Abstract
The existence of two distinct hotspot reference frames in motion relative to one another now appears to be conclusively demonstrated over the past 80 m.y. Lithospheric plate (“lithoplate”) motions relative to the two reference frames are recorded not only by hotspot traces, but by intracontinental stress field orientations, oceanic gravity lineations, and reconstructions of Caribbean island arcs. The reference frames represent plate-like layers within the upper mesosphere, termed “mesoplates”. Mesoplates, of which three appear to exist, are in motion relative to one another as well as relative to overlying lithoplates. Subduction zones appear to define part of the boundary between mesoplates and depending on the motion of the upper lithoplate of a zone, control their relative motion. Mesoplates are not merely the habitat of embedded hotspots, they may also be the environment in which plumes responsible for the hotspots develop, in response to passive or active lithospheric thinning and consequent isostatically-driven depressurization melting.
Introduction
Fundamental to the success of plate tectonic theory is the kinematic rigidity of the lithosphere. Internal deformation of lithospheric blocks (plates) is empirically minor compared with displacement of plates relative to one another. A secondary contributor to plate tectonics’ predictive success is the applicability of a theorem of Euler’s, by which displacements of kinematically rigid bodies confined to the surface of a sphere can be concisely described in terms of finite rotations about axes through the center of the sphere. Thus, plate tectonics remains a fundamentally kinematic theory; dynamic explanations must reproduce displacement observations (however they are parameterized).
Kinematic rigidity is sometimes confused with dynamic rigidity. Part of the informal doubt that plate tectonics first encountered among the larger geoscientific community was based on the assumption that the lithosphere is not dynamically rigid enough to escape deformation given the large amounts of displacement inferred between plates. As empirical evidence accumulated, providing documentation of the boundaries and parameterization of displacements between plates and the relatively minor levels of internal deformation within plates, it became apparent that the lack of deformation within plates (with few exceptions) implied that high levels of stress are largely restricted to narrow zones along plate boundaries. Should plate interiors be subjected to high stress magnitudes, the typical result is the formation of smaller plates. For example, Africa appears to be fragmenting along the East African Rift, while the internally deforming Indian plate can be approximately divided into several smaller plates.
Plate reconstructions based on magnetic isochrons and fracture zones now provide a global picture of plate kinematics (with interpolation) extending into the Late Jurassic, at resolutions ranging from 5 to 30 m.y. Such reconstructions, accomplished by rotation composition, allow inference of the kinematics between plates which do not share divergent boundaries, such as the contemporary Nazca and South American plates.
tectonics, paleomagnetic measurements provided strong evidence for continental drift; divergence of similar Late Paleozoic apparent polar wander curve segments between North America and Europe implied their subsequent separation. However, despite decades of subsequent work, variability in the magnetic field, combined with inadequate precision in isotopic dating, has precluded incorporation of paleomagnetic measurements into high resolution reconstructions and kinematic inferences. In addition, the relation of the magnetic poles to the rotational axis of the Earth remains a subject of active research; during the Recent, proximity of averaged magnetic poles to the rotational poles could imply genetic correspondence of the magnetic field and the rotational axes (such correspondence has yet to be demonstrated in the pre-Recent, as well).
There are other possible indicators of plate motions relative to shallow mantle beneath the lithosphere. Hotspots and their interpreted traces (Wilson, 1963) were postulated to represent a fixed internal (“absolute”) reference frame (Morgan, 1971, 1972); further, Morgan suggested the hotspots represented deep mantle “plumes”. Stresses within stable continental interiors are oriented consistently with plate motions in the hotspot reference frame (e.g., Zoback et al., 1989); if the stress orientation representing coupling between continental lithosphere and underlying shallow mantle (thin asthenosphere and underlying mesosphere), then a reflection of relative motion of lithosphere and underlying mantle is implied by the stress orientations. Cross-grain gravity lineations in the Pacific parallel island-seamount chains (Haxby and Weissel, 1986); their wavelength suggests a shallow origin (McAdoo and Sandwell, 1989). Yet to be explored are the implications of subduction zone configurations and mantle plume projections inferred from seismic tomography (e.g., Karason and van der Hilst et al., 2000; Montelli et al., 2003); conceivably, plate kinematics and particular mantle rheologic models might provide some insight into relative motion of lithosphere and underlying mantle.
This review explores the implications of hotspot-plate reconstructions for the relative motion of plates and underlying shallow mantle in the context of the mesoplate hypothesis (Fig. 1, Pilger, 2003a, 2003b, 2004). It is proposed that three “mesoplates”, Hawaiian, Tristan, and Icelandic (Fig. 2), underlie lithospheric plates (“lithoplates”), separated by partially molten asthenosphere of variable thickness. The solidus separates lithoplates from the asthenosphere and the asthenosphere from the mesosphere. The latter surface is the upper limit of mesoplates, as well. The thickness of mesoplates is uncertain, but their bases might correspond to the 410, 600, or 1000 km seismic discontinuities. Mesoplates may move relative to one another, with primary boundaries corresponding with deeply subducting lithosphere. Secondary boundaries are inferred to be almost entirely kinematically-determined. Convergence between mesoplates is largely accommodated along convergent lithoplate boundaries. Relative motion of the Tristan and Pacific mesoplates over the past ~45 m.y. is inferred to be relatively minor; between 45 and 70 Ma, a much greater rate of displacement between the two mesoplates is calculated.
Figure 1. Cartoon illustrating principal features of mesoplate-lithoplate hypothesis, modified after Pilger (2004). Two mesoplates, four large lithoplates, one microlithoplate, and one hotspot (with trace) are shown. Bullseye symbol implies relative fixity of the indicated mesoplate and microlithoplate; that is, other objects are moving relative to them. Note divergence of Mesoplate B relative to A in the vicinity of the subduction zone bounding Lithoplates 2 and 3, with low-angle subducting Lithoplate 3 filling the gap between the mesoplates. Note convergence of Mesoplates A and B in the vicinity of the Lithoplate 4 subduction zone beneath the microlithoplate and Lithoplate 1, forming a “double” subduction zone. Spreading center between Lithoplates 3 and 4 occurs entirely above Mesoplate B; isostatic rise of asthenosphere results in partial melting of Mesoplate B beneath ridge, progressively converting mesoplate to asthenosphere. Hotspot a is moving relative to Mesoplate A, while Lithoplate 3 is moving relative to both Mesoplates B (with embedded hotspot) and A.
Figure 2. Inferred boundaries of the three principal mesoplates (modified after Pilger, 2004). Dotted curves are boundaries taking into account satellite gravity field (Lui et al., 2003). Dash-dot curves are boundaries initially inferred by Pilger (2003a).
In order to demonstrate the viability of the mesoplate hypothesis and its implications for mantle convection, the origin of hotspots, and plate tectonics in general, it is important to critically examine several of its fundamental underpinnings. How good are relative plate reconstructions? Do hotspots form one or more reference frames, or do hotspots move relative to one another? How good are plate-hotspot reconstruction models? Do their kinematic correspondences with other indicators have plate tectonic significance? Is there particular evidence that bears on the origin of hotspots themselves?
Do hotspots form (a) reference frame(s)?
The culmination of decades of work on plate motions relative to hotspots has settled on two distinct hotspot sets: those underlying the Pacific plate (e.g.; Harada and Hamano, 2000 [Harada and Hamano (2000)]; Norton, 2000 [Norton (2000)], Raymond et al. , 2000 [Raymond et al. (2000)]), and those beneath most of the plates of the Central North and South Atlantic and Indian Oceans (Müller et al. , 1993 [Müller et al. (1993)]). The very nature of the two hotspot sets has produced very different approaches to modeling of overlying plate motions relative to each set.
In Morgan’s (1972) first Pacific-hotspot model, an additional assumption, besides fixity of hotspots and rigidity of the plate, was explicitly included: fixity of stage poles that describe plate motion in the hotspot reference frame. For the Hawaiian-Emperor trace, one stage represents the Hawaiian island-seamount chain; the previous stage corresponds with the Emperor seamount chain. Numerous workers subsequently applied the same assumptions, as more and more data (principally isotopic dates) became available. As a result different sets of island-seamount chain segments were incorporated into later models. For example, Morgan had inferred that the Easter trace included the Line Islands, while he did not incorporate the Louisville seamounts. Harada and Hamano (2000), Norton (2000), and Raymond et al. (2000) all rely primarily on the Hawaiian-Emperor and Louisiville traces. Further, the most recent models appear to less explicitly include the fixed stage pole assumption, or at least break up the original stages into smaller intervals. In fact, Harada and Hamano (2000) explicitly excluded the fixed stage pole constraint.
