Semiquantitative Modeling of Strain and Kinematics Along the Caribbean&Sol;North America Strike&Hyphen;Slip Plate Bounda

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. B5, PAGES 8293-8308, MAY 10, 1993

SemiquantitativeModeling of Strain and KinematicsAlong the Caribbean/North America Strike-Slip Plate BoundaryZone

ERIC CALAIS1 AND BERNARDMERCIER DE LI•PINAY

Institut de G•odynamique,CNR$, Valbonne,France

Recent structural and geophysicalstudies conducted along the northern Caribbean plate boundary, at sea and on land, have led to a precise description of the geometry and the tectonic regimes along this major transcurrent zone which separates the Caribbean and North American plates. In order to interpret its tectonic features in terms of plate motion, we use a simple numerical model of strike-slip faulting to test previously proposed kinematic models and to compute new motion parameters. We show that none of the previously proposed models correctly accounts for the observed deformation along the whole plate boundary. On the basis of the deformation pattern obtained from geological data we compute a motion parameters set that integrate rigid plate rotation and "a plate boundary zone deformation component." Our results show that the deformation along the northern Caribbean plate boundary zone is controledby regional kinematics (i.e., the Caribbean/North America relative motion) rather than by local effects(e.g., small block rotation, intraplate deformation).

INTRODUCTION the previously proposed kinematic parameters are used as boundary conditions to compute the predicted deformation The Caribbean domain and Central America form a small along the plate boundary. Comparison of this theoretical lithosphericplate (Caribbean plate) between North and result with the observeddeformation permits evaluation of South America (Figure 1). The Caribbeanplate is moving the tested kinematic parameters. In the secondstep, the eastwardrelative to North and South America alongtwo ma- geometry of the plate boundary trace and the observedde- jor strike-slip fault zones: Boconoand E1 Pilaf faults along formation pattern are usedas boundary conditionsto calcu- its southernboundary and the Polochic-Motagua,Swan, and late motion parameters. A simple direct formulation is used Oriente faults along its northern boundary. It is bounded to to minimize the scatter between observed and theoretical the east by the Lesser Antilles subduction zone, to the west deformation patterns. This approach leads to the motion by the Central America subduction zone. Despite this ap- parameters that best fit the observed deformation. parently well-constrainedkinematic frame, the present-day Our approachallows the useof qualitative geologicaldata motion of the Caribbean plate is one of the more poorly to control and compute quantitative kinematic parameters. knownamong all major plates[DeMets et al., 1990]. It is to be noted, however,that this approachdoes not rely Due to recent field and marine investigations,the North- on the fit of absolutedeformation values (usually not ob- ern Caribbean plate boundaryis one of the world'smost de- tainable from field observations)but on the fit of an overall tailedstudied transcurrent plate boundaryzones [e.g., Mann deformationpattern at the scale of the entire plate bound- et al., 1984, 1991; De Zoeten, 1988; Mercier de Ldpinay ary. et al., 1989; Heubeck et al., 1990; Calais and Mercier de Ldpinay, 1991; Masson and Scanlon,1991; Speedand Larue, NORTHERN CARIBBEAN PLATE BOUNDARY ZONE 1991; Rosencrantzand Mann, 1991]. Its geometryand associated structuresare known well enoughthat they cannot be From west to east, the northern Caribbean plate bound- neglectedin evaluating kinematic models of the Caribbean ary consistsof two major faults, the Oriente and Swanfaults, plate motion[Heubeck and Mann, 1991]. connectedto eachother by the Mid-Cayman spreadingcen- The first step of this paper is to test the proposed kine- ter (Figure 1). The Swanfault is the eastwardmarine exten- matic modelsof the Caribbean plate [Mac Donald, 1976; sion of the Polochic-Motaguafault zone of Central America. Minster and Jordan, 1978; Sykes et al., 1982; Stein et al., The Oriente fault continues to the east on land into the 1988]. The secondstep is to computenew kinematicparam- northern Dominican Republic and mergesfarther east with eters accordingto the most recentgeological and geophysical the Puerto Rico trench. The following discussionsumma- results in the northern Caribbean. For these purposes,we rizes the main structural features of these fault segments, used a simple numerical model of strike-slip faulting that basedon the most recent investigationsalong them. These takes into account the precise trace of the main faults and features are the basic constraintson plate boundary geom- the deformationpattern along the entire plate boundary. In etry and tectonics that we used in our numerical model. the first step, the geometry of the plate boundary trace and Polochic-Motagua and Swan Faults Few geologicaldata are available concerningrecent tec- 1Nowat ScrippsInstitution of Oceanography,La Jolla, California. tonics along the Polochic-Motaguafault zone. Schwarz et al. [1976]describe Quaternary alluvial terracesaffected by Copyright 1993 by the American Geophysical Union. left-lateral displacements.They attribute the significantup- Paper number 92JB03026. lift of the mountain ranges located north of these faults to 0148-0227/93/92JB-03026505.00 active reverse and strike-slip faulting. These observations

8293 8294 CALAIS AND DE LI•,PINAY:STRAIN AND KINEMATICSIN THE CARIBBEAN

NORTH AMERICA PLATE

CARIBBEAN PLATE spreading ..... i:i:i:!ii!iiiiii!iiiiiiiiiiiiiii!i!i".... ':i:i:i:i:E:i:i:i:!:i:!:i:i:i:i:i:i:i:i:!:i-:.:;:;;:::;;;:;:

....

...... subduction •x

strike-slip Vz

crust SOUTH AMERICA PLATE

...... -i:i:!:i:!-i:!:i:!:i.i:!:i:i:!.!...... 0 500km ======...... •

Fig. 1. a. Geodynamicsetting of the Caribbeanplate; b. The northernCaribbean plate boundary in its geodynamicframe (Mex, Mexico; Ho, Honduras;LI, Lagode Isabal;SF, Swanfault; RGI, Roatanand Guajira Islands;SI, SwanIslands; CT, Caymantrough; YB, Yucatanbasin; NR, Nicaraguarise; Ja, Jamaica;CCB, Cabo Cruz basin;SDB, Santiagodeformed belt; WP, Windwardpassage; TC, Tortuechannel; CS, Cordillera Septentrionalof DominicanRepublic; MAW, Muertosacretionnary wedge; PMF, PolochicMotagua fault zone; PR, PuertoRico; PRT, PuertoRico Trench; LA, LesserAntilles; GB, Grenadabasin; VB, Venezuelabasin; BR, Beataridge; CB, Colombiabasin; Co, Columbia; Eq, Ecuador;Cu, Cuba;GM, Gulfof Mexico;OF, Orientefault zone;Hisp, Hispaniola).

