4. Climatic forcing of asymmetric orogenic evolution

4. Climatic forcing of asymmetric orogenic evolution in the Eastern Cordillera of Colombia Andrés Mora†, Mauricio Parra, Manfred R. Strecker, Edward R. Sobel Institut für Geowissenschaften. Universität Potsdam, Karl Liebnecht-Str 24, D14476 Potsdam-Golm, Germany Henry Hooghiemstra, Vladimir Torres Institute for Biodiversitity and Ecosystem Dynamics (IBED), Paleoecology and Landscape Ecology, Faculty of Science, University of Amsterdam, Kruislaan 318, 1098 SN Amsterdam, The Netherlands. Jaime Vallejo Jaramillo Petrobras-Colombia Carrera 7, No 71-21. Torre B. Edificio Bancafé. Bogotá, Colombia. Submitted to Geological Society of America Bulletin, January 2007

4.1 ABSTRACT New apatite fission track data, paleo-elevation estimates from paleobotany, and recently acquired geological data from the Eastern Cordillera of Colombia document the onset of increased exhumation rates in the northeastern Andes at about 3 Ma. The Eastern Cordillera forms an efficient orographic barrier that intercepts moisture-laden winds sourced in the Amazon lowlands, leading to high rainfall and erosion gradients across the eastern flank of the range. In contrast, the drier leeward western flank is characterized by lower rates of deformation and exhumation. In light of the geological evolution of the Eastern Cordillera, the combination of these data sets suggests that the orographic barrier reached a critical elevation between ca. 6 and ca. 3 Ma, which ultimately led to protracted, yet more focused erosion along the eastern flank. Sequentially restored structural cross sections across the eastern flank of the Eastern Cordillera indicate that shortening rates also have increased during the past 3 Ma. From fission track and structural cross section balancing, we infer that accelerated exhumation led to increasing tectonic rates on the eastern flank, creating a pronounced topographic and structural asymmetry in the Eastern Cordillera. The tectonic and climatic evolution of this orogen thus makes it a prime example of the importance of climatic forcing on tectonic processes.

4.2 INTRODUCTION Important advances have been made in the understanding of climatic forcing of orogenic evolution over the last years (e.g., Koons, 1989; Molnar and England, 1990; Beaumont et al., 1992; Masek et al., 1994; Willet, 1999; Molnar, 2004). Probably one of the most important conclusions has been that active shortening and tectonic uplift may be localized in the course of orogenic evolution, if effective, and protracted erosion impacts tectonically active landscapes in areas of pronounced climatic gradients. Here, focused high precipitation and exhumation result from orographic barriers that intercept moisture and generate powerful erosional regimes on the windward flanks of an orogen (Horton, 1999; Reiners et al., 2003; Thiede et al., 2005; Barnes and Pelletier, 2006). For example, Beaumont et al. (1992) and Willet (1999), argued that the degree of asymmetry of an orogen may be highly dependent on climatic (e.g., precipitation) gradients. Their conclusions rely on numerical modeling of an idealized orogen with long-term precipitation focused on one side. Such predictions can also be reproduced in numerical and sandbox - 62 -

4. Climatic forcing of asymmetric orogenic evolution

models, showing that long-term erosion and sedimentation patterns are fundamental in determining structural geometries (Dahlen and Suppe, 1988; Mugnier et al., 1997; Koyi et al., 2000; Whipple and Meade, 2004; Hoth et al., 2006). If such conditions are associated with wedge-shaped thrust belts, focused precipitation and erosional removal of rock may ultimately lead to out-of-sequence thrusting (Mugnier et al., 1997; Horton, 1999; Koyi et al., 2000; Hilley and Strecker, 2004). In contrast, more arid, interior sectors of an orogen may experience cease tectonic activity due to the increase of lithostatic stresses resulting from failure to evacuate erosional materials (e.g., Sobel et. al., 2003; Hilley and Strecker, 2005). The Himalayas, the southern and central Andes, and the New Zealand Alps are premier examples of orogens, where interactions between tectonics and climate have been documented and where the interplay between tectonics and climate may have fundamentally influenced the evolution of individual mountain ranges and intraorogenic plateaus (Koons et al., 2002; Sobel and Strecker, 2003; Thiede et al., 2005). In the tectonically active northern Andes, where strong precipitation gradients also exist, however, such relations have not been explored, although this sector of the orogen receives one of the highest amounts of precipitation on Earth. This, coupled with the orographic effect of the NNE oriented mountain ranges, has created important precipitation gradients across the Western and Eastern cordilleras (Fig. 34). This sector of the Andes thus lends itself to a study of the interplay between tectonic activity, the spatiotemporal evolution of climatic gradients, and exhumation patterns.

Figure 34. (a) General features of the Colombian Eastern Cordillera. a. Regional structure with the main thrusts bounding the major topographic breaks in the Eastern Cordillera. River discharges (values in red in m3/sec) of the main rivers in the eastern and western flanks clearly showing higher values in the eastern side rivers flowing towards the Orinoco. Western rivers flowing towards the Magdalena do not manage to cut across the structures. SbB, Sabana de Bogotá Basin. The box in the smaller map shows the location of the bigger area. EC, Eastern Cordillera. (b) Precipitation map compiled from data of 424 meteorological stations for the past 20 years from IDEAM. The eastern slopes of the Eastern Cordillera clearly constitute a present day orographic barrier for winds, therefore concentrating precipitacions to the E.

- 63 -

4. Climatic forcing of asymmetric orogenic evolution

The eastern boundary faults of the Eastern Cordillera, here referred to as the Guaicaramo fault system (Fig. 34), constitutes a major plate boundary separating the structural domain of the Northern Andes microplate from the South American plate (Aggarwal, 1983). Between 4 and 5° N lat, the eastern flank of the Eastern Cordillera has accommodated significantly more shortening, and basement rocks crop out at elevations about 2 km higher compared to the western flank (Toro et al., 2004; Mora et al., 2006; Fig. 35). In addition to the structural asymmetry, the distribution of rainfall is also asymmetric and the more humid eastern flank strongly contrasts with the drier western parts (Fig. 34), but interactions between climate and tectonics remain poorly known. If significant feedback between tectonics and climate indeed existed in the Eastern Cordillera, it would likely be detectable through an erosional gradient. Therefore, we evaluated the role of erosional denudation in the context of the styles and timing of the structures that constitute this mountain range. In a first step, we collected field structural data to construct cross sections in order to assess the degree of overall shortening and removal of cover units. Secondly, we carried out apatite fission track thermochronology to quantify the long-term role and timing of erosional denudation. In a third step, we synthesized published data on the paleo-floristic evolution of the area, the late Cenozoic sedimentary record, and thermochronology data to reconstruct range wide late Cenozoic surface uplift, denudation and tectonism. In our attempt to unravel tectonics/climate interactions we compare different domains in the Eastern Cordillera with remarkably different PlioPleistocene denudation/uplift histories and explain their patterns in light of interactions between tectonically and climatically controlled migration of deformation. Our investigation reveals that the uplift of the Eastern Cordillera generated an effective orographic barrier for easterly moisture bearing winds, resulting in strong precipitation gradients across the range. Focused precipitation and removal of rocks conspire in sustaining high exhumation rates on the windward flanks, whereas the drier leeward side is characterized by low rates, emphasizing the pivotal role of climate in the evolution of mountain ranges.

