The well-studied Galápagos Archipelago is a small part of the much larger Galápagos Volcanic Province (GVP) consisting of the Cocos, Carnegie, Coiba and Malpelo aseismic ridges and related seamount provinces. In order to establish how well the GVP fits with the predictions of the "standard" fixed hotspot and mantle plume hypotheses we undertook the first systematic dredge-TV Grab sampling of the largely un-sampled submerged regions of the GVP (Figure 1) using the RV SONNE (Werner et al., 2003). We report here ⁴⁰Ar/³⁹Ar ages for SO144/3 dredge and two previously undated DSDP Leg 16 drill samples (Werner et al., 2000) and evaluate their implications for the many geochemical and geophysical models (summarised in O’Connor et al., 2007) that have been put forward to explain the origin of the GVP. We show that the temporal and spatial distribution of new and published ages is compatible with GVP formation via time-progression of broad regions of long-lived, and possibly concurrent, volcanism. We propose that such observations are most consistent with GVP formation via a process involving Cocos and Nazca tectonic plate divergence across a very broad Galápagos hotspot melting anomaly.

Figure 1: Ocean floor topography (Smith & Sandwell, 1997) showing the Galápagos Volcanic Province (GVP). Yellow dots indicate new sample sites (SO 144/3 Expedition) and DSDP Leg 16 sites. White dots show sample locations for published ⁴⁰Ar/³⁹Ar dated rock samples (Werner et al., 1999, Hoernle et al., 2002) and K/Ar dated subaerial samples sites on the Galápagos Islands (White et al., 1993). White boxes show PLUME-2 sample sites (Sinton et al., 1996). Click here or on figure for enlargement.

2. Results

Time-progressive GVP volcanism?

Correlating new Carnegie Ridge ⁴⁰Ar/³⁹Ar and published Galápagos Archipelago K/Ar (White et al., 1993) sample site ages with distance from the western leading-edge of the Galápagos Archipelago (active Cerro Azul volcano, SW Isabela Island) suggests an overall trend of time-progressive GVP volcanism on the Nazca tectonic plate (Figure 2a). Similarly, a broad overall trend of time-progressive GVP volcanism on the Cocos tectonic plate is indicated in a correlation of Cocos Ridge/seamount provinces/island sample site ages with distance from the western edge of the Galápagos Archipelago (Figure 2b). Combining new and published (Werner et al., 1999, Sinton et al., 1996, Hoernle et al., 2002) ⁴⁰Ar/³⁹Ar ages for Cocos and Carnegie ridges and seamounts (Figures 2c & d) reinforces our inference that, fundamentally, volcanism migrated across the GVP time-progressively. Evidence for volcanism between ~17 and ~11 Ma on Malpelo and Coiba ridges (Figure 2d) suggests that these ridges formed together with older regions of the Cocos and Carnegie ridges and seamounts. Malpelo and Coiba ridges have since rifted away from Carnegie and/or Cocos ridges due the complex spreading history of the CNS (Meschede & Barckhausen, 2001, Hoernle et al., 2002, Gahagan & Mann, 2002, Werner et al., 2003, Sallares & Charvis, 2003, MacMillan et al., 2004). Moreover, older parts of the Cocos and Carnegie ridges are likely to have been subducted (e.g., Gahagan & Mann, 2002, MacMillan et al., 2004).

Figure 2: (A) Correlation of new Carnegie Ridge ages with distance from the western leading edge of the Galápagos Archipelago. (B) Correlation of new Cocos Ridge/Seamounts/Island ages with distance from the western edge of the Galápagos Archipelago. (C) Correlation of combined Carnegie (A) and Cocos Ridge/Seamounts (B) ages with distance from the western edge of the Galápagos Archipelago. Also shown are published data for the Costa Rica Seamount Province (Werner et al., 1999). Dashed regression line is fitted to Carnegie Ridge samples SO144/3-17TVG, DSDP 157 and oldest K/Ar ages for the Galápagos Islands (White et al., 1993). The solid thin line shows broad agreement between GPS measured migration rate of 83 ± 3 km/My for the Cocos plate with that indicated by the age-distance correlation for Cocos plate GVP lavas erupted closest to the western edge of the Galápagos Archipelago. (D) Correlation of published ages for Malpelo and Coiba ridges (SO144/3 and DSDP 155), the Galápagos Platform (Sinton et al., 1996) and Malpelo Island (Hoernle et al., 2002) with distance from the western edge of the Galápagos Archipelago. Click here or on figure for enlargement.