and the high and prominent topographyon the northern derson,1982 ; Manton, 1987]. The Swan fault, which is side of the fault zone suggest significant active transpres- the eastward marine prolongationof the Polochic-Motagua sionin this area [Burkart,1983]. The geometryof the fault faults, is locatednorth of the Bay and Swan Islands,where it zone described here takes into account the activity of both trends N65øE and bends into a N100øE direction around lon- the Polochicand Motagua fault zones,as assumedfrom seis- gitude85ø30'W [Mann et al., 1989, 1991]. Fifty kilometers mological[Cart and $toiber,1977; Plafker, 1976; Guzmdn- to the east, the major fault bends and follows a N65øE di- $pezialeet al., 1989]and geologicMdata [Erdlacand An- rection until its connectionwith the Mid-Cayman spreading CALAIS AND D•, LI•,PINAY: STRAIN AND KINI•MATICS IN THI• CARIBBI•AN 8295

center at about longitude 81ø40'W. The present-daymotion MODELING PROCEDURE along the Swan fault, therefore, occurssouth of the Swan Is- Principle and Basic Hypothesis lands, not to the north, as formerly assumed.Multichannel seismic reflection profiles perpendicular to the Swan fault Transcurrentplate boundariesare definedas segmentsof [Rosencrantzand Mann, 1991]display spectacular compres- small circles around a rotation pole, along which the rela- sivestructures (folds and reversefaults) in an active trans- tive motionof both platesis purelystrike-slip [ Wilson,1965; pressivearea (150 km long by 20 km wide) betweenthe Bay Morgan, 1968]. Sincethis theoreticaldefinition, many stud- and Swan islands. ies, on land and at sea, have shownthat transpression(re- suiting in en •chelon folds, reverse and thrust faults, up- Oriente and Septentrional Faults lift), transtension(resulting in initiation and subsidenceof elongatedbasins), and pure strike-slipcan occur alongthe From the Mid-Cayman spreading center to the western sametranscurrent plate boundary[Sylvester, 1990]. The edge of the Cuban margin, the geometry of the Oriente distribution of these various tectonic regimesis related to fault is poorly constrained. Along the Cuban margin, how- (1) geometriceffects which can be dividedinto local (fault ever, its trace and associatedstructures have been mapped geometry,i.e., its relaysor bends)and regionaleffects (rela- in detail [Calais and Mercier de Ldpina•t,1991]. Alongthe tive motion along the main fault, i.e., the kinematics of the western Cuban margin, the Oriente fault is mainly a sys- fault); and (2) mechanicaleffects (variations of thermody- tem of sinistrally offset en •chelon fault segments,respon- namical behavior of the lithosphere with depth and along sible for the subsidenceof pull-apart basins(Cabo Cruz, the fault strike). Chivirico, and Baitiquiri basins). In contrast, the eastern In the following, we consider a plate boundary zone cut- part of the Cuban margin is characterizedby active compres- ting a lithosphereof constant physical properties along its sionMtectonic structures (folds, reversefaults and soutverg- wholestrike and thus usea lithosphereof homogeneousthe- ing thrusts) forming the Santiagodeformed belt [Calais and ology. Consequently, the model will not take mechanical Mercier de Ldpinay, 1990]. Its developmentis related to a effects into account. We thus consider that the strain dis- transpressive tectonic regime along the Oriente strike-slip tribution within the strike-slip zone is essentiallydue to the fault. In the Windward Passage, the Oriente fault trace combination of local and regional geometrical effects. The is associatedwith slightly transpressivestructures [Edgar, transcurrent plate boundary model becomesa simple "three 1989; Calais and Mercier de Ldpinagt,1993], characterized clues enigma", defined by its geometry, its kinematics and by smooth en •chelon folds. It continues to the east into the various strain regimes along it. The objective of this the Tortue Channel, as clearly imaged by recent extensive work is to link quantitatively these three "clues" in two dif- side-scansonar mapping of the WindwardPassage area (B. ferent ways (Figure 2): Mercier de Ldpina•t, personal communication,1991). Seis- 1. Assumingthat the geometryof the strike-slipfault and mic reflection profiles located at the western end of this its kinematics are known, we calculate the predicted strain channel reveal the existenceof active compressivestructures regime along each fault segment. The comparisonbetween [Edgar, 1989; Calais and Mercier de L4pinay, 1993]. The this theoretical model of deformation and field observations recent uplift of the northern Haitian coast may be related allows us to test the validity of the imposed kinematics. to a transpressive tectonic regime. Farther east, the plate 2. Assuming that the geometry and the various tec- boundary trace follows the northern Haitian coastline and tonic regimes(transtension, transpression or pure strike- continueson land, in Hispaniola, after a transtensive relay slip) along the strike-slipfault are known, we calculatethe in the western edge of the Cibao valley. Through north- set of motion parameters that best fit the observed defor- ern Hispaniola, the plate boundary trace correspondsto mation. the Septentrionalstrike-slip fault zone[Mann et al., 1984], which is responsible for the active transpressivestructures Dej'ormation Parameters of the Cordillera Septentrional of the Dominican Republic [De Zoeten,1988]. The deformationin this area is marked In order to compare observed and calculated strain, we by uplift of the Cordillera Septentrionaland intense folding make the following assumptions: and faulting associatedwith significantseismicity [Calais et 1. The deformationis finite, homogeneousand continuous al., 1992a,1992b]. (without dislocations). 2. The deformation regime along the strike-slipfault zone Northern Puerto Rico Margin is the combinationof pureshear and simpleshear (see equation (1) below). The precise geometry of the northern Caribbean plate 3. The vertical distribution of matter is conserved dur- boundary along the northern Puerto Rico Margin remains ing deformation(i.e., every motion createsdeformation that unclear. We have followed the most recent results of Mas- can be observedon the surface). We will thus considerthe son and Scanlon[1991] and Speedand Larue [1991] and deformation in an horizontal plane; i.e., we make a plane consideredthe plate boundary trace to be the "19 ø strike- strain approximation. slip fault". This assumptionis in good agreementwith the 4. The fault zone exhibits no volume changeand is lat- seismotectonicdata in this area [McGannand Sttkes,1984; erally confined(no stretchalong the zone);thus, shortening Galaiset al., 1992a],which indicate shallow strike-slip earth- (or elongation)of a givenvolume perpendicular to the tran- quakesalong this trace. Concerningthe tectonic regime in scurrentfault is compensatedby a correspondingthickening this area, all the availablestructural data showa significant (or thinning)of the hthosphere. subsidenceof the Puerto Rico trench inner wall along great Let us consider a cubic-shaped volume of lithosphere, lo- normal faults [Le Pichon et al., 1985; Heezenet al., 1985; catedalong a linearsegment of a strike-slipfault [cf. Sander- Massonand Scanlon,1991; Speed and Larue, 1991]. son and MarchJul,1984]. For the sake of simplicity of the 8296 CALAIS AND DE L•PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN

