Abstract

We integrate analysis of present-day topography with a synthesis of current knowledge of the geology, deformation history, exhumation history, and the pattern of erosion rates to address the controversies surrounding the surface uplift history of the Bolivian Andes and the relative roles of climate and tectonics in the evolution of this mountainous landscape. Using metrics of channel steepness (ksn, a measure of channel slope normalized by drainage area), local relief (over a 2.5 km radius), and hillslope gradient we identify and map a suite of previously unrecognized perched, low-relief upland landscape patches in northern Bolivia that define a long-wavelength (∼300 km) topographic ramp with an ∼3.5 km elevation drop from SE to NW. We interpret these low-relief patches as the remnants of a formerly continuous low-relief landscape formed on grade with the foreland that has been uplifted and warped since formation. The 11–7 Ma Cangalli Formation on-laps the northern end of this ramp and suggests a shared history and common baselevel with the well-known ca. 12–9 Ma San Juan del Oro erosion surface in southern Bolivia. Patterns of rock and surface uplift rate implied by this interpretation are consistent with those inferred independently from analysis of channel steepness (ksn), the distribution of fluvial hanging valleys, reconstruction of channel profiles, and the distribution of published low temperature thermochronometric ages. These data reinforce earlier interpretations for 2–3 km surface uplift in the Bolivian Andes in the last 12 Ma. The marked contrast in topography across the Santa Cruz bend (∼18°S) appears largely controlled by differences in tectonics conditioned by inherited geologic contrasts and we show that post–12 Ma erosion and exhumation in northern Bolivia are controlled primarily by tectonics, not climate. However, the pattern and style of uplift in many locations suggests warping over deep structures in absence of significant shortening, consistent with the observation that exhumation patterns cease to be coupled to patterns of contractile deformation post-11–15 Ma. Tectonically controlled topography strongly focuses rainfall and may enhance erosional efficiency and thus erosion rates in zones of high rainfall, but no data from this study area demands this. Moreover, we find no evidence for a tectonic response to the modern rainfall pattern. We speculate that differences in the pattern of deformation and uplift across the Santa Cruz bend appear to be accommodated by a crustal-scale relay ramp.

1. MOTIVATION

Study of the deformation, erosion, and landscape evolution of the flanks of the Central Andean Plateau (CAP)—defined by Allmendinger et al. (1997) as the region above 3 km elevation between 13°S and 27°S that includes the Puna plateau, the Altiplano, and parts of the Eastern and Western Cordilleras and thus not restricted to internally drained regions—is critical to testing provocative hypotheses regarding the timing, magnitude, and driving mechanisms of plateau surface uplift (e.g., Isacks, 1988; Allmendinger et al., 1997; Garzione et al., 2006; Oncken et al., 2006; Barnes and Ehlers, 2009). As such, the style, rate, and timing of deformation in the Bolivian Andes—the eastern flank of the Central Andean Plateau (Fig. 1)—is of considerable interest to geologists, geophysicists, and geomorphologists interested in the interaction of processes in Earth’s interior with those acting on its surface and, consequently, has been much debated. Much has been learned about (1) the style and timing of contractile deformation (e.g., Kley, 1996, 1999; McQuarrie and DeCelles, 2001; McQuarrie, 2002; Elger et al., 2005; Horton, 2005; Oncken et al., 2006; Barnes and Ehlers, 2009); (2) the rate and timing of exhumation via study of erosion surfaces (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006), low-temperature thermochronology (Benjamin et al., 1987; Barnes et al., 2006; Gillis et al., 2006; Safran et al., 2006; Ege et al., 2007; Barnes et al., 2008; McQuarrie et al., 2008a; Barnes et al., 2012), cosmogenic radionuclide concentrations (Safran et al., 2005; Insel et al., 2010), and sediment yield (Guyot et al., 1990; Guyot et al., 1993; Aalto et al., 2006; Barnes and Pelletier, 2006); and (3) modern strain fields via GPS (e.g., Horton, 1999; Lamb, 2000; Brooks et al., 2003; Kendrick et al., 2003; Allmendinger et al., 2005; Brooks et al., 2011). Despite this progress, fundamental questions remain unanswered and key hypotheses incompletely tested.

Masek et al. (1994) first suggested that the dramatic climatic contrast between the northern and southern Bolivian Andes (Bookhagen and Strecker, 2008) is primarily responsible for the marked contrast in topography and depth of exhumation on either side of the Santa Cruz bend at ∼18°S (Figs. 1 and 2). Although this idea has been further explored and advanced in numerous studies (e.g., Horton, 1999; Montgomery et al., 2001; Strecker et al., 2007; McQuarrie et al., 2008b; Norton and Schlunegger, 2011; Schlunegger et al., 2011), and has well-established theoretical appeal (e.g., Willett, 1999; Whipple, 2009), it remains controversial. Differences in geologic history reflected in the paleogeography of Paleozoic depositional basins on either side of the bend provide a plausible alternate explanation (Sheffels, 1995; Allmendinger and Gubbels, 1996; Baby et al., 1996; Kley et al., 1999; Giraudo and Limachi, 2001; Oncken et al., 2006) and thus both the idea that climate has importantly influenced the topography on either side of the bend and the more intriguing notion that this difference in erosional efficiency and topography induced a change to the rate, locus and style of deformation remain unproven. Similarly it remains unclear whether the plateau and its flanks share a common Neogene uplift history and whether the rise of the plateau has been slow and steady since ca. 45 Ma or largely accomplished since ca. 12 Ma (Barnes and Ehlers, 2009).

This paper is motivated by the many unanswered questions fundamental to both the tectonic history of the Bolivian Andes and our understanding of the interactions among topography, climate, erosion rates, and tectonics in general that are highlighted in the literature cited above. In particular, we focus on the following six questions. (1) Has there been significant (>2 km) surface uplift post–12 Ma? (2) Does the spatio-temporal pattern of uplift of the margin constrain the mechanism(s) responsible for plateau uplift? (3) What factors dominantly control the marked contrast in topography across the Santa Cruz bend (∼18°S)? (4) How are any differences in the pattern and history of deformation and uplift across the Santa Cruz bend accommodated? (5) Are erosion and exhumation patterns and rates in northern Bolivia since 12 Ma controlled primarily by climate or tectonics? (6) Is there evidence of a tectonic response to either the climatic gradient across the Santa Cruz bend or to the pattern of intense orographic precipitation in northern Bolivia? We argue that when integrated with a detailed analysis of the topography using the tools of the tectonic geomorphology of erosional landscapes (e.g., Wobus et al., 2006b; Kirby and Whipple, 2012; Whittaker, 2012) sufficient data are now available to answer these fundamental questions.

2. APPROACH AND SCOPE

We focus on the eastern flank of the Central Andean Plateau in northern Bolivia between 15°S and 18°S in relation to the well-studied plateau margin in southern Bolivia between 18°S and 22°S (Figs. 1 and 2). We integrate analysis of present day topography with a synthesis of current knowledge of the geology, deformation history, exhumation history, and the pattern of erosion rates (measured at millennial timescales). Using metrics of channel steepness (ksn, a measure of channel slope normalized by drainage area), local relief (measured over a 2.5 km radius), and hillslope gradient (all measured on a 90 m resolution digital elevation model or DEM) we follow established methodologies to extract information about the pattern and history of rock uplift rate relative to baselevel (foreland plains) from topography (e.g., Kirby et al., 2003; Whipple, 2004; Wobus et al., 2006b; Whittaker et al., 2008; Kirby and Whipple, 2012; Whittaker, 2012).

Importantly, our analysis of topography is both constrained by a wealth of data on the spatio-temporal patterns of deformation, exhumation, and erosion rates, and guided by a growing body of theory surrounding the evolution of bedrock river profiles and their role in landscape evolution (e.g., see reviews by Whipple, 2004; Kirby and Whipple, 2012; Whittaker, 2012). An essential part of our analysis that is novel to studies of the northern Bolivian Andes is to map the distribution of low-relief landscape remnants perched high above deep river canyons. Our mapping of low-relief perched landscape patches reveals a previously unrecognized but striking physiographic feature: a SE-NW–oriented, long-wavelength (∼300 km) topographic ramp along which mean elevation descends from ∼4500 m near Cochabamba to ∼1000 m near Tipuani (Fig. 2A). We argue that the Cochabamba-Tipuani ramp records much about the amount, timing, and style of uplift in the region. Following Spotila et al. (1998), Schoenbohm et al. (2004), Clark et al. (2005), and Whittaker et al. (2007) among others, we interpret the low-relief landscape patches that define the Cochabamba-Tipuani ramp as relict landscapes that record past conditions (e.g., lower rock uplift rates or more efficient erosion reflecting a different climate state). We explore the potential causes for preservation of these perched low-relief landscape remnants and discover that the formation of previously unrecognized fluvial hanging valleys is responsible (Wobus et al., 2006a). A recently developed theory predicts that the uplift rate required to trigger the formation of hanging tributary valleys scales with the square root of tributary drainage area (Crosby et al., 2007; Gasparini et al., 2007). Comparison of the distribution and height of fluvial hanging valleys with this theoretical prediction strongly supports a tectonic interpretation for the formation of the Cochabamba-Tipuani ramp, which we combine with other indicators of the spatial and temporal history of rock uplift in our analysis. Finally, we study the relation between these low-relief landscape remnants in the northern Bolivian Andes and the remnants of both the 11–7 Ma Cangalli Formation in northern Bolivia (Fornari et al., 1987; Mosolf et al., 2011) and the ca. 12–9 Ma San Juan del Oro erosion surface in southern Bolivia (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006).

