Abstract

Recent ocean-bottom geophysical surveys, dredging, and dives, which complement surface data and scientific drilling at the Island of Hawaii, document that evolutionary stages during volcano growth are more diverse than previously described. Based on combining available composition, isotopic age, and geologically constrained volume data for each of the component volcanoes, this overview provides the first integrated models for overall growth of any Hawaiian island. In contrast to prior morphologic models for volcano evolution (preshield, shield, postshield), growth increasingly can be tracked by age and volume (magma supply), defining waxing alkalic, sustained tholeiitic, and waning alkalic stages. Data and estimates for individual volcanoes are used to model changing magma supply during successive compositional stages, to place limits on volcano life spans, and to interpret composite assembly of the island. Volcano volumes vary by an order of magnitude; peak magma supply also varies sizably among edifices but is challenging to quantify because of uncertainty about volcano life spans. Three alternative models are compared: (1) near-constant volcano propagation, (2) near-equal volcano durations, (3) high peak-tholeiite magma supply. These models define inconsistencies with prior geodynamic models, indicate that composite growth at Hawaii peaked ca. 800–400 ka, and demonstrate a lower current rate. Recent age determinations for Kilauea and Kohala define a volcano propagation rate of 8.6 cm/yr that yields plausible inception ages for other volcanoes of the Kea trend. In contrast, a similar propagation rate for the less-constrained Loa trend would require inception of Loihi Seamount in the future and ages that become implausibly large for the older volcanoes. An alternative rate of 10.6 cm/yr for Loa-trend volcanoes is reasonably consistent with ages and volcano spacing, but younger Loa volcanoes are offset from the Kea trend in age-distance plots. Variable magma flux at the Island of Hawaii, and longer-term growth of the Hawaiian chain as discrete islands rather than a continuous ridge, may record pulsed magma flow in the hotspot/plume source.

INTRODUCTION

This overview, inspired by the 100th anniversary of the U.S. Geological Survey (USGS) Hawaii Volcano Observatory (HVO) in 2012, focuses on results of underwater studies of Hawaiian volcanoes that provide new perspectives on the growth of intraplate volcanoes. Recent studies have been especially productive for the Island of Hawaii (Fig. 1), where sonar surveys, dives, and dredging by the University of Hawaii, Monterey Bay Aquarium Research Institute, National Oceanic and Atmospheric Administration (NOAA), and USGS, and collaborations with the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) have complemented on-land scientific drilling and abundant data from HVO.

These results, in combination with a vast body of older data, provide new insights about volcano growth on Hawaii. More prior studies than can be acknowledged have evaluated growth of Hawaii in relation to longer-term evolution of the Hawaiian Ridge. Notable were early recognition of the southeastward younging of volcanoes and distinction of the parallel Kea and Loa volcanic trends (Dana, 1849; Jackson et al., 1972), and of course the insights about ocean-island volcanism that emerged in the 1960s from the plate-tectonic paradigm. Among recent critical observations and interpretations are: quantifying propagation rates along the Hawaiian-Emperor Ridge by isotopic dating (McDougall and Swanson, 1972; Jackson et al., 1972; Clague and Dalrymple, 1987), volume estimates from submarine bathymetry (Bargar and Jackson, 1974; Robinson and Eakins, 2006), recognition of an early-alkalic (“preshield”) stage at Loihi Seamount (Moore et al., 1982; Garcia et al., 1995a), insights into volcano growth based on ages of submerged slope breaks and coral reefs (Moore and Campbell, 1987; Ludwig et al., 1991), and geodynamic models for growth rates and compositional evolution in response to plate motion over a hotspot (Moore and Clague, 1992; DePaolo and Stolper, 1996; Ribe and Christensen, 1999; DePaolo et al., 2001).

Until recently, compositions and ages bearing on growth of Hawaiian volcanoes have largely come from subaerial sampling of late eruptive stages, and estimates of inception and early evolution have been heavily model dependent, using volcano spacing and plate motion to infer propagation rates and duration of edifice growth. The present interpretive synthesis, by combining recent chemical and 40Ar/39Ar isotopic-age data (mainly from underwater and drill-hole samples), revised edifice volumes, limitations from volcano structures and eruptive processes, and geodynamic constraints, attempts to interpret the growth histories of individual volcanoes and the composite growth of the entire island. While focused on construction of Hawaii Island, some data from older islands are referenced briefly where helpful to constrain volcano-growth models. In part, this analysis is a sequel to the impressive synthesis by Moore and Clague (1992), while benefiting from compilations for the geologic map of Hawaii Island (Wolfe and Morris, 1996a, 1996b) and the state map of Hawaii (Sherrod et al., 2007). Particularly useful samples, compositional data, and imaging of deep structure have come from the Hawaii Scientific Drilling Project (HSDP; Stolper et al., 1996, 2009) and ∼100 submersible dives and bathymetric surveys during JAMSTEC cruises in 1998–2002 (Takahashi et al., 2002; Robinson et al., 2003; Coombs et al., 2006a).

Reliable age determinations for old Hawaiian lavas remain sparse, but recent application of 40Ar/39Ar methods to young basalts, especially underwater and subsurface samples, has improved controls on volcano growth (Table 1). Alkalic lavas have yielded stratigraphically coherent results for Mauna Kea and Kilauea (Sharp and Renne, 2005; Calvert and Lanphere, 2006), but dating of low-K tholeiites continues to be problematic because of low radiogenic-argon yields. In such tholeiites, K resides mainly in glassy selvages next to groundmass minerals, leaving samples vulnerable to argon loss and/or argon-recoil problems in the reactor. Some tholeiite samples fail to yield meaningful ages without apparent petrographic or chemical reasons. In the most detailed 40Ar/39Ar study to date of Hawaiian tholeiites, on the 1.5 km scarp along the submarine southwest rift zone of Mauna Loa, only 14 of 45 analyzed samples yielded “successful ages” (Jicha et al., 2012). Less precise K-Ar determinations remain the main data for subaerial lavas on Kohala and Mauna Kea, and attempts to improve resolution by unspiked K-Ar methods for Kilauea and Loihi basalts (Guillou et al., 1997a, 1997b) yielded some results that are internally contradictory or inconsistent with 40Ar/39Ar dates (Calvert and Lanphere, 2006). Appendix A lists the published 40Ar/39Ar and some recent K-Ar age determinations used for estimating volcano growth rates for the Island of Hawaii and underwater slopes.

Volumes of individual volcanoes are also difficult to estimate, with probable uncertainties of 10%–20%, but constrained by the composite island construct (213,000 km3; Robinson and Eakins, 2006). Deep subsidence along the Hawaiian Ridge, defined by seismic profiling (Hill and Zucca, 1987), requires volumes nearly twice early estimates that assumed growth on flat ocean floor (Bargar and Jackson, 1974). The prior estimates used vertical contacts between volcanoes; here, edifice volumes are adjusted for sloping and interfingering boundaries (Table 2; Fig. 2; and discussion below). Effects of old seamounts and other irregularities of the ocean floor are neglected, as in prior estimates, but are unlikely to increase significantly uncertainties about total volume of the island construct. Positive gravity anomalies at volcano summits and proximal rift zones (Kinoshita et al., 1963; Kauahikaua et al., 2000), interpreted as recording dense intrusions and olivine cumulates, also help locate volcano boundaries where concealed by younger deposits. Each edifice is divided into subsections for which volumes can be calculated from simplified models, as done previously for Kilauea (Lipman et al., 2006). Further constraints come from eruption and lava-accumulation rates, especially for younger volcanoes. For volcanoes that onlap older edifices, an additional adjustment is made for the volume of deep intrusions and olivine cumulates. Details of volume calculations are tabulated in Appendix B.

The age, composition, structural, and volume data for individual volcanoes can then be used to model changing magma supply during sequential compositional stages, to place limits on the duration of volcano growth, and to evaluate the composite assembly of the island (Table 3). These data show that the growth stages of Hawaiian volcanoes are more diverse than previously documented, define inconsistencies at various scales with geodynamic models, and indicate that composite volcanic growth at Hawaii peaked ca. 800–400 ka.

MORPHOLOGIC AND COMPOSITIONAL GROWTH STAGES

Building on pioneering insights by Stearns (1946), Hawaiian volcanoes have commonly been discussed in terms of preshield, shield, postshield, and rejuvenated stages (e.g., Clague and Dalrymple, 1987; Peterson and Moore, 1987; Clague and Sherrod, in press). These terms have been used, somewhat ambiguously, concurrently to reference both morphology and composition. As chemical and age data become more abundant, especially for underwater samples that record early growth, tracking volcanic evolution primarily by composition and time seems increasingly desirable. Much more is known about later stages than earlier ones, although reliable age and eruption-rate data remain sparse. Modeling of growth commonly has assumed uniformly changing compositions and magma supply as the Pacific plate moves across a hotspot source, but recent data document considerable variation in stage durations, transitions between stages, periods of quiescence, and interactions between concurrently active edifices.

This paper distinguishes three main compositional stages: waxing early alkalic, sustained main tholeiite, and waning late alkalic, versus morphologic evolution of the edifice (submarine, subaerial shield, late submergence). No volcano on Hawaii Island contains highly alkalic late volcanism considered characteristic of the rejuvenated stage, and the age significance of this stage at older Hawaiian volcanoes currently seems uncertain. Rocks assigned to this stage at volcanoes like East Maui (Haleakala) form an age continuum with prior waning-alkalic eruptions, while similar late-erupted rocks are separated from waning-alkalic lavas by long intervals at West Maui or are absent at volcanoes such as Lanai (Sherrod et al., 2007; Clague and Sherrod, in press).

All Hawaiian volcanoes are broadly shield shaped in morphology regardless of composition, with sustained slopes rarely >15° both underwater and on land, except along fault and landslide scarps. Slopes are steeper underwater (Mark and Moore, 1987), in part because much shoreline-generated hyaloclastite breccia accumulates at angle of repose, in part because of steep slope-failure scarps at heads of submarine landslides.

The most pronounced change in topographic profile, from steeper submarine slopes to more gentle subaerial deposition as a seamount becomes an island (Mark and Moore, 1987), typically occurs during eruption of relatively uniform tholeiite that constitutes >90% of volcano volume. Although subaerial slopes are generally low during the tholeiitic stage (commonly <5°), much variation is present as a function of eruptive style. Sustained tube-fed pahoehoe sheet flows that grow by inflation have dips of only a few degrees (Hon et al., 1994), while slopes reach 10° or steeper where built by small tholeiite eruptions as on upper slopes of Mauna Loa (above ∼3000 m; Mark and Moore, 1987, their figure 3.2).

Independently of morphology and whether on land or underwater, edifice growth can be tracked by composition and magma supply. Tholeiitic lavas typically define coherent major-element arrays varying mainly in olivine content, as well documented for Kilauea and Mauna Loa (Wright, 1971; Clague et al., 1995; Sisson et al., 2002). Early- and late-alkalic lavas are more variable. Compositions that plot between the dominant tholeiite array and the alkali-basalt boundary of MacDonald and Katsura (1964) have commonly been designated transitional basalt (Wolfe and Morris, 1996a, 1996b; Sisson et al., 2002; Coombs et al., 2006b), a usage continued here (see Fig. 4). Such transitional basalts, which are abundant during shifts between compositional stages at some volcanoes, have also been described as low-silica tholeiite, especially in HSDP studies (e.g., Stolper et al., 2004; Rhodes et al., 2012). Even more subtle compositional variations among tholeiites at individual volcanoes, over varied time scales, are documented by trace-element and isotopic studies (Frey and Rhodes, 1993; Kurz et al., 1995; Pietruszka and Garcia, 1999; Marske et al., 2007; Weis et al., 2011).

The change from sustained-tholeiite (“shield”) to waning-alkalic eruptions (“postshield”) typically coincides with declining eruption rates, accompanied by submergence of the shoreline as volcano loading outpaces lava accumulation (Moore and Clague, 1992). Late-alkalic lavas also form steeper slopes than during the sustained tholeiite stage (up to 20°; Mark and Moore, 1987), a change probably resulting from smaller and briefer eruptions of alkalic basalt and more silicic lavas (hawaiite, mugearite) that form thicker and more viscous flows. Both compositional and morphologic changes have been widely referenced as the shield-postshield transition. However, capacity of a volcano to sustain subaerial growth is a function of volcano size in relation to magma supply. At a large volcano such as Mauna Loa, subsidence can outpace coastal lava accumulation late during the tholeiitic stage (Lipman, 1995; Lipman and Moore, 1996). In contrast, Mauna Kea continued subaerial growth well after the change to late alkalic volcanism (ca. 330 ka; Sharp and Renne, 2005), with submergence beginning to outpace growth only at ca. 130 ka (Moore and Clague, 1992).

As a result of such competing processes, the dueling balance between growth and subsidence at the shoreline can terminate at different stages of compositional evolution. Accordingly, the record of slope-break (shoreline) submergence (Moore and Campbell, 1987; Moore and Clague, 1992) provides critical evidence for declining eruption rates, but does not necessarily coincide with the shift from main-tholeiite to late-alkalic stage. Additional factors modulating volcano growth include changes in eruption sites and duration: distal segments of rift zones can shut down as volcano size increases (e.g., Mauna Loa southwest rift zone [Moore et al., 1990b], Hilo Ridge of Kohala [Lipman and Calvert, 2011]), eruptions become focused higher on the edifice, and shorter-lived eruptions with high proportions of a’a to pahoehoe tend to generate steeper slopes higher on volcanoes.

