Geokinematics and Lithoplate Structure: Controls on Hotspot Origin and Evolution
December 2, 2003
Rex H. Pilger Jr.
Abstract
Whatever the origin of “hotspots”, the reference frames they define are shallow, based on several distinct lines of evidence: (1) Occurrence of minor hotspot chains in the Pacific and South Atlantic appears to be controlled by plate age (therefore, thickness) and sublithospheric heterogeneities. (2) Similarly cross-grain gravity lineations in the Pacific seem to be age-controlled. (3) Intracontinental stress fields are consistent with hotspot kinematic models.
There are three principal hotspot reference frames: Hawaiian, Tristan, and Icelandic, none of which is fixed relative to the spin axis. Within each reference frame, the hotspot traces imply little internal motion of the melting anomalies responsible for them, but the different reference frames (defined by internally stable sets of hotspots) move relative to one another.
The motions of plates and the hotspot reference frames imply that the continental plates encircling the Pacific Ocean control the motion of the sub-Pacific Hawaiian hotspot reference frame. The kinematic history of the reference frames and plates suggest that the reference frames can be thought of as subasthenospheric “mesoplates”. Like the well known near surface “lithoplates”, mesoplates do not internally deform to a significant extent.
The kinematic arguments for mesoplates are consistent with top-down mechanisms of plate tectonics. Gravitational instability of the lithosphere drives motion in the upper mantle, accommodated by displacement of mesoplates. Mesoplates are bounded by lithoplate subduction zones; other lateral boundaries are kinematically determined. Mesoplates feed spreading centers by vertical transfer across phase change boundaries
Heterogeneities in the upper mantle produce “hotspots” when sudden reductions in overburden pressure occur over a fertile region, due to extensional thinning of the lithosphere or passage of lithosphere of variable thickness. Heterogeneities could conceivably reflect mantle plumes, but there is no intrinsic reason, based on kinematic arguments, for invoking a plume mechanism. Other heterogeneities could represent ancient subduction zone remnants or abandoned spreading centers.
Introduction
The hard work of so many plate tectonic workers over the past four decades has delineated the structure and kinematic evolution of the plates since the mid-Mesozoic, including some measures of their relative kinematic rigidity. Over the same period of time, kinematic models of the plates relative to presumed hotspots and the limitations of these models have also been progressively refined. With this work in the background, the question of the origin of “hotspots” (particularly the mantle plume hypothesis) has become a vigorous debate, largely involving seismic tomography, thermal modeling, and petrological and geochemical evidence, but not so clearly involving plate kinematics and structure.
In addition to tomography and numerical modeling, plate kinematics and structure need to be considered in the hotspot-plume debate. Plate kinematics relative to hotspots (and/or plumes) was a key component of the original mantle plume proposal (Morgan, 1971, 1972), with deep mantle plumes defining a global reference frame. Recently plate kinematics has been recognized as relevant to the debate in an entirely different way. Similarly, plate structure has been largely ignored except in its controls on channeling of inferred plume flow, or, for those who prefer a non-plume origin, in the influence of plate structure on stress-induced fracturing and consequent volcanism. Other aspects of plate structure and kinematics, as outlined in this contribution, may have a more direct influence on the development of “hotspots”.
For the purposes of this paper, the term “hotspot” is used without genetic significance. For some, the term implies an underlying mantle plume; such implications are not to be inferred herein. Rather, “hotspots” are defined as contemporaneously or formerly active foci of anomalous igneous activity – magmatism that cannot be readily explained in the context of “normal” plate boundary processes (seafloor spreading, subduction-related volcanic arcs, or “leaky” transforms). “Hotspot traces” are curvilinear chains of intrusive and extrusive volcanic edifices inferred to be related to one (sometimes more than one) hotspot. Hotspots almost surely have their origin beneath the base of the lithosphere, but whether the underlying asthenosphere and mesosphere are anomalously hot is part of the debate. Some have proposed usage of the term “melting spot” instead. In the current paper’s usage, the “hot” refers to the anomalous magmatic center in the lithosphere and its implied, deeper melting spot, which may (or may not) be anomalously hot.
In order to address the relevance of plate structure and kinematics to origin of hotspots it is desirable to first better define the quality of hotspot kinematic models relative to data from the inferred hotspot traces. Then, in this context, the relation of the traces to plate structure might prove illuminating. Finally, a new and literally deeper kinematic conceptualization of plate tectonics and its relation to deeper earth structure emerges, with significant implications for the mantle plume hypothesis.
Plate-Hotspot Reference Frames
Evidence assembled, especially in the past few years, now demonstrates that the hotspot reference frames of the Pacific Ocean, of the Atlantic and Indian Oceans, and of the Iceland region are distinct from each other (e.g., Raymond et al., 2000; Norton, 2000; Gaina et al., 2000). Pilger (2003) has proposed names for each of the reference frames: Hawaiian (Pacific Ocean), Tristan (central North and South Atlantic and Indian Oceans, plus bordering continents), and Icelandic (northernmost North Atlantic and Arctic Oceans, Greenland, northeastern North America, and much of Eurasia). The concept of “absolute” motion as applied to the hotspots is no longer applicable. None of the three hotspot reference frames is preferred, and all, in general, may move or have moved relative to the earth’s spin axis and, presumably, the geomagnetic reference frame. In a nutshell, the evidence for the distinctiveness of each set of hotspots is their mutual inconsistencies when global relative plate reconstructions are calculated.
Molnar and Atwater (1973) and Molnar and Francheteau (1975) were first to show the apparent inconsistencies between the Pacific and Atlantic-Indian hotspots, despite barely adequate sets of relative plate reconstructions. As the quality of oceanic plate reconstructions grew, the discrepancy still persisted (e.g., Duncan, 1981; Pilger, 1982; Molnar and Stock, 1987). In order to preserve a global hotspot reference frame, deformation of Antarctica was invoked; evidence was assumed to be, unfortunately, buried under its thick icecap. Cande et al. (1999) broke the ice by undertaking a study of marine magnetic anomalies southeast of Australia, and south of the Antarctic-Australian-Pacific triple plate junction, demonstrating thereby only a minor amount of movement between east and west Antarctica in the mid to late Cenozoic (since ~43 Ma). The amount of deformation is inadequate to explain the discrepancy between Pacific Ocean and Atlantic-Indian Ocean hotspot sets (Raymond et al., 2000). Similarly, discrepancies between hotspot and paleomagnetic reference frames have been known since Morgan (1981) for the Atlantic continents, with more recent work by Tarduno et al. (2003) for the Pacific requiring movement of the Hawaiian hotspot relative to the geomagnetic poles.
In order to assess the internal quality of the Pacific and Tristan reference frames, there are several approaches that might be taken. Loci of plate motion relative to the hotspots can be constructed for each inferred hotspot trace and plotted in local plate coordinates along with dated sample locations. Distance versus age plots for isotopic ages and calculated loci for each trace can also be prepared. Either or both approaches have been taken by most workers who have presented reconstruction models for plate-hotspot motions (Pilger, 2003, provides such maps and plots for most documented hotspot traces). Alternatively, the samples can be restored to their inferred location in the hotspot frame for their measured isotopic (inferred cooling/crystallization) age (a modification of the “hot-spotting” technique of Wessel and Kroenke, 1997) using one or more reconstruction model sets (e.g., Koppers et al., 2003).
Trace Restoration
Restoration implies rotation of a sample location to its inferred position at the time of cooling and/or crystallization (depending upon the dating technique used). For this study rotation parameters are derived by interpolation from reconstruction parameters that describe plate motions in a particular hotspot reference frame; the basic reconstruction parameters are derived by interpolation and combination of plate-hotspot and plate-plate parameters at a regular interval (5 m.y.). The interpolation technique used is that of Pilger (2003): vector spline interpolation of total rotation poles (converted to Cartesian coordinates) with vector magnitudes equal to average rotation rate for each pole; for the present-day, instantaneous rotation parameters are applied. (Hanna and Chang, 2000, have developed a different spline interpolation method with splines applied to quaternion representations of rotations; the smoothed parameter locus is constrained to be close to the unit quaternion (4-D) hypersphere. Parker and Denham, 1979, attempted a similar numerical interpolation method as applied to unit vectors (therefore a 3-D sphere). Pilger’s, 2000, method produces parameterization of rotation poles confined exactly (within machine error) to the unit 3-D sphere since the magnitude of the splined pseudovectors is, by definition, equal to the rotation rate.)
The parameters of Raymond et al. (2000) for the Pacific plate (with corrections, Stock, 2003), Müller et al. (1993) (0-130 Ma) and Morgan (1983) (130-180 Ma; the finite difference parameters for this time interval are combined with Müller et al.’s 130 Ma reconstruction) for the central and northern African plates, and Gripp and Gordon (1991) for contemporary global kinematics are the basis of this analysis, supplemented by relative rotation parameters derived from the global isochron map of Müller et al. (1997) for Pacific-Nazca plates, African-Indian plates, and Antarctic-Central Indian plates, East-West Antarctica (Cande et al., 1999; Raymond et al., 2000), plus other relative motion parameters for North America-Africa (Klitgord and Schouten, 1986; Müller et al., 1990), South America-Central Africa (Cande et al., 1988) and the time scale of Cande and Kent (1995); the remaining relative motion parameters were derived from the hotspot model of Müller et al. (1993). The resulting plate-hotspot parameters, including modifications for the Pacific plate by Pilger (2003), are listed in Table 1.