Widespread GVP volcanism

The widespread nature of GVP volcanism is evident in a correlation of measured ages with distance and in the geographical distribution of new and published (Figure 1) ⁴⁰Ar/³⁹Ar and K-Ar (Galápagos Islands) assigned to 1 My time intervals or "bins" (Figure 3). This observation is further supported by existing evidence for widespread volcanism across the Galápagos Archipelago (White et al., 1993) and platform (Figure 3; Sinton et al., 1996).

Figure 3: Geographical distribution of new and published (Figure 1) ⁴⁰Ar/³⁹Ar and K/Ar ages assigned to 1 My time intervals or "bins". Colour coded intervals also suggest GVP development via a process leading to time-progression of widespread, long-lived and possibly synchronous volcanism. Volcanism continued for ~6 My (except DSDP 155), on Malpelo and Coiba ridges, ~5 My at the NE end of Cocos Ridge (except for much younger adjacent 46TVG and 47DR sample sites), ~5 My on Carnegie Ridge (excluding much younger 26 TVG and 28DR samples), ~5 My in the Cocos Seamount Province (except for the much younger <2 My old Cocos Island, 4 Cocos seamounts and 1 co-latitudinal Cocos Ridge sample). We therefore infer a ~5 My average for the duration of the main phase of volcanism (shield to post-shield?) across the GVP together with rarer post-erosional (rejuvenescent) phases linked to localised tectonics, e.g., reactivation of small-degree Galápagos "mantle plume melts" (Castillo et al., 1988), late stage spreading west of Cocos Island (Werner et al., 2003) or alkalic volcanism lasting for several million years after a ridge jump ("post-abandonment"; Batiza & Vanko, 1985). See text for discussion.

Long-lived GVP volcanism

Although we infer an average of ~5 My duration for volcanism across the GVP (Figures 2 and 3). Measured ages for the Galápagos Islands do not seem to support such an inference since volcanism apparently began ~3 Ma in the eastern part of the archipelago and has continued until <1 Ma (White et al., 1993, their Figure 5). However, the Galápagos Islands are still volcanically active so we cannot rule out that volcanism will continue across the archipelago into the future such that the final time span of volcanism may be ultimately in the same range as those observed for older GVP centres.

Coeval GVP volcanism?

While age data for the GVP remains limited there is nevertheless evidence for GVP formation via a process that leads to broad regions of coeval volcanism (Figures 2 & 3). The most notable observation is a 10-11 Ma region of dispersed concurrent volcanism on both the Cocos and Nazca tectonic plates (Figure 2). Moreover, these coeval regions overlap in a correlation of sample age with distance from the leading edge of the Galápagos Archipelago Platform suggesting a ~700-m-wide region of 10-11 Ma coeval GVP volcanism (Figures 2 & 3). This notion of broad regions of coeval regional volcanism is supported by evidence for recent (<1 Ma) synchronous volcanism across the Galápagos Archipelago (Figure 3).

3. Discussion

How broad is the Galápagos "hotspot melting anomaly"? Measured ages point to GVP formation via a process leading to time-progression of broad regions of long-lived, possibly coeval, volcanism. Such a process is more compatible in our view with the notion of a much broader Galápagos hotspot melting anomaly (300 to 400 km) than is incorporated into current models put forward to explain GVP origin and development (Figure 4).

Figure 4: Cartoon showing GVP development via the time-progression of widespread, long-lived and possibly coeval volcanism reflecting Cocos and Nazca plate migration (and divergence) across a broad Galapagos hotspot melting anomaly. Two end-member scenarios are: hotspot volcanism is initiated as the Cocos and Nazca plates pass across the leading edge of a broad anomaly and continues until the affected region has drifted downstream beyond the influence of the anomaly (panels A and B), and GVP volcanism is initiated across the anomaly and continues (randomly or episodically) as the Cocos and Nazca plates migrate across the anomaly (panels C and D). See text for further discussion. Scenario 1: (A) Cocos Plate: A hypothetical age-distance correlation for GVP volcanism initiated at the leading edge of a broad hotspot melting anomaly. Volcanism continues for ~5 Ma as the Cocos plate migrates across the "melting zone" at a rate of ~81 mm/yr. The melting anomaly diameter of ~400 km is inferred from a Cocos plate velocity of ~81 mm/yr and an average of ~5 Ma of volcanism at any particular locality across the GVP (Figure 2). The 0-5 Ma insert illustrates the development of coeval lines of dispersed volcanism orientated in the direction of plate motion and corresponding in size to the width of the melting anomaly (numbers represent millions of years since the lava cooled). Long solid line reflects age-distance correlation for initially erupted lavas. Long dashed sub-parallel lines illustrate age-distance correlation for lavas formed in million-year increments after initial volcanism. Short solid blue lines show time-progression of coeval lines and their proposed time-progression across the GVP. (B) Nazca Plate: Shows the corresponding age-distance correlation for GVP volcanism on the Nazca plate. The slower Nazca plate drift of ~60 mm/yr suggests a narrower ~300-km-wide melting anomaly leading to correspondingly shorter lines of coeval volcanism. Other details as in A. Scenario 2: Volcanism initiated across a broad hotspot melting anomaly. (C) Cocos Plate: Cartoon showing GVP formation by ongoing volcanism (random or episodic) across a broad Galápagos hotspot melting anomaly. Other details as in (A). (D) Nazca Plate: Shows the corresponding model in the case of the Nazca tectonic plate. Other details as in B.

