JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 86, NO.

Bll, PAGES 10753-10770,NOVEMBER 10, 1981

Subduction of the Eastern Panama Basin and Seismotectonics

of Northwestern South America

WAYNE D. PENNINGTON !

Geophysical
and Polar ResearchCenter,Departmentof Geologyand Geophysics
Universityof Wisconsin-Madison,
Madison, Wisconsin53706

Spreadingalongthe Cocos-Nazcaplate boundarysincethe breakupof the Farallon plate in the Mio-

cenehas resultedin the formationof the Panamabasinand a complexinteractionof platesin and near
northwesternSouth America. Current plate boundarieshave been defined,and segmentsof subducted
lithosphereidentifiedthroughselectionof hypocentrallocationsof earthquakes,consideringonly well-
locatedevents,and throughfocal mechanismdeterminations. The existenceof relict plate boundaries,
bathymetricfeatures,and the Panamanianisthmushas affectedthe subductionprocessof the Nazca
plate beneathSou_thAmericaand determinedthe present-dayconfigurationof the subductinglithos-
pheric plate. There is no singletriple junction separatingthe Caribbean, South American, and Nazca
plates.Instead,the Panamanianisthmusand surroundingareasare accommodating east-westcompres-
sion(and a lesserdegreeof north-southcompression) along a seriesof thrustfaults strikingNW to NE,
and the Andeanrangesof Ecuador,Colombia,and Venezeulaare movingasa block NNE relativeto the
restof the SouthAmericanplates,alonga systemof faultsfollowingthe front of the EasternCordillera.
The subductedportionsof the Panamabasinand old Farallon plate have becomesegmentedinto three
piecesrecognizedin this study.From north to south,they are (1)a 'Bucaramanga'segmentcontinuous
with the Caribbeanseafloornorthwestof Colombia,(2) a 'Cauca'segmentcontinuouswith oceaniccrust
(Nazca plate) currentlybeing subductedbeneathSouth America at the Colombia-Ecuadortrench, and
(3) an 'Ecuador'segmentat the northernend of the subductedlithosphericplate which is dipping at a
smallangleto the eastbeneathnorthernPeru.The segmentation of the subductedplate can be explained
by the buoyancyof bathymetricfeatureswhich have beenpartially subducted.
_

INTRODUCTION schumacher,1976], now representedby a scarp along the

southernlimit of the 'Galapagosgore' (Figure 1). The Gala-
The Andeanmargin of SouthAmericaand the circum-Car- pagosgore representsthe limits of crustcreatedby spreading
ibbeanregionhavelongbeenareasof specialinterestto geol- on the Cocos-Nazcaplate boundary[Holdenand Dietz, 1972].
ogistsand geophysicists.
Sincethe later 1960's,studiesboth on The configurationof the spreadingcenter-transformfault sys-
land andat seahavefocusedon determiningthe structureand tem within the Panama basin has been determined by bath-
historyof the area within the frameworkof plate tectonics. ymetric and marine magneticstudiesreportedby vanAndel et
Thesehavebeenespeciallysuccessful alongthe Andean mar- al. [1971],Hey et al. [1972, 1977],Handschumacher
[1976],and
ginsouthof theequator[Barazangi
andIsacks,1976],along Lonsdale and Klitgord [1978]; the active and relict plate
theMiddle
American
trench
[Molnar
andSykes,
1969;
Deanboundaries
asdetermined
bythose
studies
areshown
'mFig-
andDrake,1978],
andneartheGalapagos
triplejunction
[Hey ure2.
etal.,1972,
1977]. Themajorfaults
control
theshape oftheAndeanranges
North oftheequator
along
theAndean margin,alongthe andthevalleys
whichseparate
them (Figure
2)andaresome-
southernCaribbean,
andinPanama,theapplication
ofplatetimes
credited
forcontrolling
present-day
tectonics
(forex-
tectonic
concepts
hasbeen
difficult.
Inthisregion
theCocos,ample theproposed'Dolores
megashear'
[Campbell,
1968],
Nazca, SouthAmerican,
andCaribbeanplatesinteract,
but whichincludes
theRomeralfault);
yetmany ofthemapped
notalloftheplateboundaries
aregenerally
agreedupon.faults
seemtobepresently
inactive
[Feininger,
1970].
Lonsdale
From
regional
and
global
evidence,
Jordan
[1975]
andMinster
and
Klitgord
[1978]
have
pointed
outthat
theseismicity
attrib-
andJordan
[1978]
havedetermined
thattheNazcaandCarib-utedtotheDolores megashear occurs
at intermediate
focal
beanplates
arepresently
movingeastwardwithrespect
to depthsandisassociatedwitha subducted
pieceoftheNazca
SouthAmerica;
boththeNazca andSouth American
platesplateratherthanthecontinental (South American)litho-
areslowly
converging
withtheCaribbean;
andtheCocos
and sphere.Thus, althoughmajor faults
controlthesurfacefea-
Nazcaplates
aremovingapart. tures
oftheAndes, theydosobecausetheyseparaterocksof
ThePanama basin
istheareaenclosed
bythecontinental
quitedifferent
lithologies
(theDoloresmegashear separates
shelves
ofColombia
andPanama
andtheCocos
andCarnegieoceanic
rocks
in theWestern
Andes
fromcontinental
rocksin
ridges,
tracks
oftheGalapagos
'hotspot'
[Johnson
andLowrie,theCentralAndes [Case
etal.,1971]),
notnecessarily
because
1972].
Thebasin
wasformedbyseafloorspreading
alongthe theyarecurrently
active
features.
Cocos-Nazca
plate
boundary
sincethebreakup
oftheFatal- Subduction of oceanic
lithospherebeneath
thenorth-
lonplate
27m.y.B.P.
Plate
separation
occurred
along
a pre- westernpartofSouthAmerica hasbeena matter
ofsome
existing
zoneofweakness,
theGrijalva
Fracture
Zone[Hand-controversy.
Santo[1969]
concluded
thata south-dipping
Benioff zone occurs beneath northern Colombia; Dewey
[1972]showedthat thisapparentzoneresultedfrom a system-
Now at Departmentof GeologicalSciences,
Universityof Texas
at Austin, Austin, Texas 78712.
atic error in locationsdeterminedby arrival times reported
mostly from North American stationsand concludedthat if a
Copyright
1981by theAmerican
Geophysical
Union. Benioffzoneexists,it strikesN-S and dipsto the east.Isacks
Paper number lB 1273. 10753
0148-0227/81/001 B- 1273501.00
10754 PENNINGTON:TECTONICSOF NORTHWESTERN
SOUTHAMERICA

90 80 70 60
[ i I

NORTH AMERICA

20 20 N

CARIBBEAN

I0 I0

cocos
/,
VENEZUELA
/
/
/
/

/
/

CRR
COLOMBIA

'-..... /'"\
f....-c 0 o

BRAZIL

/ i
./ t------
-'"")i
z.-xj
,,

\ PERU '
,J SOUTH AMERICA
,.,. o

NAZCA

90 W 80 70 60

Fig. 1. Major featuresof northwestern

SouthAmericaand neighboringareas.Generallyagreeduponplate boundaries
are shown.The Panamabasinis the area enclosedby the Cocosand Carnegieridges,tracksof the Galapagos'hot spot'
now centeredbeneaththe Galapagosislandsat 1S, 91W. Other featuresare WA, WesternAndes;CA, Central Andes;
EA, Eastern Andes; SM, Serannia de Merida; PFZ, Panama Fracture Zone; CRR, Costa Rica rift; ER, Ecuador rift; GR,
Galapagosrift; GG, Galapagosgore;CET, Colombia-Ecuadortrench.

and Molnar [1971]concludedthat a Benioffzone exists,dip- the Cocos-Nazcaplate boundary,and that a systemof strike
pingto theeastbeneathEcuadorandto thesoutheast beneath slip faults in northernVenezuela(mappedby Rod [1956])
northernColombia. formspart of the Caribbean-SouthAmericanplateboundary
Seismicityand focalmechanism studiesof the tectonicsof (althoughthis interpretationhas recentlybeen questioned
northwestern South America and the eastern Panama basin [Rial, 1978]).
sufferfrom a scarcityof data. Studiesof seismicityand focal In the time sinceMolnar and Sykes'[1969]study,the num-
mechanismshave not been able to defineaccuratelythe tec- ber of well-locatedearthquakeshasmorethan doubled.In ad-
tonics within northwesternSouth America (including Pan- dition, a number of large-magnitudeeventsfor which focal
ama) nor to determinethe geometryof subductedplatesin the mechanismscan be determinedhave occurredin recentyears.
area. Molnar and Sykes[1969]were able to showthat the Pan- This paper presentsthe resultsof a new analysisof the seis-
ama Fracture Zone (Figures 1 and 2) is an active portion of micity and focal mechanismsof earthquakeswithin the area
 PENNINGTON: TECTONICSOF NORTHWESTERNSOUTH AMERICA 10755

