Introduction and background

Understanding the nature of mantle plumes is a critical goal of modern Earth sciences. The extent to which hotspots conform to the Wilson–Morgan fixed plume hypothesis (Wilson, 1963; Morgan, 1971) fundamentally constrains the assumptions used in models of mantle convection. To date, studies of the Hawaiian–Emperor Seamount Trail have dominated our thinking about hotspot volcanism, and as a consequence, models for the construction and evolution of intraplate volcanoes, plate motion, and hotspot motion are strongly biased toward the Hawaiian hotspot. Without comparable data from any other important hotspot trail, many key questions remain unanswered. The Louisville hotspot trail (Figs. F1, F2) is one of only three primary hotspots (together with Hawaii and Easter) in the Pacific (Courtillot et al., 2003; Koppers and Watts, 2010), and it has great potential for providing answers to these questions. New results from IODP drilling of the Louisville Seamount Trail during Expedition 330, together with existing data and future drilling of other hotspot trails in the Atlantic and Indian oceans, will provide the best available opportunity to assess the importance of the motion between hotspots (or groups of hotspots) and true polar wander. These data, in turn, will provide valuable information about Earth's convection regime and will allow for a crucial calibration of current mantle flow models and global plate circuit reconstructions.

Hotspots, as well as the (deep) mantle plumes presumed to be their underlying cause (e.g., Morgan, 1971), are essential features in geodynamic models of Earth's mantle. One of the attributes frequently assigned to mantle plumes is their fixity in the mantle. This fixity contrasts distinctly with the motion of the overlying plates, which move at speeds up to 100 mm/y. However, plume theory does not demand fixity (e.g., Steinberger and O'Connell, 1998; Koppers et al., 2001), and paleomagnetic evidence collected during Ocean Drilling Program (ODP) Leg 197 from four Emperor seamounts (Detroit, Suiko, Nintoku, and Koko) indicates that the Hawaiian hotspot has moved at a speed similar to that of plate motion for tens of millions of years (Kono, 1980; Petronotis et al., 1994; Tarduno et al., 2003, 2009; Duncan et al., 2006). Three-dimensional mantle-convection computations confirm this notion by generating mantle plumes with long, narrow thermal conduits that may migrate at speeds of 10 mm/y and higher. From these simulations it follows that plume migration primarily depends on the assumed viscosity contrast between the lower and upper mantle and the configuration of the subducting plates (Lowman et al., 2004).

These observations raise critical questions. Is the rapid motion documented at the Hawaiian hotspot an isolated event, or does motion occur at other hotspots as well? If motion is not an isolated event, do other hotspots move in a sufficiently coherent fashion that subsets can be used as a moving reference frame for reconstructing past plate motion? To address these questions, we must distinguish between the following geodynamic end-member models:

  1. The primary hotspots in the Pacific (Hawaii, Louisville, and Easter) move coherently over geologic time and thus show minimal interhotspot motion, as argued by Courtillot et al. (2003) and Wessel and Kroenke (1997).

  2. The primary hotspots have very different motions that result in increased interhotspot motion, as predicted by mantle flow model calculations (Steinberger et al., 2004; Koppers et al., 2004; Steinberger, 2002; Steinberger and Antretter, 2006; Steinberger and Calderwood, 2006).

The first model predicts the motion of the Louisville hotspot to be equivalent to the 15° southern motion documented for the Hawaiian hotspot between 80 and 50 Ma. The second model predicts an essentially eastward motion of the Louisville hotspot over the last 120 m.y., with a maximum shift in paleolatitude not exceeding 2°–6° between 80–50 Ma and the present day, depending on the various assumptions used in the applied mantle flow models (Fig. F3). Both models will be tested by drilling the Louisville Seamount Trail during Expedition 330 and by accurately determining paleolatitudes (from detailed paleomagnetic measurements on individual lava flows) and 40Ar/39Ar age dates for seamounts between 80 and 50 Ma. For this purpose, Expedition 330 will provide a direct comparative test that mirrors Leg 197 drilling in the Emperor seamounts as closely as possible (Tarduno et al., 2003; Duncan and Keller, 2004; Duncan et al., 2006, 2007) by targeting guyots equivalent in age to the Detroit (76–81 Ma), Suiko (61 Ma), Nintoku (56 Ma), and Koko (49 Ma) seamounts (Fig. F2).

