Age Distribution of Volcanism

An important development in quantifying hotspot dynamics has been the recent improvement in acquisition of accurate and precise age determinations by 40Ar-39Ar incremental heating methods (Koppers et al., 2000). Many of the radiometric ages published into the 1990s, which have been used to infer the time-space distribution of volcanic activity in prominent chains of islands and seamounts (e.g., Duncan and Clague, 1985), were based on total fusions (one-step heating) or incremental heating experiments on minimally treated, altered whole rocks. As Koppers et al. (2000) have shown, acid-leached crystalline groundmass and mineral separates (usually plagioclase feldspar) from seawater-altered tholeiitic and alkalic basalts often produce much more precise and reproducible plateau and isochron ages from incremental heating age spectra. Ages based on these improved methods often show significant differences from published ages from the same samples (Koppers et al., 2004).

Duncan and Keller (2004) report age determinations from deep drilling samples at three of the Emperor Seamounts (Fig. F1): Detroit (Sites 1203 and 1204), Nintoku (Site 1205), and Koko (Site 1206). Combined with new dating results from Suiko Seamount (DSDP Site 433) and dredged rocks from volcanoes near the prominent Hawaiian-Emperor bend (Sharp and Clague, 2002), the overall trend is increasing volcano age from south to north along the Emperor Seamounts, consistent with the hotspot model. However, there are important departures from the earlier modeled simple linear age progression (Clague and Dalrymple, 1987). Specifically, Pacific plate velocity (relative to the Hawaiian hotspot) was variable, from ~5 cm/yr at the northern end of the Emperor seamounts to ~10 cm/yr at the southern end, decreasing abruptly to ~5 cm/yr after the change in orientation at the bend before increasing again to 9 cm/yr along the Hawaiian Ridge (Fig. F9).

The measured paleolatitude change between Detroit and Koko Seamounts amounts to 4–5 cm/yr southward drift of the Hawaiian hotspot with respect to the geomagnetic pole (Tarduno et al., 2003). The relative motion between the Pacific plate and the hotspot during this time was ~5 cm/yr in a direction defined by the Emperor Seamounts. We must conclude that the Pacific plate showed very little northward motion (relative to the geomagnetic pole) between 81 and 61 Ma.

Koppers et al. (2004) report radiometric ages from a reanalysis of 10 samples from the Watts et al. (1988) study of the Louisville Seamounts and two new samples. The pattern of northwestward-increasing ages observed by Watts et al. (1988) was confirmed, but rates of migration of volcanism vary considerably along the chain. Ages for seamounts younger than 47 Ma match closely with those of the Watts et al. (1988) study, but plate over hotspot velocities from volcanoes older than 62 Ma are significantly slower than proposed earlier (Fig. F9). Whereas the distribution of age determinations along the Louisville chain is less dense than that for the Hawaiian-Emperor chain, common intervals of faster and slower plate motion relative to the hotspots are apparent. This implies that, while hotspots may drift, the volcanic lineaments can still provide useful information on Pacific plate motion. In fact, Koppers et al. (2004) combined the Steinberger et al. (2004) predicted motion of the Hawaii and Louisville hotspots with an optimized set of Pacific plate stage rotations to closely match the observed geometry and newest age progressions determined for these volcanic chains (Fig. F10).

Geochemical Variations

Geochemical, petrological, and physical volcanological studies of the Hawaiian-Emperor volcanic lineament provide information about (1) the chemical and mineralogical character of the mantle source for melting at any given location, (2) long-term variability in the flux and composition of mantle sources for volcano construction, and (3) the geometry and dynamics of mantle melting, melt migration, and evolution through the crustal zone. Our models for Hawaiian volcano development are very much biased toward processes and timescales of activity that can be observed at the young, southeast end of the >80-m.y.-old chain, where the islands record only the last 5 m.y. of hotspot activity. It is of great interest to know whether the older volcanoes of the Hawaiian Ridge and Emperor Seamounts were constructed from the same types of magmas, at similar rates of eruption, and in a similar plate tectonic setting as the Hawaiian Islands or whether there are important changes in the nature of the hotspot or interaction of the Pacific lithosphere and the underlying mantle over the whole of Hawaiian-Emperor lineament construction.

