existed, divided by the Santander Floresta horst or high. The location of small mac intrusions coincides with areas of thin crust(crustal stretching factors >1.4) and maximum stretching of the subcrustal lithosphere. During the Aptianearly Albian, the basin

thickness suggest the reverse or thrust faults that now dene the eastern and western borders of the EC were originally normal faultswith a strike-slip component that inverted during the Cenozoic Andean orogeny. Thus, the Guaicaramo, La Salina, Bituima,Magdalena, and Boyaca originally were transtensional faults. Their oblique orientation relative to the Mesozoic magmatic arc of

* Corresponding author. Tel: +2345657.E-mail address: luis.sarmiento@ecopetrol.com.co (L.F. Sarmiento-Rojas).

Journal of South American Earth Sciences 21 (2006) 383411extended toward the south in the Upper Magdalena Valley. Dierences between crustal and subcrustal stretching values suggest somelowermost crustal decoupling between the crust and subcrustal lithosphere or that increased thermal thinning aected the mantle lith-osphere. Late Cretaceous subsidence was mainly driven by lithospheric cooling, water loading, and horizontal compressional stressesgenerated by collision of oceanic terranes in western Colombia. Triassic transtensional basins were narrow and increased in widthduring the Triassic and Jurassic. Cretaceous transtensional basins were wider than TriassicJurassic basins. During the Mesozoic,the strike-slip component gradually decreased at the expense of the increase of the extensional component, as suggested by paleomag-netic data and lithosphere stretching values. During the BerriasianHauterivian, the eastern side of the extensional basin may havedeveloped by reactivation of an older Paleozoic rift system associated with the Guaicaramo fault system. The western side probablydeveloped through reactivation of an earlier normal fault system developed during TriassicJurassic transtension. Alternatively, theeastern and western margins of the graben may have developed along older strike-slip faults, which were the boundaries of the accre-tion of terranes west of the Guaicaramo fault during the Late Triassic and Jurassic. The increasing width of the graben system likelywas the result of progressive tensional reactivation of preexisting upper crustal weakness zones. Lateral changes in Mesozoic sedimentMesozoic transtensional basin history of the Eastern Cordillera,Colombian Andes: Inferences from tectonic models

L.F. Sarmiento-Rojas a,*, J.D. Van Wess b, S. Cloetingh c

a Ecopetrol-Empresa Colombiana de Petroleos, P.O. Box 5938-6813-Bogota, Colombiab Netherlands Institute of Applied Geoscience TNO National Geological Survey, Prins Hendriklaan 105 P.O. Box 80015 3508 TA Utretch, The Netherlands

c Tectonics Group, Faculty of Earth Sciences, Free University, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands

Received 1 October 2004; accepted 1 March 2006

Abstract

Backstripping analysis and forward modeling of 162 stratigraphic columns and wells of the Eastern Cordillera (EC), Llanos, andMagdalena Valley shows the Mesozoic Colombian Basin is marked by ve lithosphere stretching pulses. Three stretching events aresuggested during the TriassicJurassic, but additional biostratigraphical data are needed to identify them precisely. The spatial distri-bution of lithosphere stretching values suggests that small, narrow (180 km) wide, asymmetrical, transtensional half-rift basin

www.elsevier.com/locate/jsames0895-9811/$ - see front matter 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.jsames.2006.07.003

the Central Cordillera may be the result of oblique slip extension during the Cretaceous or inherited from the pre-Mesozoic structuralgrains. However, not all Mesozoic transtensional faults were inverted. 2006 Elsevier Ltd. All rights reserved.

Keywords: Rifting; Lithosphere stretching; Tectonic subsidence; Colombia; Mesozoic; Eastern Cordillera

1. Introduction

This article focuses on the tectonic basin-forming pro-cesses of the Colombian Eastern Cordillera (EC, Fig. 1)during Mesozoic times, in terms of the geodynamic pro-cesses that govern deformation of the lithosphere. We com-pile local data into a regional geological model, analyzesubsidence, and quantitatively model tectonic subsidence.

We have studied tectonic subsidence signals that giveimportant information about basin formation mechanisms.For this purpose, we analyze temporal and spatial basinsubsidence patterns, quantitatively analyze tectonic subsi-dence, and forward model it to explain these patterns fromwithin the framework of the geodynamic processes thatformed the Mesozoic EC basin. In doing so, we addressissues such as the relationship among basin development,

extensional episodes, plate-tectonic events, magmaticevents, and basin geometry. Many features of these exten-sional basins and their underlying mechanisms are practi-cally unknown.

Fabre (1983a,b, 1987) and Hebrard (1985) study thesubsidence of the eastern ank of the EC during the Creta-ceous, identify the basin as produced by lithosphere exten-sion, calculate tectonic subsidence curves, and, followingthe uniform instantaneous stretching model developed byMcKenzie (1978), calculate lithosphere stretching factorsclose to 2. They distinguish an Early Cretaceous subsidencephase produced by rifting and Late Cretaceous subsidenceproduced by thermal decay after rifting. We assume severalevents of lithosphere stretching of nite duration in ourstudy of tectonic subsidence and examine the possibilityof dierentiating between crustal and subcrustal stretching

384 L.F. Sarmiento-Rojas et al. / Journal of South American Earth Sciences 21 (2006) 383411Fig. 1. Location map. SL, Serrana de San Lucas; MA, Serrana de La Macarena. Eastern Cordillera regions: SM, Santander Massif; SF, Santander-Floresta high; MT, Magdalena Tablazo inverted subbasin; CO, Cocuy inverted subbasin; MF, Magdalena Valley foothills; LF, Llanos foothills; CU,

Cundinamarca inverted subbasin; SE, Southern Eastern Cordillera; QM, Quetwestern boundary of the Central cordillera and the Lower Magdalena Valley.ame Massif; GM, Garzon Massif; Romeral paleosuture in this map is the

th AL.F. Sarmiento-Rojas et al. / Journal of Southat occurred in the Colombian Basin throughout theMesozoic. An extensive data set of 162 stratigraphic col-umns and wells from the EC, Magdalena Valley (MV),and Llanos Orientales (LLA) areas (Fig. 2, see referencesin table 2.1 of Sarmiento, 2001) extracted from the litera-ture, as well as data from Ecopetrol, are used.

2. Tectonic setting

The EC is the eastern branch of the Colombian Andes(Fig. 1), which comprises three mountain ranges: the East-ern, Central, and Western Cordilleras, which merge south-ward into a single range. The MV separates the Easternand Central Cordilleras, and the Cauca Valley separatesthe Central and Western Cordilleras. The EC and itsbounding basins, the LLA in the east and MV in the west,dene the area studied herein.

During the Mesozoic, the area of the EC was an exten-sional basin. During the Paleogene, according to someauthors (e.g., Van der Hamen, 1961; Roeder and Chamber-lain, 1995; Gomez et al., 1999; Sarmiento, 2001), upthru-sted blocks and/or incipient inversion of the Mesozoicextensional basin may have occurred in the area of theEC. However, another view posits a unique simple foreland

Fig. 2. Location of stratigraphic columns and wells and stratigraphic regionatransect labels (for details, see Sarmiento, 2001).merican Earth Sciences 21 (2006) 383411 385basin existed, related to the topographic load of the CentralCordillera (e.g., Cooper et al., 1995). General consensusindicates that during the Neogene, the Mesozoic extension-al basin became inverted, deformed, and uplifted to formthe EC (Cooper et al., 1995).

In the study area, during the Triassic and Jurassic, con-tinental and volcanic facies were deposited in extensionalbasins (Mojica et al., 1996). During the Triassic, thesebasins seem related to Pangea rifting (Pindell and Dewey,1982; Ross and Scotese, 1988; Cediel et al., 2003), and sincethe Jurassic, they developed behind a magmatic arc relatedto the subduction of the Pacic plates under the westernborder of South America (McCourt et al., 1984; Fabre,1987; Toussaint and Restrepo, 1989; Cooper et al., 1995;Meschede and Frisch, 1998). During the Early Cretaceous,marine facies accumulated on a wide extensional basin sys-tem. Shallow-marine sedimentation ended at the end of theCretaceous, due to the accretion of the oceanic terranes ofthe Western Cordillera.

2.1. Triassic and Jurassic plate tectonic interpretations

The tectonic settings discussed subsequently assumethat accretion of tectonic terranes east of the Romeral

l sections listed in Table 2. Numbers along sections refer to stratigraphic

th Apaleosuture and west of the Llanos Orientales Basin(Fig. 1) occurred during the Paleozoic. Following thisassumption for evolution in Colombia, the plate tectonicinterpretations rely on two hypotheses. Although theseinterpretations dier in plate tectonic setting, an alternativehypothesis suggests these settings formed at dierent times.

1. Intracontinental rifting related to the breakup of Pangea(Pindell and Dewey, 1982; Ross and Scotese, 1988; Ced-iel et al., 2003) occurred during Triassic and Early Juras-sic times. This hypothesis probably is more applicable tothe northern part of Colombia and Venezuela and theirseparation from North America.

2. Backarc extension occurred behind a subduction-relatedmagmatic arc (Maze, 1984; McCourt et al., 1984; Pindelland Erikson, 1993; Toussaint, 1995a,b; Pindell and Tab-butt, 1995; Meschede and Frisch, 1998). According tothis hypothesis, the study area was located at the marginof the continent when active subduction of oceanicPacic plates was occurring. This interpretation explainsthe Triassic and Early Jurassic extensional basins in thestudy area as backarc basins.

However, a recent paleomagnetic investigation sug-gests that along-plate margin translations of terranesmight have taken place during the early Mesozoic (Bay-ona et al., 2005), as inferred by Toussaint and Restrepo(1994). Bayona et al. (2005), on the basis of paleomag-netic data, suggest a signicant northward translation(at least 10) of terranes west of the Bucaramanga fault(Bucaramanga area, Floresta Massif and Upper Magda-lena Valley terranes) with respect to the craton duringthe Early Jurassic and no signicant paleolatitude anom-alies since then. Additional paleomagnetic data are need-ed to test and quantify the magnitude of translation oftectonic terranes (Bayona, personal communication,2005).

2.2. Cretaceous plate tectonic interpretations

For the Cretaceous, three alternative hypotheses suggestthe processes that might have operated at dierent times:

1. Backarc extension (McCourt et al., 1984; Fabre, 1987;Toussaint and Restrepo, 1989; Cooper et al., 1995;Meschede and Frisch, 1998). Key evidence for thishypothesis is the existence of a subduction-related mag-matic arc.

2. Passive margin (Pindell and Erikson, 1993; Pindell andTabbutt, 1995). The scarcity of magmatic rocks in thebasin seems to support this hypothesis, but a poorlydened Cretaceous magmatic arc (i.e., San Diego, Alta-vista, and Cambumbia stocks; Restrepo et al., 1991;Toussaint and Restrepo, 1994) in the Central Cordillerais dicult to explain. Alternatively, the Central Cordille-

386 L.F. Sarmiento-Rojas et al. / Journal of Soura may have been located farther south of its presentposition during the Cretaceous.3. Intracontinental rifting related to the opening of theCaribbean. Some authors (e.g., Geotec, 1992; Cedielet al., 2003) suggest a NW-SE graben developed in thenorthern part of the Central Cordillera during the EarlyCretaceous. A poorly dened Cretaceous magmatic arcin the Central Cordillera is dicult to explain with thishypothesis.