Sparse age-volume results suggest modest asymmetry in volcano growth, with rapid early increase in magma supply, followed by a more protracted waning stage. As detailed later, the only documented duration for early-alkalic stage volcanism (Kilauea) is fairly brief (∼150 k.y.), but volume and average magma supply are relatively large at Kilauea (∼2500 km3, 0.017 km3/yr) and Loihi (∼1000 km3 in ∼125 k.y., 0.008 km3/yr), in comparison to late-stage alkalic volcanism at Mauna Kea (∼800 km3 in 330 k.y., 0.0025 km3/yr), Kohala (∼300 km3 in ∼230 k.y., 0.0013 km3/yr), Haleakala (300 km3 in 950 k.y.: 0.0003 km3/yr), and none at Lanai (Sherrod et al., 2007).

Shifts between compositional stages are probably all gradational to varying degree (Clague and Sherrod, in press). The shift from waxing-alkalic to tholeiitic stage is in progress at Loihi Seamount, where compositional types interfinger on upper slopes (Moore et al., 1982; Garcia et al., 1995a). Thick lava sequences also interfinger during prolonged transitions from sustained-tholeiite to waning-alkalic stage (described as “late-shield”; Sherrod et al., 2007) at Mauna Kea (Wolfe et al., 1997; Rhodes and Vollinger, 2004) and Kohala (Lanphere and Frey, 1987).

Where stage transitions involve prolonged interfingering, dating of the change is inherently approximate. Perhaps the shift from early-alkalic to main-tholeiite stage should be defined by the initial appearance of abundant tholeiite (as currently at Loihi)—a time of increasing magma supply, when the continued eruption of alkalic basalt becomes volumetrically overwhelmed by tholeiitic lavas. For the change from main-tholeiite to late-alkalic stage, the transition could similarly be defined at the initial appearance of abundant transitional and alkalic lava. For the growth models in this overview, however, uncertainties about transition ages are rarely significant at the precision of available age control.

GROWTH AND MAGMA-SUPPLY MODELS

Modeling growth of Hawaiian volcanoes is complicated by many interacting processes. Factors favoring augmented growth in edifice size and height include increasing lava-accumulation and magma-supply rates during the progression from early-alkalic to main-tholeiite stage, along with intrusion-driven inflation and expansion. At large tholeiite-stage edifices, volcano height can be negatively impacted by load-driven subsidence, summit deflation, caldera collapse, flank spreading, and catastrophic slope failures. All Hawaiian volcanoes likely increase in height rapidly during early submarine growth because of initially small size. Subaerial volcanoes rise more slowly, even when eruption rates are higher, as the volcano area becomes large and subsidence modulates growth by lava accumulation.

Most previous volcano-growth models for Hawaii have portrayed age-volume relations as variants of a flattened bell curve, in which magma-supply and lava-accumulation rates increase during early growth, peak during the tholeiitic stage, and diminish during late alkalic volcanism. A perceptive early model for Mauna Kea (Wise, 1982; Fig. 3A) has been proposed with only modest differences for other volcanoes (Clague, 1987; Garcia et al., 1995a; Lipman, 1995).

Duration of eruptive stages has also been evaluated by geodynamic models involving steady-state plate motion over a fixed hotspot, as recorded by volcano spacing (Fig. 3C–3D; Moore and Clague, 1992; DePaolo and Stolper, 1996; DePaolo et al., 2001), but eruptive behavior in Hawaii appears to be non–steady state over a wide range of scales. Magma supply has increased markedly during the last few million years (Bargar and Jackson, 1974; Clague and Dalrymple, 1987), volcano spacing along the young end of the Hawaiian Ridge varies by at least a factor of two (40–80 km), major fluctuations in magma-generation and eruptive processes are recorded by the gaps between islands and seamounts, volcano volumes vary by an order of magnitude (Table 2), propagation rates are inconsistent for some adjacent volcanoes, historical magma-supply and eruption rates have varied at individual volcanoes, and life spans of volcanoes also may vary substantially as discussed later. Simple time-volume models, such as depicted in Figs. 3A and 3B, likely are generalizations of magma-supply fluctuations with fractal geometry on time scales from decades or less to that for growth of individual volcanoes, entire islands, and the ocean-channel gaps that separate them.

Determining long-term magma supply is especially challenging (Wright and Klein, 2013; Poland et al., in press). Short-term shallow magma supply at volcanoes like Kilauea has been estimated by combining historical observations, rates measured during eruptions, and intrusion volumes determined from geodetic data (Swanson, 1972; Dzurisin et al., 1984; Dvorak and Dzurisin, 1993; Cayol et al., 2000; Wright and Klein, 2013). Changes in magma supply also have been inferred from lava-accumulation rates at dated stratigraphic sections (e.g., Lipman, 1995; Sharp et al., 1996; Quane et al., 2000), but accumulation rates inevitably vary greatly with distance from vents and relation to local topography.

Late growth histories of the older Hawaiian volcanoes that are extinct or nearly so are constrained by data from subaerial lavas, but reconstructions of early evolution depend heavily on deep-water sampling. Even with recent drill-hole and submarine sampling, no Hawaiian volcano exposes a complete record of all growth stages. Accordingly, to evaluate magma supply during assembly of Hawaii, eruption rates that have been determined for a stage at one volcano are used to approximate volume-age evolution at others. For each volcano, one or more growth models are developed for 100 k.y. intervals (25 k.y. for Kilauea and Loihi), constrained by available composition, age, and volume data, and also by analogies with growth stages at other volcanoes, to permit inter-volcano comparisons and to interpret overall growth of Hawaii (Table 3). As these growth models are variably subjective and dependent on data availability, uncertainties are accordingly large. One major uncertainty involves volcanoes of differing size and volume: do smaller volcanoes have briefer life spans than large ones, or are they characterized by lower eruption rates, especially during the main-tholeiite stage? Nevertheless, available age, compositional, and volume data provide a framework to infer overall growth histories and make comparisons with geodynamic models. The time-volume distributions can be adjusted to varying degrees without violating available data, but application of consistent assumptions to the entire suite of volcanoes potentially provides insights about their diverse histories and the composite assembly of Hawaii.

Existing data are inadequate to evaluate whether the main-tholeiite stage is characterized by sustained near-constant magma supply (Fig. 3A; Wise, 1982), by a bell-curve peak (Fig. 3B; Holcomb et al., 2000), or by major variability from volcano to volcano. For the historical time frame at Kilauea and Mauna Loa, eruption and magma-supply rates have fluctuated on intervals of decades to centuries, perhaps antithetically, with periods of intense eruptions alternating with sustained intervals of reduced activity (Stearns and Macdonald, 1946; Lipman, 1980a; Klein, 1982; Swanson et al., 2011; Wright and Klein, 2013; Gonnermann et al., 2012; Poland et al., in press). Similar or longer-wavelength fluctuations are likely to have characterized earlier activity, but data to evaluate long-term trends are sparse.

Because the volcanoes of Hawaii differ substantially in volume (by an order of magnitude or more; Table 2), either the duration of volcano growth or peak magma supply must vary greatly. Several alternatives are explored for growth of less-constrained volcanoes: (1) near-steady-state progression of volcano inception, in accord with plate-motion models; (2) semi-equal durations (∼1100 k.y.) but varied peak-eruption rates; and (3) shorter durations at smaller volcanoes that maximize peak-eruption rate during the tholeiite stage (Table 3). In addition, recent ages suggest that volcano progression has been asynchronous between the Kea and Loa trends (∼N35°W on Hawaii). Measured and modeled propagation rates and growth stages, discussed in later sections, suggest that volcanoes grew earlier along the Loa trend than for similar positions along the Kea trend, at least for the more recent volcanoes. Accordingly, growth along each trend is summarized separately, in general order from younger volcanoes to less-documented older ones.

These discussions of available age, composition, and volume data, which are the framework for proposed growth models of individual edifices, provide the basis for evaluating overall island growth and resulting implications for geodynamic models of the Hawaiian hotspot/plume. Readers mainly interested in general interpretations and conclusions may prefer to go directly to the sections “Assembly of the Island of Hawaii” and “Discussion.”

KEA-TREND VOLCANOES

Because age and volume data are more robust for the Kea trend, these volcanoes are discussed first, starting with Kilauea where composition, age, and eruptive evolution are constrained by study of its young subaerial deposits, abundant seismic and other geophysical data on three-dimensional structure, several multi-kilometer-deep drill holes, and especially the submersible dives and samples obtained during the Japan-USA research supported by JAMSTEC during 1998–2005 (Takahashi et al., 2002; Coombs et al., 2006a).

Kilauea

Subaerial and underwater slopes of Kilauea display strikingly different records of growth. The on-land surface is mantled by tholeiite lava varying mainly in olivine content (Wright, 1971), mostly erupted <1.5 ka (Holcomb, 1987; Neal and Lockwood, 2003). Interlayered thin tephra deposits record prolonged intervals of volumetrically minor explosive activity, during which lava eruptions were sparse (Fiske et al., 2009; Swanson et al., 2012). Drill holes along Kilauea’s subaerial east rift zone have penetrated similarly uniform tholeiites to depths as great as 1700 m below present sea level (Quane et al., 2000).

Offshore of Kilauea, all sampled pillow lavas along Puna Ridge, the submarine continuation of the east rift zone, are similar tholeiite (Clague et al., 1995; Johnson et al., 2002). In contrast, no outcrops of Kilauea tholeiite have been found along the submarine south flank downslope from the summit. Below a prominent mid-slope bench at ∼3000 mbsl (meters below sea level) (MSB, Fig. 1), bedded volcaniclastic rocks interpreted as debris-flow deposits from ancestral Kilauea (Lipman et al., 2002) contain clasts of diverse submarine-erupted (high sulfur) alkali basalt, including nephelinite and tephriphonolite (Sisson et al., 2002), that are more compositionally diverse than known elsewhere on Hawaiian volcanoes except during the late “rejuvenated” stage. These have been interpreted as recording initial growth of Kilauea, broadly comparable to the current Loihi Seamount but including less-evolved alkalic compositions. Breccia-matrix and turbidite sands interbedded with the debris-flow breccias contain glass grains of submarine-erupted alkali basalt, mixed with degassed tholeiitic grains generated by shoreline entry of subaerially erupted lavas. This submarine volcaniclastic sequence thins westward against breccias of Loa-type tholeiite interpreted as the underlying flank of Mauna Loa. Above the mid-slope bench, to the shallowest exposures at 1800 mbsl, scattered bedrock ribs expose only weakly alkalic to transitional pillow basalts (Fig. 4). The change to subaerial-type tholeiitic lavas must lie concealed in shallower water, beneath the angle-of-repose mantle of shoreline-derived hyaloclastite.

Age and Volume

Prior geometric analysis of Kilauea’s volume, including contrasts between subaerial and submarine lava compositions, suggested a volume of ∼10,000 km3 for the edifice, with about one-quarter emplaced during the waxing-alkalic stage (Lipman et al., 2006). This volume for the alkalic part of the edifice, substantially larger than that of Loihi Seamount at present, may have been modestly overestimated, because subaerially erupted shoreline-derived tholeiitic sand forms matrix between alkalic clasts in some deep debris-flow deposits. The prior estimate of total Kilauea volume also neglected deep parts of associated summit and rift-zone intrusions emplaced within the underlying Mauna Loa flank, here roughly approximated as an additional 750–1000 km3 (Appendix A, Table A1); further interpretation assumes a total Kilauea volume of ∼11,000 km3.

Multiple 40Ar/39Ar incremental-heating ages on early-alkalic and transitional basalts from Kilauea’s submarine south flank provide especially tight constraints on ancestral growth of this volcano, as well as a possible template for early evolution of other Hawaiian volcanoes (Calvert and Lanphere, 2006). Inception of Kilauea no earlier than ca. 250–275 ka is inferred from high-precision plateau ages of 234 ± 9 and 238 ± 10 ka on phlogopite from nephelinites that record low magma supply generated by small degrees of source melting at initial stages of volcano growth. Ages on weakly alkalic pillow basalt above the mid-slope bench range down to 135 ka. Two ages from a thick breccia section of transitional lava are as young as 65 ± 28 ka (Hanyu et al., 2010), suggesting that main-stage tholeiites only became dominant at ca. 100 ka or even later (Fig. 4). These young ages for initial eruptions and shift to the tholeiite stage contrast with prior inference of earlier volcano inception (600–700 ka) based on plate-motion models (DePaolo and Stolper, 1996).

Geothermal drill holes along Kilauea’s east rift have penetrated tholeiite sections 1700 m or more thick (Quane et al., 2000), documenting proximal emplacement of this basalt type at depths nearly to that of the shallowest alkalic pillows on the offshore slope, helping to bracket the shift between compositional stages and suggesting that the change may have been fairly abrupt. Drill-hole samples have ages as old as 351 ± 12 ka by the unspiked K-Ar method (Guillou et al., 1997b), but their reliability has been questioned because of inconsistency with 40Ar/39Ar dates from submarine alkalic rocks and potential for excess Ar and K loss to disturb ages in low-K samples (Calvert and Lanphere, 2006).

