- Email: [email protected]
Thermal anomalies and magmatism due to lithospheric doubling and shifting
322
Earth and Planetary Science Letters, 65 (1983) 322-330 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
[21
Thermal anomalies and magmatism due to lithospheric doubling and shifting N.J. Vlaar Vening Meinesz Laboratory, Instituut voor Aardwetenschappen, University of Utrecht, Budapestlaan 4, 3584 CD Utrecht (The Netherlands)
Received March 8, 1983 Revised version accepted July 11, 1983
We present some thermal and magmatic consequences of the processes of lithospheric doubling and lithospheric shifting. Lithospheric doubling concerns the obduction of a cold continental or old oceanic lithospheric plate over a young and hot oceanic lithosphere/upper mantle system, including an oceanic ridge. Lithospheric shifting concerns the translation and rotation of a lithospheric plate relative to the upper mantle. In both cases the resulting thermal state of the upper mantle below the obducting or shifting lithosphere may be perturbed relative to a "normal" continental or oceanic geothermal situation. The perturbed geothermal state gives rise to a density inversion at depth and thus induces a vertical gravitational instability which favours magmatism. We speculate about the magmatic consequences of this situation and infer that in the case of lithospheric doubling our model may account for the petrology and geochemistryof the resulting magma. The original layering and composition of the overridden young oceanic lithosphere may strongly influence magmatic processes. We dwell shortly on the genesis of kimberlites within the framework of our lithospheric doubling model and on magmatism in general. Lithospheric recyclingis inherent to the mechanism of lithospheric doubling.
1. Introduction W h e n the n o r m a l c o m p o n e n t of the convergence velocity at a destructive m a r g i n is larger than the c o m p o n e n t in the same direction of the half spreading velocity of the adjacent oceanic lithosphere, the latter is b e c o m i n g younger with increasing time when being subducted. Eventually, s u b d u c t i o n of the oceanic ridge itself is b o u n d to occur. This may have h a p p e n e d throughout geological history whenever a passive a n d an active m a r g i n adjacent to an active oceanic spreading ridge approach, at the expense of the oceanic lithosphere in between being consumed. I n recent geological times this has taken place at the A l e u t i a n - A l a s k a n , the S u m a t r a a n d the 0012-821X/83/$03.00
© 1983 Elsevier Science Publishers B.V.
South Chile trenches, a n d along the west coast of N o r t h America, as m a y be witnessed from the magnetic a n o m a l y p a t t e r n of the adjacent oceanic lithosphere. Vlaar a n d Wortel [1] and Wortel [2] d e m o n strated c o n v i n c i n g l y that the s u b d u c t i o n behaviour d e p e n d s crucially o n the age of the oceanic lithosphere when subducted, the deepest earthquakes occurring in a s u b d u c t i o n zone being a u n i q u e f u n c t i o n of age of the oceanic lithosphere, j u s t prior to subduction, a n d that y o u n g oceanic lithosphere is s u b d u c t e d subhorizontally. Vlaar [3] a n d O x b u r g h and P a r m e n t i e r [4] proposed gravitational stability of y o u n g oceanic lithosphere with respect to the u n d e r l y i n g upper m a n tle. The latter authors gave evidence that stability
323 persisted for 20-40 Ma after creation of oceanic lithosphere at the ridge. A vertical stratification of the oceanic lithos p h e r e / u p p e r mantle system of the oceanic ridge and its adjacent young flanks then must result in subhorizontal subduction underneath the upper lithospheric plate. This process, leading to horizontal layering of an upper continental or oceanic lithosphere on top of a young oceanic lithosphere/ upper mantle system, we term lithospheric doubling. We assume implicitly that the lower, hot and weak lithosphere/upper mantle system is being subducted integrally, or alternatively, that the upper continental or oceanic lithospheric plate is obducted over a stationary young oceanic lithosphere. In the following we will take the obduction model to represent reality most closely. We postulate that the process of lithospheric doubling, i.e. obduction of an old, cool and strong upper lithosphere over a young, stably stratified, buoyant and weak oceanic lithosphere upper mantle system is a key element in global tectonics and has occurred throughout the geological history. In the present paper we study some thermal and magmatic consequences of this process and of the alternative one of lithospheric shifting.
2. The thermal model
In the present model we assume lithospheric doubling to take place instantaneously, thereby avoiding the complications of relative motion of parts of the system, while cooling takes place. Thus, a spatially stationary model is considered. This appears to be justified by assuming that, after a short period of active doubling, a stationary state sets in. This might be due, for instance, to an overall reorganization of the plate tectonic setting over larger areas. We also assume that spreading at the ocean ridge ceases when being overridden. Slab pull vanishes as adjacent lithosphere is buoyant and also ridge push must become negligible, as the topography, characteristic for an oceanic lithosphere is suppressed by the obducting stiff plate. The thermal model, therefore, involves the cooling of the instantaneously doubled system.
We shall consider the two cases of obduction of a mature oceanic lithosphere of 100 Ma, and of a continental shield lithosphere, both on a young ( < 20 Ma) oceanic lithosphere/upper mantle system. The two cases differ with respect to their geotherms to be employed as part of the initial geotherm of the cooling doubled system. For the young ( < 20 Ma) oceanic system, prior to instantaneous doubling, we take a layer of 100 km depth, which has cooled upon spreading from an oceanic ridge. The surface temperature of oceanic ridge basalts can be put between 1200 and 1350°C [51. Diapirism emplaces fresh crustal material at the ridge. The rising column is sufficiently buoyant at depth to allow for diapirism. Ito [6] presents a model of dry subsolidus diapiric uprising of peridotite until partial melting takes place above solidus temperatures. Latent heat of fusion lowers the temperature slightly in disfavour of buoyancy. The dry peridotite solidus gives a temperature of about 1600°C at 100 km depth, according to Ito and Kennedy [7]. In our model we take tentatively for the initial geotherm underneath the ridge a reasonable 1300°C at the surface, a constant geothermal gradient of 3°C km -1 and hence a temperature of 1600°C at 100 km depth. The cooling of this layer is governed by a one-dimensional heat conduction equation which is solved numerically for t' = 5 and 20 Ma where t' = 0 is the instant of spreading. We, therefore, consider young oceanic lithosphere. We have taken a thermal diffusivity of 0.9 × 10 -2 cm 2 s -1, a value which will also be employed for the cooling of the doubled system. For the case of 100 Ma oceanic lithosphere, according to Crough [8] we take a layer thickness of 100 km, a temperature of 1200°C at its base, and a corresponding geotherm fitting oceanic heat flow data. For the case of an obducting shield lithosphere we take the rather arbitrary thickness of 100 km, and adapt for our purpose the usual shield geotherm with a temperature of 900°C at 100 km depth. Instantaneous doubling takes place at t = 0. We consider two separate doubled systems, differing in the obducted lithospheric plate being a mature oceanic or a shield lithosphere, respectively. The
324 ~rJ~P. "c
~ ",. ~ ...
I
,.:.c
\ , ,\~ \\ \\) ~
~r,,~
i
Ii I I
' ............
~oo~ \~°~%-~o Wet
II
%,
I
//
%
\\ \\
%% \\
(b)
%. '.
\
\
\
\\\
',.
I
".....
