Sediment Evidence of Plumes

Evidence that the Earth’s mantle is in a state of circulation is well documented, yet the relative influence of deep-seated mantle plumes, some possibly originating at the core-mantle boundary is still hotly contested (e.g., Anderson, 2003; Foulger and Natland, 2003). Areas of elevated topography linked to greater than normal magmatism in the ocean basins have been historically been linked to greater melting above a steady state plume, or to the impacting of a newly initiating plume head on the base of the lithosphere (e.g., Campbell and Griffiths, 1990). New models however advance excess melting due to lithospheric processes above essentially normal upper mantle (Anderson, 2003). Resolution of this debate is essential to understanding the major controls on terrestrial magmatism and the geochemical evolution of the Earth.

Although different plume models propose different magnitudes of thermal anomaly, ranging up to +350˚C (Farnetani and Richards, 1994), a common characteristic of deep-seated mantle plumes is their excess heat relative to the surrounding ambient asthenosphere. Greater heat results in the plume mantle being less dense than normal, allowing this material to rise to the base of the lithospheric plates. This same excess heat and buoyancy is also inferred to cause long-wavelength uplift and shallowing of the seafloor, which can be seen in modern bathymetry around active hotspots or superswell areas (Crough, 1978; Detrick and Crough, 1978; McNutt and Judge, 1990). Alternatively, ancient plume-driven uplift can be preserved in the sedimentary rocks deposited in hotspot regions (e.g., White and Lovell, 1997). As a result of plate motion crust that is affected by plume activity is removed from over the mantle thermal anomaly. Continued upper mantle convection then removes the hotter, less dense material and allows the crust to subside to normal depths over periods of 10’s of millions of years (Olson, 1990; Griffiths and Campbell, 1991).

   Although the effects of plume activity have been interpreted from sedimentary records from continental margins, most notably in the NE Atlantic and Kerguelen Plateau (Brodie and White, 1994; White and Lovell, 1997; Clift et al., 1995; Coffin and Gahagan, 1995), anomalous subsidence linked to plume activity is most readily identified within oceanic crust because the subsidence behavior of oceanic lithosphere is relatively simple and well characterized (Parsons and Sclater, 1977; Stein and Stein, 1992). Although the absolute depth of any given piece of oceanic crust will vary depending on the thickness and density of the crust, the rate at which subsidence occurs after the end of crustal accretion can be used to identify the effects of mantle buoyancy, because this is controlled by cooling and thickening of the mantle lithosphere (Parsons and Sclater, 1977).

   Isolating the effect of hot asthenosphere on plate subsidence is simplest if the hotspot volcanic crust was emplaced close to the crest of a mid ocean ridge, as this allows a simple comparison between the subsidence reconstructed from the sedimentary record and that of normal oceanic crust. Clift (1997) used this approach to show anomalously shallow depths within the oldest oceanic crust adjacent to the volcanic rifted margins of the NE Atlantic, shortly after continental break-up in the Eocene. If all the anomalous depth is assigned to temperature rather than compositional buoyancy or dynamic flow then the degree of misfit with the oceanic model can be used to estimate maximum temperature anomalies.

   The subsidence of lithosphere affected by plume activity has been modeled in a variety of ways, treating the plume as either an isostatic anomaly or as a region of dynamic flow (e.g., Ribe and Christensen, 1994; Sleep, 1987, 1990). Dynamic models generate uplift through the viscous normal stresses that would be imposed on the base of the lithosphere by an upwelling column of material and mimic modern bathymetry accurately with no requirement for a temperature anomaly. However, results consistent with the evolving bathymetry along the Hawaiian seamount chain have also been generated by simple isostatic, conductive cooling models (Sleep, 1987), such as that of Ito and Clift (1998), which uses a one-dimensional thermal diffusion approach. The model allows the hot cushion of plume material to diffuse away after the initial emplacement, resulting in a gradually reducing thermally induced buoyancy with time. Figure 1 shows a series of models generated from this type of model that shows the difference in subsidence behavior predicted for oceanic crust emplaced above normal mantle, above a plume of moderate strength (T = 150˚C, 100 km thick plume head) and above a plume at the high end of current estimates (T = 350˚C, 200 km thick). In each case the end point of the subsidence history is known, because it is the present day depth of the basement, but the initial elevation is predicted to differ significantly depending on the original mantle temperature. As a result the predicted subsidence curve for crust generated above a plume mantle might be expected to be steeper than that above normal oceanic mantle.

