Italy subduct.pdf

Subduction rollback, slab breakoff, and induced strain in the

uppermost mantle beneath Italy

Francesco Pio Lucente

Lucia Margheriti

Istituto Nazionale di Geoﬁ sica e Vulcanologia, Centro Nazionale Terremoti (CNT), Rome 00143, Italy

ABSTRACT

Differences in seismic anisotropy revealed by split SKS waves that traverse the upper mantle

beneath the Italian region reveal four areas of internally coherent anisotropic strength, provid-

ing evidence for mantle strain partitioning. When compared to uppermost mantle structure

imaged by tomography, the sequence of these areas displays a straightforward parallelism.

Under proper assumptions, the correspondence between areas of coherent splitting degree

and areas of coherent velocity perturbations offers new insights into the late evolution of sub-

duction in Italy and their effect on mantle strain.

the incoming rays (see Margheriti et al., 2003),

and their time delays (δt) were interpolated

using a nearest neighbor algorithm (Fig. 1). The

pattern obtained by interpolating the splitting

time delays reveals a sharp partitioning of δt

along the Italian region: δt values are relatively

high in the northern Apennines and in the south-

ern Tyrrhenian area, and are markedly lower

in the south-central Apennines. High δt values

are also observable in Sardinia. A straight cor-

respondence exists between areas of higher δt

and areas where subducting lithosphere in the

uppermost mantle beneath Italy was imaged

by P- wave tomography (Lucente et al., 1999;

Piromallo and Morelli, 2003) (Fig. 2). These

two sets of observations are almost uncorre-

lated, because teleseismic P waves typically

arrive with steep incidence, and anisotropy with

a horizontal symmetry, seen by SKS, would not

affect traveltimes greatly.

We divided our study area into subregions

based on the presence or absence of subduct-

ing lithosphere inferred from the mean Vp

perturbation values in the uppermost 250 km

of the mantle calculated from the model of

Lucente et al. (1999) (Fig. 2). Three of the

subregions are along the Apennines (northern

Apennines, south-central Apennines, and the

southern Tyrrhenian area), which is the accre-

tionary wedge exposed during the last episodes

of the subduction process in Italy (Lucente

et al., 2006, and references therein). The fourth

subregion, the Sardinia area, was the locus of

this subduction between ca. 15 and 10 Ma,

during its retreat toward its current position

(e.g., Faccenna et al., 2001). This partitioning

was plotted on the splitting time delay map of

Figure 1, and mean quantitative attributes of

the splitting indicators in each of the subareas

were computed (Table 1 and Fig. 1).

The anisotropy fast axes, ϕ, are uniformly

oriented with an E-W trend only in the Sardinia

(box SI in Fig. 1) area, while along the Apen-

nines (boxes NA, SCA, and ST in Fig. 1), the

directions of ϕ cover almost the whole azimuth

range. The azimuthal spread of ϕ in the Apen-

nines likely reﬂ ects the complexity of dynam-

ics (and strain) in the underlying mantle as a

response to the hinged kinematics of subduction

in its most recent stages (Lucente at al., 2006).

However, despite the “uniform variability” of

ϕ through the Apennines, the measured strength

of anisotropy, as indicated by the δt proxy, dif-

fers considerably between subareas (Fig. 1).

Keywords: anisotropy, tomography, subduction, upper mantle, Italy.

INTRODUCTION

A powerful tool to quantify the strain in the

deep Earth is the measurement of anisotropy

from seismic wave propagation. In the upper

mantle, seismic anisotropy primarily results from

elastic anisotropy of olivine, which is its main and

most anisotropic mineral (Ben Ismail and Main-

price, 1998). Solid-state deformation of poly-

crystalline aggregates causes crystallographic

a -axes [100] of olivine to align in the direction of

maximum extension (Zhang and Karato, 1995).