Solius~~~ d
(O)
%
~
-
\
\
(c)
Fig. 1. Oceanic lithosphere of 100 Ma on oceanic lithosphere of 0, 5 and 20 Ma, 0, 2, 10, 50 and 200 Ma after doubling. Dry and wet peridotite solidi from Green and Liebermann [20]. No radiogenic heat production taken into account.
initial geotherm at t = 0 is determined by the geotherms of the obducting plates, and of the subducted oceanic l i t h o s p h e r e / u p p e r mantle system at various ages t' after its creation. The doubled cooling system for t > 0 will be demonstrated to have a very large thermal time
constant. We, therefore, put a realistic b o u n d a r y condition at its base (at 200 k m in the doubled system). This condition, after transient cooling has taken place, determines the steady state continental geotherm as t ~ o¢. W e take as a boundary condition 2.5°C k m -1 at 200 k m depth, as is
IEH?. "C
[IMp. "0
i-, I '%
I "... I .. I " I
f
\,,
~F
iN§I a~
"\
Hydrated
I
, ~
..~ ..¢
I I I
~r
,....
~'\,\ 500",
/
\ ,
\\
\
\\
\'\
\
eo0\,\ '\\
\
(a)
.,
[
(b)
~[
(c)
'\
\\
\,\
Fig. 2. Continental shield or craton lithosphere on oceanic lithosphere of 0, 5 and 20 Ma, 0, 2, 10, 50, 200 Ma after doubling. Dry and wet peridotite solidi from Green and Liebermann [20]. No radiogenic heat production taken into account.
325 appropriate for a shield geotherm. As only cooling is our concern we neglect radiogenic heat production at this stage. The cooling history for t > 0 (Figs. 1, 2) for various ages of the underlying young oceanic system, t' = 0, 5, 20 Ma, is given by the geotherms at t = 0, 2, 10, 50, 200, 500 Ma after doubling. For t ~ oo geotherm is linear with a gradient of 2.5°C km -1. From Figs. 1 and 2 it pertains that stationarity has not been attained after 200 Ma and not even after 500 Ma. The extremely large thermal relaxation time is compatible with the 800 Ma reported by Sclater et al. [9] for the time to reach constant heat flow in continental regions. Taking thicker layers, the relaxation time increases as the square of the upper lithospheric layer thickness. From Figs. 1 and 2 several conclusions may be drawn: (1) The underlying oceanic lithosphere acquired a hydrated top layer when it was emplaced at an oceanic spreading centre. We put the thickness of this layer tentatively at 10 km. Taking its solidus temperature at the solidus of wet peridotite, it appears that melting occurs. According to Fig. la, at the location of the former oceanic ridge, obducted by a mature oceanic lithosphere, melting takes place between t = 0 and t = 100 Ma after doubling. However, at the flanks of the former ridge at t ' = 5 Ma (Fig. lb), melting may take place marginally between 10 and 50 Ma after doubling, and at t' = 20 Ma (Fig. lc), the melting point is not reached at all. In the case under consideration, though magmatism may result from melting, the small temperature contrast between upper and lower lithospheric plates may not favour large-scale diapiric uprising in the solid state from the lower one. (2) Obduction of a continental (Fig. 2) shield over an oceanic system, the upper plate being cool, results in stronger cooling of the top layers of the lower plate. Notwithstanding this circumstance, melting takes place during the first 50 Ma after doubling near the overridden oceanic ridge, hence a shorter period than in (1). At the flanks, melting may not take place at all. However, given the larger temperature contrast between upper and lower lithospheric plates, and
the resulting density inversion at depths exceeding 100 km, large-scale diapirism of the initially solid material is favoured. (3) We have made several simplifying assumptions. The thermal effects of transient dynamic doubling have been neglected as the thermal time constant of the cooling doubled system is large compared to the transient dynamical period. For convergence velocities of 5 cm a - l , the transient relative displacement for a period of 10 Ma is 500 kin. Relative to the thermal time constant of the system, transient phenomena, in general, may be neglected. However, particularly during the initial phase of dynamic doubling complications may' arise which have not been taken into account in the present study. We have not taken into account the effect of cooling due to the endogenic character of fusion. As the total volume of melted material is small relative to the total volume being thermally affected, we consider this effect to be negligible in general. On the other hand, we also neglected shear heating in the dynamical system, which is difficult to estimate. It appears, however, that the hydrated top layers of the lower plate are ideally suited to facilitate relative motion of the plates. We did not take into account either the effect of adiabatic compression, which would add another 50°C to the temperature in the lower plate; this amount probably falls within the uncertainty of absolute temperatures of the initial oceanic system. Allowing for a variable thickness of the upper plate the thermal effects may be altered. Taking for instance a smaller value of the thickness of the upper plate, gravitational instability is enhanced as the temperature contrast between upper and lower plate is increased. A 70-km-thick obducting mature oceanic lithosphere would give rise to a larger density inversion at depth, thus favouring mantle diapirism also in this case.
3. Magmatic implications In the following we look into possible magmatic consequences of the mechanism of lithospheric doubling as presented in the foregoing. For ex-
326
plaining many tectonic and magmatic phenomena often reference is made to enigmatic occurrences of " h o t spots", "mantle plumes" and the like, thereby implying thermal anomalies in the upper mantle. Our model of lithospheric doubling provides a rationale for the existence of just these thermal anomalies and the resulting capability of mantle diapirism and associated magmatism. Continental magmatism often is associated with the occurrence of rift structures. Without going into the matter of the tectonic evolution of rift and graben structures, we restrict ourselves in the present paper to magmatism proper, and in this section in particular to continental magmatism. Our model has the advantage of constraining several parameters which dominate magmatic processes and also of discriminating between heterogeneous mantle sources. Moreover, it gives insight into the thermal evolution of possible magma sources and its influence on magma generation. Our gross upper mantle model consists of two superposed layered lithospheric plates, characterized by a repetition of essentially the same vertical petrological structure. Each plate from top to bottom consists of thin layers of sediment, basalt and depleted peridotite resting on a thick layer of undepleted peridotites, where we have simplified the continental crust of the upper plate. The basalt top layer of the lower plate, if not melted, under ambient pressures at 100 km depth is transformed into eclogite. A parameter which has been kept fixed in the present study is the thickness of the upper lithospheric plate which is put at 100 km. Varying this parameter would result in varying pressure conditions in the hydrated top layers of the lower lithospheric plate, and also in the buoyant capacity of the dry, undepleted peridotite of this lower plate. The hydrated top layers of the lower plate favour local melting, which facilitates uprising of magma. If the upper lithospheric plate has been subject to weakening or fracturing in an earlier tectonic event, magmatism will be facilitated considerably. The circumstance being singularly suited for magmatism to take place, the character of the latter will be strongly influenced by local deviating quantities, the most obvious being the temperature
at which magmas are allowed to ascend and the composition of the sedimentary deposits near the ridge prior to subduction. This composition is variable. A most pertinent aspect of our model is the thermal evolution at the site of prospective magma generation and the inherent heterogeneity of the magma source. In the following we will develop a tentative magmatic scenario consistent with the model of lithosperic doubling and its thermal evolution. It is not ruled out that mixing of sources may give rise to magmas of large complexity. 3.1. Kimberlites
Kimber]ites or diamond pipes are virtually confined to continental platforms and cratons. Their fill consists of a fragmented matrix, rich in incompatible elements and in volatiles, carrying a varying collection of xenoliths brought to the earth's surface from great depth. Some kimberlites are diamondiferous. Eruption appears to have taken place explosively. 0 ~
500 Temp°C I
tO00 1500 I • diamond I 0 depleledperidotite • undepletedperidotite
150 .