3. SETTING: GEOLOGY, TOPOGRAPHY, AND CLIMATE

The geologic, topographic, and climatic setting of the Bolivian Andes is summarized in Figure 2. As described first by Masek et al. (1994), the topography and climate of the northern Bolivian Andes are dramatically different from the long-wavelength topographic taper and semi-arid conditions south of the Santa Cruz bend at ∼18°S (Figs. 1 and 2). The northern Bolivian Andes are characterized by two dramatic but discontinuous topographic escarpments with up to 3–4 km of elevation gain (here termed the Chapare escarpment, CE, and the Beni escarpment, BE) and associated orographic rainfall locally in excess of 3 m/yr (Fig. 2A, 2C). Both topographic escarpments are characterized by deep canyons, high local relief, and steep channels. Figure 2D shows local relief measured as the range of elevations within a 2.5 km radius of each pixel in the DEM (following Montgomery and Brandon, 2002). Channel steepness (ksn), defined by the relation forumla, where S is local river slope [m/m], A is upstream drainage area [m2], and θref is the reference concavity, here set equal to 0.45 (e.g., Wobus et al., 2006b; Kirby and Whipple, 2012), is also shown for all streams with drainage area greater than 50 km2.

Given the north-south contrast in topography and climate (Figs. 2A, 2C), it is perhaps surprising that the tectonostratigraphic architecture of the Bolivian Andes is remarkably consistent along strike. The main difference in the outcrop pattern is the greater depth of exhumation, including exposure of granites and high grade metamorphic rocks in the Cordillera Real and absence of the Mesozoic and Paleocene section, evident in northern Bolivia (Fig. 2B, Pareja et al., 1978; Guarachi et al., 2001). Both the northern and southern Bolivian Andes are commonly divided into four tectonic/physiographic provinces: the Altiplano, Eastern Cordillera (EC), Interandean Zone (IA), and Subandes (SA) (Fig. 2B; e.g., Kley, 1996; McQuarrie, 2002). We note, however, that the IA is best defined in southern Bolivia where it generally corresponds to the thrust contact between Ordovician and Upper Paleozoic rocks (Fig. 2B). North of the bend, the EC/IA contact is often shown on regional maps as sub-parallel to, but ∼25 km SW of, this contact (e.g., McQuarrie et al., 2008a; Insel et al., 2010; Barnes et al., 2012) and the parts of the IA on either side of this thrust have distinct deformation and exhumation histories (Fig. 2C, 2D; McQuarrie et al., 2008a; Barnes et al., 2012)—a complication we discuss in section 7. Published apatite fission-track (AFT) ages (Benjamin et al., 1987; Barnes et al., 2006; Gillis et al., 2006; Safran et al., 2006; Ege et al., 2007; Barnes et al., 2008; McQuarrie et al., 2008a; Barnes et al., 2012) provide the most extensive record of the exhumation of the Bolivian Andes (Figs. 2C, 2D). The most striking north-south difference in exhumation history found by these authors is the rapid late Miocene-Pliocene exhumation recorded in AFT data from the Chapare and Beni escarpments and associated deep canyons (Figs. 2C, 2D), restricted to high relief areas north of the Santa Cruz bend and unmatched south of Cochabamba (Barnes et al., 2012). We will argue below that the topographic, climatic, and exhumation differences reflect post–12 Ma uplift and that the overall geologic similarity reflects the Eocene–Early Miocene contractile deformation of the EC and IA.

4. REGIONAL HISTORY OF DEFORMATION AND EXHUMATION

We focus on the last 12 m.y. of deformation, exhumation, and uplift of the northern Bolivian Andes with the aim of evaluating evidence for rapid surface uplift of the Central Andean Plateau and understanding the different morphologic and tectonic histories on either side of the Santa Cruz bend. To place our analysis in context, we begin with a brief synopsis of the Eocene–Middle Miocene history of deformation and exhumation of the Bolivian Andes.

The deformation histories of the EC, IA, and SA in Bolivia are now fairly well known and although there are some differences in timing between the southern and northern segments, a generally coherent picture has emerged over the past decade or so. The timing of deformation is directly constrained by the relation to folding and faulting of Tertiary sediments (e.g., Elger et al., 2005; Horton, 2005; Oncken et al., 2006; Barnes and Ehlers, 2009; Murray et al., 2010; Mosolf et al., 2011) and intrusions (e.g., Gillis et al., 2006). In addition, the Cenozoic exhumation history is constrained by extensive low-temperature thermochronometric data (mostly AFT, Figs. 2C, 2D) in both northern and southern Bolivia (Benjamin et al., 1987; Barnes et al., 2006; Gillis et al., 2006; Safran et al., 2006; Ege et al., 2007; Barnes et al., 2008; McQuarrie et al., 2008a; Barnes et al., 2012).

Although some details remain to be reconciled and there remains room for alternate interpretations of the thermochronometric data (Barnes and Ehlers, 2009), the constraints on timing of Cenozoic deformation and exhumation are generally concordant and, in conjunction, reveal a simple history. The Eastern Cordillera (EC) was a doubly vergent, contractile orogenic wedge active between ca. 45 and ca. 25 Ma (e.g., Horton, 2005; Barnes et al., 2012), with peak exhumation between 36 and 27 Ma in the central EC of southern Bolivia (Elger et al., 2005; Oncken et al., 2006; Ege et al., 2007; Barnes et al., 2008) and between 40 and 25 Ma in northern Bolivia (Benjamin et al., 1987; Barnes et al., 2006; Gillis et al., 2006; Safran et al., 2006; McQuarrie et al., 2008a). In southern Bolivia, the eastern EC is thought to have been the foredeep associated with this Eocene-Oligocene orogenic wedge, as recorded in thick Eocene-Oligocene megafan deposits preserved in the Camargo, Incapampa, Torotoro, Otavi, and Morochata synclines (Horton and DeCelles, 2001; Horton, 2005), and thus resided at low elevation at this time, consistent with paleobotanical estimates of paleoaltimetry (Gregory-Wodzicki et al., 1998; Gregory-Wodzicki, 2000). Much of the foredeep sediment cover was removed during Oligocene-Middle Miocene exhumation of the eastern EC. Shortening in the EC slowed by 25 Ma, largely ending by 21 Ma, and shallow intermontane basins formed between 25 and 17 Ma (Horton, 2005). Deformation and exhumation of the Interandean Zone (IA) overlaps with that in the EC, though it is generally younger, with the onset of deformation estimated between 25 and 22 Ma in the south (Elger et al., 2005; Oncken et al., 2006; Ege et al., 2007; Barnes et al., 2008).

From 17 to 9 Ma, the EC and IA of southern Bolivia experienced little deformation and the extensive low-relief San Juan del Oro (SJdO) erosion surface (Fig. 2) was formed and associated sediments and volcanic rocks deposited (Gubbels et al., 1993; Kennan et al., 1997; Horton, 2005; Barke and Lamb, 2006). In northern Bolivia, the multi-chronometer and age-elevation study of Gillis et al. (2006) documented a hiatus in exhumation between 25 and 11 Ma in the EC (specifically within the Cordillera Real) that is consistent with, if not demanded by, the AFT data of Barnes et al. (2006) and McQuarrie et al. (2008a), and corroborated by the timing of deformation of sedimentary basins and granitic intrusions (Gillis et al., 2006; Murray et al., 2010). In southern Peru, thermochronometric data and map relations (Lease and Ehlers, 2013) imply that a low-relief erosional bench at the plateau margin had to be beveled in the interval between 15 and 5 Ma after cessation of activity in the EC backthrust zone. In addition, the easternmost EC and westernmost IA in northern Bolivia is beveled by low-relief erosion surfaces associated with the 11–7 Ma Cangalli Formation (Fornari et al., 1987; Mosolf et al., 2011) (Figs. 2A, 2B), documenting a similar period of tectonic quiescence as that recorded by the contemporaneous SJdO erosion surface of southern Bolivia.

The cessation of deformation within the EC and IA generally corresponds with the onset of deformation within the Subandean fold and thrust belt (SA), thought to record the onset of underthrusting of the Brazilian Shield (e.g., Allmendinger and Gubbels, 1996; Allmendinger et al., 1997; Oncken et al., 2006). The onset of exhumation in the SA is only broadly constrained by available AFT thermochronologic data to between 20 and 8 Ma (Barnes et al., 2008; McQuarrie et al., 2008a), but has been directly tied to the onset of deposition of the Yerba Formation at 12.4 Ma in southern Bolivia by Uba et al. (2009). Since ca. 12 Ma the undeforming EC and IA have served as backstop to the SA fold and thrust belt (e.g., Brooks et al., 2011). Despite the lack of internal deformation in the EC, in northern Bolivia the onset of SA deformation at ca. 12 Ma roughly coincides with renewed rapid exhumation in the Cordillera Real and canyons of the Beni Escarpment at ca. 11 Ma (Gillis et al., 2006; McQuarrie et al., 2008a; Barnes et al., 2012). How this rapid exhumation and associated carving of deep canyons is related to tectonics, surface uplift, and climate is a central question addressed here.