Rare transitional flows and tephra with atypically high TiO2 and alkalis at low SiO2 (3 analyses of 437 tabulated for Kilauea; Wolfe and Morris, 1996b), erupted in the last few thousand years at Kilauea, support inference that the change to main-tholeiite stage is complex (Fig. 4) and may still be incomplete, consistent with the relatively modest current volume estimated for the growing volcano. Examples include the A.D. 600–1000 Kulanaokuaiki tephra (Dzurisin et al., 1995; Fiske et al., 2009), an associated lava flow (Lipman et al., 2006, their table 5), older flows of transitional basalt in Hilina fault scarps (Chen et al., 1996), and a young alkalic flow at the base of the distal Puna Ridge (Clague et al., 1995; Johnson et al., 2002).

Magma-Supply and Growth Models

A prior effort to model growth of Kilauea (Lipman et al., 2006), based on compositions and ages of submarine samples collected during JAMSTEC research, a revised edifice volume, and published estimates of late-20th-century magma supply (∼0.1 km3/yr; Swanson, 1972; Dvorak and Dzurisin, 1993), became the starting point for this summary that refines the Kilauea result and applies similar methods to the older volcanoes.

Diverse observations now suggest that magma supply at Kilauea has varied sizably in geologically recent time. Data from the continuing east rift eruption (since 1983) have documented varied eruptions rates, up to 0.2 km3/yr (Wolfe, 1988; Wright and Klein, 2013; Poland et al., 2012). Interpretation of geodetic data suggests that the total magma supply has been close to 0.18 km3/yr since 1961, including intermittent dike intrusions along rift zones during this interval (Cayol et al., 2000). Evaluation of the longer-term historical record of Kilauea eruptions, in conjunction with analysis of seismic data on magma-accumulation sites, also suggests supply rates to ∼0.18 km3/yr since ca. 1960, increasing from 19th- and earlier 20th-century rates of only 0.01–0.08 km3/yr (Pietruszka and Garcia, 1999; Wright and Klein, 2013). Examination of prehistorical eruptive deposits has begun to document even more complex variability in eruptive rates at Kilauea, with multi-hundred-year periods of lava eruption at high rates alternating with similarly long intervals dominated by explosive eruptions of only modest volume (Swanson et al., 2011, 2012).

Age-volume relations for the overall growth of Kilauea, as modeled here (Table 4; Fig. 5), show that magma-supply rates as high as 0.2 km3/yr must be recent, intermittent, or both. Such a rate, if representative during the ∼100 k.y. duration of main-stage tholeiite eruptions, would have yielded a volcano volume almost twice that estimated from geometric modeling. Early growth during the waxing-alkalic stage at Kilauea is modeled to fit the estimated volume for this interval (∼2.5 × 103 km3, during 275–100 ka; Lipman et al., 2006), so even a present-day rate of 0.2 km3/yr, constrained by a total Kilauea volume of ∼11,000 km3, requires rapidly increasing average magma supply since 100 ka, with a convex-upward slope that projects toward higher future rates (Fig. 5B). Such a supply rate also seems improbably high to be applicable for the sustained-tholeiite stages at older volcanoes. As presently modeled, a volcano as large as Mauna Loa could maintain a rate of 0.2 km3/yr for at most 100 k.y., even if its sustained-tholeiite stage were relatively brief (∼700 k.y.; see section on Mauna Loa, especially Fig. 12). Even a lower present magma supply, reaching 0.1 km3/yr at Kilauea after 100 k.y. in the sustained-tholeiite stage, produces a growth rate as high or higher than modeled for a modestly asymmetrical magma-time plot for other volcanoes (see sections on Kohala, Mauna Kea, and Hualalai, especially Figs. 9, 10, and 13; also Wise, 1982; Frey et al., 1990; Garcia et al., 1995a; Lipman, 1995). More rapid onset of tholeiitic magma supply to rates as high as 0.2 km3/yr would require a strongly asymmetrical growth-time curve, relatively brief period of peak magma supply, and prolonged decline in supply rates.

As an additional factor, the high supply rates estimated from geodetic and seismic data for rift extension during the past 50 years (Cayol et al., 2000; Wright and Klein, 2013) omit any component of passive flank motion and slumping driven by gravitational spreading (Fiske and Jackson, 1972; Borgia et al., 2000; Morgan et al., 2003; Byrne et al., 2013). The model of Cayol et al. (2000) infers average dike-induced rift opening of 40 cm/yr, which seems unsustainable for the prolonged duration of the tholeiitic stage at Kilauea. Such a dike-intrusion rate, if active since inception of tholeiite eruptions at ca. 100 ka, would have produced a zone 40 km wide of 100% dikes along Kilauea’s proximal east rift. In contrast, the intense dike swarm forming >40% (to 70%) of rock along the northwest rift at the deeply eroded Koolau volcano on Oahu, which appears geometrically analogous to Kilauea, is ∼10 km wide adjacent to its caldera and decreases to ∼5 km width 15 km down rift (Walker, 1986, 1987).

Because of these complexities and uncertainties, the preferred model for long-term magma supply at Kilauea is that in Fig. 5A, gradually reaching a multi-thousand-year average of 0.1 km3/yr since inception of its main-tholeiite stage at ca. 100 ka or younger.

Kohala

Kohala is discussed before Mauna Kea because its overall growth history is better constrained by dating, providing a possible template for modeling early evolution of other Hawaiian volcanoes. Recent underwater studies provide unique information on early edifice growth at Kohala, long recognized as the oldest subaerial volcano on Hawaii, and show that this volcano is larger than previously thought (Table 2).

Prior to the JAMSTEC-supported dives during 1998–2002, virtually all published compositional and age data for Kohala had been obtained on land, where mixed tholeiitic to weakly alkalic basalts (Pololu Volcanics) are capped by waning alkalic-stage lavas of the Hawi Volcanics (Stearns and Macdonald, 1946; Lanphere and Frey, 1987). However, many of the analyzed subaerial samples, especially tholeiites of the Pololu Volcanics, have been affected by variable to extreme alkali leaching and exchange (Lipman et al., 1990). Potassium-argon ages that have sizable uncertainties suggest that exposed Pololu rocks range from greater than 450 to ca. 300 ka, and that the overlying waning-alkalic lavas were erupted from ca. 280 to 120 ka, possibly as recently as 60 ka (McDougall and Swanson, 1972; Sherrod et al., 2007). Largely or entirely tholeiitic lower parts of the Pololu Volcanics become more compositionally diverse upward (Lanphere and Frey, 1987), recording a broad “late-shield” transition, here estimated at ca. 350 ka (280 ka in Moore and Clague [1992]).

A prominent slope break at 1000–1100 mbsl, continuously traceable for at least 60 km around the north submarine flank of the volcano, records submergence associated with waning of the sustained-tholeiitic stage at ca. 400 ka (Moore and Clague, 1992; Smith et al., 2002) and shows that Kohala was once much higher than its present summit elevation of 1678 m. Two large submarine slope failures of Kohala’s north flank, the Laupahoehoe and Pololu slumps, occurred late during the sustained-tholeiite eruptions, then were onlapped by younger lavas from Mauna Kea (Smith et al., 2002). Only a few tholeiite lavas have been sampled from underwater slopes, but abundant turbidite sandstones from the submarine north flank (37 samples from 3 dives) have uniform tholeiitic glass compositions without intermixed transitional or alkalic compositions, providing an indirect record of a long-lived tholeiite stage at Kohala (Lipman and Calvert, 2011; M.L. Coombs, 2010, written commun.).

The southeast-trending subaerial rift zone of Kohala is interpreted as continuing beneath Mauna Kea to reappear as the submarine Hilo Ridge (Figs. 1 and 6), as initially proposed by Holcomb et al. (2000) based on correlation of submarine slope breaks. This interpretation, in contrast to the more common depiction of Hilo Ridge as a rift zone of Mauna Kea (Fiske and Jackson, 1972; Moore and Clague, 1992; Wolfe et al., 1997), is supported by a residual-gravity anomaly along the Hilo Ridge that projects more directly toward Kohala than toward Mauna Kea (Kauahikaua et al., 2000) and by evidence for early inception of the ridge. Transitional to weakly alkalic pillow lavas that are overlain by tholeiitic picrite at the toe of Hilo Ridge (Fig. 7), interpreted to mark the change from waxing-alkalic to tholeiitic volcanism, have yielded 40Ar/39Ar plateau ages of ca. 1150 ± 35 ka (Lipman and Calvert, 2011). The ridge also has an overall reverse magnetic direction, requiring the bulk of rift growth before 760 ka (Naka et al., 2002, p. 46), much earlier than any dated tholeiitic lavas on subaerial Kohala (ca. 400 ka). By analogy with the 150–175 k.y. span interpreted for the waxing-alkalic stage at Kilauea, inception of early-alkalic volcanism at Kohala is estimated at 1300 ka (Table 5). These ages require the growth of Hilo Ridge before plausible inception of Mauna Kea related to semi-steady propagation along the Kea trend. For Hilo Ridge to be part of Mauna Kea would require rapid propagation from Kohala to Mauna Kea (at least 20 cm/yr, even if Kohala began as early as 1.5 Ma), then slowing greatly from Mauna Kea to Kilauea (∼5 cm/yr).

The reinterpreted Kohala east rift zone, as inferred from the summit to the distal toe of Hilo Ridge, is the longest among volcanoes on Hawaii Island (135 km). In comparison, Kilauea’s east rift zone, including its submarine extension along Puna Ridge, is 115 km long. The only longer Hawaiian rift zone would be the east rift zone of Haleakala and its submarine continuation along Hana Ridge, with an overall length of 150 km.

Hana Ridge offers an instructive geometric analog for evaluating size and geometry of the inferred rift-zone connection from subaerial Kohala to Hilo Ridge. This comparison can be illustrated (Fig. 8) by transposing major morphologic features of eastern Haleakala onto Kohala (present-day shoreline, submerged slope break at ∼2000 mbsl that marks decline in tholeiitic-stage eruptions, and approximate base of Hana Ridge adjusted on its north side for large-scale slumping). The distance from the present-day summits of the two volcanoes to submerged slope breaks along their ridge crests is similar (80–90 km), even though Hana Ridge continues 15 km farther underwater than the distal Hilo Ridge. The north-flank slope break is convex northward for both volcanoes, despite the presence of large submarine flank failures (Laupahoehoe slump for Kohala, Hana slump for Haleakala). The northward convexity along the northeast coast and submerged slope break of Hawaii results from younger infilling of lavas from Mauna Kea and also probably from late tholeiitic-stage Kohala (Smith et al., 2002, their figure 4). The geometrically similar convexity on the north flank of Hana Ridge probably records continued lava accumulation along this originally subaerial segment after slope failure generated the Hana slump (Eakins and Robinson, 2006). In contrast, the embayed south-flank slope break on Hana Ridge, which is asymmetrically close to the ridge crest, suggests that this side of the ridge was modified by late slope failures, and the resulting deposits are now concealed beneath younger rocks from the Island of Hawaii. Curvature of Hana Ridge appears somewhat less than that projected for the Kohala rift zone, but even so, the south flank of the transposed Hana Ridge would pass beneath Hilo and project beneath the summit of Mauna Kea.

Age and Volume

Dates from subaerial samples of the waning-alkalic stage document a broad transition from tholeiitic eruptions at ca. 350–300 ka and probable termination of Kohala volcanism at ca. 120 ka (Sherrod et al., 2007). Interpretation of Hilo Ridge as the distal east rift of Kohala (Holcomb et al., 2000), in conjunction with ages of transitional-composition basalt at its toe (ca. 1150 ka; Lipman and Calvert, 2011), allows the first estimated duration for the sustained-tholeiite stage (∼800–850 k.y.) of a Hawaiian volcano. This duration is substantially longer than that inferred from prior plate-motion models (∼500 k.y. [Moore and Clague, 1992]; 600 k.y. [DePaolo and Stolper, 1996]).

These results also imply that the 135-km-long east rift zone developed to near-total length early during growth of Kohala and imply a volume (Table 2) substantially larger than the prior estimate of 36,000 km3 (Robinson and Eakins, 2006). Any geometrically simple topographic profile connecting Hilo Ridge to Kohala requires the rift zone to have been subaerial at shallow depth beneath the north flank of Mauna Kea, limiting Mauna Kea to a much smaller volcano perched on the south slope of the large Kohala rift zone (Fig. 6). Kohala would have begun largely or entirely on ocean floor as an elongate northwest-southeast edifice without significant interference from pre-existing volcanoes. Rapid early growth of Hilo Ridge (Lipman and Calvert, 2011) suggests that at least distal parts of Kohala reached near-present size prior to major growth of Mauna Kea. Late interfingering of lavas from these volcanoes may have been relatively minor, as Kohala eruptions increasingly became focused closer to its present summit.

A more modest addition to the total volume of Kohala results from reduced estimates for Mahukona, as discussed for that volcano. If the volume of Mahukona is ∼6000 km3 (Garcia et al., 2012), or if this construct were the distal west rift zone of Kohala onlapped by a Hualalai rift, then half or more of its previously estimated volume (15,500 km3; Robinson and Eakins, 2006) becomes part of Kohala. Based on a simple geometric model of elongate ellipsoidal prisms for Hilo Ridge and northwest rifts of Kohala, while retaining the smaller Mahukona estimate from Garcia et al. (2012), a revised volume for Kohala is ∼64,000 km3 (Table 2; Appendix B, Table B2), nearly as large as previously estimated for Haleakala (69,800 km3; Robinson and Eakins, 2006). Of this increased Kohala volume, 7500 km3 was previously included with Mahukona and 20,000 km3 with Mauna Kea (Hilo Ridge and landward continuation).