.
.
.
.
\
Fig. 3. Graphite-diamond equilibrium (after Carmichael et al. [13]). Continental shield lithosphere on a young oceanic lithosphere ( < 5 Ma), less than 5-10 Ma after doubling. Upper lithospheric thicknesses of 100 km and 150 km are considered. Hydrated layer of 10 km thickness indicated by dashed horizontal lines.
327
In the present study the cases are considered of 100-km and 150-km-thick upper continental shield lithospheres obducted on a young oceanic system ( < 5 m.a.) (Fig. 3). We study in particular the situation shortly (5-10 Ma) after doubling took place. The evolving geotherm for this initial period has a kinky shape, the geotherm in the upper layer (10 km) of the lower lithosphere being lower than the continental shield geotherm and rises rapidly along a very steep gradient to much higher temperatures for greater depths. For the case of the 150-km-thick upper plate, from about 1000 ° to 1400°C, our theoretical geotherm bears a striking resemblance with the geotherm derived by Boyd [10] from pyroxene geothermometry applied to peridotite nodules brought up by kimberlites. These nodules give further confirmation of our model. Kimberlite magma, when rising rapidly through the upper lithospheric plate, samples the wall of the conduit. We, therefore, expect to find undepleted spinel peridotite xenoliths in kimberlites. The main constituents of kimberlites, according to our model are: (1) original sediments of the ocean bottom rich in incompatible elements and fluids and volatiles such as H20, CO 2, CH 4, C1, and F and organic remnants, (2) basalt (eclogite) of the former oceanic crust, (3) depleted peridotite of the next deeper layer, and (4) undepleted peridotite of the bottom layer, initially having risen buoyantly along a very steep thermal gradient in the solid state. This material may have been mixed in a "magma" chamber before kimberlite ascend. We do not wish to speculate on the amount of mixing having taken place nor on its petrological implications. However, temperatures have been fairly low in the initial stages after lithospheric doubling, probably below the solidus of kimberlite magma. The transport must be based on fluidization of the solid material by means of the liquids and volatiles present. Upon decompression, when rising, the magma may become explosive and, apart from xenoliths, liquid. The model description given above is confirmed by petrological findings [11,12]. Leaving apart the peridotite nodules sampled by the rising magma along the conduit walls, the remaining peridotite xenoliths fall into two categories: (1) cooler ( -
IO00°C), coarse-grained periodites depleted in their basaltic constituents, containing a lower concentration of incompatible elements relative to the undepleted kind, and (2) undepleted garnet peridotite having been subject to large deformation in the solid state and having higher temperatures ( - 1400°C). The latter deformation must have taken place by transport along the steep part of the transient geotherm in our model. The coarse-grained texture of the depleted peridotite is to be ascribed to its original emplacement at the oceanic spreading center without subsequent deformation. The difference in incompatible element concentration must be inherent to the segregation having taken place in the original oceanic ridge setting. The kimberlite matrix characterizes also the original ocean bottom environment, which is reflected by a large variation from place to place. The high volatile and incompatible element content are typical for kimberlites. Xenoliths are the peridotite nodules described above, and furthermore nodules of eclogite, calcite, apatite and serpentinite, minerals to be expected in association with an ocean bottom environment. We, therefore, speculate that diamond, CO 2, C H 4 and apatite are organic remnants, also typical for the same setting. The incompatible element content then is indicative for the sedimentary deposits. The occurrence of diamond in kimberlites, from Fig. 3, appears to be determined by some critical parameters: (1) the time elapsed since doubling, and (2) the thickness of the upper lithospheric plate. As the occurrence of kimberlites is tied to continental shield provinces, Fig. 3 has been derived from the case where the upper lithosphere is a continental shield. We consider the cases of 100-km and 150-km-thick upper lithosphere. The transient geotherm is in the diamond field only for a short period (5-10 Ma) after doubling, and for an upper lithospheric thickness in excess of 140 km. After a longer period (10-15 Ma), taking radiogenic heat production into account, and because of the perturbed high-temperature state of the lower lithosphere, the transient geotherm is to be expected to shift towards higher temperatures compared to the standard continental geotherm. From Fig. 2 it
328 follows that the kink in the geotherm is a transient phenomenon of short duration, to which the occurrence of diamonds is related. Peridotite nodules brought to the surface at a later stage after doubling are to be expected not to show a kink in their pyroxene geotherm. We therefore propose: (1) diamonds occur exclusively in association with a kinky pyroxene geotherm, and (2) are tied to a cratonic province with a lithosphere in excess of 140 km, which indeed appears to be the case in nature. A thinner upper lithosphere, i.e., a younger continental platform, would prohibit the occurrence of diamonds. Magmas which appear to be genetically related to kimberlites, are earbonatites, ultrapotassic lavas, and nephelinites [13]. We speculate that these magmas are strongly dominated by the sedimentary setting and contaminated by fluids and incompatible elements of the oceanic environment. Variable mixing of locally different sedimentary remnants with basic and ultrabasic magmas could give rise to a large spectrum of lavas, thus suggesting a heterogeneous upper mantle magma source distribution. We propose that these differences are circumstantial and that the varying sedimentary origin of part of the magma plays a major role with respect to the apparent heterogeneity of lavas.
3.2. Continental flood basalts We consider continental magmatism for a larger part to be caused by magma, originating in our model from the hydrated upper layers of the lower oceanic lithospheric plate. After an initial kimberlitic stage as described in section 3.1, which may be governed by the presence of weakness zones in the upper lithospheric plate on the one hand a n d / o r by the strongly buoyant character caused by a high concentration of volatiles and low density of the magma on the other hand, the evolution of the thermal regime with time gives rise to large-scale melting of the eclogite layer which originally constituted the basaltic oceanic crust. This may take place between 10 and 50 Ma after lithospheric doubling and must be confined to the vicinity of the overridden oceanic ridge. The hydrated basalt layer may have been transformed into hornblende eclogite or water may
have been stored into amphibolite. In any case, we assume that the material of the eclogite layer may be transported diapirically either in the solid state or as a partial melt to shallower levels where complete melting may take place if this did not yet occur.
3.3. Mantle diapirs The magmatism described in the foregoing was associated with fluidization and hydration of the top layers of the lower lithosphere. We should now call attention to the thermally anomalous state which prevails for depths in excess of 100 km, particularly near the former oceanic ridge, persisting for longer periods of time. This thermal anomaly implies a highly gravitationally unstable configuration which facilitates diapirism. We are dealing with the undepleted and depleted peridotite of the lower lithosphere, which is assumed not to be affected by hydration. These rocks are well below the dry solidus of peridotite and hence have to force their way up by deformation in the solid state. Their velocity of ascent then should be low compared to the magma considered in the foregoing. If ever mantle diapirism is to take place, circumstances are most favourable in the situation of lithospheric doubling. The slow rising of the diapir, and consequently its adiabatic and conductive cooling, favours the material to remain in the solid state at subsolidus dry peridotite temperatures. When arriving at the Moho, the latter acts as a barrier against further rising. The high temperature of the diapir results in a lower density than the density of the "normal" temperature peridotite mantle at Moho depth. However, its density is large relative to the density of crustal rocks. The diapir, therefore, may be expected to assume a cushionlike form underneath the Moho resulting in an upbulging of the latter. This combined with anathexis of the lower crust by advective heating, may result in crustal thinning and associated rift formation. Though, at present, we will not speculate further on the consequences of mantle diapirism as described here, its reality appears to be well confirmed by petrological studies as have been summerized by Den Tex [12].