This property affects the propagation of seismic

waves, causing differential velocity, scattering,

and birefringence. Thus, quantifying these phe-

nomena serves to detail upper mantle anisotropy

structure. Measurement of anisotropy in the upper

mantle is commonly done by using the split-

ting of core-refracted shear waves SKS (Savage,

1999), which are nearly vertically incident and,

in the absence of anisotropy, would be radially

polarized (i.e., vibrating in the plane containing

the seismic ray). Anisotropy causes these waves

to split into two orthogonally polarized waves,

one traveling faster than the other. This splitting

effect produces two measurable quantities (Silver

and Chan, 1991): the azimuth, ϕ, of the fast wave

polarization plane, which is aligned with the

horizontal component of the average a -axis ori-

entation, and the time delay, δt, between fast and

slow polarized waves, which basically depends

on the quantity of anisotropy encountered and on

its orientation with respect to the ray path. Both

these observables have a direct relationship with

the crystallographic fabric of mantle rocks, and

hence with the mantle deformation; nevertheless,

retrieving mantle strain from splitting indicators

can be problematic: a trade-off exists between the

thickness of the anisotropic volume traversed and

the strength of intrinsic anisotropy, the location

of the anisotropy along the ray path is undeﬁ ned,

and olivine fabrics with average a -axes oriented

along the Earth radius are almost invisible to

SKS splitting (i.e., vertical strain = no measurable

strain). Further complexities may arise because

of alternative causes of seismic anisotropy, such

as the inﬂ uence of water on olivine deformation

(Jung and Karato, 2001), the effects of dynamic

recrystallization (Kaminski and Ribe, 2001), or

the presence of a signiﬁ cant melt fraction (Holtz-

man et al., 2003). All these complications may

occur in subduction environments, where the

radial component of deformation is signiﬁ cant,

and where chemical, physical, and kinematic

complexities are reﬂ ected in an equally complex

variability in SKS splitting observations (i.e.,

Park and Levin, 2002).

Thus the equation, splitting parameters =

anisotropy = mantle strain, always requires a set

of a priori assumptions, and interpretations of

splitting observations in terms of mantle strain

mainly rely on ϕ, arguably the less ambigu-

ous of the two anisotropy indicators. This has

been true for previous studies of anisotropy in

the subduction zone extending along the Italian

peninsula (Margheriti et al., 2003; Civello and

Margheriti, 2004; Lucente at al., 2006). Here,

we reconsider the well-established set of SKS

splitting data in the Italian region, focusing on

splitting delay times, δt. This approach reveals

details of the distribution and partitioning of

anisotropy intensity in the area, and, given a set

of plausible assumptions, provides new clues

for interpreting the strain conﬁ guration in the

upper mantle beneath Italy.

LOOKING AT SPLITTING DATA FROM

A δ t-CENTRIC POINT OF VIEW

The splitting data set we consider derives

from previous studies (Margheriti et al., 2003;

Civello and Margheriti, 2004; Lucente at al.,

2006), and includes more than 300 good-quality,

non-null, single-event measurements resulting

from analysis of SKS phases at 54 seismic sta-

tions located in Italy and Corsica.

Individual splitting measurements were pro-

jected to the 150-km-depth piercing point along

Geology , May 2008; v. 36; no. 5; p. 375–378; doi: 10.1130/G24529A.1; 3 ﬁ gures; 1 table.

375

GEOLOGY, May 2008

To assess the statistical signiﬁ cance of the dif-

ferences between the populations of δt, we

derive the respective 95% conﬁ dence inter-

vals (Table 1). At this conﬁ dence level, the δt

population in the south-central Apennines box

has values signiﬁ cantly lower than in the other

boxes, because its 95% conﬁ dence interval

does not overlap the others (Table 1). As pre-

viously observed, the pattern of δt displays a

straightforward relationship with the conﬁ gura-

tion of subduction in the upper 250 km of the

mantle beneath the Apennines (Fig. 2).

