the caribbean plate: pulled, pushed, or dragged? · table 1 gives a summary of all forces that we...

The Caribbean plate: Pulled, pushed, or dragged?

S. van Benthem1 and R. Govers1

Received 3 September 2009; revised 14 April 2010; accepted 10 May 2010; published 13 October 2010.

[1] Mechanical coupling between the lithosphere and the asthenosphere remains acontroversial topic in the geosciences. Beneath the Caribbean plate, shear wave splittingmeasurements indicate EW strain in the asthenosphere, which can be interpreted as mantleflow driving or resisting motion of the overlying lithosphere. Here, we constrain the averageshear traction on the base of the Caribbean plate by balancing all torques. These torquesresult from body forces that act on the Caribbean (slab pull, ridge push, lateral densityvariations), from plate boundary friction and from basal shear tractions. We obtain a rangeof physically realistic torque solutions, which we examine further by computing thecorresponding stresses and rotations within the Caribbean plate for comparison withobservations. The deformation field for the Caribbean is particularly sensitive to the amountof friction on intraplate faults. Representative models have a good fit with observations andare characterized by (1) a near!zero basal shear traction (!0.3 MPa), (2) (lithosphere!averaged) plate boundary friction ! 10MPa, (3) local forces due to indenters and trench pull,(4) a net pull by the Caribbean slab and (5) intraplate fault shear stresses on the order oftens of megapascals. We conclude that the mechanical coupling of the Caribbean plate to theunderlying asthenosphere is small.

Citation: van Benthem, S., and R. Govers (2010), The Caribbean plate: Pulled, pushed, or dragged?, J. Geophys. Res., 115,B10409, doi:10.1029/2009JB006950.

1. Introduction

[2] The lithospheric stress/deformation field is an expres-sion of a variety of forces that act on the lithosphere. Thoseforces drive deformation on both geological and human timescales, e.g., both fault slip and earthquakes. Here we aimconstrain the most significant of these forces for the Carib-bean plate. Researchers debate the relative importance ofasthenospheric flow and lithospheric forces like ridge push orslab pull. Little support remains today for the most extremeviewpoints, that the lithosphere is either driven entirely bymantle convection or entirely by plate internal body forces.Low basal shear stresses (<0.3 MPa) acting on the base of thelithosphere have been found by Forsyth and Uyeda [1975],Richardson et al. [1979], Cloetingh and Wortel [1985],Meijer and Wortel [1999] and Govers and Meijer [2001].Low lithosphere!mantle coupling is also suggested by globalgeophysical data which require a significant low!viscosityregion in the upper mantle [Forte andMitrovica, 2001]. Otherstudies conclude however that significant (1–30 MPa) shearstresses are needed at the base of the plates [Steinberger et al.,2001; Conrad and Lithgow!Bertelloni, 2002; Negredo et al.,2004; Barba et al., 2008; Bird et al., 2008].[3] The Caribbean plate (Figure 1) is well suited to study

coupling to asthenospheric flow; shear wave splitting ob-servations reveal significant anisotropy in its upper mantle,

which has been interpreted as eastward (return) flow aroundthe northern edge of the Nazca slab [Russo et al., 1996;Schellart et al., 2007]. In a model study of the Caribbeanplate, Negredo et al. [2004] best fitted surface observationswhen they assumed such eastward mantle flow to bemechanically coupled to the overlying lithosphere. Theirfinding is at odds with studies by Bird [1998], Alvarez [1982]andConrad and Lithgow!Bertelloni [2006], who suggest thatcoupling is not expected for plates lacking a significant lith-ospheric root, such as the Caribbean plate. Bird et al. [2008]also find high (6 MPa) stresses at the base of the Caribbeanplate but small plates probably fall outside the resolution oftheir model. We aim to resolve the controversy about high orlow basal shear stresses by directly estimating the shearstresses on the base of the Caribbean plate from torque bal-ance. This yields a suite of force models. All models arecharacterized by a low (<0.4 MPa) average shear stress on thebase of the plate. Shear coupling of the Caribbean plate to theunderlying mantle is thus low.[4] For all force models, we compute deformation gra-

dients and the sense of slip on regional faults within theCaribbean domain. Based on the differences between theresulting stress models, we find that only two parametersproduce significantly different model classes: The intraplatefault friction and the amount of trench suction by the Antillesslab. Based on the fit with observations we find that values onthe order of tens of MPa for the fault friction give best results.However, we cannot discriminate between regional forces,such as the amount of trench suction or higher frictionalcoupling at subducting bathymetric highs. These forcesimprove the fit to the regional stress field by an amount that is

1Faculty of Geosciences, Utrecht University, Utrecht, Netherlands.

Copyright 2010 by the American Geophysical Union.0148!0227/10/2009JB006950

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, B10409, doi:10.1029/2009JB006950, 2010

B10409 1 of 14

http://dx.doi.org/10.1029/2009JB006950

insignificant on the overall plate scale. Our research questionof the title of this paper is thus only partially answered. Shearstresses from mantle flow are low (on average) i.e., theCaribbean plate is not dragged. It is pulled (by slab pull, andpossibly by the trench suction) and is pushed (sheared) bysurrounding plates, possibly with a locally higher coefficientof friction. The most important parameter changing thedeformation field is the intraplate fault friction.

2. Caribbean Plate

[5] The Caribbean plate is a relatively small tectonic plate(Figure 1). The mechanical entity that we refer to as the“Caribbean plate” includes independent moving blocks alongits plate boundaries. We based the location of plate bound-aries and interior faults on the geological map of French andSchenk [2004].[6] Neighboring plates are the North American plate in the

north and east, the South American plate in the south and east,and the Cocos plate in the west. Plate boundaries in the northand southeast are predominantly strike!slip, whereas at theeastern plate boundary, the North and South America platesare subducting under the Lesser Antilles island arc [Molnarand Sykes, 1969]. At Hispaniola, Puerto Rico, and theVirgin islands, relative plate motion (RPM) is highly oblique.In the west, the Cocos plate is subducting at the Middle

America Trench. The Caribbean plate itself subducts underthe North Andes block in the southwest near the coast ofColombia. The northern and southern plate boundaries arenot characterized by a single fault but rather by a diffusesystem of faults, ranging from strike!slip to thrust faulting,thus accommodating relative plate motion [Clark et al., 2008;Magnani et al., 2009]. The discrete plate boundary in ournumerical model is an approximate representation of thisshear zone. We investigated the sensitivity of our resultsto this approximation.[7] The Cayman Trough in the north is a pull!apart basin,

where new oceanic lithosphere has been continually createdsince the Eocene [Rosencrantz et al., 1988]. Oceanic agesfor the remainder of the Caribbean are unknown, due to thedeposition of flood basalts on top of the Caribbean litho-sphere in the late Cretaceous [Hoernle et al., 2004]. TheGrenada basin is suggested to be the result of back!arc orfore!arc spreading in the early Tertiary [Bird et al., 1993;Aitken et al., 2009] or a piece of trapped Atlantic lithosphere,without any extension taking place [Kearey, 1974]. Alongits northern boundary the Caribbean plate is fragmented intothe smaller Gonave block [Rosencrantz and Mann, 1991;DeMets and Wiggins!Grandison, 2007], the Hispaniolablock [Mann et al., 2002], the Puerto Rico block [Byrneet al., 1985; Jansma et al., 2000] and possibly the NorthernAntilles fore!arc block [López et al., 2006]. In the southwest,

Figure 1. Tectonic setting of the Caribbean plate. Major tectonic elements and faults are AF, AnagedaFault; AR, Aves Ridge; BER, Beata Ridge; BP, Bahamas Platform; BR, Barracuda Ridge; CAF, CentralAmerican Fore!arc; CAFSS, Central American Fore!arc Strike!Slip zone; CB, Colombia Basin; CR, CocosRidge; CT, Cayman Trough; EPGF, Enriquillo Plantain Garden Fault zone; G, Gonave block; GB, GrenadaBasin; H, Hispaniola block; HF, Hispaniola Fault zone; LA, Lesser Antilles island arc; LAB, Lesser Antillesfore!arc Block; MB, Maracaibo Basin; MS, Maracaibo Subduction zone; MT, Muertos Trough; NR, Nor-mal Ridge; NPDB, North Panama Deformed Belt; P, Panama block; PR, Puerto Rico block; SF, Septentri-onal Fault; TR, Tiburón Ridge; VB, Venezuela Basin. Interpreted principal stress directions from the WSMProject [Heidbach et al., 2008] are shown in blue. Principal strain rate directions from the Global Strain ratemodel [Kreemer et al., 2003] are shown in red. Bottom right inset shows vertical axis rotation fromKreemeret al. [2003]. Top left inset shows the (not interpreted) maximum compression axes of the WSM.