329
3.4. The plate tectonic setting
3.5. Lithosperic shifting
The evidence for the plate tectonic hypothesis is virtually restricted to active spreading centers and subduction zones. Intracontinental tectonics and magmatism and orogenic processes still require a definitive geodynamical explanation. Age-dependent subduction has been thoroughly investigated [1,2,14] particularly for the Pacific/ South American convergence zone over the last 12 Ma. These studies demonstrate an intimate connection between age, and subduction behaviour, including the geological setting in terms of sediment accretion at the trench and of associated volcanism in the continental plate. Dating of oceanic lithosphere is possible only for ages less than at most 200 Ma. A subducted lithospheric plate is seismically active for at most 10 Ma. A fossil subduction zone thus may only be identified by geological evidence. As long as geological consequences of deviations from standard Benioff subduction are not properly known, we do not know where to look for. Therefore, even if at present no discriminating evidence is available, subduction in the past cannot be excluded. This, in particular, holds for lithospheric doubling as presented here, the consequences of which are beginning to be explored. Though the correspondence between subduction and orogeny is not yet properly understood, it is generally assumed to be implicit. As we wish to relate the presence of kimberlites to lithospheric doubling we must infer a fossil subduction zone. Restricting ourselves to the South African kimberlite province we relate it to the Gondwanide orogeny which has been active into the Cretaceous. Cox [15] ascribed the Karroo flood basalts and a related extensional tectonic regime extending to far within the foreland to the presence of a subduction zone at the Pacific-Gondwana margin. We propose that lithospheric doubling is a most suitable model for understanding these geological phenomena. The genesis of kimberlites fits the same model, though we have to infer repeated lithospheric doubling during the Jurassic and Cretaceous.
In the foregoing we presented the mechanism for creating upper mantle thermal anomalies by means of lithospheric doubling. Magmatism may be facilitated when the upper plate is weakened by older tectonic events. In particular, older rift and transform structures may be reactivated by repeated lithospheric doubling. The doubled state need not to be restricted to continents. Being characterized by a thermal anomaly with a large thermal time constant, situated at greater depth, it need not directly be related to the upper plate structure. In view of the life time of the thermal anomaly concerned, lithospheric doubling may be manifest also in the oceanic realm. " H o t spots" may be manifest at the intersections of older tectonic lineations and a linear thermal anomaly which once was produced by lithospheric doubling and overriding of an oceanic ridge. We now introduce another mechanism leading to upper mantle thermal anomalies, which we term "lithospheric shifting". We envisage this to take place when a lithospheric plate translates (or rotates) over an upper mantle which has a fossil geothermal regime not compatible with the geothermal regime of the lithosphere moving over it. This, in particular, may lead to the presence of a thermal anomaly in the upper mantle when an oceanic spreading center is forced to move with respect to the underlying mantle. The spreading process induced an anomalous thermal state at depth due to buoyant rising of hot mantle material. In our oceanic lithosphere model this implies a temperature of some 1600°C at a depth of 100 km. Subsequent translation of the lithosphere (and spreading center) away from this thermally perturbed upper mantle causes cold oceanic or continental lithosphere to reside on hot upper mantle. Though we did not perform any model calculations, it is evident that the shifting process causes stronger thermal anomalies, as temperatures are higher by a few hundreds of degrees, than in the case of doubling. The process of shifting thus may give rise to linear thermal anomalies at depth, not compatible with the thermal state of the overlying lithosphere. The latter then subsequently is thermally affected. If a thermal anomaly as described here hits an old lithospheric zone of
330
weakness, oceanic or continental magmatism may result and may be manifest as a "hot spot" 4
4. Concluding remarks
5 6
The processes of lithospheric doubling and shifting are adequate to account for thermal anomalies in the upper mantle, and probably their magmatic expressions. In this light, the variability of continental and oceanic magmas could be tested. In this context, the role played by former ocean sediments and their organic fluid and gaseous content may lead to new insights in magmatism and the geochemical cycle. The geothermal perturbation of the upper mantle persists over geological periods, conform to the thermal history of continents. No recourse has to be made to hypothetical deep mantle plumes or whole mantle convection current systems. The thermal perturbations are confined to a tectosphere of at most a few hundreds of kilometers thickness, in accordance with Anderson's [16] findings. The hypothesis presented in the present paper may have a strong bearing on the concept o f lithospheric recycling for which increasing geochemical evidence is being presented [17-19].
7 8 9
10 11
12
13 14
15 16 17
References
18 19
1 N.J. Vlaar and M.J.R. Wortel, Lithospheric aging, instability and subduction, Tectonophysics 32, 331-351, 1976. 2 M.J.R. Wortel, Age dependent subduction of oceanic lithosphere, Thesis, State University of Utrecht, 1980. 3 N.J. Vlaar, The driving mechanism of plate tectonics: a qualitative approach, in: Progress in Geodynamics, G.J.
20
Borradaile, A.R. Ritsema, H.E. Rondeel and O.J. Simon, eds., North-Holland, Amsterdam, 1975. E.R. Oxburgh and E.M. Parmentier, J. Geol. Soc. London 133, 343, 1979. K.F. Scheidegger, Temperatures and composition of magma, J. Geophys. Res. 78, 8735, 1973. K. Ito, Analytical approach to estimating the source rock of basaltic magmas, J. Geophys. Res. 78, 412, 1973. K. Ito and G.C. Kennedy, Melting and phase relations in a natural peridotite to 40 kilobars, Am. J. Sci. 265, 519, 1967. S.T. Crough, Thermal model of oceanic lithosphere, Nature 256, 388, 1975. J.G. Sclater, C. Jaupart and D. Galson, The heat flow through oceanic and continental crust, Rev. Geophys. Space Phys. 18, 269, 1980. F.R. Boyd, A pyroxene geotherm, Geochim. Cosmochim. Acta 37, 2533, 1973. A.M. Boullier and A. Nicolas, Classification of textures and fabrics of peridotite xenoliths from South African kimberlites, Phys. Chem. Earth 9, 467, 1975. E. den Tex, Dynamothermal metamorphism across the continental crust/mantle interface, Fortschr. Mineral. 60, 57, 1982. J.S. Carmichael, F.J. Turner and J. Verhogen, Igneous Petrology, McGraw-Hill, New York, N.Y., 1974. S.A.P.L. Cloetingh, M.J.R. Wortel and N.J. Vlaar, Evolution of passive continental margins and initiation of subduction, Nature 297, 139 1982. K.G. Cox, Flood basalts, subduction and break-up of Gondwanaland, Nature 274, 47, 1978. D.L. Anderson, Hotspots, basalts, and the evolution of the mantle, Science 213, 82, 1981. C.G. Chase, Oceanic island Pb: two-stage histories and mantle evolution, Earth Planet. Sci. Lett. 52, 277, 1981. A.W. Hofmann and W.M. Wight, Mantle plumes from ancient ocean crust, Earth Planet. Sci. Lett. 57, 421, 1982. L. Dosso and V. Rama Murthy, Recycled old lithosphere as the source of modern ocean basalts (abstract), EOS 613, 1982. D.H. Green and R.C. Liebermann, Phase equilibria and elastic properties of a pyrolite model for the oceanic upper mantle, Tectonophysics 32, 61, 1976.
