Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Evolution of the Indian Monsoon and rise of C₄ photosynthesis in the Miocene and Pliocene
(USC Thesis Other)
Evolution of the Indian Monsoon and rise of C₄ photosynthesis in the Miocene and Pliocene
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EVOLUTION OF THE INDIAN MONSOON AND RISE OF C 4 PHOTOSYNTHESIS IN THE MIOCENE AND PLIOCENE by Hannah Liddy A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (GEOLOGICAL SCIENCES) May 2017 Copyright 2017 Hannah Liddy Dedication To Bill Ballou Cranshaw ii Acknowledgments I would like to thank... My advisor: Dr. Sarah J. Feakins, who is a constant inspiration for what it means to be a rigorous and passionate scientist. Thank you for introducing and guiding me through these fascinating research projects. Your work ethic, dedication, and fundamental love for science will always be a source of inspiration to me. My dissertation and qualifying exam committees: Dr. Frank Corsetti, Dr. David Bottjer, Dr. Doug Hammond, Dr. Caleb Finch, Dr. Wiebke Ziebis for their guidance both scientifically and professionally. Lab mates, friends and colleagues along the way: Audra Bardsley, Danielle Monteverde, Jotis Baronas, Adam Holt, Hyejung Lee, Mong Sin Wu, Camilo Pon- ton, Bernd Hoffmann, Bernhard Aichner, Elias Karkabi, Annie Tamalavage, Mark Peaple, Sylvia Dee, Joyce Yager, Paulina Pinedo, Gen Li, Caty Tems, Jianghao Wang for laughs, support, and scientific discussions. USC Earth Science staff: Cynthia Waite, John McRaney, and John Yu for their constant and dependable support for all matters administrative, technical, and financial, and Miguel Rincon for showing me the ropes in the lab during my first two years at USC. iii USC Earth Science professors: Will Berelson, Josh West, Julien Emile-Geay, and Lowell Stott for teaching the classes that helped me build a foundation in geochemistry, paleoceanography, and statistics. My support network: Dave, Nancy, Eddie, Alana, Liza, Maureen, Lizzy, Paul, Sam, Beth, Ryan, Nevin, and Noel. iv Contents Dedication ii Acknowledgments iii List of Tables ix List of Figures x Abstract xii 1 Introduction 1 1.1 Modern dynamics of the South Asian Monsoon . . . . . . . . . . . 1 1.2 South Asian Monsoon on tectonic timescales . . . . . . . . . . . . . 4 1.3 Miocene-Pliocene evolution of the South Asian Monsoon . . . . . . 4 1.4 C 4 Expansion in the late Miocene . . . . . . . . . . . . . . . . . . . 7 1.5 Marine sediments of the Arabian Sea . . . . . . . . . . . . . . . . . 9 1.5.1 Gulf of Aden, DSDP Site 231 . . . . . . . . . . . . . . . . . 9 1.5.2 Indus Fan, IODP Site U1457 . . . . . . . . . . . . . . . . . . 9 1.6 Plant wax abundance and carbon isotopes . . . . . . . . . . . . . . 11 1.7 Plant wax hydrogen isotopes . . . . . . . . . . . . . . . . . . . . . . 13 1.8 Contents of this dissertation . . . . . . . . . . . . . . . . . . . . . . 15 2 Cooling and drying in northeast Africa across the Pliocene 17 2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3 Regional Climate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.3.1 Precipitation amount and isotopic composition . . . . . . . . 23 2.3.2 Atmospheric circulation and transport of wind-blown proxies 23 2.3.3 Oceanography of the Gulf of Aden . . . . . . . . . . . . . . 24 2.3.4 Modern vegetation distribution in northeast Africa . . . . . 25 2.4 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.4.1 Marine sediments . . . . . . . . . . . . . . . . . . . . . . . . 25 2.4.2 Lipid Extraction . . . . . . . . . . . . . . . . . . . . . . . . 26 v 2.4.3 TEX 86 analysis . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.4.4 Compound specific carbon and hydrogen isotopic analysis . . 28 2.5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.5.1 Warm early Pliocene supported C 4 grasslands . . . . . . . . 29 2.5.2 Drying at 4.3 Ma . . . . . . . . . . . . . . . . . . . . . . . . 33 2.5.3 Indian Ocean influence on east African aridity . . . . . . . . 34 2.5.4 TEX 86 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.5.5 Alkenones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.5.6 Mg/Ca . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 2.5.7 Analyzing trends in Pliocene Indian Ocean temperatures . . 36 2.5.8 Ecosystem change and hominin evolution . . . . . . . . . . . 41 2.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 2.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 3 Late Miocene C4 Expansion in the Indus River Catchment 46 3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.3 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.3.1 Sedimentology . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.3.2 Modern climate and vegetation . . . . . . . . . . . . . . . . 50 3.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.4.1 Site location . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.4.2 Age model and sample selection . . . . . . . . . . . . . . . . 52 3.4.3 Bulk organic carbon analysis . . . . . . . . . . . . . . . . . . 53 3.4.4 Lipid extraction . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.4.5 Leaf wax quantification . . . . . . . . . . . . . . . . . . . . . 56 3.4.6 Compound specific carbon and hydrogen isotopic analysis . . 57 3.4.7 GDGT quantification . . . . . . . . . . . . . . . . . . . . . . 58 3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.5.1 Concentration and δ 13 C of bulk organic carbon . . . . . . . 60 3.5.2 Plant wax concentration and distribution of n-alkanes . . . . 60 3.5.3 Plant wax concentration and distribution of n-alkanoic acids 62 3.5.4 Plant wax δ 13 C and δD compositions . . . . . . . . . . . . . 62 3.5.5 GDGTs concentrations and temperature reconstructions . . 64 3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.6.1 Fidelity of paleoclimate records during shifts in marine and terrestrial sedimentation . . . . . . . . . . . . . . . . . . . . 66 3.6.2 Transport pathways of n-alkanes and n-alkanoic acids . . . . 67 3.6.3 Transport pathways of pollen and leaf wax . . . . . . . . . . 69 3.6.4 Vegetation reconstructions . . . . . . . . . . . . . . . . . . . 70 3.6.5 C 4 expansion in Indo-Arabian region pre and post 7 Ma . . 72 3.6.6 Indus catchment vegetation changes pre and post 7 Ma . . . 73 vi 3.6.7 Reconstructions of terrestrial hydroclimate . . . . . . . . . . 74 3.6.8 Late Miocene C 4 expansion . . . . . . . . . . . . . . . . . . 77 3.6.9 Did precipitation changes drive the C 4 expansion? . . . . . . 80 3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.8 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 4 Photosynthetic pathway of grass fossils from the upper Miocene Dove Spring Formation, Mojave Desert, California 83 4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.3 Stratigraphic context . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4.3.1 Geologic setting . . . . . . . . . . . . . . . . . . . . . . . . . 87 4.3.2 Tephra and biostratigraphic age control . . . . . . . . . . . . 90 4.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 4.4.1 Sample collection . . . . . . . . . . . . . . . . . . . . . . . . 90 4.4.2 Fossil imaging . . . . . . . . . . . . . . . . . . . . . . . . . . 91 4.4.3 Carbon isotope measurements . . . . . . . . . . . . . . . . . 92 4.4.4 Inorganic carbon isotopic measurements . . . . . . . . . . . 93 4.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 4.5.1 Plant fossil descriptions . . . . . . . . . . . . . . . . . . . . 93 4.5.2 Stable carbon isotopes . . . . . . . . . . . . . . . . . . . . . 97 4.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 4.6.1 Grass fossils from the Dove Spring Formation . . . . . . . . 98 4.6.2 Miocene carbon isotope values . . . . . . . . . . . . . . . . . 99 4.6.3 C 4 grass fossil revisited . . . . . . . . . . . . . . . . . . . . . 100 4.6.4 Miocene-agepaleoecologicalreconstructionsfromthepresent day Mojave Desert . . . . . . . . . . . . . . . . . . . . . . . 102 4.6.5 North American grassland expansion . . . . . . . . . . . . . 104 4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.8 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 5 Conclusions 107 Reference List 111 A Appendix 127 A.1 Carbon isotopic composition of long chain n-alkanoic acids . . . . . 127 A.2 Ocean temperature calibrations . . . . . . . . . . . . . . . . . . . . 128 A.2.1 Alkenone calibration . . . . . . . . . . . . . . . . . . . . . . 128 A.2.2 Mg/Ca SST calibration . . . . . . . . . . . . . . . . . . . . . 129 A.2.3 Ocean temperature anomalies . . . . . . . . . . . . . . . . . 133 vii B Appendix 154 B.1 Chapter 3 plant wax data tables . . . . . . . . . . . . . . . . . . . . 154 C Appendix 163 C.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 C.1.1 Lipid extraction . . . . . . . . . . . . . . . . . . . . . . . . . 163 C.1.2 Leaf wax quantification . . . . . . . . . . . . . . . . . . . . . 164 C.1.3 Compound specific hydrogen isotopic analysis . . . . . . . . 165 C.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 C.2.1 Plant wax abundance and distribution n-alkanoic acid . . . 166 C.2.2 n-Alkanoic acid hydrogen isotope ratios . . . . . . . . . . . . 167 viii List of Tables 4.1 Grass fossil carbon isotopes . . . . . . . . . . . . . . . . . . . . . . 98 A.1 Carbon isotopic composition of C 28 , C 30 and C 32 n-alkanoic acids from DSDP Site 231. . . . . . . . . . . . . . . . . . . . . . . . . . . 134 A.2 Hydrogen isotopic composition of C 28 n-alkanoic acid from DSDP Site 231 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 B.1 Carbon isotopic composition of C 26 , C 28 , C 30 , C 32 , and C 34 n- alkanoic acids from IODP Site U1457. . . . . . . . . . . . . . . . . 154 B.2 Carbon isotopic composition of C 25 , C 27 , C 29 , C 31 , C 33 , and C 35 n-alkanes from IODP Site U1457. . . . . . . . . . . . . . . . . . . . 154 B.3 Hydrogen isotopic composition of C 24 , C 26 , C 28 , C 30 , and weighted mean average of C 24 -C 30 n-alkanoic acid from IODP Site U1457. . . 154 B.4 HydrogenisotopiccompositionofC 25 ,C 27 ,C 29 ,C 31 ,C 33 ,andweighted mean average of C 25 -C 33 n-alkane from IODP Site U1457. . . . . . 154 ix List of Figures 1.1 Indian monsoon circulation . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Late Miocene C 4 expansion in the Siwaliks . . . . . . . . . . . . . . 8 1.3 C 3 and C 4 Bulk plant carbon isotopes values . . . . . . . . . . . . . 14 2.1 Map of DSDP Site 231 in the Gulf of Aden . . . . . . . . . . . . . . 21 2.2 Biomarker and pollen data from DSDP Site 231 . . . . . . . . . . . 31 2.3 Pliocene global temperatures, Northern Indian Ocean temperatures and upwelling indices. . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.4 Indian Ocean temperature anomalies relative to 5-4Ma. . . . . . . . 38 2.5 Terrestrial vegetation records from DSDP Site 231 and hominin diets. 43 3.1 Map of IODP Site U1457 in the Arabian Sea . . . . . . . . . . . . . 53 3.2 IODP Site U1457C Shipboard Age Model . . . . . . . . . . . . . . 54 3.3 Plant wax relative abundance distributions extracted from Indus Fan sediments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.4 Plant wax and organic carbon δ 13 C and Plant Wax δD compositions 63 3.5 Plant wax bivariate plots . . . . . . . . . . . . . . . . . . . . . . . . 64 3.6 Organic and inorganic proxy records from IODP Site U1457C . . . 66 3.7 Vegetation records from IODP Site U1457C. . . . . . . . . . . . . . 71 3.8 Paleoprecipitation δD reconstructions from IODP Site U1457C . . . 75 x 3.9 Late Miocene C 4 expansion . . . . . . . . . . . . . . . . . . . . . . 80 4.1 Regional map, hand sample image, stratigraphic column of Dove Spring Formation, CA, USA . . . . . . . . . . . . . . . . . . . . . . 89 4.2 Thin section images 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.3 Thin section images 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 97 4.4 Fossil image reprint . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 A.1 δ 13 C of C 28 , C 30 , C 32 n-alkanoic Acids from DSDP Site 231. . . . . 129 A.2 U k 0 37 Comparison plot from ODP Site 722 . . . . . . . . . . . . . . 130 A.3 Mg/Ca of seawater data . . . . . . . . . . . . . . . . . . . . . . . . 132 A.4 Mg/Ca seawater correction . . . . . . . . . . . . . . . . . . . . . . . 133 A.5 BIT index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 C.1 IODP Site U1456 plant wax n-alkanoic acid relative abundance dis- tributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 C.2 IODP Site U1456 plant wax δD n-alkanoic acid . . . . . . . . . . . 168 xi Abstract C 4 photosynthesis is a geologically recent adaptation to the ancestral C 3 pho- tosynthetic pathway used to fix CO 2 in plants. The evolution of grasses using the C 4 pathway and their rise to ecological dominance in grasslands and savan- nas of the tropics and subtropics in the late Miocene is enigmatic, and its drivers remain an ongoing debate. In this dissertation, I present three studies focusing on the late Miocene to Pliocene expansion of C 4 photosynthetic plants. Using plant leaf wax carbon and hydrogen isotopes, it is possible to document not only when the isotopic transition to a C 4 dominant ecosystem occurred but also assess the role of hydrological change in driving the biome shift using the hydrogen iso- tope composition of the same molecules. Therefore, this proxy is applicable in monsoon sensitive regions to assess the interplay between vegetation and hydrol- ogy. In Chapter 2, we analyzed samples from a marine sediment core (DSDP Site 231) extracted from the Gulf of Aden to determine the Plio-Pleistocene expansion of C 4 biomass over the Horn of Africa and assess its relevance to early human evolution. We find that arid C 3 shrublands expanded as ocean temperatures pro- gressively cooled and rainfall decreased. These changes occurred prior to the onset of Northern Hemisphere glaciation suggesting major changes ocean circulation and the strength of the Indian Monsoon. In Chapter 3, we determined the late Miocene expansion of C 4 grasslands in the Indus River floodplain using sediments collected xii from the Indus Fan (IODP Site U1457). Despite the complex depositional envi- ronment of the Indus Fan, we can infer that the late Miocene C 4 transition is widespread throughout the Indus Catchment and surrounding regions. In Chapter 4, we address the sparse grass fossil record by revisiting the site of one of the old- est known C 4 grass fossils of late Miocene-age. We find evidence via taphonomy, microstructure and isotopic composition of exclusively C 3 grass fossils from this location and present evidence to revise the previously published C 4 designation. Together these investigations inform the late Neogene record of C 4 photosynthesis in monsoon sensitive regions and offer an important addition, and revision, to the sparse grass fossil record. xiii Chapter 1 Introduction In this dissertation, I reconstructed changes in vegetation and climate during the late Miocene and Pliocene using plant wax molecules preserved in marine sed- iment archives. Marine sediments from the Gulf of Aden and Indus Fan sediments from the western margin of the Arabian Sea represent very different depositional settings but reflect aspects of the same climatic phenomenon: the Indian Summer Monsoon. I then address the sparse record of grass fossils to identify the photo- synthetic pathway of late Miocene fossil grasses from the Dove Spring Formation, CA, USA. 1.1 Modern dynamics of the South Asian Mon- soon Broadly, monsoons are defined by atmospheric circulation systems character- ized by the seasonal reversal of prevailing winds and wet and dry conditions (Fig. 1). For the Indian Summer Monsoon, the effects of this system include the strong cross-equatorial transport of water vapor, intense upwelling along the western coasts of Africa and the Arabian Peninsula, dramatic increase in precipitation over the India, Nepal, and coastal Bangladesh and the marked seasonal nature of these phenomena. The seasonal cycle of the South Asian Monsoon system begins in the Maritime Continent of Indonesia and its surrounding islands. In this region 1 during the spring equinox, large-scale ascent occurs in association with a precip- itation maximum. Progressively, this precipitation maximum shifts towards the Indochina Peninsula, Bay of Bengal, and the South China Sea, and by early June, this zone of large-scale ascent strengthens and forms deep convection over India and the southern regions of the Himalaya (Molnar et al., 2010). The onset of the monsoon is correlated with a distinct change in wind shear in which easterly winds flow aloft and westerly winds flow at the surface (Webster and Yang, 1992). This seasonal asymmetric flow is markedly different from the typical Hadley circulation between the tropics and subtropics and delivers substantial moisture and energy to the Asian landmass. 40 60 80 100 120 140 10 m s –1 10 m s –1 2 4 6 8 10 12 14 16 18 20 22 t s u g u A – e n u J y r a u r b e F – r e b m e c e D −20 −10 0 10 20 30 40 50 40 60 80 100 120 140 Latitude −20 −10 0 10 20 30 40 50 Precipitation and 850 hPa wind Precipitation (mm day –1 ) Figure 1.1: The seasonal changes in wind (850 hPa pressure level) and precipita- tion (1998-2006 TRMM 3B43V6 data set). Images are reprinted from Annu. Rev. Earth Planet. Sci., 38, Peter Molnar, William R. Boos, and David S. Battisti, Oro- graphic Controls on Climate and Paleoclimate of Asia: Thermal and Mechanical Roles for the Tibetan Plateau, 77-102. (2010). An important component of the Indian Summer Monsoon is the development of the Somali Jet, the western branch of monsoonal atmospheric circulation over 2 the Arabian Sea. The Somali Jet flows at ~1500 m and is channeled by the high topography of the Ethiopian Highlands (Chakraborty et al., 2009). The Somali jet travels eastward at 10-15 ◦ N across the Arabian Sea toward the western coast of India and delivers 2-4 m of summer rainfall over the north-south oriented mountain range of the Western Ghats (Molnar et al., 2010). This atmospheric pattern is important for the climates of northeast Africa by creating an anomalously dry semi-arid climate over the Horn of Africa as other regions along the same latitude promote lush tropical rainforests, such as the Congo rainforest in west-equatorial Africa. Traditionally, the monsoon has been described as a large-scale land-sea breeze caused by the seasonal differential heating between the broad and high Tibetan Plateau and the Indian Ocean (Kutzbach et al., 1989). However, recently the dynamical understanding of the monsoon has evolved. One issue with the land-sea breeze model was that the differential heating should result in a gradual onset of the monsoon, as opposed to its characteristic rapid onset. Aqua planet experi- ments found that the land-sea contrast is not necessary to drive a monsoon, rather that interactions between tropical circulation and extra-tropical eddy circulation can independently cause the rapid initiation of the monsoon (Bordoni & Schnei- der, 2008). Furthermore, observations of maximum summer heating south of the Tibetan Plateau called into question the role of the continental land mass as the dominant heat source driving summer monsoonal flow (Boos & Kuang, 2011). Other processes have been invoked shifting the importance from the high Tibetan Plateau to the insulating effect of the Himalayan mountain range for blocking cold, dry air from descending into the subtropics (Boos and Kuang, 2011). Under- standing the role of the monsoon and its relationship with the high and dynamic 3 topography of South Asia not only can be explored in the modern but also through thelensofpaleoclimateinvestigationsintoclimate-tectonicinteractionsinthepast. 1.2 South Asian Monsoon on tectonic timescales On tectonic timescales, the complex and multi-faceted nature of the South Asian monsoon presents a challenge to the paleoclimate community. Beginning with the timing of the collision between India and Eurasia, while controversial, likely began in the Eocene (Garzanti et al., 1987; Rowley, 1996; Najman et al., 2010). Terrestrial proxy records become more sparse and difficult to independently interpret further back in time, but some proxy records place the beginning of the Indian Summer Monsoon in the Eocene to coincide with the history of the uplift of the Tibetan plateau (Dupont-Nivet et al., 2007). However competing hypotheses of the initiation of include the uplift and growth of the Tibetan Plateau (Molnar et al., 1993), the uplift of the Greater Himalaya (Boos & Kuang, 2011), or the retreat of the shallow inland seas from Central Asia (Ramstein et al., 1997). 1.3 Miocene-Pliocene evolution of the South Asian Monsoon Lack of consensus also extends to when the monsoon intensified to its modern strength and whether this is coincident with the onset of the monsoon. Some proxy records suggest an early intensification of ~22 Ma (Clift et al., 2008; Guo et al., 2002) while others suggest that an intensification occurred between 8-7 Ma based on records of upwelling indices from the Arabian Sea (Kroon et al., 1991; Prell & Kutzbach, 1992). Terrestrial changes also document the emergence 4 and expansion of C 4 biomass in South Asia due to reduced precipitation since ~8 Ma (Cerling et al., 1997; Huang et al., 2007; Quade et al., 1989). While many paleoclimate records document a strong response or change in conditions ca. 8- 7 Ma, just how the monsoon changed at this time is ambiguous. Evidence for stronger winds are offered by the increased abundance of planktonic foraminifera, Globigerina bulloides, records from coastal Oman. Reduced sedimentation rates andweatheringintensityisobservedinvariousriverdrainagesfromregionsaffected by the South Asian Monsoon (Clift, 2010). Mollusk fossils from the Siwalik group, the same terrestrial archives that captured the late Miocene C 4 expansion, indicate that a strong seasonal cycle was in place by 10.7 Ma suggesting little variability in the strength of the monsoon since that time (Dettman et al., 2001). Furthermore, the intensification of monsoon winds at 8 Ma has been proposed to be an artifact of uplift of the Owen Ridge, suggesting that the classic records in the key monsoonal upwelling zone along the western coast of Oman may be a record of enhanced preservation as opposed to enhanced production of G. bulloides (Rodriguez et al., 2014). Indeed, studies since then using only foraminifera of large sizes found only modest changes in abundances and sensitivity to the depositional environments (Gupta et al., 2015; Huang et al., 2007). Part of the controversy is due to the spatially heterogeneous nature of the proxy records as well as the differences in the timing captured by these records. Atmospheric circulation over the Arabian Sea that would cause an enhancement of winds and increased upwelling results in moisture transported to the western Indian coast where it rains over the Western Ghats, limiting further rainfall over the interior. In contrast, northern India, Nepal and Bangladesh receive moisture from the Bay of Bengal. Therefore, wind strength over the Arabian Sea may not directly be reflected in enhanced rainfall over these core monsoon regions. 5 While records from the Arabian Sea present the most convincing evidence for the development of monsoonal winds near 8-7 Ma, records from the Bay of Bengal and terrestrial environments in the core of the monsoonal rainfall will reflect changes in precipitation. Therefore, it is important to take into consideration the different histories of the monsoon that the proxy records. High-resolution records from the Plio-Pleistocene indicate that Northern Hemi- sphere Glaciation affected the strength of the Indian Monsoon (Clemens et al., 1996). Lithogenic grain size decreased in sediment cores from upwelling regions in the Arabian Sea since the onset of the Northern Hemisphere ice sheets at ~3.5 Ma reflecting a weakened intensity of the Indian Summer Monsoon winds. Despite this long term trend, periodicity in paleoclimate records (δ 18 O, opal, lithogenic grain size, % G. bulloides) indicate that the strength of the Indian Summer Monsoon is dependent on orbital forcing, particularly by precession (23 kyr) and obliquity (41 kyr) cycles modulated by eccentricity (100-400 kyr) cycles (Clemens et al., 1996). This scale of variability is also observed in terrestrial records from the Gulf of Aden, reflecting a terrestrial response in vegetation and dust availability on the East African landscape (deMenocal, 1995; Feakins et al., 2005). During the Pliocene-Pleistocene, changing ocean gateways also likely played an important role in the strength of the Indian Monsoon. The opening and closing of deep-ocean gateways alter circulation and the transport of heat to and from the polar regions and the tropics (Cane & Molnar, 2001; Haug et al., 2001). In partic- ular, the shallowing of the seas surrounding the Maritime Continent of Indonesia is hypothesized to have prevented the influx of cool Artic waters into the Ara- bian Sea, altering atmospheric circulation and affecting the strength of the Indian Summer Monsoon (Cane & Molnar, 2001). 6 1.4 C 4 Expansion in the late Miocene Key to this region and many grassland and savanna biomes across the tropics and subtropics, the development of C 4 grasslands was an important ecological fea- tureofthelateMiocene. Quadeetal., (1989)firstdocumentedthedramaticexpan- sion of C 4 in the paleosols sampled from the proto Indus river in the Himalayan foreland in northern Pakistan (Fig. 2). This trend was replicated in many regions of the globe (Cerling et al., 1997). Initially a decline in pCO 2 was hypothesized to be the driver of this ecological transformation due to the success of C 4 in low CO 2 conditions (Cerling et al., 1997; Ehleringer et al., 1997). However, paleo CO 2 proxies do not indicate a decline in pCO 2 at this time. Reconstructing pCO 2 is notoriously challenging, and is an ongoing area of research. New alkenone-based reconstructions from the late Miocene display a shift in vital effects suggesting a decrease in pCO 2 conditions in the late Miocene (Bolton & Stoll, 2014). The pos- sible revitalization of this initial hypothesis linking the expansion of C 4 vegetation with declining pCO 2 in the Miocene may still be legitimate with the ongoing refine- ment of pCO 2 record and emergence of new techniques. In the absence of a clear single driver for late Miocene C 4 expansion, significant heterogeneity in regional patterns affecting C 4 expansion suggest that regional factors such as aridity, sea- sonality of precipitation, and/or disturbance (e.g. herbivory, fire) likely played an important role in amplifying or suppressing C 4 expansion. Some of the earliest evidence of C 4 on the landscape comes from Africa (Uno et al., 2011). Broad trends toward overall larger percentages of C 4 continue from the Miocene into the Pliocene amid large-scale variability (Feakins et al., 2005). The predominance of C 4 resources in the Pliocene and Pleistocene coincides with important advances in human evolution. During the Pliocene, bipedalism became 7 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 −15 −10 −5 0 Age (Ma) Siwaliks Tooth Enamel Paleosol more C 4 δ 13 C (‰, VPDB) Figure 1.2: Carbon isotopic records from the Siwalik Group in the Himalayan Foreland Basin. Black and white symbols refer to soil carbonate carbon isotope values where black symbols are from Nepal sequences Quade et al. (1995) and white symbols are from Pakistan sequences (Quade & Cerling, 1995). Tan symbols refer to carbon isotopes from fossil tooth enamel from the same sequences. Soil carbonates capture a distinct shift to 13 C-enriched values at ca. 7Ma while earlier signs of C 4 are suggested by herbivore tooth enamel records. (Nelson, 2005, 2007; Morgan et al., 1994; Cerling & Harris, 1999). All ages have been updated to GTS 2012. a dominant trait and hominin diets diversified compared with those of their ances- tors and other great apes (Levin et al., 2015). This manifested as morphological differences by 4.2 Ma (Levin et al., 2015). While the direct relationship between 8 the availability of C 4 resources and the evolution of Homo sapiens is currently untenable, C 4 was newly abundant on the landscape at similar time frames of key evolutionary development in hominin evolution. Testing potential evolution- ary pressures requires broad culling of information from ecological, climatic, and tectonic histories. 1.5 Marine sediments of the Arabian Sea 1.5.1 Gulf of Aden, DSDP Site 231 The Gulf of Aden is a young oceanic region situated between the Horn of Africa and the Arabian Peninsula. For the past 10 million years, Arabia has moved away from Somalia, forming new mafic crust from the seismically active spreading center, Sheba Ridge (Matthews et al., 1967). DSDP Site 231 is located near the southern shore of the Gulf of Aden, 80 km from Somalia in ~2160 meters water depth. Hemipelagic sediment was recovered and characterized by nearly uniform nannofossil ooze (Fisher et al., 1972). However, rotary drilling implemented at this time reduced recovery such that only 425 m of sediment were obtained from 584 depth (Fisher et al., 1972). 1.5.2 Indus Fan, IODP Site U1457 TheIndusFanliesonthepassivemarginbetweenPakistanandIndia. Spanning 1.1 x 106 km 2 (1,500 km length and 960 km maximum width) and depths greater than 9 km thick in the northern-most region, the Indus Fan is the second largest sedimentary fan in the world (CLift et al., 2001; Kolla & Coumes, 1987; McHargue & Webb, 1986). It is bounded to the south by the Carlsberge Ridge, to the west by the Owen-Murray Ridges, and to the east by Chacos-Laccadive Ridge. 9 The Laxmi Ridge, arguably considered a relic continental sliver remaining from the separation of India and the Seychelles, is a topographic high that separates the Laxmi Basin from the main fan feature (Miles et al., 1998). This region has largely been tectonically inactive as active seafloor spreading occurred southwest of Laxmi Ridge since the continental fragmentation of Gondwana at the end of the Cretaceous (Royer et al., 2002). The ultimate collision of India and Asia and the establishment of sufficiently high topography is evidenced by the presence of Pb isotopes indicative of the Indus Suture Zone in middle-Eocene fan sediments from Owen Ridge, which supports the initiation of Indus Fan sedimentation by this time (CLift et al., 2001). The Indus River is the primary source of sediment delivery to the Indus Fan. The Indus River is 2,900 km long and drains an area of 966,000 km 2 (Krish- nan, 1968). After draining the high elevation Himalaya and Karakoram regions, the Indus River passes ~1000-1,200 km through alluvial and deltaic plains before reaching the Arabian Sea. The extent of the continental shelf averages 100 km from the present-day delta shoreline (Giosan et al., 2006). The Indus canyon cre- ates a bathymetric break in the continental shelf, and evidence of at least two major canyons are observed to the west of the present day Indus canyon (Kolla & Coumes, 1987). Through these shelf breaks, sediment is delivered from the Indus River and continental shelf to the Indus Fan. The annual sediment discharge before damming of the Indus River in the 1950’s was ~250 x 10 6 T yr − 1 (Milliman et al., 1984), although this likely represents a minimum estimate (Giosan et al., 2006). ThishighsedimentdischargemakestheIndusRiverthe5 th largestsediment load in the world (Wells and Coleman, 1984), which can be attributed to the ero- sion of high-relief, rapidly uplifting mountains with source material characterized by unconsolidated glacial and fluvially-reworked detritus (Milliman et al., 1984; 10 Giosan et al., 2006). Peak river discharge of ~30,000 m 3 /s occurs seasonally due to both summer monsoonal rains and glacial meltwater from the Karakoram and Himalayas (Karim, 2002; Garzanti et al., 1987; Milliman et al., 1984). At times of active deposition to the east of the Laxmi Basin, contributions from peninsular India are possible by riverine contributions from the Naramada and Tapti Rivers. Deposition of the Indus River sediment load into the fan occurs as a series of channel/levee systems collectively referred to as the “Indus Fan Megasequence” (Droz & Bellaiche, 1991). Large, mud-rich fans are characterized by deep-water channels that travel far at low gradients before their turbidite systems give way to sheeted sand lobes in their distal areas (Kolla & Coumes, 1987). On passive margins, deposition into the fan is influenced by sea level in which periods of high deposition are associated with sea-level low stands when rivers are directly connected to the canyon heads and result in well-developed turbidite sequences. During sea-level high stands, deposition is dominated by mass-transport deposits andlesswelldevelopedturbiditesystems(Posamentier&Kolla,2003). Althoughit is possible for high sediment loads to bypass the continental shelf and travel basin- wards driven by climate pressures or if the river-canyon connection is maintained (Ducassou et al., 2009). 