VAN BENTHEM AND GOVERS: CARIBBEAN—PULLED, PUSHED, OR DRAGGED? B10409B10409

2 of 14

there is the Panama block [Kellog and Vega, 1995] andCentral American Fore!arc (CAF) block [DeMets, 2001].These blocks are separated by fault systems from the stableCaribbean plate interior.[8] On the subducting Cocos plate, southwest of Panama

the Cocos ridge is subducting under, or indenting thePanama block. It likely affects the dynamics of the overridingPanama plate and the lateral northwestern escape of the CAFblock [Correa!Mora et al., 2009; LaFemina et al., 2009].Other potentially relevant subducting ridges affecting thedynamics of the Caribbean plate are the Bahamas platform[Mann et al., 2002;Manaker et al., 2008] and Normal Ridge[Grindlay et al., 2005] in the northeast at the Puerto RicoTrench and the Barracuda and Tiburon ridges [McCann andSykes, 1984] at the northeastern Lesser Antilles Trench.

3. Forces on the Caribbean Plate

[9] Three types of forces act on the Caribbean plate(Figure 2): (1) intraplate body forces or driving forces (slabpull, ridge push and other lateral density variations, trenchsuction), (2) coupling stresses to the sublithospheric mantle(shear stresses on the base of the lithosphere and on sub-ducting slabs, corner flow) and (3) friction on plate bound-aries and faults (subduction contacts, transforms, indenters.Table 1 gives a summary of all forces that we consider (formore detail, see Forsyth and Uyeda [1975], Wortel et al.[1991] and Govers and Meijer [2001]). Here we summa-rize the parameterization of the forces. We discretized theCaribbean domain into a finite element grid (Figure 7a), tofacilitate the computation of the forces and torques.

3.1. Driving Forces3.1.1. Slab Pull[10] The Caribbean plate subducts under the North Andes

block along the coast of Colombia and the western end ofVenezuela at the South Caribbean Deformed Belt (SCDF).

This slab is referred to as Maracaibo slab [van der Hilst andMann, 1994]. Further east along the SCDF convergence istaking place, but maximally 65 km since the Miocene [Clarket al., 2008], and no large scale subduction takes place. TheMaracaibo slab exerts a slab pull on the Caribbean plate,which is caused by the negative buoyancy of the descendinglithospheric plate relative to the surrounding hot mantle andacts approximately perpendicular to the trench. To computethe slab pull, we follow the approach of Govers and Meijer[2001]. Relevant parameters include the slab depth (275 km)and dip (17°) which we take from tomography [van derHilst and Mann, 1994]. Reasonable variations in the para-meters used for the slab pull calculation, for example,changing the length or dip of the slab or changing the mantletemperature, do not significantly influence the slab pull(<10%).3.1.2. Ridge Push[11] We use available age information [Müller et al., 2008],

which we average over the elements in our grid, to calculatethe ridge push. The ridge push for a unit column is calculatedusing Pratt isostasy, following the approach of Richter and

Figure 2. Forces acting on lithosphere.

Table 1. Summary of Forces Working on the Lithosphere

Notation Name DirectionMagnitudeKnown?

Frp Ridge Push Age gradient YesFsp Slab Pull ? plate boundary YesFcf Corner Flow ? plate boundary YesFgr Gravitational stresses Topography

gradientYes

Ftf Transform friction RPM From "Ti = 0Fpcr Plate Contact Resistance RPM From "Ti = 0Fsr Shear Resistance APM or RPM From "Ti = 0Fcb Compositional Buoyancy ? plate boundary AssumedFbs Basal shear stress # to APM AssumedFind Oceanic Ridge

subductionRPM 1–10 ! Fpcr

Fts Trench suction RPM Yes


3 of 14

McKenzie [1978]. Ridge push is explicitly defined for oceaniclithosphere, where age information is available—and thusinformation on lateral density variations. In the Caribbean,there is only age information for the Cayman trough (0–49Myr old) [Leroy et al., 2000]. The rest of the plate consistsof Cretaceous or older lithosphere [Pindell and Barrett,1990], covered by flood basalts. Where neither age nor heatflow information is available, we use topography and theassumption of Airy isostasy to obtain information aboutlateral density variations. Lateral density contrasts cause apressure gradient. For a unit column, the difference in pres-sure compared to a reference crust (thickness 35 km, topog-raphy at sea level) can be calculated by

Pgr ! g 12h

2!l!m

!m " !cr# L!crh

! "for h > 0 $1%

Pgr ! g "12h

2$!cr " !w%!m

!m " !cr# L$!cr " !w%h

# $for h < 0 $2%

where g is the gravitational acceleration; h(t) the age depen-dent topography; h the topography; rm, rcr rw are averagedensities of the upper mantle (3200 kg/m3), oceanic crust(3000 kg/m3) and water (1024 kg/m3), respectively; L is thereference thickness for the crust (35 km); and H(t) is the agedependent thickness of the lithosphere. We assume isostaticequilibrium, so that the pressure gradient equals the hori-zontal stress between two adjacent columns [Meijer andWortel, 1992].3.1.3. Trench Suction[12] The Grenada basin is suggested to be either a Creta-

ceous!Eocene back!arc or fore!arc basin, resulting from slabroll!back [Bird et al., 1993; Aitken et al., 2009] or a trappedpiece of Atlantic lithosphere [Kearey, 1974]. No activespreading is taking place at present, although Schellart et al.[2007] suggest in their global compilation that trenchmigration still takes place at the Antilles. Using the globalplate boundary model of Bird [2003], they estimated thelocation of the trench and obtained a trench migration rate of18 mm/yr, suggesting a tensile force acting on the overridingplate. We name this tensile force the trench suction (Fts), andit is determined using the sea anchor model of Scholz andCampos [1995]. In this model the trench suction is com-puted from the resistance to a face!perpendicular translationof an ellipsoid in a viscous fluid given by Lamb [1993]:

Fts ! "6"#Rhv$

$3%

with m the average upper mantle viscosity (1 ! 1021 N m), vis the velocity of the slab with respect to the mantle (which isnot the same as the trench migration rate), l is the length ofthe trench (400 km) and Rh is the effective radius of a spherewith the same viscous resistance (262 km). Rh is derivedfrom the dimensions of the ellipsoid and thus the slab. Wevary v between zero and 5 mm/yr, corresponding to a trenchsuction force 0 $ 9 ! 1011 N/m.