Although determining paleolatitudes in the context of a high-resolution 40Ar/39Ar age framework is the main objective of Expedition 330, we also seek to understand the eruptive cycle and geochemical evolution of typical Louisville volcanoes. The Hawaiian and Louisville hotspots have been labeled as "primary" hotspots in the Pacific Ocean based on the presence of obvious linear age progressions, long-lived and continuous volcanism, large buoyancy fluxes, and high (in the case of Hawaii) 3He/4He ratios (Courtillot et al., 2003; Koppers et al., 2003). Such hotspots are theorized to represent plumes rising from deep in the mantle, possibly from near the core/mantle boundary (Clouard and Bonneville, 2001; Davaille et al., 2002; Courtillot et al., 2003). Unlike earlier studies that postulated that many (or even all) hotspots represent plumes originating from the core/mantle boundary, Courtillot et al. (2003) have argued that most hotspots arise from relatively shallow levels and only a small number of primary plumes ascend from the core/mantle boundary. In the case of the poorly studied Louisville Seamount Trail, only its remarkably linear age progression (Watts et al., 1988) and its long-lived ~80 m.y. volcanic record have been used to label it a primary hotspot; no 3He/4He analyses and only sparse other geochemical data are available at present.

Nonetheless, some marked differences in geochemistry and volcanic evolution are apparent between the primary Pacific hotspots. For example, the almost exclusive recovery of alkali basalts in the Louisville Seamount Trail (Hawkins et al., 1987; Vanderkluysen et al., 2007) raises the question of whether Louisville volcanoes have an alkalic shield-building phase instead of the tholeiitic shield-building phase that is a trademark of Hawaiian volcanoes. One possibility is that the shield stage of a Louisville volcano reflects systematically lesser amounts of partial melting than in Hawaiian volcanoes because of deeper melting under a uniformly thicker lithosphere over 80 m.y. for the entire Louisville Seamount Trail. In addition, isotopic and trace element data from Louisville suggest a long-lived and remarkably homogeneous mantle source (Cheng et al., 1987; Hawkins et al., 1987; Vanderkluysen et al., 2007) that apparently does not include any of the depleted source material that typically produces mid-ocean-ridge basalt (MORB). If Louisville volcanoes prove to have entirely alkalic shield-building phases that are isotopically homogeneous over 80 m.y., this will have major implications for how we think volcanism works for the Louisville Seamount Trail and intraplate volcanism in general.

In the following sections, we review current thinking on mantle geodynamics and hotspot motion, the unique geochemical evolution of the Louisville volcanoes and their mantle sources, and why the Louisville Seamount Trail is key in meeting our science objectives. We then summarize previous research and site surveys on the Louisville Seamount Trail. Finally, we explain how our hypotheses can be tested by coring four seamounts using a riserless drilling program for the Louisville Seamount Trail.

Mantle geodynamics and hotspot motion

Recently, it has been proposed that plume conduits may become strongly tilted at times because of large-scale mantle flow (e.g., Steinberger and O'Connell, 1998), which may possibly explain the fast (~40 mm/y) southward hotspot motion observed at the Hawaiian hotspot between 80 and 50 Ma (Tarduno et al., 2003). Such strong tilts may occur if the conduit is affected by a lower mantle return flow between cold downwellings associated with subduction and large-scale upwellings in the neighborhood of "superplumes." The capturing, bending, and releasing of the Hawaiian mantle plume by an ancient ridge system may also explain these observations (Tarduno et al., 2009). All of these ingredients are present at the Hawaiian hotspot, including a zone of past subduction to the north, a large-scale upwelling related to the Superswell to the south (Tarduno et al., 2003), and the waning of the Kula–Pacific ridge system north of the hotspot, which would allow a possibly captured Hawaiian plume to quickly return to its original straight position in a dominant southward flow (Tarduno et al., 2009). However, because the Louisville hotspot is located south of the Superswell, the closest subduction system has always been located west of the hotspot, and no spreading center is in close proximity or located to the north of it for most of its geological history, the Hawaiian pattern taken at face value is not compatible with a similar rapid southward motion of the Louisville hotspot.