For the plate-hotspot models of the Atlantic-Indian Ocean (beginning with Morgan, 1981, and including Duncan, 1981, Morgan, 1983, and Duncan and Richards, 1991, as well as others, and culminating with Müller et al., 1993), no workers have described the methods by which the parameters were derived. It is apparent, however, that fixed stage poles through sequential reconstructions are nowhere present (compare adjacent, finite difference stage poles for each of the principal Atlantic-Indian Ocean plates; see, e.g., Pilger 2003a). Rather, as multiple plates are involved, successful models must fit traces distributed among six major plates (African, Australian, Indian, North American, South American, Antarctic). As McKenzie and Morgan (1969) showed, in a system of three or more plates, constancy of motion (a fixed stage pole) between one pair requires, in general, continuous motion of the instantaneous rotational pole(s) describing motions of the third and other plates. Thus, there is no intrinsic reason to infer constant stage poles for interplate motions. More broadly, there is no clear reason to infer constancy of stage poles (or rates) for describing plate motions relative to hotspots.
A recent analysis of the fixity of hotspots in the Pacific (Koppers et al., 2001) relied on fixity of stage poles and rates for prolonged periods of time. A range of possible stage poles was fit to inferred traces for distinct time intervals. Independently, rates inferred from isotopic ages of traces within the same time intervals were similarly fit to possible poles. From these analyses, an inference of non-fixity of hotspots was derived. However, the models of others (e.g., Harada and Hamano, 2000; Norton, 2000; and Raymond et al., 2000) do not demonstrate fixity of stage poles, or, more critically, stage rates. Further, as the rate intervals used by Koppers et al. overlapped the different stage poles derived by the other models, the critical assumption – fixity of poles and rates – cannot be sustained, and the derivative conclusion of non-fixity of hotspots cannot be supported.
Is there another way to evaluate the viability of a fixed-hotspot hypothesis? Intrinsic non-linearity requires that all that can be done is to evaluate the applicability of particular plate-hotspot models. Iterative search methods, such as those under development by Wessel et al. (2004) have promise for single plate systems, such as the Pacific, but their applicability to multi-plate systems is unclear, especially with uncertainty in age constraints. Oceanic plate reconstructions are largely based on magnetic isochrons and fracture zones. Age uncertainty in a hotspot trace is magnified into reconstruction uncertainty, since the age of the reconstruction needs to fit corresponding points along restored traces.
Pilger (2003a) assembled maps and distance-versus-age plots of isotopic dates for each inferred hotspot trace utilized in Müller et al. (1993) and Raymond et al. (2000) and incorporated the predicted distance/age locus into the maps and plots -- an approach that has been applied repeatedly since isotopic ages first became available from such traces. Because of the multidimensionality of such data sets (latitude, longitude, and age) and changes in loci orientation, maps and distance/age plots may not be enough to evaluate the “fit” of predicted loci to isotopic age dates, especially in light the added complication of age uncertainties.
Hotspot Reference Frames-A Reconstruction Test
In recent years, Wessel and Kroenke (1997) introduced a technique for identifying hotspots (“hot-spotting”) beneath the Pacific plate. From a plate-hotspot model, without reference to age, they “back-tracked” islands and seamounts in the hotspot (rather than plate) frame, as multiple loci, and inferred hotspot locations from the intersections of such loci. Such a technique requires prior reconstruction/kinematic models and distinct changes in the orientation of hotspot traces in order to produce unambiguous intersections. Recently Wessel et al. (2004) reported progress in integrated kinematic modeling that combined the methods of Harada and Hamano (2000) for deriving hotspot-kinematic models with hot-spotting (finding the best-fit location of hotspots).
Koppers et al. (2003) applied a restoration technique to dated seamounts (from 134 to 69 Ma in age) from the western Pacific, relying on the analytical ages. If the dates, reconstruction parameters, interpolation technique, and the fixed hotspot model are correct, such restorations should produce clustering of data points from the same trace at the contemporary hotspot location. They concluded that “long-lived, deep and fixed mantle plumes cannot explain intraplate volcanism in the South Pacific mantle over the last 140 Myr.” However, they did identify two restored clusters of data points that are close to recently active hotspots, despite uncertainty in the reconstruction parameters.
A combination of the kinds of analyses applied by Koppers et al. (2003) to their own data and Wessel and Kroenke (1997) to seamounts provides a means for the evaluation of other plate-hotspot models, such as those of Harada and Hamano (2000), Norton (2000), and Raymond et al. (2000) (for the Pacific), and Müller et al. (1993) (for the Atlantic-Indian), when applied to younger data sets. There are both analytical and geological uncertainties implicit in age dates from inferred hotspot traces. In general, the geological samples must, by necessity, postdate the first encounter of a point on a plate with a hotspot. Intrusion and emplacement follow the first encounter, followed by recurrent magmatism as the magma chamber is replenished and then ultimately depleted and/or cooled. Dredging and drilling most commonly sample the youngest volcanic strata, or, more rarely, vents or dikes. With the additional contribution of analytical uncertainty, the kinematic “age” of a hotspot-generated island or seamount could be in error by several millions of years.
For the procedure presented here, published ages were restored (rotated back) to model locations according to the various (and appropriate) kinematic models [Müller et al. (1993), Harada and Hamano (2000), Norton (2000), Raymond et al. (2000)], interpolated using the spline methods of Pilger (2003a) [Pilger (2003a)]. In addition, loci in the local plate reference frame were calculated for +/- 5 m.y. at 2.5 m.y. intervals. For example, if the analytical age were 25.3 Ma, restored locations assuming ages of 30.3, 27.8, 25.3, 22.8, and 20.3 Ma were calculated (except that negative ages, for dates less than 5 Ma, or for ages greater than the maximum age of the tested model, were not included). The resulting five-point “loci” are equivalent to segments of loci constructed according to the method of Wessel and Kroenke (1997). Pilger’s (2003a) compilation of isotopic ages (78 Ma and younger), supplemented by the USGS compilation (Zartman, et al., 1995), Davis et al. (2002), Koppers and Staudigel (2005), is the basis of the initial analysis presented here (Fig. 3-11).
Figure 3a. Pacific Ocean hotspot trace dates and loci: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003). Reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of Harada and Hamano (2000), interpolated by method of Pilger (2003a). Magnetic isochrons and boundaries from Müller et al. (1997).
Figure 3b. As Figure 1b, with parameters of Norton (2000).
Figure 3c. As Figure 1b, with parameters of Raymond et al. (2000).
For the Pacific region as a whole, the reconstructed hotspots according to each of the three models are shown in with loci (+/- 5 m.y.) in Figs. 3. Restorations based on Harada and Hamano (2000) are shown in Fig. 3a, Norton (2000) in Fig. 3b, and Raymond et al. (2000) in Fig. 3c. Similarly, the same order of models is shown with combined reconstructed hotspots and loci for individual regions for the major Pacific hotspots: Hawaii (Fig. 4), Louisville (Fig. 5), Easter and Foundation (Fig. 6), and Austral-Cook (Fig. 7).
Figure 4a. Hawaiian-Emperor trace dates: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of Harada and Hamano (2000), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997). Seamounts (open diamonds) from Wessel (2004).
Figure 4b. As Figure 4a, with parameters of Norton (2000).
Figure 4c. As Figure 4a, with parameters of Raymond et al. (2000).
Figure 5a. Louisville trace dates: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003). Reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of Harada and Hamano (2000), interpolated by method of Pilger (2003a). Magnetic isochrons and boundaries from plate Müller et al. (1997). Seamounts on Pacific plate (open diamonds) from Wessel (2004).
Figure 5b. As Figure 5a, with parameters of Norton (2000).
Figure 5c. As Figure 5a, with parameters of Raymond et al. (2000).
Figure 6a. Easter (north cluster) and Foundation (south cluster) trace dates: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003). Reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of Harada and Hamano (2000), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997). Seamounts on Pacific plate (open diamonds) from Wessel (2004).
Figure 6b. As Figure 6a, with parameters of Norton (2000).
Figure 6c. As Figure 6a, with parameters of Raymond et al. (2000).
Figure 7a. Austral-Cook trace(s) dates: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a, supplemented by Koppers and Staudigel, 2005). Reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of Harada and Hamano (2000), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997). Seamounts on Pacific plate (open diamonds) from Wessel (2004). Note intersections of pre-Bend and post bend loci, forming two clusters. (Cluster in center-north is Societies Islands.)
Figure 7b. As Figure 7a, with parameters of Norton (2000).
Figure 7c. As Figure 7a, with parameters of Raymond et al. (2000).
Figure 8. Reunion (north cluster) and Kergulen (south cluster) trace dates, Indian Ocean: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares), filtered and recalculated data points (open squares; Baksi, 1999) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
Figure 9. Tristan trace dates, South Atlantic Ocean : Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
Figure 10. Trinidade trace dates, South Atlantic Ocean: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
Figure 11a. Central North Atlantic Ocean trace dates: Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
Figure 11b. Central North Atlantic Ocean trace dates (close-up): Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
For the Atlantic and Indian Oceans, individual reconstructed hotspots are shown together with the loci: Kerguelen and Reunion (Fig. 8), Tristan (Fig. 9), Trinidade (Fig. 10), and Great Meteor and Canary (Fig. 11). East Australia and Tasman (Fig. 12), East Africa (Fig. 13), and western North America (Fig. 14) are also included, although the reconstructions include volcanic dates which may not be hotspot related. Iceland and the British Isles are excluded because of scarcity of data and limited constraints on the motion of the Icelandic hotspot reference frame relative to the other two reference frames.