1.6 Plant wax abundance and carbon isotopes Plant biomarkers are widely used to reconstruct past environments. Specifi- cally, long chain lipids (n-alkanes and n-alkanoic acids) that form the waxy coating on plant leaves. There are multiple purposes of this waxy coating including pre- vention of water loss and protection against disease and ultraviolet light radiation 11 (Eglinton & Hamilton, 1967). These compounds are recalcitrant and readily pre- served in marine (Rommerskirchen et al., 2003), lacustrine (Garcin et al., 2014), and some terrestrial environments (Freeman & Colarusso, 2001) after transport via wind (Conte & Weber, 2002) or river (Ponton et al., 2014). Following deposition in one of these sedimentary archives, the compounds are initially subject to an active zone of degradation and are thereafter relatively resistant to microbial degradation (Meyers & Ishiwatari, 1993; Wakeham et al., 1997). The plant wax molecular abundance distribution as well as their isotopic value is indicative of their source vegetation. The interpretation of these signals in sed- imentary archives of plant wax builds on numerous studies based on living plants. Forplantwaxabundance, studieshavefoundthat, despitelargescatteracrossboth inter- and intra-species, some trends can be teased out. For example, long chain n-alkanes are generally more abundant in angiosperm trees and shrubs than in most gymnosperms, and while the n-alkane and n-alkanol molecular distribution and abundance varies between plant species, the abundance of n-alkanoic acids is generally more consistent between species of trees and shrubs (Bush & McInerney, 2013; Diefendorf et al., 2015, 2010). C 4 grasses have been observed to have higher concentrations of long chain n-alkanes (C 31 , C 33 , C 35 ) compared with C 3 grasses and woody angiosperms (Garcin et al., 2014; Rommerskirchen et al., 2006; Vogts et al., 2009). This observation has been helpful for interpreting late Miocene records of C 4 plant expansion (Uno et al., 2016). Furthermore, long chain n-alkanes C 33 may lend insight into the presence of C 4 grasses in grassland biomes (Garcin et al., 2014). The carbon isotopic value is the result of both environmental and biosynthetic processes. The carbon precursor to any plant product is CO 2 . During this initial step of carbon fixation, Rubisco strongly discriminates against 13 C resulting in 12 more negative δ 13 C values of plant products relative to the δ 13 C of atmospheric pCO 2 or other inorganic carbon found in the ocean, sediments or rock archives. The next largest fractionation effect is attributed to the photosynthetic pathway that the plant uses. The C 3 (Calvin-Benson) photosynthetic pathway has the largest net fractionation compared to the C 4 (Hatch-Slack) photosynthetic path- way. Additional factors that may influence the final carbon isotopic composition of the plant or plant waxes are environmental factors such as plant physiology, climate, and ecology. These factors must be considered when interpreting the car- bon isotopic values of the plant. As a result of these processes, plants that use the C 3 and C 4 photosynthetic pathways have isotopically distinct values, where modern C 3 plants range between -20 and -35%, and C 4 plants range between -10 and -14% (Fig. 3). Relative to the bulk plant, plant wax δ 13 C values are more negative due to lipid biosynthesis where simple organic precursor molecules are built into larger more complex lipids (Chikaraishi et al., 2004). Over long geologic timescales, it is important to constrain changes in the δ 13 C value of atmospheric CO 2 , particularly during the late Miocene when atmospheric pCO 2 concentrations were about 2% more enriched relative to preindustrial values. Tipple & Pagani (2010) provides a comprehensive review of Neogene changes in the δ 13 C value of the atmosphere based on benthic foraminiferal estimates. 1.7 Plant wax hydrogen isotopes Avoiding the plaguing problem of exchangeable hydrogen, covalently bonded H in plant wax lipids are valuable records of ancient environmental water. With the advent of compound-specific analytical techniques in the late 1990’s, plant wax molecules have provided important insights into the hydrological cycle that 13 Frequency C 3 plants 0 0 -5 -10 -15 -20 -25 -30 -35 -40 40 80 120 160 C 4 plants δ Cplant 13 Figure 1.3: C 3 and C 4 plants have distinctly different carbon isotope values. Image is reprinted from Annu. Rev. Earth Planet. Sci., 35, Brett J. Tipple and Mark Pagani, The Early Origins of Terrestrial C 4 Photosynthesis, 435-61. (2007). Data is from Cerling & Harris (1999) . build on nearly sixty years of research on water isotopes. This research provides important systematics of isotopes in precipitation as a result of atmospheric cir- culation and precipitation patterns (Craig, 1961; Dansgaard, 2016). The Global NetworkofIsotopesinPrecipitation(GNIP)monitorprecipitationisotopes, result- ing in decades worth of precipitation data that has been synthesized by Bowen & Wilkinson (2002). The isotopic signatures of precipitation are reflected by the δD value of plant wax as the source of hydrogen for the biosynthesis of lipids is the environmental water that the plants have access to, which likely originates from precipitation. 14 Numerous fractionation steps occur between soil-water and leaf-water, reflected by an overall net fractionation of the plant (Sachse et al., 2012). Controls on the plant wax δD values include (1) precipitation, (2) physiological and climatic influences, (3) life form and photosynthetic pathways, and (4) interspecies variability. Despite largevariabilityinthemodern,thetransportofplantwaxmoleculestosedimentary archives act to filter the signal from the landscape over temporal and spatial scales. This results in an attenuation of the ecosystem signal to reflect the dominant plant sources. In monsoonal regions, the intense summer rainfall results in distinctive D-depletion as a result of physical processes including Rayleigh distillation and other effects such as moisture source regions and orographic cooling of air masses also can result in D-depletion. This relationship has been established through time in the regions surrounding the Arabian Sea using theδ 18 O stalagmites (Fleitmann, 2003) and the δD of leaf wax n-alkanes (Huang et al., 2007). 1.8 Contents of this dissertation The motivation for the research presented in this dissertation was to docu- ment the dynamics between the climate and vegetation during the Miocene and Pliocene. These time periods also coincided with the expansion of C 4 vegetation, one of the most dramatic ecological transitions in the rock record (Edwards et al., 2010). In Chapter 2, Cooling and drying in northeast Africa across the Pliocene, I document the changes in vegetation, aridity and ocean temperatures over the Horn of Africa and in the Gulf of Aden from 5.3-2 Ma. Carbon isotopes of plant leaf wax are sampled at high resolution (3ka) to assess how precession affects ecological variability through C 3 /C 4 transitions. Hydrogen isotopes from the same plant leaf 15 waxes provide insight into the hydrological cycle. Ocean temperature reconstruc- tions are determined from TEX 86 and then compared to the broader context of Indian Ocean temperatures during the Pliocene. In Chapter 3, Late Miocene C 4 Expansion in the Indus River Catchment, plant leaf wax compounds, n-alkanoic acids and n-alkanes, were analyzed for their carbon and hydrogen isotopic compo- sition revealing important insights into what regional information is encoded in the different molecules including vegetation production (e.g. C 4 grasses) and sourcing effects (local vs. regional). This record is the result of turbidite deposition and reflects controls in sedimentary processes; however, important large-scale processes are revealed. In the last chapter, Photosynthetic pathway of grass fossils from the upper Miocene Dove Spring Formation, Mojave Desert, California, I address the sparse record of grass evolution. The rare preservation of grass fossils from Red Rock Canyon State Park, California, USA offer valuable insights into the history of grass evolution and shed new light on the oldest reported C 4 grass fossils. Finally I conclude with a chapter summarizing the major results from this research. 16 Chapter 2 Cooling and drying in northeast Africa across the Pliocene This manuscript was published in 2016 as: Liddy, H.M., Feakins, S.J., Tierney J., 2016. Cooling and drying in northeast Africa across the Pliocene. Earth and Planetary Science Letters 449, 430-438. doi:10.1016/j.epsl.2016.05.005 2.1 Abstract Terrestrial records suggest that Northeast Africa experienced drying during the Pliocene; however, these records are often incomplete in time and space, and open questions about this shift in climate remain. Here, we use marine sediments from Deep Sea Drilling Project (DSDP) Site 231 in the Gulf of Aden to generate a multi-proxy organic geochemical record of northeast African climate spanning 5.3-2 Ma. This new record provides a regional perspective on climate and serves as context for the fossil record of early hominin evolution. We measured leaf wax car- bon (δ 13 C wax ) and hydrogen (δD wax ) isotopic composition and TEX 86 (tetraether index of 86 carbons) to investigate past changes in vegetation, aridity, and ocean temperature, respectively. In the earliest Pliocene, we infer warm subsurface ocean temperatures from TEX 86 , semi-arid conditions on land and extensive C 4 grass- lands based onδD wax ,δ 13 C wax and previously published pollen. After 5 Ma, ocean 17 temperatures gradually cooled, and at 4.3 Ma there was a transition to arid condi- tions on land based onδD wax and previously published pollen. Grasslands yielded to a mid Pliocene landscape of dry shrublands. This drying appears to be an atmospheric response to cooling ocean temperatures, which may reflect changes in tropical ocean circulation, the intensification of Indian Monsoon winds or perhaps other changes associated with Pliocene cooling. 2.2 Introduction The late Pliocene witnessed cooling of global climate after the warm climates that characterized the early to mid Pliocene. Climatically relevant geological changes that ensued across the Pliocene are thought to have included changes in ocean circulation (e.g., Cane & Molnar (2001),Haug & Tiedemann (1998)), declining atmospheric carbon dioxide levels (e.g., Bartoli et al. (2011)), and the initiation of high latitude glacial cycles. It has been proposed that changes in sub- surface ocean temperatures (Cane & Molnar, 2001), surface temperature gradients (Brierley & Fedorov, 2010) as well as the onset of Northern Hemisphere Glacia- tion (deMenocal, 1995) may have led to a drying of east African climate. Here we focus on northeast African terrestrial environments during the Pliocene. We do so from the perspective of marine sediment downwind of northeast Africa, offering a regional sampling of terrestrial vegetation cover and climatic conditions, while also reconstructing ocean temperatures. Ocean conditions, rainfall changes, and the resulting vegetation responses, may have dramatically altered the terrestrial landscapes that were home to early human ancestors. 18 While most continents were in roughly present-day locations by 5 Ma, ongo- ing plate motions transformed tropical ocean gateways and uplifted new land- masses. The Central American Seaway closure was previously a candidate for climate change during the Pliocene (Haug & Tiedemann, 1998), but a more recent study has suggested completion in the mid Miocene (Montes et al., 2015). The shifting channels connecting the Pacific and Indian Oceans remain a viable mecha- nism for Pliocene climate change. Northward motion of the Indo-Australian plate (Dalyetal.,1991)constrictedandshiftedcirculation,andwashypothesizedtoalter the Indonesian Throughflow leading to Indian Ocean subsurface cooling ((Cane & Molnar, 2001); and see Molnar & Cronin (2015) for a recent review of the lit- erature on bathymetric changes and dates). This hypothesis posits that as the seaway became more restricted, cooler water from the North Pacific would feed into intermediate depths in the Indian Ocean, leading to cooler upwelled water in the Arabian Sea (Cane & Molnar, 2001). This would be expected to cause region- ally drier conditions, as cooler sea surface temperatures in the western Indian are generally associated with reduced rainfall in East Africa (Tierney & deMeno- cal, 2013). Exposure of more land on the emergent Maritime Continent (i.e., the Indonesian archipelago and nearby shallow seas) associated with tectonic motion of the Indo-Australian plate may have additional climatic implications, including a strengthening of the Walker Circulation (Molnar & Cronin, 2015), with conse- quences for African climate. In addition, the uplift of fresh basaltic crust and exposure to chemical weathering in tropical latitudes has been proposed as a pos- sible mechanism for pCO 2 drawdown and global cooling trends across the last 5 Myr (Molnar & Cronin, 2015). Indian Ocean temperature reconstructions have found signs of the predicted cooling across the Pliocene. In particular, changes appear to have occurred in the 19 subsurface, with a freshening at 4.3 Ma and cooling between 3.5 Ma and 3.0 Ma evident at DSDP Site 214 based on foraminifera Globorotalia crassaformis δ 18 O and Mg/Ca (Karas et al., 2011). This shift in subsurface conditions in the east- ern Indian Ocean occurs close to the timing of emergence of Timor (Nguyen et al., 2013). Cooling in the Leewin Current (Karas et al., 2011) and the Benguela Current (Marlow et al., 2000) have been linked to this transition, indicating the broad regional implications of Indonesian Throughflow changes. In the northwest- ern Indian Ocean, in the upwelling region of the Oman margin, Pliocene cooling has also been detected by alkenone reconstructions (Huang et al., 2007), but the cooling may be underestimated as the proxy is close to its upper limit of 28 ◦ C in the early Pliocene. Additional information comes from alkenone and total organic carbon contents in the sediments (Huang et al., 2007) that both indicate increased productivity between 4 and 3 Ma, although Globigerina bulloides, often assumed to be an upwelling indicator (Gupta et al., 2015), decreases at the same time. The organic productivity rise may be consistent with intensification of monsoonal windsovertheArabianSeaenhancingupwelling. Morepuzzlingistheforaminiferal Mg/Ca record from the central western Indian Ocean ODP Site 709 (Fig. 1) that shows sea surface temperatures (SST) warming out of the early Pliocene warm period (Karas et al., 2011), while global temperatures cooled. However, questions about Mg/Ca proxy fidelity associated with possible Mg/Ca seawater chemistry changes have elsewhere suggested upward revision of early Pliocene SST estimates (O’Brien et al., 2014). Delving further into Indian Ocean changes is of interest not only to resolve general questions of proxy fidelity (O’Brien et al., 2014), but also because those temperature patterns influence precipitation changes in East Africa (Tierney & deMenocal, 2013). 20 ! ! ! 70°E 60°E 50°E 40°E 20°N 10°N 0° El. (km) >2.5 2 1.5 1 0.5 0 ODP SITE 709C ODP SITE 722 DSDP SITE 231 Lothagam Afar Triangle S. Ethiopian Escarpment Red Sea ! Indian Ocean Hadar Figure 2.1: Map of DSDP Site 231 in the Gulf of Aden (11.89 ◦ N, 48.25 ◦ E, 2152 m water depth) and locations of SST records and terrestrial sites discussed in the text. East African rainfall changes, and resulting vegetation responses, may have dramatically altered the terrestrial landscapes that were home to species of the hominin family tree, including Ardipithecus ramidus and Australopithecus afaren- sis. Those fossil specimens, as well as the wider record of hominin fossil fragments, dentition and most recently a surge of information on their dietary preferences (Cerling et al., 2013; Levin, 2015; Wynn et al., 2013) provide anthropological 21 interest motivating climate and vegetation reconstructions. Existing marine and terrestrial records (see Levin, (2015) for a review) leave plenty of room to fill in additional details of northeast African environments during the Pliocene. How- ever, the active tectonics in the East African Rift Valley graben includes complex changes in topography, hydrology and lake formation (Maslin et al., 2014), which may not be clearly linked to regional precipitation change. Marine sediments of the Gulf of Aden provide a downwind repository of ter- restrial proxies blown off northeast Africa, and have yielded valuable paleoclimate evidence from analyses of dust (deMenocal, 1995), pollen (Bonnefille, 2010), and plant leaf waxes (Feakins et al., 2013; Tierney & deMenocal, 2013), over various timescales from Miocene to recent. Here, we present a study of Pliocene climate based on organic geochemical analyses of sediment from DSDP Site 231. Using δ 13 C wax and δD wax and previously published pollen data (Bonnefille, 2010), we assess how changes in northeast African hydrology influenced vegetation cover. To test the role of Indian Ocean temperatures on northeast African hydrology, we generated a TEX 86 record from the Gulf of Aden. We evaluate changes in western Indian Ocean SSTs based on prior alkenone (Huang et al., 2007; Herbert et al., 2010) and Mg/Ca (Karas et al., 2011) reconstructions from the Arabian Sea and western Indian Ocean. We seek to connect oceanographic conditions and East African climate through an examination of ocean temperatures and regionally significant terrestrial environmental reconstructions in the same sediments. Specif- ically, we ask whether the ocean adjacent to Africa cooled during the Pliocene and whether there was, at the same time, a regional drying of terrestrial environments. 22 2.3 Regional Climate 2.3.1 Precipitation amount and isotopic composition The Horn of Africa today receives 100-200 mm yr-1 of precipitation (Nicholson, 2000). Rainfall occurs during the biannual passage of the ITCZ with long rains from March to May and short rains from October to November (Nicholson, 2000). Precipitation isotopes (stable oxygen and hydrogen isotopes; δ 18 O and δD) act as tracers for moisture sources. Moisture derived from the Indian Ocean is generally more depleted in the heavier isotopes by Rayleigh distillation from prior rainout, while rare incursions of westerly moisture are isotopically enriched due to the recy- cling of moisture in west African rainforests (Levin et al., 2009). Isotope-enabled climate model simulations support the use of hydrogen isotopes as a hydroclimate indicator in the region (Tierney et al., 2011). Although there is a paucity of precip- itation isotope collections in this region, theory and model simulations suggest that lower values ofδD in precipitation indicate wetter conditions, i.e. more prior rain- out from Indian Ocean sources. Conversely, higher values of δD occur as a result of less antecedent rainout and increased evaporative enrichment during raindrop descent in arid conditions, especially in the Afar Triangle. 2.3.2 Atmosphericcirculationandtransportofwind-blown proxies Atmospheric circulation over northeast Africa is dominated by the seasonal reversal of monsoon winds. During the boreal winter, northwesterly wind speeds average 2-4 m s −1 , whereas from May to September southwesterly winds of the Somali Jet in the lower troposphere can reach wind speeds of 30 m s −1 (Ramage et 23 al., 1972). The summer season is thus expected to dominate the year round wind- transport of terrigenous material to the Gulf of Aden. The Somali Jet is steered by the Ethiopian Highlands, and this circulation has likely been prevalent throughout the Pliocene as the surface uplift of the Highlands predates the Pliocene (Wichura et al., 2010). Observations of dust with the Earth Probe Total Ozone Mapping Spectrometer (TOMS) suggest that terrestrial sources to the Gulf of Aden include the Somali peninsula and Afar Triangle (Prospero, 2002). Pollen taxa from DSDP Site 231 further constrain the source region to the South Ethiopian escarpment, Afar Triangle and northerly contributions from the Horn of Africa (Bonnefille, 2010;Feakinsetal.,2013). Similarsourceregionsareassumedforwind-transported leaf waxes and pollen, although the mechanism for dispersal differs. 2.3.3 Oceanography of the Gulf of Aden The seasonal wind reversal determines ocean circulation and exchange between the Red Sea, Gulf of Aden and open Indian Ocean. In winter, northeasterly winds drive surface currents from the Indian Ocean into the Gulf of Aden and into the Red Sea. An undercurrent of outflow from the Red Sea carries a high salinity water mass (36.5 psu, 1σ = 0.3 psu) into the Gulf of Aden. During the summer, winds and surface currents reverse so that Red Sea outflow occurs in surface waters. At the location of DSDP Site 231 (Fig. 1) SSTs vary between 25.5 and 30.7 ◦ C seasonally (Locarnini et al., 2010), with salinity of 36.1-36.6 psu (Antonov et al., 2010). Subsurface water at 50-200 m depth (18.6 ◦ C, 1σ = 3.6 ◦ C, 35.8 psu, 1σ = 0.2 psu) is similar to the open Indian Ocean. Red Sea Overflow Water characterized by high salinity (36.5 psu, 1σ = 0.3 psu) lies at a depth of 400-900 m, underlain by deep waters similar to those of the open Indian Ocean (Antonov et al., 2010; Locarnini et al., 2010). 24 2.3.4 Modern vegetation distribution in northeast Africa Today, Northeast Africa is dominated by dry shrubland, grasslands and <10% tree cover (Linder, 2014). The Horn of Africa is sparsely vegetated with a low species diversity (Linder, 2014) characterized by Acacia-Commiphora bushland (Bonnefille, 2010). The Ethiopian Highlands are cooler, wetter and forested and not a major source of leaf waxes or pollen to DSDP Site 231 (Feakins et al., 2013; Bonnefille, 2010). The similarity between modern vegetation and fossil pollen taxa in the Gulf of Aden marine sediment core (DSDP Site 231) indicate that the major present-daybiomesandspecieshavebeenpresentsincethemid-Mioceneinvarying proportions (Bonnefille, 2010). 2.4 Materials and Methods 2.4.1 Marine sediments We sampled marine sediments from DPSP Site 231 in the Gulf of Aden (11.89 ◦ N, 48.25 ◦ E, 2152 m water depth; Fig. 1) with contiguous 10 cm scrape samples from 181.5 to 254.4 mbsf spanning 5.3 and 2 Ma, with each sample inte- grating 3,000 years (3 ka). In total, 744 samples were analyzed for carbon isotopic compositions. A subset of 106 samples were analyzed for hydrogen isotopic com- positions at approximately 30 ka intervals. For TEX 86 , we analyzed 215 samples at an average resolution of every 15 ka. Hydrogen isotopes were sampled at the lowest resolution due to reduced sample amounts after carbon isotopic analyses. Age control is derived from linear interpolation of tephrostratigraphic and nanno- fossil datums (Feakins et al., 2013). Although the sedimentation rate appears to be constant throughout 12 Myr, age control in the Pliocene is hampered by the 25 lack of suitable tephrostratigraphic horizons, despite extensive searches for trace tephra, in the mid and late Pliocene (Feakins et al., 2007) and beyond to the early Pliocene. We do not attempt to resolve orbital variability in our reconstruc- tions as the gaps in DSDP rotary-drilled sediment recovery preclude precise age control. Sampling resolution integrating 3 ka should be sufficient to capture vari- ability across orbital cyclicity such as that associated with precession, however as the precise timing is not known we do not interpret such variability. For δD and TEX 86 , we only analyze a subset of the samples, which may alias high frequency variability. For all proxies we focus our interpretations on the long-term trends. 2.4.2 Lipid Extraction Sediment samples were freeze dried, crushed and extracted with an Accel- erated Solvent Extraction system (ASE 350 R , DIONEX) with 9:1 ratio of dichloromethane (DCM): methanol (MeOH) at 100 ◦ C and 1500 psi. Extracts were separated over NH 2 -coated silica gel columns eluted with 2:1 ratio of DCM: isopropanol (neutral) and 4% formic acid in diethyl ether (acid fraction). The acid fraction was methylated with methanol of known isotopic composition with 95:5 MeOH: hydrochloric acid at 70 ◦ C for 12 hours. Methylated products were extracted using liquid-liquid extraction with 1 mL milli-Q water and hexane, and the hexane extract was passed through anhydrous sodium sulfate. Samples were further purified over silica gel column eluting with hexane and DCM resulting in non-polar and FAME fractions, respectively. 26 2.4.3 TEX 86 analysis The neutral fractions were dissolved in hexane:isopropanol (99:1) and filtered through a 0.45 micron PTFE filter prior to injection on an Agilent 1260 High- Performance Liquid Chromatography (HPLC) coupled to an Agilent 6120 mass spectrometer. Glyceroldialkylglyceroltetraethers(GDGTs)wereanalyzedaccord- ing to the method of Schouten et al., (2007). Briefly, GDGTs were separated on a Grace Cyano column with a 5 minute isocratic elution using A, and then a gradient of 90% A/10% B to 82% A/18% B over 35 minutes, where A = hexane and B = hexane:isopropanol (9:1). The column was then backflushed with 100% B for 10 minutes, and re-equilibrated with90%A/10%B foranother10minutes. SingleIon Monitoring (SIM) mode of the M + H + ions was employed to detect and quantify the isoprenoidal GDGTs produced by marine archaea including GDGTs with 0-3 cyclopentane moieties (GDGT-0 to GDGT-3); crenarchaeol (Cren) with an addi- tional cyclohexane moiety and its regioisomer crenarchaeolÂť (CrenÂť). Following Schouten et al. (2007b), TEX 86 units were calculated using the equation: TEX 86 = [GDGT − 2] + [GDGT − 3] + [Cren 0 ] [GDGT − 1] + [GDGT − 2] + [GDGT − 3] + [Cren 0 ] (2.1) Repeat analyses of a laboratory standard yields a long-term precision of 0.004 TEX 86 units. We converted the TEX 86 record to integrated subsurface (0-200 meters) temperature using BAYSPAR, the spatially-varying Bayesian regression approach (Tierney & Tingley, 2014, 2015). For our prior distribution, we chose modern mean annual integrated subsurface temperatures (0-200 m) with a gamma- weighted(a=4.5, b=15)peakcontributionat50mdepthatthecoresite(22.4 ◦ C) and a standard deviation of 10 ◦ C. To monitor for possible terrestrial soil inputs 27 we calculated the BIT index, a measure of the proportional abundance of non- isoprenoidal GDGTs-I to III (soil bacteria-derived) relative to marine crenarchaeol (see Fig. S5). Archaeal communities differ between the open Indian Ocean and those endemic to the northern Red Sea in their membrane response to temperature (Trommer et al., 2009; Tierney & Tingley, 2014). In the Gulf of Aden, however, coretop TEX 86 measurements fall within the global coretop calibration and show no evidence of influence from the northern Red Sea (Trommer et al., 2009; Tierney & Tingley, 2014). 2.4.4 Compound specific carbon and hydrogen isotopic analysis Thecompoundspecificcarbonandhydrogenisotopiccompositionsoflongchain C 28 to C 32 n-alkanoic acids were analyzed using a Thermo Scientific Trace GC equipped with a Rxi R -5 ms column (30 m × 0.25 mm, film thickness 1μm) with a PTV injector operated in solvent-split mode, coupled to a Delta V Plus iso- tope ratio mass spectrometer (IRMS) via an Isolink combustion/pyrolysis furnace (1000/1400 ◦ C). Each sample was run with an internal standard of known isotopic value and bracketed with CO 2 /H 2 reference peaks, two of which were used for stan- dardization. Forδ 13 C,areplicatewasrunforeveryfifthsample; themeanstandard deviation of replicates is 0.4%. For δD, every sample was run in duplicate, and the mean standard deviation of replicates is 1.0%. Samples were interspersed with external standard runs. For carbon, standards included a mixture of eight fatty acid methyl esters with δ 13 C values ranging from −30.92 to −23.24% or a mixture of 15 n-alkanes (C 16 to C 30 ) withδ 13 C values ranging from 33.3 to 26.2% 28 and δD values ranging from −254.1 to −9.1% (F 8 and A mix standards sup- plied by A. Schimmelmann, Indiana University, USA). The isotopic composition of MeOH was determined by offline combustion and dual-inlet IRMS (δ 13 C MeOH = −25.24 ± 0.43%, n=3), and by the esterification of phthalic acid for analysis by GC-IRMS (δD MeOH = −198.3 ± 3.9%, n=7). The addition of the methyl group was corrected for by mass balance. The H 3 factor was monitored daily and averaged 6.34 (1σ = 0.88) over a range of 1-9 V. To monitor the source and degradation of the long-chain fatty acids, we also measured the carbon preference index (CPI), calculated as CPI = 1 2 + P [C 25−35odd ] P [C 24−34even ] + P [C 25−35odd ] P [C 26−36even ] (2.2) CPI is high (7.0 ± 2.5) indicating confidence in a terrestrial higher plant source. For carbon and hydrogen isotopic analysis, we report results for the most abundant compound, the C 28 n-alkanoic acid, hereafter δ 13 C wax and δD wax . We also report carbon isotopic compositions for the C 30 and C 32 n-alkanoic acids where possible (Table S2, Fig. S1). 2.5 Results and Discussion 2.5.1 Warm early Pliocene supported C 4 grasslands In this study, we use and interpret TEX 86 as a subsurface temperature indi- cator. Although TEX 86 was initially developed as a SST proxy, the organisms that produce GDGTs, marine Thaumarchaeota, are nitrifiers and tend to inhabit somewhat deeper waters (ca. 50-200 m) especially in upwelling zones (Schouten et al., 2012). Insofar as subsurface temperatures typically correlate with surface 29 variability, TEX 86 may still be used as a SST proxy. However, on longer timescales under changing boundary conditions, this assumption likely breaks down. TEX 86 data from the Arabian Sea and the Gulf of Aden show clear evidence of subsurface temperature behavior across the last deglaciation (Tierney et al., 2016); hence we chose to use the subsurface TEX 86 calibration (Tierney & Tingley, 2015). The sub-T calibration converts TEX 86 values into a function-weighted distribution of temperatures from 0-200 meters, with a maximum near 50 meters (Tierney & Tingley, 2015). During the early Pliocene, we find subsurface temperatures in the Gulf of Aden at DSDP Site 231 that are similar to modern temperatures (Fig. 2a). Mea- sured TEX 86 values of 0.71 at 5 Ma yield temperature estimates of 22.6 ◦ C (1σ = ±2.2 ◦ C), which is close to the modern gamma-weighted mean annual value (22.4 ◦ C, 0-200 m) (Locarnini et al., 2010; Tierney & Tingley, 2015). While the error bounds on this estimate are large, they include propagation of both cali- bration and analytical uncertainties, i.e. more comprehensive than reported for many other techniques. Within the error estimates, the early Pliocene subsurface temperatures are warmer than the early Pleistocene temperatures and are compa- rable to late Holocene subsurface temperatures in the Gulf of Aden (Tierney et al., 2016). Warm subsurface temperatures relative to the late Pleistocene imply that the thermocline was warmer and perhaps deeper in the early Pliocene, possibly due to weaker monsoonal atmospheric circulation or as may have been typical of deep tropical thermoclines during early Pliocene warmth (Philander & Fedorov, 2003). The early Pliocene was wetter in northeast Africa than the late Pliocene based upon our evidence from plant wax δD values from DSDP Site 231. Between 5.3 and 4.3 Ma (Fig. 2b), δD wax values were depleted (ca. -150%, but as low as 30 2 3 4 5 TEX86 Sub-T (°C) Age (Ma) 16 18 20 22 24 26 δDwax (‰) −170 −160 −150 −140 −130 −120 C D −29 −27 −25 −23 δ Cwax (‰) 13 0 20 40 60 80 100 Pollen (%) Amaranthaceae Grass A B Figure 2.2: Comparison of records from marine sediment core DSDP Site 231. (A) TEX 86 calibrated using BAYSPAR for subsurface temperature (red line; 1σ calibration uncertainty, shading). (B) plant leaf waxδD record of n-C 28 acid (blue line; 1σ instrumentprecision, shadingandblackline; lowpass(>75kyr)filter). (C) plant leaf waxδ 13 C record of n-C 30 acid (green line; 1σ instrument precision, error bars and black line; low pass (>75 kyr) filter) (D) Pollen percentages of Poaceae (grass) and Amaranthaceae (shrub) calculated relative to total pollen abundances (Bonnefille, 2010; Feakins, 2013) 31 -165%), relative to later in the Pliocene (ca. -130%). We infer that the early Pliocene was relatively wet (enhanced rainout and more D-depleted precipitation) than the late Pliocene. If we compare values to the more familiar Holocene, we find the early Pliocene slightly wetter than the African Humid Period (10-5 ka at ca. -45%; (Tierney & deMenocal, 2013)) and the late Pliocene analogous to the late Holocene (ca. -135%;(Tierney & deMenocal, 2013)). We reconstruct terrestrial vegetation based on a combination of leaf wax evi- dence for photosynthetic pathway and pollen (Bonnefille, 2010), which informs on vegetation type. In the early Pliocene (5.3-4.7 Ma), δ 13 C wax values are 13C- enriched (ca. -22.8%). In modern African plants,δ 13 C wax is ca. -19.9% (σ = 2.8; n = 44) for C 27 n-alkanes from C 4 grasses and ca. -33.6% (σ = 3.5; n = 79) for C 27 n-alkanes from savanna and rainforest C 3 tree ecosystems, with more deple- tion with increasing canopy closure (Garcin et al., 2014). Carbon isotopes cannot therefore be unambiguously interpreted in terms of ecosystem or C 4 grassland pro- portions, and pollen is extremely helpful to elucidate the vegetation reconstruction (Feakins et al., 2013). In the early Pliocene pollen record, grass pollen percentages are high (33-45%), and arid-adapted shrubs (Amaranthaceae) pollen are generally low (Fig. 2c, d). We interpret δ 13 C wax and pollen evidence as indication that the early Pliocene proxies reflect a mixed vegetation community dominated by C 4 grasslands, although longer records show that grasslands (up to 59% grass pollen) and C 4 plants (δ 13 C wax up to -22.3%) have been more prevalent at other times in the Miocene and Pleistocene (Bonnefille, 2010; Feakins et al., 2013). Although many assume that grasslands prevail in arid climates, in fact some shrubs, in particular Amaranthaceae, thrive in yet more arid climates than those where C 4 grasses are common. Together the proxy records (Fig. 2) indicate that the early Pliocene was wet enough to sustain C 4 grasslands typical of semi-arid 32 climates, whereas the late Pliocene was dominated by the Amaranthaceae, today typical of the arid areas of northeast Africa. Today C 4 grasslands peak in pro- portional abundance at 450 mm a −1 (Guan et al., 2012), and if these modern grasslands are analogous, then this provides a quantitative guide to conditions sustaining C 4 grasslands in the early Pliocene. 