Restoration of sample locations to the hotspot frame should produce a distinct pattern if the fixed hotspot model is correct. For a particular location along a trace, there could be several distinct ages (in the context of the fixed hotspot model): (1) Kinematic—the age at which time the location was located directly above the inferred melting spot in the underlying asthenosphere/mesosphere. The kinematic age cannot, in general, be dated directly, since we cannot sample the base of the lithosphere; there are only a few cases where such an age can be inferred. One example would be the location of a spreading center above a melting spot, producing “mirror-image” traces on either plate. Magnetic isochrons over the traces provide kinematic ages; the Tuamotu-Nazca ridges of the east-central Pacific fall in this category (e.g., Pilger and Handschumacher, 1981; this ridge set also has implications for the origin of the melting anomalies, as described below). (2) Initial emplacement—the age of initial intrusion of magma and extrusion onto the crustal surface of lava generated by the melting spot. (3) Terminal: Subsequent late extrusion of remaining lava from the intraplate magma chamber. The relative ages of the three stages should be Kinematic > Emplacement > Terminal. Conceivably, the three stages could range over several million years, depending on the size of the hotspot, plate velocity, persistence of the magmatic conduit from source to magma chamber, and extrusion history from the magma chamber.
Assuming that a period of a few million years would be recorded by the magmatic sequence, restoration of dated (by isotopic methods or reconstructed magnetic isochrons) samples from each stage should produce an elliptical pattern (elliptical because of minor departures from the inferred locus), elongate in the direction of plate motion at the time of emplacement. The hotspot location should be at the “older” end of the ellipse (that is, oldest ages from a particular location along the trace should have moved farther between emplacement and present location). Figure 1 illustrates this concept.
Figure 1a. Hypothetical hotspot trace, map view; trace includes an apparent distinct change in plate motion relative to hotspot reference frame as well as slight perturbations in location.
Figure 1b. Expected age-distance plot of hypothetical hotspot trace, with two distinct age-distance trends and persistence of activity after volcanic inception.
Figure 1c. Reconstructed data points, rotated to time of origin for hypothetical hotspot trace. Note two elongate “elliptical” clusters oriented in direction of plate motion at time of “cooling”. The scatter is a result of minor scatter in location as well as in age.
There are several possibilities that could produce unsatisfactory reconstructions relative to the hotspot model: erroneous isotopic dates, reconstruction parameters, or geomagnetic time-scale calibration points, incorrect interpolation methodology, departures from plate rigidity, relative movement among hotspots, or non-fixed hotspot origin of the inferred traces. In some cases, hotspot trace Ar/Ar dates from the Tristan hotspot set have been reinterpreted or even rejected (Baksi, 1999), allowing for tests based on a filtered subset of recalculated dates. Errors in the geomagnetic time-scale might be indicated if systematic discrepancies are apparent in a variety of reconstructed traces. Multiple clusters of dates from a single trace might indicate a non-hotspot origin.Figs. 2-9 show data point restorations for several regions: South-central Pacific, Hawaii, Gulf of Alaska, central Indian Ocean, Tristan da Cunha, Great Meteor, and Trinidade/Northeastern South America on equatorial equidistant maps relative to either the Hawaiian or Tristan reference frame. Included are magnetic isochrons of Müller et al. (1997) and, for the Pacific plate, the seamounts of Wessel and Lyons (1997). Because of the different age ranges in each area, symbol conventions vary among the figures. The hotspot data set compiled by Pilger (2003) is used for the restorations; parameters for the restoration are spline-interpolated from Table 1.
Figure 2. Hotspot traces of the east-central Pacific: Magnetic isochrons, seamounts (open diamonds), hotspot trace data points (x), calculated loci anchored at suggested hotspots (lines incremented every five m.y.), restored hotspot trace data points (blue circles: 0-10 Ma, light blue squares: 10-19 Ma; green diamonds: 19 to 29 Ma; light green up triangles: 29 to 38 Ma; light yellow down triangles: 38-48 Ma; yellow left triangles: 48-57 Ma; orange right triangles: 57-67 Ma; red top semicircles: 67-76 Ma; brown bottom semicircles: 76-86 Ma; lavender right semicircles: 86-96 Ma). Note clusters of restored points near inferred hotspots (Easter, Marquesas, Society, Macdonald, Rapa, Foundation) as well as clustering along Austral-Cook trace). Many unrotated data points are not shown as they are outside the mapped area. Latitude-longitude graticules (+) at 15°.
The Easter, Society, Marquesas, Foundation, and Pitcairn traces show elliptical clusters when data points are restored, as expected by the hotspot hypothesis (Fig. 2). In some cases, the inferred hotspot has been relocated to within the younger (usually southeastern) part of the cluster. The Macdonald hotspot shows some clustering; however, there is a scatter of other clusters from southeast of Macdonald seamount well to the west along and to the north of the hotspot locus. Foundation seamounts show a secondary cluster parallel with and to the north of the locus. A few scattered restored points (between 47 and 77 Ma) are derived from the Line Islands (there are older data points available from the Line Islands and other regions of the Pacific, beyond the age of the kinematic model).
Figure 3. Hawaiian-Emperor trace: Seamounts (open diamonds), hotspot trace data points (x), calculated locus anchored at suggested hotspots (lines incremented every five m.y.), restored hotspot trace data points (blue circles: 0-14 Ma, green squares: 14-28 Ma; yellow diamonds: 28 to 42 Ma; orange up triangles: 57 to 71 Ma; red down triangles: 57 to 71 Ma). Most unrotated data points are not shown as they are outside the mapped area. Latitude-longitude graticules (+) at 5°.
For the Hawaiian hotspot (Figure 3), note clusters of restored points near inferred hotspot; red/orange cluster is of data points older than the Hawaiian-Emperor bend; blue/green/yellow points are younger. Orientation of clusters is as expected, parallel with the modeled plate motion, although the older cluster is offset slightly from the older (southeasterly) end of younger cluster. Either model parameters are slightly in error, or there is slight relative motion between the Hawaiian-Emperor and Louisville traces, which are the principal controls on the model (Raymond et al., 2000).
Figure 4. Hotspot traces of the Gulf of Alaska: Seamounts of Pacific and Juan de Fuca plates (open diamonds), hotspot trace data points (x), calculated loci anchored at suggested hotspots (lines incremented every five m.y.), restored hotspot trace data points (blue circles: 0-19 Ma, green squares: 19-38 Ma; yellow diamonds: 38 to 57 Ma; red triangles: 57 to 76 Ma). Most unrotated data points are not shown as they are outside the mapped area. Latitude-longitude graticules (+) at 5°.
Traditionally, two hotspot traces have been inferred in the Gulf of Alaska (Fig. 4). Note clusters of restored points near inferred hotspots (Cobb and Vancouver Island) as well as separate clusters between and along the trace loci); at a minimum, an additional hotspot could be inferred between the two others. Aligned restored points represent ranges of dates from the same location.
Figure 5. Louisville trace: Seamounts (open diamonds), hotspot trace data points (x), calculated locus anchored at suggested hotspot (lines incremented every five m.y.), restored hotspot trace data points (blue circles: 0-13 Ma, green squares: 13-25 Ma; yellow diamonds: 25 to 38 Ma; orange up triangles: 38 to 50 Ma; red down triangles: 50 to 63 Ma; brown left triangles: 63 to 75 Ma). Most unrotated data points are not shown as they are outside the mapped area. Latitude-longitude graticules (+) at 5°.
The Louisville trace has the second largest range of dates from the Pacific plate (after Hawaii). Two outliers, to the southeast and south of the main cluster are apparent. Within the main cluster, two ellipses can be inferred, representing younger and older (more northerly trend) date ranges, as expected.
Figure 6. Hotspot traces of the central Indian Ocean (Reunion and Kerguelen): Hotspot trace data points (x), calculated loci anchored at suggested hotspots (lines incremented every five m.y.), restored hotspot trace data points (small symbols: original dates, large: dates filtered and recalculated by Baksi (1998); blue circles: 0-25 Ma, green squares: 25-50 Ma; yellow diamonds: 50-75 Ma; orange up-triangles: 75-100 Ma; red down-triangles: 100-125 Ma). Dashed lines connect equivalent data points with age range from Baksi. Most unrotated data points are not shown as they are outside the mapped area. Latitude-longitude graticules(+) at 15°.
For the two principal hotspots (Reunion and Kerguelen) of the Indian Ocean, note clusters of restored points near the inferred hotspots (Fig. 6). Also note a few outliers among both the original and Baksi’s (1999) data sets. A 52 Ma age (recalculated by Baksi) from the Laccadive ridge is restored well to the south of Reunion. Similarly, a sample from Ninetyeast Ridge, inferred to have an age range of 80 to 100 Ma is restored well to the south of Kerguelen. However, the rest of the recalculated ages (including the younger age for those with a range) show fair clustering around the two inferred hotspots.
