High-quality radiometric age determinations are crucial for understanding the chronology and rates of Kerguelen hotspot magmatism, the dynamic mantle processes responsible for the magmatism, and temporal relationships between magmatism and potentially related environmental changes. Leg 183 added five sites to the preexisting four igneous basement drill sites on the Kerguelen Plateau and provided the first drilled basement samples from Broken Ridge. Basalt from 10 of the 11 drill sites on the two features has now yielded high-quality radiometric (40Ar/39Ar) ages (Fig. F2) (Whitechurch et al., 1992; Coffin et al., 2002; Duncan, 2002). Eruption ages are also available for other mafic lavas that have been attributed to the Kerguelen hotspot, that is, the Ninetyeast Ridge (Duncan, 1978, 1991), the Kerguelen archipelago (Nicolaysen et al., 2000), northeast India (Baksi, 1995; Coffin et al., 2002; Kent et al., 2002), southwest Australia (Frey et al., 1996; Coffin et al., 2002), and Antarctica (Coffin et al., 2002).
In general, ages young northward on the Kerguelen Plateau (Figs. F1, F2). On the southern Kerguelen Plateau (SKP), ages are 118-119 Ma (Site 1136), ~112 Ma (Site 750), and ~110 Ma (Site 749) (Whitechurch et al., 1992; Coffin et al., 2002; Duncan, 2002). Elan Bank yielded an age of 107-108 Ma (Site 1137), and the CKP 100-101 Ma (Site 1138) (Duncan, 2002). Broken Ridge, prior to Eocene breakup, abutted the CKP and has yielded ages of 94-95 Ma (Sites 1141 and 1142) (Duncan, 2002); younger ages reported for basalt dredged from Broken Ridge (Duncan, 1991) are not considered reliable because of alteration (loss of 40Ar) (Duncan, pers. comm., 2002). The age of Skiff Bank is 68 to 69 Ma (Site 1139) (Duncan, 2002) and that of the northern NKP is 34 to 35 Ma (Site 1140) (Duncan, 2002). Age determinations for Ninetyeast Ridge basalt range from ~82 to ~38 Ma from north to south, respectively (Duncan, 1978, 1991). The Bunbury Basalt of southwest Australia (~123-132 Ma) (Frey et al., 1996; Coffin et al., 2002), the Rajmahal Traps of northeast India (~117-118 Ma) (Baksi, 1995; Coffin et al., 2002; Kent et al., 2002), and Indian and Antarctic lamprophyres (~115 and ~114 Ma, respectively) (Coffin et al., 2002) are believed to be continental expressions of the Kerguelen hotspot (Storey et al., 1992; Ingle et al., 2002b).
The radiometric age determinations described above have been combined with crustal volumes determined from wide-angle seismic data and gravity modeling to calculate the magma output rate of the Kerguelen hotspot through time (Fig. F3) (Coffin et al., 2002). Initial output rates were low from ~132 to ~123 Ma with eruption of the Bunbury Basalt. Between ~120 and ~110 Ma, rates increased by several orders of magnitude, to ~0.9 km3/yr, with emplacement of the SKP, Rajmahal Traps, and lamprophyres on the Indian and Antarctic continental margins. Following this peak rate, output appears to have waned to ~0.1 km3/yr for several million years, although this may be an artifact of sparse sampling or errors in assessing the contribution of hotspot magmatism to the crustal volume of Elan Bank. The CKP formed between ~105 and ~100 Ma at a rate of 0.9 km3/yr, similar to that of the SKP. Next, Broken Ridge formed from ~100 to ~95 Ma at a slightly lower rate of 0.8 km3/yr. Between ~120 and ~95 Ma, Kerguelen hotspot magma output rates exceed those of most known hotspot tracks (White, 1993), although an estimate of the magma flux during emplacement of the Ontong Java Plateau is an order of magnitude higher, at 8.9 km3/yr (Eldholm and Coffin, 2000).
No rocks have been definitively identified as products of the Kerguelen hotspot between ~95 and ~82 Ma, although the oldest part of Ninetyeast Ridge, buried beneath the Bengal Fan, may have formed during this interval. Whether magma flux diminished abruptly or gradually by nearly an order of magnitude, to ~0.1 km3/yr between ~95 and ~82 Ma, is open to interpretation. Coffin et al. (2002) assumed an abrupt change at ~95 Ma. Between ~82 and ~38 Ma, the Kerguelen hotspot generated the Ninetyeast Ridge and Skiff Bank at a rate of ~0.1 km3/yr. Finally, from ~40 Ma to present, the hotspot has produced the NKP, including the Kerguelen archipelago and Heard and McDonald Islands, at about the same rate as during the ~82- to 38-Ma interval. Tertiary and Quaternary output rates of ~0.1 km3/yr for the Kerguelen hotspot are typical of many hotspots, including Hawaii (White, 1993). Since ~40 Ma, some changes in magma flux are suggested by a ~30- to 24-Ma episode of volcanism in the Kerguelen archipelago (Nicolaysen et al., 2000). Since ~130 Ma, ~2.5 x 107 km3 of magma has been attributed to the Kerguelen hotspot. This estimate excludes normal thickness oceanic crust and volumes that contain dominantly continental material (Coffin et al., 2002). A previous estimate that also excluded normal thickness oceanic crust but did not subtract any continental crust was 75% larger, at 4.4 x 107 km3 (Saunders et al., 1994).
