Introduction and background

Knowledge about large igneous provinces (LIPs) has played a fundamental role in shaping the prevailing view of mantle geodynamics, that of largely upper mantle plate-driven flow punctuated by rising thermally driven plumes from the lower mantle (e.g., Davies, 1992). The largest LIPs, the oceanic plateaus, continental flood basalts, and volcanic passive margins, reach volumes of several 106 to several 107 km3 and are apparently the product of relatively short lived massive magmatic episodes that represent the largest nonridge volcanic process on Earth (e.g., Coffin and Eldholm, 1994). In terms of magma flux, volume, and extent, such LIPs dwarf even the most prodigious present-day hotspots, such as Iceland and Hawaii. Magma production rate for the largest LIPs rivaled or even surpassed that of the global mid-ocean-ridge system for short periods of time (e.g., Tarduno et al., 1991; Duncan and Richards, 1991; Mahoney et al., 1993; Coffin and Eldholm, 1994). Moreover, because many of the largest LIPs formed during the Mesozoic, they may represent a mantle convection regime different from that of the ridge-dominated Cenozoic (e.g., Stein and Hoffman, 1994; Machetel and Humler, 2003).

A widely accepted explanation for plateaus and continental flood basalts is the plume head hypothesis, which posits large (from several hundred to ~2000 km in diameter), bulbous, primarily thermal diapirs that are created at depth in the mantle, probably within the core/mantle boundary zone, and which rise toward the surface, causing cataclysmic volcanism when they impact the lithosphere (e.g., Richards et al., 1989; Griffiths and Campbell, 1990). Like the related plume hypothesis for volcanism at hotspots (Morgan, 1972, 1981; Sleep, 2007), the plume head hypothesis has been accepted by many workers because it provides a simple framework that seems to tie together many observations. Moreover, the plume head phenomenon occurs naturally in numerical and laboratory experiments, given appropriate rheologic conditions (e.g., Whitehead and Luther, 1975; Griffiths and Campbell, 1990, 1991). The trouble is that there is currently no unequivocal geological evidence proving that the plume head mechanism has operated within Earth. Many existing data are indirect indicators of eruptive rate and magmatic volume and could be explained by alternative hypotheses. A growing debate about the number, characteristics, and even existence of mantle plumes (e.g., Smith and Lewis, 1999; Anderson, 2001; Foulger, 2002, 2007; Courtillot et al., 2003; Sleep, 2003; Foulger and Natland, 2003; DePaolo and Manga, 2003) makes it desirable to consider alternative explanations for plateaus. Because ocean plateaus are argued to be the most direct expression of mantle plume heads (i.e., unlike continental LIPs, where magmas must pass through continental lithosphere), understanding oceanic plateau formation is thus critical to understanding mantle dynamics.

Operations during Ocean Drilling Program (ODP) Legs 183 and 192 drilled Kerguelen and Ontong Java plateaus, respectively, seeking evidence that would test the plume head hypothesis (Frey et al., 2003; Fitton et al., 2004). These two plateaus were targeted because they are the largest, and prior data suggested that each formed within a narrow range of ages, as predicted by the plume head model. However, both expeditions uncovered complications that do not fit the simple plume head model (e.g., Coffin et al., 2002; Fitton et al., 2004), so debate over plume heads continued.

In order to address the plume head versus alternative hypotheses, it is necessary to study a plateau for which the relation of the plateau to contemporaneous mid-ocean ridges is known. Unfortunately, this condition is not met for plateaus formed during the Cretaceous Long Normal Superchron (also known as the Cretaceous Quiet Period), such as the Ontong Java, Manihiki, and Kerguelen plateaus, because of the lack of magnetic reversals and thus linear seafloor magnetic anomalies to mark the locations of spreading ridges. Shatsky Rise, located in the northwest Pacific (Figs. F1, F2), is the only large intra-oceanic plateau formed at a time of magnetic reversals. Contemporaneous magnetic lineations exist around and within the plateau, providing a framework that allows development of a tectonic model (Nakanishi et al., 1999; Sager et al., 1999). This model is currently based on geophysical inference with little solid geological evidence from sampling.