Our current model of Hawaiian volcano development is based on field studies, submarine mapping, and sampling of Loihi Seamount and deep drilling at the Hawaii Scientific Drilling Project site on the island of Hawaii (Fig. F11). Essentially, volcanic activity starts with submarine eruptions of low-volume alkalic lavas, moving to mixed alkalic and tholeiitic lavas with increasing rates of eruption (preshield stage), progressing to dominantly tholeiitic lavas at peak eruption rates (main shield stage), changing to dominantly alkalic lavas with dramatically decreasing eruption rate (postshield stage). The timescale for this construction is on the order of 0.5 m.y. After a quiescent period of several million years, extensional tectonics may allow small-volume, small-degree mantle melts to make their way to the surface and erupt as undersaturated posterosional lavas. An addition to this well-known volcano development model is the finding reported by Kerr et al. (2005) that old volcanoes (Detroit Seamount, 76 m.y.) continued to collapse under extension at least through late Miocene time. Volcanic edifices occur on the summit of this coalesced set of shields that, based on the age of sediments they pierce and ash layers recorded in Leg 197 drilling, must have been active throughout much of the Eocene (~52–34 Ma). Hence, rejuvenescent volcanism resulting from gravitational collapse or regional extension may be possible at any of the Hawaiian volcanoes.

The volcanoes that comprise the islands of Hawaii, Maui, Molokai, Lanai, and Kahoolawe record Hawaiian magmatism over the last 2 m.y. The lavas erupted here span the compositional range observed from all Hawaiian volcanoes over the past 5 m.y. (out to Kauai) and can be related to source end-member compositions for melting (Gaffney et al., 2005). Major and trace element and isotope (Sr, Nd, Pb, Hf, and Os) compositions of shield-stage lavas from these volcanoes are consistent with ancient, recycled oceanic crust and primitive plume material in the mantle source (so-called, Kea-, Koolau- and Loihi-type magmas) (Lassiter and Hauri, 1998; Blichert-Toft et al., 1999; Ren et al., 2005). Gaffney et al. (2005) examine trace element models that include melting of recycled ocean crust (eclogite pods and pelagic sediment) mixed within primitive plume mantle (Koolau type) or melting of recycled ocean crust that has been first fertilized by reaction with partial melts from associated crustal gabbro, without mixing with primitive plume mantle (Kea type). Chen and Frey (1985) propose that Hawaiian postshield-stage lavas were formed by melts from the plume mixing with melts derived from the oceanic lithosphere, which is the wallrock of the melt conduits to the volcanoes. In this view, the contribution from the plume decreases with time (up section) as the plate moves away from the hotspot.

The time and space distribution of the volcano compositions has led to the notion that the Hawaiian plume is concentrically zoned, with Koolau- and Loihi-type volcanoes over the core of the plume and Kea-type volcanoes over the edge of the plume (Hauri, 1996; Lassiter and Hauri, 1998; DePaolo et al., 2001) (Fig. F1). Compositional heterogeneity in the Hawaiian hotspot exists, then, at many scales, yet there is evidence that the full range and pattern of variability may persist for very long periods (>5 m.y.). Deep sampling of the Emperor Seamounts provides the means to examine the range and scale of compositional variability of the Hawaiian hotspot over much longer periods (>80 m.y.) and through changes in plate tectonic setting (near-spreading ridge to midplate). The sections of lava flows sampled during Leg 197 drilling are also significantly longer than earlier sampling by dredging or drilling (with the exception of DSDP drilling at Suiko Seamount) and provide considerable detail about short-term changes in lava composition and volcanic processes. Observations of lava flow thickness, vesicularity, crystallinity, and morphology, together with analysis of volcanic sediment, have provided a picture of eruptions in subaerial to shallow-water conditions at Detroit and Koko Seamounts and waning subaerial activity at Nintoku Seamount (Fig. F12). Combining petrological and geochemical data with such physical volcanological observations leads to interpretation of volcanic stages within the Hawaiian volcano development model for cored samples from the Emperor Seamounts.

Major and trace element compositions for basaltic rocks cored during Leg 197 conform closely with compositions known from the Hawaiian volcanoes (Fig. F13). Detroit seamount lavas range from alkalic to tholeiitic compositions, consistent with preshield to main-shield stages; Nintoku Seamount lavas are dominantly alkalic compositions, consistent with waning postshield stage; Koko Seamount lavas are dominantly tholeiitic compositions, consistent with main-shield stage eruptions. Schafer et al. (2005) report major and trace element and Sr and Nd isotopic compositions for alkalic, postshield stage lavas cored at Site 1205, Nintoku Seamount (~56 m.y.). These are similar to those of young, postshield stage lavas from the Hawaiian Islands. A modified version of the Chen and Frey (1985) model for postshield stage lavas, in which the Pacific Ocean lithosphere, previously metasomatized by small-degree upper mantle melts, is mixed with melts from the Hawaiian plume mantle appears to explain the trace element and isotopic compositions of the Nintoku Seamount lavas without the need to invoke unreasonably small degrees of mantle melting.

