Research Collection

Doctoral Thesis

Holocene Climate-and Anthropogenically-Driven Mobilization of Terrestrial Organic Matter

Author(s): Usman, Muhammed Ojoshogu

Publication Date: 2018

Permanent Link: https://doi.org/10.3929/ethz-b-000310419

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH NO. 25305

Holocene climate- and anthropogenically-driven mobilization of terrestrial organic matter

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZÜRICH (Dr. Sc. ETH Zürich)

Presented by

Muhammed Ojoshogu Usman Master of Science in Earth Sciences, ETH Zürich born on November 21, 1987 citizen of the Federal Republic of Nigeria

accepted on the recommendation of

Prof. Dr. Timothy I. Eglinton Dr. Maarten R. Lupker Dr. Francien Peterse Dr. Dirk Sachse

2018

Thesis abstract:

The Indian monsoon, a part of the Asian monsoon system and one of the most important components of the earth’s climate system, affects the livelihood of over a billion people. Thus, it is imperative to understand the forcing and response mechanisms associated with long-term monsoon variability. Owing to the integrative property of rivers, continental margin sediments deposited at the mouth of rivers often serve as excellent recorders of past environmental condition on the continent. This thesis provides high-resolution records of Indian monsoon variability during the Holocene, using three sediment cores collected from fluvial-dominated margins of the Bay of Bengal (off the Mahanadi, Godavari, and Krishna-Godavari Rivers). In addition, riverine sediments and soils were collected from the modern-day Godavari river basin to document present changes within the basin and reconcile them those recorded in offshore sedimentary record. The sedimentological parameters (such as mineralogy, grain-size distribution, mineral surface area), geochemical characteristics of bulk (total organic carbon contents, stable and radio-isotopes of carbon) and biomarkers (abundance and isotopic composition of fatty acids) were investigated to detect changes in terrestrial vegetation composition, sediment provenance, and nature of terrestrial organic matter exported to the adjacent margin. Carbon isotopic composition of terrestrial biomarkers allowed a source attribution for the margin sedimentary organic matter revealing a change in provenance from a lower basin source in the early Holocene to predominantly upper basin signatures in the later part of the Holocene. This is consistent with the expansion of aridity-adapted vegetation during mid to late Holocene. The comparisons with other regional records revealed that the terrestrial vegetation changes over millennial timescales were primarily driven by changes in Indian monsoon intensity. When superimposed on high-resolution radiocarbon measurements of terrestrial biomarkers, the results show an increasing apparent age of organic matter in the last ~4.7 ky, suggesting that increasing aridity has impacted the age structure of organic matter by inducing changes in terrestrial residence time and/or carbon mobilization dynamics. Exploring the results within archaeological context indicates that increasing aridity coincided with the expansion of sedentary agriculture in central and south India that arose from the demise of the Indus Valley civilization. Taken together, these results suggest that anthropogenic

i perturbation of the landscape arising from agricultural activities resulted in large-scale mobilization of pre-aged soil organic matter during the mid to late Holocene. Integrating marine sedimentary archives and continental paleo-environmental conditions, as this thesis has done, provides important insights into sedimentation processes and organic matter transport dynamics over millennial time scales.

ii

Zusammenfassung:

Der indische Monsun, ein Teil des asiatischen Monsunsystems und einer der wichtigsten Komponenten des Erdklimasystems, beeinflusst das Leben von über einer Milliarde Menschen. Es ist deshalb notwendig die Kräfte und Reaktionsmechanismen der assoziierten langzeit Variabilität des Monsuns zu verstehen. Aufgrund der integrierenden Eigenschaft von Flüssen, abgelagerte Sedimente auf dem Kontinentalrand dienen häufig als hervorragende Aufzeichner von kontinentalen Umweltbedingungen der Vergangenheit. Diese Dissertation beinhaltet hochauflösende Aufzeichnungen der Variabilität des indischen Monsuns im Verlaufe des Holozäns, stützend auf drei Sedimentkernen beprobt von fluvial geprägten Kontinentalrändern des Golfs des Bengalen (abseits von den Mahanadi, Godavari und Krishna-Godavari Flüssen). Dazu wurden Flusssedimente und Böden vom rezenten Einzuggebiet des Godavaris beprobt, um deren geochemischen Signaturen und ihre Veränderungen mit denen aufgezeichnet in den Meeressedimenten zu vergleichen. Die sedimentologische Parameter (zum Beispiel Mineralogie, Korngrössenverteilung, Mineraloberfläche), geochemische Eigenschaften des Gesamtprobenmaterials (gesamter organischer Kohlenstoffgehalt, stabile und Radioisotope des Kohlenstoffs) und Biomarker (Häufigkeit und Isotopenzusammensetzung von Fettsäuren) wurden untersucht, um Veränderungen in der Zusammensetzung der terrestrischen Vegetation, Herkunft von Sediment und organischer Materie und die Eigenschaften des terrestrischen organischen Materials, welches auf dem angrenzenden Kontinentalrand exportiert wird, zu verfolgen. Kohlenstoffisotopenzusammensetzung von terrestrischen Biomarkern erlaubte eine Quellenzuordnung der organischen Materie der Kontinentalrandsedimenten und offenbarte eine Verlagerung der Quelle vom unteren Einzuggebietes im frühen Holozän zu überwiegenden Zufuhr aus dem oberen Einzugsgebiet im späteren Holozän. Dies stimmt mit der Ausdehnung von ariditätsangepasster Vegetation während des mittleren bis späten Holozäns über ein. Der Vergleich mit anderen regionalen Aufzeichnungen zeigt, dass die terrestrische Vegetation über Jahrtausende hauptsächlich durch Veränderungen der Intensität des indischen Monsuns getrieben wurden. Der Vergleich mit hochauflösende Radiokarbonmessungen von terrestrischen Biomarkern, zeigen die Resultate, dass ein zunehmendes scheinbares Alter der organischen Materie während den letzten ~4.7 ky stattgefunden hat, und suggeriert, dass die zunehmende Trockenheit die

iii

Altersstruktur der organischen Materie beeinflusst hat, indem die terrestrische Verweilzeit und oder die Kohlenstoffmobilisierungsdynamik verändert wurde. In einem archäologischen Kontext betrachtet deuten diese Resultate darauf hin, dass zunehmende Trockenheit mit der Ausdehnung von sesshaften Ackerbau in Zentral- und Südindien zeitlich übereinstimmt, welches entstanden ist nach dem Zerfall der Indus-Kultur. Zusammenfassend deuten diese Resultate daraufhin, dass die anthropogene Perturbation der Landschaft infolge von ackerbautreibende Tätigkeiten verursachte die grossräumige Umlagerung von gealterter organischen Bodenkohlenstoff während des mittleren bis späten Holozäns. Die Integration von marinen Sedimentarchiven und kontinentale paläoumwelt Bedingungen, so wie in dieser Dissertation vorgelegt, bietet wichtige Erkenntnisse über Sedimentationsprozesse und Transportdynamik von organischer Materie über Zeitskalen von Jahrtausenden.

iv

Acknowledgements:

I am forever grateful to my friends, colleagues, and mentors for the innumerable support I have received throughout my time at ETH. First and foremost, I would like to express my sincere gratitude to Tim Eglinton for being a wonderful advisor and mentor over the years. I have benefitted immensely from our discussions and meetings both on a professional and personal level. Thanks for always being supportive, without which I would not have been able to complete this dissertation. I am eternally thankful for giving me the opportunity. I would also like to acknowledge my “field” supervisors Maarten Lupker and Francien Peterse for your guidance and support. Thank you to Maarten and Francien for always answering my questions and for the incredible times in the Godavari. I will forever cherish those moments in India. Further thanks to Dirk Sachse for the valuable contribution and the unbiased perspectives you brought as an external examiner. Equally important is the tremendous support I have received from all the members of the Eglinton Biogeoscience lab. A big thank you to Daniel Montlucon, a friend and a mentor, without whom none of the laboratory work would have been possible. A special thank you also to Cameron McIntyre, Negar Haghipour, and the LIP folks up at Hönggerberg for their laboratory support. My gratitude to the awesome BGS crew such as Chantal, Tessa, Melissa, Blanca, Lena, Hannah, Julia, Gaby, Alysha, Uli who all made the lab such an exciting place to work. A big thank you to Kathrin who helped me navigate through the administrative nightmares over the years. Also worthy of mention is Michael Plötze and his team at the clay lab. Thanks for your instrumental support and time spent with me trying to figure out the clay mineral assemblages in my samples. Thank you to Thomas Blattmann for being an incredible friend and colleague all rolled into one. Your help over the years is simply too numerous to list here. Thanks for everything. I also thank my office mates and colleagues such as Benny, Rong, Alejo, Sascha, Erica, Katie, Franzi, Yuezhi, and Ricardo who made the office a delightful place to work. A special thank you to the folks across the Atlantic. I owe a debt of gratitude to Liviu Giosan and Camilo Ponton for the insightful discussions over the years. A special thank you to Jordon Hemingway for all your help whenever I’m in

v

Massachussetts. A great thank you to Brad Rosenheim, Ryan Venturelli and the folks at USF St. Petersburg for the great introduction to the world of Dirt Burner at Florida. My gratitude also goes to Dave Hollander who facilitated my participation on the C-IMAGE cruise in the Gulf of Mexico where I met incredible people like Ben, Patrick, Erin, Amy, Sherryl, Dan and all the crew and scientist on the Weatherbird. I have also been fortunate to live with truly wonderful people at the SeeWeeGee over the years. Yannick, Markus, Roman, Florian, Michelle, Nadine, Hannah, Larissa, Pascal, the two Joses, Felix, Alex, Simone, and of course Tamara, who has been my rock and special companion through the years. Thank you so much for everything. Last but not the least, a big thank you to my family for their immeasurable support.

vi

Table of Contents

Thesis abstract: ...... i Zusammenfassung: ...... iii Acknowledgements: ...... v Table of Contents ...... vii

Chapter 1: General Introduction ...... 1 1.1. Introduction ...... 2 1.2. The global carbon cycle ...... 3 1.3. Rivers - the “arteries” of the planet ...... 4 1.4. Motivation and objectives ...... 6 1.5. Thesis outline...... 7 References ...... 9

Chapter 2: Reconciling drainage and receiving basin signatures of the Godavari River system ...... 12 Abstract ...... 13 2.1. Introduction ...... 14 2.2. Materials and methods ...... 16 2.2.1. Study area ...... 16 2.2.2. Sampling...... 17 2.2.2.1. River Basin ...... 17 2.2.2.2. Offshore ...... 17 2.2.3. Sample treatment and measurements ...... 18 2.2.3.1. Mineral-specific surface area ...... 18 2.2.3.2. Grain size ...... 18 2.2.3.3. Sediment mineralogy ...... 18 2.2.3.4. Bulk elemental and isotopic analysis ...... 18 2.2.3.5. Compound-specific stable carbon isotopic analysis ...... 18 2.3. Results ...... 19 2.3.1. Surface and deep soils ...... 19 2.3.2. Riverbed and riverbank sediments ...... 19 2.3.3. River-proximal marine sediments ...... 22 2.4. Discussion ...... 22 2.4.1. Evolution of organic matter-mineral associations in the Godavari River basin .22 2.4.2. Linkages between Godavari drainage basin and marine sedimentary signals ..26 2.5. Conclusions ...... 27 Acknowledgements ...... 27 References ...... 28

vii

Chapter 3: Basin-wide climatic and anthropogenic driven mobilization of soil organic carbon to the Bay of Bengal...... 32 Abstract ...... 33 3.1. Introduction ...... 34 3.2. Materials and methods ...... 36 3.2.1. Study area ...... 36 3.2.2. Methods ...... 38 3.2.2.1. Age Model...... 39 3.2.2.2. Bulk elemental and isotopic analysis ...... 41 3.2.2.3. Compound-specific stable isotopic analysis ...... 41 3.2.2.4. Compound-specific 14C analysis ...... 41 3.3. Results ...... 43 3.4. Discussion ...... 47 3.4.1. OC and biomarker loadings and distribution ...... 47 3.4.2. Foraminifera-based chronology ...... 48 3.4.3. Radiocarbon contents of planktic foraminifera and organic fractions ...... 48 3.4.4. Contrasting bulk OC and LCFA age offsets – A mixing of different OC pools? .51 3.4.5. Terrestrial C3 versus C4-plants contributions to bulk OC...... 52 3.4.6. Quantification of different OC source and contributions using coupled isotopic mixing models ...... 55 3.4.7. Implications for OC transport and cycling ...... 58 3.5. Conclusions ...... 61 Acknowledgements ...... 61 References ...... 62 Supplementary Information ...... 70

Chapter 4: Evolution of the Indian Summer Monsoon and its impact on the origin and fate of terrestrial organic matter in the Bay of Bengal ...... 72 Abstract ...... 73 4.1. Introduction ...... 74 4.2. Materials and Methods ...... 78 4.2.1. Oceanographic Setting ...... 78 4.2.2. Mahanadi Basin ...... 79 4.2.3. Krishna-Godavari Basin...... 79 4.3. Methods ...... 80 4.3.1. Sampling...... 80 4.3.2. Age Model...... 80 4.3.3. Grain size and mineral surface area ...... 82 4.3.4. Bulk elemental and isotopic measurements ...... 82 4.3.5. Plant wax fatty acid extraction ...... 82 4.3.6. Compound-specific 13C and 14C measurements ...... 82 4.4. Results ...... 83 4.4.1. Linear sedimentation rates ...... 83 4.4.2. Grain size distribution and mineral-specific surface area...... 85 4.4.3. Bulk organic and biomarker characteristics ...... 85 4.4.4. Plant wax fatty acids ...... 88 4.5. Discussion ...... 89 4.5.1. Sedimentological control on OC loadings ...... 89

viii

4.5.2. Carbon isotopic composition of Krishna-Godavari and Mahanadi System: implication for the provenance and fate of OC...... 95 4.5.3. Comparison with other leaf-wax record ...... 99 4.5.4. Regional Synthesis of Holocene monsoon records...... 100 4.5.5. Potential relationship between climate (i.e. ISM intensity) and anthropogenic activities. 102 4.6. Conclusions ...... 103 References ...... 104

Chapter 5: Conclusions and outlook ...... 113 6.1. Conclusions...... 114 6.1.1. Provenance and evolution of organic matter in the Godavari River Basin and linkage to the marine sedimentary archive ...... 114 6.1.2. Timescale of terrestrial OC export ...... 115 6.1.3. Long-term Indian monsoon variability: impacts on anthropogenic activity and carbon cycling...... 116 6.2. Outlook ...... 117 References ...... 118

ix

Chapter 1: General Introduction

1

1.1. Introduction

The Asian monsoon system influences about 12 of the world’s largest 20 rivers ranked by highest total sediment discharge to the oceans (Milliman and Meade, 1983, Milliman and Farnsworth, 2011), and affects the most densely populated regions of the planet, thus rendering the monsoon an important driver of terrigenous sedimentation. The Asian summer monsoon is one of the most dynamic components of the Earth’s climate system and it is generally subdivided into the Indian summer monsoon (ISM), East Asian summer monsoon, (EASM), and Western North Pacific summer monsoon (WNPSM) (Wang et al., 2001). The ISM is a coupled ocean- atmosphere-land climate system that is driven by cross-equatorial pressure gradients. It is amplified by land-sea thermal gradients resulting in the low-level advection of warm, moisture-laden air from the Indian Ocean (Wu et al., 2012) and insulation of these air masses from the extratropics by the Himalayas (Boos and Kuang, 2010). This generates intense precipitation to the Bay of Bengal, the southern flank of the Himalayas, and the otherwise arid Indian subcontinent during the peak of the boreal summer (Hein et al., 2017). This creates a marked seasonality characterized by high annual precipitation during the summer months (i.e. June to September) (Gadgil, 2003). Variability in monsoon offset, magnitude and/or duration has been invoked as a major factor controlling flood versus drought phases that have plagued much of human history on the Indian peninsula (Mooley and Parthasarthy, 1983). Recently, global climate change has dominated scientific, socio-economic, and socio-political debates owing to growing perturbations of the Earth’s natural systems. It is now well- established that a myriad of natural processes (such as volcanism, rock weathering, metamorphic outgassing, and subduction of marine sediments) as well anthropogenic activities (such as fossil fuel combustion and land use changes) impart changes on the global climate (Marty and Tolstikhin, 1998; Walker et al., 1981; Becker et al., 2008; Kelemen and Manning, 2015). Central to this change in climate is the carbon cycle, thus understanding the carbon cycle represents a topical issue in biochemistry and earth system science. Most of the aforementioned natural processes operate on geological time scales. However, on shorter timescales, global climate change is a result of the net balance of carbon between the atmosphere and the biosphere. The balance is created through intricate networks of sources and sinks that transfer carbon between reservoirs such as soils and vegetation, the ocean, and marine sediments

2

(Sarmiento and Gruber, 2006). By integrating terrestrial carbon sources and transferring this material to the ocean, rivers form a crucial link between the geologic and biospheric carbon cycles and the efficiency of this terrestrial to ocean transfer is controlled by a combination of climatic (e.g. precipitation), tectonic (e.g. uplift and erosion), and anthropogenic (land use change) processes (Molnar and England, 1990; Syvitski et al., 2005). Rivers deliver an estimated 200 megatons (Mt) of particulate organic carbon (POC) to the oceans annually (Milliman and Farnsworth, 2011; Schlünz and Schneider, 2000), rendering rivers as major conduits of organic carbon (OC) transfer. About 90% of global OC burial in the modern oceans occurs on the continental margins (Hedges and Keil, 1995). Therefore, continental margins serve as a nexus between fluvial and marine realms and studying carbon dynamics on fluvial- dominated continental margins provides significant insights into sources of OC, including the processes controlling its nature and type as it transits through the land- ocean continuum, and its ultimate fate upon burial in marine sedimentary archives. In this thesis, a detailed investigation of sediments collected on the eastern continental margin of India is performed to assess OC sources and composition through time. These margin sediments are complemented by fluvial sediments and soils collected from a major modern river system to assess the spatial evolution of OC composition from source to sink. With the aid of organic geochemical tools, the relationship between climate change, anthropogenic perturbations, and OC dynamics from the early Holocene to the present is explored. This relationship between climate, geological and human activities provides a better understanding of monsoon variability and carbon cycle dynamics especially in the face of globally warming world.

1.2. The global carbon cycle

The concentration of CO2 in the atmosphere plays a crucial role in modulating global climate. This is because the atmosphere is relatively small in size relative to other carbon reservoirs (e.g. soils, ocean etc.) and has a short turnover time (Drenzek et al., 2009). Thus, any small perturbation in other carbon pools can produce large changes in the atmospheric carbon inventory. Recently, the carbon cycle has become a topical issue in society and policy formulation since the effect of elevated atmospheric CO2 on global climate has been recognized. In fact, atmospheric levels exceeded the 400 ppm mark in 2015, similar to levels last seen millions of years ago (Graven, 2015).

3

This significant increase in atmospheric CO2 levels has been attributed to fossil fuel combustion and anthropogenic land use change beginning during the Industrial Revolution (circa 1850)- a period aptly named “the anthropocene” (Crutzen, 2002). Over geological timescales, the net balance between processes that consume or produce CO2 regulates its concentration in the atmosphere. Silicate weathering and OC burial in marine sediments are important carbon sinks whereas volcanic activity, carbonate and OC-rich sediment weathering are important sources of CO2 to the atmosphere (Berner, 2003). As centres of OC burial, continental margins play a significant role in modulating Earth’s atmospheric chemistry and global climate (Berner, 1982). Continental margin settings are major OC repositories and include some of the most important sites of active organic matter burial on earth (Hedges and Keil, 1995) as they represent a dynamic region where rivers, estuaries, land, atmosphere, and ocean interact (Canuel et al., 2012). By virtue of the co-deposition of terrestrial and marine OC, and high sedimentation rates, investigations on continental margins sediments constrain past land and ocean conditions. A substantial proportion of sediment and organic matter eroded from the continents is deposited and stored on continental margins. Indeed, more than 85% of the global burial flux of terrestrial OC has been estimated to occur on continental margins, underscoring their role in the global carbon cycle (Hedges and Oades, 1997). Owing to the integrative property of rivers (Vier at al., 2009), sediments deposited on river dominated continental margins provide integrated records of both terrestrial and marine processes and provide important insights into paleoenvironmental conditions as well as source to sink processes (Hedges et al., 1997). Terrestrial OC is transported to the ocean via rivers and eolian processes with rivers accounting for more than 80% of the land-ocean flux (Drenzek 2007). The sources of this OC include a mixture of vascular plant debris, soils, rock erosion, biological productivity within the river, and anthropogenic emissions (Blair et al., 2010)

1.3. Rivers - the “arteries” of the planet

Rivers constitute a vital component of the land-ocean-atmosphere loop of the global carbon cycle by providing a crucial conduit for the transfer of carbon (as particulate or dissolved) between land and ocean; and discharging more than 20 billion tons of

4 sediment to the global ocean (Milliman and Farnsworth, 2011). They serve as a mechanism by which CO2 molecules, captured by terrestrial plants and incorporated into tissues during photosynthesis, are delivered and sequestered in continental margin sediments. A conceptual model showing the various pathways for terrestrial particulate carbon is presented in Fig. 1.1. The model is comprised of a series of interconnected pools where terrestrial OC resides and is transformed and/or degraded before moving to the next reservoir and eventually discharged to, and preserved in, the receptacle (i.e. sedimentary basin). The relative fluxes of materials through and around reservoirs, and the residence time within reservoirs vary and considered key controls on OC character (Blair et al., 2004). While some terrestrial OC may bypass a reservoir completely (i.e. instantaneous transfer of vascular plant OC to the continental margin, Fig. 1.1), others may reside in soils or lowland storage for thousands of years before being eroded, transported and ultimately buried. In the former case, the OC may experience little to no alteration while in the latter; the bulk of the OC is significantly altered. Earlier estimates of the flux of sediment (and associated OC) to the global ocean were based on extrapolation of measurements from a few large river basins (mostly tropical rivers with wide continental shelves) to the entire of Earth’s surface (e.g. Holland, 1981). The Indian summer monsoon feeds some of the largest sediment-carrying rivers in the world (Syvitski and Saito, 2007) including the Ganges, Brahmaputra, Indus, Irrawady, Mahanadi, Godavari, Krishna, and Kaveri. These large sediment loads contribute to the development of river-dominated margins around the Bay of Bengal (BoB) which are characterized by high sedimentation rates. These high accumulation rates facilitate the development of high-resolution temporal records that allow for a more detailed reconstruction of the Indian monsoon at a regional scale.

5

Figure 1.1. Conceptual model of the source to sink transport of terrestrial OC to the oceans (Modified from Blair et al., 2004).

1.4. Motivation and objectives

Prior to the period when human impact on the natural system became significant, fluvial systems and the landscapes they drained were solely influenced by the geologic and climatic conditions of the drainage basins. However, since the late Holocene, both human activity and/or climatic variations have become major driving factors that influence sediment compositions and fluxes. For instance, while climate-driven processes (e.g. changes in precipitation) can lead to a significant increase or decrease in sediment yield, anthropogenic activities (e.g. damming, land-use change) can attenuate or amplify this process to produce a negative, near-zero, or positive net effect on sediment export from the drainage basin. Therefore, a thorough understanding of the balance and interplay between climate and anthropogenic activity in controlling the amount, nature and provenance of materials discharged from river basins is required for a holistic assessment of OC export to the continental and

6 its impacts on margin on the global carbon cycle. To this end, this thesis focuses on a detailed geochemical investigation coupled with sedimentological characterization of OC spanning the entire Holocene. In addition, it provides new insights on the key processes that are involved in the mobilization and translocation of terrestrial OC to continental margin sediments; and assessed geomorphic, climatic, and anthropogenic controls on the dynamics of OC export from river basins. With the aid of bulk inorganic, organic, and molecular proxies that record these processes; insights into the magnitude and timescales of these perturbations were gained. Particular emphasis is placed on reconciling regional Holocene changes in terrestrial OC signatures with those recorded in adjacent continental margin sediments as well as providing regionally extensive paleoenvironmental reconstructions of the Indian peninsula.