Method 1. Method 2. o = (7) Vl known I GEOMETRYI [KINEMATICS I parameters Kinematics and Deformation unknown/ predicted As shownon Figure 3b, a -• and 7 can be expressedas parameters functions of 6). For a fault segment of length F, their expression becomes test

ß Fig. 2. Simplified principles of the methods presentedin this a-•7 = v= ßFcos - O v. sin 6) (8) paper. Substitutingthe valuesof a-• and 7 (8) in the expression of the strain tensor (2), we obtain mathematical expressions,we consider its sides parallel to the X,Y, and Z axes (Figure 3a). Sincethe deformation regime is assumedto be the combinationof pure shear and simple shear, the general expressionof the strain tensor E, assumingan isovolumedeformation with no length variation Mong the X axis, is given by

A E= o o , (]) 0 0 where c•-• representsthe shorteningalong the Y axis, 7 represents the shear strain, and a represents the vertical stretchß Our third hypothesis(see above) allowsus to ex- pressthe strain tensorin the X, Y horizontalplane (Figure 3). It thus becomes

z = o ' (2) y

Kinematic Parameters

The relative motion of rigid plates on a sphere is com- monly describedby a rotation pole P (latitude ll-, longitude Ll-) and an instantaneousangular velocity w (Figure 3..c).This motion can also be describedby a velocityfield V(l•,, L•,, w) [Minster and Jordan,1984]. We deforma given elementaryarea (So) according to thisvelocity field 1•. As- suming this area to be small with respect to its distance from the rotation pole, the velocity field can be considered I 7 X as constant over its whole surface. In this case, the instantaneous motion for each point of the area is described by a unique vector •Y.The modulus of • is given by I v = R. w ßsin D, (3) ,,, M L v-] where R = mean Earth radius, w = angular velocity in ra- dians per time unit, and D = angular distance between the rotation pole and the So area. So can be considered as a C single point of latitude I and longitude L, then D is given by D = cosI. cosI•,. cos(L- L •, ) q- sinI. sinL •,, (4) on a sphere. The value of v along the meridian is

v, = R.w. cos/•,. sin(L - L•,). (5) I. P The value of v along the parallel is L.

vz = R.w. [sinb, .cos/- cos/•,. sin/. cos(L- L•,)]. (6) Fig. 3. Notations and conventionsused in the text. a. Trans- pressive deformation of a cubic-shaped volume under the combi- The angle 6), representingthe obliquity of the relative nation of pure shear and simple shear. b. The same deformation motion vector •7 with respect to the azimuth of the strike- consideredin the horizontal (Ox, Oy) plane. c. Kinematic con- slip fault segment, is given by ventions used for a rotation around the rotation pole P. CALAIS AND DE L•PINAY: STRAIN AND KINEMATICS IN THE CARIBBEAN 8297

the width of the deformedzone along the main strike-slip E = 01 (F-v.sin©)-v.cos©)F - v. sin© ' (9) faults. NumericM deformation values (S parameter)along the successiveplate boundary segmentsare then obtained Let us consider the { vector, diagonal of the So initial that areexpressed in twoways (Figures 4 and5): area, whose coordinate are (F, F). After deformation, { becomes {', and its coordinates become 1. An histogram displays S variations along the fault. Each step givesthe S value for the correspondingfault seg- u =E.,7 ment. $ is positive in case of transpression,negative in case of transtension. This representation allows us to check if the computed deformation pattern along the entire plate boundary is realistic, by comparison with field observations. y - F(F- v. sinO) ' 2. A map displaysthe relative motion vectors. The length x- F-F.v. cosO(F - v.sin O) (10)of the vector is proportional to the velocity and its direction After deformation, the final area S' of the quadrilateral givesthe azi•nuth of the relative •notion of the movingplate is given by relative to the fixed one. S'= (x. y) -[(x - 1). y] -- y (11) A comparable approach to evaluate different sets of rotation parameters for the Caribbean/North America rela- S' = F(F- R.w. sinD. sin6)). (12) tive motion was used by Heubeckand Mann [1991]. Us- ing a graphicscomputer, they comparedoverlaps and gaps The ratio of areal changeto initial area, $ (dimension- along the strike-slipplate boundary producedby rotating less),for a squareof side F, is given by the Caribbean plate according to various published kinematic parameter sets. Our $ parameter is the quantitative So - $' R ßw ßsin D -sin 6) $ = S0 -- F . (13) equivalentof the graphic overlapsand gaps that they obtain. Heubeckand Mann's [1991]method, however,is en- The $ parameter, therefore, represents the superficial tirely graphic and does not allow quantitative evaluation of area change of a given lithosphere volume, given the dis- the tested models. placement imposed by the relative motion vector •7. Assum- ing the validity of above mentioned hypothesisconcerning Results the strain regime and the theology, we considerthat this $ parameter can be compared with the geological strain de- Four kinematic parameter sets have been proposed so far duced from field observations.A strongly transpressivearea to describe the motion of the Caribbean relative to North will thus correspondto high positive $ values(as can be America(Table 1) (Stein et al. [1988],Minster and Jordan expected for the Cordillera Septentrional of Dominican Re- [1978](RM-2 model),MacDonald [1976], Sykes et al. [1982] public, for example), whereasa strongly transtensivearea (Nuvel-1 model)). In attempting to test these models,we will correspondto negative$ values(as can be expectedfor digitized the trace of the northern Caribbean plate bound- the Mid-Cayman spreadingcenter, for example). We pro- ary using all the available Seabeam, seismic reflection, and poseto call the S parameter"relative areal strain" [cf. De geological field data. Paor, 1990]. The computed deformation models for each tested ro- Since no numerical value equivalent to this S parameter tation parameter set and the corresponding relative mo- can be obtained from field observationsof deformed rocks, tion vector maps are shown on Figures 4 and 5. Figure the $ valueshave to be interpreted in the followingway: if 6 displays the four models together, with the same verti- $ value is positive and greater in zone A than in zone B, cal scale. The model of Sykeset al. [1982] has the high- then one should observe, in the field, evidences of more in- est areal strain values, probably due to an overestimation tense transpressive deformation in zone A than in zone B. If of the Caribbean plate velocity [DeMets et al., 1990]. The this is not the case, then the initial conditions of the model modelsof MacDonald[•976] and of Sykeset al. [1982]dis- are not valid (either the fault geometry or the kinematic play contradicting predictions, except in the westernmost parameters).It shouldbe rememberedthat this approach part of the plate boundary. The first one overemphasizes relates strain to plate kinematics on the scale of an entire transtension whereas the second overemphasizestranspres- plate boundary. Therefore, the interpretation of the mod- sion. Between those two models, the parameters of Minster els must be based(1) on regionaltectonic structures rather and Jordan[1978] and Stein et al. [1988]tend to minimize than on local features; (2) on the analysisof a pattern of the compressive and extensive deformation associated with deformation along the entire plate boundary length. the strike-slip motion along the northern Caribbean plate boundary. They display similar predictions along the two SEMIQUANTITATIVE TEST OF KINEMATIC PARAMETERS strike-slip faults (Swan and Oriente faults) bounding the Method Cayman Trough between longitudes 88øW and 78øW. We examine in the following discussionthe four •nodels After digitizing the main fault trace, we calculate the val- in more detail (Figures4 and 5): ues of D, v, v•, vr•, O, and S for each fault segment. Since 1. All models predict significant transpressionalong the the motion vector (•') variesin azimuth and modulusalong Polochic-Motaguafault zone (92øW to 89øW), decreasing any given fault segment, each segment length (F) has to from west to east, in accordancewith geologicalobserva- be as small as possible with respect to the distance from tions. The models predict the transition between transpres- the tested rotation po]e. This condition is easi]y respected sion and transtension occurs around the eastern coast of by choosinga small enough step when digitizing the fault Guatemala, in agreement with the location of the actively trace. This step averages35 km in our model and is about subsiding Lago de Izabal area. The model of Minster and 8298 CALAISAND DE L•PINAY: STRAINAND KINEMATICS IN THE CARIBBEAN

