Subduction of oceanic lithosphere

Subduction of oceanic lithosphere 

• Evolution of the continental crust 

• New crust is formed in marginal basins, island arcs, and oceanic plateaux 

• New crust is accreted to the continents along the active continental margins. 

Crust is also added by magmatic underplating 

• Continental crust is deformed in collision zones. It is eroded and recycled in 

the mantle in subduction zones. Some crust can be subducted in collision 

zones. 

• What is the budget? 

Wednesday, April 14, 2010

• The age of continental crust 

appears younger at the 

margins than at the center 

of the continent. 

• Suggestion by Wilson that 

continents grew from a 

central core is now 

superseded. 

• However, new continental 

crust has been and is 

presently being added at 

the active margins, like the 

North American Cordillera. 


Subduction 

• Ocean-Ocean subduction (island 

arcs, marginal basins) Western 

pacific 

• Ocean-Continent subduction. 

Eastern Pacific. The Cordilleras of 

N and S America. 

• Continental collision. The Alps. 

Himalayas. 

• Formation of future continental 

crust 

• Accretion of new crust by docking 

of terranes and magmatic 

underplating 

• Recycling and Destruction of 

continental crust. 


Location of convergent plate margins on Earth. Active back arc 

basins are also shown. 


Wednesday, April 14, 2010 

Subduction of the oceanic crust

• The western Pacific contains 

most of Earth marginal 

basins and oceanic island 

arcs. 

• This is where new 

continental crust is being 

formed 

• It will eventually be docked 

on the active margins of the 

continents 



W Pacific. Topography Bathymetry


W Pacific Free Air Gravity (from satellite data)

• Tectonic setting of Southeast Asia. Thick lines 

with closed triangles are modern active trenches 

and those with open triangles are inactive. Thick 

arrows show the direction of plate movement in a 

fixed hotspot reference frame after Engebretson et 

al. (1985) and Royer and Sandwell (1989). Length 

of bar is motion for 10 Ma. Spreading centers are 

shown in double lines. Thin lines are active 

structural boundaries and thin broken lines are 

traces of offshore or buried structural boundaries. 

Abbreviation: T: Trench or Trough, JB: Japan 

Basin, NA: Nankai Trough, OT: Okinawa Trough, 

DRB: Daito ridges and basins, SB: Shikoku Basin, 

OD: Ogasawara Depression, MA: Mariana 

Trough, PB: Parece Vela Basin, WPB: West 

Philippine Basin, SCB: South China Basin, AB: 

Andaman Basin, PA: Palawan Trough, NE: 

Negros Trench, SU: Sulu Basin, CO: Cotobato 

Trench, CB: Celebes Basin, NS: North Sulawesi 

Trench, MB: Makassar Basin, BB: Banda Basin, 

MO: Molucca Collision, SO: Sorong Fault, CE: 

Ceram Trough, CAB: Caroline Basin, NG: New 

Guinea Trench, WE: West Melanesia Trench, NB: 

New Britain Trench and TR: Trobriand Trench, 

RF: Red River Fault, SF: Seribu Fault. 



Australia: Topography Bathymetry


Australia: Free Air Gravity.


Scotia Arc: Topography Bathymetry


Scotia Arc: Free Air Gravity


South America Topography and bathymetry


Free Air gravity in the oceans around South America


NE Pacific Topography Bathymetry


NE Pacific Free Air Gravity


Free Air gravity from GRACE (29/10/2004). 

Paired gravity anomalies


Sea floor age 

Note subduction occurs at all ages.

Distribution of ages of the sea floor 

The area of Sea floor of given 

age decreases with age. This 

indicates that sea floor is 

destroyed at constant rate 

regardless of age. 


Oceanic subduction zones 

• (top) Schematic section through the upper 

140 km of a subduction zone, showing the 

principal crustal and upper mantle 

components and their interactions. Note that 

the location of the “mantle 

wedge” (unlabeled) is that part of the 

mantle beneath the overriding plate and 

between the trench and the most distal part 

of the arc where subduction-related igneous 

or fluid activity is found. MF stands for 

magmatic front. 

• (bottom) Schematic section through the 

center of the Earth, which shows better the 

scale of subduction zones. Subducted 

lithosphere is shown both penetrating the 

660 km discontinuity (right) and stagnating 

above the discontinuity (left). A mantle 

plume is shown ascending from the site of 

an ancient subducted slab. Dashed box 

shows the approximate dimensions of the 

shallow subduction zone of Figure 1b. 


• Structure of the western Pacific subduction zones with a back-arc basin. 