For the Tristan hotspot (Fig. 7), there are clusters of restored points less than 100 Ma, mostly from the African plate, near inferred hotspot, including both original and Baksi’s (1999) filtered data. Most of the older data points (greater than 125 Ma) are from Parana and Etendeka flood basalt fields. The wide geographic dispersion of the older flood basalt data points and tighter clustering of younger data could be cited as support for the plume head and tail hypothesis. Alternatively, this could reflect initiation of magmatism as a response to focused crustal extension, with the subsequent volcanism representing residual hotspots.
Figure 8. Central North Atlantic, Great Meteor Seamount: Calculated loci (relative to African plate, extending to the east, north, and then west, and North American plate, extending to the northwest) anchored at suggested hotspot (“H”: 30°N, 28.5°W; lines incremented every five m.y.), restored hotspot trace data points from North American plate (small symbols: original dates, large: filtered and reinterpreted dates by Baksi (1998); blue circles: 0-25 Ma, green squares: 25-50 Ma; yellow diamonds: 50-75 Ma; orange up-triangles: 75-100 Ma; red down-triangles: 100-125 Ma; brown left-triangles: 125-150 Ma). No unrotated hotspot trace data points are shown as they are all to the west of the mapped region. Latitude-longitude graticules (+) at 5°.
The restored data around Great Meteor Seamount (Figure 8) are derived from the New England Seamounts, White Mountain magma series, and Monteregian Hills intrusives – all from the North American plate. Note clusters of restored original data points to the southeast of the inferred hotspot (implying motion of the hotspot) and, in contrast, better correspondence of Baksi’s (1999) filtered data (only two points; minimal hotspot motion is required). The older data points (~130 Ma) are ~300 km from the inferred hotspot. Because some of the original dates (between 75 and 100 Ma) come from the New England seamounts, whose trend was used to derive the reconstruction model (Muller et al., 1993), the apparent discrepancies are among the more disturbing for a fixed hotspot reference frame. However, Baksi’s rejection of these data points as unreliable could be invoked. The discrepancies among the older ages (from White Mountain magma series and Monteregian Hills) could indicate a need for Müller et al.’s (1993) model to be modified, since only a few traces greater than 100 Ma exist within the Atlantic-Indian Ocean set. 
Figure 9. Trinidade – Western South Atlantic: Unrotated hotspot trace data points (x), calculated locus (relative to South America plate, extending to the west) anchored at inferred hotspot (21°S, 28.5°W; lines incremented every five m.y.), restored hotspot trace data points; green squares: 25-50 Ma; yellow diamonds: 50-75 Ma; orange up-triangles: 75-100 Ma; red down-triangles: 100-125 Ma; brown left-triangles: 125-150 Ma). Not all unrotated hotspot trace data points are shown as some are to the west of the mapped region. Latitude-longitude graticules(+) at 5°. Note that no age data are available from oceanic portion of inferred trace.
For Trinidade and onshore eastern South America (Fig. 9), a number of Cretaceous and early Cenozoic intrusives (largely dated by K/Ar) show a broad dispersion, and the calculated locus of the Trinidade hotspot shows virtually no correspondence with most of the dated samples. As would be expected, then, the oldest points are restored well to the southeast of the inferred Trinidade hotspot. No data are available from the offshore ridge associated with the postulated hotspot. Like the volcanics presumed to be precursors of Tristan da Cunha (Fig. 7), the older continental volcanics occur over a wide area with much narrower (and younger) apparent loci occurring on oceanic lithoplate.Regional Precursors, Outliers, and Overprints
While, as noted above, there a few outliers among the restored data points for the five areas studied which require explanation, additionally there is the question of the origin of the clusters of younger data points along the calculated loci in the east-central Pacific and Gulf of Alaska. The latter problem is addressed first.
The en echelon faulting mechanism proposed by Herron (1973) could well be as an explanation for the multiple and over-printed chains of the east-central Pacific and Gulf of Alaska, an explanation that has appealed to a number of workers for hotspots, in general (e.g., Solomon and Sleep, 1974, Pilger and Handschumacher, 1981, Winterer, 2003; Favela and Anderson, 2000). The problem with intraplate extension as a generalized mechanism for the formation of hotspots is the existence of “mirror-image” hotspot traces, especially the Tuamotu and Nazca ridges, inferred to have formed from the same (Easter) hotspot (Morgan, 1972; Pilger and Handschumacher, 1981). It is difficult to see how intraplate extension could form a hotspot beneath a ridge. Additional structural controls and/or sublithospheric heterogeneities probably also need to be considered.
Pilger (2003) has shown that in both regions (east-central Pacific and Gulf of Alaska) each of the minor hotspot traces (except for the Marquesas) originates in a common structural environment: on the south side of a fracture zone that separates older plate on the north from younger on the south (Fig. 10). As the Pacific plate moves to the northwest, asthenosphere and lithosphere previously beneath older (and therefore, thicker) lithosphere is subsequently overlain by younger, thinner plate. Consequent isostatic rise of asthenosphere and mesosphere could result in enhanced partial melting, producing plate-penetrating magmatism. The pattern could reassert itself, depending on the fertility of the asthenosphere-mesosphere, resulting in production of numerous, and sometimes overprinting, volcanic chains.
Figure 10. Island-seamount chains of the South Pacific: Dated sample locations (x) (compiled by Pilger, 2003), calculated Pacific (or Nazca)-Hotspot loci, circles at 5 m.y. intervals (parameters in Table 1), magnetic isochrons (Müller et al., 1997), bathymetry (lavender-shallow, green- deep) (Smith and Sandwell, 1997).
The Hawaiian island-seamount chain shows an apparent, similar, structurally-controlled eruption pattern, recognized by Phipps Morgan et al. (1995). Reduced volcanic volumes are observed to the southeast of younger (and thinner)-on-north fracture zones while increased volumes are observed to the southeast of older-on-north fracture zones. Phipps Morgan et al. interpreted the pattern as representing variations in distance between the top of the plume inferred to feed the Hawaiian hotspot and the lithosphere. However, the pattern could also be interpreted as a similar isostatic effect to that proposed for the minor hotspots, whatever the origin of the hotspot. Passage of thinner plate over the asthenosphere and mesosphere enhances melting; thicker plate retards melting. [Note: Raddick et al., 2002, JGR, 107, doi: 10.1029/2001JB000617., provide an extended development of this idea, not referenced in the initial version of this manuscript.]Wessel (1993) has observed that the segment of the Hawaiian swell between the Murray and Molokai Fracture zones, which is underlain by relatively younger lithosphere than adjacent segments to the northwest and southeast, is anomalously shallow. The mechanism proposed by Pilger (2003) for explaining enhanced volcanism by variations in lithospheric thickness could also explain the enhanced elevation of the Hawaiian swell – the presence of greater amounts of partial melt in the asthenosphere due to thinner lithosphere adds elevation to the swell.
What about the outliers – restored data points that are “too old”. Again, restored data points that are on the young, downstream side of the inferred hotspot, could be interpreted as terminal volcanic episodes, representing the last eruptions from the magma chamber introduced by the hotspot. Older data points are more difficult to explain in the context of the hotspot hypothesis. Within Baksi’s (1999) filtered and recalculated data points, there are only two such “too old” outliers; one relative to Reunion and the other relative to Kerguelen. If the additional original data points are added, a large set of such outliers is found near Great Meteor and a few more near Reunion and Kerguelen.
Two alternative hypotheses to regionally fixed hotspots, origin by propagating fractures or relative movement of the hotspot in the direction of plate movement, present their own challenges. Fracturing is difficult to assess; what pattern might be expected? What about reactivation? Some hotspot traces show age propagation with no reactivation. Others have apparent reactivation, but not necessarily along the same volcanic trends (e.g., Austral-Cook). The older dates from the Great Meteor trace come from the diffuse White Mountain magma series and Monteregian Hills intrusives, which do not appear to be related to a through-going fault or fracture zone.
Movement of the hotspot itself in the direction of plate motion is even more difficult to rationalize. What mechanism might be responsible? Hot lines, as advocated for some traces, and associated with inferred longitudinal convection, have yet to be independently documented.
Of course, one could question the hotspot reconstruction parameters themselves. However, the two Indian Ocean outliers correspond with ages in other oceans that fit the model in the North and South Atlantic. Is there still another possibility?
The broad zone of magmatism that seems to be a precursor to hotspot traces has, of course, been a primary argument in favor of plume heads. The restored dates from the Parana and Edenteka basalts cover a wide area. However, its there an alternative, non-plume model that could explain them and their presumably associated traces?
Generalization…
The minor hotspots of the Pacific appear to originate adjacent to fracture zones separating thick and thin lithosphere. The volumetric variation in volcanism of the Hawaiian trace appears to be controlled by plate thickness as well. Could these apparent controls be generalized to all hotspot traces, including outliers and precursors, as well as overprints?
The loci of intrusion of the Parana volcanics is offset to the east of the subsequent rift that developed between South America and Africa. Models of continental rifting suggest that low-angle, crust-penetrating normal faults may be involved. What if such faults further penetrated the lithosphere, resulting in apparent lithospheric thinning (Fig. 11)? In such a case, enhanced melting of consequently uplifted underlying asthenosphere and mesosphere might well occur, again controlled by relative fertility of the source material. The remnant(s) of such a mechanism could be subsequent hotspot trace(s).