Figure 12. East Australia trace dates (close-up): Sample location (crosses), with age (Ma) (compilation of Pilger, 2003a). Reconstructed original data points (solid squares) and calculated loci (+/- 5 m.y.) according to parameters of Müller et al. (1993), interpolated by method of Pilger (2003a). Magnetic isochrons and plate boundaries from Müller et al. (1997).
Figure 13. K/Ar ages from East Africa. Sample locations (crosses, compilation of Pilger, 2003), reconstructed data points (solid circles) and calculated loci (+/- 5 m.y.) according to parameters of M2000, interpolated by method of Pilger (2003a).
Figure 14a. K/Ar ages of basalts (<40 Ma) from western North America. Larger and lighter circles are younger; compilation of USGS: Zartman et al., 1995), Loci of North America in Tristan (closed circles) and Hawaiian (open diamonds) reference frames, anchored near inferred Yellowstone and Raton hotspots (5 m.y. intervals) are also shown.
Figure 14c. Reconstructed data points as in Fig. 13a, in North America to Hawaiian reference frame according to parameters of Raymond et al. (2000) interpolated by method of Pilger (2003a). Larger and lighter circles are younger.
Expected Patterns
Assuming a fixed hotspot frame of reference, accurate isotopic ages, and approximately correct reconstruction and interpolation models, restored dates should cluster around the hotspot location in elliptical patterns, with one end of each ellipse corresponding with the hotspot, while the other end represents continued volcanism after initial inception. Similarly, such elliptical patterns would be oriented in the direction of plate motion at the time of magmatic emplacement. The degree of dispersion would represent the finite dimensions of the hotspot, deviation from fixity, and inaccuracies in the model.
In the case of the multi-plate restorations, rotation parameters that fit traces on one plate (e.g., the Pacific lithoplate) may misfit adjacent plates to the extent that the ages for the primary plate reconstructions, if not the parameters, might be in error. Thus, the age of the Hawaiian-Emperor bend (along with other points along the trace) is critical in terms of fitting traces on the Nazca plate, whose reconstructions are based on magnetic isochrons. Pilger and Handschumacher (1981) reviewed these considerations in proposing an age for the bend greater than the then widely-accepted 43 Ma.
Pacific Ocean Hotspots
Only two Pacific Ocean hotspot traces (Hawaiian and Louisville) produce isotopic ages that clearly range over 70+ m.y. Dates from the Austral-Cook trace(s) extend over more than 70 m.y., when the new data of Koppers and Staudigel (2005) are included and restored, with evidence of recurrent volcanism. However, continuity of island-seamounts of the Austral-Cook with the older data points is not obvious.
The Easter-Tuamotu/Nazca produce age ranges of ~45 m.y., if the magnetic isochrons that intersect the Tuamotu and Nazca ridges provide accurate ages of the intersection points. The Foundation seamount chain ranges up to ~ 40 Ma. The Line Islands appear to be composites of multiple volcanic events between 90 and 35 Ma. Remaining, lesser chains produce age ranges from as much as 30 m.y. (Gulf of Alaska chains) to between 6 and 10 m.y. (Marquesas, Pitcairn, and Society). On the Nazca plate (in addition to the Easter-Nazca Ridge trace), isotopic ages range from 9 Ma to present for the Galapagos and over the past 4 m.y. for Juan Fernandez Ridge.
Restoration of dated locations and loci vary somewhat among the three Pacific models (Harada and Hamano, 2000, Norton, 2000, and Raymond et al., 2000). The models are particularly distinctive when data points from the Nazca plate are included. In comparing reconstructed data points and loci for Pacific Ocean hotspot traces, differences among the models need to be taken into account. Harada and Hamano (2000) includes the greatest variability in parameters, calculated at 2 m.y. intervals from 70 Ma to present. Norton’s (2000) parameters are the fewest and cover the longest average time intervals (reconstructions at 25, 43, 81, and 90 Ma). Raymond et al. ’s (2000) parameters span variable length intervals, intermediate between Harada and Hamano (2000) and Norton (2000), from 78.78 Ma to present (reconstructions at 5, 10.83, 21.26, 25, 33.3, 40.22, 43, 43.87, 47.86, 53.25, 63.3, 68.68, and 78.78 Ma). Each of the three models could be “retuned” to take into account a likely older age for the Hawaiian-Emperor bend (Pilger and Handschumacher, 1981; Sharp and Clague, 1999; Clague, 2003).
The orientations of restored loci are indicative of the direction of plate motion relative to the underlying hotspot at the reconstructed age according to each model. Thus, for example, modeled loci at Hawaii and younger than the Hawaiian-Emperor bend have a WNW orientation; older loci (corresponding with the Emperor Seamounts trend) are more northerly trending.
Hawaiian-Emperor Island-Seamount Chain
As would be expected, Norton (2000) (Fig. 4b) shows the least dispersion in orientation of predicted loci and the largest dispersion parallel with the loci for the Hawaiian-Emperor trace compared with Harada and Hamano (2000) (Fig. 4a) and Raymond et al. (2000) (Fig. 4c); this is attributable to the longer duration of each parameter stage (i.e., fewer reconstructions). With only three stages in Norton (2000) from 81 Ma to present, constancy of motion rate will not take into actual velocity changes, if present. Raymond et al. (2000) demonstrates the tightest clustering of restored data points, with just a few outliers. Harada and Hamano (2000) has most of the same outliers as Raymond et al. (2000), but with less clustering of restored data points than Raymond et al. (2000).
Norton (2000) and Raymond et al. (2000) accommodate a few older data points than Harada and Hamano (2000). Loci from pre- and post- Hawaiian-Emperor bend intervals intersect close to Hawaii, but with not enough resolution to precisely predict the hotspot location. Revision of dates from the trace wouldn’t significantly affect the loci intersections, based on prelimary results (Clague, 2003), except for data points close to the bend.
Louisville Seamounts
Harada and Hamano (2000) (Fig. 5a) shows the tightest clustering of restored data points and loci intersections, with two outliers. Norton (2000) (Fig. 5b) shows more dispersion along restored loci, as also observed at Hawaii (Fig. 4b); it too, has two outliers. Raymond et al. (2000) shows tight clustering of younger restored dates, with dispersion of older dates (but outside of Harada and Hamano’s (2000) range) and two outliers – the same as the other two models. One of the outliers (in all three models) comes from a seamount that is not clearly aligned with the rest of the seamounts that comprise the inferred trace.
Easter-Tuamotu Ridge-Nazca Ridge and Foundation Seamounts
Restored data points and loci for the Easter hotspot are derived from both the Pacific and Nazca plates (Fig. 6a, b, c); loci from data points on the Nazca ridge are northeasterly oriented while those from the Sala-y-Gomez island-seamounts are more easterly oriented. Loci from Pacific data points are less age-sensitive than those from the Nazca plate. Since the reconstruction parameters for all three models are derived from only the Pacific plate, extension of the model to the Nazca plate introduces additional uncertainty due to possible errors in the geomagnetic time scale as well as the isotopic ages. If, for example, the Hawaiian-Emperor bend were older (e.g., ~47 Ma; Clague, 2003), rotation compositions would introduce a higher rate of Nazca plate motion in the Raymond et al. (2000) hotspot model between the bend and 25 Ma in Raymond et al. (2000) (if parameters with latter date were left unchanged).
Raymond et al. (2000) shows the best clustering, with younging dispersion along loci (as expected) for both Easter and Foundation. The younging directions are on the expected sides of the Easter hotspot in all modeled cases. Incidentally, the oldest dated sample(s) from of the inferred Foundation trace are close to the youngest extent of the inferred Macdonald trace (Fig. 7).
Austral-Cook (Macdonald) Island Seamounts
Isotopic ages from the Austral-Cook trace have been known to be anomalous for some time (e.g., Pilger and Handschumacher, 1981). With inclusion of new data from Gilbert Ridge and Tokelau Seamounts (Koppers and Staudigel, 2005; Fig. 7a, b, c), there are clear implications of two long-lived hotspots (Pilger, 2005b) – the more eastern corresponding with Macdonald Seamount (implicitly recognized by Koppers and Staudigel).
Each of the three models provides varying degrees of support for the two-hotspot interpretation. Harada and Hamano (2000) appears to provide the tightest clustering of loci intersections. If the Hawaiian-Emperor bend is in fact closer to ~47 Ma, then Raymond et al. (2000) would appear to provide the better clustering of loci intersections (an older age for the bend would move the Gilbert and Tokelau loci more to the north-northwest relative to the younger loci).
Other Pacific Ocean Chains
As noted above, the remaining island-seamount chains of the Pacific plate (Fig. 1) cover variable age ranges. The Gulf of Alaska chains show two linear zones of restored data points with the Raymond et al. (2000) models showing the tightest clustering. If the dates are approximately correct, they imply recurrent activity after inception. The small age ranges for the Marquesas, Society, and Pitcairn trends have no obvious additional interpretive significance other than consistency with the youngest portion of the models.