Recent modeling by Steinberger et al. (2004) shows the expected results of this configuration, whereby the Louisville hotspot is predicted to have moved in an easterly direction between 130 and 60 Ma and only ~2.5° southward since 60 Ma (Fig. F3A). In these models, a large-scale mantle flow field is first calculated from mantle-density heterogeneities (as derived from seismic S-wave speed anomalies), simultaneously applied with a radial mantle rheology structure (with the lower mantle assumed to be more viscous) and tectonic plate motions that both serve as boundary conditions (Steinberger and O'Connell, 2000; Steinberger and Calderwood, 2006). Within this modeled mantle flow field, an initially vertical plume conduit is inserted that gets advected over time, resulting in sometimes strong tilting of mantle plumes and drifting of hotspots (Steinberger and O'Connell, 2000; Steinberger and Antretter, 2006). Advection dominates the motion of a plume, typically in the lower mantle where it rises relatively slowly in comparison to the overall mantle flow field. As a result, hotspot motion in these models appears in many cases (including Louisville) to be similar to the horizontal flow components in the mid-mantle, in which a transition from advection-dominated motion to more vertical motion dominated by the buoyant rising of mantle plume materials occurs, regardless of whether these mantle plumes originate at the core/mantle or 670 km boundary layers (Steinberger, 2000). In spite of the large uncertainties in data and the assumptions on which these mantle flow models are built, these models provide an excellent basis for placing geologic data from Louisville and other seamount trails into a more complex geodynamic context.

For example, using new high-resolution 40Ar/39Ar age data for the sample collection used in Watts et al.'s (1988) study, Koppers et al. (2004) found that the age progression for the Louisville Seamount Trail is not linear after all (blue squares in Fig. F4A). By using the modeling approach of Steinberger et al. (2004) and including the primary Pacific hotspots only (Hawaii, Louisville, and Easter), revised mantle flow models can fit this updated nonlinear age trend for the Louisville Seamount Trail by allowing for a slowdown and a different rotation of the Pacific plate before 62 Ma and by decreasing the initiation age of the Louisville hotspot from 120 to 90 Ma. These revised models show an eastward motion of ~5° for the Louisville hotspot between 90 and 30 Ma, which is very different from the ~15° southward motion of the Hawaiian hotspot during the same time interval. The primarily longitudinal Louisville hotspot motion is followed by only a minor ~2° latitudinal shift to the south over the last 30 m.y. (Fig. F3B; Model 5 in Fig. F3C).

Model predictions, however, vary largely depending on the assumptions used, including plume initiation age, root depth, viscosity structure, plume buoyancy and rising speed, plate motion history, and mantle viscosity. Antretter et al. (2004) and Steinberger and Antretter (2006) considered the possible effects of these assumptions for the Louisville hotspot in more detail and predicted southward paleolatitude shifts between almost 0° and ~8° over the last 80 m.y. However, the majority of their model runs (see six representative examples in Fig. F3C) show limited latitudinal motion for Louisville that is significantly less than that observed for Hawaii. The models that do show rather fast Louisville hotspot motion predict a more eastward motion, away from the subduction zone and toward the spreading ridge. Drilling the Louisville Seamount Trail will thus provide an essential calibration of these numerical mantle flow models.

Preliminary 40Ar/39Ar age data from the SO167 and AMAT02RR site surveys underline the nonlinear character of the Louisville age progression (Fig. F4A). These new data point toward slower motion of the Pacific plate relative to the Louisville hotspot prior to 62 Ma rather than faster relative motion, as would be expected with a substantially southward-moving hotspot, thus indirectly supporting the above predictions for a minor paleolatitude shift. Results from the latest incremental heating experiments (Fig. F4B) plot above the original 64 mm/y linear age progression suggested by Watts et al. (1988) and have 2σ uncertainties as low as 0.2–0.5 Ma. Deviations from previously reported ages become noticeable at ~35 Ma but are most significant toward the oldest end of the trail, where some guyots are dated to be ~15 m.y. older than the ages predicted by the 64 mm/y age progression. For now, pending new results from Expedition 330, we presume that the age progression is best approximated by the purple line in Figure F4A, which simply envelops the oldest ages and follows the model of Koppers et al. (2004). Additional preliminary age information from SO167 dredges (averaged from the analyses of multiple plagioclase mineral separates) highlights the complex age distribution in the older portion of the chain, while adding to and confirming the monotonous, almost linear, age progression between 25 and 60 Ma.