Next we examine the cause of the observed

anisotropy strength partitioning in terms of

strain induced in the mantle by the kinematics

of subduction.

the SKS splitting necessarily require an inter-

pretative framework. This is made by formulat-

ing a series of plausible a priori assumptions:

(1) Kinematics of the subduction zone beneath

the Italian region evolved during slab rollback.

The progressive eastward migration of the trench

(Fig. 2) causes the opening of the Tyrrhenian

basin starting at ca. 10 Ma (e.g., Malinverno

and Ryan, 1986). According to paleogeographic

reconstructions (Faccenna et al., 2001), the rate

of retreat increased from the northern to the

southern boundary of the subduction system

(upper left inset in Fig. 2), as a consequence of

changes in the type of lithosphere arriving at the

trench. This has obvious effects on the degree

of extension, which is higher in the southern

Tyrrhenian area (ST in Figs. 1 and 2).

(2) The anisotropy we observe (Fig. 1) is

caused by lattice-preferred orientation of olivine

crystals (LPO), induced by present and past

ﬂ ow in the asthenosphere. The dominant source

of anisotropy is located in the uppermost mantle

(the anisotropic mantle as deﬁ ned by Gaherty

and Jordan [1995]), and contributions from the

Apennine lithosphere and the Adriatic-Ionian

subducting plate are subordinate.

(3) The anisotropic fabric acquired by the

upper mantle at a given time is preserved as long

as no tectonic and/or thermal episode occurred

subsequently (Vauchez et al., 1999).

(4) The nature, composition, and thickness of

the anisotropic mantle do not vary signiﬁ cantly

along the subduction system; therefore, coef-

ﬁ cients in the equation splitting parameters =

anisotropy = mantle strain, remain constant

through the study area.

(5) Finally, we do not enter into the olivine

fabric types argument (e.g., Kneller at al.,

2007), because we do not have grounds to

discuss this topic.

FORMULATING A SET OF PLAUSIBLE

A PRIORI ASSUMPTIONS

Due to the intrinsic ambiguity of SKS split-

ting quantities as strain indicators, any infer-

ences concerning mantle deformation based on

POSSIBLE CAUSES OF δ t

PARTITIONING: QUALITATIVE

INTERPRETATIONS IN TERMS

OF MANTLE STRAIN

Hypothetical causes of the observed aniso-

tropic strength partitioning imply assigning a

geodynamic signiﬁ cance to the combination of

splitting delay times (Fig. 1) and Vp anomalies

(Fig. 2) in each of the study subregions, and

reconciling them to a picture consistent with

the supposed geodynamic evolution of this part

of the Mediterranean. We consider the splitting

observed in the Sardinia area (SI in Fig. 1) as

representative of the horizontal strain induced in

the backarc mantle by subduction in its “steady-

state” phase. We assume the subduction to be

in a “steady-state” before the slab fragmentation

episodes documented by tomography (Fig. 2)

occurred as a response to the heterogeneous

nature (continental or oceanic) of the lithosphere

arriving at the trench along the Apenninic mar-

gin (Faccenna et al., 2004). To establish this

link, we rely on the azimuthal property of the

splitting. In particular, the close match between

the uniformly E-W–oriented ϕ in Sardinia (SI)

and the prevailing direction of extension in

the Tyrrhenian basin indicate the importance

of the subduction rollback-type kinematics for

the state of strain in the mantle. Beginning in

Tortonian time (ca. 10 Ma), the Tyrrhenian

basin opened with a northward decreasing rate

of extension (Faccenna et al., 2001) as a con-

sequence of the counterclockwise retreat of

the trench, whose strike changed from ~N-S to

NW-SE at present (upper left inset in Fig. 2).