Semilog scale plotting the depth dimension for subduction zones from 1 km below the surface to the core-mantle 

boundary. Mean depth of ocean (3880 m) is from Kennett [1982]. Typical crustal thickness (∼20 km) for juvenile 

arc crust is from Suyehiro et al. [1996]; crust associated with Andean-type convergent margins may be up to 70 

km thick. Mean distance of arc volcanoes to trench (166 ± 60 km) and of depth beneath arc volcano to seismic 

zone (124 ± 38 km) are from Gill [1981]. Greatest depth of subducted material returned to surface (∼100 km). 


Seismic structure, focal mechanisms 

• Wadati Benioff zone 

• Local seismic tomography 

• Stresses in the subducted plate 

• Stresses in the overriding plate 

• Seismic tomography and the fate of the subducted slab. 


• Plate flexure causes a bulge (outer rise) up from the trench. 


• The dipping seismogenic zone (Wadati-Benioff zone) delineates the 

subducting slab. 

• Earthquakes show that the slab is cold and brittle. 


• Quality factor Q measures 

how well energy is 

transmitted. 

• Attenuation of seismic waves 

(loss of energy) is due to 

“friction”, i.e. viscous mantle. 

• Comparing seismic signals 

that travel through the slab or 

below the volcanic arc show 

that waves that travel in the 

asthenosphere are more 

attenuated than waves that 

travel through the slab. 

• The attenuation affects mostly 

the high frequencies that are 

attenuated 


• The Q factor (quality) measures the loss of energy of the seismic waves. Quality is 

high if there is no loss of energy. The attenuation of seismic waves usually is 

caused by partial melting. 


Velocity anomalies related to temperature 

• The subducting plate is seismic because it remains cold and brittle. 

The higher velocities are also due to the plate being cold. 



• Tomographic images show that the 650km 

discontinuity is a temporary barrier for some plates 

that can not penetrate into the lower mantle, but most 

eventually do.

• The dip of the slab is variable. 

• Stresses seem mostly compressive in the shallow dipping slabs otherwise 

tensile. 


• Geophysical images of subduction 

zones. (a) P wave tomographic image 

of NE Japan (modified after Zhao et al. 

[1994]). There is no vertical 

exaggeration. (b) P wave tomographic 

image of Tonga subduction zone 

(modified after Zhao et al. [1997]). 

There is no vertical exaggeration. For 

both Figures 6a and 6b, red and blue 

colors denote regions where the P 

wave velocities are relatively slow and 

fast, respectively, compared to average 

mantle at the same depth. (c) P wave 

velocity structure across the Cascadia 

Subduction Zone (modified after 

Parsons et al. [1998]). Yellow dots 

show earthquakes during 1970–1996 

between 45° and 47° latitude, >M1 at 

depths >25 km and >M4 at shallower 

depths. Note vertical exaggeration is 

2x. Note that the subduction zones in 

Figure 6a and 6b subduct old, cold 

lithosphere, which is relatively easy to 

identify tomographically, whereas the 

subduction zone in Figure 6c subducts 

young lithosphere, which differs less in 

velocity relative to the surrounding 

mantle and is more difficult to image 

tomographically. See color version of 



Mantle structure beneath some subduction zones.

Contrasting tectonic environments in subduction zones. 

Extension or compression in the overiding plate 

Tectonic variability found behind magmatic 

arcs, using the Bolivian Andes and the Lau 

Basin as examples. Dark shading denotes 

lithosphere; underlying white is asthenosphere. 

(a) Tectonic setting of Lau Basin (modified 

after Zhao et al. [1997]), emphasizing 

lithospheric structures. Dashed box shows 

region of detail shown in Figure 20b. (b) 

Tectonic setting of Andes between 20°S and 

24°S (modified after Yuan et al. [2000]) 

emphasizing crustal structure. Area of detail 

shown in Figure 20d lies partly within the openended 

dashed rectangle. (c) Cross section across 

the Lau Basin, emphasizing lithospheric 

structure, at the latitude investigated by Ocean 

Drilling Progdram Leg (∼20°S) (modified after 

Hawkins [1994]). Extension rate is from GPS 

measurements of Bevis et al. [1995] for 

velocities of Tonga from Australia; note that the 

highest rate is at the northern end of Lau Basin 

(∼17°S), and the rate decreases to the south. (d) 

Cross section across Andean back arc region, 

emphasizing crustal structure, about 20°S 

(modified after Gubbels et al. [1993]). 

Shortening rate is from GPS measurements of 

Bevis et al. [2001]. Figures 20c and 20d have 

the same horizontal scale. There is no vertical 

exaggeration on any of the sections. 



• Compression in the eastern 

Pacific. 

• Focal mechanisms in the Peruvian 

Andes. 

• Stress in the overriding plate is 

compressive in the Peruvian Andes 

where subduction is shallow. 

• It suggests that there is crustal 

shortening in the overriding plate 

in the Andes.

• Deformation in the Peruvian Andes. 