Shatsky Rise is also unique because it has characteristics that suggest both plume head and ridge-controlled origins (Sager, 2005). The plateau’s size, morphology, apparent eruption rate, and age progression are consistent with a plume head origin (Sager and Han, 1993; Nakanishi et al., 1999; Sager et al., 1999). In contrast, the plateau formed at a triple junction during a time of ridge reorganization, which suggests a link to ridge tectonics (Sager et al., 1999; Sager, 2005). Furthermore, existing Nd-Pb-Sr isotopic data for the few basalts cored and dredged from Shatsky Rise show a Pacific mid-ocean-ridge basalt (MORB)–type signature, not the expected ocean island–type signature of a plume head eruption (Mahoney et al., 2005). Whether this MORB affinity is representative of the rise or characterizes only a few minor, late-stage magmas is unknown. However, the fact that existing data for Shatsky Rise can be interpreted both ways suggests that this plateau is uniquely suited for testing plume head versus ridge tectonics models. Moreover, because several, perhaps many, oceanic plateaus formed at triple junctions (e.g., Winterer, 1976; Larson et al., 2002; Sager, 2005; Ishikawa et al., 2005; Smith, 2007), Shatsky Rise probably represents a significant class of ocean plateau, if indeed it is not representative of all.

In the following sections, we briefly review the state of plateau-formation hypotheses, existing knowledge of Shatsky Rise, and why the rise is uniquely suited for testing plume head and ridge tectonics hypotheses. We explain how these hypotheses can be tested by drilling and detail a single-leg nonriser drilling program to sample Shatsky Rise.

Plateau formation hypotheses

Ocean plateaus are remote and difficult to sample. The resulting geological ignorance has led investigators to propose several plateau-formation mechanisms. One class of explanation calls upon anomalous behavior of tectonic plates, such as leaky transform faults (Hilde et al., 1976) or spreading ridge reorganizations (e.g., Anderson et al., 1992; Foulger, 2007). Another class invokes a mantle plume, either as a steady-state plume "tail" beneath a spreading axis or a plume head (Richards et al., 1989; Mahoney and Spencer, 1991; Duncan and Richards, 1991; Coffin and Eldholm, 1994). A third type of mechanism explains plateau formation as a result of a large meteorite impact (Rogers, 1982; Roddy et al., 1987).

The mantle plume hypothesis has been widely accepted, in part because of known shortcomings or lack of development of other hypotheses. The meteorite impact hypothesis was first proposed before discovery of the Chicxulub impact crater (e.g., Hildebrand and Boynton, 1990) and several other large impact sites that were subsequently documented on the continents. Combined with a lack of evidence linking plateaus and impacts, the idea lay fallow for many years. However, this hypothesis has been revisited for the Ontong Java Plateau (Ingle and Coffin, 2004; Tejada et al., 2004) because evidence from the plateau does not neatly fit other hypotheses. Plate boundary mechanisms have gained only limited support, partly because they require the assumption of extensive regions of shallow near-solidus asthenosphere that differ geochemically from the shallow asthenosphere that forms mid-ocean ridges and partly because they may not be able to produce the volumes of magma required for the largest plateaus, such as the Ontong Java Plateau. Creating LIPs through cracks, even in a thick part of an oceanic plate, requires that a seemingly small perturbation unleash a massive volcanic event. Consequently, large volumes of anomalously warm asthenosphere or unusually chemically "fertile" (fusible) mantle primed to undergo massive decompression melting must be assumed (Anderson et al., 1992; Foulger, 2007).

Plume-based explanations for plateaus have been bolstered by a wide acceptance of the mantle plume hypothesis for oceanic islands. The idea that thermal (and perhaps chemical) instabilities from the lower mantle rise to the base of the plate and cause hotspot volcanism initially became popular because it provided a neat explanation for age-progressive volcanic chains (Wilson, 1963; Morgan, 1971, 1972; Glen, 2005; Anderson and Natland, 2005). As more age-progressive seamount chains have been found, this explanation has been used repeatedly, with one result being an unlikely large number of proposed plumes. In part, this problem stems from loose application of the plume definition. Recent reexamination of hotspots led Courtillot et al. (2003) and Anderson (2005) to conclude that only a small number fit the original plume concept, that of a thermal diapir originating at or near the core/mantle boundary. Instead, many hotspots, especially smaller ones, likely have shallower sources that may or may not be related to significant thermal upwelling.

The plume head hypothesis arose as an offshoot of the traditional plume hypothesis. It was observed experimentally that if viscosity conditions are appropriate, then perturbations in a gravitationally unstable fluid layer form large bulbous heads that rise through the overlying fluid and that the heads are followed by tails of rising lower-layer material (Whitehead and Luther, 1975; Richards et al., 1989; Griffiths and Campbell, 1990). Such models led to the idea that mantle plumes form near the core/mantle boundary, begin with massive diapirs (plume heads) that rise through the mantle and are fed and followed by a narrow conduit of the same lower-layer material (plume tail). Other versions of the plume head model start plumes from a shallower level, which serves either as the primary source region (e.g., Allègre and Turcotte, 1985; White and McKenzie, 1989; Kellogg et al., 1999) or as a barrier to a lower-mantle plume head, which then creates an upper-mantle plume head by heating from below (Tackley et al., 1993). All of these hypotheses are similar in that they require large thermal (and/or chemical) anomalies that arise at depth and carry deep-mantle material to the base of the plate.