Given our assumptions, these kinematics should

have affected the state of deformation of the sur-

rounding mantle proportionally, i.e., generating

a progressively lower amount of the horizontal

strain (and hence δt) going from the southern

boundary (southern Tyrrhenian) of the subduc-

tion system to the northern boundary (north-

ern Apennines). Splitting δt values are indeed

12°

14°

16°

3.2

3.0

2.8

Tyrrhenian

sea

2.6

2.4

2.2

2.0

1.8

1.6

1.4

1.2

SCA

1.0

0.8

0.6

0.4

0.2

0.0

Figure 1. Map of the SKS splitting time delays ( δ t) in the Italy region. Individual splitting time

delays (white dots) were interpolated using a nearest neighbor algorithm which averages,

over a 10 ′ (latitude) × 10 ′ (longitude) grid, measurements within a 1° radius, if there is at

least one value inside each of the three 120° azimuthal sectors. Boxes NA (northern Apen-

nines), SCA (south-central Apennines), ST (southern Tyrrhenian), and SI (Sardinia area) are

imported from Figure 2. Rose diagrams show SKS fast directions in each of the four boxes:

the length of each petal is proportional to the average δ t in the corresponding direction.

Measurements included in rose diagrams and in Table 1 are circled.

376

GEOLOGY, May 2008

highest in the southern Tyrrhenian area, where

the mean value equals the mean δt in Sardinia,

and decrease northward (northern Apennines

and south-central Apennines; see Table 1). How-

ever, splitting delays in the south-central Apen-

nines are lower than in the northern Apennines.

This breaks the symmetry between the rollback

amount (and corresponding extension in the

Tyrrhenian basin; upper left panel in Fig. 2) and

horizontal mantle strain, since we would expect

a mean δt in the range between 1.84 s (southern

Tyrrhenian) and 1.60 s (northern Apennines)

for the south-central Apennines, where it is

much lower (1.35 s; see Table 1).

This apparent inconsistency is counter-

balanced by the close correspondence between

areas with relatively high δt values and areas

where subduction is continuous up to the sur-

face (cf. Figs. 1 and 2). Together, these incon-

sistencies and correspondences furnish a key

to explain our observations. Based on the last

evolutionary phases of the subduction sys-

tem as inferred from the tomographic images

(Fig. 2), and organizing our observations in a

logical sequence, we propose, as preferred, the

following scenario:

(1) The Tyrrhenian opens as a backarc basin

of the subduction zone, which at present is

located below the Italian peninsula (Fig. 2).

(2) The eastward retreat of the subduction

system, in a regime of slab rollback, causes the

horizontal straining of the uppermost mantle

in a corresponding direction, which is repre-

sented by the splitting parameters in Sardinia

(SI in Fig. 1).

(3) The migration of the subduction system

follows a counterclockwise trajectory (upper

left inset in Fig. 2), implying a differential hori-

zontal straining of the uppermost mantle, which

is reﬂ ected in the relatively higher δt values in

the southern portion of the subduction system

(ST in Fig. 1) with respect to the northernmost

ones (NA and SCA in Fig. 1).

(4) The arrival of continental material at the

trench in the south-central Apennines leads to the

segmentation of the subducting plate (Lucente

and Speranza, 2001), and while the northern

Apennine and southern Tyrrhenian narrow seg-

ments continue their backward migration at dif-

ferent speeds, in the south-central Apennines,

slab breaks off and starts to sink passively into

the mantle (Fig. 3).

(5) The further retreat of the northern Apen-

nine and southern Tyrrhenian slab fragments

maintains the deformation in their neighboring

mantle prevalently horizontally, while the gravi-

tational descent of the south-central Apennines

250–550 km depth range

10°

1 2°

1 4°

1 6°

SCA

–1

–2

35–250 km depth range

–3

–4

Figure 2. Tomographic map of the mean Vp perturbation values in the uppermost 250 km

of the mantle beneath Italy calculated from the model of Lucente et al. (1999). Boxes divide

the mapped area in subregions based on the presence or absence of subducting litho-

sphere. In the upper right inset, the mean Vp perturbation values, in the 250- to 550-km-depth

range, reveal a continuous subduction zone running along the Apennines. In the upper left

inset, the notched black line represents the present-day location of the trench, while the

dashed gray lines approximately indicate its past positions at two stages of the retreating

motion (from Faccenna et al., 2001). Red arrows represent the direction and the amount of

extension—or trench retreat—in the Tyrrhenian basin, increasing from the northern to the

southern bounds of the subduction system. NA—northern Apennines; SCA—south-central

Apennines; ST—southern Tyrrhenian; SI—Sardinia area.