Figure 35. (a) Digital elevation model of the Eastern Cordillera including the deeply dissected Eastern flank, the central flat lying Bogotá Savannah and the topographically lower western flank. (b) Topographic and generalized geologic map of the area studied in detail (see location in Figure 34). Apatite fission track sample locations are shown in red and vitrinite reflectance values are shown in black.

- 64 -

4. Climatic forcing of asymmetric orogenic evolution

4.3 REGIONAL GEOLOGY The Eastern Cordillera is the easternmost range in the northern Andes (Fig 34). It is an inversion orogen that coincides with a Lower Cretaceous rift (Colleta et al., 1990; Sarmiento-Rojas, 2001; Mora et al., 2006). Structurally, the Eastern Cordillera can be divided into three major areas perpendicular to strike, including the western and eastern marginal thrust belts with associated basement uplifts and the Bogotá Basin (Fig 34). Situated between two mountain ranges, the Bogotá Basin forms a low-relief central highland with outcrops of Cretaceous and Tertiary sedimentary units. Cenozoic shortening in the Bogotá Basin is minor and primarily related to folding and to a lesser degree to thrusting (Julivert, 1970). Nonetheless, the basin is topographically high, with elevations of more than 2500 m above sea level (Figs. 34 and 35). The western flank of the Eastern Cordillera, constitutes westvergent basement thrusts with offsets on the order of more than 20 km (Fig 34). The eastern flank is similar to the western flank as it forms an east-vergent thrustbelt where shortening is concentrated. In the east thrusting has uplifted basement rocks to higher elevations than on the western flank, resulting in basement exposure at elevations of ca. 4 km (Figs. 34 and 35). In contrast, the low relief and lack of major internal structures in the Bogotá Basin illustrate that this region was not uplifted by tectonic stacking within the basin, but rather passively uplifted by the principal thrusts on both flanks of the ranges (see cross section in Fig. 34). At the latitude of the study area, these are the Guaicaramo fault system on the eastern flank, and the Bituima, Cambao and Honda faults in the west (Fig. 34) A late Paleogene to Neogene foreland basin sequence to the east reflects the uplift history of the adjacent basement highs in the Eastern Cordillera. This sequence overlies late Cretaceous to early Paleogene transitional, to shallow marine facies and is characterized by pronounced facies changes in its lower part (Parra et al., 2005). The upper section of the sequence, ranging from Early Miocene to Late Miocene (Parra et al., 2006), is composed of continental strata whose depositional environment records the transition from marine to terrestrial conditions that were characterized by meandering rivers and proximal alluvial fan systems (Parra et al., 2006). 4.4 TOPOGRAPHY AND FLUVIAL NETWORK Between 3.5 and 6° N lat, the eastern flank of the range is remarkable for its topographic and climatic contrasts compared to the Sabana the Bogotá and the western branch of the Eastern Cordillera. For instance, the eastern flank has the highest mean elevations and a jagged topography with deep river canyons that have dissected the basement rocks (Figs. 34, 35 and 36). In addition, the eastern flank constitutes an effective orographic barrier that enhances precipitation on this side of the mountain belt and leaves leeward sectors dry (Fig. 34). For example, while the eastern flank receives about 5000 mm/yr of rainfall, the Sabana de Bogotá and the western flank receive about 1000 mm/yr and 2000 mm/yr respectively. Accordingly, the most important fluvial discharge in the Eastern Cordillera is concentrated on the east and feeds tributaries of the Orinoco River (Fig. 34). Rivers on the eastern flank are shorter than rivers in the Sabana de Bogotá and the western flank, channel slopes are steeper and are either perpendicular or moderately oblique to the structural grain (Figs. 34, 35 and 36). Along the western flank, the principal rivers have only succeeded to cross-cut major structures in a few locations and after having flowed a long distance parallel to sub-parallel to them (Fig. 34).

- 65 -

4. Climatic forcing of asymmetric orogenic evolution

Figure 36 (a) Swath topographic profile across the central segment of the Eastern Cordillera showing higher minimum, maximum and mean elevations in the eastern flank compared to the western flank. (b) Topographic profiles along main rivers of the western flank (Bogotá, Negro and Sumapaz Rivers) and eastern flank (Guatiquía, Guayuriba and Humea rivers) of the Eastern Cordillera. The Bogotá River can be identified in Figure 34a as the river having a discharge of 40 m3/sec; conversely the Negro River has been marked with a discharge of 125 m3/sec. The Sumapaz River is located south of the Bogotá Basin. At the eastern flank the Guayuriba in Figure 34a. has a discharge of 159 m3/sec, the Guatiquía 89 m3/sec and the Humea 129 m3/sec. Rivers along the eastern side of the chain are shorter and have higher slopes.

- 66 -

4. Climatic forcing of asymmetric orogenic evolution

4.5 METHODS 4.5.1 Structural and stratigraphic data We have compiled the most comprehensive surface stratigraphic and structural data base in the study area (e.g., Julivert, 1963; Guerrero and Sarmiento, 1996; Mora and Kammer, 1999; Mora and Parra, 2004; Parra et al., 2005; Parra et al., 2006), including all available oil industry seismic and well-data sets provided by the Agencia Nacional de Hidrocarburos (ANH) of Colombia in order to retrodeform and estimate amounts of shortening in two cross sections across the eastern flank of the Eastern Cordillera, as well as two regional cross sections across the entire orogen. Shortening estimates were obtained using conventional line-length balance techniques (e.g., Dahlstrom, 1969). Depths to detachment were estimated using seismic reflection profiles and geometrical approaches, such as area-balance methods (e.g., Mitra and Namson, 1989) and basement-uplift balance methods, similar to techniques applied in the Rocky Mountains (e.g.,, Erslev, 1986), see Fig. 37 and appendix for details).

Figure 37 (a) Cross section (location in Fig. 34) and digital elevation model of the Eastern Cordillera at the latitude of the Bogotá basin (Sabana de Bogotá). The inferred subsurface geometry implies that the Bogotá basin is a piggy back basin uplifted on top of the western marginal thrust faults like the Cambao and Bituima faults. (b) Retrodeformation of cross section in a, and deformed state cross section. The comparison of both allows calculating a total shortening of 58 km in the Eastern Cordillera. The area marked on the deformed state cross section is the excess area used (location in Fig. 34) in order to estimate depths to detachment (see appendix for details).