3.2. Coupling Stresses to the Sublithospheric Mantle3.2.1. Corner Flow[13] Three plates subduct under the Caribbean plate

(Figure 1): theNorth and SouthAmerican plates (at the Lesser

Antilles), the North American plate (at the Puerto RicoTrough) and the Cocos plate (at the Middle America Trench)[Molnar and Sykes, 1969]. At a subduction zone, the slab!parallel velocity of the descending slab induces flow in thesurrounding viscous mantle. The resulting shear stress actingon the base of the lithosphere, on the overriding side ofthe slab, is then computed following Tovish et al. [1978].The viscosity of the mantle wedge is taken to be 1 _! 1019 Nm[Forte and Mitrovica, 2001]. Choosing a viscosity isimportant, since it is linearly related to the amount of shearstress acting on the overriding plate. However, varying itbetween reasonable limits does not alter the torque balancesignificantly. In the calculation we also need informationabout the RPM at the trench (11, 12 and 90 mm/yr) and theaverage slab dip (50°, 60° and 50°), for the Puerto Rico,Lesser Antilles and Cocos subduction zones, respectively.This is a simplification in that both the Cocos and the LesserAntilles subduction zones are characterized by different slabdips along strike [Wadge and Shepherd, 1984; Funk et al.,2009]. However, this simplification does not affect the totaltorque due to corner flow, because these slabs are relativelynarrow. The total torque due to corner flow is small in com-parison to other torques in this study.3.2.2. Basal Shear Stress[14] The basal shear stress (sbs) acts on the base of the

lithosphere. It results from differential motion of the litho-sphere and the underlying mantle, so it can be derived fromabsolute plate motion, mantle convection, or both. Absoluteplate motion of the Caribbean plate is almost zero [Mülleret al., 1999]. Therefore the basal shear stress is expected tobe in the direction of the mantle flow. Russo et al. [1996]suggest, from east!west oriented SKS!shear wave splittingresults, that mantle flow under the Caribbean is caused byreturn flow around the Nazca slab and is predominantlyuniformly distributed and eastward directed. AlthoughPiñero!Feliciangeli and Kendall [2008] found some differentorientations for SKS!shear wave splitting results, the pre-dominantly east!west direction has been confirmed by theSKS!shear wave splitting measurements of Growdon et al.[2009]. Therefore we choose the basal shear stress underthe Caribbean to be uniformly distributed and directed eitherwest or east:

sbs ! %bsvsks $4%

with vsks the unit vector of the direction of shear wave split-ting observations.

3.3. Resistive Forces3.3.1. Resistive Forces on the Maracaibo Slab[15] A shear stress (ssr) results from the downdipmotion of

a slab relative to the surrounding mantle. The shear resistanceforce Fsr is the shear stress integrated over the contact surface(upper and lower surface of the slab) and acts into the relativemotion direction of the slab and the surrounding mantle.Assuming a stagnant mantle and trench, the shear stresses actinto the direction opposite to APM (vapm):

ssr ! "%srvapm $5%


4 of 14

[16] Here we assume that APM is negligible so that shearresistance acts in the direction of relative motion betweensubducting and overriding plates

ssr ! %srvrpm $6%

In the slab pull calculation, the petrological differencebetween oceanic crust and mantle is not taken into account,hence we need to apply a correction. The compositionalbuoyancy force (Fcb) corrects for the subduction of the morebuoyant crust. This force is varied between 3 ! 1012 N/m and6 ! 1012 N/m [England and Wortel, 1980; van den Beukel,1990]. It acts into the updip direction of the slab.[17] The plate contact resistance force (Fpcr) is the integral

of shear stresses (spcr) along the plate contact between theoverriding and subducting plates, and is directed into thedirection of the RPM. We assume that the plate contact zoneis a subduction channel [Abers et al., 2003; De Franco et al.,2008] with a Newtonian viscosity so that the applied spcrdepends linearly on the relative velocity:

spcr & vrpm $7%

3.3.2. Other Friction Forces[18] We expect the mechanical coupling between overrid-

ing and subducting plate to be higher if a buoyant oceanicridge arrives at the trench. Several of such ridges are sub-ducting or indenting the Caribbean plate: The Tiburón,Barracuda, Normal ridges and Bahamas platform in thenortheast and the Cocos ridge in the southwest (Figure 1).We model the increased friction by increasing the shearstress magnitude by a factor R. R is varied between 1 (Findentis equal to Fpcr) and 10 (Findent is 10 times larger then Fpcr).For R > 10 force magnitudes remain similar.

sind ! Rspcr $8%

[19] Transform friction is modeled as acting into thedirection of the RPM:

stf ! %tf vrpm $9%

With stf being the average shear stress acting on the fault.The northern and southern plate boundaries of the Caribbeanplate are predominantly strike!slip, and here transform fric-tion acts on the Caribbean plate.

3.4. Torque Balance[20] The Caribbean plate is in mechanical equilibrium, i.e.,

the sum of all torques is zero [Forsyth and Uyeda, 1975]:ZZ

Arp

r' srpda#Z

Ltr

Z

Lslab

r' sspdxdl # jFcbjZ

Lcb

r' Fcbdl

#Z

Lts

r' Ftsdl #Z

Acf

r' scfda# %tf

Z

Arf

r' stfda# %pcr

(Z

Apcr

r' spcrda# R%pcr

Z

Aindent

r' spcrda# %bs

Z

Abs

r' sbsda

# %sr

Z

Asr

r' ssrda ! 0 $10%

where r is the position vector of the force (taken from thecenter of the earth). Lslab is the downdip length of the slab,which is constrained by seismic tomography and Ltr is thelength of the trench. Lcb is the downdip length of the intervalbetween the surface and the depth of the basalt!eclogitetransition (150 km). Arp is the area of the Cayman trough. Ltsis the trench perpendicular width over which trench suctionacts on the Caribbean plate (200 km). Acf is the surface areawhere the plate contact force acts (200 km). Atf is the sur-face area of transform faults (is equal to transform faultlength on map times lithosphere thickness of 100 km). Apcris the surface area of the subduction plate contact, except forindenters (represented by Aindent) (is equal to trench lengthtimes lithospheric thickness over the sine of the fault dip(30°). Abs is the surface area of the Caribbean plate. Asr isthe total surface area of the subducted slab. |Fcb|, stf, spcr, sbs,ssr and R are scalar unknowns. We systematically vary thevalues for R, |Fcb| and sbs and solve (14) for the magnitudeand sign of spcr, sbs and ssr.

4. Torque Balance Results

[21] We systematically varied the model parameters withinaccepted limits (as outlined above). We coded our solutionsas follows: (1) LOTSUC, “a lot of trench suction”; i.e., trenchsuction at the southern Lesser Antilles Trench, modeled by anoutward directed suction force of 9 ! 1011 N/m, (2) TSUC,“trench suction”; trench suction at the southern LesserAntilles Trench, modeled by an outward directed suctionforce of 45 ! 1010 N/m, (3) NOTSUC, “no trench suction”;no trench suction at the southern Lesser Antilles Trench, and(4) the number stands for R.[22] For example, TSUC5 stands for a model where R = 5

and the trench suction force amounts to 45 ! 1010 N/m.Figure 3 and Figure S1 in the auxiliary material show 12representative torque balance results.1 The restriction thatfrictional forces cannot be driving the Caribbean platereduces the range of possible solutions (gray areas inFigures 3 and S1). Figures 3 and S1 show that the solutionsshare similar characteristics;[23] 1. The solutions center about low magnitudes for

basal shear stress, which can be either eastward or westward.(!0.4 MPa)[24] 2. Pull by the Maracaibo slab is nearly canceled by

resistive forces (Fcb, Fsr, Fpcr).[25] 3. The plate contact resistance and transform fault

resistance have small magnitudes.[26] The basal shear stress is found to be at most 0.4 MPa.

The compositional buoyancy is found to range between 3 !1012 and 6 ! 1012 N/m. This corresponds to subductingcrustal thickness (assuming a constant thickness of thedepleted upper mantle) of 5 to 13 km [Oxburgh andParmentier, 1977]. This is consistent with subduction ofrelatively thick Caribbean crust; i.e., > 8 km [Mauffret andLeroy, 1997]. The average shear stress acting at this trenchis about 12 MPa. Average shear stresses due to frictionalforces at the Cocos trench are 5 MPa, at the Lesser Antillestrench they are around 2 MPa, and at transform boundariesthey are 8 MPa. Plate boundaries of the Caribbean plate are

1Auxiliary materials are available in the HTML. doi:10.1029/2009JB006950.


5 of 14

thus weak, in agreement with the studies of Lyon!Caen et al.[2006], Álvarez!Gómez et al. [2008], Manaker et al. [2008]and Correa!Mora et al. [2009].[27] Since all models share these features, we consider

them to be robust.