4.1. Evidence for Post–12 Ma Surface Uplift

The mid-Miocene cessation of deformation and exhumation in the EC and IA is most definitive in the south where volcanic deposits draping low-relief paleo-landscapes form the San Juan del Oro (SJdO) erosion surface and attest to a lack of deformation and erosion between 12 and 9 Ma (Gubbels et al., 1993; Kennan et al., 1997; Horton, 2005; Barke and Lamb, 2006). Notably, the erosion surface remnants are beveled into both the EC and IA in southern Bolivia (Figs. 2A, 2B). Formation of this low-relief landscape by 12 Ma requires a period of tectonic quiescence of at least a couple million years (e.g., Baldwin et al., 2003), consistent with available data cited above. In addition, erosional beveling of a broad region requires either a stable high-elevation baselevel maintained over millions of years (such as internal drainage) or formation on a gentle grade above the foreland. Given clear evidence for integrated drainage to the foreland, this reasoning has been the grounds for the geomorphic argument that the EC of southern Bolivia has experienced significant post–12 Ma surface uplift in the absence of shortening (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006). In addition, evidence of a climate change at ca. 8.5 Ma in southern Bolivia is consistent with changes to atmospheric circulation and orographic enhancement of precipitation predicted to accompany Miocene surface uplift of the EC and IA (Ehlers and Poulsen, 2009; Uba et al., 2009; Mulch et al., 2010).

Barke and Lamb (2006) employed several approaches to estimate the amount of surface uplift recorded by the SJdO erosion surface. They reported a composite estimate of 1.7 ± 0.7 km surface uplift as a figure that brackets most of their data. This composite estimate, however, is greatly influenced by low and internally inconsistent estimates derived from the elevation of modern knickpoints on major rivers that are deeply incised into the SJdO surface and thus provide only a crude minimum constraint on the amount of surface uplift since beveling of the SJdO landscape. Using more reliable estimates of the height of formation of the SJdO surface relative to the paleo-foreland and downstream projections of reconstructed paleo-drainages, Barke and Lamb (2006) derive an internally consistent estimate of 2.4 ± 0.4 km post–9 Ma surface uplift of the erosionally beveled EC and IA.

Barke and Lamb’s geomorphic estimate of surface uplift compares favorably with three independent paleoaltimetry estimates. Gregory-Wodzicki et al. (1998) studied the 20.8–13.8 Ma Potosi flora deposited in a shallow basin on the erosionally beveled EC of southern Bolivia and determined that it recorded >3 km post-depositional surface uplift. Graham et al. (2001) later reached similar conclusions (∼2.4 km surface uplift) for the younger Pislepampa flora preserved near the EC/IA boundary ∼20 km northeast of Cochabamba. Similarly, in the central Altiplano (near the 68°W18°S graticule, Fig. 2), Ghosh et al. (2006) and Garzione et al. (2006) present evidence for plateau surface uplift of 3.4 ± 0.6 km from the clumped isotope method and, independently, 3.0 ± 0.5 km from the oxygen isotope method since 10.3 Ma. As summarized by Barnes and Ehlers (2009), and as will be discussed later, similar estimates for the timing and magnitude of surface uplift have been published for the Central Andean Plateau based on the geology and geomorphology of the Western Andean Escarpment in Peru and Chile (Wörner et al., 2002; Victor et al., 2004; Hoke et al., 2007; Schildgen et al., 2007; Thouret et al., 2007; Hoke and Garzione, 2008; Schildgen et al., 2009a, 2009b; Jordan et al., 2010). Although some contrary evidence has been published (e.g., Barnes and Ehlers, 2009; Evenstar et al., 2009; Hartley and Evenstar, 2010) and interpretations of δ18O records have been challenged (e.g., Ehlers and Poulsen, 2009; Insel et al., 2012), much independent geomorphic and paleoaltimetric evidence is consistent with ∼3 km surface uplift since ca. 11 Ma.

Despite this considerable evidence for late Miocene-present surface uplift of the CAP and its margins, uncertainty remains over the degree to which these data could record climate change as opposed to surface uplift: cooling could bias paleoaltimetry estimates, an increase in flood runoff could trigger incision of previously uplifted surfaces (Schildgen et al., 2007; Lease and Ehlers, 2013), and even gradual uplift of the CAP above a threshold elevation could trigger an abrupt change in rainfall and the distribution of moisture sources that could significantly bias estimates based on δ18O measurements (e.g., Ehlers and Poulsen, 2009; Insel et al., 2012). Accordingly we ask whether there is any unequivocal evidence for surface uplift of this magnitude and seek to evaluate the uplift history of the northern Bolivian Andes for comparison with the well-known, if controversial, evidence for surface uplift in southern Bolivia. Other than the pattern and timing of late Miocene-present exhumation and millennial-scale erosion rates, no direct estimates of the timing, pattern, and amount of surface uplift in northern Bolivia have been published. In a related study, Gasparini and Whipple (2014) study the uplift history of the Beni escarpment (BE, Figs. 2A, 2C) in detail. We provide no new constraints on the age of landforms, rates of erosion, or timing of events, but tie our analysis to published data. A key question we pursue in this study is whether there is a relationship between the San Juan del Oro (SJdO) erosion surface of southern Bolivia and the contemporaneous Cangalli Formation and associated low-relief erosion surfaces in the foothills of the northern Bolivian Andes.

5. TECTONIC GEOMORPHOLOGY OF THE NORTHERN BOLIVIAN ANDES

In this section, we present our analysis of the tectonic geomorphology of the northern Bolivian Andes. Our topographic analysis focuses on the spatial distribution of channel steepness, local relief, low-relief erosion surfaces, and fluvial hanging valleys in context of available geologic, climatic (modern rainfall), thermochronologic, and cosmogenic nuclide data. The section is subdivided into four parts in which we present (1) interpretation of the rock uplift pattern and candidate active structures as inferred from topography, erosion rate patterns, low temperature thermochronology, and the mapped distribution of the 11–7 Ma Cangalli Formation; (2) the identification and distribution of perched low-relief landscape patches; (3) the interpretation of recent (post–ca. 12 Ma) surface uplift patterns from these low-relief landscape remnants and their association with the Cangalli Formation; and (4) analysis of fluvial hanging valleys that may have played a critical role in the preservation of these low-relief landscapes. Methods used are described in each of these sections as appropriate.

5.1. Relief, Channel Steepness, and Inferred Rock Uplift Patterns

As summarized in a recent review, where rock strength and climate are uniform, patterns of both the channel steepness index, ksn, and local relief (at the 1–10 km scale) reflect the spatio-temporal history of rock uplift, with higher rock uplift rates producing steeper channels and greater local relief (Kirby and Whipple, 2012, and references therein). Variations in both rock strength and climate, however, must always be considered. In this regard, geologic constraints on rock uplift patterns and data on patterns of erosion and exhumation rates are important components of robust interpretation of tectonics from topography (Wobus et al., 2006b; Kirby and Whipple, 2012; Whittaker, 2012).

Rock strength can potentially influence topography as much as rock uplift rates (e.g., Duvall et al., 2004), but for most of the area of interest here (IA and EC of the northern Bolivian Andes), rock strength appears to vary mostly at the scale of alternating sandstone and shale beds within the Paleozoic (mostly Ordovician to Devonian) sedimentary sequence (Safran, 1998; Syrek and Barnes, 2011) that outcrops over nearly the entire area (Fig. 2B). Although we lack definitive data, our interpretation is that except as noted below in regard to the weak rocks of the Subandes and the stronger granites and high grade metamorphic rocks of upper reaches of the Beni Escarpment—in the Cordillera Real (Safran, 1998; Syrek and Barnes, 2011; Gasparini and Whipple, 2014)—lithology is unlikely to much influence the regional topography. We find no indication of a correlation between the zones of high relief and channel steepness (Fig. 2D) and lithology (Fig. 2B, Guarachi et al., 2001).

The strong spatial variability of modern rainfall in the northern Bolivian Andes (Fig. 2B) likewise does not, for the most part, greatly impact our ability to interpret rock uplift patterns, though it—and the likelihood of significant climate changes on the million-year timescale—does complicate any attempts to invert topography for a quantitative estimate of rock uplift rates. This follows because, all else held equal, areas experiencing higher rates of rainfall are expected to develop lower relief and lower channel steepness indices (Kirby and Whipple, 2012 and references therein). The strong positive correlation between areas with high local relief and high ksn values and areas of orographically enhanced modern rainfall (Figs. 2C, 2D; Bookhagen and Strecker, 2008), therefore strengthens the interpretation that high rock uplift rates are required to support such steep and rugged topography despite the vigorous erosional environment. The much greater efficiency of glacial erosion (generally above ∼3800 m in the northern Bolivian Andes) constitutes an important exception: areas of low-relief and channel steepness within the glaciated zone need not imply low rates of rock uplift and erosion, as discussed below.

Figure 3 shows a higher resolution map of the topographic analysis presented in Figure 2D (patterns of local relief and channel steepness index values) together with the distribution of the 11–7 Ma Cangalli Formation and published AFT ages discussed below. For clarity, patterns of millennial erosion rates determined from cosmogenic isotope concentrations in river sands are not show in Figure 3, but maps have been published by Safran et al. (2005), Insel et al. (2010), and Hippe et al. (2012) and are included in Gasparini and Whipple (2014). In combination these data allow us to interpret the patterns of recent rock uplift and thus tentatively identify the major active deformational structures (also shown on Fig. 3) using the principles outlined above. In short, active rock uplift is interpreted to support areas of high local relief, high channel steepness indices, young AFT ages, and high erosion rates; active faults are inferred where an abrupt change in a combination of local relief, channel steepness, AFT age, and erosion rate is observed across known faults, or based on prior work.