• GPS measurement 

• Geological rates of deformation 


• Extension in the western 

Pacific 

• Focal mechanisms show tensile stress 

in the overriding plate in the Tonga 

Kermadec region. 

• Stress consistent with back arc 

extension and marginal basin 

formation 


Stress regime determines tectonic style in the overriding plate 

• End-member types of subduction 

zones, based on the buoyancy of 

lithosphere being subducted (modified 

after Uyeda and Kanamori [1979]). 

• Usually young lithosphere is buoyant, 

but old lithosphere could be also be 

light if it includes oceanic plateaus or 

fossil ridges (e.g. Ontong-Java in the 

western Pacific, the northern Andes). 


Back arc spreading 

• Several 

mechanisms have 

been proposed 

• Convection 

entrained by 

subducting plate 

and enhanced by 

melting 

• Slab rollback 

(Scotia plate, and 

others) 


Example of back arc extension-east Scotia sea 

The back arc spreading forms normal 

sea floor with well developed marine 

magnetic anomalies. 

Interpreted magnetic anomaly map of 

the East Scotia Sea. Anomaly 

identifications are essentially the same 

as those of Barker [1995], but recently 

identified pseudofaults [Livermore et 

al., 1994, 1997], formed by ridge 

segment propagation [Hey, 1977], are 

incorporated in this interpretation. The 

central Bruhnes anomaly as well as 

anomalies 2A and 3 are shaded. Ridge 

crest segments E2–E9 are numbered. 

The 2500-m contour (dashed line) 

locates the South Sandwich island arc 

and the North and South Scotia Ridge. 

The ornamented line represents the 

trench. The islands identified are 

Zavodovski (Z), Candlemas (C), 

Montagu (M), and Southern Thule 

(ST). 



Back arc extension in the Philippine Basin at the Izu-Bonin trench 

Bathymetry Magnetics Gravity 




• Free-air gravity anomalies of the Ontong Java Plateau–Solomon Islands convergent zone derived from Geosat and ERS-1 

altimetry data (Sandwell and Smith, 1997). Arrows indicate direction and rate of Pacific plate relative to adjacent plates 

from DeMets et al. (1994). Large yellow areas are known or inferred oceanic plateaus compiled by Coffin and Eldholm 

(1994); dashed yellow lines represent hotspot tracks or "tails". Key to abbreviations: NB=Nauru basin; ER=Eauripik Rise; 

LP=Louisiade Plateau; OJP=Ontong Java Plateau; VAT=Vanuatu trench; VT=Vitiaz trench; NFP=North Fiji Plateau; 

MP=Manihiki Plateau; SHS=Samoa hotspot; TT=Tonga trench; LR=Louisville ridge; HP=Hikurangi Plateau; 

THS=Tasminid hotspot; LHS=Lord Howe hotspot. Note the general difficulty in correlating individual hotspot tracks to 

oceanic plateaus because of intervening subduction zones. 


Evolution of the Tonga Fiji subduction? 

Tectonic reconstruction of the 

New Hebrides - Tonga region 

(modified and interpreted from 

Auzende et al. [1988], Pelletier 

et al. [1993], Hathway [1993] 

and Schellart et al.(2002a)) at (a) 

~ 13 Ma, (b) ~ 9 Ma, (c) 5 Ma 

and (d) Present. The Indo- 

Australian plate is fixed. DER = 

d'Entrcasteaux Ridge, HFZ = 

Hunter Fracture Zone, NHT = 

New Hebrides Trench, TT = 

Tonga Trench, WTP = West 

Torres Plateau. Arrows indicate 

direction of arc migration. 

During opening of the North Fiji 

Basin, the New Hebrides block 

has rotated some 40-50° 

clockwise [Musgrave and Firth, 

1999], while the Fiji Plateau has 

rotated some 70-115° 

anticlockwise [Malahoff et al., 

1982]. During opening of the 

Lau Basin, the Tonga Ridge has 

rotated ~ 20° clockwise [Sager et 

al., 1994 


Volcanic activity in the overriding plate 

• Magmatic arc complexities. Only an 

idealized section through an 

intraoceanic arc is shown; similar 

processes are expected beneath 

Andean-type arcs. Note that the 

asthenosphere is shown extending up 

to the base of the crust; delamination 

or negative diapirism is shown, with 

blocks of the lower crust sinking into 

and being abraded by convecting 

mantle. Regions where degassing of 

CO2 and H2O is expected are also 

shown. 


• What is the cause of the 

magmatic activity? 

• Simple thermal models show that the 

down-going plate remains cold and could 

not melt. 

• It is likely that melting occurs because 

fluids are expelled from the downgoing 

slab that lower the melting temperature 

in the mantle. 