SCA

S T

SCA

STSZ

TABLE 1. STATISTICAL ATTRIBUTES OF THE SPLITTING INDICATOR δ t IN THE STUDY AREA

0.0

1.0

2.0

3.0

t (s)

Number of

Mean time

Standard

95% conﬁ dence

Data set

measurements (n)

delay ( δ t) (s)

deviation ( α ) (s)

interval

Figure 3. Simpliﬁ ed sketch illustrating the

correlation between the subduction conﬁ gu-

ration beneath Italy, as deduced by tomog-

raphy, and the partitioned distribution of the

SKS splitting time delays (color scale as in

Fig. 1). Arrows represent, in our interpre-

tation, the directions wherein subduction

kinematics strain the upper mantle (red—

horizontal; blue—vertical).

Whole data set

309

1.59

0.55

1.53–1.65

1.60

0.54

1.44–1.76

SCA

137

1.35

0.51

1.27–1.43

1.84

0.49

1.72–1.96

1.84

0.50

1.72–1.96

( )

is the standard

error of the mean (SE), and 1.96 is the 0.975 quantile of the normal distribution. NA—northern Apennines;

SCA—south-central Apennines; ST—southern Tyrrhenian; SI—Sardinia area.

Note : The 95% conﬁ dence interval is deﬁ ned as: mean ( δ t) ±

σ n ∗

. , where

σ n

GEOLOGY, May 2008

377

196

slab fragment strains the mantle in the vertical

direction (Fig. 3), thus inducing a reorganization

of the olivine fabric accordingly.

(6) Given the assumption that the thickness of

the anisotropic layer does not vary, the process

of reorientation of the olivine fast axes in the

vertical direction erases part of the previously

horizontal fabrics (which are those seen by

SKS waves), producing the relatively lower δt

observed in the south-central Apennines sub-

region (Fig. 3). Assuming a mean δt value of

~1.7 s for the south-central Apennines (inter-

mediate between mean δt values in the northern

Apennines and the southern Tyrrhenian area;

cf. Table 1) before the slab fragment sinking

process starts to inﬂ uence the mantle deforma-

tion, and since the relationship between S-wave

anisotropy and olivine fabric strength is almost

linear in the range of anisotropy percentages gen-

erally assumed for the upper mantle (up to 7%;

Ben Ismail and Mainprice, 1998; Savage, 1999),

we can derive a rough estimate of the percentage

of fast olivine axes reoriented in a vertical direc-

tion that is approximately the same amount as

the δt deﬁ cit (~20%). It is noteworthy that, in this

hypothesis, the vertical orientation of fast axes

would affect the result of the isotropic inversion

(Fig. 2) by speeding up the teleseismic rays at

the south-central Apennines with respect to the

neighboring subareas. As an outcome, the differ-

ence in the isotropic elastic properties between

the subducting lithosphere (in the northern Apen-

nines and the southern Tyrrhenian area) and the

ambient mantle (in the south-central Apennines)

in Italy is possibly larger than that inferred by the

P- wave velocity tomography.