Plate motions in the Indian Ocean region since ~130 Ma have resulted in both continental and oceanic lithosphere of variable thickness transiting over the Kerguelen hotspot's asthenospheric source region. Furthermore, paleolatitudes of Kerguelen Plateau and Ninetyeast Ridge rocks suggest 3°-10° of southward motion of the Kerguelen hotspot relative to the rotation axis since 100 Ma, a result that can be numerically modeled with large-scale mantle flow affecting the location of the plume conduit (Antretter et al., 2002). Plate reconstruction models for the Indian and Southern Oceans since 130 Ma (Coffin et al., 2002, without southward plume motion; Kent et al., 2002, with southward plume motion) indicate that seafloor spreading initiated between Western Australia and Greater India at ~133 Ma (Fig. F4). The age of initiation of seafloor spreading between India and Antarctica, however, is more problematic (e.g., Kent et al., 2002). Because of a lack of definitive geophysical or geological data, the initial age of continental breakup between India and Antarctica could be as young as ~132 Ma (e.g., Gaina et al., in press; Kent et al., 2002) or as old as ~165 Ma (e.g., Roeser et al., 1996). As described later, the presence of continental lithosphere within the Kerguelen Plateau (Elan Bank and SKP) and possibly Broken Ridge and of apparent oceanic crust in the Enderby Basin and Princess Elizabeth Trough between the Kerguelen Plateau and Antarctica (Fig. F1) indicates that when India and Antarctica first broke up, continental portions of Elan Bank and the SKP were attached to Greater India. Subsequently, at least portions of the Early Cretaceous mid-ocean-ridge system between Antarctica and India jumped northward toward India one or more times as the Enderby Basin continued to open, creating the Elan Bank microcontinent and dispersing continental fragments in the SKP (Coffin et al., 2002; Kent et al., 2002). One or more ridge jumps, in turn, suggest the existence of one or more abandoned Early Cretaceous spreading centers between Antarctica and the Kerguelen Plateau and that the Enderby Basin began opening prior to ~132 Ma. Uncertainties in the regional tectonic development make evaluation of the initial effects of various plume models, including impact and incubation, difficult, especially with respect to cause and effect of plume activity and continental breakup.
The first appearance of magmas possibly related to the Kerguelen hotspot, the Casuarina (~128-132 Ma) and Gosselin (~123 Ma) types of Bunbury Basalt in southwest Australia (Figs. F1, F4), correlates both temporally and spatially with continental breakup between Australia and Greater India and between Australia and Antarctica, respectively. Lower Cretaceous volcanic rock capping the Naturaliste Plateau (Figs. F1, F4) may be correlative with the Bunbury Basalt (Coleman et al., 1982). Continental rocks dredged from the southern flank of the Naturaliste Plateau suggest that the feature is cored by continental crust (Beslier et al., 2001), so the volume of volcanic rock potentially related to Bunbury volcanism is unknown. Apparently, neither the area nor the volume of the Bunbury Basalt are of flood basalt province dimensions, which are typically 105-106 km2 and 106 km3, respectively. Also, volcanic rocks are not abundant in the continental margin of the Perth Basin (Symonds et al., 1998). So if indeed the Bunbury Basalt ± Naturaliste Plateau volcanic rocks represent the initial output of the Kerguelen hotspot, the underlying anomalous mantle was not significantly hot, wet, or voluminous enough to produce either a continental flood basalt province or a strongly volcanic passive margin despite the enhancing effects of thin, weakened lithosphere on decompressional melting. In marked contrast, voluminous magmatism accompanied ~136- to ~158-Ma continental breakup between Greater India and Australia to the north, creating volcanic margins and perhaps the younger Cuvier and Wallaby Plateaus (F1, F4) northward along the margin of western and northwest Australia (Symonds et al., 1998). To date, no hotspot source or sources have been identified for that voluminous magmatism.