Inferred active structures include the Subandean Thrust (SAT), parts of the Main Frontal Thrust (CFP—Cabalgemiento Frontal Principal [Guarachi et al., 2001]) and Main Andean Thrust (CANP—Cabalgemiento Andino Principal), and an unmapped structure at the foot of the Beni Escarpment (Fig. 3B). The SAT is known to be the active thrust front (e.g., Baby et al., 1997; Guarachi et al., 2001; Brooks et al., 2011) and while the Subandes have a distinct morphological expression, the generally erodible rocks, narrow anticlinal uplifts, and gentle taper of the fold and thrust belt mean there is little expression in either our channel steepness or 2.5 km local relief maps given that color ramps are chosen to capture the much stronger variation in the EC and IA. The Main Frontal Thrust (CFP) and Main Andean Thrust (CANP) appear active in the southern part of Figure 3B where they (or deep structures beneath them) combine to build the Chapare Escarpment (defined in Figs. 2A and 2C and clearly visible on Fig. 3), including an ∼1.5-km-high escarpment in the immediate hanging wall of the CANP along the margin of the low-relief surface NE of Cochabamba (Figs. 3 and 4B). The AFT ages in the Chapare transect of Barnes et al. (2012) include some of the youngest in the Bolivian Andes and are all in the hanging wall of these faults and mostly, but not all, in the hanging wall of the CANP. Recent activity on the CANP along the Chapare escarpment is strongly suggested by the abrupt change in both relief and ksn immediately across the fault. Indicators of active uplift along both faults, however, fade away to the north as the Chapare escarpment tapers off, with no morphological expression northwest of Tipuani. Miocene AFT ages between the CFP and CANP in the Barnes et al. (2006) and McQuarrie et al. (2008a) transects (Fig. 3B, inset) attest to Neogene activity on the CFP at least this far north. There is no evidence for Neogene activity on the CANP northwest of Tipuani: indeed even south of Tipuani AFT ages between the Beni escarpment and the CANP are all Oligocene (Fig. 3B, inset); cosmogenic nuclide-derived erosion rate estimates in this zone are uniformly low (Safran et al., 2005; Insel et al., 2010); the 11–7 Ma Cangalli Formation and associated low-relief erosion surfaces are largely undeformed (Fornari et al., 1987; Mosolf et al., 2011); and the uplift and incision of these low-relief surfaces (discussed in section 5.2 below) is minimal northwest of Tipuani.

All data point to relatively high uplift and erosion rates along the Beni escarpment, although the combination of stronger rocks in the core of the Cordillera Real (granites and high grade metamorphic rocks), the pattern of orographic precipitation, and glaciation at high elevation complicate detailed interpretation (e.g., Schlunegger et al., 2011; Gasparini and Whipple, 2014). First, mean elevation, local relief, and channel steepness indices all increase abruptly within the Beni escarpment (Figs. 3 and 4). Second, millennial erosion rates from cosmogenic isotope concentrations are considerably higher within the Beni escarpment (Safran et al., 2005; Gasparini and Whipple, 2014). Third, Neogene AFT ages are restricted to the deep canyons of the Beni escarpment and glaciated landscapes of the Cordillera Real (Fig. 3B). Fourth, the mapped distribution of the 11–7 Ma Cangalli Formation (Fornari et al., 1987; Guarachi et al., 2001; Mosolf et al., 2011) implies recent uplift of the Beni escarpment (Figs. 3, inset, and 4C).

A structural boundary at the foot of the escarpment appears to be required (solid unlabeled black line in Figure 3B, dashed to indicate attenuation and uncertain extent to the southeast), but whether this boundary is a fault, shear zone, or fold and how it relates to deeper structure (décollement ramp, duplex, or basement thrust sheet) is unclear, as will be discussed later. Models presented by Gasparini and Whipple (2014) further suggest that this structure must remain active well into the Quaternary, possibly to the present—incision into a previously uplifted margin is not consistent with landscape morphology. Mean elevation, local relief, and channel steepness indices of rivers of greatly differing size all change abruptly across this boundary (Figs. 3 and 4). Although this physiographic transition is similar to that between the lesser and higher Himalaya in Nepal (e.g., Masek et al., 1994), the foot of the Beni escarpment is not marked by abrupt knickpoints on channels as expected for an active thrust fault, but rather by a 30-km-wide high-concavity zone (Schlunegger et al., 2011)—a gradient in channel steepness rather than a step-function change. This broad high-concavity zone could reflect a lateral gradient in rock uplift rate (e.g., Kirby and Whipple, 2001), the influence of the orographic distribution of precipitation and runoff (e.g., Craddock et al., 2007; Schlunegger et al., 2011), a temporal decline in rock uplift rate (e.g., Whipple and Tucker, 2002; Baldwin et al., 2003) as would be the case if uplift precedes a climate-triggered phase of river incision, or a combination of these. Gasparini and Whipple (2014) analyze topography, climate, and erosion rate patterns along the Beni escarpment in detail and compare it to synthetic topographies using the CHILD landscape evolution model (Tucker et al., 2001) to quantitatively constrain the relative contributions of all potential factors. For our purpose here, we simply note that both the erosion rate pattern (Safran et al., 2005) and the deformation of the Cangalli Formation (Fig. 4C; Fornari et al., 1987) strongly suggest both an active tectonic boundary at the foot of the Beni escarpment and an associated gradual increase in rock uplift rate to the southwest (see Gasparini and Whipple, 2014). Implications of this interpretation will be discussed in section 6 below.

5.2. Distribution of Perched, Low-Relief Landscape Remnants

As noted in southern Bolivia (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006), isolated patches of perched, high-elevation, low-relief landscapes often record landscape morphology prior to a change in conditions that results in river canyon incision and relief production (e.g., Kirby and Whipple, 2012). These perched remnants of former landscapes are termed “relict” as they record prior conditions of uplift and incision that established the low-relief condition. Similar relict landscape remnants have been used to identify, and in some cases to quantify the timing and magnitude of, surface uplift in diverse geologic settings including the Tibetan Plateau and its margins (Schoenbohm et al., 2004; Clark et al., 2006; Harkins et al., 2007), the Basin and Range Province (Clark et al., 2005; Kirby and Whipple, 2012), the transverse ranges of southern California (Spotila et al., 1998; Wobus et al., 2006b), the Alps and Pyrenees (Babault et al., 2005; van der Beek and Bourbon, 2008), the Sila Massif in Italy (Olivetti et al., 2012), the southern margin of the Anatolian plateau (Schildgen et al., 2012), the Woodlark Basin of Papau New Guinea (Miller et al., 2012), the Cordillera de Talamanca in Costa Rica (Morell et al., 2012), and the Appalachian Mountains in the eastern United States (Miller et al., 2013).

Although low-relief landscape remnants have been little discussed north of 18°S (Cochabamba) in Bolivia, Fornari et al. (1987) identified low-relief erosional landscapes at 1–2 km elevation just west of the Main Andean Thrust (CANP) that they associated with the 11–7 Ma Cangalli Formation, interpreting that these erosional landscapes graded to the top of Cangalli fill (surfaces in Fig. 3, inset, 500–1000 m above incised modern rivers). Given that these surfaces and the Cangalli Formation are contemporaneous with the San Juan del Oro (SJdO) surface and associated deposits (Barke and Lamb, 2006; Mosolf et al., 2011) we studied the landscapes of northern Bolivia using a 90 m resolution DEM (http://srtm.csi.cgiar.org/) to determine the extent and distribution of these low-relief landscape remnants.

Following the methods described in Schoenbohm et al. (2004) and Clark et al. (2006), and employed in most of the examples cited above, we utilize a combination of maps of local slope (measured over ∼270 m), local relief (within a 2.5 km radius of each pixel), and channel steepness (ksn) to identify, and map the spatial distribution of, perched low-relief landscape remnants. Perched low-relief landscape patches are identified here as areas of relatively low mean slope (local slopes ranging between 12 and 25°), low local relief (<800 m), and low channel steepness (<120 m–0.9) isolated within, and surrounded by, regions characterized by threshold slopes (mean slope >28°), higher local relief (>1000 m), and higher channel steepness values (Figs. 2 and 3). Rivers draining these perched, low-relief patches in all cases have abrupt increases in channel steepness (slope-break knickpoints) that demarcate the boundary of the remnant landscape (Fig. 3B, discussed in more detail in section 5.4). Once areas meeting all these criteria were identified, watershed boundaries were used to map the margins of perched low-relief drainage basins. The three northeasternmost low-relief remnants are the surfaces described and roughly mapped by Fornari et al. (1987) as associated with the 11–7 Ma Cangalli Formation (Fig. 3A).