90 85 80 75 70 65

lOO

a -lOO

Mac Donald (1976) ß lat ß -15.0 ø Ion ß -75.0 ø ang ß 0.40ø/Ma

90 85 80 75 70 65

lOO

-,•

:•:• • "• 0

b -100

Sykes et al. (1982) ' lat ß 66.0 ø Ion ß 132.0 ø ang ß 0.36ø/Ma

Fig. 4. Theoreticaldeformation models computed using rotation parameters determined by previousauthors. MacDonald[1976]; b. Sykeset al. [1982];c. Minsterand Jordan [1978]; d. Steinet al. [1988]. CALAIS AND DE L•PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN 8299

9O 85 8O 75 7O 65

lOO

- •

_ c -lOO

Minster and Jordan (1978) ß lat ß -34.18' Ion ß -70.40' ang ß 0.22ø/Ma

90 85 80 75 70 65

lOO

" d , t -100

Stein et al. (1988, N l-G) ß lat ß -55.20 ø Ion ß -60.80 ø ang ß 0.11ø/Ma

Fig. 4. (continued) 8300 CALAIS AND DE L•PINAY: STP•AIN AND KINEMATICS IN THE CAP•IBBEAN

90 85 80 75 70 65 i i i i i i

0 ,.•i::i::i::{ii!?:iii::i::i::iii::iii::i::i::iii::i::i::i:ii!•?•:?•:::•:•::-":...... •:':.....• -

Mac Donald (1976)' lat ß -15.0' Ion ß -75.0 ø ang ß 0.40ø/Ma

Sykes et al. (1982) ß lat ß 66.0 ø Ion ß 132.0 ø ang ß -0.36ø/Ma

9O 85 8O 75 7O 65 i i i i i i

19 •o:zo • :•

Minster and Jordan (1978) : lat :-34.18 ø Ion :-70.40 ø ang : 0.22ø/Ma Fig. 5. Mapsof therelative motion vectors of theCaxibbean plate relative to NorthAmeric• along the northern Caribbeanplate boundaxy, using rotation parameters determined by previousauthors. a. MacDonald[1976]; b. S3/keset al. [1982];c. Minsterand Jordan [1978]; d. Steinet al. [1988]. CALAIS AND DE L•PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN 8301

90 85 80 75 70 65 i i i i i i

Stein et al. (1988, N1-G) : lat :-55.20 ø Ion :-60.80 ø abg : 0.11ø/Ma

Fig.5 (continued)