- 67 -

4. Climatic forcing of asymmetric orogenic evolution

4.5.2 Vitrinite reflectance analysis Macerals (like vitrinite) are major components of coal or organic matter in sedimentary rocks. Macerals increase their reflectance (i.e. reflectivity as measured under the microscope) with increasing temperatures (e.g., Bustin et al., 1990). Vitrinite reflectance (Ro) is the most widespread measurement of the maturity of organic matter as related to heating, commonly associated with burial. Thus, vitrinitereflectance analysis can provide the maximum paleo-temperatures reached by sedimentary rocks during burial. Normal vitrinite reflectance values in organic matter (Ro) typically range between 0.2 at maximum paleo-temperatures of < 30°C, to 6 for rocks that experienced maximum paleo-temperatures > 250°C. Intermediate values can be converted to paleo-temperature estimates using an adequate kinetic model (e.g., Barker and Pawlewicz, 1994). In our study, vitrinite measurements were made at 100 points in polished whole-rock blocks with a Leica MPV3 photomicroscope. Polarized reflected light was used to measure the mean random percentage reflectance in oil. Most of the samples were from coals. Our vitrinite data from the lowermost Cretaceous rocks (Berriasian) were compared with vitrinite data from equivalent units available in the national oil industry (Chevron, Ecopetrol, unpublished data). The two data sets are consistent. In a following step, the vitrinite data was used to estimate the maximum paleo-temperatures and to derive amounts of removed overburden. Samples from units, ranging from Barremian to Oligocene age, were also used to assess the regional paleo-thermal history and amounts of removed overburden. Geothermal gradients were also available from about 678 wells, mostly from the foreland areas. 4.5.3 Apatite fission tracks analysis Fission-track dating relies on crystal damage produced by the spontaneous fission of 238U in apatite or zircon (Wagner and Van den Haute, 1992). The number of spontaneous tracks is proportional to the age and the uranium content of the crystal (e.g., Green et al., 1989a). A typical total annealing temperature for apatite is about 120°; above this temperature on geological timescales, tracks are removed completely (annealed). This temperature is variable, depending principally on the F/Cl ratio of the apatites and the cooling rate (Gallagher et al., 1998; Carlson et al., 1999; Ketcham et al., 1999). Apatite fission tracks (AFT) remain stable only below 60°C (Gleadow et al., 1986). At temperatures in the partial annealing zone (PAZ), between 60°C and the total annealing temperature, tracks are partially or totally erased. If the apatites cooled sufficiently fast through the PAZ, the calculated age would coincide with the time when the apatite bearing rocks cooled through the closure temperature. However, if a sample spends a significant time in the PAZ, the age and confined track-length distribution will be modified. In regions undergoing contraction, erosional denudation following thrusting is one of the primary mechanisms of cooling (e.g., Sobel et al., 2006). In this study, we assess cooling driven by erosional denudation using AFT (e.g., Sobel and Strecker, 2003) in samples that cooled from temperatures higher than the total annealing temperatures, as determined by vitrinite reflectance data. Therefore, we only obtain information about the last cooling event (e.g., Green et al., 1989a; Green et al., 1989b). Samples were collected along two elevation profiles in order to decipher age/elevation trends and derive apparent exhumation rates (e.g., Sobel and Strecker, 2003; Sobel et al., 2006). Thermochronologic data alone do not provide rock uplift or paleo-elevation. We therefore compare surface uplift history

- 68 -

4. Climatic forcing of asymmetric orogenic evolution

deduced from published paleofloristic data with the denudation history obtained from apatite fission track analysis. In the following sections we present 17 new AFT ages from the eastern flank of the Eastern Cordillera and compare them to published AFT data from the western flank in order to examine the Late Cenozoic exhumation histories of the two regions. Data and methodologic details are presented in Tables 1 and 2. 4.6 STRUCTURAL STYLES AND TOTAL SHORTENING ESTIMATES OF THE EASTERN CORDILLERA A correct assessment of the structural styles of the Eastern Cordillera is required to support shortening estimates. In our cross sections we assume that the principal marginal thrusts in the Eastern Cordillera flatten at depth (Fig. 37) in order to transfer deformation throughout the middle crust from the west (e.g., the Central Cordillera), as has been observed in other inverted intra-plate basins (e.g., Roberts, 1989; Chapman, 1989; Sinclair, 1995; Deeks and Thomas, 1995) and other inversion orogens (Beauchamp et al., 1996). Based on this consideration, we deduced the depth to a mid-crustal detachment using a similar method as applied by Colleta et al. (1990) and Cortés et al., (2006) (see appendix and Fig. 37). In our cross sections, we modeled the Eastern Cordillera as a bivergent orogen with two main detachments. The assumption that the eastern detachment is the principal structure and the western, east-dipping detachment is a backthrust (Fig. 34 and 37), is supported by a major change in the character of the basement in the eastern Llanos foreland. The eastern Guaicaramo fault system (Fig. 34) separates a phyllitic basement domain of the Eastern Cordillera from the migmatite and igneous basement rocks of the Guyana shield. In a subsequent step, additional geometric considerations allow us to model the Sabana de Bogotá as an area passively transported on top of the gently dipping backlimb of the basement thrusts of the western marginal thrustbelt (Fig. 37). Accordingly, we modeled the boundary of the westward dipping panel east of the Sabana de Bogotá as an active hinge that coincides at depth with a bend in the main west dipping detachment (Fig. 37). Therefore, in our cross section A-A´ , surface uplift in the Sabana de Bogotá is not necessarily dependent on movement along the eastern Guaicaramo fault system (see appendix and Fig. 37). The validity of this assumption is underscored by observation in the Magdalena thrust system. Here, where the western marginal thrust system (detaching at ca. 20 km depth) is superseded by the Magdalena thrust system with a flat detachment at < 10 km toward the south (compare cross sections in Figs. 35b and 37), the Eastern Cordillera becomes an asymmetric block, tilted to the west by the eastern marginal thrusts and still detaching at more than 20 km. Inversion tectonics plays a fundamental role in the present-day structure of the eastern flank of the Eastern Cordillera (Mora et al., 2006). We summarize the structural style of this area from west to east. East of the Sabana de Bogotá, several basement highs are bounded by Lower Cretaceous normal faults, such as the San Juanito and Naranjal faults (Figs. 38, 39 and 40). To the east, the Lower Cretaceous Servitá normal fault was reactivated during the Andean orogeny as a major basement thrust, in contrast to the San Juanito and Naranjal faults, which have undergone only minor contractional reactivation. Therefore, the latter two faults are passively uplifted by the Servitá fault. The Farallones anticline (Fig. 40) is a typical inversion structure (Mora et al., 2006) in the hangingwall of the Servitá fault and east of the San Juanito fault. East of the Servitá fault lies the Guaicaramo thrust (Figs. 34 and 35), inferred to have formed in response to the Servitá fault hanging-wall block pushing from behind - 69 -

4. Climatic forcing of asymmetric orogenic evolution

(Mora et al., 2006). The Guaicaramo and Servitá faults are an integral part of the Guaicaramo fault system, which converges into the basal west-dipping main detachment (Fig. 37).

Figure 38. Interpreted migrated seismic line near the Anaconda-1 well. It shows clearly the low angle basement sheet of the frontal Mirador thrust. See location in Figure 34.

The Anaconda exploratory well and seismic lines close to it provide good evidence that the main basement thrusts on the eastern flank (e.g., Mirador shortcut) dip between 20 and 30° (e.g., Narr and Perez, 1993; Mora et al., 2006 and Fig. 38). Using these fault geometries, we obtained consistent shortening estimates of ca. 30 - 70 -

4. Climatic forcing of asymmetric orogenic evolution

km in three cross sections along the eastern flank of the Eastern Cordillera (Fig. 37, 39b and 40 see appendix for details). Similar low-angle geometries are supported by seismic lines in the frontal faults along the western marginal thrust belt (e.g., Cambao and Honda thrusts). Relatively gently dipping fault angles demonstrate that vertical uplift estimates can only represent a minor component of much larger amounts of slip along the fault planes. These and all other previously mentioned considerations allow us to calculate a total of ca. 58 km shortening for the northern cross section (A-A´) along the Eastern Cordillera (Fig. 37). It is remarkable that ca. 30 km of shortening along the eastern flank contrasts with ca. 23 km along the western flank, whereas shortening in the Bogotá basin is only about 5 km (Fig. 37).

