40–42, 64–72, Cocos Island)
Cocos Island and the Cocos Province seamounts exhibit many similar characteristics to the Costa Rica Seamount Province, except that Cocos Island and some of the seamounts extend to higher (La/Sm)n and (Sm/Yb)n (Table 3; Fig. 6b). They are also distinct from lavas observed in the present-day Galápagos because of their extreme ITE ratios coupled with PLUME-like isotopic signatures. These characteristics suggest that the lavas of Cocos Island and the surrounding seamounts are derived from a predominantly PLUME-like source that has experienced low extents of melting (<1%) in the garnet stability field and considerable subsequent fractional crystallization.
Southwest Seamount Province (dredges 30–33)
Although data are sparse in this region, the SW seamounts appear to be derived from a PLUME + DUM mantle source, similar to that for most of the Cocos Ridge lavas (Fig. 6b). The SW seamounts are the products of moderate (2–>5%) extents of melting, probably in the garnet facies. If WD contributes to this set of lavas, it is only to a minor extent, unlike the northern Galápagos lavas across the GSC.
Tectonic implications for melt generation and plume–ridge interaction
Tectonic history of the Galápagos plume–GSC system
In their isotopic study of the aseismic ridge lavas, Werner et al. (2003) proposed that the Galápagos plume has maintained its compositional zonation for the past 20 Myr. Although they confirmed the striped zonation along the NE Cocos Plate hotspot track, the southwestern part of the track does not exhibit the same pattern, a result of the young age of the samples collected on the SW Cocos Ridge (Werner et al., 2003). They explained deviations from the predicted pattern with a tectonic model for the region that takes into account plate reconstructions (e.g. Wilson & Hey, 1995; Meschede & Barckhausen, 2000; Barckhausen et al., 2001), the changing relationship between the migrating GSC and the Galápagos hotspot, including the relatively recent formation of the 91°W transform fault and a pre-existing large-offset transform (Wilson & Hey, 1995; Meschede & Barckhausen 2000), and variations in the morphology of the volcanic structures along the hotspot track.
Briefly, Werner et al. (2003) proposed the following sequence of events: (1) the GSC initiated a southward jump to the northern edge of the Galápagos hotspot at 19·5 Ma; (2) from 19·5 to 14·5 Ma, the GSC was centered above the hotspot, erupting plume material onto both the Cocos and Nazca Plates; (3) at 14·5 Ma, a second ridge jump relocated the GSC south of the hotspot, such that the majority of the plume products were erupted onto the Cocos Plate; (4) between 11 and 12 Ma, the plume was centered beneath an offset along the GSC, when (5) at 4 Ma, the GSC relocated north of the hotspot, leaving the plume isolated beneath the Nazca Plate, as is the case today. Currently, the northeastward migration of the GSC continues, increasing the distance between the spreading center and the hotspot (e.g. Wilson & Hey, 1995).
Whereas the Werner et al. (2003) model is consistent with the distribution of isotopic signatures observed along the aseismic ridges and with changes in ridge volume over time, it does not adequately account for the variations in partial melting as determined from trace element analyses. In particular, two fundamental observations cannot be reconciled with the model based on isotopic ratios alone. First, according to Werner et al. (2003), the Costa Rica Seamounts are considered genetically similar to lavas from the northern Galápagos archipelago [the ‘Northern Domain’ of Hoernle et al. (2000)], but their alkalic, highly ITE-enriched compositions bear little resemblance to lavas erupted along the Wolf–Darwin Lineament (Fig. 1).
The second unexplained observation is the formation of the lavas erupted on and around Cocos Island. According to Werner et al. (2003), the Cocos Island Province is associated with compositions observed in the middle of the Galápagos archipelago today [the ‘Central Domain’ of Hoernle et al. (2000)], because of their isotopic resemblance to lavas erupted at Fernandina Island. Yet both Cocos Island and the surrounding seamount lavas have ITE concentrations far more enriched than any observed on the Galápagos Platform and exhibit almost exclusively alkalic compositions. Moreover, Bellon et al. (1984) established that Cocos Island was produced within the past 2 Myr, hence it must have been erupted over 500 km from the presumed plume center in the western Galápagos archipelago. Castillo et al. (1988) concurred that Cocos Island was constructed far from the Galápagos plume center, possibly as the result of reactivation of small-degree Galápagos plume melts. Meschede & Barckhausen (2000) took these observations a step further and attributed the Cocos Island lavas to a second hotspot isolated beneath the Cocos Plate.
