Plate boundaries and extensional tectonics

Plate boundaries and extensional tectonics

Tectonoph_wics. 8 I (1982) Elsevier Scientific 239 239-256 Publishing Company, Amsterdam-Printed in The Netherlands PLATE BOUNDARIES AND EXTE...

1MB Sizes 2 Downloads 97 Views

Tectonoph_wics. 8 I (1982)








in The Netherlands




f ~~ru~o~~e


Je ~~~~a~~~e,





de GCotectonique. Universit+ p. er Al. Curie, 4 Piace Jusieu,

75130 Paris, Cedex 05 (France) ’ Centre Oc&nologique


de Bretagne, Plouzun~, B.P. 337, 29-773 Brest, Cedex (Frunce)

July 9, 1981)


Le Pichon,

X., Angelier,

J. and Sibuet, J.-C.,

Hales (Editor),


The behaviour

of the lithosphere


this level, extensive asthenospheric



is extended


For high strain


and probably


The pattern




Similar intra-continental for oceanic



many orogens

of the Western


accretion whenever

and oceanic

continental brittle


in these widely extended

of the Below





in the African uniformly, portion


of the

IO km thick, is extended

areas is compared

the rate.

may be thinned

The greatest



on the strain

as occurs

is thinned


starts. Above


to a pack of cards

a slight angle with the preceding


the case of the Pindos basin of the Hellenides. 150 to 400 km wide may be created

to begin. the stress

are localized

the surface

the level is reached, probably

rates, the whole lithosphere on many

In: A.L.

in its absence. accretion

the lower lithosphere

at an angle on a plane, with each card (tilted block) making



the upper one is much less affected,

while the upper

of faulting

is different

The thinning


81: 239-256.

by the asthenosphere

For low strain rates:


Using this simple model, we discuss



may be rapidly

of the old lithosphere

and lithospheric

in the Aegean



may break out to’tbe surface.

cases must be distinguished.


Rifts or the Rhine


of the old lithosphere



Two extreme



and extensional


is above or below the level that would be reached

this level, the continuity


1982. Plate boundaries

Final Symposium.

We point system


out that these basins


from extensional

along such zones of earlier extension,

the subsidence

are especially



to compressional.

as shown by geological



Alpine system.


When the Geodynamics project started, global plate kinematics were already fairIy well known and, as a result, the plate tectonic model had been established (Le Pichon et al., 1973). Two major accomplishments during the Geodynamic program were a better understanding of the lithosphere as a dynamic and not a static entity ~-1951/82/~-~/$02.75

Q 1982 Elsevier Scientific




(e.g., Parsons and Richter, 1979) and the unraveling boundaries (Arcyana, 1975; Ballard et al., 1978). Althou~ it had been realized that the evolution controlled made olivine

by its rheology


when the formulation

1967), it is only progressively

led to a better


of the fine structure of the oceanic

of the plate tectonic that

of the complex

the study

of plate


model was first

of the rheology





of the

lithosphere. In a similar way, as most of the well-defined plate boundaries lie hidden below 2.5 to I1 km of water, data obtained from the sea-surface had first led to deceptively simple structural models of these narrow tectonic structures. The FAMOUS project showed that the surface width of an accreting plate boundary is about 20 to 30 km. It is determined by a rather complex distribution of strain, the stress-strain relationship being controlled by the mechanical properties of the lithosphere which are continuous across the plate boundary (Tapponnier and Francheteau, 19’78; Tarantola et al., 1979). It was then progressively realized that the basic problem of global tectonics is the understanding of the distribution of stress and strain over the thickness of the lithosphere along the different plate boundaries. Since the FAMOUS project, other ~~-resolution geological, geophysical and geochemical studies of portions of accreting plate boundaries have been conducted over a wider range of spreading rates (e.g., Cyamex Scientific Team, 198 1). Although such studies have not yet been made on fast-spreading boundaries, we now have a fair knowledge of the structure of accreting plate boundaries. Recently, similar fine-scale studies have been undertaken across consuming plate boundaries (Le Pichon et al., 1980, 1981) which also have a complex distribution of strain which may change in time (Nakamura and Uyeda, 1980). However, in this paper, we wish to discuss a different boundaries, the nascent extensional the intrusion of any asthenospheric

type of active


ones during the continental rifting stage, prior to material. Some of the continental extensional

zones may evolve into passive continental margins once the continuity of the continental material is broken. Quite often, they do not proceed to that extreme stage and form intracontinental extensional zones which play quite a significant role in the evolution of continental areas and of orogens. It may well be that the most significant progress in geodynamics in the next few years will occur in the comprehension of these intracontinental extensional zones. CONTINENTAL