Figure 39. (a). Cross section showing the projected location of AFT samples along the Guatiquía River profile. (b) Cross section along the Guayuriba river profile showing the projected samples and an age elevation plot. Samples 5 to 8 constitute the vertical profile sampled whose ages were used in the age elevation plot above. Calculated mean exhumation rates in this profile are 1.5 ± 0.5 mm/yr.

- 71 -

4. Climatic forcing of asymmetric orogenic evolution

4.7 NEW AFT DATA FROM THE EASTERN FLANK OF THE EASTERN CORDILLERA The Guayuriba and Guatiquía river profiles (Figs. 35, 39 and 40) were collected in deeply incised valleys where the shallowly dipping lower Cretaceous sedimentary sequence unconformably overlies the basement near the drainage divide. Therefore, samples from lower elevations in the river valleys should correspond to deeper crustal levels. Samples were collected in lower Cretaceous units. Additional material was collected from basement rocks below the basal Cretaceous unconformity and at lower elevations. The set of samples analyzed represents the youngest ages reported to date in the Eastern Cordillera.

Figure 40. Regional cross section across the eastern flank of the Eastern Cordillera with projected average values of vitrinite reflectance measurements and projected additional AFT ages along the Guatiquía River profile.

Along the Guatiquía river profile (Figs 39, 40 and Table 1) we analyzed 6 samples collected at elevations ranging from ~1650 to ~3650 m. The samples have no clear age/elevation trend. However, ages are younger in the hanging-wall block of the San Juanito fault compared to the footwall. East of this structure, all ages are younger than 1.6 Ma (maximum 1.2 ± 0.4 and minimum of 0.8 ± 0.2 Ma) and in the footwall to the west, all of them are older than 1.5 Ma (maximum age 2.8 ± 0.4 and minimum of 2.3 ± 0.8 Ma). Dpar measurements show similar kinetic characteristics for all of the samples along this profile; these samples are close to the Fish Canyon apatite composition (Ketcham et al., 1999). Therefore, we assume that age differences are not due to compositional differences (Table 2). In the Guayuriba River profile (Fig. 39), located south of the Guatiquía River profile and along strike, four basement samples were collected between ~1200 m and ~2950 m (samples 5 to 8, Figs 35 and 39). Ages increase with elevation from 1.6 ± 0.4 Ma at 1200 m to 2.6 ± 0.5 Ma at 2954 m (Fig. 39). The age/elevation plot yields an apparent denudation rate of 2 ± 1 mm/year. In addition to these samples, we collected material from the lowermost Cretaceous units (samples 3 and 10) or - 72 -

4. Climatic forcing of asymmetric orogenic evolution

- 73 -

4. Climatic forcing of asymmetric orogenic evolution

basement units located very close to the basal Cretaceous unconformity in different structural positions (samples 1, 2, 4 and 9). The ages from those samples have similar values between 2.6 ± 0.5 and 3.0 ± 0.4 Ma. These cooling ages are independent of the tectonic block sampled. Furthermore, ages from samples of nearly the same stratigraphic horizon are very similar in the different blocks of the major structures. Basement samples from lower elevations in the Guayuriba River valley (e.g., samples 5 and 6, Figs. 35 and 39) have younger cooling ages than samples collected in Cretaceous or basement rocks at higher elevations in this transect. The youngest sample stratigraphically comes from a horizon about 1 km above the base of the Cretaceous strata and has the oldest cooling age (3.8 ± 0.7). It appears that cooling-age trends are conditioned by elevation and stratigraphic position, suggesting that younger rock units were exhumed prior to older units, as expected in the simplest structural setting. Nonetheless, structures do not appear to have conditioned cooling age patterns in this profile. However, Dpar data from samples east of the Servitá fault have larger values, closer to the Fish Canyon apatite composition, whereas Dpar values to the west are closer to the Durango composition (Table 2). Therefore, the eastern samples likely have a closure temperature about 20°C higher than the western samples (Ketcham et al., 1999).

4.8 INTERPRETATION OF THE NEW AFT DATA In order to constrain the amount of overburden removed during cooling, estimates of apatite-closure temperature and past geothermal gradients are critical variables. Ketcham et al., (1999) estimated closure temperatures for Durango and Fish Canyon apatites. Our samples in the Guayuriba River profile west of the Servitá fault (Figs. 35 and 39b) have Dpar values close to that for Durango apatite (Table 2); therefore, we assume closure temperatures between 107 and 126°C for cooling rates between 10 to 100°C/Ma, respectively. In contrast, Dpar values close to those of Fish Canyon apatite in the Guatiquía River transect and in the Guayuriba River profile east of the Servitá Fault suggest closure temperatures between 128 to 148°C for cooling rates between 10 to 100°C/Ma. - 74 -

4. Climatic forcing of asymmetric orogenic evolution

Based on maximum paleo-temperatures estimated from vitrinite reflectance data, compared with the thickness of the sedimentary units (Figs. 41 and 42; Table 3), we calculate a Campanian to Oligocene geothermal gradient of approximately 20°C/km. This value is similar to the present average geothermal gradient obtained from wells in the eastern foreland regions (Bachu et al., 1995). However, Mancktelow and Grasseman (1997) demonstrated that in an active tectonic regime with high exhumation rates, normal geothermal gradients may increase by 60% due to advection. Therefore, we use an approximate value of 30±2°C/km for the geothermal gradient of areas undergoing denudation at high rates along the eastern flank of the orogen. A compilation of unpublished data from Ecopetrol and new vitrinite reflectance data shows Ro (vitrinite reflectance, see our new data in Table 3) values greater than 4% for the Lower Cretaceous rocks that hosted our AFT samples, implying maximum paleotemperatures above 250 °C (Barker and Pawlewicz, 1994) (Table 3 and Figs. 35, 39 and 40). Clearly, AFT data from the same units represents cooling of completely reset samples, and therefore only recording the last episode of the cooling event. Given the assumptions concerning closure temperatures and geothermal gradient, our data suggests about 3 to 5 km of overburden exhumation during the last 3 Ma (Figs. 43 and 44), implying cooling rates of about 1.0 ± 0.5 mm/yr. This rate is slightly lower than that obtained from the elevation profile (1.5 ± 0.5 mm/ year; Fig. 39). Mancktelow and Grasemann (1997) argue that apparent cooling rates on the order of 2 ± 1 mm/year derived from elevation profiles can be overestimated and even doubled in settings with high relief expressed over 10 to 20 km wavelengths due to the influence of non-planar isotherms. Therefore, we rely more on the rate of 1 ± 0.5 mm/yr derived from estimates of overburden removal.