   The predicted subsidence history is slightly more complex if the emplacement of hot mantle material and associated hotspot volcanism occurs away from an oceanic spreading center. In this case the water depth shallows compared to equivalent age crust away from the hotspot, as a result of crustal thickening and the buoyancy of the mantle plume material. The rate of subsidence of the crust after removal from the area of the hotspot and its swell is faster than crust of similar age that was not influenced by the hotspot. Detrick and Crough (1978) noted such behavior in the Hawaiian seamount chain, a phenomenon that they assigned to a thermal erosion of the Pacific lithosphere after it passes over the Hawaiian plume. Although the Hawaiian Islands are much shallower than the surrounding abyssal seafloor they subside at a faster rate after they pass away from the immediate influence of the thermally erosive plume. Alternatively Phipps Morgan et al. (1995) explained the bathymetry of the Hawaiian Islands as being the result of melt extraction in the asthenosphere producing a buoyant mantle pillow centered under the hotspot. This buoyant mantle was partially entrained with the lithospheric plate as this moved away from the hotspot. In the case of old hotspot provinces it is not always clear whether a hotspot edifice was emplaced on axis or not, and this may be decided by comparing hotspot and lithospheric ages. In this study I infer the age of the surrounding oceanic lithosphere on the basis of the nearest identified magnetic anomaly, while the age of hotspot volcanism is taken from the radiometric age of the basement lavas, or from the biostratigraphic age of the overlying sediment if the former is not available.

   In order to assess the magnitude of mantle temperature anomalies associated with hotspot magmatism I have reconstructed the subsidence history of a series of oceanic plateaus, ridges or seamounts using their sedimentary covers. These features span all the major ocean basins and a wide period of geologic time (Fig. 2). Their analysis is made possible through the sampling of the Deep Sea Drilling Project (DSDP) and the Ocean Drilling Program (ODP), which recovered the sediments overlying the volcanic sequences. The sediments and fossils found in these cores allow the depositional environment, and especially the water depth of deposition to be constrained. These data were compiled from the cruise reports and associated postcruise papers, particularly those on benthic foraminifer fauna, which are considered sensitive water depth indicators. Water depth estimation is the principle uncertainty in the reconstruction of the vertical motions of the igneous basement and is generally better where water depths are shallow (<200 m), as this is the range where sedimentary structures and benthic foraminifers shows the greatest resolution. Nonetheless, some resolution is possible at greater depths. For the purpose of this analysis I assumed that the present depth of the features is the maximum during their history. The stratigraphy at each drill site was “backstripped” in order to isolate the tectonic component of the subsidence history. This process involves the correction of sediment loading and restoration of the seafloor to the estimated depth in the past for each dated interval, assuming local isostacy (Sclater and Christie, 1980). In doing so the vertical motion of the igneous basement at each drill site can be reconstructed since the end of volcanism. No correction was made for sealevel variations because of debate about the timing and magnitude of this effect. However, the uncertainties introduced by this omission are small (~100 m) compared to the much greater values of total subsidence and the general uncertainties in the water depths.

   Figures 3, 4 and 5 shows the results of the backstripping analysis and compares each reconstructed history with that of normal oceanic crust (Stein and Stein, 1992). In each case the normal oceanic subsidence curve was shifted to match the modern depth to basement in order to compare rates, and not the total depth of subsidence. It is rate of subsidence that is affected by the mantle thermal state. Absolute depth is influenced by crustal thickness and thus the degree of melting, which is partially controlled by mantle temperature, but also by the chemical state of the mantle, and the nature of upwelling under the hotspot (Mutter et al.). In the case of features emplaced off-axis a second oceanic subsidence curve is shown indicating the predicted rate of subsidence if the lithosphere had been totally reset to zero age at the time of eruption of the basement. Some of the drill sites show clear evidence for shallower than predicted depths early in their history, such as would be predicted for emplacement above hotter than normal mantle. Sites in the NE Atlantic, the Hatton Bank, SE Greenland and the Vøring Plateau and even SE Newfoundland show early anomalously shallow depths, i.e. steeper than expected subsidence curves, similar to that seen in modern Hawaii. Modeling by Clift (1997) indicated that the moderate nature of the depth anomaly was consistent with temperature anomalies of around +100˚C. Outside this region definitive evidence for shallower depths following eruption is scarce, as many of the sites fall within the predicted trend of crust overlying normal upper mantle temperatures. Where water depth uncertainties are large the possibility of hot mantle exists but is not required. There appears to be as many if not more sites where the reconstructed depth to basement after emplacement is shallower than predicted for even normal mantle temperatures. This raises the possibility that the asthenosphere under the sites has actually heated, not cooled since emplacement, though this is not considered likely.