Growth Model

By these volume interpretations, Kohala is among the largest Hawaiian volcanoes. Its lifespan (∼1200 k.y.) and duration of main-tholeiitic stage (∼800–850 k.y.) are relatively well constrained by the recent ages from the Hilo Ridge (Lipman and Calvert, 2011), in conjunction with prior K-Ar dating of its late-alkalic stage and analogy with duration of the early-alkalic stage at Kilauea. The Kohala ages provide the only measured duration for the main-tholeiite stage at a Hawaiian volcano, and simple geometric modeling of its growth as dominated by a mildly asymmetric trend of sustained tholeiite eruptions (Fig. 9A) yields peak magma-supply rates of ∼0.10 km3/yr that are similar to estimates of long-term historical rates at Kilauea (Swanson, 1972; Dzurisin et al., 1984; Wright and Klein, 2013). A rate this high can characterize only a fraction of Kohala’s tholeiite stage; otherwise, its total volume would be even larger than the 64,000 km3 estimated here (Table 2). An alternative growth model, in which magma supply becomes comparable to the 0.20 km3/yr peak rate during recent Kilauea activity, produces a high-amplitude short-wavelength growth curve (Fig. 9B) that could persist for only a brief interval without exceeding the total volume of this volcano. Even Mauna Loa, with its greater total volume, yields a compressed growth curve at such magma production rates. Such a short-duration high magma supply would also be inconsistent with a large-diameter hotspot source (100–150 km), as commonly inferred for the Hawaiian chain from diverse geochemical and geophysical evidence (e.g., Ribe and Christensen, 1999; DePaolo et al., 2001).

No estimates have been published for volumes of the late-alkalic lavas at Kohala, but no more than a few hundred cubic kilometers seems likely, judging by the widely exposed transitional and tholeiitic flows low on subaerial slopes and absence of alkalic clasts or sand grains in landslide and turbidite deposits sampled on the north submarine flank during the JAMSTEC dives. For a volume of ∼300 km3, erupted between 350 and 120 ka, average late-alkalic magma supply would have been ∼0.0013 km3/yr, significantly lower than for the shorter duration of waxing-alkalic eruptions at Kilauea or Loihi.

Despite uncertainties, the time-volume plots for Kohala provide a possible template for inferring growth rates at other less-constrained volcanoes. Based on its interpreted inception at ca. 1300 (± 50?) ka, and that of Kilauea at 275 ka, the distance between these two volcanoes (88 km) yields a propagation rate of 8.6 ± ∼0.4 cm/yr. This rate, similar to that for the entire Hawaiian Ridge (8.6 ± 0.2 cm/yr; Clague and Dalrymple, 1987) and to current motion of the Pacific plate (∼7 cm/yr; http://sideshow.jpl.nasa.gov/post/series.html), can then be used to infer inception ages for other volcanoes along the Kea trend. As discussed later, interpretation of propagation rates is more complex for volcanoes of the Loa trend.

Mauna Kea

Based on reinterpretation of the size and shape of Kohala including the Hilo Ridge, Mauna Kea is interpreted as a topographically high edifice (4205 m) of relatively modest volume (22,000 km3) among the volcanoes of Hawaii, which onlaps the flank of a much larger Kohala. The late eruptive record of Mauna Kea has been deciphered in detail, both from surface geology (Wolfe et al., 1997) and from the impressively documented HSDP holes targeted to penetrate Mauna Kea lavas near Hilo (DePaolo and Stolper, 1996; Sharp and Renne, 2005; Stolper et al., 2004; Rhodes and Vollinger, 2004; Stolper et al., 2009; among many publications). In contrast, the early growth history of Mauna Kea remains poorly known.

The 3506 m HSDP2 hole (Stolper et al., 2009) contains an ∼500 k.y. record of upper parts of the sustained-tholeiite stage and the change to late-alkalic lavas. The lowermost 2 km of the HSDP hole consist of submarine-emplaced tholeiitic pillow lavas and hyaloclastite breccias, all tholeiitic but including more variable compositional subgroups than characterize other main-stage tholeiites such as at Mauna Loa (Stolper et al., 2004; Rhodes and Vollinger, 2004). Extrapolation of 40Ar/39Ar ages suggests that the submarine section accumulated from ca. 635 to 400 ka (Sharp and Renne, 2005), when (at 1080 m depth in the hole) an abrupt change to subaerial lavas of similar tholeiitic composition continues to 352 m (ca. 330 ka). The entire 3.1 km tholeiitic sequence accumulated at an average rate of 8–9 mm/yr. Above 352 m to the top of the Mauna Kea section at 245 m (>200 ka), interlayered alkalic to tholeiitic lavas (Hamakua Volcanics) define evolution from the main-stage tholeiite to more uniform late-alkalic lavas and tephra (Laupahoehoe Volcanics) that are largely confined to upper subaerial slopes of the volcano. These compositionally transitional flows accumulated at a much lower average rate, ∼0.9 mm/yr (Sharp and Renne, 2005), and were capped at ca. 100 ka by Mauna Loa lavas in the upper 245 m of the hole.

The oldest subaerial lavas on Mauna Kea, the Hamakua Volcanics, include interbedded tholeiitic and alkalic flows with ages (K-Ar determinations only) of ca. 300–65 ka (Wolfe et al., 1997; Sherrod et al., 2007). The Hamakua Volcanics therefore are at least largely correlative with the transitional-composition lavas at 352–245 m in the drill hole. The alkalic Laupahoehoe Volcanics on upper slopes of Mauna Kea have ages from 65 ka to 4–5 ka (summarized by Sherrod et al., 2007); no deposits in the HSDP hole are equivalent to the Laupahoehoe rocks.

Diverse compositional, morphologic, and structural features of Mauna Kea are consistent with its interpretation as a relatively small volcano in volume, despite the high elevation of its summit. Without Hilo Ridge (here interpreted as a distal part of Kohala), Mauna Kea lacks morphologically or geophysically defined sizable rift zones. Late-alkalic vents scatter widely on its flanks without clear alignment that might reflect rift geometry, in contrast to more linear trends of late vents on volcanoes such as Haleakala, Kohala, and Hualalai. The volume of waning-stage alkalic lavas on Mauna Kea, estimated at 875 km3 (Wolfe et al., 1997), also seems large relative to overall size of this volcano, in contrast to 300 km3 for the much bigger Haleakala edifice (Sherrod et al., 2003). The apparent less-complete development of a sustained-tholeiite stage at Mauna Kea, in comparison to other Hawaiian volcanoes, may be related to the lower magma-supply rates modeled in following sections.

Additionally, the HSDP hole encountered compositional complexities in the 3.1 km main-tholeiite sequence (Stolper et al., 2004; Rhodes et al., 2012) that could be consistent with relatively small edifice volume, low magma supply, or interfingering with concurrently growing Kohala (compare projected location of Hilo Ridge’s south flank beneath Mauna Kea, at the HSDP site; Fig. 8). Almost a third (31%) of the submarine glass samples from the drill hole are low-Si tholeiite (<50% SiO2) that are atypical in the main-tholeiite stage at Kilauea and Mauna Loa (<1% of the analyses for these volcanoes, as tabulated by Wolfe and Morris [1996b]). Only about a third of the section, at 800–1950 m depth corresponding to ages of ca. 380–515 ka (Sharp and Renne, 2005), consists mainly of high-Si tholeiite; within this interval are two excursions to low-Si tholeiite. Deeper in the hole, proportions of high- and low-Si tholeiite are subequal, and the longest interval without low-Si samples (2650–2800 m) has been interpreted to span only ∼20 k.y. (ca. 595–615 ka; Stolper et al., 2004; Sharp and Renne, 2005). These tholeiites are compositionally more variable than thick main-stage sequences where sampled on other volcanoes such as Mauna Loa (Garcia et al., 1995b) or Kilauea (Quane et al., 2000). The abundance of low-Si tholeiite at Mauna Kea is consistent with lower partial melting and magma supply during shield growth than typical of other Hawaiian volcanoes (Sisson et al., 2002).

These compositional variations, and repeated alternation of subaerially erupted hyaloclastites with pillow basalts in the drill hole, might also in part result from interfingering with another concurrently growing volcano. In the drill hole at 2000–2900 m depth (ca. 520–625 ka; Sharp and Renne, 2005), degassed hyaloclastites (erupted on land, quenched at the shoreline) interfinger with pillow lavas that erupted underwater. The Hilo Ridge crest, projected 15 km north of the HSDP site, would have been near sea level in the vicinity of the drill site during formation of the bench at 1100 mbsl (425–450 ka, estimated from isostatic subsidence at 2.4–2.6 mm/yr [Moore, 1987]). Hyaloclastites in the drill hole could have come from the shoreline of one volcano, while pillow lavas erupted from the other (Lipman and Calvert, 2011). High-Si tholeiites of Kea-trend volcanoes are petrologically similar, making sources challenging to distinguish, but isotopic studies also suggest lavas from more than one volcano in the drill hole: the “lost volcano” of Blichert-Toft and Albarede (2009).

Age and Volume

Based on results from subaerial and the HSDP samples (Wolfe et al., 1997; Sharp and Renne, 2005), Mauna Kea progressed from main-tholeiite to late-alkalic stage during a lengthy transition interval from 330 to 65 ka, and is now near the end of its lifetime, at least in terms of total eruptive volume. The subaerial edifice continued to grow during much of this interval, as indicated by the submerged offshore slope break that marks decline in growth, at a water depth of 400 mbsl with an estimated age of 130 ka (Moore and Clague, 1992). The prolonged transitional volcanism and delayed decline in subaerial growth at Mauna Kea is in contrast to these events at Kohala, where the compositional change and shoreline submergence appear to have been nearly concurrent at ca. 350 ka.

The downward-revised volume estimate, from ∼42,000 to 22,000 km3 (Table 2; Appendix B, Table B3), results mainly from exclusion of Hilo Ridge and its landward projection beneath Mauna Kea. Also excluded is the Laupahoehoe slump area, depicted as part of Mauna Kea by Robinson and Eakins (2006) but interpreted here, and by Smith et al. (2002), as the northeast flank of Kohala overlain at a submerged slope break (1100 mbsl) by Mauna Kea lavas. The southwestern boundary between Mauna Kea and Hualalai is inferred to involve steep interfingering, because these two volcanoes are interpreted to have grown concurrently, at similar distance along the trend of the Hawaiian Ridge, and with similar ages of eruptive decline as marked by 130 ka submerged slope breaks (400 mbsl). Geometry of the deep boundary with Mauna Loa to the south is less certain, but also inferred to be steep in most sectors. Mauna Loa is only 25 km farther along the Hawaiian Ridge propagation trend, and the volcano-propagation model developed in a later section predicts nearly concurrent inception of these two volcanoes, resulting in large overlap in the sustained-tholeiite stages. In addition, Mauna Loa as the larger volcano may have a longer overall life span, and initial eruptions there could have begun earlier than projected from simple plate-motion models. Finally, because of higher magma-supply rates, Mauna Loa would likely have grown more rapidly and may have encroached on a concurrently active Mauna Kea.

A relatively minor uncertainty involves the probability that deep solidified intrusions of Mauna Kea magma, which generate the positive gravity anomaly below the summit, continue at depth into the underlying flank of Kohala. Assuming that the total thickness of the Kohala flank could be as much as 10 km, reaching as shallow as sea level (Fig. 6), and that intrusive feeders and solidified tholeiite-stage reservoirs were as much as 5 km in radius beneath Mauna Kea, this volume could be ∼800 km3, but well within uncertainties for the total volume. In the absence of geophysically detectable rift zones, no sizable volume of deep dike intrusions seems likely beneath the Mauna Kea edifice.

Growth Models

As a mid-sized volcano, for which age and volume are well constrained only for late growth, the early history of Mauna Kea is modeled by analogy with information from other Hawaiian volcanoes. Based on propagation of 8.6 cm/yr along the Kea trend as bracketed by results from Kilauea and Kohala (Table 3), volcanism at Mauna Kea is modeled to have begun at 850 ka, with a relatively brief ∼400 k.y. duration for its sustained-tholeiite stage, and a likely peak eruption rate of ∼0.07 km3/yr (Table 6; Fig. 10A). Alternatively, if overall lifespan (1100 k.y.) and duration of the sustained-tholeiitic stage (700 k.y.) were closer to that for Kohala, the peak eruption rate would have been lower, ∼0.06 km3/yr (Fig. 10B). Such an early inception and long lifespan, however, would require divergent propagation rates between Mauna Kea and adjacent volcanoes, rapid from Kohala and much slower to Kilauea. In either case, the highest eruption rate is about two-thirds the 0.1 km3/yr inferred for peak 100 k.y. intervals at Kohala or observed at present-day Kilauea. For Mauna Kea to have achieved even a brief interval of tholeiitic eruptions at a rate as high as 0.1 km3/yr, its modeled eruptive duration would have been shorter, ∼700 k.y. with a sustained-tholeiite stage of only 250–300 k.y. (Fig. 10C), timing that appears inconsistent with the ages from deep in the HSDP hole (Sharp and Renne, 2005). With an estimated volume of 875 km3 (Wolfe et al., 1997) and a 330 k.y. duration for the prolonged transitional and waning-alkalic stage at Mauna Kea (Sharp and Renne, 2005), its late-stage eruption rate is 0.003 km3/yr, 1–2 orders of magnitude lower than that modeled for its earlier tholeiite stage (Table 6).