The Deccan Traps occupy a similar structural and geohistorical setting to the Parana volcanics and subsequent rifts. The Traps, too, could represent a similar rift-controlled phenomenon. Other continental flood basalts are also proximal to or within zones of crustal extension: Rajmahal, Broken Ridge, Columbia River, and, possibly, Siberia.
The continental intrusives and volcanics associated with the inferred Great Meteor and Trinidade traces (Figs. 8 and 9) are problematic, as they postdate rifting. The locus of the Trinidade trace may extend into the Jurassic rift zone between North and South America (combining Morgan’s, 1983, pre-130 Ma parameters with Müller et al.’s, 1993, 130 Ma parameters). However, the reconstructed Great Meteor trace apparently extends into the Canadian Shield. Further, the restored White Mountain and Monteregian data points are still offset well to the southeast of the inferred hotspot (Fig. 8).
The minor hotspot traces of the South Atlantic occur in structural positions analogous to those of the Pacific, except that instead of reflecting older-on-north African plate fracture zones, the reconstructed traces indicate passage of asthenosphere originally beneath older South American plate under fracture zones to beneath younger African plate (Fig. 10). Thus, an analogous isostatic mechanism to that inferred for the Pacific is operable.
Fig. 12. Northeastern South Atlantic: Free-air gravity (Sandwell and Smith, 1997); magnetic isochrons with age in Ma italicized (Müller et al., 1995); reconstructed isochrons in Atlantic-Indian hotspot frame with age in Ma (Pilger, 2003); locus of possible hotspot relative to Central African plate, 0-130 Ma, anchored at 17°S, 10°W, 5 m.y. increments; dated volcanic locations (x’s); restored hotspot data points: blue circles: 0-16 Ma, green squares: 16-32 Ma; yellow diamonds: 32-49Ma; orange up-triangles: 49-65 Ma; red down-triangles: 65-81 Ma). Note relative northward movement of Africa in hotspot frame between 90 and 75 Ma (shown both with locus and reconstructed isochrons); as consequence, African plate near locus passes over asthenosphere originally located beneath older South American plate.
The reconstructed outliers of the Indian Ocean (Fig. 6), whether the original data points, or the smaller recalculated subset (Baksi, 1999), are also problematic. However, it is interesting to note that all of the outliers are from aseismic ridges adjacent to and subparallel with fracture zones of large offset. Thus, slight variations in the azimuth of movement of the Indian (or Central Indian) plate relative to the fracture zone azimuth could produce the thick-to-thin plate effect, with subsequent overprinting by the more northerly major hotspots that produced the Deccan and Rajmahal Traps. The outliers could in effect represent multiple aligned hotspots produced by the fracture zones. Like the minor traces of the Pacific, such multiple overprinted hotspots need not persist to the present. There is, of course, the peculiar circumstance of primary hotspots produced by crustal extension overprinting minor hotspots produced along major fracture zones. However, the geometry of rifting of the Indian Ocean implies some correlation between the configuration of the rifted continental margins and the subsequently formed transform faults.To summarize, lithospheric thickness variations, combined with mesospheric heterogeneities, may well provide the fundamental explanation for hotspot traces. Further, available data in this context are consistent with the inference of two (or three, counting the Icelandic) hotspot reference frames. Mantle plumes do not appear to be a required component of this construct. Evidence for shallowness of the reference frames rules out a global reference frame, from mass and volume considerations – subduction zones inhibit communication between shallow mesosphere beneath the plates of the Pacific Ocean and surrounding Atlantic and Indian Ocean plates. Thus, inconsistency of the two parameterized hotspot reference frames is not surprising. Further implications of the three hotspot reference frames need to be more fully considered.
More Shallow Evidence
The patterns of magmatism implied by structural controls on hotspot traces is essential evidence for the shallowness of the hotspot reference frames. Additional evidence for the shallowness of the reference frames is also available.
Stress Fields
Pilger (2003) has shown that intracontinental paleostress indicators from North America and Africa are consistent with motion directions in the Tristan hotspot reference frame (derived from Müller et al., 1993). Like contemporary intracontinental stress indicators (e.g., Zoback et al., 1989), the maximum principal horizontal compressive stress parallels the direction of plate motion in the reference frame for the past 130 m.y. (North America) and 80 m.y. (Central African plate).
Paleostress indicators from Western Europe, however, are not consistent with the Tristan hotspot model of Müller et al. (1993), when extended to Eurasia (Pilger, 2003). The hotspot model also fails to fit the Icelandic hotspot traces on Eurasian and Greenland-North American plates (Pilger, 2003). Since Eurasia is separated from the Pacific, African, Arabian, Indian, and Australian plates by a major zone of convergence, including subduction zones, by analogy with the relation of the Hawaiian and Tristan reference frames, it seems reasonable to postulate a third reference frame beneath Eurasia, the northernmost Atlantic and Arctic Oceans, Greenland, and, perhaps, northeasternmost North America. Norton (2000) has suggested a third such reference frame, recorded in the Icelandic hotspot traces in the Norwegian-Greenland Sea. Perhaps a combination of the Icelandic traces and Eurasian paleostress indicators could be used to refine Norton’s parameters describing plate motions in the Icelandic frame.
Cross-Grain Gravity Lineations
Another line of evidence that is indicative of shallowness of the Hawaiian hotspot reference frame, is the present of cross-grain gravity lineations in the eastern Pacific (Fig. 13; sources: gravity, Sandwell and Smith, 1997 isochrons, Müller et al., 1997 flowlines, calculated from Gripp and Gordon, 1991; loci calculated from parameters of Raymond et al., 2000, using method of and modifications by Pilger, 2003). The gravity lineations, first recognized by Haxby and Weissel (1986), have been subjected to significant subsequent study and speculation. What is perhaps most distinctive is the recognition that the lineations, while cutting across the seafloor spreading fabric of the region, are parallel with the direction of Pacific plate motion in the Hawaiian reference frame, as Haxby and Weissel observed. Wessel et al. (1996) have applied significant analysis to the trends of the lineations and concur that they are consistent with the contemporary direction of plate motion in the hotspot frame.
The origin of the lineations remains debatable. Haxby and Weissel (1986) suggested longitudinal convection cells, after the model of Richter (1973). A number of workers have suggested plate extension, with the lineations representing plate-scale boudinage or plate fracturing with associated magmatism forming dikes (e.g., Winterer and Sandwell, 1987; Wessel et al., 1996).
What has not been noted, however, is the geographic restriction of the gravity lineations to settings comparable to those of the minor island-seamount chains of the eastern Pacific. That is, the lineations occur within segments of seafloor located on the south side of older-on-north fracture zones (Fig. 13). Only a few apparent isolated lineations occur in the opposite structural setting; for example, south of Hawaii; there the lineation is probably a manifestation of the Hawaiian flexural moat and rise. Thus the thick-to-thin mechanism that appears to contribute to the formation of the minor island-seamount chains also may contribute to the process responsible for the formation of the gravity lineations. Whatever the origin of the lineations, their characteristic wavelengths imply a shallow origin (e.g., McAdoo and Sandwell, 1989; Wessel et al., 1996). Parallelism with the hotspot reference frame conversely implies a shallow depth for the Hawaiian hotspot reference frame.
East African Rift Volcanism
Turcotte and Oxburgh (1978) recognized a pattern of progressive inception of Cenozoic volcanism, from north to south, in East Africa, which they interpreted in terms of membrane tectonics. When available data are assembled and plotted, this pattern of inception can be shown to match the rate of African plate motion in the Atlantic-Indian Ocean reference frame (Fig. 14; Pilger, 2003). In general, volcanism postdates earliest rifting.
Figure 14. K/Ar ages of volcanic rocks from East Africa plotted against Latitude, together with arbitrarily located hypothetical hotspot (6°S, 30°W) locus in Atlantic-Indian hotspot reference frame (data sources in Pilger, 2003). Note parallelism of locus with onset of magmatism.
Parallelism of the locus of inception with volcanic inception (in latitudinal view) implies that the volcanism cannot be attributed to a spreading plume head; in such a case, onset of volcanism would be expected to occur much more rapidly. There appear to be two other possibilities: (1) Volcanism represents the point at which lithospheric extension has thinned the plate enough for depressurization melting to begin; therefore the onset of extension propagates at the same rate as plate motion in the hotspot reference frame. (2) Crustal extension, combined with a sublithospheric inhomogeneity, induces depressurization melting in the fertile region; therefore the progressive onset of volcanism marks the moving intersection of thinned African lithosphere with the southern boundary of the inferred zone of fertile mantle (within the hotspot reference frame).Reference Frames Equal Mesoplates
“Rigid plates” is the remarkable approximation that characterizes plate tectonics. The rigidity is kinematic and geometric to the extent that plate interiors do not deform significantly; there are exceptions, but they are well known from either diffuse seismicity or measurable deformation. From a dynamic perspective, plate rigidity implies that shear stresses are not large enough to deform the plates; where stresses achieve a magnitude sufficient to deform a plate, the plate usually fractures, creating two or more new (smaller) plates. There are limitations to kinematic rigidity, as reviewed by Gordon (2000), but the approximation is still applicable to a significant collective area of contemporary lithoplates.