2.5.2 Drying at 4.3 Ma We find a transition at 4.3 Ma when δD wax increased by as much as +42% (to ca. -122%), with values similar to âĂŞ130% persisting throughout the rest of the record to 2 Ma. The D-enriched δD wax signal in the 4.3 to 2 Ma period likely records lower precipitation amounts (Levin et al., 2009; Tierney et al., 2011). This interpretation is supported by the pollen record of Bonnefille (2010), which shows an increase in the proportion of arid taxa, Amaranthaceae, indicative of xeric shrublands (Fig. 2). These lines of evidence are not completely indepen- dent as arid conditions could produce D-enrichment through smaller fractiona- tions between the isotopic composition of precipitation and leaf wax as we shift from grasses with larger fractionations to xeric shrubs with smaller fractionations (Sachse et al., 2012). As the vegetation-based isotopic effect could be as large as the observed δD wax shift, it remains possible that the δD wax values reflect the vegetation fractionation without any change in the isotopic composition. However, the pollen provides solid evidence for drying on the landscape. We therefore infer a precipitation isotope shift, associated with a decline in rainfall amounts. From 4.9 to 3.8 Ma, δ 13 C wax values trend from ca. -22.9% toward more 13C- depleted isotopic values (ca. -25.8%, with a low of -27.2% at 3.8 Ma). This could reflect a contraction of C 4 grassland and increase of either C 3 forest (wet) or shrubland (dry). Pollen again resolves the ambiguity: Amaranthaceae display a 33 coevalincreasefrom6%to70%, whilegrasspollendecreasesfrom43%to15%(Fig. 2) (Bonnefille, 2010). Together the pollen and leaf wax δ 13 C evidence indicate a contraction of C 4 grasslands and replacement by C 3 desert shrublands. Animal diets corroborate this reconstruction: hippopotamuses show a shift to eating less C 4 between two deposits dating to 6.5 and 4.2 respectively from the Lothagam site in Turkana region of the Rift Valley (Uno et al., 2011). From 3.8 to 3 Ma, δ 13 C wax values return to values ca. -22.8% (Fig. 2). This 13C-enrichment cannot be attributed to grasses, as Amaranthaceae shrubs continue to dominate with up to 90% of total pollen abundance (Fig. 2). The combined pollen and isotopic evidence indicates that it was too dry to support widespread grasslands. Either the isotopic shift must be confined to increasing water stress in C 3 plants or C 4 must either be appearing within the shrubs or in localized pockets of grasslands. Despite regional dry conditions, from 3.6 to 3 Ma there were large lakes at Turkana (Lokochot) and in the Ethiopian Rift and Awash Basins (Maslin et al., 2014). This paints a stark contrast between water resources in the East African Rift Valley and desert shrublands beyond. It has previously been suggested that the onset of Northern Hemisphere glacia- tion from ca. 2.7 Ma promoted a drying of northeast African environments (deMenocal, 1995). Here we find no evidence of a shift at this time. The combined δD wax and pollen data suggest an aridity shift occurred at 4.3 Ma and no major changes occurred between 3.3 and 2.7 Ma when high-latitude cooling led to the first major Northern Hemisphere glaciations. 2.5.3 Indian Ocean influence on east African aridity Now we turn to conditions in the western Indian Ocean for possible proximal causes of drying. Today warm western Indian Ocean, cool eastern Indian Ocean 34 and cool western Pacific SSTs generate wet conditions over east Africa (Tierney et al., 2013). To assess the broader spatial pattern of SSTs in the western Indian Ocean basin, we compare our Gulf of Aden TEX 86 records to Pliocene-Pleistocene alkenone records from ODP Site 722 in the northern Arabian Sea off the Omani coast (Herbert et al., 2010; Huang et al., 2007) and a Mg/Ca SST record from ODP Site 709C from the central tropical Indian Ocean (Karas et al., 2011) (Fig. 3, Fig. S3-S4). 2.5.4 TEX 86 Our TEX 86 based subsurface temperatures cool monotonically from 5.3 to 2 Ma by 4.4 ◦ C (TEX 86 values decrease from 0.73 to 0.62), yielding subsurface tempera- ture reconstructions that decrease from an average 22.6 ◦ C (1σ = 2.2 ◦ C) at 5 Ma to 18.0 ◦ C (1σ = 2.2 ◦ C) at 2.2 Ma (Fig. 3). Contribution of terrestrial GDGT lipids to the marine signal is negligible as BIT index values are low (0.06±0.01) (Fig. S5), suggesting soil inputs were minimal, and that rivers were likely internally draining during the Pliocene, as they are today. 2.5.5 Alkenones We combined the two available Pliocene alkenone SST records from ODP Site 722 (see supplementary text and Fig. S2; Herbert et al., 2010; Huang et al., 2007). Similar to the 4.4 ◦ C cooling observed in the TEX 86 -SST trend in the Gulf of Aden, U37k’-SSTs cool by 3 ◦ C from 29 ◦ C to 26 ◦ C in the northern Arabian Sea (ODP Site 722) from 5.3 to 2 Ma with uncertainties on calibrations on the order of ±2.0 ◦ C (Fig. 3), making the cooling trend only slightly larger than the uncertainty on the calibration. We note that as the U k 0 37 index approaches saturation (ca. 28 ◦ C), the temperature response becomes nonlinear. Therefore it is possible that the 35 ODP Site 722 record is underestimating the full magnitude of temperature change throughout the Pliocene as temperatures may have been warmer than 29 ◦ C in the early Pliocene. 2.5.6 Mg/Ca The Mg/Ca content of planktonic foraminifera Globigerinoides sacculifer from ODP Site 709C (Fig. S4) reveals a slight warming from 5 to 2 Ma (Karas et al., 2011). Mg/Ca seawater adjusted estimates (SST sw−corr ; see Fig. S3 and S4 and supplemental text for explanation and methods) suggest reconstructed tem- peratures that are 1.3 ◦ C warmer than the uncorrected values yielding similar to modern SSTs (28.6±0.7 ◦ C). SST sw−corr estimates maintain a slight warming trend from 4.8-4 Ma (from 25 ◦ C to 27.5 ◦ C) followed by relatively stable SSTs from 4-2 Ma (ca. 27.5 ◦ C; Fig. S4). 2.5.7 Analyzing trends in Pliocene Indian Ocean temper- atures To facilitate comparison of trends in ocean temperature reconstructions that may include surface and subsurface temperature recorders, we compare the records as temperature anomalies, similar to the approach used in Tierney et al., (2016). We define anomalies relative to the early Pliocene (5-4 Ma) mean value from the proxy records at each core location (Fig. 4). This approach allows us to avoid some of the uncertainties of depth habitat and absolute temperature calibration issues but does not exclude temporal changes in seawater chemistry, temperature responses or depth habitats. We find that western tropical Indian Ocean SSTs remained within 1 ◦ C of early Pliocene values while subsurface temperatures in the 36 20 22 24 26 28 30 32 U37 SST (°C) k’ more polar ice Restriction of Indonesian Throughflow ONHG δ 18 O (‰) Benthic 2 3 4 5 ODP 722 A D 2 3 4 5 Age (Ma) TEX86 Sub-T (°C) DSDP 231 16 18 20 22 24 26 B C E 0 1 2 3 4 5 Corg (%) 0 2000 4000 6000 8000 Alkenone (ng gdw ) -1 Figure 2.3: Comparison of global temperatures, northern Indian Ocean temper- atures and upwelling indices. (A) Benthic foraminiferal δ 18 O values that reflects changing deep sea and high latitude temperature and ice volume (Zachos et al., 2001). Labeled arrows indicate major global climatic events and ocean circulation eventsdiscussedandreferencedinthetext,includingonsetofNorthernHemisphere glaciations (ONHG). (B) TEX 86 BAYSPAR calibrated subsurface temperatures at DSDP Site 231 from the Gulf of Aden (red line; 1σ calibration uncertainty, shad- ing). The red diamond indicates the gamma-weighted average of 0-200 m water depth at DSDP Site 231 of 22.4 ◦ C (Locarnini et al., 2010; Tierney & Tingley, 2015). (C) Alkenone-based sea surface temperatures at ODP Site 722 from the Arabian Sea (blue; ODP 722 combined record; Fig. S2; (Herbert et al., 2010; Huang et al., 2007) with various SST calibrations (as described in supplemental text). Shading represents 1σ of the calibration error. The blue diamond indicates themodernmeanannualseasurfacetemperatureof26.9 ◦ C(Locarninietal.,2010). (D) Alkenone abundance (ng gdw-1) at ODP Site 722 (Huang et al., 2007). (E) Percent of organic carbon at ODP Site 722 (Huang et al., 2007). 37 Gulf of Aden (DSDP Site 231) and SSTs near the Omani Margin (ODP Site 722) cooled by ca. 3 ◦ C by 2 Ma (Fig. 4). Mg/Ca SSTs in the western Indian Ocean at ODP Site 709C increase by 2 ◦ C out of the early Pliocene warm period and remain stable throughout the Pliocene into the Pleistocene (Fig. 4), even after corrections for changing seawater Mg/Ca (section 4.4.3). This may mean the warming trend is robust to seawater changes or that other factors such as diagenesis, salinity, or pH effects (see supplementary text) may mask the cooling that is expected based on the other records. 2 3 4 5 −6 −5 −4 −3 −2 −1 0 1 2 3 Temperature Anomaly (°C) Age (Ma) Figure 2.4: Indian Ocean temperature anomalies relative to 5-4Ma. DSDP Site 231 TEX 86 subsurface temperatures (red line; this study), ODP Site 722 alkenone based SSTs (blue line; Herbert et al., 2010; Huang et al., 2007) and ODP Site 709C Mg/Ca sw−corr based SSTs (black line; after Karas et al., 2011, modified with sw-corr). Data are plotted using the calibrations and corrections described in the supplementary text. Site locations are shown in Fig. 1. Cooling at the margins of the Indian Ocean, observed in both the subsurface record at DSDP Site 231 and the SST record from ODP Site 722 could reflect 1) intensification of upwelling and/or 2) cooling of subsurface upwelled water masses. Both are possible in the Pliocene for a variety of reasons. First, let us consider 38 upwelling (1). Intensification of summer-season upwelling, without cooling of the associated water mass, may be sufficient to explain the observed cooling. Such upwelling-wind strength feedbacks may be linked to the Somali Jet and may have contributed to increasing moisture export out of northeast Africa and the observed drying (Brierley & Fedorov, 2010). Total organic carbon and alkenone abundances increase (Fig. 3d, e) across the same period (Huang et al., 2007), and may be evi- denceforincreasedproductivityandintensificationofupwellingacrossthisinterval. While abundances of the planktonic foraminifer, G. bulloides, have previously been usedtoreconstructupwellingintensity, amorerecentstudyfrommultiplelocations on the Oman Margin indicate spatial differences in production and preservation (Gupta et al., 2015). Thus we consider the organic content as more robust evidence for upwelling than G. bulloides. Alternatively (2), cooling of subsurface water masses during the Pliocene could entirely or partially explain the cooling trend in the upwelling areas and margins of the Indian Ocean. The subsurface cooling to 20-18 ◦ C after 3 Ma is comparable to Gulf of Aden subsurface temperatures during the last glacial period (Tierney et al., 2016). The cooling to this temperature could be associated with a more widespread phenomenon discussed by Philander & Fedorov (2003) with a shoaling of the thermocline perhaps related to slow changes in the deep thermohaline cir- culation (changes which have been implicated in the overall global cooling). There are also Indian Ocean specific mechanisms for subsurface cooling. As the north- ward shift of New Guinea introduced fresher, cooler intermediate water from the North Pacific from 4.3 to 2.5 Ma, subsurface waters cooled by ca. 4 ◦ C (Karas et al., 2011) and this is sufficient to explain the magnitude of cooling recorded in DSDP Site 231 waters. Theoretical calculations suggest that zonal flow across the equatorial basin at ca. 10 ◦ S would have transported cooler waters from the 39 Indonesian Throughflow region to the tropical Indian Ocean thermocline including the upwelling regions along the coast of Africa (Cane and Molnar, 2001). Modern observations demonstrate that the Somali and Omani upwelling regions are also partially fed by Indonesian Throughflow water (Song et al., 2004). During the boreal summer, 62±5% of the Indonesian Throughflow water flows northward and upwells along the Somali Coast (Song et al., 2004). Increasingly cool upwelled water would have exerted a positive feedback on land-sea temperature contrasts, promoting the stronger winds and increased upwelling. It is therefore quite possi- ble that the observed cooling reflects both a cooling of subsurface waters and an intensification of upwelling, however either mechanism alone could be sufficient to explain the observed trends. Inadditiontotheocean-atmospherecirculationchangesdescribedabove, global climate change (Zachos et al., 2008) and low latitude vegetation distributions are strongly affected by atmospheric pCO 2 (Ehleringer et al., 1997). Recent studies provide suggestions that pCO 2 may have decreased since 5 Ma (e.g., Bartoli et al. 2011), however uncertainties and proxy divergence remains an issue. This atmospheric pCO 2 decline may have been caused by carbon cycle changes as feed- backs associated with the onset of Northern Hemisphere glaciations. Alternatively weathering of the uplifted Maritime Continent has been proposed as a mechanism to drawdown pCO 2 since 5 Ma (Molnar and Cronin, 2015). Although this remains uncertain, any such pCO 2 decline may potentially have contributed to global tem- perature trends and the cooling of Indian Ocean temperatures observed here. This cooling is of particular interest because it likely suppressed rainfall on land as recorded in the pollen and leaf wax evidence. 40 2.5.8 Ecosystem change and hominin evolution The evidence for drying on land captured in this multi-proxy reconstruction fromDSDPSite231offersregionalcontextforthefossilrecordofhumanevolution. While local context remains of primary interest to understanding the habitat of the individual fossil, the marine record enlarges the perspective which may be helpful to consider as the speciesâĂŹ and individualsâĂŹ range are likely larger than the loci of fossil preservation. We estimate that wind-blown plant leaf waxes from marine sediments provide an averaging of conditions across the source area (likely 1000s km2). Comparing the marine core record to soil carbonates long-measured at hominin sites, we find that the density of sampling of soil carbonates is not sufficient to detect if there is a long term trend or variability within the Pliocene (Feakins et al., 2013). What is apparent from the soil carbonates is that at each locality sam- pled there are a range of environments, as the soil carbonates record the very fine spatial scale on the order of (<1 m2), with patterns dependent on sample cover- age (Cerling et al., 2012). We note that there is no pure grassland category at either of the Ethiopian or Kenyan rift valley reconstructions between 5 and 2 Ma except for the Ardipithecus ramidus site at Hadar, Ethiopia which has extensive C 4 grasslands beyond riparian corridor forest (Cerling et al., 2011). This grass- land at 4.4 Ma is consistent with the evidence for early Pliocene grasslands in the marine core record from 5.3 to 4.3 Ma. After that time we find a shift to drier and more shrubby vegetation and a decrease in the grasses and the use of the C 4 pathway (Bonnefille, 2010 and this study), and that is matched by a lack of pure C 4 grassland areas at any of the hominin localities sampled to date (Cerling et al., 2012). While the work is far from complete to characterize the terrestrial 41 mosaic for the Pliocene landscapes, the marine record provides a regional con- text. This regional context suggests that the large lakes reported from the East African Rift Valley in the late Pliocene (Maslin et al., 2014) occurred during dry times for northeast Africa beyond the Rift Valley as recorded by the marine core. Such a scenario is not unexpected, as a prominent dipole between the coastal and interior regions has been observed in modern climatological observations and last millennium hydroclimate reconstructions (Tierney et al., 2013). More recently the δ 13 C of individual hominin teeth have also been sampled to record dietary preferences of individuals (Cerling et al., 2013). Carbon isotopic analysis of hominin teeth provide a new window into hominin diet revealing the inclusion of more 13C enriched foods in hominin diets after 3.4 Ma (Fig 5b) show- ing the data for individual teeth reflecting variations in each individualâĂŹs diet (Cerling et al., 2013). For example, specimens of Australopithecus afarensis incor- porated more C 4 biomass in their diet at Hadar and Dikika, Ethiopia after 3.4 Ma, than did hominins prior to that time (Wynn et al., 2013). Intriguingly this uptake of C 4 foods coincides with the regional 13C enrichment between 4 and 3 Ma (Fig 5). There is as yet no evidence for C 4 foods in diets of sampled individuals prior to that time, including during the early Pliocene extensive C 4 grasslands. It is well known that different fauna adapt their diet to take advantage of new resources at different times (Uno et al., 2011). We tentatively speculate that the timing of the dietary uptake of C 4 foods by hominins might be an evolutionary response to pres- sures on limited food resources during regionally dry conditions, and we propose that the marine core record be considered as one line of evidence when evaluating the context of these evolutionary shifts. 42 Northeast Africa δ C Plant Leaf Wax 2 3 4 5 13 Age (Ma) Ethiopia Kenya 0 20 40 60 80 100 Pollen (%) Amaranthaceae Grass −30 −28 −26 −24 -8 -6 -4 -2 0 −14 −12 −10 East Africa δ C Hominin Tooth Enamel 13 A B C Figure 2.5: Terrestrial vegetation records from DSDP Site 231 and hominin diet. (A) Plant leaf wax δ 13 C wax nC 3 0 from DSDP Site 231. (B) δ 13 C of hominin tooth enamel from Turkana Basin (yellow symbols) including from Australopithecus ana- mensis (diamonds), Kenyanthropus platyops (circles), Homo Indenti (upside down triangle), Paranthropus aethiopicus (triangles) and Homo sp. (star) (Cerling et al., 2013). The δ 13 C of hominin tooth enamel from Ethiopia (green symbols) include Ardipithecus ramidus (triangles), Australopithecus afarensis (squares), and Homo Indenti (upside down triangle) (Levin et al., 2015; Wynn et al., 2013; White et al., 2009). (C) Pollen as in Fig. 2. 2.6 Conclusions Marine sediments from the Gulf of Aden provide a continuous, multi-proxy, high-resolution perspective on changing climate conditions and vegetation commu- nities over northeast Africa during the Pliocene-Pleistocene (5.3-2 Ma). Leaf wax 43 and pollen together indicate that widespread C 4 grasslands were present in north- east African lowlands during the early Pliocene, before drying led to an expansion of arid-adapted C 3 shrubland. Mixed C 3 and C 4 vegetation including arid shrub- lands and minor inputs from C 4 grasslands characterized the later Pliocene during times when some hominins incorporated C 4 in their diet. The marine core evidence indicates aridification occurred well prior to the first major Northern Hemisphere glaciation. TEX 86 indicates that Gulf of Aden subsurface temperatures cooled by 4.4 ◦ C across 5.3 to 2 Ma, with similar cooling in SSTs recorded by alkenones in the north- ern Arabian Sea (Huang et al. 2007; Herbert et al. 2010). This cooling is con- sistent with the Indonesian Throughflow hypothesis (e.g., Cane and Molnar 2001) and supported with evidence for subsurface cooling in the Indian Ocean (Karas et al., 2009) (Karas et al. 2009). It is possible that the cooler waters, once upwelled, exerted a positive feedback on atmospheric circulation thereby strengthening the Indian Monsoon and promoting further upwelling. Additional global mechanisms for cooling, thermocline shoaling and upwelling intensification may alternatively explain the observed trends. These oceanic changes are likely to explain the drying and vegetation change in northeast Africa. 2.7 Acknowledgements This research used samples provided by the Integrated Ocean Discovery Pro- gram. Funding for this research was provided by the IODP Schlanger Ocean Drilling Fellowship 2013-2014 to HL; the University of Southern California and the Women in Science and Engineering program at USC to SF for compound spe- cific isotope analyses and the US National Science Foundation (OCE-1203892) to 44 JT for TEX 86 analyses. We thank Raymonde Bonnefille, Kevin Uno, Tim White, Naomi Levin, Rich Pancost, Tim Herbert, Tim Cronin and Peter Molnar for dis- cussions of aspects of this work. We would like to thank Miguel Rincon and USC undergraduates Alexa Sieracki, Jeff Zhang, Susan Oh, Jack Seeley, Vincent Nguyen, Tiffany Kao, Kyle McAhlaney, Jacob Leonard, Megan McDonald, and Clara Hua for laboratory assistance. 45 Chapter 3 Late Miocene C4 Expansion in the Indus River Catchment This manuscript is in preparation for submission to the journal Paleoceanogra- phy to include coauthors: Sarah J. Feakins, Lisa Tauxe, Sophie Warny, Jessica E. Tierney, Valier Galy, Denise Kulhanek, Peter Clift 3.1 Abstract TheIndusRiverhassuppliedsedimentstotheArabianSeaduringtheCenozoic ultimately resulting in the accumulation of the world’s second largest sedimentary fan. Sediments from the Indus Fan mainly represent turbidite deposition of terres- trial material derived from the Indus River catchment with intervals of hemipelagic sedimentation and wind-blown terrestrial inputs. IODP Expedition 355 recovered late Miocene sediments that capture a record of erosion, vegetation and climate in the paleo-Indus River catchment and lesser contributions from other regions surrounding the Arabian Sea. We report the carbon isotopic composition (δ 13 C) of leaf wax n-alkanes, n-alkanoic acids and bulk organic carbon as well as pollen to reconstruct vegetation change. Additional terrestrial versus marine biomark- ers and microfossils provide complementary information on changes in production, erosion and sourcing of sediments, which is needed for accurate interpretation of 46 the fan record. Transitions are abrupt given the episodic nature of fan sedimenta- tion, but comparisons to terrestrial and open ocean sediments reveal a regionally coherent picture of vegetation change across a broad region. We find that regard- less of source, C 4 grasslands expanded after 7 Ma. We also measured the leaf wax hydrogen isotopic composition as a tracer of the isotopic composition of pre- cipitation. After C 4 grassland expansion, the δD of plant wax sourced from the paleo-Indus catchment remains constant while plant wax source from the broader region suggest reduced precipitation after 7 Ma. The late Miocene Indus Fan sed- iments preserve episodic and sometimes voluminous inputs from the paleo-Indus River catchment and surrounding continents, and our multi-proxy investigation reveals how those sedimentary changes resolve regional environments during the late Miocene. 3.2 Introduction Terrestrial records, including the Siwalik Group in the Himalayan foothills, have documented the emergence and expansion of C 4 biomass in South Asia ~7 Ma (Cerling et al., 1997; Quade et al., 1995; Quade & Cerling, 1995). While many proxies in these sediments document a vegetation shift (including evidence from soil carbonates, soil organic matter and fossil teeth of herbivores), just how the monsoon changed at this time is ambiguous. Oxygen isotopes from soil carbon- ates document gradual shift to more positive values amidst large scatter associated with evaporation that can alter the isotopic composition of soil waters (Quade et al., 1989). The oxygen isotope record preserved in mollusk fossils from the Siwalik range suggest that while a seasonal monsoonal regime was in place by the late Miocene, summer rainfall decreased after 7.5 Ma (Dettman et al., 2001). Offshore 47 evidenceforchangesinthemonsoonsystemdorevealchangesinatmosphericcircu- lation; however foraminiferal evidence for intensification of wind-driven upwelling at 8 Ma (Kroon et al., 1991) have recently been questioned because of evidence for uplift of the Owen Ridge at that time and thus enhanced carbonate preservation (Rodriguez et al., 2014). Furthermore, upwelling indicator species from different locations reveal sensitivity to bathymetry (Gupta et al., 2015; Kroon et al., 1991). The nature and timing of the late Miocene C 4 expansion remains of great interest because of likely, but still uncertain, connections to changes in the monsoon. Part of the controversy is due to the nature of the available proxy records, providing a partial view of environmental change, as well as the differences in the timing and locations captured by these records. There is therefore considerable value in developingmulti-proxymarinecorearchivesofvegetationandprecipitationproxies together. Sediments originating from the Indus River are transported through canyons bisecting the continental shelf, delivering turbidity currents into the Indus Fan along channels and building associated levees (Kolla & Coumes, 1987). The locus of deposition has shifted laterally over hundreds of kilometers, and one of these transport channels supplied material to the Laxmi Basin, located in the eastern- most region of the fan (between the continental shelf of peninsular India and the Laxmi Ridge) during the late Miocene, the time period of interest for C 4 expan- sion and monsoon change. Turbidite deposition is an efficient carrier of terrestrial carbon to the deep sea (France-Lanord & Derry, 1994), however during periods of hemipelagic deposition, eolian material can be the dominant conveyor of ter- restrial material (Prins & Postma, 2000). During these periods, Eastern Africa, Arabian Peninsula, and Pakistan/Northern India may all contribute material to the Arabian Sea due to prevailing seasonal winds (Dahl et al., 2005; Sirocko & 48 Lange, 1991). Indus Fan sediments thus offer a crucial yet complicated history of regional vegetation change in the paleo-Indus catchment and surrounding regions, and thus it is important to constrain the sediment source through time. In this study, we present the results from analyses of late Miocene sediments recovered during IODP Expedition 355 to the Indus Fan (Pandey et al., 2015). We focus on the evidence for vegetation and precipitation change. We draw upon mul- tiple proxy lines of evidence, in particular the plant leaf wax molecules, biomark- ers for terrestrial vegetation, and their carbon and hydrogen isotopic composition that provide evidence for vegetation change and the isotopic composition of pre- cipitation in the same molecule. We also present biomarker evidence for ocean temperatures, marine productivity and terrestrial:marine inputs. Palynology pro- vides constraints on marine dinoflagellate cysts versus terrestrial pollen and helps to diagnose the vegetation assemblages and origins of the plant-derived debris. Additionally, we compare geochemical and sedimentary evidence for sourcing to further trace shifts in erosion within the paleo-Indus catchment. These various lines of evidence, together with age constraints from paleomagnetism, nannofossils and foraminifera help to build the history of terrestrial environmental change in the paleo-Indus catchment of late Miocene environmental change. 3.3 Background 3.3.1 Sedimentology IODP Expedition 355 drilled two sites in the eastern Indus Fan within the Laxmi Basin. Seismic data show reflective sequences that reveal small internal onlapping relationships and complex cut-and-fill channel bodies suggestive of the “Indus Fan Megasequence” observed in the upper fan (Droz & Bellaiche, 1991). 49 These features correspond to sediments recovered from Units I-III (0-730 mbsf) at Site U1456 and Unit III (385-835 mbsf) from Site U1457 (Pandey et al., 2015). Another feature of the seismic data was an acoustically transparent, highly reflec- tive unit with prominent reflectors at the top and base. It was recently identified as part of the Nataraja Submarine Slide, a giant mass wasting deposit originating from the Indian continental margin and spanning 330 km from Gujurat-Saurashtra margin to the Laxmi Basin (Calvès et al., 2015). This feature corresponds to the recovered sediment from Unit IV (730-1,109 mbsf) at Site U1456 and Unit IV (835-1,060 mbsf) at Site U1457 (Pandey et al., 2015). For this study we targeted the late Miocene sedimentary sequence at Site U1457 interpreted as bedded sheet lobes with intervals of hemipelagic nannofossil ooze (Pandey et al., 2015). Within the sequence, silty sandstone layers have sharp erosive bases with sand grading upward to clay, and were cored with discontinuous recovery (52%) (Pandey et al., 2015). 3.3.2 Modern climate and vegetation During the summer months, the Indian low-pressure cell pulls moist cross- equatorial flow from the Arabian Sea and Bay of Bengal into Bangladesh, eastern India, and the central and eastern Ganges plains with little rainfall penetrating further west and into the Indus catchment. In the winter months, atmospheric circulation reverses, with northeasterly flow on average, interrupted by storms called “Western Disturbances” that move into the Himalayan region of north- western India and Pakistan and deliver wintertime rainfall originating from the Mediterranean region (Barros et al., 2006). Up to 72% of Indus discharge derives from this Mediterranean source (Karim, 2002). 50 The Indus catchment includes arid, semi-arid and sub-humid regions, with 461 mm of mean annual precipitation (MAP; Karim (2002)). In the upper regions of the catchment, vegetation communities are determined by altitude and pre- cipitation. Above the tree line, junipers and alpine scrubs can be found. Below the treeline and above 3,300 m Himalayan dry coniferous forests include cedars (Cedrus spp.), pines (Pinus spp.), and spruces (Picea spp.). Between 3,300-1,500 m, Himalayan moist temperate forests occur including pines, cedars, spruces, firs and some broad-leaf deciduous trees (Quercus spp., Alnus spp., Acer spp.). Dry subtropical mixed scrub forests (Olea cupidata, Acacia spp., and Dodonaea vis- cosa) are found from 1,500 to 1000m. In the Indus lowlands from 1,000 m to sea level, xeric communities of tropical thorn forests include small trees (Acacia spp., Prosopis cineraria, Tamarix dioica, Salvador persica) and shrubs (Amaranthaceae, Calligonum polygonoides, Cassia spp.) in riparian zones, while beyond the reaches of the river are sand dune deserts. The delta includes mangrove (Rhizophora spp.) forest (Ivory & Lézine, 2009). The Naramada and Tapi Rivers draining the Indian subcontinent also con- tribute sediment to the Indus Fan. Although sedimentology suggests only as recently as 2 Ma to the Laxmi Basin (Pandey et al., 2015). Their catchments have 800-1600 mm MAP, with most precipitation occurring during the Indian summer monsoon. Although still distinctly seasonal, the catchments of the Nara- mada River and Tapti River are more humid than the Indus River catchment due to dominant monsoonal climate patterns. 51 3.4 Methods 3.4.1 Site location IODP Expedition 355 Site U1457 is located in the Laxmi Basin in the eastern- most region of the Indus Fan (Figure 3.1). Site U1457 was drilled in the distal fan at 17 ◦ 90.95’N and 67 ◦ 55.81’E in 3523 m water depth. This site is located at the foot of the slope leading to the Laxmi Ridge and is ~760 km south of the modern Indus River mouth and ~490 km west of the Indian continent. 3.4.2 Age model and sample selection Age control includes nannofossil and foramineral datums and paleomagnetic reversals(Figure3.2)ontheGTS2012timescale(Gradsteinetal.,2012). InU1457, between998and834mbsf(CSF-Adepthscale)sedimentationratesaverage~17cm ky-1, followed by a hiatus of ~0.5 Ma. Between 834 and 515 mbsf, sedimentation rates decrease to ~10 cm −1 with an interval of hemipelagic sedimentation between 670-600 mbsf with slower sedimentation rates of 7 cm ky −1 . Following 505 mbsf, a second hiatus lasting ~2 Ma spans the early Pliocene (Pandey et al., 2015). We sampled a total of 52 samples from Site U1457 Hole C, targeting silty clay lithologies to capture a terrestrial signal. We required samples of ~50g dry weight (gdw) to obtain measurable quantities of leaf wax n-alkanoic acid and n-alkanes. We sampled from 505 to 869 mbsf every ~7 m spanning 5-10 Ma, and increased the resolution up to 0.5 m during the interval of slower deposition between 670 and 600 mbsf. Graded bedding was a common sedimentary structure with deposits ranging from 5 to 15 cm thickness and thin bedding cycles less than 2 cm thick; therefore our sampling resolution of ~7 m suggests that each sample corresponds to a distinct turbiditic event. 