During the latest Cretaceous (post-Santonian), all platetectonic interpretations propose a convergent margin westof Colombia. The Caribbean plate was moving northeast-ward relative to South America, while the Farallon platewas subducting west of southern Colombia (Pindell andErikson, 1993; Pindell and Tabbutt, 1995).

3. Stratigraphy

3.1. Triassic and Jurassic synrift sedimentation

The Triassic and Jurassic sedimentary record is presentin several isolated outcrops (Fig. 3). Continental depositswith redbeds and volcanic eusive and pyroclastic depositsare dominant, though some marine facies appear locally.Triassic and Jurassic rocks were deposited in extensionalbasins mainly located in the Upper Magdalena Valley(UMV), Serrana de San Lucas, and the western ank ofthe EC (Mojica et al., 1996).

Fig. 4 shows a stratigraphic synthesis modied afterMojica et al. (1996; Fig. 6). Triassic and Jurassic sedi-mentary rocks formed a sequence bounded by unconfo-rmities. The lower contact is marked by anunconformity, which is dominantly angular. The uppercontact is dominantly unconformable but locally con-formable. Jurassic deposits consisting of clastic faciesdeposited in dominantly continental environments arewidely distributed. In those sections where there aresome marine facies, they are under- and overlain by con-tinental clastic facies. The ne-grained muddy marinefacies record local marine incursions during the Late Tri-assicEarly Jurassic. Volcaniclastic, pyroclastic, and vol-canic lavas are mainly restricted to the upper part ofthe Upper Triassic to the lower part of the Middle Juras-sic (Mojica et al., 1996, Fig. 6).

Facies and thickness similarities related to geographicpositioning suggest that TriassicJurassic sedimentationoccurred in two separate basin compartments, each withits own subsidence history and sedimentary ll (Figs. 3and 4).

1. Upper Magdalena, Cienaga de Morrocoyal, and SierraNevada terrane (region A, Fig. 3). This section corre-sponds with the western part of the Chibcha terrane,as dened by Toussaint (1995a,b), or the Payande, SanLucas, and Sierra Nevada terranes, as proposed byEtayo-Serna et al. (1983). In this compartment, two

merican Earth Sciences 21 (2006) 383411marine incursions are recognizable: in the Late Triassic(Norian?Rhetian, Chicala member of the Saldana

th AL.F. Sarmiento-Rojas et al. / Journal of SouFormation, UMV) and in the Early Jurassic (Sinemurian,Morrocoyal, and Los Indios formations; ages accordingto Mojica et al., 1996). Continental sedimentation fol-lowed shallow-marine deposition in this compartment.Volcanic-related facies are volumetrically more impor-tant here than in the compartment formed by outcropsin the EC area (region B, Fig. 3). On the basis of paleo-magnetic data from the UMV, Bayona et al. (2005) pro-pose that the UMV and San Lucas (Payande-San Lucasterranes; Etayo-Serna et al., 1983) were located south ofthe equator before the Middle Jurassic and north of itduring the late Jurassicmiddle Cretaceous, with anorthward movement of at least 10 relative to the SouthAmerican craton.

Fig. 3. Location of TriassicJurassic strata outcrops and stratigraphic sections.E represents the TriassicJurassic sedimentary record of the eastern part of therepresents the TriassicJurassic sedimentary record of the western part of the Cterranes of Etayo-Serna et al. (1983) (modied from Toussaint, 1995b). Inset:merican Earth Sciences 21 (2006) 383411 3872. Eastern Cordillera (region B, Fig. 3). This area corre-sponds to the eastern side of the Chibcha terrane, asdened by Toussaint (1995a,b). Marine inuence ondeposition is located within the Lower Jurassic (Monte-bel Formation; ages according to Mojica et al., 1996,Fig. 4). This compartment is characterized by an absenceof Triassic-dated rocks and thick deposition of siliciclas-tic deposits on continental extensional basins (e.g., Gir-on Formation). Bayona et al. (2005), on the basis ofpaleomagnetic data from the Bucaramanga and Florestaareas, suggest part of this terrane located west of theBucaramanga fault was located close to the equator dur-ing the early Middle Jurassic and moved northwardabout 4 relative to the stable craton. According to their

Labeling of stratigraphic sections according to Fig. 4. Stratigraphic sectionChibcha terrane according to Toussaint (1995b). Stratigraphic section Whibcha terrane, equivalent to the Payande, San Lucas, and Sierra Nevadalocation of study area.

th A388 L.F. Sarmiento-Rojas et al. / Journal of Soureconstruction, during the Late TriassicMiddle Juras-sic, the UMV and Cienaga de Morrocoyal (Payande-San Lucas terranes) basin compartment was locatedsouth of the EC compartment.

3.2. Cretaceous sedimentation

Most exposed rocks in the EC are Cretaceous in age.Fig. 5 shows a traverse time-stratigraphic cross-section ofthe basin.

Cretaceous rocks, including locally the uppermostJurassic and Paleocene deposits, form a megasequencebounded by regional unconformities that are at least local-ly angular. On a broad scale, Cretaceous rocks represent amajor transgressiveregressive cycle with a maximumooding surface close to the CenomanianTuronianboundary, corresponding to the maximum Mesozoiceustatic level (Fabre, 1985; Villamil, 1993; Fig. 5). Superim-posed on this large-scale trend, several smaller transgres-siveregressive cycles are present, suggesting an

Fig. 4. TriassicJurassic stratigraphic sections. Location of transects in Fig.present-day horizontal distance (km) without palinspastic restoration (modifossiliferous facies (El Sudan Formation) in the Cienaga de Morrocoyal area (Formation of the Payande area.merican Earth Sciences 21 (2006) 383411oscillating relative tectonoeustatic level. Subsidence wasrapid (Fabre, 1983a,b, 1987), but shallow-water sedimenta-tion suggests deposition kept pace with it.

The basin was a wide graben system oriented approxi-mately NNE-SSW, divided into two subbasins (Tablazoand Cocuy, Fig. 6) by the Santander-Floresta paleo-Mas-sif. To the north, these subbasins continued to the Machi-ques trough in the Merida Andes of Venezuela andUribante trough in Serrana de Perija (Julivert, 1968; Fab-re, 1985, 1987). To the south, these subbasins joined as theCundinamarca subbasin (Burgl, 1961), where the thicknessof the Cretaceous sections reaches a maximum (Figs. 2 and5). Fabre (1987) and Sarmiento (1989) suggest the Cundi-namarca trough was limited to the south and north byNW-SE transfer paleofaults (Fig. 6). The N-S lateralchanges of thickness and an extensional relay system southof the Cundinamarca trough support the existence of NW-SE-striking transfer faults.

On the basis of the presence of Lower Cretaceous sedi-mentary rocks in the northern part of the Central Cordille-ra, Geotec (1992) suggests the existence of a NW-SE

3. Vertical axis represents geological time, and horizontal axis representsed from Mojica et al., 1996). Geyer (1982), however, suggests poorlynorth of Serrana de San Lucas) are Triassic and correlate with the Luisa

ertim)ordon a

th Agraben, connected with the Cundinamarca subbasin. Com-

Fig. 5. Cretaceous and Tertiary stratigraphic section. Location in Fig. 2. VOgg (1996); horizontal axis represents present-day horizontal distance (kearlier versions by Fabre (1985, 1987) and Cooper et al. (1995), modied accet al., 1993; Villamil, 1993, 1994; Etayo-Serna, 1994; Ecopetrol, 1994; Rol

L.F. Sarmiento-Rojas et al. / Journal of Soupiled available thicknesses of these sediments for similarchronostratigraphic intervals, however, are signicantlythinner than those of the EC (Figs. 5, 7b, and 8a). If sucha graben existed, in terms of subsidence, it was a minor fea-ture compared with the grabens in the EC. These LowerCretaceous marine sedimentary rocks in the Central Cor-dillera are strongly deformed and associated with macvolcanic rocks of ocean anity, which collectively havebeen named the Quebradagrande Complex by Nivia et al.(1996) and Nivia and Gomez (2005). These authors inter-pret the rocks as originated within a marginal basin withoceanic crust formation developed along an older suturebetween dierent metamorphic terranes of the Central Cor-dillera (between the eastern Cajamarca Complex and thewestern Arqua Complex). Compiled thicknesses suggesta current N-S basin in agreement with Nivia and Gomezs(2005) interpretation.

3.2.1. Early cretaceous synrift sedimentationSedimentation started in the Tablazo subbasin in Juras-

sic times and continued during the Early Cretaceous local-ly, without a tectonic-related angular unconformity (e.g.,Rio Lebrija section, Cediel, 1968). In other areas, Creta-ceous sedimentary rocks rest with angular unconformityon earlier Mesozoic, Paleozoic, or even pre-Cambrianrocks. In the Tablazo subbasin, basal beds correspond tosandstones (Los Santos, Tambor, and Arcabucoformations) deposited in uvial environments, usually withdetrital sediments derived from uplifted fault blocks(Etayo-Serna and Laverde-Montano, 1985). The amalgam-

cal axis represents geological time according to the scale of Gradstein andwithout palinspastic restoration. This section has been constructed froming to sequence stratigraphy interpretations (Pimpirev et al., 1992; Fajardond Carrero, 1995; Villamil and Arango, 1998).

merican Earth Sciences 21 (2006) 383411 389ation of channel beds indicates that the sediment supplyexceeded the rate of basin subsidence at the JurassicCreta-ceous boundary. Burgl (1967) suggests that an initial mar-ine incursion in the Cundinamarca subbasin ooded acontinental area with a desert climate, which provided con-ditions for evaporite formation during the early stages ofmarine transgression. McLaughlin (1972) cites paleonto-logical evidence of a BerriasianValanginian age for someevaporite occurrences. During the Berriasian, the sea ood-ed the basin from the northern part of the Central Cordil-lera toward the Cundinamarca subbasin (Etayo-Sernaet al., 1976). Then, the sea advanced from the Cundinamar-ca subbasin to the north into two subbasins, while theSantander-Floresta paleo-Massif remained emergent(Etayo-Serna et al., 1976; Fabre, 1985, 1987; Sarmiento,1989; Figs. 5, 7a, b, and 8a).

The Tablazo and Cocuy subbasins started to form a sin-gle wide basin during the Hauterivian due to ooding ofthe Santander-Floresta paleohigh (Fabre, 1985) and baselevel rise. However, this intrabasinal high was a signicantbarrier to sediment movement until the Aptian (Figs. 7a, b,and 8a) .