Effect of water on melting of 

peridotite. (a) Pressuretemperature 

diagram for H2O 

undersaturated (0.2–0.5 wt % 

H2O) and anhydrous partial 

melting of mid-ocean ridge basalt 

(MORB) and pyrolite mantle 

composition (modified after 

Ulmer [2001]). Dashed lines 

indicate the stability limits of 

garnet peridotite (gar), spinel 

peridotite (sp), plagioclase 

peridotite (plg), and amphibole 

peridotite (amph). The dash-dot 

line corresponds to the average 

curent mantle adiabat (ACMA) 

corresponding to a potential 

temperature of 1280°C. (b) Plot 

of melt fraction versus 

temperatures of anhydrous batch 

melts of a depleted MORB 

mantle source at 1.5 GPa with the 

liquidus temperatures of hydrous 

batch melts from peridotite 

containing 0.15% H2O and 

0.32% of simplified Mariana 

subduction component, which is 

an aqueous fluid with dissolved 

solutes [Stolper and Newman, 

1994]. Upper axis shows 

forsterite content of equilibrium 

olivine. Figure 13b is from 

Gaetani and Grove [1998]. 

Effect of water on the solidus 


Melting occurs only in a very narrow range. It 

determines the location of volcanic arc. 

• This model accounts for the entrainment of the 

asthenosphere by the downgoing slab (which 

makes it colder at given depth. There is a narrow 

window where the geotherm intersects the solidus 

for wet mantle. 

• (a) Material movement (arrows) and temperature 

structure (dashed lines) of a simplified convergent 

margin (modified after finite element models of 

Davies and Stevenson [1992]). Crust and mantle 

lithosphere of the subducting plate move at a 

constant velocity of 7 cm yr-1, dragging the 

asthenosphere with it beneath the overriding plate. 

Note also that the isotherms remain approximately 

parallel with the original seafloor, even deep into 

the subduction zone. The shaded area in the 

mantle wedge approximates the location of detail 

shown in Figure 12. 

• (b) Temperature profile beneath the arc volcanoes 

(bold line) and melting curve for wet mantle (thin 

line). Note that the temperature reaches a 

maximum at ∼80 km depth and then decreases as 

the subducted lithosphere is approached, reaching 

a minimum at a depth of ∼120 km. The 

temperature is high enough for mantle melting 

only at a depth of ∼80 km, ∼40 km above the 

subducted plate. 


Volcanic Arc 

Model for dehydration of subducted materials 

(modified after Schmidt and Poli [1998]). 

Dehydration of subducted peridotite and oceanic 

crust occurs continuously to ∼150–200 km; thus 

water is continuously supplied to the overlying 

mantle. The shaded region in the mantle wedge 

labeled “partially molten region” is expected to 

melt to a significant degree. The volcanic front 

forms where the amount of melt can separate and 

rise to the surface. Open arrows indicate rise of 

fluid, solid arrows indicate rise of melts. Long 

arrows indicate flow of asthenosphere in the mantle 

wedge. Stippled area marks stability field of 

amphibole beneath the forearc. Dashed lines 

outline stability fields of hydrous phases in 

peridotite. The horizontally ruled region shows 

where talc is stable. For some thermal structures a 

portion of the peridotitic lithosphere will be 

• Detailed cartoon of the mantle wedge (region of detail 

depicted by shading in Figure 11), showing how fluids 

might move from the subducted slab into a region where 

melts could be generated (modified after Davies and 

Stevenson [1992] and Stern [1998]). Water is carried in 

the descending slab (A), and dense aqueous fluids are 

continuously released from subducted sediments, crust, 

and serpentinites (B). These fluids rise into the overlying 

mantle to form hydrous phases in mantle peridotite (C). 

Amphiboles are shown forming here, but it could also be 

another hydrous mineral. Metasomatized mantle 

descends with the subducted slab. At the maximum 

depth of stability for amphibole peridotite, ∼100 km, it 

breaks down to anhydrous peridotite and aqueous fluid 

(D). The fluid rises vertically, moving away from the 

subducted slab and toward hotter regions in the mantle. 

At some point the rising water will react with anhydrous 

peridotite to form amphibole again (E). This descends 

until the amphibole breaks down again (F). Amphibole 

peridotite forms (G), descends, and breaks down again 

(H). The dark shaded area is mantle that can melt if 

sufficient water is provided. Above point H the mantle is 

sufficiently hot that water added to it leads to melting (I). 

The mantle is still moving downward, so melt will not be 

able to rise until enough of it accumulates to form a 

sufficiently large diapir (K) as a result of Rayleigh- 

Taylor instability. Partially molten diapirs are less dense 

than the surrounding mantle and can rise through it at a 

rate of perhaps 1 m yr-1 (L). At the base of the overlying 

mantle lithosphere (M), magma separates from the 

unmelted part of the diapir to feed an arc volcano. 


• Temporal evolution of island arc 

magmatism. 