Shifts in paleolatitude of the Louisville hotspot can also be predicted by transferring plate motion from Indo-Atlantic hotspots to the Pacific using global plate circuits. In this approach (Fig. F5), the plate circuit may go through East and West Antarctica (EANT-WANT) (Cande et al., 1995) or, alternatively, through the Lord Howe Rise (LHR) (Steinberger et al., 2004). Also critical is the current location of the Louisville hotspot, which is still a matter of debate because of its faint expression at the younger end of this seamount trail (Lonsdale, 1988; Raymond et al., 2000; Wessel and Kroenke, 1997). All plate reconstructions (using different combinations of plate circuits and present-day hotspot locations) yield predictions that are significantly different from the position of the Louisville chain prior to 45 Ma (Fig. F5). As with the mantle flow models, a large longitudinal shift is apparent at 78 Ma when the oldest seamounts in the trail are compared to the reconstructed position for Chron C33, which plots markedly west of the Louisville Seamount Trail. However, the predicted potential southward shift in these plate tectonic reconstructions seems less pronounced for the oldest seamounts (3.8° and 4.0°, depending on the plate circuit used) in the Louisville Seamount Trail, is largest (8.9° and 9.8°) for seamounts ~53 m.y. old (which actually indicates a possible 4°–5° northward motion of the plume between 80 and 50 Ma), and is nonexistent for seamounts younger than ~45 m.y. old.

The above geodynamic and plate tectonic models thus provide us with different predictions for the latitudinal history of the Louisville hotspot. These models will be groundtruthed during Expedition 330, proving one of the following:

  1. The Louisville hotspot shows an increasing southward motion (as much as 15°) that is comparable to the motion of the Hawaiian hotspot between 80 and 50 Ma, providing evidence for a common motion of the mantle underlying the Pacific plate with respect to Earth's spin axis.

  2. The Louisville hotspot shows an insignificant cumulative latitudinal shift (<2°–6°), supporting the mantle flow models of Steinberger et al. (2004) and Steinberger and Antretter (2006) that predict minimal latitudinal motion and more pronounced easterly longitudinal motion.

  3. The Louisville hotspot shows a variable (but significant) latitudinal motion between 80 and 50 Ma in combination with a large longitudinal shift, reconciling the observations made with global plate circuit models.

By comparing paleolatitudes derived from paleomagnetic measurements on cored basalt flows and high-precision 40Ar/39Ar ages for the Hawaiian and Louisville seamount trails, Expedition 330 drilling results will offer strong constraints for one of these possibilities. Finding a large latitudinal shift between 80 and 50 Ma would clearly indicate that current assumptions made in mantle flow models are wrong. Finding no appreciable shift, on the contrary, would indicate significant interhotspot motion between the Hawaiian and Louisville hotspots and a stronger local control on the mantle flow regime.

In addition, paleolatitude measurements of the Louisville Seamount Trail can be used to compare interocean motion between hotspots in the Pacific, Indian, and Atlantic oceans. Previous studies reported discrepancies in motion as high as 20° between hotspot groups in the Pacific and Atlantic–Indian oceans (Cande et al., 1995; DiVenere and Kent, 1999; Raymond et al., 2000). These observations have led to models of large-scale motion of the mantle underneath each ocean and to explanations involving true polar wander, which is defined as a coherent shift of the entire mantle relative to Earth's spin axis (Goldreich and Toomre, 1969; Gordon, 1987; Besse and Courtillot, 1991; Torsvik et al., 2002; Argus and Gross, 2004). Drilling the Louisville Seamount Trail will be pivotal in establishing the necessary global paleolatitude and 40Ar/39Ar age databases that, in turn, will help us to evaluate the above scenarios. In fact, Expedition 330 will collect paleolatitude data for a period of time similar to that sampled in the Emperor Seamount Trail during Leg 197 (Kono, 1980; Tarduno et al., 2003), and in the future, similar data may be collected from ocean drilling expeditions to the Chagos-Laccadive and Ninetyeast ridges in the Indian Ocean and the Walvis Ridge in the Atlantic Ocean. In the end, future ocean drilling may provide us with a state-of-the-art paleolatitude database covering five major hotspot trails in three oceans, all between 80 and 50 Ma in age. At that point, it will be clear whether the observed paleolatitude shifts in hotspots are best explained by coherent motion of all hotspots relative to the spin axis, by coherent motion of hotspot groups within each ocean domain but with relative motion between these groups, or by incoherent motion of all individual hotspots.