Earth and Planetary Science Letters, 65 (1983) 322-330 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
[21
Thermal anomalies and magmatism due to lithospheric doubling and shifting N.J. Vlaar Vening Meinesz Laboratory, Instituut voor Aardwetenschappen, University of Utrecht, Budapestlaan 4, 3584 CD Utrecht (The Netherlands)
Received March 8, 1983 Revised version accepted July 11, 1983
We present some thermal and magmatic consequences of the processes of lithospheric doubling and lithospheric shifting. Lithospheric doubling concerns the obduction of a cold continental or old oceanic lithospheric plate over a young and hot oceanic lithosphere/upper mantle system, including an oceanic ridge. Lithospheric shifting concerns the translation and rotation of a lithospheric plate relative to the upper mantle. In both cases the resulting thermal state of the upper mantle below the obducting or shifting lithosphere may be perturbed relative to a "normal" continental or oceanic geothermal situation. The perturbed geothermal state gives rise to a density inversion at depth and thus induces a vertical gravitational instability which favours magmatism. We speculate about the magmatic consequences of this situation and infer that in the case of lithospheric doubling our model may account for the petrology and geochemistryof the resulting magma. The original layering and composition of the overridden young oceanic lithosphere may strongly influence magmatic processes. We dwell shortly on the genesis of kimberlites within the framework of our lithospheric doubling model and on magmatism in general. Lithospheric recyclingis inherent to the mechanism of lithospheric doubling.
1. Introduction W h e n the n o r m a l c o m p o n e n t of the convergence velocity at a destructive m a r g i n is larger than the c o m p o n e n t in the same direction of the half spreading velocity of the adjacent oceanic lithosphere, the latter is b e c o m i n g younger with increasing time when being subducted. Eventually, s u b d u c t i o n of the oceanic ridge itself is b o u n d to occur. This may have h a p p e n e d throughout geological history whenever a passive a n d an active m a r g i n adjacent to an active oceanic spreading ridge approach, at the expense of the oceanic lithosphere in between being consumed. I n recent geological times this has taken place at the A l e u t i a n - A l a s k a n , the S u m a t r a a n d the 0012-821X/83/$03.00
© 1983 Elsevier Science Publishers B.V.
South Chile trenches, a n d along the west coast of N o r t h America, as m a y be witnessed from the magnetic a n o m a l y p a t t e r n of the adjacent oceanic lithosphere. Vlaar a n d Wortel [1] and Wortel [2] d e m o n strated c o n v i n c i n g l y that the s u b d u c t i o n behaviour d e p e n d s crucially o n the age of the oceanic lithosphere when subducted, the deepest earthquakes occurring in a s u b d u c t i o n zone being a u n i q u e f u n c t i o n of age of the oceanic lithosphere, j u s t prior to subduction, a n d that y o u n g oceanic lithosphere is s u b d u c t e d subhorizontally. Vlaar [3] a n d O x b u r g h and P a r m e n t i e r [4] proposed gravitational stability of y o u n g oceanic lithosphere with respect to the u n d e r l y i n g upper m a n tle. The latter authors gave evidence that stability
323 persisted for 20-40 Ma after creation of oceanic lithosphere at the ridge. A vertical stratification of the oceanic lithos p h e r e / u p p e r mantle system of the oceanic ridge and its adjacent young flanks then must result in subhorizontal subduction underneath the upper lithospheric plate. This process, leading to horizontal layering of an upper continental or oceanic lithosphere on top of a young oceanic lithosphere/ upper mantle system, we term lithospheric doubling. We assume implicitly that the lower, hot and weak lithosphere/upper mantle system is being subducted integrally, or alternatively, that the upper continental or oceanic lithospheric plate is obducted over a stationary young oceanic lithosphere. In the following we will take the obduction model to represent reality most closely. We postulate that the process of lithospheric doubling, i.e. obduction of an old, cool and strong upper lithosphere over a young, stably stratified, buoyant and weak oceanic lithosphere upper mantle system is a key element in global tectonics and has occurred throughout the geological history. In the present paper we study some thermal and magmatic consequences of this process and of the alternative one of lithospheric shifting.
2. The thermal model
In the present model we assume lithospheric doubling to take place instantaneously, thereby avoiding the complications of relative motion of parts of the system, while cooling takes place. Thus, a spatially stationary model is considered. This appears to be justified by assuming that, after a short period of active doubling, a stationary state sets in. This might be due, for instance, to an overall reorganization of the plate tectonic setting over larger areas. We also assume that spreading at the ocean ridge ceases when being overridden. Slab pull vanishes as adjacent lithosphere is buoyant and also ridge push must become negligible, as the topography, characteristic for an oceanic lithosphere is suppressed by the obducting stiff plate. The thermal model, therefore, involves the cooling of the instantaneously doubled system.
We shall consider the two cases of obduction of a mature oceanic lithosphere of 100 Ma, and of a continental shield lithosphere, both on a young ( < 20 Ma) oceanic lithosphere/upper mantle system. The two cases differ with respect to their geotherms to be employed as part of the initial geotherm of the cooling doubled system. For the young ( < 20 Ma) oceanic system, prior to instantaneous doubling, we take a layer of 100 km depth, which has cooled upon spreading from an oceanic ridge. The surface temperature of oceanic ridge basalts can be put between 1200 and 1350°C [51. Diapirism emplaces fresh crustal material at the ridge. The rising column is sufficiently buoyant at depth to allow for diapirism. Ito [6] presents a model of dry subsolidus diapiric uprising of peridotite until partial melting takes place above solidus temperatures. Latent heat of fusion lowers the temperature slightly in disfavour of buoyancy. The dry peridotite solidus gives a temperature of about 1600°C at 100 km depth, according to Ito and Kennedy [7]. In our model we take tentatively for the initial geotherm underneath the ridge a reasonable 1300°C at the surface, a constant geothermal gradient of 3°C km -1 and hence a temperature of 1600°C at 100 km depth. The cooling of this layer is governed by a one-dimensional heat conduction equation which is solved numerically for t' = 5 and 20 Ma where t' = 0 is the instant of spreading. We, therefore, consider young oceanic lithosphere. We have taken a thermal diffusivity of 0.9 × 10 -2 cm 2 s -1, a value which will also be employed for the cooling of the doubled system. For the case of 100 Ma oceanic lithosphere, according to Crough [8] we take a layer thickness of 100 km, a temperature of 1200°C at its base, and a corresponding geotherm fitting oceanic heat flow data. For the case of an obducting shield lithosphere we take the rather arbitrary thickness of 100 km, and adapt for our purpose the usual shield geotherm with a temperature of 900°C at 100 km depth. Instantaneous doubling takes place at t = 0. We consider two separate doubled systems, differing in the obducted lithospheric plate being a mature oceanic or a shield lithosphere, respectively. The
324 ~rJ~P. "c
~ ",. ~ ...
I
,.:.c
\ , ,\~ \\ \\) ~
~r,,~
i
Ii I I
' ............
~oo~ \~°~%-~o Wet
II
%,
I
//
%
\\ \\
%% \\
(b)
%. '.
\
\
\
\\\
',.
I
".....