5. Stress, Rotation, and Fault Slip

[28] We use finite element package GTECTON [Goversand Meijer, 2001] to compute stress, vertical axis rotationand slip on regional faults from the force sets that we obtainedin the previous section. Below, we first discuss the modelsetup, and the method that we use to recognize similarnumerical model results. We then present representativemodels, and will discuss their implications in terms of stress,vertical axis rotations and fault slip.

5.1. Model Setup5.1.1. Domain, Equations, and Material Properties[29] The finite element domain is identical to the Caribbean

plate domain that we used to compute the torques (Figure 7a).Element size was selected on the basis of convergence testswhere we demonstrated the independence of the results tofurther grid refinement. We solve the 2D plane stressmechanical equilibrium equations using a uniform elasticspherical shell (Young’s modulus 70 GPa, Poison’s ratio0.25) for a reference lithospheric thickness of 100 km. Ourmotivation for this simple rheology is that elastic stressesshow the potential for permanent deformation. Incorporatinga more sophisticated rheology also implies the introductionof more uncertain parameters. Moreover, if the lithosphericrheology is isotropic, stresses and permanent strains areparallel, which is relevant for comparison with observations.The exact type of plate boundary between Central Americaand the North Andes block is not known, and we apply a plateboundary!perpendicular movement restriction here instead,

since both Central America and the Northern Andes consistof thick continental lithosphere, what could inhibit the con-vergence between the two blocks [Wadge and Burke, 1983].5.1.2. Regional Faults/Shear Zones in the CaribbeanPlate[30] Intraplate faults are incorporated in the finite element

models using a slippery node technique [Melosh andWilliams,1989] to allow fault parallel slip if the shear stress exceedsfault friction. Transform faults in the model are idealized(infinitely thin, smoothed and continuous) representations offinite width, anastomosing fault fragments that constitute realfault zones. In these natural fault zones, stress accumulation attips of finite length fault fragments andmisalignment of faultswill result in resistance to slip that is higher than the frictionalshear stress on individual faults. This resistance would mostlycome from (elastic) deformation of intact rocks that surroundthe fault fragments, and is expected to increase with theamount of slip on the shear zone. As a simple first orderapproximation of this behavior, we take the frictional shearstress (sxy) on model faults to be proportional (kfr) to shearstrain ("xy):

sxy ! kfrexy 0 ) kfr ) 2G sxy !kfrW

Ds $11%

where G is the elastic shear modulus, W is the shear zonewidth, and Ds is the slip on the shear zone. kfr is expected todecrease with the total amount of slip in natural shear zonesystems, but we ignore this in the present study and assume auniform proportionality constant for all fault zones. Thrustfaults are incorporated by slippery nodes with both a strike!parallel slip component and a strike!normal slip component.Strike!normal slip requires work against gravity. To esti-mate this resistance, we assume a fault geometry as shownin Figure 4. Horizontal convergence x of the two fault blocksis transferred in vertical displacement by an amount that is

Figure 3. Torque balance solutions for models (a) NOTSUC5, (b) TSUC5 and (c) LOTSUC5. The verticalaxis shows resistive force magnitudes (per meter horizontal plate boundary length) from the torque balancesolution (transform fault resistance Ftf, subduction plate contact resistance Fpcr, and shear resistance alongthe slab Fsr). The horizontal axis shows the basal shear stress magnitude. Lines denote the solutions for theforces denoted above, as function of the basal shear magnitude.We plot this for different values of Fcb. Phys-ically realistic solutions (no driving resistive forces, i.e., their magnitude should be less than zero) are shownin gray; unrealistic solutions are transparent. The dots denote the values of Fdr and Fcb used to calculate theelastic deformation field.


6 of 14

controlled by the fault dip angle and by Airy isostasy. Theresulting resistive force to convergence is then given by:

F ! g tan &$ %H!cr!m " !cr

!m

! "x ! kupliftx $12%

with g gravitational acceleration (9.8m/s2),H the thickness ofthe thrust slab (5 km) and a the fault dip angle (30°). We addthis extra resistance to the frictional shear stress, discussedabove.5.1.3. Boundary Conditions[31] Appropriate values for the varied forces, Fcb and Fbs,

need to be chosen. Subducting Colombian basin crust has athickness of %8 km at the trench [Mauffret and Leroy, 1997],corresponding to a Fcb of 4 ! 1012 N/m [England and Wortel,1980]. This is the best estimate we can have for the thick-ness of already subducted Caribbean crust and we thereforegave preference to this value. Possible values for sbs rangebetween %$0.4 MPa to 0.4 MPa (Figures 3 and S1). Wechoose a value in between these two extremes. We alsoexamined other choices, but this doesn’t change the modelssignificantly (details below). The dots in Figures 3 and S1show the values of Fcb and Fbs that we used. Figure 7ashows the resulting nodal point forces applied to the gridfor model TSUC5.[32] The solution to the partial differential equations is

fully determined by the above choices for domain, rheologyand boundary conditions, except for a net translation and/orrotation. We therefore specify additional displacement bound-ary conditions at nodes where they evoke the smallestdeformation gradient. Furthermore we dampen the effect ofthis pinning, by applying proportional!to!displacement for-ces at all nodes inside the stable interior of the plate.

5.2. Grouping Models[33] To recognize whether two models, k and model l are

significantly different, we compute average differencesbetween their deformation gradient fields, defined by

DF$kl%ij * 1

NUMEL

XNUMEL

n!1

@ui@xj

! "$k%

n

" @ui@xj

! "$l%

n

%%%%%

%%%%% $13%

NUMEL is the number of elements. The indices i and jdenote mutual perpendicular direction. For each combination

of models k and l we compute the (scalar) norm of the 2 ! 2tensor DF(kl) as the sum of its invariants:

C$kl% *X2

i!1

DF$kl%ii

& '2# DF$kl%

11 DF$kl%22 "DF$kl%

21 DF$kl%12

& '$14%

Identical models haveC(kl) = 0 and by definitionC(kk) = 0.Wethus obtain “distances” between all possible model combi-nations. This allows us to make a two!dimensional spatialplot of the models (Figure 5). Applying the 2D Kolmogorov!Smirnov (KS) test [Peacock, 1983], we can estimate theprobability (P) that differences between two distributionsare by chance alone.[34] Dividing the models into LOTSUC, NOTSUC and

TSUC groups and applying the 2D KS test gives P!values <0.001. Literally this means that it is highly unlikely that dif-ferences between the three groups are coincidental, whichsuggests that the three groups are different from each other.Similarly, dividing the models on the base of the value for kfrgives low P!values, especially for the lower values of kfr. Adivision based on the collision factor R gives a higher P!valueof 0.173, suggesting that variation of R does not producesignificantly different models. The likely reason for this isthat indenters evoke local stress changes only due to thepresence of faults. A division based on basal drag (Fbs) oron compositional buoyancy (Fcb) gives both high and lowP!values, and we therefore conclude that neither force con-vincingly divides the models into subgroups. In summary:the fault friction kfr and trench suction Fts are the param-eters that produce significant different deformation models.

5.3. Results5.3.1. Quantitative Comparison With Observations[35] We computed the average difference in direction of the

principal axes (obtained from focal mechanisms and boreholebreakouts) with the World Stress map (WSM) [Heidbachet al., 2008]:

D!& ! 1nobs

Xnobs

1

&wsm " &modj j:

nobs is the number of observations, awsm and amod are thedirections of maximum compression from the WSM and ourmodels, respectively. We also calculated slip directions onexisting fault planes of earthquakes (taken from the CentroidMoment Tensor (CMT) catalog and computed the averagedifference with the original direction of slip: D!' [Meijer,1995; Álvarez!Gómez et al., 2008].5.3.1.1. Fault Friction[36] Figure 6 shows the fit with the WSM and focal

mechanisms as function of models. A first observation is thatfit with the WSM and focal mechanisms varies strongly withkfr . High friction and thus high shear stresses on the faultplane result in less fault slip, and thus less elastic deformationof irregularities along the fault. This implies a decrease ofcontinuum stresses along the fault. For low kfr!values how-ever shear stresses along the fault are low, slip is high andcontinuum deformation along the fault is high. This explainsthe dependency of the deformation field on the kfr!value. Itcan also be seen in Figure 5a, where kfr is the main parameterproducing different deformation fields. For each value of kfr

Figure 4. Geometry of modeled faults. B is a column oflithosphere at the fault, where thickening occurred, depend-ing on amount of convergence x and dip a. The thickenedlithosphere is isostatically compensated. Column A is regularlithosphere.


