Mio-Pliocene geology of the Southern Puna Plateau Margin, Argentina

The Mio-Pliocene geologic record is investigated along the southeastern margin of the Puna plateau in northwestern Argentina. The Puna plateau is the southernmost extension of the high elevation, internally drained, Central Andean plateau. A series of intermontane basins at the plateau margin preserve thick stratigraphic sections spanning ~10-3 Ma. The strata in these basins were investigated between 25°30?S and 28°S latitude, with geochronological and paleoenvironmental objectives. The stratigraphy, composition, and age of volcanic ash beds provide age control. Fossil vertebrates and soils interbedded with these volcanic ash beds provide material for stable isotopic proxies of environment. This approach permits analysis of spatial and temporal patterns in the tectonic and climatic evolution of the landscape at the Puna margin. Stable isotope analysis of fossil tooth enamel from the 2.5 km thick section at Puerta de Corral Quemado documents the late Miocene expansion of plants using the C4 photosynthetic pathway. Tooth enamel was analyzed by conventional methods, and by laser ablation which incorporates small bodied taxa in the study. These results indicate the presence of C4 plants in the region by at least 8.5 Ma and a shift to C4 diets among fossil notoungulates between 7-5.5 Ma. Fossil rodents exhibit a less dramatic diet change across this interval, and all taxa document transient enrichment in 13C and 18O approximately coincident with the Miocene-Pliocene boundary. Interbasin correlation of ash beds demonstrate that conglomeratic deposits are conformable with stratified sections and initiated between 4-3 Ma. These deposits lag uplift of basin margin mountain blocks by several million years and precede contractional deformation of Mio-Pliocene strata. This sequence is characteristic, but diachronous between basins. This Mio-Pliocene pulse of deformation typifies the evolution of a broken foreland at the plateau margin. Interbasin comparison of isotopic proxy data from fossil soils identifies regional 18O enrichment concurrent with topographic growth at the plateau margin. Carbon isotope data from fossil soils demonstrate strong ecological gradients at the landscape-scale and at local scales. The record of C4 expansion in fossil soils is significantly influenced by the growth of complex topography and possibly by regional reorganization of precipitation systems.

MIO-PLIOCENE GEOLOGY OF THE SOUTHERN PUNA PLATEAU MARGIN, ARGENTINA by Scott Anthony Hynek A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Geology Department of Geology and Geophysics The University of Utah August 2011 Copyright © Scott Anthony Hynek 2011 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Scott Anthony Hynek has been approved by the following supervisory committee members: Francis H. Brown , Chair April 6, 2011 Date Approved Ronald L. Bruhn , Member April 11, 2011 Date Approved Thure E. Cerling , Member April 7, 2011 Date Approved Barbara P. Nash , Member April 11, 2011 Date Approved W. James Steenburgh , Member April 6, 2011 Date Approved and by D. Kip Solomon , Chair of the Department of Geology and Geophysics and by Charles A. Wight, Dean of The Graduate School. ABSTRACT The Mio-Pliocene geologic record is investigated along the southeastern margin of the Puna plateau in northwestern Argentina. The Puna plateau is the southernmost extension of the high elevation, internally drained, Central Andean plateau. A series of intermontane basins at the plateau margin preserve thick stratigraphic sections spanning ~10-3 Ma. The strata in these basins were investigated between 25°30′S and 28°S latitude, with geochronological and paleoenvironmental objectives. The stratigraphy, composition, and age of volcanic ash beds provide age control. Fossil vertebrates and soils interbedded with these volcanic ash beds provide material for stable isotopic proxies of environment. This approach permits analysis of spatial and temporal patterns in the tectonic and climatic evolution of the landscape at the Puna margin. Stable isotope analysis of fossil tooth enamel from the 2.5 km thick section at Puerta de Corral Quemado documents the late Miocene expansion of plants using the C4 photosynthetic pathway. Tooth enamel was analyzed by conventional methods, and by laser ablation which incorporates small bodied taxa in the study. These results indicate the presence of C4 plants in the region by at least 8.5 Ma and a shift to C4 diets among fossil notoungulates between 7-5.5 Ma. Fossil rodents exhibit a less dramatic diet change across this interval, and all taxa document transient enrichment in 13C and 18O approximately coincident with the Miocene-Pliocene boundary. Interbasin correlation of ash beds demonstrate that conglomeratic deposits are conformable with stratified sections and initiated between 4-3 Ma. These deposits lag uplift of basin margin mountain blocks by several million years and precede contractional deformation of Mio-Pliocene strata. This sequence is characteristic, but diachronous between basins. This Mio-Pliocene pulse of deformation typifies the evolution of a broken foreland at the plateau margin. Interbasin comparison of isotopic proxy data from fossil soils identifies regional 18O enrichment concurrent with topographic growth at the plateau margin. Carbon isotope data from fossil soils demonstrate strong ecological gradients at the landscape-scale and at local scales. The record of C4 expansion in fossil soils is significantly influenced by the growth of complex topography and possibly by regional reorganization of precipitation systems. iv TABLE OF CONTENTS ABSTRACT……………………………………………………………………………...iii LIST OF TABLES……………………………………………………………………….vii ACKNOWLEDGMENTS………………………………………………………………viii CHAPTER 1 INTRODUCTION: TECTONIC AND GEOLOGIC SETTING OF THE PUNA PLATEAU MARGIN, NORTHWESTERN ARGENTINA………………….1 Purpose of the dissertation…………………………………………………………….1 Tectonic setting….…………………………………………………………………….2 What is the Puna plateau?……………………………………………………………..5 How did the Puna plateau form?..................................................................................11 Geologic record of the southeastern Puna plateau margin……………………….......18 Structure of the dissertation………….……………………………………................36 2 STABLE ISOTOPE ECOLOGY ACROSS THE MIOCENE-PLIOCENE BOUNDARY; PUERTA DE CORRAL QUEMADO………………………………39 Introduction………………………………………………………………………......39 Materials and methods………………………………………….……………………42 Stratigraphic framework….………………………………………………………….45 Stable isotope results………..……………………………………………………......52 Discussion of stable isotope results……..……………………………………….......64 Conclusions………………………………………………………………………......83 3 MIO-PLIOCENE SEDIMENTATION AND DEFORMATION CONSTRAINED BY IDENTIFICATION OF WIDESPREAD VOLCANIC ASH BEDS……………85 Introduction…………………………………………………………………………..85 Scientific approach, material studied, and analytical methods………………………88 Major volcanic ash beds of the southern Puna plateau………………………………91 Proposed tephrostratigraphic correlations……...…………………………………...119 Accuracy and precision of the chronological framework…...……………………...130 Regional lithostratigraphic framework…..…………………………………………135 Deposition of conglomeratic units…….……………….…………………………...141 Structural and stratigraphic relationships in the Corral Quemado basin…………...144 Structural and stratigraphic relationships in the Fiambalá basin…………………...147 Timing, style, and locus of plateau-margin deformation…………………………...150 4 PALEOSOL CARBONATE RECORDS OF CLIMATE AND ENVIRONMENT: REGIONAL SYNTHESIS…………………………………….152 Introduction…………………………………………………………………………152 Material sampled and methods employed…………………………………………..155 Fiambalá basin………………………………………………………………….......156 Puerta de Corral Quemado……………………………………………………….....159 Vallé Santa Maria…………………………………………………………………..160 Mio-Pliocene 13C enrichments……………………………………………………...171 Mio-Pliocene 18O enrichment………………………………………………………177 Mio-Pliocene ecological gradients in northwestern Argentina…………………......181 Conclusions…………………………………………………………………………187 5 DISCUSSION AND CONCLUSIONS..…………………………………………...189 Paleoenvironmental observations………………………………………………......189 Observations of sedimentation and deformation at the plateau margin…………….190 The stratigraphic record of volcanism (observations)…..……………………….....190 Tectonics, climate, and the origin of the Puna plateau……………………………..192 Conclusions…………………………………………………………………………202 Outstanding questions………………………………………………………………203 APPENDIX A STRATIGRAPHIC AND GEOCHEMICAL DATA FOR VOLCANIC ASH BEDS…………………………………………………….206 B STABLE ISOTOPE DATA………………………………………………………..277 REFERENCES…………………………………………………………………………301 vi LIST OF TABLES TABLE 2.1 Stratigraphic framework at Puerta de Corral Quemado…………………………….48 2.2 Summary of stable isotope data grouped by stratigraphic interval and data filtering method…………………………………………………………….53 3.1 EPMA data for latest Miocene sequence at Puerta de Corral Quemado…………...96 3.2 EPMA data for latest Miocene sequence at Río Guanchin…………………………99 3.3 EPMA data for tuff of sandstone #12 sequence at Puerta de Corral Quemado…...102 3.4 EPMA data for the 4.0 Ma sequence at Fiambalá………………………………...107 3.5 EPMA data for the Toba Corral Quemado sequence (<3.8 Ma)………………….112 3.6 ICP-MS data for volcanic glass in widespread ash beds……………………….....122 4.1 Stable isotope data for pedogenic carbonates in the Corral Quemado basin……...161 4.2 Stable isotope data comparative with that of Kleinert and Strecker (2001)………163 4.3 EPMA data for Entre Rios section #1 ash beds and correlative samples…………167 4.4 ICP-MS data for volcanic glass in correlated and dated ash beds in VSM……….169 ACKNOWLEDGMENTS This research was initiated with support from the Packard Foundation, and the University of Utah, Vice President of Research. In 2002, field assistance was provided by Graciela Esteban, Rafael Herbst, Claudia Muruaga, Norma Nasif (Universidad Nacional de Tucumán), Gustavo Gómez and Pamela Steffan (Universidad Nacional del Centro), and is gratefully acknowledged. Thure Cerling, Ben Passey, Pepe Prado, and Jay Quade were instrumental in the early stages of the project. In 2007, exceptional field assistance was provided by Lars Battle. In the laboratory numerous people provided advice and assistance. Most notably: Frank Brown taught me to process volcanic ash samples efficiently and effectively, Barb Nash expertly maintains the Cameca SX50 electron microprobe and works to provide the highest quality data, and Diego Fernandez spent considerable effort developing a method for the digestion and trace element analysis of volcanic glass separates by ICP-MS. My committee is thanked for their support, flexibility, and collective wisdom. Frank Brown must be singled out as a patient, hardworking and thoughtful advisor and as a collaborator of immense value. Jim Steenburgh provided an opportunity to get involved with mountain meteorology. Senior faculty at the University of Utah are thanked for friendship and mentoring, especially Thure Cerling, Dave Chapman, Jim Ehleringer, and Bill Parry. My parents, my wife, and my friends all contributed to keeping me on the straight and narrow path forward. CHAPTER 1 INTRODUCTION: TECTONIC AND GEOLOGIC SETTING OF THE PUNA PLATEAU MARGIN, NORTHWESTERN ARGENTINA Purpose of the dissertation The data collected for this dissertation, and the themes discussed herein relate to the late Miocene and Pliocene geologic history at the southeasternmost margin of Central Andean plateau. In northwestern Argentina, modern interactions between climate and topography are striking features of the landscape and profoundly influence geological processes. The evolution of this system before, during, and after a strong Pliocene pulse of uplift in the region is investigated. Several lines of inquiry are followed. A stable isotope proxy approach to environmental change during this period addresses questions of landscape scale ecology and its relationship with growing topography and global ecological change. Stratigraphic study of volcanic ash beds provides chronological control for stable isotope data and permits analysis of the locus and timing of plateau margin uplift. Integration of volcanic ash and stable isotope stratigraphies is a means of addressing local and regional variations in environmental change. A composite record of climate and ecology in the region permits analysis of the relative importance of global and regional events on the evolution of Central Andean plateau and its environment. This chapter provides an introduction to the region of study, formulates the questions addressed, and describes the methods employed. A general tectonic framework for the Andes is followed by a description of the study area and a summary of hypothesized mechanisms for plateau uplift. The methods employed are then discussed, highlighting the assumptions and realistic limits of data interpretation. Having established the scientific approach, the organization of the dissertation is presented. Tectonic setting The Andes mountains span ~7500 km along the western margin of the South American continent (Figure 1.1). In places, the Andes are a single narrow spine; in others, a broad amalgamation of ranges, with each range having its own structure and morphology. It has been hypothesized that at the continental scale, the Andes owe their morphological variation to climatic patterns of zonal atmospheric circulation (Montgomery et al., 2001). The focus of this dissertation is the southern terminus of the broad, high elevation Puna-Altiplano plateau which is well developed in the subtropical belt of deserts (~15-33°S). As the type example of subduction related continental mountain building, cordilleran fold-thrust belt tectonics, and volcanic arc processes, the Andes remain atypical in many respects. Subduction zones are common geologic features known from many locations on Earth, but only along the western margin of South America does subduction of oceanic lithosphere beneath a continent result in topography and crustal thickness comparable to the Himalaya. The Central Andes in particular, have a variable but thick (up to 75 km) continental crust underlying the plateau portion of the mountain belt (Beck et al., 1996; Yuan et al., 2002). 2 Figure 1.1: Shaded relief map of South America from GTOPO30 data. The Central Andean plateau, highlighted in the box, corresponds to the area of Figure 1.3. The approximate location of subduction is indicated by the trench. Kilometers of sediment-fill thickness (from Bangs and Cande, 1997) are denoted along the length of the trench. 3 Except at the northernmost and southernmost ends, the subduction zone is comprised of the oceanic Nazca plate moving to the northeast at 3-4 cm/year and the overriding South American plate moving to the west at comparable rates (Marret and Strecker, 2000). Sediment supply to the oceanic trench demarcating the Nazca-South America plate boundary varies greatly along strike (Bangs and Cande, 1997). This results in dramatic differences in sediment-fill thickness and potentially important differences in the coupling between the two plates. Lamb and Davis (2003) have proposed that arid climates reduce sediment supply to the trench resulting in stronger plate coupling, more efficient transfer of shear stress to the mountain belt, and ultimately enhanced uplift. The geometry of the subducting Nazca plate beneath South America changes along strike. Some segments of the Nazca plate dip eastward at ~30°; whereas other segments dip much more shallowly. Among other factors, the flat-subducting sections of the Nazca plate have been correlated to aseismic ridges, seamount chains, and associated seafloor topography (Pilger, 1981; von Huene and Ranero, 2009). Regardless of cause, flat-subducting segments are associated with notable changes in continental tectonics (Jordan et al., 1983). Cessation of arc volcanism and changes in the style of continental deformation are spatially related to flat-subducting segments. The study area of this thesis, ~26.5-28°S, overlies a gradual transition zone in the dip of the subducting Nazca plate. To the north of the study area lies the Altiplano, bounded on the east by a thin-skinned fold-thrust belt and underlain by a slab subducting at a dip of 30° (Figure 1.2a). To the south, inland topography is dominated by the thick-skinned deformation of the Sierras Pampeanas and a flat-subducting slab is present (Figure 1.2b). 