15ON
I I I

IO

!
!
FR?
Oo

J
//J/ I I 4S
85 W 80 75 70 65

Fig. 2. Tectonicfeaturesof the studyarea,asfollows:PFZ, PanamaFractureZone;CFZ, CoibaFractureZone;FR,

'Fossil'
ridge;MR, Malpeloridge;EX, 8-m.y.extinct
spreading
centers;
RF, Romeral
fault(Dolores
megashear);
SMBC,
SantaMarta-Bucaramanga fault; BF, Boconofault. Theseand othermajor featuresshownon the map are from the fol- .
lowingsources:
LonsdaleandKlitgord[1978],.CaseandHolcombe[1977],Irving [1975],Shagam[1975],Case[1974],the
geologicmapsof Panama[Direccion Generalde Recursos Minerales,1976],Colombia[Ingeominas,1976],and Ecuador
[Servicio
Nacionalde Geologio
y Mineria,1969],andP. Lonsdale(personal communication, 1978).

of northwesternSouthAmerica, leadingto a determinationof request.All eventsreportedby the Bulletinof the International

the current tectonic features and permitting speculationon SeismologicalCentre(BISC) within the time peod January
their origins. 1964throughAugust 1976have been considered.Only events
In this paper, the term 'subduction'refersto the descentof with magnitude(mb) -> 4.2 and locatedby reported arrivals
a lithosphericplate into the asthenosphere.'Normal' sub- from at least 10 stations were used. Obvious aftershock se-
ductionimpliesthat the subductedplate is descendingalonga quenceshave been, in general, omitted. (The geographic
dip angleof 35 or moreand that asthenospheric materiallies boundsof the study include the Panama Fracture Zone to the
abovethe plate. 'Shallow'subductionoccursalong anglesless west but not its junction with the Middle America trench nor
than35.'Underthrusting'
isa'term
used
toimplythatthetwo any otherportionof that trench;Figures1 and 2.)
platesare in contactand no asthenospheric materialseparates A recentfield study of the highly active Bucaramanganest
them. This occursin somecasesof very shallowsubductionas of earthquakes[Penningtonet al., 1979]has placed the nest at
well as at the trench in all casesof subduction. 6.8N, 73.1W and at a depth of 158 kin. In the presentstudy,
The tectonicfeaturesdescribed in thispaperwereobtained 230 events associatedwith the Bucaramanganest were not
from straightforwardinterpretationsof the seismicityand consideredindividually but were assumedto be at that loca-
focalmechanism solutions. Speculation on the originof these tion. All these earthquakes were within 6.5N-7.1N,
featuresemploysa modelin whichbathymetri
featuresare 72.7W-73.3W, with depths from 140 to 175 kin. Earth-
consideredto bemorebuoyantthan thesurrounding
oceanic quakeswithin thisvolumehavebeenshownby Dewey[1972]
lithosphere.
Thismodel[Kelleher
andMcCann,
i976;Vogtet to be teleseismicallyindistinguishablefrom eventsat a single
al., 1976] not completely
accepted
amonggeophysicists.
point source.
Whether
bathymetric
features
sucasaseismic
ridges
andex- One deep event occurredwithin the study area at 653 km
tinct volcanicarcsare indeedbuoyantor not is immaterial to depth on July 31, 1970.All other eventswithin the time period
the descriptionof the tectonicfeatures, of the study were shallowerthan 300 kin.
The earthquake locationsused have been assignedone of
TECHNIQUES USED IN THIS STUDY
thre.e quality indices,determinedby the azimuthal distribu-
Selectionof EarthquakeHypocenters tionof theobserving stations
andthequalityof the depthcon-
Hypocentraldatausedin the seismicity
mapsandprofiles trol. The method of assigningqualities is similar to that em-
presentedin this study are available from the author on ployedby Cardwell andIsacks[1978].Specifically, the best-
10756 PENNINGTON:TECTONICSOF NORTHWESTERNSOUTHAMERICA

located group ('A' locations) have stationsreporting from used for obtaining S wave polarizationscan be found in the
threeto four quadrants(very few earthquakesin northwestern work by Herrmann [ 1975].
South America have coveragefrom four quadrants)and, for
earthquakesgreaterthan 80 km deep, at least four consistent
OBSERVATIONS OF SHALLOW SEISMICITY
pP readings or local observationswith small residuals,for
AND FOCAL MECHANISMS'
which the depth/distance ratio from the nearest station was
PLATE BOUNDARIES
greater than or equal to 0.6. 'B' locationswere assignedto
eventswith azimuthal coveragein three quadrants,and two
consistentpP readingsor a depth/distanceratio for the near- PanamaFractureZone and NeighboringAreas
eststationof 0.2-0.6. 'C' locationshad stationsreportingfrom The distributionof earthquakes
(Figure3) near82.5W and
fewer than three quadrants,no consistent pP readings,and a north of 4.5N clearly defines the Panama Fracture Zone
depth/distanceratio for the neareststationof lessthan 0.2. (PFZ) as a seismicallyactivefeature.Focal mechanismswere
The pP-determineddepth was chosenwhen it differed from obtainedfor sixof the events(4-9, Figure4), and demonstrate
the depthdeterminedfrom P readingsalone.The depthcon- north-southfight-lateralstrikeslip faulting,as expected.(All
straintswere relaxed for eventsshallowerthan 80 kin. If but one of the eventsrequiresomesmallcomponentof east-
nearby stationswere assignedlarge residualsby the location westcompression or a nodalplaneorienteda few degreeseast
procedure,the quality of the event was further downgraded. of north. However, the peculiar crustal and upper mantle
Eventswith locationsdeterminedby reportsfrom fewer than structurenear the fracturezoneand alongray pathsto North
10 stations,or with magnitudeslessthan 4.2, were rejected. America,which passthroughthe subductedCocosplate be-
Roughly one third of the 665 hypocentrallocationsused fell neath Central America, may resultin the lateral refractionof
in each of the three categories. rays to producethis effect.)
The quality of the locationswith an A classificationwas Althoughthe PFZ is clearlythe currentCocos-Nazcaplate
testedby comparisonwith a group of eventswhich had been boundarysouthof Panama,there is an ongoingtransferof
relocatedby Dewey(1972).For all eventsin our A group,and motionfrom the PFZ to the next fracturezoneswest,as sug-
to whichDeweyalsoassigned
an A .rating,
the differences
in gestedby vanAndelet al. [1971].Events1 and 41 (Figure4
depthsand epicentralcoordinateswere negligible(_+0.1o, _+5 and the appendix) representfaulting similar to that on the
km in depth).(For shallowevents(lessthan 80 kin) the depth PFZ but are on extensions of fracturezonesseparatingthe Ec-
determinationsare not reliable in either case.)The locations uadorrift fromthe Galapagos and CostaRica rifts(Figure1).
in the B groupsdiffered by up to 30 km in depth and 15-20 Both events also lie along or near the flanks of the Cocos
km in epicenters.The differencesin the locations of the C ridge. Van Andel et al. [1971] suggestedthat the transform
groupwere up to 45 km in depth and 30-40 km in epicenters, boundarybetweenthe Nazca and Cocosplatessouthof Pan-
with someeventshavingeven greaterdepth separations. ama has repeatedlyjumped westward,transferringmaterial
from the Cocosplateto the Nazcaplateand causingsegments
Focal Mechanisms
of spreading centers to become extinct. The occurrence of
Focal mechanisms for 56 eventswithin the studyarea (and theseearthquakeson the extensionsof ridge-ridgetransforms
one additionaleventnearby)were coRRected from the existing is consistentwith the suggestion
of vanAndelet al. [1971]that
literature(24 events)or determinedfor thisstudy(33 events), thesejumps occurfrom the ridge-trenchtransformto ridge-
summarizedin the appendix.Ten eventshave solutionsdeter- ridge transforms,rejuvenating the segmentsbeyond the
mined by more than one study;the best-constrainedsolution spreadingcenters,where they exist as major zonesof weak-
is the one used.In most cases,the multiple solutionsfor indi- ness.The currentjump may be occurringin responseto the
vidual eventsare quite similar. Some of the earlier studiesre- pluggingof the subductionzone by the collisionof the Cocos
lied in part on reported readingsof first motionsor on short- Ridge with the Middle America trench.
period records.New solutionsdeterminedfor this studyused A small spreadingcentermay exist with the fracture zone
long-periodWorld-Wide StandardizedSeismograph Network separatingthe Ecuador and Costa Rica rifts. Event 2 is within
(WWSSN) film chipsalmostexclusively,with only occasional this fracturezone and indicatesnormal faulting, suggesting
short-periodrecordsfor nearby stationsfrom small eventsand that the fracturezone is a 'leaky' one, and small spreading
then only when the first motion was unambiguous.Only first centerswithin it havenot yet beenrecognized.In this paper,
motions read by the author were used in the new determina- the fracturezone is representedas two transformfaultswith a
tions.Recordsof the high-gainlong-periodstationswere used smallspreadingcenteroffsettingthem at the locationof event
for some recent events. 2.
The nodal planeswere fitted by eye and constrainedto be Lonsdaleand Klitgord [1978]have suggested that the east-
orthogonal.(For many of the eventsfor which nonorthogonal ern Panamabasinis movingas a smallplate, somewhatinde-
planes would have provided a better fit, the use of non- pendentlyfrom the Nazca plate. One focal mechanism(re-
standardcrustaland upper mantle velocitieswould have per- portedby Molnar and Sykes[1969])supportstheir contention,
mitted an orthogonalfit.) Most solutionsare primarily con- althoughthere is very little seismicityin the regionbetween
strainedby P wavefirstarrivals;someare supplemented by $ the Cocos-Nazcaplate boundaryand the Colombia-Ecuador
wave polarizations,as indicatedin the diagrams(seethe ap- trench. Event 3 lies south of the Costa Rica rift and is not
pendix). Angles of incidence were determined by inter- along any recognizedmajor fault. Its mechanism demon-
polation from the Hodgsonand Storey [1953] extendeddis- strates left-lateral strike slip faulting along a north-south
tance tables,basedon the Jeffreysand Bullen[1940]velocity plane(the senseof motionis oppositeto that of the PFZ) or
structurefor the earth. A crustalvelocityof 6.5 km/s was as- right-lateralfaulting along an east-westplane. The tectonics
sumedfor earthquakeslessthan 33 km deep.The generalpro- here may be complicatedby the proximity of active and ex-
cedure for determination of focal mechanisms and the method tinct ridges,transforms,and the Colombia-Ecuadortrench.
 PENNINGTON:
TECTONICS
OF NORTHWESTERN
SOUTHAMERICA 10757