7 of 14

we summed the misfit over the models, and thus determinedwhich value for kfr gives the lowest total misfit. This value(7 GPa) is shown by the black vertical line in Figure 6. Thegray area around it denotes these values of kfr with a misfitof less then one standard deviation away from the lowestmisfit. These kfr!values are much higher then kuplift, whichis then negligible. In the remainder of this manuscript wewill use the value of 7 GPa. Seven GPa corresponds to ashear stress of 7 ! 104 Pa per meter displacement at thefault. Fault displacements are in the order of kilometers, socorresponding shear stresses on the faults are in the orderof tens of MPa (Figure 7b).5.3.1.2. Fit as Function of Model Parameters[37] A second observation from Figure 6 is that the fit

with WSM and focal mechanisms is independent of theother parameters plotted (R, Fts). Taking kfr as 7 GPa, D!& isaround 20°, within the error bounds of the WSM data whileD!' lies around 40°. An earthquake slip direction difference

of 40° corresponds to a deviation of %15–20° of maximumstress direction. This is labeled as quality ‘A’ or ‘B’!data inthe WSM [Zoback et al., 1989]. Because, apart from kfr,model parameters do not produce a better fit (Figure 6), it isnot possible to decide between the parameter and find apreferred model. Instead we show the most representativestress model. To find this most representative model, all thedistances from each particular model k to another model l areadded:

Ck !XN

l!1

Ckl

where N is the total number of models. Ck indicates howsimilar a particular model is compared to other models. Wefind that TSUC5 is the most representative model. Figure 7cshows the effective stress (%E *

((((((((((((12%

0ij%

0ij

q) as color contours,

principal stress direction arrows, vertical axis rotations andfault slip for model TSUC5.

Figure 5. Distance between the models. (a) Colors of the circles denote the value of kfr. (b) Symbolsdenote the values of R; symbol color denotes the magnitude of Fts. Axes are mm/m.


8 of 14

[38] Although total fit with observations is constant,regional changes in fit may occur. We showed that there arethree significantly different groups, based on the amount ofapplied trench suction. We will now explore the changesin the stress field caused by a varying trench pull. Figure 8shows the fit of the maximum compressional axis for mod-els TSUC5, LOTSUC5 and NOTSUC5 with the WSM data.The models with trench suction, LOTSUC5 and TSUC5,have an almost equal fit with the WSM everywhere. Thisimplies that we cannot make conclusions about themagnitudeof the trench suction. However, the fit near the SouthernLesser Antilles is very different for model NOTSUC5. Themaximum compression axes are oriented more trench per-pendicular, when compared with the LOTSUC5 and TSUC5.This leads to a better fit along the southern PB, i.e., theEl Pilar!San Sebastian fault, where there is a strike slip regimewith maximum compressional axes directed northwest!southeast [Audemard et al., 2005]. In contrast TSUC5 andLOTSUC5 have a better fit along the Lesser Antilles islandarc, where maximum compression axes are directed almostnorth!south. The inability of our models to produce the cor-rect stress field everywhere is probably due to the choicefor an elastic rheology. Incorporating a viscous rheologywould localize the effect of the trench pull.5.3.2. Model Stresses[39] Stresses for model TSUC5 are shown in Figure 7c.

Lithosphere averaged stress magnitudes are < 40 MPa. Themodel is characterized by low stresses in the east, and higherstresses in the southwest and along faults in the north. In thenorth, modeled stresses are mainly (north)east!(south)westcompressive, while in the south near the Colombian sub-duction zone they are north!south tensional. Along trans-form fault borders we find a strike!slip regime, where theCaribbean plate is overriding another plate the regime ismainly compressive and compressional axes are trench per-pendicular (apart from the Southern Lesser Antilles). North!south compressive stresses near the Cayman trough resultfrom ridge push, which is locally oriented perpendicular to itssouthern and northern boundaries due to the high differencein age between the trough and the adjacent lithosphere. Faultirregularities and restraining bends result in (local) increasesof stress magnitudes, e.g., southwest of the Cayman trough,at the Jamaica restraining bend and east of the Maracaibosubduction zone.Stresses near indenters, most notably theCocos and Bahamas indenters are higher. At the Bahamasmost of the higher stresses are accommodated within thesmall fore!arc sliver, bounded by the septrentional fault andthe PB itself. At the Cocos indenter the higher stresses agreewith the higher strain rates predicted by the world strain ratemap from Kreemer et al. [2003], which is calculated fromGPSmeasurements and seismic moment release. The stresseshave a radial pattern around the indenting Cocos ridge, asis also seen in the regional stress compilation of Kolarskyet al. [1995].[40] In the northeast we predict a strike slip regime with

maximum compressive axes approximately NW!SE, as alsoHuérfano et al. [2005] find from examining shallow earth-quakes in southwest Puerto Rico.5.3.3. Model Rotations[41] The bottom left inset of Figure 7c shows vertical axis

rotations. The modeled rotations are elastic rotations thatshow the potential for geological rotation to occur. The actual

Figure 6. Misfit of the models with (a) the WSMmaximumcompression directions and (b) Slip direction of focalmechanisms, as function of kfr. The black line denotes thevalue for kfr that gives the best fit (averaged over the models)with theWSM. The gray area around it gives the one standarddeviation boundaries. The black dot shows the value andmodel used for constructing Figure 7a.


9 of 14

Figure 7


10 of 14

geological rotations will occur in regions with a weakerrheology or bounded by faults. Our models have counter-clockwise (CCW) rotations in the northeastern blocks. Thesoutheast is characterized by low and clockwise (CW) rota-tions. The western boundary is characterized by differentopposing rotation directions, due to the Cocos indenter. TheColombia basin shows clockwise rotation, as opposed to theneighboring Venezuela basin. Puerto Rico experienced largeCCW rotations in the Cenozoic [Reid et al., 1991] and this

therefore agrees with the CCW rotation potential that weshow for the northeast. The rotations in the northeast alsoagree with rotations predicted by the deformation model ofKreemer et al. [2003] (bottom right inset in Figure 1). Ourpredicted rotation pattern shows in!plane bending of theCaribbean plate around the Beata ridge region. Fault slipmotions [Heubeck and Mann, 1991] and seismic data[Mauffret and Leroy, 1999] suggest a similar rotation pattern,although Driscoll and Diebold [1998] and DeMets et al.

Figure 7. (a) Boundary forces acting on the Caribbean finite element domain for model TSUC5. Magnitudes correspond tothe red dots in Figure 3b, whereas directions are discussed in the text. The different colors of the plate boundary denotedifferent force types (shown in the legend). (b) Shear stresses acting on the fault surfaces of intraplate faults in the Caribbean,for model TSUC5, with kfr = 7 ! 109 kg/m s2 (corresponding to the dot in Figure 5). For the abbreviations of tectonic features,see Figure 1. (c) Principal stress directions and color contours of the effective stress (second invariant) of the reference modelTSUC5. Insets show predicted fault slip and vertical axis rotations.

Figure 8. Misfit of models (top) LOTSUC5, (middle) TSUC5 and (bottom) NOTSUC5 with the WSM.Bars show the predicted directions of the maximum compressive axes. The colors of the bars show themisfitwith the WSM: Red (dish) denotes negative (CCW) and blue colors positive (CW) rotations of predictionswith respect to theWSM. Abbreviations are S!A, Southern Lesser Antilles; PMSS, Pilar!Morón Strike!Slipfault system.