1.5. Thesis outline

This thesis provides new and improved Holocene records of Indian monsoon variability using sediment cores from river-dominated continental margins in the BoB. In addition, a regional overview of organic carbon and sediment dynamics was produced through the collection of river basin sediments and the interplay between human and climate and its effects on organic carbon export and burial was explored. Results from this doctoral project are structured in three main results chapters (2-4) and a synthesis/outlook chapter (5): Chapter 2 presents results of the sedimentological, bulk elemental and isotopic composition of organic matter as well as compound-specific carbon isotopic analysis of soils and riverine sediments collected from the modern Godavari river catchment. These samples are complimented by offshore sedimentary sediments collected at the mouth of the Godavari River to reconstruct the Holocene paleoclimate of the river basin. By reconciling offshore sedimentary signatures with those of the modern river system, information on the provenance and the degree of organic matter-mineral relationship during OC mobilization from source to sink are gleaned in more detail. Chapter 3 focuses exclusively on the high-resolution reconstruction of Holocene climate in the Godavari Basin using a sediment core retrieved off the mouth of the Godavari River in the northwestern BoB. Bulk organic geochemical and sedimentological parameters were discussed briefly in the previous chapter while this

7 chapter explores the compound-specific geochemical characteristics of sediments exported by the Godavari River. Fatty acids are employed exclusively as the biomarker of choice because they are highly abundant in marine sediments, thus yielding sufficient mass for compound-specific radiocarbon analysis and are relatively easily extracted for isotope analysis. In addition, they contain information about different sources within the same homologous series and are insensitive to fossil contamination resulting from anthropogenic influences. This allows for coupled molecular isotope (d13C and D14C) assessment of OC sources and evolution through the Holocene in the Godavari. This approach is then used to resolve the pre-aged terrestrial component of the OC delivered to the Godavari margin. Chapter 4 presents high-resolution records of sediment cores recovered at the mouth of the Krishna-Godavari (southwest of core 16A in the previous chapter) and the Mahanadi Rivers. Comparisons between the sedimentological parameters as well as bulk organic and compound-specific geochemical signature of both locations provide insights into the timing and magnitude of the ISM in both basins. Results were then compared with regionally available records of ISM variations. This comparison facilitates the regional scale assessment of latitudinal variations in ISM impacts and its timing across the Indian subcontinent. In addition, a record of a sediment core recovered at the mouth of the Cauvery/Kaveri River on the southwestern flank of the Indian peninsula is examined. Radiocarbon measurements on organic and inorganic sediment components reveal high heterogeneity across the entire Holocene, indicative of slumping and drift deposit, effectively rendering this record unsuitable for paleoclimatic reconstruction of the Kaveri basin. Chapter 5 summarizes the main findings of the thesis, highlights the key challenges, poses questions that are beyond the scope of this research, and offers some perspectives on future research avenues. This thesis presents the first high-resolution organic geochemical record of the Indian summer monsoon from the three largest non-Himalayan rivers in the Eastern continental margin of India. It builds upon previous works in the northwestern BoB (e.g. Ponton et al., 2012; Giosan et al., 2017, Cui et al., 2017; Usman et al., 2018) and contributes towards a more holistic understanding of the Indian monsoon throughout the entire Holocene. By combining field observation of the modern river system and

8 sedimentary archives of climate proxies, this thesis explores linkages between climatic and anthropogenic impacts on organic matter export and/or burial.

References

Becker, J. A., Bickle, M. J., Galy, A., Holland, T. J. B. (2008). Himalayan metamorphic

CO2 fluxes: Quantitative constraints from hydrothermal springs. Earth and Planetary Science Letters 265, 616-629. Berner, R. A. (1982). Burial of organic-carbon and pyrite sulfur in the modern ocean – its geochemical and environmental significance. American Journal of Science 282, 451-473. Berner, R. A. (2003). The long-term carbon cycle, fossil fuels, and atmospheric composition. Nature 426, 323-326. Blair, N. E., Leithold, E. L., Aller, R. C. (2004). From bedrock to burial: The evolution of particulate organic carbon across coupled watershed-continental margin systems. Marine Chemistry 92, 141-156. Blair, N. E., Leithold, E. L., Brackley, H., Trustum, N., Page, M., Childress, L. (2010). Terrestrial sources and export of particulate organic carbon in the Waipaoa sedimentary system: problems, progress and processes. Marine Geology 270, 108- 118. Boos, W.R. and Kuang, Z. (2010). Dominant control of the South Asian monsoon by orographic insulation versus plateau heating. Nature 463, 218-223. Canuel, E. A., Cammer, S. S., McIntosh, H. A., Pondell, C. R. (2012). Climate change impacts on the OC cycle at the land-ocean interface. Annual Review of Earth and Planetary Sciences 40, 685-711. Crutzen, P. J. (2002). Geology of mankind. Nature 415, 23. Cui, M., Wang, Z., Nageswara Rao, K., Sangode, S. J., Saito, Y., Chen, T., Kulkarni, Y. R., Naga Kumar, K. Ch. V., Demudu, G. (2017). A mid- to late-Holocene record of vegetation decline and erosion triggered by monsoon weakening and human adaptations in the south-east Indian Peninsula. The Holocene 27, 1976-1987. Drenzek, N. J. (2007). The temporal dynamics of terrestrial organic matter transfer to the oceans: initial assessment and application. Ph.D. thesis, MIT/WHOI Joint program in Oceanography/Applied Ocean Science and Engineering.

9

Drenzek, N. J., Hughen, K. A., Montlucon, D. B., Southon, J. R., dos Santos, G. M., Druffel, E. R. M., Giosan, L., Eglinton, T. I. (2009). A new look at old carbon in active margin sediments. Geology 37, 239-242. Graven, H. D. (2015). Impact of fossil fuel emissions on atmospheric radiocarbon and various applications of radiocarbon over this century. Proceedings of the National Academy of Sciences 112, 9542-9545. Gadgil, S. (2003). The Indian monsoon and its variability. Annual Review of Earth and Planetary Sciences 31, 429-467. Giosan, L., Ponton, C., Usman, M., Blusztajn, J., Fuller, D. Q., Galy, V., Haghipour, N., Johnson, J. E., McIntyre, C., Wacker, L. and Eglinton, T. I. (2017). Short communication: Massive erosion in monsoonal central India linked to late Holocene land cover degradation. Earth Surface Dynamics 5, 781-789. Hedges, J. I., and Keil, R. G. (1995). Sedimentary organic matter preservation: an assessment and speculative synthesis. Marine Chemistry 49, 81-115. Hedges, J. I. and Oades, J. M. (1997). Comparative organic geochemistries of soils and marine sediments. Organic Geochemistry 27, 319-361. Hedges, J. I., Keil, R. G., Benner, R. (1997). What happens to terretsrial organic matter in the ocean. Organic Geochemistry 27, 319-361. Hein, C. J., Galy, V., Galy, A., France-Lanord, C., Kudrass, H., Schwenk, T. (2017). Post-glacial climate forcing of surface processes in the Ganges-Brahmaputra river basin and implications for carbon sequestration. Earth and Planetary Science Letters 478, 89-101. Holland, H. D. (1981). River transport to the oceans, in: Emiliani, C. (Ed.). The sea. The ocean lithosphere. 7:763-800. Wiley, New York. Kelemen, P. B. and Manning, C. E. (2015). Re-evaluating carbon fluxes in subduction zones, what goes down, mostly comes up. Proceedings of the National Academy of Sciences 112, 3997-4006. Marty, B. and Tolstikhin, I. N. (1998). CO2 fluxes from mid-ocean ridges, arcs, and plumes. Chemical Geology 145, 233-248. Milliman, J.D. and Meade, R. H. (1983). World-wide delivery of river sediment to the ocean. Journal of Geology 91, 1-21. Milliman, J. and Farnsworth, K. (2011). Runoff, erosion, and delivery to the coastal ocean, in: River discharge to the coastal ocean: A global synthesis. Cambridge University Press, Cambridge. 10

Molnar, P. and England, P. (1990). Late Cenozoic uplift of mountain ranges and global climate change: chicken or egg? Nature 346, 29-34 Mooley, D.A., Parthasarathy, B. (1983). Droughts and floods over India in summer monsoon seasons 1871-1980. In: Street-Perrott, A et al. (eds.) Variations in the global water budget. 239-252. Ponton, C., Giosan, L., Eglinton, T. I., Fuller, D. Q., Johnson, J. E., Kumar, P. and Collett, T. S. (2012). Holocene aridification of India. Geophysical Research Letters 39, 6.

Sarmiento, J. L. and Gruber, N. (2006). Carbon cycle, CO2, and climate, in: Ocean Biogeochemical Dynamics. Princeton University Press. Schlünz, B and Schneider, R. R. (2000). Transport of terrestrial organic carbon to the oceans by rivers: Re-estimating flux and burial rates. International Journal of Earth Sciences 88, 599-606. Syvitski, J.P. M., Vörösmarty, C. J., Kettner, A. J., Green, P. (2005). Impacts of humans on the flux of terrestrial sediment to the global coastal ocean. Science 308, 376-380. Syvitski, J. P. M. and Saito, Y. (2007). Morphodynamics of deltas under the influence of humans. Global Planetary Change 57, 261-282. Usman, M. O., Kirkels, F. M. S. A., Zwart, H. M., Basu, S., Ponton, C., Blattmann, T. M., Plötze, M., Haghipour, N., McIntyre, C., Peterse, F., Lupker, M., Giosan, L., Eglinton, T. I. (2018). Reconciling drainage and receiving basin signatures of the Godavari River system. Biogeosciences. Accepted. Vier, J., Dupre, B., Gaillardet, J. (2009). Chemical composition of suspended sediments in world rivers: New insights from a new database. Science of the Total Environment 407, 853-868 Walker, J., Hays, P. B., Kasting, J. F. (1981). A negative feedback mechanism for the long-term stabilization of Earth’s surface temperatures. Journal of Geophysical Research 86, 9776-9782. Wang, Y., Cheng, H., Lawrence Edwards, R., Kong, X., Shao, X., Chen, S., Wu, J., Jiang, X., Wang, X., An, Z. (2001). Millennial- and orbital-scale changes in the East Asian monsoon over the past 224,000 years. Nature 451, 1090-1093 Wu, G. X., Liu, Y., He, B., Bao, Q., Duan, A., Jin, F.-F. (2012). Thermal controls on the Asian summer monsoon. Scientific Reports 2, 404.

11

Chapter 2: Reconciling drainage and receiving basin signatures of the Godavari River system

This chapter originally appeared as:

Usman, M.O., Kirkels, F.M.S.A., Zwart, H. M., Basu, S., Ponton, C., Blattmann, T.M., Ploetze, M., Haghipour, N., McIntyre, C., Peterse, F., Lupker, M., Giosan, L., Eglinton, T. I. 2018. Reconciling drainage and receiving basin signatures of the Godavari River system. Biogeosciences, Vol. 15, 3357-3375. Copyright, 2018, Authors.

Reproduced with the permission of the European Geoscience Union.

12

Biogeosciences, 15, 3357–3375, 2018 https://doi.org/10.5194/bg-15-3357-2018 © Author(s) 2018. This work is distributed under the Creative Commons Attribution 4.0 License.

Reconciling drainage and receiving basin signatures of the Godavari River system

Muhammed Ojoshogu Usman1, Frédérique Marie Sophie Anne Kirkels2, Huub Michel Zwart2, Sayak Basu3, Camilo Ponton4, Thomas Michael Blattmann1, Michael Ploetze5, Negar Haghipour1,6, Cameron McIntyre1,6,7, Francien Peterse2, Maarten Lupker1, Liviu Giosan8, and Timothy Ian Eglinton1 1Geological Institute, ETH Zürich, Sonneggstrasse 5, 8092 Zürich, Switzerland 2Department of Earth Sciences, Utrecht University, Heidelberglaan 2, 3584 CS Utrecht, the Netherlands 3Department of Earth Sciences, Indian Institute of Science Education and Research Kolkata, 741246 Mohanpur, West Bengal, India 4Division of Geological and Planetary Science, California Institute of Technology, 1200 East California Boulevard, Pasadena, California 91125, USA 5Institute for Geotechnical Engineering, ETH Zürich, Stefano-Franscini-Platz 3, 8093 Zürich, Switzerland 6Laboratory of Ion Beam Physics, ETH Zürich, Otto-Stern-Weg 5, 8093 Zürich, Switzerland 7Scottish Universities Environmental Research Centre AMS Laboratory, Rankine Avenue, East Kilbride, G75 0QF Glasgow, Scotland 8Geology and Geophysics Department, Woods Hole Oceanographic Institution, 86 Water Street, Woods Hole, Massachusetts 02543, USA

Correspondence: Muhammed Ojoshogu Usman ([email protected])

Received: 12 January 2018 – Discussion started: 8 February 2018 Revised: 18 May 2018 – Accepted: 24 May 2018 – Published: 7 June 2018

Abstract. The modern-day Godavari River transports large sediment mineralogy, largely driven by provenance, plays an amounts of sediment (170 Tg per year) and terrestrial organic important role in the stabilization of OM during transport carbon (OCterr; 1.5 Tg per year) from peninsular India to the along the river axis, and in the preservation of OM exported Bay of Bengal. The flux and nature of OCterr is considered to by the Godavari to the Bay of Bengal. The stable carbon have varied in response to past climate and human forcing. In isotopic (13C) characteristics of river sediments and soils order to delineate the provenance and nature of organic mat- indicate that the upper mainstream and its tributaries drain ter (OM) exported by the fluvial system and establish links to catchments exhibiting more 13C enriched carbon than the sedimentary records accumulating on its adjacent continen- lower stream, resulting from the regional vegetation gradient tal margin, the stable and radiogenic isotopic composition and/or net balance between the upper (C4-dominated plants) of bulk OC, abundance and distribution of long-chain fatty and lower (C3-dominated plants) catchments. The radiocar- 14 acids (LCFAs), sedimentological properties (e.g. grain size, bon contents of organic carbon (1 COC) in deep soils and mineral surface area, etc.) of fluvial (riverbed and riverbank) eroding riverbanks suggests these are likely sources of “old” sediments and soils from the Godavari basin were analysed or pre-aged carbon to the Godavari River that increasingly and these characteristics were compared to those of a sed- dominates the late Holocene portion of the offshore sedimen- iment core retrieved from the continental slope depocenter. tary record. While changes in water flow and sediment trans- Results show that river sediments from the upper catchment port resulting from recent dam construction have drastically exhibit higher total organic carbon (TOC) contents than those impacted the flux, loci, and composition of OC exported from the lower part of the basin. The general relationship from the modern Godavari basin, complicating reconciliation between TOC and sedimentological parameters (i.e. mineral of modern-day river basin geochemistry with that recorded surface area and grain size) of the sediments suggests that in continental margin sediments, such investigations provide

Published by Copernicus Publications on behalf of the European Geosciences Union.

13

3358 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system important insights into climatic and anthropogenic controls The Godavari River basin (Fig. 1) is an example of on OC cycling and burial. a monsoon-influenced low-latitude river basin and, as the largest non-Himalayan river in India, is of special interest due to its large catchment size and sediment flux to the ocean (Kale, 2002). Draining central peninsular India, the river in- 1 Introduction tegrates rainfall within the core monsoon zone of central In- Rivers form a key component of the global carbon cycle, dia, both reflecting the mean monsoon regime and captur- transporting about 200–400 Tg of particulate organic carbon ing fluctuations in monsoonal rains over the sub-continent. (POC) to the oceans annually (Degens et al., 1991; Ludwig With over 90 % of discharge from the Godavari deriving et al., 1996; Schlünz and Schneider, 2000), with the major- from summer monsoon precipitation (Rao et al., 2005), cor- ity of this POC deposited on the continental margins (Berner, responding offshore sedimentary sequences record past vari- 1989; Hedges, 1992). Much of this POC is mobilized from ations in continental climate as well as anthropogenic activ- soils (Meybeck, 1982; Tao et al., 2015) and augmented by ity within the drainage basin (Cui et al., 2017; Giosan et al., recently biosynthesized higher plant debris, recycled fossil 2017; Ponton et al., 2012; Zorzi et al., 2015). OC derived from erosion of sedimentary rocks, and in situ Prior studies, spanning the Holocene, of a Godavari River- aquatic productivity within the rivers (Hedges et al., 1986). proximal sediment core NHGP-16A from the Bay of Ben- Rivers act not only as conduits linking terrestrial and marine gal (BoB) have revealed marked geochemical and sedimen- tological variations that have been interpreted in the con- reservoirs but also as reactors where terrestrial OC (OCterr) is subject to a myriad of processes resulting in degradation text of both evolving regional hydroclimate and accompa- and modification of the suspended OC load (Aufdenkampe nying changes in land use within the Godavari catchment et al., 2011; Wu et al., 2007; Cole et al., 2007). Although (Giosan et al., 2017; Ponton et al., 2012). Specifically, a dis- 13 a general framework for describing the origin and evolution tinct change in the stable carbon isotopic ( C) composition of molecular markers of terrestrial vegetation implies an in- of OCterr in different types of river basins is emerging (e.g. Blair and Aller, 2012), a detailed understanding of the im- crease in the proportion of aridity-adapted C4 vegetation be- pact of the diverse and complex array of processes occurring ginning around 4.5 kyr BP (Ponton et al., 2012). This shift in vegetation type is accompanied by increased variability in within river basins on the amount and composition of OCterr that is ultimately exported offshore is still developing. the oxygen isotopic composition of planktonic foraminiferal The flux and nature of OC discharged to the ocean is de- carbonate, suggesting enhanced hydrological variability, po- pendent on a number of factors, including the composition tentially reflecting less frequent (“break monsoon”) but in- of underlying bedrock, geomorphologic properties, and cli- tense rainfall activity within the drainage basin (Ponton et matic factors like temperature and precipitation (Hilton et al., al., 2012). Subsequent down-core geochemical and sedimen- 2008; Leithold et al., 2006). Climate variability on millen- tological measurements on the Godavari-proximal BoB sed- nial and longer timescales is considered to exert an impor- iments have served to paint a more comprehensive picture of tant influence on the export of OC from the terrestrial bio- past changes within the Godavari basin (Giosan et al., 2017). sphere and burial in ocean sediments, with important feed- Notably, sharp increases in sediment accumulation rate dur- ing the late Holocene imply a concomitant increase in flu- backs on atmospheric CO2. Variations in exhumation, oxi- dation, and burial of bedrock OC exported from river basins vial sediment discharge, despite the onset of increasingly arid is considered to exert fundamental controls on atmospheric conditions. Furthermore, detrital Neodymium (Nd) isotopic compositions indicate a shift in sediment provenance at ca. CO2 / O2 balance over longer (> million year) timescales (Berner, 2003). 4.5 kyr BP from a relatively unradiogenic signature consis- Tropical and subtropical rivers are estimated to account for tent with lower-basin bedrock as the primary detrital mineral sources prior to increased contributions from the more ra- more than 70 % of the global OCterr delivery to the oceans (Ludwig et al., 1996; Schlünz and Schneider, 2000) and thus diogenic rocks in the upper basin (Deccan Plateau). Finally, comprise major vectors in land–ocean carbon transport (Auf- these changes recorded in the sediment core are also asso- 14 denkampe et al., 2011; Galy and Eglinton, 2011; Hedges et ciated with increased C age offsets between bulk OC and al., 1986; Schefuss et al., 2016; Spencer et al., 2012). The coeval planktonic foraminifera, suggesting enhanced erosion discharge of such rivers is sensitive to variations in climate, and export of pre-aged OCterr exhumed from deeper soil lay- such as the location and intensity of monsoonal rains and ers (Giosan et al., 2017). Collectively, these different lines dry-season droughts. Fluvially derived OC, deposited and of evidence are consistent with an overall scenario in which preserved in adjacent continental margins, serves as a rich increasing aridity results in a shift in the type (from decid- archive of information on past perturbations in continental uous to shrub or grass) and extent (reduced) of vegetation climate and fluvial dynamics (Bendle et al., 2010; Schefuss coverage, while changes in the pattern and frequency of sea- et al., 2011; Weijers et al., 2007). sonal monsoons promote enhanced soil erosion in the dri- est regions of the upper basin. Increased soil loss may have been exacerbated by human activity through intensification

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

14

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3359

Godavari River

------

-

Figure 1. (a) Location of the Godavari River basin in central peninsular India. (b) Sampling locations along the river basin. Upper- (UB) and lower-basin (LB) samples are shown in red and blue colours, respectively (Modified from Pradhan et al., 2014). (c) The Godavari drainage basin in its ecological (i), hydroclimatic (ii), geological (iii), and soil cover (iv) context (Modified from Giosan et al., 2017).