4 Figure 1.2: Topographic profiles and structural cross sections, reprinted with permission from Jordan et al. (1983). What is the Puna plateau? The Puna is a high volcanically active plateau in northwestern Argentina whose topography is characterized by basins and ranges (Figure 1.3). Volcanism on the Central Andean plateau is extensive and diverse, being dominated on the west by stratovolcanoes of the frontal magmatic arc. Young back-arc mafic rocks are widely distributed throughout the plateau (Kay et al., 1994). Both large and small silicic calderas are also present (de Silva and Francis, 1991). One of the world's largest silicic calderas, Cerro Galán, is an important feature of the southern Puna plateau that was first recognized with spacecraft imagery (Francis et al., 1978, 1983, 1989). An accurate geochronology of the most recent explosive silicic volcanism at Cerro Galan and associated mafic flows is just emerging (Kay et al., in press; Risse et al., 2008). The youngest silicic ignimbrite from this volcanic center is very precisely dated at 2.06 Ma (Hynek et al., 2011). 5 Figure 1.3: Tectonic and climatic setting of the study area reprinted with permission from Strecker et al. (2007a); a) shaded relief map and principal morphotectonic provinces (after Jordan et al., 1983), b) Mean annual rainfall distribution from the Tropical Rainfall Measurement Mission (TRMM) satellite, calibration of rainfall amounts described in Strecker et al. (2007a). Transects highlighted in a) correspond to those presented in Figure 1.2 of this dissertation, and boxes shown in a) denote the locations depicted in the following two figures of this chapter. The basins and ranges of the Puna create high relief which is in marked contrast to the Altiplano of Bolivia. The basins commonly contain volcanic, clastic, and evaporite deposits 3-5 km thick (Alonso et al., 1991). Internally drained, evaporite depositional centers have existed in the region since ~15 Ma (Vandervoort et al., 1995). This dissertation focuses on the geologic record contained in structurally similar intermontane basins along the plateau margin, which are currently externally drained. These basins are associated with contractional deformation during the late Miocene and Pliocene, and may represent an early stage of plateau growth (Coughlin et al., 1998). 6 Concentration of precipitation along the eastern flanks of the plateau is documented by climatological data (WMO, 1975), and by more recent Tropical Rainfall Measurement Mission (TRMM) satellite observations (Figure 1.3b; Bookhagen and Strecker, 2008). Most of this precipitation falls during the summer months, and is believed to originate in the Amazon lowland, being advected southward by the South American low-level jet (Strecker et al., 2007a). The localization of precipitation on the eastern flanks of a topographic barrier is especially obvious across the Sierra Aconquija at ~26°S latitude where vegetative cover responds strongly to precipitation (Figure 1.4). Stable isotope proxy data from the leeward side of the Sierra Aconquija (Kleinert and Strecker, 2001) and apatite fission track thermochronology of the Sierra Aconquija bedrock (Sobel and Strecker, 2003) are both believed to record initial uplift at ~6 Ma and the development of this orographic rainshadow by ~3 Ma. This example of topographic control on climatic and ecologic gradients is extreme, yet characteristic of the region. Figure 1.4: Representative vegetation on the windward and leeward side of the Sierra Aconquija; a) subtropical rainforest at ~1,300 m elevation along the east facing (windward) flank of the range, b) semi-arid intermontane basin at ~1,800 m elevation on the west facing (leeward) side of the range. These two sites are separated by less than 50 km along an east to west transect which corresponds to an approximately 3-4 fold decrease in mean annual rainfall. 7 Thus, tropical moisture drawn to the latitude of the southern Puna is effectively blocked from the interior of the plateau by the series of orographic barriers on its eastern margin. Moisture flux to the plateau appears to be funneled through topographic lows and to respond to short term climatic variations (Bookhagen and Strecker, 2008). However, sedimentary and geomorphic records of lake type and lake level indicate that the Puna remained relatively arid during the last glacial cycle compared to the Chilean Atacama Desert or the Bolivian Altiplano (Godfrey et al., 2003; Placzek et al., 2006). A glacial chronology for the region is essentially nonexistent; however, during Pleistocene glaciations moisture flux to the plateau appears to have increased. Glacial landforms and the modern snowline document westward increases in elevation, suggesting that easterly moisture sources have remained important at this latitude throughout the recent past (Haselton et al., 2002). In the plateau interior and along its western margin, evidence for glaciation is lacking for many peaks attaining elevations between 5,500-6,000 m. The general east to west aridity gradient continues to the hyper-arid climate of northern Chile which has existed intermittently for the last 10-15 Ma (Houston and Hartley, 2003). Westerly circulation over the plateau is unlikely to be moisture bearing, but its influence on the plateau is profound nonetheless. Wind sculpted features in ignimbrite sheets of the southern Puna indicate strong northwesterly oriented wind regimes since at least 2 Ma (Greene, 1995). In some localities the wind sculpted ignimbrites are associated with the largest known wind ripples on Earth (Milana, 2009). These extreme geomorphic features have been explored as terrestrial analogs to Martian geomorphic features (de Silva et al., 2010; de Silva, 2010; Milana et al., 2010). This strong westerly circulation is manifested as dune fields in intermontane basins at the 8 plateau margin, further afield as loess deposits, and possibly as significant dust contributions to South Atlantic sediments and East Antarctic ice (Gaiero, 2007). The modern tectonic and geomorphic setting of the southern Puna margin is particularly well exemplified by the Cerro Blanco volcanic complex (Figure 1.5). The Cerro Blanco volcanic complex, is characterized by at least two young collapse calderas and several lava domes (Arnosio et al., 2005). The Cerro Blanco caldera was observed to subside rapidly (2.5 cm/year) during the 1990s and is the only subsiding volcano of more than 900 Central Andean volcanic centers studied (Pritchard and Simons, 2002). Pritchard and Simons (2004) suggest that the subsidence is caused by a cooling magma chamber (9-14 km depth) and associated hydrothermal systems. Recent volcanic activity is attested to by the Campo de la Piedra Pómez ignimbrite, and younger unconsolidated ignimbrites to the south and west of the volcanic center (Arnosio et al., 2008). High rates of eolian erosion on the Puna are documented by the Campo de la Piedra Pómez ignimbrite (Figure 1.5b), which is dated at 44.1 ± 2.2 ka by 40Ar/39Ar analysis of sanidine (Appendix A). This young, moderately welded, ignimbrite is extensively exposed on the southernmost Puna plateau where it is dissected into northwest-southeast oriented yardangs. Erosional relief of 10 m is observed for mature yardangs and an estimated 5 m of thickness has been removed from the ignimbrite since eruption (de Silva et al., 2010). If this erosion is averaged over the 44 ka since emplacement, minimum long-term eolian erosion rates of >10 cm/ka are implied. The influence of northwesterly atmospheric circulation over the Puna plateau and resulting erosion and sediment evacuation is well documented in the Corral Quemado basin (Figure 1.5c, d). Extensive dune fields are present in hanging valleys of the plateau 9 Figure 1.5: Field photographs and ASTER (http://ava/jpl.nasa.gov/) images of the southern Puna plateau margin: a) image depicting the Cerro Blanco volcanic complex (CB), the Campo de la Piedra Pómez ignimbrite (ig), and Quaternary basaltic and basaltic-andesite flows (B), b) photograph of the ignimbrite illustrating wind erosion into elongated swales (yardangs), c) image depicting transport of wind eroded material to the southeast, d) the intermontane Corral Quemado basin at the Puna margin, 1) eolian deposits derived from the Puna, 2) Quaternary range front gravels and terraces disconformably overlying 3) Miocene and Pliocene continental strata. margin and eolian deposits are found throughout the modern basin. Quaternary pediments and fluvial terraces host soils with extremely thick Av horizons, indicating high rates of eolian deposition (cf. McFadden et al., 1987). The modern and recent eolian systems of the region provide an important analog for the ~2.5 km thick Puerta de Corral Quemado stratigraphic section, which records Mio-Pliocene eolian deposition, with extensive dune fields present from ~7-6 Ma (cf. Chapter 2). 10 How did the Puna plateau form? Along the margins of the southern Puna plateau, vigorous Pliocene deformation of the upper crust is documented by deposition of syntectonic conglomerates, as well as folding and faulting of Mio-Pliocene deposits. This surficial expression of tectonic activity is broadly synchronous along the plateau margin throughout the study area encompassed by this dissertation (~28-26.5°S). In contrast to the plateau margins, relatively minor deformation is observed within the plateau after 10 Ma, although abundant evidence exists for surface uplift on the order of 2 km between 10-6 Ma (Hoke and Garzione, 2008). Most data regarding plateau uplift pertain to the Altiplano; substantially less information exists regarding the uplift of the Puna. Therefore much of the following discussion is comparative. The Puna and Altiplano share many similarities, including a dramatic Mio-Pliocene uplift history; however, their paths to the present state differed (cf. Allmendinger et al., 1997). Several processes have been implicated in formation of the Puna-Altiplano plateau. These processes, by no means mutually exclusive, have been put forward as general models to explain the genesis of this enigmatic geologic province. The mechanics of building and supporting a vast high altitude plateau are poorly understood and simple models are useful for conceptualizing relevant processes. Regarding the southern Puna, Allmendinger (1986) summarized potential plateau uplift mechanisms (Figure 1.6). Observations of upper crustal fault geometries in this study, favor a structural model of distributed shortening leading to crustal thickening rather than underthrusting of the stable craton beneath the orogenic belt. 11 Figure 1.6: Cartoon models of continental plateau uplift (reprinted with permission from Allmendinger, 1986.) Comprehensive models for deformation within the Central Andes and uplift of the Puna-Altiplano plateau rely on a combination of crustal thickening and lithospheric thinning (Isacks, 1988). A majority of crustal thickening (70-90%) is attributed to shortening, leaving a minor role for magmatic addition in supporting the expansive topography of the plateau (Allmendinger et al., 1997). Isacks (1988) proposed a two-stage model for crustal thickening, the first stage dominated by distributed shortening and the second characterized by underthrusting of the foreland beneath the plateau by shortening of the ductile lower crust. This two-stage model is attractive because it produces recent uplift of a low relief, internally drained plateau with minor surface deformation of the plateau interior. This two-stage model has been shown to be viable for the Altiplano (Gubbels et al., 1993), but unlikely for the Puna, which exhibits little evidence of underthrusting (Allmendinger and Gubbels, 1996). 12 Recent geophysical data also suggest that the two-stage model, which relies on underthrusting of the craton beneath the plateau, is viable for the Altiplano. Of particular interest is a regionally extensive, rheologically weak layer observed as a thin horizon at ~20 km depth below the Altiplano (Chmielowski et al., 1999; Beck and Zandt, 2002; Oncken et al., 2003). This weak mid-crustal layer may serve to decouple the upper and lower crust, effectively permitting underthrusting from the east and geodynamically isolating the brittle upper crust. Beck and Zandt (2002) indicate that the crust underlying high topography is thick and has a felsic to intermediate bulk composition. They further suggest that the lower crust is ductile, and that underthrusting from the east has penetrated beneath the plateau margin, but does not penetrate beneath the entire Altiplano. It is also suggested that dense sub-crustal lithosphere is delaminating beneath the Altiplano, thereby reducing lithospheric thickness, accommodating inflow of upper and mid-crustal material beneath the plateau and resulting in the "felsification" of continental crust (Beck and Zandt, 2002). In a recent review of evidence for the rapid Mio-Pliocene uplift of the Altiplano, Hoke and Garzione (2008) provide an updated conceptual review of the possible geodynamic processes responsible for plateau formation (Figure 1.7). This analysis was aimed at determining which geophysical models are capable of producing a hypothesized 2.5 km of surface uplift between 10-6 Ma. In this instance, proposed models of plateau uplift are essentially constrained by the rate at which they proceed. Crustal shortening alone appears incapable of producing plateau uplift. As proposed by Hoke and Garzione (2008), the second, underthrusting, stage of Isacks' (1988) model represents a combination of crustal thickening (Figure 1.7a) and mass redistribution by 13 Figure 1.7: Conceptual models for rapid, large magnitude, surface uplift of the Altiplano plateau (reprinted with permission from Hoke and Garzione, 2008). 