11 of 14

[2000] show that there is no present differential plate motionat the Beata ridge. However, the present force configurationand geometry has been more less stable from the Eocene andonwards, and thus the Beata ridge could have resulted ofpaleo!bending of the Caribbean plate.5.3.4. Model Fault Slip[42] For most faults the models succeed in reproducing the

observed fault slip. At the Mona rift however, the LOTSUCand TSUC models produce strike slip, with a minor compo-nent of divergence, instead of the observed divergence[Jansma and Mattioli, 2005]. The NOTSUC!models mainlyhave a diverging Mona rift. Fault slip on the modeled fault,separating the Lesser Antilles block from the Caribbeaninterior is small, agreeing with GPS observations [Manakeret al., 2008].[43] Models with a normal/ridge subduction friction ratio

R < 8 fail to reproduce the correct fault slip at the CentralAmerica Fore!Arc Strike!slip System (CAFASS), but showinstead dextral motion in the South (Nicaragua) and sinistralmotion in the North (San Salvador). This could be a conse-quence of the modeled geometry of the fault, however theexact geometry of the transition from the Nicaraguan strike!slip system to the San Salvador strike slip fault is not verywell known. Models with R > 8 reproduce the correct faultslip. Thus, adding a relatively large indenter force reproducesthe observed fault motions fault motions.5.3.5. Comparison With Earlier Studies[44] Negredo et al. [2004] did a kinematics!based model-

ing study of the Caribbean plate. They found high coupling ofthe Caribbean plate with the underlyingmantle, agreeing withthe global model of Bird et al. [2008]. Such a high couplingwould be expected for a plate with a lithospheric root [Bird,1998; Conrad and Lithgow!Bertelloni, 2006], but not forthe Caribbean plate. This discrepancy can partly be explainedby the fact that these models do not account for regionalforces acting on the Caribbean plate. Our study accounts forthese regional forces, and predicts low shear stresses on thebase of the Caribbean. This agrees with earlier studies ofoceanic plates [Cloetingh and Wortel, 1985; Govers andMeijer, 2001; Conrad and Lithgow!Bertelloni, 2006] thatalso predict low coupling. Most studies predicting lowmantle!plate coupling are dynamical models, and their out-come might depend on the assumed force sets [Bird, 1998].However, our sensitivity studies show that for the case of theCaribbean plate all applied force sets produce similar resultsand that the low coupling is a robust feature. The highdependency of the deformation field on the intra plate faultfriction was also found by Negredo et al. [2004], who foundthat fault friction is low. We find values in the order of tensof megapascals for the fault friction.

6. Conclusions

[45] We evaluated upper and lower limits of driving andresisting forces acting on the Caribbean plate. We use adynamical approach and we want to determine whethermantle drag or local forces such as interaction with sur-rounding plates are deforming the Caribbean plate. We foundthat important forces contributing to the dynamics of theCaribbean plate are (1) pull by the Maracaibo slab which islargely counteracted by slab resistive forces, (2) Frictionalforces with the surrounding plates, perhaps combined with

(3) trench suction at the southern Lesser Antilles. After cal-culating the deformation field we were able to divide ourmodels into different groups. Parameters that control thedeformation field are the trench pull and the strength ofintraplate faults. For a intraplate fault strength of 7GPa,models have an average misfit of 18–22° with the WorldStress Map. An extra push by the indenting Cocos ridge (andsmaller ridges in the northeast) reproduces observed faultmotions better, but does not affect the fit with the (local)stress field. A consistent feature of all models is the very lowvalue for the basal shear stress (0–0.4 MPa). We concludetherefore, that the influence of asthenospheric flow on thedynamics of the Caribbean plate is small or absent.

[46] Acknowledgments. We wish to thank Corné Kreemer for pro-viding the vertical axis rotations (Figure 1), Paul Meijer for his help withthe torque balance and calculation of earthquake slip, and Rinus Wortel forhis help with the torque balance and reviewing the manuscript. Thoughtfulreviews by two anonymous reviewers and the Associate Editor greatlyimproved the manuscript.

ReferencesAbers, G. A., T. Plank, and B. R. Hacker (2003), The wet Nicaragua slab,Geophys. Res. Lett., 30(2), 1098, doi:10.1029/2002GL015649.

Aitken, T., P. Mann, A. Escalona, and G. L. Christeson (2009), Evolutionof the Grenada and Tobago basins and implications for arc migration,Mar. Pet. Geol., doi:10.1016/j.marpetgeo.2009.10.003, in press.

Alvarez, W. (1982), Geological evidence for the geographical patternof mantle return flow and the driving mechanism of plate tectonics,J. Geophys. Res., 87, 6697–6710, doi:10.1029/JB087iB08p06697.

Álvarez!Gómez, J. A., P. T. Meijer, J. J. Martínez!Díaz, and R. Capote(2008), Constraints from finite element modeling on the active tectonicsof northern Central America and the Middle America Trench, Tectonics,27, TC1008, doi:10.1029/2007TC002162.

Audemard, F. A., G. Romero, H. Rendon, and V. Cano (2005), Quaternaryfault kinematics and stress tensors along the southern Caribbean fromfault!slip data and focal mechanism solutions, Earth Sci. Rev., 69(3–4),181–233, doi:10.1016/j.earscirev.2004.08.001.

Barba, S., M. M. C. Carafa, and E. Boschi (2008), Experimental evidencefor mantle drag in the Mediterranean, Geophys. Res. Lett., 35, L06302,doi:10.1029/2008GL033281.

Bird, D. E., S. A. Hall, J. F. Casey, and P. S. Millegan (1993), Inter-pretation of magnetic anomalies over the Grenada Basin, Tectonics, 12,1267–1279, doi:10.1029/93TC01185.

Bird, P. (1998), Testing hypotheses on plate!driving mechanisms withglobal lithosphere models including topography, thermal structure, andfaults, J. Geophys. Res., 103(B5), 10,115–10,129, doi:10.1029/98JB00198.

Bird, P. (2003), An updated digital model of plate boundaries, Geochem.Geophys. Geosyst., 4(3), 1027, doi:10.1029/2001GC000252.

Bird, P., Z. Liu, and W. K. Rucker (2008), Stresses that drive the platesfrom below: Definitions, computational path, model optimization,and error analysis, J. Geophys. Res., 113, B11406, doi:10.1029/2007JB005460.

Byrne, D. B., G. Suarez, and W. R. McCann (1985), Muertos Troughsubduction—Microplate tectonics in the northern Caribbean?, Nature,317, 420–421, doi:10.1038/317420a0.

Clark, S. A., A. Levander, M. B. Magnani, and C. A. Zelt (2008), Negligibleconvergence and lithospheric tearing along the Caribbean!South Ameri-can plate boundary at 64°W, Tectonics, 27, TC6013, doi:10.1029/2008TC002328.

Cloetingh, S., and R. Wortel (1985), Regional stress field of the Indianplate, Geophys. Res. Lett., 12, 77–80, doi:10.1029/GL012i002p00077.

Conrad, C. P., and C. Lithgow!Bertelloni (2002), How mantle slabs driveplate tectonics, Science, 298, 207–209, doi:10.1126/science.1074161.

Conrad, C. P., and C. Lithgow!Bertelloni (2006), The influence of conti-nental roots and asthenosphere on plate!mantle coupling, Geophys.Res. Lett., 33, L05312, doi:10.1029/2005GL025621.

Correa!Mora, F., C. DeMets, D. Alvarado, H. L. Turner, G. Mattioli,D. Hernandez, C. Pullinger, M. Rodriguez, and C. Tenorio (2009),GPS!derived coupling estimates for the Central America subduction zoneand volcanic arc faults: El Salvador, Honduras and Nicaragua, Geophys.J. Int., 179(3), 1279–1291, doi:10.1111/j.1365!246X.2009.04371.x.