Failed rifts and alkalic seamount provinces
Both the Costa Rica and Cocos Island Provinces are located on or near extinct spreading centers (Meschede & Barckhausen, 2000). The northeastern end of the Cocos Ridge is the site of an extinct spreading center abandoned when the GSC began a southward jump at 19·5 Ma (Meschede & Barckhausen, 2000). Consequently, the Costa Rica Seamounts are clustered around the NW end of the abandoned rift.
Werner et al. (1999) dated lavas dredged along the eastern half of the Costa Rica Seamount Province, parallel to the Costa Rican coast, revealing ages ranging from 12·97 ± 0·21 Ma to 14·46 ± 0·32 (1
) Ma. According to paleomagnetic data, the lithosphere underlying this seamount province is older than 15 Ma (Meschede & Barckhausen, 2000). The seamounts, therefore, were not produced at the active ridge axis, but later, after the GSC had jumped south. All the Costa Rica seamounts are clustered within 200 km of the abandoned spreading center (Fig. 1).
Likewise, the Cocos Island Province is also located adjacent to a failed spreading center. Unfortunately, the ages of the seamounts around Cocos Island are not known, but K–Ar ages of Dalrymple & Cox (1968) and Bellon et al. (1984) indicate that the subaerial parts of the island are <2 Ma, at least 5 Myr younger than the underlying lithosphere [based on plate motion vectors of Wilson (1996) and Barckhausen et al. (2001)]. It is not unreasonable to posit that the seamounts clustered around Cocos Island were formed during broadly the same timeframe, because the compositions of the seamount lavas closely resemble those of Cocos Island, in their extreme ITE enrichment, alkalic compositions, and nearly identical isotopic and ITE ratios (Castillo et al., 1988; Werner et al., 2003).
According to Meschede & Barckhausen (2000), the Cocos Island region is the site of a failed rift. An east–west-trending spreading center became active at 3·5 Ma, but went extinct by 2 Ma, leaving an abandoned rift extending westward c. 200 km from the region near Cocos Island. Thus, if the Cocos Island Province formed nearly contemporaneously with Cocos Island, then the seamounts were erupted shortly after the failure of the rift, and they all formed within 200 km of the abandoned segment.
Abundant post-abandonment alkalic volcanism, lasting for several million years after a ridge jump, appears to be a common feature of failed rifts. Batiza & Vanko (1985) proposed that as rifting activity slows, magma supply to the ridge becomes sporadic and a steady-state magma chamber can no longer exist, resulting in abundant strongly evolved lavas, just as is observed in the Cocos Province. After the cessation of spreading, alkali basalts are produced by low degrees of melting, which are a manifestation of the progressively thickening lithosphere as the region cools.
The Mathematician Ridge is a 800 km long failed rift located north of the Cocos Ridge in the eastern Pacific (e.g. Mammerickx et al., 1988). At c. 3·5 Ma, activity along the ridge shifted eastward nearly 600 km to what is now the East Pacific Rise at 10°–20°N (e.g. Mammerickx & Klitgord, 1982). The region around the Mathematician Ridge is notable for abundant alkalic seamounts and islands (Socorro, San Benedicto, Clarion, and Roca Partida) that form the Revillagigedo Archipelago (e.g. Bohrson & Reid, 1995). Fieldwork and dredging operations have revealed that the subaerial and submarine volcanoes throughout the area remain active today (Farmer et al., 1993; McClelland et al., 1993; Bohrson & Reid, 1995; Bohrson et al., 1996), producing alkalic lavas that appear to be derived from similar mantle sources (Batiza & Vanko, 1985; Bohrson & Reid, 1995, 1997). The active volcanic region is broad, extending across the entire failed rift and transform system and up to 350 km SW of Clarion Island (Batiza & Vanko, 1985).
The lavas of both the Costa Rica and Cocos Island Provinces bear a striking resemblance to those erupted from the Mathematician Ridge and the related islands (Batiza & Vanko, 1985; Bohrson & Reid, 1995, 1997, 1998; Bohrson et al., 1996): (1) with the exception of a few tholeiitic lavas in the Costa Rica Province, all lavas erupted in the failed rift regions are alkalic; (2) subaerial lavas exhibit greater degrees of differentiation than the seamounts, extending to trachytic compositions on Cocos Island (e.g. Castillo et al., 1988; Socorro Island lavas include abundant peralkaline trachytes and rhyolites; Bohrson & Reid, 1995, 1997); (3) magmatism persisted at the same place long enough for the volcanic structures to emerge as islands; (4) both the subaerial and the seamount lavas in a given province are derived from either the same or similar mantle sources, with limited ranges in radiogenic isotope and ITE ratios (Bohrson & Reid, 1995, 1997); (5) high field strength element (HFSE) concentrations increase systematically with decreasing MgO content, whereas ratios of HFSE remain relatively constant for the submarine samples (Cocos Island lavas exhibit significant variation in HFSE ratios with MgO; Fig. 7).