Impingement of a plume head on the lithosphere is thought to lead to voluminous production of basaltic magma, forming an oceanic plateau or continental flood basalt, depending on the type of lithosphere (e.g., Richards et al., 1989; Campbell, 1998). Wide acceptance of this hypothesis rests on radiometric ages indicating that several flood basalts and plateaus were formed rapidly, on the ocean island–like Nd-Pb-Sr-Hf isotopic signatures of many flood basalt sources, and on several long-lived seamount chains that can be traced back to a flood basalt province (e.g., Campbell, 1998). Recently, however, dating results from Leg 183 on Kerguelen Plateau and the Caribbean LIP indicate a longer, more complex emplacement history than previously thought (Duncan, 2002; Coffin et al., 2002; Hoernle et al., 2004). Also, although Ontong Java basalts have an ocean island–type isotopic signature and most of the plateau appears to have formed rapidly (at ~120 Ma), an associated postplateau seamount chain is lacking; plus, the initial depth of much of the plateau was well below that predicted by the plume head model and the amount of posteruptive subsidence has been less than predicted (e.g., Mahoney, Fitton, Wallace, et al., 2001; Fitton et al., 2004, Roberge et al., 2005). The effect of such complications for the plume head model is still being sorted out.

One possible explanation for complex geologic histories for plumes comes from the thermochemical plume hypothesis, in which plume buoyancy is fueled not only by a difference in temperature between the plume and surrounding mantle but also by density differences resulting from chemical composition (Davaille et al. 2003, 2005; Farnetani and Samuel 2005; Lin and van Keken 2006a, 2006b). The primary implications of this type of plume are that it may not behave as would a simple thermal plume, potentially having an extended residence in the lower mantle, perhaps stalling at intermediate mantle depths and not resulting in the magnitude of uplift expected from a thermal plume, and having more than one pulse of flood basalt volcanism.

Why study Shatsky Rise?

During the mid-1990s, ocean drilling studies focused on Kerguelen and Ontong Java plateaus because they are by far the largest, most outstanding examples of LIPs. A serious limitation to understanding the origin of these two plateaus is that they formed mainly during the Cretaceous Long Normal Superchron, so their relationship to contemporaneous spreading ridges cannot be determined. Shatsky Rise is important because it is the only large intra-oceanic plateau that formed during a time of magnetic reversals and the magnetic lineations that run through the plateau (Fig. F3) imply that it formed at a triple junction. Knowledge of oceanic plateaus is still so rudimentary that we cannot be certain whether Shatsky Rise and Ontong Java Plateau, for example, formed by the same mechanism. Many other plateaus have formed at triple junctions (Sager, 2005), so Shatsky Rise probably represents a class of ridge-related plateaus, if not all ocean plateaus.

If one wishes to study plume volcanism at a mid-ocean ridge, why not just study Iceland? Several observations imply that the two LIPs have significant differences. Iceland and Shatsky Rise differ in estimated magmatic flux by more than an order of magnitude. If one accepts the plume hypothesis for Iceland (which is not universally accepted; e.g., Foulger, 2002; Foulger et al., 2005), Iceland is well suited for studying present-day plume-ridge interaction; however, Shatsky Rise permits study of the interaction between a plume head and a ridge. Furthermore, Shatsky Rise formed at a triple junction, whereas Iceland did not. In addition, the initial formation of Shatsky Rise appears coincident with a plate reorganization and large (800 km) jump of the triple junction, implying a connection with large-scale plate tectonics. Although small ridge jumps have occurred at Iceland, they have been far smaller than the scale of the jump associated with Shatsky Rise.

Although there are a dozen or so large oceanic plateaus, Shatsky Rise is unique in its setting and holds critical clues to understanding plateau formation. It is a high priority for study for the following reasons:

  1. With an area of ~4.8 x 105 km2 (about the same as Japan or California) and total volume of ~4.3 x 106 km3, Shatsky Rise is one of the largest ocean plateaus (Sager et al., 1999). Moreover, bathymetric ridges and lava geochemistry suggest Shatsky Rise and Hess Rise (Fig. F1) may have arisen from the same source (Bercovici and Mahoney, 1994), which would nearly double the magmatic output. Magmatism of this scale requires something significantly unusual about the physical and/or chemical state of the source mantle.