14 crustal flow (Figure 1.7b). Various modes of mass redistribution by crustal flow have been proposed for the Altiplano (Husson and Sempere, 2003; Hindle et al., 2005), but Hoke and Garzione (2008) prefer to implicate removal of the lower lithosphere (lower crust and upper mantle; Figure 1.7c) in the late Miocene uplift of the Altiplano. In discussing mechanisms of lower lithosphere removal, they disregard ablative subduction (e.g., Pope and Willett, 1998) and focus instead on mechanisms driven by gravitational instability (Kay and Kay, 1993; Molnar and Houseman, 2004). Removal of dense lower lithosphere and influx of hot, buoyant, asthenospheric mantle is equivalent to lithospheric thinning (Figure 1.6). Lithospheric thinning and thermal isostatic effects are smaller than those of lithological density variations, but conform well to geologic observations for the plateau (Froidevaux and Isacks, 1984). Disentangling the various geodynamic models of plateau formation is not an especially tractable problem given the data presented in this dissertation; however, an interesting paradox presents itself. The Altiplano portion of the Central Andean plateau is hypothesized to have uplifted rapidly by isostatic adjustment to removal of dense lower lithosphere, but the strongest evidence for removal of the lower lithosphere is documented in the southern Puna portion of the Central Andean plateau. It has been suggested that magmatic rocks provide the most enduring and unequivocal evidence for delamination (Kay and Kay, 1993). Mafic lavas in particular have been explored as a proxy for lithospheric delamination and influx of fresh asthenospheric mantle. Basaltic rocks of intraplate type are found in a zone overlying the proposed lower lithospheric delamination of the southern Puna. This zone is ringed by 15 mafic rocks of calc-alkaline affinity, and north of approximately 24°S, mafic lavas are shoshonitic (Kay et al., 1994). These high potassium basaltic rocks continue across the Altiplano (Davidson and de Silva, 1992) to its northernmost termination, where mafic rocks of ultra-potassic composition are described (Carlier et al., 2005). It has been argued, that the zone of basaltic and silicic volcanic rocks centered about 26°S is a signal of lithospheric delamination (Drew et al., 2009; Kay, 2010). Approximately 1/3 of the available age estimates for volcanic rocks in the Central Andes are samples from the southern Puna between 25-27°S (cf. Trumbull et al., 2006). These radiometric data provide a first approximation of the most recent volcanic history in the region (Figure 1.8). Several observations relevant to delamination, volcanism, and regional geology over the last 10 Ma can be made: 1) the age distribution of rhyolitic and dacitic rocks associated with the volcanic arc is approximately constant, 2) eruption of basaltic and andesitic rocks in the volcanic arc may document a transient increase between 6-5 Ma, and 3) mafic back-arc volcanism and the Cerro Galán caldera commenced eruption at ~6 Ma. This period also produced intraplate volcanic rocks 300 km west of the main arc, and 170 km west of the plateau margin (Gioncada et al., 2010). While excellent age control exists for volcanic rocks in the southern Puna plateau, geophysical data are sparse by comparison. Existing geophysical data demonstrate that the southern Puna is the region of thinnest crust beneath the plateau, perhaps as little as 40 km thick (Yuan et al., 2002; Tassara et al., 2006; McGlashan et al., 2008). This is consistent with a lower crustal delamination event as an important mechanism for the generation of widespread back-arc mafic volcanism on the plateau since 7 Ma. 16 Figure 1.8: Frequency-age histograms and cumulative percentages plots for volcanic rocks <10 Ma in the study area; a) data from Trumbull et al. (2006), b) data from Risse et al. (2008), Kay et al. (2010 and in press), and Gioncada et al. (2010). The back-arc magmatic pulse documented by mafic lavas between 25-27°S, and by volcanic history at Cerro Galan caldera currently provides the strongest evidence for lower lithospheric delamination of any locality in the region (Kay and Coira, 2009). This is ~750 km removed from the observations indicating rapid surface uplift of the Altiplano, and any mechanistic link between the two is precluded. Geophysical understanding of the Altiplano is significantly greater, as is the direct evidence of plateau 17 uplift; however, the evidence for large scale delamination events beneath the Altiplano is not strong (Beck and Zandt, 2002; Hoke and Garzione, 2008). Conversely, the magmatic understanding of the southern Puna is significantly more advanced. The age data shown in Figure 1.8 provide an opportunity to compare surficial geologic data directly with the proposed tectonic models for plateau formation, by using volcanic rocks as a proxy for tectonic processes. Constraining the timing of geological proxy records for surficial tectonic and climatic events will yield more meaningful comparisons with the magmatic record of the southern Puna and with the paleoenvironmental record of the Altiplano. Developing a high resolution paleoenvironmental record for the southern Puna margin and facilitating its comparison with other data is the primary goal of this dissertation. Geologic record of the southeastern Puna plateau margin This dissertation is guided by the principle that models and theories must not violate geological observations. Extensively exposed and exceptionally thick sedimentary records preserved along the southern and eastern margins of the Puna plateau are investigated. These sedimentary records contain continental deposits commonly in excess of 1 km thick and generally span much of the late Miocene and Pliocene epochs (~10-3 Ma). As discussed, this temporal interval includes the proposed rapid uplift of the Central Andean orogenic plateau (Garzione et al., 2008) and is also a period of global ecologic change (Cerling et al., 1997). Utilizing the geologic record of continental sedimentation, two primary types of information are documented; 1) paleoenvironmental proxy data relating to climate, depositional environment, and the ecology of plants and 18 herbivores, and 2) geochronology of volcanic ash beds intercalated with paleoenvironmental proxy data. Paleoenvironmental proxy data Stable isotope proxies for climate and ecology provide the most widespread, abundant, and high resolution data available for the region. Materials studied include soil carbonate, tooth enamel, snail shell, diagenetic carbonate, and water. The isotopic composition of carbonate was measured in CO2 liberated by phosphoric acid digestion. A subset of tooth enamel samples were analyzed by laser ablation, which, unlike acid digestion,samples both the carbonate and phosphate components of the enamel. Temporal intervals sampled. Three temporal intervals are represented by stable isotope data. Wherever possible, paleosols and fossil tooth enamel were sampled from localities with interstratified volcanic ash beds. In the absence of absolute age estimates, these data still provide a temporal record of climate and ecology based upon the principle of stratigraphic superposition. The extent of these data is over the interval from 9-3 Ma. The second temporal interval investigated is from ~2 Ma-present; these samples are associated with geomorphic features interpreted as relict land surfaces. These features are pediments or fluvial terraces high above modern river level, comprising a veneer of coarse-grained sediments disconformably overlying Mio-Pliocene strata (cf. Figure 1.5d). In most cases the ages of these geomorphic surfaces are poorly known at best, so assembling a precise time-series of stable isotope data is precluded. The third group of samples presents modern isotopic data for a variety of materials. Samples of water, terrestrial gastropods, and tooth enamel can reliably be assumed to represent the modern 19 environment. Soil carbonate on modern land surfaces can not be strictly interpreted as modern, but most likely represents the Holocene (<10 ka). Stable isotopes in soil carbonate. Soil and paleosol carbonate are interpreted in the framework put forth by Cerling (Cerling; 1984; Cerling et al., 1989; Cerling and Quade, 1993). The carbon isotopic composition of soil carbonate is interpreted in terms of the proportion of C3 and C4 vegetation present at a location, recognizing that formation depth of soil carbonate and soil respiration rates are extremely important considerations. A first order assumption is that C4 plant ecosystems are dominated by tropical grasses, and are most successful under conditions of water stress, low elevation, low concentrations of atmospheric CO2, and warm growing season (Ehleringer et al., 1997). The oxygen isotope composition of soil carbonate relates to the isotopic composition of soil water and the temperature of carbonate formation. These confounding variables are further complicated by processes that modify the composition of soil from that of meteoric water (i.e., evaporation). Thus, oxygen isotopic data must be interpreted as a balance between precipitation input and soil water evaporation in which the signal is modified by the temperature of carbonate formation. The isotopic composition of soil water is, at best, difficult to reconstruct and reconstructing the isotopic composition of meteoric water is one step further removed. Spatial scale represented by soil carbonate data. A distinct advantage of isotope proxy data from paleosol carbonates is that the signal is integrated over a relatively small land area, probably 10-100 m2 (Quade et al., 2007). The estimated proportion of C3 to C4 vegetation is a function of the below ground biomass at this scale, and landscape variations in ecosystem composition can be recorded by carbon isotopic 20 data (cf. Behrensmeyer et al., 2007). Input to the soil water oxygen isotope system is at a similar spatial scale, assuming precipitation as the dominant isotopic contribution. Age of soil carbonate on abandoned geomorphic surfaces. Interpreting soil carbonate formed on long-lived geomorphic surfaces presents its own set of challenges. The age of the surface and the age of the carbonate are two different parameters, which do not have a strict relationship. Soil formation continues from the inception of these land surfaces to present, and isotopic data may realistically integrate this whole period, or some unknown portion of it. Given the age of many of theses surfaces, soil carbonate can record 0.1-1.0 Ma of time. This is a period long enough to record long-term ecological change as well as incorporate several glacial-interglacial transitions in plant communities. Often, the soil carbonate on these surfaces is in the form of laminated pendants which underlie cobbles or boulders. These laminations have measurable isotopic differences, and may provide an internal stratigraphic record of carbonate formation. Temporal control on isotopic records derived from ancient geomorphic surfaces in northwestern Argentina is poorly established, but given the internal lamination of many soil carbonates and accounting for the relative height of a geomorphic surface above the modern river level, a crude time series can be constructed. Concerted effort to establish ages for geomorphic surfaces and associated carbonate would improve this situation. Lack of age control on geomorphic surfaces does not present the biggest problem with extending the isotopic record from ~3 Ma-present. The soils formed on relict land surfaces are fundamentally different from those in the stratigraphic record (Figure 1.9). These two types of soil carbonate form in different landscape positions. Paleosols in the stratigraphic record are generally interpreted to have formed on floodplains or alluvial 21 Figure 1.9: Modern soil forming environments in northwestern Argentina; a) floodplain soil in which water availability results in a riparian ecosystem, b) soil on a relict land surface, isolated from the modern depositional system and from range-front topography, resulting in protracted soil formation under water limiting conditions. The surface in b) is composed of basalt boulders unconformably overlying Pliocene (~3.5 Ma) sandstones. plains. These soils form in aggradational sedimentary settings during periods of reduced sedimentation. Eventually a sedimentary event(s) buries the soil and largely quenches formation processes. In this manner, paleosols represent a discrete period of formation that is associated with fluvial or alluvial sedimentary systems at topographically low points in the landscape. These sedimentary systems implicate a nearby water source, at least intermittently, and exert some control over vegetation. This is quite distinct from soil formation on abandoned geomorphic surfaces, which is protracted and occurs on local topographic highs isolated from modern hydrological catchments. The ecology of these surfaces is different than that on adjacent floodplains. Isotopic interpretation of soil carbonate on geomorphic surfaces. As discussed, the interpretation of carbonate isotopic data from soils on long-lived geomorphic surfaces is hampered by problems of age control and landscape ecology. Two specific issues related to the isotopic composition of soil carbonate can be identified. Firstly, the ecology of high geomorphic surfaces can be quite different from that of 22 nearby environments; this is particularly notable in the Santa Maria valley where the cactus Trichocereus atacamensis (pasacana) is limited to upland hillslopes, pediments, and terraces (cf. Figure 1.4b). This large cactus employs Crassulacean Acid Metabolism (CAM) photosynthesis. All cacti use this photosynthetic pathway to acquire atmospheric CO2 for growth, yet the specific implications of CAM photosynthesis for carbon isotope studies of plant tissue are still being delineated (Dodd et al., 2002; Black and Osmond, 2003; Griffiths et al., 2007; English, 2008). Carbon isotope discrimination of CAM plants is intermediate to that of C3 and C4 plants, thus, soil carbonates with significant CAM contributions are not interpretable with a C3/C4 mixing model. Practically speaking, humid C3 ecosystems can be positively identified, but the isotopic signatures of water-stressed C3 plants, CAM plants, and C3/C4 mixtures are very similar. Trichocereus atacamensis is a particularly drought tolerant species which prefers well drained soils. The surfaces which host T. atacamensis are covered by coarse alluvium and consequently well drained. This soil physical property raises two difficulties in interpreting isotopic data: 1) these high permeability soils are much more susceptible to soil water evaporation (oxygen isotopic effect), and 2) the combination of high permeability and low soil respiration rate raises concerns of atmospheric CO2 contributions to soil carbonate (carbon isotopic effect). Stable isotopes in tooth enamel. Both observational (Cerling and Harris, 1999) and experimental data (Passey et al., 2005a) document a predictable carbon isotope relationship between diet and tooth enamel. This implies that tooth enamel is a faithful recorder of diet in modern environment. Fossil tooth enamel retains a robust record of dietary composition, even under strong diagenetic influence (Wang and Cerling, 1994). 23 As such, a number of ecological and paleoecological questions can be generated and addressed. Dietary specialization is commonly documented, and it is shown that grazing animals such as horses were among the first herbivores to adapt to the ecological expansion of C4 grasses in the late Miocene (Wang et al., 1994; Passey et al., 2002). By selective feeding, herbivores can document the presence of C4 vegetation when soil carbonate data are difficult to interpret. Behavioral differences among herbivores provide a proxy for mixed C3/C4 ecosystems that is also unambiguous. Comparison of tooth enamel and soil carbonate data yields complementary information on the landscape scale distribution of plants using different photosynthetic pathways. This approach is necessary to understand local and regional paleoecologic complexities, cast them in a global context, and to address the influence of growing topography on the late Miocene and Pliocene ecology of northwestern Argentina. Isotopic time series in tooth enamel. Serial analysis of modern and fossil teeth provides valuable insights into short term temporal changes in the diet and body water composition of herbivores (Passey and Cerling, 2002; Passey et al., 2005b). The ability to recover a high resolution time series from a single tooth presents several opportunities for understanding paleoclimate and paleoecology. Temporal variation within a tooth may record ancient seasonality, though with larger mammals migration is a confounding variable. Analysis of small mammal fossils can increase confidence that the dietary signal is locally derived and that migration is a secondary concern. The home range of a small mammal is greater than the land area typically integrated by soil carbonate isotopic data and significantly less than that for large mammals. Given the richness of the South 24 American fossil rodent record (Simpson, 1980), and their under-representation in isotopic datasets, an effort was undertaken to incorporate small mammals into the study. Laser ablation isotopic analysis of tooth enamel. Small teeth can be difficult to analyze by common phosphoric acid digestion methods, which typically require in excess of 500 μg of tooth enamel. Careful analysis by laser ablation produces carbon isotope results in agreement with conventional analyses and permits high spatial resolution sampling of small teeth (Passey and Cerling, 2006). More than 1/3 of the fossil teeth in this dissertation were analyzed by laser ablation. This has reduced the bias of the isotopic record towards large mammals and successfully identified high resolution variation is the diet of small mammals (Figure 1.10). These data have also filled in the stratigraphic record, enabling higher temporal resolution across >5 Ma of geologic time. Figure 1.10: Laser ablation of small teeth: a) sample Arg-246, ~3.5 Ma hegetothere notoungulate tooth, each row of ablation craters corresponds to an isotopic analysis, and b) resultant isotopic profile, size of symbols approximates precision. 