www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

15

3360 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system of agriculture and the implementation of irrigation practices b), strongly affecting the precipitation pattern over peninsu- that amplify soil disturbance and destabilization (Giosan et lar India, with monsoonal rains falling preferentially between al., 2017). the coast and the Ghats, leaving much of the inland region Although the above interpretations appear to be consistent with lower precipitation (Gunnel et al., 2007). As a result, with available geochemical observations, support is lacking the upper river catchment, spanning the Deccan Plateau, is from direct observations on spatial variations in geochem- characterized by arid to semi-arid vegetation and lower an- 1 ical, mineralogical, and sedimentological properties within nual precipitation (< 800–1200 mm yr ), while moist and the Godavari drainage basin. Furthermore, direct attribution deciduous vegetation and higher annual precipitation (1600– 1 of signatures observed in the sediment record with those of 3200 mm yr ) typifies the lower basin (Asouti and Fuller, the drainage basin remains elusive. In the present study, we 2008) (Fig. 1c, panels i and ii). assess the extent to which terrestrial signatures recorded in The underlying rock formations exert significant control river-proximal continental margin sediments can be recon- on sediment and solute transport by rivers. Based on their ciled with those within (specific regions of) the river basin. erodibility, rock formations in the Godavari are categorized In particular, we seek to establish whether OC characteristics as follows (Biksham and Subramanian, 1988): (a) Deccan of the basin are consistent with those of distal sediments de- Traps, which are volcanic in origin and of Tertiary age, are posited during the Holocene on the adjacent continental mar- known for their distinct spheroidal weathering and high flu- gin. In addition to bulk and molecular characteristics of par- vial erosion (Subramanian, 1981). The whole Deccan Plateau ticulate organic matter, we explore quantitative and composi- (representing 48 % of the basin area; Fig. 1) is covered by ⇠ tional relationships to mineral phases in soils, river and ma- 10–40 cm thick black clay loam, which serves as a source rine sediments. Such an approach comparing drainage basin of riverine sediments; (b) sedimentary rocks (mostly sand- and adjacent continental margin signatures may prove crucial stones) of Mesozoic–Cenozoic age located in the central and in delineating the nature and provenance of signals preserved lower part of the catchment ( 11 % of basin area) are known ⇠ in marine sedimentary sequences in the receiving basin, and for their high degree of erodibility; (c) Precambrian gran- hence for informed interpretation of corresponding down- ites, charnockites, and similar hard rocks ( 39 % of total ⇠ core records. Specific questions include the following: (i) to basin area) are characterized by low erodibility. River trib- what extents do offshore sedimentary signatures reflect char- utaries draining through these relatively stable rock forma- acteristics of the modern-day basin, and what is their prove- tions (e.g. Sabri and Indravati) carry low sediment loads. nance? (ii) How and to what degree are organic and mineral Compared to the Deccan volcanic rocks, soils derived from matter (de-)coupled during mobilization and transfer from the erosion of sedimentary and Precambrian rocks preva- source to sink? By addressing these questions, we aim to im- lent in the eastern segment of the basin are generally thinner prove our understanding of carbon flow through river basins, (< 15 cm) and reddish–yellowish in colour (Bhattacharyya as well as to better inform interpretation of geochemical sig- et al., 2013). Sediments transported by the Godavari are thus nals preserved in river-dominated sedimentary sequences. mostly derived from the Deccan Traps and from granitoids of the Indian Craton (Biksham and Subramanian, 1988). These contrasting bedrocks manifest themselves in corresponding 2 Materials and methods isotopic signatures, where relatively young Deccan volcanic rocks are characterized by highly radiogenic mantle-derived 2.1 Study area material ("Nd 1 5, 87Sr / 86Sr 0.701), while the rel- = ± = atively old Indian Craton is unradiogenic ("Nd 35 8, = ± The Godavari River is the largest monsoon-fed river basin 87Sr / 86Sr 0.716) (Giosan et al., 2017; Tripathy et al., = and the third largest river (behind Ganges and Brahmaputra) 2011) (Fig. 1c, panel iii). of India, delivering 170 Tg per year of sediment and 1.5 Tg Spatial variations in soil types and coverage of the basin per year of OC to the BoB (Biksham and Subramanian, 1988; are described by Gupta et al. (1997). Black soils (Vertisols, Ludwig et al., 1996). It originates from Sahyadris in the Vertic Inceptisols, and Entisols) are prevalent in the central Western Ghats and flows toward the east-south-east across and western parts of the basin. The eastern part of the basin the Indian peninsula, traversing various geological and vege- is dominated by red–yellow soils (Alfisols and Luvisols), and tation gradients before emptying into the BoB (Fig. 1). Four in the estuarine–deltaic region, soil type varies over relatively major tributaries (Purna, Pranhita, Indravati, and Sabri) drain short distances (Gupta et al., 1997) (Fig. 1c, panel iv). over 60 % of the basin area, and the modern-day catchment The Godavari River emerges from the Eastern Ghats on ( 3 105 km2) supports a population of about 75 million the coastal plain near Rajahmundry, from where it has built ⇠ ⇥ people (Pradhan et al., 2014). The basin experiences pro- a large delta in conjunction with the neighbouring Krishna nounced seasonality with marked wet and dry seasons, and River that empties into the BoB and delivers sediment to the majority of annual rainfall occurs during June–September the pericratonic Krishna–Godavari Basin (Manmohan et al., and is associated with the moist south-west monsoon winds. 2003). The latter, located in the central part of the eastern The Western Ghats act as an orographic barrier (Fig. 1a and continental margin of peninsular India, formed as a result

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

16

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3361 of the down-warping of the eastern segment of the Indian sites were chosen to represent the dominant soil type of the Shield subsequent to the break-up of Gondwanaland (Murthy given region, and were sampled on level ground and close et al., 1995). Unlike the Himalayan rivers that adjust to large- to rivers. Surface soils and litter (0–5 cm) were collected us- magnitude monsoon floods by increasing their width and ing a small hand shovel. Additionally, undisturbed soil pro- width–depth ratio (Coleman, 1969), the incised channel of files were obtained at some targeted locations (Fig. 1b) using the Godavari responds to the increase in discharge by de- a metre-long coring device, and where possible were sam- creasing its width–depth ratio (Kale, 2002). Because of co- pled to bedrock. Soil cores were then sub-sectioned into a 0– hesive banks and incised channel morphology in the lower 5 cm (“shallow/surface”) interval, and every 10 cm thereafter basin, shifts in channel position are rare, resulting in lim- (“deep”). These depths were chosen to represent the likely ited overbank sediment deposition and restricted areal extent sources of shallow (surface run-off) and deeper (e.g. bank) of the floodplain. As a consequence of this limited accom- soil erosion and supply to nearby streams. At a few sites, modation space in the lower basin, fluvial sediments either road constructions provided access to complete soil sections accumulate in the delta or are exported to the BoB. Further- that were sampled at 10 cm intervals. more, sediment trapping on the continental shelf is minimal Riverbed sediments were collected from the middle of the because the shelf in front of the Godavari Delta is narrow stream either with a Van Veen grab sampler from bridges or (generally < 10 km), promoting more rapid and direct trans- with a hand shovel where the river was very shallow. The port of fluvial sediments to the continental slope. Also, satel- sampling sites were selected as being representative of the lite images reveal a plume of suspended river sediments from local depositional settings of the rivers and its tributaries, and the Godavari mouth out into the BoB past the continental they mostly comprise areas dominated by bedload sediments shelf, confirming delivery of riverine sediments to the slope (channel thalweg) with particle sizes ranging from < 2 µm (Sridhar et al., 2008). Therefore, no major lags in or modifi- (clay) to 2 mm (coarse sand) and minor proportions of peb- cations to the fluvial signals between discharge from the river bles and plant debris. Where a tributary joins the mainstem of and deposition on the continental slope are expected. This the Godavari, sampling was conducted before the confluence sedimentary regime of the Godavari system thus allows for of the two rivers and shortly downstream of the confluence relatively straightforward interpretation of sediment sources so as to assess the integrated signal of the sub-catchments. and transfer processes (Giosan et al., 2017) and facilitates Where present, riverbank sediments that represent loose direct comparison between characteristics of drainage basin and unconsolidated freshly deposited suspended sediments and BoB sediments. were also collected with a hand shovel and as close to the Damming of the Godavari River and its tributaries has main river stem as possible. Upon arrival at ETH Zurich, all increased tremendously over the past several decades, with sediment and soil samples were stored frozen ( 20 C), then more than 300 hydrologic projects of various sizes currently freeze-dried and subsequently dry-sieved to < 2 mm to re- in operation that regulate water discharge and sediment trans- move the rock fragments and plant debris. About 20 soils port to the BoB. For the purpose of our study, we divide the and sediment samples were further milled to powder using drainage basin into two major sections: the upper-basin (UB) an agate ball mill. section (source to Pranhita River tributary) and lower-basin (LB) section (Pranhita to the BoB) as this captures the major 2.2.2 Offshore contrast in bedrock lithology and vegetation between the two segments, allowing for assessment and attribution of signals A piston offshore sediment (OS) core NGHP-01-16A emanating from these major parts of the river basin. It should (16.59331 N, 82.68345 E, 1268 m water depth) was col- be noted that though the Pranhita River has half of its catch- lected near the mouth of the Godavari River in the BoB (Col- ment in the upper basin, about 94 % of its total suspended lett et al., 2014) (Fig. 1b). The 8.5 m long core spanning the entire Holocene ( 11 kyr; Ponton et al., 2012) was analysed particulate matter (SPM) flux is derived from the Wardha and ⇠ Wainganga rivers in the lower reaches of the catchment (Bal- for sedimentological, mineralogical, and geochemical char- akrishna and Probst, 2005; Fig. 1), justifying the classifica- acteristics. Due to the top 25 cm of the core being exhausted tion of Pranhita into the lower basin. by prior investigations, our results are augmented with those from Ponton et al. (2012) and Giosan et al. (2017). The sedi- ment depth corresponding to 4.5 kyr BP (ca. 515 cm), rep- 2.2 Sampling ⇠ resenting the onset of the vegetation shift in peninsula In- 2.2.1 River basin dia during the Holocene, was designated as the boundary be- tween the early (EH) and late Holocene (LH). River sediments (flood deposits from the flank of the river and riverbed deposits) and soil sampling was carried out in February–March 2015, coinciding with the dry season. Sam- pling locations are shown in Fig. 1b, with additional de- tails provided in Table S1 in the Supplement. Soil sampling www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

17

3362 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system

2.3 Sample treatment and measurements diffractometer using Co-K↵ radiation. The instrument was equipped with an automatic theta-compensating divergence 2.3.1 Mineral surface area and anti-scattering slit, primary and secondary Soller slits and a Sol-X solid state detector. The phase composition was About 1 g dry weight (gdw) from each soil and sediment then determined using the DIFFRACplus software. Mineral sample (unground) was combusted at 350 C for 6 h in or- phases were identified on the basis of the peak position and der to remove the organic matter. The samples were then relative intensity in comparison to the PDF-2 database (In- outgassed at 350 C for 2 h in a vacuum oven to remove ad- ternational Centre for Diffraction Data). Quantification of sorbed moisture on the surface before analysis. Prior to anal- minerals was achieved with the BGMN/AutoQuan software ysis, samples were homogenized in an agate mortar, using a using Rietveld refinement (Bergmann and Kleeberg, 1998; plastic pestle to avoid crushing mineral grains. Surface area Bish and Plötze, 2011). of the mineral components of the sediment was analysed by the multi-point BET N2 adsorption method using a Quan- 2.3.4 Bulk elemental and isotopic analysis tachrome Monosorb Analyzer (Wakeham et al., 2009). The precision on duplicates of alumina standards was better than Aliquots of freeze-dried sediment or soil samples ( 50– ⇠ 1 %. 200 mg) were weighed into pre-combusted silver boats (Ele- mentar) and fumigated in a closed desiccator in the presence 2.3.2 Grain size of 12M HCl (70 C, 72 h) to remove inorganic carbon (Bao et al., 2016; Komada et al., 2008). The samples were subse- An aliquot ( 0.5 gdw) of combusted (350 C, 6 h) sediment ⇠ quently neutralized and dried over NaOH pellets to remove and soil samples processed for mineral surface area analysis 1 residual acid. The sample was then wrapped in tinfoil boats was treated with 10–15 mL of dissolved (40 g L ) sodium (Elementar), pressed, and analysed using a combined ele- pyrophosphate (Na P O 10H O) for about 12 h to disag- 4 2 7· 2 mental analyser, isotope ratio mass spectrometer, and accel- gregate the sediment grains. Sediment and soil grain size erator mass spectrometer (EA-IRMS-AMS) system at ETH distributions were measured using a Malvern Mastersizer Zurich (McIntyre et al., 2016; Wacker et al., 2010). The 2000 Laser Diffraction Particle Analyser that characterizes instrumental set-up, blank assessment, accuracy, and repro- particle sizes ranging from 0.04 to 2000 µm. Sediment and ducibility for the data presented here have been previously soil samples were measured in triplicate, with average me- reported in McIntyre et al. (2016). dian (d50) values reported. The standard deviation on tripli- 14 For down-core sediments, COC values were decay- cate analysis was better than 1 %. corrected for 14C loss since time of deposition (Eq. 1; Stu- iver and Polach, 1977). This decay correction is necessary 2.3.3 Sediment mineralogy to facilitate comparison of 14C values between the sediment 14 Eight sediment and soil samples were selected to represent core and C signatures in the modern river basin. The decay- varying regions and lithologies of the Godavari basin, and corrected radiocarbon level, 1, is calculated as follows: 12 samples taken from various depths in offshore core 16A were selected for X-ray diffraction (XRD) analysis. About 14 (1950 x) 1 (F Ce 1) 1000, (1) 1 g of bulk sediment was wet-milled in ethanol using a Mc- = · Crone Micronising Mill. The milled sample was then passed where F 14C measured fraction modern value of 14C, = through a 20 µm sieve and transferred into a ceramic bowl. (ln2) / 5730 yr 1 (5730 years is the true half-life of 14C), = Mineral grains larger than 20 µm were reintroduced into the and x year of deposition. The year of sediment deposi- = mill and the process was repeated. The milled samples were tion is estimated from the age model of Ponton et al. (2012) dried overnight at 65 C. The dried sample was pulverized (Table S2). Henceforth, all bulk 14C values for the offshore and homogenized using a Fritsch Pulverisette 23 milling de- sediment core refer to the 1 value. However, it should be vice. The resulting sample was then gently loaded onto a noted that the influence of “bomb 14C”, resulting from above- sample holder and packed continuously using razor blades to ground nuclear weapons testing in the mid-20th century, on form a randomly oriented powdered specimen with a smooth modern-day Godavari Basin 14C values is not accounted for surface which minimizes preferential orientation (Zhang et in this calculation al., 2003). A second sample preparation was carried out, producing textured specimens for enhancement of the basal 2.3.5 Compound-specific stable carbon isotopic reflections of layered silicates and thereby facilitating their analysis identification. The changes in the basal spacing in the XRD pattern by intercalation of organic compounds (e.g. ethylene Freeze-dried and homogenized sediment samples (30– glycol) and after heating (2 h, 550 C) were used for iden- 100 g) were microwave-extracted with dichloromethane tification of smectite and kaolinite, respectively. XRD mea- (DCM) methanol (MeOH) (9 1 v/v) for 25 min at 100 C : : surements were performed on a Bruker AXS D8 Theta-Theta (MARS, CEM Corporation). The 20 selected milled sam-

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

18

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3363 ples were extracted, using an accelerated solvent extractor Illite + chlorite (%) (ASE 350, Dionex, Thermo-Scientific), with DCM MeOH 100 : (9 1 v/v) at 100 C and 7.6 MPa. The total lipid extracts : (TLEs) were dried under N2 and then saponified with 0.5 M 80 20 potassium hydroxide (KOH) in MeOH (70 C for 2 h). A “neutral” fraction was obtained by back-extraction with hex- ane after the addition of Milli-Q water with sodium chlo- 60 40 ride (NaCl) to aid separation. The “acid” fraction was ob- tained by back-extraction of the hydrolysed solution with hexane DCM (4 1 v/v) after adjusting the pH to 2. 40 60 : :  The acid fraction was transesterified with MeOH HCl (hy- Late Holocene (LH) : Early Holocene (EH) drochloric acid) (95 5 v/v) of known isotopic composition Riverbank sediment (UB Rba) : 20 Riverbed sediment (UB Rbe) 80 at 70 C for 12–16 h in order to yield corresponding fatty Surface soil (UB SSo) acid methyl esters (FAMEs). The resulting FAMEs were then Deep soil (UB DSo) purified using silica gel-impregnated silver nitrate (AgNO3– 100 100 80 60 40 20 SiO2) column chromatography to remove unsaturated homo- Smectite (%) Kaolinite (%) logues. Aliquots of the FAMEs obtained from each sample Phyllosilicate mineral composition of the Godavari River were measured in duplicate by gas chromatography–isotope Figure 2. basin and offshore sediments. ratio mass spectrometry (GC-IRMS) using an HP 6890 GC coupled with a Thermo-Delta V IRMS system. The 13C val- ues of fatty acids (FAs) were subsequently corrected for the Lipid analyses from river sediments and soils produced contribution of the added methyl carbon and respective errors LCFA with an average chain length (ACL) consistently > 28 were propagated (Tao et al., 2015). The average uncertainty and similar stable carbon isotope values among C26–C32 FA is 0.3 ‰ for the FAs. Results are reported relative to Vienna homologues (Fig. S1). Thus, isotopic values are reported as Pee Dee Belemnite (VPDB) (Craig, 1953). mean weighted averages of C26–C32 FA (Table S1). LCFAs 1 of soils range from 4 to 264 µg g OC with extremely low concentrations in the surface soils of the lower basin 3 Results (mean 10 3 µg g 1 OC, n 8). SA-normalized C –C = ± = 26 32 3.1 Surface and deep soils FA concentrations (FA loadings) of surface soils range from 0.1 to 4.6 µg LCFA m 2 (mean 2.0 and 0.3 µg LCFA m 2 = Both surface and deep soils from the upper basin are for upper-basin and lower-basin soils, respectively) and highly enriched in smectite (30–50 % of total minerals) with decrease progressively towards the estuary. The average 13 lesser abundances of kaolinite and illite chlorite (Fig. 2). COC value of upper-basin soils ( 17.9 3.1 ‰; n 51) + ± = On the other hand, soils from the lower basin are mostly contrasts sharply with that of soils from the lower basin quartzo-feldspathic (25–40 % of total minerals) with mi- ( 23.2 2.0 ‰; n 16, Fig. 3c). A similar 5 ‰ differ- ± = 13 ⇠ nor amounts of kaolinite (Fig. 2; see also Kulkarni et al., ence was observed in corresponding CLCFA values, which 2015; Subramanian, 1981). Total organic carbon (TOC) con- average 24.1 ‰ ( 0.3 ‰, n 39) in the upper basin and ± = tents of Godavari River basin soils range from 0.1 to 1.8 % 30.6 ‰ ( 0.3 ‰ , n 8) in the lower basin. ± = (mean 0.6 0.4 %, n 67; Table S1). The highest TOC The soil depth profiles generally show a decrease in TOC = ± = values were found for surface soils close to the headwaters of contents from top to bottom, accompanied with relatively in- 13 the river (Fig. 3a). The highest and lowest values for median variant (upper basin) or increasing (lower basin) CLCFA 14 grain size (GS) (970 and 5.9 µm, respectively) are recorded in values (Fig. S2). Corresponding 1 COC values of soils surface soils from the upper part of the basin (Table S1). High range from 337 to 132 ‰, with the most depleted values + mineral surface area (MSA) values are common in upper- recorded in deeper soil horizons (Fig. 3d). basin soils (mean 42 18 m2 g 1, n 51), with lower val- = ± = ues in those from the lower basin (mean 21 11 m2 g 1, 3.2 Riverbed and riverbank sediments = ± n 16; Fig. 3b). MSA-normalized OC (a term that expresses = OC loading on mineral surfaces) values of soils range from The median GS of riverbed and riverbank sediments var- 0.03 to 0.84 mg OC m 2 (mean 0.20 0.15 mg OC m 2). ied between 8 and 851 µm (Table S1). Generally, the upper = ± Due to the relatively low TOC values and high MSA val- basin is characterized by fine-grained sediments (9–50 µm; mean 23 11 µm, n 12) and the lower basin by coarse- ues, the majority of the soils plot outside the range of = ± = grained sediments (136–852 µm; mean 456 288 µm, typical river-suspended sediments as defined by Blair and = ± n 6) (Table S1), except in the delta where finer-grained ma- Aller (2012). = terial (8–116 µm, mean 51 57 µm, n 3) again predom- = ± = inates. Similarly, MSA values show consistently high val- www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

19

3364 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system

2.0 UB LB OS UB LB OS (a) (b) 80 1.6

-1 60

1.2 2

LH EH 40 OC (%) OC 0.8 UB Rba MSA (mMSA g ) UB Rbe UB SSo 0.4 UB DSo 20 LB Rbe LB SSo 0 0 1600 1200 800 400 0 -400 1600 1200 800 400 0 -400 Distance to/from coast (km) Distance to/from coast (km)

-12 200 UB LB (c) UB LB OS (d) OS

-16 0 (‰) (‰) OC OC -20 C C 14 13 ∆ δ -200 -24

-28 -400 1600 1200 800 400 0 -400 1600 1200 800 400 0 -400 Distance to/from coast (km) Distance to/from coast (km)

2 1 13 Figure 3. (a) Total organic carbon (OC; %). (b) Mineral surface area (MSA) (m g ). (c) Bulk OC stable carbon isotope COC (‰). 14 (d) Bulk OC radiocarbon contents (1 COC) (‰) in the Godavari basin and adjacent margin. Symbols key and abbreviations are as for Figs. 1 and 2. Average values of early (EH) and late Holocene (LH) sediment samples are shown. ues (19–60 m2 g 1; mean 39 11 m2 g 1, n 12) in the comprises illite and chlorite (25 and 27 % for riverbed and = ± = upper basin, markedly lower values in the lower basin (2– riverbank, respectively; Fig. 3). 2 1 2 1 14 m g ; mean 6 4m g , n 6), and intermediate Concentrations of long-chain (C26 32) FAs in riverine = ± = values in the delta (12–37 m2 g 1; mean 28 13 m2 g 1, sediments range from 7 to 187 µg g 1 OC, with the low- = ± n 3; Table S1, Fig. 3b). There is a weak positive lin- est concentrations (7–16 µg g 1 OC) found in the lower- = ear correlation between MSA and GS (r2 0.33 and 0.36 basin riverbed sediments, and the highest concentrations = 1 for riverbank and riverbed sediment, respectively). Samples (67–187 µg g OC) in upper-basin riverbank sediments (Ta- with lower GS and higher MSA generally have higher TOC ble S1). The lower-basin riverbed sediments have very low 1 contents (0.3–1.6 %; Fig. 3a and b). Conversely, sediments C26 32 FA concentrations (mean 9 4 µg g OC, n 4), = ± = with coarser GS and lower MSA have low TOC contents while those of upper-basin riverbed sediments are below de- 13 (0.1–0.4 %). OC loading values, which range from 0.09 to tection. Bulk COC values for all river sediments range 0.80 mg OC m 2 (mean 0.29 0.18 mg OC m 2) are gen- from 17.3 to 25.2 ‰. The most enriched 13C value = ± OC erally low compared to typical river sediments (Fig. 4; Blair was observed in the upper basin (Fig. 3c), where samples and Aller, 2012; Freymond et al., 2018). The relative phyl- yielded an average value of 20.6 ‰ ( 0.3 ‰, n 12). 13 ± = losilicate mineral content of the two upper-basin sediment The most depleted COC value was recorded in the samples that were analysed shows that smectite predomi- lower basin, where fluvial sediments averaged 24.1 ‰ nates (65 and 72 % for riverbed and riverbank, respectively) ( 0.3 ‰, n 9). Compound-specific 13C analysis of the ± 13 = while the kaolinite content is higher in the riverbed (10 % LCFAs ( CLCFA) of river sediments yielded values be- vs. 1 % for riverbank). The remainder of the phyllosilicates tween 24.8 and 32.8 ‰. Similar to 13C , the most OC enriched 13C value ( 24.8 ‰, mean 27.4 0.3 ‰, LCFA = ±

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

20

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3365

3.0 (a) LH Typical river- EH 1.0 suspended sediments UB Rba

2.5 /SA = UB Rbe High productivity org Non-deltaic shelf UB SSo upwelling regions C UB DSo LB Rbe 2.0 /SA = 0.4 LB SSo C org

1.5 OC (%)

1.0 Energetic deltaic sediments Deep sea

0.5

0.0 20 40 60 80 Mineral surface area (m2 g -1 ) Sand

Clay

3.0 (b) Typical river- 1.0 suspended sediments

2.5 High productivity /SA =

org Nondeltaic shelf upwelling regions C

2.0 /SA = 0.4 C org

1.5 OC (%)

1.0 Energetic deltaic sediments Deep sea

0.5

0.0 20 40 60 80 Mineral surface area (m2 g -1 ) Sand

Clay

Figure 4. (a) Organic carbon loadings for river sediments and marine sediments. (b) OC vs. MSA for soils samples within the basin. Symbol key and abbreviations are as for Figs. 1 to 3. Blue shaded area corresponds to range typical for river-suspended and non-deltaic sediments as described by Blair and Aller (2012).