Only a few data points from the Line Islands are restored (remaining published dates are older than the plate hotspot models of Harada and Hamano, 2000, Norton, 2000, and Raymond et al., 2000). The restored points fall between the Marquesas and Easter Islands and to the north of the Marquesas.
For the Nazca plate (exclusive of the Easter-Nazca trace), two additional hotspots are inferred. Harada and Hamano (2000) shows the best clustering for the Galapagos, while the small age range for Juan Fernandez doesn’t allow for meaningful distinctions among the three models. It should be noted that there is substantial uncertainty in plate reconstructions between isochrons 3 and 7 for the Pacific and Nazca plates. Thus, Nazca plate isotopic dates between ~5 and ~25 Ma are restored with far less certainty than younger and older dates. In particular, the older dates from the Galapagos-Carnegie Ridge trace, which show greater dispersion for Norton (2000) and Raymond et al. (2000) than Harada and Hamano (2000), could be attributed to this uncertainty.
Atlantic-Indian Ocean Hotspot Traces
The Müller et al. (1993) plate-hotspot model fits the physiography of the inferred hotspot traces of the Atlantic and Indian Ocean significantly better than preceding models (compare with, e.g., Morgan, 1981, 1983; Duncan, 1981; and Duncan and Richards, 1991). There is, therefore, no reason to make comparisons of the kind done for the Pacific hotspots with the previous Atlantic-Indian ocean models. However, there are apparent misfits to published isotopic dates (Pilger, 2003a) which warrant further analysis.
As noted above, for the expected pattern of restored data points, assuming validity of an internally fixed hotspot frame and the reconstruction parameters, dates that are “too young” would extend in an elliptical pattern (in the younging direction) away from the hotspot location in the direction of plate motion at the time of volcanic emplacement. Trace dates that are “too old” would tend to violate the assumptions, as they imply emplacement before the “future” trace location encountered the hotspot; such dates would occur in the “olding” direction away from the hotspot.
For some of the hotspot traces incorporated into the Müller et al. (1993) analysis, Ar/Ar dates have been reanalyzed by Baksi (1999). In some cases Baksi rejected the original dates; in others, he recalculated dates using different assumptions. Thus, in addition to original published dates, the filtered, recalculated dates of Baksi were also restored.
Kerguelen (Ninetyeast Ridge) and Reunion (Chagos-Laccadive Ridges)
Relative to the inferred Müller et al. (1993) hotspot locations, there are several original data points which are “too old” (Fig. 8). However, when Baksi’s (1999) filtered data set is used, only one data point (from the Ninetyeast Ridge, dated between 80 and 100 Ma) is clearly “too old”.
A single data point from Laccadive Ridge is restored south of Reunion (on the “olding” side). However, possible age uncertainties (not calculated by Baksi) of 2 to 3 m.y. would place it in closer proximity to the inferred hotspot.
Tristan da Cunha (Walvis Ridge and Rio Grande Rise)
Restoration of data points in the South Atlantic (using Müller et al. (1993), combined with Morgan’s, 1983, pre-130 Ma African-hotspot stage parameters) produces a broad clustering of Cenozoic dates (Figure 9), but a much broader dispersion of continental margin Cretaceous and Late Jurassic dates. There is evidence for recurrent magmatism along Walvis Ridge after inception.
Trinidade and Eastern Brazil
Virtually no data are available from an unequivocal Trinidade hotspot trace. Restored data (all are K/Ar) come from eastern Brazil (Fig. 10). Many of the dates postdate rifting of the South Atlantic. The wide dispersion of activity makes it difficult to ascribe a simple hotspot origin to the magmatism, even without reconstruction.
Great Meteor (New England Seamounts) and Canary Islands
Restored data points from the New England Seamounts show a large number of “too old” positions relative to the inferred hotspot and younger portion of the trace (Fig. 11a, b). However, Baksi (1999) filtered out many of these data points and recalculated a few. His recalculated points are much closer to the hotspot.
The few data points from the Canary Island trace restore well, in a younging direction (Fig. 11a, b). A few older data points, which may or may not be part of the trace, restore significantly to the east.
East Australian Highlands, Tasmania, and Tasman Seamounts
The few dated Tasman Seamounts don’t restore to a common cluster using Müller et al. (1993) parameters (Fig. 12). Pilger (1982) and Gaina et al. (2000) suggested that the Tasman trace may belong to the Pacific Ocean (Hawaiian) hotspot set, for which better clustering results (not shown).
The volcanics of the East Australian highlands, when restored, do not produce a single cluster (Fig. 12). Rather, there is a suggestion of a linear alignment at ~ 40º South latitude, plus separate north-south elongate clusters farther south and a small cluster in Northern Queensland. Pilger (1982) inferred a time-transgressive cessation of activity along the Highlands at latitudinal rate equal to that of Australian plate motion in the hotspot reference frame.
East Africa
When reconstructed, isotopic ages from East Africa (Fig. 13) cluster in one primary group, largely to the north of single line of latitude (~5ºS). A second cluster appears to exist just to the north of the Equator. Unlike marine examples, the original data points do not appear align along one or more “hotspot” loci.
Pilger (2003a; 2005b) noted that the time-transgressive onset of magmatism occurs at the same latitudinal rate as motion of Africa according to the hotspot reconstruction parameters of Müller et al. (1993). Magmatism, once initiated, appears to have continued for tens of millions of years in some parts of East Africa. These patterns are not consistent with a spreading plume-head inferred by some workers, as spreading would be faster than plate motions.
Western United States
Isotopic ages (<40 Ma) from basalts in the western United States (Fig. 14a) are reconstructed according to both Müller et al. (1993) (Fig. 14b) and Raymond et al. (2000) (Fig. 14c). No attempt has been made to apply additional palinspastic reconstructions of the data points to account for Cenozoic crustal deformation, however.
Restored clusters anchored on Yellowstone and Raton are apparent with both parameter sets (Fig. 14b, c). Of course, wide dispersion of other data points results from this analysis – from basalts that are inferred to be extension-related. What is particularly striking is the tighter clustering of younger data points, particularly along Yellowstone trend, reconstructed according to Raymond et al. (2000) (Hawaiian hotspots) rather than Müller et al. (1993) (Tristan hotspots). Conversely, Müller et al. (1993) produces tighter clustering for the Raton trend than Raymond et al. (2000).
When calculated loci for the two models are compared with original sample locations (Fig. 14a), the arcuate trend of Raymond et al. (2000) parallels that of data from the Snake River plain southwest of Yellowstone over the past ~20 Ma. No such arcuate trend is apparent in data along the inferred Raton trend; rather, available data from that trend are closer to the Müller et al. (1993) locus.
Caribbean Island-Arcs
While it is not hotspot-related, volcanism associated with the Lesser and Greater Antilles subduction zones provides intriguing insight into the nature of hotspot reference frames when restored to the Atlantic-Indian Ocean hotspot set (Pilger 2005c). Using Pindell and Kendall’s (2002, 2003) reconstruction of the paleo-arcs relative to North America, Pilger restored the inferred paleo-volcanic centers to the hotspot reference frame according to the Müller et al. (1993) model as extended and interpolated by Pilger (2003a) (Fig. 15).
The resulting restoration implies that the Antillean subduction zones have been essentially fixed relative to the Atlantic-Indian Ocean hotspot set, although the lateral extent has changed over this period of time. Assuming that the subduction zones extended into the mesosphere, this implies some degree of stationarity of the upper mesosphere as well.
Restoration Patterns
What do the restored data signify? First, consider the most obvious departures from any kind of simple hotspot model. The few restored data points from the Line Islands are quite dispersed, as would be expected given the age-distance patterns (e.g., Davis et al., 2002). Clearly, a single hotspot cannot explain the Line Islands. Even multiple hotspots are problematic.
Note that the restored locations align in the direction of plate motion at the time of cooling (if not emplacement). The depressurization-melting model of Raddick et al. (2002) is possibly applicable to the Line Islands. Limited published data and interpretations (e.g., Müller et al. , 1997) suggest that the islands and seamounts of the Line trend tend to occur on the younger (therefore thinner) side of older-on-north fracture zones. Further, the source regions could represent remnant sublithospheric mantle formed beneath abandoned spreading centers formed during the Late Cretaceous plate reorganization to the west of the Line Islands.
Similarly, the older portions of the other hotspot traces of the South Pacific, with the apparent exception of the Marquesas, also occur in a similar tectonic position to the Line islands. Major fracture zones separating older-on-north lithosphere occur to the north of the older ends of the volcanic chains. And, even the Marquesas Islands, can be shown to overlie mesosphere that once underlay significantly older lithosphere, possibly even beneath an abandoned spreading center.