|
Some of the trace element compositions of Cocos and Costa Rica seamount lavas share features that Bohrson & Reid (1995
, 1997) attributed to assimilation of oceanic crust: (1) incompatible elements including Ba and Y do not correlate well with incompatible HFSE, instead exhibiting two distinct trends (Fig. 7b), an indication of assimilation taking place in addition to straightforward fractional crystallization (Bohrson & Reid, 1995); (2) like Socorro Island lavas, the Cocos and Costa Rica Seamount Province lavas exhibit highly variable Ba, P2O5, and Y contents for a given MgO content, as well as a spectrum of negative Ce anomalies, a phenomenon Bohrson & Reid (1995) attributed to variable assimilation of ocean crust.
Given the extensive petrologic and geochemical similarities to lavas erupted in the Mathematician Ridge region, we propose that the Costa Rica and Cocos Island Provinces are also the manifestation of failed rifts. This conclusion is further strengthened by the proximity of the seamount provinces to abandoned spreading center scars (Meschede & Barckhausen, 2000), as well as the timing of their eruption relative to that of the major spreading center jumps (beginning at 19·5 Ma: Costa Rica Province; 2–3 Ma: Cocos Island Province; Bellon et al., 1984; Werner et al., 1999; Meschede & Barckhausen, 2000).
Melt generation and failed rift systems
We propose that the predominantly alkalic Cocos and Costa Rica Seamount Provinces are the result of post-abandonment volcanism, following jumps and the failure of a spreading ridge in each of the regions. The mechanism for widely dispersed alkalic volcanism after a spreading ridge jump has occurred can be explained as the result of the decrease in the upwelling rate. As spreading ridge activity wanes, decreased upwelling will result in progressively lower extents of melting. Furthermore, far-field, deviatoric stresses generated by major tectonic rearrangements may also cause extensive cracking of the lithosphere in the vicinity of the failing rift, as has been observed in Iceland (e.g. Fujita & Sleep, 1978; Clifton et al., 2000). The stress field could result in the initiation of mantle upwelling, causing widespread, localized volcanic activity over a period lasting up to several million years from initial spreading rift failure (Batiza & Vanko, 1985).
In this model, each volcanic center taps only the immediately underlying mantle, consistent with geochemical observations that the seamounts are derived from variable amounts of partial melting of similar mantle sources. The elongate structures and linear arrangements of many of the seamounts in the Costa Rica and Cocos Provinces further support the idea that they may result from volcanism caused by rifting and upwelling; such asymmetric structural features are relatively rare in off-axis seamount provinces (Batiza & Vanko, 1983; Smith & Cann, 1992) and are indicative of unusually strong deviatoric stresses on a regional scale. It should also be noted that both the Costa Rica and Cocos Island Provinces are located on the northern flanks of the main Cocos Ridge, not on the crest itself, where the bulk of the hotspot-produced lavas have presumably accumulated. The off-axis location of the seamounts is therefore more consistent with the failed rift origin than as direct products of the Galápagos plume. Nevertheless, as described by Werner et al. (2003), we believe that the Galápagos plume is responsible for the majority of the aseismic ridges' volume.
Broadly similar, tensional volcanism has also been invoked to explain the distribution and composition of islands and seamounts in the northern Galápagos archipelago (Harpp & Geist, 2002; Harpp et al., 2002). An important difference between the northern Galápagos and the aseismic ridge seamount provinces, however, is that only tholeiitic lavas have been produced in the northern Galápagos and none of the volcanoes' compositions achieve the ITE enrichment of the Cocos Ridge seamounts (Harpp & Geist, 2002). We attribute this distinction to the fundamental difference between the two tectonic settings. In the northern Galápagos, the active region is bounded by the Galápagos plume to the south and an active spreading center to the north. Consequently, the ambient temperature of the mantle throughout the northern Galápagos should be higher than average, resulting in abundant melt supply to the volcanic centers. Lavas in the northern Galápagos exhibit (Sm/Yb)n >1, an indication that the melts were partially generated in the garnet stability field.