Fig. 3. Shallow-focus
seismicity
(depthlessthanor equalto 80 km).LargecirclesareA locations;
smallcirclesare B
locations;
C locationsand eventsrejectedfrom the studyare not shown.Bathymetricand topographic
contoursare from
Figure 1.

Colombia-Ecuador Trench Panama-ColombiaBorder Region

There is scatteredand diffuseseismicityin the area of Pan-
The trench is the site of active underthrustingof the Nazca
plate beneathSouthAmericafrom the Panama-Colombiaarea and the Panama-Colombiaborder.Apparent lineationsof
border to where it joins the active Peru-Chiletrench (the eventsand regionaltectonicconsiderations have led earlier
southernlimit of the study).Seismicityis greatestnorth of workersto proposean activetrenchnorth of Panama [Bowin,
5N, and mostof the largesteventsduringthe studyperiod 1976]or a transformfault southof Panama[Jordan,1975].
occurred since 1970. Events 17-25 are located at or near the The focal mechanismsreportedin this study,however,show
Colombia-Ecuadortrench and, with the single exceptionof that there is no singleboundary and that east-westcompres-
event 21, have at leastone nodal plane which is parallel or sion is being accommodated along many faults strikingNW
nearlyparallelto the trench.Of theseeightevents,four are to NE. The principalaxesfor events10-16 are includedin
normalfaultingand four are thrustfaulting;both typesof Figure4 on an equal-areastereonetand demonstrate that the
mechanisms are typicalfor trenchand near-trenchenviron- greateststress
is east-west
compression,
with a north-south
in-
ments,but only two eventsare of the shallowunderthrust- termedlate stressaxis. Most of the motion being taken up by
typetypicallyassociatedwith convergentmargins.The recent earthquakesin and aroundPanamais east-west,and signifi-
(December 12, 1979)Tumacoearthquake of M, -- 7.9had a cantlylessmotionis takingup north-southcompression. Pan-
focalmechanism[Herd et al., 1981]very similarto, and was ama is a complexregionbetweenthe Caribbeanand Nazca
locatedonly 75 km SSW of, event23 (Figure4), demonstrat- plates.Thesetwo plateshave been estimatedto be moving
ing thatovera longertimescalethanthatof thisstudy,this eastwardwith respectto South America at 5-6 cm/yr and
coastof Colombia exhibitsa more typically convergentbe- converginguponeachotherat lessthan 2 cm/yr [Minsterand
havior.The thrust-faultingeventslikely occurat the bound- Jordan,1978],a conclusionat leastqualitativelysupportedby
ary betweenthe underthrust plates;somemay this study.
andoverthrust
Polygonalfracturingof Panamahasbeendescribed
occurwithin the deeperportionsof the oceanicplate near the from
trench,whereit is beingcompressed dueto bending.Normal- observationsof radar imagery [Wing and MacDonald, 1973]
faultingeventsare believedto occurin the upperportionof and geologicstudies[Terry, 1956],and is consistentwith the
the bendingslab,whereextensional may dominate focal mechanismobservations.A similar pattern of faults has
stresses
[Stauder,1968].Extremelyaccurateearthquake depthdetermi- beenrecognizedin Jamaica[Burkeet aL, 1978b],wherethey
nationsfor eventslessthan 80 km deep,necessaryfor identi- occurin response
to shearingalongthe North American-Car-
areoutside ibbeanplateboundary.In Panama,the shearmotionbetween
fyingthelocationof eventswithinthelithosphere,
the scopeof this study. the Caribbean and Nazca platesis coupledwith major corn-
10758 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

15N
i I

10--16

I0

29 30

0o

27

26 -35
26

4S
8.5 W 80 7.5 70 65

Fig. 4. Focalmechanism solutionsof shallow-focusearthquakes,

depthslessthanor equalto 80 km. Shadedquadrants
representcompressional seismicwave arrivals;open quadrantsrepresentdilatationalarrivals.Hachuredquadrantsare
compressional,
for poorly constrainedsolutions.The T axesare indicatedby white dots,and P axes by black dots. The
insetsshowP, T, and B axes(small crosses)for selectedsubgroupsof events.The focal mechanismsand insetsare lower
hemisphereprojectionequal-areaplots.

pressiveforcesassociatedwith the convergenceof the Nazca tectonicsare not readily understood.The seismicityis scat-
and South American plates, and all the seismicallyactive tered, and alignmentsunclear (Figure 3). Focal mechanism
faultsappearto be dominantlythrusting. solutionsfrom body wave arrivalsexistfor three events(36-
38, Figure 4) which may representinteractionsat the inter-
ContinentalEarthquakes sectionof boundaries.Event 36 may representeastwardmo-
An alignment of shallow hypocentersto the east suggests tion of the Caribbean plate with respectto South America.
that a major boundary followsthe easternfront of the Andes Molnar and Sykes [1969] suggestedthat event 38 occurred
(Figure 3). This alignment is not convincingby itself, so we along the east-westplane, but lial [1978]has shownthat mo-
usethe focal mechanismsfor eventsalongthe boundary(26- tion occurredalong the north-southplane. Two other focal
35 in Figure 4). These events have one nodal plane each mechanism
solutions(39 and 40, Figure4) are not readilyin-
which is nearly parallel to the NE strike of the hypocentral terpretedin termsof large-scaletectonicsand may resultfrom
alignmentand the Andean front and is likely to be the plane locally complicatedstresspatternsnear the intersectionof the
Eastern Andean Frontal Fault Zone and the southern Carib-
along which slip occurred. Six of the 10 events are pre-
dominantly right-lateral strike slip, and four are pre- bean margin.
dominantly thrusting;all are consistentwith east-westcom- OBSERVATIONS OF INTERMEDIATE-DEPTH SEISMICITY
pression.The earthquakesdefining the boundary seemto be AND FOCAL MECHANISMS:
along severalfaults, and events33 and 30 showthat motion is SEGMENTED SUBDUCTED LITHOSPHERE
occurringboth within the Andes and the SouthAmerican cra-
ton (Guyana shield) as well as along the easternfront. This Seismicityand Focal Mechanisms
systemof faults along which the thrusting and strike slip The intermediate-depthseismicity of northwesternSouth
earthquakesoccurwill be referredas the "EasternAndean America is clusteredinto three distinctzones(Figure 6), inter-
Frontal Fault zone' (Figure 5). . preted as representingsegmentsof subduetedlithosphere.
East-west transcurrent faults are to the northeast of the !Thesesegments will be discussed
separately.Figure7 shows
Eastern Andean Frontal Fault Zone in the region of 10N, depth contoursof the Benioff zones beneath northwestern
70W[1od,!956; MolnarandSykes,1969].The intersectionSouthAmerica.Figure 8 showsverticalcrosssectionsof the
of thesetwo setsof faultsformsthe only part of the studyarea seismicitywithin each segment.
in which the shallowseismicityis difficultto interpretand the The orientationof major stressaxesof intermediate-depth
 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA 10759