Huang et al. (2005) report major and trace element and isotopic compositions for lava flows from Detroit Seamount (76-81 m.y.), one of the oldest existing Hawaiian volcanic systems, now cored at six locations. Tholeiitic basalts on the eastern flank and alkalic basalts on the summit plateau have been identified. At summit Site 1203 subaerial pahoehoe alkalic lavas occur beneath submarine-erupted tholeiitic lavas, indicating a possible preshield–shield stage transition. The surprising upward change from subaerial to submarine eruption conditions implies rapid subsidence of the volcano, consistent with an inferred near-spreading ridge setting of the seamount at ~80 m.y. Such an interpretation is supported by the lack of age difference between the youngest lavas and oldest sediments at this site. A shallower depth of melt segregation for Detroit Seamount lavas relative to Hawaiian Island lavas is also consistent with construction on thinner oceanic lithosphere (near ridge axis). Huang et al. (2005) confirm the previously reported decrease in 87Sr/86Sr for shield lavas northward along the Emperor Seamounts (Keller et al., 2000). Indeed, tholeiitic lavas from Detroit Seamount have the most extreme isotopic composition within the Emperor seamounts, and overlap the 87Sr/86Sr–143Nd/144Nd field for Pacific mid-ocean ridge basalt (MORB). However, in contrast to Pacific MORB, Detroit Seamount lavas trend to extremely unradiogenic Pb isotopic compositions (Fig. F14) and contain relatively high Ba/Th that is characteristic of Hawaiian Island lavas. Huang et al. (2005) conclude that Detroit Seamount lavas sample an intrinsic component present in the Hawaiian hotspot for at least 81 m.y.

Frey et al. (2005) focus their attention on the Hf isotopic composition of the geochemically depleted component in Hawaiian-Emperor lineament lavas. As noted earlier, the Chen and Frey (1985) model recognizes the involvement of old oceanic lithosphere in producing characteristics of the postshield stage lavas in Hawaiian volcanoes. The most depleted shield-stage lavas at Detroit Seamount have many similarities to Pacific MORB, but important differences exist, namely the trends to relatively unradiogenic Pb isotopic compositions (Huang et al., 2005) and steep trends of 176Hf/177Hf vs. 143Nd/144Nd. Frey et al. (2005) propose that a depleted component, intrinsic to the hotspot, has contributed to these shield-stage lavas throughout the Hawaiian-Emperor lineament and is a persistent feature of the long-lived plume (Fig. F15).

Keller et al. (2004) measured He isotopic compositions in olivines from picrites and basalts from the Hawaiian-Emperor lineament and report that high 3He/4He is a persistent characteristic of the hotspot for its entire history. Picrites erupted at 76 m.y. at Detroit Seamount have 3He/4He ratios at the lower end of the range for the Hawaiian Islands but still above the range of modern MORB. During this time volcanism occurred on thin oceanic lithosphere close to a spreading ridge and produced lava compositions that, except for the elevated He and low Pb isotopic compositions, would be indistinguishable from Pacific MORB. The 3He/4He of Hawaiian-Emperor volcano lavas increased dramatically during the Late Cretaceous as the distance between the spreading ridge and the Hawaiian hotspot increased (Fig. F15). The 3He/4He isotopic composition of shield lavas stabilized at high values by 61 m.y. (Suiko Seamount), while trace element and Sr isotopic compositions continued to change before arriving at ~45 m.y. at values similar to those of the Hawaiian Islands. The contribution of the plume component to Hawaiian-Emperor chain volcanoes is therefore more sensitively recorded by 3He/4He than by other geochemical indicators through the Emperor Seamount phase of hotspot activity. A poor correlation between 3He/4He and calculated plume flux based on bathymetry and residual gravity anomalies (Van Ark and Lin, 2004) suggests that variations in delivery of primitive plume mantle does not control the total flux of hotspot melting.