Isotopically, however, the depleted mantle plays a significant role in all the northern Galápagos compositions (Harpp & White, 2001). The elevated sub-lithospheric temperatures throughout this region, therefore, not only cause partial melting of fertile mantle heterogeneities, such as pockets of PLUME source dispersed in the mantle (Harpp & Geist, 2002), but they also initiate melting of the depleted upper mantle, which dilutes the magmas to the observed intermediate enrichment levels. In the seamount provinces on the Cocos Ridge, ambient mantle temperatures are lower, minimizing the contribution from the more refractory depleted mantle source and allowing the small-degree, alkalic melts of enriched material to dominate erupted compositions.
Extensive studies of the present-day GSC have established that lavas erupted along the mid-ocean ridge axis within a few hundred kilometers of the hotspot are contaminated by the Galápagos plume (e.g. Schilling et al., 1982, 2003; Detrick et al., 2002). Lithosphere and sublithospheric mantle produced at the GSC in the past, when the plume–GSC separation was equal to or less than it is today, should similarly be infused with plume material. Therefore, the volcanic centers of the Cocos and Costa Rica Seamount Provinces tap mantle that is variably contaminated with plume material. Because the enriched plume source is more fertile than the surrounding depleted mantle, it should be the first to melt, and in a region of only limited heat supply such as a failed rift, may be the only material that is ultimately erupted.
Revised tectonic model
We propose the following sequence of tectonic and volcanic events that summarize the 20 Myr history of interaction between the Galápagos plume and Galápagos Spreading Center (Fig. 8), amended from Werner et al. (2003).
|
(1)
19·5–14·5 Ma. At c. 19·5 Ma, the Galápagos Spreading Center began a 20° strike shift in a major tectonic rearrangement (Meschede et al., 1998; Barckhausen et al., 2001). The alkalic volcanoes of the Costa Rica Seamount Province then erupted over an interval of several million years as the abandoned spreading center failed (e.g. Batiza & Vanko, 1985). The waning upwelling rate resulted in small amounts of deep melting of a predominantly PLUME source, yielding alkalic lavas highly enriched in incompatible trace elements. During this period, the bulk of the hotspot products were being deposited onto the Nazca Plate, forming the combined Carnegie and Malpelo Ridges from moderate degrees of partial melting in the garnet stability field of a predominantly PLUME–DUM mantle source, much like lavas erupted today in the central Galápagos archipelago.
(2) 14·5–12 Ma. The GSC again jumped south, with a minor change in strike (Meschede et al., 1998; Barckhausen et al., 2001), causing the ridge to overlap only the southern edge of the hotspot. Consequently, the volume of eruptive material deposited on the Nazca Plate decreased, manifested as a narrowing of the Carnegie Ridge between 85°W and 87°W (Fig. 8). The southward GSC jump also caused the Malpelo Ridge to initiate rifting away from the Carnegie Ridge. Beginning at this time, the bulk of the hotspot products were erupted onto the Cocos Plate, the results of slightly higher extents of melting in the garnet stability field (2–10%), with a greater contribution from PLUME than in the past, much like lavas erupted at the western Galápagos shield volcanoes today. Because the plume was located predominantly beneath the Cocos Plate during this interval, the Cocos Ridge axis reflects the geochemical zonation of the Galápagos plume (Hoernle et al., 2000; Werner et al., 2003).
(3) 12–11 Ma. Werner et al. (2003) proposed that the Galápagos plume was centered beneath an offset in the GSC at this time, because only lavas with PLUME-like compositions are observed along the Cocos Ridge. This minor change in the GSC configuration is reflected in a shift in the strike of the magnetic sea-floor anomalies for this period (Barckhausen et al., 2001) and probably marks the initiation of northward migration of the GSC that continues today (Wilson & Hey, 1995).
(4) 9·5 Ma. Spreading between the Malpelo and Carnegie Ridges ceased around this time.
(5) 5·2–3·5 Ma. The GSC moved north of the plume center during this period (Wilson & Hey, 1995), shifting the eruption of hotspot material onto the Nazca Plate in its entirety, similar to the current hotspot–ridge configuration.
(6) 3·5–2 Ma. A short-lived, east–west-trending spreading center became active north of the GSC (Meschede & Barckhausen, 2000). Shortly thereafter, this rift failed, resulting in the post-abandonment alkalic volcanism responsible for the 2 Ma formation of Cocos Island (Castillo et al., 1988) and the surrounding Cocos Island Province seamounts. Once again, only deep, small-degree melts of a predominantly PLUME mantle source were generated as the rift's magma supply was cut off.