15N
I I I

IO

0o

GUYANA SHIELD

I I 4os
85 W 80 75 70 65

Fig. 5. Summarydiagramof theseismotectonics of northwestern SouthAmerica.All of thefocalmechanism solutions

for eventslessthan 80 km deepoccuralongoneof the majorfeaturesshownhereor are individuallyrepresented by fault-
ing symbols. Normal faultingis shownby an opencircleand arrowspointingoutwardalongthe axisof tension;thrust
faultingis shownby a solidcircleand arrowspointinginwardalongtheaxisof pressure; strikeslipfaultingis shownby a
singlecouplewherethefaultplanehasbeenassumed, otherwise bothcouplesareshown.Individualeventsare numbered.
The brokendoublelinesstraddlingthe strikeslip and thrustingsymbolof the EasternAndeanFrostalFault Zone are in-
tendedto conveythat motion is accommodated over a zone of faults,not a singlefault.

earthquakes
canprovide
additional
evidence
for theorienta along or near the plane of seismicity,and the T axesare ap-
tions of subducted lithospheric plates. Isacks and Molnar proximatelydowndip.Two other eventsoccurredwithin the
[1971],in a globalstudywhich includedfive eventsfrom Ec- Bucaramanganest(45 and 47, Figure 9); for these,the T axes
uador and Colombia,demonstratedthat tensional(T) axesare were not downdip but were oriented along the strike of the
generally oriented parallel to the dip directionsof the sub- subductedplate, parallel to the B axes of the other three
ductedplate containingthe eventsfor plateswhich penetrate events(as noted by Isacksand Molnar [1971]for one of the
to depthslessthan 300 km, asin northwesternSouthAmerica. events).The significanceof this is not clear but must be re-
Figure 9 .summarizes15 focal mechanismsolutionsfor inter- lated to whateverfeatureis responsiblefor the existenceof the
mediate-depthearthquakesin this area. nest itself.
Bucararnangasegment. The northernmost zone, from At the northernmost end of the Bucaramanga segment,
about 5.2N to 11N, containsthe Bucaramanganest, treated event 42 exhibits a T axis located within the plane of the seg-
in this study as a point source.The seismicityoutsideof the ment but which is neither downdip nor along strike. It seems
nest is sparse;yet the alignment of hypocenterssuggestsa likely that lateral stresses
of the samemangitudeas the gravi-
singleBenioff zone (Figure 8). The subductedsegmentcon- tational (downdip)stresses may be setup near the edgeof a
taining the eventsmay be somewhatmore complicatedthan subductedplate, and event42 may representa combinationof
pictured,but it mostlikely dips at 200-25 toward N109E. those two stresssystems.There may also be some unrecog-
(Precisionto 1o is not implied; this is simply the orientation nized distortionof the plate at its northern edge which is re-
which producesthe best alignment of hypocenters.)It may sponsiblefor the peculiarorientationof the focal mechanism
steepensomewhatat depthsgreaterthan 140 km. The sub- for event 42.
ducted lithosphereis apparently continuouswith the Carib- Cauca segment. The intermediate-depthseismicitysouth
bean seafloor northwest of Colombia. of 5.2N definesa pieceof subductedlithosphere,called here
Two earthquakeswithin the Bucaramanganest and one the Caucasegment. The subducted pieceis apparentlycontinu-
nearby (43, 44, and 46, Figure 9) have nearly identicalfocal ous with the oceanic crust underthrustingSouth America at
mechanisms.The P (pressure),T (tension),and B (null) axes the trenchfrom the Panama-Colombiaborder (7N) to 1.5N.
for theseeventsare plotted togetherin an insert to Figure 9, The segmentis dippingat 35 towardN 120E,as seenin Fig-
alongwith the planeof the Benioffzone.The T and B axeslie ure 8.
10760 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

15N
1 I

I0

BUCARAMANGA
SEGMENT

CAUCA
SEGMENT

[-i [] 81-120
,0 121-160
,,A i61-200
O0 20i +
,, 65:
ECUADOR
SEGMENT

I 4S
85 W 80 75 70 65

Fig. 6. Intermediate-depth
seismicityof northwestern
SouthAmerica;hypocentraldepthsare indicatedby type of
symbol.The largersymbolsareA locations;smallersymbolsareB locations.
The threesegments
of subducted lithosphere
are labeled, and the lines of projectionusedin Figure 8 are shown.

15ON
I I I

i0 o

BUCARAMANGA
SEGMENT

CAUCA SEGMENT

0o

ECUADOR SEGMENT

4S
85 W 80 75 70 ' 65

Fig.7. Contours
to theBenioffzonesbeneathnorthwestern
SouthAmerica(labeledin kilometers).
Thethreesegments
of
subductedlithosphereare identified.
 PENNINGTON:
TECTONICS
OFNORTHWESTERN
sOUTH
AMERIcA 10761

NW SE
16 11 12 15 13 14 33 32 31 34 35 36

lOO-
-O 0
o
0
c OOC
0
o% o1

T -

BUCARAMANGA SEGMENT
100 KM---

300 I I , I i I I I i I I I , '1 I I i I [ I I

NW SE
I,___7-
23 um

' I0o ' o I 'Oo--

-3_'
35o
0 o 0o

48 49 50 -200-
0
. _

CAUCA SEGMENT ECUADOR

SEGMENT
, I I I I I I , -300-

Fig. 8. Crosssections
of theseismicity
of thethreesegments
of subdUCted
oceaniclithosPhere,
includingthe seismicity
oftheoverlying
continental
lithosphere.
Novertical
orhorizontal
exaggeration.
LargerCir:leS
areA andB quality
loCa-
tions.The positions
of volcanoesasprojectedontothe line areshownat t-hetop of eachdiagram,wheretheyexist.Num-
beredeventsreferto individualfocalmechanism solutions. The Bucaramanga nestis shownasa solidsymbolin the top
diagram.

Two earthquakes,locatedvery near eachother and clearly Five earthquakes with'mthe Ecuadorsegmentexhibit
within the Caucasegment,havefocalmechanisms (49 and 50, downdiptension,
with B axesalongthe strikeof the Betdoff
Figure9) summarizedin an insetin Figure9. The T axesfor zone(events52,56;FigUre9), The remarkableconsistency
of
these eventsare clearly downdip, and the B axes are nearly the-orientationof axesfrora the eventsat various depthspro-
horizontal along the strike of the segment. vides reassurance:that:the :Ecuadorsegmentis, indeed, dip-
Bucaramanga-Caucahinge fault. Event 48 occurred at ping to the northeast,(Event 51 is the northernmostevent of
117-kindepth at the northernedgeof the Cauca segmentand theECuador
Segment;
andlikethenorthermost
Bucaramanga-
exhibited nearly purely strike slip faulting (Figure 9). The segmentevent,it isnot easilyexplained.)This segmentis ap-
strike of one of the nodal planes(110) is nearly parallel to parentlyConfmuous with the fiat-lyingPerusegment[Bara-
the dip directionsof the Bucaramangaand Cauca segments zangiand Isacks,1976,1979].to.the south.
(109 and 120, respectively)
and is mostreadilyinterpreted
asthe fault plane,with the shallow-dippingBucaramangaseg- Alte.mate
Interpretations.
ment moving past the more steeplydipping Cauca segment. It isPOssible
to"interpret
'the.
in.termediate,depth
seismicity
This eventprovidesindirectevidencefor the physicalcontinu- of theEcuadr'CaUca2Bucaramanga
regions
'in a different
ity of the Bucaramangasegmentwith the crust of the Pan- manner,
althougha 'nl--urn-
.bet
of:problems
arise.TheCaucaand
amanian isthmus. Bucaramangasegments maybe interpreted
asonecontinuous
Ecuador segment. The seismicityat intermediatedepths segment
withsomease.'mmic
regions
Withinit. The 'hinge
beneath Ecuador exhibits a remarkable feature: the subducted fatlt'
'event asaninconsistent
plate, as defined by the hypocentraldistribution(Figure 8), event;the segment must"becontorted'inthoseregionswhere
dips at 35 toward N35E. The NNE dip of the segmentis the seismicity.definesthe Betdoffzone and becomesless
wholly unexpectedsincethe Nazca plate ispresumedto move steeplydiPp'mgto the-north.-The.ECuadorsegmentmaybein-
nearly due east relative to South America beneath southern terpretedasa tear-between thefiat Perusegmentanda more
Ecuador and the subductedportionsbeneath Colombia (the steeply/dipp.ingsegment to the north,whichwouldincludeall
Cauca segment)and Peru dip generallyeastward[Stauder, the events.atdepthsgreaterthan 150 kin. In this inter-
1975;Barazangiand Isacks, 1976]. pretatio.n,aH of the focalmechanisms mustbe interpretedas
10762 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