52 Kabul Jammu Mumbai Karachi 90° E 90° E 80° E 80° E 70° E 70° E 60° E 60° E 50° E 50° E 40° N 4 30° N 3 20° N 2 10° N 1 0° 0 0 1,000 Kilometers U1457 Figure 3.1: Site location of IODP Site U1457 (17 ◦ 90.95’N and 67 ◦ 55.81’E in 3522.7 m water depth) in the Arabian Sea and locations discussed in the text. The tan shaded area indicates the Indus River catchment. 3.4.3 Bulk organic carbon analysis Samples for total organic carbon (OC) quantification were acidified to remove carbonates(Whitesideetal.,2011). Specifically, samplescontainingapproximately 100μg organic carbon were weighed into methanol-rinsed Ag capsules (4 x 6 mm, Costech). Glassplates(combusted4hat450 ◦ C)holdingthesesampleswereplaced in a vacuum desiccator with an open dish with 50 mL 12N HCl. An inverted watch 53 5 6 7 8 9 10 500 600 700 800 900 Age (Ma) Depth (mbsf) Nanno Foram Pmag Figure 3.2: Nannofossil and paleomagnetic age constraints for Site U1457C (Pandey et al., 2015). Red triangles represent nannofossils datums, blue triangles represent foraminiferal datums, and yellow circles represent paleomagnetic rever- sals. The black wavy line represents a ~0.5 Myr hiatus. The black bars represent core recovery. In the stratigraphic column, brown represents clay/claystone, pink represents silt/silt stone, yellow represents sand/sandstone, and blue represents carbonate ooze/stone. glass was placed over the samples to protect them from drips. The desiccator was evacuated to ~0.5 atm and placed in an oven at 62 ◦ C (sufficient to cause the acidification of dolomite and siderite but low enough to maintain silver boat integrity) for 60 to 72 h. The samples were then transferred to an evacuated vacuum desiccator charged with indicating silica gel (Fisher S162-500, activated by heating to 450 ◦ C for 4 h) for at least 24 h. Immediately prior to analysis, the samples were wrapped in Sn capsules (4 x 6 mm, Costech). Samples were analyzed at the Woods Hole Oceanographic Institution using a Carlo Erba / Fisons 1108 flash elemental analyzer (EA) equipped with a Costech "ZeroBlank" air-excluding 54 carousel. Effluent from the EA passed via a Finnigan MAT "Conflo II" interface to a DeltaPlus stable light isotope mass spectrometer. A set of a blank and three standards (IAEA-N1 [Ammonium sulfate], USGS-40 [Glutamic Acid], and a well characterized glycine) was run between every 16 samples. NBS-19 [limestone] and a calcite laboratory standard were run as unknowns to confirm bulk measurement quality. The mass spectrometer signals (M/Z 44, 45, 46) were used for amount determination as well as isotopic analyses. Nitrogen blanks were approximated using a method similar to that of Polissar et al. (2009). Calibrations of elemental contents were accurate and precise to better than 2 and 4% of the measured value, respectively. Carbon isotopic compositions of standards were accurate and precise to better than 0.3%. We report OC% and δ 13 C OC in permil (%) units on the VPDB scale. 3.4.4 Lipid extraction Samples were freeze dried in a Virtis 2k unit and homogenized with a mortar and pestle. Dry, powdered sediment samples (31.7-95.1 gdw) were extracted with an Accelerated Solvent Extraction system (ASE 350 R , DIONEX) with 9:1 ratio of dichloromethane (DCM): methanol (MeOH) at 100 ◦ C and 1500 psi for two 15- minute cycles. Extracts were separated over a NH 2 sepra column (5 cm x 40 mm Pasteur pipette, 60 Å) eluted with 2:1 ratio of DCM: isopropanol (neutral fraction) and 4% formic acid in diethyl ether (acid fraction). The acid fraction was methy- lated with methanol of known isotopic composition with 95:5 MeOH: hydrochloric acid at 70 ◦ C for 12 hours. Methylated products were extracted using liquid-liquid extraction with 1 mL milli-Q water and hexane, and the hexane extract was passed through anhydrous sodium sulfate. Samples were further purified over a silica gel column (5 cm x 40 mm Pasteur pipette, 5% water-deactivated silica gel, 100-200 55 mesh) eluted with hexane and DCM resulting in non-polar and FAME fractions, respectively. The neutral fraction was loaded onto a silica gel column (5 cm x 40 mm Pas- teur pipette, 5% water-deactivated silica gel, 100-200 mesh) and eluted with hex- ane, DCM and MeOH resulting in the separation of aliphatic hydrocarbons (n- alkanes), ketones (alkenones), and a more polar fraction (GDGTs), respectively. The aliphatic hydrocarbon containing the n-alkane fraction was treated to remove elementalsulfurbypassingthesamplethroughanactivatedcopperpipettecolumn. Urea adduction was performed as necessary to remove the cyclic and branched alkanes compounds Wakeham et al. (1997). 3.4.5 Leaf wax quantification The fatty acid methyl ester fractions were identified and quantified using gas chromatography coupled with both a mass-selective detector and flame ioniza- tion detection (GC-MSD/FID Agilent) at the University of Southern California. 1/100 μL of the sample was analyzed by gas chromatography with injection via a split/splitless inlet in splitless mode, to a capillary column (Rxi R - 5ms 30m × 0.25mm, film thickness 0.25mm) with a constant He flow rate of 4mL/min. Initial temperature of 50 ◦ C was held for 3.5 minutes followed by a temperature ramp of 20 ◦ C min −1 to 300 ◦ C held for an additional 10 minutes. Quantification was achieved using an in-house standard comprising a mixture of 4 n-alkanes and 3 n-alkanoic acids of varied and known concentration, with the calibrations deter- mined separately for the two compound classes. The n-alkane and n-acid fractions are reported relative to the organic carbon content (μg g-1 OC) of each sample. 56 After quantification of the individual peak areas, we calculated the carbon preference index (CPI) for C 25 -C 33 n-alkanes as: CPI = 1 2 + P [C 25−35odd ] P [C 24−34even ] + P [C 25−35odd ] P [C 26−36even ] (3.1) The C 36 n-alkane was below the detection limit and assumed to be 0. For n- alkanoic acid, the same CPI equation was applied using the C 24 -C 34 in the numer- ator and C 23 -C 33 and C 25 -C 35 in the denominator. The C 35 n-alkanoic acid was below the detection limit and assumed to be 0. To determine changes in the average chain length (ACL) of land plants, we calculated the ACL of the concentration-weighted abundances of long-chain homologues using the following equation for n-alkanes: ACL = 25× [C 25 ] + 27× [C 27 ] + 29× [C 29 ] + 30× [C 30 ] P C 25−30 (3.2) The same formula was applied for n-alkanoic acids over the even-numbered homo- logue range of C 26 -C 34 n-alkanoic acids. 3.4.6 Compound specific carbon and hydrogen isotopic analysis The compound specific δ 13 C and δD isotopic composition of n-alkanoic acid and n-alkanes were analyzed using a Thermo Scientific Trace GC equipped with a Rxi R -5 ms column (30 m × 0.25 mm, film thickness 1μm) with a PTV injec- tor operated in solvent-split mode, coupled to a Delta V Plus isotope ratio mass spectrometer (IRMS) via an Isolink combustion/pyrolysis furnace (1000/1400 ◦ C). 57 Isotopic linearity was monitored daily across a range of peak amplitude (1-8V) CO 2 /H 2 gas pulses. For carbon, the isotopic linearity of δ 13 C averaged 0.05%. For hydrogen, the H 3 factor averaged 4.6 ppm mV −1 . Two out of five CO 2 /H 2 reference peaks were used for standardization of the isotopic analysis and the remaining peaks were used to assess precision. We used external standard runs containing a mixture of 15 n-alkanes (C 16 to C 30 ) with δ 13 C values ranging from 33.3 to 26.2% and δD values ranging from -254.1 to -9.1% (A mix standards supplied by A. Schimmelmann, Indiana University, USA) to normalized the data to the Vienna Pee Dee Belemnite (VPBD) carbon isotopic scale and to the Vienna Standard Mean Ocean Water (VSMOW) - Standard Light Antarctic Precipitation (SLAP) hydrogen isotopic scale. The RMS error of the external standard repli- cates throughout the total run time was 4.5%. Samples were run in triplicate, and the average standard deviation was 0.5% for carbon and 2.2% for hydrogen. The isotopic composition of C added during methylation of FAs to methyl esters was determined by offline combustion and dual-inlet IRMS (δ 13 C MeOH = -24.7 ± 0.2%, n=6), and the isotopic composition of H added during methylation was determined by the esterification of phthalic acid for analysis by GC-IRMS (δD MeOH = -187 ± 4%, n=6). The addition of the methyl group was corrected for by mass balance. The results are reported using conventional delta notation (δD%). 3.4.7 GDGT quantification The neutral fractions were dissolved in hexane:isopropanol (99:1) and fil- tered through a 0.45 micron PTFE filter prior to injection on an Agilent 1260 High-Performance Liquid Chromatography (HPLC) coupled to an Agilent 6120 mass spectrometer. Glycerol dialkyl glycerol tetraethers (GDGTs) were analyzed 58 according to the method of Schouten et al. (2007a). Briefly, GDGTs were separated on a Grace Cyano column with a 5 minute isocratic elution using A, and then a gradient of 90% A/10% B to 82% A/18% B over 35 minutes, where A = hexane and B = hexane:isopropanol (9:1). The column was then backflushed with 100% B for 10 minutes, and re-equilibrated with 90% A/10% B for another 10 minutes. Single Ion Monitoring (SIM) mode of the M + H+ ions was employed to detect and quantify the isoprenoidal GDGTs produced by marine archaea including GDGTs with 0-3 cyclopentane moieties (GDGT-0 to GDGT-3); crenarchaeol (Cren) with an additional cyclohexane moiety and its regioisomer crenarchaeolÂť (CrenÂť). Following Schouten et al., (2007), TEX 86 units were calculated using the equation: TEX 86 = [GDGT − 2] + [GDGT − 3] + [Cren 0 ] [GDGT − 1] + [GDGT − 2] + [GDGT − 3] + [Cren 0 ] (3.3) Repeat analyses of a laboratory standard yields a long-term precision of 0.004 TEX 86 units. We converted the TEX 86 record to sea surface temperature using the BAYSPAR- No Red Sea (NRS) calibration, the spatially-varying Bayesian regression approach (Tierney & Tingley, 2014). For our prior distribution, we chose modern mean annual integrated sea surface temperatures at the core site (28 ◦ C) and a standard deviation of 10 ◦ C (Locarnini et al., 2010). We report the total branched (brGDGT) and isoprenoidal (isoGDGTs) moieties including crenarchaeol (ng/gdw) as measure of terrestrial and marine inputs respectively and calculate the BIT index: BIT = [I] + [II] + [III] [I] + [II] + [III] + [IV ] (3.4) 59 where I, II and III represent the abundances of brGDGT and IV represents the abundance of crenarchaeol in each sample (Hopmans et al., 2004). 3.5 Results 3.5.1 Concentration and δ 13 C of bulk organic carbon The organic carbon (OC) content ranges from 0.12 to 0.76% with a mean of 0.29%. δ 13 C OC values range from -24.4 to -16.8% with a mean of -21.1%. Hemipelagic sediments between 670 and 600 mbsf yield the highest concentrations of OC, and lower concentrations are associated with sand between 820 and 670 mbsf and after 600 mbsf (Figure 3.6f). 3.5.2 Plant wax concentration and distribution of n- alkanes Total C 16 -C 35 n-alkane concentrations range between 46 and 1525 ng/g of dry sediment (ng/gdw) or 18 and 417 μg/g OC. As opposed to TOC concentrations, n-alkane concentrations do not appear to change in response to varying lithology. n-Alkane abundance distributions are typical of higher plant sources, with modal chain lengths C 29 and C 31 (Figure 3.3a). Odd chain lengths in the range of C 27 -C 35 account for 40 to 78% of total n-alkanes. CPI varies between 3.4 and 11.0 (except for 1 sample at 869 mbsf, CPI = 1.4) indicating that petrogenic input is negligible. ACL is high (29.9) between 832-869 mbsf. Following a hiatus, ACL decreases to 29.2 and subsequently increases to a maximum value of 31.3 at 521 mbsf, towards the end of the record. 60 0 0.1 0.2 Relative abundance n−Alkanoic Acid C16 C18 C20 C22 C24 C26 C28 C30 C32 C34 0.3 0.3 0 0.1 0.2 Relative abundance n−Alkane C17 C19 C21 C23 C25 C27 C29 C31 C33 C35 (n = 51) (n = 51) A B Figure 3.3: Average histogram distribution (relative abundance normalized to sum of 1) of (A) n-alkanes and (B) n-alkanoic acid. Total number of quantified samples is reported in parentheses. Error bars represent standard deviation. 61 3.5.3 Plant wax concentration and distribution of n- alkanoic acids n-alkanoic acid concentrations are an order of magnitude greater than n-alkane concentrations. Total C 16 -C 34 n-alkanoic acid concentrations range between 221 and 3,319 ng/g of dry sediment (ng/gdw) or 78 and 838μg/g OC. n-alkanoic acid concentrations also do not appear to vary in response to changes in lithology. n- alkanoic acids are dominated by mid to long-chain C 26 -C 28 homologues with an even-carbon number predominance (Figure 3.3b). Even chain C 28 -C 34 account for 11 to 57% of the total n-alkanoic acids. C 16 and C 18 are also abundant. CPI varies between 2.0 and 5.6. ACL is ~29 throughout, except for a minimum of 26.3 in one interval of hemipelagic sedimentation. 3.5.4 Plant wax δ 13 C and δD compositions We reportδ 13 C andδD values for long chain n-alkane (C 25 -C 35 ) and n-alkanoic acid (C 36 -C 36 ) homologues. n-Alkane homologues have an average δ 13 C range of 8.8%. For example, C 31 n-alkanes span -33.6 to -25.1%. We note there is a systematic 13 C -enrichment of the long chain n-alkanes especially C 35 n-alkane which ranges from -31.8 to -21.5%. Long chain n-alkanoic acids show on average 10.0% variability within a single chain length, for example δ 13 C 30 varies between -30.8 to -20.9%. δD of long-chain n-alkanes and n-alkanoic acid homologues display similar downcore trends (Fig 3.4). For the C 31 n-alkanes, the δD value varies between -214 to -126%, with an averageδD range for n-alkane homologues being 79.1%. For the C 28 n-alkanoic acid, the δD value varies between -213 to -126%, and the average δD range within a single chain length is 86%. Due to the coherence 62 between the long chain δD values of the two compound classes, we calculated the weighted mean average of δD n-alkane (δD alk ) and δD n-alkanoic acid (δD acid ) using the relative abundance of C 25 -C 31 and C 24 -C 30 , respectively (Figure 3.4b). A bivariate plot ofδ 13 C andδD values of the C 28 n-alkanoic acid, reveals 3 clus- ters of data and we demarcate an expected positive relationship, although another cluster have high δ 13 C values but low δD and are excluded from the correlation (Fig. 3.5A). No correlation is observed for the n-alkanes (Fig. 3.5B). −36 −34 −32 −30 −28 −26 −24 −22 −20 −18 −16 500 550 600 650 700 750 800 850 900 Depth (mbsf) TOC C27 C29 C31 C33 C35 C28 C30 C32 C34 C36 δ 13 C wax (‰) −220 −200 −180 −160 −140 −120 δD wax (‰) A. B. C24 C26 C28 C30 WM C25 C27 C29 C31 WM Figure 3.4: (A) Carbon isotope values of long chain n-alkanes (C 27 -C 35 ), long chain n-alkanoic acids (C 28 -C 36 ), and organic carbon plotted using dashed green, solid green, and black lines, respectively. Analytical uncertainty for plant wax carbon isotopes is ±0.5% and for OC is ±0.11% (B) Hydrogen isotope values of n- alkanes (C 25 -C 31 ) and n-alkanoic acids (C 24 -C 30 ) are plotted using blue solid lines and blue dashed lines. The δD weighted mean average for C 25 -C 31 is plotted as a solid purple line andδD weighted mean average for C 24 -C 30 is plotted as a dashed purple line. Analytical uncertainty for plant wax hydrogen isotope is ±2.2% for hydrogen. 63 A. B. −220 −200 −180 −160 −140 −120 −34 −32 −30 −28 −26 −24 −22 −20 δD C28 (‰) δ 13 C C28 (‰) R 2 =0.88, n = 42 −220 −200 −180 −160 −140 −120 −34 −32 −30 −28 −26 −24 −22 δD C31 (‰) δ 13 C C31 (‰) Figure 3.5: Carbon and hydrogen bivariate plots of the most abundant homologue of n-alkanoic acids (C 28 ) and n-alkanes (C 31 ). 3.5.5 GDGTs concentrations and temperature reconstruc- tions IsoGDGT concentrations are typically ~1033 ng/g or ~256 ng/g OC, but increase in pulses to 5775 ng/g or 959 ng/g OC in OC-rich hemipelagic sedi- ments from 600 to 650 mbsf (Figure 3.6e). brGDGTs concentrations are generally lower and more stable than isoGDGTs, but still range over an order of magnitude between 16 and 284 ng/g or 7 and 112 ng/g OC. As a result, isoGDGT varia- tions dominate the BIT index downcore. We identify three intervals of low BIT (<0.3) between 869-832, 667-605, and 521-505 mbsf, and two intervals of high BIT (>0.3) between 823-673 and 601-533 mbsf. During periods of high BIT, we do not interpret TEX 86 in terms of temperature. During intervals of low BIT, TEX 86 estimates indicate high SSTs (31-32 ◦ C) ca. 9 Ma, decreasing to 28 ◦ C by 6 Ma. 64 −34 −32 −30 −28 −26 −24 −22 −20 500 550 600 650 700 750 800 850 900 δ 13 C wax (‰) Depth (mbsf) −25 −23 −21 −19 −17 δ 13 C OC (‰) C. D. E. F. −220 −200 −180 −160 −140 −120 δD (‰) Pandey et al., 2015 0 0.4 0.8 BIT Index 0 0.2 0.4 0.6 0.8 OC (%) Dominant lithology Carbonate ooze/stone Clay/ Claystone Silt/ Siltstone Sand/ Sandstone 0 2000 4000 6000 GDGT (ng/g) isoprenoid branched hemipelagic hiatus sands turbidite turbidite hiatus? sands turbidite hiatus more marine productivity C 30 C 35 WMA alkane WMA acid OC A. B. 65 Figure 3.6: Organic and inorganic proxy records from the Indus Fan IODP Site U1457C. (A) δ 13 C plant wax isotope records of n-alkane C 31 and n-alkanoic acid C 30 (B) organic carbon isotope values (C) Weighted mean average δD plant wax (D) BIT Index (E) GDGT of branched and isoprenoidal concentrations (F) organic carbon content (%). Dominant lithology is characterized by turbidite deposition with an interval of slower hemipelagic deposition (Pandey et al., 2015). Dashed lines refer to hiatuses in deposition. 3.6 Discussion 3.6.1 Fidelity of paleoclimate records during shifts in marine and terrestrial sedimentation Although terrestrial inputs dominate the Indus Fan in general, several proxies reveal abrupt variations in the proportions of terrestrial vs. marine inputs in the Indus Fan deposits. We have microfossil evidence for terrestrial pollen and marine palynomorphs (dinoflagellate cysts) and biomarker evidence from the branched and isoprenoidal GDGTs, as well as sedimentary and carbonate microfossil strati- graphic records evaluating the dominant lithology of terrestrial detrital material or marine nannofossil origins (Pandey et al., 2015). It is important to identify changes in sedimentation as the source of terrestrial material may change between periods of turbidite deposition and hemipelagic deposition as observed at other times in the Indus Fan (Deplazes et al., 2014; Prins & Postma, 2000). We reconstructed the Branched and Isoprenoidal Tetraether (BIT) index from GDGT concentrations toassess the relativeinputs ofsoil-derivedto marinederived organic material (Hopmans et al., 2004). We find that GDGT concentrations are dominated by in situ production of isoGDGTs (Figure 3.6e), but shifts in the pro- portions of brGDGTs to isoGDGTs resolve the changing dominance of terrestrial 66 versus marine GDGT inputs that follow fan sedimentation changes between tur- bidic and hemipelagic sediment reflected in the dominant lithology (Pandey et al., 2015). As a result of these changes, the BIT index varies abruptly, and we find a strong relationship between BIT index and δD values of n-alkanoic acids (Figure 3.6). During periods of high BIT, we observe δD acid values of -200%, while in intervals of low BIT,δD acid values are more enriched (ca. -140%). These shifts in deposition andδD acid reflect changes in sedimentation to Site U1457. The fan sed- imentary record is dominated by episodic transport to the mid and distal fan, thus we sampled an episodic record of terrestrial export at our core site. We assume that during times of turbidic lithology, that the biomarker record reflects the paleo- Indus River export, and sedimentology confirms this (Pandey et al., 2015); whereas during intervals of weak terrestrial inputs and hemipelagic sedimentation, we sus- pect that eolian deposition may dominate the biomarker record. Eolian plant wax deposition were inferred from coretop sediments across the Arabian Sea together with geochemical evidence for eolian sediment provenance (Dahl et al., 2005) and it is entirely plausible that wind-blown waxes would also reach the Indus Fan given dust loading in the atmosphere over the adjacent landmasses (Prospero, 2002) and given favorable wind-transport events during northerly winter winds (Sirocko & Lange, 1991). 3.6.2 Transport pathways of n-alkanes and n-alkanoic acids Plant wax includes various biochemical components, however differences in plant production (Diefendorf et al., 2015; Feakins et al., 2016; Garcin et al., 2014) soil storage, mobilization and transport between different biomarkers (Giri et al., 2015; Hemingway et al., 2016) may alter their representation in sediments. 67 Recent reviews rank our knowledge of sediment transport of leaf waxes as limited (Diefendorf & Freimuth, 2017; Sachse et al., 2012), however studies of wind-blown transport (Gao et al., 2014; Schefuß et al., 2003; Yamamoto et al., 2013) and fluvial transport (Galy et al., 2011; Hemingway et al., 2016; Ponton et al., 2014) have con- strained transport processes in modern environments. While in the sedimentary record, downcore records that compare the preserved signal of multiple compound classes are an important aspect of understanding the fate of these compounds and how they archive environmental information. Here we present an unprecedented downcore comparison of multiple homologues of two compound classes including both carbon and hydrogen isotopic compositions (Figure 3.4). Overall the coher- ence between δ 13 C wax and δ 13 COC strongly suggest that most of the OC in the Indus Fan is terrestrially derived. We find that n-alkanes are overall more 13 C- enriched relative to n-alkanoic acid: prior to grassland expansion this difference is small (+1.6%) and increases after grassland expansion are larger (+3.0%). Within the n-alkanes we find the C 35 n-alkane is enriched relative to other chain lengths after grassland expansion. Although this is not associated with a clear offset between δD n-alkane and n-alkanoic acid values. We find that δD values exhibit larger differences between compound classes, with the n-alkanes showing less downcore variability and the n-alkanoic acids responding abruptly to source switching as indicated by terrestrial vs. marine indices. Offsets may be as large as +36.4% as observed at the onset of hemipelagic sedimentation at 675 mbsf (7.25 Ma), which likely reflect a difference in the transport mechanism of n-alkanes and n-alkanoic acid. We find n-alkanoic acid to be an order of magnitude more concentrated than n-alkanes. Similarly, n-alkanoic acids are reported to be more abundant in sus- pendedriver sediment thann-alkanesin boththeGanges-Brahmaputra andCongo 68 Rivers (Galy et al., 2011; Hemingway et al., 2016). In contrast in plants, n- alkanes often dominate in angiosperms, whereas n-alkanoic acids often dominate in gymnosperms (Diefendorf et al., 2011). Leaf wax export by the Indus, Ganges- Brahmaputra and Congo Rivers are unlikely to be dominated by gymnosperm sources, thus we instead infer that erosion and transport by rivers favors n-alkanoic acids. Similarly, n-alkanoic acids were found more responsive to runoff in Congo River seasonal time series data (Hemingway et al., 2016). Overall, we find that n-alkanes and n-alkanoic acids capture generally consistent trends in δ 13 C and δD, with some divergences that reveal differences in sourcing, mobilization and transport processes. 3.6.3 Transport pathways of pollen and leaf wax Pollenlendsimportantinsightintocarbonisotoperecords, revealingnuancedinfor- mation about vegetation community assemblage. Complications can arise in the interpretations of ecological changes that cannot be clarified without pollen data. However, different transport pathways of plant wax and pollen can complicate comparison of the two proxies. Dispersal and production mechanisms differ as a reflection of the different roles pollen and plant wax serves for the plant. Different pollen have different production and transport pathways too, for example, tropical trees are underrepresented in the fossil pollen record due to limited pollen produc- tion and reliance on animal transport (Bush, 1995), whereas grass pollen may be more ubiquitous because it is dispersed by wind (Dupont, 2011). Pollen grains can transport long distances and typically follow the most dominant pathway of trans- port (e.g. eolian or fluvial) (Dupont & Wyputta, 2003). Long-range transport of plant wax has also been observed and is an important component of dust (Conte & Weber, 2002; Huang et al., 2000; Schefuß et al., 2003). Here, we interpret pollen 69 and leaf wax data to reflect both local and distal sources and use the BIT Index to interpret changes in source. We use pollen counts to clarify trends inδ 13 C wax andδ 13 C OC . We present ratios of terrestrialpollenincludingGramineae(grasses), bryophytes/pteridophytes(mostly mosses and ferns), non-arboreal angiosperms (shrubs and herbaceous flowering plants), andgymnosperms(pines, spruces, firs). Ratiosofterrestrialpollen, marine dinoflagellate cysts, and deltaic pollen counts (including freshwater algae, man- groves, aquatic plants, and wetland plants) reflect dominant sources of microfossil input. Most pollen and marine microfossils are penecontemporaneous, although in some samples reworked proportions are high (Figure 3.7c). Dinoflagellate cysts are the largest contributor of the reworked microfossils and are identified as Juras- sic or Cretaceous in age. Likely sources of reworked microfossils from within the paleo-Indus catchment would be erosion of Himalayan rocks as part of the Tethyan Himalayan Sequence (Yin, 2006). 3.6.4 Vegetation reconstructions We identify intervals in time that correspond to periods of low BIT hemipelagic deposition and high BIT turbidic deposition, which we interpret to reflect changes in the environments of the broader Indo-Arabian region and within the Indus catchment, respectively. We define the intervals of low BIT as 869-832 mbsf (9.8- 8.3 Ma), 667-605 mbsf (7.2-6.0 Ma), and 521-505 mbsf (5.6-5.5 Ma) and intervals of high BIT as 823-673 mbsf (8.2-7.3 Ma) and 601-533 mbsf (5.9-5.5 Ma). These intervalsbracketanexpansionofC 4 thatisobservedregardlessofchangesinsource and provides an opportunity to characterize the associated changes in vegetation and hydroclimate both within the Indus catchment and surrounding regions. 70 −34 −30 −26 −22 500 550 600 650 700 750 800 850 900 δ 13 C (‰) Depth (mbsf) −25 −23 −21 −19 −17 δ 13 C (‰) 0 50 100 Non−arboreal Angiosperm Bryophyte/ Pteridophytes Arboreal Angiosperm Gymnosperm Terrestrial Marine Deltaic Grass Terrestrial (%) 0 50 100 Reworked (%) Sourcing (%) 0 50 100 0 0.8 0.4 BIT Index Dominant lithology Sand/Sandstone Pandey et al., 2015 Clay/Claystone Carbonate ooze/ stone Silt/Siltstone C 30 C 31 TOC more C 4 C 35 C. D. E. A. B. Figure 3.7: Vegetation records from IODP Exp 355 Site U1457C. (A) Carbon isotope records of n-alkanoic acid C 30 , n-alkane C 31 and C 35 (B) Carbon isotope record of OC (C) Pollen records describe the dominant terrestrial vegetation; rela- tive terrestrial, marine and deltaic sourcing of pollen; and percentage of reworked pollen (D) BIT Index (E) Dominant lithology. 71 3.6.5 C 4 expansion in Indo-Arabian region pre and post 7 Ma Intervals of low BIT share other common characteristics including large (up to 91% relative to terrestrial and deltaic inputs) dinoflagellate percentages and more enriched δDwax values of ~-140%. These characteristics are consistent in both intervals (869-832 mbsf (9.8-8.3 Ma) and 667-605 mbsf (7.2-6.0 Ma)), while the main changes occur in the vegetation proxy records. Between 869-832 mbsf, δ 13 CC 31 values of ~-30%, reflect a mixture of C 3 and C 4 biomass sources (Fig- ure 3.7a). Grass pollen is not a major component, indicating that extensive C 4 grasslands are unlikely at this time, and instead we infer the 13 C-enrichment may be due to a mixture of C 3 plants in arid conditions, C 3 trees, and minor contribu- tions from C 4 grasses. The pollen assemblages are diverse and include near equal proportions of grasses, mosses and ferns, and gymnosperm trees and varying pro- portions of nonarboreal angiosperms, and arboreal angiosperms. We also note an influx of pollen from the aquatic freshwater plant, Potamogeton (up to 31% of the total terrestrial pollen) that we presume washes out of delta distributaries, and thus suggests a strong deltaic production and source of microfossils to the fan at this time. This interval is also characterized by relatively large contributions of marine dinoflagellate cysts relative to terrestrial pollen (41-91%). During the second low BIT interval from 667-605 mbsf (7.2-6.0 Ma), δ 13 C wax reflects C 4 values (ca -23%). Grass pollen increases to greater than 50% of the terrestrial pollen counts, whereas tree percentages decline. Arid-adapted shrubs andherbaceousplants(AmaranthaceaeandAsteraceae)alsoincreaseinabundance during this interval reflecting drier conditions in the source region. These changes on the landscape persist during the last interval of low BIT between 505 - 520 mbsf 72 (5.5-5.7 Ma) indicating that C 4 grasslands had become an established feature of the landscape. 3.6.6 Indus catchment vegetation changes pre and post 7 Ma During the first interval of high BIT between 822-653 mbsf (8.2-7 Ma), C 3 biomass is consistently recorded by δ 13 C values of bulk organic matter, as well as leaf wax n-alkanes and n-alkanoic acids (Figure 3.7a,b). This period is characterized by fast accumulation of sandy sediments. Deltaic input makes up a significant proportion of pollen inputs (31-68%): with maxima in deltaic vegetation such as mangroves and typha. Freshwater algae, Pediastrum, suggest efficient river transport of plant material from freshwater sources. Pollen from both deltaic and freshwater sources in addition to leaf wax n-alkane and n-alkanoic acidδD values of -206% indicate vegetation sources from higher in the catchment, which would be consistent with more depleted δD values and C 3 vegetation. Marine inputs (dinoflagellate cysts < 1%) and a high BIT index indicate that marine contributions are minor, either due to reduced productivity or more likely indicating rapid emplacement of this unit. During the second interval of high BIT sedimentation (600-533 mbsf or 6-5.7 Ma), elevatedδ 13 Calkane values, high ACL and high proportions of grass pollen (>50% grasses) suggests C 4 grasslands expanded to dominance. Wenote inparticular that chain lengths increase, consistent with high ACL in C 4 grasses relative to other plant types (Rommerskirchen et al., 2006; Vogts et al., 2009). Furthermore, the δ 13 CofthelongestchainlengthC 35 n-alkanesbecomesprogressivelymoreenriched relative to the other n-alkane chain lengths (C 27 -C 31 ; Figure 3.4). Similarly it has been noted that C 4 grasses contribute proportionally more to longer chain lengths 73 as seen in modern ecosystems for C 33 n-alkanes (Garcin et al., 2014; Hemingway et al., 2016) and in Miocene and Pliocene sediments for the C 35 n-alkanes (Uno et al., 2016), whereas angiosperm trees produce C 29 and C 31 dominantly (Feakins et al., 2016). We infer that C 4 grassland communities have become an established, perhaps dominant feature of the lowland paleo-Indus catchment. 3.6.7 Reconstructions of terrestrial hydroclimate To determine how hydroclimate changed during these distinct intervals of time, we can relate plant wax δD to the δD of precipitation after a large net or appar- ent fractionation. The interpretation of δDwax to δD of precipitation (δD precip ) constitutes a major uncertainty inδD precip reconstructions given the unknown con- tributions from different plant sources and the lack of modern plant calibration in the region. However, in woody, sub-humid to semi-arid southern California, the apparentfractionationisapproximately-90% (Feakins&Sessions,2010). Wecan furthertestthesensitivityofthereconstructiontoshiftsincommunitycomposition as suggested by pollen data (with the caveat that pollen and leaf wax production and transport differ) and can modulate the apparent fractionation with a pollen and C 3 /C 4 correction following Feakins et al. (2013). When we do this we find that the apparent fractionation varies between -112 and -140%. δD acid has two modes centering on -140% and -200%, whereas the calculated δD precip falls into three modes due to changes in vegetation communities. We infer that shifts in sourcing are the main driver of isotopic changes in this record. Values of ca. -75 to -82% suggest high altitude sources within the paleo-Indus catchment while ca. 2.4% and ca. -37% may be dominated by lowland sources within the paleo-Indus catchment or wind-blown transport from adjacent regions (Figure 3.8). 