Solius~~~ d
(O)
%
~
-
\
\
(c)
Fig. 1. Oceanic lithosphere of 100 Ma on oceanic lithosphere of 0, 5 and 20 Ma, 0, 2, 10, 50 and 200 Ma after doubling. Dry and wet peridotite solidi from Green and Liebermann [20]. No radiogenic heat production taken into account.
initial geotherm at t = 0 is determined by the geotherms of the obducting plates, and of the subducted oceanic l i t h o s p h e r e / u p p e r mantle system at various ages t' after its creation. The doubled cooling system for t > 0 will be demonstrated to have a very large thermal time
constant. We, therefore, put a realistic b o u n d a r y condition at its base (at 200 k m in the doubled system). This condition, after transient cooling has taken place, determines the steady state continental geotherm as t ~ o¢. W e take as a boundary condition 2.5°C k m -1 at 200 k m depth, as is
IEH?. "C
[IMp. "0
i-, I '%
I "... I .. I " I
f
\,,
~F
iN§I a~
"\
Hydrated
I
, ~
..~ ..¢
I I I
~r
,....
~'\,\ 500",
/
\ ,
\\
\
\\
\'\
\
eo0\,\ '\\
\
(a)
.,
[
(b)
~[
(c)
'\
\\
\,\
Fig. 2. Continental shield or craton lithosphere on oceanic lithosphere of 0, 5 and 20 Ma, 0, 2, 10, 50, 200 Ma after doubling. Dry and wet peridotite solidi from Green and Liebermann [20]. No radiogenic heat production taken into account.
325 appropriate for a shield geotherm. As only cooling is our concern we neglect radiogenic heat production at this stage. The cooling history for t > 0 (Figs. 1, 2) for various ages of the underlying young oceanic system, t' = 0, 5, 20 Ma, is given by the geotherms at t = 0, 2, 10, 50, 200, 500 Ma after doubling. For t ~ oo geotherm is linear with a gradient of 2.5°C km -1. From Figs. 1 and 2 it pertains that stationarity has not been attained after 200 Ma and not even after 500 Ma. The extremely large thermal relaxation time is compatible with the 800 Ma reported by Sclater et al. [9] for the time to reach constant heat flow in continental regions. Taking thicker layers, the relaxation time increases as the square of the upper lithospheric layer thickness. From Figs. 1 and 2 several conclusions may be drawn: (1) The underlying oceanic lithosphere acquired a hydrated top layer when it was emplaced at an oceanic spreading centre. We put the thickness of this layer tentatively at 10 km. Taking its solidus temperature at the solidus of wet peridotite, it appears that melting occurs. According to Fig. la, at the location of the former oceanic ridge, obducted by a mature oceanic lithosphere, melting takes place between t = 0 and t = 100 Ma after doubling. However, at the flanks of the former ridge at t ' = 5 Ma (Fig. lb), melting may take place marginally between 10 and 50 Ma after doubling, and at t' = 20 Ma (Fig. lc), the melting point is not reached at all. In the case under consideration, though magmatism may result from melting, the small temperature contrast between upper and lower lithospheric plates may not favour large-scale diapiric uprising in the solid state from the lower one. (2) Obduction of a continental (Fig. 2) shield over an oceanic system, the upper plate being cool, results in stronger cooling of the top layers of the lower plate. Notwithstanding this circumstance, melting takes place during the first 50 Ma after doubling near the overridden oceanic ridge, hence a shorter period than in (1). At the flanks, melting may not take place at all. However, given the larger temperature contrast between upper and lower lithospheric plates, and
the resulting density inversion at depths exceeding 100 km, large-scale diapirism of the initially solid material is favoured. (3) We have made several simplifying assumptions. The thermal effects of transient dynamic doubling have been neglected as the thermal time constant of the cooling doubled system is large compared to the transient dynamical period. For convergence velocities of 5 cm a - l , the transient relative displacement for a period of 10 Ma is 500 kin. Relative to the thermal time constant of the system, transient phenomena, in general, may be neglected. However, particularly during the initial phase of dynamic doubling complications may' arise which have not been taken into account in the present study. We have not taken into account the effect of cooling due to the endogenic character of fusion. As the total volume of melted material is small relative to the total volume being thermally affected, we consider this effect to be negligible in general. On the other hand, we also neglected shear heating in the dynamical system, which is difficult to estimate. It appears, however, that the hydrated top layers of the lower plate are ideally suited to facilitate relative motion of the plates. We did not take into account either the effect of adiabatic compression, which would add another 50°C to the temperature in the lower plate; this amount probably falls within the uncertainty of absolute temperatures of the initial oceanic system. Allowing for a variable thickness of the upper plate the thermal effects may be altered. Taking for instance a smaller value of the thickness of the upper plate, gravitational instability is enhanced as the temperature contrast between upper and lower plate is increased. A 70-km-thick obducting mature oceanic lithosphere would give rise to a larger density inversion at depth, thus favouring mantle diapirism also in this case.
3. Magmatic implications In the following we look into possible magmatic consequences of the mechanism of lithospheric doubling as presented in the foregoing. For ex-
326
plaining many tectonic and magmatic phenomena often reference is made to enigmatic occurrences of " h o t spots", "mantle plumes" and the like, thereby implying thermal anomalies in the upper mantle. Our model of lithospheric doubling provides a rationale for the existence of just these thermal anomalies and the resulting capability of mantle diapirism and associated magmatism. Continental magmatism often is associated with the occurrence of rift structures. Without going into the matter of the tectonic evolution of rift and graben structures, we restrict ourselves in the present paper to magmatism proper, and in this section in particular to continental magmatism. Our model has the advantage of constraining several parameters which dominate magmatic processes and also of discriminating between heterogeneous mantle sources. Moreover, it gives insight into the thermal evolution of possible magma sources and its influence on magma generation. Our gross upper mantle model consists of two superposed layered lithospheric plates, characterized by a repetition of essentially the same vertical petrological structure. Each plate from top to bottom consists of thin layers of sediment, basalt and depleted peridotite resting on a thick layer of undepleted peridotites, where we have simplified the continental crust of the upper plate. The basalt top layer of the lower plate, if not melted, under ambient pressures at 100 km depth is transformed into eclogite. A parameter which has been kept fixed in the present study is the thickness of the upper lithospheric plate which is put at 100 km. Varying this parameter would result in varying pressure conditions in the hydrated top layers of the lower lithospheric plate, and also in the buoyant capacity of the dry, undepleted peridotite of this lower plate. The hydrated top layers of the lower plate favour local melting, which facilitates uprising of magma. If the upper lithospheric plate has been subject to weakening or fracturing in an earlier tectonic event, magmatism will be facilitated considerably. The circumstance being singularly suited for magmatism to take place, the character of the latter will be strongly influenced by local deviating quantities, the most obvious being the temperature
at which magmas are allowed to ascend and the composition of the sedimentary deposits near the ridge prior to subduction. This composition is variable. A most pertinent aspect of our model is the thermal evolution at the site of prospective magma generation and the inherent heterogeneity of the magma source. In the following we will develop a tentative magmatic scenario consistent with the model of lithosperic doubling and its thermal evolution. It is not ruled out that mixing of sources may give rise to magmas of large complexity. 3.1. Kimberlites
Kimber]ites or diamond pipes are virtually confined to continental platforms and cratons. Their fill consists of a fragmented matrix, rich in incompatible elements and in volatiles, carrying a varying collection of xenoliths brought to the earth's surface from great depth. Some kimberlites are diamondiferous. Eruption appears to have taken place explosively. 0 ~
500 Temp°C I
tO00 1500 I • diamond I 0 depleledperidotite • undepletedperidotite
150 .