In the same context as lithospheric plates, the three hotspot reference frames can be thought of as kinematic plates. That is, the minimal apparent movement between hotspots implies little internal deformation, allowing for their characterization as reference frames. Thus, Pilger (2003) has proposed that the three reference frames be termed “mesoplates” of the same name (Hawaiian, Tristan, Icelandic; Fig. 15). Their upper surfaces correspond with the upper surface of the mesosphere (the deep solidus that forms the lower surface of the asthenosphere) and the lower surface is probably no deeper than the 660 km discontinuity (it may correspond with the 410 km discontinuity), mesoplates are inferred to extend over larger areas than do plates.
Figure 15. Approximate boundaries between the three major mesoplates, Hawaiian, Tristan, and Icelandic (after Pilger, 2003).
The divergent boundary between lithospheric plates (“lithoplates”), spreading centers, is fed by asthenosphere and vertically rising mesosphere, but no corresponding deeper boundary in the underlying mesoplate exists (Fig. 16). Part of the boundaries between mesoplates is occupied by descending plates in subduction zones. The remaining boundaries between mesoplates are inferred to largely be kinematically determined, conceivably consisting of changing small circles around the instantaneous pole of motion between the mesoplates, as well as zones of divergent or convergent (with the latter two generally corresponding with subduction boundaries). Divergent mesoplate boundaries corresponding with convergent plate boundaries may seem contradictory. However, the kinematics of the Hawaiian and Tristan mesoplates imply divergence accommodated somewhere between the Cordilleran subduction zone and established Tristan mesoplate in the latest Cretaceous and Early Cenozoic; paleostresses in the western continental interior have a Tristan orientation during this period of time. Therefore, the extensional zone between the two mesoplates must be beneath the Cordillera. Oddly, this divergent zone corresponds with the inferred low-angle subduction zone of the Laramide event (e.g., Cross and Pilger, 1978).
Figure 16. Cartoon illustrating mesoplate and lithoplate interaction in the early Cenozoic, North American, Kula, Farallon, and Pacific lithoplates and Hawaiian and Tristan mesoplates, looking to northeast. Tristan mesoplate is arbitrarily fixed. Arrows indicate motion relative to Tristan.
A secondary argument for the mesoplate concept (at least in terms of the separate reference frames) was recognized by Pilger (2003): Derivation of the relative motion of the Hawaiian and Tristan mesoplates implies a strong correspondence with the motion of North and South America in the Tristan reference frame. Particularly in the latest Cretaceous through the early Cenozoic, the two Americas moved in the same direction at nearly the same rate as the Hawaiian mesoplate, all relative to the Tristan mesoplate. North America, in particular, appears to be nearly fixed relative to the Hawaiian mesoplate during this time interval. As a side effect, the Hawaiian-Emperor bend is paralleled by the locus of Pacific plate motion relative to North America (Fig. 17), a paradox recognized by Norton (2000). The mesoplate hypothesis provides a mechanism for understanding this correspondence. One of the objections to the hotspot hypothesis has been the near-absence of a significant change in motion of the Pacific plate relative to adjacent oceanic plates corresponding with the Hawaiian-Emperor bend (e.g., Norton, 1995). As Norton (2000) observes, the Pacific plate does show a change in motion relative to the North American and South American plate close in age to the bend. In effect, North America, South American, and the Hawaiian mesoplate all changed their motion, producing the bend, without affecting the motion of the Pacific or Farallon plates. Thus the absence of a change of motion among the Pacific Ocean plates at the time of the bend does not contradict the hotspot hypothesis (other than the hotspot reference frame itself is moving).
Figure 17. Loci of relative motion of four lithoplates (North and South American, Eurasian, and Australian) and two mesoplates (Hawaiian and Tristan) relative to the Pacific plate. Hawaiian, almost by definition, corresponds with the Hawaiian-Emperor island-seamount chain. Note similarity in shape of North American-Pacific and South American-Pacific to Hawaiian-Pacific.
Expansion of the Atlantic Ocean requires contraction of the Pacific Ocean, a phenomenon recognized by Wilson (1966) in the nascent plate tectonic period. As North America advances on the Pacific, the subduction zones that swallow the Kula and Farallon plates are also displaced in front of the advancing continent. As a consequence, the shallow mesosphere beneath the Pacific plates is displaced to the southwest by the shrinking northern margin of the Pacific Ocean Basin (Fig. 18).
Figure 18. Reconstructions of the boundaries between the Tristan and Hawaiian mesoplates for past 90 m.y. at 10 m.y. intervals, along margin of western North America. Solid: North America relative Tristan. Dashed: Hawaiian relative to Tristan. Note partial parallelism of reconstructed boundaries between 90 and 30 Ma and especially between 70 and 30 Ma.
The advancing subduction zone effect gradually is lost as progressively younger plate is subducted beneath the Americas (resulting in progressively shallower subduction; Cross and Pilger, 1982), until subduction actually ceases over much of the margin with the encounter of North America with the East Pacific Rise and the Pacific plate. Thus, after about 25 Ma, the coupling of North America with the Hawaiian mesoplate is largely lost. In fact, part of the Hawaiian mesoplate may now exist beneath the western United States. Pilger (2003) has shown that the Yellowstone-Snake River Plain hotspot trace corresponds with the locus of North American plate motion relative to the Hawaiian reference frame for the past ~25 m.y.Refining Mesoplate Boundaries
The preliminarily-defined boundaries between mesoplates (Fig. 15) were made by inferential leaps and intervening guesses. Could the near kinematically rigid regions (mesoplates) be much smaller, with wider and more diffuse zones of deformation between mesoplates?
Clearly, if we rely on the limits of inferred hotspots which comprise an apparent stable reference frame, the resulting boundaries of mesoplates define much smaller areas than initially inferred. However, the other indicators of motion of lithoplates relative to mesoplates provide additional constraints. Fig. 19 shows the originally proposed mesoplate boundaries (Pilger, 2003), hotspots with documented Cenozoic traces, other proposed hotspots (UTIG, 2003), and coarse hulls around the documented hotspots plus mesoplate-interaction indicators (contemporary stresses, recent paleostresses, and cross-grain gravity anomalies) for the three mesoplates. Note that the coarse hulls are based on the hotspot locations, not their traces. A trace formed above one mesoplate could, in theory, presently exist above another mesoplate. The coarse hull for the Icelandic mesoplate is the smallest, based on Iceland and the region of documented regionally-consistent (north-northwest-south-southeast) compressive stresses in Western Europe; few intraplate stress indicators (except for earthquake focal mechanisms) are available for the rest of Eurasia.
Figure 19. Hotspots with document Cenozoic traces: blue circles; other proposed hotspots (UTIG, 2003): red circles. Coarse hulls (solid lines) around hotspots, stress indicators, and cross-grain gravity anomalies for mesoplates -- Hawaiian: green; Tristan: teal, and Icelandic: gold. Original mesoplate boundaries (Pilger, 2003): dash-dot line. Modified mesoplate boundaries (this paper): dotted line.
In the context of evidence for mesoplates there could, indeed, be non-kinematic rigidity in the regions between the coarse hulls. On the other hand, there is no evidence either way, primarily because hotspots are missing and stress indicators are either missing or ambiguous. Modified mesoplate boundaries, based on the coarse hulls and the presumption of correspondence with deep (below 200 km) subduction boundaries, are also indicated on Fig. 19, as a basis for further analysis. However, it is interesting to note that the continental portion of Australian plate, unlike many other continents, does not show stress fields that correspond with the hotspot motion model (e.g., Zoback et al., 1989), while the deformation of the Indian plate (e.g., Gordon, 2000), is the largest compressional deviation from kinematic rigidity of any lithoplate; could the boundary between Tristan and Hawaiian mesoplates be beneath Australia, and the boundary between Tristan and Icelandic mesoplates be beneath the northern Indian plate? There is no evidence of a contemporary deep subduction zone beneath the Himalayas and/or Tibet to constrain the latter boundary.Mesoplate Objections
There could be objections raised to the introduction of this new term, “mesoplates”, into the plate tectonic construct. Traditionally, the upper mantle beneath the lithosphere has been assumed to be actively convecting. Convection is assumed to be driven by the plates themselves, especially subduction, as well as lateral and vertical variations in temperature. To expect a kind of kinematic rigidity in this environment is counterintuitive.
Further, there is the definition of “rigidity”. From the earliest days of plate tectonics, the idea of plate rigidity encountered a kind of geoscientific rigidity. Many earth scientists argued that the lithosphere is “weak”, not rigid. The empirical response to this objection is that plates behave in a kinematically (and dynamically) rigid manner, because shear stresses within plates do not achieve great enough magnitudes to significantly deform the plates. Where high enough shear stress levels are achieved, plates do indeed deform; they fracture, forming two or more new (and smaller) plates.
The same type of empirical evidence used to define lithoplates can be seen to apply to mesoplates. The two hotspot reference frames imply minimal relative movement among the hotspots in each system. Just as lithoplates do not show significant internal deformation, for the most part, so mesoplates are inferred to demonstrate little internal deformation.