The total thickness of the Meso-Cenozoic sedimentary sequence in areas adjacent to our sampling locations (Fig. 41) is between 7.5 and 10 km (Julivert, 1963; Mora and Kammer, 1999; Mora et., 2006). Our vitrinite-reflectance data suggest that a comparable thickness of overburden once covered the lower Cretaceous units in the AFT sample locations prior to the onset of denudation (Table 3 and Figs. 41 and 42). Sedimentological and basin analysis show that denudation of these ranges has been active since middle Oligocene time (Parra et al., 2007; Parra et al., 2006). In addition, Parra et al. (2007) calculated a total of 3 km of unroofing between that time and the late Miocene. Our thermochronological data documents that 3 to 5 km of rock - 75 -

4. Climatic forcing of asymmetric orogenic evolution

Figure 41. Measured stratigraphic thicknesses in two different structural domains across the study area. To the left the sequence west of the Servitá fault and in the Bogotá basin shows that the youngest widespread (pre-deformation) formation is of late Eocene age, whereas Oligocene and Miocene rocks are absent or have a reduced thickness. To the right, the sequence east of the Servitá fault with a condensed Paleogene thickness but an increased Neogene thickness.

was exhumed from the studied section over the last 3 Ma. Comparing these two datasets reveals that exhumation rates must have accelerated at some point during the Pliocene (Fig. 43). The exhumation history documented in our study involves rocks eroded from above the Servitá fault and the frontal Mirador shortcut (Figs. 38, 39, 40 and 44). The 4-5 km of eroded material on top of the Mirador shortcut during the last 3 Ma roughly equals its total vertical throw, calculated by comparing the position of the Lower Cretaceous hanging-wall/footwall cutoffs (Figs. 38, 39 and 44). Assuming that the Pleistocene surface elevation at the position of the most distal thrust sheets, such as the Mirador shortcut, was roughly the same as the present-day elevation of approximately 1000 m, the Mirador shortcut was probably initiated during the last 4 Ma (Figs. 39 and 44). Data from seismic reflection profiles and the Anaconda-1 exploratory well reveal that the Mirador shortcut dips about 20 to 30°W (Fig. 38). - 76 -

4. Climatic forcing of asymmetric orogenic evolution

Using this information, we estimate that about 13 to 15 km of shortening has occurred during the last 3 Ma. This is ca. 50% of the total shortening across the eastern foothills. These values are based in incrementally balanced retrodeformations of the cross sections in figures 39 and 40 (see incremental restorations in Fig. 44). Compared with total shortening estimates in the Eastern Cordillera (Fig. 37) the calculated Plio-Pleistocene shortening along the eastern flank represents ca. 25% of the total accumulated Cenozoic shortening of the entire orogen. The similar AFT ages from the hanging-wall of the Mirador shortcut and the hanging-wall block of the Servitá fault demonstrate that both structures formed in a forward breaking or synchronous sequence of thrusts, with the Servitá hanging-wall block pushing from behind (e.g., Mora et al., 2006). The ongoing tectonic activity of these structures is underscored by numerous fault scarps and tilted river terraces (Robertson, 1989).

Figure 42. Inferred maximum paleo-temperatures from vitrinite samples in table 3 (using Barker and Pawlewicz, 1994 kinetic model) versus theoretical amounts of overburden on top of them, assuming that all the measured amounts of burial covering the different units (Fig. 41 and Table 3) in areas were those rocks are preserved equals the eroded thickness in areas were they have been eroded and where we performed vitrinite analysis. The thick lines in the plot show inferred Campanian to Oligocene geothermal gradients (using samples from units of those ages). The same procedure was followed to calculate Berriasian to Campanian geothermal gradients. The two different trends roughly coincide first with the Berriasian to Aptian syn-rift phase, thus with high geothermal gradients (Sarmiento-Rojas, 2001) and second with the post-Alban post-rift phase (Sarmiento-Rojas, 2001).

In contrast, in the Guatiquía River vertical profile the only marked jump in AFT ages coincides with a fault (see Figs 39 and 40 samples 11 to 16). Our Dpar data shows that apatite composition is very similar on both sides of the structure. Therefore, the western footwall block of the San Juanito fault apparently commenced exhumation about 1 Ma prior to the hanging-wall.

- 77 -

4. Climatic forcing of asymmetric orogenic evolution

Figure 43. Compilation of inferred amount of eroded material versus time in the study area as deduced from our data and data from Parra et al. (2007).

Interestingly, east of the San Juanito fault the Lower Cretaceous unconformity in the hinge zone of the Farallones anticline lies at the same elevation as west of the San Juanito fault (Figs. 35 and 40). Younger ages east of this fault may indicate that the Lower Cretaceous unconformity in the Farallones anticline may have reached similar elevations later compared to the equivalent horizon in the footwall of the San Juanito fault. Thus, at least part of the fold amplification process in the Farallones anticline may be subsequent to movement along the San Juanito fault. Drainage patterns in this area may support this assessment. For example, the Guatiquía River flows east from the footwall block of the east-dipping San Juanito normal fault, cutting through basement lithologies (Fig. 35). Farther east, the river reaches the hangingwall block and is deflected almost 90° to the south near the trace of the San Juanito fault, cutting weak black shales. If an imaginary straight line parallel to the trace of the Guatiquía River in the footwall block of the San Juanito fault were extended ESE, the headwaters of the Humea River would be reached farther east (Fig. 35). We hypothesize that the Guatiquía River was formerly linked to the Humea flowing directly to the east. In such a scenario incision through basement lithologies during growth of the Farallones anticline may have caused the Humea to be beheaded and the Guatiquia River to be deflected or even captured by the present-day lower Guatiquía River (Fig. 35). After such a capture, the deflected Guatiquía River may have begun to erode the less resistant lithologies of the synrift shales adjacent to the San Juanito fault in the downdip sectors of the tilted hanging wall. This hypothetic geomorphic scenario requires earlier exhumation of the San Juanito footwall block and later folding of the Farallones Anticline, an assessment which is supported by our AFT data. - 78 -

4. Climatic forcing of asymmetric orogenic evolution

Figure 44 (Continue)

- 79 -

4. Climatic forcing of asymmetric orogenic evolution

Figure 44. (a). Incremental retrodeformation of cross section in Figure 37 using the constraints from paleoaltimetry (from palynoflora), exhumation (AFT) and total burial from Ro data. On the other hand many published data (Julivert, 1963; Helmens, 1990; Gomez et al., 2005; Kammer, 2003) constitute pieces of evidence showing that most of the folding in the western part of the section (Bogota area) is older than Pliocene at least. Shortening at each stage is calculated with line length balancing. The methods to deduce fault geometry at depth, where there is no seismic control, are explained in the appendix. (b). Incremental retrodeformation of cross section A, in Figure 36, using similar constraints as in the previous one. The frontal shortcut geometry is defined with seismic profiles and the Anaconda-1 well.