Geochemical evolution of the Louisville hotspot

The construction and geochemical history of an intraplate seamount is often envisioned to resemble that of a typical Hawaiian hotspot volcano (Clague and Dalrymple, 1988; Staudigel and Clague, 2010). There is, however, little empirical evidence for a similar evolutionary sequence in the Louisville Seamount Trail. Almost all igneous rocks dredged from the Louisville Seamount Trail are alkalic basalts, basanites, or tephrites containing normative nepheline (Fig. F6A) (Hawkins et al., 1987; Vanderkluysen et al., 2007). In addition, isotopic and trace element data from this seamount trail suggest a long-lived and remarkably homogeneous mantle source equivalent to the proposed "common" components FOZO (Focal Zone) or C (Fig. F6B) (Cheng et al., 1987; Hawkins et al., 1987; Vanderkluysen et al., 2007). The minor variations in major and trace elements appear to be controlled mostly by variable extents of melting and fractional crystallization, with little influence from mantle-source heterogeneities.

This raises several first-order questions that can be addressed by geochemical studies of samples cored during Expedition 330. Do Louisville volcanoes evolve through geochemically distinct shield, postshield, and posterosional (or rejuvenated) stages similar to those of Hawaiian volcanoes? If so, is the volumetrically dominant shield stage characterized by eruption of tholeiites or, as suggested by presently available samples, alkalic lavas? Tholeiitic basalts generally represent greater amounts of partial melting than more alkalic lavas do. One possibility is that the typical shield stage of a Louisville volcano reflects systematically less partial melting than that of a Hawaiian volcano. Alternatively, dredging may have sampled only later stage lavas that in many locations may cover shield-stage flows, which would explain why the least alkalic lavas obtained by dredging are from Osbourn Guyot near the Kermadec Trench, where extensive faulting may have exposed older shield-building lavas (Hawkins et al., 1987). Although drill core penetration of an entire Louisville seamount would be necessary to conclusively demonstrate that no tholeiitic shield-building occurs in the Louisville Seamount Trail, the proposed basement penetration of ~350 m during Expedition 330 will provide the best approach for sampling the waning part of the shield-building stage (whether tholeiitic or alkalic), while maximizing the number of flows to be used for determining paleolatitudes.

Hawaiian shield-stage lava flows also possess a wide range of isotopic and incompatible element compositions, and the range is even greater when data for postshield and posterosional lavas are included. In contrast, isotope and incompatible element ratios (e.g., Zr/Y and Nb/Y) for the Louisville Seamount Trail are surprisingly homogeneous (Fig. F6B). These data indicate that the mantle (plume) source of Louisville hotspot lavas has been unusually homogeneous for a very long time. This raises important questions, such as how geochemically variable has the Louisville mantle source been over the last 80 m.y., and why has it stayed homogeneous? An important difference between the Louisville and Hawaiian–Emperor seamount trails may be the age of the underlying seafloor at the time of volcano formation. It has been assumed that all Louisville volcanoes have generally been erupting onto seafloor that is ~40–50 m.y. old (Lonsdale, 1988; Watts et al., 1988). This includes the older (northwestern) portion of the Louisville Seamount Trail that formed close to the Osbourn Trough paleo-spreading center, which ceased activity between 115 and 121 Ma (Downey et al., 2007). This notion is confirmed by recent three-dimensional flexural studies of the Louisville Seamount Trail (Lyons et al., 2000), which also suggest crustal ages of 40–50 Ma and perhaps older. However, a new study based on seismic refraction provides the first ever detailed two-dimensional tomographic image of the internal structure of the oceanic lithosphere beneath one of the older Louisville seamounts (Fig. F7). This image shows the internal intrusive structure of this Louisville seamount, with intrusions visible as shallow as 1.5 km beneath its top, but it also shows a downward-flexed Mohorovicic seismic discontinuity (MOHO) by ~2.5 km that can be explained only by an elastic plate model whereby this particular seamount was emplaced upon oceanic lithosphere that is only ~10 m.y. old (Contreras-Reyes et al., 2010). Similar ~10 m.y. age differences between seafloor and volcanic eruption are observed for the oldest seamounts in the Hawaiian–Emperor Seamount Trail, yet the age differences are much larger (>100 m.y.) for the younger Hawaiian volcanoes (Keller et al., 2000; Caplan-Auerbach et al., 2000). Overall, however, lithospheric thickness seems to have been less variable for the Louisville hotspot than for Hawaii, particularly because oceanic lithosphere tends to thicken more slowly after ~40 m.y. (e.g., Stein and Stein, 1993). Lithospheric thickness is a key control on partial melting because it determines the minimum depth of the top of the melting column and limits the extent of decompressional melting that occurs in the upwelling mantle (e.g., McKenzie and Bickle, 1988). Other things being equal, greater amounts of partial melting occur under thin lithosphere, and in a mantle containing isotopically and chemically distinct lithologies, different mantle components likely begin to melt at different depths (Sun and Hanson, 1976; Ellam, 1992; Phipps Morgan and Morgan, 1999; Hoernle et al., 2000; Niu et al., 2002; Ito and Mahoney, 2005; Devey et al., 2003). For the Emperor seamounts, much of the observed isotopic and chemical variation may be related to changing proximity to a spreading center and related changes in lithospheric thickness (Keller et al., 2004; Regelous et al., 2003). In contrast, for most of the Louisville Seamount Trail, the limited isotopic variation likely reflects a remarkably homogeneous plume source or relatively uniform melting conditions over 80 m.y.