74 −225 −200 −175 −150 −125 −100 −75 −50 −25 0 25 50 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 Age (Ma) δD (‰) Figure 3.8: The relationship between ÎťD wax andδD precip can be determined based on a net fractionation factor, however changes in vegetation can alter this relation- ship. To account for possible biases, we followed the pollen correction approach of Feakins (2013). Using the pollen and C 3 /C 4 estimates of δ 13 C 30 , we used the following equation: corr = h f Amarantheaceae * shrub ] + h f grass * C 3 % * C3grass ] + h f grass * C4% * C4grass ] + h f tree * tree ] + alkane−acid . The values of different plant types are reported in Sachse et al., (2012), and an alkane−acid of 25±16% is reported in Chikaraishi & Naraoka (2007). Pollen counts were calculated for shrubs, trees, and grass. C 3 and C 4 percentages were estimated based on a two endmember mixing model using values from the data presented here, where a C 3 end member value of -32.7% and C 4 grassland end member value of -20.9% was used. The correction was applied to the WMA δD n-alkanoic acid. Compounded standard deviation includes uncertainty in analytical measurements, end member and the standard deviation of . To compare our δD precip to modern values, we calculated a mean weighted precipitation values of -26% at coastal Karachi (grey diamond), -31% at Jammu located in the Himalayan foothills (blue diamond), and -34% at Kabul, which lies further west of Jammu in Afghanistan (purple diamond). The coastal city of Mumbai is most enriched at -3% (red diamond). 75 Inthepaleo-Induscatchment, ourestimatesforδDprecipremainconsistentdespite changes in vegetation communities (Fig. 3.8). Between 823 and 673 mbsf (8.2- 7.3 Ma),δD precip values average -75%, and between 601-533 mbsf, δD precip values average-82% aftergrasslandexpansion. Consistencybetweenthesetwotimeperi- ods suggests that hydroclimate was not changing in the Indus catchment despite an expansion of C 4 grasslands. Increased input from the upper Himalayas would be consistent with a depleted δD value although C 4 plants are limited to below 3,000 m and are typically found at much lower elevations (Tieszen et al., 1979). It is possible that this range extended higher in the late Miocene under warmer temperatures. In the today, amount-weighted δD precip values from GNIP stations do not reflect values as negative as -80% (Figure 3.8). However, these values are observed during the winter months in Kabul, Afghanistan located in the Hindu Kush Mountains adjacent to the modern Indus catchment. The more negative δD precip estimates could reflect an increase in westerly-derived moisture sources due to the continental rainout effect on precipitation isotopes. In contrast, low BIT intervals are overall more enriched relative to high BIT inter- vals. Prior to C 4 expansion δD precip values average -37%. After C 4 expansion, δD precip values become more enriched by +39% and these values persist in the third low BIT interval. These more enriched δD precip values more closely reflect coastal precipitation sources that could be associated with Indian Summer Mon- soon rainfall. While changes in rainfall could drive C 4 expansion, changes in hydro- climatearenotexclusivelyrequiredforC 4 expansionafter7Maaswitnessedduring the high BIT intervals. Interestingly, we could be witnessing two separate hydro- logical regimes due to changes in source. 76 3.6.8 Late Miocene C 4 expansion The Indus Fan record of C 4 expansion adds to the evidence from marine records of the Indian ocean and terrestrial records from the Himalayan lowlands. Based on the shipboard age model, our Indus Fan late Miocene sediments range from 10-5.5 Ma. This interval spans the late Miocene ecological shift observed in the Siwaliks in Northern Pakistan and Nepal (Quade et al., 1995; Quade & Cerling, 1995). Between 10 and 8 Ma, Siwalik sediment record an unchanging C 3 -dominant landscape, and after 7 Ma, C 4 expands. Pollen was not preserved in Siwaliks sediments in Pakistan, but in the Nepalese sector pollen records confirm that by 7 Ma, grasslands replaced subtropical and broadleaf forests in the paleo-Ganges floodplain (Hoorn et al., 2000). Herbivore tooth enamel from the Siwaliks reveal an earlier appearance of C 4 on the landscape with horses reflecting a mixed C 3 /C 4 diet by ~8.5 Ma (Nelson, 2005, 2007). These trends are also reflected in the Indus sediments, in which regardless of sourcing, C 3 -dominant environments with large percentages of trees are present prior to 7.5 Ma, and C 4 grasslands become domi- nant after 7 Ma in both the Indus catchment and regions beyond. Shiftingfromtheterrestrialtothemarinerealm,lateMioceneleafwaxrecordsfrom the Arabian Sea and Bay of Bengal show coherence that is remarkable given the different source regions and dominant transport pathway (wind-blown vs. river export). ODP Site 722, located off the Oman Margin on Owen Ridge, receives dominantly terrestrial input via wind transport from regions including Pakistan, Iran, Afghanistan, the Arabian Peninsula and minor inputs from East Africa (Dahl et al., 2005; Huang et al., 2007; Van Campo, 1991). Direct comparison of n-alkane δ 13 C C 31 captures coeval trends in Site 722 and U1457 (Figure 3.9). In our Indus Fan record, δD n-alkanoic acid responds abruptly with changes in sedimentation patterns. During periods of low BIT and presumed wind-blown inputs to the Indus 77 Fan (~9.8-8.28 Ma and 7-6 Ma), δD values are similar between the two cores, whereas they diverge by up to 40% during intervals of high BIT on the Indus Fan when transport mechanisms differ between the two locations. Between 6 and 5.5 Ma, despite differences inδD wax and therefore source regions, C 4 persists in both records, highlighting the regionally expansive nature of C 4 biomass by this time in the regions surrounding the Arabian Sea. 78 −220 −180 −140 −100 Plant Wax δD (‰) −15 −10 −5 0 Tooth Enamel Paleosol Paleosol and Tooth Enamel δ 13 C (‰) −34 −30 −26 −22 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 Age (Ma) Plant Wax δ 13 C (‰) −12 −10 −8 −6 −4 Paleosol δ 18 O (‰, PDB) Oman Margin Indus Fan (this study) Bay of Bengal Oman Margin Indus Fan (this study) C. D. A. B. 79 Figure 3.9: Regional record of late Miocene vegetation change of plant leaf waxes from the Oman Margin (Huang et al., 2007), Bay of Bengal (Freeman & Colarusso, 2001), and Indus Fan (this study). Terrestrial records from the Himalayan foreland based on paleosols (Quade & Cerling, 1995; Quade et al., 1995) and fossil tooth enamel (Nelson, 2005, 2007; Morgan et al., 1994; Cerling & Harris, 1999). 3.6.9 Did precipitation changes drive the C 4 expansion? Plant wax records from the Gulf of Aden and the Bay of Bengal indicate C 4 expansion across a broad region during the late Miocene. Gulf of Aden sediment cores capture northeast African terrestrial conditions. Plant wax records (C 30 n- alkanoic acids) reveal ~4% δ 13 C enrichment shortly after 10 Ma (Feakins et al., 2013) and n-alkane C 31 show 4.5% around the same time (Uno et al., 2016). From theBayofBengal,sedimentsreflectingterrestrialmaterialdeliveredbytheGanges- Brahmaputra River system shift to more 13 C enriched values by ~6 Ma in leaf wax n-alkanes (Figure 3.9) (Freeman and Colarusso, 2001), as well as earlier records of δ 13 C shifts in the bulk organic carbon (France-Lanord & Derry, 1994). Collectively these records reveal that C 4 rose to dominance in the regions surrounding the Arabian Sea in the late Miocene following 7 Ma (Figure 3.9). Many hypotheses have been put forth to explain the drivers of this C 4 expansion including a late Miocene decline in pCO 2 , increase in aridity, seasonality of pre- cipitation, increase in disturbance such as fire and herbivory (Cerling et al., 1997; Dettman et al., 2001; Pagani et al., 1999; Sage, 2001). It can be challenging to directly quantify these various drivers with the geologic record and most likely these changes are interrelated. In the late Miocene, aridity became an expansive feature of the African/Indo-Monsoon region. Aeolian dune deposits mark the first evidence of desert conditions in the Sahara at 7 Ma. Dust flux records adjacent to north Africa increase and δD wax records from the Oman Margin reflect more 80 arid conditions on land (deMenocal, 1995; Huang et al., 2007). Modeling studies show that the Paratethys acted to moderate the thermal contrast between ocean and the continental interior. With the exposure of more land as the Paratethys sea reduced in size, a stronger summer thermal low could enhance summer monsoonal circulation by altering the atmospheric energy balance and creating a near-surface cyclonic anomaly that strengthens the Somali jet (Ramstein et al., 1997; Zhang et al., 2015). Furthermore, the exposure of the Saharan desert and drying of the Paratethys and expansion of highly reflective grasslands, carry albedo and transpi- ration feedbacks that may lead to further aridification. We find regional coherence in terrestrial and marine δ 13 C records that support a wide-spread expansion of C 4 vegetation and large-scale changes in atmospheric circulation and precipita- tion. The changes in plant leaf wax sources during intervals of turbidic sedimenta- tion and hemipelagic sedimentation provide insight into two different hydrological regimes. WithintheInduscatchment, hydrologicalchangesarenotassociatedwith C 4 expansion, however we observed reduced coastal precipitation from plant wax likely sourced from the Arabian Peninsula. 3.7 Conclusions The Indus Fan is characterized by episodic turbidite deposition. We present a multi-proxy record of late Miocene vegetation change. Taken together, organic, inorganicandmicrofossilproxiesprovideacomplimentaryinterpretivewindowinto the depositional history of Indus Fan sedimentation. We find that hemipelagic sed- iments and low BIT values are associated with higher inputs of marine microfossils and may be periods when wind-transported plant waxes dominate the biomarker 81 record. Turbidite deposits and high BIT intervals with negligible marine micro- fossil inputs and high proportions of terrestrial or deltaic pollen sources indicate efficient export and burial of terrestrial material primarily exported by the paleo- Indus River. Plant wax biomarkers and pollen record a distinct expansion of C 4 grasslands after 7 Ma regardless of changes in source. WhileδD precip sourced from the paleo-Indus catchment do not significantly change after C 4 expansion,δD precip sourced from the broader Indo-Arabian region reflect drier conditions after C 4 expansion, highlighting that changes in hydroclimate are not necessary for regional C 4 expansion. With this multi-proxy knowledge of fan sedimentation, we compare theIndusFanrecordwithsurroundingarchivestoprovideclearevidenceforregion- wide C 4 expansion throughout the Indo-Arabian region in the late Miocene. 3.8 Acknowledgements This research was funded by the US National Science Foundation (OCE 14-50528 to Consortium for Ocean Leadership, sub-award GG0093093-01 to S. Feakins and sub-award GG0093093-01 to S. Warny). This research used samples collected by the International Ocean Discovery Program, supported by funding from the US National Science Foundation and other member nations. We thank all participants of the shipboard science party and crew on Expedition 355 in particular D. Pandey, S. Ando, James Bendle and S. Bratenkov. We thank the following for laboratory assistance: Alexandra Figueroa, Jeremy Sunwoo, and Carl Johnson. 82 Chapter 4 Photosynthetic pathway of grass fossils from the upper Miocene Dove Spring Formation, Mojave Desert, California This manuscript is under review in 2017 as: Liddy, H.M., Feakins, S.J., Corsetti, F.A., Sage, R., Dengler, N., Whistler, D.P., Takeuchi, G.T., Faull, M., Wang, x. Photosynthetic pathway of grass fossils from the upper Miocene Dove Spring Formation, Mojave Desert, California. Palaeogeography, Palaeoclimatology and Palaeoecology 4.1 Abstract The spread of grasslands in the Miocene and of C 4 grasses in the late Miocene- Pliocene represents a major development in terrestrial plant evolution that affected the climate system and faunal evolution. The fossil record is sparse, likely due to the limited preservation potential of grasses. Diagnosis of the C 3 or C 4 photosyn- thetic pathway depends upon preservation of both cellular structures and organic carbon for isotope analysis. Here we analyze the anatomical and isotopic composi- tion of grass fossils from the Dove Spring Formation, Red Rock Canyon State Park, 83 California, USA, located in the El Paso Basin on the western side of the Basin and Range Province, a site previously identified as one of the earliest known C 4 fossil localities. We analyzed anatomical and geochemical characteristics of grass fossils dated to 12.01-12.15Ma. Thefossils in this study include grass shoots and in cross- section display anatomy indicative of the C 3 photosynthetic pathway. We isolated organic carbon from the stem fossils and determined the carbon isotopic compo- sition to be −24.8 ± 0.5%. Together, the anatomical and geochemical analyses confirm that these plants used the C 3 photosynthetic pathway. Our findings are consistent with dietary evidence based on tooth enamel from grazing mammals of available C 3 resources in the same sections. These newly reported Miocene- age C 3 grass fossils contribute to a sparse fossil record of grass evolution and the reevaluation of the previously reported C 4 grasses from this locality suggest that these fossils were likely using the C 3 pathway as well. Overall, paleoecological reconstructions at this site indicate more humid conditions during the Miocene compared to the modern Mojave Desert with widespread grasslands and diverse grazing mammals. 4.2 Introduction Grasses found intropical and subtropical grasslandor savanna ecosystems predom- inately use the C 4 photosynthetic pathway (Sage, 2001). This pathway concen- trates CO 2 around the photosynthetic enzyme, Rubisco, conferring a competitive advantage over the C 3 photosynthetic pathway in warm, dry and/or low CO 2 con- ditions (Ehleringer et al., 1997). In C 3 plants, both carbon assimilation and carbon reduction occur in the mesophyll cells. In C 4 plants, these processes are spatially separated resulting in the internal concentration of CO 2 , where CO 2 is assimilated 84 into mesophyll cells and then shuttled into vascular bundle sheath cells where carbon reduction occurs via the Calvin cycle (Hatch, 1987). As a result of this process, the δ 13 C org of C 4 plants is typically more 13 C enriched than co-occurring C 3 plants. Anatomically, the C 4 photosynthetic pathway is commonly manifested as a double-celled ring described as Kranz (wreath) anatomy (Brown, 1975). How- ever, additional anatomical configurations have been documented, including iden- tification of the C 4 pathway within a single photosynthetic cell (Voznesenskaya et al., 2001). Kranz anatomy and other documented anatomies suggest the use of the C 4 pathway, but this is more securely confirmed in combination with isotopic evidence. The history of grasses and grassland evolution is largely determined by geochem- ical and microfossil evidence. This includes geochemical records of stable carbon isotopes of vegetation recorded in paleosols (Fox & Koch, 2003; Levin, 2013; Quade & Cerling, 1995), herbivore tooth enamel (Cerling et al., 1997; Passey et al., 2002; Tipple & Pagani, 2010), plant waxes (Feakins et al., 2013; Tipple & Pagani, 2010) , microfossil records of plant silica (phytolith) assemblages (McInerney et al., 2011; Strömberg, 2005) and pollen (Bonnefille, 2010). The earliest evidence of grasses based on phytolith evidence is from the late Cretaceous ca. 67 Ma (Prasad et al., 2005). Although grass dominated habitats did not become widespread on many continents until the Miocene (Strömberg, 2011). In the Oligocene, ca. 30 Ma, a major drop in CO 2 may have driven the evolution of alternative strategies for carbon fixation including C 4 and Crassulacean acid metabolism (CAM) photosyn- thetic pathways (Sage et al., 2012). Molecular dating using fossilized grass pollen and inflorescences place initial C 4 appearances in the Oligocene (Christin et al., 2008; Vicentini et al., 2008). Despite evidence of C 4 origins in the Oligocene, grasslands were dominantly C 3 throughout the early-middle Miocene as evidence 85 for C 4 grass is modest (<30% of the landscape), if present (Edwards et al., 2010). Beginning ca. 10 Ma, C 4 grasslands expanded, largely replacing C 3 grasslands throughout the tropics and subtropics (Cerling et al., 1997; Feakins et al., 2013). Few studies offer both anatomical evidence of the photosynthetic pathway as well as isotopic evidence. Grass fossils from the Middle Miocene (~14 Ma) were found at the Fort Ternan Formation in Kenya (Retallack et al., 1990). Although internal anatomy was not preserved, the cuticle morphology was proposed to be analogous to modern C 4 grasses (Retallack et al., 1990). However, the carbon isotopic value of paleosols from that locality are comparable to modern C 3 -dominated environ- ments, and no additional evidence of C 4 grasses was found (Cerling et al., 1991). Grass fossils from the western Mojave Desert, California, from the Ricardo Forma- tion (now Dove Spring Formation) of the El Paso Basin were initially interpreted as Pliocene-age and later revised as upper Miocene-age (Nambudiri et al., 1978; Tidwell & Nambudiri, 1989). These fossils were reported to have Kranz anatomy and small interveinal distances indicative of C 4 photosynthesis (Nambudiri et al., 1978; Tidwell & Nambudiri, 1989). Aδ 13 C value of−13.7% was obtained for bulk carbon within the fossil, however the hydrofluoric digestion methods for prepara- tion of the fossil would remove silicates but not carbonates, and thus carbonates present in the matrix of this fossil would bias the result towards more enriched, C 4 - like values. A third late Miocene grass fossil has been reported from the Ogallala Formation (7-5 Ma) in northwestern Kansas (Thomasson et al., 1986). The grass fossil was inferred to be C 4 based on the presence of Kranz anatomy. The fossils were assessed to be completely permineralized and carbon isotopic measurements were not attempted (Thomasson et al., 1986). This amounts to a sparse fossil record of C 4 grass fossils. 86 In order to more robustly investigate the presence of early C 4 plants via anatomical and isotopic evidence, we revisited the grass fossil site in the Dove Spring Forma- tion (formerly Ricardo Formation), western Mojave Desert, California (Nambudiri et al., 1978; Thomasson et al., 1986). The grass fossil in that study had been donatedandwascollectedfromanunspecifiedlocalitywithinLastChanceCanyon. The authors reported that the fossils were destructively analyzed and additional efforts to collect more specimens were unsuccessful (Tidwell & Nambudiri, 1989). The unknown locality is problematic in placing age constraints on this grass fossil because Last Chance Canyon refers broadly to a region in Red Rock Canyon State Park, CA, USA, which includes Pleistocene-age strata. Indeed the first publica- tion reported Pliocene (Nambudiri et al., 1978) and the second publication on the same fossil reported upper Miocene-age (Tidwell & Nambudiri, 1989). Therefore, we returned to Last Chance Canyon specifically to the Dove Spring Formation and found additional grass fossils in stratigraphic context that allow us to date these new fossils to the upper Miocene. We describe the taphonomy, microstructure and isotopic composition of these fossils. The anatomical description and isotopic composition of these grass macrofossil samples provide the means to diagnose the photosynthetic pathway by the same criteria applied to modern plants. 4.3 Stratigraphic context 4.3.1 Geologic setting The fossil-bearing Dove Spring Formation is located within the El Paso Basin at thewesternendoftheBasinandRangeProvinceandisboundedtothewestbythe Sierra Nevada (Fig. 1A). Uplift of the El Paso Mountains along the El Paso Fault exposed >6.2 km of sedimentary and volcanic rock. The Ricardo Group is divided 87 intotwoMiocene-ageformations: theCudahyCampFormationcomprisedof450m of dominantly volcanic rocks and the Dove Spring Formation comprised of 1800 m of fluvial, lacustrine and volcanic rocks (Dibblee, 1952; Loomis & Burbank, 1988). TheDoveSpringFormationliesdisconformablyovertheCudahyCampFormation, andQuaternarydepositsoriginatingfromtheSierraNevadalieunconformablyover the Dove Spring Formation (Whistler et al., 2009). The Dove Spring Formation, divided into six members spanning 8-12.5 Myr, was deposited in an elongate, fault-bound trough (Loomis & Burbank, 1988; Whistler etal.,2009). Thesedimentaryandvolcanicfillconsistsofconglomerate, sandstone, mudstone, chert, basalt, and tuff. Specifically, five major paleoenvironments are identified including (1) fine-grained lacustrine deposits, (2) coarse-grained fluvial deposits of channel sandstone and channel conglomerates, (3) finer-grained over- bank and floodplain silts, (4) poorly sorted alluvial fan deposits, and (5) paleosol caliche and silicified hardpan deposits (Whistler et al., 2009; Whistler & Burbank, 1992). These sedimentary deposits contain a diverse fossil assemblage of at least 86 species of fossil vertebrates that are Clarendonian though earliest Hemphillian- aged (Merriam, 1919; Tedford et al., 1987; Wood et al., 1941). The fluvial and lacustrine rock sequence of the Dove Spring Formation was deposited by ephemeral braided streams draining into a semi-permanent lake in the center of a broad basin (Loomis & Burbank, 1988; Whistler et al., 2009). In the lower part of the formation, paleocurrent orientation is preferentially directed toward the north-northwest suggesting a source to the south from the uplifted Mojave block (Loomis & Burbank, 1988). At 10 Ma with the initiation of sinistral shear on the Garlock fault, the El Paso Basin began to rotate counter clockwise, and by 9 Ma, basin and range style extension began in this region (Burbank & Whistler, 1987; Loomis & Burbank, 1988). In the upper strata of the formation 88 8.5 ± 0.15 Ar/Ar (a) MEMBER 6 MEMBER 5 MB4 MEMBER 3 MEMBER 2 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 Time (Ma) 8.4 ± 1.8 FT (b) 9.7 ± 0.2 Ar/Ar (c,d,e) Celetron 2 10.2 ± 0.2 Ar/Ar (c,d,e) OC3 10.6 ± 0.2 Ar/Ar (c,d,e) OC2 10.4 ± 1.6 FT (b) , 11.01 ± 0.03 Ar/Ar (c,d,e) CPT XIII 10.5 ± 0.25 Ar/Ar (a) 11.2 ± 0.1 Ar/Ar (c,d,e) CPT XII 11.64 ± 0.05 Ammonia Tanks (c,e) 11.8 ± 0.9 FT (f) 11.83 ± 0.05 Ar/Ar (c,e) Ranier Mesa 11.7 ± 0.2 Ar/Ar (g) 12.01 ± 0.03 Ar/Ar (c,d,e) 12.15 ± 0.04 Ar/Ar (c,e) unconformity 15.1 ± 0.5 K/Ar (b) 200 400 600 800 1000 1200 1400 1600 1800 Thickness (m) AGE CONTROL B. ^ _ -117° W -117° W -118° W -118° W -119° W -119° W -120° W -120° W -121° W -121° W 39° N 39° N 38° N 38° N 37° N 37° N 36° N 36° N 35° N 35° N 34° N 34° N 33° N 33° N California Nevada Fossil Locality Pacific Ocean Elevation (masl) 0 - 500 500 - 1,500 1,500 - 2,500 2,500 - 3,500 3,500 - 4,500 ¯ 0 100 200 Kilometers A. 39°N 38°N 37°N 36°N 35°N 34°N 33°N 121°W 117°W 120°W 119°W 118°W C. Figure 4.1: (A) Regional map of the fossil locality (star) at Red Rock Canyon, California, USA. (B) Fossil plant embedded hand sample collected the Natural History Museum of Los Angeles County (LACM) locality 8019. (C) Stratigraphic section of the Dove Spring Formation modified from Whistler et al. (2009). The age chronology includes dating methods such as fission track (b) Cox and Diggles, (1986); (f) Loomis & Burbank (1988), Ar/Ar radiometric dates (e) Bonnichsen et al. (2008); (c) Perkins et al. (1998); (d) Perkins & Nash (2002); (g) Smith et al. (2002); (a) Whistler & Burbank (1992), K/Ar radiometric dates (b) Cox and Diggles, 1986, and biostratigraphic correlations (e) Bonnichsen et al. (2008); (c) Perkins et al. (1998). (Members 5 and 6), paleocurrent evidence is sparse and a transition in lithology to coarser clast sandstones and conglomerates is interpreted to be poorly-sorted allu- vial fan deposits indicative of a Sierra Nevada detrital source (Loomis & Burbank, 1988) (Fig. 1C). Within these uppermost sections (upper 200 m), well-developed 89 paleosol, caliche and silicified hardpan deposits indicate a shift to more arid con- ditions in the basin at this time (Whistler et al., 2009). Throughout the Dove Spring Formation, the sedimentary fill is punctuated by pyroclastic and basaltic volcanic flows, which provide material for radiometric dating of the Dove Spring Formation sequence (Bonnichsen et al., 2008; Cox and Diggles, 1986; Perkins et al., 1998; Perkins and Nash, 2002; Smith et al., 2002; Whistler et al., 2009). 4.3.2 Tephra and biostratigraphic age control A total of 18 volcanic air-fall vitric ashes have been mapped throughout the Dove Spring Formation (Whistler et al., 2009). Active volcanism was occurring in the southern Great Basin and the Yellowstone ‘Hot Spot’ in the late Miocene (Bon- nichsen et al., 2008; Perkins et al., 1998; Perkins & Nash, 2002). Two basalt flow sequences, each including several individual flows, are sourced south of the Gar- lock Fault in the Lava Mountains (Monastero et al., 1997; Smith et al., 2002). The radiometric dates span 12.15 ± 0.04 Ma (Cougar Point Tuff V in Member 2) to 8.5 ± 0.15 Ma (near the top of the formation in Member 6) (Perkins et al., 1998; Bonnichsen et al., 2008; Whistler & Burbank, 1992) (Fig. 1C). The oldest volcanic strata that provide the most basal age constraint for the Dove Spring Formation is dated to 12.15 ± 0.04 Ma followed by the Ibex Hollow tephra dated to 12.01 ± 0.03 Ma (Perkins & Nash, 2002; Whistler et al., 2009). 4.4 Methods 4.4.1 Sample collection We revisited the Red Rock Canyon fossil locality that has yielded a fossil grass thought to be the oldest known C4 grass fossil (Nambudiri et al., 1978). The 90 fossil was reported from the Ricardo Formation (now Dove Spring Formation) in the Last Chance Canyon area. Unfortunately, the precise stratigraphic location of the Nambudiri et al. (1978) fossil is unknown, but is thought to have been collected from Members 4 and 5 (as named by Dibblee (1952) and later updated to Member 2 by Loomis & Burbank (1988), which spans an age range of ~12.01 ± 0.1 Ma to 10.5 ± 0.25 Ma (Loomis & Burbank, 1988; Tidwell & Nambudiri, 1989; Whistler et al., 2009). Our fossils were recovered from the Natural History Museum of Los Angeles County (LACM) locality 8019, Member 2 of the Dove Spring Formation. This locality lies between two ash layers: below the Ibex Hollow tephra dated to 12.01 ± 0.03 Ma (Bonnichsen et al., 2008; Perkins et al., 1998; Perkins & Nash, 2002) and in proximity to the Cougar Point Tuff V dated to 12.15 ± 0.04 Ma (Bonnichsen et al., 2008; Perkins et al., 1998; Whistler et al., 2009). Five thin sections were catalogued and housed in the Department of Vertebrate Paleontology, Natural History Museum of Los Angeles County: LACM 160024 (RR1A and RR1B), LACM 160025 (RR2B), LACM 160026 (RR4A and RR4B). Two additional hand samples were collected but did not yield viable thin section images for anatomical analysis. 4.4.2 Fossil imaging Field-excavated samples were cut with a rock saw to remove weathered surfaces. Samples were cut to hand-sample size to reveal as many leaf stems in cross section as possible. Surfaces were polished to identify well-preserved stems. Thin sections were prepared to 7.62cm × 5.08cm with 30 μm thickness and clear epoxy (by Wagner Petrographics, Lindon, Utah). Thin section samples were analyzed using light microscopy (Zeiss Imager M2m with a Zeiss Axiocam MRc) at USC and 91 images prepared with MOSAIC software to reveal the internal anatomy of the fossil leaves on features larger than single microscope images. 4.4.3 Carbon isotope measurements Hand-sampleswerebrokenaparttoseparatefossilsfromthematrixinajawcrusher that was cleaned of possible organic material using igneous rock blanks. Silicified grass stems were powdered in a solvent-rinsed ball mill, and the matrix was drilled to powder. Up to 5 g of sediment was added to 40 mL of 1M hydrochloric acid and heated to 70 ◦ C in a water bath for 4 hours to remove carbonates including calcite as well as more refractory carbonate phases (e.g., dolomite) (Brodie et al., 2011; Galy et al., 2007; Ward et al., 2007). Samples were rinsed with deionized water three times then dried in a drying oven at 50 ◦ C for 24 hours. The low total organic carbon (TOC) content of the samples and matrix required sample sizes of 257−308 mg (packaged in Sn capsules). The TOC and isotopic composition of organic carbon (δ 13 C org ) was determined using a Costech Elemental Combustion System (EA 4010) connected to a Picarro cavity ring down spectrometer (G2131-i) via a PicarroLiaison(A0301). Twoblanksandtwostandards(USGS-40[GlutamicAcid] δ 13 C org = −26.6%) and bulk calcite (δ 13 C org = −37.9%) were run in replicate at the beginning and end of the sample run to standardize measurements to the Vienna Pee Dee Belemnite (VPDB)-Lithium carbonate standard prepared by H. Svec (LVSEC) isotopic scale. δ 13 C org values are reported using delta notation in per mil (%) units relative to the VPDB-LVSEC isotopic scale. Replicate standard precision was better than 0.23%, and values were within 0.24% of known values. Standards were diluted with kaolinite to test reproducibility when measuring low TOC materials, with reproducibility of these large mass aliquots within 0.6%. 92 Replicates of different aliquots of the matrix and stem sample were reproducible to within 0.37%. 4.4.4 Inorganic carbon isotopic measurements The matrix of the embedded plant material was drilled to powder. Of this powder, 12 mg was weighed into 10 mL glass Exetainer vials with rubber septa caps. Vials were evacuated and pre-acidified with 1 mL 30% H 3 PO 4 . Samples and standards were heated for 80 minutes in a water bath at 70 ◦ C to ensure that all carbon asso- ciated with more refractory phases (e.g. dolomite) were released as CO 2 . Samples were run on a Picarro CRDS coupled to an Automate preparative device. Two blanks and two standards (USGS-40 [Glutamic Acid] δ 13 C = −26.4%) and bulk calcite (δ 13 C = 2.47%) were run in replicate at the beginning and end of the sample run to standardize measurements to the VPDB)- LVSEC isotopic scale. Replicate standard precision was better than 0.022%, and values were within 0.000067% of known values. 4.5 Results 4.5.1 Plant fossil descriptions Grass shoots were abundant in the excavated sections. Here we present the results of thin sections that reveal preserved internal anatomy. Each sample represents a subsample of the specimens collected in the field as preservation of internal struc- tures varied between the hand samples. In general, the permineralized plant fossils LACM 160026 (RR4B1), LACM 160025 (RR2B), and LACM 160024 (RR1A) are likely from the Poaceae family (Fig. 2A-C). Shoots with concentrically arranged leaves range in diameter from 3 to 4.4 mm. Prominent mid-veins (500 μm in 93 diameter) are observed within ring-shaped leaves that surround a central culm. Additional smaller vascular bundles ( 200μm) occur within individual leaf sheaths best observed in samples LACM 160026 (RR4B1), LACM 160025 (RR2B), LACM 160024 (RR1A), and LACM160024 (RR1B) (Fig. 2A-C and Fig. 3C). LACM 160026 (RR4B1) is approximately 3.75 mm in diameter (Fig. 2A). The midvein is a circular feature that is approximately 500 μm in diameter consisting ofbundlesheathcellsandinternalvasculartissue. Themidribisthethickenedpart of the grass sheath that surrounds the midvein. In this sample it is oval shaped (1.5× 0.75 mm). Interior to the thickened midrib is parenchymatous tissue that forms the triangular area in the center of the specimen and likely represents the grass culm or stem (Fig. 2A). Surrounding the culm are flattened organs that are likelyleafsheathswithwidelyspacedvascularbundleswithsmallerdiameters(200 μm) than the midvein. The wide spacing between the vascular bundles suggest the section of the grass stem was preserved near the base, below the horizon of the stem in which leaf blades form. LACM 160025 (RR2B) has a similar anatomy to LACM 160026 (RR4B1) (Fig. 2A and B). The sample is less symmetrical with an approximate length of 4.4 mm and width of 2.4 mm. The internal tissues are of comparable sizes with a prominent midvein of 500μm and a thickened midrib that surrounds the midvein. The stem forms a triangular shape that consists of parenchymatous tissue. Vascular bundles located within the innermost tissue of the leaves are 200 μm in diameter. They are spaced approximately 1.5 mm apart. Although the preservation is less clear, leaf sheaths seem to again wrap around the culm with well-preserved vascular bundles that represent the midvein of the leaf sheaths. These vascular bundles are of comparable size (~200μm) to the midveins of the leaf sheaths in LACM 160026 (RR4B1). 94 C C M L V V V V 500 μm B M V P V L 500 μm A B M A M V L V V P 500 μm Figure 4.2: Thin cross-section images of Miocene grass shoot fossils (A) LACM 160026 (RR4B1), (B) LACM 160025 (RR2B), and (C) LACM 160024 (RR1A). Arrows point to vascular bundles (V), large midvein (M), and leaf sheaths (L) that are arranged concentrically around the central culm. Scale bar represents 500 Îijm unless otherwise noted. The inset box in C enlarges two adjacent leaf sheaths that surround the central culm and highlight a larger vascular bundle with two vascular bundle sheaths including an inner mestome sheath and an outer parenchymatous bundle sheath. 