A remarkable feature of the AFT data is the similarity of young ages of two samples east of the San Juanito fault. The first sample is from a lower Cretaceous sandstone with an AFT age of 0.8 ± 0.2 (sample 16). The second is from a preSilurian granite located to the east (Fig 40), with an AFT age of 0.9 ± 0.2 Ma (sample 17). The granite underlies the Paleozoic sequence (Segovia, 1965). Our cross sections show that today the granite sample lies stratigraphically 3 km below the lower Cretaceous sample (Figs. 40 and 44). Similar ages in these samples that underwent very different amounts of exhumation can be explained if the frontal Servita fault is interpreted as a listric fault that first uplifted and exhumed the frontal, easternmost parts of the hanging-wall block rather than the backlimb (Figs. 40 and 44a). In addition, the reconstructed amount of exhumed material in the cross sections (Fig 44a), calculated from vitrinite-reflectance data and measured thicknesses, depicts progressively increasing amounts of total exhumation from the backlimb to the frontal areas of the eastern flank. In support of our interpretation, structural - 80 -

4. Climatic forcing of asymmetric orogenic evolution

models show that pronounced and progressively increasing differential exhumation between the backlimb of a structure and the deformation front (Fig. 44a) is expected and can be considered to be typical of reactivated listric faults (e.g., Mitra, 1993; McClay, 1995). A listric geometry for the eastern flank frontal thrusts has been previously proposed based on cross-section balancing (Restrepo-Pace et al., 2004; Cortés et al., 2005; Mora et al., 2006; Cortés et al., 2006). Samples with ages younger than 1 Ma imply removal of 3 to 4 km of rock in approximately 1 Myr. One possibility explaining this phenomenon is that such an increase in denudation may have been caused by accelerated incision. However, some additional effect due to isotherm advection cannot be discarded (e.g., (Mancktelow and Grasemann, 1997). 4.9 APATITE FISSION TRACK DATA FROM THE WESTERN FLANK Gomez (2001) and Gomez et al. (2003) analyzed 15 apatite samples from different Meso-Cenozoic stratigraphic units from the Guaduas Syncline and the Villeta Anticlinorium at 4.5°N (Fig. 34) to unravel the timing of shortening, exhumation, and related syn-orogenic sedimentation on the western flank of the Eastern Cordillera. These authors used apatite fission track dating, track-length distributions, and vitrinite-reflectance analysis to assess the thermal events associated with tectonism. Their conclusions are mostly based on modeling rather than individual ages. The first event they identify is cooling from maximum temperatures of around 130 to 100°C between 65 and 30 Ma, associated with erosion of more than 3 to 4 km of sedimentary cover. A second event, bracketed between 15 and 5 Ma, indicates cooling from temperatures of about 120 to 90°C and erosion of ca. 3 km of rock in the hanging wall of the main thrusts of the western flank. Only four of the samples (Gómez, 2001) have totally reset ages. None of them provided ages younger than 5 Ma. It is worth emphasizing again that our samples from the eastern flank have younger cooling ages and represent a larger magnitude of late Cenozoic exhumation than samples from the western flank. Most of our samples are from structurally deeper levels and were obtained either from basal Cretaceous units or from basement rocks. The only comparable samples from the western flank are the two lower Cretaceous samples of Gomez (2001). In these samples, apatites were totally reset, but the ages are substantially older than the average age of our samples. Combining both data sets reveals a pronounced asymmetry in the degree of exhumation, indicating a significantly younger onset and higher rate of exhumation along the eastern, compared to the western flank of the orogen (Figs. 34 and 35, Table 1). Gomez et al. (2003) also present data for a shortening event along the western flank that occurred between late Miocene and prior to 5 Ma. They argue that this was associated with exhumation documented in the AFT data. Gomez et al. (2003) date the deposition of the Honda Group, in the hanging wall of the more frontal Honda thrust (Fig. 34), between 12.7 and 6 Ma. This unit is tilted about 30° to the west. Overlying undeformed volcanic-ash bearing terraces of late Pliocene age provide a minimum age for motion along this fault (Gómez et al., 2003). Furthermore, based on detailed mapping and detailed stratigraphic work in this region, RestrepoPace et al. (2004) suggest that ca. 80% of the total shortening and uplift along the western flank of the Eastern Cordillera must have occurred before the late Miocene. 4.10 PALEOALTIMETRY AND PALEOFLORA

- 81 -

4. Climatic forcing of asymmetric orogenic evolution

Paleoaltimetry data in the Andes is sparse and often subject to significant errors and uncertainties (e.g., Gregory-Wodzicki, 2000; Blisniuk and Stern, 2005; Garzione et al., 2006). Paleoflora data have been used in the assessment of paleo-topography in Bolivia (e.g., Graham et al., 2001) and Colombia. In the Northern Andes, Van der Hammen et al. (1973), Wijninga (1996) and Hooghiemstra et al. (2006) analyzed the fossil floras of the late Miocene to late Pliocene fluviatile Tilatá Formation in outcrops exposed in the Bogotá Basin above 2400 m. Using the nearest living relatives method, they correlated fossil pollen spectra from a 6 to 3-m.y.-old sedimentary section (Andriessen et al., 1993) with similar pollen spectra from modern vegetation and found a clear trend of environmental evolution (Fig. 45). The oldest pollen assemblages (late Miocene, Salto de Tequendama sections) including Alchornea, Bombacaceae, Humiriaceae, Ilex Iriartea, Mauritia Malpighiaceae, Melastomataceae, and Protium, indicate a tropical lowland environment. In contrast, the youngest pollen assemblages from the ca. 3 Ma Guasca section mainly comprise taxa such as Eugenia, Hedyosmum, Myrica Weinmannia Oreopanax, characteristic of high-elevation (>2000 m) Andean forest.

Figure 45. Inferred paleoelevation of stratigraphic profiles in the Bogotá Basin (after Wijninga, 1996).

Wijninga (1996) argued that the present topography of the Eastern Cordillera does not prevent pollen from the tropical lowland forest in the Magdalena Valley from reaching the high plain of the Sabana de Bogotá. Probably a similar situation has prevailed since at least late Miocene time as pollen of all adjacent vegetation zones are represented in younger sediments of the Bogotá Basin (Wijninga, 1996). However, the proportion of Andean pollen in the Pliocene record of the Bogotá Basin - 82 -