www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

21

3366 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system n 6) was recorded in the upstream section and the most de- 4 Discussion = pleted value ( 32.8 ‰, mean 31.3 0.3 ‰, n 4) was = ± 14 = observed in the downstream segment. The 1 COC values 4.1 Evolution of organic matter–mineral associations of river sediments vary between 151 and 97 ‰ (Fig. 3d) in the Godavari River basin with no clear systematic difference between the upstream and downstream sections of the basin. Soil and sediment samples analysed from the Godavari River and its major tributaries reveal a wide range of grain sizes, 3.3 River-proximal marine sediments mineral surface areas, and TOC contents and compositions. This diversity in characteristics encompasses the range of The TOC content of sediments from core 16A varies values reported in previous studies of river sediments and from 1.2 to 2.1 %, and generally decreases from the bot- soils within the Godavari catchment (e.g. Balakrishna and tom to the top of the core, resulting in mean values Probst, 2005; Pradhan et al., 2014; Kulkarni et al., 2015; of 1.9 0.1 % (n 10) and 1.6 0.3 % (n 37) for the ± = ± = Cui et al., 2017). The average TOC content of the upper- early and late Holocene, respectively (Table S2, Fig. 3a). basin riverbed sediments is a factor of 2 higher than that The GS and MSA are both fairly uniform, with val- of lower-basin sediments (0.9 0.5 % vs. 0.4 0.4 %, Ta- ues ranging from 4.0 to 5.2 µm (mean 4.5 0.3 µm) ± ± = ± ble S1), and this distribution reflects the geochemical and and 54 to 72 m2 g 1 (mean 63 4m2 g 1), respec- = ± sedimentological characteristics of the basin. The relatively tively. OC loadings decrease progressively from the 2 high TOC values in the upper basin are likely due to low sus- early (mean 0.31 0.03 mg OC m ) to late Holocene = ± 2 pended sediment loads and/or greater proportions of organic (mean 0.25 0.04 mg OC m ). The range OC loading = ± 2 debris (Ertel and Hedges, 1985). The modern-day upper Go- values of the core sediments (0.19–0.35 mg OC m ) are davari is characterized by low suspended sediment load and within the range of values expected for deltaic and deep- relatively high phytoplankton production, resulting in rela- sea sediments (Fig. 4; Blair and Aller, 2012). Relative abun- tively high OC contents in riverbed sediments (Pradhan et dances of phyllosilicate minerals of the analysed core sed- al., 2014). In contrast, the lower catchment is more heav- iments show that early Holocene sediments have slightly ily charged with suspended sediments primarily derived from higher kaolinite and illite chlorite contents than the late + the Pranhita and Indravati rivers draining the Eastern Ghats Holocene sediments, whereas the smectite contents of late (Balakrishna and Probst, 2005), with dilution by lithogenic Holocene sediments are slightly higher (Fig. 2). materials resulting in lower observed TOC values in sedi- Concentrations of LCFA in the core vary between 49 ments from the lower Godavari basin. The lithological con- and 519 µg g 1 OC (mean 181 95 µg g 1 OC), but = ± trast between the upstream and downstream part of the basin remain relatively invariant despite the down-core variations may also play an important role in OM distribution between in TOC (Table S2). LCFA loading ranges from 0.82 to 2 both parts of the basin (see Fig. 1). Erosion of the basalt 4.93 µg LCFA m , with slightly higher loadings in the in the upper basin produces high-MSA, smectite-rich clay late Holocene (mean 2.73 µg LCFA m 2) than the early = mineral assemblages, whereas the erosion of granitic rocks Holocene (mean 2.15 µg LCFA m 2). These are similar to = outcropping in the lower basin yields lower-MSA, kaolinite- the range of LCFA loading values observed in the Danube rich assemblages. This lithological contrast likely accounts Basin (Freymond et al., 2018). Stable carbon isotopic for the spatial offset in the MSA between the upstream and compositions of bulk OC (13C ) range from 19.9 to OC downstream Godavari (Fig. 2b). The higher TOC values are 18.2 ‰ (mean 18.8 0.5 ‰, n 35) and 20.8 to = ± = likely a result of availability of a large mineral surface that 19.8 ‰ (mean 20.3 0.5 ‰, n 8) for the late and = ± = provides substrate for OM sorption, stabilization, and pro- early Holocene, respectively; 13C range from 26.75 LCFA tection (Keil et al., 1997; Arnarson and Keil, 2007; Gordon to 23.43 ‰ (mean 24.89 1.16 ‰, n 34) and = ± = and Goni, 2004; Mayer, 1994b). Organic matter first devel- 28.90 to 26.84 ‰ (mean 28.01 0.48 ‰, n 11) = ± = ops associations with minerals during soil formation (Mayer, for the late and early Holocene, respectively. There is a 13 13 1994a), and these organo-mineral associations that evolve gradual increase in both the COC and CFA values during soil mobilization and erosion are considered to in- towards the top of the core. The 1 values of the measured fluence the balance of preservation and oxidation (Marin- samples (corrected for decay since deposition) vary between Spiotta et al., 2014; Wang et al., 2014). 194.6 and 52.1 ‰, and generally increase with increasing Plotting mineral-surface-area normalized OC (OC / MSA) depth (Table S2). 13 vs. COC for the river basin sediments and soils reveals marked differences between upper- and lower-basin signals 13 (Fig. 5a). Relatively low OC loading and higher COC values characterize the upper basin whereas high OC load- 13 ings and lower COC values typify the lower basin. This likely reflects the spatial contrast in vegetation and sedimen- 13 tology or mineralogy within the river basin. Higher COC

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

22

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3367

1.0 1.0 (a) (b)

0.8 0.8 LH -2 EH -2 UB Rba UB Rbe 0.6 UB SSo 0.6 UB DSo LB Rbe LB SSo 0.4 0.4 OC/MSA (mg OC m ) OC (mg OC/MSA m OC ) (mg OC/MSA

0.2 0.2

0 0 -28 -24 -20 -16 -12 -400 -200 0 200

13 14 δ COC (‰) ∆ COC (‰)

13 14 14 Figure 5. Organic carbon loading vs. (a) COC and (b) 1 COC, for Godavari river basin and marine sediment core 16A. The 1 C values for marine sediments refer to the age-corrected value (1). Symbol key and abbreviations are as for Figs. 1 to 3. values in the upper basin reflect a greater preponderance the lower-basin sediments and soil plot within the vicinity of of C4 vegetation in the upper basin while lower OC load- the C3-plant end-member. This implies that OC in the upper- ings are attributed to a wide range of factors, including ero- basin sediments mostly derive from C4-plant-derived soil sion of heavily weathered soils that are relatively depleted OM with a minor C3 plant contribution, as evidenced by the in OC and notably enriched in high-surface area smectite- clustering of sediments around the C4 end of the soil domain. rich secondary minerals due to erosion of basalts of the Dec- In the same vein, lower-basin samples point to increased con- can Plateau (Table S1). This interpretation is consistent with tribution of C3-plant-derived terrestrial OM (soil). other independent observations within the Godavari Basin The spatial decoupling of upper and lower-basin geochem- and its adjacent margin (Kessarkar et al., 2003; Philips et ical signatures of river sediments has been largely attributed al., 2014; Shrivastava and Pattanayak, 2002; Srivastava et to the vegetation gradients in the basin. However, the appar- al., 1998). Assessment of relationships between OC load- ent lack of upper-basin signatures in fluvial sediment from 14 ing and 1 COC show that samples with higher OC load- the lower reaches could also be a consequence of in-river ings are generally more enriched in 14C (Fig. 5b). In contrast processes such as loss or replacement of OC and/or sedi- to bulk OC loadings, LCFA loadings are generally higher in ment dilution. The general increase in 114C values from up- the upper basin (Fig. 6a). These low OC–high LCFA load- per to lower basin (Fig. 8) indicates that preferential loss ings in the upper basin suggest that a large proportion of the of a younger, more reactive fraction is unlikely. Modern OC stabilized onto mineral surfaces derives from terrestrial sediment and OC flux data show the highest POC yield 13 2 1 plants, even at low OC contents. Furthermore, like COC ( 12 t km yr ) in the Indravati and Pranhita rivers mostly 13 ⇠ values, CLCFA values are relatively high in the upper basin as a consequence of high runoff that carries large amount of (Fig. 6b), indicating a predominant C4 plant origin. (younger) plant detritus and loose (top) soil from the forest Coupled plots of 13C vs. 114C have been widely used to the mainstream (Balakrishna and Probst, 2005). Presently, 2 1 to elucidate potential sources of OC in riverine systems, and more than 470 km year are lost in the lower basin due to to delineate various end-member contributions to OC (e.g. deforestation and forest fire, with maximum forest denuda- Goni et al., 2005; Marwick et al., 2015). The Godavari basin tion taking place in the state of Orissa (Silviera, 1993), which 13 samples exhibit a broad range of COC values, indicative is drained by the Indravati River. These processes may desta- of mixed vegetation signatures of savanna, tropical grass- bilize soils and enhance the loss of associated OM to the flu- lands, and tropical forests, as well as aquatic productivity vial network. In contrast, the general decrease in TOC con- 13 13 and bedrock inputs, with higher COC and CLCFA val- tents towards the lower basin (Fig. 3a) and the downstream ues of upper-basin sediments and soils reflecting the greater increase in SPM (Gupta et al., 1997) points towards dilution 13 proportion of C4 (vs. C3) vegetation. When plotted in C of riverine OC with mineral matter derived from soil ero- vs. 114C space (Fig. 7), the majority of the upper-basin sedi- sion in the lower basin. As a result, the OC signatures in ments and soils plot within the “soils” end-member and gen- the modern-day Godavari river sediments appear to not only erally cluster around the C4-plant domain, whereas most of reflect the biogeographic and geochemical make-up of the www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

23

3368 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system

6 (a) (b) -20 ) -2

UB Rba 4 UB SSo UB DSo -24 LB Rbe (‰) LB SSo LCFA C 13

δ -28 2 LCFA loadings (μg C m C (μg loadings LCFA

-32

0 16001200800 400 0 16001200 800 400 0 Distance to coast (km) Distance to coast (km)

13 Figure 6. (a) LCFA loading vs. distance to coast. (b) CLCFA vs. distance to coast for river basin samples. Symbol key and abbreviations are as for Figs. 1 to 3.

400

Modern C3 plants Modern C4 plants 200

0

-200 Phytoplankton (‰) OC C 14

∆ -400 Soil

-600 LH EH UB Rba UB Rbe -800 UB SSo UB DSo LB Rbe LB SSo -1000 Kerogen

-35 -30 -25 -20 -15 -10

13 δ COC (‰)

Figure 7. Identifying major sources of organic carbon to the Godavari River and the offshore sediment core using stable (13C) and radio (114C) isotopes (Modified after Marwick et al., 2015). The soil end-member is defined based on ranges of values observed within the Godavari and other tropical river systems (e.g. Pessenda et al., 1997; Shen et al., 2001; Trumbore et al., 1989). Symbol key and abbreviations are as for Figs. 1 to 3.

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

24

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3369

(a) 0.8 -2

0.6

( =27) 0.4 (n=28) n

(n=11) 0.2 (n=36) (n=12) (n=3) (n=9) (n=8) OC loadings (mg OC m (mg loadings OC ) 80 (b) -1

2 60 (n=12) (n=3) (n=11) (n=36) (n=9) 40 MSA (mMSA g ) 20 (n=9) (n=28) (n=27)

(c) -14 -16 -18 (‰)

-20 OC (n=12) ( =9) n (n=35) C (n=8) -22 13 (n=27) δ (n=3) -24 (n=9) (n=28) -26 -2 5 (d)(n=16)

4 (n=27) 3 (n=4) 2

1 (n=8) (n=7) ( =32) (n=11) n LCFA loadings (μg C loadings m LCFA )

(e) -20 -22

-24 (‰)

-26 LCFA C

( =7) (n=34) -28 13 n δ (n=4) (n=11) -30 (n=25) (n=6) (n=15) -32 100 (f)

0 (‰)

OC -100 (n=11) C

14 (n=3) (n=9) (n=12) (n=9) ∆ -200 ( =24) n (n=36) -300 (n=31)

UB SSo UB DSo UB Rbe UB Rba LB SSo LB Rbe OS EH OS LH

Figure 8. Box-and-whisker summary of the geochemical and sedimentological data for the Godavari River basin and offshore sediments. (a) Organic carbon loadings, (b) mineral surface area (MSA), (c) bulk stable carbon isotope ratio, (d) long-chain fatty acid loadings, (e) mean weighted stable carbon isotope ratio of long-chain (C26 32) fatty acid, and (f) bulk radiocarbon signature. The box represents the first (Q1) and third quartiles (Q3), and the line in the box indicates the median value. The whiskers extend to 1.5 (Q3–Q1) values, and outliers are · shown as points. SSo surface soil; DSo deep soil; Rbe riverbed; Rba riverbank; EH early Holocene; LH late Holocene. ======www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

25

3370 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system basin, but also the processes (loss and replacement vs. sedi- position of biogenic carbonate and opal (Hobert and Wet- ment dilution) that influence the nature of OC. zel, 1989). However, sediment trap data from the central BoB suggest that modern-day carbonate and opal fluxes to 2 4.2 Linkages between Godavari drainage basin and BoB are relatively low (0.03–3.1 g m per year; Sarin et al., marine sedimentary signals 1979). In addition, low foraminifera abundances and high sedimentation rates supported by detrital sediment inputs The Holocene record from core 16A (Figs. 5, 7, 8; Table S2) (Giosan et al, 2017), especially during the late Holocene, shows that increasing long-chain plant wax 13C values from minimize the effect of carbonate and opal influences on MSA the early to late Holocene coincide with other lines of evi- measurements at this location. Consequently, the measured dence indicating a transition to drier conditions on the Indian MSA was interpreted as exclusively reflecting fluvially de- and Arabian peninsulas (Ponton et al., 2012; Prasad et al., rived lithogenic materials. In this context, we do not find 2014). Because C4 vegetation is adapted to more arid condi- any systematic difference in MSA between early and late tions, the marked isotopic change beginning at 4.5 kyr BP, Holocene sediments, with OC loadings that plot within the ⇠ accompanied by a shift in neodymium isotopic composition general range that is characteristic of deltaic and deep-sea towards Deccan bedrock signatures (Tripathy et al., 2011) in sediments (Fig. 4; Blair and Aller, 2012). 13 14 detrital phases, has been interpreted to reflect a shift in sed- Bulk OC loading vs. COC and 1 C show that at iment provenance associated with changes in basin hydrol- higher loading, OC is relatively 13C-depleted and enriched ogy, resulting in increased sediment flux from the upper Go- in 14C, whereas the reverse is the case at lower loading davari catchment to the adjacent continental margin (Giosan (Fig. 5). Direct comparisons of bulk OC loadings between et al, 2017). marine sediment core and river basin soils and sediments In contrast to the river basin sediments and soils, the uni- are not straightforward, as the likely addition of marine car- form distribution of grain size and mineral surface area in bon to offshore sediments introduces a layer of complexity to receiving basin sediments is likely a result of hydrodynamic such comparison. In contrast, LCFAs derive exclusively from sorting during fluvial transport and export of sediments to the higher terrestrial plants, enabling more direct comparison of BoB. Thus, in order to compare and contrast signals ema- loadings between riverine and offshore sediment. Adopting nating from the Godavari drainage basin with those in sed- the biomarker loadings concept described by Freymond et iments deposited on the adjacent continental margin, it is al. (2018), we find elevated LCFA loading in the upper basin important to take into account processes that may induce compared to the lower basin (Figs. 6a and 8), and a similar particle mobilization, transformation, and sorting. Normal- range of LCFA loadings in sediments deposited during the ization to MSA may provide a means to address this prob- early and late Holocene to that observed in soils and sed- lem, as it eliminates hydrodynamic sorting effects due to iments of the lower and upper basin, respectively. This sug- GS, particle density, and shape (Freymond et al., 2018). gests that the loading signatures in early vs. late Holocene are Marked differences between early vs. late Holocene offshore likely a consequence of the changes in sediment provenance sediments that mimic upper vs. lower-basin signals, respec- previously inferred from neodymium isotopic data (Giosan tively, are evident when MSA normalized OC (OC / MSA) et al., 2017). 13 is plotted vs. COC (Figs. 5a and 8). The mean OC load- The progressive increase in stable carbon isotopic val- ings of early (0.33 0.03 mg OC m 2) and late Holocene ues of bulk (13C ) and long-chain fatty acids (13C ) ± OC LCFA (0.25 0.04 mg OC m 2) sediments are similar to mean from the marine sediment core towards the late Holocene has ± loading values observed in lower- (0.31 0.12 mg OC m 2) been interpreted as an enhanced supply of C -derived OC ± 4 and upper-basin (0.24 0.15 mg OC m 2) riverbed sedi- sourced from the Deccan Plateau during the late Holocene ± ments, respectively. In addition, these values are similar to triggered by changes in Indian monsoon strength and/or lo- the OC loadings of soils from the respective source regions in cation (Gadgil et al., 2003; Sinha et al., 2011; Webster et al., the basin (0.29 0.14 and 0.23 0.17 mg OC m 2 for lower 1998). Our new results from within the drainage basin lend ± ± and upper-basin soils, respectively). The early Holocene part support for a significant reorganization in sediment and OC of the record is characterized by relatively high OC load- provenance from lower to upper-basin sources. 13 14 ing and lower COC values that progressively shift towards The decay-corrected C values of the sediment core are 13 lower OC loading and relatively higher COC values during bracketed by the range of values of surface and deep soils the latter part of Holocene (Fig. 5). from the upper and lower basin (Figs. 7 and 8). This suggests There have only been limited investigations on the lon- that the core consists of a mixture of pre-aged carbon sourced gitudinal evolution of OM–mineral interactions during tran- from deep soils and fresh carbon from plant litters and pos- sit through river basins (Freymond et al., 2018). However, sibly freshwater algae. It should, however, be noted that soil evidence suggests that loss and replacement of OM may be samples, including deeper soil layers, have likely been im- substantial within floodplains, estuarine, and deltaic systems pacted by “bomb 14C” (see Trumbore et al., 1989; van der (Galy et al., 2008; Keil et al., 1997). Estimates of MSA in Voort et al., 2017). There is a general decrease in the 1 val- marine sediments are complicated by the production and de- ues towards the late Holocene, and the ranges of 1 values of

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

26

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3371 the late Holocene sediment are only observed in the deeper 5 Conclusions sections of upper-basin soils (Fig. S2). Taken together, these sedimentological and geochemical In this study, we sought to reconcile previously observed geo- results suggest that export of OC-poor C4-dominated and chemical variations in the Holocene sediments deposited in smectite-rich mineral soils intensified in the late Holocene, the BoB offshore of the Godavari River with those observed with the observed shift ca. 4.5 kyr BP reflecting a shift in sed- in soils and sediments within the modern drainage basin. iment provenance from lower basin to upper basin. Further- Distinct contrasts were observed in the abundance and more, much of these upper-basin sediments were likely de- characteristics of OM and mineral components of soils and rived from deeper, older, more degraded Deccan soils. This fluvial sediments in the upper and the lower basin. The for- apparent shift in the loci and nature of soil mobilization mer (upper basin) are characterized by C4-dominated OM is also accompanied by a 3-fold increase in sediment flux associated with high-surface-area Deccan-sourced mineral (Pradhan et al., 2014), implying extensive soil loss from the phases, whereas those of the lower basin contain higher pro- upper catchment (e.g. Van Oost et al., 2012). This loss may portions of C3-plant-derived OM. have stemmed from both natural (aridification and associated The strong links between OM characteristics and sediment reduction in vegetation cover) and anthropogenic (agriculture mineralogy (GS, MSA) suggest that OM–mineral interac- and irrigation) causes, the latter potentially being triggered tions play an important role in OC stabilization throughout by changes in regional climate. the Godavari source-to-sink system, from mobilization to ex- For the period spanning the late Holocene, perturbations port. within river basins due to natural climate variability have Comparison of bulk and molecular-level characteristics of become intertwined with those stemming from human ac- drainage basin and marine sediment core show that a marked tivity. This is particularly so for subtropical river basins of mid-Holocene transition is consistent with a change in sed- central Asia, where the influence of anthropogenic activity iment provenance towards a greater contribution of Deccan- on the landscape and watersheds extends back several mil- sourced material in the upper basin, although extensive an- lennia (e.g. Van Oost et al., 2012). Within the past 2 cen- thropogenic perturbation of the modern Godavari system turies, humans have imparted particularly dramatic changes limits the effective transmission of the upper signal to the on drainage basins both in terms of land use (e.g. deforesta- deltaic region and offshore. However, given the limited ac- tion, agricultural practices) and modification of water net- commodation space that restricts upstream trapping and pro- works through dam construction and other major perturba- motes rapid export, anthropogenic influences on the flux and tions (Syvitski et al., 2005). nature of OC exported from the Godavari basin may be sub- Both the landscape and hydrological characteristics of the ject to marked future changes. Godavari basin have been dramatically altered over the past Our findings suggest that reconstruction of past continen- century. For example, in the past few decades, there has been tal conditions based on terrestrial biomarker proxy records a 10-fold decrease in OC flux from the Godavari to the BoB in marine sediments need to consider potential shifts in sig- due to reduced monsoon rainfall and to dam constructions nal provenance as a consequence of both natural and anthro- (Gupta et al., 1997; Pradhan et al., 2014). However, the late pogenic forcing. Holocene section generally mimics modern-day upper-basin signatures in high fidelity, suggesting that the perturbations of the modern Godavari had little impact on sediment and Data availability. All underlying research data associated with this OC mobilization. study are available in the Supplement. The general agreement between signals emanating from the river basin and those recorded in the sedimentary archive provides valuable insights into understanding the major The Supplement related to this article is available online mechanisms of sediment and OC mobilization, the dynamics at https://doi.org/10.5194/bg-15-3357-2018-supplement. and interactions of organic matter and sedimentary minerals during fluvial transport, and their impact on the provenance and nature of signals exported from the drainage basin. Fur- thermore, such studies that seek to reconcile drainage and Competing interests. The authors declare that they have no conflict receiving basin characteristics, as well as climate and an- of interest. thropogenic influences on these connections, are necessary to determine the factor(s) controlling the nature and fate of OC preserved in sedimentary archives. Acknowledgements. We thank the associate editor Markus Kienast and two anonymous reviewers for their comments. This project was supported by the Swiss National Science Foundations (“CAPS LOCK” grant no. 200021-140850 and “CAPS-LOCK2” grant no. 200021-163162). Francien Peterse received funding from www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