95 Much of the internal structure of LACM 160024 (RR1A) was not preserved (Fig. 2C). However, the features preserved seem to be similar to LACM 160026 (RR4B1) andLACM160025(RR2B).Whiletheinternalfeaturesofthecentralculmwerenot preserved, well-preserved vascular bundles are intercalated with flattened organs that likely represent leaf sheaths surrounding the culm (Fig. 2C). The diameters of the vascular bundles are comparable to their counterparts (200 μm) in LACM 160026 (RR4B1) and LACM 160025 (RR2B). However, some smaller bundles are observed with diameters of 150 μm. The inset in Fig. 2C displays four vascular bundles with one showing two bundle sheath layers: an inner mestome sheath and an outer parenchymatous bundle sheath. The spacing between vascular bundles is more clustered than in the previous samples, which may be a result of taphonomic alteration. LACM 160026 (RR4B2) is more degraded with some preserved cross sections of grass leaf sheaths and vascular bundle (Fig. 3A). The vascular bundles range in diameter from 100−200 μm. While it is common in these samples for the more labile components of the leaf to be degraded, the more resilient suberized mestome sheath are more likely to persist after other tissues degrade. LACM 160026 (RR4A) is comparable anatomically to LACM 160026 (RR4B1) and LACM 160025 (RR2B) but is approximately half the diameter at ~2 mm. The internal structure is not well preserved; however, vascular bundles of 150−250 μmindiameterareembeddedwithinlayersofcellsthatlikelyrepresentleafsheaths (Fig. 3B). While the tissue in the center is not well preserved, it does not give the appearance of a parenchymatous culm as LACM 160026 (RR4B1) and LACM 160025 (RR2B). The small size and possible lack of development suggests that this sample reflects a developing grass inflorescence, which would appear as small structures with a variety of orientations. 96 ѥP ѥP ѥP A B C V V V M V L V V V V v 100 μm 100 μm 100 μm Figure 4.3: Thin cross-section image of Miocene grass shoot fossils. Scale bar represents 100 Îijm. (A) LACM 160026 (RR4B2). Arrows point to preserved grass bundle sheaths (V) that are surrounded by degraded tissue. (B) LACM 160026 (RR4A). Arrows point to vascular bundles (V), large midvein (M), and leaf sheaths (L) that are arranged concentrically around the central culm. (C) LACM 160024 (RR1B). May represent a grass culm with four larger vascular bundles (V) and one small vascular bundle (v). LACM 160024 (RR1B) has a stem with the smallest diameter of 475 μm (Fig. 3C). This sample exhibits radial symmetry. The five circular cell arrangements inside the stem may represent vascular bundles. Four large vascular bundles have diameters of ~100 μm while one small vascular bundle has a diameter of ~50 μm. 4.5.2 Stable carbon isotopes The stable carbon isotope value of the fossil material and surrounding matrix are listed in Table 1. The fossil material and surrounding matrix had low TOC values of 0.001% and 0.003%, respectively. Theδ 13 C org value of the grass fossil is−24.4± 0.03%, andtheδ 13 C org valueofthematrixδ 13 C org is−27.3±0.42%. Thematrix is significantly more negative than the fossil material (p < 0.01), which suggests that the organic carbon from the fossil is not contaminated by the organic carbon 97 Samples δ 13 C (%) std. dev. (%) n Grass fossil -24.4 0.03 2 Matrix TOC -27.3 0.42 2 Matrix TIC -3.79 0.05 3 Table 4.1: Carbon isotopic composition of fossil material and matrix material from the same hand sample that the thin section images (LACM 160024-LACM 160026) were taken. of the matrix. The TIC content of the matrix is 7.75% and theδ 13 C value is−3.79 ± 0.05%. Two other hand samples had imbedded fossilized plant material but did not yield viable thin section images. The δ 13 C org value of these plant fossils is −25.3 ± 0.03% and −24.6 ± 0.5%, and the matrix δ 13 C org value is −26.9 ± 0.34% and −31.3 ± 0.89% (1σ, n = 3), respectively. The matrix is also significantly more negative than the fossil material (p < 0.01), which suggests that the organic carbon from the fossil is not contaminated by the organic carbon of the matrix. The TIC content of the matrix is 10.2% and 9.06% and the δ 13 C value is −4.97 ± 0.02% and −4.27 ± 0.05%. 4.6 Discussion 4.6.1 Grass fossils from the Dove Spring Formation We revisited the locality of the oldest known C 4 grass fossil from Red Rock Canyon State Park, CA, USA to recover additional grass fossils. Grass fossils were not ubiquitous throughout the Dove Spring Formation, and a search for additional fos- sils yielded only the material presented in this study. Anatomically, these fossils differ from the previously reported grass fossils (Nambudiri et al., 1978; Tidwell 98 & Nambudiri, 1989). Flattened organs surrounding the main culm are sugges- tive of sheathing leaf bases, rather than photosynthetic leaf blades. Most of the structures preserved are lignified sections of whole shoots, and the preservation of photosynthetic tissues was limited. The plant fossils reported here are likely C 3 grass shoot cross-sections from the Poaceae family with well-preserved vascular bundles and leaf sheaths surrounding a central culm. We infer that the plant fossils presented here display C 3 plant anatomy. No C 4 grass fossils were recovered from this locality. 4.6.2 Miocene carbon isotope values The grass fossils reported in this study have a meanδ 13 C value of−24.8± 0.5%, which indicates the use of the C 3 photosynthetic pathway based on the range of modern C 3 plants (Cerling and Harris, 1999; Tipple and Pagani, 2007). However, the carbon isotopic composition of plant material reflects both the photosynthetic pathway used by the plant to fix carbon dioxide from the atmosphere as well as the carbon isotopic composition of atmospheric carbon dioxide (Farquhar et al., 1989), which has varied across geological time. The late Miocene average δ 13 C of atmospheric CO 2 reconstructed from the δ 13 C values of benthic foraminifera from North Atlantic ODP Sites 553, 558, 563, and 601 is −6.0 ± 0.2% (maximum = −5.4% and minimum =−6.5%) (Tipple & Pagani, 2010). Other reconstructions of the δ 13 C of CO 2 based on planktonic foraminifera come to similar conclusions with values varying between âĂŞ5.0âĂř to ÂňâĂŞ6.2âĂř (Passey et al., 2002). Therefore, the Miocene δ 13 C of CO 2 was on average higher than pre-industrial (−6.5%) (Friedli et al., 1986) and modern (−8.5%, 2015) (Keeling et al., 2001). Assuming late Miocene atmospheres of−6.0%, we can adjust the C 3 endmembers of modern plants collected when δ 13 C of CO 2 was close to −8.0% by 2% such 99 that C 3 plant values would shift from −20% and −35% to −18% and −32% (Cerling and Harris, 1999; Tipple and Pagani, 2007). Therefore we have high confidence that the Miocene grass fossil carbon isotopic value of ~−25% indicates these grasses used the C 3 pathway. 4.6.3 C 4 grass fossil revisited Grass fossils previously reported from the Dove Spring Formation were interpreted to use the C 4 photosynthetic pathway based on anatomical and isotopic evidence (Nambudirietal., 1978; TidwellandNambudiri, 1989). Herewere-visittheimages from these studies. Our analysis of the interveinal distances and bundle sheath size suggests that the fossilized grasses were in fact using the C 3 as opposed C 4 photosynthetic pathway. In C 4 grasses, the distance between veins is characterized by 60-150 μm or 1-4 mesophyll cell diameters, while this distance in C 3 grasses is generally >200μm or more than 5 mesophyll cell diameters (Dengler et al., 1994). Additionally, C 4 plants generally have an enlarged bundle sheath cell relative to the mesophyll cells (Dengler and Nelson, 1998; Griffiths et al., 2012; Hattersley, 1984; Sage, 2004). As shown in Fig. 4A and B, the interveinal distance between vein centers is overall greater than 150 μm. This value is appropriate for grass leaf interveinal distance but is outside the range for C 4 grasses (Dengler et al., 1994). In Fig. 4C, the vascular tissue of the veins are surrounded by sclerenchyma tissue with identifiable parenchymatous bundle sheath cells at the periphery. The parenchymatous bundle sheath cells are indistinguishable from C 3 bundle sheath tissue. The anatomy of the fossil images presented in these studies do not present unequivocal evidence for C 4 anatomy. One of the defining characteristics of the previously reported fossil grasses was the isotopically 13C-enriched values (−13.7%) (Nambudiri et al., 1978). However, we 100 ȝP A ȝP ȝP B C 100 μm Figure 4.4: (A) Cross section of the culm. Circular features inside the culm rep- resent vascular bundles. (B) Two vascular bundles with a scale bar of 100 μm. (C) Vascular tissue of veins surrounded by sclerenchyma tissue with identifiable parenchymatous bundle sheath cells at the periphery. Images are reprinted from Review of Palaeobotany and Palynology, 60, William D. Tidwell and E.M. Nam- budiri, Tomlinsonia Thomassonii, Gen. et sp. Nov., A permineralized grass from the Upper Miocene Ricardo Formation, California, 165-177. (1989) with permis- sion from Elsevier. caution that the method of preparation used hydrofluoric acid to digest silicates may not have been sufficient to remove carbonates, which have a 13 C-enriched composition relative to organic matter. Assuming a small TOC concentration in the sample and that the δ 13 C of TIC is substantially heavier than typical organic matter, a small amount of recalcitrant carbonate would bias the organic carbon analyses towards heavier values. It is not possible to verify this result with high- temperature decarbonation methods to establish theδ 13 C of organic matter as the fossil material was destructively sampled for the initial carbon isotopic analysis (Nambudiri et al., 1978; Tidwell & Nambudiri, 1989). Although for our sample, given that theδ 13 C of TIC is−4%, as little as 0.0005% of recalcitrant carbonate could have produced similar C 4 -like values. Overall, the evidence for the C 4 grasses at this upper Miocene locality is not firmly established and additional grass fossils 101 recovered from the Dove Spring Formation conclusively used the C 3 photosynthetic pathway. 4.6.4 Miocene-age paleoecological reconstructions from the present day Mojave Desert From the Dove Spring Formation, additional silicified plant remains are limited to petrified wood assemblages and further evidence of other C 4 clades were not found. Tree species characteristic of low elevations in close proximity of streams includes palms (Palmoxylon mohavensis) and locusts (Robinia alexanderi), whereas oaks (Quercus ricardensis), pines (Pinus kelloggi) and cypress (Cupressus) trees could be found on upstream slopes farther away from the streambeds (Webber, 1933). Fluvial and lacustrine sediment facies suggest a lake environment with braided streams (Whistler et al., 2009). In the fossil images, evidence of open channels such as the large space in the center of LACM160026 (RR4A) and the possible veins or lacticifers of LACM160024 (RR1B) may represent aerenchyma air channels, a common feature of wetlands plants (Fig. 3B and C). The robust size of the plant fossils suggests that they grew in a wet, marshy environment, consistent with the paleoenvironmental interpretation of the lithology. While it is possible for C 4 grasses to be associated with aerially restricted ground-water fed springs with water available in summer months or around saline lakes, we confirmed that these grasses used the C 3 photosynthetic pathway. The presence of widespread grasses at this locality can be inferred from the preser- vation of grazing mammals including equids and antilocaprids throughout the Dove Spring Formation. Herbivore tooth enamel records the availability and/or uptake of dominantly C 3 resources (Bowman et al., 2015). Taxa including grazers, browsers and mixed feeders have enamel values ranging from −13.3 to −6.7% 102 with an average value of −10.1% (Bowman et al., 2015). Enamel values greater than −8.0% only accounted for 1% of the total individuals sampled. A δ 13 C enamel value of -8.0% is considered the modern cut off of C 3 plants (Feranec and Pagnac, 2013), however this value would shift to −6% considering the more 13 C enriched atmosphere of the Miocene (Bowman et al., 2015). Therefore, the possibility of the consumption of C 4 plants is unlikely given the more 13 C enriched Miocene atmospheres at this time (Bowman et al., 2015). Overall grazers had the most positive values, and these values may reflect water stressed C 3 plants. Serial sampling of the herbivore teeth did not reveal a seasonal signal in δ 13 C indicat- ing a consistent diet of C 3 resources among the individuals sampled (Bowman et al., 2015). Herbivore diets indicate that C 3 grasslands were a readily available resource on the landscape, and C 4 grasslands were either absent or not exploited by the wide range of taxa sampled from the Dove Spring Formation. Mid Miocene age (14-13.4 Myr) herbivore tooth enamel from the Barstow For- mation were interpreted to support an early C 4 expansion in the Mojave region (Feranec and Pagnac, 2013). While the diets of most browsers and grazers includ- ing antilocapridae, camelidae, and gomphotheriidae consisted of pure C 3 resources, the horses, equids, had diets slightly more enriched in 13 C (−7.8 ± 0.8%) with values up to −6.2% (Feranec and Pagnac, 2013). The authors applied a linear mixing model assuming a C 3 end member of −8% to estimate the percentage of C 4 plants consumed by equidae. Using this approach, they approximated that up to 18% of C 4 plants were included in the diets of some ungulates. However, linear mixing models fail to account for the large range of 13 C values of C 3 plants, and propagated error associated with the C 3 isotopic range (2.3%) and uncertainty associated with 13 C CO 2 estimates (0.2%) translates into an error estimate of ± 20% C 4 . Therefore, we cannot assume with confidence that C4 was incorporated 103 in equidae diets from the Barstow Formation during the upper Miocene. Taken together, late Miocene herbivores from the Mojave Desert likely did not exploit C 4 resources either due to the absence of C 4 on the landscape or a C 3 dietary preference of all the taxa sampled. 4.6.5 North American grassland expansion The development of North American grasslands occurred in two phases (Edwards et al., 2010; StrÃűmberg, 2011). The late Oligocene and early Miocene (ca. 25-20 Myr) witnessed the initial spread of C 3 grasslands and open woodlands (StrÃűm- berg, 2011; 2002). Early evidence of the presence, but not dominance, of C 4 plants is observed in the early-middle Miocene. By 19 Ma, phytoliths indicative of C 4 Chloridoideae species are present in grassland assemblages from the Great Plains (StrÃűmberg, 2005). Leaf wax 13 C records from the Gulf of Mexico document a peak in 13 C-enriched 13 C values during the Middle Miocene Climatic Optimum ca. 15 Ma (Tipple and Pagani, 2010). Soil carbonates from the Great Plains also document a corresponding increase in 13 C values (Fox and Koch, 2003). Evidence of an early C4 expansion is not regionally uniform, and tooth enamel records from southwestern North America do not capture a significant increase in C 4 plants at this time (Bowman et al., 2015; Feranec and Pagnac, 2013). Beginning in the late Miocene through to the Pliocene, the replacement of C 3 grasslands and dominance of C 4 grasslands occurs in several regions. By 6 Ma, herbivores readily adapt to C 4 resources in the Great Plains (Passey et al., 2002). In the Pliocene, carbon isotopes in soil carbonates records document the widespread abundance of C 4 vegetation in the Great Plains, Arizona and New Mexico (Fox and Koch, 2003; Mack et al., 1994; Wang et al., 1993). 104 The grass fossil record accompanying this history of grasslands remains sparse. Here we report C 3 grass fossils recovered from one of the key C 4 grass fossil locali- ties, and the reevaluation of the C 4 evidence suggests that the grass may have used the C 3 photosynthetic pathway. To date, C 4 grass fossils have only been recovered fromthelateMiocenedespiteearlieroriginsestimatedin theEarlyOligocene(~30- 32 Myr). Identifying grass fossils with sufficient anatomical and organic carbon preservation for C 3 or C 4 identification is a high priority to secure our reconstruc- tions of grass evolution, C 4 origins and paleoecological reconstructions. In light of the findings presented here, we question previous reports of the oldest C 4 grass fossils with anatomical and isotopic evidence from this region. Carbon isotopic analysis included methods that may not have removed carbonate from the sam- ple, and carbonate contamination would erroneously bias the carbon isotopic value towards C 4 -like. Unfortunately, the material no longer exists to reanalyze the sam- ples with appropriate high-temperature decarbonation methods. The location of the specimen is also vague and may have been collected from strata up to Pleis- tocene in age. We revisited the Dove Spring Formation to collect additional fossils, and for all the samples we collected, the results definitively indicate the use of the C 3 photosynthetic pathway. 4.7 Conclusion In this study, we describe grass fossils collected from the upper Miocene-age Dove Spring Formation in southern California dating from ca. 12 Ma. Anatomical descriptions based on microscopy of grass stems in cross sectionand organic carbon isotopic compositions after high-temperature carbonate removal provides conclu- sive evidence that these fossils are C 3 grasses. We revisited previously published 105 grass fossil images from this site and found that the C 4 designation is equivocal. This interpretation reduces the C 4 grass macrofossil record to only one established example. Our detailed anatomical and geochemical information present an impor- tant addition, and revision, to an exceedingly sparse grass fossil record. 4.8 Acknowledgments These samples were collected under the collection permit of Xiaoming Wang, with assistance from Dave Whistler, Gary Takeuchi, and Mark Faull. We thank Thure Cerling for drawing our attention to this interesting research question and locality. We thank Josh West and Will Berelson for access to facilities for high tempera- ture decarbonation and carbon isotopic analyses. We thank Nick Rollins, Yadira Ibarra, Dylan Wilmeth and Joyce Yager for laboratory assistance. This work was supported by funding from the University of Southern California to SF and HL. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. 106 Chapter 5 Conclusions The evolution of grasslands using the C 4 photosynthetic pathway and their sudden expansion in the late Miocene has been an active area of research since of 13 C- enriched soil carbonate material was first discovered in the rock record. While this dramatic biome shift implicitly suggests a global trigger, its heterogeneous expansion demands other explanations such as local shifts in hydrology, herbivory, fire regime, or other forms of habitat disturbance. For these reasons, the history of C 3 /C 4 is often tied to changes in regional climate. Plant wax carbon and hydrogen isotopes from the same molecules provide a unique record of not only vegetation changes but also hydrological change, reflected by the δD of plant wax. In this dissertation, I reconstructed late Miocene-Pleistocene environments in the regions surrounding the Arabian Sea. These environments are strongly affected by changes in the strength of the Indian Summer Monsoon, which is inherently tied into the interpretation of shifting C 3 /C 4 conditions. While our understanding of the evolutionary history of C 4 grasses has largely been built on indirect methods suchascarbonisotopereconstructionsfromleafwax,soilcarbonates,andherbivore tooth enamel, the direct record of grass fossils is extremely limited, amounting to only one study with both anatomical and isotopic evidence. Fossil grasses collected by the Natural History Museum of Los Angeles County from the Dove Spring Formation in southern California were sampled for additional information and insight into the grass fossil record. Taken together, these studies offer important insights into late Miocene C 4 expansion and the evolution of the Indian Monsoon. 107 The Pliocene-Pleistocene regional expansion of C 4 biomass in northeast Africa has direct and indirect consequences on resource availability for faunal and hominin evolution. Terrestrial records document local changes; however marine sediments from the Gulf of Aden capture a regionally integrated signal that reflects the dom- inant biomass signal from the landscape. In Chapter 2, we found that widespread C 4 grasslands werepresent in northeastAfrican lowlands during theearly Pliocene, before drying led to an expansion of arid-adapted C 3 shrubland, highlighting that a C 4 signal is not simply diagnostic of aridity. This shift in aridity was supported by a positive shift in δD plant wax and increase in the arid-adapted shrub pollen, Amaranthaceae (Bonnefille, 2010). The shift to drier conditions on land occurred in conjunction with a subsurface cooling of 4.4 ◦ C in the Gulf of Aden reflected in the TEX 86 record from the same sediments. This oceanographic change is also observed in upwelling regions in the northern Arabian Sea (Herbert et al., 2010; Huang et al., 2007), implicating shifting oceanic gateways though the Maritime Continent and its effect on Indian Ocean temperatures and the acidification of surrounding terrestrial environments (Cane & Molnar, 2001). In the Siwalik Group, fluvial sediments found in the foreland basin of the Himalayas, document an abrupt positive shift in δ 13 C by 7 Ma accompanied by a long-term positive trend in δ 18 O, implying a change in the hydrological cycle (Quade et al., 1989). Additional questions persist regarding this strong ecosys- tem shift: the regional extent of the ecological transition and its relation to the Indian Monsoon. From April-May 2015, I sailed with IODP Expedition 355 to recover sediments from the Indus Fan. In Chapter 3, we present a Late Miocene record from these sediments. Plant leaf waxδ 13 C of plant leaf wax n-alkanoic acid and n-alkanes, δ 13 C OC and pollen from the fan sediment record two phases of C 4 expansion: an early mixed C 3 /C 4 wooded grassland from 9.8-8.2 Ma and a later 108 expansion of C 4 -dominant grasslands from 7-5.5 Ma. To make sense of the episodic sedimentationwithintheIndusFan, additionalterrestrialandmarineorganicprox- ies and detrital geochemical proxies provide complementary evidence for changing depositional histories. We find that intervals characterized by hemipelagic sedi- mentation, low BIT values, and larger proportions of marine microfossils are inter- vals associated with more wind-blown material likely originating from a larger source region. Conversely, turbidite deposition, high BIT, and larger proportions of deltaic and terrestrial pollen are intervals of efficient transport and burial of terrestrial sediments originating from the Indus Catchment. Guided by this inter- pretation, we were able to compare our Indus Fan record with other plant wax and terrestrial carbon records to provide a regionally consistent record of C 4 expansion within the monsoon-sensitive regions throughout the Indo-Arabian region in the late Miocene. The evolution of C 4 photosynthesis is largely inferred by indirect means: molecular sequencing, isotopic records, and phytoliths. Molecular sequencing and phytolith assemblages suggest that C 4 photosynthesis first evolved in the Oligocene ~20-30 million years ago (Kellogg et al., 1999; Strömberg, 2005). Isotopic records indi- cate that C 4 grasses became a dominant feature of the tropics in the late Miocene (~8-7 Myr) and of the subtropics into the Pliocene (~3 Myr) (Cerling et al., 1997). However, direct fossil evidence of the distinctive C 4 Kranz anatomy and C 4 iso- topic signature is extremely scarce. One fossil recovered from Red Rock Canyon in California displays both the distinctive C 4 Kranz anatomy and the more enriched isotopic signature of a C 4 plant (Nambudiri et al., 1978). A second grass fossil was found from the Miocene Ogallala Formation in northwestern Kansas with dis- tinctive morphology indicative of Chloridoideae grasses (Thomasson et al., 1986), 109 however the isotopic composition of the fossil was not measured. The oldest tenta- tiveexampleofC 4 photosynthesisat~14MaisfromFortTernanofKenyabasedon cuticle morphology (Retallack, 1992). However, distinctive C 4 Kranz anatomy or pristine organic material was not preserved. In Chapter 5, we described grass fos- sils collected from the Dove Springs Formation at Red Rock Canyon and present a thorough taphonomic, microstructural, and isotopic description of a Miocene-age grass fossil. We find that all evidence unequivocally supports a C 3 grass fossil and review of the previously published C 4 grass fossil from this locality highlights potential issues in decarbonation methods and anatomical inferences that suggest that C 3 grasses have only ever been recovered from this location. Therefore, after this revision to the C 4 fossil record, unequivocal macrofossil evidence of C 4 grasses is reduced to one study and highlights an important gap for future research in C 4 grass evolution. 110 Reference List Antonov J, Sediov D, Boyer T, Locarnini R, Mishonov A, Garcia H (2010) World OceanAtlas2009, Volume2: Salinity InLevitusS,editor, NOAA Atlas NESDIA 69, pp. 1–184. U.S. Government Printing Office, Washington, D.C. Barros AP, Chiao S, Lang TJ, Burbank D, Putkonen J (2006) From weather to climate—Seasonal and interannual variability of storms and implications for ero- sion processes in the Himalaya In Willett SD, Hovius N, Brandon MT, Fisher DM, editors, Tectonics, Climate, and Landscape Evolution, pp. 17–38. Geologi- cal Society of America Special Paper 398. Bartoli G, Hönisch B, Zeebe RE (2011) Atmospheric CO2 decline during the Pliocene intensification of Northern Hemisphere glaciations. Paleoceanogra- phy 26:PA4213. Bolton CT, Stoll HM (2014) Late Miocene threshold response of marine algae to carbon dioxide limitation. Nature 500:558–562. Bonnefille R (2010) Cenozoic vegetation, climate changes and hominid evolution in tropical Africa. Global and Planetary Change 72:390–411. Bonnichsen B, Leeman WP, Honjo N, McIntosh WC, Godchaux MM (2008) Miocene silicic volcanism in southwestern Idaho: geochronology, geochem- istry, and evolution of the central Snake River Plain. Bulletin of Volcanol- ogy 70:315–342. Boos WR, Kuang Z (2011) Dominant control of the South Asian monsoon by orographic insulation versus plateau heating. Nature 463:218–222. Bordoni S, Schneider T (2008) Monsoons as eddy-mediated regime transitions of the tropical overturning circulation. Nature Geoscience . Bowen GJ, Wilkinson B (2002) Spatial distribution of δ18O in meteoric precipi- tation. Geology . 111 Brierley CM, Fedorov AV (2010) Relative importance of meridional and zonal sea surface temperature gradients for the onset of the ice ages and Pliocene- Pleistocene climate evolution. Paleoceanography 25:PA2214. Brodie CR, Leng MJ, Casford JSL, Kendrick CP, Lloyd JM, Yongqiang Z, Bird MI (2011) Evidence for bias in C and N concentrations and δ13C composition of terrestrial and aquatic organic materials due to pre-analysis acid preparation methods. Chemical Geology 282:67–83. Burbank DW, Whistler DP (1987) Temporally constrained tectonic rotations derived from magnetostratigiraphic data: Implications for the initiation of the Garlock fault, California. Geology 15:1172–1175. Bush MB (1995) Neotropical plant reproductive strategies and fossil pollen repre- sentation. The American Naturalist 145:594–609. Bush RT, McInerney FA (2013) Leaf wax n-alkane distributions in and across modern plants: Implications for paleoecology and chemotaxonomy. Geochimica et Cosmochimica Acta 117:161–179. Calvès G, Huuse M, Clift PD, Brusset S (2015) Giant fossil mass wasting off the coast of West India: The Nataraja submarine slide. Earth and Planetary Science Letters 432:265–272. Cane MA, Molnar P (2001) Closing of the Indonesian seaway as a precursor to east African aridification around 3–4 million years ago. Nature 411:157–162. Cerling TE, Harris JM (1999) Carbon isotope fractionation between diet and bioa- patite in ungulate mammals and implications for ecological and paleoecological studies. Oecologia 120:347–363. Cerling TE, Manthi FK, Mbua EN, Leakey LN, Leakey MG, Leakey RE, Brown FH, Grine FE, Hart JA, Kaleme P, Roche H, Uno KT, Wood B (2013) Stable isotope-based diet reconstructions of Turkana Basin hominins. Proceedings of the National Academy of Sciences 110:10501–10506. Cerling TE, Quade J, Ambrose SH, Sikes NE (1991) Fossil soils, grasses, and carbon isotopes from Fort Ternan, Kenya: grassland or woodland? Journal of Human Evolution 21:295–306. Cerling TE, Harris JM, MacFadden BJ, Leakey MG, Quade J, Eisenmann V, Ehleringer JR (1997) Global vegetation change through the Miocene/Pliocene boundary. Nature 389:153–158. 112 Cerling TE, Wynn JG, Andanje SA, Bird MI, Korir DK, Levin NE, Mace W, Macharia AN, Quade J, Remien CH (2012) Woody cover and hominin environ- ments in the past 6 million years. Nature 476:51–56. Chikaraishi Y, Naraoka H (2007) δ 13 C and δ D relationships among three n- alkyl compound classes ( n-alkanoic acid, n-alkane and n-alkanol) of terrestrial higher plants. Organic Geochemistry 38:198–215. Chikaraishi Y, Naraoka H, Poulson SR (2004) Hydrogen and carbon isotopic fractionations of lipid biosynthesis among terrestrial (C3, C4 and CAM) and aquatic plants. Phytochemistry 65:1369–1381. Christin PA, Besnard G, Samaritani E, Duvall MR, Hodkinson TR, Savolainen V, Salamin N (2008) Oligocene CO2 Decline Promoted C4 Photosynthesis in Grasses. Current Biology 18:37–43. Clemens SC, Murray DW, Prell WL (1996) Nonstationary phase of the Plio- Pleistocene Asian monsoon. Science . CLift PD, Shimizu N, Layne GD, Blusztajn J (2001) Tracing patterns of erosion and drainage in the Paleogene Himalaya through ion probe Pb isotope analysis of detrital K-feldspars in the Indus Molasse, India. Earth and Planetary Science ... . Clift PD (2010) Enhanced global continental erosion and exhumation driven by Oligo-Miocene climate change. Geophysical Research Letters 37:1–6. Clift PD, Hodges KV, Heslop D, Hannigan R, Van Long H, Calvès G (2008) CorrelationofHimalayanexhumationratesandAsianmonsoonintensity. Nature Geoscience 1:875–880. Conte MH, Weber JC (2002) Long-range atmospheric transport of terres- trial biomarkers to the western North Atlantic. Global Biogeochemical Cycles 16:89–1–89–17. Craig H (1961) Isotopic variations in meteoric waters. Science 133:1702–1703. Dahl KA, Oppo DW, Eglinton TI, Hughen KA, Curry WB, Sirocko F (2005) Ter- rigenous plant wax inputs to the Arabian Sea: Implications for the reconstruc- tion of winds associated with the Indian Monsoon. Geochimica et Cosmochimica Acta 69:2547–2558. Daly MC, Cooper MA, Wilson I, Smith DG (1991) Cenozoic plate tectonics and basin evolution in Indonesia. Marine and Petroleum Geology 8:1–21. Dansgaard W (2016) Stable isotopes in precipitation. Tellus 16:436–468. 113 deMenocal PB (1995) Plio-Pleistocene African climate. Science 270:53–59. Deplazes G, Luckge A, Stuut J, Pätzold J (2014) Weakening and strengthen- ing of the Indian monsoon during Heinrich events and DansgaardâĂŘOeschger oscillations. ... . Dettman DL, Kohn MJ, Quade J, Ryerson FJ, Ojha TP, Hamidullah S (2001) Seasonal stable isotope evidence for a strong Asian monsoon throughout the past 10.7 my. Geology 29:31–34. Dibblee J TW (1952) Geology of the Saltdale quadrangle, California. California Division of Mines Bulletin 160:5–43. Diefendorf AF, Mueller KE, Wing SL, Koch PL, Freeman KH (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proceedings of the National Academy of Sciences 107:5738–5743. Diefendorf AF, Freeman KH, Wing SL, Graham HV (2011) Production of n-alkyl lipids in living plants and implications for the geologic past. Geochimica et Cosmochimica Acta 75:7472–7485. Diefendorf AF, Freimuth EJ (2017) Organic Geochemistry. Organic Geochem- istry 103:1–21. Diefendorf AF, Leslie AB, Wing SL (2015) Leaf wax composition and car- bon isotopes vary among major conifer groups. Geochimica et Cosmochimica Acta 170:145–156. Droz L, Bellaiche G (1991) Seismic facies and geologic evolution of the central por- tion of the Indus Fan In Seismic Facies and Sedimentary Processes of Submarine Fans and Turbidite Systems, pp. 383–402. Springer-Verlag, New York. Ducassou E, MIGEON SB, Mulder T, Murat A, Capotondi L, Bernasconi SM, Mascle J (2009) Evolution of the Nile deep-sea turbidite system during the Late Quaternary: influence of climate change on fan sedimentation. Sedimen- tology 56:2061–2090. Dupont LM, Wyputta U (2003) Reconstructing pathways of aeolian pollen trans- port to the marine sediments along the coastline of SW Africa. Quaternary Science Reviews 22:157–174. Dupont L (2011) Orbital scale vegetation change in Africa. Quaternary Science Reviews 30:3589–3602. 