(7) 2·6 Ma. A major transform fault formed immediately north of the hotspot, probably the result of a southward jump by the GSC segment closest to the plume (e.g. Small, 1995; Wilson & Hey, 1995). The regional stresses initiated widespread, dispersed volcanism throughout the northern Galápagos, resulting in the formation of the tholeiitic islands and seamounts of the Wolf–Darwin Lineament and northern Galápagos (Harpp & Geist, 2002; Harpp et al., 2002). The elevated mantle temperatures caused by the adjacent ridge and plume systems initiated melting of the depleted mantle and lowered average depths of melting.
(8) Present. The geochemical zonation of the Galápagos plume is detectable along the western periphery of the platform (White et al., 1993; Kurz & Geist, 1999; Harpp & White, 2001). According to Hoernle et al. (2000), the central part of the archipelago is covered by lavas with a more depleted signature as the Nazca Plate moves eastward, leaving the altered, southern plume component detectable on the southern flank of the Carnegie Ridge (Werner et al., 2003).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Trace element analyses of dredged samples collected along the Cocos, Carnegie, Coiba, and Malpelo Ridges during the PAGANINI expedition provide important insights into the history of the Galápagos hotspot and plume–ridge interaction over the past
20 Myr. The compositional and spatial trends in the aseismic ridge data suggest that the Galápagos plume has been compositionally zoned for the past 18–20 Myr (Werner et al., 2003), exhibiting the same spatial zonation as the present-day Galápagos archipelago (e.g. Geist et al., 1988; White et al., 1993; Harpp & White, 2001). The dynamic relationship between the plume and the ridge, however, adds an additional set of complexities to the system, resulting in deviations from the predicted distribution of compositions along the Cocos and Carnegie Ridges caused by a zoned plume (Hoernle et al., 2000).
On the basis of regional variations in melting conditions along the ridges, as determined from the trace element contents of the dredged lavas, the relationship between the Galápagos plume and the Galápagos Spreading Center can be elucidated. Trace element content variations are generally consistent with the tectonic history of the plume–GSC system as proposed by Werner et al. (2003; Fig. 8), with some modifications to take into account volcanism following the failure of spreading ridge systems.
The Costa Rica Province seamounts probably formed within a few million years after the major ridge jump at 19·5 Ma (e.g. Meschede & Barckhausen, 2000). They are the result of waning mantle upwelling in response to regional deviatoric stresses after the jump. Similarly, Cocos Island and the surrounding seamounts may have formed in the wake of a ridge failure at 2 Ma (e.g. Bellon et al., 1984) instead of a second hotspot center (e.g. Meschede & Barckhausen, 2000). Tests of this hypothesis would include geochronological analysis of the lavas from the Costa Rica and Cocos Island Provinces, which should each cluster within a few million years of the ridge jumps at 19·5 and 2 Ma.
Aseismic ridges and their associated seamounts, therefore, are not simply the surficial manifestation of plume activity, but also reflect subsequent regional tectonic events, including spreading center jumps and migration. They may serve as valuable sources of information about the history of a mantle plume. In the presence of a proximal spreading center, aseismic ridges preserve a record of the interaction between the spreading center and the hotspot that permits reconstruction of the regional tectonic history.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We would like to thank the scientific team and shipboard crew of the R.V. Sonne during the SO 144-3 PAGANINI expedition for their assistance in carrying out this study. We also thank L. Mayhew for sample preparation efforts. In addition, Britta Lissinna provided the Colgate students with invaluable assistance and hospitality while they were processing samples for analysis at GEOMAR. We extend our gratitude to the governments of Costa Rica and Ecuador for granting access to their territorial waters and to Cocos Island. Thanks go to Wendy Bohrson, Dennis Geist, and Pat Castillo for their thoughtful and constructive reviews. This study was supported by NSF grants CHE-9996141 and CHE-9996136 to K.S.H.; cruise funding was provided by the German Ministry of Education and Research (BMBF; Grant PAGANINI).
|
FOOTNOTES |
|---|
Present address: Department of Geology and Geophysics, SOEST, University of Hawaii, Honolulu, HI 96822, USA. 
Present address: University of Miami, Rosenstiel School of Marine and Atmospheric Science, Miami, FL 33149, USA.
* Corresponding author. Telephone: (315)228-7211. Fax: (315)228-7187. E-mail: kharpp{at}mail.colgate.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUND INFORMATIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Barckhausen, U., Ranero, C. R., von Huene, R., Cande, S. C. & Roeser, H. A. (2001). Revised tectonic boundaries in the Cocos plate off Costa Rica: implications for the segmentation of the convergent margin and for plate tectonic models. Journal of Geophysical Research 106, 19207–19220.[CrossRef]
Batiza, R. & Vanko, D. A. (1985). Petrologic evolution of large failed rifts in the eastern Pacific: petrology of volcanic and plutonic rocks from the Mathematician ridge area and the Guadalupe trough. Journal of Petrology 26, 564–602.