4. Climatic forcing of asymmetric orogenic evolution

steadily increases in the younger sections, until reaching amounts similar to those in the present-day high elevation Andean forests. Van der Hammen et al. (1973), Wijninga (1996) and Hooghiemstra et al. (2006) proposed that the observed change in pollen spectra implies that a major temperature change must have affected the Bogotá Basin. In their interpretation this change cannot be solely explained by the effects of late Cenozoic global cooling (Hooghiemstra et al., 2006). Rather, the pollen data is interpreted to reflect a major change in paleo-elevation between 6 and 3 Ma, from less than 1000 m to the present-day elevation of about 2600 m (Fig. 45; Wijninga, 1996). After the main phase of topographic growth, simultaneous with folding in the Bogotá Basin, the deposition of approximately 600 m of undeformed fluvio-lacustrine sediments (Helmens and Van der Hammen, 1994; Hooghiemstra and Cleef, 1995; Hooghiemstra et al., 2006) from ca 3.2 Ma illustrates that the Bogota Basin evolved into a subsiding, isolated intermontane basin. The most recent study of the sedimentary and palynological record of these sediments was published by Torres et al. (2005) based on the Funza-2 borehole, located in the depocenter of the Bogotá Basin. The chronology was initially established through tephrochronology (Andriessen et al., 1993), but more recently astronomical tuning to the pollen record was applied (Torres, 2006). The Funza-2 site provides one of the best terrestrial records of environmental change for the last 3.2 Ma. This data set documents climate change during the last 3 Ma, but no further surface uplift in the Bogotá Basin or adjacent areas can be deduced (Hooghiemstra, 1984; Hooghiemstra and Cleef, 1995; Van´t Veer and Hooghiemstra, 2000; Torres, 2006; Hooghiemstra et al., 2006). We therefore conclude that the main phase of topographic growth in the Eastern Cordillera must have occurred between 6 and 3 Ma. 4.11 DISCUSSION Our new thermochronological data reveal a complex interaction between tectonics, climate and exhumation in the Eastern Cordillera of Colombia. We document an acceleration of exhumation rates on the eastern flanks of this range by about 3 Ma, compared to the previous 30 Ma. However, compared with similar data sets from the western flank of the orogen it appears that focused and accelerated denudation solely characterize the eastern deformation front. The spatial pattern of exhumation has thus been asymmetric in the Eastern Cordillera from late Pliocene to present. Asymmetry is also visible at different levels in the geological evolution of both areas. We document that greater total amounts of shortening have occurred on the eastern versus the western flank. In addition, basement exposures occur in the east but are absent in the west (Fig. 34). Accelerated shortening rates of about 5mm/year (see appendix and Figs. 37 and 44 for shortening calculations) are restricted to the east during the Pliocene. Finally, topography, precipitation patterns, and the characteristics of the fluvial network follow this disparate spatiotemporal evolution. An important question is whether climate and exhumation have been active or passive with respect to the spatial and temporal trends in tectonic evolution. In the following we propose a scenario involving an interaction between climate and structures in order to explain the geological evolution. Areas with high mean elevation typically coincide with thickened crust; therefore the timing of crustal thickening presumably coincides with the timing of growth of topography and the establishment of high relief conditions (e.g., Isacks, 1988). As in many other areas of the Andes (e.g., Allmendiger et al., 1997), crustal thickening in the Eastern Cordillera was mainly accomplished through shortening, as - 83 -

4. Climatic forcing of asymmetric orogenic evolution

the contribution of magmatic addition can be considered to be minor and uplift related to lithospheric delamination appear unlikely based on the overall tectono-magmatic evolution of this part of the Andean orogen (e.g., Sarmiento-Rojas, 2001). The specific structural style of the Eastern Cordillera (Figs 34, 34 and 37) shows that crustal shortening is mostly concentrated on both margins of the mountain belt. Activity along the marginal thrusts raised the central plain of Bogotá and the timing of topographic growth in the central plain can be related to active shortening along both marginal thrust systems, which set the stage for the subsequent accelerated climate driven processes Any hypothesis about the influence of climate on the temporal patterns of shortening and topographic growth however, has to take into account the paleoclimatic history prior to and during uplift. Molnar and Cane (2002) suggested that the global climate system during Mio-Pliocene time was in a permanent paleo-El Niño state. According to their predictions the northwestern corner of South America would have been much drier than today during such conditions. If this were true for the Neogene of northern South America, present-day precipitation patterns have to be used with extreme caution in characterizing interactions between orography and precipitation during the Plio-Pleistocene. However, palynological, macroscopic paleoflora evidence, and stable isotope data from fossil and present-day growth bands in mollusks suggest that foreland areas east of the Eastern Cordillera were humid and characterized by similar precipitation patterns, at least since Middle Miocene time (e.g., Lorente, 1986; Hoorn, 1994; Van der Hammen and Hooghiemstra, 2000; Kaandorp et al., 2005; Hoorn, 2006). Therefore, we posit that the impact of global climate change on erosion processes in the Pliocene (e.g., Molnar, 2004) was neglible in this region. In keeping with the regional paleo-climate data we suggest that the Eastern Cordillera at the latitude of Bogotá was at low elevation and was characterized by a tropical lowland climate during late Miocene time, as proposed by (Wijninga, 1996) and (Hooghiemstra et al., 2006). Topography increased between 6 and 3 Ma in the Eastern Cordillera (Wijninga, 1996) as a consequence of movement along the main boundary thrusts in the eastern and western foothills. Presumably when topography reached a critical elevation, an orographic barrier was created that forced easterly moisture-bearing winds to focus precipitation along the eastern side of the orogen, as happens in the present. The regional precipitation map (Fig. 34) shows that fault-bounded ranges 0.7 to 1 km higher than the adjacent undeformed foreland, like the Serranía de las Palomas, apparently have no pronounced effect on amount and distribution of rainfall in this environment. In other tectonically active regions, such as the Sierras Pampeanas or the Patagonian Andes (Sobel and Strecker, 2003; Blisniuk et al., 2005), the Eastern Cordillera of Bolivia (Masek et al., 1994) or on the Indian Subcontinent (Bookhagen et al., 2005) higher threshold elevations between 2000 and 2500m are needed to generate pronounced precipitation gradients. We thus infer that such conditions were probably attained toward the end of the phase of relief growth between 6 and 3 Ma. If true, focused precipitation on the eastern flank would have generated an effective, eastward flowing fluvial system with higher discharge that cut deeper canyons compared to the modern rivers of the western flank. Accordingly, we suggest that some of the differences in AFT ages associated with the different fault blocks can also be linked to differential river incision, such as in the Guatiquía River profile. Thus, some of the documented cooling occurred through river incision. If such relationships indeed had been in place during the last 3 Ma, then climate-driven focused denudation and accelerated rates would be the expected result. Accelerated - 84 -

4. Climatic forcing of asymmetric orogenic evolution

denudation rates during the last 3 Ma along the eastern side of the Eastern Cordillera would have caused unloading of the fault-bounded ranges, prompting faster movement along the main thrusts, as proposed by (Hilley et al., 2005) for orogens strongly conditioned by inherited structures like the Eastern Cordillera. Support for the viability of this model comes from the comparison of timing and amount of shortening along the western (Gomez et al., 2003; Restrepo-Pace et al., 2004; Montes et al. 2005; Cortés et al., 2005) and eastern flanks (Fig. 44). Crustal shortening has been concentrated on the eastern flank during the last 10 million years, with peak values of about 5 mm/yr during the last 3 Ma (Figs. 34, 35, 37 and 44). Therefore, peak shortening rates would have occurred subsequent to uplift of the Bogotá Basin, as supported by the pollen data (Wijninga, 1996). Thus, enhanced mass removal may have favored, and probably accelerated motion along the shortcut faults on the eastern flank since mid-Pliocene time. Such accelerated shortening, involving ca. 25% of a total 60 km orogenic shortening and 50% of the total shortening along the eastern flank, is expected to generate a significant amount of lithospheric flexure in adjacent areas that are not being uplifted. The exact thickness of corresponding units in the adjacent Llanos foreland basin to the east has never been established, due to poor chronological constraints. However, the Funza 2 drill hole from the Bogotá Basin records about 0.6 km of lacustrine sediments deposited in an area undergoing no deformation during that time (Helmens and Van der Hammen, 1994; Torres, 2006). We speculate that the onset of continuous deposition and subsidence in the Bogotá Basin at 3.2 Ma (Helmens and Van der Hammen, 1994; Torres, 2006) reflects enhanced tectonic thickening in adjacent, actively deforming areas, such as the eastern foothills. The lacustrine conditions in the Bogota Basin prevailed until about 17.000 BP (Hooghiemstra, 1984). We infer that this was the result of internal drainage, aided by the drier conditions in the highlands, which prevented fluvial systems from efficient downcutting and sediment evacuation. We suggest that reduced rainfall caused inefficient drainages in the leeward highlands to the west, while active tectonism associated with high rainfall along the eastern flank caused deep incision of high-gradient rivers. Probably because of the ongoing tectonic activity along the east, these rivers have not been able yet to reach the Bogota Basin by headward erosion. Intense, localized precipitation may thus have resulted in higher rock uplift rates and a profound influence on the drainage system compared to areas to the west. 4.12 CONCLUSION Based on the analysis of new structural and thermochronologic data combined with published paleoflora and thermochronologic information we document contrasting deformation, exhumation and surface uplift histories in different areas of the Eastern Cordillera of Colombia since the the late Pliocene. The fundamental trigger causing asymmetry in the orogenic processes is the initial topographic growth between 6 and 3 Ma that built an orographic barrier that intercepted easterly moisture-bearing winds leading to focused erosion. Our data suggest that this focused erosion increased mass removal along the eastern flank of the orogen to values comparable to tectonically controlled advection of material, therefore enhancing tectonic mass flux. In contrast, due to the eastern orographic barrier precipitation was reduced in the western sectors of the orogen during the same period, and the basin became internally drained. Consequently, mass removal and tectonic flux were reduced. We propose that such disparate behaviour ultimately - 85 -