12 of 14

De Franco, R., R. Govers, and R. Wortel (2008), Nature of the plate contactand subduction zone diversity, Earth Planet. Sci. Lett., 271(1–4), 245–253, doi:10.1016/j.epsl.2008.04.019.

DeMets, C. (2001), A new estimate for present!day Cocos!Caribbean platemotion: Implications for slip along the Central American volcanic arc,Geophys. Res. Lett., 28, 4043–4046, doi:10.1029/2001GL013518.

DeMets, C., and M. D. Wiggins!Grandison (2007), Deformation ofJamaica and motion of the Gonave microplate from GPS and seis-mic data, Geophys. J. Int., 168(1), 362–378, doi:10.1111/j.1365!246X.2006.03236.x.

DeMets, C., P. Jansma, G.Mattioli, T. Dixon, F. Farina, R. Bilham, E. Calais,and P. Mann (2000), GPS geodetic constraints on Caribbean!North Amer-ican plate motion, Geophys. Res. Lett., 27, 437–440, doi:10.1029/1999GL005436.

Driscoll, N. W., and J. B. Diebold (1998), Deformation of the Caribbeanregion: One plate or two?, Geology, 26(11), 1043–1046, doi:10.1130/0091!7613(1998)026<1043:DOTCRO>2.3.CO;2.

England, P., and R. Wortel (1980), Some consequences of the subductionof young slabs, Earth Planet. Sci. Lett., 47, 403–415, doi:10.1016/0012!821X(80)90028!X.

Forsyth, D., and S. Uyeda (1975), On the relative importance of the drivingforces of plate motion, Geophys. J. R. Astron. Soc., 43, 163–200.

Forte, A. M., and J. X. Mitrovica (2001), High viscosity deep mantle flowand thermochemical structure inferred from seismic and geodynamicdata, Nature, 410, 1049–1056, doi:10.1038/35074000.

French, C. D., and C. J. Schenk (2004), Map showing geology, oil and gasfields, and geologic provinces of the Caribbean region, U.S. Geol. Surv.Open File Rep., 97!470!K.

Funk, J., P. Mann, K. McIntosh, and J. Stephens (2009), Cenozoic tectonicsof the Nicaraguan depression, Nicaragua, and Median Trough, El Salvador,based on seismic!reflection profiling and remote!sensing data, Geol. Soc.Am. Bull., 121(11–12), 1491–1521, doi:10.1130/B26428.1.

Govers, R., and P. T. Meijer (2001), On the dynamics of the Juan de Fucaplate, Earth Planet. Sci. Lett., 189, 115–131, doi:10.1016/S0012!821X(01)00360!0.

Grindlay, N. R., P. Paul Mann, J. F. Dolan, and J.!P. van Gestel (2005),Neotectonics and subsidence of the northern Puerto Rico!Virgin Islandsmargin in response to the oblique subduction of high!standing ridges, inActive Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands,and Offshore Areas, edited by P. Mann, Spec. Pap. Geol. Soc. Am., 385,31–60.

Growdon, M. A., G. L. Pavlis, F. Niu, F. L. Vernon, and H. Rendon (2009),Constraints on mantle flow at the Caribbean!South American plateboundary inferred from shear wave splitting, J. Geophys. Res., 114,B02303, doi:10.1029/2008JB005887.

Heidbach, O., M. Tingay, A. Barth, J. Reinecker, D. Kurfeß, and B. Müller(2008), The 2008 release of the World Stress Map, http://www.world!stress!map.org, Helmholtz Centre Potsdam, Potsdam, Germany.

Heubeck, C., and P. Mann (1991), Geologic evaluation of plate kinematicmodels for the northern Caribbean plate boundary zone, Tectonophysics,191, 1–26, doi:10.1016/0040!1951(91)90230!P.

Hoernle, K., F. Hauff, and P. van de Bogaard (2004), A 70 Myr history(139–69 Ma) for the Caribbean large igneous province, Geology, 32,697–700, doi:10.1130/G20574.1.

Huérfano, H., C. von Hillebrandt!Andrade, and G. Báez!Sanchez (2005),Microseismic activity reveals two stress regimes in southwestern PuertoRico, in Active Tectonics and Seismic Hazards of Puerto Rico, the VirginIslands, and Offshore Areas, edited by P. Mann, Spec. Pap. Geol. Soc.Am., 385, 81–101.

Jansma, P., A. Lopez, G. Mattioli, C. DeMets, T. Dixon, P. Mann, andE. Calais (2000), Microplate tectonics in the northeastern Caribbean asconstrained by Global Positioning (GPS) geodesy, Tectonics, 19, 1021–1037, doi:10.1029/1999TC001170.

Jansma, P. E., and G. S. Mattioli (2005), GPS results from Puerto Rico andthe Virgin Islands: Constraints on tectonic setting and rates of activefaulting, in Active Tectonics and Seismic Hazards of Puerto Rico, theVirgin Islands, and Offshore Areas, edited by P. Mann, Spec. Pap. Geol.Soc. Am., 385, 13–30.

Kearey, P. (1974), Gravity and seismic reflection investigations intothe crustal structure of the Aves Ridge, eastern Caribbean, Geophys.J. R. Astron. Soc., 38, 435–448.

Kellog, J. N., and V. Vega (1995), Tectonic development of Panama, CostaRica, and the Colombian Andes: Constraints from global positioning sys-tem geodetic studies and gravity, in Geologic and Tectonic Developmentof the Caribbean Plate Boundary in Southern Central America, edited byP. Mann, Spec. Pap. Geol. Soc. Am., 295, 75–90.

Kolarsky, R. A., P. Mann, and W. Montero (1995), Island arc response toshallow subduction of the Cocos Ridge, Costa Rica, in Geologic andTectonic Development of the Caribbean Plate Boundary in Southern

Central America, edited by P. Mann, Spec. Pap. Geol. Soc. Am., 295,235–262.

Kreemer, C., W. E. Holt, and A. J. Haines (2003), An integrated globalmodel of present!day plate motions and plate boundary deformation,Geophys. J. Int., 154, 8–34, doi:10.1046/j.1365!246X.2003.01917.x.

LaFemina, P., T. H. Dixon, R. Govers, E. Norabuena, H. Turner, A. Saballos,G. Mattioli, M. Protti, and W. Strauch (2009), Fore!arc motion and CocosRidge collision in Central America, Geochem. Geophys. Geosyst., 10,Q05S14, doi:10.1029/2008GC002181.

Lamb, H. (1993), Hydrodynamics, 6th ed., 738 pp., Cambridge Univ. Press,New York.

Leroy, S., A. Mauffret, P. Patriat, and B. M. de Lepinay (2000), An alter-native interpretation of the Cayman trough evolution from a reidenti-fication of magnetic anomalies, Geophys. J. Int., 141(3), 539–557,doi:10.1046/j.1365!246x.2000.00059.x.

López, A. M., S. Stein, T. Dixon, G. Sella, E. Calais, P. Jansma, J. Weber,and P. LaFemina (2006), Is there a northern Lesser Antilles forearcblock?, Geophys. Res. Lett., 33, L07313, doi:10.1029/2005GL025293.

Lyon!Caen, H., et al. (2006), Kinematics of the North American!Caribbean!Cocos plates in Central America from new GPS measure-ments across the Polochic!Motagua fault system, Geophys. Res. Lett.,33, L19309, doi:10.1029/2006GL027694.

Magnani, M. B., C. A. Zelt, A. Levander, and M. Schmitz (2009), Crustalstructure of the South American!Caribbean plate boundary at 67°W fromcontrolled source seismic data, J. Geophys. Res., 114, B02312,doi:10.1029/2008JB005817.

Manaker, D. M., E. Calais, A. M. Freed, S. Ali, P. Przybylski, G. Mattioli,P. Jansma, C. Prépetit, and J. B. de Chabalier (2008), Interseismic platecoupling and strain partitioning in the northeastern Caribbean, Geophys.J. Int., 174(3), 889–903, doi:10.1111/j.1365!246X.2008.03819.x.