27

3372 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system

NWO-Veni grant (grant no. 863.13.016). Liviu Giosan thanks Biksham, G. and Subramanian, V.: Sediment transport of the Go- colleagues and crew from the NGHP-01 expedition for intellectual davari River Basin and its controlling factors, J. Hydrol., 101, interactions leading to pursuing work on fluvial–continental margin 275–290, https://doi.org/10.1016/0022-1694(88)90040-6, 1988. systems of Peninsular India and to grants from the National Science Bish, D. and Plötze, M.: X-ray powder diffraction with emphasis Foundation (OCE-0841736) and Woods Hole Oceanographic on qualitative and quantitative analysis in industrial mineralogy, Institution. We also wish to thank Daniel Montluçon for laboratory in: Advances in the characterization of industrial minerals, edited assistance, and we acknowledge the logistical support of Prasanta by: Christidis, G., EMU and Mineralogical Society, London, 36– Sanyal and Chris Martes with sampling. Further thanks to Michael 76, 2011. Strupler for help with grain size measurements. This paper benefit- Blair, N. and Aller, R.: The Fate of Terrestrial Organic Carbon ted from discussions with Chantal Freymond. in the Marine Environment, Ann. Rev. Mar. Sci., 4, 401–423, https://doi.org/10.1146/annurev-marine-120709-142717, 2012. Edited by: Markus Kienast Cole, J. J., Prairie, Y. T., Caraco, N. F., McDowell, W. H., Tranvik, Reviewed by: two anonymous referees L. J., Striegel, R. G., Duarte, C. M., Kortelainen, P., Downing, J. A., Middleburg, J. J., and Melack, J.: Plumbing the global carbon cycle: Integrating inland waters into the terrestrial carbon bud- get, Ecosystems, 10, 171–184, https://doi.org/10.1007/s10021- 006-9013-8, 2007. References Coleman, J.: Brahmaputra River: Channel processes and sedimenta- tion, Sediment. Geol., 3, 129–139, https://doi.org/10.1016/0037- Arnarson, T. and Keil, R.: Changes in organic matter-mineral 0738(69)90010-4, 1969. interactions for marine sediments with varying oxygen ex- Collett, T., Boswell, R., Cochran, J., Kumar, P., Lall, M., Mazum- posure times, Geochim. Cosmochim. Ac., 71, 3545–3556, dar, A., Raman, M., Ramprasad, T., Riedel, M., Sain, K., Sathe, https://doi.org/10.1016/j.gca.2007.04.027, 2007. A., Vishwanath, K., and Party, N. E.: Geologic implications of Asouti, E. and Fuller, D.: Trees and Woodlands in South India: Ar- gas hydrates in the offshore of India: Results of the National chaeological Perspectives, Left Coast Press, Walnut Creek, Cali- Gas Hydrate Program Expedition 01, Mar. Pet. Geol, 58, 3–28, fornia, 2008. https://doi.org/10.1016/j.marpetgeo.2014.07.021, 2014. Aufdenkampe, A. K., Mayorga, E., Raymond, P. A., Melack, Craig, H.: The geochemistry of the stable carbon J. M., Doney, S. C., Alin, S. R., Aalto, R. E., and Yoo, isotopes, Geochim. Cosmochim. Ac., 3, 53–92, K.: Riverine coupling of biogeochemical cycles between land, https://doi.org/10.1016/0016-7037(53)90001-5, 1953. oceans, and atmosphere, Front. Ecol. Environ., 9, 53–60, Cui, M., Wang, Z., Nageswara Rao, K., Sangode, S. J., Saito, Y., https://doi.org/10.1890/100014, 2011. Chen, T., Kulkarni, Y. R., Naga Kumar, K. C. V., and Demudu, Balakrishna, K. and Probst, J.: Organic carbon transport and G.: A mid- to late-Holocene record of vegetation decline and ero- C / N ratio variations in a large tropical river: Godavari sion triggered by monsoon weakening and human adaptations in as a case study, India, Biogeochemistry, 73, 457–473, the south-east Indian Peninsula, The Holocene, 27, 1976–1987, https://doi.org/10.1007/s10533-004-0879-2, 2005. https://doi.org/10.1177/0959683617715694, 2017. Bao, R., McIntyre, C., Zhao, M., Zhu, C., Kao, S.-J., and Degens, E., Kempe, S., and Richey, J.: Summary: biogeochemistry Eglinton, T.: Widespread dispersal and aging of organic of major world rivers, in: Biogeochemistry of major world rivers, carbon in shallow marginal seas, Geology, 44, 791–794, edited by: Degens, E., Kempe, S., and Richey, J., Wiley, Chich- https://doi.org/10.1130/G37948.1, 2016. ester, 323–348, 1991. Bendle, J. A., Weijers, J. W. H., Maslin, M. A., Sinninghe-Damsté, Ertel, J. and Hedges, J.: Sources of sedimentary humic substances: J. S., Schouten, S., Hopmans, E. C., Boot, C. S., and Pancost, vascular plant debris, Geochim. Cosmochim. Ac, 49, 2097–2107, R. D.: Major changes in glacial and Holocene terrestrial temper- https://doi.org/10.1016/0016-7037(85)90067-5, 1985. atures and sources of organic carbon recorded in the Amazon Freymond, C. V., Kündig, N., Stark, C., Peterse, F., Buggle, B., fan by tetraether lipids, Geochem. Geophy. Geosy., 11, Q12007, Lupker, M., Plötze, M., Blattmann, T. M., Filip, F., Giosan, L., https://doi.org/10.1029/2010GC003308, 2010. and Eglinton, T.: Evolution of biomolecular loadings along a Bergmann, J. and Kleeberg, R.: Rietveld analysis of dis- major river system, Geochim. Cosmochim. Ac., 223, 389–404, ordered layer silicates, Material Science Forum, 278–281, https://doi.org/10.1016/j.gca.2017.12.010, 2018. 300–305, https://doi.org/10.4028/www.scientific.net/MSF.278- Gadgil, S.: The Indian monsoon and its variabil- 281.300, 1998. ity, Ann. Rev. Earth Planet. Sci., 31, 429–467, Berner, R.: Burial of organic carbon and pyrite sulfur in the modern https://doi.org/10.1146/annurev.earth.31.100901.141251, 2003. ocean: its geochemical and environmental significance, Am. J. Galy, V. and Eglinton, T.: Protracted storage of biospheric car- Sci., 282, 451–473, https://doi.org/10.2475/ajs.282.4.451, 1989. bon in the Ganges-Brahmaputra basin, Nat. Geosci., 4, 843–847, Berner, R.: The long-term carbon cycle, fossil fuels https://doi.org/10.1038/ngeo1293, 2011. and atmospheric composition, Nature, 426, 323–326, Galy, V., France-Lanord, C., and Lartiges, B.: Loading and fate https://doi.org/10.1038/nature02131, 2003. of particulate organic carbon from the Himalaya to the Ganga– Bhattacharyya, T., Pal, D. K., Mandal, C., Chandran, P., Ray, S. K., Brahmaputra delta, Geochim. Cosmochim. Ac., 72, 1767–1787, Sarkar, D., Velmourougane, K., Srivastava, R., Tiwary, P., Nagar, https://doi.org/10.1016/j.gca.2008.01.027, 2008. A. P., and Nimkhedkar, S. S.: Soils of India: historical perspec- Giosan, L., Ponton, C., Usman, M., Blusztajn, J., Fuller, D. tive, classification and recent advances, Curr. Sci., 104, 1308– Q., Galy, V., Haghipour, N., Johnson, J. E., McIntyre, C., 1323, 2013.

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

28

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3373

Wacker, L., and Eglinton, T. I.: Short communication: Mas- Leithold, E., Blair, N., and Perkey, D.: Geomorphologic controls sive erosion in monsoonal central India linked to late Holocene on the age of particulate organic carbon from small mountain- land cover degradation, Earth Surf. Dynam., 5, 781–789, ous and upland rivers, Global Biogeochem. Cy., 20, GB3022, https://doi.org/10.5194/esurf-5-781-2017, 2017. https://doi.org/10.1029/2005GB002677, 2006. Goni, M., Yunker, M., Macdonald, R., and Eglinton, T.: Ludwig, W., Probst, J., and Kempe, S.: Predicting the oceanic input The supply and preservation of ancient and modern of organic carbon by continental erosion, Global Biogeochem. components of organic carbon in the Canadian Beau- Cy., 10, 23–41, https://doi.org/10.1029/95GB02925, 1996. fort Shelf of the Arctic Ocean, Mar. Chem., 93, 53–73, Manmohan, M., Rao, M., Kamaraju, A., and Yalamarty, S.: Origin https://doi.org/10.1016/j.marchem.2004.08.001, 2005. and occurence of Lower Cretaceous high gamma-high resistivity Gordon, E. and Goni, M.: Controls on the distribution and accumu- shale- a key stratigraphic sequence for hydrocarbon exploration lation of terrigenous organic matter in sediments from the Mis- in the Krishna-Godavari basin, A.P., J. Geol. Soc. India, 62, 271– sissippi and Atchafalaya river margin, Mar. Chem., 92, 331–352, 289, 2003. https://doi.org/10.1016/j.marchem.2004.06.035, 2004. Marin-Spiotta, E., Gruley, K. E., Crawford, J., Atkinson, E. E., Gunnell, Y., Anupama, K., and Sultan, B.: Response of the Miesel, J. R., Greene, S., Crardona-Correa, C., and Spencer, South Indian runoff-harvesting civilization to northeast mon- R.: Paradigm shifts in soil organic matter research affect inter- soon rainfall variability during the last 2000 years: instrumen- pretations of aquatic carbon cycling: transcending disciplinary tal records and indirect evidence, The Holocene, 17, 207–215, and ecosystem boundaries, Biogeochemistry, 117, 279–297, https://doi.org/10.1177/0959683607075835, 2007. https://doi.org/10.1007/s10533-013-9949-7, 2014. Gupta, L., Subramanian, V., and Ittekkot, V.: Biogeochem- Marwick, T., Tamooh, F., Teodoru, C., Borges, V., Darcham- istry of particulate organic matter transported by the beau, F., and Bouillon, S.: The age of river-transported carbon: Godavari River, India, Biogeochemistry, 38, 103–128, A global perspective, Global Biogeochem. Cy., 29, 122–137, https://doi.org/10.1023/A:1005732519216, 1997. https://doi.org/10.1002/2014GB004911, 2015. Hedges, J.: Global biogeochemical cycles: progress and prob- Mayer, L.: Relationships between mineral surfaces and organic lems, Mar. Chem., 39, 67–93, https://doi.org/10.1016/0304- carbon concentrations in soils and sediments, Chem. Geol., 4203(92)90096-S, 1992. 114, 347–363, https://doi.org/10.1016/0009-2541(94)90063-9, Hedges, J., Clark, W., Quay, P., Richey, J., Devol, A., and San- 1994a. tos, U.: Composition and fluxes of particulate organic mate- Mayer, L.: Surface area control of organic carbon accumula- rial in the Amazon river, Limnol. Oceanogr., 31, 717–738, tion in continental shelf sediments, Geochim. Cosmochim. Ac., https://doi.org/10.4319/lo.1986.31.4.0717, 1986. 58, 1271–1284, https://doi.org/10.1016/0016-7037(94)90381-6, Hilton, R., Galy, A., Hovius, N., Chen, M.-C., Horng, M.-J., 1994b. and Chen, H.: Tropical-cyclone-driven erosion of the terres- McIntyre, C., Wacker, L., Haghipour, N., Blattmann, T., Fahrni, S., trial biosphere from mountains, Nat. Geosci., 1, 759–762, Usman, M., Eglinton, T., and Synal, H.: Online 13C and 14C gas https://doi.org/10.1038/ngeo333, 2008. measurement by EA-IRMS-AMS at ETH Zuürich, Radiocarbon, Hobert, L. and Wetzel, A.: On the relationship between silica and 59, 893–903, https://doi.org/10.1017/RDC.2016.68, 2016. carbonate diagenesis in dep-sea sediments, Int. J. Earth Sci., 78, Meybeck, M.: Carbon, nitrogen, and phosphorus trans- 765–778, https://doi.org/10.1007/BF01829321, 1989. port by world rivers, Am. J. Sci., 282, 401–450, Kale, V.: Fluvial geomorphology of Indian rivers: https://doi.org/10.2475/ajs.282.4.401, 1982. an overview, Prog. Phys. Geogr., 26, 400–433, Murthy, K., Subramanam, A., Lakshminarayana, S., Chan- https://doi.org/10.1191/0309133302pp343ra, 2002. drasekhar, D., and Rao, T.: Some geodynamics aspect of the Keil, R., Mayer, L., Quay, P., Richey, J., and Hedges, J.: Loss of Krishna-Godavari basin, east coast of India, Cont. Shelf Res., 15, organic matter from riverine particles in deltas, Geochim. Cos- 779–788, https://doi.org/10.1016/0278-4343(94)00044-N, 1995. mochim. Ac., 61, 1507–1511, https://doi.org/10.1016/S0016- Pessenda, L., Gouveia, S., Aravena, R., Gomes, B., Boulet, R., 7037(97)00044-6, 1997. and Ribeiro, S.: 14C Dating and stable carbon isotopes of soil Kessarkar, P. M., Rao, V. P., Ahmad, S. M., and Babu, G. A.: Clay organic matter in forst-savanna boundary areas in the south- minerals and Sr-Nd isotopes of the sediments along the west- ern Brazilian Amazon region, Radiocarbon, 40, 1013–1022, ern margin of India and their implication for sediment prove- https://doi.org/10.1017/S0033822200018981, 1997. nance, Mar. Geol., 202, 55–69, https://doi.org/10.1016/S0025- Philips, S. C., Johnson, J., Underwood, M. B., Guo, J., Giosan, L., 3227(03)00240-8, 2003. and Rose, K.: Long-timescale variation in bulk and clay min- Komada, T., Anderson, M., and Dorfmeier, C.: Carbonate eral composition of Indian continental margin sediment in the removal from coastal sediments for the determination of or- Bay of Bengal, Arabian Sea, and Andaman Sea, Mar. Pet. Geol., ganic carbon and its isotopic signatures, 13C and 114C: 58, 117–138, https://doi.org/10.1016/j.marpetgeo.2014.06.018, comparison of fumigation and direct acidification by hy- 2014. drochloric acid, Limnol. Oceanogr.-Meth., 6, 254–262, Ponton, C., Giosan, L., Eglinton, T., Fuller, D., Johnson, J., Kumar, https://doi.org/10.4319/lom.2008.6.254, 2008. P., and Collett, T.: Holocene aridification of India, Geophys. Res. Kulkarni, Y. R., Sangode, S. J., Blomendal, J., Meshram, D. C., and Lett., 39, L03704, https://doi.org/10.1029/2011GL050722, 2012. Suresh, N.: Mineral Magnetic Characterization of the Godavari Pradhan, U., Wu, Y., Shirodkar, P., Zhang, J., and Zhang, G.: Multi- River and Western Bay of Bengal: Implications to Source to Sink proxy evidence for compositional change of organic matter in the Relations, J. Geol. Soc. India, 85, 71–78, 2015. largest tropical (peninsular) river basin of India, J. Hydrol., 519, 999–10009, https://doi.org/10.1016/j.jhydrol.2014.08.018, 2014.

www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

29

3374 M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system

Prasad, S., Anoop, A., Riedel, N., Srakar, S., Menzel, P., Subramanian, V.: Geomorphology of the Deccan volcanic province, Basavaiah, N., Krishnan, R., Fuller, D., Plessen, B., Gaye, in: Deccan Volcanism and Related Basalt Provinces in Other B., Röhl, U., Wilkes, H., Sachse, D., Sawant, R., Wiesner, Parts of the World, edited by: Rao, S., Hindustan Publisher, New M., and Stebich, M.: Prolonged monsoon droughts and links Delhi, India, p. 467, 1981. to Indo-Pacific warm pool: A Holocene record from Lonar Syvitski, J. P. M., Vörösmarty, C. J., Kettner, A. J., and Lake, central India, Earth Planet. Sci. Lett., 391, 171–182, Green, P.: Impacts of humans on the flux of terrestrial sed- https://doi.org/10.1016/j.epsl.2014.01.043, 2014. iment to the global coastal ocean, Science, 308, 376–380, Rao, K., Sadakata, N., Malini, B., and Takayasu, K.: https://doi.org/10.1126/science.1109454, 2005. Sedimentation processes and asymmetric development Tao, S., Eglinton, T., Montlucon, D., McIntyre, C., and Zhang, of the Godavari delta, Sepm. Spec. P., 89, 433–449, M.: Pre-aged soil organic carbon as a major component https://doi.org/10.2110/pec.05.83.0435, 2005. of the Yellow River suspended load: Regional significance Sarin, M., Borole, D., and Krishnaswami, S.: Geochemistry and and global relevance, Earth Planet. Sci. Lett., 414, 77–86, geochronology of sediments from the Bay of Bengal and the https://doi.org/10.1016/j.epsl.2015.01.004, 2015. equatorial Indian Ocean, P. Natl. Acad. Sci. USA, 88, 131–154, Tripathy, G., Singh, S., Bhushan, R., and Ramaswamy, V.: Sr-Nd 1979. isotope composition of the Bay of Bengal Sediments: Impact of Schefuss, E., Kuhlmann, H., Mollenhauer, G., Prange, M., Climate on Erosion in the Himalaya, Geochem. J., 45, 175–186, and Pätzold, J.: Forcing of wet phases in southeast https://doi.org/10.2343/geochemj.1.0112, 2011. Africa over the past 17,000 years, Nature, 480, 509–512, Trumbore, S., Vogel, J., and Southon, J.: AMS 14C measure- https://doi.org/10.1038/nature10685, 2011. ments of fractionated soil organic matter: an approach to de- Schefuss, E., Eglinton, T., Spencer-Jones, C., Rullkötter, J., De ciphering the soil carbon cycle, Radiocarbon, 31, 644–654, Pol-Hols, R., Talbot, H., Grootes, P., and Schneider, R.: Hy- https://doi.org/10.1017/S0033822200012248, 1989. drologic control of carbon cycling and aged carbon dis- van der Voort, T. S., Zell, C. I., Hagedorn, F., Feng, X., McIn- charge in the Congo River basin, Nat. Geosci., 9, 687–690, tyre, C. P., Haghipour, N., Graf Pannatier, E., and Eglin- https://doi.org/10.1038/ngeo2778, 2016. ton, T. I.: Diverse Soil Carbon Dynamics Expressed at Schlünz, B. and Schneider, R.: Transport of terrestrial or- the Molecular Level, Geophys. Res. Lett., 44, 11,840–850, ganic carbon to the oceans by rivers: re-estimating https://doi.org/10.1002/2017GL076188, 2017. flux-and burial rates, Int. J. Earth Sci., 88, 599–606, Van Oost, K., Verstraeten, G., Doetterl, S., Notebaert, B., https://doi.org/10.1007/s005310050290, 2000. Wiaux, F., Broothaerts, N., and Six, J.: Legacy of human- Shen, C., Yi, W., Sun, Y., Xing, C., Yang, Y., Yuan, C., Li, Z., Peng, induced C erosion and burial on soil–atmosphere C ex- S., An, Z., and Liu, T.: Distribution of 14C and 13C in forest soils change, P. Natl. Acad. Sci. USA, 109, 19492–19497, of the Dinghushan biosphere reserve, Radiocarbon, 43, 671-678, https://doi.org/10.1073/pnas.1211162109, 2012. https://doi.org/10.1017/S0033822200041321, 2001. Wacker, L., Bonani, G., Friedrich, M., Hajdas, I., Kromer, Shrivastava, J. P. and Pattanayak, S. K.: Basalts of the Eastern B., Nemec,ˇ M., Ruff, M., Synal, H.-A., and Vock- Deccan Volcanic Province, India, Gondwana Res., 5, 649–665, enhuber, C.: MICADAS: Routine and High-Precision https://doi.org/10.1016/S1342-937X(05)70636-5, 2002. Radiocarbon Dating, Radiocarbon, 52, 252–262, Silviera, D.: India Book, Classic Publishers, Goa, India, 1993. https://doi.org/10.1017/S0033822200045288, 2010. Sinha, A., Stott, L., Berkelhammer, M., Cheng, H., Edwards, Wakeham, S., Canuel, E., Lerberg, E., Mason, P., Sampere, T., and R. L., Buckley, B., Aldenderfer, M., and Mudelsee, M.: Bianchi, T.: Partitioning of organic matter in continental margin A global context for megadroughts in monsoon Asia dur- sediments among density fractions, Mar. Chem., 115, 211–225, ing the past millenium, Quaternary Sci. Rev.,30, 47–62, https://doi.org/10.1016/j.marchem.2009.08.005, 2009. https://doi.org/10.1016/j.quascirev.2010.10.005, 2011. Wang, X., Cammeraat, E., Cerli, C., and Kalbitz, K.: Soil ag- Spencer, R., Hernes, P., Aufdenkampe, A., Baker, A., Gul- gregation and the stabilization of organic carbon as affected liver, P., Stubbins, A., Aiken, G., Dyda, R., Butler, K., by ersosion and deposition, Soil Biol. Biochem., 72, 55–65, Mwamba, V., Mangangu, A., and Wabakanghanzi, S.: An ini- https://doi.org/10.1016/j.soilbio.2014.01.018, 2014. tial investigation into the organic matter biogeochemistry of Webster, P. J., Magana, V. O., Palmer, T. N., Shukla, J., Tomas, the Congo River, Geochim. Cosmochim. Ac., 84, 614–627, R. A., Yanai, M., and Yasunari, T.: Monsoon: processes, pre- https://doi.org/10.1016/j.gca.2012.01.013, 2012. dictability, and the prospects for prediction, J. Geophys. Res., Sridhar, P. N., Ali, M. M., Vethamony, P., Babu, M. T., Ramana, 103, 14451–14510, https://doi.org/10.1029/97JC02719, 1998. I. V., and Jayakumar, S.: Seasonal Occurrence of Unique Sedi- Weijers, J. W. H., Schefuss, E., Schouten, S., and Sinninghe- ment Plume in the Bay of Bengal, Eos Trans. AGU, 89, 22–23, Damsté, J. S.: Coupled thermal and hydrological evolution of https://doi.org/10.1029/2008EO030002, 2008. tropical Africa over the last deglaciation, Science, 315, 1701– Srivastava, P., Parkash, B., and Pal, D. K.: Clay Minerals in 1704, https://doi.org/10.1126/science.1138131, 2007. Soils as Evidence of Holocene Climatic Change, Central Indo- Wu, Y., Dittmar, T., Ludwichowski, K., Kattner, G., Zhang, J., Zhu, Gangetic Plains, North-Central India, Quatern. Res., 50, 230– Z., and Kock, B.: Tracing suspended organic nitrogen from the 239, https://doi.org/10.1006/qres.1998.1994, 1998. Yangtze River catchment inot the East China Sea, Mar. Chem., Stuiver, M. and Polach, H.: Reporting of C-14 107, 367–377, https://doi.org/10.1016/j.marchem.2007.01.022, data – Discussion, Radiocarbon, 19, 355–363, 2007. https://doi.org/10.1017/S0033822200003672, 1977. Zhang, G., Germaine, J., Martin, R., and Whittle, J.: A simple sample-mounting method for random pow-

Biogeosciences, 15, 3357–3375, 2018 www.biogeosciences.net/15/3357/2018/

30

M. O. Usman et al.: Reconciling drainage and receiving basin signatures of the Godavari River system 3375

der X-ray diffraction, Clay Clay Miner., 51, 218–225, https://doi.org/10.1346/CCMN.2003.0510212, 2003. Zorzi, C., Goni, M., Anupama, K., Prasad, S., Hanquiez, V., John- son, J., and Giosan, L.: Indian monsoon variations during three contrasting climatic periods: The Holocene, Heinrich Stadial 2 and the last interglacial-glacial transition, Quaternary Sci. Rev., 125, 50–60, https://doi.org/10.1016/j.quascirev.2015.06.009, 2015.

www.biogeosciences.net/15/3357/2018/ Biogeosciences, 15, 3357–3375, 2018

31

Chapter 3: Basin-wide climatic and anthropogenic driven mobilization of soil organic carbon to the Bay of Bengal.

Muhammed O. Usman1, Camilo Ponton2, Negar Haghipour1,3, Maarten Lupker1, Liviu Giosan4, Timothy I. Eglinton1.