A significant progress in this domain has been the recent introduction by McKenzie (1978a,b) of a simple model of uniform stretching of the lithosphere. Of course, it had long been realized that stretching is the mechanism responsible for both the formation of Rift Valleys and continental margins. Actually, since Heezen (1962), it had been explicitly assumed that the basic genetic sequence is one which by continued stretching evolves from a continental rift valley, similar to the African


ones, to an open ocean continental Bott (1971) had proposed layer


formation genetic



of rift valleys


is because,


that necking,

in the lower



with brittle



the measured



from rift valleys to continental



and Artyushkov behaviour

be invoked

But, curiously, margins

(1971) and

in the upper both

crustal for

led to an impasse.

in the upper brittle


the proposed This

layer of rift valleys

is small, of the order of a few kilometers, it is accompanied by an important thinning of the lithosphere but a much smaller one of the continental crust. Thus, no unique simple process can be assumed to model the evolution of continental rifts. On the other hand, continental rifts are not the only type of structures corresponding to extensional tectonics on the continents. Fairly wide continental areas are affected by rather uniform extensional tectonics, such as the Basin and Range, the region of the Aegean sea, the Danakil and Aisha blocks in Afar. It was an analysis of the evolution of the Aegean Sea which led McKenzie to propose that its subsidence could be explained by uniform stretching of the lithosphere by a factor as high as two (McKenzie, 1978a and b). He pointed out that this model also produced a phase of thermal subsidence, controlled by the thermal time constant of the lithosphere, following the initial relatively short stretching phase. This could then explain the formation of many continental basins in which an initial phase of extension is often present. Christie and Sclater (1980) applied this model to the formation of the North Sea basin; Royden et al. (1980) and Royden and Keen (1980) applied it to the formation of passive continental margins. The main problem in applying McKenzie’s model to the formation of continental margins is that it implies stretching factors as high as 3 or 4. Le Pichon and Sibuet (1981) have discussed the geometry of extension in the brittle upper layer of the Armorican



analysis shows that McKenzie’s They consequently felt justified


the data

of Montadert

et al. (1979).


model is a good first approximation to the data. in exploring its consequences. They pointed out in

particular that the supposed genetic sequence from rift valleys to the oceanic continental margins may not be justified. In this paper, we explore further some consequences of this model for continental extensional tectonics. THE INITIAL STRETCHING


We shall base the following discussion on Le Pichon and Sibuet (1981) using however a different presentation. Let us consider a lithosphere composed of a crust portion of average density p, and a mantle portion of average density p, (Fig. 1); p, and p,,, depend on the actual distribution of temperatures. The corresponding average density of the lithosphere is pr, with: PC

The lithosphere

lies on the asthenosphere

of average density

pa, which is made of the


a Fig. I. Initial Vertical








lent with aa =a,_). b. Lithosphere



c. Lithosphere

with lithosphere

This enables



are a, {asthenosphere),



pm (mantle



p,,, (water). replaced

the “mantle

and h,:



by asthenosphere geoid”


(which is isostatically

and the “asthenosphere

of crust

and lithosphere,


geoid” (see text). respectively).


to (a) is E.

just after inst~taneous

(a) is E,. The subsidence


to define


with respect



pC (crust), pL (whole lithosphere),

a. Hypothetic






by a factor /3_ The new surface elevation

to (b) is Z, (explanations

with respect to

in text).

same material as the mantle portion of the lithosphere but has a higher average temperature, so that: PC -c Pa += &xl

The ratio between the thickness h, of the crust and the total thickness h, of the lithosphere (Fig. lb) determines whether p, is smaller or larger than pt. For a value R of the ratio h,/h,, pa = pL and the lithosphere has no buoyancy. If k,/h, #R, paf-p,_. The ~thosphere has a buoyancy B given by: ~=M4l-PL~ which is positive if h,/h, > R (i.e. pa > pL) and negative otherwise. The surface elevation E of the lithosphere with respect to the elevation E, of the reference lithosphere of density pL = pa is given by: E=4+,


where p, is the density of water and assuming for simplicity of discussion that the highest possible elevation is the sea level.


If uniform


of the whole

r = 0, in such a way that the thickness density


pi_ does not change and the buoyancy

new elevation

En with respect


of the lithosphere is reduced




at time the average

in the same ratio. Thus, the

to E is (Fig. lc):

E,, = E/B Consequently,

the change in surface elevation


Zi = E(l - I/p) There is subsidence ( Zi > 0) if pi_ < p,, and uplift ( Zi < 0) if pL > p,, which reflects the fact that E,, tends asymptotically to 0 when the stretching factor tends toward infinity and the lithosphere is completely replaced by the asthenosphere (Fig. la). Using the constants chosen in a previous paper, based on the isostatic equilibrium between the crest of a mid-ocean ridge 2.5 km deep and a continental crust 30 km thick in a lithosphere 125 km thick at 0 surface elevation (see discussion by Le Pichon

and Sibuet,

198 1, and Table I):