Jordan[1978] predictsa transpressiontwice as intenseas 68ø), correspondingto the observedtranspressive deforma- the one of Stein et al. [1988]. tion of the Cordillera Septentrional[Calais et al., 1992b 2. To the east, along the Swan fault, all models predict ; DeZoeten and Mann, 1991]. The model of Sykeset al. transtension up to the longitude of 85ø30'W, in accordance [1982]seems to overestimatethe predictedtranspression in with the most recent structural observations in this area this area, whereasthe one of Stein et al. [1982]tends to [Mannet al., 1989;Rosencrantz and Mann, 1991]. However, minimize the compressivedeformation. none of the kinematic models accounts for the transpres- 7. Along the Puerto Rico trench (68øW to 62øW), the sive structures observedbetween the Bay and Swan Islands modelsof Stein et al. [1988]and S•tkeset al. [1982]predict (85ø30'W to 83ø30'W). To the east of the Swan Islands transpressionup to the longitude of 64øW and beyond that (83øW to 81ø30'W), all the modelspredict transpression, transtension.The modelsof MacDonald[1976] and Min- although neither seismic reflection profiles, nor SeaMARC ster and Jordan[1978] predict transtension from .theMona II recordingsdisplay the correspondingstructures [Mann et Passageeastward. This second group seemsin better ac- al., 1989, 1991]. cordancewith the observednormal faulting along the inner 3. All modelsexcept the one of MacDonald[1976] pre- wall of the Puerto Rico trench. dict transpressivedeformation along the easternpart of the The graphic representation of the relative motion vectors southernCuban margin (from 77øW to 75øW), in accor- along the northern Caribbeanplate boundary (Figure 5) dance with the existence of the Santiago Deformed Belt clearly shows the obliquity of the predicted displacement [Calaisand Mercier de Ldpinagt,1991]. All the modelsalso along the main strike-slip fault. All models display relative displaythe structuralconsequences of threeleft-stepping re- motion vectors very close to the azimuth of the Swan and lay areas, correspondingto the Cabo Cruz, Chivirico and Oriente faults in the vicinity of the Mid-Cayman spread- Baitiquiribasins [Calais and Mercier de Ldpinagt,1991]. In ing center. As a matter of fact, the youngest parts of the Cabo Cruz area (79øW to 75øW), the modelof Stein et these faults can be considered as flow lines of the recent al. [1988]minimizes the tensionaldeformation, whereas the Caribbean/North America relative motion. East and west oneof MacDonald[1976] overemphasizes it. of the plate boundary, however, the predicted motion vec- 4. All models predict transpressionin the Windward Pas- tors of the various models are significativelydifferent. When sagearea (74øW to 73øW), as also indicated by the avail- usingthe rotation parametersof MacDonald[1976],the pre- ablestructural data [Mercierde Ldpinagtet al., 1992].The observedtranspression, however, is comparativelylow with TABLE 1. Instantaneous Rotation Parameters of the Caribbean respect to the transpressionobserved along the southern Plate Relative to North America Proposedby Previous Authors Cuban margin and in the northern Dominican Republic. Only the modelof Minster and Jordan[1978] accounts for Authors Latitude Longitude Angular velocity this differencein deformationintensity. 5. All models predict transpression within the Tortue 1 -15.0 -75.0 0.4 2 -34.18 -70.4 0.225 Channeland along the northernHaitian coast (73øW to 3 66.0 132.0 -0.36 71ø30'W), as suggestedfrom the few structural data avail- 4 -55.2 -60.8 0.11 able in this area. These data, however, are not detailed enough to allow a more precise comparisonwith the pre- 1. MacDonald,[1976]; 2. Minster and Jordan,[1978]; dicted deformation models. 3. Sykeset al., [1982];4. Stein et al., [1988]. Latitudes 6. All models predict a westwardincreasing transpres- and longitudesin degrees,angular velocities in degreesper sion alongthe northernDominican Republic (71ø30'W to million years, clockwise sense. 8302 CALAIS AND D•, L•PINAY: STRAIN AND KINEMATICS IN THE CARIBBEAN

-10

-20

-30 95 -85 -75 -65 -90 -80 -70 -60

longitude (deg.) [] MacDonald

+ Minster and Jordan ½ Stein et al. Sykes et al.

Fig. 6. Areal strain parameterplotted alongthe plate boundary,according to the testedmodels. The vertical scale is the same for the four models.

dicted motion variesbetween N60øE in the westernpart of ern Caribbean plate boundary, one must conclude that they the plate boundary,to N l10øE at its eastern end (Figure are inaccurate descriptionsof the actuM present-daymotion 5a). On the other extreme, the parametersof Sykeset ai. between the Caribbean plate and North America, as Mso [1982]predict the sameazimuth of relative motion Mong stated by Heubeckand Mann [1991]. The explanationfor the entire plate boundary. In the vicinity of the Puerto Rico this discrepancy probably lies in the basic hypothesislead- trench,only the modelsof MacDonald[1976]and of Minster ing to classical plate motion calculation. As presented by and Jordan[1978] are in accordancewith a southerlycom- Stein et al. [1988],classical kinematic models only take into ponent of motion of the Caribbean plate relative to North accountspreading rates, transform azimuths and earthquake Americasuggested by Speedand Larue [1991]on the basis slip vectorsas the basicdata for their calculations.Neglect- of marine structural observations in this area. ing the structures associatedwith the transform fault and using an assumedrectilinear fault trace allows these models Discussion to describethe motion only as a pure and rigid plate rotation. Since these models do not correctly account for the A comparative analysis of the tested kinematic param- observeddeformation, we concludethat they do not repre- eters shows that each of the predicted deformation mod- sent the total plate motion, but only one componentof it els displays correlations and mismatches with the observed that we call "rigid plate rotation component". deformationpattern. The best fitting rotation parameters The hypothesisdeveloped in the followingdiscussion to seemto be thosefrom Minster and Jordan[1978], as also interpret the discrepancy between the actual motion and stated by Stein et al. [1988],Heubeck and Mann [1991], the rigid motion is the existenceof significantdeformation and Calais and Mercier de Ldpinay[1991]. However,since within the plate boundary zone. This "plate boundary zone none of the tested rotation parameter sets correctly ac- deformation component" is not taken into account in classi- countsfor the whole deformationpattern along the north- cM kinematicmodels [Minster and Jordan,1984]. An alter- CALAIS AND DE LI•,PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN 8303

9O 85 8O 75 7O 65

lOO

.t o

-lOO

constructed deformation model

Fig. 7. Constructedmodel of deformationalong the northern Caribbean transcurrentplate boundary used to calculate the best fitting set of rotation parameters.