Haleakala

As the youngest volcano north of Hawaii and the largest edifice in the Kea trend, Haleakala provides a framework for interpreting growth of the younger Kea-trend volcanoes. Haleakala has a volume estimated at ∼70,000 km3 (Robinson and Eakins, 2006), an original summit probably 4–5 km above sea level—1–2 km higher than the present summit (3055 m) as result of subsidence, and an exceptionally long rift zone—the 150 km Hana Ridge (Fig. 1).

No observational data bear on early growth, but based on the size of Haleakala, propagation rate along the Kea trend, and analogy with younger volcanoes like Kohala, volcano inception is modeled at 2000–2200 ka. A 900–1000 k.y. tholeiite stage ended at ca. 1000 ka, with estimated peak magma supply at ∼0.12 km3/yr. Lava compositions alternated for at least 100 k.y. during the shift to waning-alkalic stage on land (available dates, 1100–970 ka; Chen et al., 1991), after which only alkalic lavas erupted. A widespread submerged slope break, at ∼2000 mbsl, is interpreted to record submergence of the subaerial shoreline at ca. 950 ka (Faichney et al., 2009), near the end of the tholeiite stage (Moore et al., 1990a; Eakins and Robinson, 2006). In contrast to Mauna Kea, some tholeiitic volcanism continued after inception of shoreline submergence, as indicated by dredged tholeiitic hyaloclastite along the ridge crest (Moore et al., 1990a).

Diverse alkalic lavas then erupted for more than 900 k.y. at Haleakala (youngest flow, ca. A.D. 1600). The waning-alkalic rocks on Haleakala, accumulating concurrently with growth of Hawaii, formed a subaerial cap as much as 1 km thick but only ∼300 km3 in volume (Sherrod et al., 2007). No distinct time break accompanies the compositional shift from lavas designated “postshield” (Kula Volcanics) to the more alkalic Hana Volcanics (Sherrod et al., 2003), which previously had been interpreted as a “rejuvenated stage” (Clague and Dalrymple, 1987). The average magma supply for this prolonged interval of late-alkalic volcanism is only 0.0003 km3/yr, almost three orders of magnitude lower than during the main-tholeiite stage.

LOA-TREND VOLCANOES

Growth and magma supply for the Loa-trend volcanoes are discussed separately from the Kea trend, because the two groups appear to have contrasting eruptive histories, propagation rates, and inception ages that diverge on Hawaii (see Discussion, especially Table 11 and Fig. 15). A common propagation rate for the two trends would require the younger volcanoes along the Loa trend (Loihi, Mauna Loa) to have begun more recently than seems possible, as evaluated in a later section.

Loihi Seamount

Upper slopes of Loihi, the youngest and smallest of the clustered volcanoes that form Hawaii Island, with a summit about a kilometer below sea level, contain interlayered alkalic, transitional, and tholeiitic basalts. These provided the first compelling evidence for a waxing-alkalic stage in Hawaii and showed that this volcano is currently in early transition to the sustained-tholeiite stage (Moore et al., 1982; Garcia et al., 1995a, 2006). No lavas have thus far been sampled at Loihi that are as mafic and primitive as the nephelinites and other highly alkalic rocks from the submarine south flank of Kilauea, and the increased proportions of Loa-trend tholeiite high on Loihi suggests a stage comparable to that of Kilauea at ca. 100–125 ka. Many published Loihi glass analyses that have been described as tholeiite are less silicic (<50% SiO2, ∼3% total alkalis) than typical Loa-trend tholeiites from Mauna Loa or Hualalai; these would be classed as transitional or low-Si tholeiites in plots for other Loa or Kea volcanoes (Sisson et al., 2002; Stolper et al., 2004).

Age and Volume

Inception of volcanism at Loihi was estimated at ca. 100 ka by Moore and Clague (1992), based on plate-motion modeling and inferred duration of early-alkalic volcanism. In contrast, from unspiked K-Ar age determinations (Guillou et al., 1997b), Garcia et al. (2006) inferred that Loihi eruptions had already begun on the sea floor at depths of ∼5000 m by 330–400 ka. Some reported Loihi ages are internally inconsistent with sample depth, however, and ages obtained by this method for young basalts can be too old because of potential effects of excess Ar (Calvert and Lanphere, 2006). Alternatively, based on rates of volcano propagation discussed in a later section and analogy with duration of the early-alkalic stage at Kilauea, Loihi is inferred to have commenced at ca. 100–150 ka. Although no 40Ar/39Ar isotopic ages have been determined yet for Loihi basalts, alkalic rocks from deep on the landslide-scarred eastern and western flanks could be sufficiently old to yield reliable dates.

The volume of Loihi has been estimated as 1700 km3, with lower parts of the edifice concealed beneath the volcaniclastic apron derived from subaerial Hawaii Island (Garcia et al., 2006; Robinson and Eakins, 2006). By analogy with age-volume parameters for the waxing-alkalic volcanism at Kilauea, however, this volume at Loihi could have accumulated within 150 k.y. or less, also suggesting that the unspiked K-Ar ages are too old. In addition, the Loihi volume estimate seems high. Rather than initiated directly on deep-sea floor, Loihi appears to have grown over lower flanks of the large Punaluu slump derived from the south flank of Mauna Loa, where the latest major downslope movement is recorded by faulting in the Ninole Hills at 100–200 ka (Lipman et al., 1990; Jicha et al., 2012). If constructed on the Punaluu slump, much less of Loihi’s lower slopes would be concealed by younger volcaniclastic deposits, the erupted volume of the volcano would be significantly smaller at ∼800 km3 (Appendix B, Table B4), and it could have grown to present size in 125 k.y. or less. This smaller volume estimate neglects dense olivine-rich cumulates within the Mauna Loa flank beneath the summit of Loihi. As discussed for Kilauea, such deep cumulate may be as much as 20% of total magma supply, and accordingly, the estimated total Loihi volume is reduced to 1000 km3 (Table 2). Because Loihi is relatively young, its volume small, and magma supply low, such volume uncertainties play little role in modeling the overall growth of Hawaii.

Growth Model

With likely inception at 100–150 ka, and a volume of ∼1000 km3, Loihi’s growth would be dominated by increasing eruption rates during the waxing-alkalic stage (Fig. 11). Its in-progress shift to main-tholeiite stage, after erupting only about half the total volume as Kilauea at a similar stage, suggests that it may become a smaller or shorter-lived volcano. Loihi’s current volume would be comparable in size to Kilauea at ca. 150 ka (125 k.y. after initial Kilauea growth), and an inception age of 125 ka is modeled for Loihi (Table 7). For an earlier inception at 150–175 ka that would be most comparable to the early-alkalic duration at Kilauea, Loihi’s magma supply would be lower and seemingly less likely to have begun to erupt tholeiite. More recent inception at 100 ka would be consistent with a rapidly increasing magma supply, perhaps more appropriate for the shift to tholeiite. No data exist for present-day magma supply at Loihi, but by analogy with Kilauea’s transition to tholeiitic eruptions, the modeled rate of 0.03 km3/yr seems plausible.

A more significant problem in relation to Kilauea is the volcano-propagation rate; Loihi is too distant for its inferred age. Projected along the composite Kea-Loa trend (N35°W), Loihi is 45 km southeast of Kilauea. Based on an inception age (125 ka) only 150 k.y. younger than for Kilauea, the propagation rate between these volcanoes would be an improbable 30 cm/yr (see Discussion, especially Table 11B). Even for inception of Loihi at 75 ka, the propagation rate would be 22.5 cm/yr. Alternatively, at the Kea-trend propagation rate of 8.6 cm/yr, activity at Loihi should not have commenced until 525 k.y. after Kilauea, ∼250 k.y. in the future (see Discussion, especially Table 11A). A related problem with propagation rates between Mauna Loa and Kilauea is discussed next.

Mauna Loa

As the largest volcano on Earth, rising ∼15 km above oceanic crust down-bowed beneath the Hawaiian Ridge, Mauna Loa dominates growth models for Hawaii. Its volume has been approximated at 80 × 103 km3 (Lipman, 1995), as much as 105 × 103 km3 (Garcia et al., 1995b), and as little as 74 × 103 km3 (Robinson and Eakins, 2006) by a model that did not include the substantial subsurface flank onlapped by Kilauea (Lipman et al., 2006). Despite its size amounting to 35%–50% of the volcanic volume of Hawaii, and in part because of it, the growth history of Mauna Loa is perhaps the least constrained of the island’s volcanoes.

Historical eruptions, surface exposures of lavas as old as several hundred thousand years, on-land and underwater fault and landslide scarps as much as 1.6 km high, and the 245 m Mauna Loa section in the HSDP drill core, expose only tholeiitic lavas that have similar compositions modulated by olivine-control trends (Wright, 1971; Garcia et al., 1995b; Rhodes and Vollinger, 2004). No deep samples recovered to date provide compositional or age information bearing on an early-alkalic stage. While Mauna Loa continues to erupt frequently, diverse evidence suggests it is late in its main-tholeiite stage (Moore et al., 1990b; Lipman, 1995). Only three subaerial lavas (of 468 analyses; Wolfe and Morris, 1996b) have transitional compositions, but several young-appearing underwater cones on Mauna Loa’s west submarine flank are alkalic (Wanless et al., 2006), hinting that a transition to late-alkalic volcanism may be imminent. Additional evidence for eruptive decline comes from ages and lava-accumulation rates in relation to island subsidence.

Age and Volume

Several hundred surface flows from Mauna Loa have radiocarbon ages back to 30–40 ka, the effective resolution for this method (Lockwood, 1995), but older tholeiites have been a geochronometric challenge. Early attempts to analyze tholeiitic samples by K-Ar methods yielded ages with large analytical uncertainties, as well as some dates that are spurious on geologic grounds (Dalrymple and Moore, 1968; Lipman et al., 1990).

Recent results by 40Ar/39Ar methods have met with greater success, although analyses of low-K tholeiites remain difficult and uncertainties large. A flow from the base of the Mauna Loa section in the HSDP core yielded a groundmass date of 132 ± 32 ka (Sharp et al., 1996), consistent with the age (ca. 100 ka) inferred from lava-accumulation and coastal-subsidence rates at this site (Lipman and Moore, 1996). In the most substantial effort thus far, 11 dates on submersible and dredge samples from the 1.6-km-long scarp along Ka Lae Ridge (underwater southwest rift zone, Fig. 1) are broadly consistent with stratigraphic position, defining slowing of eruptions at ca. 400 ka while documenting that Mauna Loa was already a large subaerial volcano by that time (Jicha et al., 2012). A deeper sample dredged from the distal ridge yielded an age of 657 ± 175 ka, further helping delimit inception of tholeiitic volcanism at Mauna Loa. On the submarine west flank, a few ages in the range 240–460 ka, with large analytical uncertainties, have been obtained from tholeiites that have compositions similar to recent Mauna Loa lavas (Morgan et al., 2007). In addition, two tholeiite samples from the Ninole Hills on Mauna Loa’s south flank yielded ages of 227–108 ka, demonstrating that by this time, the subaerial Mauna Loa edifice was approaching its present size. This geochronologic evidence for early rapid growth to form Ka Lae Ridge, followed by a lengthy period of eruptive decline later during the main-tholeiite stage at Mauna Loa, support a general model of asymmetric growth during the lifespan of Hawaiian volcanoes, as initially diagrammed perceptively by Wise (1982).

The volume of Mauna Loa, estimated here as 83 × 103 km3 (Appendix B, Table B5), adjusts the Robinson and Eakins (2006) value of 74 × 103 km3, based on a substantially reduced volume of Kilauea and increased volume of Hualalai. Because Kilauea onlaps the south flank of Mauna Loa (Lipman et al., 2006), the volume of Kilauea is here estimated at 11 × 103 km3, in contrast to the previous 32 × 103 km3 (Robinson and Eakins, 2006). Inferred to offset some of this component, however, is evidence from gravity data that the south rift zone of Hualalai continues beneath the west flank of Mauna Loa for at least 20 km beyond any surface outcrops of Hualalai lavas (Kauahikaua et al., 2000). Accordingly, Hualalai is interpreted to be larger than previously estimated, at the expense of Mauna Loa volume (Table 2).

Two additional difficult-to-evaluate uncertainties further complicate these volume estimates: (1) a plausible but untested inference that the relatively short present northeast rift of Mauna Loa formerly may have continued beneath the east rift and Puna Ridge of Kilauea (Lipman, 1980b, p. 772; Flanigan and Long, 1987), and (2) a proposal that Hualalai’s south rift zone might once have continued as far south as Ka Lae Ridge (Holcomb et al., 2000). Neither of these alternatives can be evaluated unambiguously from available data. As possible support for alternative 2, a drowned seacliff along Ka Lae Ridge (Moore et al., 1990b; Garcia et al., 1995b) is similar in depth (430–450 m) to a submerged reef that drapes the northwest (Kiholo Ridge) of Hualalai (Moore and Clague, 1992). Alternative 1 would further increase the volume of Mauna Loa relative to that of Kilauea because much of Kilauea’s volume as currently interpreted resides in its long east rift zone (∼3000 km3, 30% of total Kilauea volume in just the submarine Puna Ridge; Lipman et al., 2006). Alternative 2, discussed further in the section on Hualalai, if valid, would increase that volcano’s volume substantially and reduce Mauna Loa’s. If both alternatives were valid, impact on the volume of Mauna Loa could be modest. Either alternative would transfer volume to an older adjacent edifice, thereby augmenting conclusions concerning timing of peak overall growth of Hawaii.