15N
I I

IO

43, 44, 46

5o

49, 50

0o

52-56

I 4S
85 W 80 75 70 65

Fig. 9. Focalmechanism solutions for intermediate-depthearthquakesbeneathnorthwestern

SouthAmerica.Symbols
aredescribedin Figure4. The insetsincludethe planesof the subducted
lithospheresegments
asdeterminedfrom the seis-
micity.

hingefault events,with the implicationthat the steepsegment 1956]are locatednear the Caribbeancoastalongthe Panama-
was, in fact, swingingup to a lesssteepdip. Colombia border. These suggestthat at least a part of the
Panamanianisthmusrepresentsan ancientislandarc (or some
DISCUSSION
other volcanicterrane), which may have resultedfrom con-
The previoussectiondealt with observationsof seismicity temporaneous subductionof the Panamabasinfrom the south
and focal mechanismsand their interpretationin terms of re- or of the Caribbean seafloorfrom the north. The buoyancyof
gionaltectonics.The followingsectionpresentsspeculationon bathymetricfeaturessuchas aseismichot spottracesor vol-
the originsof the tectonicfeatures.Thesespeculations depend canic island chains appearssufficientto resistnormal sub-
on the recognitionof variousfeatureswithin the Panama ba- duction [Kelleherand McCann, 1976; Vogt et al., 1976].It is
sin, especiallyas they intersectthe trench. The shallow sub- likely that the volcanicfeaturewhich comprisesa portion of
duction of oceaniclithosphericsegmentsis related to the pres- Panama, no longer active as an island arc, and the buried
ence of aseismic ridges on them. The aseismic region aseismicridge have been carried along on the Nazca (Faral-
separatingthe Cauca and Ecuador segmentsis a result of lon) or Caribbeanplate and partially subductedbeneathCol-
plate separationalong a zone of subductedextinct spreading ombia. Due to its buoyancy,normal subductioncouldnot oc-
centers. The width of the Andes is correlated with the shallow cur, and the plate directly underthrustthe South American
subductionof plate segments.The existenceof a separate'An- lithosphere.Active underthrustingof the oceaniccrustat the
dean block' is due to the collisionof the Carnegieridge with trench then ceasedor was drasticallyreduced,and the north-
the Colombia-Ecuador trench. westernmostportion of Colombia is now at least looselyat-
_

tachedto the Caribbeanplate. This model can be contrasted

$ubductionof theEasternPanama Basinand with that of Burke et al. [1978a], in which it is proposedthat
the Causesof Shallow$ubductionand Tearsin theLithosphere subduction of the Caribbean ceased due to the presumed
The shallowdip of the Bucaramangasegmentand the lack buoyancyof the entire Caribbeanseafloor.
of present-dayseismicityalongthe northwestern coastof Col- It appearsthat the Bucaramangaand Caucasegmentsrep-
ombia may both be due to the collisionof Panama and/or a resentportionsof one continuousplate which wastorn in two
buried aseismicridge in the Colombian basin 100km north of during subduction.The Caucasegmentis beingsubductedbe-
Panama, with South America. The ridge has been described neath Colombia in a manner typical of many trench regions,
by Houtz andLudwig[1977];its originis not known.Although at leastwhere the seismicityis sufficienttO definethe Benloft
the basementof most of easternPanama is made up of typical zone.
oceanic crust [Case, 1974], several extinct volcanoeswhich Between2N and 2S, the seismicityis sparseat all depths.
were activein the late Tertiary and early Quaternary[Terry, _ The absenceof shallowearthquakescould be attributedto
 PENNINGTON:TECTONICSOF NORTHWESTERNSOUTHAMERICA 10763

I00 o
i
90
i
80' w
quiescence betweenthe great 1906 earthquakewhich may
I0 N
have rupturedthis area [Kelleher,1972]and the 1979earth-
quakeat 1.6N,but that doesnot explainthe paucityof inter-
mediate-depthearthquakes.There may be more than one ex-
planationfor their absence.
1. The crustbeing subductedis extremelyyoung and hot
0o-
andquicklylosesthe strength.
necessary
to produce
earth-
GHS quakes.Near 2N, for example,a spreadingcenterwhichwas
FARALLON activeuntil 8 m.y.B.P. is presentlyat the trench (P. Lonsdale,
personal communication,1978); similarly young crust has
30 m.y.B.P.
probablybeensubducted.
Youngoceanic
crustshouldrequire
I0 S" I I very little time to be assimilatedinto the mantle and flow
plastically.
ION I 2. The presenceof the Ecuador segmentto the southmay
also explain the absenceof intermediate-depthevents,since
the Ecuadorsegmentmay, M itself, accountfor someof the
COCOS lithosphericmateri,al subductedbenea[h northern Ecuador
and southernColombia.In thiscas,the gapin 'totermediate-
depth seismicityreflectsa separationbetweenthe subducted
0 o-
Ecuador
andCaucasegments,
Thisseparatio.n
wouldlikely
have occurredalong the subducted(extinct) Spreadingcen-
ters,which may be expectedto have very little tensilestrength.
PACIFIC
NAZCA In eithercase,thedeepest
(200+km) earthquakes
probably
occurin rather old crustwhich waspart of the Farallon plate.
I0 S --
'' 20m.y.B.P.
I ' I The peculiarorientationof t.he Ecuadorsegmentcan be ex-
plained by a speculativemodel, relating it to the actifity of
IOeN - I I I the Galapagos hot spot. This hot spot is currently located
some 100 km south of the presently active Cocos-Nazca

spreadingcenter.If it had beenactiveprior to breakup of the

Farallon plate, its hypothesizedtracewould lie just outsideof
the Galapagosgore(Figure 1) on ihe crus,t
of the Farallon
0 o-
plate.Figure10 tracesthe possiblehistoryof the hot spot
trace and shows that its current location could be on the sub-
ductedplate beneathsouthernEcuador.This bathymetricfea-
ture could then be an inhibitor of normal subduction,causing
PACI F this plate segmentto be subductedat a very smallanglebe-
I0 S-
, I
\\ NAZCAI0m.y.B.P. i i
neath the Ecuador-Peruborder, as observed[Barazangi and
Isacks, 1976].The subductedlithospherenorth of the ancient
hot spottrace is not buoyant,butis dense,like normal oceanic
I0Nt I i '.te,
i:.,_ I lithosphereat trenches,and is sinking. If extinct spreading
// centers
".-.:...,.:,.;:::'
':',,: havebeen'Subducted
nearIN-2Ni
asproposed
cocos above,theweakness
at thisrelictplateboundfiry
provided
a
I .-'' f/ location for the separationof the plate into two segmentsat
depth.
Southof the subducted
hot spottrace,s,everal
hundredkilo-
meters
of$ceanic
lithoSphere
beneath
Perui being
subducted
at a shallowangle[Barzangi
andIsacks,1976](Figure11).
Theshallow tinglemaybedueto yetanothersubducted bath-