• Early stage expulsion of water (80-100km) 

causes partial melting of mantle with 

tholeitic magmas. 

• At greater depths, the partially molten 

region ascents as a diapir and adiabatic 

decompression causes more melting and 

differentiated calc-alkaline magmas. 


• Andean style batholiths 

• New continental crust is added 

directly to the continent through 

magmatic activity. 

• Sr 87 /Sr 86 indicate mantle origin 

with continental crust 

contamination. 

• Be 10 (1/2 life = 1.5 My) was 

found in some of the lavas. It 

could come only from sediments 

that were subducted, melted, and 

included in the magmas. 

• Note large batholiths in 

continents used to detect ancient 

Andean type margin (e.g.dePas 

batholith in NQO). 


Proterozoic examples: 

Is the elongated de Pas 

batholith in the Core Zone 

anologuous to the batholith in 

the Andes? 

Suggests subduction of the 

Superior plate beneath the core 

zone during the NQO 



Wathaman batholith on margin of Hearne. No batholith in the 

Superior => No subduction beneath Superior during closure of 

Maniwekan ocean?


• Andes elevation ~ 6km 

• Crustal thickness is 

estimated 60-70km 

• Crustal thickening by 

tectonic shortening and by 

magmatic underplating


• Receiver function (RF) images and crustal 

models of the Central Andes along an east– 

west profile. In the images, red indicates 

positive, velocity increase downwards; blue 

indicates negative, velocity reduction 

downwards. a, Time domain RFs averaged 

over a 30-km-wide moving window, 2–10-s 

band pass filtered. b, Possible crustal S-wave 

velocity (vS) models resulting from the 

inversion of the RFs in a. c, Interpretative 

cartoon with depth-migrated RF image as 

background averaged over the Altiplano and 

Puna regions. Black plus signs show Moho 

depth data from wide-angle reflection 

studies4. Local crustal models and a global 

mantle reference model were used for 

migration. The TRAC1 and TRAC2 

converters in a and c border the Andean lowvelocity 

zone (ALVZ) which dips westwards 

from below the Eastern Cordillera fault 

system (indicated in c by black lines at the 

surface) across the entire Altiplano/Puna to the 

Precordillera (red points at surface indicate 

Cenozoic volcanoes). Figure S1 in 

Supplementary Information shows this data 

set split in two at 23° S for judging north– 

south variations of the TRAC2 converter. 

Figure S2 in Supplementary Information 

shows the lateral extension of the ALVZ. The 

Nazca converter (thick black-white dashed 

line indicated in c) is interpreted as an image 

of the oceanic Moho. The upper boundary of 

the oceanic crust (slab shear zone) is set 

10 km above the oceanic Moho in agreement 

with the results of waveform inversion (see 

text and Fig. 4). Oceanic crust is clearly 

visible from converted waves only down to

Electrical conductivity of the mantle 

is measured by MT 


Contours: depth of Nazca plate


High conductivity = hydrated mantle

Metamorphism in subduction zones 

• Metamorphism in subduction zones 

• High pressure in the downgoing slab 

• Low pressure in the overiding plate 


Paired metamorphic belts in Japan 

Note paired metamorphic belts are found around the Pacific (with the right polarity) 


Thermal models of end-member (young and hot 

versus old and cold) subduction zones. (a) NE Japan 

arc, a good example of a cold subduction zone. (b) 

SW Japan arc, a good example of a hot subduction 

zone. Note the great difference in the temperature of 

the slab interface at 50 km depth (T50km) and 

beneath the volcanic front (Tvf) but the small 

difference in the maximum temperature of the mantle 

wedge beneath the volcanic front (Tmw). (c) 

Pressure-temperature diagram showing metamorphic 

facies and melting relations for basaltic oceanic crust, 

along with trajectories for crust subducted beneath 

NE and SW Japan Roman numerals identify fields for 

metamorphic facies: I, greenschist; II, epidote 

amphibolite; III, amphibolite; IV, granulite; V, epidote 

blueschist; VI, lawsonite blueschist (green field 

shows location of the eclogite field, with the dark 

dashed line separating this from blueschist); VII, 

chloritoid-amphibolite-zoisite eclogite; VIII, zoisitechloritoid 

eclogite; and IX, lawsonite-chloritoid 

eclogite. Note that oceanic crust subducted beneath 

SW Japan enters the eclogite field at ~40 km depth 

and, if hydrous, begins to melt at ~90 km, whereas 

oceanic crust subducted beneath NE Japan barely 

enters the eclogite field at 120 km depth. Figures 6a– 

6c are modified after Peacock and Wang [1999] and 

Peacock [2002]. See color version of this figure at 

back of this issue. 


• The Canadian Cordillera 

• Accreted terranes in the Canadian 

Cordillera. 

• Some terranes are of North 

American affinity 

• Others are of oceanic affinity. 