An additional proposal, that an ancient buried “Ninole rift zone” underlying the south flank of Mauna Loa predates the present-day configuration of the southwest rift (Morgan et al., 2010), is inconsistent with isotopic ages that are older for distal southwest rift lavas (400–650 ka) than for the Ninole Basalt (100–200 ka; Lipman et al., 1990; Jicha et al., 2012). Profiles of high seismic velocity and positive gravity, cited as evidence for a Ninole rift (Morgan et al., 2010, their figures 2 and 4), appear to have juxtaposed the large velocity/density anomaly associated with the summit magma reservoir of Mauna Loa and analogous geophysical expression of Kilauea’s distal southwest rift.

Growth Models

Multiple scenarios are possible for Mauna Loa’s growth, depending on estimates of its lifespan. As the largest Hawaiian volcano, would duration of its growth also be relatively lengthy? The recent 40Ar/39Ar ages from Ka Lae Ridge (Jicha et al., 2012) appear to document 600 k.y. or more of sustained eruption of compositionally similar tholeiite, without any hint of inception. If a duration of 100–150 k.y. is assumed for an early-alkalic stage by analogy with Loihi and Kilauea, and a future late-alkalic stage of ∼200 k.y., the lifespan of Mauna Loa would be at least 900 k.y.

Mauna Loa and Kilauea are the only Hawaiian volcanoes for which historical records and geologic mapping of young prehistoric flows provide quantitative data on eruption rates and magma supply. Total lava output for the 170 years of historical record (A.D. 1843–2012) is ∼4.1 km3 (Lockwood and Lipman, 1987), or 0.024 km3/yr. Detailed mapping of prehistoric lava flows on Mauna Loa suggests a roughly similar rate for at least the past 3000 years (Trusdell, 2010).

Intrusive contributions to the recent magma supply of Mauna Loa are difficult to estimate but seem likely to be less than at Kilauea where the rift-bounded south flank is spreading seaward much more rapidly than its geometric counterpart on Mauna Loa (Miklius et al., 1995; Miklius and Cervelli, 2003). The long-term intrusive contribution to Kilauea’s magma supply has been estimated at 30%–50% (Dvorak and Dzurisin, 1993; Cayol et al., 2000; Wright and Klein, 2013; Poland et al., in press); an assumed 30% for Mauna Loa would yield a historical magma supply of ∼0.035 km3/yr, only about a third of the estimated average historical rate at Kilauea (∼0.1 km3/yr).

In addition to isotopic and historical age data, coastal lava-accumulation rates and ages of submerged shorelines and coral reefs provide evidence for decline in magma supply late during the tholeiite stage. While Mauna Loa has been interpreted to be continuing vigorous growth because its historical lava volume (4.1 km3), if spread uniformly over the subaerial volcano (5125 km2), would have an average lava-accumulation rate of ∼5 mm/yr (Jicha et al., 2012), coastal accumulation has been insufficient to grow the subaerial edifice. About a quarter of historically erupted lava has ponded within the summit caldera (Lockwood and Lipman, 1987), some historical eruptions crossed the shoreline to deposit on submarine slopes, and much of the on-land lava accumulated preferentially in proximity to vents high on the edifice. With adjustments for these factors, overall coverage rates for subaerial slopes during the historical period averages only ∼3 mm/yr, and coastal rates would necessarily be lower. With shoreline subsidence at ∼2.6 mm/yr (Moore, 1970, 1987; Ludwig et al., 1991), the historical period thus provides little evidence for continued vigorous growth, or even maintaining the subaerial size of the edifice.

Along much of the present coastline, average lava accumulation is only approximately keeping pace with subsidence (Lipman, 1995), even though the on-land area of Mauna Loa has decreased by ∼20% as Kilauea has grown above sea level and overlapped the south flank of its large neighbor. The low rate of subaerial lava accumulation for Mauna Loa is well documented in the HSDP core, where average accumulation has been balanced by subsidence since at least ca. 100–120 ka (Lipman and Moore, 1996), despite funneling of Mauna Loa flows toward the drill site, from its northeast rift into the broad valley between Mauna Kea and the growing Kilauea shield.

The dated decline in lava accumulation at ca. 400 ka along the submarine southwest rift zone (Jicha et al., 2012) may result mainly from the apparent tendency for Hawaiian rift zones to be established early during volcano growth, then to become less active (“drying up”) as tholeiite eruptions become focused higher on the growing edifice (Moore and Clague, 1992; Lipman and Calvert, 2011). Alternatively, reduced activity along the lower southwest rift may have been caused by dislocations in response to large-scale landslides and slumps along Mauna Loa’s west flank (Lipman, 1980b; Lipman et al., 1990). Whatever the initial cause, decline in lava-accumulation rate along the lower southwest rift is further documented by a submerged paleo-shoreline, marked by an ∼10-m-high sea cliff with wave-rounded boulders at its base, 450 mbsl, with an interpreted age of 170 ka (Moore et al., 1990b; Garcia et al., 1995b).

Growth of the submarine Ka Lae Ridge (Fig. 1) to near present size by 400 ka (Jicha et al., 2012) and maintenance of its subaerial shoreline 8.5 km seaward of the present one until ca. 170 ka demonstrate that Mauna Loa already was a large subaerial edifice by these times, with topographic profiles projecting close to its present slopes. Accordingly, growth models developed here (Table 8) assume substantial decline in tholeiite eruption rates, starting ca. 400 ka. Even if volcano inception began as early as ca. 950 ka, about concurrently with Mauna Kea (and requiring a long gap before initial volcanism at Kilauea at ca. 275 ka), peak sustained magma supply for 100 k.y. periods at Mauna Loa likely approached 0.17 km3/yr (Fig. 12B), 50% greater than for present-day Kilauea. Alternately, if inception of Mauna Loa were younger, at ca. 850 ka, to allow for a more nearly constant age progression among the island’s volcanoes, the time interval for main-stage tholeiite magma supply becomes shorter and the estimated peak rate higher, ∼0.20 km3/yr (Fig. 12A).

The timing of volcano inception at Mauna Loa, as at Loihi, is a special problem in relation to propagation along the Kea trend. While Loihi is too distant from Kilauea to have a propagation rate consistent with motion of the Pacific plate, the proximity of Mauna Loa to Kilauea (23 km, along the composite volcano trend) would imply volcano inception at only ca. 540 ka, at the average progression rate (8.6 cm/yr) for the Kea trend (see Discussion, especially Table 11). Modeling of young inception of Mauna Loa (at ca. 540 ka) would also require extremely high magma production during the peak-tholeiitic stage (∼0.5 km3/yr at 450 ka) that then declined rapidly to the present-day rate of ∼0.035 km3/yr (Fig. 12C). Such a recent inception for Mauna Loa seems inconsistent, however, with dates as old as 657 ± 175 ka (Jicha et al., 2012) on submarine tholeiite from Ka Lae Ridge. Alternatively, Mauna Loa and Loihi could record volcano inception asynchronously older by several hundred thousand years than for counterparts along the Kea trend. Such an interpretation seems required by the young ages for waxing-alkalic growth of Kilauea (Calvert and Lanphere, 2006).

Hualalai

Hualalai appears to have begun growing before Mauna Kea even though it has erupted more frequently in recent time. The location of Hualalai slightly farther northwest along the Hawaiian Ridge suggests earlier inception, it appears to have grown mainly on ocean floor rather than on the flank of another volcano, and its estimated volume is larger than Mauna Kea. Its relatively small surface exposures are deceptive because of widespread cover by Mauna Loa, probable submarine overlap of Mahukona, and large-scale flank failure along the west coast. Hualalai remains active, in its late-alkalic stage (Moore et al., 1987), but the relatively infrequent eruptions are not keeping pace with shoreline subsidence (Moore and Clague, 1992). Future volumetric growth will be modest.

Subaerial Hualalai exposes only waning-stage alkalic lavas, but tholeiite crops out offshore and has been penetrated above sea level in water wells (Moore et al., 1987; Cousens et al., 2003). The topographic summit of alkalic basalt lies 5–7 km north of the gravity maximum that likely images intrusions associated with the tholeiitic stage (Kauahikaua et al., 2000), perhaps reflecting late vent migration in response to buttressing by Mauna Loa (Lipman, 1980b). Northwest- and south-trending rift zones are marked by late-alkalic vents that coincide with residual-gravity highs, in contrast to absence of similar alignments on Mauna Kea. The underwater continuation of the northwest rift zone (Kiholo Ridge) has bathymetric expression for >70 km from the subaerial summit (Fig. 1), and the associated gravity high continues to the Mahukona platform (Garcia et al., 2012). The south rift zone can be traced for at least 40 km, including its gravity expression beyond exposed Hualalai lavas (Kauahikaua et al., 2000).

Age and Volume

As for most old volcanoes on the island, direct information is unavailable for inception of the early-alkalic stage or the change to tholeiitic volcanism. A submerged slope break and coral reef, widely traceable at ∼400 mbsl and dated at ca. 130 ka, mark decline in main-stage activity at Hualalai, although some tholeiitic flows drape the reef (Moore and Clague, 1987). Trachyte lavas at Puu Waawaa on Hualalai’s north flank, inferred to record beginning of the waning-alkalic stage, have K-Ar ages as old as 114 ka (Clague, 1987; Cousens et al., 2003). The shift from tholeiite to waning alkalic stages is thus closely bracketed, <130 to >114 ka.

The volume of Hualalai, previously estimated at 14,200 km3 (Robinson and Eakins, 2006), is particularly uncertain because of cover by Mauna Loa lavas, gravitational failure of its southwest flank (North Kona slump), difficulty in tracing extent of rift zones, and likely complex interfingering with Mauna Kea to the east. A larger estimate of 26,000 km3 (Table 2; Appendix B, Table B6) is based on assumption that the perimeter of subaerial Hualalai lies 5–15 km beyond present exposures, onlapped by Mauna Loa flows, and that distal rift zones continue farther than estimated by Robinson and Eakins (2006).

An additional uncertainty is the speculative proposal that deeper parts of Ka Lae Ridge could be the distal continuation of Hualalai’s south rift (Holcomb et al., 2000). No petrologic distinctions are known between Mauna Loa versus Hualalai tholeiites that could help evaluate this alternative, but a Hualalai connection could be consistent with the abrupt decline in eruption rate at ca. 400 ka at Ka Lae Ridge (Jicha et al., 2012). Gravity data show a density gap between Hualalai and the southwest rift of Mauna Loa (Kauahikaua et al., 2000); however, making such an interpretation seem improbable. In addition, a lengthy south rift of Hualalai all the way to Ka La Ridge would likely have formed a structural barrier to large slope failures (as on the northeast side of Mauna Kea), but the west side of Mauna Loa instead has been the site of numerous giant landslides and slumps. If Ka Lae Ridge were part of Hualalai, its volume would be considerably larger, perhaps 45,000–50,000 km3, and that of Mauna Loa accordingly smaller. Broader implications of this uncertain hypothesis (Holcomb et al., 2000) are considered further in the section on overall assembly of Hawaii.

Growth Models

As a mid-sized volcano, for which age and volume are well constrained only for late growth, Hualalai’s earlier eruptive history is modeled by analogy with other volcanoes. Based on distance from Mauna Loa (37 km) and a Loa-trend propagation rate (10.6 cm/yr), Hualalai would have begun at ca. 1100 ka, with a relatively long sustained-tholeiite stage (Table 9A; Fig. 13A). Without inclusion of Ka Lae Ridge, peak tholeiitic magma supply would have been ∼0.05 km3/yr (Fig. 13A), about half that inferred for peak 100 k.y. intervals at Kohala or observed at present-day Kilauea. For Hualalai to have a tholeiitic stage approaching 0.1 km3/yr, its eruptive duration would have been several hundred thousand years shorter than at Kohala (Table 9B; Fig. 13B). With a rough volume estimate of 500 km3 and a 115 k.y. duration for the waning-alkalic stage at Hualalai, the late-stage eruption rate averages 0.004 km3/yr, similar to that for Mauna Kea but an order of magnitude or more lower than during the sustained-tholeiite stage.

Mahukona

The broad Kohala platform extending ∼50 km offshore from northwest Hawaii (Fig. 1), first proposed as the site of a submarine volcano by Stearns and Macdonald (1946, p. 56), was named Mahukona volcano and inferred to mark initial volcanism of Hawaii (Moore and Campbell, 1987). Dredging and submersible sampling have recovered tholeiitic, transitional, and weakly alkalic basalt along the broad ridge at the west end of the platform, which has been interpreted as the “missing volcano” along the Loa trend between Kahoolawe and Hualalai (Garcia et al., 1990; Clague and Moore, 1991). Most sites yielded tholeiite; transitional to weakly alkalic basalt has been recovered mainly from the large cone that forms the youngest part of Mahukona as currently interpreted. The eastern extent of Mahukona is hidden beneath the Kohala platform, which contains a stepped succession of at least six drowned coral reefs. Isotopic ages from the coral constrain late growth and termination of volcanism at Mahukona, as well as island subsidence rates (Moore and Campbell, 1987; Moore and Clague, 1992).