To the south, both the Tablazo and Cocuy subbasinsshow a gradual increase in dark shale deposited in poorlyoxygenated shallow-marine environments (Caqueza andVilleta groups; Fabre, 1985; Sarmiento, 1989). In the Cun-dinamarca subbasin, Cretaceous sedimentation startedduring the Tithonian?BerriasianValanginian with turbi-dite deposits in both the eastern (lower Caqueza Group;

th A390 L.F. Sarmiento-Rojas et al. / Journal of SouPimpirev et al., 1992) and western (lower part of Uticasandstone, Murca Formation; Sarmiento, 1989; Moreno,1990, 1991) anks (Figs. 7a, b, and 8a). Turbidite deposi-tion prevailed until the Hauterivian at the eastern borderof the basin (Caqueza Group, Pimpirev et al., 1992).

During the earliest Cretaceous, basin subsidenceexceeded sediment supply, resulting in retrogradation ofthe turbidite fan system, so distal fan sediments coveredmiddle fan mouth channel deposits. Post-Berriasian, sed-iment supply increased and overwhelmed basin subsi-dence, resulting in progradation of the turbidite fansystem (Pimpirev et al., 1992) and locally in progradationof deltaic sands during the Hauterivian (upper Uticasandstone; Sarmiento, 1989; Moreno, 1990). Dierentialsubsidence related to syn-sedimentary normal faultingcaused unstable slopes on basin margins. These processesfavored turbidite deposition during the early CretaceousAptian (lower Utica sandstone, Murca Formation, Sar-miento, 1989; Moreno, 1990, 1991; Socota Formation,Polana and Rodrguez, 1978; Caqueza Group, Pimpirevet al., 1992; Figs. 7a, b, and 8a).

An important transgression followed a relative sea-levelrise during late Aptian time. The sea ooded all the area ofthe present EC, even south of the Cundinamarca subbasin

Fig. 6. Cretaceous basin compartments (subbasins) and their tectonic subsiden(B) Tectonic subsidence of the Cundinamarca subbasin. Cities: M, Medelln; MB, Bogota; V, Villavicencio; Y, Yopal; A, Arauca.merican Earth Sciences 21 (2006) 383411(Etayo-Serna et al., 1976; Etayo-Serna, 1994). Southwardof the paleo-UVM, Cretaceous sedimentation started dur-ing the Aptian (Vergara and Prossl, 1994) in an extensionalbasin formed initially in Jurassic times.

In the whole EC basin, abrupt lateral thickness changesand fault-related alluvial or turbidite deposition attest tolocal tectonic/dierential subsidence depositional condi-tions in the Early Cretaceous. Regional correlation of Ear-ly Cretaceous relative tectonoeustatic cycles is dicult toestablish because of local active extensional tectonics. Sincethe Aptian, these relative tectonoeustatic cycles havebecome more tractable than the pre-Aptian cycles (Fig. 5).

In the Central Cordillera, Lower Cretaceous sedimenta-ry rocks (mudstones, chert, feldspathic sandstones, andconglomerates) of the Quebradagrande Complex are asso-ciated with an ophiolite suite of ultramac rocks, gabbros,basalts, breccias, and pyroclastic rocks usually aected bydynamic metamorphism (Moreno and Pardo, 2003). Depo-sitional environments vary among uvial, coastal, deltaic,and marine shelf (Rodrguez and Rojas, 1985). Evidenceindicates that during the middle Cretaceous, these rocksbegan compressive deformation (Rodrguez and Rojas,1985), possibly as a result of the closure of the Quebrada-grande marginal basin (Nivia et al., 1996).

ce in meters. (A) Tectonic subsidence of the Tablazo and Cocuy subbasins.z, Manizales; I, Ibague; N, Neiva; C, Cucuta; Bu, Bucaramanga; T, Tunja;

th AL.F. Sarmiento-Rojas et al. / Journal of Sou3.2.2. Cretaceous postrift sedimentation

Several relative tectoeustatic level cycles have been pro-posed during the Late Cretaceous (Fig. 5).

During the Albian, a relative base-level fall favored pro-gradation of deltaic and littoral sands in the UMV (Cabal-los Formation, Etayo-Serna, 1994) and the eastern borderof the basin (lower Une Formation, Fabre, 1985). Duringthe middlelate Albian, the transition from near-shore

Fig. 7. (a) BerriasianValanginian and (b) HauterivianBarremian thicknessbelieved to be active during BerriasianValanginian/HauterivianBarremian,calculated through forward modeling for the BerriasianHauterivian (144Distribution of main Early Cretaceous faults and mac intrusions shown withbasaltic lava, 4. Rio Cravo Sur microgabbro, 5. Pajarito, 6. Q. La Esperanza,Chorrera, 12. La Chunchalita, 13. Q. La Fiebre, 14. Caceres, 15. La Corona, 1lithosphere stretching factors calculated through forward modeling for the Bpalinspastic restoration. Distribution of main Early Cretaceous faults is shownBucaramanga; T, Tunja; B, Bogota; V, Villavicencio; Y, Yopal; A, Arauca.merican Earth Sciences 21 (2006) 383411 391marine facies (e.g., Caballos, San Gil Inferior, and Socotaformations) to outer shelf facies (e.g., Villeta Group, SanGil Superior, and Hilo formations) recorded a rise inrelative tectonoeustatic levels (Villamil, 1993; Etayo-Serna,1994). During late Albianearly Cenomanian, a relativetectonoeustatic level fall was recorded by progradation ofthe upper part of Une Formation and a generalized shal-lowing-upward facies trend (e.g., transition of San Gil

(m) without palinspastic restoration. Thick lines represent paleofaultsrespectively. (c) Contour map of crustal (d) lithosphere stretching factors127 Ma, Cretaceous) stretching event without palinspastic restoration.circles: 1. Rio Nuevo diorite, 2. Rodrigoque micrograbbro, 3. Porritic

7. Q. Las Palomas, 8. Q. La Culebra, 9. Marl, 10. Q. Grande, 11. Q. La6. Pacho, and 17. Rio Guacavia diorite. (d) Contour map of subcrustal (b)erriasianHauterivian (144127 Ma, Cretaceous) stretching event without. Cities: M, Medelln; Mz, Manizales; I, Ibague; N, Neiva; C, Cucuta; Bu,

th A392 L.F. Sarmiento-Rojas et al. / Journal of SouSuperior to Churuvita and Hilo to unnamed shale; Villa-mil, 1993).

During the late Cenomanian, Turonian, and Coniacian,the tectonoeustatic base level reached its maximum Meso-zoic level. The sea ooded the entire northwestern cornerof South America, and dark gray shale was deposited fromVenezuela to northern Peru (Thery, 1982, in Fabre, 1985).In contrast to the EC, where the Cretaceous maximumooding surface occurred at the CenomanianTuronianboundary, maximum ooding in the LLA occurred during

Fig. 8. (a) Aptian thickness (m) without palinspastic restoration. Thick lines rmap of crustal (d) lithosphere stretching factors calculated through forward mopalinspastic restoration. Distribution of main Early Cretaceous faults is alsocalculated through forward modeling for the Aptian (121102.6 Ma, CretaceoEarly Cretaceous faults also shown. Cities: M, Medelln; Mz, Manizales; I, IVillavicencio; Y, Yopal; A, Arauca.merican Earth Sciences 21 (2006) 383411the Campanian (CS at the top of Gacheta Formation; Faj-ardo et al., 1993; Cooper et al., 1995, Fig. 4).

Villamil (1993) recognized smaller relative tectonoeustat-ic level cycles during late Cenomanian, Turonian, andConiacian times. A relative tectonoeustatic base level riseduring the late Cenomanian (Villamil, 1993) induced a slightdeepening of the basin and a notorious decrease of detritalsupply to the basin (e.g., Frontera and lower part of SanRafael formations; Villamil, 1993). During the TuronianConiacian, the present-day LLA foothills ooded (Cooper

epresent paleofaults believed to be active during Aptian time. (b) Contourdeling for the Aptian (121102.6 Ma, Cretaceous) stretching event withoutshown. (c) Contour map of subcrustal (b) lithosphere stretching factorsus) stretching event without palinspastic restoration. Distribution of mainbague; N, Neiva; C, Cucuta; Bu, Bucaramanga; T, Tunja; B, Bogota; V,

quantitative framework for the pulsating rift evolution ofthe lithosphere during Mesozoic basin formation, we quan-tify extension rates by forward modeling tectonic subsi-dence. We use an automated forward modelingtechnique (Van Wees et al., 1996b), which we explain brief-ly next.

4.2.1. Numerical model

The forward modeling approach is based on lithosphericstretching assumptions (McKenzie, 1978; Royden and

th Aet al., 1995) but not the entire LLA (Fig. 5). From theMiddleTuronian to late Coniacian, a gradual progradation andshallowing upward (e.g., upper San Rafael Formation andVilleta Group in the upperMV) was related to a relative tec-tonoeustatic level fall (Villamil, 1993).

In the UMV and during the late ConiacianSantonian,deepening of the basin and relative tectonoeustatic levelrise occurred (transition from inner shelf uppermost VilletaGroup to middle shelf lower chert unit of the Olini Group;Etayo-Serna, 1994, Fig. 2).

During the Santonian, Campanian, Maastrichtian, andPaleocene, a general regression and progradation wasrecorded by littoral to transitional coastal plain facies(e.g., Guadalupe Group, Guaduas Formation). GuadalupeGroup sands represent two cycles of westward shorelineprogradation, aggradation, and retrogradation, dominatedby high-energy, quartz-rich shoreface sandstones derivedfrom the Guyana Shield (Cooper et al., 1995; Fig. 5).Regression did not occur continuously but with minortransgressive events recorded by ne-grained siliceous andphosphatic facies (Plaeners Formation, Olini Group, andupper La Luna Formation, Follmi et al., 1992; Fig. 5).

At the end of Cretaceous (Maastrichtian), the gradualuplift of the western margin of the MV supplied clasts ofmetamorphic rocks accumulated by uvial systems closeto the sea in a braided delta (Cimarrona Formation,Gomez and Pedraza, 1994).

In the Central Cordillera east of the QuebradagrandeComplex, no Late Cretaceous sedimentary record has beenfound. This area may have been uplifted during the latestCretaceous and started to supply clastics to the east.

4. Methods

4.1. Subsidence analysis

The stratigraphic record provides information aboutvertical crustal movements in a basin. Basin subsidence isthe result of both a thermomechanical component calledtectonic subsidence and a sediment and water loading com-ponent. Tectonic subsidence is undistorted basin subsi-dence in the absence of sedimentation and therefore isrelated to the geodynamics of the basin.

To quantify the tectonic component of subsidence of thestudied basin, we used the one-dimensional (1D) backstrip-ping technique (Steckler and Watts, 1978; Bond andKominz, 1984), as explained by Sclater and Christie (1980),Bond and Kominz (1984), and Bessis (1986). For this pur-pose, we calculate tectonic subsidence from the stratigraphicrecord, adopting localAiry isostasy to correct for the eect ofsediment loading. The corrections for compaction followporositydepth relationships on the basis of the observedlithologies using standard mean exponential relations andmaterial parameters (cf. Sclater and Christie, 1980).