• Main geological belts of the 

Canadian Cordillera and Bouguer 

gravity anomaly. 

• Intermontane and Coast belts are 

accreted terranes. 

• Omineca and Foreland belts are of 

north American affinity. 



Schematic cross section of the Canadian Cordillera

Geodynamic model of the Canadian Cordillera after the LITHOPROBE studies 

• Intermontane belt transported on North America basement. 

• Coast belt docked on the side. 

• Sediments 


a, Unmigrated section showing the reflection, labelled JdF, from the top of the subducting Juan de Fuca plate west of 

Vancouver Island, the E reflection zone, and the F reflection which marks the top of the subducting Juan de Fuca plate further 

east. Reflections from the Leech River crustal fault, LRF, are truncated by the E reflections. 

b, Migrated section superimposed on a display of P wave velocities derived by 3D tomographic inversion of first arrivals from 

local earthquakes and wide-angle airgun shots around Vancouver Island9. The velocities were extracted from the 3D velocity 

model along the composite seismic profile shown by the red line in Fig. 1 and projected in the same way as the seismic 

reflection data. Earthquakes, which were relocated in the tomographic inversion, are shown by filled black circles. The nearlinear 

alignment of east-dipping inslab earthquakes just above the oceanic Moho at 45–55 km depth may indicate delaminatio 

of the oceanic crust 



• Comparison of scattered wave inversion results 

with thermal model. a, S-velocity perturbations 

below the array, recovered from the inversion of 

scattered waves in the P-wave coda of 31 

earthquakes recorded at teleseismic distances. The 

image represents a bandpass-filtered version of the 

true perturbations to a one-dimensional, smoothly 

varying reference model. Discontinuities are 

present where steep changes in perturbation 

polarity occur. b, Thermal model of Cascadia 

subduction zone corresponding approximately to 

the profile in a. The cool subducting plate 

depresses isotherms in the forearc, rendering 

serpentine stable within that portion of the mantle 

encompassed by the dashed rectangle; solid lines 

indicate locations of subducting oceanic crust and 

continental Moho. Note temperature contour 

interval is 200 °C. c, Interpretation of structure in 

a. High degrees of mantle serpentinization where 

the subducting oceanic crust enters the forearc 

mantle results in an inverted continental Moho 

(high-velocity crust on low-velocity mantle), 

which gradually reverts eastward to normal 

polarity by -122.3° longitude. The signature of the 

subducting oceanic Moho diminishes with depth 

as a result of progressive eclogitization below 

45 km. Inverted triangles in a and c show 

instrument locations.


What happened to the Farallon plate? 

It is still down there!


• Cross-section through the Cascadia subduction zone on the northwest coast of 

the United States. Water released from the subducting oceanic crust rises up and 

hydrates the forearc mantle; water released deeper down causes partial melting 

of rock to magma that erupts in the arc volcanoes. In the water-altered — or 

'serpentinized' — forearc rocks, seismic waves travel more slowly and the usual 

gradient of wave speed is reversed, producing an effect known as an 'inverted 

Moho', which has been detected by Bostock et al.1. The slippery nature of the 

serpentinized forearc mantle limits the seismogenic depth of coupling between 

the two plates, and so might also limit the impact of earthquakes in the region.


• Foreland basins are caused by the bending of the plate under 

the load of the accreted terranes

The accretion of terranes over the western margin of Canada caused the 

formation of the Alberta basin 


Forearc basin-Accretionary prism 

End-member forearc types: (a) accretionary 

forearc (modified after Dickinson [1995]) 

and (b) nonaccretionary forearc (P. Fryer, 

personal communication, 2001). Note that 

the abundance of sediments associated with 

accretionary forearcs is manifested as an 

accretionary prism and as a thick forearc 

basin and that the relative lack of sediments 

leaves the nonaccretionary forearc exposed. 



• Mechanical model of an accretionary wedge. 

• Balance of forces implies a positive slope for the wedge

Erosion and subduction of sediment in crust could exceed accretion of new 

crust. 

• Map showing the distribution of accreting versus eroding subduction zones 

considered within this study. Accretionary margins are shown with solid 

barbs on the plate boundary, while open barbs mark erosive margins. 


Compilation of profiles across 

accretionary plate margins. Profiles are 

redrawn and resized to a common scale 

in order to allow direct comparison of 

different margins. Sources for the 

original data are shown next to each 

profile. 


Compilation of profiles across nonaccretionary and erosive plate margins. 

Profiles are redrawn and resized to a common scale in order to allow direct 

comparison of different margins. 


Budget of all continental margins 

Diagram showing the integrated growth 

or erosion rate of each active plate 

margin in relation to the global average 

growth rate required to maintain the 

continental freeboard. Note that several 

erosive plate margins are actively 

growing crust despite rapid loss at the 

trench. 