25 Geochronology of volcanic ash beds Volcanic ash beds are common in the late Miocene and Pliocene continental deposits along the southern margin of the Puna plateau. Age data for volcanic ash beds are not as numerous as for volcanic rocks on the Puna, but radiometric age estimates for the eruption and deposition of high silica volcanic ashes are available. This constrains the age of deposits and fossils in the region, but few studies have undertaken a systematic approach to dating strata in the region. At Puerta de Corral Quemado, 2 km of strata with a magnetic polarity stratigraphy and multiple dated ash beds (via the K-Ar and 40Ar/39Ar methods) is the principal section of this dissertation (Butler et al., 1984; Latorre et al., 1997). To the south, in the Fiambalá basin, recent stratigraphic work has been undertaken with U-Pb zircon geochronology; these results constrain the age of the strata (Carrapa et al., 2008), and have improved chronological control of formation boundaries (McPherson, 2008). To the north, in the Angastaco basin, U-Pb zircon geochronology has also successfully improved age control of Mio-Pliocene strata and documents time-transgressive formation boundaries (Bywater-Reyes et al., 2010). Integration of this age data into a coherent framework for the region is the principal aim of the geochronological efforts in this dissertation. When possible, stratigraphic sequences of volcanic ash beds have been collected and geochemically analyzed. Sequences of ash beds which contain one or more absolute ages can provide chronological control at other localities via identification and correlation of chemically similar ash beds. Volcanic glass has been separated and analyzed for its major, minor, and trace element composition to accomplish this. In concert, selected ash beds have been dated by 40Ar/39Ar geochronology. 26 40Ar/39Ar geochronology. When the strata of interest are of unknown or poorly constrained age, minerals have been separated for 40Ar/39Ar geochronology from intercalated tuffs. Mineral separates were analyzed by single crystal laser fusion at the Rare Gas Geochronology Laboratory, University of Wisconsin-Madison. When possible sanidine was used, and provided excellent results. In the age range between 2.0-5.5 Ma, fully propagated external uncertainty at the 2σ level is ~0.2% of the measured age. A sanidine dated at 9.1 Ma is slightly less precise, yielding an uncertainty > 0.3%. Sanidine from a young ignimbrite (44 ka) exposed on the Puna plateau yielded an uncertainty of ~5%. For 40Ar/39Ar work, the precision and accuracy of sanidine is unsurpassed. Unfortunately, sanidine is not always present, and other minerals must be employed for age dating. Coeval biotite-sanidine ages provide a basis for testing the fidelity of biotite 40Ar/39Ar geochronology in the region. The oldest age measured was a 9.4 Ma biotite age from the same sample that yielded a 9.1 Ma sanidine age. The biotite age is nearly as precise, but its accuracy is in question. The > 300 ka discrepancy is similar to bioitite-sanidine age discordances identified throughout the region (Hora et al., 2010) and suggests that 40Ar/39Ar analysis of biotite should be avoided if possible. Plagioclase feldspar is an alternative to sanidine; however, 40Ar/39Ar ages are an order of magnitude less precise, yielding external 2σ uncertainties of 2-3% at 4 Ma. The uncertainty obtained for plagioclase (±100 ka) is equivalent to that obtained by U-Pb zircon results, and the two methods are in excellent agreement within the study area. Plagioclase is the most common and abundant mineral in the volcanic ash beds sampled; therefore, its inherent precision may dictate the temporal resolution of the geologic record in unfavorable circumstances. 27 Major element composition of volcanic glass. Broad patterns are discernable in the chemical composition of volcanic glass separated from ash beds (Figure 1.11). Nearly all analyzed samples are rhyolites or dacites following the total alkalis vs. silica classification of Le Bas et al. (1986). Dacitic ash beds are characterized by the highest Fe2O3 and CaO values, in addition to high Al2O3, TiO2, and MgO concentrations. Rhyolitic ash beds have the lowest Fe2O3 and CaO values, and the highest SiO2 content. Ash beds of intermediate composition and character are termed rhyodacites in this dissertation. The average composition and the mineralogy of ash beds are valuable descriptors, but microanalysis is imperative to document and account for shard-to-shard variation in composition. The intershard pattern of Fe2O3 and CaO, as determined by electron probe microanalysis, can be characteristic for a given ash bed, and is especially useful for identifying multiple compositional modes (Figure 1.11). Rhyolite ash beds. Typically, rhyolite ash beds contain only a single compositional mode and can be differentiated from one another on the basis on Fe2O3 and CaO alone. In some instances, ash beds with nearly identical patterns in Fe2O3-CaO space can be differentiated by minor element concentrations (cf. Figure 1.11a). In this dissertation TiO2, MnO, MgO, BaO, Cl, and F are shown to be sufficient to successfully distinguish rhyolitic ash beds. Most rhyolite ash beds fall within a limited range of Fe2O3 and CaO contents (~0.4-1.0 weight %), so positively identifying unique compositions in this field requires as much information as possible. Some ash beds have truly unique chemistry. In the Fiambalá basin, a low CaO composition has the highest measured MnO content and in the Santa Maria valley a sequence of high Fe2O3 rhyolites have the highest measured F contents (Figure 1.11d,f). 28 Figure 1.11: Intershard variation of volcanic glass as documented by electron probe microanalysis; a) two rhyolitic ashes indistinguishable by patterns of Fe2O3 and CaO variation, but differentiable by Cl and slightly different CaO content, b) rhyodacite ash bed with bimodal composition, c) dacite ash beds illustrating polymodal composition, less common unimodal composition, and typical TiO2/MgO covariance, d) all shards from the Fiambalá basin, predominantly rhyolite and dacite compositions, e) all shards from the Corral Quemado basin demonstrating numerous occurrences of all three types of ash beds, and f) all shards from Santa Maria valley illustrating predominatly rhyodacite and rhyolite compositions and the absence of dacite ash beds. 29 Rhyodacite ash beds. Although not strictly defined, the term rhyodacite as used herein is a useful descriptor for ash beds which have compositions intermediate to those strictly defined as rhyolites or dacites. Glass of rhyodacitic composition has higher Fe2O3 and CaO values than rhyolite glass. Several patterns of shard-to-shard variation also serve to distinguish these ash beds. Rhyodacite ash beds may contain glass composition characterized by two discrete compositional modes (Figure 1.11b). These compositional modes are generally apparent on the basis of SiO2, Al2O3, Fe2O3 and CaO. TiO2 or MgO may also identify the compositional modes, but this is not always the case and variation of these elements is greatly reduced relative to true dacites. It is hypothesized that strongly bimodal rhyodacite ash beds are the products of zoned, perhaps stratified, magma chambers. Another common pattern of rhyodacite ash beds is strong intershard variation in Fe2O3 and/or CaO, such that the range could easily encompass several compositional modes, but none are identified. For these ash beds, intershard patterns in Fe2O3-CaO space can have a distinctive orientation, which may reflect processes of mineral growth in the evolving magma. Dacite ash beds. The low SiO2 content (often <70 weight %) of dacitic volcanic glass is inversely correlated with Al2O3, Fe2O3, CaO, TiO2, and MgO contents. These variations are strong, and can span more than one weight % in Fe2O3-CaO space. This variation is often manifested in a series of compositional modes, such that some of these ash beds are termed polymodal (Figure 1.11c). Insufficient analyses have been obtained to adequately define these ‘modes'. Such patterns may emerge with additional data; however, some dacite ash beds have unimodal compositions. High, and covarying, TiO2 and MgO values may be the most universal characteristic of dacite glass. 30 Regional patterns in ash bed composition. Several simple observations regarding the geographic distribution of volcanic ash beds are instructive. Firstly, the majority of ash beds are of rhyolitic composition. Rhyodacite and dacite ash beds are subordinate. Of these, rhyodacite ash beds are thicker and more widespread in their distribution. Dacite ash beds are typically preserved as thin lenses and no long distance correlations have been established. Based upon these observations, it is suggested that dacite eruptions are typically smaller, and perhaps less explosive. Dacite volcanoes may be restricted to the volcanic arc or to isolated back-arc centers; their distribution appears to reflect this. The Santa Maria valley is furthest from the volcanic arc and no dacite ash beds are documented there (Figure 1.11f). The Santa Maria valley is characterized by the highest proportion of rhyodacite ash beds, and by the presence of a short sequence of distal rhyolite ashfall deposits characterized by high Fe2O3 and F contents. Most ash beds in Santa Maria valley are assumed to have derived from large silicic eruptive centers on the Puna plateau. Ash beds preserved in the Fiambalá basin are mostly rhyolites, with rhyodacite and dacite ash beds approximately equivalent in abundance. Fiambalá is the closest locality to the volcanic arc and therefore interpreted as indicative of ashes derived from the arc. Geographically, the Corral Quemado basin is intermediate between Fiambalá and Santa Maria, and the record of explosive volcanism there reflects this. Ash beds at Corral Quemado yield correlations to both Fiambalá and Santa Maria, but no direct correlations have yet been identified between Fiambalá and Santa Maria. Each basin appears to yield a record of explosive volcanism which is dominantly local; these changes are observed at a spatial scale of ~100 km. 31 Trace element composition of volcanic glass. Some trends in trace element composition of volcanic glass are observed, but data are not nearly as extensive as for major elemental composition. Glass data confirm the basic compositional patterns determined from analysis of ignimbrite samples in the region (Schnurr et al., 2007; Kay et al., 2010). Comparison to data from whole rock ignimbrite samples is not equivalent because the samples are fundamentally different material, but some potential correlations from distal ash beds to proximal ignimbrites have been identified on this basis. In the future this may become more tractable as additional source regions are identified and trace element concentrations in glass can be used predictively. Ash beds with similar major element composition often have similar trace element chemistry. For this reason, differentiating ash beds on trace element data is not always straightforward, though small differences in trace element content appear to reliably distinguish ash beds. Building confidence in the robustness of small trace element variations within and between ash beds is crucial. Importantly, trace elemental compositional variation, laterally or vertically, within a volcanic ash bed is observed to be small, even for ash beds which are zoned in major element composition. Trace element data have proven very useful, but the methodology employed relies on samples with well characterized intershard variation. This is because 30 mg aliquots of volcanic glass concentrates are digested and analyzed as a bulk sample. Trace element analysis of individual shards by laser ablation might be employed when several compositional modes are documented in an important sample, but this still relies on abundant electron probe data to characterize the population of glass shards. Electron probe analysis must be used to understand how data are being averaged and which samples carry a primary signal. 32 Sedimentary mixing of volcanic glass. Most of the volcanic ash beds in northwestern Argentina display some evidence of sedimentary reworking. That is, very few ash beds are primary ashfall deposits. Fluvial and eolian processes mix foreign material, volcanic or otherwise, with primary eruptive material. These processes can be assessed by analysis of individual glass shards and the resultant data can be used to determine whether the measured compositions are petrologically related, or the result of sedimentary mixing. These interpretations influence potential correlations, and determine which samples are promising for trace element work. With broad petrologic patterns of volcanic glass chemistry established, several examples are drawn upon here to illustrate the situation (Figure 1.12). These examples reinforce the need for careful sample selection, analysis of large glass shard populations, and detailed analysis of data. At the principal section in the Corral Quemado basin, there is a ~300 m thick interval of laterally extensive coarse sandstone bodies. This interval is interpreted as strata of a braided river system, and numerous volcanic ash beds are present. The stratigraphic relationships between many of these ash beds are difficult to document because they are exposed for only several hundred meters along strike in an almost 5 km wide panel of coarse sandstone. A 2 km long lateral transect in this interval, assumed to be the same ash bed, yielded at least three distinct compositions (Figure 1.12a). The first two samples are unimodal, but their Fe2O3, TiO2, and MgO contents are distinct. The third sample yields a composition equivalent to the second sample, and another population of glass shards that appears to derive from an unrelated polymodal dacite. This is interpreted as resulting from sedimentary mixing. This interpretation is supported by the fourth sample, which contains only a few shards similar to the first sample. 33 Figure 1.12: Compositional populations within volcanic ash beds resulting from sedimentary mixing, a) samples evenly spaced along a 2 km lateral transect assumed to be the same ash bed, b) vertical transect through an ash bed containing finely laminated sediment and paludal carbonate, c) samples with low initial glass concentration. Similarly, a vertical transect through an ash bed yields at least one definite compositional mode and some ambiguous compositional groups (Figure 1.12b). The basal sample constitutes a well defined population, but four samples in the 8 m above it have minimal overlap with this population. The upper samples may define additional compositional modes associated with the base of the ash bed, but sedimentary processes have clearly also mixed in a number of apparently unrelated compositions. Samples of ash beds containing >50% volcanic glass are generally less likely to yield complicated results, but this is not always the case. Alternatively, some samples with a minor volcanic glass component can yield robust and interpretable results. One such example is a thin tuffaceous horizon in Pliocene gravels which was essentially unimodal and can be correlated to other localities (Figure 1.12c). This stands in contrast to a Miocene eolian sandstone with ~1% glass, which appears to be a complex mixture. 