We identify numerous, previously unrecognized, low-relief landscape patches perched above deeply incised canyons in the northern Bolivian Andes in addition to those noted by Fornari et al. (1987) (Figs. 2 and 3). These low-relief remnants are organized into two roughly strike-parallel swaths. One group of low-relief patches runs along the eastern flank of the Cordillera Real (the top of the Beni Escarpment) at a mean elevation of ∼4000 m. These surfaces are roughly concordant with the well-developed low-relief erosional surface described by Lease and Ehlers (2013) ∼200 km to the NW in southern Peru and are plausibly correlative. A second group runs just west of the CANP in a line stretching from Cochabamba in the SE to Tipuani in the NW (Figs. 2 and 3), dropping in elevation from ∼4500 m to ∼1000 m over that distance. This band of low-relief surfaces demarcates a landform we term the Cochabamba-Tipuani ramp that is well expressed in the topographic swath profile in Figure 4A (swath location shown in Fig. 3A). The on-lapping relationship between the Cangalli Formation and the three northern most low-relief surfaces, plus the concordance of the highest and southern-most surface near Cochabamba with the remnants of the contemporaneous San Juan del Oro (SJdO) surface to the south (Fig. 2A) (Barke and Lamb, 2006; Hoke and Garzione, 2008), together suggest the possibility that the Cochabamba-Tipuani ramp records the warping and uplift (the ∼3.5 km elevation difference ΔZC-T, Fig. 4A) of a formerly continuous low-relief landscape with a common baselevel—a possibility we explore in the following sections.

5.3. Interpretation of Low-Relief Landscape Remnants

Patches of low-relief landscapes perched above deeply incised canyons generally may be produced in four ways: (1) exhumation of a sub-horizontal weak over strong lithologic contact (e.g., Oskin and Burbank, 2005; Oskin and Burbank, 2007); (2) beveling at or near the glacial equilibrium line altitude (the “glacial buzzsaw”) (Brozovic et al., 1997); (3) maintenance of a stable or rising high-elevation local baselevel due to either internal drainage or active rock uplift (Sobel et al., 2003); (4) uplift of a low-relief surface formed at low elevation in response to a change in conditions that leads to a period of relief production such as an increase in rock uplift, or possibly a change in climate (e.g., Bonnet and Crave, 2003; Schoenbohm et al., 2004; Clark et al., 2006; Schildgen et al., 2007; Jeffery et al., 2013). An increase in rock uplift rate has been interpreted by most authors for the landscape remnants associated with the San Juan del Oro erosion surface in southern Bolivia (Fig. 2A) (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006; Hoke and Garzione, 2008). All mapped low-relief landscape patches cut across highly deformed folded and faulted rocks, mostly Ordovician-Devonian meta-sediments (Fig. 2B); there is no evidence for a lithologic control on erosion surfaces in the Bolivia Andes. The band of low-relief patches along the flank of the Cordillera Real could be either glaciated remnants of an uplifted low-relief landscape akin to the San Juan del Oro surface, or could merely reflect efficient glacial erosion: the 3800 m topographic contour effectively demarcates the extent and location of these surfaces, the elevation of knickpoints on transverse rivers that dissect them, and separates glaciated landscapes above from fluvial landscapes below (see Gasparini and Whipple, 2014). Although it is impossible to ascertain which is the correct interpretation on the basis of landscape morphology alone, the lack of high peaks and sharp arêtes suggests slowly eroding glaciated benches (e.g., Brocklehurst and Whipple, 2007). Indeed, erosional beveling in a low-erosion rate environment is suggested by (1) the concordance of these surfaces with the low-relief erosional surface described by Lease and Ehlers (2013) in southern Peru and (2) the very low (<20 m/m.y.) erosion rates reported in similar landscapes on the margin of the plateau near La Paz (Hippe et al., 2012). Still, the glaciation of these surfaces does leave considerable uncertainty.

Several lines of evidence suggest that the band of low-relief patches that form the Cochabamba-Tipuani ramp (Figs. 3 and 4A) can be more confidently interpreted to reflect recent surface uplift in response to a period of accelerated rock uplift. First, only the two highest surfaces show any evidence of glaciation. From geologic maps (Guarachi et al., 2001) and examination of satellite images, only the highest surface (here termed the Cochabamba surface) experienced significant glaciation, but even so much of this surface has fluvial morphologies and the low-relief landscape extends downstream of terminal moraines. Second, preservation of gravels of the Cangalli Formation on the northernmost three surface remnants attest to both uplift increasing to the SE and low erosion rates on the surface remnants—far slower erosion than in the surrounding canyons with angle of repose walls, as has been confirmed with analyses of detrital cosmogenic isotope concentrations (Safran et al., 2005). Indeed, even the partially glaciated Cochabamba surface preserves some remnants of sediments thought to date from the mid to late Miocene (hosting the Pislepampa flora of Berry [1922] and Graham et al. [2001]), attesting to slow erosion in spite of glaciation. Third, the Pislepampa flora corresponds to warmer (+ ∼10°C) and wetter (+ ∼950 mm/yr) conditions consistent with a paleo-environment equivalent to that found today at 1200–1400 m (basal elevations of the cloud forest), implying up to 2.4 km of surface uplift since deposition (less if accompanied by notable cooling and drying) (Graham et al., 2001). Fourth, the river knickpoints that demarcate the margins of each these low-relief patches (stars on Fig. 3) are particularly dramatic, meeting the criteria for fluvial hanging valleys (Wobus et al., 2006a) which, as shown in the following section, puts important constraints on the conditions of knickpoint formation (Crosby et al., 2007) that independently support formation in response to an increase in rock uplift rate, with the new rate increasing to the southeast along the ramp (section 5.4 below). Thus, the Cochabamba-Tipuani ramp appears to consist of remnants of a formerly continuous landscape now dissected by numerous transverse drainages (Figs. 3 and 4) that records a change in rock uplift rate that post-dates deposition of the Cangalli Formation (11–7 Ma), at least at the northwest end of the ramp—lateral propagation from SE to NW is possible.

The geometry of the Cochabamba-Tipuani ramp implies a long-wavelength warping of the surface with a total of ∼3.5 km surface uplift to the southeast (ΔZC-T, Fig. 4A). Following Schoenbohm et al. (2004) and Barke and Lamb (2006) we can independently estimate net surface uplift of the Cochabamba surface by projecting headwater river profiles beyond the abrupt knickpoints that demarcate the margins of this highest low-relief surface. To do this we determine the mean channel steepness and best-fit concavity indices upstream of surface-bounding knickpoints and use these to compute an estimate of the pre-uplift channel profile assuming negligible change in drainage area and network structure. Uncertainties on the channel steepness and concavity indices, and on the length of channel used in their determination, are used to bound the estimated amount of net surface uplift (ΔZ, see Kirby and Whipple, 2012 for discussion). For this analysis we choose the two largest rivers that drain to the NE, nearly perpendicular to the structural grain, off the large low-relief landscape patch near Cochabamba (marked S1 and S2 on Fig. 3B). This analysis is illustrated in Figure 5 where ΔZ, ΔZ1, ΔZ2, and their uncertainties place bounds on the estimated surface uplift. In the case of river S1, ΔZ1 provides a minimum bound as the surface uplift relative to the point where the river turns to run ∼200 km NW (S1′, Fig. 3B), mostly along the trace of the CFP, before joining the Beni River at S1″ and crossing the SA to its foreland baselevel. Conversely, ΔZ2 provides a maximum bound by measuring uplift at point S1′ relative to the foreland baselevel. Together, these results bracket net surface uplift (rock uplift minus river incision) on these two rivers to between 2.2 and 2.9 km. Climate change at ca. 8.5 Ma (Uba et al., 2009; Mulch et al., 2010) could accelerate the incision of canyons (Jeffery et al., 2013) and exhumation of rock (Barnes et al., 2012) and also reduce channel steepness below the knickpoints, but such an effect would not influence this estimate of net surface uplift. This estimate is more definitive than the modern stream profile projections performed by Barke and Lamb (2006) because modern rivers are not noticeably incised into the Cochabamba surface, unlike the major rivers draining the San Juan del Oro surface in southern Bolivia, which explains the much lower bounds of their analysis. Our analysis of unincised modern rivers is consistent, however, with their estimate of 2.4 ± 0.4 km surface uplift derived from reconstructed paleo-river profiles. While this profile projection estimate is less than the ∼3.5 km of surface uplift implied by the elevation drop on the Cochabamba-Tipuani ramp (Fig. 4A), it does substantiate the need for significant (2–3 km) recent (post low-relief surface formation) surface uplift.

5.4. Hanging Fluvial Valleys: Formation and Interpretation

Although only recently recognized, fluvial hanging valleys have now been identified in numerous tectonically active landscapes (Wobus et al., 2006a; Crosby et al., 2007; Goode and Burbank, 2009). Fluvial hanging valleys violate Playfair’s Law in that they are dramatically oversteepened, often into a cascade of waterfalls, as they merge with mainstem rivers. They only occur where tributaries enter a trunk stream with much greater drainage area (Wobus et al., 2006a). Fluvial hanging valleys have only been identified in areas of active rock uplift and appear to be associated with sudden increases in the rate of trunk stream incision. Importantly, the formation of a hanging valley insulates the landscape upstream from ongoing baselevel fall on the trunk stream, delaying landscape response and preserving relict landscapes much longer than otherwise expected (Wobus et al., 2006a; Crosby et al., 2007). The formation of hanging valleys at the margins of the perched low-relief landscape remnants described above (Fig. 3) is likely responsible for the preservation of these surfaces. In this area, hanging valleys have formed in two ways: (1) in response to rapid incision of large, transverse rivers that run roughly perpendicular to the Cochabamba-Tipuani ramp (Fig. 3A), and (2) along the trace of the Main Andean Thrust (CANP) in response to rapid rock uplift of its hanging wall (in the vicinity of rivers S1 and S2, Fig. 3B). The more gradual spatial variation in rock uplift rate within the Beni Escarpment (BE) may have prevented the formation of hanging valleys there (see Gasparini and Whipple, 2014). We show in this section how the characteristics of hanging valleys as a function of position along strike can be used to independently test the hypothesis that the Cochabamba-Tipuani ramp represents a tectonic warping of an initially continuous low-relief and low-elevation surface. In essence we ask whether the size of hung tributaries increases from NW to SE along the Cochabamba-Tipuani ramp in a manner consistent with the hypothesized rock uplift pattern and the prediction (Crosby et al., 2007; Gasparini et al., 2007) that the drainage area of the largest tributary that can be hung scales with the square of rock uplift rate.