E, =3.6km this is the “mantle

geoid” of Turcotte

et al. (1977). The surface elevation

of a crust is

equal to E, when pL = p,, which gives: /2,/h,

= 0.13

For a lithosphere 125 km thick, h, = 16 km. Consequently, whenever the initial thickness of the crust is greater than 13% of the initial thickness of the lithosphere, which is in general close to 16 km, initial subsidence occurs. When it is smaller, initial uplift occurs. This notion of a reference lithosphere is critical in geodynamics because it is also the boundary between a lithosphere lighter than the asthenosphere, which can only be forcibly subducted, and a lithosphere denser than the asthenosphere, which is in a state of gravitational instability. Furthermore, for a new mid-ocean ridge to form within an old lithosphere, it is in general necessary that the surface elevation of the old lithosphere is brought down to

TABLE I Parameters


h,=125km h,=30km pm =3.35

gcme3 (at O’C)

pc = 2.78 gem-’

(at O°C)

pw = 1.03 gem-’ c~=3.38XlO-~


T, = 1333°C No heat production.





the hydrostatic asthenosphere However, fusion

level to which the asthenosphere is not


to break


there is a slight complication

occurs resulting

in segregation

will rise by itself. Otherwise, the old lithosphere


as the asthenosphere

of an oceanic


to the surface. rises, partial

crust. Consequently,

the hydro-

static level of the asthenosphere

is 2.5 km instead

of 3.6 km so that, actually,




of the old lithosphere

in general


the water



becomes at least equal to 2.5 km before the asthenosphere can break to the surface, the 2.5 km level being the real “asthenosphere geoid” (Fig. 2). This is substantiated by the numerous seismic refraction profiles across continental margins which do not reveal a significant discontinuous change in basement depth when passing from oceanic to continental crust. Consequently, if the level of implacement of new oceanic crust was close to the 2.5 km water depth, that is if it was implaced in local isostatic equilibrium, then subsidence must have proceeded until this 2.5 km level was reached. The main factor controlling the buoyancy of the lithosphere with respect to the asthenosphere is the ratio between thickness of crust h, and thickness of lithosphere h,. A change in this ratio produces very significant changes in the evolution of the subsidence with thinning. Although the subsidence always tends to reach asymptotically 3.6 km (with a “normal” asthenosphere) when the stretching factor becomes infinite, the rate of subsidence of course depends on the initial elevation. Thus, a wide range of different geological cases can be produced by variations in initial elevation, h,/h, ratio and asthenosphere conditions. Figure 3 shows the relation between amount of stretching (1 - l//3) and initial subsidence under water when the initial elevation is 0. In this case, E = 3.6 km and consequently: Zi = 3.6 (1 - l/P)km Note that this value on& depends on the initial elevation and the depth of the mantle geoid, provided the lithosphere is in isostatic equilibrium. It can for example be obtained

for a continental

Fig. 2. The asthenosphere

crust 30 km thick within

geoid. Hatchured


the 2.5 km level when it can break to the surface.


a lithosphere

Black: asthenospheric

125 km thick, as




Fig. 3. Initial subsidence Zi and total subsidence at infinite time Z, as functions of (I- I/p). tions in text.


proposed by Le Pichon and Sibuet (1981) for the Bay of Biscay continental margin. Figure 3 shows that the crust and lithosphere should be reduced in thickness by at least 70% (/3 = 3.27) before the asthenosphere geoid level (2.5 km) is reached and oceanic accretion can begin. Then, continental crust should be thinned to less than 10 km. Note also that if the initial elevation is above sea level: E=3.6+(&-3.6)(p,


and the rate of subsidence shows an increase due to water loading when sea level is passed. THE THERMAL SUBSIDENCE PHASE

Stretching is supposed to be instantaneous so that no cooling can occur. However, as time passes after the initial phase of stretching, the lithosphere progressively returns to its initial thickness by cooling. Accepting a value of 125 km for the equilibrium of the oceanic lithosphere made of asthenospheric material (Parsons and Sclater, 1977), the asymptotic value reached for the infinite stretching case will be obtained by replacing the hot asthenosphere corresponding to the 3.6 E, level by a lithosphere made of mantle material with a linear distribution of temperature. It is easy to compute that, with the values of Table I: E, = 7.8 km

Of course, this zero thickness crust case is actually never reached and the maximum depth is unlikely to exceed significantly the maximum depth of the oceanic lithosphere (6.4 km), for reasons developed above. To a good approximation, then: z,

= E’(I - l/q)

E’ being the surface elevation of the lithosphere with respect to the elevation E, and

assuming that everything is under water. If the initial elevation is at sea-level: z,