Bellon, H., Saenz, R. & Tournon, J. (1984). K–Ar radiometric ages of lavas from Cocos Island (eastern Pacific). Marine Geology 54, M17–M23.[CrossRef][Web of Science]
Blichert-Toft, J. & White, W. M. (2001). Hf isotope geochemistry of the Galápagos Islands. Geochemistry, Geophysics, Geosystems 2, paper number 2000GC00138, posted at http://www.g-cubed.org.
Bohrson, W. A. & Reid, M. R. (1995). Petrogenesis of alkaline basalts from Socorro Island, Mexico: trace element evidence for contamination of ocean island basalt in the shallow ocean crust. Journal of Geophysical Research 100, 24555–24576.[CrossRef]
Bohrson, W. A. & Reid, M. R. (1997). Genesis of silicic peralkaline volcanic rocks in an ocean island setting by crustal melting and open-system processes: Socorro Island, Mexico. Journal of Petrology 38, 1137–1166.[CrossRef][Web of Science]
Bohrson, W. A. & Reid, M. R. (1998). Genesis of evolved ocean island magmas by deep- and shallow-level basement recycling, Socorro Island, Mexico: constraints from Th and other isotope signatures. Journal of Petrology 39, 995–1008.[CrossRef][Web of Science]
Bohrson, W. A., Reid, M. R., Grunder, A. L., Heizler, M. T., Harrison, T. M. & Lee, J. (1996). Prolonged history of silicic peralkaline volcanism in the eastern Pacific Ocean. Journal of Geophysical Research 101, 11457–11474.[CrossRef]
Castillo, P., Batiza, R., Vanko, D., Malavassi, E., Barquero, J. & Fernandez, E. (1988). Anomalously young volcanoes on old hot-spot traces: I. Geology and petrology of Cocos Island. Geological Society of America Bulletin 100, 1400–1414.
Christie, D. M., Duncan, R. A., McBirney, A. R., Richards, M. A., White, W. M., Harpp, K. S. & Fox, C. G. (1992). Drowned islands downstream from the Galápagos hotspot imply extended speciation times. Nature 355, 246–248.[CrossRef]
Clifton, A. E., Schlische, R. W., Withjack, M. O. & Ackermann, R. V. (2000). Influence of rift obliquity on fault-population systematics: results of experimental clay models. Journal of Structural Geology 22, 1491–1509.[CrossRef][Web of Science]
Dalrymple, G. B. & Cox, A. (1968). Paleomagnetism, potassium–argon ages and petrology of some volcanic rocks. Nature 217, 323–326.[CrossRef]
Detrick, R. S., Sinton, J. M., Ito, G., Canales, J. P., Behn, M., Blacic, T., Cushman, B., Dixon, J. E., Graham, D. W. & Mahoney, J. J. (2002). Correlated geophysical, geochemical, and volcanological manifestations of plume–ridge interaction along the Galápagos Spreading Center. Geochemistry, Geophysics, Geosystems 3, 8501, doi: 10.1029/2002GC000350, posted at http://www.g-cubed.org.[CrossRef]
Doherty, W. (1989). Internal standardization procedure for the determination of yttrium and the rare earth elements in geological materials by inductively coupled plasma-mass spectrometry. Geochimica et Cosmochimica Acta 44B, 263–280.
Eggins, S. M., Woodhead, J. D., Kinsley, L. P. J., Mortimer, G. E., Sylvester, P., McCulloch, M. T., Hergt, J. M. & Handler, M. R. (1997). A simple method for the precise determination of 40 trace elements in geological samples by ICP-MS using enriched isotope internal standardization. Chemical Geology 134, 311–326.[CrossRef][Web of Science]
Farmer, J. D., Farmer, M. C. & Berger, R. (1993). Radiocarbon ages of lacustrine deposits in volcanic sequences of the Lomas Coloradas area, Socorro Island, Mexico, Radiocarbon 35, 253–262.[Web of Science][Medline]
Fujita, K. & Sleep, N. (1978). Membrane stresses near mid-ocean ridge–transform intersections. Tectonophysics 50, 207–221.[CrossRef][Web of Science]
Gallahan, W. E. & Nielsen, R. L. (1992). The partitioning of Sc, Y, and the rare earth elements between high-Ca pyroxene and natural mafic to intermediate lavas at 1 atmosphere. Geochimica et Cosmochimica Acta 56, 2387–2404.[CrossRef][Web of Science]
Geist, D. J., White, W. M. & McBirney, A. R. (1988). Plume asthenosphere mixing beneath the Galápagos Archipelago. Nature 333, 657–660.[CrossRef]
Geldmacher, J., Hanan, B. B., Blichert-Toft, J., Harpp, K., Hoernle, K., Hauff, F., Werner, R. & Kerr, A. C. (2003). Hafnium isotopic variations in volcanic rocks from the Caribbean Large Igneous Province and Galápagos hot spot tracks. Geochemistry, Geophysics, Geosystems 4(7), 1062, doi: 10.1029/200296000477.[CrossRef]
GERM Website, http://earthref.org/GERM/, last updated 29 May 2003; accessed 6 July 2003.