Most stratigraphic columns are from published litera-

L.F. Sarmiento-Rojas et al. / Journal of Souture (see Sarmiento, 2001); well data are from Ecopetrol.We carefully checked the thickness of each stratigraphiccolumn with available geological maps to avoid structuralrepetitions. We also veried the consistency of thicknessbetween neighboring sections. The eects of paleobathyme-try have been taken into account, using sedimentary faciesand faunal content as interpreted in the literature. Ages arebased on data in the literature, mainly the regional strati-graphic cross-sections presented by Cooper et al. (1995).To express ages in Ma, we used the geological time scaleproposed by Gradstein and Ogg (1996). Unconformitiesare also included with ages and durations in Ma. Addition-al parameters for forward modeling are the initial crust andlithosphere thickness and densities. Table 1 provides theadditional parameters we used in forward modeling, whichare the average accepted values for normal continental lith-osphere (Table 2).

Pulses of fast tectonic subsidence of the basement havebeen interpreted in terms of tectonic processes. Only thosepulses of fast tectonic subsidence correlatable to the nor-mal movement component of fault activity are interpretedas produced by lithosphere extension or transtension,though normal faulting also may correspond to transte-sional settings. Fast subsidence events related to crustalor lithosphere thinning also is common in transtensionalbasins or intra-arc basins (Ingersoll and Busby, 1995). Inthese settings, the extensional component is a major con-tributor to tectonic subsidence. The slower, generally later,rate of subsidence has been interpreted as produced bythermal reequilibration of the lithosphere following thethermal anomaly created by stretching.

4.2. Forward modeling of basin evolution

To quantify the horizontal extensional movementsresponsible for the observed subsidence and establish a

Table 1Parameters used to calculate tectonic subsidence in the forward models

Model parameters Value

Initial lithospheric thickness 120 kmInitial crustal thickness 35 kmAsthenospheric temperature 1333 CThermal diusivity 1 106 m2 s1

Surface crustal density 2800 kg m3

Surface mantle density 3400 kg m3

Seawater density 1030 kg m3

Thermal expansion coecient 3.2 105 C

merican Earth Sciences 21 (2006) 383411 393Keen, 1980). The extension factor d is used for crustalstretching and b for subcrustal stretching. For thermal

Table 2Cretaceous stretching events and stretching lithosphere factors from stratigraphic columns where the Cretaceous sedimentary record is present

Basin compartmenta Stratigraphic columnb BerriasianHauterivian Event Aptianearly Albian event

Start (Ma) End (Ma) Stretchingfactor

Start (Ma) End (Ma) Stretchingfactor

b d b d

2 72 Casabe-199 144 127.8 1.522 1.146 114 109.3 1.03 1.032 74 Cascajales-1 c 144 127.5 0.987 0.987 114 109.3 1.132 1.0942 73 Infantas-1613 138 127.8 1.301 1.117 114 109.3 1.328 1.1132 53 Lebrija-1 144 127 1.625 1.213 114 109.3 1 12 71 Llanito-1 138 127.8 1.313 1.098 114 109.3 1.478 1.1312 70 Sabalo-1K 144 127.8 1.37 1.113 114 109.3 1.039 1.0393 4 Arcabuco c 140 127.5 1.099 1.0993 8 Chima c 144 127.5 1.47 1.261 114.8 112.2 1.063 1.0193 10 Cimitarrra 143 127.5 1.354 1.31 114.8 112.2 1.329 1.1113 24 Los Medios c 142 127.5 1.043 1.043 114.8 112.2 1.096 1.0433 25 Los Santos 136.5 127.5 1.131 1.088 114.8 112.2 1.009 1.0093 47 Simacota 144 127.5 1.251 1.251 114.8 112.2 1.064 1.013 52 Tablazo c 144 127.5 1.052 1.052 114.8 112.2 1 13 57 Vadorreal 134.5 127.5 1.585 1.179 114.8 112.2 1.374 1.083 58 Velez 142 127.5 1.151 1.245 114.8 112.2 1.219 1.0723 59 Villa de Leiva 142 127.5 2.016 1.332 114.8 112.2 1.052 1.0524 66 Chitasuga-1 142 130 1.196 1.196 121 112.2 1.536 1.1874 22 La Calera 142 130 1.331 1.206 121 112.2 1.58 1.1784 39 Quipile c 141 130 1.511 1.147 121 112.2 1.301 1.1154 48 Simijaca 142 130 1.247 1.247 121 112.2 3.238 1.3184 75 Suba-2 c 142 130 1.085 1.085 121 112.2 1.221 1.1424 76 Suesca-1 142 130 1.217 1.217 121 112.2 2.476 1.3044 77 Suesca Norte-1 142 130 1.214 1.214 121 112.2 1.824 1.2244 50 Sutamarchan 142 130 1.234 1.234 121 112.2 1.825 1.2524 51 Tabio 142 130 1.227 1.227 121 112.2 1.587 1.2064 60 Villeta c 141 127.5 1.01 1.01 121 112.2 1.52 1.4924 61 Yacop c 134.2 127.5 1.449 1.449 121 112.2 1.77 1.2995 3 Apulo 127.6 127 1.792 1.164 121 112.2 1.261 1.0495 15 Fusagasuga 127.6 127 1.571 1.13 121 112.2 1.466 1.0716 2 Alpujarra c 115.1 102.6 1.411 1.2226 12 Coello c 121 102.6 1.11 1.0666 16 Girardot 121 102.6 1.366 1.1576 19 Guataqui 121 102.6 1.366 1.1736 21 Itaibe 108.5 102.6 1.176 1.0876 17 Melgar 121 102.6 1.18 1.1216 29 Neiva c 121 102.6 1.312 1.0826 31 Ortega 121 102.6 1.658 1.2846 35 Prado c 121 102.6 1.422 1.1156 36 Q Calambe 119.1 102.6 1.477 1.16 37 Q El Cobre c 121 102.6 1.275 1.0246 38 Q Olini c 121 102.6 1.166 1.0497 56 Chivata 132.8 127 3.144 1.2157 64 Cormichoque-1 122.4 122.3 2.029 1.1797 14 Floresta 132.6 127 1.024 1.0377 18 Guaca 132 127 1.259 1.0757 26 Matanza 128.8 127 1.716 1.137 54 Tibasosa 132.8 127 2.046 1.1817 55 Tunja 133.5 127 2.204 1.1857 78 Tunja-1 132 127 2.576 1.2138 63 Bolivar-1 Corrales-1 132 127 2.071 1.1868 6 Caqueza c 144 127.5 1.215 1.2048 28 Nazareth 131.5 127.5 1.493 1.1558 34 Paz de Rio 144 127 1.385 1.1548 44 Servita 132 127.5 1.565 1.1389 33 Aguazul 142 127.5 1.205 1.3099 9 Chita 139 127.5 2.525 1.4829 7 Cocuy 139 127.5 3.489 1.6579 20 Guateque 138 127.5 1.035 1.4059 23 Labateca 132 127.5 1.478 1.2859 69 Medina-1 144 127.5 3.605 1.118

394 L.F. Sarmiento-Rojas et al. / Journal of South American Earth Sciences 21 (2006) 383411

vian

(Ma

aph

th ATable 2 (continued)

Basin compartmenta Stratigraphic columnb BerriasianHauteri

Start (Ma) End

9 27 Mojicones 139 127.59 32 Pajarito 142 127.59 40 R Cusay 136.5 127.59 46 San Luis de Gaceno c 144 127.59 49 Sogamoso 136.5 127.510 67 Cusiana-1X-2

Basin compartment numbers as shown in Fig. 6. Location of each stratigra Numbers indicate Cretaceous subbasins shown in Fig. 6.b Stratigraphic column number shown in Fig. 2.c Modeled using Triassic and Jurassic actual or inferred stratigraphy.

L.F. Sarmiento-Rojas et al. / Journal of Soucalculations, we use a 1D numerical nite dierence model,which allows us to incorporate nite and multiple stretch-ing phases. To handle the large number of wells and stretch-ing phases in the forward model, we apply a numericaltechnique (Van Wees et al., 1996b), which automaticallynds the best t stretching parameters for (part of) the sub-sidence data. In this procedure, we must specify the timingand duration of the rift phase, whereas the best t stretch-ing values emerge by searching for the minimum of themean square root F of the deviation in predicted andobserved subsidence (Fig. 9), as a function of d and b, asfollows:

F d; b 1num

Xinumi1

spp;i so;i2vuut ; 1

where num is the number of subsidence data used in the t-ting procedure, and sp,i and so,i are predicted and observedsubsidence values, respectively. For a rift phase, either uni-form lithospheric stretching (d = b) (McKenzie, 1978) or

Fig. 9. Outline of the forward modeling technique. ExpEvent Aptianearly Albian event

) Stretchingfactor

Start (Ma) End (Ma) Stretchingfactor

b d b d

1.36 1.231.169 1.3031.643 1.351.076 1.3521.459 1.354

112.3 112.2 1.095 1.025

ic column identied by number shown in Fig. 2.

merican Earth Sciences 21 (2006) 383411 395two-layered stretching (d b) (e.g., Royden and Keen,1980) can be used. For uniform stretching, the solutionto Eq. (1) requires that at least one observed subsidencedata point is given after the onset of rifting, whereas two-layered stretching requires at least two data points. Forpolyphase stretching, the t is accomplished in sequentialorder. Initially, using a steady state thermal and composi-tional lithospheric conguration (cf. McKenzie, 1978),stretching parameters of the rst phase are determined bytting data points in the syn- and postrift time interval tothe onset of the following phase. Subsequently, using theperturbed lithosphere conguration predicted at the onsetof the second rift phase, stretching parameters are deter-mined using subsidence data from its syn- and postrift timeintervals to the next rifting phase.

4.2.2. Modeling procedure

In the tting procedure, initial lithospheric congurationand thermal parameters are adopted as listed in Table 1.To t the data, we assume that each observed phase of

lanation in the text (from Van Wees et al., 1996b).

th Arapid tectonic subsidence should correspond to a stretchingphase in the forward model. For these phases, we adopt atwo-layered stretching model of the lithosphere (d b) toobtain the highest degree of freedom. However, we preferto use a uniform stretching model (d = b) when uncertaintyexists in estimating the stretching factors due to fewer datapoints or for relatively large age uncertainties, as is the casefor the Triassic and Jurassic sedimentary record.

Using the starting and nishing times previously deter-mined for the stretching events, we calculate the litho-sphere stretching factors that would produce theoreticalsubsidence curves similar to those observed. For the for-ward modeling, we include for most of the modeled loca-tions the complete Mesozoic sedimentary section sincethe Triassic, even in those columns in which the pre-Me-sozoic section is probably deep and does not crop out. Inthese cases, we use thicknesses interpolated from isopachmaps.

5. Results

In Fig. 10, we show the tectonic subsidence curves. Theforward-modeled tectonic subsidence curves (Fig. 11) indi-cate a remarkably good t with the subsidence data. Thelithosphere, crustal, and subcrustal stretching factors calcu-lated for each stretching phase also are plotted in map view(Figs. 7, 8, and 12).