4. Climatic forcing of asymmetric orogenic evolution

enhanced or eventually produced the regional tectonic asymmetry. Climate-driven focused denudation in the Eastern Cordillera is therefore a phenomenon that has profoundly modified the rates and the location of tectonic deformation. Acknowledgments: The authors are indebted to Peter Molnar, Richard W. Allmendinger, Brian Horton, John Suppe and Jose María Jaramillo for fruitful suggestions, and Birgit Fabian for graphic work. Gerold Zeilinger helped generating the DEM and the swath profiles. Seismic and well data were kindly provided by Javier Cardona and Mauricio Blanco (Agencia Nacional de Hidrocarburos) and German Rodriguez (Sipetrol). Sandra Passos (Ecopetrol) allowed us access to confidential vitrinite reflectance data from Ecopetrol. Dr. Carlos Costa Posada from IDEAM Colombia provided data of precipitation and river discharges for the last 20 years from the Eastern Cordillera. Vitrinite reflectance analysis was carried out at the Departamento de Geociencias (Universidad Nacional) and Ingeominas, with the cooperation of Gladys Valderrama, Luis Jorge Mejía, Andreas Kammer and Luis Ignacio Quiroz. Oscar Fernandez from Midland Valley provided a 2d Move license. Elias Gomez kindly provided us some of his original data from the western flank. Financial support for field work and analysis came from Petrobras Colombia and the German Science Foundation to M. Strecker (Str 373/19-1). A. Mora and M. Parra thank the DAAD for funding their studies at Potsdam University. M. R. Strecker also thanks the A. Cox fund of Stanford University for additional support.

APPENDIX Fault geometries and depth to detachment calculations A correct calculation of shortening depends on a well constrained fault geometry. Our interpretation of a listric fault shape is favored with the exhumation pattern documented in this work (Fig. 41), reminiscent of that simulated in inverted listric normal faults (McClay, 1995; Mitra, 1993). Otherwise the forward breaking sequence of deformation from W to E along the eastern deformation front, documented with AFT in the Guayuriba profile, does not favor the geometrical assumption of steeply dipping fault planes at depth for the basement faults while the frontal thrusts are low angle. Second it has been observed by means of deep seismic reflexion profiles, reaching more than 7 sec TWT (e.g., Deek and Thomas, 1995; Sinclair, 1995) that many inverted master faults in intra-plate inverted grabens have a listric shape. We further assume inversion reactivating mostly the same extensional fault planes, following observations by (Mora et al., 2006). With these assumptions an approximate depth to detachment can be found based on the fact that, assuming plain strain deformation, the area of cross sectional structural relief equals total shortening times depth to detachment (e.g., Mitra and Namson, 1991; Hossack, 1979, Fig 37). This relationship has been used already in the Eastern Cordillera by Colleta et al. (1991) and Cortés et al. (2006). With a total shortening of 58 Km calculated from line length balancing in our cross section (Fig. 37) and an excess area of 1340 Km² a depth to detachment of about 23 km in the undeformed state was obtained assuming the west dipping fault as a main detachment. Through an iterative process using 2d Move©, we tried to extrapolate the dip data from the area adjacent to the Anaconda well to the west and at depths bigger than 5 km. We searched for a fault plane that cutting sequence up fits better to the observed spatial and temporal evolution of rock uplift, exhumation and surface uplift during the late Cenozoic and the predicted depth to detachment as well. We tried different fault related deformation mechanisms but found that a rigid block rotation along a circular fault, like proposed by (Erslev, 1986) in the Rocky Mountains, reproduced better the observations at different moments in time (Fig. 37). This geometry was used also by Jordan and Allmendinger (1986) in the Sierras Pampeanas and Kley and Monaldi (2002) in the Santa Barbara System. As proposed by Erslev (1986) the backlimb dip angle of a given basement uplift (Δ tilt) can be - 86 -

4. Climatic forcing of asymmetric orogenic evolution

related to fault curvature (1/R where R is the radius of curvature) and slip along the circular fault segment by the following relationship (Fig. 46). Δ tilt = 180 SRA/πR

Figure 46. Backlimb dip (7°), rotation angle (8°) and total slip along the Servitá fault, taken to constrain fault geometry in section E-E´.

Using as an example cross section E-E` ´(Fig 40), we only know the backlimb dip angle (7°) from drawing a tangent line to the base of the best constrained horizon, that is the base of the Aptian Une Formation in section E (Fig 40 and 46). Then we assumed a radius of curvature of a circle that fits approximately the surface trace of the fault and the depth to detachment, where the rotation axis is located above the place where the backlimb is horizontal, in our case the axis of the Bogotá Basin (Figs. 37, 40, 44 and 46). Given the uncertainities we obtained a range of values but we finally choosed those variables that had the best fit with depth to detachment and rock uplift, surface uplift and exhumation through time. With such radius of curvature we calculated the slip along the main listric inverted fault assuming an additional - 87 -

4. Climatic forcing of asymmetric orogenic evolution

constrain: A retrodeformed state where lowermost Cretaceous hangingwall and footwall cutoffs fit to each other (which equals backrotating 7° the backlimb to zero dip) plus some differential subsidence in the downthrown block of the Servitá fault during the Lower Cretaceous, assuming the Farallones Anticline as an amplified rollover of the listric detachment as interpreted by Mora et al. (2006). This means an additional 1° rotation along the fault (Fig. 44 and 46). Finally we found the best fit with a total rotation of 8°, ca. 21Km of slip along the main fault and a radius of curvature of about 158.5 Km (Figs. 44 and 46). We used those geometries to calculate the shortening values obtained in our cross sections. Another subsequent step was to incrementally retrodeform the deformed state cross sections (Fig. 44). In this step we used the mentioned considerations plus the data of total amount of overburden from vitrinite reflectance and removed overburden at each stage calculated from our thermochronological data.

- 88 -