.
.
.
.
\
Fig. 3. Graphite-diamond equilibrium (after Carmichael et al. [13]). Continental shield lithosphere on a young oceanic lithosphere ( < 5 Ma), less than 5-10 Ma after doubling. Upper lithospheric thicknesses of 100 km and 150 km are considered. Hydrated layer of 10 km thickness indicated by dashed horizontal lines.
327
In the present study the cases are considered of 100-km and 150-km-thick upper continental shield lithospheres obducted on a young oceanic system ( < 5 m.a.) (Fig. 3). We study in particular the situation shortly (5-10 Ma) after doubling took place. The evolving geotherm for this initial period has a kinky shape, the geotherm in the upper layer (10 km) of the lower lithosphere being lower than the continental shield geotherm and rises rapidly along a very steep gradient to much higher temperatures for greater depths. For the case of the 150-km-thick upper plate, from about 1000 ° to 1400°C, our theoretical geotherm bears a striking resemblance with the geotherm derived by Boyd [10] from pyroxene geothermometry applied to peridotite nodules brought up by kimberlites. These nodules give further confirmation of our model. Kimberlite magma, when rising rapidly through the upper lithospheric plate, samples the wall of the conduit. We, therefore, expect to find undepleted spinel peridotite xenoliths in kimberlites. The main constituents of kimberlites, according to our model are: (1) original sediments of the ocean bottom rich in incompatible elements and fluids and volatiles such as H20, CO 2, CH 4, C1, and F and organic remnants, (2) basalt (eclogite) of the former oceanic crust, (3) depleted peridotite of the next deeper layer, and (4) undepleted peridotite of the bottom layer, initially having risen buoyantly along a very steep thermal gradient in the solid state. This material may have been mixed in a "magma" chamber before kimberlite ascend. We do not wish to speculate on the amount of mixing having taken place nor on its petrological implications. However, temperatures have been fairly low in the initial stages after lithospheric doubling, probably below the solidus of kimberlite magma. The transport must be based on fluidization of the solid material by means of the liquids and volatiles present. Upon decompression, when rising, the magma may become explosive and, apart from xenoliths, liquid. The model description given above is confirmed by petrological findings [11,12]. Leaving apart the peridotite nodules sampled by the rising magma along the conduit walls, the remaining peridotite xenoliths fall into two categories: (1) cooler ( -
IO00°C), coarse-grained periodites depleted in their basaltic constituents, containing a lower concentration of incompatible elements relative to the undepleted kind, and (2) undepleted garnet peridotite having been subject to large deformation in the solid state and having higher temperatures ( - 1400°C). The latter deformation must have taken place by transport along the steep part of the transient geotherm in our model. The coarse-grained texture of the depleted peridotite is to be ascribed to its original emplacement at the oceanic spreading center without subsequent deformation. The difference in incompatible element concentration must be inherent to the segregation having taken place in the original oceanic ridge setting. The kimberlite matrix characterizes also the original ocean bottom environment, which is reflected by a large variation from place to place. The high volatile and incompatible element content are typical for kimberlites. Xenoliths are the peridotite nodules described above, and furthermore nodules of eclogite, calcite, apatite and serpentinite, minerals to be expected in association with an ocean bottom environment. We, therefore, speculate that diamond, CO 2, C H 4 and apatite are organic remnants, also typical for the same setting. The incompatible element content then is indicative for the sedimentary deposits. The occurrence of diamond in kimberlites, from Fig. 3, appears to be determined by some critical parameters: (1) the time elapsed since doubling, and (2) the thickness of the upper lithospheric plate. As the occurrence of kimberlites is tied to continental shield provinces, Fig. 3 has been derived from the case where the upper lithosphere is a continental shield. We consider the cases of 100-km and 150-km-thick upper lithosphere. The transient geotherm is in the diamond field only for a short period (5-10 Ma) after doubling, and for an upper lithospheric thickness in excess of 140 km. After a longer period (10-15 Ma), taking radiogenic heat production into account, and because of the perturbed high-temperature state of the lower lithosphere, the transient geotherm is to be expected to shift towards higher temperatures compared to the standard continental geotherm. From Fig. 2 it
328 follows that the kink in the geotherm is a transient phenomenon of short duration, to which the occurrence of diamonds is related. Peridotite nodules brought to the surface at a later stage after doubling are to be expected not to show a kink in their pyroxene geotherm. We therefore propose: (1) diamonds occur exclusively in association with a kinky pyroxene geotherm, and (2) are tied to a cratonic province with a lithosphere in excess of 140 km, which indeed appears to be the case in nature. A thinner upper lithosphere, i.e., a younger continental platform, would prohibit the occurrence of diamonds. Magmas which appear to be genetically related to kimberlites, are earbonatites, ultrapotassic lavas, and nephelinites [13]. We speculate that these magmas are strongly dominated by the sedimentary setting and contaminated by fluids and incompatible elements of the oceanic environment. Variable mixing of locally different sedimentary remnants with basic and ultrabasic magmas could give rise to a large spectrum of lavas, thus suggesting a heterogeneous upper mantle magma source distribution. We propose that these differences are circumstantial and that the varying sedimentary origin of part of the magma plays a major role with respect to the apparent heterogeneity of lavas.
3.2. Continental flood basalts We consider continental magmatism for a larger part to be caused by magma, originating in our model from the hydrated upper layers of the lower oceanic lithospheric plate. After an initial kimberlitic stage as described in section 3.1, which may be governed by the presence of weakness zones in the upper lithospheric plate on the one hand a n d / o r by the strongly buoyant character caused by a high concentration of volatiles and low density of the magma on the other hand, the evolution of the thermal regime with time gives rise to large-scale melting of the eclogite layer which originally constituted the basaltic oceanic crust. This may take place between 10 and 50 Ma after lithospheric doubling and must be confined to the vicinity of the overridden oceanic ridge. The hydrated basalt layer may have been transformed into hornblende eclogite or water may
have been stored into amphibolite. In any case, we assume that the material of the eclogite layer may be transported diapirically either in the solid state or as a partial melt to shallower levels where complete melting may take place if this did not yet occur.
3.3. Mantle diapirs The magmatism described in the foregoing was associated with fluidization and hydration of the top layers of the lower lithosphere. We should now call attention to the thermally anomalous state which prevails for depths in excess of 100 km, particularly near the former oceanic ridge, persisting for longer periods of time. This thermal anomaly implies a highly gravitationally unstable configuration which facilitates diapirism. We are dealing with the undepleted and depleted peridotite of the lower lithosphere, which is assumed not to be affected by hydration. These rocks are well below the dry solidus of peridotite and hence have to force their way up by deformation in the solid state. Their velocity of ascent then should be low compared to the magma considered in the foregoing. If ever mantle diapirism is to take place, circumstances are most favourable in the situation of lithospheric doubling. The slow rising of the diapir, and consequently its adiabatic and conductive cooling, favours the material to remain in the solid state at subsolidus dry peridotite temperatures. When arriving at the Moho, the latter acts as a barrier against further rising. The high temperature of the diapir results in a lower density than the density of the "normal" temperature peridotite mantle at Moho depth. However, its density is large relative to the density of crustal rocks. The diapir, therefore, may be expected to assume a cushionlike form underneath the Moho resulting in an upbulging of the latter. This combined with anathexis of the lower crust by advective heating, may result in crustal thinning and associated rift formation. Though, at present, we will not speculate further on the consequences of mantle diapirism as described here, its reality appears to be well confirmed by petrological studies as have been summerized by Den Tex [12].