The restored data from the Austral-Cook and Gulf of Alaska chains imply multiple “hotspots”. Again, however, the chains occur on the thinner side of paleofracture zones, in accord with the depressurization hypothesis, although recently published data from the Gilbert Ridge and Talekau seamounts, which reconstruct close to apparent hotspots based on younger data, are not in such an obvious position.
If the depressurization hypothesis is applicable to the hotspot traces of the Pacific, what do the reconstructions imply? The clustering of restored data points imply that the volcanic sources are indeed relatively fixed in the sublithospheric Pacific mantle. The ability to construct kinematic models of Pacific and Nazca plate motion from hotspot traces implies little relative motion among the volcanic sources over at least the past 70 Ma.
Could the apparent hotspot reference frames be virtual constructs, representing propagating extensional fractures? Within continental plates, certainly, there is evidence that stress field orientations correspond in part with plate motion models relative to hotspots (e.g., Zoback et al., 1989; Pilger, 2003a).
The dilemma with the intraplate stress hypothesis for the origin of hotspot traces is evidence for hotspots passing from beneath one plate to beneath another. For example, Kerguelen hotspot appears to have passed from beneath the Indian plate to beneath the Antarctic; similarly, Reunion hotspot appears to have passed from beneath the Indian plate to beneath the African. Finally, there is the case of Easter hotspot. While its present location is uncertain, it is apparent that between magnetic isochrons 21 and 11 the hotspot was located beneath a segment of the Pacific-Farallon spreading center and generated the “mirror-image” eastern Tuamotu and Nazca ridges (Pilger and Handschumacher, 1981). Subsequent to isochron 11, the hotspot passed beneath a fracture zone and apparently to the present has remained beneath the Farallon (and its successor) Nazca plates. It is difficult to see how a stress node could pass from one plate to another across a spreading center, especially since stress orientations at the spreading center would have been oblique to the inferred stress fields responsible for the two aseismic ridges. Rather, a sublithospheric melting anomaly must be invoked to explain the mirror-image ridges.
Further, there are the examples of the Gilbert Ridge and Tokelau Seamounts (Fig. 7). The two areas are not clearly connected with the younger portions of the Austral-Cook island-seamount chains, even though, when restored, the older data points are consistent with two possible hotspots responsible for the younger traces. The intra-Pacific stress field would have not only change orientation significantly at ~47 Ma, it would have allowed for fractures and accompanying magmatism propagating at consistent rates throughout the plate. And, the postulated propagating fractures would be void of magmatism for thirty million years between the youngest portions of the Gilbert and Tokelau trace segments and the oldest portions of the overprinted Austral-Cook traces.
Mesospheric Reference Frame(s)
The restored hotspot trace evidence assembled above remains consistent with distinct hotspot reference frames: beneath the Pacific (although it is difficult to resolve which reconstruction model best fits the inferred traces) and beneath the central-North Alantic, South Atlantic, and Indian Oceans (as parameterized by Müller et al. , 1993). Appropriate further questions are: (1) Do the two hotspot reference frames have any other manifestation? (2) Are they consistent with one another?
Reference Frame Manifestations
Contemporary intracontinental stress fields appear to correspond with the direction of plate motion in the hotspot reference frame (e.g., Zoback et al., 1989; see also Fig. 16). Similarly, paleostress indicators also appear to correspond with plate-hotspot kinematics (Pilger, 2003a). The principal paleostress indicators are: (1) isotopically dated igneous dikes and veins, and (2) faults and fractures tightly constrained in age. Figs. 17 and 18 illustrate parallelism of North American instantaneous maximum horizontal compressive paleostress measurements with flow-lines derived from plate-hotspot reconstruction parameters (Müller et al. , 1993) using the method of Pilger (2003a), especially between 30 and 100 Ma. As noted above, the model of Müller et al. might require revision for the period prior to 100 Ma. Since 30 Ma, onset of extension of the U.S. Cordillera is apparent in more northerly-striking stress indicators, although a few indicators are consistent with the flow-lines, especially on the eastern margin of the Cordillera.
Figure 17a.
Figure 17. Maps of paleostress measurements and flow-lines 0 to 90 Ma in 10 m.y. intervals, Western United States, in 30 m.y. montages (a, b, c). Bars: Maximum principal horizontal compressive paleostress measurements (s1; compiled by Pilger, 2003a). Flow-lines (small circles): Instantaneous direction of motion of North American (to the west) relative to the global hotspot reference frame of Müller et al. (1993), calculated using method of Pilger (2003a). Solid curves: 0, 20, 40, 60, and 80 Ma; dashed curves: 5, 15, 25 … Ma; dotted curves: 10, 30, 50, 70, 90 Ma. Contemporary stress measurements are not included in 0 to 10 Ma map.
Figure 17b.
Figure 17c.
Figure 18. Maps of paleostress measurements and flowlines 90 to 130 Ma in 10 m.y. intervals, United States and Canada
In the hotspot debate, there has been little reference to the significance of intraplate stress field orientation beyond the present-day observations. One implicit exception is for models for explaining hotspot traces in terms of propagating fractures. If traces are consistent in orientation, this could be inferred to represent stress fields controlled by plate motions, with propagation occurring at near-constant angular rates over the entire plate. Further, consistency of trace orientations with multi-plate reference frame models would imply that the operable mechanism for controlling the plate motions is sublithospheric. Reference Frame Inconsistency
Inconsistency between hotspot traces in the Pacific with those of the Atlantic and Indian Oceans has long been a theme of critics of the fixed hotspot and plume hypotheses (beginning with Molnar and Atwater, 1973). The recent models of Müller et al. (1993) (extended to the Pacific plate via the circuit: hotspots-Africa-Antarctica-Pacific) and Raymond et al. (2000) (as well as Harada and Hamano, 2000, and Norton, 2000) are clearly inconsistent prior to ~ 10 Ma, and grossly so prior to ~45 Ma. (e.g., Fig. 19).
Figure 19. Data points from Hawaiian-Emperor island-seamount chain with loci of calculated motion of Pacific plate relative to Pacific (Hawaiian; PCFC-HAWA) hotspot reference frame, according to Raymond et al. (2000) reconstruction model, interpolated using methods of Pilger (2003a). Also shown are loci of Pacific plate relative to Atlantic-Indian Ocean (Tristan; TRIS-PCFC) hotspot reference frame, and North American (NOAM), South American (SOAM), Antarctic (ANTA), and Eurasian (EURA) plates. Loci from 0 to 80 Ma. Circles at 10 m.y. intervals.
A persistent explanation for the inconsistency of the two hotspot sets is unrecognized internal deformation within Antarctica. Cande et al. (2000) demonstrated from marine magnetic anomalies that, while East and West Antarctica can be distinguished as separate plates between magnetic isochrons 13 and 16 (33 to 39 Ma); however the amount of relative movement between the two plates is inadequate to explain the apparent discrepancy between the two hotspot sets over the past 39 Ma (Raymond et al. (2000); Pilger, 2003a). More importantly, significant displacement between two Antarctic plates would also be recorded in marine magnetic isochrons older than 39 Ma on the southern portions of the Australian and Pacific plates – such displacement is not apparent in mapped isochrons of the region.
A more telling point is to consider how much deformation, and of what sense, is required to have occurred within Antarctica in order to demonstrate relative fixity of the hotspot sets of the Pacific and Atlantic-Indian Oceans. Fig. 20 illustrates reconstructions of West Antarctica (semi-arbitrarily defined) relative to all of Antarctica at 10 m.y. increments from present to 80 Ma. The reconstructions follow the circuit Antarctica-Africa-Hotspots-Pacific-West Antarctic, where the Africa-hotspot parameters are based on the model of Müller et al. (1993) and Hotspot-Pacific parameters are based on those of Raymond et al. (2000), with interpolations by the method of Pilger (2003a). To be explicit, the two hotspot reference frames are assumed to be equivalent for just this exercise.
Figure 20. Map of reconstructed West Antarctica relative to present-day Antarctica coast, using parameters of Müller et al. (1993) and Raymond et al. (2000) interpolated using method of Pilger (2003a); darker outlines are younger. Adjacent seafloor with magnetic isochrons and plate boundaries from Müller et al. (1997) and reconstructions at 10 m.y. increments from 0 to 80 Ma based on Müller et al. (1993) and Raymond et al. (2000), interpolated using method of Pilger (2003a). Note increasing overlap of reconstructed West Antarctica for earlier reconstructions.
The resulting apparent overlap of reconstructed West Antarctica with East Antarctic (Fig. 20) is unacceptable, unless significant extension within Antarctica has occurred over the past 80 m.y. – extension for which there is no clear evidence. There is also no evidence for distribution of the “missing” deformation between Antarctica and Africa. Conversely, and more probably, the inconsistency implies that the two hotspot references frames are not fixed relative to one another. Thus the calculated relative motion between West and East Antarctica is largely nonexistent (except for that documented by Cande et al., 2000) and, more profoundly, is a measure of the relative motion between the two hotspot sets.