A theory of hanging valley formation (Crosby et al., 2007; Gasparini et al., 2007) based on the saltation-abrasion river incision model (Sklar and Dietrich, 2004) predicts that mainstem river incision rate (I) must exceed a critical threshold to form tributary hanging valleys. The critical incision rate is a function of tributary drainage area, old (pre-perturbation) uplift rate, and the efficiency of both gravel transport and bedrock abrasion. The analytical hanging valley theory of Gasparini et al. (2007) is based on a generalization and simplification of the Sklar and Dietrich (2004) saltation-abrasion river incision model (see Whipple, 2004): 
graphic
where Ksa is a bulk erodibility coefficient that depends on rock susceptibility to abrasion, the threshold of bedload mobilization (grain size), and factors relating drainage area and slope to bed shear stress (runoff coefficient, channel width coefficient, and roughness) (Sklar and Dietrich, 2004; Gasparini et al., 2007). The negative exponents on drainage area (A) and slope (S) reflect the fact that all else held equal, erosion rate is actually expected to decrease with increasing bed shear stress because saltation hop length increases with shear stress, reducing the frequency of grain impacts on the bed (Sklar and Dietrich, 2004). The term in brackets represents the dual influence of sediment flux (tools and cover effects) on bedrock channel incision (Sklar and Dietrich, 2004) and ensures that despite the negative exponents on A and S, there is a positive, monotonic relationship between steady-state (where I = uplift rate, U) channel slope and rock uplift rate up to a maximum uplift rate (Sklar and Dietrich, 2004; Sklar and Dietrich, 2006). Beyond this maximum rock uplift rate (for a given climate, rock strength, and drainage area), channels are predicted to become tool-starved and steady-state channel slope is predicted to become infinite (Sklar and Dietrich, 2004; Sklar and Dietrich, 2006).
A related instability is predicted away from steady-state if a landscape is subjected to a sufficiently large temporal increase in rock uplift rate (Gasparini et al., 2007). Crosby et al. (2007) have shown that this instability may be responsible for the formation of fluvial hanging valleys (Wobus et al., 2006a) if the new rock uplift rate exceeds a critical value that depends on tributary drainage area, At, which represents the downstream increase in both water discharge and sediment flux. The critical rock uplift rate, Ucr, for hanging valley formation is derived by Gasparini et al. (2007) (solving their Equation 32 for the new rock uplift rate for channels with steady-state concavity index ∼0.5 as observed in most landscapes): 
graphic
where β is the fraction of eroded debris delivered to channels as bedload, Kt is the bedload transport coefficient and Uo is the old, or pre-perturbation, rock uplift rate. Importantly, Equation (2) determines whether a river crosses this instability threshold at the onset of uplift and mainstem river incision (Crosby et al., 2007; Gasparini et al., 2007). Although there are too many unknowns in Equation 2 to allow a quantitative comparison between model prediction and observations in the Bolivian Andes, we can usefully test whether field observations are consistent with the fundamental prediction in Equation 2 that, assuming invariant β, Kt, and Ksa, forumla, as illustrated below.

The test is quite simple. If our hypothesis that the Cochabamba-Tipuani ramp records differential uplift along this transect is correct, then the elevation of each relict landscape remnant relative to the incised rivers at the NW end of the ramp should be proportional to the average rock uplift rate each has experienced since uplift began. Since each relict landscape patch is rimmed by hanging valleys, the rock uplift rate it has experienced should be equal to or greater than Ucr for all hanging tributaries. To test this, we determine the drainage areas of the largest hanging tributary and of the smallest tributary river that did not form a hanging valley in the vicinity of each landscape remnant. These two data points then should bracket Acr, the critical drainage area for the formation of a hanging tributary valley, defined by solving Equation (2) for drainage area: forumla. We also identify a few intermediate cases where the rivers are notably over-steepened downstream of knickpoints but do not meet the criteria for hanging valleys (Wobus et al., 2006a) for which the drainage area must be close to, but just larger than Acr (for a discussion of how this phenomenon may have influenced landscape evolution in the Beni Escarpment see Gasparini and Whipple, 2014). The test for internal consistency, then, involves evaluating whether the critical drainage area for the formation of hanging valleys (Acr) varies along strike in a manner consistent with the rock uplift pattern implied by the rise in elevation of low-relief surface remnants along the Cochabamba-Tipuani ramp (Fig. 4A).

One important complication in this simple analysis is that we do not know how to quantitatively account for climatic differences from one surface remnant to the next, as β, Kt, and Ksa likely vary with climate in complex and potentially different ways—presumably as a function of changes in mean annual runoff, runoff variability, bedload grainsize, dominant hillslope processes, and the weathering state of both bed and bedload clasts (Sklar and Dietrich, 2004; Tucker, 2004; Lague et al., 2005; Sklar and Dietrich, 2006; DiBiase and Whipple, 2011). Modern rainfall does vary along the ramp, ranging between ∼0.3 m/yr and ∼1.5 m/yr (Fig. 2). However, the theory for whether a hanging valley forms or not addresses channel response to the onset of uplift and incision—according to this theory, a hang either does or does not form in response to the initial perturbation of the system (Crosby et al., 2007; Gasparini et al., 2007). The hypothesis we are testing here is that the low-relief surfaces along the ramp started at a common low elevation in a similar topographic position along a low-relief range front, and were subjected to an along-strike gradient in rock uplift rate. If supported, this would imply that all surface remnants likely experienced very similar climates at the onset of uplift and incision. For this reason and the lack of significant along-strike changes in lithology, we proceed with the assumption that β, Kt, and Ksa were uniform along the ramp at the moment of hanging valley formation and therefore that the problem can be reduced to testing the prediction that forumla.

We estimate the amount of rock uplift (in meters) relative to local baselevel as the elevation difference between the hypothesized warped low-relief surface (thick gray dashed line in Fig. 4A) and the incised rivers near Tipuani at the NW end of the Cochabamba-Tipuani ramp (ΔZH, Fig. 4A). For each river (largest hanging tributary, any oversteepened but not hanging tributary, and the smallest nearby graded tributary) we determine drainage area and differential rock uplift amount at the along-strike position of its confluence with the incised mainstem river (i.e., position of the confluence projected into the swath profile in Fig. 4A). By plotting the uplift amount inferred for each of these tributaries against their drainage area we can easily determine whether our bracketed estimates of Acr scale with the square of our uplift rate estimates (Fig. 6C). The hanging-valley theory matches data remarkably well, which both strongly supports our interpretation of the Cochabamba-Tipuani ramp as a tectonic warp, and also validates the hanging-valley theory of Gasparini et al. (2007) and Crosby et al. (2007).

6. REGIONAL UPLIFT PATTERNS AND INFERRED STRUCTURAL ARCHITECTURE

We infer from a combination of available exhumation rate and erosion rate data, channel steepness and local relief patterns, and the elevation of perched low-relief erosion surfaces that post–12 Ma rock uplift in northern Bolivia has been concentrated along the Chapare and Beni escarpments. As discussed above, this pattern implies significant uplift on, or beneath, the Main Andean Thrust (CANP) and probably the Main Frontal Thrust (CFP) in the southeastern part of Figure 3 that fades away to the northwest as uplift picks up along the sub-parallel Beni escarpment. Rock uplift on the CANP and probably the CFP is associated with the formation of the rugged topography of the Chapare escarpment (an ∼3–4 km topographic step, Figs. 2A, 3A, 4B, and 5) where rapid exhumation commenced by ca. 8 Ma (Barnes et al., 2012). This recent activity is also recorded in the surface uplift and long-wavelength warping (∼3.5 km of surface uplift over a 250 km distance) of preserved remnants of a low-relief landscape that now defines the Cochabamba-Tipuani ramp (Figs. 3 and 4), suggesting a preceding period of tectonic quiescence and erosional beveling. Moreover, the geometry of the Cochabamba-Tipuani ramp and the relationship between the perched low-relief landscape patches near Tipuani and the 11–7 Ma Cangalli Formation suggests that the erosional beveling of the EC and IA in northern Bolivia was concordant with the contemporaneous 12–9 Ma San Juan del Oro erosion surface of southern Bolivia. This implies a common erosional baselevel—clearly the foreland for the Cangalli Formation (Fornari et al., 1987; Mosolf et al., 2011)—an observation that strongly supports previous interpretations of the San Juan del Oro surface and its paleo-baselevel (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006).