6. Tectonic subsidence and lithosphere stretching during

Triassic and Jurassic time

For the tectonic subsidence of the Triassic sedimenta-ry record, we assume, following Geyer (1982), that thepoorly fossiliferous sediments (El Sudan Formation) ofthe Cienaga de Morrocoyal north of the Serrana deSan Lucas with lithology and relative stratigraphicalposition similar to the Luisa Formation of the Payanderegion are Triassic and time correlative and that Trias-sic sediments accumulated in the western ank of theEC. Using subsidence patterns, we attempt to dierenti-ate fast subsidence events. However, these attemptsshould be considered preliminary, because the continentalTriassic and Jurassic poorly fossiliferous sedimentaryrecord has relatively scarce and low-quality biostrati-graphic data that make it dicult to establish clear timeboundaries between the events.

6.1. Basin compartments

Dierences in the shape of the subsidence curves in dif-ferent areas conrm that TriassicJurassic sedimentationoccurred in two separate basin compartments, each withits own subsidence history and sedimentary ll:

1. Payande-San Lucas terranes (UMV and Cienaga de

396 L.F. Sarmiento-Rojas et al. / Journal of SouMorrocoyal, region A, Fig. 3).2. Eastern Cordillera (region B, Fig. 3).6.2. TriassicJurassic subsidence and lithosphere stretching

events

6.2.1. Early? Triassic event (variable in dierent columns,

248 to 235 Ma, time scale of Gradstein and Ogg, 1996)This event is best represented in the UMVCienaga de

Morrocoyal with uniform stretching factors; b = d reachvalues of 1.23. Subsidence curves (Fig. 10), thicknessvariations, and the spatial distribution of stretching val-ues (Fig. 12a) and paleomagnetic data (Bayona et al.,2005) suggest that small, narrow (150 km), transcurrentrift basins formed. Abrupt lateral changes of sedimentthickness and facies in the UMV suggest dierential sub-sidence in dierent faulted blocks (Bayona et al., 1994;Mojica et al., 1996), which is common in transcurrentfaults. Jurassic normal faults are described by Guillande(1988).

The transtensional event correlates in time with mag-matic arc activity in the Central Cordillera, interpretedby Aspden et al. (1987) as related to oblique subduction.The event also may have been related to intracontinentalrifting (breakup of Pangea), as proposed elsewhere, par-ticularly the separation between South and North Amer-ica (e.g., Pindell and Dewey, 1982; Ross and Scotese,1988; Cediel et al., 2003). However, considering that rif-ting and subsequent sea oor spreading, which originatedthe Gulf of Mexico, occurred during the Late Jurassic(Pessagno and Martin, 2003), this event probably relatesto local transtensional events along the continentalmargin.

6.2.2. Latest TriassicMiddle Jurassic event (208 to185 Ma)

Subsidence curves (Fig. 10), stratigraphic thickness, mapdistribution of stretching values, paleomagnetic data (Bay-ona et al., 2005), and fault distribution (Fig. 12b) suggesttwo narrow (

and subsequent sea oor spreading, which originated theGulf of Mexico, occurred during the Late Jurassic(Pessagno and Martin, 2003), so this event probably isrelated to local transtensional events along the continen-tal margin.

6.2.3. Middle Jurassic event (180 to 176 Ma)Paleogeography and stratigraphic thickness distribu-

tions indicate continued widening of transtensional basins,though they remained relatively narrow. Major depocen-ters developed in the MV and western ank of EC within

ord.

L.F. Sarmiento-Rojas et al. / Journal of South American Earth Sciences 21 (2006) 383411 397Fig. 10. Tectonic subsidence curves from the Mesozoic sedimentary rec

subsidence in meters obtained from backstripping analysis. Vertical shaded striin Fig. 6. Note vertical bars represent the fast tectonic subsidence events.Horizontal axis represents age in Ma. Vertical axis represents tectonic

ps represent fast subsidence events. Numbers refer to basin compartments

th A398 L.F. Sarmiento-Rojas et al. / Journal of Souelongated NNE transcurrent grabens on opposite sides ofthe MagdalenaLa Salina fault system. The distributionof stretching b = d factor values (Fig. 12c) and paleomag-

Fig. 10 (conmerican Earth Sciences 21 (2006) 383411netic data (Bayona et al., 2005) indicate rift basins locatedalong the present-day western ank of the EC, with b = dvalues up to 1.39, and the paleo-MV. Moreover, large

tinued)

th AL.F. Sarmiento-Rojas et al. / Journal of Soupostrift Cretaceous subsidence along the western ank ofthe EC northwest of Bogota can be explained only as ther-mal subsidence after a Jurassic stretching event. Therefore,these basins extended southward into the Cundinamarcaregion, possibly limited by the paleo-La Salina, paleo-Suarez, and paleo-Boyaca fault systems. Small, isolatedgrabens developed in the Santander Massif (Geotec,1992; Kammer, 1993), Perija (Shagam, 1975; Maze,1984), Merida Andes (Ricardi et al., 1990, in Mojicaet al., 1996), LLA (Numpaque, 1986, in Cooper et al.,1995; Geotec, 1992), and Maracaibo (Shubert and Ricardi,1980, in Mojica et al., 1996) areas. Volcanic activitydecreased at this time, mainly occurring in the MeridaAndes (basalts in La Quinta Formation, Maze, 1984).

These TriassicJurassic events correlate with magmaticarc activity in the Central Cordillera and Santander Massif.The calc-alkaline composition of these magmatic rocks sug-gests a magmatic arc related to high-angle subduction (Asp-den et al., 1987). Themost developed calc-alkalinemagmaticplutonic bodies of the Payande-San Lucas terranes (i.e.,Ibague and Segovia batholiths) tend to be adjacent to thewest of these transtensional basins, which indicates thesetranstensional basins were located in a backarc setting closeto the arc or an intraarc setting. Alternatively, these eventsmay have been related to intracontinental rifting (breakupof Pangea), as proposed elsewhere, particularly the separa-tion between South and North America (e.g., Pindell andDewey, 1982; Ross and Scotese, 1988; Cediel et al., 2003).

Fig. 11. Forward modeled tectonic subsidence (continuous linmerican Earth Sciences 21 (2006) 383411 399However, considering that rifting and subsequent sea oorspreading, which originated the Gulf of Mexico, occurredduring the Late Jurassic (Pessagno andMartin, 2003), theseevents probably relate to local transtensional events alongthe continental margin. Paleomagnetic data suggest thatduring the Late Triassic andEarly Jurassic, the SantaMarta,SantanderMassif, FlorestaMassif, San Lucas, and Payande(UMV)magmatic arc segments aligned into amagmatic arc,striking parallel to the subduction zone (Bayona et al., 2005).This magmatic arc belt formed as a result of subductionalong western Pangea, which resulted in the southward con-tinuation of the early Mesozoic continental magmatic arcfrom the southwestern United States to Guatemala (Bayonaet al., 2005). In the eastern part of the Chibcha terrane (ECeast of the Bucaramanga fault), the main calc-alkaline plu-tonic bodies developed on the Santander Massif are posi-tioned between the transtensional basins, suggesting aninterarc setting.

7. Tectonic subsidence and lithosphere stretching during

latest Jurassic and Cretaceous

7.1. Basin compartments

Tectonic subsidence curves (Fig. 10) and restored thick-ness maps (Figs. 7a, b, and 8a) indicate several basin com-partments (Fig. 6). In the northern part of the EC (Fig. 6),we recognize

e) and observed tectonic subsidence (dots) curves (in m).

th A400 L.F. Sarmiento-Rojas et al. / Journal of Sou1. Two subbasins: the Cocuy (region 9) andTablazo (region 3) graben subbasins, separatedby the less subsiding Santander-Floresta block(region 7), and

2. A regional westward decrease in tectonic subsidencewith a maximum in the Cocuy subbasin (region 9)and a minimum in the Middle MV (region 2), whichsuggests a regional half-graben geometry for the wholebasin.

Fig. 12. (a) Contour map of lithosphere stretching factors (b = d) calculated thassuming there are Triassic sediments in the Cienaga de Morrocoyal area (GeyeEC, without palinspastic restoration. Distribution of main early Mesozoic faucalculated through forward modeling for the Early Jurassic (208185 Ma) strMesozoic faults is shown. (c) Contour map of lithosphere stretching factors (bstretching event without palinspastic restoration. Distribution of main Early Met al., 1976), western part of Chibcha terrane (Toussaint, 1995a); (B) eastern pMedelln; Mz, Manizales; I, Ibague; N, Neiva; C, Cucuta; Bu, Bucaramanga;merican Earth Sciences 21 (2006) 383411In the southern part of the Cordillera at the latitude ofBogota, we recognize

1. A single extensional basin, the Cundinamarca subbasin(region 4, Fig. 6), and

2. BerriasianHauterivian tectonic subsidence, maximal inthe eastern side of the Cundinamarca subbasin (region8, Fig. 6), which indicates that a rst stretching event

rough forward modeling for the Triassic (248.2235 Ma) stretching event,r, 1982) and that Triassic sediments accumulated in the western ank of thelts also shown. (b) Contour map of lithosphere stretching factors (b = d)etching event without palinspastic restoration. Distribution of main early= d) calculated through forward modeling for the Jurassic (180.1176 Ma)esozoic faults also shown. (A) Payande, San Lucas terranes (Etayo-Sernaart of Chibcha terrane (Toussaint, 1995a) and Guyana shield. Cities: M,T, Tunja; B, Bogota; V, Villavicencio; Y, Yopal; A, Arauca.

th Amainly aected the eastern Guaicaramo normal faultsystem. During the Aptian, however, subsidence wasmaximal in the western side (region 4, Fig. 6), so a sec-ond stretching event mainly aected the western Bituimafault system. The total tectonic subsidence during theCretaceous was slightly greater on the western side ofthe basin (region 4).

In the western part on the Central Cordillera, a N-S(present day geometry) depocenter represents the Quebra-dagrande Complex marginal basin (Nivia et al., 1996;Nivia and Gomez, 2005).

Tectonic subsidence in the UMV (region 6, Fig. 6),where marine sedimentation started in Aptian times, is sig-nicantly less than that of the EC and Middle MV (region2). In the easternmost LLA (region 10, Fig. 6), sedimenta-tion started during the Late Cretaceous, and total tectonicsubsidence during the Cretaceous was small compared withthe EC and MV. Subsidence probably occurred throughexural thermal subsidence (Watts et al., 1982) and waterloading due to an increase in paleowater depth.