A number of models have been proposed to account for the lack of subsidence in old hotspot regions. The relative lack of subsidence and initial submarine eruption of lavas in the Ontong Java and Manihiki Plateaus lead Ito and Clift (1998) to propose growth of these features in two or more stages. In practice they argued that the plateaus are inflated through time by underplating stretching from 120 to 90 Ma. In this case uplift caused by crustal growth counteracted the subsidence driven by lithospheric cooling. This model is practical in that setting because these features remained approximately stationary over the proposed plume source during that time. However, this model is not appropriate when applied to many other igneous provinces, which move rapidly relative to their source hotspots, e.g., NinetyEast Ridge. The MIT Guyot and the Mid Pacific Mountains are good examples of features that appear to represent excess melting, move rapidly within the hotspot reference frame after emplacement and which are not associated with a hotspot track. Their subsidence histories are slower than predicted for normal oceanic crust of their age, the opposite of that predicted for emplacement over hot asthenosphere. Long-term crustal accretion is not a reasonable explanation for the slow subsidence seen in this type of hotspot. Instead mantle processes must be inhibiting subsidence. Most likely the presence of a buoyant, depleted mantle residue from melting is acting as a cushion under these features. Mantle from which melt has been extracted is known to have a lower density than normal depleted upper mantle. If this material was not advected away from under the seamounts then this could slow subsidence (Robinson, 1988). Such a compositional-derived density model for hotspots was invoked by Phipps Morgan et al. for the Hawaiian swell and may be applicable to many Pacific and Indian Ocean hotspot features. What is unusual is that although Phipps Morgan et al. (1995) consider the depleted residue to be more viscous that normal asthenosphere, it is predicted to disperse over long periods of geologic time, resulting in a reduction in the height and width of the Hawaiian Swell. That prediction is at odds with the long-term lack of subsidence seen in many of the reconstructions presented here, which would require much of the buoyant residue to remain under the hotspot edifice until the present day, i.e. that it had a high viscosity.

The North Atlantic margins where subsidence anomalies do exist are consistent with mantle temperature anomalies of ~100˚C, but preclude very hot plume models. This conclusion is supported by inversion modeling of rare earth element data from hotspot volcanic rocks that predict temperature anomalies not exceeding 150˚C (e.g., Watson, 1993; Kent and McKenzie, 1994; Kerr, 1994). If any of the anomalous shallow depths seen after emplacement are caused by buoyancy resulting from mantle composition or dynamic mantle upwelling then the temperature anomalies inferred from the subsidence anomalies would be lower. The subsidence behavior of the North Atlantic margins and the Hawaiian seamount chain appears to differ from many other hotspots. This may be linked to a different origin that is linked to hot mantle that is not required in many other areas. Fast subsidence in these regions begs the question of why these provinces are not affected by a buoyant mantle residual root. Perhaps vigorous mantle circulation of the type proposed by Mutter et al. to increase melting was responsible for the removal of any residue that did form. There is little evidence to support the presence of a buoyant root in the NE Atlantic related to Tertiary magmatism. Temporary uplift of the NW European shelf away from the immediate vicinity of the volcanic margins during their formation is consistent with the moderate thermal anomaly derived from the oceanic crust subsidence histories (Clift et al. 1998), modern heatflow (Stein and Stein, 2003) and the modeling of volcanic rare earth element data (Kerr, 1994).

In summary, the sedimentary records of hotspot provinces worldwide frequently do not require hotter than normal mantle to be present at the time of magmatic crustal growth. In the few examples where hotter mantle is required the degree of the temperature anomaly is usually no more than ~100˚C and could be much less if dynamic flow and compositional buoyancy is an major factor in driving temporary uplift.