Both scenarios predict similar GVP spatial and temporal development, and are compatible with our observation that the volcanism is:

1. broadly time-progressive,
2. widespread,
3. long-lived (we infer an ~5 Ma average lifespan), and
4. possibly formed in bands of coeval volcanism orientated in the direction of Cocos and Nazca plate motions, respectively (Figure 4).

Since both end-member possibilities lead to indistinguishable patterns of widespread, long-lived and possibly coeval volcanism it is not possible, from our data, to determine which might be the dominant mechanism or whether interplay between both is involved. Future studies aimed at distinguishing between these two end-members are important in our view. Scenario 1 is more compatible with current models requiring a focus of upwelling at the leading edge of the hotspot (e.g., a narrow plume conduit). Scenario 2 involves initiation of volcanism across the anomaly so suggesting the need to propose and test a range of possible mantle upwelling (plume?) shapes, sizes and dynamics.

4. Conclusions

New and published isotopic basement ages suggest that the GVP formed via a process leading to time-progression of broad regions of long-lived and possibly concurrent zones of volcanism. We infer an average of ~5 Ma of shield to post-shield volcanism across the GVP together with much rarer occurrences of significantly later post-erosional (rejuvenated) volcanism linked to localised tectonic control. Moreover, our inferred migration rates for GVP volcanism on the Cocos and Nazca tectonic plates, although poorly constrained nevertheless seem to correspond with their respective GPS-measured present-day relative plate velocities. We conclude that the complex history and distribution of GVP volcanism might be explained by Nazca and Cocos tectonic plate divergence/migration across a broad hotspot melting anomaly on the order of 300-400 km wide.

The distribution of GVP isotopic ages is equally well explained by

1. onset of volcanism at the leading edge of a broad hotspot melting anomaly, and
2. random or episodic volcanism across a broad anomaly.

Both end-member scenarios predict long-lived (~5 Ma) volcanism across the GVP reflecting the time taken for the Nazca and Cocos plates to drift across hotspot. While both scenarios require a broad Galápagos melting anomaly it is not possible to establish which might be the dominant mechanism. Distinguishing between these end-member possibilities is important since in our view Scenario 1 - onset of volcanism at the leading edge of such a melting anomaly - is a more passive process that fits better with formation of the GVP by a narrow deep-seated continuous plume "tail" or "conduit"’ at the western leading edge of the Galápagos Archipelago platform. Such a posited mantle plume conduit is thought to be tilted by shear in the upper mantle due to (Nazca) plate motion and upper mantle flow. Thus, widespread long-lived volcanism across the Galápagos Archipelago and underlying platform has been attributed to long-lived melting downstream from the leading edge of the Galápagos hotspot (i.e., "plume conduit") or "post-shield extensional volcanism" across the platform region at earlier formed shield volcanoes (White et al., 1993, Sinton et al., 1996). Both of these mechanisms require the existence of narrow mantle plume "conduit" or "tail" at the leading western edge of the Galápagos archipelago on the scale of a single island (e.g., Isabela Island).

In contrast, Scenario 2 - random or episodic volcanism across the anomaly - does not require a narrow focused upwelling anomaly (plume conduit). Our notion of a broad Galápagos hotspot melting anomaly therefore opens the way to testing a greater range of possible mantle upwelling (plume?) shapes, sizes and dynamics (e.g., thermo-compositional?) via, for example, combined dredge-, ROV- and drill-sampling and tomography, seismic and numerical studies.