Comparable to other primary hotspots (e.g., Hawaii), potential temperatures of the Louisville mantle plume sources are expected to be 100°– 300°C higher than the 1350° ± 50°C temperature of an upper mantle MORB source (Putirka, 2008; Courtier et al., 2007). Olivine-phyric rocks will likely be encountered within 350 m of drilling into the Louisville seamounts and can be used to determine Mg-Fe compositions of olivine phenocrysts and melt-inclusions therein. In turn, these compositions will yield information about source temperatures by relating the Mg/Fe ratio of olivines directly to that of the liquid from which they crystallized (e.g., Putirka et al., 2007). The challenge here is to determine the most magnesian-rich olivines that come closest to the parental magma compositions, a task that may be more complicated for the Louisville seamounts because all samples studied so far are relatively evolved and typically alkalic. Picritic basalts tend to be found deeper in the stratigraphy of a volcanic pile, providing another key reason for IODP drilling in the Louisville Seamount Trail. For example, during Leg 197, olivine-rich basalts were recovered only after drilling >250 m (Duncan et al., 2006). Melt inclusions also provide key insights into the "true" (lack of) heterogeneity in the mantle source from which the Louisville magmas have been generated.

Courtillot et al. (2003) have argued that most hotspots arise from relatively shallow levels and that no more than three primary plumes (Hawaii, Easter, and Louisville) ascend from the core/mantle boundary in the Pacific Basin. These authors suggest several criteria by which primary plumes may be assessed, the chief geochemical criterion being high 3He/4He ratios in hotspot lavas. Although agreement is not universal (e.g., Meibom et al., 2003), high 3He/4He is considered by the great majority of researchers to be a sign of deep-mantle origin (e.g., Allègre et al., 1983; O'Nions, 1987; Farley and Neroda, 1998). Hawaiian basalts, for example, have 3He/4He values as high as 35 RA (where RA is the atmospheric 3He/4He ratio measured today), and even higher values have been reported for samples from Iceland. In comparison, MORB typically has values of only 7–10 RA (e.g., Graham, 2002). No He isotope data have been published for the Louisville Seamount Trail. Another geochemical indicator of a deep-mantle origin is high 186Os/188Os, which is interpreted by some to signify Os derived from the outer core. Only a few studies of Os isotopes in oceanic hotspot lavas have been performed, but anomalously high 186Os/188Os and 187Os/188Os ratios have been discovered in at least some of the primary hotspots, such as Hawaii (e.g., Brandon et al., 1999). Although the interpretation of elevated Os isotope ratios is disputable (e.g., Smith, 2003), combined studies of Os and He isotopes have the highest potential to reveal a deep-mantle signature in oceanic lavas. Drilling is critical for obtaining samples with relatively unaltered olivine crystals and oxide minerals that can be used to successfully test these geochemical criteria and prove that, indeed, the Louisville hotspot is a primary hotspot. If these tests indicate that Louisville does not have a deep (lower) mantle origin, this outcome will place limits on the mantle flow models by forcing a shallower root for its mantle plume. Although geodynamic modeling likely will be unable to resolve the depth of origin between primary and secondary hotspots, geochemical results may be able to, particularly via characteristic high 3He/4He and 186Os/188Os deep-mantle plume-source ratios.