• Pie chart showing the relative proportions 

of the major inputs and outputs from the 

global subduction systems with respect to 

the crust. Note the dominance of arc 

magmatism over subduction accretion as a 

source of new material. 


When did subduction processes begin on Earth ? 

• Evidence for late Archean subduction? 

• Seismic reflector in the mantle interpreted as a relict subducted slab. 


Mantle reflectors could be fossil 

Archean subduction zones. 

(example from the Slave Province, LITHOPROBE 

SNORCLE transect). 


Continental crust evolution? 

• Today continental crust is being destroyed at about the same rate it is accreted. 

• We do not know how the total volume of the continental crust has changed with 

time. 

• Model 4 with a rapid growth of the continental crust at the end of Archean is 

plausible but so are the other models 

• Episodic growth models have also been proposed. 


Summary 

• Subduction of sea floor independent of age 

• But tectonic style might depend on age of the subducted slab (western vs 

eastern Pacific) 

• Plate remains cold 

• Expulsion of volatiles causes partial melting in the mantle 

• Magmas in volcanic arc from mantle (but …) . Main source of continental 

growth 

• High pressure low temperature metamorphism in subducted slab 

• Accretion of terranes in North American cordillera 

• Plate tectonic and subduction in late Archean at 2.7 Ga 

• Budget of continental crust? 


Open questions 

• What initiates subduction? 

• Mechanism of accretion. 

• When did present day style of subduction start? 

• Meaning of Archean and Proterozoic subcrustal reflectors. 

• What is crustal accretion destruction present budget? 

• How did total volume of continental crust change with time? 


• Importance of convergence rate and age 

of subducted lithosphere. (a) Plot of 

length of seismic zone as a function of 

the product of convergence rate and age 

of subducted lithosphere. Approximate 

uncertainties are given by error bars. 

Dashed line corresponds to length of 

seismic zone equal to convergence rate 

multiplied by age divided by 10 

(modified after Molnar et al. [1979]). 

(b) Plot of convergence rate versus 

lithosphere age, showing the strong 

influence this relationship has on 

seismicity (modified after Ruff and 

Kanamori [1980], using GPS-defined 

convergence rates for Pacific-Tonga 

[Bevis et al., 1995]. The number at each 

subduction zone is the associated 

maximum Mw (seismic moment 

magnitude), and the contours of 

constant Mw are defined by regression 

analysis. Shaded area outlines 

subduction zones associated with back 

arc spreading or interarc rifting. Note 

that most of the biggest earthquakes 

occur at Andean-type margins. 


• Diagnostic trace element differences 

between adakites and more common 

andesite-dacite-rhyolite (ADR) suites 

(modified after Defant and Drummond 

[1990]). Adakites form by melting at 

high pressure in equilibrium with 

garnet (garnet has a high partition 

coefficient for Y and low partition 

coefficients for Sr) and so have low-Y 

contents and high Sr/Y. ADR suites 

form at lower pressure in equilibrium 

with plagioclase (plagioclase has a 

high partition coefficient for Sr and 

low partition coefficients for Y) and so 

have high Y contents and low Sr/Y. 



• Incompatibility plot (spidergram) for important subduction zone inputs 

(global subducting sediment (GLOSS) and normal mid-ocean ridge basalt 

(NMORB)) and typical outputs (boninite, arc tholeiite, back arc basin 

basalt, dacite, and continental crust) from Table 2. Elements on the 

horizontal axis are listed in order of their incompatibility in the mantle 

relative to melt; elements on the left are strongly partitioned into the melt, 

whereas those on the right are strongly partitioned into peridotite minerals. 

Composition of NMORB and element order is after Hofmann [1988]. Note 

the characteristic enrichments of GLOSS and subduction zone outputs 

relative to MORB with respect to fluid-mobile large-ion lithophile elements: 

Rb, Ba, U, K, Pb, and Sr; note the relative depletion of these in high field 

strength elements: Nb, Ta, Zr, Ti, Y, and heavy rare earth elements. Note 

also the greater enrichment and overall similarity of the “continental 

suite” (GLOSS, continental crust, and Chilean dacite) on the one hand and 

the “oceanic suite” (boninite, arc tholeiite, back arc basin basalt, and 

MORB) on the other. Shaded field encompasses the continental suite.

• Model of subducted slabs (modified after Stein and Rubie [1999]). (top) Predicted mineral phase boundaries 

and (bottom) resulting buoyancy forces in a downgoing slab with (left) equilibrium mineralogy and (right) 

for a nonequilibrium metastable olivine wedge. Assuming equilibrium mineralogy, the slab has significant 

negative thermal buoyancy (dark shading in bottom graphs), due to both its colder temperature and the 

elevated 410 km discontinuity, and significant positive compositional buoyancy (cross hatching) associated 

with the depressed 660 km discontinuity. If a metastable wedge is present, it adds positive buoyancy and 

hence decreases the net negative buoyancy force driving subduction. Units are in 103 N m-3. 