Much of the modern relief on the Beni escarpment likewise appears to post-date this phase of erosional beveling, which may be recorded in the 25–11 Ma hiatus in the cooling history of the Cordillera Real (Gillis et al., 2006). Consistent with this interpretation is the lack of correspondence between the modern physiography and the tectonostratigraphic architecture expressed in the boundaries of the EC and the IA in northern Bolivia; whereas the EC/IA contact as typically shown on regional maps (10–25 km SW of the CANP) (e.g., McQuarrie et al., 2008a; Insel et al., 2010; Barnes et al., 2012) has no topographic expression, the prominent Beni escarpment (BE) is an ∼3 km high topographic step entirely within the EC (indeed entirely within the outcrop belt of the Ordovician) the foot of which does not generally correspond to any mapped structure with regional significance (Fig. 2B; Guarachi et al., 2001; McQuarrie et al., 2008a). Both the 30 km-wide high concavity zones that characterize river profiles draining the Beni Escarpment (Fig. 3; Schlunegger et al., 2011; Gasparini and Whipple, 2014) and the mapped distribution of the Cangalli Formation (Figs. 3 and 4C) imply a 30-km-wide zone of increasing rock uplift rate, consistent with a broad shear zone or a crustal-scale monoclinal fold. It appears plausible, even likely, that much of the EC and IA of northern Bolivia (in the hanging wall of the CANP), but likely excepting the core of the Cordillera Real, were beveled to low relief by ca. 11 Ma and simply warped, uplifted, and incised by deep canyons during Late Miocene to recent deformation.

Our speculative interpretation of the post–12 Ma tectonic architecture of the northern Bolivian Andes is illustrated in Figure 7. The 3-D perspective image of the topography (highlighting low-relief erosion surface remnants and the spatial pattern of channel steepness indices) clearly highlights the Cochabamba-Tipuani ramp and suggests that it may represent a large-scale relay ramp transferring deformation from the CFP and CANP to the structure responsible for the Beni escarpment (Fig. 7A)—an hypothesis worthy of further exploration. In addition, the 3-D perspective image of the geologic map draped over modern topography visually reinforces the interpretation that whereas the erosionally beveled EC and IA has been simply uplifted to 3–4 km elevation in southern Bolivia (Fig. 1), in northern Bolivia it appears to have been uplifted on thrust faults in the Chapare escarpment, warped along the Cochabamba-Tipuani ramp, and warped and uplifted across the Beni escarpment (Figs. 4C and 7B). A schematic cartoon of inferred structural configuration is shown in Figure 7C. The presence of such a large-scale relay ramp appears to be broadly consistent with available geologic maps and interpretive cross-sections, though the implied along-strike differences in uplift history have not been evaluated with detailed structural data (e.g., Baby et al., 1996, 1997; Kley, 1999; Guarachi et al., 2001; McQuarrie et al., 2008a; Eichelberger et al., 2013).

7. DISCUSSION/IMPLICATIONS

Early workers concluded that the Eastern Cordillera (EC) of southern Bolivia experienced significant (>2 km) surface uplift since the Mid to Late Miocene on the grounds of geomorphic (Gubbels et al., 1993; Kennan et al., 1997) and paleobotanical evidence (Gregory-Wodzicki et al., 1998; Gregory-Wodzicki, 2000; Graham et al., 2001). Large potential uncertainties on these estimates, however, limited their acceptance in the literature (e.g., Barnes and Ehlers, 2009). Barke and Lamb (2006) improved quantification of the geomorphic evidence for post–12 Ma surface uplift but reported a very cautious overall estimate of 1.7 ± 0.7 km surface uplift that encompasses all their varied estimates, including low and inconsistent estimates based on projections of profiles of modern rivers that are deeply incised into the San Juan del Oro surface. This overall estimate has been carried forward in the literature citing Barke and Lamb (2006) though we find their most reliable methods yield a composite 2.4 ± 0.4 km post–ca. 12 Ma surface uplift, consistent with both earlier geomorphic and paleobotanical work and our own analysis. Both the preservation of a chain of perched low-relief landscape remnants along the Cochabamba-Tipuani ramp and the projections of profiles of rivers not deeply incised into these surfaces (Fig. 5) provide more direct geomorphic evidence for ∼3 km post–ca. 12 Ma surface uplift. Moreover, the relation between these low-relief landscape remnants and the 11–7 Ma Cangalli Formation strongly supports the interpretation of earlier workers that the 12–9 Ma San Juan del Oro erosion surface was formed at low elevation, on grade with the foreland. Further, the Cochabamba-Tipuani ramp documents how differences in the pattern of rock uplift experienced on either side of the Santa Cruz bend at ∼18°S was accommodated (Fig. 7C).

A swath of perched low-relief surfaces preserved along the flank of the Cordillera Real at the crest of the Beni escarpment at ∼4 km elevation similarly suggest ∼3 km of surface uplift (Fig. 3A and ΔZmax in Fig. 4B). This interpretation is subject to much greater uncertainty, however, because these surfaces have been etched by glaciers and plausibly merely reflect the action of the “glacial buzzsaw” (Brozovic et al., 1997). However, these surfaces lack the glacial horns and dramatic headwalls characteristic of glacial carving of actively uplifting ranges (e.g., Brozovic et al., 1997; Brocklehurst and Whipple, 2007) and may reflect erosional beveling during the hiatus in cooling 25–11 Ma recorded in the Cordillera Real (Gillis et al., 2006). We note that late Miocene–Pliocene exhumation along the Beni escarpment is only recorded in the deep canyons incised below these low-relief benches, and not in high-elevation samples on the benches themselves (Figs. 2 and 3). In addition, glacial erosion has been limited on the concordant low-relief erosional surface (erosionally beveled between 15 and 4 Ma) that defines the plateau margin ∼200 km to the NW in southern Peru as demonstrated by preservation of thin Tertiary volcanic deposits (Lease and Ehlers, 2013). Moreover, interpretation of these low-relief benches as glacially etched remnants of a Miocene low-relief landscape is consistent with evidence that the Beni escarpment is a monoclinal fold over a deeper structure (Figs. 4C and 7).

The complex topography of the northern Bolivian Andes thus appears to be primarily controlled by tectonics—the pattern and amount of topographic relief is dictated by uplifted low-relief erosion surfaces (e.g., Fig. 7A). As demonstrated by Bookhagen and Strecker (2008) the dramatic climatic difference between the northern and southern Bolivian Andes is a consequence of two scales of topographic control on modern rainfall: (1) the west-east promontory of high topography that defines the northern limb of the Santa Cruz bend intercepts south-directed moisture transport, and (2) rainfall is greatly enhanced by orographic effects in areas of high relief (Fig. 2). Thus in the Bolivian Andes there is clear evidence for tectonic control of topography, and topographic control of local climate. Further, the patterns of tectonic uplift appear to be strongly influenced by the paleogeography of Paleozoic depositional basins, as has been previously suggested (Sheffels, 1995; Baby et al., 1996; Kley et al., 1999; Giraudo and Limachi, 2001; Oncken et al., 2006). First, the east-west promontory of the Santa Cruz bend appears to follow a paleogeographic boundary that dictates a change in depth to basement and, consequently, structural style (dashed white line, Fig. 2B). Second, the segments of the Main Frontal Thrust (CFP) and Main Andean Thrust (CANP) that appear to have accommodated significant rock uplift in the Chapare escarpment generally correspond with the zone where the Chapare basement high impinges on the thrust front. Consequently, the Cochabamba-Tipuani ramp plausibly reflects the impingement of the Chapare basement high (Fig. 2).

Although most models predict enhanced erosion where orographic rainfall is concentrated, this anticipated effect is not evident in cosmogenic erosion rates (Safran et al., 2005; Insel et al., 2010), sediment yield (Guyot et al., 1990, 1993; Aalto et al., 2006; Barnes and Pelletier, 2006), or in the distribution of AFT ages (Fig. 2 and 3, discussed below). More important, we see no indication of a tectonic response to the spatial distribution of modern rainfall in northern Bolivia; although a tectonic response is permissible in the Chapare escarpment and possibly the Cordillera Real, we do not know when the modern rainfall pattern was established and the extensive preservation of Miocene low-relief surfaces appears to rule out sufficient erosional exhumation to trigger a significant tectonic response.

Our interpretation is thus at odds with the Barnes et al. (2012) conclusion that since ca. 15–11 Ma climate, rather than tectonics, has been the principle driver of erosion and exhumation in the northern Bolivian Andes—a divergence that merits further scrutiny. In essence Barnes et al. (2012) demonstrate that (1) prior to ca. 15 Ma exhumation is closely tied to the spatio-temporal patterns of contractile deformation, and (2) a second phase of rapid exhumation begins by ca. 15–11 Ma in all tectonostratigraphic domains (the EC, IA, and SA) despite a lack of shortening within the EC and IA over this interval. Citing the general correspondence between the locus of intense modern rainfall and young AFT ages (Fig. 2C), Barnes et al. (2012) make the logical inference that climate is driving this young phase of exhumation, and therefore that the higher rainfall in northern Bolivia is responsible for the younger AFT ages there. We disagree with three facets of this logical argument. First, although they interpret “a second rapid exhumation phase…spanning the entire EC to SA,” the data specifically documents not a truly range-wide behavior, but rather that rapid Late Miocene to recent exhumation occurs in parts of all tectonostratigraphic domains. This second phase of rapid exhumation is restricted to the fold and thrust belt east of the Main Andean Thrust (CANP) and the high relief areas of the Beni and Chapare escarpments (Figs. 2 and 3)—only discrete parts of the EC and IA are involved. For instance, all AFT ages in the IA (as mapped by Barnes et al., 2012) west of the CANP record Oligocene exhumation and appear to stratigraphically underlie low-relief surfaces associated with the 11–7 Ma Cangalli Formation (Fig. 3B, inset). Second, although modern rainfall is naturally concentrated in areas of high relief, Late Miocene–Pliocene AFT ages are restricted to areas of high relief but not necessarily to areas of high modern rainfall, as evidenced by young ages in the relatively dry but deeply incised La Paz river valley just south of La Paz (Figs. 2 and 3). Third, an absence of shortening is not equivalent to an absence of deformation and rock uplift. As we have documented, there is considerable evidence for post–12 Ma uplift and warping of parts of the EC and IA in northern Bolivia that shows a strong correspondence to the distribution of Late Miocene-Pliocene AFT ages (Figs. 2 and 3). Late Miocene exhumation in the EC is restricted to deep canyons and glaciated areas in the Cordillera Real and along the Beni escarpment, which we interpret to be associated with tectonic rock uplift and relief production, just not shortening. For these reasons—and because channels exhibit high steepness values in zones of high modern rainfall rates, which requires active or recently active rock uplift—we conclude that tectonics, and not climate, dictates the pattern of rapid exhumation in the Bolivian Andes.