  2. The fact that Shatsky Rise formed during a time of magnetic reversals makes it easier to understand than any other large ocean plateau. Magnetic reversals recorded in Shatsky lavas provide constraint on the rise’s structure and original tectonic setting (e.g., Sager and Han, 1993; Sager, 2005). Magnetic anomalies can be used not only to date the plateau and surrounding lithosphere but also to understand how plateau morphology is related to ridge tectonics (e.g., Sager et al., 1999; Nakanishi et al., 1999).

  3. Morphology, apparent age progression, and magnetic lineations together indicate that the rise volcanism was spread out laterally, perhaps owing to rapid movement of the Pacific plate over the source mantle (Nakanishi et al., 1999; Sager et al., 1999). In contrast, the volcanic record of larger plateaus formed on a slowly moving plate (e.g., Ontong Java Plateau) may mainly consist of a vertically stacked pile. Therefore, the tectonic and geochemical evolution of Shatsky Rise is easier to address through drilling (i.e., shallow holes distributed laterally).

  4. Shatsky Rise formed at a ridge-ridge-ridge triple junction of rapidly spreading ridges; consequently, the lithosphere was young and thin, so lithospheric contamination of magmas should be minimal. Likewise, variations in lithospheric thickness should not have influenced magma compositions significantly and shallow melting beneath a thin lithosphere potentially allowed for greater partial melting.

  5. Because of its location exactly along the track of a migrating triple junction and the fact that it appears to share both ridge and plume characteristics, Shatsky Rise is uniquely suited to testing plume head versus ridge-controlled hypotheses of plateau genesis.

Why is drilling required? Although the main edifices of Shatsky Rise have some basaltic outcrops, dredged samples of igneous basement suitable for geochemical and geochronological work have proven difficult to get and harder to study. Because of the Late Jurassic to Early Cretaceous age of the oldest part of the plateau, outcrops on these edifices have probably been exposed for long periods. All outcrops dredged to date are coated with thick ferromanganese oxide deposits that make basement rock recovery difficult. Likewise, all existing dredge samples are highly altered. Although dredged basalts were recovered during the 1994 site survey cruise, the recovered samples are highly altered and did not produce reliable radiometric dates, even with the most modern techniques (M. Pringle, unpubl. data). Likewise, all but a very few samples were unsuitable for chemical or isotopic studies (Tejada, 1998; Tatsumi et al., 1998). In addition, the dredge sites were all promontories, ridges, and other high points that may not be representative of the main plateau-building lavas. In sum, the only way to obtain samples that can address the origin of Shatsky Rise is by drilling a series of holes across the plateau and recovering hundreds of meters of basement igneous rock.

Prior research on Shatsky Rise

By the late 1960s it was known that Shatsky Rise is ancient because Early Cretaceous sediments were cored from its summit (Ewing et al., 1966). Although seismic refraction experiments have not yet imaged the Mohorovicic Discontinuity beneath the high parts of the plateau, they have revealed anomalously thick crust with a similar velocity structure to normal oceanic crust but several times thicker (Den et al., 1969; Gettrust et al., 1980). Seismic profiling showed that the tops of the rise edifices hold thick piles of pelagic sediments (up to 1.2 km), whereas sediments on the rise flanks are thin or absent in places (Ewing et al., 1966; Ludwig and Houtz, 1979; Neprochnov et al., 1984; Sliter and Brown, 1993).

Several Deep Sea Drilling Project (DSDP) and ODP cruises have cored Shatsky Rise over a span of 32 years. In succession, operations during DSDP Legs 6 (Sites 47–50), 32 (Sites 305 and 306), and 86 (Site 577) as well as ODP Legs 132 (Site 810) and 198 (Sites 1209–1214) cored atop the highest southern massif of the rise (Tamu Massif) (Fig. F2). Many of the holes had only shallow penetration. Drilling during Leg 32 probed deep into the sedimentary cap, recovering Berriasian (earliest Cretaceous) sediments ~50 m above the expected level of basement at Site 306 (Fig. F2). This finding was significant because it implied that Tamu Massif formed during latest Jurassic or earliest Cretaceous time. Recently, operations during Leg 198 cored sediments from all three of the Shatsky Rise massifs (Bralower, Premoli-Silva, Malone, et al., 2002), including Ori Massif (Site 1208) and Shirshov Massif (Site 1207). At the last two sites, only the upper part of the sedimentary section was cored, reaching Late Cretaceous sediments. Igneous basement has been reached only twice. During Leg 6, drilling stopped at the top of supposed basement at Site 50, recovering only a few pebbles of basalt, perhaps from a basal conglomerate (Fischer et al., 1971; Melson, 1971). At Site 1213 on the southwest flank of Tamu Massif (Fig. F2), operations during Leg 198 cored a 46 m section of little-altered basaltic sills intruding earliest Berriasian sediments (Shipboard Scientific Party, 2002). These basalts produced the first reliable radiometric date for Shatsky Rise, as well as valuable chemical and Nd-Pb-Sr isotopic data (Mahoney et al., 2005).