I0

SI I00 90
I 8 W
ymetricfeature;
theNazcaridge.Thisaseismic
ridgeintersects
thetrenchat about15S,bui dueto obliqueconvergence,
providesbuoyancyfrom about 10S-15S
it
[Pilger,1981].
Thus, the segmentbeneathPeru consistsprimarily of normal
oceaniclithosphere
beingheldup betweentwobuoyantfea-
Fig. 10. The historyof the breakup of the Farallon plate into the tures:
inthenorththeancient
Galapagos
hotspot
tracewhere
Cocosand Nazca plates, modified from Hey [1977]. The Galapagos it erupted onto the Farallon crust,and in the south,the Nazca
gore('GG') istheareaenclosed
bythedashed
lines.Thelocations
of
the South American and Central American coastlines are shown for ridge. Southof the Nazcaridge,the tearbetweenthe shallow-
reference
purposes oftheGalapa- subducting
o!aly.Thesoliddots'thelocation' segment anda moresteeplydippingsegment be-
goshotspot.At 30m,y.B.P.,Hey[1'977]hadassumed thehotspotto neathsouthernPeruis sharp[Barazangiand Isacks,1976].
liebeneath
theOpening fracture
zone..Inthis
study,
thehotspotis Th0 deep focusearthquakeof 1970(event 57) presumably
proposed'
tohavebeenlocated justsouthofthefracturezone,creating lieswithin
,lithosphere
whichwassubducted priortothe'shaI-
a bathymetricfeature,whichis traced'rough Hey'srccOn.
structionto
thepresent, whenit should
liebeneath
Ecuador.
ContourstotheBen- lowing"of theperusegment. Mendiguren andAki [1978] note
ioffzone',beneathColombia and Ecuador
areshownfor thepresent ihat the 1970rupturepropagatedat least153km in the direc-
time, from Figure 7. tion of an activedeepfocusseismicregionin Braziland con-
10764 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

There is no volcanismassociatedwith the shallow-dipping

I0 N Bucaramangasegment,even above the portions of the slab
that reach depthsof 150-200 kin. This could simply be due to
the brief time that the slab has occupiedits presentposition,
0 o
sincethere is almost certainly a time lag of one to a few mil-
CARNEGIE
lion years before eruptions break through the surface. The
Bucaramangasegmentmay have movedto its presentposition
I0 S
quite recently, and any magma chamberswhich could give
rise to volcanismmay not have yet reachedthe surface.
The past locationsof subductionzonesand the configura-
tions of the subductedplate may be inferred from the timing
of earlier volcanism.For example, early Cenozoic intrusives
and volcaniclasticdeposits,north of 5N, at 74-76W, sug-
gestthat the Bucaramangasegmentwas then subductingnor-
30 mally and extendedto depthsof 100-200 km beneath what is
now the Central Cordillera of northern Colombia.
The width of the northern Andesmay also be related to the
shallow subductionof the Bucaramangasegment.The uplift
of the northern Andes began in the Pliocene,roughly at the
same time as the collision with the feature which is now Pan-
ama and, possibly,simultaneouslywith the initiation of shal-
low subductionfollowing the earlier normal subductionof the
50 Bucaramangaslab. The width of the northern Andes is great-
est where the subductionangle is the least. This is also ob-
served in Peru, where shallow subduction correlates with a
broad zone of recent deformation and uplift [Barazangiand
90 W 70 50
Isacks, 1979]. Similarly, shallow subductionbeneath the west
coastof North America is frequentlycited as the causeof the
Fig. 11. Map of South America showinglocationsof volcanoes Laramide Orogeny, when the mountain building progressed
(activeor showingfumarolicactivityin the recentpast)as soliddots, hundreds of kilometers inland from the trench.
from the VolcanologicalSociety of Japan [1971], and the 150 km The northern Andes constitutea separatecrustal block, or
depthcontourto the Benioffzone,determinedfor northwesternSouth
America in this study and for Peru-Chile by Barazangi and Isacks
small plate, which is being compressedand underthrustfrom
[1976].A bold line indicatesthe locationof a major tear in the sub- the eastby the South American plate and from the westby the
ductedlithosphere. Nazca plate and is 'pinching out' to the NNE. Since the con-
vergenceof the South American and Nazca plates is not per-
pendicular to the Colombia-Ecuadortrench, perpendicular
cludethat the subductedlithospherecontainingtheseeventsis
thrusting at the trench and transcurrentfaulting behind the
continuous beneath Colombia and Brazil. If this is true, then
trench may be expected,as observedin the southwestPacific
that subductedlithospherelies continent-ward and beneath
[Fitch, 1972]. It is tempting to invoke an additional driving
the entire length of the fiat-lying Peru segmentand its north-
force for the pinching out of the Andean block: the collision
em edge,the Ecuador segment,and must representa portion
of the Carnegieridge with the Ecuadortrench.As it resistssub-
of the subductedFarallon plate which has torn off from the
duction, the Carnegie ridge may be providing compressive
Peru segmentand continuedsubducting'normally.' The rela-
stressesin addition to thoseresultantfrom typical subduction
tively aseismicarea separatingthe Colombianand Brazilian which are transmitted across the trench to the continental lith-
portionsis probablymorecompletelyassimilated
into the sur-
osphere.This could, in a manner similar to the 'rigid indenter'
roundingmantle (althoughnot completely,for it still behaves
model of India-Asia collisionproposedby Molnar and Tap-
as a stressguide) than the seismicallyactive portions. This
ponnier[ 1975],resultin increasedeast-westcompressive forces
variation is assimilationrates can be explained by varying
and NNE displacementof the Andes.
agesof oceaniccrust at time of subduction,differing crustal
properties,heatingrates,or even heterogeneities
within the SUMMARY
surroundingmantle.
Analysis of seismicityand earthquakefocal mechanisms
has led to the determinationof plate boundaries(both sharp
$eismotectonicsof NorthwesternSouth4merica, and diffuse)and recognitionof segmentsof subductedlithos-
the Widthof the Northern4ndes,and the pheric plates in the region of northwesternSouth America.
4bsenceof VolcanismOver Theseconclusionshave in turn led to speculationon the ori-
Shallow-DippingBenioffZones gin of the segmentationof subductinglithosphereand pro-
Figure 11 showsthe location of the active volcanoeswithin vided indirect evidencefor the timing of the inceptionof Ga-
South America and the 150-km depth contour of the sub- lapagoshot spot activity in relation to the breakup of the
ductedplate. Active volcanoesexistonly over lithospherethat Farallon plate.
is berg subductedat anglesof 35 or more, and the endsof The following conclusionscomedirectly from the observa-
the volcanic chains are located over major tears or bends in tions of seismicityand focal mechanisms(seeFigure 12):
the subductedplate segments. 1. The PanamaFractureZone is a major plate boundary,
 PENNINGTON:
TECTONICS
OFNORTHWESTERN
SOUTH
AMERICA 10765

9O 8O 7O 60

NORTH AMERICA

20 20 N

CARIBBEAN
x

I0 o I0 o
COCOS
VENEZUELA

COLOMBIA

ECUADOR
,/
/
/

PERU

SOUTH AMERICA

NAZCA

90 W 80 70 60

Fig.12. Majortectonic
features
of northwestern
SouthAmerica
andneighboring
areas,
fromrgure1 andtheresults
of thisstudy.
Theboundaries
nearthenorthern
edgeoftheAndean
blockandthesouthern
Caribbean
arestillunclear;
the
regionsof currentseismicity
at thisjunctionare shownby smallcrosses.

separatingthe Cocosand Nazca plates.Nearby seismicactiv- 3. TheColombia-Ecuador trenchisthesiteofcurrent

un-
ity supportsthe hypothesis[van Andel et al., 1971] that the derthrustingof the Nazca plate beneathSouthAmerica.
plateboundaryisjumpingwestward,reactivatingextinctarms 4. The Andeanblock is separatedfrom the restof South
of fracture zones. America by the Eastern Andean Frontal Fault Zone. It is
2. The Isthmusof Panamaand neighboringareasform a moving NNE with respectto the rest of South America and
diffuseplate boundarywith thrustingalonga systemof faults being compressedin an E-W direction.
striking NW to NE. This zone of deformation accommodates 5. The Bucaramanga
segment
of oceanic
lithosphere
con-
(primarily) the east-westshorteningresultingfrom the con- tainsthe Bucaramanga nestof intermediate-depthearth-
vergenceof the Nazca and Caribbean plates with the South quakes.The segmentdips at 200-25 toward N109E and is
American plate and (secondarily)north-southcompression apparently continuouswith the oceaniccrust of the Carib-
betweenthe Nazca and Caribbeanplates. beanseafloornorthwestof Colombiaand the zoneof defor-
10766 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