34 Volcanic ash beds as structural datums. Volcanic ash beds provide chronological horizons for stratigraphic studies; however, once their age has been established the attitude (strike and dip) of the ash bed provides valuable structural information as well (Figure 1.13). Assuming that deposition was close to horizontal, inferences made from current orientation can be used to constrain the location, timing, and magnitude of upper crustal deformation. This is the most promising approach to study deposition and deformation of poorly bedded conglomerates in the region. Figure 1.13: Volcanic ash beds in the Corral Quemado basin, illustrating their utility as structural datums; a) prominent ash bed at the 482 m level in the Puerta de Corral Quemado (PCQ) stratigraphic section, b) latest Miocene ash bed, dipping ~15° to the SSW at San Fernando, c) view from the upper PCQ section to the prominent Toba Corral Quemado (8 km), which delineates a syncline in coarse sandstones and conglomerates. 35 Structure of the dissertation This dissertation is organized into three principal chapters, each focused on a different set of questions relating to the Miocene-Pliocene development of the southeastern Puna plateau margin. Data collected during the course of this dissertation are fully documented in the appendices, but not necessarily discussed in the text. Data are drawn upon as needed in each chapter. As a whole, the goal of this dissertation is to provide geological constraints on the tectonic, climatic, and ecologic evolution of the southeastern Puna plateau margin during the late Miocene and Pliocene epochs. Chapter 2 develops a synoptic paleoecologic reconstruction in the region between 9-3.5 Ma. This work is undertaken in the 2.5 km thick stratigraphic section exposed at Puerta de Corral Quemado. This was the location of early biostratigraphic studies (cf. Marshall and Patterson, 1981), in addition to the most detailed chronostratigraphic study in the region (Butler et al., 1984) and a long paleosol carbonate record with the highest resolution in the region (Latorre et al., 1997). For these reasons, a detailed paleoecologic analysis was pursued at this locality. Stable isotope data from diagenetic, paleosol, and fossil tooth enamel carbonate are compared throughout the section with the following questions in mind: 1) is the late Miocene expansion of C4 grasslands in northwestern Argentina synchronous with the global record?, 2) how was this ecological expansion modulated by growing orographic barriers?, and 3) what landscape-scale ecological patterns existed in the region during the late Miocene and early Pliocene? Chapter 3 puts forth a regional geochronological framework for the latest Miocene and early Pliocene. This framework is based upon isotopic dating and 36 geochemical characterization of volcanic ash beds. A composite tephrostratigraphy is developed for the upper section at Puerta de Corral Quemado and correlations to volcanic ash beds in the Fiambalá basin are proposed. Identification of widespread ash beds permits their use as both stratigraphic and structural datums. When each datum is considered within a coherent geochronological framework, analysis of spatial and temporal variation in tectonic and sedimentary processes becomes tractable. Following this approach several questions are addressed at Puerta de Corral Quemado and Fiambalá. The timing and style of the Plio-Pleistocene contractional deformation that exposed the study sections is addressed, as is the timing of deposition for coarse conglomeratic deposits capping most stratigraphic sections in the region. Both of these are interpreted as proxies for plateau margin uplift; and thereby constrain the recent tectonic history of the southeastern Puna. Chapter 4 extends the geochronological framework northward to Vallé Santa Maria, and explores regional variations in paleosol carbonate isotopic data within this framework. Short sequences of paleosol carbonate data from Fiambalá are compared to the Puerta de Corral Quemado record. The longer paleosol carbonate record from Vallé Santa Maria is also assessed, with emphasis on the portions of the record that can be confidently correlated to Puerta de Corral Quemado. In total, data spanning ~200 km kilometers from north to south are incorporated into a regional stable isotope record spanning 10 Ma-present. This regionally integrated record is most robust between 8.5- 3.5 Ma, a period of time coincident with global change in terrestrial ecosystems (Cerling et al., 1997) and with the hypothesized rapid uplift of the Altiplano plateau (Hoke and Garzione, 2008). 37 Chapter 5 summarizes the results of Chapters 2, 3, and 4. These results are contextualized with respect to the geologic and tectonic setting as laid forth in this chapter. Major conclusions derived from the late Miocene and Pliocene geological record of the southern Puna plateau margin are synthesized. Outstanding problems are identified, and exciting questions are formulated. Appendix A hosts the background stratigraphic and geochemical framework of the dissertation. The location, identity, composition, and age of important stratigraphic markers are documented. Electron probe microanalysis (EPMA) data for volcanic glass are presented in a stratigraphic framework. Trace element data for volcanic glass, measured by Inductively coupled plasma mass spectrometry (ICP-MS), and 40Ar/39Ar age estimates for minerals are reported by sample number and can be referenced to the stratigraphic framework in this way. Appendix B contains all stable isotope data gathered from northwestern Argentina. This includes analysis of a variety of materials (water, carbonate, and tooth enamel) ranging in age from Miocene to modern. Geographic coordinates, and stratigraphic placement where appropriate, are linked to each sample. 38 CHAPTER 2 STABLE ISOTOPE ECOLOGY ACROSS THE MIOCENE-PLIOCENE BOUNDARY; PUERTA DE CORRAL QUEMADO Introduction The late Miocene hosts a rich record of environmental change in terrestrial environments. The global nature of the environmental change is evidenced, in part, by stable isotope proxies from most continents. Late Miocene expansion of plants using the C4 photosynthetic pathway is frequently documented in locales where these plants currently comprise an important proportion of modern ecosystems (Cerling et al., 1997; Latorre et al., 1997; Passey et al., 2009). C4 plants are particularly successful in tropical and temperate latitudes where monsoonal climates (abundant summer rainfall) dominate. A high proportion of plant cover in grassland and savanna environments is attributable to the C4 pathway. The late Miocene expansion of C4 ecosystems is a global phenomenon; yet potential differences in the timing, magnitude and rate of ecologic change raise questions regarding regional and local response to global change (Fox and Koch, 2004). Having established a detailed stratigraphic framework through a section in northwestern Argentina it is possible to compare stable isotope proxies for plant cover and mammalian diet across the Miocene-Pliocene boundary (cf. Koch, 1998). The stratigraphy is based on volcanic ash beds collected from a 2.3 km thick section at Puerta de Corral Quemado (PCQ) in Catamarca Province, northwestern Argentina (Figure 2.1). Figure 2.1: Study area and geologic setting; a) Location map denoting morphotectonic provinces of the southern Central Andes (after Jordan et al., 1983), b) Shaded relief map (from Shuttle Radar Topography Mission 3 arc second data). 40 Volcanic ash beds and sandstone marker beds which have known relations with the paleosol carbonate record from PCQ (Latorre et al., 1997) permit analysis at high stratigraphic resolution. Within this framework, the nature of Miocene-Pliocene ecological change in the region is addressed through analyses of δ13C and δ18O in enamel of fossil teeth collected from the PCQ section. Stable isotope ratios were analyzed by conventional methods, and also by laser ablation. Laser ablation enables analysis of small and delicate teeth, thereby removing some of the sampling bias inherent to stable isotope studies of fossil collections. The presence of plants using the C4 photosynthetic pathway, as detected by the carbon isotope ratio of paleosol carbonate or fossil tooth enamel, is a proxy for environmental conditions in which C4 plants are ecologically successful. These parameters include low mean annual precipitation, low atmospheric CO2 concentrations and warm growing season (Ehleringer et al., 1997). Warm growing seasons are a result of numerous factors including low latitude, low elevation, and monsoonal precipitation. The modern geographic and climatic setting of the studied section is favorable for recording the Miocene-Pliocene history of C4 biomass in the region. Regional studies indicate a strong elevational gradient in modern C4 plant abundance, and PCQ is currently near the upper limit for C4 dominated ecosystems (Cabido et al., 1997). At present, C4 plants comprise a small proportion of subsurface biomass locally. Regionally, paleosol carbonate studies have documented a complex ecological history over the last 10 Ma; possibly modulated by the growth of orographic barriers during the Pliocene (Kleinert and Strecker, 2001; Latorre et al., 1997). Comparison of stable 41 isotope proxies for C4 vegetation within discrete stratigraphic intervals is an opportunity to evaluate the spatial and temporal dynamics of ecologic change. The goals of this study are to: 1) assess the proliferation of C4 ecosystems in the southern Central Andes as a means of constraining Miocene-Pliocene climate and ecology in the region, 2) compare this to the global record of C4 expansion and evaluate possible local or regional controls on C4 plant distribution , and 3) provide a coherent and accessible chronostratigraphic framework, based upon the geochemical characterization of volcanic ash beds, for use in studies by other workers concerning tectonic, climatic, or ecologic questions. Materials and methods Geochemical characterization of volcanic ash beds Volcanic ash beds were characterized and correlated based upon the elemental composition of volcanic glass. Glass separates were prepared for electron probe microanalysis (EPMA) by ultrasonic treatment of the 125-250 μm fraction in 10% HNO3, 5% HF, and distilled water. When necessary, the glass fraction was concentrated by standard magnetic and density methods prior to being mounted in epoxy, polished, and carbon coated. Analysis of major and minor elements was carried out on a Cameca SX-50 microprobe using methods described by Nash (1992). Typically, 20 glass shards per sample were analyzed and incorporated into average values (Appendix A). Stable isotope analysis of carbonate by acid digestion Paleosol, diagenetic, and tooth enamel carbonate were analyzed using an online carbonate device, the Finnigan Carboflo. Samples were reacted in 100% H3PO4 at 90°C 42 and the evolved CO2 was cryogenically purified prior to analysis via a dual inlet and micro-volume coldfinger on a Finnigan MAT 252 at the University of Utah. Internal carbonate and tooth enamel standards, calibrated to NBS-19, were used to correct measured values relative to the PDB scale. This correction is small for carbon isotope ratios and indicates comparable results irrespective of analytical session. Apparent acid fractionation factors for the oxygen isotope ratio of CO2 liberated from tooth enamel apatite were calculated separately for modern and fossil enamel using equations presented in Passey et al. (2007). Tooth enamel standards analyzed during the study had a 1σ standard deviation of 0.09‰ for δ13C and 0.16‰ for δ18O. This estimate of precision is in agreement with the average absolute difference between duplicate analyses of unknowns (δ13C = 0.07‰, δ18O = 0.31‰, n = 7). A subset of tooth enamel powders was chemically pretreated in ~3% H2O2 for 15 minutes and 0.1 M glacial acetic acid for 15 minutes. Pretreatment is intended to remove organic material and non-enamel carbonate, but in many cases has little effect on the δ13C values (Levin et al., 2008; Passey et al., 2002). In this study, treated enamel is depleted in 13C by ~0.13 ± 0.20 ‰ and enriched in 18O by ~0.39 ± 1.00 ‰ relative to untreated enamel (n = 20). Comparison between untreated enamel and various organic materials demonstrates an opposite relationship. Enrichment in 13C (1.05 ± 0.71‰) and depletion in 18O (1.77 ± 1.58 ‰) is observed in dentin, cementum, and the red coating on rodent incisors relative to adjoining tooth enamel (n = 11). Non-enamel tooth materials may have different isotopic compositions initially, or they may be diagenetically altered. Fossil enamel is less affected by treatment than modern enamel, suggesting that contamination by organic tooth material may be a more important consideration than 43 non-structural carbonate. Our effort to sample as many teeth as possible resulted in most analyses deriving from untreated tooth powders. When a tooth was too small or delicate to physically isolate tooth enamel from the adjoining organic material laser ablation methods were employed. Stable isotope analysis of tooth enamel by laser ablation A subset of fossil tooth enamel samples was analyzed using thermal laser ablation methods. Analysis by laser ablation allows incorporation of small mammals into the study, and additionally permits intratooth analysis of these specimens. Of the 43 samples analyzed for this study, 9 individuals were previously reported in a publication detailing the analytical method (Passey and Cerling, 2006). These data showed that laser δ13C values are comparable to conventional phosphoric acid digestion within 0.5‰ and sufficient for the reconstruction of the fraction of C3 vs. C4 plants in the diet. Later work on the same system observed a 1.9 ± 0.4 depletion in 13C relative to acid digestion (Podlesak et al., 2008). A similar depletion in laser δ13C values is noted for some samples analyzed herein (Arg-103, Arg-106; Appendix B). Passey and Cerling (2006) clearly show that quantitative CO2 yield is not achieved during ablation and that depletion of 13C occurs at low laser power. Fractionation of 12C into the CO2 analyte and 13C into melt-forming condensate may occur during ablation, particularly at low laser power. The analytical strategy was to utilize laser power near the low end of optimal analytical conditions (~5 W) and to increase the number of ablation events as needed to achieve sufficient CO2 for analysis. Primarily, this strategy compromises spatial (temporal) resolution. Any fractionation during ablation should lead to systematically depleted δ13C values and thus does not jeopardize conservative identification of fossil 44 taxa consuming C4 plants. Laser ablation δ13C data are reported relative to the PDB standard and these results are discussed together with those from acid digestion. δ18O laser values are measured in CO2 deriving O from the phosphate, carbonate, and potentially the hydroxyl components of tooth enamel. The mixing of these components and the nonstochiometric yield of oxygen in the CO2 analyte obfuscates the relationship between laser values and conventional acid digestion data. The majority of the O analyzed is from the phosphate component of enamel (Passey and Cerling, 2006). In general, the phosphate oxygen is believed to be more resistant to diagenetic alteration than either the carbonate or hydroxyl oxygen (Zazzo et al., 2004). Deviation from the predicted δ18O offset between laser and acid digestion values may result from incomplete mixing of phases or isotopic fractionation during ablation. Further, the observed difference between fossil and modern enamel analyzed by laser ablation may result from isotopic exchange of carbonate oxygen or replacement