Mann, P., E. Calais, J. C. Ruegg, C. DeMets, P. E. Jansma, and G. S.Mattioli (2002), Oblique collision in the northeastern Caribbean fromGPS measurements and geological observations, Tectonics, 21(6),1057, doi:10.1029/2001TC001304.

Mauffret, A., and S. Leroy (1997), Seismic stratigraphy and structure ofthe Caribbean igneous province, Tectonophysics, 283(1–4), 61–104,doi:10.1016/S0040!1951(97)00103!0.

Mauffret, A., and S. Leroy (1999), Neogene intraplate deformation of theCaribbean Plate at the Beata Ridge, in Sedimentary Basins of the World,vol. 4, Caribbean Basins, edited by P. Mann, p. 627–669, Elsevier Sci.,Amsterdam.

McCann, W., and L. Sykes (1984), Subduction of aseismic ridges beneaththe Caribbean plate: Implications for the tectonics and seismic potentialof the northeastern Caribbean, J. Geophys. Res., 89, 4493–4519,doi:10.1029/JB089iB06p04493.

Meijer, P. T. (1995), Dynamics of active continental margins: The Andesand the Aegean region, Geol. Ultraiectina, 130, 218.

Meijer, P. T., and M. J. R. Wortel (1992), The dynamics of motion of theSouth American plate, J. Geophys. Res., 97, 11,915–11,931, doi:10.1029/91JB01123.

Meijer, P. T., and M. J. R. Wortel (1999), Cenozoic dynamics of the Africanplate with emphasis on the Africa!Eurasia convergence, J. Geophys. Res.,104, 7405–7418, doi:10.1029/1999JB900009.

Melosh, H. J., and C. A. Williams (1989), Mechanics of graben forma-tion in crustal rocks: A finite element analysis, J. Geophys. Res., 94,13,961–13,973, doi:10.1029/JB094iB10p13961.

Molnar, P., and L. R. Sykes (1969), Tectonics of the Caribbean and middleAmerica regions from focal mechanisms and seismicity, Geol. Soc. Am.Bull., 80, 1639–1684, doi:10.1130/0016!7606(1969)80[1639:TOTCAM]2.0.CO;2.

Müller, R. D., J.!Y. Royer, S. C. Cande, W. R. Roest, and S. Maschenkov(1999), New constraints on the Late Cretaceous/Tertiary plate tectonicevolution of the Caribbean, in Sedimentary Basins of the World, vol. 4,Caribbean Basins, edited by P.Mann, p. 33–59, Elsevier Sci., Amsterdam.

Müller, R. D., M. Sdrolias, C. Gaina, and W. R. Roest (2008), Age, spread-ing rates, and spreading asymmetry of the world’s ocean crust, Geochem.Geophys. Geosyst., 9, Q04006, doi:10.1029/2007GC001743.

Negredo, A. M., I. Jiménez!Munt, and A. Villaseñor (2004), Evidence foreastward mantle flow beneath the Caribbean plate from neotectonic mod-eling, Geophys. Res. Lett., 31, L06615, doi:10.1029/2003GL019315.

Oxburgh, E. R., and E. M. Parmentier (1977), Compositional and densitystratification in oceanic lithosphere causes and consequences, J. Geol.Soc., 133, 343–355, doi:10.1144/gsjgs.133.4.0343.

Peacock, J. A. (1983), Two!dimensional goodness!of!fit testing in astron-omy, Mon. Not. R. Astron. Soc., 202, 615–627.

Pindell, J. L., and S. F. Barrett (1990), Geologic evolution of the Carib-bean: A plate!tectonic perspective, in The Geology of North America,vol. H, The Caribbean Region, edited by G. Dengo and J. E. Case,pp. 405–432, Geol. Soc. of Am., Boulder, Colo.


13 of 14

Piñero!Feliciangeli, L. T., and J. M. Kendall (2008), Sub!slab mantle flowparallel to the Caribbean plate boundaries: Inferences from SKS splitting,Tectonophysics, 462(1–4), 22–34, doi:10.1016/j.tecto.2008.01.022.

Reid, J., P. Plumley, and J. Schellekens (1991), Paleomagnetic evidencefor late Miocene counterclockwise rotation of north coast carbonatesequence, Puerto Rico, Geophys. Res. Lett., 18, 565–568, doi:10.1029/91GL00401.

Richardson, R. M., S. C. Solomon, and N. H. Sleep (1979), Tectonic stressin the plates, Rev. Geophys. Space Phys., 17, 981–1019, doi:10.1029/RG017i005p00981.

Richter, F. M., and D. P. McKenzie (1978), Simple plate models of mantleconvection, J. Geophys., 83, 441–471.

Rosencrantz, E., and P. Mann (1991), SeaMARC II mapping of transformfaults in the Cayman Trough, Geology, 19, 690–693, doi:10.1130/0091!7613(1991)019<0690:SIMOTF>2.3.CO;2.

Rosencrantz, E., M. I. Ross, and J. G. Sclater (1988), Age and spreadinghistory of the Cayman Trough as determined from depth, heat flow, andmagnetic anomalies, J. Geophys. Res., 93, 2141–2157, doi:10.1029/JB093iB03p02141.

Russo, R. M., P. G. Silver, M. Franke, W. B. Ambeh, and D. E. James(1996), Shear!wave splitting in northeast Venezuela, Trinidad, and theeastern Caribbean, Phys. Earth Planet. Inter., 95(3–4), 251–275,doi:10.1016/0031!9201(95)03128!6.

Schellart, W. P., J. Freeman, D. R. Stegman, L. Moresi, and D. May(2007), Evolution and diversity of subduction zones controlled by slabwidth, Nature, 446, 308–311, doi:10.1038/nature05615.

Scholz, C. H., and J. Campos (1995), On the mechanism of seismic decou-pling and back!arc spreading at subduction zones, J. Geophys. Res., 100,22,103–22,115, doi:10.1029/95JB01869.

Steinberger, B., H. Schmeling, and G. Marquart (2001), Large!scale litho-spheric stress field and topography induced by global mantle circulation,Earth Planet. Sci. Lett., 186, 75–91, doi:10.1016/S0012!821X(01)00229!1.

Tovish, A., G. Schubert, and B. P. Luyendyk (1978), Mantle flow pressureand the angle of subduction: Non!Newtonian corner flows, J. Geophys.Res., 83(B12), 5892–5898, doi:10.1029/JB083iB12p05892.

van den Beukel, J. (1990), Thermal and mechanical modelling of conver-gent plate margins, Geol. Ultraiectina, 62, 126 pp.

van der Hilst, R., and P. Mann (1994), Tectonic implications of tomo-graphic images of subducted lithosphere beneath northwestern SouthAmerica, Geology, 22, 451–454, doi:10.1130/0091!7613(1994)022<0451:TIOTIO>2.3.CO;2.

Wadge, G., and K. Burke (1983), Neogene Caribbean plate rotation andassociated Central American tectonic evolution, Tectonics, 2, 633–643,doi:10.1029/TC002i006p00633.

Wadge, G., and J. B. Shepherd (1984), Segmentation of the Lesser!Antillessubduction zone, Earth Planet. Sci. Lett., 71, 297–304, doi:10.1016/0012!821X(84)90094!3.

Wortel, M. J. R., M. J. N. Remkes, R. Govers, S. A. P. L. Cloetingh, andP. T. Meijer (1991), Dynamics of the lithosphere and the intraplate stressfield, Philos. Trans. R. Soc. London A, 337, 111–126, doi:10.1098/rsta.1991.0110.

Zoback, M. L., et al. (1989), Global patterns of tectonic stress, Nature, 341,291–298, doi:10.1038/341291a0.

R. Govers and S. van Benthem, Faculty of Geosciences, Utrecht University,Budapestlaan 4, NL!3584 CD Utrecht, Netherlands. ([email protected])


14 of 14

the caribbean plate: pulled, pushed, or dragged? · table 1 gives a summary of all forces that we...

Documents