Finally, the Ontong Java Plateau (OJP) has been proposed to be a large igneous province (LIP) that is the product of the Louisville hotspot's initial plume-head phase (e.g., Richards and Griffiths, 1989; Mahoney and Spencer, 1991; Tarduno et al., 1991), even though this would require a significant amount of true polar wander and a 6° southward hotspot migration (Antretter et al., 2004). On the other hand, existing isotopic data for Louisville dredge samples (Cheng et al., 1987; Vanderkluysen et al., 2007) offer no support for such a connection (Mahoney et al., 1993; Tejada et al., 1996), and recent paleolatitude data for basalts recovered from the plateau during ODP Leg 192 differ significantly from the present-day latitude of the Louisville hotspot (Riisager et al., 2003). For example, the OJP and Louisville samples have similar age-corrected Nd and Sr isotope values, but the Louisville lavas have significantly higher Pb isotope ratios than the OJP basalts (Fig. F6B). This difference is more than can be accounted for by ingrowth of radiogenic Pb in the mantle source between 120 and 80 Ma (Vanderkluysen et al., 2007). However, as noted above, the existing Louisville dredge samples may come from late-stage flows that are not representative of the bulk of these seamount edifices. The ~350 m of basement penetration at the four Expedition 330 drill sites will provide a much more rigorous test of any geochemical connection between the OJP and the Louisville Seamount Trail.

Hydrothermal and seawater alteration

Each drilling project has its own scientific and technical challenges. In drilling the Louisville Seamount Trail we will face challenges related to hydrothermal and seawater alteration present in the cored seamount basalts. Most of the dredged samples from the Louisville Seamount Trail that are now available to us are highly to completely altered by seawater interaction. Although drilling will provide us with fresher basaltic material, alteration remains problematic and thus requires special analytic attention in order to maximize the amount of high-quality data on rock ages and original (erupted) compositions. Holocrystalline groundmass samples that have been carefully handpicked and acid leached to remove alteration have provided ages consistent with 40Ar/39Ar ages of co-magmatic minerals and can be interpreted as eruption ages (Koppers et al., 2000, 2004). Preliminary data on basalts from the AMAT02RR site survey emphasize the suitability of this technique for Louisville (Fig. F4B), with excellent reproducibilities between groundmass and plagioclase separates. This makes the 40Ar/39Ar groundmass dating technique perfectly suited to date and resolve the duration of multiple lava units in a single drill site, particularly if samples are enriched in potassium and are taken from relatively unaltered drill core material. Many studies also have demonstrated that altered rocks can be effectively used for determining a hotspot's geochemical characteristics, particularly data for elements (lanthanides, Nb, Ta, Zr, Ti, Fe, and Al) and isotopic systems (Sm–Nd and Lu–Hf) that are resistant to seawater alteration. In addition, useful data can be obtained for the more sensitive Sr and Pb isotopic systems by applying mineral separation and acid-leaching of rocks to remove secondary minerals (Cheng et al., 1987; Mahoney et al., 1998; Koppers et al., 2003; Regelous et al., 2003). Finally, magmatic compositions can be inferred from the microanalysis of major and trace elements in typically (relatively) unaltered portions of various phenocrystic phases and melt inclusions.

Paleosecular variation

Robust paleolatitude estimates require a sufficient number of flows be sampled to average out the secular variations of the geomagnetic field, which shows small deviations from a geomagnetic axial dipole on timescales ranging from years to millennia. The proposed penetration depth of at least 350 m during Expedition 330 is therefore primarily dictated by the necessity to obtain accurate paleolatitude data. Likely errors in paleolatitude estimates at a ~50° southern latitude appropriate for the Louisville hotspot can be illustrated via Monte Carlo simulations of field directions, as drawn from two recent paleosecular variation models based on a global database of directions from recent lava flows (0–5 Ma). These simulations suggest that we must recover >42 independent flow units at each site to achieve a nominal 2σ error of 4° in the paleolatitude estimates (Fig. F8) and only 25–30 flow units if we aim more conservatively for 5° errors. These results can also be compared to a compilation of Deep Sea Drilling Program (DSDP)/ODP drilling statistics from other expeditions to seamount trails and LIPs (Fig. F8B) that on average recovered >20 flow groups from ~200–300 m of volcanic basement coring. Measured uncertainties (Fig. F8C) in these DSDP/ODP cores are generally compatible with the Monte Carlo simulations but scatter significantly between 3° and 7° uncertainties for paleolatitude estimates based on five or more flow units. Because mantle flow models and global plate circuit reconstructions predict small paleolatitude shifts for the Louisville hotspot, we require a basement penetration of at least 350 m to achieve a paleolatitude uncertainty better than 5°.