SUPERSWELLS Marcia K. McNutt Monterey Bay Aquarium Research Institute Moss Landing,California Abstract. Geophysicists have long sought direct, in- controvertible evidence from surface observables for convection in Earth's mantle. One of the best candidates for a phenomenonof convective origin is the South Pacific Superswell, a broad area of uplifted seafloor containing numerous volcanoesin French Polynesia. Other proposed examples of the superswell phenome- non include seafloor now located in the North Pacific that was anomalously shallow back in Cretaceous time and the present-day high topography of eastern and southern Africa. Researchers suggest that superswells form over mantle that is hotter than the global average on account of long-term absence of cooling by subduct- ing lithosphere. Superswell mantle is melt rich, as evi- denced by a fourfold increase in the rate of volcanism as compared with that of normal lithosphere. Seismic to- mography suggests that the source for this diffuse,ra- diogenically enrichedvolcanism without long age pro- gressions is a hot layer abovethe transition zone, rather than numerous deep-mantle plumes. Dynamic upwelling of this buoyant material in a low-viscosity zone immedi- atelybeneath the plate is responsible for both the shal- low seafloor and the dip in the elevationof the Earth's sea level equipotentialsurfaceover French Polynesia. Thermalmodels consistent with the seismic tomography, depth, and geoid data predict extremely minor pertur- bations to the temperature structure in the upper 50 km of the lithosphereand thus unresolvable anomalies in both heat flow and the stiffness of the elastic plate supporting the volcanoes. Superswell volcanism is dis- tinct from other typesof volcanism by not being imme- diately attributable to plate separation, plate conver- gence, or deep-mantle plumes. 1. INTRODUCTION Earth scientists do not doubt that the mantle con- vects. Simpleextrapolation of surface thermal gradients into the upper mantleyields temperatures at which rocks will flow on geologic timescales when subjected to stresses of the order of 1 MPa. It has been far more difficult, however, to point to specific geologic features caused by that convection. Although plate tectonics itself is surely evidence for convection, in that the plates represent the cold, upper thermal boundary layer of a convecting system, most geologic eventscan be inter- pretedasthe interactions of these plates, without requir- ing anyknowledge of the scale or pattern of convection beneath them. The highviscosity and thermal inertia of the plates effectively shield surface geologic processes from convective forces below.For example, overthe vast expanse of the ocean basins the depthof the seafloor is almost completely controlled by thermal contraction of the platesas they conductively cool, not by the viscous stresses from mantle upwellings and downwellings im- pinging on the baseof the plates. One exceptionto this rule is hotspots, which are thoughtto be plume-like thermal upwellings from the mantle that produceseamounts, oceanic plateaus, and broad areas of anomalously shallow topography around sites of active midplate volcanism [Wilson, 1963; Morgan, 1971].Within the theory of plate tectonics, there is no explanation for hotspots. They do not move with the plates, but theydo drift slowly with respect to eachother [Molnarand Stock,1987; Acton and Gordon,1994; Tar- duno and Gee, 1995;Steinberger, 1996]. In this paper, I make a distinctionbetween a hotspot, which is any meltinganomaly that produces more than the usual5-7 km of igneous oceanic crust,and a plume, which is one particular explanation for a hotspotthat invokes a nar- row, isolatedupwelling from the deep Earth. The type example of a hotspot is the Hawaiian-Emperor island and seamount chain, a line of volcanoes that eruptedin the middle of the North Pacific far from any plate boundary. The age-progressive nature of the Hawaiian chain (volcanoes getmonotonically olderto thewest and north) has led to the popularity of the plume explana- tion: The chainerupts asthe plate driftspassively over a hot, risingplume fixed with respect to the core-mantle boundary. The fact that the plume model so successfully fitsthe observations from Hawaii [Clague andDalrymple, 1989] has led to the generalbelief that all hotspots are plumes despite lack of any conclusive evidencefor a deep-mantleplume beneath Hawaii or anywhereelse. The scale of a hotspot is so smallthat only one, Iceland, hasbeen imaged with seismic tomography [Wolfe et al., 1997], and it is possible to explain anomalous depth, gravity, heat flow, and seismic velocity near hotspots asa perturbationto the thermal structure of the upper 400 km or less of the mantle [Detrick and Crough, 1978; Copyright 1998 by the AmericanGeophysical Union. 8755-1209/98/98 RG-00255515.00 02110 Reviews of Geophysics, 36, 2 / May 1998 pages 211-244 Papernumber 98RG00255