nativeexplanation is proposedby Heubeckand Mann [1991]. usedin the computationsas a referencefor relativequantita- They suggestthat the Caribbeanplate has brokenup into tive comparisons,the absolutevalue given to eachfault seg- four subplatesand that its composition(mainly transitional ment has no significance.However, the relativeheight of the crust) makesit prone to ductile deformation. In this hy- histogrambars for the successivefault segmentsrepresents pothesis,the equationsof rigid plate kinematics would be as closelyas possiblethe observedtectonic pattern alongthe unable to solvefor a unique rotation parameters set for the plate boundary. Each strain valuecan be weightedby an er- Caribbeanplate motion. Beforemaking any definitivecon- ror coefficient(e• in equation(13)), whichallows us to dis- clusion,our next step is thus to answerthe question"does card in the computationfault segmentswhose strain regime a rotation parameter set exists that can fit the observed de- is poorlyconstrained (lack of structuraldata or ambiguous formationpattern alongthe entire northern Caribbeanplate tectonicinterpretation). The modelwe obtain (Figure 7) boundary?" representsas accurately as possiblethe actual present-day In the following, the observeddeformation in the plate tectonic pattern along the whole plate boundary. boundary zone is taken into account to calculate new kine- The computationconsists of calculatingone theoretical matic parametersfor the Caribbean/North Americarelative deformationmodel for each of n possiblerotation poleslo- motion. If a singlemotion parameter set can be found, which cated at the nodesof a given latitude/longitudegrid. The is able to describethe whole plate boundary deformation comparisonbetween each of these computed models and pattern, we can consider that it represents the actual mo- the constructedone is quantifiedby a residualmean square tion betweenboth plates, i.e. the combinationof the rigid (rms) function basedon the differencebetween the com- plate rotation componentand of a "plate boundary defor- putedareal strain values (Scomp,i in equation(13)) andthose mation" component. The existenceof such a parameter set imposedin the constructedmodel (S½o,s,i in equation(13)), would be a contribution to a direct relation between overall taking into accountthe errors on these imposedvalues. For plate kinematics and plate boundary zone deformation. k faults segmentsand a given plate motion model, the rms function is given by CALCuLATION OF INSTANTANEOUS k MOTION PARAMETERS Tins--E (Scons,i- Scomp,i)/ei 2 General Principle i=1 We built the deformationmodel displayed on the Figure The valuesobtained at eachlatitude/longitude node are 7 basedon the currentknowledge of the northernCaribbean then gridded and contouredto establishan iso-rms map. tectonics.An arealstrain value (S) wasassigned to eachdig- It is assumedthat the most accuratemotion parameter set itized fault segment. Since this deformation model will be correspondsto the minimum rms value. Moreover, the iso- 8304 CALAIS AND DE Li•,PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN rms contoursgive a confidencesurface associatedwith this information deduced from published and unpublished data. rotation pole, bounded by the rms value calculated usingthe All the references that we used are cited in the previous best fitting rotation parameters,assuming that all errors on discussion.We also refer the reader to Mann et al. [1991] the constructed model axe equal to zero. for a summary of the available geological data along the It is to be noted that this approach is not based on the northern Caribbean plate boundary. fit of more parameters than in classical kinematic calcula- 4. The linear velocityused is 0.02 m/yr at 18.7øN/81.4øW tions. As a matter of fact, the rms is only a function of fault (point located on the Caribbean plate, 1 km to the east of segmentazimuth, fault segmentlocation, and rotation pole the mid-Cayman spreadingcenter). This value is given by location (parameterscontained in F, D and O (see equa- Rosencrantzet al. [1988]from the interpretationof the most tion (12)). On the contrary, the accuracyof our model is recent Cayman Trough magnetic anomalies. It is in good based on the integration of numerous observedstr•ictural agreementwith the value of 1.1 cm/yr obtainedby Mocquet data distributed along the whole plate boundary trace. [1984]from seismologicaldata alongthe Oriente fault. The result of this computation is shown in Figure 7. The Application to Present-Day Motion of the Caribbean plate minimum of the rms function corresponds to the follow- Relative to North America ing Euler vector (Caribbeanplate relative to North Amer- ica)' latitude 33.9øS, longitude 69.0øE, angular velocity We applied the above described method to the northern 0.10ø/m.y.. Caribbean plate boundary using the following data:

1. The main strike-slip fault was digitized in 86 segments Discussion (k in equation(13)), on the basisof all the availableSea Beam, seismic reflection, and geologicalfield data. These The motion parameters determined in this work differs segmentsare the same as those used for the test of rotation from all the previouslyproposed (Figure 8). It is, however, parameters. very dose to the RM-2 pole of Minster and Jordan[1978] 2. The constructed deformation model is displayed in located within the confidence axea. This result is in ac- Figure 7. It takes into account all the available geological cordance with the Minster and Jordan's model to account • 75'W 65'W 55'W•

20'S 3

30'S

40'S

50'S S 1. MacDonald,1976 2. Minster and Jordan, 1978 (RM-2) 3. Sykes et al., 1982 4. Stein et al., 1988 (Nuve11-G)

60'S minimal rms: lat. 33.9øSlong. 69.0øW ang. velocity0.10 ø/Ma confidence surface O MinsterandRM-2Jordan [1978] •......

iso-rms contour interval : 5% 70'S Steinet al. [1988] Nuvel-1 Fig. 8. a. Iso-rms map obtained using the constructed deformation model shown on Figure 6. b. Location of the rotation poles proposed by previous authors in comparison with the new pole. CALAIS AND DE L•PINAY: STRAIN AND KINEMATICS IN THE CARIBBEAN 8305

90 85 80 75 70 65

lOO -

. • ..-- - E _.• _

- o

-lOO

90 85 80 75 70 65 I I I I I I

this study' lat ß -33.9 ø Ion ß -69.0 ø ang ß 0.10ø/Ma Fig. 9. a. Theoreticaldeformation model obtainedusing the rotation parameterscomputed in this study: 1, strong transpressionalong the Polochic-Motaguafault zone; 2, transtensionfrom the Lago de Isabal to the Bay Island;3, little transpressionalong the Swanfault; 4, oceanicopening of the mid-Caymanspreading center; 5, unknownarea; 6, transtensionand openingof the Cabo Cruz basin;7, transpressionand uplift of the Santiago deformedbelt; 8, little transpressionthrough the Windward Passage;9, transpressionin the Tortue channel;10, strongtranspression and uplift of the CordilleraSeptentrional of DominicanRepublic; 11, transtensionand strong downfaultingalong the inner wall of the Puerto Rico. b. Map of relative motion vectorsof the Caribbeanplate relative to North America obtained using the rotation parameterscomputed in this study. 8306 CALAIS AND DE I3•PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN for the observeddeformation along the northern Caribbean parameters calculated in this work minimize the deforma- plate boundary,as previouslystated. The pole that we pro- tion (eithertranspression or transtension)associated with posedisplays a muchgreater uncertainty in latitude than in the main strike-slipfault. Figure 10 alsosuggests that, at longitude,typical for an almosteast-west trending strike-slip least for somesegments of the plate boundary,Stein et al.'s plate boundary. model is the one that best fits the resultspresented in this We tested the motion parameters calculatedin this work paper, although Minster and Jordan's rotation pole is much usingthe processdescribed in the first part of this paper. closerto the one proposedhere. This apparentgood fit is in Figure 9 showsthe correspondinghistogram and relative fact the result of the influenceof w in equation(13). As a motion vector map. One can observethat the calculated matter of fact, Minster and Jordan[1978] overestimate the model of deformation is the one which best accounts for the velocity of the Caribbean plate by a factor of 2 in compar- observeddeformation along the entire northern Caribbean isonwith the mostrecent results [Rosencrantz et al., 1988]. plate boundary. The calculated deformation pattern is The Euler vector proposed here representsa compromisebe- in agreementwith the observedone along the Polochic- tween Minster and Jordan's pole location and Stein et al.'s Motagua fault zone (high transpression),along the Swan plate velocity. fault (transtensionto longitude85ø30'W, little transpression to the east),along the southernCuban margin (transtension CONCLUSIONS in the Cabo Cruz basin, transpression in the Santiago deformedzone), north of Hispaniolaand throughthe Domini- As demonstrated above, classical kinematic parameters canRepublic (transpression in the CordilleraSeptentrional), are unable to describe the actual motion between the and alongthe Puerto Rico trench (transtensionand collapse Caribbean and the North American plates, because they of its inner wall). cannot produce the deformation pattern actually observed Figure 10 displaysa comparisonbetween the modelsof along the plate boundary. These kinematic parameters, Minster and Jordan[1978] and Stein et al. [1988]and the by only taking into accountspreading rates, transformaz- results presented in this paper. Figure 10 shows that the imuths,or earthquakesslip vectors,and neglectingthe struc-