= ?.8(I - l/P)

which is the case shown in Fig. 3. The subsidence is multiplied by a factor iarger than 2 (2.16) by the thermal cooling which occurs with a time constant of 62.8 my. according to Parsons and Sclater (1977). Thus, the elevation change due to thermal cooling which is: Z,, = 4.2(1 - l//3) reaches 63% of its maximum vaIue in 63 m-y. and 878 in 125 m.y. Going back to Fig. 3, we see that the 2.5 km initial water depth at which the asthenosphere is able to break through the surface will progressively increase through time to 5.4 km through cooling. We atso see that, after thermaf e~~~briurn has been reached, the 3.6 km level below which the lithosphere has no positive buoyancy and becomes gravitationally unstable occurs for a stretching factor fi of 1.86 or a crust 16 km thick (with the values given in Table I), as discussed previously. This depth corresponds to an initial depth immediately after stretching of 1.66 km. GENERAL



These simple considera~ons have important geodynamic implications and demonstrate the significance of the mantle geoid (3.6 km) and even more of the asthenosphere geoid (2s km), If the surface eIevation of a lithosphere is significantly deeper than 2.5 km (but of course smaller than 3,6 km), as stretching proceeds, the asthenospheric material will tend to rise and its ascension will be helped by its density lighter than the density of the adjacent lithospheric material. It will be helped further by the buoyancy of the melted portion produced by partial fusion during the ascension. By the time it reaches the mantle-crust interface, the melted portion is so large (about 30% see Ahem and Turcotte, 1979) and its density so low (2.6 gem-‘) that it should penetrate within the crust and rise up to the surface. Thus no significant thinning of the lithosphere is likely to occur and the transition to accretion should be rather sharp.

If the surface elevation is much less than 2.5 km, on the other hand, the asthenospheric material cannot reach the surface level and break the continuity of the lithosphere. In addition, as the crust-mantle interface is situated deeper, its density contrast is a more efficient barrier with respect to any possible asthenospheric intrusion which contains a much smaller proportion of partial fusion than in the previous case. Thus, stretching will be able to proceed until the water depth gets in the vicinity of 2.5 km. It is unlikely that it will greatly exceed this amount as the likelihood of the asthenosphere breaking to the surface increases rapidly beyond this point. Note that, in addition, when the water depth is larger than 3.6 km, uplift of the surface results, whereas when it is smaller subsidence occurs. Consequently, energy considerations suggest that stretching is likelier to occur in the shallow case than in the deep case. Note further that, similarly, under compression, simple subduction will tend to occur when the water depth is deeper than 3.6 km, whereas it is increasingly difficult above this level. It is interesting to consider what happens if the mantle portion of the lithosphere lets a large diapir of asthenosphere rise in the absence of significant stretching. This is the case corresponding for example to the African Rift Valley which we may describe by a normal thickness crust (30 km) overlying a greatly thinned mantle portion of ~thosphere (30 km). The corresponding surface elevation then is 1.3 km and a stretching factor of 5 is now required to produce the 2.5 km water depth which permits oceanic creation to take place, instead of 3.2 in the normal thickness lithosphere case. Oceanic accretion is thus less likely to occur. Actually, the African rifts as well as the Rhine Graben have been active extensional structures for tens of millions of years (e.g., Le Pichon et al., 1973) but very limited extension has occurred. This limited extension acting over geologically long periods of time may lead to intrusions of asthenospheric diapirs accompanied by sinking of portions of deep lithosphere. On the other hand, if a large amount of extension is applied in a time which is geologically short (an average of 1.4 over a width of 300 km and a time of 13 m.y. in Aegea: Le Pichon and Angelier, 1981), the entire lithosphere appears to be thinned at a uniform rate (McKenzie, 1978a,b) and rapid subsidence occurs, eventually leading to oceanic accretion. Intermediate situations with larger thinning in the lower portion of the lithosphere than in the upper one may occur for intermediate rates of stretching. The extensional rates suggested are 10 -I5 see-’ for uniform stretching in the Aegean case and in the Rhine Graben case 10 -I6 set-’ or less for lower lithosphere thinning. We suggest consequently that the evolution of extensional geological structures depends on the rate at which extension is applied, The Rhine Graben and African Rift Valley are not prototypes of early stages of continental margin formation. The extensional rate applied is too small and has resulted instead in an overall uplift which renders later evolution to an oceanic ~thosphere less likely. Rather, the early


stages of continental margin formation correspond to areas under geologically rapid extension as Aegea nowadays although intermediate stages may occur between these two extreme stages. GEOMETRY OF EXTENSION