Mesoplates and Plumes
The concepts of deep mantle plumes, fixed hotspots, and absolute motion are tightly bound in much of the ongoing debate on the origin of anomalous volcanism and mantle convection. Ironically, however, a number of workers who focus on plume and convection modeling argue that hotspots cannot form a fixed reference frame; their modeling produces too much intra-hotspot displacement (see references in Duncan and Richards, 1991, and Steinberger and O’Connell, 2000). Conversely, with the emergence of evidence of displacement of the Hawaiian hotspot relative to the paleomagnetic pole (e.g., Tarduno et al., 2003) can be cited as evidence against not only “absolute motion” but, because of their original coupling by Morgan (1971, 1972), against mantle plumes as well.
While not explicitly developed by Morgan (1971, 1972), there is an implicit argument in support of mantle plumes based upon “absolute motion”. That is, geometric considerations imply that a reference frame shallower than the deepest extent of subduction zones cannot exist. Subduction zones not only serve as a means of returning oceanic lithosphere into the mesosphere, as an accommodation to seafloor spreading, they also serve as boundaries between shallow mesosphere on either flank of each subduction zone. As the upper plates of subduction zones move towards the plate boundary, shallow mesosphere beneath the ocean plate must be displaced, particularly if the global budget of plate motions is considered. Enlargement of the Atlantic and Indian Oceans requires shrinkage of the Pacific Ocean basin, at least those portions of the basin that are bounded by subduction zones. As a consequence, hotspots located in the shallow mesosphere will be displaced also. Therefore, hotspots cannot form a single, shallow global reference frame.
Morgan’s (1972) evidence for a global reference frame came was contemporary and instantaneous; a similar model emerged from Minster et al. (1974). The geometric argument would require that such a reference frame be deeper than the deepest subduction zones. In order to produce surface volcanism from hotspots embedded in the deep reference frame, rapid vertical transport of material would be required: plumes.
With confirmation of the non-existence of a global hotspot reference frame (e.g., Raymond et al., 2000, using the East-West Antarctic evidence of Cande et al., 2000), the implicit argument for plumes evaporates. Further, the evidence cited here for shallow origin of many, if not all, hotspots removes anomalous magmatism as direct evidence for plumes. The plume hypothesis might still be invoked to explain the heterogeneities in the shallow mesosphere that are implied by the hotspot traces, but evidence of such cryptic plumes is not to be found in age-distance patterns of the traces.
Deep mantle plumes were implied by a global reference frame at first, although that expectation was removed by modeling. Evidence of shallowness of three reference frames (mesoplates) and, more critically, evidence for shallow controls (plate thickness) on the formation and evolution of hotspot traces removes another rationale for plumes.
Tomographic evidence for plumes is debatable. Isotopic and rare-element chemistry of hotspot volcanics need not necessarily imply deep origin. Perhaps it is time to focus on shallow models for the formation of hotspots and their traces, and consider mechanisms by which heterogeneities in the upper mesosphere might be produced at shallow depths.
Plate Stresses and Hotspot Traces
To this point, this paper has emphasized the role of variations in plate thickness combined with sublithospheric heterogeneities as the primary mechanism for the production of hotspot traces. Yet, as Pilger (2003) has shown, within continents at least, paleostress orientations are consistent with the direction of plate motions in the hotspot frame. Could there be a genetic relationship between intraplate stresses and the formation of hotspots (as Solomon and Sleep, 1975, suggested, and Pilger and Handshumacher, 1981, advocated)?
Clearly, lithoplates must fracture in order for magma to reach the surface. Further the state of stress of the lithosphere will have a determining role in the orientation of the first, dike-forming, fractures to develop. In addition, lithospheric thinning due to pronounced crustal extension can result in significant magmatism due to depressurization.
But, do hotspot traces themselves represent propagating fractures, with the volcanism entirely a consequence of depressurization, or do the melting anomalies, whatever their origin, play a role in subsequent magmatism?
Pilger and Handshumacher (1981) proposed that hotspots initially formed due to crustal extension and thinning, but, subsequently, the resulting melting anomaly becomes a focus of intraplate stress. The Tuamotu and Nazca Ridges of the southeast Pacific demonstrate an apparent history that requires the melting anomaly be self-perpetuating. Fig. 20 is a cartoon based on Pilger and Handschumacher’s (1981) rigorous reconstructions of the Pacific and Nazca (Farallon) plates for several discrete magnetic isochrons. The reconstructions use only the isochrons and offset paleotransform faults. Also shown are outlines of the Nazca and Tuamotu ridges. Note that for Chron 11 and 13 reconstructions, the two aseismic ridges intersect at the spreading center. This coincidence implies that a melting anomaly was responsible for both ridges until anomaly 11 (Pilger, 1981, 1984, infers that the melting anomaly was beneath the spreading center by at least Chron 18 time, and perhaps as early as Chron 25). Subsequently, the melting anomaly passed beneath a fracture zone was from that point on entirely beneath the Nazca plate.
Figure 20. “Hotspot” model for the origin of the Easter-Saly y Gomez island seamount chain and Nazca and Tuamotu ridges (modified from Pilger and Handschumacher, 1981). Nazca Ridge (N.R.), Sala y Gomez Island (S.G.), Tuamotu Ridge (T.R.), Easter Island (E.I.). Magnetic isochrons with ages (Ma) in parentheses.
Propagating fractures centered on a ridge would seem to be inadequate to explain such a melting anomaly, since the strains associated with such a fracture would be significantly less than those due to the seafloor spreading process. Rather, it seems more likely that the melting anomaly represents some sort of heterogeneity. Heterogeneities
Anderson and co-workers (e.g., Anderson, 1994, 2001; Foulger and Anderson, 2003; Meibom and Anderson, 2003) have argued for a heterogeneous upper mesosphere and asthenosphere as part of their comprehensive earth model. The heterogeneities in their framework are largely attributable to millions of years of prior subduction – remnants of subducted lithoplate are distributed in the upper mantle, providing a cryptic record of ancient plate tectonics. Foulger and Anderson (2003), for example, suggest that Iceland represents melting of pre-Caledonian oceanic lithosphere.
This paper, building upon Pilger (2003), argues that minor hotspots (and perhaps major hotspots, as well), are a consequence of two circumstances: rapid depressurization due to passage from thick to thin lithosphere and presence of fertile mantle in the area of depressurization. Major hotspots, then, represent more extreme depressurization combined with a locally fertile mantle. In other words, isostatic uplift of asthenosphere and shallow mesoplate may result in the formation of a melting anomaly, depending upon the presence of fertile material.
Are there other sources of fertile heterogeneities in the upper mantle? Plumes remain a possibility, but there is nothing intrinsic in the kinematic and lithospheric structure arguments assembled here to support the idea. On the other hand, the kinematic modeling used in the first part of this paper provides predicted loci of plate motion relative to melting anomalies in mesoplates – melting anomalies inferred to be controlled by lithoplate structure. Of particular interest in this regard is the location of the loci relative to plate structures older than the oldest portions of the hotspot traces.
Tuamotu and Line Islands
The calculated locus of the inferred Easter hotspot not only fits the Nazca ridge (on the Nazca plate) and the eastern Tuamotu ridge (of the Pacific plate), it extends to the north across a small fracture zone that demarcates the northern extent of the latter ridge, from ~44 to ~48 Ma lithoplate (Figure 21). Farther to north, the locus crosses the Marquesas fracture zone, passing from ~50 Ma lithoplate to ~64 Ma lithoplate. Still farther north, it crosses the Galapagos fracture zone, passing from ~70 Ma to ~80 Ma lithoplate and a region, interpreted by Muller et al. (1997) to consist of an abandoned spreading center approximately the same age as that of points along the locus in the same location. Could the remnant asthenosphere/mesosphere initially trapped beneath the extinct ridge provide the fertile source for the future Easter hotspot? The melting anomaly would have come into existence once the “fertile zone” was overlain by the thinner, younger lithoplate to the south of the Marquesas fracture zone.
Figure 21. Tuamotu Islands region. Bathymetry from Smith and Sandwell (1997). Magnetic isochrons from Muller et al. (1997). Calculated hotspot loci at 5 m.y. intervals.
The western Tuamotus are also on the south side of the older-on-north Galapagos fracture zone. The other hotspot traces of the South Pacific may also originate from abandoned spreading ridges. However, evidence is largely missing, since the older portions of the hotspot loci extend into the Cretaceous normal polarity zone (Fig. 2).Conclusion
Despite the history of their development, hotspot reference frame and mantle plume concepts need not be necessarily coupled hypotheses. Evidence marshaled here argues for the existence of three distinct reference frames, identified with “mesoplates” in which hotspots are lodged. Additional, complementary evidence indicates that many, if not all hotspots, are of shallow origin. Their formation is a consequence of isostatic uplift and depressurization of fertile zones in the upper mesosphere. The evidence for lithospheric thickness controls on hotspot formation and evolution has further implications for the thermal structure of the asthenosphere. As Anderson (1994) has argued, a slightly hotter asthenosphere provides the necessary heat to produce all of the anomalous magmatism observed at the earth’s surface.
The arguments for shallow origin of hotspots complement dynamic arguments assembled by Anderson (2001). The outer shell of the earth is the active element in mantle/crust dynamics, rather than deep, internally driven convection. Hotspots are a side-effect of the top-down model of plate dynamics. He further emphasizes the role of intraplate extension in the formation of hotspot traces. Ideas advanced in the current paper emphasize the additional effects of passive and active variation in lithospheric thickness as critical controls on hotspot formation and evolution.