Intermediate Summary
The analysis above establishes that there are two distinct and well-defined hotspot sets with traces on the plates of the Pacific Ocean and much of the Atlantic and Indian Oceans and adjacent continents. Further, these two sets each can be described in terms of distinct reference frames – that is within each of the two sets hotspots do not appear to have move significantly relative to one another for at least the last 80 m.y.
The two hotspot reference frames are distinct from one another. Reconstructions and kinematic/geometric constraints imply that there has been significant motion between the two hotspot sets over the past 80 m.y., especially between 80 and 40 Ma (Fig. 19, 20). That is, alternate relative plate reconstructions do not appear to account for the discrepancy between the two reference frames.
For the Atlantic-Indian Ocean reference frame, contemporary and paleostress measurements are consistent with continental movement relative to the same reference frame as hotspots. For the Pacific Ocean, distinct lineations in the satellite gravity field are consistent in orientation with Pacific plate motion in the Pacific hotspot reference frame; wavelength of the lineations implies a relatively shallow sublithospheric origin. The stress fields and gravity lineations both imply that the reference frames shared with hotspots extend to relatively shallow depths. These observations have implications for the origin of hotspots and the nature of mantle convection.
Mesoplates and Lithoplates
Lithospheric plates, as original defined early in plate tectonics, are thin, kinematically rigid bodies, segments of spherical shells, in largely circumferential motion relative to one another. Conveniently, their relative motions and reconstructions can be described as instantaneous and finite rotations.
The two defined hotspot reference frames, Atlantic/Indian and Pacific, also provide a basis of describing plate motions (recorded by the hotspot traces, Pacific gravity lineations, and intracontinental stress fields) using rotational parameters. In other words, each hotspot set is semi-rigid and, hypothetically, then, embedded in a plate-like environment. For this reason, Pilger (2003a, b) suggested that the habitat of the hotspot reference frames consists of “mesoplates” in contrast with “lithoplates” (lithospheric plates). As noted above, three such mesoplates are proposed to have existed for at least the past 80 m.y.: Tristan (named after Tristan da Cunha, one of the hotspots of the Atlantic/Indian Ocean set), Hawaiian (the Pacific Ocean set), and Icelandic (beneath Iceland, Eurasia, the Arctic, and portions of northeastern North America and the northernmost Atlantic Ocean).
Mesoplate Boundaries
If the hypothesis that two (or three) distinct hotspot reference frames exist, and that the reference frames are “embedded” in “mesoplates”, what are the boundaries of mesoplates? What are the top and bottom? What are the boundaries between mesoplates? What happens along the boundaries?
Empirically, it is apparent that fundamental boundaries could separate the two well-defined hotspot sets (Tristan and Hawaiian): subduction zones. That is, the Nazca-South American, Cocos-Central American, and Pacific-Australian (Indian) subduction zones separate the two hotspot sets (with the possible exception of the Tasman and East Australian Highlands hotspots). Further, the Pacific-Eurasian and Indian/Australian-Eurasian convergence zones provide a loose boundary between the Tristan and Hawaiian sets from sub-Eurasian mantle (part of the inferred Icelandic mesoplate). To the extent that subduction zones extend well into the mesosphere, such subducting plates provide a “natural” wall separating mesosphere on either side.
Since stress field and gravity lineation evidence suggest a shallow upper surface of “mesoplates”, it can be inferred that medium-to-deep subduction zones extend well below the upper surface of mesoplates. Further, assuming that partially molten asthenosphere separates lithosphere from “solid” mesosphere, it is reasonable to further infer that the upper surface of the mesosphere is the upper boundary of mesoplates.
What of boundaries away from subduction zones? It seems likely that such boundaries are kinematically and geometrically determined. That is, the location and configuration of a shear boundary between mesoplates might well be analogous to transform faults between lithoplates. While the rheology at upper mesospheric depths remains speculative, it is possible that the shear boundary could be constantly changing as the relative kinematics varies. Or, shear boundary segments could connect divergent boundaries, with isostatic upwelling of deeper mantle to fill the gap. More questionable is what happens at convergent boundaries between mesoplates. A kind of deeper subduction might result, although if the convergence of mesoplates corresponds with converging lithoplates, a “double” subduction could occur.
What of the lower boundaries of mesoplates? The 410, or 600 km seismic velocity discontinuities are each likely candidates. While phase or chemical changes are variously invoked to explain these boundaries, might they also be tectonic? That is, like the boundary between lithoplate and underlying asthenosphere, might one or even all of these discontinuities represent surfaces of parallel displacement?
Mesoplate Relative Motions
The discrepancy between the Tristan and Hawaiian hotspot reference frames cannot be accounted for by deformation within the circuit Africa-East Antarctica-West Antarctica-Pacific, as shown above. Therefore, the two reference frames, interpreted as mesoplates, are in motion relative to one another. Along with Fig. 20, Fig. 19 also includes a measure of the relative motion. The locus of Pacific plate motion relative to the Tristan reference frame implies a small amount of displacement over the past ~25 m.y., with significant displacement between ~25 and ~80 Ma.
The question has been raised as to whether the Hawaiian-Emperor and other island seamount chains of the Pacific really record plate motions, since the bend in the former chain, if 43 Ma in age, is not accompanied by other relative plate motion changes in the circum-Pacific (e.g., Norton, 1995). With an older age of the bend now seeming likely (e.g., Clague, 2002), this argument is no longer relevant. As Norton (2000) noted, there are bends in loci of relative motion of the Pacific plate and adjacent continental plates, especially North and South American (e.g., Fig. 19). Subparallelism of the two Pacific-American loci with Pacific-Hawaiian loci is striking, with bends of the same age (47-48 Ma) as the revised age of the Hawaiian-Emperor bend. The plate-to-plate loci are constructed entirely independent of the loci of Pacific plate to hotspot motion.
How could there be a genetic connection between continental-oceanic plate motion and oceanic plate-hotspot motion. Pilger (2003a,b) has suggested that westward motion of the two American plates relative to Tristan between ~75 and ~25 Ma has not only resulted in the advance of the convergent zones between the American continental plates and the Farallon and Kula oceanic plates, it defected the sublithospheric mantle beneath those two plates and the adjacent Pacific plate – the Hawaiian mesoplate, with the subduction zone between the Pacific and Eurasian plate serve as a “guide” or constraint on the motion. Thus, the locus of Pacific plate motion relative to the Tristan mesoplate is subparallel with plate motion of North and South America relative to the Tristan mesoplate over the same time period. Further, the period ~75 to ~40 Ma corresponds with the Laramide event in the western North American Cordillera, during which low-angle subduction beneath the central and southern Rocky Mountains is inferred to have occurred. The low-angle subducting slab may have “filled the gap” between the diverging Tristan and Hawaiian mesoplates – a gap implied by the relative mesoplate motions (Pilger, 2003b).
Speculation?
The section above might seem entirely speculative, but it is important to recognize that the speculation is rooted in empirical observation. There are two (maybe three) distinct hotspot sets “embedded” in upper mesosphere; each hotspot set produces traces that record lithoplate motion relative to not merely the hotspots but the embedding mesosphere (as implied by stress indicators and gravity lineations). Medium to deep subduction zones do not occur in the middle of hotspot sets. Seismic discontinuities occur at appropriate depths to provide lower bounds of mesoplates.
Implications for the Origin of Hotspots
Two possible origins for hotspots are actively advocated at present: deep mantle plumes and lithospheric fracturing. A third hypothesis has a smaller following: isostatically- induced depressurization melting. Consider the implications of the mesoplate plate hypothesis (which is largely a kinematic proposal, with geodynamic consequences).
Plumes and Fixed Hotspots
Morgan (1972) coupled apparent fixity of hotspots with the mantle plume hypothesis. However, the rationale for this coupling was not completely elaborated. Subsequent physical and numerical modeling suggested, in contrast, that mantle plumes could not produce a fixed hotspot reference frame. Nevertheless, there is a geometric and kinematic argument that could support the deep mantle plume hypothesis if, indeed, hotspots form a globally-fixed reference frame.
The thought experiment-argument is as follows: Continent-dipping subduction zones bound much of the circum-Pacific throughout the Cenozoic and, perhaps, at least the late Mesozoic. Expansion of the Atlantic Ocean by seafloor spreading results in contraction of the Pacific Ocean basins. Shallow mesosphere beneath is thereby compromised, and its displacement by asymmetrically encroaching subduction zones is required. If hotspots are embedded in the shallow mesosphere beneath the Pacific, they must be displaced along with the embedding mesosphere relative to mesosphere beneath the surrounding continental plates (and above the subduction zones). If, however, plate reconstructions and hotspot traces imply that Pacific and Atlantic/Indian Ocean hotspots are fixed relative to one another, then the hotspots cannot be embedded. Rather, jet-like plumes, derived from the deeper mantle and fixed in location relative to one another, could penetrate shallow mesosphere, and not be displaced by the shallow “moving” mantle.
Of course, the evidence assembled above argues against hotspots forming a single distinct relatively fixed reference frame. At least two, probably three, distinct reference frames exist. And, as shown below, the relative displacement of two of the reference frames further implies embedding of the hotspots within shallow mesosphere and displacement by precisely the mechanism postulated in the gedankenexperiment in the previous paragraph.