Magnetic lineations mapped in the northwest Pacific revealed that Shatsky Rise sits at the confluence of two lineation sets: the northeast-trending Japanese lineations and the northwest-trending Hawaiian lineations (Figs. F1, F2, F3) (Larson and Chase, 1972; Hilde et al., 1976). This circumstance indicates that the plateau formed at a triple junction separating the Pacific, Farallon, and Izanagi plates (Larson and Chase, 1972). Subsequent studies revealed that the triple junction jumped repeatedly during the time it occupied the location of the rise and that it must have been geometrically unstable to follow the path of the rise (Sager et al., 1988, 1999; Nakanishi et al., 1999). Furthermore, age constraints (Cretaceous sediments and the Site 1213 radiometric date), seismic stratigraphy, and isostatic compensation all indicate that the age of the rise is near that of the adjacent seafloor (Sager et al., 1999), implying that the triple junction and rise formation are linked. Current thought is that a plume head is the link, a source of heat, uplift, and volcanism that both created the rise and captured the triple junction (Sager et al., 1988, 1999).

Magnetic data were also instrumental in supporting the idea that Shatsky Rise formed from a plume head. Sager and Han (1993) postulated that the rise formed rapidly, based on modeling of the magnetic anomaly over Tamu Massif. They noted that the magnetic anomaly implies mainly reversed polarity, in turn implying that most of the edifice may have formed during a single interval of reversed polarity. With simple calculations using the massif volume and an estimate of the length of the single polarity period, the authors inferred that the massif formed with an eruption rate similar to those of several large flood basalts (~1.8 km3/y).

Recent analyses have refined and expanded these conclusions. Paleomagnetic analysis of Site 1213 basalt samples gives inclination values that are most consistent with reversed magnetic polarity (Tominaga et al., 2005). Furthermore, the mean 40Ar-39Ar age from two basalt samples from the sills is 144.6 ± 0.8 Ma (2σ error) (Mahoney et al., 2005), a value identical to the age of the Jurassic/Cretaceous boundary and that correlates with magnetic Anomaly M19 in the Gradstein et al. (1995) timescale. This result limits the formation of much of the Tamu Massif to between Anomalies M21 and M19, a period of 1.5 m.y. If Tamu Massif formed during a single polarity interval, it is likely either Anomaly M20 or M19, with durations of 0.4 and 0.75 m.y., respectively. Assuming the volume of the massif between Anomalies M21 and M19 formed in 1.5 to 0.4 m.y. (and making the conservative assumption that it formed on existing [very young] 7 km thick crust) implies volcanic emplacement at rates of 1.2 to 4.6 km3/y (Sager, 2005). Again, such values are in the range of estimates for several large continental flood basalts (e.g., Richards et al., 1989; Johnson and Thorkelson, 2000). Although these estimates are intriguing, they were made by very indirect means and require confirmation from radiometric ages of igneous basement samples, particularly from other locations on the rise.

Formation and tectonic history of Shatsky Rise

Much of what is known about the tectonics of Shatsky Rise is based on the magnetic lineations that surround the plateau and in some places transect it (Fig. F3) (Sager et al., 1988; Nakanishi et al., 1989, 1999). The lineations range from Anomaly M21 (147 Ma; polarity ages from Gradstein et al., 1995), bordering the southwest edge of the plateau, to M1 (124 Ma) at the northern tip of Papanin Ridge (Figs. F1, F2). Magnetic lineations have been mapped on the southeast flank of Tamu Massif, on flanks all around Ori and Shirshov massifs, in the basins between massifs, and all through Papanin Ridge (Fig. F3); indeed, little of Shatsky Rise is without magnetic lineations. This observation led to the conclusion that the rise consists of three large edifices (Tamu, Ori, and Shirshov massifs) surrounded by lithosphere that is not greatly modified by plateau-building igneous activity (Sager et al., 1999; Nakanishi et al., 1999).