It was pointed out that the simple uniform stretching model is a good approximation to the history of subsidence of Aegea, as shown by McKenzie (1978a,b), Le Pichon and Angelier (198 1) and Angelier (1981). There, the surface extension produced by normal faulting in the upper layer is in fair agreement with the extension predicted by the model (Angelier, 1979, 1981). In the Aegean region, the largest coefficient of surface extension measured at regional scale on the basis of field observations is only j? = 1.4 to 1.5, as demonstrated in Crete (Angeher, 1979). The areas affected by larger extension lie under water where field observations cannot be made; it is likely, however, that even in these submerged areas of Aegea, /I is generally smaller than 2. Morton and Black (1975) have shown that in the Aisha area of the Afar region, surface extension may reach values as high as p = 3. Similarly, in the Basin and Range, Proffett (1977) has measured surface extension reaching a p as high as 2.4. Thus, there is no doubt that /3 as large as 3 can be obtained by normal faulting in the upper brittle layer of the crust, Le Pichon and Sibuet (1981), using data of Montadert et al. (1979) have shown, on the basis of seismic profiles, that the extension, on the Armorican continental margin, can reach values larger than 3. They have shown further that, to a first appro~mation, the surface extension is equal to the one predicted by the uniform stretching model for the whole lithosphere. Figure4 is a theoretical model which fits the Armorican margin results of Le Pichon and Sibuet (1981). It can be compared to a pack of cards resting at an angle on a plane, with each card making a slight angle df? with the preceding one. The original angle of faulting is 45” over the main part of the fault. As the angle of tilting 8 approaches the maximum value, 30”, the horizontal extension increases very rapidly with: sin (Y

P- sin(a+B)

(Ybeing the supplement of the angle of the fault plane with respect to the bedding plane. Thus, with cw= 135’, and 8 = 30”, p = 2.73. The increase in B from one block to the other requires brittle deformation in the upper portion and probably plastic deformation in the lower portion which produce an apparent concave curvature of the fault planes, but without corresponding rotation of the blocks. Consequently, although Uric faulting does occur, it is not an essential part of the model. Near the region where the initial break occurred, the





0 1 2 3 4 5 6 7


k;n t

’ 9








Fig. 4. Simplified geometric model of a continental margin, based on the interpretation of the Armorican margin by Le Pi,chon and Sibuet, 1981. The volume remains constant during extension. Plastic deformation may occur only in the lower layers (in white), whereas upper layers (stratified) are faulted. u, h and c are different parts of the margin from the continent toward the ocean, with increasing extensional rates. Faults are ptane, except for their lower part where plastic defo~ation occurs. As the tiit of blocks siigbtty increases from invent to ocean, the existence of steeper normal faults is geometrically indispensable. Together with the lower parts of main faults, these faults may resemble concave listric faults. The deformation is more complex in C. The thinning is increased by the interfingering of the two sets of normal faults with opposite dips. In addition, oceanic crust is created where p becomes greater than 3.3.

extension is not due mainly to tilting of the bIocks but increasingly to i~te~finge~ng of the two main sets of faults (see Marton and Black, 1975, for examples of interfingering of the two sets of faults).

We do not wish to further discuss here the role played by the width of the blocks (from an average of 5-7 km on the Armor&an continental margin to 1 to 2 km in the Yerington district of the Basin and Range) and by the original angle of faulting, which we believe to be close to 45” in the main part of the brittle crust from observations made in the Bay of Biscay and in Basin and Range. This will be done in detail elsewhere. However, the important point is that the observations imply an original thickness, prior to thinning of about 10 km for the brittle layer extended by normal faulting (Le Pichon and Sibuet, 1981) which implies some kind of plastic behaviour below this level.

Figure5 is a cross-section, without vertical exaggeration, of a basin under extension just prior to asthenospheric intrusion, 8s the water depth to the axis reaches 2.5 km. The total width is about 160 km, corresponding to twice the width of the continental margin shown in Fig. 4. This value is a reasonable one for continental margins, although widths of up to 400 km are not exceptional (based on the present width of the northeast American (off USA} thinned portion of continental crust). The m~mum /? is about 3.2 at the axis and consequently the asthenosphere rises to less than 40 km. As pomted out by Le Pi&on and Sibuet (1981) this unusual configuration, in which highly different vertical. mass distributions are juxtaposed, leads to horizontal deviator& stresses which are directed toward the axis of the basin, tending to pull material from the thicker to the thinner hthosphere. Thus, if for some reason, the superimposed extensional stresses which created this basin cease, the stresses due to mass distribution, which are of the order of several hundred bars, tend to suppress the thinned lithospheric basin. This is consequently a geodynamically significant situation, where a hot and thin lithosphere, pervasively cut by new faults, is submitted to relatively large compressional stresses of gravitational origin. Thus, any additional compressional stresses may lead to major shortening of this i~tracon~nenta~ basin. The portion which has been submitted to an extension larger than about 2, may be entirely subducted, as it has a negative buoyancy. The tilted blocks in the upper brittle layer may be remobilized and form the future nappes of the orogen. Consider far example the history of the Western Alpine system. The Triassic phase is one of active distension followed by subsidence in a probably entirely-or almost entirely- intracontinental basin. In the Hellenic area, geological studies show that in the Late Triassic, the Apulian continental margin had just been the location of a generalized extension, which probably explains the creation of the Pindos (and hypothetical Mahac) basins between the Apulian platform to the southwest and the Pelagonian ridge to the northeast (Aubouin et al., 1977). The extensional processes created these basins during the Triassic and were probably still active during the Jurassic, while the Vardar oceanic basin, which formed the western end of the

251 0










150-I km

Fig. 5. Cross-section the surface.

of a thinned

The dimensions




prior to asthenosphere

were chosen on the basis of the Bay of Biscay continental



margin results. No

vertical exaggeration.