mation
inandnearthepanamanian
isthmus.
6. The Caucasegmentis separatedfrom the Bucaramanga
segmentby a tear in the plate and dips at 35 toward
N120E.The subducted
caucasegment
is continuous
With
the Nazca plate currently underthrusting South America at
the Colombia-Ecuador trench.
7. The Ecuadorsgment,beneathcentraland southern
Ecuador,
dipsat35towardN35Eandisseparated
fromthe
Caucasegment
by a zoneo'verysparse
intermediate-depth
seismicity.
TheEcuador
Segment
is apparently
theNE dip-
ping northern edge of the plate segmentsubductedbeneath
northernPeru,whichgenerallydipsto the eastat a very small
angle.
The origin of some of these features,particularly the seg-
mented nature of the subductedlithosphere,is a matter of
speculation.Kell'eherand McCann [1976]and Vogt et al.
[1976]
have
proposed
thatthbuoyancy
ofcertain
b,athymet-
ric featurescan inhibit the subductionprocess,but they were
unable to apply theseprinciples satisfactorilyto the South
American subduction regime. Although the .buoyancy of
bathymetric
features
isnotproven,
thefollowing
arehypo-
theticalexplanationsfor the observedfeatures,making useof
the buoyancyhypothesis:
1. TheBucaramanga
segmen
t most
likelycontains
a por-
tion of a volcanicfeature,probably an ancientislandarc, that
also exists on the Panamanian isthmus as well as an aseismic
ridge north of Panama. These features are resisting sub-
ductionand resultingin both the shallowangleof the Bucara-
manga slab and the diffuseplate boundaryrepresentedby
Panama and nearby areas.
2. Because
theCauca
segmen
doesnotcontain
anysub-
ducted
aseismic
ridgeorislandarc,it isbeingsubducted
in a
normal manner.
3. The Ecuadorsegmentconttinsa subductedhot spot
trace,whichresultedfrom the activityof the Galapagoshot
spotpriorto thebreakupof the Farallonplateinto the Cocos
andNazcaplates.Thistracewascreatedon crustof theFaral-
lon plate southof the Grijalva Fracture Zone and therefore
lies outsideof the Galapagosgore.
4. The plate beneathPeru is held up by two aseismic
ridges,onebeneathEcuador(discussedabove)andthe Nazca
ridge at 10-15S.
5. A zone of weaknessexisted200-400 km north of the
subductedaseismicridge beneath Ecuador,where extinct
spreadingcentersare beingand havebeensubducted.This
provideda line alongwhichthe EcuadorandCaucasegments
separated.
Rather
thanseparate
bytearing,
thenormal
oce-
anic crust,whichis subductedbetweenthe asiesmicridgeand
the spreadingcentersand which is pulled away from the
Caucasegment, is descending into the asthenosphere
(as in
normalsubduction),held up by the subductedaseismic
ridge
to the SW.
6. TheAndean
blockis beingdrivento theNNE by the
collisionof the Carnegieridge with the Colombia-Ecuador +++ I I +++++ I I
trench.

APPENDIX
FOCAL MECHANISMS USED IN THIS STUDY

Table A1 listsall eventsfor which body wave focal mecha-

nismsexistandwereusedin thisstudy.FigureA 1 shows
focal <o<Z
mechanism solutionsdetermined in this study.
PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA 10767
10768 PENNINGTON:
TECTONICS
OF NORTHWESTERN
SOUTHAMERICA

N
N

F' 5 17 JUL 76 E' 6 15 JUN 68 E 9 12 AUG 72 E I0 20 JAN 71

E 12 II JUL 76 E -I$ 13. JUL 74 E 14 21 NOV 75 E 15 25 JAN 75

/' ,l
E 16 31 AUG 77 E 17 27 SEP 70 E 18 26 SEP 70 E 19 16 DEC 70
.. -

E 20 ii NOV 76 E ?_2 .2:4 APR 73 E 23 o AUG 76 E ?_4 09 APR 76

-.
..
.

i I X
t

E 25 I0 DEC70 E 28 06 OCT 76

E 29 09 FEB67 E 0 27 SEP74

Fig.A1. Plots
ofallfocal
mechaffism.
Slutins
determined
inthisstudy.
Diagrams
arelower
hemisphere,
equal-area
projections;
solid
symbols
repreSent
cmpreSSional
firstarrivals;
opensymbols
represent
dilatational
firstarrivals;
crosses
represent
first
motions
ofuncrtpolaritY;
Short
linesare$ wavepolarizations.
PandTaxesareindicated;
Baxesarethe
intersections
ofthenodalplanes.
N.dal
Planeswereconstrained
tobeorthogonal;
somesolutions
wouldexhibit
anim-
proved
fitif nonstandard
crustal
andupper mantle
seismic
velocities
wereassumed.
(AJeffreys-Bullen
[Jeffreys
andBullen
1940]
velocitystructure
was aSsUmed,
i:tha Crustal
velocity
of6.5km/s.)
Numbers
refer
toTable
A1.
 PENNINOTON:
TECTONICS
OFNORTHWESTERN
SOUTHAMERICA 10769

N N N

E 32 27 JAN 70 E $6 05 APR75 E 39 20 OCT69 E 41 29 MAR 76

E 42 17 NOV 68 E 45 30 AUG73 E 47 08 JUL73 E 48 24 APR 73

N N

E 49 03 APR73 E 50 19 MAY76 E 55 17 MAY71 E 54 '28 MAY69

N N

E 55 30 MAY73 E 56 27 JUL 71

Fig. A1. (continued)

,lcknowledgments.
This studybenefitedgreatlyfrom many long NationalScience
Foundation
grantsEAR-77-09582,
INT-77-222240,
discussions
with R. Meyer, W. Mooney,W. McCann, A. Lowtie, and and EAR-76-14840.This studyprobablyneverwould have beenun-
P. Lonsdale.Mooney,Lowrie,Lonsdale,
R. Hey, R. Pilger,and M. dertakenif it were not for the continuinginterestand supportof
Barazangialsokindlyprovidedcopiesof unpublished manuscripts COLCIENCIAS in cooperativeand internationalstudiesof the seis-
and preprints. The ISC bulletins were consulted at the Lamont-Do- micity, structure,and tectonicsof northwesternSouthAmerica, and
hetty GeologicalObservatory
and the Universityof Michigan,and the diligenteffortsof the Institutode losAndesColombianos,Univer-
readingsof the WWSSN film chipswere madeat Lamont, wherethe sidad Javeriana, Bogota. Geophysicaland Polar ResearchCenter
permissionof L. Sykes and the assistanceof E. Wellmon and M. publication 390.
Yamasaki are gratefullyacknowledged.Valuable assistance
in com-
puting was generouslyprovidedby J. Luetgertand B. Karsh of the
REFERENCES
Universityof Wisconsin
Geophysical
Computing
Facility,anda com-
puterprogramto displayvariousprojectionsof hypocenters
waswrit- Atwater, T., Implicationsof plate tectonicsfor the Cenozoictectonic
tenby J. Luetgert.Helpfulreviews
of themanuscript
weregivenby R. evolution of western North America, Geol. Soc. Am. Bull., 81,
Meyer,C. S. Clay,andC. Craddock. Somecomputerfundswerepro- 3513-3535, 1970.
vided by the University of WisconsinGraduate School.The alterna- Barazangi,M., and B. L. Isacks,Spatialdistributionof earthquakes
tive interpretations
of the intermediate-depth
seismicitywere sug- and subductionof the Nazca plate beneath South America, Geol-
gestedby an anomymous reviewer.Thisworkwassupported by the ogy, 4, 686-692, 1976.
10770 PENNINGTON: TECTONICS OF NORTHWESTERN SOUTH AMERICA

