Moving hotspots

The assumption that stationary hotspots underlie the Earth’s
lithospheric plates has been most important in the development of
the theory of plate tectonics. According to the fixed hotspot
hypothesis seamount trails are formed by volcanism penetrating the
lithospheric plates whilst moving over ”hotspots”of upwelling
mantle. In turn, the azimuths and age progressions of seamount
trails can be used to quantify plate motions with respect to an
independent reference frame of hotspots in the mantle. Also,
assuming fixed hotspots, the direction of characteristic remanent
magnetization in the basalts acquired during cooling should always
be the same. Even if due to plate motion the products of the
hotspot are located far away from the position of the hotspot
itself, paleomagnetic studies on the basalts must always provide
the position of the hotspot itself. Recently the question arose,
why a hotspot with its origin deep in the mantle would not get
advected in the convecting mantle of the Earth. - In this thesis a
possible motion of the Kerguelen hotspot in the southern Indian
Ocean and of the Louisville hotspot in the Pacific has been studied.
The Kerguelen hotspot is active since approximately 117 Ma. Since
then it formed the Kerguelen Plateau and the Broken Ridge in the
southern Indian Ocean as well as the Ninetyeast Ridge, which is the
hotspot track going north up to India, and the Ramajal Traps in
India. Drilling into basement rocks of Broken Ridge and the
Kerguelen Plateau was aim of the Ocean Drilling Program, Leg 183,
from December 1998 to February 1999. Eight sites have been drilled.
In seven of the sites also the sediments have been recovered. In
this thesis, a possible motion of the Kerguelen hotspot has been
studied by determining its paleolatitudes. First, basalts from the
Kerguelen Plateau have been studied paleomagnetically to compare
the paleolatitudes with the latitude of the hotspot itself.
Basement from a drillsite on the central Kerguelen Plateau (Site
1138) and of a site on the northern Kerguelen Plateau (Site 1140)
were suitable for a determination of paleolatitudes. A sufficient
number of independent lavaflows has been penetrated and sampled
there to properly average out paleosecular variation, an important
requirement for determining paleolatitudes. The characteristic
magnetization from the subaerial Site 1138 with AA- and Pahoehoe
lava and of the submarine Site 1140 with its pillow basalts is
carried by magnetite and titanomagnetites and -maghemites and
consists of a single remanence component with sometimes a small
viscous overprint, that could easily be removed during
demagnetization. Stepwise demagnetization in an alternating field
and stepwise heating of the specimens provided the inclination
value of the characteristic magnetization very precisely with small
error. Conversion of the mean-site inclination into the
paleolatitude of a site provided a latitude of λ = 43.6S (max.:
47.8S; min.: 37.9S) for Site 1138 on the central Kerguelen
Plateau and a latitude of λ = 35.8S (max: 43.0S; min.: 28.9S)
for Site 1140 on the northern Kerguelen Plateau. In Site 1136 on
the southern Kerguelen Plateau only two lava flows have been
sampled. Therefore paleosecular variation could not be averaged out
properly. Site 1142 on the Broken Ridge has been tilted and
deformed tectonically after its formation, as was found from
seismic explorations prior to drilling, and the inclination of the
magnetization could therefore not be used for a determination of
paleolatitudes. Compared to the latitude of the Kerguelen hotspot
at 49S, the paleolatitudes of the central and northern Kerguelen
Plateau are further north. This result agrees with previous
paleomagnetic studies on the southern Kerguelen Plateau and the
Ninetyeast Ridge, where paleolatitudes have been found that
indicate also a formation north of the present-day hotspot
position. This difference indicates a southward movement of the
hotspot since the Cretaceous relative to the spin axis of the
Earth. The motion can be explained with a rotation of the whole
mantle of the Earth relative to the spin axis (true polar wander)
or with a motion of the hotspot within the Earth’s mantle.