Is Mahukona Really a Volcano?

Despite much elemental and isotopic chemistry, 40Ar/39Ar age determinations, and detailed bathymetric and gravity surveys, uncertainty continues about the shape and volume of the Mahukona construct, location of its summit, whether the edifice ever rose above sea level, causes of compositionally diverse lavas, timing of the change from tholeiitic to late-alkalic stages, and even whether this feature constitutes a discrete volcano. Analogous interpretive ambiguities and possible alternative origins also exist for several other shallow elongate platforms offshore of older Hawaiian islands: Penguin Bank southwest of Molokai, Kaena and Waialu Ridges west of Oahu, and Pauwela Ridge north of Maui (Robinson et al., 2006).

Garcia et al. (1990, 2012) inferred a relatively small Mahukona seamount, having an area of ∼1600 km2, and a summit location marked by a large steep-sided cone that grew only to ∼270 mbsl. In contrast, Clague and Moore (1991) and Clague and Calvert (2009) interpreted a larger edifice that formerly rose above sea level, as now marked by a submerged platform ∼30 km across that contains a filled caldera. These contrasting views are closely tied to alternative interpretations of the submarine slope breaks that mark the end of sustained-tholeiitic growth in relation to ages of coral reefs on the platform (Fig. 1). Clague and Moore (1991) interpreted reef 6 and associated slope break as the paleo-shoreline of Mahukona, and correlated the Kohala tholeiitic shoreline with a shallower coral reef at 950 mbsl (reef 4) on the platform. They also interpreted “trains of basalt rubble in chutes” (Clague and Moore, 1991, p. 161) on an intermediate-depth reef (∼1150 mbsl, reef 5) as erupted from Mahukona, requiring that its summit lay to the east. All the reefs on the Kohala terrace are tilted southward in response to volcanic loading on the Hawaiian Ridge (Moore and Campbell, 1987), however, and the deeper reefs (5–6) rise and merge northward with a single slope break at a depth of ∼1000 mbsl. This break continues clockwise around the north flank of Kohala volcano and marks the decline of tholeiitic eruptions (Moore and Clague, 1992; Smith et al., 2002).

Recent geophysical, bathymetric, and petrologic data provide additional perspectives on complexities and uncertainties concerning the summit location and shape of Mahukona. An elliptical positive residual-gravity anomaly that trends northwest from Hualalai (Garcia et al., 2012, their figure 4) coincides with neither of the proposed summit locations nor the morphologically expressed west-trending Mahukona ridge. Absence of a dense core sufficient to generate a gravity anomaly at Mahukona was inferred to result from slow growth and small size (Garcia et al., 2012), but other Hawaiian volcanoes, and even the relatively small Loihi Seamount at an early growth stage, have large positive gravity anomalies associated with summits and proximal rift zones (Kinoshita et al., 1963; Strange et al., 1965; Kauahikaua et al., 2000). Absence of a positive anomaly marking a Mahukona summit would seem especially problematic for a relatively large edifice hosting a summit caldera and accompanying shallow magma chamber, such as proposed by Clague and Moore (1991). In addition, the broad Kohala platform seems puzzling morphologically for a waning-alkalic stage of a Hawaiian volcano; such late volcanism typically generates steeper slopes than the tholeiite stage.

As an additional complexity, tholeiitic lavas attributed to Mahukona are unusually diverse in elemental and isotopic composition, including subequal proportions of Loa and Kea types (Clague and Moore, 1991; Garcia et al., 2012). This variability, which differs from the other volcanoes of Hawaii Island, might be related to a diffuse magmatic system without a long-lived central reservoir, or possibly to overlapping by nearby volcanoes. For this last speculative alternative, based on geometry of the gravity anomaly and the mixture of Kea and Loa compositions, no discrete Mahukona volcano need exist. Instead, the western ridge of the Kohala platform perhaps could be an early-formed broad rift zone from Kohala (Kea composition), onlapped by continuation of the northwest rift zone from Hualalai (Loa composition). A larger analog for such distal widening of a submarine rift could be the landside-modified tip of the Hana Ridge of Haleakala (Eakins and Robinson, 2006). If part of the Mahukona edifice were a west ridge of Kohala, this rift would be ∼100 km long, and Kohala in broad form would be a 230-km-long ridge, paralleling the arcuate south flank of the clustered older volcanoes of Maui Nui (submerged platform of Maui, Molokai, Lanai, Kahoolawe).

Volume and Age

The volume of any Hawaiian volcano is inherently difficult to determine, because of uncertainties about edifice overlaps and effects of crustal subsidence beneath the Hawaiian Ridge. For Mahukona, these complexities are augmented by the uncertainties concerning volcano area, summit location, and origin. Possible alternatives include: (1) small western edifice, volume of ∼6000 km3 (Garcia et al., 1990, 2012); (2) larger edifice with summit farther east and a concealed flank onlapped by Kohala (Clague and Moore, 1991; Clague and Calvert, 2009), volume here estimated as ∼20,000 km3; or speculatively (3) even absence of Mahukona as a discrete volcano. In absence of positive evidence for alternative 3, or gravity data in support of alternative 2, models for this overview are based conservatively on the ∼6000 km3 estimate of Garcia et al. (2012). Because the volume of the Mahukona construct would be relatively small by whatever alternative, none of them substantially affects broad conclusions of overall magma supply and eruption rates during composite growth of Hawaii, discussed later.

Age estimates are also problematic, because of uncertainties concerning which dates are reliably associated with growth at Mahukona. Based on a coral age from the 1350-mbsl shelf break at reef 6, Clague and Moore (1991) estimated that tholeiite-stage eruptions began to decline by ca. 470 ka but persisted to at least 430 ka, the coral age of reef 5 that is draped by tholeiitic rubble inferred from Mahukona. Recent 40Ar/39Ar ages from transitional basalt on the shallow cone along the western Mahukona ridge are 300–350 ka (Clague and Calvert, 2009; Garcia et al., 2012), probably representing the near termination of eruptions. An age of 481 ± 37 ka for transitional basalt from deeper on the flank of the same cone suggests a possibly prolonged interval of waning volcanism, but no highly alkalic lavas have been sampled at Mahukona. An age of 654 ± 36 ka on tholeiite from a deeper cone farther west provides another limit on late activity (Garcia et al., 2012). Generalizing from these results, the putative Mahukona volcano is interpreted to have changed from main-tholeiite to late-alkalic stage at ca. 400–450 ka and to have terminated by 300 ka. No data exist for waxing-alkalic stage at Mahukona, but the bulk of this edifice must be tholeiite; only two sites have yielded transitional samples.

Growth Model

A 6000 km3 volume for Mahukona would imply growth markedly different from larger Hawaiian volcanoes. If its lifespan were comparable to volcanoes such as Kohala (∼1200 k.y.), then even with prolonged (100–200 k.y.) waxing and waning stages, peak magma supply would have been ∼0.01 km3/yr, an order of magnitude less than for larger volcanoes or present-day Kilauea. Peak tholeiite activity at even half the rate of larger volcanoes would require a much briefer lifespan (600 k.y. or less), but this would delay inception of Mahukona until ∼400 k.y. later than initial activity at adjacent Kohala or predicted from geodynamic models of volcano propagation.

Kahoolawe

In parallel to Haleakala on Maui Nui, the growth history of Kahoolawe bears on initiation of the Loa trend on Hawaii, but modern geologic data are sparse because of island use as a military bombing range. Available K-Ar ages roughly define the end of sustained-tholeiite eruptions at ca. 1200 ka, followed by late-alkalic volcanism until ca. 900 ka (Sherrod et al., 2007); analogies with other volcanoes suggest inception at 2100–2250 ka (Table 3).

ASSEMBLY OF THE ISLAND OF HAWAII

Composition, age, and volume data for individual volcanoes, while of uneven quality, quantity, and completeness, provide a framework for modeling the composite growth of Hawaii Island. Ages and volumes have been combined for three alternative magma-supply models for 100 k.y. intervals from each volcano: (1) near-constant propagation of volcano inception; (2) near-equal lifespan, varied peak-tholeiite rate; and (3) varied duration, high peak-tholeiite rate (Table 10). These models then can generate age-volume growth plots for the entire island (Fig. 14).

Somewhat unexpectedly, the three growth models yield generally similar results despite the varied assumptions and inputs. Each model has a broad peak of high magma supply (20–35 × 103 km3/100 k.y.) from ca. 400 to 800 ka, when four Hawaiian volcanoes (Kohala, Hualalai, Mauna Kea, Mauna Loa) were erupting tholeiite voluminously. The lower magma volumes for earlier time intervals reflect ramping up of volcanism at Hawaii as it waned on Maui Nui, and an interval of diminished magma supply along the Hawaiian Ridge as recorded by the inter-island channel. The lower volumes for younger intervals largely reflect existence of only a single volcano, Kilauea, with a recent magma supply of 0.1 km3/yr or higher. For the other two highly active volcanoes, Loihi is transitioning from its waxing-alkalic stage while tholeiitic eruption rates have been declining at Mauna Loa, resulting in lower overall magma supply. Because Kilauea only entered its main-tholeiite stage at ca. 100 ka or more recently, magma volumes were lower for the preceding few hundred thousand years as tholeiite eruptions diminished at Kohala, Hualalai, Mauna Kea, and Mauna Loa.

In comparison to the near-constant propagation plot (Table 10A; Fig. 14A), the age-volume distribution of magma supply shifts to a slightly earlier peak time for near-constant volcano life spans (Table 10B), because these models lengthen eruptive duration and lower peak magma supply for volcanoes with small total volume. In contrast, models of higher tholeiite production and variable lifespan (Table 10C) shift peak magma supply to younger times, because the reduced life spans require younger inception ages and reduced durations of the main-tholeiite stage. The high volume at 400 ka in this model is dominated by Mauna Loa input, a result probably inconsistent with available ages as discussed earlier. Although the differences among the plots are relatively modest, they incorporate substantially different growth models for some volcanoes. For example, the inception age for Mauna Kea varies from 1100 to 750 ka, and duration of its main-tholeiite stage 650–300 k.y.; for Mauna Loa, 950–550 ka, and 850–450 k.y., respectively (Table 3). These time-volume plots for composite assembly of the Hawaii construct thus broadly mirror, on different scales, the growth histories of the individual component volcanoes (Figs. 3A, 9–10, 12–13), and define an intense pulse of magma supply that has diminished during the last few hundred thousand years.

The growth models for individual volcanoes can be adjusted to varying degrees without violating available age data and uncertainties in volume estimates, and multiple iterations were explored. However, no seemingly reasonable models produced substantial changes to the growth geometries plotted in Figure 14, and some adjustments lead to improbable models. For example, models that raise peak-tholeiite magma supply for smaller-volume volcanoes, closer to the current 0.1 km3/yr rate for Kilauea and that modeled for large volcanoes like Kohala, reduce the time span for this stage and yield a narrow steep growth curve, rather than a broad plateau of sustained growth (e.g., Mauna Kea, Fig. 10C; Mauna Loa, Fig. 12C). Such abbreviated intervals of peak growth would likely require a hotspot magma source of much smaller magma-capture diameter than the 100–150 km inferred for the Hawaiian Ridge, based on duration of activity at individual volcanoes and distribution of concurrently active ones (Ribe and Christensen, 1999; Quane et al., 2000; DePaolo et al., 2001). A small hotspot locus would also make the sustained compositional contrast between the Kea and Loa trends (Weis et al., 2011) even more difficult to interpret, as well as the cause for eruptive loci along separate trends rather than a single one centralized over the propagation axis.

Evidence for decline in the overall magma supply during the last few hundred thousand years seems robust unless the models for Mauna Loa were drastically in error. No obvious alternatives seem adequate. Even if Mauna Loa were much younger than the models preferred here or previously proposed (800–900 ka; Lipman, 1995; DePaolo et al., 2001), with inception as recent as 550 ka and thereby requiring very high tholeiite-stage magma supply (up to 0.5 km3/yr; Table 8C; Fig. 12C), the island-wide magma-supply rate (at 300–500 ka) would still be about triple that since 200 ka (Table 10C; Fig. 14A). An even more speculative alternative might be the previously noted possibility that Ka Lae Ridge could be the distal south rift of Hualalai (Holcomb et al., 2000), thereby permitting eruptive decline at Mauna Loa to have begun as recently as 100 ka, rather than decreasing in growth at ca. 400 ka as implied by age determinations at Ka Lae (Jicha et al., 2012). This interpretation, although seemingly inconsistent with gravity expression for the rift zones of these volcanoes (Kauahikaua et al., 2000), could lower peak magma production and broaden the time span for growth at Mauna Loa, especially if combined with a young inception age as in Figure 12C. Such a model would also reduce the total volume of Mauna Loa and greatly augment that of Hualalai, however, precluding large increase in total magma supply for the island in the interval 400–100 ka. In this seemingly extreme model (not illustrated separately), the main-tholeiite stage during sustained growth of Mauna Loa would have been atypically brief (400–450 k.y.) compared to that documented for Kohala, and the island-wide magma supply would still have peaked at ca. 400 ka with a rate more than double that since 200 ka.