Shatsky Rise volcanism displays a progression in both age and volume along the trace of the triple junction. Rise volume decreases markedly with distance from Tamu Massif. This edifice has an estimated total crustal volume of 2.5 x 106 km3, whereas Ori and Shirshov massifs each have volumes of 0.7 x 106 km3. Papanin Ridge, at the north end of the plateau, has a volume of 0.4 x 106 km3 and the low ridge implies a low volcanic flux over a long period (Sager et al., 1999). Age also apparently decreases with distance from Tamu Massif, with the ages of the volcanic edifices close to those of the underlying lithosphere, as suggested by isostasy (Sandwell and MacKenzie, 1989). The 144.6 Ma age for the Site 1213 sills is coincident with Anomaly M19, implying the bulk of the massif is Anomaly M19 age or older. The Ori and Shirshov massifs must be younger than Tamu Massif because they reside on lithosphere younger than M19. The youngest magnetic lineation beneath both Ori and Shirshov massifs is M14 (136 Ma), and Papanin Ridge is underlain by Anomalies M10 to M1 (from 131 to 124 Ma). These observations are consistent with a northeastward-younging trend and volcanism following the triple junction path.

Magnetic lineations also show that a geometrically stable triple junction was moving northwest (in a Pacific plate reference frame) prior to Anomaly M22 time (Fig. F4). At Anomaly M21 time, the triple junction began to reorganize, with the Pacific-Izanagi isochrons rotating 30°, leading to microplate formation and an 800 km eastward jump of the junction to the location of Tamu Massif (Sager et al., 1988, 1999; Nakanishi et al., 1999). Afterward until Anomaly M3 time (126 Ma) Shatsky Rise formed along the trace of the triple junction. During this time the triple junction jumped repeatedly, at least nine times (Fig. F4) (Nakanishi et al., 1999). In addition, the main volcanic massifs have sides parallel to spreading ridges and transform faults. Together, these observations imply that the rise of volcanism was episodic and tied to ridge jumps (Sager et al., 1999).

Geochemical data

Chemical and isotopic data from igneous rocks are important for understanding the formation of ocean plateaus because such data provide key information on mantle sources and the conditions of magma genesis. For Shatsky Rise, such data are few. Only a small number of dredges have recovered basalt, and all of the samples are highly altered, making the interpretation of geochemical data difficult. Tatsumi et al. (1998) concluded from Nb-Zr-Y data that a seamount within the rise has an ocean island–like composition similar to volcanoes of the South Pacific Superswell region, a finding that was interpreted as evidence for a plume head lower-mantle source. Whether or not Superswell mantle sources come from the lower mantle is a subject of debate (e.g., Janney and Castillo, 1999; Lassiter et al., 2003; Natland and Winterer, 2005), but in any case the seamount is located in a basin between the Tamu and Ori massifs (dredge D11; Fig. F3) and may have been formed after the rise itself.

In contrast, the Site 1213 basalts and two of the least-altered dredge samples from the Tamu and Ori massifs (dredges D9 and D14; Fig. F3) display distinctly MORB-type isotopic characteristics (Mahoney et al., 2005). Age-corrected Nd and Pb isotope ratios of these rocks (e.g., εNd(t) = +9.8 to +8.6) are within the range for Pacific MORB and, despite seawater alteration effects, Sr isotope values (0.70269–0.70280) are also MORB-like (Fig. F5). Furthermore, the Site 1213 basalts have broadly MORB-like incompatible element patterns (Fig. F6). The plume head model predicts ocean island–like, not MORB-like, isotopic compositions (e.g., Campbell, 1998). Thus, at face value the few existing data do not support a plume head origin. However, Site 1213 basalts are sills and the D9 and D14 dredge hauls sampled summit ridges; such late-stage volcanic products may not be representative of the main plateau-building lava pile beneath.

Sea level indicators

A plume head should produce both dynamic and constructional uplift, implying that much of the area atop a plateau will initially be subaerial, particularly if formed on young lithosphere, as with Shatsky Rise (e.g., Griffiths and Campbell, 1990, 1991). For most of Shatsky Rise, evidence on basement paleodepth is lacking; however, a dredge from the upper flank (dredge D12; Fig. F3) of Tamu Massif recovered shallow-water fossils (rudist casts and corals) (Sager et al., 1999). Because the summit of Tamu Massif is higher, it must have been at or above sea level. Furthermore, a flat summit on Shirshov Massif (beneath the sediment cap) as seen in seismic profiles (Sager et al., 1999) indicates erosion by wave action. Thus, it appears likely that conditions during emplacement were sufficient to raise some areas of the rise above sea level. The anticipated recovery of sediments (including benthic fossils) resting immediately above the igneous basement during Expedition 324 may help to constrain paleodepths of the rise summits. However, as indicated on the seismic profiles (see "Site summaries"), the oldest sediment layers often thin toward the proposed drill sites (selected to have least sediment cover) and therefore the recovered overlying sediments might be not contemporaneous to the latest volcanism.

What formed Shatsky Rise—a plume head or ridge tectonics?