-2

-4

-6

-8

-10

-12

-14

-16

-90 -80 -70 -60

longitude (deg.) Minster and Jordan

0 Stein et al. this paper Fig. 10. Comparisonofthe predicted areal strain parameter along the plate boundary bet,ween the model presentedin this paper and those of Minsterand Jordan [1978] and Stein et al. [1988]. CALAIS AND D•, L•PINAY: STRAIN AND KINEMATICSIN THE CARIBBEAN 8307

tures associatedwith the plate boundary zone, only repre- Calais, E., and B. Mercier de L&pinay, Strike-slip tectonic pro- senta "rigid plate rotation component"of the actual (total) cessesin the northern Caribbean between Cuba and Hispaniola motion. This rigid componenthas to be complementedby (Windward Passage),Mar. Geophys.Res., in press,1993. Calais E., N. Bethoux, and B. Mercier de L&pinay, From transcur- a "plate boundary deformation component". rent faulting to frontal subduction: A seismotectonicstudy of On the basisof the observedstrain regime along the suc- the northern Caribbean plate boundary from Cuba to Puerto cessivesegments of the Northern Caribbean plate boundary, Rico, Tectonics, 11, 114-123, 1992a. this work demonstrates that it is possible to calculate mo- Calais, E., B. Mercier de L&pinay, J. Butterlin, P. Saint-Marc, tion parametersthat take into accountthis "plate boundary and A. Schaaf, Evolution pal•og•gr.a.phique et structurale de la fronti•re de plaques nord caraibes en Hispaniola au deformation component". The new parameters proposedin C&nozo•'que,Bull. Soc.Geol. Ft., 163,309-324, 1992b. this paper (rotation pole latitude 33.9øS,longitude 69.0øW, Carr, M.J., and R.L. Stroiber, Geologic setting of some destruc- angularvelocity 0.10ø/m.y.) correctlyaccount for the recent tive earthquakes in Central America. Geol. Soc. Am. Bull., 88, deformation(from 2 Ma to Present)along the entirenorth- 151-156, 1977. DeMets, C., R.G. Gordon, D.F. Argus, and S. Stein, Current ern Caribbean plate boundary zone. We, therefore, believe plate motions, Geophys. J. R. Astron. Soc., 101, 425-478, that they describethe actual motion of the Caribbean plate 1990. relative to North America, i.e., the combinationof a rigid De Paor, D.G., Cross-section balancing in space and time, in plate rotation component and of a plate boundary deforma- Petroleum and Tectonics in Mobile Belts, pp. 149-154, Tech- nips, Paris, 1990. tion component. De Zoeten, R., Structure and stratigraphy of the Central The coherence of the results presented here demonstrates Cordillera Septentrional, Dominican Republic, Master thesis, that the plate boundary zone deformation in the north- 292 pp., Univ. of Tex. at Austin, 1988. ern Caribbean can be described within the framework of De Zoeten, R., and P. Mann, Structural geologyand Cenozoictec- rigid plate tectonics and thus confirms the validity of our tonic history of the central Cordillera Septentrional, northern Dominican Republic, in Geologicand Tectonic Development of approach. The fact that the rigid-plate kinematics equa- the North America- Caribbean Plate Boundary Zone in His- tions can provide an unique Eulerian vector that correctly paniola, edited by P. Mann, G. Draper and J. Lewis, Spec. describesthe plate boundary deformation shows that the Pap. Geol. Soc. Amer., 262, 265-280, 1991. Caribbean plate behaves as a rigid body, at least for the Edgar, N.T., Structure and geological development of the Cibao Valley, northern Hispaniola, paper presented at 12th time and space scalesconsidered. In other words, these re- Caribbean Geology Conference, Ste Croix, Virgin Islands, sults show that the motion responsiblefor intraplate defor- 1989. mation within the Caribbean plate is of much lower magni- Erdlac, R.J., Jr., and T. Anderson, The Chixoy-Polochic fault tude than the strike-slip motion along the plate boundary and its associated fractures in western Guatemala, Geol. Soc. Am. Bull., 93, 57-67, 1982. itself, and the only significantdeformation is concentrated Guzm•n-Speziale, M., W.D. Pennington, and T. Matumoto, The in the plate boundary zone. Consequently,we think that the triple junction of the North America, Cocos, and Caribbean deformationwithin the northern Caribbean plate boundary plates: Seismicity and tectonics, Tectonics, 8, 981-997, 1989. zoneis rather controlledby regionalplate tectonics(i.e., by Heezen, B.C., W.D. Nesteroff, M. Rawson, and R.P. Freeman- the Caribbean/North America relative motion) than by lo- Lynde, Visual evidence for subduction in the western Puerto Rico trench,in SymposiumG•odynamique Cara•be, pp. 287- cal effects(e.g., block rotationswithin the plate boundary 304, Teclmips, Paris, 1985. zone). Heubeck, C., and P. Mann, Geologicevaluation of plate kinematic Plate boundary structures along transcurrentfault zones models for the northern Caribbean plate boundary, Tectono- are therefore very reliable recorders of plate motion. Their physics, 191, 1-26, 1991. Heubeck, C., P. Mann, J. Dolan, and S. Monechi, Diachronous usein a semiquantitative way, as demonstratedin this paper, uplift and recyclingof sedimentarybasins during Cenozoictec- can lead to an accurate determination of the total motion tonic transpression,northeastern Caribbean plate margin, Sed- between two plates. One can infer that the developmentof iment. Geol., 70, 1-32, 1990. geodeticmeasurements (especially Global PositioningSys- Le Pichon, X., J. Iiyama, J. Bourgois, B. Mercier de L&pinay, tem), that allow for calculationof the combinedeffects of J. Tournon, C. Muller, J. Buttertin, and G. Glad;on,Premiers r•sultats de la campagned'essais du submersiblefran•;ais Nau- plate motion and plate boundary deformation will provide tile dansla fossede Porto Rico (GrandesAntilles), C.R. Acad. valuable assistancein understanding interactions between Sci., 301, 743-749, 1985. strain and plate motion, especially in the case of the north- MacDonald, W.D., Cretaceous-Tertiary evolution of the ern boundary of the Caribbean plate. Caribbean, paper presented at 8th Caribbean Geology Con- ference, Saint-Christophe, Curacao, 1976. Mann, P., K. Burke, and T. Matumoto, Neotectonics of Hispan- Acknowledgments. The authors are gratefully indebted to Bernard Minster and J•eanVirieux for their helpful comments on iolaplate motion, sedimentation and seismicity at a restraining bend, Earth Planet. Sci. Left., 70, 311-324, 1984. the first version of this paper. We thank Paul Mann and two anonymous referees for critical reviews of this work and for their Mann, P., E. Rosencrantz,S.A. Tyburski, B. Mercier de L•pinay, excellents comments. and E. Calais, SeaMarc II survey of the North America- Caribbean plate boundary zone between Western Jamaica and Eastern Guatemala, Eos Trans. A GU, 70, 1346, 1989. REFERENCES Mann, P., G. Draper, and J.F. Lewis, An overviewof the geologic and tectonicdevelopment of Hispaniola,in Geologicand Burkart,B., NeogeneNorth Americ•m-Caribbea across northern Tectonic Development of the North America-Caribbean Plate Central America: Offset along the Polochic-Motaguafault, Boundary Zone in Hispaniola, edited by P. Mann, G. Draper, Tectonophysics, 99, 251-270, 1983. and J. Lewis, Spec. Pap. Geol. Soc. Amer., 262, 1-28, 1991. alais, E., and B. Mercier de Lepinay, A natural model of active Manton, W.I., Tectonic interpretation of the morphologyof Hon- transpressional tectonics: The en (•chelon structures of the Ori- duras, Tectonics, 8, 633-651, 1987. ente Deep, along the northern Caribbean transcurrent plate Masson, D.G., and K.M. Scanlon, The neotectonicsetting of boundary, Rev. Inst. Ft. Pit., •5, 147-160, 1990. Puerto Rico, Geol. Soc. Am. Bull., 103, 144-154, 1991. Mc- Calais, E., and B. Mercier de L•pinay, From transtension to Cann, W.R., and L.R. Sykes, Subduction of aseismicridges transpressionalong the northern Caribbean plate boundary beneath the Caribbean plate: Implications for the tectonic and off Cuba: Implicatiorm for the recent motion of the Caribbean seismic potential of the northeastern Caribbean, J. Geophys. plate, Tectonophysics,186, 329-350, 1991. Res., 89, 4493-4519, 1984. 8308 CALAIS AND DE Li•PINAY: STRAIN AND KINEMATICS IN THE CARIBBEAN