Could plumes still be responsible for hotspots, even if they do not form a single fixed reference frame? Again, numerical and physical modeling would support such an inference. However, the evidence assembled herein argues that whatever their origin, the hotspots (plumes?) are embedded in kinematically rigid shallow mesosphere.
Propagating Fractures
That “hotspot traces” represent propagating fractures has been a popular explanation since even before plate tectonics. However, the hypothesis is faced with at least two challenges, as noted above:
1. The plate-hotspot models of both the Hawaiian and Tristan hotspot sets appear to be consistent with observed ages when “poor” data are excluded (as evidenced above). That is, propagation rates are consistent with plate motions relative to relatively fixed hotspots (in each set). Would one expect propagation rates to be spherically consistent over the entire dimensions of large plates?
2. More critically, there is evidence that plate boundaries have passed over hotspots, or even remained relatively stationary over hotspots for significant periods of time. Hotspots which “passed beneath” plate boundaries include: Great Meteor (New England Seamounts), Reunion (Chagos-Laccadive-Deccan), and Kerguelen (Ninetyeast) (e.g., Müller et al. , 1993). The Easter hotspot was beneath the Pacific-Farallon (Nazca) spreading center from between isochrons 21 (or older) and 11, before passing entirely beneath the Nazca plate (Fig. 21, after Pilger and Handschumacher, 1981). The Iceland hotspot also appears to have been “centered” beneath the Mid-Atlantic Ridge since at least 13 Ma. The Cocos and Carnegie Ridges appear to have formed from the Galapagos hotspot when it was beneath the Cocos-Nazca ridge.
Figure 21. Schematic evolution of the Nazca Ridge (N.R.) and Tuamotu Ridge (T.R.) from the Easter hotspot (after Pilger and Handschumacher, 1981). Reconstructions for indicated magnetic isochrons with ages (Ma) in parentheses. From prior to isochron 13 (perhaps as early as isochron 22) until isochron 11, the Easter hotspot was overlain by the Pacific-Farallon ridge, with “mirror-image” Tuamotu and Nazca Ridges resulting. By isochron 9 a transform fault moved over the hotspot and subsequently, up to the present, the Farallon/Nazca plate has overlain the hotspot.
It is difficult to understanding how propagating fractures alone could explain the ridge-crossing and ridge-centered effects described above. Stresses across the spreading center would be significantly (if not “infinitely”) greater than intraplate stresses inferred to produce propagating fractures.This is not to say that fracturing plays a role in hotspot magmatism. Fractures are required to allow magma to penetrate lithoplates. Further, in the context of the third model, discussed below, extensional fracturing might even contribute to the development or enhancement of a hotspot melting anomaly.
Isostatic Depressurization
In the Pacific, many of the island-seamount chains occur in a distinctive tectonic environment. Their oldest extents occur on the south side of fracture zone separating older lithosphere on the north from younger on the south. Similarly, the gravity lineations over the Pacific plate (Haxby and Weissel, 1986) appear to be best developed in a similar tectonic environment; conversely, few lineations appear on the south side of fracture zones separating younger plate on the north from older on the south.
Raddick et al. (2002) suggested that the observed tectonic position of many of the hotspot traces could be a result of depressurization melting. Asthenosphere beneath older lithosphere occurs at a greater depth than that beneath younger (and, therefore, thinner) lithosphere. Thus, as a fracture zone separating older from younger plate passes over a fertile or anomalously warmer region of the mantle, consequent relative isostatic uplift could result in enhanced partial melting of the rising asthenosphere (and, perhaps, partial melting of underlying mesosphere – converting the latter to asthenosphere).
The model of Raddick et al. could be thought of as a generalization of the depressurization melting mechanism already invoked for active rifting (e.g. White and McKenzie, 1989). In the case of the East Africa rift, the progressive inception of volcanism at the same rate as motion of the African plate, according to the model of Müller et al. (1993), could reflect depressurization melting of fertile asthenosphere/mesosphere facilitated by progressive extension of African lithosphere.
Depressurization melting alone is inadequate to explain hotspots, since anomalous melting on the younger side of faults (when plate motion is to the older side) is limited to observed traces. Some sort of fertile or warmer zone also seems to be required. One candidate for such a fertile zone could be remnant asthenosphere trapped beneath an abandoned spreading center. Others have suggested that ancient remnant subduction zones preserved within the upper mantle could also be candidates (e.g., Foulger and Anderson, 2003).
Top Down
Montelli et al.’s (2003) recent report of tomographic evidence for plumes certainly provides intriguing support for the plume hypothesis. What is striking about the images, however, is the near verticality of each of the images beneath identified hotspots. In other words, contrary to physical and numerical models, the plumes inferred from the images are not significantly “tilted” as might be expected from their being “entrained” in convecting upper mantle.
While the images provide evidence for plumes, the lack of deep vertical extent of many of the images does not necessarily support plume provenance from the lower mantle. Is it possible that plumes have a shallower origin?
A proposed mechanism is as follows: Rapid isostatic rise of shallow, relatively fertile mantle occurs as a result of either rapid lithospheric thinning or passage of thinned plate over the fertile region. The rising region as it melts is converted to asthenosphere. Melt is drained off vertically (forming a hotspot) and laterally into adjacent asthenosphere. Consequently, more isostatic uplift beneath the melting region occurs, with flow from progressively deeper mesosphere resulting. Thus, it is envisioned that a plume is formed from top of the mesosphere down into deeper mantle. The rising deeper mantle in the developing vertical conduit may, then, be explicitly more fertile, if only because it is hotter.
Implicitly, then, lateral flow of shallow mesosphere to accommodate the isostatically rising material is limited. Flow is inferred to be almost entirely vertical. This inference is compatible with the inference of stable mesoplates in the upper mesosphere. Mantle convection is of a very different sort than commonly visualized. Mesoplate motions, like lithoplate motions, are circumferential, with rise of deeper mantle into zones of divergence of mesoplates and, to a much lesser extent, in top-down isostatically driven plumes. Probably a relatively high viscosity for the upper mesosphere is implied, as long as temperatures and pressures remain static. Reduction of pressure might, then, trigger rapid reduction in viscosity accompanied by partial melting at progressively shallower depths.
This qualitative model for plumes is derived from the implications of the numerical modeling of Raddick et al. (2002) and the earlier work of White and McKenzie (1989). Clearly, further numerical modeling of this proposed mechanism is appropriate.
Another conceptually-related mechanism for plume formation builds upon an idea of Schouten et al. (1987). That is, should a spreading ridge segment become stationary relative to underlying mantle (mesosphere in the current formulation), replenishment of asthenosphere could result in tapping of progressively deeper mantle, especially if rapid spreading is occurring. Fig. 22 illustrates restoration of east-central Pacific magnetic isochrons according to Raymond et al. (2000). Note correspondence of relatively stationary segments close to the inferred Easter hotspot for times during which the “mirror-image” Tuamotu and Nazca ridges were formed.
Figure 22. Reconstructed magnetic isochrons (Müller et al., 1997) and hotspot loci relative to Easter hotspot according to Raymond et al. (2000) model, interpolated using method of Pilger (2003a). Note convergence of isochrons between 21 and 5 near the inferred hotspot. Locus on west is relative to Pacific plate; locus on left is relative to Nazca/Farallon plate.
Summary and ConclusionsHotspots are tracers in the mantle. For geochemists and petrologists this assertion has been a critical assumption in research on the nature of the material that comprises the mantle. But, hotspots are also tracers of relative motions of the upper mesosphere and the overlying lithosphere.
The inference that hotspots form a single, fixed reference frame can not longer been supported by relative plate reconstructions, hotspot trace morphology, and available isotopic dating. However, the existence of two well-defined hotspot reference frames is, nevertheless, is strongly supported for at least the past 80 m.y. When contemporary and paleostress orientations, Pacific gravity lineations, and reconstructions of the Caribbean subduction zone are added, there is compelling evidence that the outer mesosphere behaves in a plate-like manner.
The observations and inferences presented herein have significant implications for mantle convection models as well as the origin of hotspots. The more speculative hypotheses advanced herein may prove inapplicable to the earth, but the assembled evidence must be taken into account in alternative earth models. The mesoplate hypothesis, with its more tenuous corollary – top down origin of plumes – are offered as heuristic steps towards improved models for the evolution of the earth over the past few hundred millions of years.
References
Baksi, A.K. 1999. Reevaluation of plate motion models based on hotspot tracks in the Atlantic and Indian Oceans. Jour. Geol. 107, 13-26.
Cande, S.C., Kent, D.V. 1995. Revised calibration of the geomagnetic time scale for the Late Cretaceous and Cenozoic. Jour. Geophys. Res. 100, 6093-6095.
Cande, S.C., Stock, J.M., Müller, R.D., Ishihara, T. 2000. Cenozoic motion between East and West Antarctica. Nature 404, 145-150.