Gripp, A. E. & Gordon, R. G. (1990). Current plate velocities relative to hotspots incorporating the NUVEL-1 global plate model. Geophysical Research Letters 11, 1109–1112.
Harpp, K. S. (1995). Magmatic evolution of mid-ocean ridges and hotspots: isotopic and trace element studies of the East Pacific Rise, Mid-Atlantic Ridge, and Galápagos Islands. Ph.D. thesis, Cornell University, Ithaca, NY.
Harpp, K. S. & Geist, D. J. (2002). Wolf–Darwin lineament and plume–ridge interaction in northern Galápagos. Geochemistry, Geophysics, Geosystems 3, 8504, doi: 10.1029/2002GC000370, posted at http://www.g-cubed.org.[CrossRef]
Harpp, K. S. & White, W. M. (2001). Tracing a mantle plume: isotopic and trace element variations of Galápagos seamounts. Geochemistry, Geophysics, Geosystems 2, paper number 2000GC00137, posted at http://www.g-cubed.org.
Harpp, K. S., Wirth, K. R. & Korich, D. J. (2002). Northern Galápagos Province: hotspot-induced, near-ridge volcanism at Genovesa Island. Geology 30, 399–402.
Hey, R. (1977). Tectonic evolution of the Cocos–Nazca spreading center. Geological Society of America Bulletin 88, 1404–1420.
Hoernle, K., Werner, R., Phipps Morgan, J., Garbe-Schonberg, D., Bryce, J. & Mrazek, J. (2000). Existence of complex spatial zonation in the Galápagos plume for at least 14 m.y. Geology 28, 435–438.
Hoernle, K., van den Bogaard, P., Werner, R., Hauff, F., Lissinna, B., Alvarado, G. E. & Garbe-Schönberg, D. (2002). Missing history (16–71 Ma) of the Galápagos hotspot: implications for the tectonic and biological evolution of the Americas. Geology 30, 795–798.
Holden, J. C. & Dietz, R. S. (1972). Galápagos Gore, NazCoPac Triple Junction and Carnegie/Cocos Ridges. Nature 100, 266–269.
Kellogg, J. N. & Vega, V. (1995). Tectonic development of Panama, Costa Rica, and the Colombian Andes: constraints from global positioning system geodetic studies and gravity. In: Mann, P. (ed.) Geologic and Tectonic Development of the Caribbean Plate Boundary in Southern Central America. Boulder, CO: Geological Society of America, pp. 75–90.
Kurz, M. D. & Geist, D. (1999). Dynamics of the Galápagos hotspot from helium isotope geochemistry. Geochimica et Cosmochimica Acta 63, 4139–4156.[CrossRef][Web of Science]
Langmuir, C. H., Voche, R. D., Jr, Hanson, G. N. & Hart, S. R. (1978). A general mixing equation with applications to Icelandic basalts. Earth and Planetary Science Letters 37, 380–392.[CrossRef][Web of Science]
Langmuir, C. H., Klein, E. M. & Plank, T. (1992). Petrological systematics of mid-ocean ridge basalts: constraints on melt generation beneath ocean ridges. In: Morgan, J. P., Blackman, D. K. & Sinton, J. M. (eds) Mantle Flow and Melt Generation at Mid-Ocean Ridges. Geophysical Monograph, American Geophysical Union, 71, 183–280.
Lonsdale, P. & Klitgord, K. D. (1978). Structure and tectonic history of the eastern Panamá Basin. Geological Society of America Bulletin 89, 981–999.
Mammerickx, J. & Klitgord, K. D. (1982). Northern East Pacific Rise; magnetic anomaly and bathymetric framework. Journal of Geophysical Research B, 87, 6725–6750.[CrossRef]
Mammerickx, J., Naar, D. F. & Tyce, R. L. (1988). The Mathematician paleoplate. Journal of Geophysical Research 93, 3025–3040.