7.2. Latest JurassicEarly Cretaceous fast subsidence and

lithosphere stretching events

7.2.1. Latest JurassicHauterivian event (variable in

dierent stratigraphic columns, 144127 Ma)

This event occurred in the area of the EC and is best rep-resented in its eastern ank and west of the Bucaramangafault. Subsidence curves (Fig. 10), thickness maps (Figs.7a, b, and 8a), paleomagnetic data (Bayona et al., 2005),and the distribution of crustal d stretching factors(Fig. 7c) evidence a wide (>180 km), asymmetrical, trans-tensional half-graben basin divided by the Santander-Flor-esta high. Maximum tectonic subsidence and crustalstretching d up to 1.66 was associated with the pre-Guaica-ramo normal master fault system, the eastern boundary ofthe graben (Figs. 7a and c). Strong changes in the thicknessof the Giron Formation from 0 to more than 4 km acrossENE-WSW- and NE-SW-striking normal faults and localvertical axis rotations of fault-bounded blocks during thistime, as documented by paleomagnetism, suggest an impor-tant control of normal faulting on deposition and basinconguration during the deposition of continental beds ofthe Giron Formation (Bayona et al., 2005; Fig. 10,Tablazo-Lebrija). A second-order half-graben was locatedat the current location of the western ank of the EC withcrustal stretching values up to 1.45 (Fig. 7c). This minorhalf-graben probably was associated with a paleonormalfault system, approximately following La Salina-Bitumafault system that was its western border. To the south, therewas only one depocenter, limited to the south by a NW-SEvertical transfer fault. Early Cretaceous turbidites at bothanks (Murca Formation, Caqueza Group) of the exten-sional basin can be taken as evidence of tectonic instability

L.F. Sarmiento-Rojas et al. / Journal of Souassociated with normal faulting movement. Branquet(1999) presents outcrop and seismic evidence of Cretaceousnormal faulting. Normal faults imaged on seismic sections(Fig. 13) conrm extensional tectonic movements that attestthat this rapid subsidence event was produced by litho-sphere stretching. As suggested by paleomagnetic data(Bayona et al., 2005), these faults may have had a dextralstrike-slip component. Unlike the TriassicJurassic trans-tension, magmatic activity within the basin was reducedduring the Early Cretaceous. Evidence of Early Cretaceousmagmatism is limited to small, mac, igneous intrusionsdescribed by Fabre and Delaloye (1983) and Moreno andConcha (1993) and some volcanic input within Cretaceousshales (Rubiano, 1989; Villamil, 1994).

Small mac intrusions described by Fabre and Delaloye(1983) coincide with areas of thin crust (crustal stretchingfactors >1.4) and places of maximum stretching of the sub-crustal lithosphere (Fig. 7c). As a consequence of thedepth-dependent lithosphere rheology assumed by the mod-el, the results suggest thatmore intense stretching aected thesubcrustal mantle lithosphere (Fig. 7d). Dierences betweencrustal and subcrustal stretching factors suggest somedecou-pling occurred between the crust and the subcrustal litho-sphere or that increased thermal thinning aected themantle lithosphere. The latter interpretation implies a con-siderable thermal anomaly produced by mantle lithospherethinning,which seems supported by the presence ofmagmat-ic mac intrusions. During rifting, stress-induced litho-sphere thinning causes adiabatic decompression of thelower lithosphere and asthenosphere, their partial melting,and the diapiric rise of melts into the zone of thinned litho-sphere (Wilson, 1993). Although the 1D model cannot pre-dict regional isostatic eects, the Lower Cretaceousunconformity on the rift margins (e.g., LLA) and locallyon horst blocks (e.g., Santander-Floresta paleo-Massif)probably was produced by thermal uplift of rift shoulders,as suggested by the subcrustal stretching values. In addition,Jurassic magmatic activity in the Santander Massif contrib-utes to relative thermal uplift and reduces tectonic subsi-dence. According to Ziegler (1994), unconformities on riftshoulders and intrabasinal fault blocks can be attributed toa footwall uplift in response to extensional unloading ofthe lithosphere. In general, the location of subcrustal andcrustal stretched zones coincides, as a consequence of the1D model assumption of local isostasy. However, wherethere is some oset, it indicates asymmetry in the basin, assupported by the general geometry of the basin. On the basisof subsidence analyses of Cretaceous stratigraphic columnsof the EC, Fabre (1987) and Hebrard (1985), using theinstantaneous stretching model of McKenzie (1978), calcu-late uniform b = d stretching factors up to 2 for the wholelithosphere. However, they lump the Cretaceous stretchingevents into a single instantaneous stretching event withan innite extension rate. The higher stretching values theyobtain is a logical consequence of combining several stretch-ing events with nite extension rates.

According to the Cretaceous passive margin interpreta-

merican Earth Sciences 21 (2006) 383411 401tion (Pindell and Erikson, 1993), active opening of theproto-Caribbean occurred north of Colombia and west of

th A402 L.F. Sarmiento-Rojas et al. / Journal of Southe paleo-Central Cordillera.Moreno and Pardo (2003) cor-relate this latest JurassicEarly Cretaceous extensionalphase, which originated the proto-Caribbean basin accord-ing to Pindell and Erikson (1993) reconstruction, with theQuebradagrande Complex marginal basin of Nivia et al.(1996) and Nivia and Gomez (2005). According to Moreno

Fig. 13. Seismic sections in the Medina foothills area, along the eastern bordCretaceous (K) in the Guaicaramo paleofault system along the eastern bordePaleogene evidenced by lateral changes of thickness of the Paleogene Carbosedimentary ll (Linares, 1996). Location of seismic line is shown in Fig. 2.merican Earth Sciences 21 (2006) 383411and Pardo (2003), subduction of the western Quebrada-grande proto-Caribbean crust beneath the Farallon plateoriginated a magmatic arc. To the east, a passive margin dis-tant from the magmatic arc prevailed in the eastern side ofthe Quebradagrande Basin and east of the paleo-CentralCordillera (Moreno and Pardo, 2003, their Fig. 7). If such

er of the Cundinamarca subbasin. Note normal fault evidence during ther and contractional inversion of Cretaceous extensional faults during thenera and Mirador formations. Note thickness changes in the Cretaceous

th Aan interpretation is valid for the BerriasianHauterivian, theevent may be the result of stretching in the study area, whichproduced a failed-rifted arm related to a major opening ofthe proto-Caribbean oceanic basin. However, the presenceof some Early Cretaceous plutonic rocks in the Central Cor-dillera east of the Quebradagrande Complex (i.e., lower Cre-taceous K/Ar ages of the Samana and Mariquita stocks,Vesga and Barrero, 1978) is dicult to explain with a passivemargin hypothesis. One possibility is that these ages are notreliable. An alternative plate tectonic reconstruction attri-butes Early Cretaceous extension of the paleo-EC to a back-arc basin contemporaneous with reduced magmatic activityin the Central Cordillera. The proto-Caribbean eastern partof the Quebradagrande Complex basin crust may have sub-ducted beneath the Central Cordillera to develop amagmat-ic arc. Although these alternative interpretations arecontroversial, the following evidence supports an EC basinlocated behind a partially emerged, less subsiding paleo-Central Cordillera (magmatic arc?):

1. Early Cretaceous igneous intrusions in the Central Cor-dillera east and west of the Quebradagrande Complex(e.g., San Diego, Cambumbia, and Mariquita stocks,Restrepo et al., 1991) dene one or two (?) magmaticarcs (produced by subduction of the W and E bordersof the Queberadagrande basin crust?). However, suchmagmatic arcs are not well dened.

2. The presence in the western part of the Cundinamarcasubbasin of Lower Cretaceous sandstones with abun-dant volcanic lithic fragments and feldspar derived froma western detrital source area, as indicated by paleocur-rent data (Murca Formation and Utica sandstone; Sar-miento, 1989; Moreno, 1990, 1991).

3. Volcanic lithic fragments in the mid-Cretaceous Cabal-los Formation of the UMV (Guerrero et al., 2000).

4. Progressive westerly onlap terminations of Cretaceouscarbonates on the basement, observed in seismic linesin the western border of the Cesar Valley, northernColombia (in Mesozoic times, part of the EC basinalarea, Fig. 1; Audemard, 1991).

5. Stratigraphical and petrographical evidence suggestingthat during the Berriasian (?)Valanginian, clastic sedi-ments near San Felix in the western ank of the CentralCordillera (between Romeral and Palestina faults) camefrom erosion of uplifted areas with metamorphic rocksand small tectonic blocks with plutonic rocks (Rodr-guez and Rojas, 1985).

6. Cretaceous volaniclastic rocks in the Central Cordillera(Rodrguez and Rojas, 1985) consisting of mixtures ofpyroclastic and epiclastic fragments, probably derivedfrom a magmatic arc.

7. Relatively high concentration of volcanogenic clay min-erals in HauterivianBarremian (030%), middle Albian(021%), and Turonian (69%) shales of the VilletaGroup (Rubiano, 1989) and the ValanginianHauteri-

L.F. Sarmiento-Rojas et al. / Journal of Souvian Rosablanca Formation (Moreno, 1989, in Rubiano,1989) in the Cundinamarca subbasin. Thin beds of volca-of South America between 100 and 80 Ma. However, thisinterpretation contradicts strong stratigraphic evidence ofa subsiding basin at these localities. The zircon ages reectlocal uplift of faulted blocks located at rift margins, asdemonstrated by ssion-track data from several rift basins(Van der Beek, 1995). Van der Beek (1995) explains riftmargin uplift by mechanical support of rift anks, resultingfrom an upward state of exure. Evidence of magmaticactivity in the basin is limited to some small mac intru-sions (Fabre and Delaloye, 1983) and minor volcanic input(Rubiano, 1989; Villamil and Arango, 1998).

Plate tectonic interpretations by Meschede and Frischnogenic clays or bentonites in the CenomanianTuro-nian stratigraphic interval (Villamil and Arango, 1998)and Salada Member of the La Luna Formation (Patter-son, 1970, in Rubiano, 1989), as well as subaqueous vol-canic tus in La Frontera and La Luna formations in theMV (Restrepo-Pace, personal communication).

8. Jurassic (185 Ma) and Cretaceous (P77 Ma) zircon s-sion-track ages from the Central Cordillera (Toroet al., 1999; Gomez et al., 1999), evidence of uplift. Inthe Ecuadorian Andes (Rivadeneira, 1996) and CentralCordillera (Rodrguez and Rojas, 1985), uplift or defor-mation have been suggested during the Late Cretaceous.

7.2.2. Aptianearly Albian event (121102 Ma)

This fast subsidence event mainly occurred at the south-ern part of the western ank of the EC and the UMV, indi-cating asymmetry in the basin. In other localities, thermalsubsidence is due to lithospheric cooling after prior litho-spheric stretching events. During the BarremianAptian,the basin extended to the south in the UMV (Figs. 8aand 10). Turbiditic deposits of Aptian age (Socota Mem-ber, Polana and Rodrguez, 1978) indicate tectonic insta-bility associated with the rapid subsidence event. Theisopach map (Fig. 8a) suggests a master normal fault sys-tem, located approximately at the present-day BituimaMagdalena fault system, was active. In the UMV, a normalpaleo-Chusma fault system probably also was active.Crustal stretching factors up to 1.4 are associated withthe southern segment of the Bituima fault system and thoseup to 1.2 with the UMV (Fig. 8b). Because of the depth-de-pendent rheology assumed by the model, the results suggestthat stretching aected the subcrustal mantle lithospheremore strongly. Subcrustal stretching values reach 3.24 atthe southwestern ank of the EC and 1.6 at the UMV(Fig. 8c). Dierences between crustal and subcrustalstretching values suggest some decoupling between crustand subcrustal lithosphere or that an increased thermalthinning aected the mantle lithosphere. These resultsimply a thermal anomaly that probably is responsible forrift shoulder uplift, as evidenced from ssion-track databy Van der Wiel (1991) in the UMV and Garzon Massif(Fig. 1). Van der Wiel (1991) interprets these ages as relatedto an orogenic event that aected the northwestern corner

merican Earth Sciences 21 (2006) 383411 403(1998) assume the beginning of subduction of the Faral-lon/Pacic plate under the Panama-Costa Rica arc west of

th AColombia during the Aptianearly Albian. The alternativehypothesis of Pindell and Erikson (1993) assumes that dur-ing the Aptian, the proto-Caribbean lithosphere began tosubduct under the Amaime-Antilles arc, which wasapproaching thewesternmargin of northern SouthAmerica.