Clague, D. 2003. New ages from the Hawaiian-Emperor chain and relations to other Pacific linear chains. (abs.) Hawaii Focus Group. Link: http://www.mantleplumes.org/HawaiiFocusGroup.html.
Davis, A.S., Gray, L.B., Clague, D.A., Hein, J.R. 2002. The Line Islands revisited: New 40Ar/39Ar geochronologic evidence for episodes of volcanism due to lithospheric extension. Geochem., Geophys., Geosystems 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. 74, 29-42.
Duncan, R.A., Richards, M.A. 1991. Hotspots, mantle plumes, flood basalts, and true polar wander. Rev. Geophys. 29, 31-50.
Foulger, G.R., Anderson, D.L. 2005. A cool model for the Iceland hotspot, J. Volc. Geotherm. Res., 141, 1-22.
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. Am. Geophys. Union Geophys. Mon. 121, 189-210.
Gripp, A.E., Gordon, R.G. 2000. Current plate velocities relative to the hotspots incorporating the NUVEL-1 global plate motion model. Geophys. Res. Lett., 17, 1109-1112.
Harada, Y., Hamano, Y., 2000, Recent progress on plate motions relative to hotspots. Am. Geophys. Union Geophys. Mon. 121., 327-338.
Haxby, W.F., Weissel, J.F. 1986. Evidence for small-scale mantle convection from Seasat altimetry data. Jour. Geophys. Res. 91, 3507-3520.
Kárason, H., Van der Hilst, R.D. 2000, Constraints on mantle convection from seismic tomography. Am. Geophys. Union Geophys. Mon. 121, 277-288.
Koppers, A.A.P., Phipps Morgan, J., Morgan, J.W., Staudigel, H., 2001. Testing the fixed hotspot hypothesis using 40Ar/39Ar age progressions along seamount trails. Earth Planet. Sci. Lett., 185, 237–252.
Koppers, A.A P., Staudigel, H., Pringle, M.S., Wijbrans, J.R., 2003. Short-lived and discontinuous intraplate volcanism in the South Pacific: Hot spots or extensional volcanism? Geochem., Geophys., Geosystems 4, 1089, doi:10.1029/2003GC000533.
Koppers, A.A P., Staudigel, H., 2005. Asynchronous Bends in Pacific Seamount Trails: A Case for Extensional Volcanism? Science, 307, 94.
Liu, H.S., Kolenkiewicz, R., Li, J.L., Chen , J.Z. 2003. Satellite detection of the convection generated stresses in Earth. Recent Res. Developments in Geophys. 5, 117-137.
McAdoo, D.C., Sandwell, D.T. 1989. On the source of cross-grain lineations in the central Pacific gravity field. Jour. Geophys. Res. 94, 9341-9352.
McKenzie, D.P., Morgan, W.J. 1969. Evolution of triple junctions. Nature 224,125-133.
Molnar, P., Atwater, T. 1973. Relative motion of hot spots in the mantle. Nature 246, 288-291.
Montelli, R., Nolet, G., Dahlen, F. A., Masters, G., Engdahl, E. R., Hung, S-H. 2003. Finite-frequency tomography reveals a variety of plumes in the mantle. Science, 10.1126, 1092485, 1-6.
Morgan, W.J. 1971. Convection plumes in the lower mantle. Nature, 230, 42-43.
Morgan, W.J. 1972. Plate motions and deep mantle convection. Geol. Soc. Am. Mem. 132, 7-22.
Morgan, W.J. 1981. Hotspot traces and the opening of the Atlantic and Indian Oceans. In: Emiliani, C (Ed.) The Sea: The Oceanic Lithosphere. Wiley-Interscience, NY, 7, 443-475.
Morgan, W.J. 1983. Hotspot tracks and the early rifting of the Atlantic. Tectonophysics, 94, 123-139.
Müller, R.D., Royer, J.-Y., Lawver, L.A. 1993. Revised plate motions relative to the hotspots from combined Atlantic and Indian Ocean hotspot tracks. Geology 21, 275-278.
Müller, R.D., Roest, W.R., Royer, J.-Y., Gahagan, L.M., Sclater, J.G. 1997. Digital isochrons of the world's ocean floor, Jour. Geophys. Res. 102, 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 14, 1080-1094.
Norton, I.O. 2000. Global hotspot reference frames and plate motion. Am. Geophys. Union Geophys. Mon. 121, p. 339-357.
Pilger, R.H. 1982. The origin of hotspot traces: evidence from Eastern Australia. Jour. of Geophys. Res. 87, 1825-1834.
Pilger, R.H. 2003a. Geokinematics: Prelude to Geodynamics. Springer-Verlag, Berlin, 338 p.
Pilger, R.H. 2003b. Mesoplates: resolving a decades-old controversy, EOS, Trans., Am. Geophys. Union, 84, 573-576.
Pilger, R.H., 2004. Cordilleran tctonics: The roles of lithoplate and mesoplate boundaries. (abs.) Geol. Soc. Am. Ann. Meeting, Denver, CO, 217-10, http://gsa.confex.com/gsa/2004AM/finalprogram/abstract_73220.htm.
Pilger, R.H., 2005a. The Bend: origin and significance, in review.
Pilger, R.H., 2005b. Caribbean back-arcs and East African Rift volcanism: mesoplate tracers, in review.
Pilger, R.H., 2005c. Comment on: “Asynchronous bends in Pacific seamount trails: a case for extensional volcanism?” by Anthony A. P. Koppers and Hubert Staudigel, Science, 307, 94 (2005), in review.
Pilger, R.H., Handschumacher, D.W. 1981. The fixed hotspot hypothesis and origin of the Easter-Sala y Gomez trace. Geol. Soc. Am. Bulletin, 92, 437-446.
Pindell, J., Kennan, L. 2001. Kinematic evolution of the Gulf of Mexico and Caribbean, Trans., Gulf Coast Sect. Soc. Econ. Paleon. Min. Found., 21st Ann. Res. Conf. Trans., p. 193-220. Link: http://www.tectonicanalysis.com/site/download/papercarib01.pdf.
Pindell, J., Kennan, L. 2002. Plate Model for the Caribbean. Link:
http://www.ig.utexas.edu/CaribPlate/forum/pindell/pindell_kennan.htm.
Raddick, M.J., Parmentier, E.M., Scheirer, D.S., 2002. Buoyant decompression melting: A possible mechanism for intraplate volcanism. Jour. Geophys. Res. 107, doi: 10.1029/2001JB000617.
Raymond, C.A , Stock, J.M., Cande, S.A. 2000. Fast Paleogene motion of the Pacific hotspots from revised global plate circuit constraints. Am. Geophys. Union Geophys. Mon. 121, 359-375.
Sandwell, D.T., Smith W.H.F. 1997. Marine gravity anomaly from Geosat and ERS 1 satellite altimetry, Jour. Geophys. Res. 102, 10039-10054. Data: http://topex.ucsd.edu/marine_grav/mar_grav.html.
Schouten, H., Dick, H.J.B., Klitgord, K.C. 1987. Migration of mid-ocean-ridge volcanic segments. Nature 326, 835-839.
Sharp, W.D., Clague, D.A. 1999. A new older age of ~47 Ma for the Hawaiian-Emperor bend. (abs.) EOS, Trans. Am. Geophys. Union, 80. F1196.
Smith, W.H.F., Sandwell D.T. 1997. Global seafloor topography from satellite altimetry and ship depth soundings. Science, 277, 1957-1962. Data: http://topex.ucsd.edu/marine_grav/mar_grav.html.
Wessel, P., Lyons, S. 1997. Distribution of large Pacific seamounts from Geosat/ERS-1: Implications for the history of intraplate volcanism. Jour. Geophys. Res. 102, 22,459-22,475. Data: http://www.soest.hawaii.edu/pwessel/papers/1997/JGR/WL_seamounts_1997.d.
Wessel, P., Kroenke, L.W. 1997. A geometric technique for relocating hotspots and refining absolute plate motions. Nature, 387, 365-369.
Wessel, P., Harada, Y. & Kroenke, L.W., 2004, Improved Absolute Plate Motion Modeling in the Pacific. EOS Trans. Am. Geophys. Union, 85 (47), Fall Meet. Suppl., Abs. V51B-0539.
White, R.S., McKenzie, D.P. 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood basalts, Jour. Geophys. Res. 94, 7685-7729.
Wilson, J.T. 1966. Did the Atlantic close and then re-open? Nature, 211, 676-681.
World Stress Map Project, 2003. http://www.world-stress-map.org/.
Zartman, R.E., Bush, C.A., Abston, Carl, 1995.National Geochronological and Natural Radioelement Data Bases, A, National Geochronological Data Base, B, Natural Radioelement Data Base: U.S. Geol. Surv. Digital Data Ser. DDS-14. [CD-ROM]
Zoback, M.L., et al. 1989. Global patterns of tectonic stress, Nature, 341, 291-298.
© 2010 Rex H. Pilger, Jr.
No comments:
Post a Comment