Within the framework settled by our a priori

assumptions, alternative hypotheses are: 1) The

retreat of the slab fragment in the south-central

Apennines stopped farther west of the area

sampled by the SKS observations reported in

Figure 1; therefore, differences in splitting delays

are simply related to the longer retreat experi-

enced both in space and in time, by the northern

Apennines and southern Tyrrhenian areas. 2) The

slab fragments imaged in the northern Apennines

and the southern Tyrrhenian were once joined,

forming a continuous subducting body. At some

point, this body underwent a lateral tear process,

which led to the current conﬁ guration through

the progressive rollback of two separated slab

fragments. Also, in this case, the lower δt values

in the south-central Apennines result from the

absence, in this area, of the process responsible

for the horizontal deformation in the mantle, i.e.,

the retreat of the slab. Lateral tears in the slab

would also account for the lengthening of the

Apennine trench during its eastward migration

(upper left inset in Fig. 2). Although we cannot

exclude that alternative mechanisms concurred

with the anisotropy signature we observe, these

hypotheses are difﬁ cult to reconcile with both

surface and deep expressions of the subduction

beneath the Apennines: the presence of an almost

continuous subduction-related magmatic arc

along the Tyrrhenian margin of Italy (Serri et al.,

1993) implies that slab reached and possibly

passed beyond this margin also in the south-central

Apennines; the slab at depth does not show off-

sets along its strike, running almost parallel to the

Tyrrhenian coast for its entire length (upper right

inset in Fig. 2); also, the total length of the slab at

depth is greater than the combined length of the

northern Apennine and southern Tyrrhenian slab

fragments (upper right inset in Fig. 2).

Therefore, we believe that our preferred sce-

nario is more consistent than the other hypoth-

eses discussed. It can provide a reliable, if only

qualitative, explanation for the mantle strain

partitioning implied by the SKS splitting time

delays in the Italian region, and can be used

as a basis for more sophisticated, quantitative

analyses, as, for instance, on the time scales

for the evolution of anisotropic fabric in the

olivine aggregates or on the viscous interactions

between slab and surrounding mantle.

Kaminski, E., and Ribe, N.M., 2001, A kinematic

model for recrystallization and texture devel-

opment in olivine polycrystals: Earth and Plan-

etary Science Letters, v. 189, p. 253–267, doi:

10.1016/S0012-821X(01)00356-9.

Kneller, E.A., van Keken, P.E., Katayama, I., and

Karato, S., 2007, Stress, strain, and B-type

olivine fabric in the fore-arc mantle: Sensitiv-

ity tests using high-resolution steady-state sub-

duction zone models: Journal of Geophysical

Research, v. 112, doi: 10.1029/2006JB004544.

Lucente, F.P., and Speranza, F., 2001, Belt bend-

ing driven by lateral bending of subducting

lithospheric slab: Geophysical evidences

from the Northern Apennines (Italy): Tec-

tonophysics, v. 337, p. 53–64, doi: 10.1016/

S0040-1951(00)00286-9.

Lucente, F.P., Chiarabba, C., Cimini, G.B., and

Giardini, D., 1999, Tomographic constraints on

the geodynamic evolution of the Italian region:

Journal of Geophysical Research, v. 104,

p. 20,307–20,327, doi: 10.1029/1999JB900147.

Lucente, F.P., Margheriti, L., Piromallo, C., and

Barruol, G., 2006, Seismic anisotropy reveals

the long route of the slab through the western-

central Mediterranean mantle: Earth and Plan-

etary Science Letters, v. 241, p. 517–529, doi:

10.1016/j.epsl.2005.10.041.

Malinverno, A., and Ryan, W.B.F., 1986, Extension

in the Tyrrhenian Sea and shortening in the

Apennines as a result of arc migration driven

by sinking of the lithosphere: Tectonics, v. 5,

p. 227–245.

Margheriti, L., Lucente, F.P., and Pondrelli, S., 2003,

SKS splitting measurements in the Appenninic-

Tyrrhenian domains (Italy) and their relation

with lithospheric subduction and mantle con-

vection: Journal of Geophysical Research,

v. 108, doi: 10.1029/2002JB001793.