The next phase of magmatism attributed to the Kerguelen hotspot, 110 to 120 Ma, began at least 12 m.y., and perhaps as many as ~45 m.y., after seafloor spreading started between India and Antarctica (Fig. F4). During this phase, the Kerguelen hotspot produced its only features of flood basalt scale on both thinned continental crust (Rajmahal Traps) (Kent et al., 1997) and in an ocean basin (SKP) (Figs. F1, F4). Between ~120 and ~110 Ma, the SKP grew at a high rate (Figs. F3, F4). Magma output rates were also high from ~105 to 100 Ma and ~100 to 95 Ma with the formation of the CKP and Broken Ridge, respectively.
The peak output rates of the Kerguelen hotspot from ~120 to ~95 Ma (Figs. F3, F4) lag initial breakup between India and Antarctica by 12-70 m.y. Corroborating evidence is that no physiographic feature analogous to the Greenland-Scotland Ridge, Chagos-Laccadive Ridge, Walvis Ridge, and Sao Paulo/Rio Grande Plateaus connects Antarctica and the SKP across the Princess Elizabeth Trough and that the East Antarctic continental margin south of the Kerguelen Plateau does not exhibit seismic characteristics that would classify it as strongly volcanic (Stagg, 1985; P.A. Symonds, pers. comm., 2001). Neither does the conjugate continental margin of East India (e.g., Gopala Rao et al., 1997; Chand et al., 2001), although thick sediments of the Bengal Fan mask the margin's basement structure due south of the Rajmahal Traps (e.g., Kent et al., 1997; Subrahmanyam et al., 1999). Thus, unlike the Iceland hotspot and the associated North Atlantic volcanic province and the Tristan hotspot and the associated Paraná/Etendeka flood basalt province, the peak output of the Kerguelen hotspot cannot be correlated temporally or spatially with a major phase of continental breakup and volcanic margin formation. Rather, like the relationship between the Réunion hotspot, the Deccan flood basalt, and the breakup between the Seychelles and West India, (e.g., Müller et al., 2001), the peak output of the Kerguelen hotspot appears to coincide with one or more episodes of microcontinent formation as pieces of East India, such as Elan Bank, broke off and became isolated within oceanic lithosphere. This hypothesis predicts that the portion of the East India continental margin from which Elan Bank and portions of the SKP broke up and separated are to some extent volcanic, despite the observed lack of significant volcanism (e.g., Gopala Rao et al., 1997; Chand et al., 2001). For the Bengal Fan portion of the East Indian margin, however, evaluating this prediction awaits acquisition of high-quality, deep seismic data.
The areal extent of magmatism attributed to the Kerguelen hotspot was very large from ~120 to ~110 Ma. For example, at ~119 Ma, the distance between the Rajmahal Traps and the Bunbury Basalt was ~2000 km, and a similar distance separated the Rajmahal Traps and the Antarctic lamprophyres at ~110 Ma (Figs. F1, F4). The recent 40Ar/39Ar age determinations described previously indicate simultaneous volcanism in India (Rajmahal Traps and lamprophyres), Antarctica (lamprophyres), and the Enderby Basin (SKP). That volcanism occurs over such a broad region is not surprising; what is surprising is that magmatic products from a single plume, whether axisymmetric or not, would apparently bypass the relatively young oceanic lithosphere of the Enderby Basin south of the SKP, including active spreading centers and the continent-ocean transition lithosphere of East Antarctica and much of eastern India, to erupt on old continental lithosphere, albeit probably rifted and thinned, of India and Antarctica (Fig. F4).
From ~82 (possibly ~95) to ~38 Ma, the Kerguelen hotspot produced the ~5000-km-long Ninetyeast Ridge. Geochemical evidence suggests that much of it formed relatively close to or at a spreading ridge axis (Frey et al., 1977, 1991; Saunders et al., 1991; Weis et al., 1991; Frey and Weis, 1995), but the lack of any conjugate feature on the Antarctic plate (cf. the Iceland hotspot and the Greenland-Faeroe-Shetland Ridge) casts considerable doubt that the Kerguelen hotspot coincided with a spreading center from ~82 Ma until breakup between the CKP and Broken Ridge at ~40 Ma (Fig. F4). Subsequently, the Antarctic plate moved over the Kerguelen hotspot, and the NKP was constructed on relatively old oceanic lithosphere. From ~30 Ma to the present, magmatic output related to the Kerguelen hotspot has been variable and spatially widespread. The flood basalt of the Kerguelen archipelago formed from ~30 to ~24 Ma (Nicolaysen et al., 2000), and less voluminous alkalic volcanism continued in the archipelago until <1 Ma (Weis and Giret, 1994; Weis et al., 1993, 1998). Since at least 21 Ma, Kerguelen hotspot magmatism has also constructed volcanic edifices on the Cretaceous CKP, including Heard Island (Quilty et al., 1983; Coffin et al., 1986; Weis et al., 2002). Both Heard and McDonald Islands have had historical eruptions (Quilty and Wheller, 2000).