1Geological Institute, ETH Zürich, Sonneggstrasse 5, 8092 Zürich, Switzerland 2Division of Geological and Planetary Science, California Institute of Technology, 1200 East California Boulevard, Pasadena, 91125 California, USA 3Laboratory of Ion Beam Physics, ETH Zürich, Otto-Stern-Weg 5, 8093 Zürich, Switzerland 4Geology and Geophysics Department, Woods Hole Oceanographic Institution, 86 Water Street, Woods Hole, 02543 Massachusetts, USA

32

Abstract

Riverine transport of organic carbon (OC) from the terrestrial to the oceanic realm plays a crucial role in the global carbon cycle. Terrestrial organic matter (OM) temporarily stored and processed within intermediate reservoirs such as soils ultimately impacts the nature and composition of OC delivered to the ocean. Bulk and molecular level isotopic composition of OM can be used to constrain sources and mean terrestrial residence time of sedimentary OC. Here, we use paired radiocarbon measurements on terrestrial high plant-derived long-chain fatty acids (LCFA), total organic carbon (TOC) and coeval foraminifera, from a sediment core near the mouth of the Godavari River to assess the temporal evolution in OC exported from the drainage basin during the Holocene. A coupled carbon isotopic mixing model is used to evaluate the fractional contribution of the main OC pools (i.e. terrestrial C3, C4 plants, and marine OC). The results show an increased predominance of C4 plants through mid to late Holocene, likely in response to increasing aridity in the basin. The remarkable correspondence between the LCFA and bulk TOC indicates that LCFA is a representative organic component for bulk OM composition, and the latter is predominantly derived from terrestrial sources. Additionally, an increase in the radiocarbon age offset between OC and depositional age, from <400 year in the early Holocene to ~1500 years, concomitant with increased sedimentation rate, suggests enhanced mobilization and export of pre-aged (soil-derived) terrestrial OC during the latter half of the late Holocene. This change is likely driven by widespread destabilization and exhumation of (deeper) soils accompanying changing monsoon dynamics, as well as the rise of agricultural practices.

33

3.1. Introduction

Terrigenous organic carbon (OC) transport from continents to the ocean and its ultimate fate therein play a crucial role in regulating global carbon and oxygen cycles and hence past and future climate (Berner, 1989; Burdige, 2005). The nature of the OC, its mineral associations, as well as the conditions under which it is deposited exert a significant influence on the oxidation state of the atmosphere (Hedges and Keil, 1995; Blair et al., 2004; Leithold et al., 2005). The amount of carbon in recently biosynthesized terrestrial biomass and soils is about 3-5 times greater than that as

CO2 in the atmosphere (Hedges, 1992; Houghton, 2007). Thus, small changes in the rate of carbon exchange between the atmosphere and these terrestrial reservoirs can influence atmospheric CO2 inventories. Therefore, understanding controls on the composition, reactivity, and fate of terrigenous OC is of key importance. Rivers constitute an important conduit for the transfer of terrigenous OC (either in particulate (POC) or dissolved (DOC) form) between land and sea, and link about 85% of the continental surface to the ocean (Ludwig et al., 1996; Schlünz and Schneider, 2000). River-dominated continental margins provide an integrated perspective on drainage basin processes that influence the flux and nature of land–derived organic matter exported to the ocean. OC eroded from catchments to rivers derives from three primary sources: detritus of recently biosynthesiszed terrestrial biomass, soils, and sedimentary rock (Hedges et al., 1986). These inputs can be augmented by those from primary and secondary production within the river. Of these inputs, soil and sedimentary rock (petrogenic OC) may contain relatively refractory OC, having previously experienced degradation and modification during their formation, and hence influence carbon cycling over longer (thousands to million years) timescales (Tao et al., 2015). A combination of physical transport and biogeochemical processes along the river- ocean continuum determine where, and to what extent, materials are remineralized, or exported to the ocean (Vonk et al., 2008). Although most of the terrigenous OC supplied to the ocean is remineralized (Hedges and Oades, 1997) materials which escapes this fate and are buried in oceanic sediments is thought to primarily consist of refractory material that has survived decomposition in soils and floodplains as well as during transport (Keil et al., 1997). Protracted storage of refractivity terrigenous OC on the continents supplies pre-aged OC to marine sediments and can contribute to old

34 radiocarbon ages of OC in continental margin sediments (Bao et al., 2016; Griffith et al., 2010). Terrigenous OC contributions to riverine DOC and POC, and its effect on the apparent radiocarbon ages have been assessed in several studies using a wide range of approaches, including elemental and bulk carbon isotopic compositions and molecular characteristics (e.g. Hedges et al., 1986; Collister et al., 1994; Raymond and Bauer, 2001; Blair et al., 2004; Drenzek et al., 2009; Galy and Eglinton, 2011). Because of the inherently complex and heterogeneous nature of OC within drainage basins, a challenge in quantifying different OC components is to constrain respective end- member characteristics (Tao et al., 2015). Variable radiocarbon ages have been attributed to differences in morphology and geology of the respective catchments. For instance, Raymond and Bauer (2001) suggested that the main controlling factor is the relief of the headwaters. Steep-relief basins showed older radiocarbon ages interpreted as result of enhanced erosion of old soils and bedrock. Blair et al (2004) posited that OC storage in lowland soils controls the riverine POC age and overprints the headwater signals. They, however, also showed that intermediate storage in soils may be bypassed in small mountainous rivers, in which case OC age is determined by bedrock and plant isotopic signals (Blair et al., 2004). Griffith et al. (2009) have shown that in densely populated catchments wastewater effluents contributing high amounts of petroleum-derived carbon skew riverine DOC and POC radiocarbon ages. Furthermore, Longworth et al. (2007) argued that POC is older in basins draining both OC-rich lithology and agricultural land use areas. Vascular plants synthesize a range of biochemical compounds, including lignin, cellulose, and lipid epicuticular waxes (Eglinton and Hamilton, 1967). The latter, which include long-chain n-alkanes, long-chain n-alkanols, and long chain n-alkanoic acids (herein referred to as long-chain fatty acid, LCFAs), have been extensively employed as both biomolecular tracers of higher plant carbon as well as molecular proxies of past vegetation and climate change, due to their inherent resistance to degradation and ability to survive transport over long distances and timescales, while retaining the original information encoded in them (Eglinton and Eglinton, 2008). Compound- specific isotopic measurements (13C, 2H, and 14C) of lipid biomarkers have the potential to provide insights into past ecology, hydrology, and atmospheric vapour dynamics across a range of timescales. Coupled molecular 13C and 14C measurements have previously been used to establish the provenance of natural and 35 anthropogenic compounds in marine sediments (Eglinton et al., 1997), soils (Trumbore et al., 1989), aerosols (Eglinton et al., 2002), and rivers (Tao et al., 2015). Radiocarbon measurements on plant waxes have been used to study residence times of terrestrial OC in a wide variety of river/continental margin settings including tropical/subtropical (Galy and Eglinton, 2011), temperate (Kusch et al., 2010), and high latitudes (Feng et al., 2013). These studies have observed plant wax 14C ages that range from decades (Mayorga et al., 2005) to several millennia (Drenzek et al., 2009), suggesting significant variations in mean terrestrial residence time between different regions that are likely controlled by a combination of factors including climate, bedrock geology, and basin morphology. This study presents a high-temporal resolution molecular (LCFA) and bulk (total organic carbon, TOC) isotopic investigation of a sediment core collected proximal to the mouth of the Godavari River. Observations are used to constrain past variations, sources, composition, and age of terrestrial OC exported by the Godavari River to the adjacent margin, and to infer past carbon dynamics within the Godavari catchment. Findings are interpreted within the context of climatic, geologic, and anthropogenic influences on the basin, and implications for carbon cycle processes.

3.2. Materials and methods

3.2.1. Study area

The Godavari River drains the central part of the Indian peninsula, discharges into the Bay of Bengal (BoB) (Fig. 3.1) and is one of the major sources of terrigenous sediment to the adjacent margin. Extensive aridification of this region, a consequence of weakening of the Indian Monsoon over the late Holocene (Ponton et al., 2012), renders the Godavari an ideal system to examine the relationships between climate and transport dynamics of terrestrial OC from large drainage basins. The river has its source in the mountains of the Western Ghats and drains the Deccan Plateau in western-central India and traverses various geological and vegetation gradients as it traverses east-central before emptying into the BoB. The modern-day natural vegetation cover comprises a mixture of savannah, tropical grassland, and tropical forest (Asouti and Fuller, 2008). The Deccan plateau segment is dominated by C4- plants, whereas the lower basin is dominated by tropical grasslands (mixture of C3/C4 plants) as well as other types of C3-vegetation towards the estuary/delta. Crops in the 36 upper catchment of the Godavari basin is dominated by sorghum and millet (C4-plants) and the lower basin is mostly covered by rice (C3-plants; Krishna et al., 2013).

Sugarcane (a C4-plant) is also a prominent agricultural crop in the Godavari catchment. Anthropogenic changes in vegetation cover through cultivation have been considered minimal up until the 19th century when intensive deforestation of the Eastern segment of the peninsula first began (Hill, 2008). Water and sediment discharge by the Godavari are mostly controlled by precipitation associated with the Indian Summer Monsoon. This coupled ocean-land-atmosphere climate phenomenon, which is driven by cross-equatorial pressure gradients and amplified by land-sea thermal gradients, results in low-level advection of warm, moisture-laden air from the Indian Ocean and insulation of these air masses by orographic barriers (e.g. Western Ghats; Wu et al., 2012). This generates intense precipitation over western India during the peak monsoons period - June to September - when India receives more than 80% of its annual precipitation (Gadgil, 2003). Over 70% of the Godavari catchment is agricultural land and forest. Farming and land management is considered to have exacerbated the erosion and delivery of sediments to the BoB (Giosan et al., 2017). Large-scale construction of dams over the Godavari has increased since the mid- 1960s and at present, more than 300 hydrological projects of various sizes are currently in operation, limiting the total fluvial transport of sediment to the BoB (CWC, 2010). The annual sediment load from the Godavari to the BoB has declined significantly over the past three decades from 170 x 106 tons/year (in 1988) to ~15 x 106 tons/year (in 2010), and this decrease has been largely attributed to damming along the mainstream and decreasing monsoonal rainfall (Pradhan et al., 2014).

37

Figure 3.1. A map of peninsular India showing the major rivers and adjacent oceans. Location of core 16A is indicated in red.

3.2.2. Methods

A sediment core NGHP-01-16A (16.59331N, 82.68345E, hereafter referred to as core 16A) was collected on the continental margin (1268m water depth) about 40 km from the mouth of the Godavari River where it decants into the BoB (Collet et al., 2014). The continental shelf in front of the Godavari is relatively narrow (<10 km), and thus limited fluvial material is stored on the shelf and the majority is rapidly exported downslope (Giosan et al., 2017). The core was stored refrigerated at the Woods Hole Oceanographic Institution core repository. The top 25 cm of the core has been largely

38 exhausted by prior investigations (e.g. Ponton, 2012; Ponton et al., 2012; Giosan et al., 2017; Zorzi et al., 2015), and where possible we augmented our results with those from the afore-mentioned studies. We analyzed fifty-nine samples (~5 cm thick sections) of bulk sediments taken from the top 8.5 m of marine sediment core 16A at a sampling resolution of 10 cm. This correspond to average sampling interval of 130 years (~90 years near the top of the core to ~500 years near the bottom of the core). The samples were processed for measurements of total organic carbon (TOC) content, grain size (GS), and mineral- specific surface area (MSA). In addition, stable and radioisotopes of carbon were measured on TOC, long-chain (C26-32) n-alkanoic acids (as fatty acid methyl esters, FAMES) and planktic foraminifera. TOC and FAME ages are reported as conventional radiocarbon ages.

3.2.2.1. Age Model

The age model for core 16A (Fig. 3.2a) was established using 40 14C dates on planktic foraminifera. Tests of planktic foraminifera were picked from the 125-250 μm size fraction of sediments obtained by wet-sieving using Milli-Q water. These measurements were performed on mixed planktic foraminifera species (G. ruber and G. bulloides). Radiocarbon ages of planktic foraminifera were converted to calendar ages using the CALIB 7.0 radiocarbon calibration program (http://calib.org) with the calibration dataset Marine13 (Reimer at al., 2013). A ∆R value of -65 ± 53 and a reservoir correction of 229 years were applied based on values from the northwestern segment of BoB (Dutta et al., 2001). Linear sedimentation rates for the Holocene (Fig. 3.2b) are based on the calibrated age of planktic foraminifera.

39

Figure 3.2. (a) Age model for core 16A. Uncertainty associated with age determinations are denoted by the grey envelope enclosing the data points. (b) Linear sedimentation rates for core 16A.

40

3.2.2.2. Bulk elemental and isotopic analysis

Detailed sample preparation for bulk elemental and isotopic (stable and radioactive) composition of carbon as well sedimentological measurements have been previously reported in McIntyre et al. (2016) and Usman et al. (2018).

3.2.2.3. Compound-specific stable isotopic analysis

The lipid extraction procedure has been previously reported in Usman et al. (2018). LCFAs were analyzed as fatty acid methyl ester (FAME) derivatives. For stable carbon isotopic measurements, approximately 10% of the FAME fractions of each sample were analyzed in duplicate on an HP 6890 gas chromatograph (GC) equipped with a CP-Sil 5 CB lowbleed/MS column coupled to a Finnigan Delta-V Isotope Ratio Mass Spectrometer (IRMS) system via a modified combustion interface. Standard mixtures of n-alkanes with known δ13C values were measured in triplicate between every 10 injections. Results are reported as the mean of duplicate analyses (with 1σ analytical uncertainties) in δ13C (‰) notation on the VPDB scale and have been corrected for the addition of the carboxyl methyl group during esterification via an isotope mass balance.

3.2.2.4. Compound-specific 14C analysis

Individual nC16, nC24, nC26, nC28 homologues as well as the combined nC18+20+22 and nC30+32 homologues were isolated (as corresponding FAMEs) by preparative capillary gas chromatography (PCGC, an Agilent 7890 GC equipped with Gerstel CIS 4 injection system coupled to a Gerstel preparative fraction collector; see Eglinton et al., 1996 and Tao et al., 2015 for a detailed description). PCGC traps were eluted with DCM and brought to a known volume for purity check and quantification for recovery calculations on a gas chromatograph-flame ionization detector (GC-FID). Individual FAMEs were generally >99% pure, and recoveries averaged 62%. After isolation, the FAMEs were eluted over a silica gel column using DCM to remove potential column bleed contamination. The pure FAMEs were quantified using a GC-FID. In order to obtain sample yields large enough for radiocarbon measurements, some isolates were recombined (nC26-nC32).

41

Purified FAMEs were transferred into pre-cleaned tin capsules designed for liquid samples. Prior to sample transfer, the tin capsules were pre-rinsed with DCM and dried in a 50°C oven. The tin capsules containing the FAMEs were dried on a 40°C hot plate and subsequently wrapped neatly and placed in a pre-combusted GC vial. Compound-specific radiocarbon measurements were performed MICADAS facility at ETH Zurich using the EA-AMS system (McIntyre et al., 2016; Ruff et al., 2010). The corrected radiocarbon contents are reported as conventional radiocarbon age or fraction modern (fm). fm is also commonly referred to in the literature as ASN/AON 14 (Stuiver and Polach, 1977) or CN (Mook and van der Plicht, 1999) and refer to the 14C activity of a fraction compared to the activity of a “modern” international standard (Oxalic acid), both corrected for carbon isotopic fractionation using their respective δ13C values. Also, conventional radiocarbon ages (yr BP) are reported, that is, the ages calculated from measured radiocarbon contents using the Libby half-life of radiocarbon (Stuiver and Polach, 1977). Furthermore, the radiocarbon contents are also reported as age corrected, or initial, Δ14C values where the depositional age (in calendar age) of a sample is known. This age corrected Δ14C may also be encountered 14 in the literature as Δ (Stuiver and Polach, 1977), Δ Cinitial (Mollenhauer and Eglinton, 14 i 2007), or δ N (Mook and van der Plicht, 1999). Here we have adopted the Δ notation of Stuiver and Polach, 1977. The radiocarbon ages of the individual n-alkanoic acids were corrected for the addition of one methyl group during derivatization using isotopic mass balance. In addition, internal corrections for additional modern and dead carbon that can introduced as contamination during AMS analytical procedures have been applied before reporting final Δ14C values, based on quantitative measurements of Ox-I and coal standards as well as the long-term blank average for the facility. Since the fractional abundance of “modern” (defined as 0‰) and “dead” (-1000‰) blank contamination introduced during the various sample preparation stages increases in proportion to decreases in total sample mass (Shah and Pearson, 2007; Santos et al., 2010), measured Δ14C values and their errors require corresponding adjustments for these influences at small sample levels. This is especially so in our case where sample sizes range from 9-45 µg C. To assess the blank contribution, PCGC isolation steps were carried out with only solvents and no sample added. In addition, solvent blanks were assessed by quantifying the deviation between nominal 14 and measured Δ C of radiocarbon dead (nC28 alkane, Sigma Aldrich LOT # 42

14 BCBK0953V, Δ C = -979±4‰) and modern (nC32 alkane, Sigma Aldrich LOT # BCBJ5372V, Δ14C = +28±7‰) standards spiked into sub-fractions of the solvent blanks after PCGC isolation. This yielded 0.74±0.25 of modern C contamination and 0.26±0.10 of dead C contamination. The combined blank from this method is 1.00±0.30 µg C with a Δ14C value of -297±210 and was applied to blank correction in this study. The resulting analytical uncertainty for 14C analysis of FAMEs isolated with PCGC procedure ranges from 2 to 22% (mean = 5%), including full error propagation as presented in Hanke et al., 2017.

3.3. Results

Down-core variations in total organic carbon (TOC) contents are plotted alongside the 13 stable carbon isotopic composition of bulk sedimentary TOC (δ CTOC) for core 16A in Fig. 3.3. In addition, the age corrected 14C composition (Δ) as well as abundance and stable isotopic compositions of LCFAs are shown. The complete dataset for core 16A are given in the supplementary information. TOC contents range from 1.20 to 2.15%, and generally increases from the early Holocene to mid Holocene (~4500 yr BP), after which the trend reverses and there is a progressive decrease in TOC until present. Mineral surface area normalized-OC and LCFA (herein termed OC and LCFA loadings, respectively) show distinct pattern during the Holocene. The early Holocene is characterized by higher OC loadings with values ranging from 0.25 to 0.38 mg OC/m2 (mean = 0.32 ± 0.04 mg OC/m2). The late Holocene values range from 0.19 to 0.34 mg OC/m2 (mean = 0.25 ± 0.04 mg OC/m2; Fig. 3.4a). Conversely, the early Holocene is characterized by lower LCFA loadings (mean = 1.51 ± 0.83 μg LCFA/m2) than the late Holocene (mean = 2.80 ± 0.98 μg LCFA/m2; Fig. 3.4b). The TOC-normalized contents of LCFA concentrations generally increase from early to late Holocene (Fig. 3.3), whereas no systematic variation is evident in short-chain homologues (data not shown). LCFA contents range from 10 to 304 µg/g TOC for n-

C26 to n-C32, and the distributions display distinct molecular level patterns. Specifically, n-FAs show a typically bimodal distribution with an even carbon number predominance with carbon number maxima at C16 and at C24, C26 or C28 (Fig. S1).

Amongst the long-chain homologues (C26 to C32), C26 and C28 dominate with a general increase in LCFA from the early to late Holocene.

43

Generally, the stable carbon isotopic composition of TOC shows gradual enrichment 13 13 in C from the early to late Holocene with δ CTOC values ranging from -21.9 ± 0.2‰ to -18.2 ± 0.2‰. These values are similar to those reported in Ponton et al., 2010. The δ13C values of LCFA average -27.97 ± 1.32‰ in the early Holocene and -24.82 ± 13 1.17‰ in the late Holocene. The δ CLCFA trend is similar to that observed for bulk OC 13C (i.e. a general increase from early to late Holocene; Fig. 3.3), and with prior observations (Ponton et al., 2012). Radiocarbon measurements show a characteristic age relationship between the different sediment constituents (TOC, LCFA, and foraminifera) in core 16A (Fig. 3.4). Where all three constituents are measured, foraminifera are the youngest (most enriched in radiocarbon) of these three constituents. The conventional 14C ages of the dated sediment components generally increases with core depth. Radiocarbon ages of the planktic foraminifera, long-chain fatty acids, and TOC are offset by between 3 years to 4200 years. These age offsets are significant (greater than 2σ analytical error), even after accounting for measurement uncertainties for small samples (<100 µg C) related to sample preparation and analytical errors on AMS measurements. LCFA are older than foraminifera by 44 to ~2000 years, except for the depth intervals 618-622 cm and 659-652 cm where LCFA are younger than foraminifera by 23 and 236 years, respectively. These values are within the 2σ error margin and are therefore not considered significant. Similarly, TOC ages are older than foraminifera by 12 to ~1300 years. There is a systematic increase in the age offset between TOC and foraminifera from early to late Holocene. However, unlike TOC there is no systematic offset between LCFA and foraminifera. pronounced age offsets are observed for one sample near the top of the core (42-46 cm) where LCFA is ~4700 years older than the corresponding foraminifera. Likewise, the LCFA date at the top of the core (0-6 cm) is ~5000 years older than other sediment constituents (Ponton 2012).

44

13 Figure 3.3. Down-core profiles for core 16A showing (a) TOC content (b) δ CTOC (c) 14 D values of bulk OC (i.e. Age-corrected D COC) (d) Concentrations of LCFA (e) 13 δ CLCFA. Data from Ponton et al. (2012) are superimposed on the topmost panel.

45

Figure 3.4. (a) Bulk OC and (b) LCFA loadings for core 16A.

46

3.4. Discussion

3.4.1. OC and biomarker loadings and distribution

Mineral surface associations have been shown to influence OM stability in marine sediments (Keil et al., 1997). Measured OC/SA in core 16A sediments range between 0.19 - 0.38 mg C/m2 (Fig. 3.4a) and are significantly lower than values reported for many continental margin sediments. However, our values are similar to those reported for high sedimentation rate deltaic environments (Aller and Blair, 2006; Mayer 1994). The generally low OC loadings as well as the general decrease in OC loadings from early to late Holocene can be attributed to extensive and variable supply of strongly weathered soils and sediments that are low in OC. In addition, the lower loadings of late Holocene sediments reflect increase supply of high surface-area clay minerals (e.g. smectite). The latter are abundant in the Deccan region of peninsular India constituting up to 65% of the total phyllosilicate fraction (Usman et al., 2018). Decreased OC loadings could also reflect increased erosion of OC-lean deeper mineral soils during the late Holocene. In general, these observations, together with available Nd isotopic evidence (Giosan et al., 2017) point to increased contribution of Deccan-derived material in the late Holocene. Lipids comprise <1% of the bulk terrestrial OC pool, and thus there is a potential disconnect between biomarker and TOC abundance due to variability in original source biomarker/TOC ratios and differential diagenesis (Jasper and Gagosian, 1993). Nevertheless, terrestrial lipid biomarkers (e.g. LCFA) are routinely used as a tracer of terrestrial OC inputs and preservation (Blair and Aller, 2012). Building on the approach of Freymond et al (2018), we have employed the LCFA/SA biomolecular loading parameter to assess variations in loading of terrestrial OM on mineral surfaces. Unlike bulk TOC, there is a general increase in LCFA loading during the Holocene with values ranging from 0.45 μg/g OC/m2 in the early Holocene to 4.93 μg/g O C/m2 in the late Holocene (Fig. 3.4b). This indicates that despite the general decrease in bulk OC in the later part of the Holocene, a significant portion of the terrigenous OC persisted until burial with little to no degradation. In addition, the general increase in sedimentation rates throughout the late Holocene likely enhanced OC burial efficiency/preservation. The general increase in the LCFA abundance in the late Holocene corroborates this interpretation.

47

3.4.2. Foraminifera-based chronology

Planktic foraminifera are generally assumed to faithfully record surface water dissolved organic carbon (DIC) radiocarbon content and thus represent the time of sediment deposition since their size and density facilitates rapid sinking through the water column (Pearson et al., 2000; Kusch et al., 2010). For the time intervals covered by core 16A, this constituent is considered the best estimate of a “depositional age” of the sediment horizon (e.g., Mollenhauer et al., 2007; Pearson et al., 2000). As a result, radiocarbon dates on mixed planktic foraminifera was used to generate the chronology for the core and show that the core spans the past 11ky.