The origin and nature of heterogeneities (Meibom and Anderson, 2003) is the remaining question of interest. Perhaps as Foulger and Anderson (2003) have suggested, they represent remnants of previously subducted lithosphere. Alternative, at least for the minor hotspots, they might represent the residual asthenosphere beneath abandoned ridge spreading centers. The loci of a number of the minor traces in the South Pacific (Fig. 9) intersect possible abandoned ridge segments.
The mesoplate concept could be viewed by some as an unnecessary nuisance -- a non-physical approximation or earth behavior. But, then, too, lithoplates are non-physical approximations of shallow earth tectonics. As acknowledged here, the lower boundary of mesoplates is uncertain – it might correspond with either the 410 or 660 km discontinuity. Even more speculatively, each such discontinuity might represent the boundary between stacked mesoplates – shallow and deep.
Perhaps the classic conceptual flowlines of mantle convection need to be replaced with multiple layers of mesoplates. The mesoplates slip along the spherical discontinuities, accommodating the displacement induced by subducting oceanic lithoplate. Upward vertical motion occurs to accommodate seafloor spreading, with conversion of deeper mesoplate to shallower mesoplate to asthenosphere by appropriate phase changes. Downward vertical motion of at least one mesoplate is required where mesoplates converge. Figure 21 is a highly schematic cartoon illustrating this concept (assuming no communication, except heat, across the 1000 km discontinuity). Subduction of oceanic lithosphere and motion of the upper lithoplate relative to the subduction zone are the two principal drivers of mesoplate motion. Upward vertical motion of mesospheric material compensates for seafloor spreading and ocean-ward displacement of the subduction zone. Lateral movement of mesoplate compensates for displacement of the subduction zone by the overriding lithoplate and by subducted lithoplate. These concepts could be generalized with more lithoplates and stacked mesoplates, plus three dimensional movements.
Figure 22. Idealized cartoon illustrating two lithoplates, two shallow mesoplates, and two deeper mesoplates, with their relative motions (heavy arrows, assuming one shallow mesoplate fixed: bull’s-eye) and vertical motion (light arrows) with phase changes at discontinuities, compensating for seafloor spreading and displacement of the subduction zone by the upper lithoplate. Asthenosphere between lithoplates and shallow mesoplates is not shown. Surfaces separating lithoplates and mesoplates and shallow and deep mesoplates largely correspond with inferred phase change-induced seismic discontinuities (e.g., Gu and Dziewonski, 2002. In the Earth there are numerous lithoplates and three major shallow mesoplates. The number of deeper mesoplates is speculative. The lower surface of the deepest mesoplate corresponds with the 1000 km discontinuity.
With improved tomography of the mantle, might it be possible to unravel the total plate tectonic history of the earth for the last 200+ m.y.? The stacked mesoplate framework might provide a means of such an elucidation. Rather like the palinspastic restoration of a geologic cross section, a combination of lithoplate reconstructions, shallow mesoplate reconstructions using hotspot tracers, and tomographic structural interpretation could lead to historical geology of the outer 1000 km of the earth for the Mesozoic and Cenozoic.Geodynamic models need to explain the small amounts of internal deformation of plates. So, too, similar models need to reproduce not only plate motions and convection, but also anomalous volcanism and the lack of significant internal deformation of the habitat of anomalous volcanic sources – mesoplates.
There is a hierarchy of mesoplate hypotheses articulated here. Numerous lines of evidence point to the existence of three hotspot reference frames. The existence of three shallow mesoplates has is based on observational evidence, summarized here, and, for the most part, more completely documented in Pilger (2003). Insofar as the deeper mesoplates are concerned, there is very little evidence for their existence. However, conservation of mass requires that the lithoplate subduction across the 410 and 660 km discontinuities induce displacements within the Transition Zone (between the two discontinuities) and between the 660 and 1000 km discontinuities. If the two deep regions behave like fluids, then plate-like approximations are probably not relevant. However, the narrow range of variation in the topography of the 410 and 660 discontinuities (Gu and Dziewonski, 2002) could also be a manifestation of shear displacement across each discontinuity. Each boundary could be not only a phase change but a mega-fault zone.
Anderson (2002) has argued that plate tectonics is the manifestation of a far-from-equilibrium system. Plates are transient self-organized entities that serve as the principal means of cooling of the Earth. Might not mesoplates be a consequent, self-similar manifestation of the top-down mechanisms of Earth dynamics? Pastor-Satorras and Wagensberg (1998) have shown that fractal emergence (self-similarity) is a manifestation of the principle of maximum information entropy. That is, a structure tends to replicate itself on multiple levels or scales, in the most efficient way, because the structural similarity is the most probable realization. In this way, then, information entropy somehow emulates thermodynamic entropy, in a non-equilibrium system.
References
Anderson, D.L. (1994) The sublithospheric mantle as the source of continental flood basalts; the case against continental lithosphere and plume head reservoirs, Earth and Planetary Science Letters, v. 123, p. 269-280.
Baksi, A.K. (1999) Reevaluation of plate motion models based on hotspot tracks in the Atlantic and Indian Oceans, Journal of Geology, v. 107, p. 13-26.
Cande, S., LaBrecque, J.L., and Haxby, W.B. (1988) Plate kinematics of the South Atlantic: Chron 34 to present, Journal of Geophysical Research., v. 93, p. 13,479-13,492.
Cande, S. C., and Kent, D. V. (1995) Revised calibration of the geomagnetic time scale for the Late Cretaceous and Cenozoic, Journal of Geophysical Research, v. 100, p. 6093-6095.
Cande, S. C., Stock, J. M., Muller, R. D. and Ishihara, T. (2000) Cenozoic motion between East and West Antarctica, Nature, v. 404, p. 145-150.
Cross, T. A., and Pilger, R. H. (1978) Tectonic controls of Late Cretaceous sedimentation, western interior, USA. Nature, v. 270, p. 653-657.
Cross, T. A., and Pilger, R. H. (1982) Controls of subduction geometry, location of magmatic arcs, and tectonics of arc and back-arc regions, Geological Society of America Bulletin, v. 93, p. 545-562.
Davis, A. S., Gray, L. B., Clague, D. A., and Hein, J. R. (2002) The Line Islands revisited: New 40Ar/39Ar geochronologic evidence for episodes of volcanism due to lithospheric extension, Geochemistry, Geophysics, Geosystems, v. 3, n. 3, 10.1029/2001GC000190.
Duncan, R. A. (1981) Hotspots in the southern oceans - An absolute frame of reference for motion of the Gondwana continents, Tectonophysics, v. 74, p. 29-42.
Duncan, R. A., and Richards, M. A. (1991) Hotspots, mantle plumes, flood basalts, and true polar wander, Reviews of Geophysics, v. 29, p.31-50.
Favela, J., and Anderson, D. (2000) Extensional tectonics and global volcanism, in Boschi, E., Ekstrom, G. and Morelli, A., eds., Problems in Geophysics for the New Millennium, Editrice Compositori, Bologna, p.463-498.
Foulger, G.R., and Anderson, D.L. (2003) Iceland is cool: An origin for the Iceland volcanic province in the remelting of subducted Iapetus slabs at normal mantle temperatures, submitted to Journal of Volcanology and Geothermal Research.
Gaina, C., Müller, R. D., Cande, S. C., (2000) Absolute plate motion, mantle flow, and volcanism at the boundary between the Pacific and Indian Ocean mantle domains since 90 Ma, in Richards, M. A., Gordon, R. G., and van der Hilst, R. D. , eds., The history and dynamics of global plate motions, American Geophysical Union Geophysical Monograph 121, p. 189-210.
Gordon, R. G. (2000) Diffuse oceanic plate boundaries: Strain rates, vertically averaged rheology, and comparisons with narrow plate boundaries and stable plate interiors, in Richards, M. A., Gordon, R. G., and van der Hilst, R. D. , eds., The history and dynamics of global plate motions, American Geophysical Union Geophysical Monograph 121, p .143-159.
Gripp, A. E., Gordon, R. G. (1991) Current plate velocities relative to the hotspots incorporating the NUVEL-1 global plate motion model, Geophysical Research Letters, v. 17, p. 1109-1112.
Gu, Y. J., and Dziewonski, A. M. (2002) Global variability of transition zone thickness, Journal of Geophysical Research, v. 107, p. 2135 , doi:10.1029/2001JB000489.
Haxby, W. F., and Weissel, J. F. (1986) Evidence for small-scale mantle convection from Seasat altimetry data, Journal of Geophysical Research, v. 91, p. 3507-3520.
Herron, E. M. (1972) Two small crustal plates in the South Pacific near Easter Island, Nature, v. 240, p. 35-37.
Horowitz, F.G., P. Hornby, G. Strykowski, F. Boschetti (2001) "Earthworms: multiscale (generalised wavelet?) analysis of the EGM96 gravity field model of the Earth", p. 89-98 in printed volume; paper number 320 in the proceedings website (http://www.ned.dem.csiro.au/research/solidMech/Geodynamics/ChapmanConference/AbstractsReceived/AbstractFiles/Horowitz-et-al.pdf ), American Geophysical Union Chapman Conference on Exploration Geodynamics, Dunsborough, Western Australia.