Meschede, M. & Barckhausen, U. (2000). Plate tectonic evolution of the Cocos–Nazca spreading center. In: Silver, E. A., Kimura, G. & Shipley, T. H. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 170. College Station, TX: Ocean Drilling Program [online]. Posted at http://www-odp.tamu.edu/publications/170_SR/chap_07/chap_07.htm. Accessed 7 July 2003.
Meschede, M., Barckhausen, U. & Worm, H.-U. (1998). Extinct spreading on the Cocos Ridge. Terra Nova 10, 211–216.[CrossRef][Web of Science]
Morgan, W. J. (1971). Convection plumes in the lower mantle. Nature 230, 42–43.[CrossRef]
Schilling, J.-G., Kingsley, R. H. & Devine, J. D. (1982). Galápagos hot spot–spreading center system: 1. Spatial petrological and geochemical variations (83°W–101°W). Journal of Geophysical Research 87, 5593–5610.
Schilling, J.-G., Fontignie, D., Blichert-Toft, J., Kingsley, R. & Tomza, U. (2003). Pb–Hf–Nd–Sr isotope variations along the Galápagos Spreading Center (101°–83°W): constraints on the dispersal of the Galápagos mantle plume. Geochemistry, Geophysics, Geosystems 4, 512, doi: 10.1029/2002GC000495, posted at http://www.g-cubed.org.
Sinton, C., Christie, D. M. & Duncan, R. A. (1996). Geochronology of Galápagos seamounts. Journal of Geophysical Research 101, 13689–13700.[CrossRef]
Small, C. (1995). Observations of ridge–hotspot interactions in the Southern Ocean. Journal of Geophysical Research B 100, 17931–17946.[CrossRef]
Smith, D. K. & Cann, J. R. (1992). The role of seamount volcanism in crustal construction at the Mid-Atlantic Ridge (24°–30°N). Journal of Geophysical Research B 97, 1645–1658.[CrossRef]
Smith, W. H. F. & Sandwell, D. T. (1997). Global seafloor topography from satellite altimetry and ship depth soundings. Science 277, 1956–1962.
Sun, S. S. & McDonough, W. F. (1989). Chemical and isotopic systematics of oceanic basalts; implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications 42, 313–345.
Van Andel, T.H., Heath, G. R. Malfait, B. T., Heinrichs, D. F. & Ewing, J. I. (1971). Tectonics of the Panama Basin, eastern equatorial Pacific. Geological Society of America Bulletin 82, 1482–1504.
Walther, C. H. E. (2003). The structure of the Cocos Ridge off Costa Rica. Journal of Geophysical Research 108, doi: 10.1029/2001LB000888.
Werner, R., Hoernle, K., v.d. Bogaard, P., Ranero, C., von Huene, R. & Korich, D. (1999). Drowned 14 m.y.-old Galápagos Archipelago off the coast of Costa Rica: implications for tectonic and evolutionary models. Geology 27, 499–502.
Werner, R., Hoernle, K., Barckhausen, U. & Hauff, F. (2003). Geodynamic evolution of the Galápagos hot spot system (Central East Pacific) over the past 20 m.y.: constraints from morphology, geochemistry, and magnetic anomalies. Geochemistry, Geophysics, Geosystems 4, 1108, doi: 10.1029/2003GC000576, posted at http://www.g-cubed.org.[CrossRef]
White, W. M. & Hofmann, A. W. (1978). Geochemistry of the Galápagos Islands: implications for mantle dynamics and evolution. Carnegie Institute of Washington Yearbook 77, 596–606.
White, W. M., McBirney, A. R. & Duncan, R. A. (1993). Petrology and geochemistry of the Galápagos Islands: portrait of a pathological mantle plume. Journal of Geophysical Research 98, 19533–19563.[CrossRef]
Wilson, D. S. (1996). Fastest known spreading on the Miocene Cocos–Pacific plate boundary. Geophysical Research Letters 23, 3003–3006.[CrossRef][Web of Science]
Wilson, D. S. & Hey, R. N. (1995). History of rift propagation and magnetization intensity for the Cocos–Nazca spreading center. Journal of Geophysical Research 100, 10041–10056.[CrossRef]
Wood, D. A. (1980). The application of a Th–Hf–Ta diagram to problems of tectonomagmatic classification and to establishing the nature of crustal contamination of basaltic lavas of the British Tertiary volcanic province. Earth and Planetary Science Letters 50, 11–30.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. G. Fitton The OIB paradox Geological Society of America Special Papers, January 1, 2007; 430(0): 387 - 412. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||