3.4.3. Radiocarbon contents of planktic foraminifera and organic fractions

In general, there is remarkable similarity between the trends displayed by the three sediment constituents - i.e. LCFA, foraminifera and TOC (Fig. 3.5) - suggesting that the age of LCFA is representative of the wide array of organic components that constitute the bulk OM which were subjected to the same physicochemical processes during their transport from source to the depositional basin. The 14C age of foraminifera, TOC, and LCFA are within 2σ from the early Holocene up until around 4.7 kyBP when they start to diverge and the 14C age difference gradually becomes more pronounced through the later part of the Holocene. The 14C age differences within 2σ error margin can be attributed to myriad of measurement uncertainties as sample preparation and processing often contribute small amounts of carbon contamination of unknown isotopic composition. This is especially true for small sample sizes (such as those used in this study) and compounds that require several processing steps (Drenzek et al., 2009; Kusch et al., 2010).

48

Figure 3.5. Radiocarbon age of the different sediment components (i.e. Bulk OC, LCFA, and planktic foraminifera) plotted as a function of calibrated age.

When expressed as age offset, there is a marked increase in TOC age offset (i.e., difference between TOC age and coeval sediment depositional ages) from less than 400 years (prior to ~4.7 kyBP) to more than 1500 years in the late Holocene and even up to 2500 years for the core top. Similarly, the average age offset of LCFA until 4.7 kyBP is ~125 years and increased to ~950 years for the later part of the record (Fig. 3.6). This change beginning 4.7 kyBP is especially significant as it coincides with the beginning of basin-scale changes in hydroclimatic conditions and designated as the onset of aridification of the Indian peninsular (Ponton et al., 2012). Notably, this is also the interval when the bulk TOC trend changes and coincides with the onset of

49 increasing sedimentation rate through the late Holocene (Giosan et al., 2017; Ponton 2012).

Figure 3.6. Radiocarbon age offset of bulk OC during the Holocene. Also plotted are the data from Ponton et al. (2012).

In the early Holocene, the oldest of the fatty acids is more than 2σ younger than the coeval planktic foraminifera ages. If foraminifera ages represent the best estimate of depositional age in sediment affected by transport, such age offset would be implausible since it is highly unlikely to have a carbon source fresher than directly fixed marine surface water dissolved inorganic carbon. Plant waxes derive the carbon from the atmosphere whereas foraminifera carbon is from DIC, which includes a surface ocean reservoir correction, and our calendar age calibration accounts for this reservoir correction. Solvent extraction procedures introduce no significant uncertainties to radiocarbon measurements on foraminifera shells (Ohkouchi et al., 2005) and other

50 potential causes for the disparity such as bioturbation, differential dissolution, and selective transport of either the foraminifera or organic fractions (Kusch et al., 2010) are equally unlikely. We attribute the younger LCFA age offset for this particular period to overestimation in the planktic foraminifera age calibration, or to insufficient blank correction for LCFA from these older depth intervals (where potential entrainment of modern carbon would exert the greatest influence on measured 14C ages).

3.4.4. Contrasting bulk OC and LCFA age offsets – A mixing of different OC pools?

There is consistent increase in the bulk OC and LCFA age offset beginning in the mid Holocene (~5 kyBP) with values ranging from between ~200 to ~5000 years (Fig 3.6). The observed age offset of bulk OC is consistently higher than the LCFA. For instance, the average LCFA offset is ~700 whereas that of bulk OC for the same time interval is ~1300. This systematic difference suggests a mixture of two or more components that are characterized by longer residence times within the Godavari basin. It has been observed is some alpine soils LCFA are as old or older than bulk OC, suggesting that they are a refractory component of soil OC (van der Voort et al., 2017). On the other hand, Galy and Eglinton (2011) suggest that very old soils contain low concentrations of LCFA, presumably resulting from degradation within the soil. A plot of fatty acid 14C

14 (i.e. D CLCFA) against LCFA concentration shows that sediments with generally low

14 LCFA concentrations have a positive correlation with D CLCFA whereas at higher concentrations, there is little to no correlation (Fig. 3.7). This is similar to the observation of the Galy and Eglinton (2011), suggesting a mixing of refractory more slowly cycling mineral soil components and vegetation derived component. In addition, the general decrease in bulk OC loadings and increase in sedimentation rates is consistent with increased flux from erosion of deeper soils. The declining vegetation cover associated with the progressive aridification of the Godavari during the late Holocene (Ponton et al., 2012; Zorzi et al., 2015) likely promote the development of soils that are enriched in old refractory OC. Lower LCFA offset could also result from incorporation of greater proportion of younger OC derived from surface soils, where plant wax concentrations may be higher and the increasing LCFA loadings (despite a decrease in bulk OC loadings) supports this hypothesis. Finally, the relatively higher OC age offset could be a consequence of mobilization of even older source of OC

51

(e.g. petrogenic OC) that does not contain LCFA. This latter scenario is however unlikely given the bedrock geology of the Godavari drainage basin.

Fatty acids concentration (�g/g OC) 0 50 100 150 200 250 300 350 0

-100

-200

-300

-400 C composition (‰) composition C

14 -500

-600

-700

Fatty acids ∆ -800

-900

Figure 3.7. Radiocarbon contents of LCFA Age versus LCFA concentrations. Red- early Holocene; Blue-late Holocene

3.4.5. Terrestrial C3 versus C4-plants contributions to bulk OC

13 13 13 The range of bulk (δ COC =-23 to -18‰) and LCFA (δ CLCFA = -30 to -23‰) δ C 13 values in our study suggests possible contributions from terrestrial C3 (δ COC = -23 to

13 -30‰ and δ CLCFA = -36 to -30‰; Smith and Epstein, 1971; Chikaraishi et al., 2004a), 13 13 and C4 plants (δ COC = -17 to -9‰ and δ CLCFA = -28 to -18‰; Sackett, 1989; 13 Chikaraishi et al., 2004b), as well as marine phytoplankton (δ COC = -22 to -19‰ and 13 δ CLCFA = -26 to -19‰; Fry and Sherr, 1984; Chikaraishi et al., 2004b). OM from terrestrial C3 and C4 plants are supplied to the BoB mostly via riverine transport of OC derived from vegetation and crops in the Godavari drainage basin. Erosion of natural

52 and agricultural lands and subsequent riverine transport delivers these plant detritus and soil OM to the BoB during the summer monsoon. Recent work has shown that a number of rivers export aged OC and the source of this aged material is a combination of slow-cycling pools of soil OC with long residence times and ancient OM released by eroding sedimentary rock (Raymond et al., 2004; Dickens et al., 2004; Komada et al., 2005; Leithold et al., 2006). Common to all these studies is the presence of OM-rich sedimentary rocks underlying the associated catchments. In contrast, the dominant bedrock of the Godavari comprises of basalt and Precambrian granites, accounting for ~87% of total drainage area. Thus, petrogenic OC contribution to total OC pool can be considered minimal. In addition, the TOC contents in core 16A are relatively high, suggesting that petrogenic contributions (usually characterized by low TOC contents; Marwick et al., 2015), is minor.

Apportioning bulk OC between C3 and C4 end members is relatively straightforward due to their distinct signature. However, accounting for autochthonous OC production is considerably more difficult (especially in the absence of suitable proxies such as chlorophyll a and DIC carbon isotopes). This is owing to the large variations in sources of DIC pools utilized by photoautotrophs, including atmospheric C, carbon associated with carbonate dissolution, and soil respired CO2 pumped to freshwater from the surrounding basin, itself dependent on the terrestrial photosynthetic pathway (C3 versus C4) employed to fix the initial atmospheric C (Marwick et al., 2015). Core 16A is characterized by relatively high sedimentation rate (>60 cm/kyr) through the Holocene (Fig. 2b). Given this high sedimentation regime and monsoon-driven fluvial dynamics, the Godavari River is likely very turbid, thereby limiting autotrophic and heterotrophic production. Thus, aquatic OC contribution is considered unlikely.

The relative proportions of OM from C3, and C4 plants were estimated from the end- member mixing model using the following stable isotope mass balance equations.

13 13 13 δ CC = f • (δ C4) + (1 - f) • (δ C3) (Eq. 1)

Eq. 1 can be written as:

13 13 13 13 f = (δ CC - δ C3) / (δ C4 - δ C3) (Eq. 2)

53

where f represents the fractional contribution of C4 plants; (1 - f) is the fractional 13 13 13 13 contribution of C3 plants; and δ CC, δ C3, and δ C4 are the δ C value of sediment components (i.e., bulk OC and LCFA), C3 and C4-plant end member, respectively. The values of bulk C3 and C4-plant end member used here are -25.3‰ and -13.0‰, 13 respectively, based on C measured on C3 and C4-plant collected from the modern Godavari basin (Krishna et al., 2013). Similarly, -31.3‰ and -27.4‰ have been

13 employed as LCFA δ C of C3 and C4-plant end member, respectively, based on values reported from the lower (C3-plant dominated) and upper basin (C4-plant dominated) riverine sediments (Usman et al., 2018). Using Eq. 2 above, the relative proportion of each of these sources to the sediments of core 16A can be quantified.

The relative contributions of C4 plants remain fairly uniform from prior to 5 ky BP (mean = 37%) and increase progressively (up to 57%) over the last 5 ky BP (Fig. 3.8). The

%C4 based on LCFA generally display similar, albeit muted, trend over the Holocene and can be attributed to increased sourcing from the upper catchment of the Godavari during the later portion of the Holocene (Usman et al., 2018). Thus, the relative contribution of OM transported by the Godavari to the BoB is largely controlled by the vegetation types in its drainage basin. The general agreement between LCFA and bulk OC implies that OM in sediment from the continental margin at the core location is dominated by allochthonous OM transported by the Godavari River from the Indian peninsular.

54

1

0.9

0.8 Bulk OC LCFA 0.7

0.6

0.5

0.4 plant (x100) plant - 4 0.3

% C % 0.2

0.1

0 0 2000 4000 6000 8000 10000 Calibrated Age (yr B.P.)

Figure 3.8. C4-plants contribution to bulk OC in the Holocene.

3.4.6. Quantification of different OC source and contributions using coupled isotopic mixing models

For most of the Holocene, the 14C ages of bulk OC are older than the corresponding depositional ages, implying pre-aged OC contributions to the sediment. These aged OC may have been derived from erosion of fossil OC or pre-aged higher plants OC in the basin. Also, it may have been a consequence of OC aging associated with lateral transport through the river basin. A three end-member mixing model based on the δ13C and Δ14C values of bulk and compound-specific biomarkers is used here to estimate the relative fractional contribution of modern/contemporary (fM), pre-aged soils (fS), and fossil (fF) OC in core 16A sediments. This mixing model is expressed by the following equations (Drenzek et al., 2009; Tao et al., 2015; Tao et al., 2016):

14 14 14 14 Δ COC = fM (Δ CM) + fS (Δ CS) + fF (Δ CF) (Eq. 3)

55

13 13 13 13 δ COC = fM (δ CM) + fS (δ CS) + fF (δ CF) (Eq. 4)

fM + fS + fF = 1 (Eq. 5) where f is the fractional abundance and the subscripts OC, M, S, F are bulk TOC, modern, soil, and fossil OC. Based on the source assessment above, the stable 13 14 isotopic compositions of nC16 are employed for δ CM and for Δ CM, the uncalibrated radiocarbon contents of foraminifera are used. Planktic foraminifera 14C faithfully records surface water DIC 14C (Pearson et al., 2000), thus we have opted for 14 14 foraminifera C in the absence of nC16 C data. The abundance-weighted average 13 14 isotopic compositions of LCFA are chosen for δ CS and Δ CS. In the absence of molecular-level isotopic information for the fossil end-member, values of -28‰ and - 13 14 1000 (assumed) are selected for δ CF and Δ CF. These source assignments are based on the premise that short-chain (e.g. nC16) and long chain (nC26-32) homologues of fatty acid reflect modern OC and pre-aged mineral associated OC in soils, respectively. This source assignment is rather simplistic as there are many different sources for FA, especially short-chain fatty acids (Volkman et al., 1998). A modest 5- 13 13 9‰ depletion from bulk biomass (δ Cbiomass - δ Clipid; Collister et al., 1994, Schouten 13 et al., 2008, Drenzek et al., 2009) was applied to the measured δ C values used for 13 each end-member in order to approximate the corresponding δ Cbulk values based on the work of Tao et al (2015; Supplementary information). A plot of bulk OC δ13C versus LCFA δ13C exhibits a strong positive correlation (r2 = 0.92, r <0.001; Fig. 3.9) suggesting that OC in core 16A derives mostly from terrestrial inputs. Based on this mixing model, pre-aged soil OC contribution dominates the OC budget throughout most of the Holocene. Similarly, the marine OC contribution is significant, averaging ~45% throughout the record. The relatively high contribution of marine OC is likely a consequence of the large uncertainty associated with the values chosen for the fossil end member. The lack of direct 14C (for marine component) and 13C (for petrogenic component) measurements hinder a tight constraint on the end-members. Nonetheless, a ~40% marine (with presumably modern C) input would imply that the terrestrial component derives from an even older pool than implies from the age offset.

56

13 13 Figure 3.9. Cross plot of δ CLCFA versus δ COC for core 16A.

57

3.4.7. Implications for OC transport and cycling

Terrestrial OM can undergo myriad of processes and follow several pathways before its ultimate burial in continental margin sediments (Blair et al., 2004). Thus, terrestrial OM buried in marine sediments represents a mixture of a broad suite of OC “flavours” and “vintages” derived from different carbon reservoirs (Drenzek, 2007). The bulk and biomarker abundance and isotopic data shed light on the dynamics of OC delivery to the continental margin proximal to the mouth of the Godavari. The Godavari River (and its tributaries) supplies the majority (~53%) of sediment to this sector of the BoB (Subramanian, 1987). The river’s influence is highlighted by the similarity between geochemical and sedimentological signatures of drainage basin and adjacent offshore sediments (Usman et al., 2018), suggesting that changes in the sedimentary record are a reflection of changes (natural and/or anthropogenic) within the river drainage basin. The abundance and isotopic distribution of LCFA provides insights into the sedimentological processes influencing nature and composition of vascular plant- derived OC. In contrast to bulk parameters, the abundance and loadings of LCFA increases from early Holocene to late Holocene, consistent with increasing vascular plant fluvial supply of sediment towards the late Holocene, suggesting that a copious amount of terrestrial material is exported and buried. This interplay between source and transport of terrestrial and marine OC provides a basis for understanding OC supply and deposition on the Godavari margin.

The stable isotopic composition of LCFA indicates a mixed origin from C3 and C4 plants consistent with the vegetation coverage of the Godavari River Basin (Krishna et al., 2013), signatures observed in the modern river system (Usman et al., 2018) and the stable isotope mixing model results discussed above. These compounds are significantly pre-aged upon deposition, likely due to their resistance to degradation resulting from low reactivity or stabilization on mineral surfaces (Eglinton and Eglinton, 2008; van der Voort et al., 2017). Thus, LCFAs can resist environmental degradation resulting in old radiocarbon ages determined by the respective average residence time of the compounds in active reservoirs of carbon such as soils and floodplains (Drenzek et al., 2009). In the absence of significant accumulation in alluvial deposits within the Godavari basin owing to the limited extent of the floodplain (Kale, 2002), Giosan et al. (2017) hypothesized that the age offset is primarily as a result of deeper (and older) soil

58 exhumation by gully erosion and deep channel incisions. Changes in apparent terrestrial residence time has implications for terrestrial carbon cycle. For instance, an increase in residence time could imply that under arid conditions such as those prevalent in the late Holocene, the terrestrial to marine “conveyor” system slows, thus prolonging the transfer of OC from continent to ocean. In contrast, export of pre-aged OC previously sequestered in deeper mineral soils reintroduces older carbon to the fluvial network. During the early Holocene, the relatively high precipitation and the attendant elevated discharge (Prasad and Enzel, 2006) promotes rapid OC mobilization and transport through the drainage basin. In the mid to late Holocene when conditions in the drainage basin became more arid (Ponton et al., 2012), the age offset between planktic foraminifera and LCFA increase concomitantly, especially during the last 2200 years. During this latter period, the river catchment experienced severe droughts (Zorzi et al., 2015), resulting in a reduction of fresh biospheric OC flux and an increase in exhumation of pre-aged (mineral) soil OC. The progressive increase in sedimentation rate beginning in the mid Holocene (Fig. 2b) and the gradual decrease in TOC is consistent with this interpretation. Enhanced soil erosion may have been exacerbated by diminished vegetation cover, and more abrupt and intense storms (e.g. flash floods) incise into the soils. Furthermore, the local population increased significantly since ~2000 years ago (Allchin, 1995) and the attendant intensification of agriculture has led to a significant perturbation of the landscape via mobilization of soil OC, contributing to the observed offset (van Oost et al., 2007, Doetterl et al., 2015) as temporary storage (hence, aging) of vascular plant OC prior to export and accumulation in the ocean may introduce lags in climate and vegetation signals. This may be particularly relevant for the extreme age offset of LCFA observed in selected sediment samples deposited during the last millennium. A shift in OC source may also contribute to the observed offset. The continuous increase in δ13C points to intensification of aridity in the late Holocene. Under such conditions, previously stored OC may be re-exposed due to lack of vegetation cover leading to an increase in OC degradation (e.g., Jacinthe and Lal, 2001). This represents a feedback loop whereby the relatively labile fractions are oxidized leaving the old, recalcitrant fractions behind. Ponton (2012) explored this scenario by mixing young and old OC with variable 14C age to simulate the observed age in core 16A. His model results show that more than 48% of aged OC fraction, with a nominal Δ14C = - 460‰, is required to produce the ~2000 years offset in this study. Usman et al (2018) 59 observed a Δ14C =value of -340‰ for bulk OC in the deepest (and oldest) soil collected within the modern Godavari basin. Using this Δ14C value, approximately 65% of this “old” fraction would be required to generate the observed age offset, which is a rather unlikely scenario as sediment export from the Godavari has reduced significantly over the past century (Balakrishna and Probst, 2005; Pradhan et al., 2014; Fig. 2b). However, the lower section of the Godavari drains one of the largest coal belts in peninsular India (Prachiti et al., 2011). These Permian-Triassic rocks, subaerially exposed in the Odisha region of the lower Godavari (Acharya, 2000), are “14C-dead”. The extent of fossil OC contribution from the Godavari to the adjacent margin is poorly understood but using the mixing approach employed here, a “modest” 21% mixing of 14C-dead OC with 79% of fresh/modern OC, with a nominal Δ14C = 0‰, would produce a ~2000 years offset. Mixing with 14C-dead OC has been well documented in other basin systems (e.g. Drenzek et al., 2009; Tao et al., 2016; Galy et al., 2008, Gustaffson et al., 2011, Raymond and Bauer, 2001). This implies that a meager addition of 14C- dead OC to the relatively aged OC emanating from deep soil erosion within the Godavari can potentially lead to significant age offset between the sediment OC age and depositional age. The drawback with this however, is that fossil OC contribution from the coal belts would only explain the bulk OC age offset and not LCFA offsets as these sedimentary rocks are likely devoid of LCFA. Furthermore, the largest age offsets are observed in the late Holocene and during this period, Neodymium isotope (Giosan et al., 2017), pollen assemblages (Zorzi et al., 2015), as well as magnetic mineral assemblages (Cui et al., 2017) all point towards the upper basin as the dominant source of pre-aged OC. Therefore, several facets of the terrestrial OC cycle (especially pre-aged OC sources and contributions) need to be carefully assessed and end members better constrained. Overall, the mobilization and transport processes within the basin (surficial runoff versus deep gully erosion) may have profound implications for organic carbon cycling and transfer dynamics. The mixing estimates here show that fluctuations in the source and type of aged OC pools resulting from climate variability and anthropogenic activity can impact the terrestrial residence times of OC in the drainage basin and hence, the carbon cycle. Such factors require further consideration in the face of ongoing changes in climate and human activity.

60

3.5. Conclusions

This detailed study of core 16A from the mouth of the Godavari River has revealed insights into past variations in the isotopic composition of OC exported from peninsular India during the Holocene. Large-scale changes in continental climate and its impact on OC provenance and transport are evident from carbon isotopic composition of bulk OC and LCFA, which together reveal an increase in the flux and age of terrestrial OC delivered to the Bay of Bengal during the late Holocene. This suggests that changing climate (i.e., increased aridity) have impacted OC cycling, primarily through shifts in mobilization pathways of terrestrial OC within the drainage basin. A coupled isotopic mixing model based on bulk and compound-specific δ13C and Δ14C was employed to estimate the fractional contribution of different OC types in the core sediment. The model results show that during the early Holocene, the majority (>54%) of OC pool is dominated by a modern/contemporary fraction (combined terrestrial and marine) suggesting a shorter residence times and rapid OC transfer from source to margin. In contrast, the late Holocene is dominated by pre-aged (46%) terrestrial OC. This suggests a longer terrestrial residence time and/or enhanced exhumation of old soil OC/petrogenic OC as a result of the intensification of erosion during arid climate that pervades the later part of the Holocene. This erosion may have been further exacerbated by anthropogenic activity (agriculture) during the last 2000 years. The findings presented here have implications for understanding climate-carbon cycle interactions, as well as for paleoclimate reconstructions based on fatty acids and other terrestrial proxies.

Acknowledgements

We would like to thank Daniel Montlucon for his laboratory assistance. We also acknowledge the ETH AMS laboratory member for their help with radiocarbon and bulk stable isotope measurements. This study was funded by the Swiss National Science Foundation (“CAPS LOCK” Grant no: 200021-140850 and “CAPS LOCK 2” Grant no: 200021-163162). This manuscript benefitted from discussions with Thomas Blattmann.

61

References

Acharya, S. K. (2000) Coal and lignite resources of India- an overview. Geological Society of India, Bangalore. p. 50. Allchin, F. R. (1995) Early cities and states beyond the Ganges valley, in: Allchin, F. R. (Eds.), The Archaeology of Early Historic South Asia: The Emergence of Cities and States. Cambridge University Press, Cambridge, pp 123-151. Aller, R.C., Blair, N.E. (2006) Carbon remineralization in the Amazon-Guianas tropical mobile mudbelt: a sedimentary incinerator. Continental Shelf Research 26, 2241- 2259. Asouti, E. and Fuller, D. (2008) Trees and Woodlands in South India: Archaeological Perspectives. Left Coast Press, Walnut Creek, California. Balakrishna, K. and Probst, J.L. (2005) Organic carbon transport and C/N ratio variations in a large tropical river: Godavari as a case study, India. Biogeochemistry 73, 457-473. Berner, R. (1989) Burial of organic carbon and pyrite sulfur in the modern ocean: its geochemical and environmental significance. American Journal of Sciences 282, 451- 473. Blair, N.E., Leithold, E.L. and Aller, R.C. (2004) From bedrock to burial: the evolution of particulate organic carbon across coupled watershed-continental margin systems. Marine Chemistry 92, 141-156. Burdige, D.J. (2005) Burial of terrestrial organic matter in marine sediments: A re- assessment. Global Biogeochemical Cycles 19, n/a-n/a. Chikaraishi, Y., Naraoka, H. and Poulson, S.R. (2004a) Carbon and hydrogen isotopic fractionation during lipid biosynthesis in a higher plant (Cryptomeria japonica). Phytochemistry 65, 323-330. Chikaraishi, Y., Naraoka, H. and Poulson, S.R. (2004b) Hydrogen and carbon isotopic fractionations of lipid biosynthesis among terrestrial (C3, C4 and CAM) and aquatic plants. Phytochemistry 65, 1369-1381. Collett, T., Boswell, R., Cochran, J., Kumar, P., Lall, M., Mazumdar, A., Raman, M., Ramprasad, T., Riedel, M., Sain, K., Sathe, A., Vishwanath, K. and Party, N.E. (2014) Geologic implications of gas hydrates in the offshore of India: Results of the National Gas Hydrate Program Expedition 01. MArine and Petroleum Geology 58, 3-28.