Mercier de L•pinay, B., V. Renard, and E. Calais, Morphology Sanderson, D.J., and W.R.D. Marchini, Transpression, J. of and structure alonga major marine strike-slipfault zone: The Struct. GeoL, 6, 49-458, 1984. northern Caribbean plate boundary between Cuba, Hispaniola Schwarz, D.P., L.S. Cluff, and T.W. Donnelly, Quaternary fault- and Jaxnaica,paper presentedat 28th International Geological ing along the Caribbean North American pl&te boundary in Congress,Geol. Soc. Am., Washington, D.C., 1989. Centram America, Tectonoph•lsics,5•, 431-445, 1976. Mercier de L•pinay, B., V. Renard, J.M. Vila, G. Buffet, and Speed, R.C., and D.K. Larue, Extension and transtension in the E. Calais, North Caribbean plate boundary: High reolution plate boundary zone of the northeastern Caribbean, Geoph•s. imagery of the Oriente fault acrossthe Winward Passageand Res. Left., 18, 573-576, 1991. south of Tortuga Island, EOS Trans. A G U, 73, 278, 1992. Stein, S., C. DeMets, R.G. Gordon, J. Brodholt, D. Argus, J.F. Minster, J.B., and T.H. Jordan, Present-day plate motions, J. Engeln, P. Lundgren, C. Stein, D.A. Wiens, and D.F. Woods, Geoph•ls.Res., 83, 5331-5354, 1978. A test of alternative Caribbean plate relative motion models, Minster, J.B., and T.H. Jordan, Vector constraints on Quaternary J. Geoph•ls. Res., 93, 3041-3050, 1988. deformation of the western United States east and west of the Sykes, L.R., W.R. Mc Cann, and A.L. Kafka, Motion of the San Andreas fault, in Tectonics and Sedimentation Along the Caribbean plate during the last 7 million years and implica- California Margin, Pubi. 38, edited by J.K. Crouch and S.B. tions for earlier Cenozoic movements, J. Geoph•ls. Res., 87, Bachmart, pp.1-16, Pacific Section, Society of Economic Pale- 10,656-10,676, 1982. ontologistsand Mineralogists,Los Angeles,California, 1984. Sylvester, A.G., Strike-slip faults, Geol. Soc. Am. Bull., 100, Mocquet, A., Vitesses de d•placement discontinu le long d'une 1666-1703, 1990. liraitcde plaques: Caral'bes-Am•rique du Nord, M•m. Diplome Wilson, J.Y., A new classof faults and their bearing on continen- Etudes Approfondies, 53 pp., Univ. of Rennes, France, 1984. tal drift, Nature, •07, 343-347, 1965. Morgan, W.J., Rises, trenches,great faults •nd crustal blocks, J. Geoph•ts.Res., 73, 1959-1982, 1968. E. Calais, Scripps Institution of Oceanography, 9500 Gilman Plalker, G., Tectonic aspects of the Guatemala earthquake of 4 Drive, La Jolla, CA 92093-0225,(eric(•pgga.ucsd.edu), February 1976, Science, 93, 1201-1208, 1976. B. Mercier de L•pinay, Institut de G•odynarnique, CNRS, Av. Rosencrantz, E., and P. Mann, SeaMarc II mapping of transform Albert Einstein, 06560 Valbonne, France. faults in the Cayman Trough, Caribbean Sea, Geolot?•, 19, 690-693, 1991. Rosencrantz, E., M.I. Ross, and J.G. Sclater, The age and spreading history of the Cayman Trough as determined from depth, (ReceivedDecember 23, 1991; heat flow and magnetic anomalies, J. Geoph•ls.Res., 93, 2141- revised December 22, 1992; 2157, 1988. &cceptedDecember 28, 1992.)