Shatsky Rise was initially attributed to plume volcanism because it is a very large, somewhat linear igneous construct (Sager et al., 1988; Nakanishi et al., 1989). Indirect evidence of a rapid eruption rate led to the proposal that the plateau formed from a plume head (Sager and Han, 1993; Nakanishi et al., 1989; Sager et al., 1999). At first blush, this explanation seems a good one. It predicts a trail of age-progressive volcanism tracking the motion of the plate over a nearly fixed source (Morgan, 1971, 1972). Shatsky Rise seems to fit this criterion because existing age constraints imply that the rise becomes younger northeastward. Aseismic ridges and seamount chains connect Shatsky Rise with Hess Rise, apparently continuing the eastward-younging trend. Moreover, a similarity of ages and trends between the Shatsky and Hess rises and the Mid-Pacific Mountains even suggests that the volcanic tracks record the motion of the Pacific plate over nearly fixed mantle sources (Sager, 2005).

Arrival of a plume head should cause voluminous flood basalt–type magmatism, with peak volcanism occurring over a brief period (<2 m.y. in several continental flood basalts) and significant amounts of initial uplift (e.g., Richards et al., 1989; White and McKenzie, 1989; Campbell and Griffiths, 1990; Duncan and Richards, 1991). As summarized above, existing evidence indeed suggests that at least the highest portions of Tamu Massif were initially shallow. Although emplacement rates are not known for most of Shatsky Rise, the radiometric age of the Site 1213 basalts combined with the nearby seafloor magnetic lineations suggests that Tamu Massif was constructed at a very high average rate between 1.2 and 4.6 km3/y. The upper value is more than a quarter of the 16.8 km3/y of new ocean crust (e.g., Larson, 1991) estimated to be formed worldwide today at ocean ridges. Moreover, the estimated 1.8 x 106 km3 volume of the initial Tamu Massif eruption implies a source volume equivalent to a sphere 224–408 km in diameter, assuming a mean melt fraction between 5% and 30% (cf. Coffin and Eldholm, 1994), a volume consistent with supply by an actively upwelling plume head.

The geometry of Shatsky Rise also appears to support the plume head hypothesis. Apparently, the emplacement rate of igneous rock waned with time, as shown by the northeastward decrease in size coupled with the ages inferred from magnetic lineations; this decrease is consistent with a transition from plume head to plume tail (Sager et al., 1999). A plume-type hypothesis is likewise an attractive explanation for the odd behavior of the Pacific-Farallon-Izanagi triple junction during the ~20 m.y. that the plateau was forming. The arrival of a plume head, a major source of heat and tensional stress on the lithosphere, is a potential reason for the initial 800 km jump of the triple junction. Heat and flux of upwelling mantle from a plume might have "pinned" the triple junction near the plume head (and later, tail), explaining the repeated triple junction jumps and the observation that the triple junction did not migrate away from the rise as it should have given the velocities of surrounding plates (Sager et al., 1988). In short, a plume head is a plausible explanation for many Shatsky Rise characteristics.

However, some important observations are not explained easily, if at all, by the plume head model. The MORB-type isotopic signature of the existing Shatsky basalts already has been noted. Another nagging point is the ridge reorganization that occurred near the time that Shatsky Rise formed. Just after Anomaly M21 time, synchronous with the beginning of Shatsky Rise eruptions, the Pacific-Izanagi Ridge rotated ~30° (Sager et al., 1988). It is generally accepted that plate motion is driven primarily by subduction (e.g., Lithgow-Bertelloni and Richards, 1998), so it is unclear how a plume head could cause plate velocity to change by acting on the trailing boundary at the ridge. Although a plume may tend to "capture" nearby ridges because it is a major source of heat and actively upwelling mantle (e.g., Kleinrock and Phipps Morgan, 1988), the ridge reorientation occurred >800 km from the alleged plume center. If plume activity and plate motions are independent or only loosely coupled, as is widely believed (e.g., Eldholm and Coffin, 2000), the temporal proximity of these two events would have to be a coincidence.

Another apparent coincidence is the proximity of plume head and triple junction. Although a ridge or triple junction may jump or reorganize to stay near a plume (e.g., Kleinrock and Phipps Morgan, 1988), this assumes that the ridges are already near the plume. How likely is a plume head to rise within 800 km of a triple junction? Assuming plumes form randomly, the probability of one striking within 800 km of a triple junction is only ~0.4%. If more than one plume head erupted within a given period, the probability can be increased by a factor of N, where N is the number of plumes. This simple calculation ignores mantle convection or basal lithosphere topography that might help steer plumes toward a ridge (e.g., Courtillot et al., 1999; Jellinek et al., 2003; Braun and Sohn, 2003). Nevertheless, having a plume head "find" a triple junction would seem a low-probability event.