7.2.3. Cenomanian subsidence event (9893 Ma)

This event occurred in the eastern ank of the EC. TheLate CenomanianTuronian global sea level maximum cor-relates with it, suggesting that an increase in subsidence wasdriven by water load. However, the event aected only theeastern ank of the Cordillera, where the maximum Creta-ceous thickness appears. At other localities, thermal subsi-dence resulted from lithospheric cooling after lithosphericstretching events. Although the maximum ooding surfacefor the Cretaceous sediments of the EC is the Cenoma-nianTuronian boundary (Villamil and Arango, 1998), it isCampanian in the eastern LLA (Fajardo et al., 1993; Cooperet al., 1995; Fig. 5). If subduction of the Caribbeans thickand buoyant lithosphere (Duncan and Hargraves, 1984;Nivia, 1987; Hill, 1993; Meschede and Frisch, 1998) underSouthAmericawas inhibited, it initiated uplift of theCentralCordillera, as evidenced by Late Cretaceous ssion-trackages from the Central Cordillera (Gomez et al., 1999; Toroet al., 1999) and exerted horizontal stresses on the northwest-ernmargin of SouthAmerica.Horizontal stresses can inducelocal exural lithosphere bending, which is maximal wherethe lithosphere is weakest (Cloetingh, 1988; Cloetingh andKooi, 1992). This process probably enhanced the relativesea level rise, creating a maximum CenomanianTuronianmarine ooding surface in the depocenter of the EC, charac-terized by weak lithosphere due to earlier stretching. In con-trast, horizontal stress produced a submarine shallow waterdepth bulge in the LLA, which partially compensated for themaximum eustatic signal.

7.2.4. MaastrichtianPaleocene event (variable in dierent

columns, 6854.8 Ma)This fast subsidence event aected the axial part of the

EC, its eastern ank, and, locally, the westernmost partof the LLA. This event correlates in time with deformationand uplift in the Central Cordillera (Jaramillo, 1978, 1981;Cooper et al., 1995).

All plate tectonic interpretations agree that during thelatest Cretaceous and probably Paleocene, accretion ofthe Western Cordillera oceanic terranes along the Cau-ca-Patia fault occurred, producing deformation and upliftof the Central Cordillera. Increased subsidence in theaxis of the Cundinamarca subbasin (Sabana de Bogota)could result from increased horizontal compressionalstress (Cloetingh, 1988; Cloetingh and Kooi, 1992) asso-ciated with the collision of the oceanic terranes of wes-tern Colombia and deformation and uplift of theCentral Cordillera. Development of normal faults inthe LLA area, as suggested by Kluth et al. (1997), could

404 L.F. Sarmiento-Rojas et al. / Journal of Soube the result of local tensional stresses in the developedexural bulge.In conclusion, Jurassic (and probably post-Triassic)fast tectonic subsidence events are related to backarc tointra-arc transtensional lithospheric stretching, with a pre-dominant dextral strike-slip component. EarlymiddleCretaceous fast tectonic subsidence events seem relatedto backarc lithosphere stretching behind a poorly devel-oped, subduction-related magmatic arc. An increase indistance from the backarc basin to the magmatic arc fromthe Jurassic to the Cretaceous has been observed else-where in basins that evolved from an initial intra-arc stageto a later backarc stage (Smith and Landis, 1995). LateCretaceous subsidence was mainly thermal subsidenceafter the previous lithosphere stretching events, and local-ized discrete subsidence events probably result from litho-sphere exural bending due to horizontal compressivestresses related to accretion of the oceanic terranes in wes-tern Colombia.

8. Discussion

8.1. Relationships between Mesozoic rifting and magmatism

In the study area, abundant Upper TriassicLowerJurassic volcanic rocks are associated with moderatestretching factors (b = d up to 1.12). In contrast, theCretaceous sedimentary record is almost devoid of volca-nic rocks (only volcanic tus and minor mac intrusions)and associated with higher stretching factors (b up to 3,d up to 1.66). Thermal processes were more importantthan mechanical stretching during Late TriassicEarlyJurassic rifting than during Cretaceous extension. Duringthe Late TriassicEarly Jurassic, abundant volcanic rockssuggest a positive thermal anomaly in the lithosphere butmoderate lithosphere stretching. TriassicJurassic uncon-formities could have been produced by thermal uplift(active rifting?). Thermal doming results from progres-sive thinning of the higher density mantle lithosphereand its replacement by low-density asthenosphere (Bott,1992). In contrast, during the Cretaceous, the less abun-dant volcanic rocks, absence of tectonically controlledunconformities, and large amount of tectonic subsidencesuggest an absence of thermal doming. Small mac intru-sions coincident with places of maximum crustal andmantle subcrustal stretching also suggest modest magma-tism as a consequence of the extension of the lithosphere(passive rifting). The abundant Late TriassicEarlyJurassic volcanic rocks vary in composition from felsicto mac. Chemical analyses of La Quinta Formation vol-canic rocks indicate a calc-alkaline composition in theAFM diagram and alkaline composition in the alkali-sil-ica diagram (Toussaint, 1995b). Chemical analyses of theSaldana Formation indicate a calc-alkaline composition,probably generated in a backarc environment (Bayonaet al., 1994). The predominance of calc-alkaline composi-tions seems to suggest a convergent-related backarc

merican Earth Sciences 21 (2006) 383411extension rather than intracontinental rifting (Toussaint,1995b).

th AL.F. Sarmiento-Rojas et al. / Journal of Sou8.2. Correlation between fast subsidence events and

subduction-related magmatic arcs

The inferred Mesozoic stretching events seem to corre-late in time with reduced magmatic activity in the CentralCordillera (Fig. 14, modied from Aspden et al., 1987;Guillande, 1988). However, this preliminarily correlationshould be tested with more and better data. If the calc-alkaline (Alvarez, 1983) plutonic belts of the CentralCordillera and Santander Massif developed as subduc-tion-related magmatic arcs during Mesozoic times, as sug-gested by Aspden et al. (1987), the extensional ortranstensional basins behind them or close to them maybe interpreted as backarc or intra-arc basins. Extensional

Fig. 14. Event correlation between lithosphere stretching in the area of the ETriassicEarly Jurassic, (3) Middle Jurassic, (4) Early Cretaceous BerriasianHpreliminary because original data are heterogeneous; 94% are K-Ar (biotite, horor whole-rock). (a) Left panel: principal structural/plutonic zones of western Cactivity in western Colombia (modied after Aspden et al., 1987). (b) Cumulatiafter Guillande, 1988). Periods of intense magmatic activity are characterdeterminations for a time interval (low slope).merican Earth Sciences 21 (2006) 383411 405backarc basins develop when the velocity rollback, due tofast subduction, exceeds the oceanward convergencevelocity of the overriding plate (Royden, 1993a,b). Ifmagmatic arc activity decreases with the oceanward con-vergence velocity of the overriding plate, during times ofreduced magmatic arc activity, a constant rollback veloc-ity exceeds that velocity, increasing extension and subsi-dence in the backarc region. According to Aspden et al.(1987), the Triassic magmatic belt was controlled alongstrike-slip faults, as also supported by Restrepo-Pace(1995). The Jurassic magmatic arc may have been con-trolled by strike-slip faults, as suggested by paleomagneticdata (Bayona et al., 2005) and its elongated shape inmap view, parallel to major strike-slip faults of major

C and magmatic activity in the Central Cordillera. (1) Triassic, (2) Lateauterivian, and (5) Aptian events. This correlation should be considerednblende, muscovite, or whole-rock), and 6% are Rb-Sr (hornblende/biotiteolombia. Right panel: age distribution of Mesozoic and Cenozoic plutonicve histogram of radiometric ages of plutonic bodies in Colombia (modiedized by rapid increase in the cumulative number of radiometric age

th Acalc-alkaline plutonic bodies (i.e., Ibague and Segoviabatholiths). Along these faults, part of the magmatic beltmight have moved northward, as suggested by paleomag-netic data (Bayona et al., 2005). Jurassic calc-alkaline plu-tonism and volcanism along the Central Cordillera andUMV is interpreted by Aspden et al. (1987) and Bayonaet al. (1994) as a subduction-related magmatic arc, butCretaceous plutonism developed sporadically only in thenorthern part of the Central Cordillera (Restrepo et al.,1991), whereas it is very extensive in Peru (Cobbing,1982, in Aspden et al., 1987). Aspden et al. (1987) suggestoblique convergence and an oset in the subduction zonealong a major NE-SW transform fault to account for thenotable absence of Cretaceous plutonism in southernColombia and Ecuador.

8.3. Geometry of transtensional basins

8.3.1. Triassic and Jurassic

On the basis of paleomagnetic data, Bayona et al. (2005)suggest that the Payande terrane (UMV) was located southof equator (close to 10 S) before the Middle Jurassic; Sem-pere et al. (2002) recognize Late PermianMiddle Jurassictranstensional rifting in the EC of Peru and Bolivia (822 S). The early Mesozoic stratigraphy of the PeruvianBolivian rift basin is very similar to that of the UMV(Payande terrane, sensu Etayo-Serna et al., 1983). In Peruand Bolivia, Late PermianTriassic, red to purple, coarse-grained, continental, synrift deposits (Mitu Group) resem-ble those of the Luisa Formation of the UMV; LateTriassicEarly Jurassic (Norian-Liassic), marine carbon-ate-dominated postrift deposits (Pucara Group) resemblethose of the Payande Formation of the UMV; and Jurassic,brown-red mudstones and coarse-grained alluvial deposits(Sarayaquillo Formation) resemble the sedimentary com-ponent of the Saldana Formation of the UMV. The simi-larity of the early Mesozoic stratigraphy in these areasinitially was recognized by Geyer (1982). Synrift depositsof Peru and Bolivia commonly are interbedded with locallydominant volcanic and volcaniclastic rocks and/or intrud-ed by subvolcanic to plutonic rocks (Sempere et al., 2002).A similar association with volcanic, volcaniclastic, and plu-tonic rocks is observed in the UMV. Paleomagnetic data(Bayona et al., 2005) and the similarity of the early Meso-zoic stratigraphy and basin tectonic setting of the Payandeterrane and PeruBolivia rift probably indicate that bothareas were adjacent before the Middle Jurassic. In Peruand Bolivia, rifting events occurred during the Late Perm-ianMiddle Jurassic interval (Sempere et al., 2002) and sep-arated during the Late TriassicEarly Jurassic, thermal sag,postrift event during deposition of the Pucara Group (Sem-pere et al., 2002). Similar punctuated rift and lithospherestretching events occurred in the UMV (Payande terrane).Geographically, however, between these Colombian andPeruvian/Bolivian transtensional basins in Ecuador during

406 L.F. Sarmiento-Rojas et al. / Journal of SouJurassic, the accretion of terranes preserved as highlydeformed metamorphic rocks has been proposed by Lither-land et al. (1994). These Ecuadorian terranes were aectedby intense ductile deformation and transcurrent faulting(Litherland et al., 1994). The early Mesozoic rift systemsof the Payande-San Lucas terranes in Colombia and thePeruvian/Bolivian EC represent transtensional segmentsseparated by a transpresional node in Ecuador, in linewith the map view geometry of a major dextral strike-slipfault system. In Colombia, the calc-alkaline magmas likelyintruded along active strike-slip faults and shear zones,developing a magmatic arc that may have resulted from ahighly oblique subduction zone.