Shortening post–ca. 12 Ma in the Bolivian Andes appears to be restricted to the Subandean fold and thrust belt and has been attributed to the onset of underthrusting of the Brazilian Shield (e.g., Isacks, 1988; Allmendinger and Gubbels, 1996; Allmendinger et al., 1997; Oncken et al., 2006). How is this change in behavior related to the deep-seated warping and monoclinal folding apparent in the topography of the northern Bolivian Andes? How is that deformation and rock uplift related to the uplift of the Central Andean Plateau (CAP) and deformation on its western flank (WAE) and do these relationships place important constraints on the driving mechanisms of plateau uplift? As documented elsewhere (e.g., Barnes and Ehlers, 2009), the magnitude and timing of surface uplift in the northern Bolivian Andes is consistent with paleobotanical and isotopic constraints on plateau uplift (Gregory-Wodzicki, 2000; Graham et al., 2001; Garzione et al., 2006; Ghosh et al., 2006; Garzione et al., 2008). In addition, most geomorphic and geologic constraints on uplift, deformation, and canyon incision suggest a similar history along the western flank of the Central Andean Plateau (e.g., Wörner et al., 2002; Victor et al., 2004; Hoke et al., 2007; Schildgen et al., 2007; Thouret et al., 2007; Schildgen et al., 2009a, 2009b; Jordan et al., 2010). Moreover, these studies have demonstrated that much of the modern relief on the WAE reflects monoclinal warping (Isacks, 1988), as we have tentatively interpreted for the Beni escarpment. Interestingly, the Beni escarpment has the same macro morphology (same regional slope, height and width) as the WAE—there is a surprising symmetry to the Central Andean Plateau north of the Santa Cruz bend, with the low-taper Subandean fold and thrust belt related to underthrusting of the Brazilian shield as the only clear difference (Fig. 1). We argue that this symmetry reinforces the interpretation that the Beni escarpment is indeed a monoclinal warp and speculate that this records a flexural response of the upper crust to mantle-driven plateau uplift (e.g., Garzione et al., 2006). However, we can only conclude with certainty that the deformation in the Beni escarpment and Cochabamba-Tipuani ramp reflect deep structures—lower crustal flow, décollement ramps, and emplacement of duplexes or basement thrust sheets are also viable explanations for the observed patterns of rock uplift in the northern Bolivian Andes and merit further exploration (e.g., Allmendinger and Gubbels, 1996; Kley et al., 1999; McQuarrie, 2002; McQuarrie et al., 2008a; Ouimet and Cook, 2010).

8. LIMITATIONS AND FURTHER RESEARCH NEEDS

The structural interpretation illustrated in Figure 7 is based on interpretation of uplift patterns on largely morphologic grounds and as such is speculative, presented here as a hypothesis that merits further exploration. While analogous in form to commonly observed structures, relay ramps at this scale are unknown to us in other localities. Further work could usefully address whether there are other structural configurations equally compatible with the topography and erosion rate patterns. The notion that the Cochabamba-Tipuani ramp represents broad-scale warping of a formerly continuous low-relief surface is critical to our structural interpretation, and depends on interpolation across an ∼75 km wide gap in surface preservation due to erosion by Rio La Paz and its tributaries (Figs. 3 and 4A). Although corroborated by our hanging-valley analysis, direct, definitive evidence that the low-relief patches are remnants of a former low-relief landscape with a common baselevel is lacking. Geologic mapping to determine whether correlative late Miocene deposits are preserved on the low-relief patches at the SE end of the Cochabamba-Tipuani ramp would be valuable. Isotopic or paleobotanical studies of the paleo-altitude of the Cangalli formation paired with similar data on contemporaneous deposits at the SE end of the ramp (such as the rocks hosting the Pislepampa flora), or on the San Juan del Oro erosion surface, appear to offer a singular opportunity to settle this question. We note, however, that our interpretation that topographic relief, erosion and exhumation rates, and patterns of modern rainfall are dominantly controlled by tectonic processes in this landscape is in no way contingent on the validity of our speculative structural interpretation of the Cochabamba-Tipuani ramp. This fundamental conclusion is, as we have shown, supported by all available data quite independent of the question of how deformation is transferred from the CANP to the Beni escarpment. This finding is elaborated in more detail by Gasparini and Whipple (2014) for the landscapes and associated erosion patterns of the Beni escarpment. Additional thermochronologic data in the lower part of the Beni escarpment—both in and between major canyons—would help resolve deformation and exhumation patterns. Finally, detailed mapping of the Cangalli formation where it on-laps the Beni escarpment (shown in Figs. 2, 3A, and 4C based on available geologic maps) holds promise to definitively determine the amount and pattern of post-Cangalli (post-7–11 Ma) deformation at the foot of the Beni escarpment.

9. CONCLUSIONS

We conclude that there is strong evidence for ∼3 km surface uplift post–12 Ma of the Eastern Cordillera (EC) in the Bolivian Andes, corroborating previous interpretations of geomorphic (Gubbels et al., 1993; Kennan et al., 1997; Barke and Lamb, 2006) and paleobotanical (Gregory-Wodzicki et al., 1998; Gregory-Wodzicki, 2000; Graham et al., 2001) evidence locally and consistent with evidence elsewhere in the Central Andean Plateau (CAP) and along its western Flank (e.g., Wörner et al., 2002; Victor et al., 2004; Garzione et al., 2006; Ghosh et al., 2006; Hoke et al., 2007; Schildgen et al., 2007; Thouret et al., 2007; Schildgen et al., 2009a, 2009b; Jordan et al., 2010). Long-wavelength surface warping suggests deformation on deep structures and is consistent with expectations for flexural deformation in response to mantle-driven uplift of the CAP (e.g., Garzione et al., 2006), but non-uniquely so. The pattern of rock uplift in the Chapare escarpment just north of the Santa Cruz bend is consistent with previous suggestions that the paleogeography of Paleozoic basins (essentially the impingement of the Chapare basement high) has importantly influenced the structural evolution of the Bolivian Andes (e.g., Sheffels, 1995; Baby et al., 1996; Kley et al., 1999).

Interestingly, the marked differences in the pattern and history of deformation and uplift across the Santa Cruz bend that is evident in the topography (Figs. 1 and 2) appears to have been accommodated by a large-scale relay ramp (the Cochabamba-Tipuani ramp; Figs. 3A, 4, and 7) that transfers deformation from the Chapare escarpment to Beni escarpment to the NW. Further, it appears that erosion and exhumation patterns in northern Bolivia since 12 Ma are controlled primarily by uplift and relief production in the Chapare and Beni escarpments. Precipitation is greatly enhanced by orographic mechanisms along much of the rugged terrain associated with these escarpments and while conventional wisdom dictates that erosional efficiency should therefore be enhanced (e.g., Willett, 1999; Whipple, 2004; Bookhagen and Strecker, 2012), erosion rates remain limited by rock uplift rates. Sustained enhanced erosion rates only result if the rock uplift rate is amplified through feedback mechanisms (e.g., Whipple, 2009). Although it cannot be altogether ruled out, there is no direct evidence of such mechanisms at work in northern Bolivia and the extensive preservation of perched low-relief landscape patches suggests a significant tectonic response is unlikely. All data available at present are consistent with the simpler, if less tantalizing, interpretation that inherited geologic structure and deep-seated processes are dictating the pattern of rock uplift which sets large scale landscape morphology, which in turn controls rainfall patterns (e.g., Bookhagen and Strecker, 2008). The degree to which the distribution of modern rainfall in the northern Bolivian Andes influences the pattern of erosion rates remains unclear (e.g., Safran et al., 2005; Aalto et al., 2006; Insel et al., 2010; Gasparini and Whipple, 2014) but no amplifying tectonic response has yet been recognized.

We thank E. Kirby, K. Hodges, B. Brooks, M. Bevis, R. Arrowsmith, and B. Horton for helpful discussions at various stages in our work on this paper. L. Safran, J. Barnes, N. McQuarrie, and S. Nesbitt kindly shared extensive GIS data sets. R. DiBiase and M. Rossi provided helpful commentary on an early draft of the manuscript. Discussions with T. Ehlers and J. Barnes and review comments from B. Bookhagen improved the manuscript.