Therefore, the possibility was studied whether true polar wander
can be responsible for the difference between the paleomagnetic data
and the present-day latitude of the hotspot. Three independently
obtained true polar wander paths have been used, that describe the
motion of the whole mantle (with the hotspots) relative to the
rotation or dipole axis. All three curves point to a shift of the
mantle at the time when the central and southern Kerguelen Plateau
formed in such a way that higher southern paleolatitudes should be
observed. This prediction is just the opposite to what was found in
the paleomagnetic studies. The Cenozoic parts of the three
experimentally obtained true polar wander paths roughly agree
within their uncertainties with a numerically calculated path that
accounts for changes of moments of inertia of the mantle. This
means that the difference between paleomagnetic data and the
present-day position of the hotspot can not be explained by true
polar wander. The next starting point to explain the discrepancy is
hotspot motion. For the determination of hotspot drift, geodynamic
modeling has been carried out. Assuming that a mantle plume rising
from the core-mantle boundary is advected in an convecting mantle,
a hotspot sould move relative to the surface of the Earth. Seismic
tomography models were converted into density models of the Earth’s
mantle. Then a velocity field derived from the mass motion due to
the density heterogeneities is calculated. The rising mantle plume
is then inserted into the model and becomes advected in the
velocity field. Seven different tomographic models have been used to
obtain velocity fields. All seven models result in a southward
motion for the Kerguelen hotspot since its first appearance
approximately 117 Ma ago. The motion is in a similar direction for
the different models, and its magnitude varies from 5 to over 10
degrees. So far, the program to model the hotspot drift assumed a
constant viscosity within the rising plume. More realistic is the
assumption of a depth-dependent plume radius, based on estimates of
temperature- and hence viscosity variations within the plume. This
has been integrated as a subroutine into the program. The plume
radius affects the buoyancy of the plume. A plume with larger radius
rises faster through the mantle, and will hence have a stronger
tendency to straighten up. In contrast, a plume with smaller radius
rises slowly and will be influenced more strongly by the velocity
field of the mantle. Allowing for the variation of viscosity within
the plume, the hotspot motion was calculated again. A comparison of
the resulting hotspot motion for various input parameters showed
that the result is rather independent of the parameters. The
calculations also yield a southward motion of 5 to 10 degrees, only
the shape of the hotspot path is somewhat changed. This southward
motion of the Kerguelen hotspot by 5 to 10 degrees can explain the
difference between the paleomagnetic data and the present-day
position of the hotspot. Even combined with true polar wander it
fits the paleomagnetic results, although true polar wander, taken by
itself, even increases the difference that has to be explained. The
consistency of paleomagnetic results with the model calculations
allows the conclusion that the Kerguelen hotspot indeed moved
southward by some degrees since its first occurence 117 Ma ago. A
magnetostratigraphy has been made using the sediments of ODP Leg
183. It yielded a contribution to the age dating of the basalts
prior to 40Ar/39Ar dating. Paleomagnetic studies on the sediments
contributed to a combined Bio/Magnetostratigraphy. The stratigraphy
helps to determine the minimal age of the underlying basalts. Using
the reversals found in the magnetization and a correlation with the
paleontological data, the lowermost sediments of Site 1136
(southern Kerguelen Plateau) are dated to have an age in the Early
Cretaceous, Site 1138 (central Kerguelen Plateau) in the Late
Cretaceous, and Site 1140 (northern Kerguelen Plateau) in the
Oligocene. These results are meanwhile confirmed by precise
40Ar/39Ar age dating of the basement yielding an age of 100 Ma for
Site 1138 and of 35 Ma for Site 1140. The Ontong Java Plateau, a
Large Igneous Province in the western Pacific, was thought to be
formed by the rising mantle plume of the Louisville hotspot
approximately 120 Ma ago. However, according to a recent plate
reconstruction, the plateau has been formed well to the north of
the location of this hotspot. In this thesis it could be shown that
the formation of the Ontong Java Plateau by the Louisville hotspot
is possible if hotspot motion in the convecting mantle is allowed.
For this purpose, the motion of the Louisville hotspot for the last
120 Ma years has been modeled, using the same method as already
applied for the Kerguelen hotspot. The calculations indicate, that
the Louisville hotspot has probably shifted by some degrees to the
south since its first occurence approximately 120 Ma ago. There is a
considerable variation between different model results, though. The
Louisville hotspot is now located too far south to be responsible
for the formation of the Plateau. However, it could have been in
the right place at the time of the formation 120 Ma ago if hotspot
motion is considered. This is an example that the drift of hotspots
can affect plate tectonics and tectonic reconstructions and that it
should be considered.