329
3.4. The plate tectonic setting
3.5. Lithosperic shifting
The evidence for the plate tectonic hypothesis is virtually restricted to active spreading centers and subduction zones. Intracontinental tectonics and magmatism and orogenic processes still require a definitive geodynamical explanation. Age-dependent subduction has been thoroughly investigated [1,2,14] particularly for the Pacific/ South American convergence zone over the last 12 Ma. These studies demonstrate an intimate connection between age, and subduction behaviour, including the geological setting in terms of sediment accretion at the trench and of associated volcanism in the continental plate. Dating of oceanic lithosphere is possible only for ages less than at most 200 Ma. A subducted lithospheric plate is seismically active for at most 10 Ma. A fossil subduction zone thus may only be identified by geological evidence. As long as geological consequences of deviations from standard Benioff subduction are not properly known, we do not know where to look for. Therefore, even if at present no discriminating evidence is available, subduction in the past cannot be excluded. This, in particular, holds for lithospheric doubling as presented here, the consequences of which are beginning to be explored. Though the correspondence between subduction and orogeny is not yet properly understood, it is generally assumed to be implicit. As we wish to relate the presence of kimberlites to lithospheric doubling we must infer a fossil subduction zone. Restricting ourselves to the South African kimberlite province we relate it to the Gondwanide orogeny which has been active into the Cretaceous. Cox [15] ascribed the Karroo flood basalts and a related extensional tectonic regime extending to far within the foreland to the presence of a subduction zone at the Pacific-Gondwana margin. We propose that lithospheric doubling is a most suitable model for understanding these geological phenomena. The genesis of kimberlites fits the same model, though we have to infer repeated lithospheric doubling during the Jurassic and Cretaceous.
In the foregoing we presented the mechanism for creating upper mantle thermal anomalies by means of lithospheric doubling. Magmatism may be facilitated when the upper plate is weakened by older tectonic events. In particular, older rift and transform structures may be reactivated by repeated lithospheric doubling. The doubled state need not to be restricted to continents. Being characterized by a thermal anomaly with a large thermal time constant, situated at greater depth, it need not directly be related to the upper plate structure. In view of the life time of the thermal anomaly concerned, lithospheric doubling may be manifest also in the oceanic realm. " H o t spots" may be manifest at the intersections of older tectonic lineations and a linear thermal anomaly which once was produced by lithospheric doubling and overriding of an oceanic ridge. We now introduce another mechanism leading to upper mantle thermal anomalies, which we term "lithospheric shifting". We envisage this to take place when a lithospheric plate translates (or rotates) over an upper mantle which has a fossil geothermal regime not compatible with the geothermal regime of the lithosphere moving over it. This, in particular, may lead to the presence of a thermal anomaly in the upper mantle when an oceanic spreading center is forced to move with respect to the underlying mantle. The spreading process induced an anomalous thermal state at depth due to buoyant rising of hot mantle material. In our oceanic lithosphere model this implies a temperature of some 1600°C at a depth of 100 km. Subsequent translation of the lithosphere (and spreading center) away from this thermally perturbed upper mantle causes cold oceanic or continental lithosphere to reside on hot upper mantle. Though we did not perform any model calculations, it is evident that the shifting process causes stronger thermal anomalies, as temperatures are higher by a few hundreds of degrees, than in the case of doubling. The process of shifting thus may give rise to linear thermal anomalies at depth, not compatible with the thermal state of the overlying lithosphere. The latter then subsequently is thermally affected. If a thermal anomaly as described here hits an old lithospheric zone of
330
weakness, oceanic or continental magmatism may result and may be manifest as a "hot spot" 4
4. Concluding remarks
5 6
The processes of lithospheric doubling and shifting are adequate to account for thermal anomalies in the upper mantle, and probably their magmatic expressions. In this light, the variability of continental and oceanic magmas could be tested. In this context, the role played by former ocean sediments and their organic fluid and gaseous content may lead to new insights in magmatism and the geochemical cycle. The geothermal perturbation of the upper mantle persists over geological periods, conform to the thermal history of continents. No recourse has to be made to hypothetical deep mantle plumes or whole mantle convection current systems. The thermal perturbations are confined to a tectosphere of at most a few hundreds of kilometers thickness, in accordance with Anderson's [16] findings. The hypothesis presented in the present paper may have a strong bearing on the concept o f lithospheric recycling for which increasing geochemical evidence is being presented [17-19].
7 8 9
10 11
12
13 14
15 16 17
References
18 19
1 N.J. Vlaar and M.J.R. Wortel, Lithospheric aging, instability and subduction, Tectonophysics 32, 331-351, 1976. 2 M.J.R. Wortel, Age dependent subduction of oceanic lithosphere, Thesis, State University of Utrecht, 1980. 3 N.J. Vlaar, The driving mechanism of plate tectonics: a qualitative approach, in: Progress in Geodynamics, G.J.
20
Borradaile, A.R. Ritsema, H.E. Rondeel and O.J. Simon, eds., North-Holland, Amsterdam, 1975. E.R. Oxburgh and E.M. Parmentier, J. Geol. Soc. London 133, 343, 1979. K.F. Scheidegger, Temperatures and composition of magma, J. Geophys. Res. 78, 8735, 1973. K. Ito, Analytical approach to estimating the source rock of basaltic magmas, J. Geophys. Res. 78, 412, 1973. K. Ito and G.C. Kennedy, Melting and phase relations in a natural peridotite to 40 kilobars, Am. J. Sci. 265, 519, 1967. S.T. Crough, Thermal model of oceanic lithosphere, Nature 256, 388, 1975. J.G. Sclater, C. Jaupart and D. Galson, The heat flow through oceanic and continental crust, Rev. Geophys. Space Phys. 18, 269, 1980. F.R. Boyd, A pyroxene geotherm, Geochim. Cosmochim. Acta 37, 2533, 1973. A.M. Boullier and A. Nicolas, Classification of textures and fabrics of peridotite xenoliths from South African kimberlites, Phys. Chem. Earth 9, 467, 1975. E. den Tex, Dynamothermal metamorphism across the continental crust/mantle interface, Fortschr. Mineral. 60, 57, 1982. J.S. Carmichael, F.J. Turner and J. Verhogen, Igneous Petrology, McGraw-Hill, New York, N.Y., 1974. S.A.P.L. Cloetingh, M.J.R. Wortel and N.J. Vlaar, Evolution of passive continental margins and initiation of subduction, Nature 297, 139 1982. K.G. Cox, Flood basalts, subduction and break-up of Gondwanaland, Nature 274, 47, 1978. D.L. Anderson, Hotspots, basalts, and the evolution of the mantle, Science 213, 82, 1981. C.G. Chase, Oceanic island Pb: two-stage histories and mantle evolution, Earth Planet. Sci. Lett. 52, 277, 1981. A.W. Hofmann and W.M. Wight, Mantle plumes from ancient ocean crust, Earth Planet. Sci. Lett. 57, 421, 1982. L. Dosso and V. Rama Murthy, Recycled old lithosphere as the source of modern ocean basalts (abstract), EOS 613, 1982. D.H. Green and R.C. Liebermann, Phase equilibria and elastic properties of a pyrolite model for the oceanic upper mantle, Tectonophysics 32, 61, 1976.