114 Dupont-Nivet G, Krijgsman W, Langereis CG, Abels HA, Dai S, Fang X (2007) Tibetan plateau aridification linked to global cooling at the Eocene–Oligocene transition. Nature 445:635–638. Edwards EJ, Osborne CP, Stromberg CAE, Smith SA, C4 Grasses Consortium, Bond WJ, Christin PA, Cousins AB, Duvall MR, Fox DL, Freckleton RP, Ghan- noum O, Hartwell J, Huang Y, Janis CM, Keeley JE, Kellogg EA, Knapp AK, Leakey ADB, Nelson DM, Saarela JM, Sage RF, Sala OE, Salamin N, Still CJ, Tipple B (2010) The Origins of C4 Grasslands: Integrating Evolutionary and Ecosystem Science. Science 328:587–591. Eglinton G, Hamilton RJ (1967) Leaf epicuticular waxes. Science 156:1322–1355. Ehleringer JR, Cerling TE, Helliker BR (1997) C4 photosynthesis, atmospheric CO2, and climate. Oecologia 112:285–299. Feakins SJ, Levin NE, Liddy HM, Sieracki A, Eglinton TI, Bonnefille R (2013) Northeast African vegetation change over 12 m.y. Geology 41:295–298. Feakins SJ (2013) Pollen-Corrected leaf wax D/H reconstructions of northeast African hydrological changes during the late Miocene. Palaeogeography, Palaeo- climatology, Palaeoecology 374:62–71. Feakins SJ, Bentley LP, Salinas N, Shenkin A, Blonder B, Goldsmith GR, Ponton C, Arvin LJ, Wu MS, Peters T, West AJ, Martin RE, Enquist BJ, Asner GP, MalhiY(2016) Plantleafwaxbiomarkerscapturegradientsinhydrogenisotopes of precipitation from the Andes and Amazon. Geochimica et Cosmochimica Acta 182:155–172. Feakins SJ, deMenocal PB, Eglinton TI (2005) Biomarker records of late Neogene changes in northeast African vegetation. Geology 33:977. Feakins SJ, Eglinton TI, deMenocal PB (2007) A comparison of biomarker records of northeast African vegetation from lacustrine and marine sediments (ca. 3.40Ma). Organic Geochemistry 38:1607–1624. Feakins SJ, Peters T, Wu MS, Shenkin A, Salinas N, Girardin CAJ, Bentley LP, Blonder B, Enquist BJ, Martin RE, Asner GP, Malhi Y (2016) Organic Geo- chemistry. Organic Geochemistry pp. 1–12. Feakins SJ, Sessions AL (2010) Controls on the D/H ratios of plant leaf waxes in an arid ecosystem. Geochimica et Cosmochimica Acta 74:2128–2141. Fisher RL, Bunce ET, Cernock PJ, Clegg DC, Cronan DS, Damiani VV, Dmitriev LV, Kinsman DJ, Roth PH, Thiede J, Vincent E (1972) Initial Reports of the Deep Sea Drilling Project Technical report. 115 Fleitmann D (2003) Holocene Forcing of the Indian Monsoon Recorded in a Sta- lagmite from Southern Oman. Science 300:1737–1739. Fox DL, Koch PL (2003) Tertiary history of C4 biomass in the Great Plains, USA. Geology 31:809–812. France-Lanord C, Derry LA (1994) δ13C of organic carbon in the Bengal Fan: Source evolution and transport of C3 and C4 plant carbon to marine sediments. Geochimica et Cosmochimica Acta 58:4809–4814. Freeman KH, Colarusso LA (2001) Molecular and isotopic records of C4 grassland expansion in the late Miocene. Geochimica et Cosmochimica Acta 65:1439–1454. Galy V, Bouchez J, France Lanord C (2007) Determination of Total Organic Car- bon Content andδ13C in CarbonateâĂŘRich Detrital Sediments. Geostandards and Geoanalytical research 31:199–207. GalyV,EglintonT,FranceLanordC,SylvaS(2011) Theprovenanceofvegetation and environment signatures encoded in vascular plant biomarkers carried by the Ganges-Brahmaputra rivers. Earth and Planetary Science Letters 304:1–12. Gao L, Zheng M, Fraser M, Huang Y (2014) Comparable hydrogen isotopic frac- tionationofplantleafwaxn-alkanoicacidsinaridandhumidsubtropicalecosys- tems. Geochemistry, Geophysics, Geosystems 15:361–373. Garcin Y, Schefuß E, Schwab VF, Garreta V, Gleixner G, Vincens A, Todou G, Séné O, Onana JM, Achoundong G, Sachse D (2014) Reconstructing C3 and C4 vegetation cover using n-alkane carbon isotope ratios in recent lake sedi- ments from Cameroon, Western Central Africa. Geochimica et Cosmochimica Acta 142:482–500. Garzanti E, Baud A, Mascle G (1987) Sedimentary record of the northward flight of India and its collision with Eurasia (Ladakh Himalaya, India). Geodinamica Acta 4/5:297–312. Giosan L, Constantinescu S, Clift PD, Tabrez AR, Danish M, Inam A (2006) Recent morphodynamics of the Indus delta shore and shelf. Continental Shelf Research 26:1668–1684. Giri SJ, Diefendorf AF, Lowell TV (2015) Organic Geochemistry. Organic Geo- chemistry 82:22–32. Guan K, Wood EF, Caylor KK (2012) Multi-sensor derivation of regional vegeta- tion fractional cover in Africa. Remote Sensing of Environment 124:653–665. 116 Guo ZT, Ruddiman WF, Hao QZ, Wu HB, Qiao YS, Zhu RX, Peng SZ, Wei JJ, Yuan BY, Liu TS (2002) Onset of Asian desertification by 22 Myr ago inferred from loess deposits in China. Nature 416:159–163. Gupta AK, Yuvaraja A, Prakasam M, Clemens SC, Velu A (2015) Evolution of the South Asian monsoon wind system since the late Middle Miocene. Palaeo- geography, Palaeoclimatology, Palaeoecology 438:160–167. Hatch MD (1987) C4 photosynthesis: a unique blend of modified biochemistry, anatomy and ultrastructure. Biochimica et Biophysica Acta 895:81–106. Haug GH, Tiedemann R, Zahn R, Ravelo AC (2001) Role of Panama uplift on oceanic freshwater balance. Geology 29:207–210. Haug GH, Tiedemann R (1998) Effect of the formation of the Isthmus of Panama on Atlantic Ocean thermohaline circulation. Nature 393:673–676. Hemingway JD, Schefuß E, Dinga BJ, Pryer H, Galy VV (2016) Multiple plant- wax compounds record differential sources and ecosystem structure in large river catchments. Geochimica et Cosmochimica Acta 184:20–40. HerbertTD,PetersonLC,LawrenceKT,LiuZ(2010) Tropicaloceantemperatures over the past 3.5 million years. Science 328:1530–1534. Hoorn C, Ohja T, Quade J (2000) Palynological evidence for vegetation develop- ment and climatic change in the Sub-Himalayan Zone (Neogene, Central Nepal). Palaeogeography, Palaeoclimatology, Palaeoecology 163:133–161. Hopmans EC, Weijers JWH, Schefuß E, Herfort L, Sinninghe Damsté JS, Schouten S (2004) A novel proxy for terrestrial organic matter in sediments based on branched and isoprenoid tetraether lipids. Earth and Planetary Science Let- ters 224:107–116. Huang Y, Dupont L, Sarnthein M, Hayes JM, Eglinton G (2000) Mapping of C4 plant input from North West Africa into North East Atlantic sediments. Geochimica et Cosmochimica Acta 64:3505–3513. Huang Y, Clemens SC, Liu W, Wang Y, Prell WL (2007) Large-scale hydrological change drove the late Miocene C4 plant expansion in the Himalayan foreland and Arabian Peninsula. Geology 35:531. Ivory SJ, Lézine AM (2009) Climate and environmental change at the end of the Holocene Humid Period: A pollen record off Pakistan. Comptes Rendus Geoscience 341:760–769. 117 Karas C, Nürnberg D, Gupta AK, Tiedemann R, Mohan K, Bickert T (2009) Mid-Pliocene climate change amplified by a switch in Indonesian subsurface throughflow. Nature Geoscience 2:434–438. Karas C, Nürnberg D, Tiedemann R, Garbe-Schönberg D (2011) Pliocene Indone- sianThroughflowandLeeuwinCurrentdynamics: ImplicationsforIndianOcean polar heat flux. Paleoceanography 26:PA2217. Karim A (2002) Water balance of the Indus River Basin and moisture source in the Karakoram and western Himalayas: Implications from hydrogen and oxygen isotopes in river water. Journal of Geophysical Research 107:4362. Keeling CD, Piper SC, Bacastow RB, Wahlen M, Whorf TP, Martin H, Meijer HA (2001) Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978 to 2000. I. Global aspects Technical report 01-06. Kellogg EA, FARNSWORTH EJ, RUSSO ET (1999) Growth Responses of C4Grasses of Contrasting Origin to Elevated CO2. Annals of ... . Kolla V, Coumes F (1987) Morphology, internal structure, seismic stratigraphy, and sedimentation of Indus Fan. AAPG Bulletin 71:650–677. Kroon D, Steens T, Troelstra SR (1991) Onset of monsoonal related upwelling in the western Arabian Sea as revealed by planktonic foraminifers Technical report, College Station, Texas. Kutzbach JE, Guetter PJ, Ruddiman WF, Prell WL (1989) Sensitivity of climate to late Cenozoic uplift in Southern Asia and the American West: Numerical experiments. Journal of ... 94:18–393–18–407. Levin NE (2013) Compilation of East Africa Soil Carbonate Stable Isotope Data Integrated Earth Data Applications. Levin NE (2015) Environment and Climate of Early Human Evolution. Annual Review of Earth and Planetary Sciences 43:405–429. Levin NE, Haile-Selassie Y, Frost SR, Saylor BZ (2015) Dietary change among hominins and cercopithecids in Ethiopia during the early Pliocene. Proceedings of the National Academy of Sciences 112:12304–12309. Levin NE, Zipser EJ, Cerling TE (2009) Isotopic composition of waters from Ethiopia and Kenya: Insights into moisture sources for eastern Africa. Journal of Geophysical Research 114:D23306. Linder HP (2014) The evolution of African plant diversity. Frontiers in Ecology and Evolution 2:1–40. 118 Locarnini R, A M, J A, T B, H G, O B, M Z, DD J (2010) World Ocean Atlas 2009, Volume 1: temperature In Levitus S, editor, NOAA Atlas NESDIS, pp. 1–184. US Government Printing Office, Washington, DC. Loomis DP, Burbank DW (1988) The stratigraphic evolution of the El Paso basin, southern California: Implications for the Miocene development of the Garlock fault and uplift of the Sierra Nevada. Geological Society of America Bulletin 100:12–28. Marlow JR, Lange CB, Wefer G, Rosell-Melé A (2000) Upwelling intensification as part of the Pliocene-Pleistocene climate transition. Science 290:2288–2291. Maslin MA, Brierley CM, Milner AM, Shultz S, Trauth MH, Wilson KE (2014) East African climate pulses and early human evolution. Quaternary Science Reviews 101:1–17. McHargue TR, Webb JE (1986) Internal geometry, seismic facies, and petroleum potential of canyons and inner fan channels of the Indus submarine fan. AAPG Bulletin 2:161–180. McInerney FA, Strömberg C, White J (2011) The Neogene transition from C3 to C4 grasslands in North America: stable carbon isotope ratios of fossil phytoliths. Paleobiology 37:23–49. Merriam JC (1919) Tertiary mammalian faunas of the Mojave Desert University of California Publications, Department of Geology, Bulletin. Meyers PA, Ishiwatari R (1993) Lacustrine organic geochemistry—an overview of indicators of organic matter sources and diagenesis in lake sediments. Organic Geochemistry 20:867–900. Miles PR, Munschy M, Segoufin J (1998) Structure and early evolution of the Ara- bian Sea and East Somali Basin. Geophysical Journal International 134:876–888. Milliman JD, Quraishee GS, Beg MAA (1984) Sediment discharge from the Indus River to the ocean: past, present and future In Haq BU, Milliman JD, edi- tors, Marine Geology and Oceanography of Arabian Sea and Coastal Pakistan, pp. 65–70. Van Nostrand Reinhold. Molnar P, Cronin TW (2015) Growth of the Maritime Continent and its possible contribution to recurring Ice Ages. Paleoceanography . Molnar P, England P, Martinod J (1993) Mantle dynamics, uplift of the Tibetan Plateau, and the Indian monsoon. Reviews of Geophysics . 119 Molnar P, Boos WR, Battisti DS (2010) Orographic Controls on Climate and Paleoclimate of Asia: Thermal and Mechanical Roles for the Tibetan Plateau. Annual Review of Earth and Planetary Sciences 38:77–102. Monastero FC, Sabin AE, Walker JD (1997) Evidence for post–early Miocene initiation of movementon the Garlock fault from offset of the Cudahy Camp Formation, east-central California. Geology 25:247–250. Montes C, Cardona A, Jaramillo C, Pardo A, Silva JC (2015) Middle Miocene closure of the Central American Seaway. Science 348:226–229. Morgan ME, Kingston JD, Marino BD (1994) Carbon isotopic evidence for the emergence of C4 plants in the Neogene from Pakistan and Kenya. Nature 367:162–165. Najman Y, Appel E, Boudagher-Fadel M, Bown P, Carter A, Garzanti E, Godin L, Han J, Liebke U, Oliver G, Parrish R, Vezzoli G (2010) Timing of India-Asia collision: Geological, biostratigraphic, and palaeomagnetic constraints. Journal of Geophysical Research 115:B12416. Nambudiri E, Tidwell WD, Smith BN, Hebbert NP (1978) A C4 plant from the Pliocene. Nature 276:816–817. Nelson SV (2005) Paleoseasonality inferred from equid teeth and intra- tooth isotopic variability. Palaeogeography, Palaeoclimatology, Palaeoecol- ogy 222:122–144. Nelson SV (2007) Isotopic reconstructions of habitat change surrounding the extinction of Sivapithecus, a Miocene hominoid, in the Siwalik Group of Pak- istan. Palaeogeography, Palaeoclimatology, Palaeoecology 243:204–222. Nguyen N, Duffy B, Shulmeister J, Quigley M (2013) Rapid Pliocene uplift of Timor. Geology 41:179–182. Nicholson SE (2000) The nature of rainfall variability over Africa on time scales of decades to millenia. Global and Planetary Change 26:137–158. O’BrienCL,FosterGL,Martínez-BotíMA,AbellR,RaeJWB,PancostRD(2014) HighseasurfacetemperaturesintropicalwarmpoolsduringthePliocene. Nature Geoscience 7:606–611. Pagani M, Freeman KH, Arthur MA (1999) Late Miocene atmospheric CO2 con- centrations and the expansion of C4 grasses. Science 285:876–879. 120 Pandey DK, CLift PD, Kulhanek DK, Andò S, Bendle JAP, Bratenkov S, Griffith EM, Gurumurthy GP, Hahn A, Iwai M, Khim BK, Kumar A, Kumar AG, Liddy HM, Lu H, Lyle MW, Mishra R, Radhakrishna T, Routledge CM, Saraswat R, Saxena R, Scardia G, Sharma GK, Singh AD, Steinke S, Suzuki K, Tauxe L, Tiwari M, Xu Z, Yu Z (2015) Expedition 355 Preliminary Report: Arabian Sea Monsoon Technical report. Passey BH, Cerling TE, Perkins ME, Voorhies MR, Harris JM, Tucker ST (2002) Environmental change in the Great Plains: An isotopic record from fossil horses. The Journal of Geology 110:123–140. Perkins ME, Brown FH, Nash WP, McIntosh WC, Williams SK (1998) Sequence, age, and source of silicic fallout tuffs in middle to late Miocene basins of the northern Basin and Range province. Geological Society of America Bul- letin 110:344–360. Perkins ME, Nash BP (2002) Explosive silicic volcanism of the Yellowstone hotspot: The ash fall tuff record. Geological Society of America Bul- letin 114:367–381. Philander SG, Fedorov AV (2003) Role of tropics in changing the response to Milankovich forcing some three million years ago. Paleoceanography 18:1045. PolissarPJ,FultonJM,JuniumCK,TurichCC,FreemanKH(2009) Measurement of 13C and 15N Isotopic Composition on Nanomolar Quantities of C and N. Analytical Chemistry 81:755–763. Ponton C, West AJ, Feakins SJ (2014) Leaf wax biomarkers in transit record river catchment composition. Geophysical research ... . Posamentier HW, Kolla V (2003) Seismic geomorphology and stratigraphy of depositional elements in deep-water settings. Journal of Sedimentary Research 73:367–388. Prasad V, Strömberg C, Alimohammadian H, Sahni A (2005) Dinosaur Coprolites and the Early Evolution of Grasses and Grazers. Science 310:1177–1180. Prell WL, Kutzbach JE (1992) Sensitivity of the Indian monsoon to forcing param- eters and implications for its evolution. Nature 360:647–652. Prins MA, Postma G (2000) Effects of climate, sea level, and tectonics unraveled for last deglaciation turbidite records of the Arabian Sea. Geology 28:375–378. Prospero JM (2002) Environmental characterization of global sources of atmo- spheric soil dust identified with the NIMBUS 7 Total Ozone Mapping Spec- trometer (TOMS) absorbing aerosol product. Reviews of Geophysics 40:1002. 121 Quade J, Cater J, Ojha TP, Adam J, Harrison TM (1995) Late Miocene environ- mental change in Nepal and the northern Indian subcontinent: stable isotopic evidence from paleosols. GSA Bulletin 107:1381–1397. Quade J, Cerling TE, Bowman JR (1989) Development of Asian monsoon revealed by marked ecological shift during the latest Miocene in northern Pakistan. Nature 342:163–166. Quade J, Cerling TE (1995) Expansion of C4 grasses in the late Miocene of northern Pakistan: evidence from stable isotopes in paleosols. Palaeogeography, Palaeoclimatology, Palaeoecology 115:91–116. Ramage CS, Miller FR, Jeffries C (1972) Meteorological Atlas of the International Indian Ocean Expedition: The Surface Climate of 1963, 1964 Technical report. Ramstein G, Fluteau F, Besse J, Joussaume S (1997) Effect of orogeny, plate motion and land-sea distribution on Eurasian climate change over the past 30 million years. Nature 386:788–795. Retallack GJ (1992) Middle Miocene fossil plants from Fort Ternan (Kenya) and evolution of African grasslands. Paleobiology 18:383–400. Retallack GJ, Dugas DP, Bestland EA (1990) Fossil soils and grasses of a Middle Miocene East African grassland. Science 247:1325–1328. Rodriguez M, Chamot-Rooke N, Huchon P, Fournier M, Delescluse M (2014) The OwenRidgeupliftintheArabianSea: Implicationsforthesedimentaryrecordof Indian monsoon in Late Miocene. Earth and Planetary Science Letters 394:1–12. Rommerskirchen F, Eglinton G, Dupont L, Güntner U, Wenzel C, Rullkötter J (2003) AnorthtosouthtransectofHolocenesoutheastAtlanticcontinentalmar- gin sediments: Relationship between aerosol transport and compound-specific δ 13Clandplantbiomarkerandpollenrecords. Geochemistry, Geophysics, Geosys- tems 4:n/a–n/a. Rommerskirchen F, Plader A, Eglinton G, Chikaraishi Y, Rullkötter J (2006) Chemotaxonomic significance of distribution and stable carbon isotopic com- position of long-chain alkanes and alkan-1-ols in C4 grass waxes. Organic Geo- chemistry 37:1303–1332. Rowley DB (1996) Age of initiation of collision between India and Asia: A review of stratigraphic data. Earth and Planetary Science Letters 145:1–13. Royer JY, Chaubey AK, Dyment J, Bhattacharya GC, Srinivas K, Yatheesh V, Ramprasad T (2002) Paleogene plate tectonic evolution of the Arabian and 122 Eastern Somali basins In Clift PD, Kroon D, Gaedicke C, Craig J, editors, The Tectonic and Climatic Evolution of the Arabian Sea Region, pp. 7–23. Geological Society (London) Special Publications. Sachse D, Billault I, Bowen GJ, Chikaraishi Y, Dawson TE, Feakins SJ, Freeman KH, Magill CR, McInerney FA, van der Meer MTJ, Polissar P, Robins RJ, Sachs JP, Schmidt HL, Sessions AL, White JWC, West JB, Kahmen A (2012) Molecular Paleohydrology: Interpreting the Hydrogen-Isotopic Composition of Lipid Biomarkers from Photosynthesizing Organisms. Annual Review of Earth and Planetary Sciences 40:221–249. Sage RF (2001) Environmental and evolutionary preconditions for the origin and diversification of the C4 photosynthetic syndrome. Plant Biology 3:202–213. Sage RF, Sage TL, Kocacinar F (2012) Photorespiration and the Evolution of C4 Photosynthesis. Annual review of plant biology 63:19–47. Schefuß E, Ratmeyer V, Stuut JBW, Jansen JHF, Sinninghe Damsté JS (2003) Carbon isotope analyses of n-alkanes in dust from the lower atmosphere over the central eastern Atlantic. Geochimica et Cosmochimica Acta 67:1757–1767. Schouten S, Forster A, Panoto FE, Damsté JSS (2007a) Towards calibration of the TEX 86 palaeothermometer for tropical sea surface temperatures in ancient greenhouse worlds. Organic Geochemistry 38:1537–1546. Schouten S, Huguet C, Hopmans EC, Kienhuis MVM, Sinninghe Damsté JS (2007b) Analytical Methodology for TEX86 Paleothermometry by High-Performance Liquid Chromatography/Atmospheric Pressure Chemical Ionization-Mass Spectrometry. Analytical Chemistry 79:2940–2944. Schouten S, Pitcher A, Hopmans EC, Villanueva L, van Bleijswijk J, Damsté JSS (2012) Intactpolarandcoreglyceroldibiphytanylglyceroltetraetherlipidsinthe Arabian Sea oxygen minimum zone: I. Selective preservation and degradation in the water columnand consequences for the TEX. Geochimica et Cosmochimica Acta 98:228–243. Sirocko F, Lange H (1991) Clay-mineral accumulation rates in the Arabian Sea during the late Quaternary. Marine Geology 97:105–119. SmithEI,SánchezA,KeenanDL(2002) Stratigraphyandgeochemistryofvolcanic rocks in the Lava Mountains, California: Implications for the Miocene develop- ment of the Garlock fault In Glazner AF, Walker JD, Bartley JM, editors, Geologic Evolution of the Mojave Desert and Southwestern Basin and Range, pp. 151–160. Geological Society of America Memoir. 123 Song Q, Gordon AL, Visbeck M (2004) Spreading of the Indonesian Throughflow in the Indian Ocean*. Journal of Physical Oceanography 34:772–792. Strömberg C (2005) Decoupled taxonomic radiation and ecological expansion of open-habitat grasses in the Cenozoic of North America. Proceedings of the National Academy of Sciences 102:11980–11984. Strömberg C (2011) Evolution of Grasses and Grassland Ecosystems. Annual Review of Earth and Planetary Sciences 39:517–544. Tedford RH, Skinner MF, Fields RW, Rensberger JM, Galusha T, Taylor BE, Mac- Donald JR, Webb SD, Whistler DP (1987) Faunal succession and biochronology of the Arikareean through Hemphillian interval (late Oligocene through earliest Pliocene Epochs) in North America In Woodburne MO, editor, Cenozoic Mam- mals of North America, pp. 153–210. University of California Press, Berkeley, California. Thomasson JR, Nelson ME, Zakrzewski RJ (1986) A fossil grass (Gramineae: Chloridoideae) from the Miocene with Kranz anatomy. Science 233:876–878. Tidwell WD, Nambudiri E (1989) Tomlisonia thomassonii, gen. et sp. nov., a permineralized grass from the upper Miocene Ricardo Formation, California. Review of palaeobotany and palynology 60:165–177. Tierney JE, deMenocal PB (2013) Abrupt Shifts in Horn of Africa Hydroclimate Since the Last Glacial Maximum. Science 342:843–846. Tierney JE, Lewis SC, Cook BI, LeGrande AN, Schmidt GA (2011) Model, proxy, and isotopic perspectives on the East African Humid Period. Earth and Plane- tary Science Letters 307:103–112. Tierney JE, Pausata FSR, deMenocal P (2016) Deglacial Indian monsoon failure and North Atlantic stadials linked by Indian Ocean surface cooling. Nature Geoscience 9:46–50. Tierney JE, Russell JM, Damsté JSS, Huang Y, Verschuren D (2011) Quaternary Science Reviews. Quaternary Science Reviews 30:798–807. Tierney JE, Smerdon JE, Anchukaitis KJ, Seager R (2013) Multidecadal variability in East African hydroclimate controlled by the Indian Ocean. Nature 493:389–392. Tierney JE, Tingley MP (2014) A Bayesian, spatially-varying calibration model for the TEX. Geochimica et Cosmochimica Acta 127:83–106. 124 Tierney JE, Tingley MP (2015) A TEX86 surface sediment database and extended Bayesian calibration. Scientific Data 2:150029. Tipple BJ, Pagani M (2010) A 35 Myr North American leaf-wax compound- specific carbon and hydrogen isotope record: Implications for C4 grasslands and hydrologic cycle dynamics. Earth and Planetary Science Letters 299:250–262. Trommer G, Siccha M, van der Meer MTJ, Schouten S, Damsté JSS, Schulz H, Hemleben C, Kucera M (2009) Distribution of Crenarchaeota tetraether membrane lipids in surface sediments from the Red Sea. Organic Geochem- istry 40:724–731. Uno KT, Cerling TE, Harris JM, Kunimatsu Y, Leakey MG, Nakatsukasa M, Nakaya H (2011) Late Miocene to Pliocene carbon isotope record of differential dietchangeamongEastAfricanherbivores. Proceedings of the National Academy of Sciences 108:6509–6514. Uno KT, Polissar PJ, Kahle E, Feibel C, Harmand S, Roche H, deMenocal PB (2016) A Pleistocene palaeovegetation record from plant wax biomarkers from the Nachukui Formation, West Turkana, Kenya. Philosophical Transactions of the Royal Society B: Biological Sciences 371:20150235. Van Campo E (1991) 15. Pollen transport into Arabian Sediments Technical report. Vicentini A, Barber JC, Aliscioni SS, Giussani LM, Kellogg EA (2008) The age of the grasses and clusters of origins of C4 photosynthesis. Global Change Biol- ogy 14:1–15. Vogts A, Moossen H, Rommerskirchen F, Rullkötter J (2009) Distribution pat- terns and stable carbon isotopic composition of alkanes and alkan-1-ols from plant waxes of African rain forest and savanna C3 species. Organic Geochem- istry 40:1037–1054. Voznesenskaya EV, Franceschi VR, Kiirats O, Freitag H, Edwards GE (2001) Kranz anatomy is not essential for terrestrial C4 plant photosynthesis. Nature 414:543–546. Wakeham SG, Lee C, Hedges JI, Hernes PJ, Peterson ML (1997) Molecular indica- tors of diagenetic status in marine organic matter. Geochimica et Cosmochimica Acta 61:5363–5369. Ward PD, Garrison GH, Williford KH, Kring DA, Goodwin D, Beattie MJ, McRoberts CA (2007) The organic carbon isotopic and paleontological record 125 across the Triassic–Jurassic boundary at the candidate GSSP section at Fer- guson Hill, Muller Canyon, Nevada, USA. Palaeogeography, Palaeoclimatology, Palaeoecology 244:281–289. Whistler DP, Burbank DW (1992) Miocene biostratigraphy and biochronology of the Dove Spring Formation, Mojave Desert, California, and characterization of the Clarendonian mammal age (late Miocene) in California. Geological Society of America Bulletin 104:644–658. Whistler DP, Tedford RH, Takeuchi GT, Wang X, Tseng JZ, Perkins ME (2009) Revised Miocene biostratigraphy and biochronology of the Dove Spring Forma- tion, Mojave Desert, California, Vol. 65 Museum of Northern Arizona Bulletin. White TD, Asfaw B, Beyene Y, Haile-Selassie Y, Lovejoy CO, Suwa G, Wolde- Gabriel G (2009) Ardipithecus ramidus and the paleobiology of early hominids. Science 326:64–86. Whiteside JH, Olsen PE, Eglinton TI, Cornet B, McDonald NG, Huber P (2011) Pangean great lake paleoecology on the cusp of the end-Triassic extinction. Palaeogeography, Palaeoclimatology, Palaeoecology 301:1–17. Wichura H, Bousquet R, Oberhansli R, Strecker MR, Trauth MH (2010) Evidence for middle Miocene uplift of the East African Plateau. Geology 38:543–546. Wood AEI, Chaney RW, Clark JmS, Colbert ED, Jepsen GL, Reesdie JB, Stock C (1941) Nomenclature and correlation of the North American continental Ter- tiary. Geological Society of America Bulletin pp. 1–48. Wynn JG, Sponheimer M, Kimbel WH, Alemseged Z, Reed K, Bedaso ZK, Wilson JN(2013) DietofAustralopithecusafarensisfromthePlioceneHadarFormation, Ethiopia. Proceedings of the National Academy of Sciences 110:10495–10500. Yamamoto S, Kawamura K, Seki O, Kariya T, Lee M (2013) Influence of aerosol source regions and transport pathway on dD of terrestrial biomarkers in atmospheric aerosolsfrom the East China Sea. Geochimica et Cosmochimica Acta 106:164–176. Zachos J, Pagani M, Sloan L, Thomas E, Billups K (2001) Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292:686–693. Zhang Z, Ramstein G, Schuster M, Li C, Contoux C, Yan Q (2015) Aridification of the Sahara desert caused by Tethys Sea shrinkage during the Late Miocene. Nature 513:401–404. 126 Appendix A Appendix A.1 Carbon isotopic composition of long chain n-alkanoic acids In the main text, we report the carbon isotopic compositions of the most abun- dant long chain n-alkanoic acid, the C 28 n-alkanoic acid, to match the hydrogen isotopic analyses, which, due to limited material, were only conducted on the C 28 homolog. Here we additionally compare the δ 13 C values of the C 30 and C 32 homologs (Fig. S1). The overall temporalδ 13 C trend across homologs is consistent (Fig. 1). Theδ 13 C values of the different chain lengths are significantly correlated and robust to serial correlation, which were determined using a nonparametric method (Ebisuzaki, 1997): C 28 : C 30 r = 0.29, p < 0.05; C 28 : C 32 r = 0.55, p < 0.05; n-C 30 : C 32 r =0.70, p < 0.05). The correlation between chain lengths is consistent with the expectation that all chain lengths are dominantly derived from terrestrial plant sources. The δ 13 C values of C 28 are consistently more enriched relative to C 30 by 1.5%. This offset between chain lengths has been previously observed in samples from the same core where 13 C-enrichment followed in the order C 26 > C 28 > C 30 and this may be due to some contributions from aquatic sources to the shorter chain lengths (Feakins et al., 2005). Here we add the C 32 n-alkanoic aciddataandfindthattheδ 13 CofC 32 exhibitthemostenrichedvaluesofthechain lengths. Rather than an aquatic source, this may reflect the long chain production patterns of tropical C 4 grasses; the data are consistent with reports that modern 127 C 4 grasses from tropical Africa produce more C 31 and C 33 than n-C 27 and n-C 29 alkanes (Garcin et al., 2014) however, no n-alkanoic acid data were reported in African vegetation. Overall, the range of C 30 (5.6%) is smaller than the isotopic range of C 28 (6.2%) or C 32 (7.5%). The smallest difference between homologs is observed during the interval with the most 13 C-depleted values (ca. 4 Ma), while there is a larger spread when 13 C-enrichment occurs (Fig. S1). A.2 Ocean temperature calibrations A.2.1 Alkenone calibration To compare our Gulf of Aden temperature record to Indian Ocean temperatures, we used previously published Pliocene-Pleistocene U k 0 37 records from the northern Arabian Sea at ODP Site 722 (Huang et al. 2007; Herbert et al. 2010). An offset of 0.04 units is observed between the two U k 0 37 records. We adjusted the Huang et al. (2007) record accordingly (Fig. S2) and used this combined and adjusted recordforcomparisontootherIndianOceanseasurfacetemperature(SST)records in this paper. As a single globally applied calibration may not feasibly account for influencing factors on haptophyte production and the U k 0 37 signal, we applied four U k 0 37 calibrations to the alkenone reconstruction at ODP Site 722 including a global core top (MÃijller et al., 1998), culture (Prahl et al., 1988), surface sediment (Conte et al., 2006), and Indian Ocean warm temperature calibration (Sonzogni et al., 1997). This conservative approach accounts for a wider range of possible calibration uncertainty than an a priori selection of a single calibration. In the main text, these various SST calibrations are represented by a short dashed line (MÃijller et al., 1998), long dashed line (Prahl et al., 1988), solid line (Conte et al., 2006), dotted line (Sonzogni et al., 1997) (Fig. 3). 128 3.8 4 4.2 4.4 4.6 4.8 5 5.2 −30 −29 −28 −27 −26 −25 −24 −23 −22 −21 −20 Age (Ma) δ 13 C wax (‰) n-C 28 n-C 30 n-C 32 Figure A.1: Time series of the carbon isotopic composition of C 28 , C 30 , C 32 n- alkanoic acids from DSDP Site 231. A.2.2 Mg/Ca SST calibration Recently several papers suggested corrections to the Mg/Ca paleothermometry in the Pliocene due to changing Mg/Ca sw concentrations (OâĂŹBrien et al., 2014; 129 U37 k’ 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 Herbert et al. 2010 Huang et al. 2007 722 combined 2 3 4 5 22 24 26 28 30 32 SST (°C) Age (Ma) Figure A.2: Comparison plot of ODP 722 U k 0 37 records (Herbert et al., 2010; Huang et al., 2007). An offset of 0.04 is observed between the two U k 0 37 records (top panel). We adjusted the Huang et al. (2007) record by this offset and applied the Indian Ocean high temperature SST calibration (bottom panel, 1σ, shading) (Sonzogni et al., 1997). Zhang et al., 2014). If Pliocene Mg/Ca sw were lower than present this would lead to underestimated sea surface temperatures if not accounted for (Medina-Elizalde et al., 2008). Evidence that Pliocene Mg/Ca sw estimates were lower than modern comes from marine evaporites (3.6 and 4.05±1.75 mol mol −1 ) (Horita et al., 2002; Lowenstein et al., 2001) and a pore-fluid chemical profile model (2.8 - 3.5 mol mol −1 ) (Fantle and DePaolo, 2006). To provide a best estimate of the current 130 understanding of Mg/Ca sw evolution, we fit a second-order exponential function to Mg/Ca sw estimates spanning 180 Ma to present using Mg/Ca sw data based on calcium carbonate veins (Coggon et al., 2010), halite fluid trapped inclusions (Horita et al., 2002; Lowenstein et al., 2001), and echinoderm ossicles (Dickson, 2004; 2002) (Fig. S3). We excluded foram-based estimates of Mg/Ca sw to provide a foram-independent assessment of the Mg/Ca sw (Fig. S3) and avoid circularity. WeextrapolatedtheestimatedMg/Ca sw ofagiventimeperiodfromthiscurveand divided each value by the modern seawater value to generate a Mg/Ca seawater correction. Prior to converting the Pliocene-age Site 709C Mg/Ca values to SST, we accounted for possible calcite dissolution effects following Karas et al., (2011). Today, Site 709C sits above the lysocline, however the modern ΔCO 3 2− is 10μmol/kg (Karas etal., 2011), whichiswellbelowthedefinedthresholdvalueof20μmol/kgatwhich primary Mg 2+ loss can occur (Regenberg et al., 2006). Therefore, we applied the species-specific correction for Globigerinoides sacculifier (Regenberg et al., 2006). As nonreductive methods were used to clean the foraminiferal tests, we assumed no loss of Mg from cleaning protocols (Karas et al., 2011). Using the dissolution corrected values and taking into account the seawater correction described above, we applied the Anand et al. 2003 multispecies calibration using the modified equation: SST sw−corr = (ln( Mg Cam )− (ln(0.38) + ln( Mg/Caest Mg/Camg ))) 0.090 (A.1) where Mg/Ca m refers to the dissolution-corrected, measured Mg/Ca ratio, Mg/Ca est to the estimated Pliocene seawater Mg/Ca ratio, and Mg/Ca mv to the present seawater Mg/Ca ratio (5.18 mol mol −1 ) (Fig. S4). The SST sw−corr values are overall warmer than the originally published SST values by 0.8-1.8 ◦ C, and the 131 0 20 40 60 80 100 120 140 160 180 0 1 2 3 4 5 6 7 Age (Ma) Mg/Ca seawater mol/mol Figure A.3: Second order exponential fit through Mg/Ca seawater data including the modern value (orange diamond) and estimates based on halite trapped fluid inclusions [pink triangles (Lowenstein et al., 2001) and blue triangles (Horita et al., 2002)], cool site calcium carbonate veins (CCV) [light blue squares (Coggon et al., 2010)], warm site CCV [red squares [(Coggon et al., 2010)], and echinoderm ossicles [yellow diamonds (Dickson, 2004; 2002)]. The fit excludes foram-based estimates from this time period to provide a foram-independent estimate of the Mg/Ca of seawater. corrected values fall within error of the modern day mean annual SST at that loca- tion (28.6 ◦ C, 1σ = 1.1 ◦ C) (Locarnini et al., 2010) (Fig. S4). This method used to account for changing Mg/Ca concentrations of seawater is conservative compared to porefluid chemical profile model based estimates and TEX 86 - Mg/Ca back cal- culation methods, which imply average SST estimates of ca. 29-30 ◦ C (Fantle and DePaolo, 2006; O’Brien et al., 2014). 