Tethys Ocean, was still actively spreading, and then progressively disappeared during the Late Jurassic by a northeastward subduction under the Euro-Rhodopian plate (Aubouin et al., 1977; Le Pichon and Blanchet, 1978). Whether or not this general reconstruction of the structural evolution of the Hellenides during Triassic and Jurassic times is correct, theie is no doubt that the Pindos basin was created by extension, most of which occurred during the Triassic, as suggested by stratigraphic and paleogeographic studies (e.g., Fleury, 1980). A noticeable amount of subsidence was still occurring during the Early Jurassic, as shown by the very large thicknesses of sediments of this age at the margins of the basin. Figure 6a is a schematic section of this basin during Early Cretaceous; the following assumptions have been made. First, we suppose that the geometry of extension and related stretching is as simple as possible, i.e. that of the model of Fig. 4. Actually, although the series of the Pindos basin are quite characteristic, the thicknesses of different strata, especially those of Upper Triassic and Lower Jurassic ages, are highly variable (Dercourt et al., 1973; Fleury, 1980). This variability could be due to the structural pattern of tilted blocks which implies large differences of thicknesses in the overlying sediments (Fig. 4), as shown also in the profiles of the Armorican continental margin (Montadert et al., 1979). However, no detailed study of these variations has yet been made, and the intense folding and thrusting of the Pindos units would make it difficult. Note also that, in the case of the Pindos basin, the morphology of tilted blocks has been partly or totally buried by the very thick Triassic volcanics. Secondly, we assume that no significant amount of oceanic crust was created within the Pindos basin, which is thus formed by two symmetric passive continental margins. Actually, these basins when developed between the Apulian






100 km


Fig. 6. Hypothetical Dimensions a. Section margins oceanic


are approximative. of the Pindos




c. Present


of the Pindos


of the Hellenides.


phase. The pattern layers


of tilted blocks on both

A very limited



figured (cross hatchured).

of the thinned

from extensionaf of the Pindos

sheets. GT=Gavrovo-Tripolitza



plastically lithosphere

times. All parts of the basin where & is greater pattern


and dips not at scale.

basin after the Triassic

with Fig. 4) overlies

crust is hypothetically

b. Probable

of the birth


to compressional. nappe.

of the Pindos


basin during

Late Cretaceou-Eocene

1.8 or 2 could be easily subducted The sense of subduction

Note that it is formed

units. PO=Pindos-Olonos

by numerous


when the stress

is uncertain. folded




nappe (v-pattern=


margin and the Pelagonian ridge had a width of about 150 to 400 km as indicated both by kinematic reconstruction (Biju-Duval et al., 1977; Fig. 7) and geological study (e.g., Fleury, 1980). This width is comparable to that predicted by our model. The maximum depth just after stretching was probably between 2 and 2.5 km and probably increased by thermal subsidence to 3-4 km or more 30 to 40 m.y. later, when the shortening of the basin first began, and to 4-5 km by Late Cretaceous at the time when this basin was being completely destroyed by subduction processes (Fig. 6b), prior to collision. In a discussion of paleobathymetric data, Fleury (1980) suggests that the depths were much smaller, and increased from about 1 km in the Late Triassic to 2.5 km in Late Cretaceous, but points out that the model of Bosellini and Winterer (1975) leads to depths which increase from 1.5 km to 4 km during the same period, as our model also suggests. It is likely that the shallower depths estimated by Fleury are valid for the edges of the basin, whereas our estimates correspond to its central part.


Fig. 7. Extensional zones (hatches) Biju-Duval et al., 1977.

of the western Afpine system during the Lower Jurassic. After

The decrease in width of the Pindos basin during Cretaceous-Eocene times is believed to be due to the subduction of its thinnest parts where the coefficient of extension p was greater than 1.8 or 2 (Fig. 6b), thus facilitating the subduction of this thinned out continental crust, as previously stated. The last stage was the Eocene collision (Fig. 6c) during which the remnants of the Pindos basin completely disappeared, and gave birth to the Pindos nappe with a complicated structure of numerous folded and imb~cat~ thrust sheets. We suggest that these thrust units could well correspond to the original tilted blocks of the extensional phase, since the thrust faults have developed preferentially along former normal faults which constituted zones of mechanical weakness. Although this hypothesis is difficult to prove on the basis of field observations, it is supported by the fact that the original width of numerous sheets of the Pindos nappe is comparable with that of tilted blocks on the Armorican margin (Le Pichon and Sibuet, 1981): this width now averages 5 km (e.g., Fleury, 1980) and was thus 8 to 15 km before folding. It is clear that, in addition, some structural units have been tectonically divided into smaller thrust sheets of smaller width, especially along the western front of the Pindos nappes. The tectonic evolution of the Pindos series was facilitated by the decollement at the level of detritic sediments of Carnian age. Further detailed comp~ative analyses of the