of OH- by F- in fossil enamel (Passey et al., 2007). Laser ablation δ18O values are reported on the PDB scale and are discussed briefly alongside the acid digestion analyses. Stratigraphic framework Lithostratigraphic framework Strata of interest belong to the Neogene Santa Maria Group, which also crops out extensively in the Santa Maria Valley 100 km NNE of the study area (cf. Figure 2.1). A historical summary and revision of the lithostratigraphic nomenclature is presented by Muruaga (2001). Stratigraphic correlations to the earliest paleontologic work are possible, though currently supported by only a few marker beds (Butler et al., 1984; 45 Marshall and Patterson, 1981). Magnetostratigraphic and 40K-40Ar data firmly place the PCQ section across the Miocene-Pliocene boundary (Butler et al., 1984). This 2300 m thick section was sampled on the north side of Río Corral Quemado. Subsequently, a 2300 m section south of Río Corral Quemado (Figure 2.2) was sampled for paleosol carbonate and correlated to the magnetostratigraphy by identification and redating of two prominent tuffs by 40Ar/39Ar (Latorre et al., 1997). Together, the paleomagnetic and radiometric age contraints on the strata at Puerta de Corral Quemado provide an integrated geochronology which bears on biostratigraphic, paleontologic, and stable isotopic data. Latorre et al. (1997) recorded the stratigraphy of part of the section at PCQ; geographic coordinates are established for some of their stratigraphic layers (Table 2.1). Ten of these layers are numbered sandstone marker beds; three are prominent tuffs. To ensure proper correlation, samples of eight volcanic ash beds from the study of Latorre et al. (1997) were also analyzed. These establish four additional stratigraphic correlations, and provide the basis for the tephrostratigraphy reported here and in Chapter 3. Tephrostratigraphic framework Compositionally, all analyzed glasses are dacite or rhyolite considering their SiO2, Al2O3, and alkali content. The ash beds at PCQ preserve a rich record of volcanism through the late Miocene and early Pliocene, and are especially numerous in the upper 1,000 m of the section. Analyses of ash beds reported here are only a skeletal framework of the potential record, but they all relate directly to previous stable isotope work. Electron probe microanalyses of volcanic glass improve correlation of this study to Latorre et al. (1997), by providing compositional information for stratigraphically 46 Figure 2.2: Stratigraphic section at Puerta de Corral Quemado. Lithologic column, sandstone marker beds, volcanic ash beds, and paleosol carbonate data from Latorre et al. (1997). Formation names and ash bed names are after Muruaga (2001). 47 Lithostratigraphic Framework m level Fm. Paleosol carbonate isotope ratio ·10 ·8 ·4 ·2 SS#,----,-----,----,----, . , ·6 2 i -V12 i ,, 11 10 9 , I, i 8 , ,, 7 , 6 _1. oba Corral Quemado • 3.66 ±0.10 Ma •.•r,. ft.· ~ • 00( •• • '----:-:-;--:~--1 O.stci"'.l 564 ± 032 Ma o • o " • 00 ,. o o •• o • #60,,1 •• ~-+~-+~.~~------~7~.1~4~±~0=.0~4~Ma~ Toba del Puerto . " . 4 bd, 3 2 -10 -8 lithostratigraphic Symbols ~;;,::::O·:· ... ·::c9:.0:J conglomerate, massive o 013Cp o POB ~: !': .~:.~ ::,,:,: ... .. . ~ . ., conglomerate, cross bedded L- ::~ · :~;2J sandstone, Huvial E!i~""'::":":"-:"-"-:) sandstone, eolian ·6 ·4 ·2 • 0180 po POB siltstone, claystone composite paleosol, > Sm 2 sandstone marker bed v volcanic ash bed Table 2.1: Stratigraphic framework at Puerta de Corral Quemado. Unita Sample ID Thickness Age Latitudeb Longitude Correlative sample EPMA (m) (Ma) (°S) (°W) tuffd Arg 267 2320 27.2210 66.9643 PCQ 806 y tuffc Arg 244 2198 27.2213 66.9631 PCQ 578, 805 y Toba Corral Quemadoc, d Arg 241 2072 3.66 ± 0.10 27.2227 66.9608 PCQ 109, 511, 582-584 y Sandstone #12c 1838 27.2224 66.9548 tuff of sandstone #12d Arg 227 1832 27.2226 66.9548 PCQ 573-575, 793-795 y tuffd Arg 215 1715 27.2239 66.9529 PCQ 570 y Sandstone #11c 1682 27.2236 66.9514 Sandstone #10c 1565 27.2263 66.9494 tuffd Arg 196 1440 5.64 ± 0.32 27.2278 66.9488 PCQ 560 y Sandstone #9c 1417 27.2280 66.9484 tuff Arg 167 1180 y tuff Arg 168 1176 tuff Arg 166 1092 aUnit, sample ID, thickness, and age (2σ) from Latorre et al. (1997). bGeographic coordinates, correlative samples, and electron probe microanalysis (EPMA) this study. cCoordinates established using photographs and field notes of J. Quade (see Appendix A; Figure A.1). dGeographic coordinates from volcanic ash samples correlated to those of Latorre et al. (1997). 48 Table 2.1 (continued): Stratigraphic framework at Puerta de Corral Quemado. Unita Sample ID Thickness Age Latitudeb Longitude Correlative sample EPMA (m) (Ma) (°S) (°W) Sandstone #8 1015 lateral tuff Arg 164 1005 tuff Arg 155 821 Sandstone #7c 630 27.2425 66.9436 tuff Arg 131 595 y Sandstone #6c 588 27.2479 66.9398 Toba del Puertoc Arg 101 486 27.241 66.931 Toba del Puertoc Arg 100 482 7.14 ± 0.04 27.241 66.931 Sandstone #5c 309 27.2420 66.9267 Sandstone #4c 202 27.2425 66.9241 Sandstone #3c 102 27.2431 66.9214 Sandstone #2c 48 27.2442 66.9203 aUnit, sample ID, thickness, and age (2σ) from Latorre et al. (1997). bGeographic coordinates, correlative samples, and electron probe microanalysis (EPMA) this study. cCoordinates established using photographs and field notes of J. Quade (see Appendix A; Figure A.1). dGeographic coordinates from volcanic ash samples correlated to those of Latorre et al. (1997). 49 important ash beds. Four samples analyzed in duplicate confirm analytical reproducibility capable of uniquely identifying ash beds. Populations of glass shards within an individual ash bed are unimodal with two exceptions. A dacitic ash bed 595 m above the base of the section contains at least five compositional modes, identified by discrete TiO2, Fe2O3, MgO, and CaO contents. A rhyodacitic ash bed 1832 m above the base has two compositional modes. All correlations are supported by calculation of the statistical distance between analyses using the concentrations of TiO2, Fe2O3, MnO, MgO, CaO, and Cl (Perkins et al., 1995; Appendix B). Full analyses are provided in Appendix A. The ultimate goal is to examine stable isotope data at high stratigraphic resolution; therefore all data are referenced to their stratigraphic level. The three 40Ar/39Ar dates on volcanic ash beds reported in Latorre et al. (1997) are known to their exact stratigraphic level. Two of these ages also have a known relationship with the magnetostratigraphy presented in Butler et al. (1984). We refrain from calculating ages for intermediate stratigraphic levels because such age estimates are subject to revision as new data become available, but the stratigraphic level of each fossil remains constant. The only stratigraphic level subject to revision is our estimate of the Miocene-Pliocene boundary (1555 m) based upon linear interpolation between two dated ash beds and a boundary age of 5.33 Ma (Van Couvering et al., 2000). In the ensuing text all stratigraphic levels are given in meters above the base of the section. Soil carbonate record Several salient features of the paleosol record frame our analysis of tooth enamel data (Figure 2). The first δ13C pedogenic carbonate (δ13Cpc) values > -8.0‰ indicating 50 the presence of C4 vegetation in a fossiliferous paleosol at 591 m. Several meters above the paleosol, a dacitic ash layer is preserved as a thin discontinuous bed (Arg 131; Table 2.1). δ13Cpc values higher than -7.5‰ further confirm the presence of C4 plants at 816 m. Latorre et al. (1997) highlight the interval between 591 and 816 m as recording minor contributions of C4 biomass to paleosol carbonate, and we adopt this interpretation terming it the δ13Cpc C4 first appearance interval. The largest range in δ13Cpc values (2.6‰) is observed between 1300 and 1400 m, coinciding with the first major conglomerates in the section. A series of paleosols beginning at 2149 m record a rapid increase in C4 vegetation. This sharp positive excursion in δ13Cpc values lies above the 3.66 Ma Toba Corral Quemado (Table 2.1). Paleosol carbonate is an in situ proxy for local environmental conditions at a relatively small spatial scale (10-100 m2). As such, sedimentary and stratigraphic observations provide complementary information to stable isotope data. At PCQ, Muruaga (2001) uses the two most prominent ash beds to demarcate formation boundaries. This approach is convenient, chronostratigraphically robust, and broadly approximates important changes in the sedimentary system. Strata of the Chiquimil Formation below Toba del Puerto record an alluvial environment characterized by sandy channels, broad floodplains, and preservation of fossil wood. Above the Toba del Puerto (~ base Andalhuala Formation) sediments deposited on a broad floodplain continue for more than 300 m upsection before grading into fining upwards gravel beds. Only a little higher in the section, eolian cross-strata and marlstones indicative of sand dunes and shallow standing water become intercalated (Muruaga, 2001). The upper Andalhuala Formation contains extensive bodies of coarse sand indicative of deposition in the distal 51 reaches of a braided river. Above the Toba Corral Quemado (~ base Corral Quemado Formation) cobble sized clasts become more common and give way upward to massive, disorganized conglomerates characteristic of the top of the section. The stratigraphic record of environmental change at PCQ is broadly similar to other localities in the central Andes (Strecker et al., 2007a). Stable isotope results Tooth enamel isotope ratios were obtained for 141 individual specimens; 21 modern and 120 fossil. Rodents and notoungulates comprise ~85% of the fossil teeth sampled. Many specimens were sampled multiple times to assess the isotopic variation within individuals. To mitigate the bias toward specimens sampled multiple times an average value can be assigned to each individual. However, this approach masks potentially important ecological variation within individuals. The use of maximum and minimum values to represent specimens sampled three or more times is explored as an effort to capture individual variation while limiting the importance of multiply sampled individuals in stratigraphic and numerical analyses (Table 2.2). This approach is also attractive because analysis of the data includes measured values, a criterion not satisfied by individual averages. Each stratigraphic interval is further characterized by the 20th and 80th percentile for each population of isotopic data. Unless otherwise noted, data presented and discussed are the max-min subset of the full data (Appendix B). 52 Table 2.2. Summary of stable isotope data grouped by stratigraphic interval and data filtering method. Stratigraphic Tooth enamel Tooth enamel Tooth enamel Paleosol carbonate interval (all data) (individual average) (max-min filter) (all data) m level δ13C δ18Ocvna δ18Olas δ13C δ18Ocvn δ18Olas δ13C δ18Ocvn δ18Olas δ13C δ18O PDB PDB PDB PDB PDB PDB PDB PDB PDB PDB PDB Modern nb 44 44 0 21 21 0 28 28 0 0 0 avg -8.2 0.7 - -8.4 0.3 - -8.5 0.3 - - - 1σ 2.7 3.3 - 2.8 3.1 - 2.8 3.1 - - - 20% -10.6 -2.7 - -10.7 -2.9 - -10.7 -2.8 - - - 80% -6.0 3.6 - -6.0 3.4 - -6.1 3.0 - - - 1555-2215 m n 99 31 68 23 13 15 40 17 29 24 24 avg -6.2 -2.3 -7.9 -7.2 -1.0 -8.2 -7.6 -1.3 -8.2 -5.7 -5.0 1σ 4.4 2.2 2.4 4.1 2.3 2.4 4.3 2.3 2.5 1.7 0.4 20% -10.3 -4.2 -9.7 -10.2 -2.5 -9.7 -10.4 -3.2 -10.1 -7.0 -5.3 80% -1.7 -0.4 -6.7 -4.1 0.0 -7.4 -4.3 -0.2 -6.6 -3.4 -4.8 591-1555 m n 124 58 66 41 32 15 62 41 30 26 26 avg -8.0 -2.0 -8.8 -8.1 -2.1 -8.6 -8.1 -2.0 -8.6 -7.5 -6.0 1σ 3.0 2.0 2.2 2.7 2.3 2.2 2.8 2.2 2.3 1.0 0.5 20% -10.7 -3.7 -10.6 -10.2 -3.9 -10.4 -10.2 -3.9 -10.6 -8.3 -6.3 80% -5.9 -0.2 -6.7 -6.3 -0.2 -7.2 -6.1 -0.1 -6.5 -6.5 -5.7 0-591 m n 234 186 48 55 46 13 85 68 26 22 22 avg -9.6 -4.2 -8.5 -9.2 -3.2 -8.1 -9.3 -3.4 -8.1 -8.9 -7.3 1σ 1.5 2.3 3.5 1.5 2.2 3.7 1.7 2.4 3.7 0.4 1.0 20% -10.6 -5.9 -10.8 -10.1 -4.9 -10.5 -10.5 -5.1 -10.3 -9.1 -7.7 80% -8.7 -2.7 -6.8 -8.1 -1.4 -6.8 -8.2 -1.7 -6.3 -8.7 -6.5 acvn=conventional phosphoric acid digestion method (cf. Passey et al., 2007), las=laser ablation method (cf. Passey and Cerling, 2006). bn=number of measurements, avg=average, 1σ =standard deviation, 20%=20th percentile of the population, 80%=80th percentile of the population. 53 δ18O composition of tooth enamel and diagenetic carbonate Paleoclimatic and paleoecologic reconstructions based upon tooth enamel necessitate that the primary stable isotope signature be recovered. Tooth enamel has low porosity and structural carbonate is expected to retain a primary carbon isotope signature in most diagenetic environments, whereas oxygen isotope signatures are more vulnerable to diagenetic alteration (Wang and Cerling, 1994). This is consistent with diagenetic studies of carbonate rocks and is believed to result from significantly higher molar water/rock ratios for oxygen than for carbon (Banner and Hanson, 1990). Paleosol carbonate often preserves primary carbon isotope signatures, but oxygen isotope alteration is demonstrated from samples which nonetheless retain a micritic texture (Leier et al., 2009). To address the potential for alteration of the primary isotopic signal, diagenetic calcite veins and disseminated carbonate were studied in addition to paleosol and fossil tooth enamel carbonate. δ18O is more variable in tooth enamel (δ18Oen) than in paleosol carbonate (δ18Opc), but the two records exhibit similar stratigraphic patterns (Figure 2.3). Paleosol and tooth enamel carbonate both record 18O enrichment from the late Miocene to the early Pliocene of at least 2‰ (Table 2.2). Diagenetic calcite veins are consistently depleted in 18O relative to paleosol and fossil tooth enamel carbonate; whereas disseminated carbonate is of similar isotopic composition to paleosol carbonate (Appendix B). Based upon paleosol carbonate δ18O values and the isotopic composition of modern water in the region, we choose -4‰ (SMOW) as a conservative initial value for diagenetic water (Appendix B). Choosing initial water values < -5‰ (SMOW) or surface temperature of 25°C both move the depth dependent δ18O calcite projections to the left 54 Figure 2.3: Oxygen isotope composition of carbonate phases at Puerta de Corral Quemado. The isotopic composition of diagenetic calcite is depicted as a function of temperature using fractionation factors from Friedman and O'Neil (1977) where surface temperature is 20°C and diagenetic fluids are assumed to have a homogenous isotopic composition of -4‰ (SMOW). 