Klitgord, K. D., and Schouten, H. (1986) Plate kinematics of the central Atlantic, in The Geology of North America, The Western North Atlantic Region, Vogt, P. R., and Tucholke, B. E., eds., Geological Society of America, Boulder, Colorado, v. M, p. 351-378.
Koppers, A. A. P., Staudigel, H., and Duncan, R. A. (2003) High-resolution 40Ar/39Ar dating of the oldest oceanic basement basalts in the western Pacific basin, Geochemistry, Geophysics, Geosystems, v. 4, n. 11, 8914, doi:10.1029/2003GC000574.
Liu, H. S. (2003) Satellites Seek Gravity Signals for Probing the Earth’s Interior, Geodynamics 2003: The Year in Review, Annual Report of the Geodynamics Branch, Laboratory for Terrestrial Physics, Goddard Space Flight Center, http://denali.gsfc.nasa.gov/annual2003/921ar_2003.htm#Satellites.
Liu, H. S., Kolenkiewicz, r., Li, J. L. and Chen, J. Z. (2003) Satellite detection of the convection generated stresses in Earth, Recent Research Developments in Geophysics, v. 5, p. 117-137.
McAdoo, D. C., and Sandwell, D. T. (1989) On the source of cross-grain lineations in the central Pacific gravity field, Journal of Geophysical Research, v. 94, p. 9341-9352.
Meibom, A., and Anderson, D. L. (2003) The statistical upper-mantle assemblage, Earth and Planetary Science Letters, 217, 123-139.Minster, J. B., Jordan, T. H., Molnar, P., and Haines, E. (1974) Numerical modeling of instantaneous plate tectonics, Geophysical Journal of the Royal Astronomical Society, v. 36, p. 541-576.
Molnar, P., and Atwater, T. (1973) Relative motion of hot spots in the mantle, Nature, v. 246, p. 288-291.
Molnar, P., and Francheteau, J. (1975) The relative motion of 'hot spots' in the Atlantic and Indian oceans during the Cenozoic, Geophysical Journal of the Royal Astronomical Society, v. 43, p. 763-774.
Molnar, P., and Stock, J. (1987) Relative motions of hotspots in the Pacific, Atlantic and Indian Oceans since Late Cretaceous time, Nature, v. 327, p. 587-591.
Morgan, W. J. (1971) Convection plumes in the lower mantle, Nature, v. 230, p. 42-43.
Morgan, W. J. (1972) Plate motions and deep mantle convection, in Studies in Earth and Space Sciences (Hess Volume), Shagam, R., et al., eds., Geological Society of America Memoir 132, p. 7-22.
Morgan, W. J. (1981) Hotspot traces and the opening of the Atlantic and Indian Oceans, in The Sea, The Oceanic Lithosphere, Emiliani, C (ed.), Wiley-Interscience, New York, v. 7, p. 443-475.
Morgan, W. J. (1983) Hotspot tracks and the early rifting of the Atlantic, Tectonophysics, v. 94, p. 123-139.
Müller, R. D., Royer, J.-Y., and Lawver, L. A. (1993) Revised plate motions relative to the hotspots from combined Atlantic and Indian Ocean hotspot tracks, Geology, v. 21, p. 275-278.
Müller, R.D., Sandwell, D.T., Tucholke, B.E., Sclater, J.G., and Shaw, P.R. (1990) Depth to basement and geoid expression of the Kane Fracture Zone: a comparison, Marine Geophysical Researches, v. 13, p. 105-129.
Müller, R.D., Roest, W.R., Royer, J.-Y., Gahagan, L.M., and Sclater, J.G. (1997) Digital isochrons of the world's ocean floor, Journal of Geophysical Research, v. 102, p. 3211-3214. Data: http://www-sdt.univ-brest.fr/~jyroyer/Agegrid/utig_report.html.
Norton, I. O. (1995) Plate motions in the North Pacific: the 43 Ma non-event, Tectonics, v. 14, p. 1080-1094.
Norton, I. O. (2000) Global hotspot reference frames and plate motion, in Richards, M. A., Gordon, R. G., and van der Hilst, R. D. , eds., The history and dynamics of global plate motions, American Geophysical Union Geophysical Monograph 121, p. 339-357.
Parker, R.L., and Denham, C. F. (1979) Interpolation of unit vectors, Geophysical. Journal of the Royal Astronomical Society, v. 58, p. 685-687
Pastor-Satorras, R., and Wagensberg, J. (1998) The maximum entropy principle and the nature of fractals, Physica A, v. 251, p. 291-302.
Phipps Morgan, J., Morgan, W. J., and Price, E. (1995) Hotspot melting generates both hotspot volcanism and a hotspot swell? Journal of Geophysical Research, v. 100, p. 8045-8062.
Pilger R. H. (1982) The origin of hotspot traces: evidence from Eastern Australia. Journal of Geophysical Research, v. 87, p. 1825-1834.
Pilger, R. H. (2003) Geokinematics: Prelude to Geodynamics, Springer-Verlag, Berlin, 338 p.
Pilger, R. H., and Handschumacher, D. W. (1981) The fixed hotspot hypothesis and origin of the Easter-Sala y Gomez trace, Geological Society of America Bulletin, v. 92, p. 437-446.
Pitman, W. E., III, and Talwani, M. (1972) Sea-floor spreading in the North Atlantic, Geological Society of America Bulletin, v. 83, p. 619-646.
Raymond, C. A , Stock, J. M., and Cande, S. A. (2000) Fast Paleogene motion of the Pacific hotspots from revised global plate circuit constraints, in Richards, M. A., Gordon, R. G., van der Hilst, R. D. , eds., The history and dynamics of global plate motions, American Geophysical Union Geophysical Monograph 121, p. 359-375.
Replumaz, A., Kárason, H, Van der Hilst, R.D., and Tapponnier, P., (2004) 4-D evolution of SE Asia’s mantle from geological reconstructions and seismic tomography, Earth and Planetary Science Letters, v. 221, p. 103-115.
Richter, F.M. (1973) Convection and the large-scale circulation of the mantle, Journal of Geophysical Research, v. 78, p. 8735-8745.
Sandwell, D. T., Winterer, E. L., Mammerickx, J., Duncan, R. A., Lynch, M. A., Levitt, D. A., and Johnson, C. L. (1995) Evidence for diffuse extension of the Pacific plate from Pukapuka ridges and cross-grain gravity lineations, Journal of Geophysical Research, v. 100, p. 15,087-15,099.
Sandwell, D. T., and Smith W. H. F. (1997) Marine gravity anomaly from Geosat and ERS 1 satellite altimetry, Journal of Geophysical Research, v. 102, p. 10039-10054. Data: http://topex.ucsd.edu/marine_grav/mar_grav.html.
Smith, W. H. F., and Sandwell D. T.. (1997) Global seafloor topography from satellite altimetry and ship depth soundings, Science, v. 277, p. 1957-1962. Data: http://topex.ucsd.edu/marine_grav/mar_grav.html.
Solomon, S. C., and Sleep, N. H. (1974) Some models for absolute plate motions, Journal of Geophysical Research, v. 93, p. 2899-2918.
Steinberger, B., and O'Connell R. J. (2000) Effects of mantle flow on hotspot motion, in Richards, M. A., Gordon, R. G., van der Hilst, R. D. , eds., The history and dynamics of global plate motions, American Geophysical Union Geophysical Monograph 121, p. 377-398.
Tarduno, J. A., Duncan, R. A., Scholl, D. W., Cottrell, R. D., Steinberger, B., Thordarson, T., Kerr, B. C., Neal, C. R., Frey, F. A., Torii, M., and Carvallo, C. (2003) The Emperor Seamounts: Southward motion of the Hawaiian hotspot plume in the earth’s mantle, Science, v. 301, p. 1064-1069.
UTIG, 2003, Present day hotspots, University of Texas at Austin Institute for Geophysics, http://www.ig.utexas.edu/research/projects/lips/data/hotspots.htm.
Wessel, P., and Lyons, S. (1997) Distribution of large Pacific seamounts from Geosat/ERS-1: Implications for the history of intraplate volcanism, Journal of Geophysical Research, v. 102, p. 22,459-22,475. Data: http://www.soest.hawaii.edu/pwessel/papers/1997/JGR/WL_seamounts_1997.d.
Wessel, P. (1993) Observational Constraints on Models of the Hawaiian Hot Spot Swell, Journal of Geophysical Research, v. 98, p. 16,095-16,104.
Wessel, P., Kroenke, L. W., and Bercovici, D. (1996) Pacific Plate Motion and Undulations in Geoid and Bathymetry, Earth and Planetary Science Letters, v. 140, p. 53-66.
Wessel, P., and Kroenke, L. W. (1997) A geometric technique for relocating hotspots and refining absolute plate motions, Nature, v. 387, p. 365-369.
Wilson, J. T. (1966) Did the Atlantic close and then re-open? Nature, v. 211, p. 676-681.
Winterer, E. L., and Sandwell, D. T. (1987) Evidence from en-echelon cross-grain ridges for tensional cracks in the Pacific plate, Nature, v. 329, p. 534-537.
Zoback, M. L., et al. (1989) Global patterns of tectonic stress, Nature, v. 341, p. 291-298.
No comments:
Post a Comment