thickness of mesozoic sediments of the Pindos units should show whether this hypothesis is correct or not. Finally, we consider that, although subduction and collision processes in the Alpine system make difficult a precise quantitative reconstruction of the previous extensional pattern (Fig. 6), the Pindos provides an illustration of the development of extensional intracontinental basins with high strain rates, where the thinning of the lithosphere facilitates further shortening whenever the stress system becomes compressional, whether the stage of oceanic accretion is reached or not. CONCLUSIONS

The behaviour of the lithosphere under extensional strain is quite different above and below the asthenosphere geoid. Below this level, continuity of the old lithosphere is likely to be rapidly broken and, as a consequence, oceanic accretion starts rapidly. Above this level, the old lithosphere will be thinned extensively until it reaches the level of the asthenosphere geoid, thus enabling the asthenospheric material to break through to the surface. The thinning however probably depends on the strain rate. For low strain rates, the lower lithosphere may be thinned by diapiric asthenospheric intrusions and lithospheric sinking whereas the upper lithosphere may be barely affected. This situation is considered typical of the African rifts or the Rhine Graben. For high strain rates, the whole lithosphere is apparently thinned rather uniformly, as in Aegea and probably on many continental margins. However, whereas the lower portion is extended plastically, the upper brittle portion, which is about 10 km thick prior to extension, is extended by normal faulting. The process of faulting can be compared to a pack of cards resting at an angle on a plane, with each card making a slight angle with the preceding one. The geometry of extension is such that intracontinental basins 150 to 400 km wide may be created before the water depth increases sufficiently for oceanic accretion to begin. It is believed that many orogens are localized along earlier zones of extension which resulted in similar intracontinent~ basins, specially susceptible to compressional failure whenever the stress system changes from extensional to compressional. This is, in particular, believed to be the case of the western Alpine system. ACKNOWLEDGEMENTS

This work was supported by the CNRS (ATP IPOD n”4228) and the Centre National pour I’Exploitation des O&ans (Contrat CNEXO n”79/5929). We are grateful for the comments made on the manuscript by John Royden.




J.L. and Turcotte,

D.L., 1979. Magma



an Ocean ridge. Earth Planet. Sci. Lett.,

45: 115-122. Angeher,

J., 1979. Neotectonique


J., 1981. Analyse



Ann. Geophys., Arcyana,


de I’Arc Egeen. Sot. Geol. Nord, Publ. No. 3, 418 pp.


des relations

Cgeenne, la subsidence

entre la deformation


et les mouvements

de la mer de Crete et al surrection

de I’arc hellenique.

37(2) in press.

1975. Transform

fault and rift valley from bathyscaph

and diving saucer.



108-I 16. Artemjev,

M.E. and Artyushkov,

of rifting. Aubouin,

J. Geophys.

J., Le Pichon,

tectonique Athens, Ballard,

X.. Winterer,

des plaques.

and isostasy

of the Baikal Rift and the mechanism

E. and Bonneau,

VI Colloquium

M.. 1979. Les Hell&ides

on the Geology

of the Aegean

dans I’optique


de la

Inst. Geol. Min. Res.

Proc. vol. III, pp. 1333-1354.

R.D., Atwater,

IV: AMAR Geophys.

T., Stakes, D., Crane,

rift valley, evidence


Biju-Duval, plate

E.V., 1971. Structure

Res., 76: 1197- 1212.





X., 1977. From the Tethys ocean to the Mediterranean of the western

and Winterer,

Basins, Split (Yugoslavia) E., 1975. Pelagic





Int. Symp.

1976. Editions


and radiolarite

seas: a

on the Structural Paris, pp. 143- 164.

of the Tethyan



3(5): 279-282.

I. Evolution of young continental



EOS, Trans.

J. and Le Pichon,

model. Geology,

Bott, M.H.P.,

78; Preliminary

of rift valleys (abstr.).

of the evolution

of the Mediterranean


C., 1978. AMAR

59: 1198.

B., Dercourt, tectonic

K. and Hopson,

for cycles in the evolution


and formation

of shelf basins. Tectonophysics,

I I: 319-327. Christie,

P.A.F. and Sclater, J.G., 1980. An extensional

the North Cyamex

Sea. Nature,


Rita): General Dercourt,


1981. First manned


Mar. Geophys.