55 approximately the same amount. Either of these changes precludes calcite precipitation at maximum burial depth. Diagenetic calcite veins sampled from strata buried more than 2 km have δ18Ocv values of -10.4 ± 1.9 ‰ (n=12). The oxygen isotope ratio of this diagenetic calcite is easily distinguished from δ18Oen values observed in fossil teeth buried to an equivalent depth (-3.4 ± 2.6 ‰; n=30). A more convincing test of the fidelity of δ18Oen values is afforded by a rodent tooth and adjoining maxillary bone buried to ~1,700 m (Arg 148; 629 m level). Diagenetic calcite crystals (δ18Ocv = -9.4‰) growing in porous bone adjacent to the tooth are distinct from δ 18Oen values (-3.7‰). δ18Ocv data provides constraints on possible deep burial diagenesis because isotopic fractionation of oxygen between calcite and water is a function of temperature (depth). Thus, a deep burial diagenetic event with an isotopically homogenous fluid could produce the observed stratigraphic trend in δ18O. To produce δ18O vs. depth slopes equivalent to those observed for both paleosol and tooth enamel carbonate would require a geothermal gradient as low as 7°C/km. Use of a more realistic range of geothermal gradients (e.g., 15-30°C/km; see Figure 2.3) indicates that calcite veins probably did not precipitate at maximum burial depth. Disseminated carbonate content in sandstones and volcanic ash beds ranges between 1-65% and yields δ18O values consistent with paleosol carbonate, indicating early diagenetic precipitation rather than deep burial diagenesis. Modern tooth enamel is enriched ~1.5-2.0‰ relative to Pliocene δ18Oen values and the standard deviation of the modern data is significantly larger (Table 2.2). Modern data were derived from samples within a 10 km radius of PCQ with two exceptions. A horse sampled on the windward flank of the northernmost Sierra Aconquija (~125 km 56 northeast of PCQ) yields δ18Oen values of -4.0 ± 0.7‰, whereas a vicuña sampled ~90 km northwest of PCQ on the Puna plateau has a δ18Oen of 3.5 ± 0.2‰. Excluding the aforementioned data from the modern PCQ subset does not significantly change δ18Oen (0.4 ± 2.9‰; n = 24). Vicuña are believed to be obligate drinkers (Vila and Roig, 1992); but the physiological considerations of strong environmental aridity and the possibility of evaporatively enriched drinking water preclude strict comparisons with horse data. However, it remains clear that PCQ occupies an intermediate position both geographically and isotopically. δ13C composition of tooth enamel Endmember C3 and C4 diets are calculated using the strategy of Passey et al. (2002; 2009), which accounts for changing isotopic composition of the atmosphere through time. This calculation utilizes the enrichment factor between enamel and diet (ε*en-diet = 14.1 ± 0.5 ‰) estimated by Cerling and Harris (1999). Taxon-specific experimental estimates demonstrate that this enrichment factor is applicable to domestic cows, ~ 1‰ high for domestic pigs, and at least 2‰ greater than observed for rabbits and voles (Passey et al., 2005a). The latter estimates agree with an ε*en-diet value of 11.0±0.1‰ calculated for woodrats on a controlled diet (Podlesak et al., 2008). This calculation thus employs a conservative approach to identifying fossil specimens with a C4 component in their diet. Many South American fossils have few extant relatives, so little can be inferred about their digestive physiology. Some fossil rodents distributed throughout the section have δ13Cen values more negative than estimates for a pure C3 diet and likely indicate a digestive physiology characterized by ε*en-diet <14.1‰. 57 Fossil tooth enamel data record C4 resource availability not observed in soil carbonate (Figure 2.4). This is notable for the extensive fossil collection older than ~7 Ma, in which 7 of the 59 individuals analyzed, provide positive evidence for the presence of C4 plants. These fossil specimens (Rodentia and Notoungulata), may have acquired C4 plants in their diet either by selective grazing of limited C4 plants locally or by consumption of potentially more abundant C4 resources at other localities. Relative to the paleosol carbonate record these fossils push the first appearance of C4 plants at PCQ back from the 591 m level nearly to the base of the section with δ13Cen >-6.0‰ at the 123, 250, 253, and 350 m levels. In the first appearance interval defined by the δ13Cpc record (591-816 m), 4 of the 20 individuals analyzed record a C4 dietary component. Again, these specimens are both rodents and notoungulates. The fossil record is especially sparse between 830 and 1200 m, a long interval of eolian strata. In total 16 specimens above the δ13Cpc first appearance interval and below the inferred level of the Miocene-Pliocene boundary (1555 m) were analyzed; all but 4 specimens have δ13Cen values indicating C4 plants in their diet. Taken together, 44% of the fossils between 591 and 1555 m consumed C4 plants; though most have δ13Cen values consistent with a minor C4 dietary component. One fossil tooth at 1462 m (a mesothere notoungulate) has values approximating a C4 endmember diet. Of Pliocene fossils, 13 of 23 indicate C4 plants in their diet. As a percentage this is not markedly greater than for the latest Miocene. Of these, 5 have δ13Cen > -2.0‰, consistent with interpretation as predominantly C4 grazers (MacFadden, 2005). The range of values for Pliocene samples far exceeds that observed for the Miocene. The 58 Figure 2.4: Tooth enamel carbon isotope composition from Puerta de Corral Quemado. Estimates of δ13Cen for taxa consuming C3 and C4 diets follow the strategy of Passey et al. (2002, 2009). Histograms represent the stratigraphic intervals identified in Table 2.2 59 distribution of δ13Cen values also records some C3 and mixed feeders. The enriched δ13Cen values may represent a group of specialized C4 feeders. Modern tooth enamel identifies predominantly C3 and mixed feeders. All modern samples were collected within ~10 km of PCQ, with two exceptions: a horse sampled 125 km to the northeast (~2750 m elevation), which yields an average δ13Cen value of -12.2 ± 0.2‰, and a vicuña 90 km to the northwest (~3100 m elevation), which yields an average δ13Cen value of -9.2 ± 0.1‰. Both of these are assumed to represent pure C3 diets owing to their elevations. Most modern taxa sampled at PCQ, i.e., horses and goats, are recent arrivals to South America. Perhaps the best analog to the PCQ fossil record are the modern rodents. Assuming ε*en-diet = 14.1‰ the data indicate a mixed diet dominated by C3 plants. Probably more realistic is an ε*en-diet closer to 11‰, which implies that modern rodents have available and consume a mixture of C3 and C4 plants locally. Intratooth profiles of δ13C and δ18O The South American record contains numerous taxa, especially within the Rodentia and Notoungulata, with high-crowned (including ever-growing) teeth (Simpson, 1980). Along its length a high-crowned tooth preserves an isotopic record of change experienced by the animal while the tooth was being formed (Balasse, 2002; Koch et al., 1989; Passey and Cerling, 2002). Thus, sample profiles along the length of fossil teeth record temporal variations in diet and body water composition (Figure 2.5). For this study, intra-tooth profiles of large teeth were drilled at a spatial resolution of ~1 mm, and the resulting powder was analyzed for δ13C and δ18O using conventional phosphoric acid digestion. Smaller teeth were sampled serially by laser ablation at a spatial resolution of 0.4-0.8 mm. This spatial resolution is approximately equivalent to 60 Figure 2.5: Intra-tooth profiles for selected samples. Laser δ18O data have been shifted +5‰ for purposes of visual comparison (cf. Passey and Cerling, 2006). Taxon, stratigraphic height, and analytical method are noted in each panel. Full data are available in Supplementary data 2. Sample numbers are as follows: a) Arg2002-1, b) Arg2002-17, c) Arg2002-22, d) Arg-246, e) Arg2001-50, f) Arg-200, g) Arg2002-5, h) Arg2002-2, i) Arg-121, j) Arg-133n, k) Arg2002-19, l) Arg2002-25. 61 62 Pliocene ~ 4 o 63' -4 Q ~ -8 "-' U -12 ~ C/O Notoungulata Mesotheriidae a) o 5 10 15 20 Miocene-Pliocene boundary 1833 m acid ~ 4 lN otoungula.ta Mesotheriidae 1462 m o ~ ~ ~ ,-.." ~ -4 Q ~ -8 "-' ~ -12 "Zo OI e~) --~--~~--~----~~~ 5 10 15 20 acid Rodentia b) 1 1 las o 2 4 1511 m Rodentia Octodontidae ~ f) las 0 2 4 6 2067 m :;:..-:: 0 QI) ......... ~ 00 -2 0 ~ ,-.." -4 ~ U t:C Notoungulata -6 "-' Hegetotheriidae -8 c) las d) las + 0 2 4 6 0 2 4 6 8 1550 m 1561 m 0 ~ ROd~ E -2§ Octodontidae -4 t§ ~ Notoungulata -6 ~ Hegetotheriidae g) las h) -8 las + 0 2 4 6 0 2 4 6 8 813CpC C4 first appearance interval ~ ,-.." ~ Q ~ "-' u ~ va 4 1_. _ 1 _ _ . I... I o ~ Notoungulata Mesotheriidae 577 m ~ ::~l_ .~ ( -12 I i) acid I o 5 10 15 20 I 1 1 1 I 1 1 1 II 1 1 1 I n ~ l 591 m ~ Rodentia 692 m1 ~ Rodentia 800 m1- -i; Notoungulata Octodontidae Chinchillidae -2 0 Hegetotheriidae ~ _ ______ _ a&-a -4 D r:1 ~~JFJ:: + o 2 4 6 2 4 6 8 02468 distance from enamel-dentin junction (mm) 1-3 days for ever-growing incisors of small mammals (Passey et al., 2005a; Podelsak et al., 2008). Laser profiles reported here may record as little as a few weeks, and should be interpreted as capturing short-term changes in diet and body water, likely on sub-seasonal timescales for many taxa. Profiles of larger teeth analyzed by conventional acid digestion methods probably represent annual or greater timescales. δ13Cen values varied little within any tooth of larger bodied mammals. Intra-tooth 1σ standard deviations rarely exceeded 0.5‰, and in every instance near endmember diets are indicated. In the lowest 500 m of the section, dinomyid rodents have average δ13Cen values of -9.9 ± 0.6 ‰ and -10.0 ± 0.3 ‰ respectively, while a typothere notoungulate has a value of -10.9 ± 0.2‰ (Appendix B). Mesotheres record a change from pure C3 to pure C4 diets between 577 and 1462 m (Figure 2.5). A Pliocene mesothere (1833 m) also has a near endmember C4 diet and confirms an abundance of C4 plants locally from the latest Miocene through the early Pliocene. Smaller bodied mammals, including hegetotheres and rodents, document more variable diets (Figure 2.5). Specifically, teeth analyzed from the hegetothere subfamily Pachyrukhinae have the largest ranges in δ13Cen. A specimen from near the Miocene- Pliocene boundary has a range of 5.5‰ and one from near the top of the section (Paedotherium sp.) has a range of 4.7‰. Both have mixed diets with substantial C4 components. Rodents from the δ13Cpc first appearance interval at 692 and 800 m have isotopically variable, but endmember C3 diets. Octodontid rodents from the Miocene- Pliocene boundary have stable diets with a minor C4 component. Pliocene rodent incisors collected immediately below the Toba Corral Quemado (3.66 Ma) appear to record 63 strong short-term variation in diet. One incisor indicates a C3 diet and the other a C4 diet, both have large intra-tooth ranges (4.0‰ and 3.4‰, respectively). Oxygen isotope compositions vary widely between individuals. Two octodontid rodents of similar age and similar diet have δ18Oen values that differ by nearly 5‰ (Figure 2.5, panel f,g). Intra-tooth variation greater than 2‰ is unusual in rodents, but hegetotheres have slightly larger ranges. Oxygen isotope variation in excess of 4‰ is noted for Pliocene rodents with large carbon isotope variations. For these fossils a strong correlation between the carbon and oxygen isotope composition of tooth enamel exists. A broad 4‰ excursion observed in the mesothere at 577 m presumably formed on annual or longer timescales. This range in δ18Oen values may record either migration or local seasonality. Discussion of stable isotope results The faunal record at Puerta de Corral Quemado The PCQ fossil record contains some of the earliest North American participants in the Great American Biotic Interchange (GABI; sensu Webb, 1991). At PCQ Procyonidae are known from below the Toba del Puerto (7.14±0.04 Ma) as well as in several higher stratigraphic levels (Butler et al., 1984). This dispersal represents an early wave of North American immigrants prior to the closure of an interoceanic seaway at the isthmus of Panama. Additionally, fossil assemblages of the Huayaquerian and Montehermosan South American Land Mammal Ages (SALMA) are in conformable superposition at PCQ. Butler et al. (1984) place this boundary very near the top of the eolian interval locally and suggest an age of 6.0 Ma for Argentina. An estimate of 6.8 64 Ma for the Huayaquerian/Montehermosan boundary in the Bolivian Altiplano highlights potential regional differences in the boundary (Flynn and Swisher, 1995). At PCQ this timespan (6.8-6.0 Ma) encompasses ~600 m of section that correspond to the poorly fossiliferous interval including and immediately overlying eolian strata. Thus, the PCQ section is not ideal for establishing this boundary. Several observations of the fossil record indicate environmental change through the section. The absolute abundance of fossil teeth decreases upsection (Figure 2.4). The proportion of rodent teeth also increases from 46% in the lowest 590 m of the section to 68% between 591-1555 m, and 83% for the early Pliocene. Larger bodied fossil mammals including dinomyid rodents, toxodont notoungulates, and litoptern ungulates are not known above the base of the δ13Cpc C4 first appearance interval (591-816 m). The fossil lineages providing the most continuous records at PCQ, hegetothere and mesothere notoungulates and octodontid and chinchillid rodents, are among the taxa that never dispersed to North America. This fact attests to the sub-tropical/temperate affinities of the PCQ fauna (Webb, 1991). Estimation of C4 dietary component identifies family level differences in the timing and extent of diet change among rodents and notoungulates (Figure 2.6). The estimates of C4 dietary component presented are conservative by design; nonetheless, strict interpretation of the calculated C4 dietary component is not pursued. Since relative comparisons are always the most robust (given δ13Cen variation due to diagenesis, digestive physiology, plant physiology, and changes in the δ13C of atmospheric CO2) calculated diets between 0-25% C4 as considered probable signs of a C4 dietary component and those >25% as confirmation of a C4 dietary component. Comparatively, 65 Figure 2.6: Family level diet changes for rodents and notoungulates as a function of stratigraphic level. The C4 dietary component is calculated using the strategy of Passey et al. (2002, 2009). the absolute range in δ13Cen values observed for the early Pliocene is strong evidence for abundant C4 plants and a stark contrast to the base of the section. However, to assess family-level stratigraphic patterns, estimation of diet composition is advantageous. Fossils of the notoungulate family Toxodontidae were among the earliest taxa to consume C4 plants, but no specimens have been collected above the 580 m level. Mesothere notoungulates (3 specimens) below 614 m had C3 diets, one at 1462 m had a 66 consistent diet of 70-75% C4 plants and one at 1833 m had a diet between 50-60% C4. Based upon dietary composition and estimated body size of mesothere notoungulates (20-60 kg; Croft et al., 2004); abundant C4 grasses during the latest Miocene and Pliocene are inferred. Hegetothere notoungulates (5 specimens) below 604 m exclusively consumed C3, one at 1437 m records ~40% C4 vegetation in its diet, one at 1561 m ranges between 25-75% C4 and one at 2215 m ranges from 20-65% C4. Hegetothere notoungulates are smaller bodied than mesotheres and are characterized by isotopically variable diets relative to mesotheres (Figure 2.5). Both families consume a greater proportion of C4 plants near the Miocene-Pliocene boundary than they do higher in the section. The largest bodied rodents sampled (Dinomyidae) record C3 diets in the lowest 500 m of the section and are not known from higher in the section. Rodent fossils of the family Caviidae are the first specimens recording consumption of C4 plants. Four caviid rodents record 20-40% C4 dietary component at the 250, 350, 629, and 729 m levels, whereas nine specimens below 760 m may have consumed pure C3 diets. Fossils caviids are not known from the middle part of the section, but three specimens above the 2100 m level consumed 10-80% C4 plants and give no indication of pure C3 diets. Octodontid rodents are poorly represented in the lower part of the section, but the record from 1500- 2120 m shows that 7 of 9 specimens consumed C4 vegetation. These octodontid rodents indicate a diet containing 0-35% C4 plants, and, as with the notoungulates, maximum C4 consumption is observed near the Miocene-Pliocene boundary. In contrast to octodontids, the Chinchillidae rodent record is richer for the lower section; 3 of 7 specimens below 860 m record minor C4 components to their diet (5-20%). Maximum 67 C4 consumption is again documented near the Miocene-Pliocene boundary; a chinchillid at 1370 m records ~15% C4 and one at 1561 m records ~25% C4. For Rodentia the dietary estimates should be viewed as extremely conservative for two additional reasons: 1) the enamel-diet isotopic enrichment factor (ε*en-diet = 14.1 ‰) employed may overestimate that for rodents by as much as 3‰ and therefore underestimate C4 dietary component by as much as 25%, 2) laser ablation methods may bias δ13Cen values by as much as -2‰ and