Map of Pacific and Caribbean LIP's compared to 

modern plate boundaries (yellow lines), 

active hotspots/mantle plumes (yellow stars), 

and rates of the Pacific and Caribbean plate 

relative to surrounding plates from global 

plate motion model of DeMets et al. (1994). 

Oceanic plateaus are subdivided into three 

tectonic settings as shown in the key to the 

upper right. Box denotes Solomon Islands– 

Ontong Java Plateau study area for papers in 

this volume. Key to abbreviations for larger 

LIP's and hotspot tracks: HE=Hawaii– 

Emperor seamount chain; SR=Shatsky Rise; 

SO=Sea of Okhotsk; OP=Ogasawara 

Plateau; MNR=Marcus Necker Ridge; 

OJP=Ontong Java Plateau; MP=Manihiki 

Plateau; HR=Hikurangi Plateau; NR=Nazca 

Ridge; CR=Carnegie Ridge; COR=Cocos 

Ridge; CP=Caribbean Plateau. Numbered 

LIP's 1–21 are keyed to Table 2A (oceanic 

plateau or hotspot tracks presently 

subducting at Pacific and Caribbean plate 

boundaries) and Table 2B (proposed oceanic 

plateaus accreted during Phanerozoic 

collision events involving island arcs or 

continental orogenic belts). (B) Schematic 

diagram summarizing four possible 

relationships between hotspot or mantle 

plume heads and hotspot tracks or "tails" in 

intraplate (1), subduction (2, 3) and in 

ancient orogenic belts (4). See text for 

discussion. 


• Potassium-silica diagram for representative arcs. 

Dashed line defines boundary between shoshonitic 

and calc-alkaline and tholeiitic suites (CATS). Izu- 

Bonin and Mariana arcs are exemplary of 

intraoceanic arcs (fields from Stern et al. [2002], 

note that volumetrically subordinate Mariana 

shoshonites [Sun and Stern, 2001] are omitted). 

Field for Andes, 16°S–26°S encompasses most of 

606 Plio-Pleistocene and younger samples from 

the Central Volcanic Zone (CVZ) (G. Wörner, 

personal communication, 2002). Abbreviations are 

as follows: M, typical MORB from Table 2; B, 

back arc basin basalt from Table 2; I, mean 

composition of Izu-Bonin arc samples; MA, mean 

composition of Mariana arc samples; CC, bulk 

continental crust from Table 2; G, GLOSS from 

Table 2; A, mean composition of Andes CVZ 

lavas; and UC, composition of upper continental 

crust [from McLennan, 2001]. Dark shading 

encompasses mean and typical compositions of 

the “oceanic suite” (MORB, back arc basin basalt, 

and mean Mariana and Izu-Bonin lavas), and light 

shading encompasses mean and typical 

compositions of the “continental suite” (bulk 

continental crust, GLOSS, upper continental crust, 

and mean Andean dacite). 


Seismic tomography of two arcs 

• Crustal structure of typical arcs, based on P 

wave velocities. Note vertical exaggeration 

is ∼5x. 

• (a) Izu arc, 33°N (modified after Suyehiro 

et al. [1996]). 

• (b) Eastern Aleutian arc (modified after 

Holbrook et al. [1999]). 

• Note differences in thickness and velocity 

structure. We do not have comparable 

tomographic images of Andean-type arc 

crust. See color version of this figure at 

back of this issue. 


• Comparison of pattern of deformation 

observed in Ontong Java Plateau–Solomon 

arc convergent zone with examples of 

Phanerozoic and Precambrian patterns of 

deformation in collision zones. Key to 

lithospheric compositions: RED=upper 

crustal, mainly igneous rocks; 

YELLOW=upper crustal, mainly 

sedimentary rocks; gray and BLACK=upper 

mantle, mainly ultramafic rocks. (A) 

Ontong Java Plateau—initial subduction– 

accretion of uppermost plateau in western 

Malaita accretionary prism (modified from 

Rahardiawan et al., 2004). Open dots 

represent ISC hypocenters (M>4.0). (B) 

Ontong Java Plateau—more advanced 

subduction–accretion in eastern Malaita 

accretionary prism. (C) Cenozoic Alpine 

crustal-scale triangle zone or 

"flake" (modified from Oxburgh, 1972); 

L=base of lithosphere; M=subducted Moho. 

(D) Precambrian Canadian crustal-scale 

triangle zones or "flakes" (modified from 

Cook et al., 1998); M2=base of Slave 

Province mantle. (E) Precambrian African 

thrust-imbricated, subduction-related prism

Subduction of oceanic lithosphere

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?