Magma-supply and eruption rates change by an order of magnitude or more at some individual volcanoes and between adjacent ones during the tholeiite stage, despite relatively uniform major-element melt compositions. These ranges suggest near-constant proportions of melting but large changes in source volume. Such source variations are likely recorded by more subtle trace-element and isotopic variations during tholeiitic growth (Frey and Rhodes, 1993; Rhodes and Hart, 1995; Pietruszka and Garcia, 1999; Marske et al., 2007; Weis et al., 2011).

DISCUSSION

Major results from this summary include recognition that no one-size-fits-all growth model accounts for the diverse age, volume, composition, and magma-supply variations among Hawaiian volcanoes. Volumes of the older volcanoes that have nearly completed their growth (Table 2) vary by up to an order of magnitude (if Mahukona is a separate edifice), volcano spacing along the Kea and Loa trends by a factor of two (Tables 11C and 11D), and peak magma supply probably by a factor of five or more during the tholeiite stage that accounts for the bulk of each volcano. The growth models also show that the evolutionary stages during volcano growth are more varied than previously discussed, demonstrate inconsistencies with prior geodynamic models, indicate that composite volcanic growth at Hawaii peaked ca. 800–400 ka, and suggest that current island growth is at reduced rates.

Variable durations of growth stages are well documented for the waning-alkalic stage on the older volcanoes: commonly a few hundred thousand years, but ranging from >950 k.y. at Haleakala to absent at Lanai (Moore and Clague, 1992; Sherrod et al., 2007). Much variability also seems likely for the earlier growth stages, for which age data are sparse. The waxing-alkalic stage is ∼50% longer at Kilauea than at Loihi. Modeling suggests that durations of tholeiitic activity were briefer at smaller-volume volcanoes like Mauna Kea than at larger ones such as Kohala (Table 3). As a result, the shifts between growth stages correlate inconsistently with propagation and inception rates estimated from volcano spacing and plate motion. For example, the tholeiite–late alkalic transition was nearly concurrent at ca. 330–350 ka for Kohala and Mauna Kea, 40 km distant. For Mauna Kea and Hualalai, the decline in morphologic shield growth (shoreline subsidence) was nearly concurrent at ca. 130 ka, while their tholeiite stages ended asynchronously at ca. 330 and ca. 120 ka, respectively.

Interpretations that have modeled growth at near-constant values for total volcano lifespan, duration of growth stages, and propagation over a fixed hotspot, thus are difficult to reconcile in detail with even the limited age and volume data now available. The constant propagation rate and growth-stage duration model for Hawaii by Moore and Clague (1992) required propagation at 13 cm/yr (Fig. 3C) that is much faster than present-day or longer-duration Pacific plate motion (7–9 cm/yr), inception ages for Kohala and Mauna Loa that are younger than can be reconciled with recent 40Ar/39Ar dates, and duration of the main-tholeiite stage of only 500 k.y. that would require magma supply at the larger volcanoes to be much higher than at present-day Kilauea (five times as high for Mauna Loa; Table 8C). Geometric models for the varied size and duration of Hawaiian volcanoes in relation to location along the trace of the hotspot by DePaolo and Stolper (1996) led to inference of a 600 ka inception age for Kilauea, more than twice that implied by recent isotopic ages (Calvert and Lanphere, 2006). Their steady-state geometric model, which related volcano lifespan and size to varied distance from the axis of hotspot progression, also lacks an explanation for alternation of large and small volcanoes on both the Kea and Loa trends. The uniform growth curves (Fig. 3B) modeled by Holcomb et al. (2000) are incompatible with the order-of-magnitude variation in volume among the island’s volcanoes (Table 2).

No single propagation rate works for the combined volcanoes of the two trends, either. The best-constrained rate for any of the Kea or Loa volcanoes is 8.6 cm/yr (Table 11; Fig. 15), calculated from time-distance relations for the waxing-alkalic and volcano-inception ages at Kilauea and Kohala (Calvert and Lanphere, 2006; Lipman and Calvert, 2011). In contrast, a propagation rate of 8.6 cm/yr from Kilauea yields an impossible future inception age for Loihi Seamount (-248 ka; Table 11A) and an implausibly young age for Mauna Loa (540 ka; Table 11B), younger than some isotopic ages on main-stage tholeiite from the distal Ka Lae Ridge (Jicha et al., 2012). At the best estimate of inception age for Loihi (ca. 125 ka), the propagation rate from Kilauea would be an improbably rapid 30 cm/yr (Table 11B). If Loihi were older than 125 ka (e.g., 330–400 ka, as inferred by Garcia et al. [2006]), the divergence between trends would be even greater.

Alternatively, a separate distance-inception age plot for Loa-trend volcanoes, both projected from Loihi inception at 125 ka and for individual volcano pairs, yields more rapid propagation rates of 10.5–10.7 cm/yr (Table 11D; Fig. 15B). This rate produces offsets to earlier inception ages for the younger Loa-trend volcanoes, relative to distance along the Kea trend, that reduce or eliminate inconsistencies in age-distance plots for these two volcano groups (Fig. 15B). Most importantly, the model inception age for Mauna Loa becomes geologically more plausible (ca. 800 ka). Such an average Loa-trend rate (∼10.6 cm/yr) would converge with the Kea trend on Maui Nui, yielding near-concurrent distances and ages for Haleakala and Kahoolawe.

A semi-constant Loa-trend propagation rate (10.6 cm/yr) yields an inception age of 2250 ka (Table 11) and a sustained-tholeiite stage of ∼900 k.y. for Kahoolawe, consistent with the slightly younger age (by only 50 k.y.) estimated independently for Haleakala, based on the lower propagation rate for the Kea trend. These inception ages for volcanoes on Maui Nui may be somewhat old, generating durations of 900–1000 k.y. for their main-tholeiite stage (Table 11, models C–D), and suggesting that propagation rates across the Maui Channel may have been more rapid, perhaps ∼11 cm/yr for both trends. The combined average Kea and Loa propagation rates agree reasonably with a prior estimate of 9 cm/yr by DePaolo and Stolper (1996), but are lower than the 13 cm/yr suggested by Moore and Clague (1992). However, such comparisons are based on limited data: ages and volumes are less known for the older Loa-tend volcanoes, and interpretation of the Kohala terrace (Mahukona) is especially problematic.

Inception ages and sites for new volcanoes along the Hawaiian Ridge, rather than following any simple geometric propagation, are likely nonlinear in detail, jumping ahead of the plate-motion progression or lagging behind in response to availability of favorable structures in the oceanic lithosphere or in adjacent volcanoes. Perhaps volcano inception can be triggered by controls other than just location relative to the melting zone. Some volcanoes could be conceived in the headwalls of slumps or slides on older volcanoes; Kilauea and Loihi both could have started this way. Complexities of the geometric models explored here may be indicating that other factors are involved in inception and growth of Hawaiian volcanoes. Although existing age data are inadequate to evaluate such alternatives, they demonstrate that volcano inception cannot be rigorously modeled based solely on motion of the Pacific plate. In addition, absolute southwestward motion of the hotspot at 4–5.5 cm/yr seems required by the divergence of direction and velocity for volcano propagation along the Kea and Loa trends on Hawaii (N35°W, 8.6–10.6 cm/yr) from the longer-term trend of the Hawaiian chain (N65°W, 9.6 cm/yr; Clague and Dalrymple, 1987) and from present GPS-measured motion of the Pacific plate (N65°W, ∼7 cm/yr; http://sideshow.jpl.nasa.gov/post/series.html). This hotspot motion thus would be about three-quarters as large as the plate motion, contributing significantly to the geometry of volcano propagation. Recent southwestward motion of the Hawaiian hotspot, also inferred from trends of isotopic and seismic data (DePaolo et al., 2001; Wright and Klein, 2006), could account for the change in trend of the Hawaiian Ridge at Maui Nui.

The variable growth histories of individual volcanoes, divergent propagation rates along the Kea and Loa trends, discontinuous magma flux during assembly of Hawaii Island, and longer-term growth of the Hawaiian chain as discrete islands rather than a continuous ridge, may record strongly pulsed magma flow in the hotspot/plume source, similar to that recorded in tomographic images of the mantle beneath Yellowstone (Schmandt et al., 2012). For the growth of Hawaii, the magma pulse took ∼500–600 k.y. to reach maximum supply, then was sustained for ∼400 k.y. before diminishing steeply. Peak growth of Hawaii Island between ca. 400 and 800 ka was probably preceded by another period of high magma supply at ca. 1.5–2.0 Ma during composite growth of Maui Nui. Such episodic magma flux and island growth along the Hawaiian chain thus may constitute an intraoceanic analog to the intermittent recurrence of ignimbrite super-eruptions and attendant caldera formation along the Snake River–Yellowstone hotspot track (Pierce and Morgan, 2009).

The conclusions presented here are largely based on sparse results from rocks that are challenging to date, while volume estimates remain poorly constrained because of limited control on dimensions of volcano onlap, the component of intrusive and cumulate bodies at depth below the associated volcano construct, and even uncertainty about the existence of Mahukona volcano. Inconsistent relations between location and inception age among the youngest volcanoes on Hawaii (Loihi, Kilauea, Mauna Loa), leading to interpreted offset between the Kea and Loa trends, suggest that comparable complexities probably characterize the older edifices for which early growth is even less constrained. Particularly needed are additional ages from early-alkalic basalts on other volcanoes, perhaps from submarine samples low along landslide scarps or distal rift zones, and controls on rate of main-stage tholeiite lava accumulation sampled from drill holes or submarine scarps. Much more could be learned from such additional sampling, along with efforts to improve capacity to determine reliable isotopic ages for low-K basalt (discussed further in Appendix A).

This overview, initiated as an invited keynote talk at the 2012 American Geophysical Union Chapman Conference on Hawaiian Volcanoes, has roots in decades of interactions with colleagues too numerous to acknowledge, but especially including Dave Clague, Michelle Coombs, Don DePaolo, Mike Garcia, Jack Lockwood, Jim Moore, Bill Normark, Tom Sisson, Don Swanson, and Bob Tilling. Much appreciated are thoughtful comments on an early draft by Bob Tilling, Jim Moore, Tom Sisson, and Brian Jicha, and helpful reviews for the journal by Dennis Geist and Dave Clague.

APPENDIX A. GEOCHRONOLOGY

Geochronology of Hawaiian lavas is extremely difficult due to modest K contents, difficult rock textures, and tropical weathering. Early results were often plagued by excess argon (implausibly old ages), negative or zero ages, and results that violated superposition. Recent studies using careful sample selection, preparation, and analytical techniques have obtained promising results. 40Ar/39Ar results from relatively high-K, crystalline-lava groundmass concentrates separated from submarine and borehole samples are reproducible, satisfy stratigraphic constraints, and appear reliable. Unspiked K-Ar samples using novel methods yield promising results, though occasionally in conflict with 40Ar/39Ar results. Tholeiitic lavas continue to yield little useable data, due to low potassium (K2O generally <0.4 wt%) and difficult textures, although Jicha et al. (2012) managed to produce 14 reasonable ages from 41 candidate tholeiitic lavas. All known published 40Ar/39Ar and unspiked K-Ar results for the Island of Hawaii are compiled in Table A1, along with representative radiogenic argon yields and bulk-rock potassium contents.

Armed with this improved chronologic framework and understanding of rock samples and analytical techniques necessary for reliable ages, we recommend several directions for future research:

(1) Loihi samples dated by Guillou et al. (1997a) should be analyzed using 40Ar/39Ar incremental-heating techniques, such as employed in laboratories at the Berkeley Geochronology Center, University of Wisconsin, Oregon State University, and U.S. Geological Survey. This sample suite contains 0.6–1 wt% K2O and yielded reproducible but stratigraphically inconsistent ages. 40Ar/39Ar techniques may solve those issues.

(2) Mauna Loa’s Ninole Hills have produced complicated K-Ar (Lipman et al., 1990) and 40Ar/39Ar results (Jicha et al., 2012) that suggest eruption at 100–200 ka. Further careful sample collection and analysis may yield more reliable results for these Mauna Loa units.

(3) Subaerial transitional and alkalic rocks from Kohala (Pololu and Hawi Volcanics) should be analyzed using 40Ar/39Ar techniques.

Finally, we encourage full disclosure of future Hawaii geochronologic work to help understand limitations of the techniques. It is tempting to publish only data that yield positive or stratigraphically consistent ages; however, Hawaiian rocks are unusually difficult to date and eventual success will require a collective effort. While discussing problematic samples often complicates and lengthens manuscripts, authors and editors are encouraged to present all data, not only those that yield satisfactory results.

APPENDIX B. ESTIMATED VOLUMES OF VOLCANOES

The following tables (Tables B1–B6) summarize geometric assumptions and other approaches used to estimate volumes for the individual volcanoes on the Island of Hawaii. Most revisions involve adjustments for inferred sloping onlap contacts between edifices, while constrained by the estimated overall island volume of 213 × 103 km3 (Robinson and Eakins, 2006). Uncertainties vary among volcanoes, but likely are about ±10% for most. Uncertainties are probably largest for the older volcanoes of the Loa trend.