consequently underestimate C4 consumption by >15%. Based upon the consistent occurrence of rodent δ13Cen values depleted beyond estimates for C3 diets (Figure 2.4) it is inferred that an enamel-diet isotopic enrichment factor less than ~14‰ is most reasonable. The evidence for 13C depletion associated with laser ablation of fossil enamel is minimal and in our view a minor consideration for this data (cf. Passey and Cerling, 2006). In the absence of additional tooth enamel records in the region, it is difficult to determine if the fossil record at PCQ effectively captures the adaptation of fossil taxa to C4 food resources or merely the local arrival of these taxa. A shift in notoungulate diet to predominantly C4 food sources is documented to occur between the 600 and 1400 m levels at PCQ. The precise nature and timing of this change is not well documented, but volcanic ash beds constrain it to 7.1-5.6 Ma on the basis of 40Ar/39Ar age estimates. Notoungulate fossils with pure C3 diets have not been identified after 5.6 Ma suggesting a true shift in diet. Rodents are among the first to consume C4 plants, but an obvious shift in diet is not observed. Several Pliocene rodents appear to have had nearly pure C4 diets, though most indicate mixed diets dominated by C3 plants. Due to their limited home ranges, rodents provide isotopic proxies that are more likely to sample an environment 68 similar to that in which they were preserved (taphonomic considerations excluded). As such Pliocene rodents indicate an abundance of both C3 and C4 plants. Understanding small scale spatial patterns in plant communities is aided by laser ablation methods. The small volume sampled by laser ablation, coupled with the high growth rates of small mammal teeth, resolves isotopic heterogeneity resulting from short-term variations in diet and body water composition. For small mammals, such intra-tooth variations can be constrained to a limited home range and thereby remove taphonomic concerns of a mixed assemblage representing two separate environments. The most convincing evidence for mixed C3/C4 environments comes from intra-tooth laser ablation of the smallest notoungulates (hegetotheres). Hegetotheres have isotopically variable (mixed) diets and also suggest a landscape with both plant types locally available. At many scales, small teeth analyzed by laser ablation methods exhibit a relationship between δ13C and δ18O. Rodent teeth analyzed from the δ13Cpc C4 first appearance interval are systematically depleted in 13C and 18O relative to those near the Miocene-Pliocene boundary (Figure 2.7). These two intervals represent the most fossiliferous strata above and below the C4 expansion at PCQ, and correspondingly afford the best opportunity to compare ecological variation as a function of time. Increased inter-tooth isotopic variation is coincident with enrichment for both δ13C and δ18O. Ecological expansion of C4 plants is expected to increase inter-tooth variation in δ13C as a new, isotopically distinct, food source becomes available. Increased inter-tooth variation in δ18O among rodents may result from increased aridity, new ecological strategies, or both (Levin et al., 2006). The observation of increased inter-tooth variation is consistent with the opening of new, isotopically varied, ecological niches between 7.0-5.5 Ma. 69 Figure 2.7: Inter-and intra-tooth isotopic variability measured by laser ablation. Data from 591-816 m represent all analyses of rodents (4 multiply sampled specimens) within the δ13Cpc C4 first appearance interval. Data from 1480-1550 m represent all analyses of rodents (7 multiply sampled specimens) from the fossil rich interval immediately below the inferred level of the Miocene-Pliocene boundary. The intra-tooth variation of three Pliocene fossils is plotted along with a linear least squares regression through the data for each individual; these three specimens correspond to panels b, c, and d in Figure 2.5. Intra-tooth isotopic variation, as measured by the variance, standard deviation and range of values from a single tooth, does not increase significantly across this interval. Rodents from the base of the section exhibit the least intra-tooth variation and no marked differences are noted from 591-2000 m. However, a subset of rodents between 2000- 2100 m exhibits strong intra-tooth variation in δ13C and δ18O characterized by positive linear correlations (Figure 2.7). In total, four rodents from this interval yield linear relationships with consistent slopes where the range of δ18O values is greater than δ13C. 70 A positive linear correlation is also noted for a hegetothere (Paedotherium sp.) at 2215 m, though the slope is significantly different (Figure 2.7). A hegetothere from the 1561 m level yields a similar slope and in both cases the range in δ13C is much greater than the range in δ18O. These observations suggest that under some circumstances a strong relationship exists between diet and body water isotopic composition for small mammals. This relationship is especially strong for fossil rodents from the upper part of the section. A host of climatic, ecological, and physiological factors can be invoked to explain the relationship between a 13C enriched diet and 18O enriched body water. The δ18O enrichment of warm season precipitation should correspond to maximum C4 abundance and result in a positive correlation between diet and body water at seasonal timescales. The growth rates of rodent incisors are probably too high to record seasonal changes, though the intra-tooth profiles of hegetotheres may do so. Leaf water of grasses is known to be more evaporatively enriched in 18O than dicotyledonous species (Helliker and Ehleringer, 2000). Short term dietary changes might explain the inferred diet-body water relationships by providing 18O enriched leaf water during periods of increased C4 consumption. This mechanism relies on a large percentage of body water deriving from plant water (at least during consumption of C4 plants) and is most attractive for small mammals with rapid body water turnover. This is supported by δ18O enrichment of a C4 rodent incisor relative to a contemporaneous C3 rodent incisor (Figure 2.7). The C4 incisor has an average δ18Olas of -7.7 ± 1.3 ‰ (Arg2002-17; 2017 m) whereas the C3 incisor has an average δ18Olas of -9.9 ± 1.5 ‰ (Arg2002-22; 2067 m). The strong intra-tooth relationship between δ13C and δ18O (r2 > 0.9 in many cases) raises some analytical concerns. This is particularly true for rodent incisors with thin 71 enamel, where lower laser power is required for analysis. The possibility of simultaneous fractionation of 13C and 18O during ablation events exists, though the range of values far exceeds the anticipated effect due to changes in laser power (Passey and Cerling, 2006). The teeth exhibiting δ13C/δ18O correlation were not analyzed at lower laser power than other teeth. Further, the δ13C/δ18O correlation was observed for teeth in several analytical sessions during which many teeth do not exhibit such a correlation. The fact that these teeth are confined to a specific stratigraphic interval suggests an ecological rather than analytical signal. The distinctly different, though equally robust, δ13C/δ18O correlation for the Paedotherium sp. from 2215 m also supports an ecological interpretation of this correlation. Larger body size, lower metabolism, and slower turnover of body water combined with different tooth growth rate and ecological strategy could explain the much shallower slope in δ13C/δ18O space. The Paedotherium sp. tooth is characterized by thick enamel and high CO2 yield per ablation event. These characteristics make similar samples attractive for future laser ablation work. If tooth growth is slow enough such samples may be ideal for addressing Pliocene seasonality. Larger bodied notoungulates (e.g., mesotheres) are also attractive targets for assessing seasonality based on potentially slower tooth growth rates. However these specimens may be less likely to record δ13C/δ18O relationships due to body size and feeding ecology (not mixed feeders). The ecological implications of the δ13C/δ18O correlations observed for laser ablation data are approached with caution. Analytical concerns are considerable and may prove difficult to control when analyzing rodent incisors with extremely thin enamel. Nonetheless, small Pliocene mammals from PCQ provide evidence that diet and body 72 water isotopic composition are positively correlated. This may prove useful for investigating physiology, environmental aridity, and seasonality. Fossils of the subfamily Pachyrukhinae (Notoungulate, Hegetotheriidae) appear to be especially well suited for intra-tooth studies using laser ablation. Climate and environment at Puerta de Corral Quemado Stable isotope data from paleosol and fossil tooth carbonate document climatic and environmental change. These two proxies record δ18O enrichment > 2‰ through the section. Globally, paleosol carbonate data document enrichments in 18O in South Asia and East Africa during the late Miocene and early Pliocene (Cerling, 1992; Quade and Cerling, 1995; Quade et al., 1995). General correlations between δ13Cpc and δ18Opc in these records led to the hypothesis that the oxygen isotope record was dominated by increased evaporation of soil water as C4 grasslands expanded. Recent studies in South Asia have concluded instead that the 18O enrichment in paleosol carbonate documents decreasing rainfall via the "amount effect" (Behrensmeyer et al., 2007; Dettman et al., 2001). Plant wax δD values from the Bengal Fan indicate that both decreased precipitation and increased evaporation between 10 and 5.5 Ma contributed to the 18O enrichment (Huang et al., 2007). The δ18O enrichment at PCQ can be constrained to the same interval as observed for East Africa and South Asia. Whether this enrichment records a globally significant shift in the hydrologic cycle remains an important question. Several potential global causes have been suggested; including increasing continental aridity and ecological disturbance worldwide (Tipple and Pagani, 2007). If earth's hydrologic cycle is a closed system at the temporal scale of 10 Ma, an increase in aridity must be balanced by similar 73 increases in precipitation. Changes in the distribution of precipitation may have been spatial, temporal, or both. Increased seasonality of precipitation is an attractive explanation for two reasons; 1) precipitation during warmer months would result in systematically enriched 18O input to the landscape (Kohn and Welker, 2005; Rozanski et al., 1993), and 2) enhanced evaporation during the dry season would lead to 18O enriched soil water (Gat, 1996). Both of these factors are likely to be recorded in paleosol carbonate isotope ratios if carbonate formation is seasonally biased towards dry periods following a rainy season (Breecker et al., 2009). Paleosol carbonate data from the Bolivian Altiplano record depletion in 18O across the same time interval (Garzione et al., 2006). These δ18Opc data, supported by other proxy data, have been interpreted to document rapid uplift of the northern Altiplano between 10-6 Ma (Bershaw et al., 2010; Garzione et al., 2008). Rapid uplift of the Bolivian Altiplano is a viable cause for the inititation of the South American monsoon by increasing moisture transport from the tropics to higher latitudes via the South American low-level jet (Insel et al., 2009). However, it is possible that slow topographic growth of the Andean plateau may have triggered similar climatic changes when a topographic threshold was crossed (Ehlers and Poulsen, 2009). In either case, the strong divergence in δ18O proxy records between PCQ and the Altiplano points to major reorganization of precipitation patterns during the late Miocene. Regional differences in δ18O proxy records may be further explained by differential uplift histories and local orographic effects. An orographic rainshadow related to growth of the Sierra Aconquija may have formed as early 6-5 Ma (Kleinert and Strecker, 2001; Sobel and Strecker, 2003), and a pulse of Pliocene uplift is well 74 documented in northwestern Argentina (Strecker et al., 2007a). By definition, orographic controls on climate and environment predict strong spatial gradients. Consequently, our ability to test the influence of local orographic barriers on the Miocene-Pliocene climate and ecology of PCQ is currently limited by the spatial density and temporal control of proxy records. Modern climatic gradients in the region are striking (Bookhagen and Strecker, 2008). A strong gradient from the easternmost Sierras Pampeanas to the Puna plateau is documented with modern isotope data. Two modern δ18Oen values from high elevation samples record a 7.5‰ westward enrichment at a spatial scale of 150 km. Both of these samples are from obligate drinkers; yet, the complexity of variables influencing δ18Oen values in an individual precludes straightforward interpretations (cf. Levin et al., 2006). The corresponding 3‰ enrichment in δ13Cen provides more convincing evidence of a strong climatic gradient. We assume that both specimens sampled environments of pure C3 plants owing to their elevation (>2.7 km). The 3‰ difference in δ13Cen is attributable to differences in carbon isotope discrimination between C3 plants in wet environments and C3 plants in water-stressed environments (Farquhar et al., 1989). Documenting Miocene-Pliocene climatic and environmental gradients in northwestern Argentina at similar spatial scales should be possible using isotope proxy data, and doing so will provide a test of local/regional modulation of a global ecological event. With regard to the vegetative history at PCQ, carbon isotope ratios of fossil and modern teeth complement the paleosol carbonate isotopic record. δ13Cen values indicate that C4 plants were present at PCQ from 9.0-3.5 Ma, and remain there today. C4 plants were consumed by herbivores in minor amounts prior 7 Ma; during this time many of the 75 largest herbivores have δ13Cen values (-8 to -11‰) indicative of C3 vegetation in mesic to xeric environments. The environment was probably not humid or densely forested, but rather a mosaic of forested and grassland environments. Substantial C3 grassland may have existed and some C4 grasses were certainly present. For plant and herbivore communities, the period from 7.0-5.5 Ma was one of transition. Locally eolian environments were present for ~ 0.5 Ma. The proportion of herbivores consuming C4 plants increases markedly across this interval, but due to the poorly fossiliferous nature of the strata details are scarce. The Pliocene fossil record strongly indicates a mixed or patchy environment with sufficient C4 vegetation to support specialized feeders. This reconstruction informs interpretation of the uppermost paleosol carbonate in the section. Between 2149 and 2219 soil carbonate nodules at 4 different levels record a dramatic positive shift in δ13C values to a maximum of -2.5‰. This entire zone is above Toba Corral Quemado (2072 m, 3.66 Ma). A prominent ash bed occurs within this interval (Arg-244, 2198 m) and another (Arg-267, 2320 m) overlies this positive shift in δ13Cpc values. Radioisotopic ages for these ash beds will be an important constraint on the rate of this change. The present interpretation is that this occurred relatively rapidly and probably represents a shift across ecotone boundaries present throughout the early Pliocene as opposed to an ecological expansion of C4 grasslands. An elevational shift in ecotone boundaries due to glacial-interglacial type climate changes is plausible (cf. Amundson et al., 1996); however the absence of a synchronous δ18Opc shift argues against this hypothesis. Similarly, the lack of a δ18Opc shift also argues against a C4 expansion resulting from the development of orographic rainshadows (Kleinert and Strecker, 2001). 76 Instead an explanation invoking lateral shifts in ecotone boundaries that