Barazangi,M., and B. Isacks,Subductionof the Nazca plate beneath Jordan, T., The present-day motions of the Caribbean plate, J.
Peru: Evidencefrom spatial distribution of earthquakes,Geophys. Geophys.Res., 80, 4433-4439, 1975.
J. R. Astron. Soc., 57, 537-555, 1979. Kelleher, M., Rupture zones of large South American earthquakes
Bowin, C. O., The Caribbean:Gravity field and plate tectonics,Spec. and somepredictions,J. Geophys. Res., 77, 2087-2103, 1972.
Pap. Geol. Soc. Am., 169, 1976. Kelleher, M., and W. McCann, Buoyant zones, great earthquakes,
Burke, K., P. J. Fox, and A.M. C. Sengot, Buoyant ocean floor and and unstableboundariesof subduction,J. Geophys.Res.,81, 4885-
the evolution of the Caribbean, J. Geophys.Res., 83, 3949-3954, 4896, 1976.
1978a. Lonsdale,P., and K. D. Klitgord, Structureand tectonichistoryof the
Burke, K., J. Grippi, and A. C. Sengot,Tectonicstyleof the northern eastern Panama Basin, Geol. Soc. Am. Bull., 98, 981-999, 1978.
boundary of the Caribbean plate (abstract),Eos Trans.A GU, 59, Mendiguren, J. A., Identificationof free oscillationspectralpeaksfor
1203,1978b. 1970July 31, Colombian deep shockusingthe excitationcriterion,
Campbell, C., The Santa Marta wrench fault of Colombia and its re- Geophys.J. R. Astron. Soc., 33, 281-321, 1973.
gional setting,Trans. CaribbeanGeol. Conf 4th, 247-261, 1968. Mendiguren,J. A., and K. Aki, Sourcemechanismof the deepColom-
Cardwell, R., and B. Isacks,Geometry of the subductedlithosphere bian earthquake 1970 July 21 from the free oscillation data,
beneath the Banda Sea in eastern Indonesia from seismicityand Geophys.J. R. Astron. Soc., 55, 539-556, 1978.
fault plane solutions,J. Geophys.Res., 83, 2825-2838, 1978. Minster, J., and T. Jordan, Present-dayplate motions, J. Geophys.
Case, J. E., Oceanic crust forms basement of eastern Panama, Geol. Res., 83, 5331-5354, 1978.
Soc. Am. Bull., 85, 645-652, 1974. Molnar, P., and L. Sykes, Tectonics of the Caribbean and Middle
Case,J. E., and T. L. Holcombe, Geologic-tectonicmap of the Carib- America regionsfrom focal mechanismsand seismicity,Geol. Soc.
bean, Eng. Min. J., 178, 49, 1977. Am. Bull., 80, 1639-1684, 1969.
Case, J. E., S. Duran, L. G. Lopez, R. Alfonso, and W. R. Moore, Molnar, P., and P. Tapponnier, Cenozoictectonicsof Asia: Effectsof
Tectonic investigationsin western Colombia and eastern Panama, a continental collision, Science, 189, 419-426, 1975.
Geol. Soc. Am. Bull., 82, 2685-2711, 1971. Pennington,W. D., W. D. Mooney, R. van Hissenhoven,Hj. Meyer,
Dean, B. W., and C. L. Drake, Focal mechanism solutions and tecton- J. Ramirez, and R. P. Meyer, Resultsof a microearthquakerecon-
ics of the Middle America arc, J. Geol., 86, 111-128, 1978. naissancesurvey of Bucaramanga,Colombia, Geophys.Res. Lett.,
Dewey, J. W., Seismicity and tectonics of western Venezuela, Bull. 6, 65-68, 1979.
Seismol. Soc. Am., 62, 1711-1751, 1972. Pilger, R., Plate reconstructions,aseismicridges, and low-angle sub-
Direccion General de RecursosMinerales, Mapa geologico,Repub- duction beneath the Andes, Geol. Soc. Am. Bull., 92, 448-456, 1981.
lica de Panama, scale 1: 1,000,000, Panama City, 1976. Rial, J., The Caracas,Venezuela, earthquake of July 1967: A mul-
Feininger, T., The Palestinafault, Colombia, Geol.Soc.Am. Bull., 81, tiple-sourceevent, J. Geophys.Res., 83, 5405-5414, 1978.
1201-1216, 1970. Rod, E., Strike-slip faults of northern Venezuela, Am. Assoc. Pet.
Fitch, T., Plate convergence,transcurrentfaults and internal deforma- Geol. Bull., 40, 457-476, 1956.
tion adjacentto southeastAsia and the westernPacific,J. Geophys. Santo, T., Characteristicsof seismicityin South America, Bull. Earth-
Res., 77, 4432-4460, 1972. quakeRes.Inst., Tokyo Univ., 47, 635-672, 1969.
Handschumacher,D., Post Eocene plate tectonicsof the eastern Pa- Servicio Nacional de Geologia y Mineria, Mapa geologicode la Re-
cific, in Geophysicsof the Pacific Ocean and its Margin, Geophys. publica del Ecuador, scale 1:1,000,000, Quito, Ecuador, 1969.
Monogr. Ser., vol. 19, edited by G. Sutton, M. Manghnani, and R. Shagam, R., The northern termination of the Andes, in The Ocean
Moberly, pp. 177-202, AGU, Washington, D.C., 1976. Basinsand Margins: The Gulf of Mexico and the Caribbean,edited
Herd, D. G., T. L. Youd, Hj. Meyer, J. L. Arango C., W. J. Person, by A. E. M. Nairn and F. G. Stehli, vol. 3, pp. 325-420, Plenum,
and C. Mendoza, The great Tumaco, Colombia earthquake of 12 New York, 1975.
December 1979, Science, 211, 441-445, 1981. Studer,W., Tensionalcharacter
of earthquake
loci beneaththe
Herrmann, R., A student'sguide to the use of P and S wave data for Aleutian trench with relation to sea floor spreading,J. Geophys.
focal mechanismdetermination,EarthquakeNotes,46, 29-39, 1975. Res., 73, 7693-7701, 1968.
Hey, R,, Tectonicevolutionof the Cocos-Nazcaspreadingcenter, Stauder, W., Subductionof the Nazca plate under Peru as evidenced
Geol. Soc. Am. Bull., 88, 1404-1420, 1977. by focal mechanismsand by seismicity,J. Geophys.Res.,80, 1053-
Hey, R., K. Defteyes,G. Johnson,and A. Lowtie, The Galapagos 1064, 1975.
triple junction and plate motionsin the East Pacific,Nature, 237, Terry,
R.,A geologicalreconnaissanceofPanama, Occas.Pap.Calif.
20-22, 1972. Acad. Sci., 23, 99 pp., 1956.
Hey, R., G Johnson,and A. Lowri,Recentplate motionsin the Ga- van Andel, T., G. Heath, B. Mallair, D. Heinrichs,and J. Ewing, Tec-
lapagosarea, Geol. Soc. Am. Bull., 88, 1385-1403, 1977. tonics of the Panama basin, eastern equatorial Pacific, Geol. Soc.
Hodgson,J., and R. Storey, Tables extendingByerly's fault plane Am. Bull., 82, 1489-1508, 1971.
techniqueto earthquakesof any focal depth, Bull. Seismol.Soc. Vogt, P. R., A. Lowrie, D. Brace, and R. Hey, Subductionof aseismic
Am., 43, 49-61, 1953. oceanic ridges: Effects on shape, seismicity,and other character-
Holden, J., and R. Dietz, Galapagosgore,NazCoPactriple junction istics of consumingplate boundaries,Spec. Pap. Geol. Soc. Am.,
and Carnegie/Cocosridges,Nature, 235, 265-269, 1972. 172, 56 pp., 1976.
Houtz, R. E., and W. J. Ludwig, Structureof Colombia basin, Carib- VolcanologicalSocietyof Japan, 1971, List of the world active volca-
bean Sea,from profiler-sonobuoy measurements, J. Geophys. Res., noes,Bull. VolcanicEruptions,Spec.Issue, 123-126, 1971.
82, 4861-4867, 1977. Wagner, D., Statistical decisiontheory applied to the focal meth-
INGEOMINAS (InstituteNacional de Investigaciones Geologico anisms of Peruvian earthquakes,Ph.D. dissertation,176 pp., St.
Mineras), Mapa geologicode Colombia,scale1:500,000,Bogota, Louis Univ., St. Louis, Mo., 1972.
Colombia, 1976. Wing, R., and H. MacDonald, Radar geology:Petroleumexploration
Irving, E. M., Structuralevolutionof the northernmostAndes, Colom- technique,easternPanamaand northwesternColombia,Am. Assoc.
bia, U.S. Geol. Surv. Prof. Pap., 846, 47 pp., 1975. Pet. Geol. Bull., 57, 825-240, 1973.
Isacks,B., and P. Molnar, Distribution of stressesin the descending
lithosphere from a global survey of focal-mechanismsolutions of
mantle earthquakes,Rev. Geophys.SpacePhys.,9, 103-174, 1971.
Jeftreys,H., and K. Bullen, Seismological Tables,48 pp., British Asso-
ciation for Advancement of Science, London, 1940. (ReceivedMarch 24, 1980;
Johnson,G. L., and A. Lowtie, The GalapagosIslands:A possible revisedJuly 7, 1981;
'hotspot'?,Earth Planet. Sci. Lett., 14, 279-280, 1972. acceptedAugust7, 1981.)