Unlike the calc-alkaline magmatism of the ColombianTriassicJurassic magmatic arc segments, the Late Perm-ianMiddle Jurassic rift system of the EC of Peru andBolivia is characterized by intense magmatism, predomi-nantly alkaline to tholeiitic, probably related to intraplatemagmatism and lithospheric thinning (Sempere et al.,2002). In southern Peru and Bolivia, the Late PermianMiddle Jurassic rift system is located behind the Jurassicmagmatic arc (Sempere et al., 2002, Fig. 3). In central Peru,plutons probably were emplaced in the rift roots (Sempereet al., 2002). The Colombian Payande-San Lucas terranesmay have originated as fragments of thinned lithosphereseparated from the PeruBolivia rift system by intraplaterifting, which later moved north along strike-slip faults.The Late PermianMiddle Jurassic rift system of the ECof Peru and Bolivia is interpreted by Sempere et al.(2002) as a transcurrent rift system in which transtensionalsegments were separated by transpressional nodes. Thistranscurrent setting also could explain Triassic plutons inPeru and Bolivia with deformation contemporaneous totheir emplacement (Sempere et al., 2002), as well as then-foliated Triassic plutons described in Colombia (Aspdenet al., 1987). However, new paleomagnetic, geochronolog-ical, isotopic, petrological, structural, and stratigraphicresearch in necessary to test this hypothesis.

8.3.2. CretaceousDuring the Cretaceous, the Colombian transtensional

basins increased in width. During the Mesozoic, thestrike-slip component gradually decreased at the expenseof increases in the extensional component, as preliminarilysuggested by paleomagnetic data (Bayona et al., 2005) andlithosphere stretching values. As in other areas (e.g., EastAfrican rift), probably during the initial Cretaceous exten-sional stages, reactivation of crustal discontinuities led tothe subsidence of isolated grabens linked by shear zones.The increase in width probably was the result of increasingstrain, with grabens propagating toward one another, coa-lescing, and evolving into a generally continuous rift sys-tem (Ziegler, 1994). Whether the Colombian Mesozoicextensional basins were pure shear or simple shear exten-sional basins is dicult to demonstrate. Probably bothmechanisms operated; these extensional basin modelsshould be viewed as end-member cases. If the orientation

merican Earth Sciences 21 (2006) 383411of preexisting crustal discontinuities is such that they couldnot be reactivated by the stress system governing the

th Aevolution of the transtensional basin, new faults woulddevelop, and pure shear deformation will likely prevail(Ziegler, 1994). This mechanism may be applicable to theTriassicJurassic extensional basin system in Colombia.However, if the upper crust is weakened by the presenceof preexisting crustal discontinuities and is oriented to favorreactivation under the prevailing tensional stress eld, theywill present zones of preferential strain concentration,which can result in simple shear deformation (Ziegler,1994). This mechanism may explain the development ofthe paleo-EC rift system during the Early Cretaceous: Theeastern side of the extensional basin developed during theBerriasianHauterivian by reactivation of an older Paleozo-ic rift system along the Guaicaramo paleofault (Hossacket al., 1999), and the western side developed by reactivationof earlier transtensional fault systems from the TriassicJurassic. Another alternative is that the eastern and westernmargins of the graben developed along older strike-slipfault systems, which bounded the accretion of tectonic terr-anes west of the Guaicaramo paleofault system during theLate Triassic and Jurassic (Bayona et al., 2005).

The rheological properties of the lithosphere control thedepth at which tensional necking occurs and whether a riftzone is exed up- or downward (Ziegler, 1994). A deep lith-osphere necking level causes upward exure of the riftzone, whereas necking at shallow crustal levels causesdownward exure and the absence of shoulder uplifts(Kooi et al., 1992; Ziegler, 1994). Similar deep levels ofnecking in the eastern side of the Early Cretaceous exten-sional basin system may have generated shoulder uplift inthe LLA area during the Early Cretaceous. Coarse detritalfragments in the Lower Cretaceous Brechas de Buenavista(Pimpirev et al., 1992) and Calizas del Guavio (Conglome-rado de Miralindo, Ulloa and Rodrguez, 1976) formationscould be derived from this graben shoulder. In contrast, inthe western margin of the Early Cretaceous extensionalbasin, sedimentation was more continuous from Jurassicto Early Cretaceous times, which implies the downwardexure of the rift shoulders and thus a shallower level ofnecking during Early Cretaceous times.

On a lithospheric scale, the location of rift systems is con-trolled by the location of weakness zones in the lithosphere,which in turn depends on its thermal state and crustal thick-ness. At crustal scales, the composition, thickness of itsmechanically strong upper layer, and availability of internaldiscontinuities (which can tensionally be reactivated) arealso important controls for rift location (Ziegler, 1994).The overall present-day pattern of these transtensionalbasins for the Cretaceous indicates several grabens orientedNNE-SSW in an en echelon pattern, in contrast with themore N-Soriented Central Cordillera (e.g., Mojica et al.,1996). Some authors (Fabre, 1987; Sarmiento, 1989; Geotec,1992; Mojica et al., 1996) suggest that some NW-SE faultsprobably represented transfer faults. Features such as theNazareth NW-SE fault (Fig. 7c), which limits the Early Cre-

L.F. Sarmiento-Rojas et al. / Journal of Soutaceous basin to the south (Fabre, 1987), or the NW-SEalignment connecting the two emerald districts of the EC(Fig. 7c, Sarmiento, 1989) probably represent Cretaceoustransfer faults. If a subduction-related magmatic arc existedat the current location of the Central Cordillera during theCretaceous, as has been proposed (Aspden et al., 1987; Tous-saint and Restrepo, 1989, 1994; Toussaint, 1995b), and theorientation of that magmatic arc and the extensional basinshas been preserved, their oblique orientation and en echelonpattern would suggest oblique slip extension with a right-lat-eral strike-slip component. However, the available data can-not rule out the hypothesis that some rift arms form acuteangles to the dominant NNE-SSW trend in a pattern similarto aborted aulacogen rifts.

If the inverse or thrust faults that now dene the easternand western borders of the EC originally were Cretaceousnormal faults with a strike-slip component, inverted duringthe Cenozoic, their geometry in map view would providesome information about Mesozoic extensional faults. Lat-eral changes of Mesozoic thickness suggest this is the case,at least for the master faults, which probably dened theregional extensional basin geometry. Adopting this hypoth-esis, we posit that theGuaicaramo,LaSalina, Bituima,Mag-dalena, and Boyaca faults represent original extensional ortranstensional faults. Analog model experiments of obliqueextension produce a similar map view fault pattern (e.g.,Tron and Brun, 1991). NW-SE transfer faults and possiblenormal faults with this orientation, as interpreted by Ecope-trol (1994) in the middle MV, were not inverted during theCenozoic. Some normal faults were passively transportedwith short-cut basement blocks during the Cenozoic inver-sion (e.g., Esmeraldas fault; Cooper et al., 1995).

9. Conclusions

High-resolution backstripping analysis and forwardmodeling show that the Mesozoic Colombian Basin ismarked by ve lithosphere stretching pulses. Periods ofextension seem to correlate in time with gaps of subduc-tion-related magmatic arc activity, as suggested by Aspdenet al. (1987), especially during the Jurassic, which supportsa hypothesis of backarc extension. However, this prelimi-narily correlation must be tested with more and better geo-chronological and biostratigraphical data. If backarcextension continued during the Early Cretaceous throughoblique plate convergence, it probably has a strongstrike-slip component. Evidence of a backarc basin locatedbehind a partially emerged, less subsiding paleo-CentralCordillera (magmatic arc?) during the Late JurassicEarlyCretaceous includes the following: (1) Cretaceous plutonicbodies in the Central Cordillera; (2) Lower Cretaceoussandstones in the western part of the Cundinamarca subba-sin with abundant volcanic lithic fragments and feldsparderived from a western detrital source area, indicated bypaleocurrent data (Murca Formation and Utica sand-stone); (3) progressive westerly onlap terminations of theCretaceous carbonates on the basement, observed in seis-

merican Earth Sciences 21 (2006) 383411 407mic lines, in the western border of the Cesar Valley innorthern Colombia; (4) petrographical evidence suggesting

th Athat Berriasian (?)Valanginian clastic sediments near SanFelix in the western ank of the Central Cordillera camefrom the erosion of nearly uplifted areas containing frag-ments of metamorphic rocks and small tectonic blocks withplutonic rocks; (5) Cretaceous volaniclastic rocks thatprobably also derived from a magmatic arc; and (6) LateCretaceous zircon ssion-track ages in the Central Cordil-lera. However, a passive margin hypothesis (Pindell andErikson, 1993) or aborted rift arms related to the Caribbe-an opening cannot be completely ruled out because of theabsence of a well-dened Cretaceous magmatic arc.

Preliminarily, three stretching events are suggested dur-ing TriassicJurassic times. These events must be consid-ered preliminary because the poor continental Triassicand Jurassic fossiliferous sedimentary record oers rela-tively scarce, low-quality biostratigraphic data that makedening clear time boundaries between events dicult.The spatial distribution of the lithosphere stretching valuessuggests that small, narrow (180 km) wide, asymmetrical, transtensional half-riftbasin existed, divided by the Santander Floresta high. Asingle depocenter in the south was limited at its southernreach by a vertical transfer fault. Small mac intrusionscoincide with areas of thin crust (crustal stretching factors>1.4) and places of maximum stretching of the subcrustallithosphere. During the Aptianearly Albian, the basinextended south in the UMV. Dierent crustal and sub-crustal stretching values suggest either some lowermostcrustal decoupling between crustal and subcrustal litho-sphere or increased thermal thinning that aected the man-tle lithosphere. Late Cretaceous subsidence was mainlydriven by lithospheric cooling, water loading, and horizon-tal