Curiously, western Pacific bathymetry and magnetic lineations seem to imply that other similar plume-ridge coincidences occurred. Some other plateaus formed along or near the paths of the Pacific-Farallon-Izanagi triple junction as well as the Pacific-Farallon-Phoenix triple junction, located on the east end of the Pacific plate. Moreover, many of these plateaus are located near proposed ridge reorganizations. After Shatsky Rise, Hess Rise may have formed near the track of the Pacific-Farallon-Izanagi triple junction as it jumped eastward. Similarly, Magellan Plateau, the oldest part of the Mid-Pacific Mountains, and probably the Manihiki Plateau were all formed near the track of the Pacific-Farallon-Phoenix triple junction (Sager, 2005). Explaining all of these plateaus by plume heads independent of ridge dynamics requires many recurrences of a low-probability event. To remain plausible, the plume head hypothesis must assume that plumes and triple junctions are somehow attracted to each other.

How could ridge tectonics lead to plateau formation? Triple junctions could be the key. Ridges that meet at a triple junction are a focal point for strong upwelling (e.g., Georgen and Lin, 2002), but present-day triple junctions are clearly not sites of plateau formation. The discrepancy between the excess volcanism associated with Late Jurassic and Early Cretaceous Pacific triple junctions and the paucity of such activity during the Late Cretaceous through Cenozoic may be explained by the "fertile" mantle hypothesis (also known as the "perisphere" hypothesis) (e.g., Anderson et al., 1992; Anderson, 1995; Smith and Lewis, 1999; Smith, 2003; Foulger, 2007). This hypothesis states that extensive regions of the shallow asthenosphere have a lower melting point (because of higher volatile content, a more mafic composition, and/or higher potential temperature) than the asthenosphere beneath the present-day ridge system. Although the fertile mantle hypothesis is rejected as a general explanation by many, the Late Jurassic–Early Cretaceous Pacific may be a very special case. During this period, much of the Pacific plate (which was then far smaller than at present) may have been located over an anomalously hot region of asthenosphere that now lies beneath the South Pacific Superswell and which has long been an area of oceanic island and seamount production (e.g., McNutt and Fischer, 1987; Staudigel et al., 1991). Today, this region is far from a spreading center and is characterized by several short-lived, poorly understood hotspots that may represent shallow-sourced plumes or entirely nonplume processes (e.g., Janney and Castillo, 1999; Lassiter et al., 2003; Courtillot et al., 2003; Koppers et al., 2003). Triple junction formation in such an area may have promoted excess melting of anomalously fusible mantle and thus plateau formation. The MORB-type isotopic ratios of the few existing Shatsky samples, all of which are from the last stages of volcanism at their sites, are explicable in this context because isotopically normal MORB-source mantle is predicted to underlie the shallow asthenosphere and to well up and gradually replace it as it melts out and advects away from the melting region (e.g., Anderson, 1995).

Finally, could the rise have been formed by meteorite impact, as Rogers (1982) suggested? This hypothesis readily accounts for the MORB-type isotopic ratios of the Shatsky Rise basalts, as removal of the lithosphere by the impacting object would cause massive melting of the underlying mantle, which normally should be MORB-type mantle (in contrast, the Ontong Java basalts all lack a MORB-like isotopic signature). However, this hypothesis requires the coincidence of a large impact (itself a rare event) within 800 km of a preexisting triple junction, and it fails to explain the 30° Pacific-Izanagi ridge reorientation at Anomaly M21 time and the lack of any evidence for the predicted massive destruction and disruption of seafloor over a very large area surrounding the impact site (Mahoney et al., 2005).

In summary, Shatsky Rise clearly formed in association with plate-velocity changes and ridge and triple junction reorganizations during a period when several plateaus appear likely to have formed near ridges in general and triple junctions in particular. Although the plume head hypothesis can explain many features of Shatsky Rise, it requires significant ad hoc coincidences or modifications. Alternatively, the rise may be explained by anomalous volcanism induced by changes in plate boundaries and lithospheric stress over a region of anomalously fusible mantle. Such a hypothesis requires no coincidence of triple junction location and site of plume impingement and can explain the MORB-type signature of late-stage basalts from Shatsky Rise. However, it also relies on unusual circumstances. Indeed, no matter what the hypothesis, unusual circumstances of some sort appear to be required, as illustrated by the dichotomy of Pacific plateau formation in the Late Jurassic and Early Cretaceous versus the paucity of such features since. At present, data for and against each hypothesis are incomplete and largely circumstantial. As a result, the mystery of how Shatsky Rise formed is still an open question.