Park, J., and Levin, V., 2002, Seismic anisotropy: Trac-

ing plate dynamics in the mantle: Science, v. 296,

p. 485–489, doi: 10.1126/science.1067319.

Piromallo, C., and Morelli, A., 2003, P wave

tomography of the mantle under the Alpine-

Mediterranean area: Journal of Geophysical

Research, v. 108, doi: 10.1029/2002JB001757.

Savage, M.K., 1999, Seismic anisotropy and mantle

deformation: What have we learned from shear

wave splitting?: Reviews of Geophysics, v. 37,

p. 65–106, doi: 10.1029/98RG02075.

Serri, G., Innocenti, F., and Manetti, P., 1993, Geo-

chemical and petrological evidence of the

subduction of deliminated Adriatic conti-

nental lithosphere in the genesis of the Neo-

gene-Quaternary magmatism of central Italy:

Tectonophysics, v. 223, p. 117–147, doi:

10.1016/0040-1951(93)90161-C.

Silver, P.G., and Chan, W.W., 1991, Shear-wave

splitting and subcontinental mantle deforma-

tion: Journal of Geophysical Research, v. 96,

p. 16,429–16,454.

Vauchez, A., Barruol, G., and Nicolas, A., 1999, Com-

ment on “SKS splitting beneath rift zones”:

Journal of Geophysical Research, v. 104,

p. 10,787–10,789, doi: 10.1029/1999JB900053.

Zhang, S., and Karato, S., 1995, Lattice preferred

orientation of olivine aggregates deformed in

simple shear: Nature, v. 375, p. 774–777, doi:

10.1038/375774a0.

ACKNOWLEDGMENTS

We are greatly indebted to Ray Russo and

Martha Savage for extremely constructive reviews and

comments. We also thank Silvia Pondrelli, Antonio

Piersanti, and Alberto Basili for useful discussions.

We are grateful to Prof. Maurizio Parotto.

REFERENCES CITED

Ben Ismail, W., and Mainprice, D., 1998, An olivine

fabric database: An overview of upper mantle

fabrics and seismic anisotropy: Tectono-

physics, v. 296, p. 145–158, doi: 10.1016/

S0040-1951(98)00141-3.

Civello, S., and Margheriti, L., 2004, Toroidal man-

tle ﬂ ow around the Calabrian slab (Italy) from

SKS splitting: Geophysical Research Letters,

v. 31, doi: 10.1029/2004GL019607.

Faccenna, C., Becker, T.W., Lucente, F.P., Jolivet,

L., and Rossetti, F., 2001, History of subduc-

tion and back-arc extension in the central

Mediterranean: Geophysical Journal Inter-

national, v. 145, p. 809–820, doi: 10.1046/

j.0956-540x.2001.01435.x.

Faccenna, C., Piromallo, C., Crespo-Blanc, A.,

Jolivet, L., and Rossetti, F., 2004, Lateral

slab deformation and the origin of the west-

ern Mediterranean arcs: Tectonics, v. 23, doi:

10.1029/2002TC001488.

Gaherty, J.B., and Jordan, T.H., 1995, Lehmann

discontinuity as the base of an aniso-

tropic layer beneath continents: Science,

v. 268, p. 1468–1471, doi: 10.1126/science.

268.5216.1468.

Holtzman, B.K., Kohlstedt, D.L., Zimmerman,

M.E., Heidelbach, F., Hiraga, T., and Hustoft,

J., 2003, Melt segregation and strain partition-

ing: Implications for seismic anisotropy and

mantle ﬂ ow: Science, v. 301, p. 1227–1230,

doi: 10.1126/science.1087132.

Jung, H., and Karato, S.I., 2001, Water-induced

fabric transitions in olivine: Science, v. 293,

p. 1460–1462, doi: 10.1126/science.1062235.

Manuscript received 26 October 2007

Revised manuscript received 11 January 2008

Manuscript accepted 19 January 2008

Printed in USA

378

GEOLOGY, May 2008

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: