The interplay between Kerguelen hotspot magmatism and Indian Ocean plate motions since the Early Cretaceous has produced a complicated record to decipher (Fig. F4). What is clear, however, from the preceding discussion, as well as from a large body of published work, is that the Kerguelen hotspot challenges many widely held assumptions about mantle plumes (Wilson, 1963, 1965; Morgan, 1971, 1972).
Current plume models predict massive magmatism coeval with continental breakup (e.g., White and McKenzie, 1989; Anderson, 1995), voluminous magmatism associated with an individual plume exploiting thin and weak lithosphere (e.g., Artyushkov et al., 1980; White and McKenzie, 1989, 1995; Thompson and Gibson, 1991; Sleep, 1996, 1997), and temporal scales of <5 m.y. for plume-head-type magmatism (e.g., Campbell and Griffiths, 1990). In contrast, peak output rates for magmatism attributed to the Kerguelen plume lasted for 25 m.y. Apparently, massive magmatism did not accompany the breakup of India and Antarctica but did accompany the breakup between India and Elan Bank. Extensive Early Cretaceous magmatic activity did not affect large areas of relatively young oceanic and transitional lithosphere, including spreading center segments between the SKP and Antarctica and along much of the margin of East India, yet produced lamprophyres by partial melting of the relatively old continental lithosphere of eastern India and East Antarctica.
Magmas associated with individual plumes typically show significant geochemical heterogeneity (e.g., in radiogenic isotopic ratios). Such heterogeneity is interpreted to represent intrinsic heterogeneities within the plume and mixing of plume material with entrained asthenosphere and overlying lithosphere. A depleted component in basalt associated with plumes is interpreted to reflect components derived from MORB-related asthenosphere and lithosphere (e.g., Galapagos; Harpp and White, 2001) or a depleted component intrinsic to the plume (e.g., Iceland; Kempton et al., 2000). For Site 1140 in the NKP, Weis and Frey (2002) favor mixing of melts derived from the Kerguelen plume and nearby (<50 km) SEIR. For lavas erupted in the Kerguelen archipelago, more distant from the SEIR at the time of eruption, the origin of the depleted component is under debate (Doucet et al., 2002; Frey et al., 2002a).
However, what distinguishes lavas associated with the Kerguelen plume is the diversity of radiogenic isotopic ratios ranging from values typical of MORB to those of continental crust (Figs. F8, F9). This range exceeds that of lavas associated with all other oceanic hotspots. Although erupted in an oceanic setting, some Kerguelen Plateau and Broken Ridge rocks contain abundant evidence for a continental lithosphere component. In particular, basalt from Site 738 in the SKP (Fig. F1) with 87Sr/86Sr = ~0.709 (Fig. F8) is interpreted "to have inherited its isotopic signature from old lithospheric mantle underlying the East Antarctica and southwestern Australia continental margins, rather than from the Kerguelen hotspot" (Alibert, 1991), and Mahoney et al. (1995) suggest that at Site 738 continental lithosphere was incorporated into the plume at "relatively shallow levels." Relative to Site 738 basalt, lavas from Site 1137 are not as extreme in Sr and Nd isotopic ratios (Fig. F8) but their Pb isotopic ratios (Fig. F9) also suggest a role for continental lithosphere (Ingle et al., 2002b). In this case, the clasts of ancient continental crust in a conglomerate intercalated with Site 1137 basaltic lava flows confirm the shallow origin of the continental component. Within the Cretaceous parts of the Kerguelen Plateau and Broken Ridge, continental components are widespread, ranging from Site 738 in the SKP to Site 747 in the CKP to Site 1137 on Elan Bank to Site 1142 on Broken Ridge. However, a continental component is not pervasive. Alkalic basalt from Broken Ridge (95 Ma) and tholeiitic basalt from Site 1136 (119 Ma in SKP) and Site 1138 (100 Ma in CKP) have Sr and Nd isotopic ratios (after correction for age differences) similar to Cenozoic lavas erupted in the Kerguelen archipelago (Fig. F8). Also, the role of a continental component in lavas associated with the Kerguelen hotspot has diminished from the Cretaceous to present; that is, a continental component is not evident in lavas forming the Ninetyeast Ridge (Frey and Weis, 1995) and the Kerguelen archipelago (e.g., Frey et al., 2002a; Weis et al., 2002), and only a single trachyte from Heard Island has a continental isotopic signature (Barling et al., 1994).
At three drill sites (747, 750, and 1139) on the Kerguelen Plateau, Pb isotopic data require a component with unusually low 206Pb/204Pb ratios (Fig. F9). The origin of this component, which is similar to the EM-1 oceanic island component, is not well constrained; Frey et al. (2002b) argue for a lower continental crust origin (Sites 747 and 750), but Kieffer et al. (2002) conclude that the "volcanic rocks of Site 1139 show no convincing evidence of continental material."
In summary, the Kerguelen hotspot shows longer-lived voluminous magmatism and more complex plume-lithosphere interactions than predicted by current plume models. As alternatives to incubating single plume head models (e.g., Kent et al., 1992; Saunders et al., 1992; Storey et al., 1992), we suggest two possibilities: (1) that the magmatism attributed to the Kerguelen plume did not arise from a single plume source, but to multiple plume sources (e.g., Burke, 2001; Wilson and Patterson, 2001); or (2) that a single plume source accounts for the magmatic products attributed to the Kerguelen plume, but that vigorous mantle circulation during Early Cretaceous time (e.g., Larson, 1991; Stein and Hofmann, 1994; Eldholm and Coffin, 2000) caused strong mantle shear flow that split the initial Kerguelen plume conduit into several "diapirs" of varying sizes, buoyancies, and mantle ascent rates (e.g., Olson, 1990; Steinberger and O'Connell, 1998).
Multiple sources for regional hotspot volcanism are suggested by tomographic images of the lowermost mantle that show heterogeneous slow regions in which slower, D'' hotspots are embedded (e.g., Garnero, 2000). Multiple fingerlike convective instabilities in the upper mantle have recently been proposed to explain the Tertiary-Quaternary volcanism of western and central Europe (Wilson and Patterson, 2001). If multiple mantle plumes can arise from individual heterogeneous slow regions at either the core mantle boundary or the 670-km discontinuity (i.e., the first alternative), then much, if not all, of the temporal and spatial variability displayed by the volcanic products attributed to the Kerguelen hotspot could be explained. This scenario might also explain some but not all of the geochemical heterogeneity in magma sources inferred from the variable radiogenic isotope ratios in basalt associated with the Kerguelen hotspot (e.g., Figs. F7, F8). If a plume is geochemically heterogeneous, as seems likely, the second alternative, splitting of a single plume into several diapirs, can also explain temporal, spatial, and geochemical variability in the volcanism. In Early Cretaceous time, separate mantle diapirs could have produced the Bunbury Basalt, the SKP, the Rajmahal Traps/Indian lamprophyres, the Antarctic lamprophyres, and the CKP/Broken Ridge. Additionally, separate mantle diapirs from the same source region could account for massive Cretaceous volcanism offshore Western Australia that has not so far been linked to the Kerguelen hotspot (e.g., Symonds et al., 1998). As mantle circulation rates slowed during the Late Cretaceous, a single plume conduit could have become continuous and long-lived, producing the Ninetyeast Ridge. Thereafter, old oceanic lithosphere, slow Antarctic plate motion relative to the plume conduit, and thickened Cretaceous Kerguelen Plateau lithosphere may account for the lack of a well-defined hotspot trace since 40 Ma.
Little is known about how the mass and energy fluxes of mantle plumes varies with time. Hawaii, the prototypical plume with its continuity of magma output over time (e.g., Duncan and Clague, 1985), is probably an end-member of the spectrum. At the other end of the spectrum are short-lived, massive bursts of flood basalt volcanism that lack any obvious subsequent volcanic record (e.g., the Central Atlantic magmatic province) (Marzoli et al., 1999). The Kerguelen plume appears to lie somewhere in the middle of this spectrum, as well-defined hotspot tracks are difficult to discern both before and after the creation of Ninetyeast Ridge. Finally, the obvious role for a continental component at several locations in the Kerguelen Plateau and Broken Ridge shows that even in an oceanic setting a plume can interact with continental lithosphere.