132 2 3 4 5 20 22 24 26 28 30 32 SST (°C) Age (Ma) Figure A.4: A comparison plot of the originally published Mg/Ca SST record from ODP Site 709C (purple line; 1σ, shading) (Karas et al., 2011) to the Mg/Ca SSTsw-corr estimate (black line; 1σ, shading). Modern SST estimate at Site 709C is 28.6 ± 1.1 ◦ C (yellow triangle) (Locarnini et al., 2009). A.2.3 Ocean temperature anomalies To calculate the ocean temperature anomalies relative to the early Pliocene, we subtracted the early Pliocene mean, defined as the mean temperature of the avail- able data between 5-4 Ma for each site. For the Mg/Ca SST record from ODP Site 709C, we used the SSTsw-corr values described in Section 2b and calculated the early Pliocene mean over the full extent of the early Pliocene record (4.87-4 Ma). For the alkenone temperature record from ODP Site 722, we used the com- bined and corrected alkenone records (Herbert et al., 2010; Huang et al., 2007) 133 described in Section 2a calibrated with the Indian Ocean warm temperature cali- bration (Sonzogni et al., 1997). 2 2.5 3 3.5 4 4.5 5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 BIT Index Age (Ma) 2 2.5 3 3.5 4 4.5 5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 BIT Index Age (Ma) Figure A.5: BIT Index data for DSDP Site 231. The BIT Index was calculated based on the ratio of branched glycerol dialkyl glycerol tetraethers (GDGTs) to theaquatic-sourcedisoprenoidalGDGTcrenarchaeol(Hopmansetal., (2004). Our marine sediment core BIT Index values are <0.1 units indicating that soil-derived branched GDGTs have a negligible contribution to the TEX 86 signal. Low BIT values also indicate that terrestrial runoff during the Pliocene was minimal, sup- porting our assumption that plant leaf wax transport was wind-blown. Table A.1: Carbon isotopic composition of C 28 , C 30 and C 32 n-alkanoic acids from DSDP Site 231. Table A.2: Hydrogen isotope composition of C 28 n-alkanoic acid from DSDP Site 231. Age estimates were determined using a tephrostratigraphic and nannofossil age model (Feakins et al., 2013) 134 Table A1: Carbon isotopic composition of C 28 , C 30 and C 32 n-alkanoic acids from DSDP Site 231. Depth (mbsf) Age (Ma) a ! 13 C C28 (‰) std. dev. (‰) ! 13 C C30 (‰) std. dev. (‰) ! 13 C C32 (‰) std. dev. (‰) Reference b 97.300 2.001 n.d. n.d. -25.9 0.2 n.d. n.d. * 97.450 2.004 n.d. n.d. -25.5 0.2 n.d. n.d. * 97.600 2.007 n.d. n.d. -26.7 0.4 n.d. n.d. * 97.750 2.010 n.d. n.d. -25.6 0.4 n.d. n.d. * 97.900 2.013 n.d. n.d. -24.9 0.8 n.d. n.d. * 98.050 2.016 n.d. n.d. -27.2 0.7 n.d. n.d. * 98.200 2.019 n.d. n.d. -26.5 0.3 n.d. n.d. * 98.350 2.022 n.d. n.d. -24.9 0.3 n.d. n.d. * 103.550 2.127 n.d. n.d. -25.1 0.2 n.d. n.d. * 103.650 2.129 n.d. n.d. -25.6 0.2 n.d. n.d. * 103.800 2.132 n.d. n.d. -26.2 0.6 n.d. n.d. * 105.150 2.159 -22.4 0.1 -24.9 n.d. n.d. n.d. 105.300 2.162 -23.3 n.d. n.d. n.d. n.d. n.d. 105.450 2.165 -22.9 n.d. n.d. n.d. n.d. n.d. 105.750 2.171 -23.2 n.d. n.d. n.d. n.d. n.d. 106.350 2.183 -25.1 n.d. -25.7 n.d. n.d. n.d. 106.650 2.189 n.d. n.d. -25.2 n.d. n.d. n.d. 106.800 2.192 n.d. n.d. -24.5 n.d. n.d. n.d. 107.400 2.204 -23.1 n.d. -25.0 n.d. n.d. n.d. 108.300 2.222 -23.9 n.d. n.d. n.d. n.d. n.d. 108.600 2.228 -22.1 n.d. n.d. n.d. n.d. n.d. 108.750 2.231 -23.7 n.d. n.d. n.d. n.d. n.d. 108.900 2.234 -22.9 n.d. n.d. n.d. n.d. n.d. 109.050 2.238 n.d. n.d. -26.6 n.d. n.d. n.d. 110.250 2.262 -25.8 n.d. -25.9 n.d. n.d. n.d. 110.400 2.265 -25.3 n.d. -25.7 n.d. -25.3 n.d. 110.550 2.268 -25.5 n.d. -27.0 n.d. n.d. n.d. 110.700 2.271 n.d. n.d. -25.9 n.d. n.d. n.d. 111.600 2.289 -25.6 n.d. -27.3 n.d. n.d. n.d. 111.750 2.292 n.d. n.d. n.d. n.d. n.d. n.d. 112.050 2.298 n.d. n.d. -26.5 n.d. -26.9 n.d. 112.350 2.304 n.d. n.d. -26.7 n.d. n.d. n.d. 113.150 2.320 -23.4 n.d. -26.6 n.d. n.d. n.d. 113.300 2.323 n.d. n.d. -26.2 n.d. n.d. n.d. 113.450 2.326 -25.2 n.d. -25.4 n.d. n.d. n.d. 113.600 2.329 -22.5 n.d. -26.3 n.d. n.d. n.d. 113.750 2.332 -25.7 n.d. -24.1 n.d. n.d. n.d. 113.900 2.335 -25.2 n.d. -26.9 n.d. n.d. n.d. 114.350 2.344 -25.4 n.d. -23.6 n.d. n.d. n.d. 114.500 2.347 -26.0 n.d. -24.1 n.d. n.d. n.d. 114.650 2.350 -25.1 n.d. -24.1 n.d. n.d. n.d. 114.950 2.356 -24.2 n.d. -25.2 n.d. n.d. n.d. 115.100 2.359 -23.8 n.d. -24.9 n.d. n.d. n.d. 115.250 2.362 -25.1 n.d. -24.1 n.d. n.d. n.d. 115.400 2.366 n.d. n.d. -25.2 n.d. n.d. n.d. 115.400 2.366 n.d. n.d. -25.5 n.d. n.d. n.d. 115.550 2.369 -22.9 n.d. -25.4 n.d. n.d. n.d. 115.700 2.372 n.d. n.d. -25.7 n.d. -24.6 n.d. 115.850 2.375 n.d. n.d. -25.6 n.d. -22.4 n.d. 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 Depth (m) Age (Ma) a D C28 (‰) stdev (‰) 105.15 2.159 -130.8 0.2 105.60 2.168 -136.1 5.8 105.75 2.171 -140.8 3.8 110.40 2.265 -131.3 0.4 112.35 2.304 -126.5 0.9 115.25 2.362 -133.3 0.2 122.50 2.520 -131.1 n.d. 124.30 2.561 -130.0 0.9 131.85 2.735 -129.6 n.d. 133.20 2.766 -125.2 2.6 134.40 2.793 -124.2 1.2 135.00 2.807 -128.7 0.3 135.60 2.821 -126.9 0.6 136.20 2.835 -132.9 4.7 136.80 2.848 -131.5 0.7 137.40 2.862 -130.3 2.5 138.00 2.876 -135.3 0.9 138.60 2.890 -135.1 2.7 139.20 2.904 -130.7 n.d. 141.05 2.946 -136.8 1.0 143.30 2.998 -125.3 5.5 144.80 3.032 -130.0 3.9 144.95 3.036 -137.8 n.d. 145.40 3.046 -141.9 1.9 146.60 3.074 -132.9 1.8 147.20 3.087 -136.0 n.d. 150.10 3.154 -129.1 n.d. 150.70 3.168 -135.9 0.1 151.30 3.182 -130.6 1.7 151.90 3.195 -131.4 0.0 152.35 3.206 -131.7 1.2 160.50 3.393 -144.0 4.8 160.95 3.403 -137.4 n.d. 162.02 3.428 -148.5 5.7 162.17 3.431 -129.0 n.d. 163.35 3.458 -133.2 0.9 169.85 3.591 -141.1 1.6 171.50 3.630 -139.1 1.3 171.80 3.643 -138.5 0.4 173.15 3.673 -134.1 0.1 173.90 3.687 -127.7 0.7 174.65 3.702 -137.4 1.4 176.45 3.737 -132.3 4.6 176.90 3.746 -127.8 0.8 177.35 3.754 -142.9 1.7 184.40 3.865 -138.8 6.4 185.60 3.882 -131.8 5.0 188.00 3.915 -142.1 0.1 190.30 3.960 -146.5 1.5 192.00 3.995 -134.2 0.7 151 192.50 4.005 -130.3 0.3 193.50 4.025 -128.7 2.7 193.90 4.033 -136.8 2.5 194.60 4.048 -129.7 9.5 198.30 4.123 -144.8 1.8 201.10 4.181 -145.3 0.0 201.90 4.197 -131.4 2.3 204.60 4.252 -152.3 0.7 205.30 4.266 -135.2 3.1 206.60 4.293 -135.9 1.4 206.80 4.297 -141.4 3.2 207.30 4.307 -133.2 2.8 208.00 4.322 -134.5 0.1 208.70 4.336 -136.1 n.d. 209.40 4.350 -135.4 1.1 210.10 4.364 -128.4 3.3 210.80 4.379 -131.7 n.d. 212.20 4.407 -147.6 0.7 212.90 4.422 -140.0 2.1 213.60 4.436 -131.5 1.7 216.50 4.495 -150.2 0.2 216.80 4.501 -133.4 0.6 217.90 4.524 -126.1 0.3 219.20 4.550 -126.9 1.7 219.90 4.565 -142.9 0.5 220.60 4.579 -134.7 n.d. 220.70 4.581 -121.9 1.6 222.70 4.622 -141.8 1.1 224.20 4.653 -138.7 0.5 224.90 4.667 -151.1 3.5 226.70 4.704 -128.5 4.2 227.40 4.718 -130.9 0.5 228.20 4.734 -134.0 2.5 228.90 4.749 -128.6 3.1 230.90 4.789 -142.0 3.2 231.60 4.804 -145.7 5.9 236.70 4.908 -154.9 1.7 236.90 4.912 -148.4 0.1 238.20 4.939 -152.5 0.3 239.60 4.967 -158.2 1.3 240.30 4.982 -147.5 0.8 241.70 5.010 -164.0 6.6 242.40 5.024 -141.7 1.4 243.10 5.039 -139.9 0.5 244.50 5.067 -131.8 0.1 245.20 5.082 -164.7 3.6 246.60 5.110 -161.3 n.d. 247.30 5.125 -144.6 2.2 248.00 5.139 -155.1 1.7 249.40 5.168 -140.0 1.6 250.10 5.182 -155.3 0.5 250.80 5.196 -131.9 0.4 251.20 5.204 -142.6 2.6 152 252.80 5.237 -139.3 n.d. 252.90 5.239 -155.7 1.1 153 Appendix B Appendix B.1 Chapter 3 plant wax data tables Data tables reported here correspond to the plant wax isotope data presented in Chapter 3, Late Miocene C4 Expansion in the Indus Catchment. Table B.1: Carbon isotopic composition of C 26 , C 28 , C 30 , C 32 , and C 34 n-alkanoic acids from IODP Site U1457. Table B.2: Carbon isotopic composition of C 25 , C 27 , C 29 , C 31 , C 33 , and C 35 n- alkanes from IODP Site U1457. Table B.3: Hydrogen isotopic composition of C 24 , C 26 , C 28 , C 30 , and weighted mean average of C 24 -C 30 n-alkanoic acid from IODP Site U1457. Table B.4: Hydrogen isotopic composition of C 25 , C 27 , C 29 , C 31 , C 33 , and weighted mean average of C 25 -C 33 n-alkane from IODP Site U1457. 154 Depth (mbsf) Age (Ma) 13 C C26 (‰) std. dev. (‰) 13 C C28 (‰) std. dev. (‰) 13 C C30 (‰) std. dev. (‰) 13 C C32 (‰) std. dev. (‰) 13 C C34 (‰) std. dev. (‰) 505.0 5.55 -22.2 0.1 -22.6 0.5 -23.9 0.2 -23.8 0.1 -22.1 0.1 532.6 5.67 -24.3 0.2 -24.2 0.2 -22.3 0.2 -23.3 0.4 -23.4 0.2 543.3 5.71 -22.1 0.1 -22.1 0.1 -21.4 0.0 -22.4 0.0 -22.8 0.2 553.5 5.76 -21.7 0.3 -21.9 0.5 -21.0 0.2 -22.3 0.0 -22.5 0.0 562.9 5.80 -22.6 0.3 -22.4 0.3 -21.5 0.4 -22.5 0.4 -23.4 0.2 572.3 5.84 -21.7 0.1 -21.9 0.1 -20.9 0.0 -22.2 0.0 -22.1 0.1 579.9 5.87 -22.5 0.2 -22.3 0.3 -21.5 0.2 -22.3 0.3 -22.6 0.3 591.7 5.92 -22.1 0.5 -22.3 0.3 -21.1 0.0 -22.1 0.1 -21.8 0.1 600.6 5.96 -22.1 0.1 -22.6 0.1 -21.7 0.1 -22.8 0.1 -22.6 0.4 604.5 5.98 -21.7 0.5 -22.3 0.0 -22.5 0.2 -21.3 0.3 -22.5 1.4 609.7 6.00 -22.9 0.2 -22.4 0.5 -23.5 0.8 -21.5 0.4 n.d. n.d. 615.2 6.02 -22.4 0.9 -22.8 0.3 -25.1 0.8 -23.1 0.3 n.d. n.d. 618.7 6.18 -22.6 0.2 -22.3 0.5 -23.2 0.5 -21.4 0.4 -22.4 0.6 619.3 6.22 -22.1 0.5 -22.5 0.1 -24.0 0.2 -27.0 4.9 n.d. n.d. 619.6 6.25 -21.9 0.3 -22.5 0.3 -23.9 0.6 -22.4 0.1 -22.1 0.5 624.5 6.31 -22.2 0.2 -21.9 0.1 -23.0 0.0 -22.1 0.0 n.d. n.d. 628.9 6.36 -22.4 0.1 -22.3 0.5 -22.7 0.3 -25.1 3.7 -24.6 n.d. 632.3 6.40 -21.7 0.1 -21.8 0.1 -23.6 0.8 -23.3 0.1 -24.4 0.7 635.3 6.44 -22.3 0.2 -22.7 0.0 -24.0 0.0 -23.3 0.2 n.d. n.d. 636.0 6.48 -22.5 0.0 -22.7 0.1 -24.5 0.2 -22.6 0.1 -21.6 0.3 636.4 6.52 -21.8 0.2 -22.0 0.3 -23.5 0.7 n.d. n.d. -23.9 n.d. 636.9 6.57 -21.7 0.1 -21.9 0.1 -22.8 0.2 -21.6 0.1 -21.2 0.4 637.3 6.61 -21.8 0.0 -22.3 0.4 -23.6 0.3 -22.7 n.d. -22.1 n.d. 637.8 6.66 -22.2 0.3 -22.4 0.3 -23.5 0.3 -22.6 0.1 -20.9 n.d. 638.2 6.70 -23.0 0.5 -23.3 0.5 -25.5 1.0 -24.6 n.d. -31.8 0.2 639.1 6.74 -21.7 0.1 -22.4 0.6 -24.2 0.6 -23.1 0.4 -23.5 n.d. 648.0 6.92 -23.1 0.0 -24.2 0.4 -26.5 0.5 -26.3 0.3 -27.2 n.d. 649.9 6.96 -21.5 0.7 -23.9 1.0 -23.8 1.4 -24.4 n.d. n.d. n.d. 651.5 6.99 -28.0 0.0 -29.1 0.3 -29.6 0.1 -30.1 0.3 -30.8 0.2 653.4 7.03 -28.5 0.4 -29.9 0.5 -30.0 0.3 -30.5 0.0 -29.8 0.1 658.1 7.13 -23.1 0.7 -25.8 0.6 -26.4 0.7 -26.6 0.5 -27.3 0.7 667.2 7.25 -22.4 0.6 -25.5 1.2 -25.2 0.9 -25.5 0.4 -26.0 0.6 672.8 7.31 -28.0 0.0 -29.9 0.2 -30.1 0.1 -30.9 0.0 -31.1 0.4 679.1 7.35 -28.4 0.4 -28.9 0.3 -29.5 0.4 -29.7 0.5 -30.2 0.3 697.0 7.46 -28.3 0.1 -29.4 0.2 -29.9 0.0 -30.6 0.1 -30.9 0.9 716.3 7.57 -28.0 0.2 -28.9 0.4 -29.2 0.2 -29.4 0.2 -30.4 0.6 738.4 7.71 -27.3 0.0 -29.2 0.1 -30.4 0.2 -29.9 0.1 -30.3 0.3 757.1 7.82 -29.0 0.4 -30.2 0.4 -30.7 0.4 -31.7 0.2 -32.3 0.3 774.1 7.92 -29.8 0.3 -30.5 0.6 -30.7 0.2 -31.5 0.2 -32.7 0.3 794.4 8.05 -28.4 0.1 -30.2 0.1 -30.5 0.1 -31.2 0.4 -32.0 0.4 155 813.2 8.16 -30.0 0.3 -31.0 0.5 -30.8 0.0 -31.4 0.1 -32.5 0.4 814.9 8.17 -28.8 0.4 -30.1 0.2 -30.4 0.4 -31.5 0.7 -32.3 0.0 822.6 8.22 -29.9 0.6 -30.6 0.9 -30.7 0.5 -31.5 0.3 -32.6 0.2 832.5 8.28 -21.8 0.1 -25.6 2.7 -24.7 0.3 -24.1 0.2 -24.6 0.1 838.8 8.83 -21.9 0.6 -23.9 0.4 -26.8 0.5 -27.0 0.3 n.d. n.d. 842.4 9.02 -23.6 0.3 -25.6 2.2 -26.5 0.3 -27.0 0.8 -27.4 0.4 848.0 9.31 -22.6 0.1 -24.2 0.0 -26.7 0.9 -25.4 0.6 n.d. n.d. 852.2 9.52 -23.6 0.5 -25.8 1.0 -27.1 0.3 -27.2 0.9 n.d. n.d. 857.1 9.73 -22.2 0.5 -25.1 0.4 -27.5 0.3 -28.4 0.5 -29.0 0.8 862.1 9.79 -24.6 0.3 -25.9 0.8 -27.3 0.2 -29.2 0.6 -30.6 0.4 868.6 9.80 -22.8 0.1 -25.6 0.9 -29.1 1.1 -29.1 1.0 -31.0 0.3 156 Depth (mbsf) Age (Ma) 13 C C25 (‰) std. dev. (‰) 13 C C27 (‰) std. dev. (‰) 13 C C29 (‰) std. dev. (‰) 13 C C31 (‰) std. dev. (‰) 13 C C33 (‰) std. dev. (‰) 13 C C35 (‰) std. dev. (‰) 505.0 5.55 -24.0 0.2 -23.8 0.1 -26.4 0.1 -26.5 0.1 -25.6 0.3 -23.2 0.4 520.5 5.61 -24.8 0.3 -23.0 0.2 -27.4 0.3 -26.5 0.4 -25.3 0.3 -22.4 0.1 532.6 5.67 -26.6 0.1 -26.9 0.1 -26.2 0.0 -27.5 0.1 -26.8 0.0 -23.7 0.1 543.3 5.71 -25.1 0.0 -24.1 0.0 -24.9 0.2 -26.2 0.2 -25.3 0.5 -22.5 0.5 553.5 5.76 -26.1 0.4 -24.5 0.1 -25.0 0.1 -26.4 0.3 -24.7 0.1 -22.6 0.2 562.9 5.80 n.d. n.d. n.d. n.d. -24.3 0.3 -25.1 0.3 -24.4 0.1 -21.5 0.2 572.3 5.84 -25.3 0.2 -24.6 0.1 -25.0 0.1 -26.5 0.2 -24.5 0.2 -22.1 0.5 579.9 5.87 -24.4 0.2 -23.4 0.2 -24.2 0.3 -25.5 0.2 -24.7 0.0 -22.7 0.8 591.7 5.92 -27.2 0.3 -25.5 0.3 -25.9 0.3 -26.6 0.4 -24.7 0.3 -21.8 0.3 600.6 5.96 -24.7 0.1 -23.7 0.2 -24.7 0.3 -25.8 0.2 -25.1 0.5 n.d. n.d. 604.5 5.98 -23.5 0.3 -23.7 0.0 -25.9 0.0 -26.2 0.1 -25.4 0.2 -23.1 0.9 609.7 6.00 -23.3 0.1 -23.4 0.1 -25.3 0.3 -25.4 0.5 -25.1 0.6 -23.9 1.0 615.2 6.02 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 618.7 6.18 -23.6 0.0 -24.2 0.1 -26.9 1.0 -25.9 0.6 -25.7 1.2 -25.4 4.0 619.3 6.22 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 619.6 6.25 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 622.4 6.28 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 624.5 6.31 -24.5 0.4 -24.4 0.3 -26.4 0.6 -27.1 0.6 -26.0 0.9 -25.3 2.6 628.9 6.36 -23.5 0.0 -23.1 0.1 -25.1 0.1 -25.5 0.1 -25.0 0.7 -24.2 2.4 632.3 6.40 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 635.3 6.44 -25.7 0.2 -24.5 0.1 -26.2 0.1 -27.2 0.2 -25.4 0.1 -23.6 0.5 636.0 6.48 -25.3 0.2 -25.1 0.5 -27.1 0.9 -28.0 1.2 -27.9 3.4 -30.7 10.5 636.4 6.52 -24.0 0.4 -24.2 0.6 -26.3 0.9 -26.8 0.1 -25.0 0.1 -24.5 2.6 636.9 6.57 -24.8 1.5 -24.2 1.0 -25.5 0.1 -26.3 0.0 -25.3 0.5 -25.5 3.8 637.3 6.61 -23.7 0.0 -23.4 0.1 -25.1 0.3 -26.2 0.2 -24.8 0.1 -22.8 0.6 637.8 6.66 -24.4 0.3 -24.1 0.2 -26.3 0.3 -27.3 0.2 -25.7 0.2 -23.4 0.3 638.2 6.70 -25.7 0.2 -25.7 0.0 -27.9 0.1 -30.5 0.7 -28.6 0.3 -27.7 0.3 639.1 6.74 -24.8 0.0 -24.6 0.0 -27.1 0.2 -28.2 0.0 -27.5 0.5 -28.7 1.5 648.0 6.92 n.d. n.d. n.d. n.d. -27.3 0.0 -28.4 0.4 -27.3 0.0 -24.4 0.1 649.9 6.96 -27.3 1.0 -26.9 0.2 -28.5 0.2 -29.5 1.0 -28.6 0.6 -26.8 1.6 651.5 6.99 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 653.4 7.03 -30.0 0.1 -30.5 0.3 -31.5 0.2 -32.7 0.3 -31.7 0.4 -29.6 0.1 658.1 7.13 n.d. n.d. n.d. n.d. -29.2 0.2 -29.8 0.1 -29.5 0.9 -27.0 0.6 667.2 7.25 -27.5 1.2 -26.8 1.0 -27.5 0.9 -28.1 0.3 -27.8 0.4 -28.3 2.9 672.8 7.31 -28.3 0.1 -29.5 1.2 -29.5 0.0 -30.3 0.1 -31.5 0.2 -31.4 0.1 679.1 7.35 -28.2 0.0 -29.1 0.1 -30.0 0.2 -31.1 0.2 -32.0 0.4 -31.8 1.4 697.0 7.46 -29.4 0.1 -29.9 0.2 -30.9 0.1 -32.0 0.1 -31.3 0.2 -28.5 0.5 716.3 7.57 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 738.4 7.71 -30.2 0.3 -30.6 0.3 -31.5 0.3 -32.3 0.1 -31.4 0.1 -29.5 0.6 157 757.1 7.82 -29.2 0.1 -30.0 0.1 -31.0 0.1 -32.3 0.2 -32.8 0.1 -31.7 0.3 774.1 7.92 -29.9 0.8 -30.3 0.8 -30.9 0.7 -32.2 0.8 -33.5 1.7 -34.7 4.1 794.4 8.05 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 813.2 8.16 -29.9 1.2 -30.7 1.5 -31.7 1.9 -33.1 2.2 -34.4 3.0 n.d. n.d. 814.9 8.17 -31.3 0.0 -31.2 0.2 -31.9 0.2 -33.6 0.4 -32.5 0.3 -30.7 1.5 822.6 8.22 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 832.5 8.28 -27.2 1.6 -28.1 1.7 -27.5 0.1 -28.8 0.4 -28.0 0.3 -26.9 1.4 838.8 8.83 -25.1 0.9 -25.6 0.8 -28.0 0.3 -28.8 1.3 -27.2 1.0 n.d. n.d. 842.4 9.02 -26.7 0.8 -27.5 0.3 -29.5 0.5 -30.1 1.0 -30.1 0.7 -27.8 1.4 848.0 9.30 -27.3 0.9 -28.6 0.6 -30.7 0.8 -31.1 0.7 -29.5 0.7 -26.1 0.6 852.2 9.52 -28.0 0.4 -29.9 1.2 -30.1 0.5 -30.1 0.3 -30.9 0.9 -32.5 4.3 857.1 9.68 -26.8 0.0 -28.2 0.1 -29.7 0.1 -30.3 0.1 -30.6 0.1 -28.6 0.5 862.1 9.79 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 868.6 9.80 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 158 Depth (mbsf) Age (Ma) D C24 (‰) std. dev. (‰) D C26 (‰) std. dev. (‰) D C28 (‰) std. dev. (‰) D C30 (‰) std. dev. (‰) D WMA (‰) std. dev. (‰) 505.0 5.55 -124 1 -136 2 -141 1 n.d. n.d. -133 2 532.6 5.67 -169 3 -179 0 -185 1 -191 1 -184 3 543.3 5.71 -182 4 -189 1 -195 1 -203 2 -193 5 553.5 5.76 -191 6 -196 4 -203 2 -201 5 -199 9 562.9 5.80 -186 2 -193 2 -197 2 -206 3 -196 4 572.3 5.84 -191 2 -196 0 -205 1 -204 2 -201 3 579.9 5.87 -191 1 -197 2 -202 4 -207 3 -200 5 591.7 5.92 -187 5 -197 1 -199 3 -199 1 -197 6 600.6 5.96 -186 5 -191 1 -197 2 -210 4 -198 7 604.5 5.98 -137 1 -129 1 -136 3 -154 2 -138 3 609.7 6.00 -138 2 -135 1 -139 3 -170 0 -143 4 615.2 6.02 -119 1 -135 3 -135 3 n.d. n.d. -130 4 618.7 6.18 -139 7 -131 2 -140 7 -157 2 -141 10 619.3 6.22 -127 1 -137 1 -139 1 -156 1 -137 2 619.6 6.25 -138 6 -139 2 -144 2 n.d. n.d. -141 7 624.5 6.31 -142 8 -129 2 -131 2 -149 3 -135 9 628.9 6.36 -125 2 -125 2 -131 4 -144 5 -128 7 632.3 6.40 -126 3 -136 1 -144 2 -149 1 -141 4 635.3 6.44 -140 0 -128 1 -132 3 n.d. n.d. -132 3 636.0 6.48 -143 2 -129 2 -138 2 -158 4 -141 5 636.4 6.52 -141 n.d. -125 n.d. -126 n.d. -162 n.d. -133 0 636.9 6.57 -146 6 -132 1 -143 3 -164 2 -146 7 637.3 6.61 -133 2 -135 0 -136 2 -157 1 -137 3 637.8 6.66 -140 2 -135 2 -145 1 -153 2 -143 4 638.2 6.70 -138 3 -131 1 -134 2 n.d. n.d. -133 4 639.1 6.74 -150 9 -132 2 -141 3 n.d. n.d. -139 9 648.0 6.92 -150 6 -139 2 -142 5 -153 1 -145 8 649.9 6.96 -149 1 -135 4 -137 1 n.d. n.d. -139 5 651.5 6.99 -199 4 -203 1 -201 2 n.d. n.d. -201 4 653.4 7.03 -197 4 -200 1 -196 3 -190 2 -196 5 658.1 7.13 -150 4 -141 2 -143 4 -201 3 -158 7 667.2 7.25 -150 4 -139 2 -146 3 -155 3 -146 6 672.8 7.31 -193 3 -200 1 -200 1 -196 3 -198 4 679.1 7.35 -200 4 -201 4 -201 4 -162 1 -190 7 697.0 7.46 -188 1 -196 1 -191 1 -174 2 -187 2 716.3 7.57 -181 0 -184 2 -186 0 n.d. n.d. -185 2 738.4 7.71 -179 1 -182 1 -187 1 -180 4 -182 4 757.1 7.82 -214 2 -216 1 -213 1 -205 2 -212 3 774.1 7.92 -205 3 -205 1 -205 1 -180 3 -197 4 794.4 8.05 -193 1 -194 1 -194 1 -193 2 -194 3 159 813.2 8.16 -202 3 -205 2 -200 2 -207 2 -204 4 814.9 8.17 -204 6 -206 6 -202 5 -199 3 -202 10 822.6 8.22 -204 2 -209 1 -204 1 -207 3 -206 4 832.5 8.28 -145 2 -132 1 -139 1 -211 4 -155 5 838.8 8.83 -140 4 -145 5 -146 4 -156 n.d. -146 8 842.4 9.02 -139 1 -142 1 -146 4 -154 2 -145 5 848.0 9.31 -147 3 -129 2 -134 6 -140 0 -136 7 852.2 9.52 -141 n.d. -136 n.d. -138 n.d. n.d. n.d. -138 0 857.1 9.73 -141 3 -145 1 -142 1 -152 2 -145 4 862.1 9.79 -143 1 -148 4 -150 1 n.d. n.d. -147 5 868.6 9.80 -146 1 -146 4 -148 0 -209 n.d. -155 4 160 Depth (mbsf) Age (Ma) D C25 (‰) std. dev. (‰) D C27 (‰) std. dev. (‰) D C29 (‰) std. dev. (‰) D C31 (‰) std. dev. (‰) D C33 (‰) std. dev. (‰) D WMA (‰) std. dev. (‰) 505.0 5.55 -142 1 -137 2 -137 2 -126 4 n.d. n.d. -134 5 520.5 5.61 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 532.6 5.67 -162 2 -163 2 -152 2 -148 n.d. n.d. n.d. -154 3 543.3 5.71 -208 2 -213 2 -211 0 -211 1 -206 3 -211 3 553.5 5.76 -173 0 -179 1 -172 1 -159 n.d. n.d. n.d. -169 1 562.9 5.80 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 572.3 5.84 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 579.9 5.87 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 591.7 5.92 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 600.6 5.96 -198 3 -206 3 -202 3 -198 1 n.d. n.d. -201 5 604.5 5.98 -158 0 -160 2 -163 2 -164 8 n.d. n.d. -162 8 609.7 6.00 n.d. n.d. -174 3 -169 0 -180 0 -177 1 -175 3 615.2 6.02 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 618.7 6.18 -182 1 -182 0 -181 1 -189 0 -182 2 n.d. n.d. 619.3 6.22 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 619.6 6.25 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 622.4 6.28 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 624.5 6.31 -157 0 -168 0 -163 0 -165 1 -162 0 -165 1 628.9 6.36 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 632.3 6.40 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 635.3 6.44 -149 2 -143 0 -147 1 n.d. n.d. n.d. n.d. -146 3 636.0 6.48 n.d. n.d. -161 1 -161 0 -163 1 -161 2 -162 2 636.4 6.52 -161 3 -175 0 -172 0 -173 0 -169 2 -172 4 636.9 6.57 n.d. n.d. -180 2 -175 2 -175 1 -168 2 -176 2 637.3 6.61 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 637.8 6.66 -181 0 -181 0 -177 1 -178 0 -177 2 -179 1 638.2 6.70 -155 7 -164 2 -158 0 -162 1 -153 0 -160 8 639.1 6.74 -166 1 -175 1 -170 1 -173 0 -170 2 -172 2 648.0 6.92 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 649.9 6.96 n.d. n.d. -167 1 -169 2 -170 0 -164 2 -169 3 651.5 6.99 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 653.4 7.03 n.d. n.d. -190 n.d. -189 n.d. n.d. n.d. n.d. n.d. -189 0 658.1 7.13 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 667.2 7.25 -182 2 -186 2 -191 1 -188 0 -188 0 -188 2 672.8 7.31 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 679.1 7.35 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 697.0 7.46 -184 1 -175 2 -175 2 -170 n.d. n.d. n.d. -174 3 716.3 7.57 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 738.4 7.71 -169 4 -184 2 -169 3 n.d. n.d. n.d. n.d. -173 5 757.1 7.82 -200 3 -200 3 -196 1 -189 n.d. n.d. n.d. -195 4 161 774.1 7.92 -218 4 -204 5 -210 2 -213 1 -211 2 -212 7 794.4 8.05 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 813.2 8.16 -217 0 -210 1 -214 3 -214 0 -208 0 -214 3 814.9 8.17 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 822.6 8.22 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 832.5 8.28 -159 3 -167 2 -176 1 -177 5 -173 5 -173 6 838.8 8.83 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 842.4 9.02 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 848.0 9.30 -146 2 -144 2 -152 1 -144 1 n.d. n.d. -147 3 852.2 9.52 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 857.1 9.68 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 862.1 9.79 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 868.6 9.80 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 162 Appendix C Appendix As part of a collaborative effort to characterize the long record of erosional and environmental changes in the Indus River Catchment, I extracted 18 samples and analyzed the hydrogen isotopic composition of plant wax n-alkanoic acids for a low resolution reconstruction from Site U1456, the first site drilled from the Indus Fan in the Laxmi Basin during International Ocean Discovery Program Expedition 355. C.1 Methods C.1.1 Lipid extraction Samples were freeze dried in a Virtis 2k unit and homogenized with a mortar and pestle. Dry, powdered sediment samples (17.1 - 31.8 gdw) were extracted with an Accelerated Solvent Extraction system (ASE 350 R , DIONEX) with 9:1 ratio of dichloromethane (DCM): methanol (MeOH) at 100 ◦ C and 1500 psi for two 15- minute cycles. Total lipid extracts were shipped to the University of Birmingham for separation. At the University of Birmingham, extracts were separated over a silica gel column where the total neutral fraction was eluted with 4ml of 1:1 DCM: isopropanol, and the total acid fractions were eluted with 4% acetic acid in ethyl- ether solution. The total neutral fraction was transferred onto a silica gel column and further separated with 4ml of n-hexane, 2ml of 2:1 n-hexane/DCM, 4ml DCM 163 and 5ml MeOH to obtain fractions of aliphatic hydrocarbons, aromatic hydrocar- bons, aldehydes and ketones, and alcohols, respectively. The total acid fraction was methylated with methanol of known isotopic composition with 95:5 MeOH: hydrochloric acid at 70 ◦ C for 12 hours. Methylated products were extracted using liquid-liquid extraction with 1 mL milli-Q water and hexane. The hexane extract was passed through an anhydrous sodium sulfate column. Samples were further purified over a silica gel column (5 cm x 40 mm Pasteur pipette, 5% water- deactivated silica gel, 100-200 mesh) eluted with hexane and DCM resulting in non-polar and fatty acid methyl ester fractions, respectively. C.1.2 Leaf wax quantification The fatty acid methyl ester fractions were identified and quantified using gas chromatography coupled with both a mass-selective detector and flame ioniza- tion detection (GC-MSD/FID Agilent) at the University of Southern California. 1/100 μL of the sample was analyzed by gas chromatography with injection via a split/splitless inlet in splitless mode, to a capillary column (Rxi R - 5ms 30m x 0.25mm, film thickness 0.25mm) with a constant He flow rate of 4 mL/min. Initial temperature of 50 ◦ C was held for 3.5 minutes followed by a temperature ramp of 20 ◦ C min −1 to 300 ◦ C held for an additional 10 minutes. Quantification was achieved using an in-house standard comprising a mixture of 4 n-alkanes and 3 n- alkanoic acids of varied and known concentration, with the calibrations determined separately for the two compound classes. The n-alkanoic acid concentrations are reported relative to the mass of dry sediment extracted (μg g-1). After quantifica- tion of the individual peak areas, we calculated the carbon preference index (CPI) 164 for C 22 -C 30 n-alkanoic acids to quantify the abundance of even over odd n-alkanoic acids: CPI = 1 2 + P [C 22−30even ] P [C 21−29odd ] + P [C 22−30odd ] P [C 23−31even ] (C.1) The C 31 n-alkanoic acid was assumed to be 0 when below the detection limit. To determine changes in the average chain length (ACL) of land plants, we calculated the ACL of the concentration-weighted abundances of long-chain homologues using the following equation for n-alkanoic acids: ACL = 24× [C 24 ] + 26× [C 26 ] + 28× [C 28 ] + 30× [C 30 ] P C 24−30 (C.2) C.1.3 Compound specific hydrogen isotopic analysis Thehydrogenisotopiccomposition(δD)ofn-alkanoicacidhomologswereanalyzed using a Thermo Scientific Trace GC equipped with a Rxi R -5 ms column (30 m ÃŮ 0.25 mm, film thickness 1Îijm) with a PTV injector operated in solvent-split mode, coupled to a Delta V Plus isotope ratio mass spectrometer (IRMS) via an Isolink pyrolysis furnace (1400 ◦ C). Isotopic linearity was monitored daily across a range of peak amplitude (1-8V) H 2 gas pulses. For hydrogen, the H 3 factor averaged 4.8 ppm mV −1 . Two out of five H 2 reference peaks were used for standardization of the isotopic analysis and the remaining peaks were used to assess precision. We used external standard runs containing a mixture of 15 n-alkanes (C 16 to C 30 ) with δD values ranging from -254.1 to -9.1%(A 3 mix standard supplied by A. Schimmelmann, Indiana University, USA) to normalize the data to the Vienna Standard Mean Ocean Water (VSMOW) - Standard Light Antarctic Precipitation (SLAP)hydrogenisotopicscale. TheRMSerroroftheexternalstandardreplicates throughout the period of analysis was 4.2%. Samples were run in triplicate, 165 and the average standard deviation was 1.8%. The isotopic composition of H added during methylation was determined by the esterification of phthalic acid for analysis by GC-IRMS (δD MeOH = -198 ± 3.9%, n=7; δ13C MeOH = -25.5 ± 0.37 , n=7). The addition of the methyl group was corrected for by mass balance. The results are reported using conventional delta notation (δD%) where: δ = h R sample −R std R std i × 1000 (C.3) C.2 Results C.2.1 Plant wax abundance and distribution n-alkanoic acid Total C 16 -C 34 n-alkanoic acid concentrations range between 412.1 and 1415.7 ng/gdw. High concentrations of n-alkanoic acid were found in clay-rich sediments, whereas lower concentrations were found in nannofossil ooze. The n-alkanoic acids have a bimodal distribution with peaks in short-chain n-alkanoic acid distribu- tions (<C 22 ) dominated by C 16 and C 18 , while mid to long-chain n-alkanoic acid distributions are generally dominated by C 24 and C 26 (Fig. 1). Chain lengths greater than C 30 were generally below detection limits. Distributions display the expected even-over-odd preference, and the carbon preference index (CPI 22-30) ranges between 3.3 and 6.6 indicative of a land plant source (Table 1). The ACL of C 24 -C 30 varies between 25.8 and 27.6. ACL values are low (ca. 26.4) between 10 and 8 Ma and increase to 27.4 at 7.79 Ma. ACL values remain high until 5.86 Ma then decrease to average values of 25.9 between 2.78 and 1 Ma (Table 1). 166 0 0.05 0.1 0.15 0.2 0.25 Relative abundance n−Alkanoic Acid C16 C18 C20 C22 C24 C26 C28 C30 C32 C34 (n = 11) Figure C.1: Average histogram distribution (relative abundance normalized to sum of 1) of n-alkanoic acids. Total number of quantified samples is reported in parentheses. Error bars represent standard deviation. C.2.2 n-Alkanoic acid hydrogen isotope ratios Of the 18 samples, only 11 samples gave viable quantities of n-alkanoic acids for isotopic measurement. The hydrogen isotopic composition of all mid to long chain length n-alkanoic acids (C 24 -C 30 ) displays a similar isotopic pattern, suggesting a common source of higher land plants. Therefore, we calculated the weighted averageδD (δD WM ) value of even numbered C 24 -C 30 using the relative abundance of each chain length. δD WM values vary between two extremes of -116.5 and -207.3%(Fig. 2). From 11-9 Ma, δDWM averages -141.9%. A negative excursion is observed in the latest Miocene at ca. 8 and 6 Ma, with values of ~-205.5%at 8 Ma and slightly more positive values of ~-195%at 6 Ma. Late Miocene trends agree with the δD n-acid record from Chapter 3 of this dissertation. In the Pliocene and 167 into the Pleistocene, δD WM shifts to more positive values (-122.3%) with the most positive values (-116.5%) occurring at 1.0 Ma. PleistoceneδD WM values are comparable to those of the late Miocene between 11-9 Ma, however the low sample resolution does not reflect the high climatic variability associated with the onset of glacial/interglacial cycles expected during this time. −220 −180 −140 −100 0 1 2 3 4 5 6 7 8 9 10 11 δD n−alkanoic acid (‰) Age (Ma) C24 C26 C28 C30 WM Miocene Pliocene Pleistocene Figure C.2: δD values of C 24 -C 30 n-alkanoic acids from IODP Site U1456 located in the Indus Fan. Error bars on the weighted mean average indicate the sum of squares error of the replicate analyses included in the calculation of the weighted mean average. Black bars represent generalized sediment recovery intervals while the white bars represent large hiatuses in sedimentation. 168
Abstract (if available)
Abstract
C₄ photosynthesis is a geologically recent adaptation to the ancestral C₃ photosynthetic pathway used to fix CO₂ in plants. The evolution of grasses using the C₄ pathway and their rise to ecological dominance in grasslands and savannas of the tropics and subtropics in the late Miocene is enigmatic, and its drivers remain an ongoing debate. In this dissertation, I present three studies focusing on the late Miocene to Pliocene expansion of C₄ photosynthetic plants. Using plant leaf wax carbon and hydrogen isotopes, it is possible to document not only when the isotopic transition to a C₄ dominant ecosystem occurred but also assess the role of hydrological change in driving the biome shift using the hydrogen isotope composition of the same molecules. Therefore, this proxy is applicable in monsoon sensitive regions to assess the interplay between vegetation and hydrology. In Chapter 2, we analyzed samples from a marine sediment core (DSDP Site 231) extracted from the Gulf of Aden to determine the Plio-Pleistocene expansion of C₄ biomass over the Horn of Africa and assess its relevance to early human evolution. We find that arid C₃ shrublands expanded as ocean temperatures progressively cooled and rainfall decreased. These changes occurred prior to the onset of Northern Hemisphere glaciation suggesting major changes ocean circulation and the strength of the Indian Monsoon. In Chapter 3, we determined the late Miocene expansion of C₄ grasslands in the Indus River floodplain using sediments collected from the Indus Fan (IODP Site U1457). Despite the complex depositional environment of the Indus Fan, we can infer that the late Miocene C₄ transition is widespread throughout the Indus Catchment and surrounding regions. In Chapter 4, we address the sparse grass fossil record by revisiting the site of one of the oldest known C₄ grass fossils of late Miocene-age. We find evidence via taphonomy, microstructure and isotopic composition of exclusively C₃ grass fossils from this location and present evidence to revise the previously published C₄ designation. Together these investigations inform the late Neogene record of C₄ photosynthesis in monsoon sensitive regions and offer an important addition, and revision, to the sparse grass fossil record.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Isotopic fractionations in plant biomarker molecules with application to paleoclimate
PDF
Antarctic climate variability from greenhouse to icehouse world
PDF
From deep lakes to deserts: Plio-Pleistocene paleoenvironmental transitions
PDF
Late Pleistocene changes in winter moisture source in the coastal southwest United States
PDF
Germanium and silicon isotope geochemistry in terrestrial and marine low-temperature environments
PDF
Recent variability in the hyrdological cycle of tropical Asia from oxygen isotopes of tree celulose
PDF
Tracking fluctuations in the eastern tropical north Pacific oxygen minimum zone: a high-resolution geochemical evaluation of laminated sediments along western North America
PDF
Tectonic and kinematic evolution of the Alborán domain during late orogenic extension
PDF
The geobiology of fluvial, lacustrine, and marginal marine carbonate microbialites (Pleistocene, Miocene, and Late Triassic) and their environmental significance
PDF
Concentration and size partitioning of trace metals in surface waters of the global ocean and storm runoff
Asset Metadata
Creator
Liddy, Hannah M.
(author)
Core Title
Evolution of the Indian Monsoon and rise of C₄ photosynthesis in the Miocene and Pliocene
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Geological Sciences
Publication Date
04/20/2017
Defense Date
03/09/2017
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Arabian Sea,biomarkers,C₃/C₄,carbon isotopes,grass fossils,Gulf of Aden,marine sediments,OAI-PMH Harvest,organic geochemistry,paleoclimate,plant wax
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Feakins, Sarah J. (
committee chair
), Corsetti, Frank A. (
committee member
), Ziebis, Wiebke (
committee member
)
Creator Email
hmliddy@gmail.com,liddy@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-363091
Unique identifier
UC11257972
Identifier
etd-LiddyHanna-5248.pdf (filename),usctheses-c40-363091 (legacy record id)
Legacy Identifier
etd-LiddyHanna-5248.pdf
Dmrecord
363091
Document Type
Dissertation
Rights
Liddy, Hannah M.
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
Arabian Sea
biomarkers
C₃/C₄
carbon isotopes
grass fossils
Gulf of Aden
marine sediments
organic geochemistry
paleoclimate
plant wax