J., Flament,

radiolarites W.M.,

J.M., Fleury,

1967. Convection

Rep., 5: 23 pp (reprinted and Planetary Fleury,







Rise, 21 “N (Project

Le Trias superieur

des couches


et le Jurassique

sous les





(abstract). and stress propagation

in: S.K. Runcorn


Wiley, New York, N.Y.,


dives on the East Pacific

J.J. and Meilliez, F., 1973. Stratigraphie

J.J., 1980. Les zones de Gavrovo-Tripoli&a

du Nord).

and Witchground

Res., 4: 345-379.

de la zone du Pinde-Olonos

Geol. Pays Hell., 25: 397-406 Elsasser,

origin for the Buchan

283: 729-732.


in the upper The Application



of Modem


to the Earth

1969, pp. 223-246.

et du Pinde-Olonos

et d’un bassin


dans leur cadre


et Peloponnese

alpin. Sot. Geol. Nord,


No. 4, 651 pp. Heezen,

B.C., 1962. The deep-sea

floor. In: SK.







New York, N.Y., pp. 235-288. Le Pichon,

X. and Angelier,

J., 1981. The Aegean

X. and Blanchet,

R., 1978. Where

Sea. Philos.


R. Sot. London,

Ser. A. 300:

357-372. Le Pichon, Geology, Le Pichon,

are the passive


of the western



6: 597-600. X. and Sibuet, J.C., 1981. Passive margins:

a model of formation.

J. Geophys.

Res., 86 (B5):

3708-3720. Le Pichon,

X., Francheteau,

Elsevier, Amsterdam, Le Pichon,

X., Huchon,

J. and Bonnin,

J., 1973. Plate Tectonics,


in Geotectonics


J.P., Dercourt,


300 pp. Ph., Angelier,

J., Lyberis,

N., Boulin,

J., Bureau,

D., Cadet,



G., Got, H., Karig. D., Mascle, J., Ricou, L.E. and Thiebault,



Continentales, Le Pichon,


Ph., Angelier.


G., Got, H., Karig,




J., Lyberis.

N.. Boulin.

D., Mascle, J., Ricou,


role of thick evaporitic


1980, London.

D., 1978a. Some remarks

J., Bureau,


D.. Cadet.

and Thiehault,

in the Margcs

and Tectonics

J.P.. Dcrcourt,

F., I980b. Subduction

layer based on Seabeam

Proc. Conf. on Trench and Fore Arc Sedimentation Zones, 23-25


Paris 1980, pp. 273-282..

X., Huchon,


F., IY81. Active tectonics

Acta, Spec. Issue. Proc. 26th Int. Geol. Congr.,

and submersible

in Modern

and Ancient


in the studies.


in press. on the deveioprnent

of sedimentary

basins. Earth


Sci. Lett..

40: 25-32. McKenzie,

D., 1978b. Active tectonics




J. R. Astron.

L., Roberts,

Northern Initial




of the Alpine-Himalayan

De Charpal.


belt: the Aegean

Sea and surrounding

Sot., 55: 217-254. 0. and Guennoc.

of the Bay of Biscay.

of the Deep Sea Drilhng


P.. 1979. Rifting

In: L. Montadert

U.S. GOY. Printing

and subsidence

and D.C. Roberts Office,

of the



D.C.. 48:

1025-1060. Morton,

W.H. and Black, R., 1975. Crustal

Afar Depression Nakamura.

of Ethiopia.

K. and Uyeda,


B. and Richter,

B. and Sclater, J.G.. 1977. Analysis

age. J. Geophys.

Royden, Royden,

F., 1979. Mantle convection


and the oceanic


of the variation


I ., pp. 55-65.


and plate



The Sea, in press.

of ocean floor bathymetry

of the Yerington

and origin of Basin and Range


L. and Keen, C.E., 1980. Rifting




L., Sclater,



in arc-back-arc

and heat fiow with

Res., 82 (5): 803-827.

J.M., Jr., 1977. Cenozoic


In: A. Pilger and A. Rosier (Editor.\),

Res., 85: 6419-6428.



in Afar.


S., 1980. Stress gradient

Parsons, Proffett,



from subsidence

J.G. and Von Henen,


in formation




and implications

and thermal


of two continental

curves. Earth Planet. Sci. Len,


1980. Continental

of petroleum

for the

Sot. Am. Bull.. 88: 247-266.





51: 343-361.


Am. Assoc.

and heat flow:

173-187. Tapponnier,

R. and

accreting Tarantola.


plate boundaries.

of the lithosphere

D.L.. Haxby,

W.C. Pitman, Geophys.



of the lithosphere.

W.F. and Ockendow,

Earth Planet. J.R.,

the mechanics

of slowly

D.C., pp. 63-69.


for rifting

in Afar:

a brittle-elastic

Sci. Lett., 45: 435-444

1977. Lithospheric

arcs and Deep Sea Trenches



Res., X3: 3955-3970.

A., Ruegg, J.C. and Lepine, J.C., 1979. Geodetic

model of the beha~our Turcotte,

J., 1978. Necking

.I. Geophys.

and Back-Arc

instabilities. basins.

In: M. Talwani

M. Ewing



1, Am.