62

Collister, J., Rieley, G., Stern, B., Eglinton, G. and Fry, B. (1994) Compound-specific d13C analyses of leaf lipids from plants with differing carbon dioxide metabolisms. Organic Geochemistry 21, 619-627. Central Water Commission, CWC India. (2010) Integrated Hydrological Data Book (Non-classified river basins). Dickens A. F., Gelinas, Y., Masiello, C. A., Wakeham, S., Hedges, J. I. (2004) Reburial of fossil organic carbon in marine sediments. Nature 427, 336-339. Drenzek, N. J. (2007) The temporal dynamics of terrestrial organic matter transfer to the oceans: initial assessment and application. Ph.D Thesis, MIT/WHOI Joint Program in Oceanography/Applied Ocean Science and Engineering. Drenzek, N.J., Montluçon, D.B., Yunker, M.B., Macdonald, R.W. and Eglinton, T.I. (2007) Constraints on the origin of sedimentary organic carbon in the Beaufort Sea from coupled molecular 13C and 14C measurements. Marine Chemistry 103, 146- 162. Drenzek, N.J., Hughen, K.A., Montlucon, D.B., Southon, J.R., dos Santos, G.M., Druffel, E.R.M., Giosan, L. and Eglinton, T.I. (2009) A new look at old carbon in active margin sediments. Geology 37, 239-242. Dutta, K., Bhushan, R. and Somayajulu, B.L.K. (2001) ΔR Correction Values for the Northern Indian Ocean. Radiocarbon 43, 483-488. Eglinton, G. and Hamilton, R.J. (1967) Leaf epicuticular waxes. Science 156, 1322- 1335. Eglinton, T., Aluwihare, L., Bauer, J., Druffel, E. and McNichol, A. (1996) Gas Chromatographic Isolation of Individual Compounds from Complex Matrices for Radiocarbon Dating. Analytical Chemistry 68, 904-912. Eglinton, T.I., Benitez-Nelson, B.C., Pearson, A., McNichol, A.P., Bauer, J.E. and Druffel, E.R.M. (1997) Variability in radiocarbon ages of individual organic compounds from marine sediments. Science 277, 796-799. Eglinton, T.I. and Eglinton, G. (2008) Molecular proxies for paleoclimatology. Earth and Planetary Science Letters 275, 1-16. Eglinton, T.I., Eglinton, G., Dupont, L., Sholkovitz, E.R., Montlucon, D. and Reddy, C.M. (2002) Composition, age, and provenance of organic matter in NW African dust over the Atlantic Ocean. Geochemistry Geophysics Geosystems 3. Feng, X., Vonk, J.E., van Dongen, B.E., Gustafsson, O., Semiletov, I.P., Dudarev, O.V., Wang, Z., Montlucon, D.B., Wacker, L. and Eglinton, T.I. (2013) Differential 63 mobilization of terrestrial carbon pools in Eurasian Arctic river basins. Proc Natl Acad Sci U S A 110, 14168-14173. Fry, B. and Sherr, E. (1984) δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contributions in Marine Science 27, 13-47. Gadgil, S. (2003) The Indian monsoon and its variability. Annual Review of Earth and Planetary Sciences 31, 429-467. Galy, V. and Eglinton, T. (2011) Protracted storage of biospheric carbon in the Ganges-Brahmaputra basin. Nature Geoscience 4, 843-847. Galy, V., France-Lanord, C. and Lartiges, B. (2008) Loading and fate of particulate organic carbon from the Himalaya to the Ganga-Brahmaputra delta. Geochimica Et Cosmochimica Acta 72, 1767-1787. Giosan, L., Ponton, C., Usman, M., Blusztajn, J., Fuller, D.Q., Galy, V., Haghipour, N., Johnson, J.E., McIntyre, C., Wacker, L. and Eglinton, T.I. (2017) Short communication: Massive erosion in monsoonal central India linked to late Holocene land cover degradation. Earth Surface Dynamics 5, 781-789. Goni, M.A., Ruttenberg, K.C. and Eglinton, T.I. (1997) Source and contribution of terrigenous organic carbon to surface sediments in the Gulf of Mexico. Nature 389, 275-278. Griffith, D., Barnes, R. and Raymond, P. (2009) Inputs of fossil carbon from wastewater treatment plants to U.S. River and Oceans. Environmental Science and Technology 43, 5647-5651. Gustafsson, O., van Dongen, B.E., Vonk, J.E., Dudarev, O.V. and Semiletov, I.P. (2011) Widespread release of old carbon across the Siberian Arctic echoed by its large rivers. Biogeosciences 8, 1737-1743. Hedges, J. and Keil, R. (1995) Sedimentary oranic matter preservation: an assessment and speculative synthesis. Marine Chemistry 49, 81-115. Hedges, J., Keil, R. and Berner, R. (1997) What happens to terretsrial organic matter in the ocean. Organic Geochemistry 27, 195-212. Hedges, J. and Oades, J. (1997) Comparative organic geochemistries of soils and marine sediments. Organic Geochemistry 27, 319-361. Hedges, J.I. (1992) Global Biogeochemical Cycles - Progress and Problems. Marine Chemistry 39, 67-93.

64

Hedges, J.I., Clark, W.A., Quay, P.D., Richey, J.E., Devol, A.H. and Santos, U.D. (1986) Compositions and Fluxes of Particulate Organic Material in the Amazon River. Limnology and Oceanography 31, 717-738. Hill, C. (2008) South Asia – An environmental history. ABC-Clio Incorporation, Santa Barbara, California. Houghton, R.A. (2007). Balancing the global carbon budget. Annual Review of Earth and Planetary Science 35, 313-347. Jacinthe, P. A., Lal, R. (2001) A mass balance approach to assess carbon dioxide evolution during erosional events. Land Degradation and Development 12, 329-339. Jasper, J.P., Gagosian, R.B. (1993) The relationship between sedimentary organic carbon isotope composition and organic biomarker compound composition. Geochimica et Cosmochimica Acta 57, 167-186. Kale, V.S. (2002) Fluvial geomorphology of Indian rivers: an overview. Progress in Physical Geography 26, 400-433. Keil, R.G., Mayer, L.M., Quay, P.D., Richey, J.E. and Hedges, J.I. (1997) Loss of organic matter from riverine particles in deltas. Geochimica et Cosmochimica Acta 61, 1507-1511. Komada, T., Druffel, E. R. M., Hwang, J. (2005). Sedimentary rocks as sources of ancient organic carbon to the ocean: an investigation through D14C and d13C signatures of organic compound classes. Global Biogeochemical Cycles 19, GB2017 Krishna, M.S., Naidu, S.A., Subbaiah, C.V., Sarma, V.V.S.S. and Reddy, N.P.C. (2013) Distribution and sources of organic matter in surface sediments of the eastern continental margin of India. Journal of Geophysical Research: Biogeosciences 118, 1484-1494. Kusch, S., Rethemeyer, J., Schefuß, E. and Mollenhauer, G. (2010) Controls on the age of vascular plant biomarkers in Black Sea sediments. Geochimica et Cosmochimica Acta 74, 7031-7047. Leithold, E.L., Perkey, D.W., Blair, N.E. and Creamer, T.N. (2005) Sedimentation and carbon burial on the northern California continental shelf: the signatures of land-use change. Continental Shelf Research 25, 349-371. Leithold, E. L., Blair, N. E., Perkey, D. W. (2006). Geomorphic controls on the age of particulate organic carbon from small mountainous and upland river. Global Biogeochemical Cycles 20, GB00267

65

Longworth, B.E., Petsch, S.T., Raymond, P.A. and Bauer, J.E. (2007) Linking lithology and land use to sources of dissolved and particulate organic matter in headwaters of a temperate, passive-margin river system. Geochimica Et Cosmochimica Acta 71, 4233-4250. Ludwig, W., Probst, J.L. and Kempe, S. (1996) Predicting the oceanic input of organic carbon by continental erosion. Global Biogeochemical Cycles 10, 23-41. Marwick, T., Tamooh, F., Teodoru, C., Borges, V., Darchambeau, F., Bouillon, S. (2015). The age of river-transported carbon: A global perspective. Global Biogeochemical Cycle 29, 122-137. Mayorga, E., Aufdenkampe, A.K., Masiello, C.A., Krusche, A.V., Hedges, J.I., Quay, P.D., Richey, J.E., Brown, T.A. (2005) Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers. Nature 436, 538-541. McIntyre, C., Wacker, L., Haghipour, N., Blattmann, t., Fahrni, S., Usman, M., Eglinton, T. and Synal, H. (2016) Online 13C and 14C gas measurement by EA-IRMS-AMS at ETH Zürich. Radiocarbon, 1-11. Mollenhauer, G. and Eglinton, T.I. (2007) Diagenetic and sedimentological controls on the composition of organic matter preserved in California Borderland Basin sediments. Limnology and Oceanography 52, 558-576. Mook, W. and van der Plicht, J. (1999) Reporting 14C activities and concentrations. Radiocarbon 41, 227-239. Ohkouchi, N., Eglinton, T.I., Hughen, K.A., Roosen, E. and Keigwin, L.D. (2005) Radiocarbon Dating of Alkenones from Marine Sediments: III. Influence of Solvent Extraction Procedures on 14C Measurements of Foraminifera. Radiocarbon 47, 425- 432. Pearson, A., Eglinton, T.I. and McNichol, A.P. (2000) An organic tracer for surface ocean radiocarbon. Paleoceanography 15, 541-550. Ponton, C., Giosan, L., Eglinton, T.I., Fuller, D.Q., Johnson, J.E., Kumar, P. and Collett, T.S. (2012) Holocene aridification of India. Geophysical Research Letters 39, 6. Ponton, C. (2012) Aridification of the Indian subcontinent during the Holocene: Implications for landscape evolution, sedimentation, carbon cycle, and human civilizations. Ph.D Thesis, MIT/WHOI Joint Program in Oceanography/Applied Ocean Science and Engineering.

66

Prachiti, P.K., Manikyamba, C., Singh, P.K., Balaram, V., Lakshminarayana, G., Raju, K., Singh, M.P., Kalpana, M.S. and Arora, M. (2011) Geochemical systematics and precious metal content of the sedimentary horizons of Lower Gondwanas from the Sattupalli coal field, Godavari Valley, India. International Journal of Coal Geology 88, 83-100. Pradhan, U.K., Wu, Y., Shirodkar, P.V., Zhang, J. and Zhang, G.S. (2014) Multi-proxy evidence for compositional change of organic matter in the largest tropical (peninsular) river basin of India. Journal of Hydrology 519, 999-1009. Prasad, S. and Enzel, Y. (2006) Holocene Paleoclimates of India. Quaternary Research 66, 442-453. Raymond, P. and Bauer, J. (2001) Use of 14C and 13C natural abundances for evaluating riverine, estuarine, and coastal DOC and POC sources and cycling: a review and synthesis. Organic Geochemistry 32, 469-485. Raymond P.A., Bauer, J. E., Caraco, N. F., Cole, J. J., Longworth, B. E., Petsch, S. T. (2004) Controls on the variability of OM and dissolved inorganic carbon ages in northeast US rivers. Marine Chemistry 92, 353-366 Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B., Buck, C.E., Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Haflidason, H., Hajdas, I., Hatté, C., Heaton, T.J., Hoffmann, D.L., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M. and van der Plicht, J. (2013) IntCal13 and Marine13 Radiocarbon Age Calibration Curves 0– 50,000 Years cal BP. Radiocarbon 55, 1869-1887. Ruff, M., Fahrni, S., Gäggeler, H.W., Hajdas, I., Suter, M., Synal, H.A., Szidat, S. and Wacker, L. (2010) On-line Radiocarbon Measurements of Small Samples Using Elemental Analyzer and MICADAS Gas Ion Source. Radiocarbon 52, 1645-1656. Sackett, W. (1989) Stable carbon isotope studies on organic matter in the marine environment, in: Fritz, P., Fontes, J. (Eds.), Handbook of Environmental Isotope Geochemistry. Elsevier, New York, pp. 139-169. Santos, G.M., Southon, J.R., Drenzek, N.J., Ziolkowski, L.A., Druffel, E., Xu, X., Zhang, D., Trumbore, S., Eglinton, T.I. and Hughen, K.A. (2010) Blank Assessment for Ultra-Small Radiocarbon Samples: Chemical Extraction and Separation Versus AMS. Radiocarbon 52, 1322-1335.

67

Schlünz, B. and Schneider, R. (2000) Transport of terrestrial organic carbon to the oceans by rivers: re-estimating flux-and burial rates. International Journal of Earth Sciences 88, 599-606. Schouten, S., Breteler, W., Blokker, P., Schogt, N., Rijpstra, W., Grice, K., Baas, M. and Sinninghe-Damsté, J.S. (1998) Biosynthetic effects on the stable carbon isotopic compositions of algal lipids: Implications for deciphering the carbon isotopic biomarker record. Geochimica et Cosmochimica Acta 62, 1397. Shah, S.R. and Pearson, A. (2007) Ultra-Microscale (5–25 μg C) Analysis of Individual Lipids by 14C AMS: Assessment and Correction for Sample Processing Blanks. Radiocarbon 49, 69-82. Smith, B. and Epstein, S. (1971) Two categories of 13C/12C ratios for higher plants. Plant Physiology 47, 380-384. Stuiver, M. and Polach, H.A. (1977) Reporting of C-14 Data - Discussion. Radiocarbon 19, 355-363. Subramanian, V. (1987). Environmental geochemistry of Indian river basins – a review. Geological Society of India 29, 205-220 Tao, S., Eglinton, T.I., Montluçon, D.B., McIntyre, C. and Zhao, M. (2016) Diverse origins and pre-depositional histories of organic matter in contemporary Chinese marginal sea sediments. Geochimica et Cosmochimica Acta 191, 70-88. Tao, S.Q., Eglinton, T.I., Montlucon, D.B., McIntyre, C. and Zhao, M.X. (2015) Pre- aged soil organic carbon as a major component of the Yellow River suspended load: Regional significance and global relevance. Earth and Planetary Science Letters 414, 77-86. Trumbore, S., Vogel, J. and Southon, J. (1989) AMS 14C measurements of fractionated soil organic matter: an approach to deciphering the soil carbon cycle Radiocarbon 31, 644-654. Usman, M.O., Kirkels, F.M.S.A., Zwart, H.M., Basu, S., Ponton, C., Blattmann, T.M., Ploetze, M., Haghipour, N., McIntyre, C., Peterse, F., Lupker, M., Giosan, L. and Eglinton, T.I. (2018) Reconciling drainage and receiving basin signatures of the Godavari River system. Biogeosciences 15, 3357-3375. Volkman, J., Barrett, S., Blackburn, S., Mansour, M., Sikes, E. and Gelin, F. (1998) Microbial biomarkers: A review of recent developments. Organic Geochemistry 29, 1163. 68

Vonk, J.E., van Dongen, B.E. and Gustafsson, O. (2008) Lipid biomarker investigation of the origin and diagenetic state of sub-arctic terrestrial organic matter presently exported into the northern Bothnian Bay. Marine Chemistry 112, 1-10. Wu, G., Liu, Y., He, B., Bao, Q., Duan, A. and Jin, F.F. (2012) Thermal controls on the Asian summer monsoon. Sci Rep 2, 404. Zorzi, C., Goni, M.F.S., Anupama, K., Prasad, S., Hanquiez, V., Johnson, J. and Giosan, L. (2015) Indian monsoon variations during three contrasting climatic periods: The Holocene, Heinrich Stadial 2 and the last interglacial-glacial transition. Quaternary Science Reviews 125, 50-60.

69

Supplementary Information

Coupled carbon isotope mixing model

The relative fractional contributions of contemporary OC, pre-aged soil OC, and petrogenic OC to the bulk OC through the Holocene was assessed using an end member mixing model according to the following equations:

14 14 14 14 Δ COC = fM (Δ CM) + fS (Δ CS) + fF (Δ CF) (Eq. 1)

13 13 13 13 δ COC = fM (δ CM) + fS (δ CS) + fF (δ CF) (Eq. 2)

fM + fS + fF = 1 (Eq. 3)

The end members were constrained using δ13C and Δ14C values of compound specific biomarkers in individual sediment horizon. The specific compounds used for this quantification are well-established, source-specific biomarkers. In the case of δ13C of petrogenic end-member where we lack adequate data, values from Tao et al. 2016 were assumed for the petrogenic end member. An important drawback of this mixing 13 13 model is the large uncertainty associated with δ Cbiomass - δ Clipid. Therefore, solving 13 13 for equation 1-3 over a large range of δ Cbiomass - δ Clipid (i.e. 5-9%) generates over 10,000 unique solutions. Since each of the OC component must integrate to 1 according to equation 3, the end members are represented by a normal distribution and the mean and standard deviation yield the “measured” and uncertainty of the different components. The mean and standard deviation are calculated by eliminating values outside the limit of the condition set forth by equation 3 (i.e. 0 ≤ x ≤1).

70

Fig. S1: TOC normalized n-fatty acids abundance of selected samples plotted as a function of carbon number.

71

Chapter 4: Evolution of the Indian Summer Monsoon and its impact on the origin and fate of terrestrial organic matter in the Bay of Bengal

Muhammed O. Usman1, Camilo Ponton2, Negar Haghipour1,3, Maarten Lupker1, Liviu Giosan4, Timothy I. Eglinton1.

1Geological Institute, ETH Zürich, Sonneggstrasse 5, 8092 Zürich, Switzerland 2Division of Geological and Planetary Science, California Institute of Technology, 1200 East California Boulevard, Pasadena, 91125 California, USA 3Laboratory of Ion Beam Physics, ETH Zürich, Otto-Stern-Weg 5, 8093 Zürich, Switzerland 4Geology and Geophysics Department, Woods Hole Oceanographic Institution, 86 Water Street, Woods Hole, 02543 Massachusetts, USA

72

Abstract

Geochemical and sedimentological measurements were performed on two marine sediment cores (NGHP-01-19A and NGHP-01-05D) collected from the mouth of the Mahanadi and Krishna-Godavari Rivers, respectively in the Bay of Bengal. In combination with previous sedimentary records, they were used to create a high- resolution composite record of paleo-environmental variability over the central peninsular India spanning the last 10 ky based on stable and radiocarbon isotopic compositions of bulk and terrestrial plant long-chain fatty acids (LCFA) biomarkers. A strong positive correlation between bulk organic carbon (OC) and long-chain fatty acids (LCFA) biomarker isotopic compositions indicates that for both systems bulk OC in the sediments is almost exclusively terrestrially derived, and corresponding geochemical signatures reflect the adjacent continental paleoclimate. Comparison with regionally extensive paleoclimatic records reveal large scale changes in Holocene paleo-precipitation and paleo-vegetation resulting from changes in the summer monsoonal intensity over the Indian subcontinent. Sediments exported by these river systems to the Bay of Bengal during the early Holocene, including the early Holocene climatic optimum (~8 - 5 ky BP), are characterized by relatively high OC contents and overall predominance of C3-vegetation. In contrast, the late Holocene sediments exhibit low OC contents and reflect development of arid-adapted C4-vegetations. Older apparent 14C ages of OC and biomarkers in the later part of Holocene suggest a change in the relative proportion of fresh versus pre-aged OC and/or an increased terrestrial residence times within the basin. This indicates a tight coupling between Indian summer monsoon variability and continental carbon cycle dynamics. Exploration of our records within an archaeological context suggests close coupling between climate and anthropogenic modulation of OC cycling in central India since the demise of the Indus civilization approximately 5,000 years ago. This highlights the potential importance of humans as drivers of terrestrial OC mobilization and export from land to the ocean on the scale of large continental drainage basins.

73

4.1. Introduction

The Asian monsoon system affects more than 3 billion people and influences some of the top 20 rivers globally with the highest total sediment discharge to the oceans (Milliman and Meade, 1983; Milliman and Farnsworth, 2011), rendering the monsoon an important driver of terrigenous sedimentation in the northern Indian Ocean. The Asian summer monsoon, one of the most energetic components of the earth’s climate system, varies in intensity and generally subdivided into the Indian summer monsoon (ISM), East Asian summer monsoon (EASM), and western North Pacific summer monsoon (WNPSM) (Wang et al., 2001). The ISM represents a reversal in wind direction, a physical manifestation of seasonal migration of the intertropical convergence zone (ITCZ), driven by variations in insolation of the Indian subcontinent and Tibetan Plateau (Gadgil, 2003). The ISM (aptly named the southwest monsoon, after the direction of the prevailing surface winds), occurs during the northward migration of the ITCZ and results in high precipitation over Indian peninsula. Conversely, the opposite migration of the ITCZ (i.e., southward) produces dry condition on the peninsula. More than 80% of India’s annual precipitation occurs from June to September during which the Arabian Sea and Bay of Bengal (BoB) branches of the monsoon delivers moisture to the most of the Indian peninsula (Gadgil, 2003). The Western Ghats mountain range, acting as an orographic barrier, limits the penetration of Arabian Sea moisture towards the interior resulting in increased rainfall between the coast and the Ghats, and leaving regions located inland of the Ghats with less precipitation (Gunnel et al., 2007). Past changes in the monsoon have been investigated through a wide range of paleoenvironmental proxies such as ice cores (e.g. Liu et al., 1998; Thompson et al., 2000), tree rings (Hughes et al.,1994; Feng et al., 1999), corals (Tudhope et al., 1996; Charles et al., 2003), speleothems (Yadava and Ramesh, 2005; Fleitmann et al., 2007), lake sediments (Enzel et al., 1999; Prasad et al., 2014), and marine sediments (Weber et al., 1997; Ponton et al., 2012). The latter represent some of the most robust, and continuous archives of paleoclimatic information as organic geochemical analysis of lacustrine and marine sediments can produce high-resolution (centennial) records of environmental change. Given that most of the modern precipitation drains into the BoB, the “primary sink” of the ISM, paleo-records from the BoB are required for a detailed understanding of past changes

74 of the ISM. Here we present the past changes of ISM based on proxy records of continental climate from the BoB, spanning the last 12 ka. Molecular biomarkers are an increasingly common tool in the reconstruction of terrestrial climate as information encoded in the composition of plant-wax lipids has proven useful as a recorder of terrestrial ecosystems both in modern systems (e.g., Diefendorf et al., 2011; Freymond et al., 2018) and in the geologic past (e.g. Pancost and Boot, 2004; Eglinton and Eglinton, 2008). One efficacious approach is to obtain high-resolution terrestrial environmental records from marine sedimentary sequences (Bendle et al., 2010). Most sediment and associated terrestrial organic matter (OM) delivered to the continental margin occurs via fluvial transport, with river-dominated continental margins accounting for the majority of the global OM burial in ocean sediments globally (~90%), underlining their prominent role in the global carbon cycle (Hedges and Oades, 1997; Burdige et al., 2005). Owing to the integrative effects of fluvial systems, continental margin records can be a valuable proxy for terrestrial environmental conditions (e.g. continental paleo-vegetation). The BoB is a major depository of terrestrial sediments from the Indian subcontinent. Indeed, biomarkers of exclusively terrestrial origin (e.g. long chain n-alkanes, fatty acid, etc.) have been found in abundance at the core locations (Fig. 4.1) (e.g., Ponton et al., 2012; Usman et al., 2018). A particular advantage of this approach is that the terrestrial material sequestered in the marine sedimentary archive reflects a regional integration of the terrestrial ecosystem (Traverse, 1989).

75

Figure 4.1. A map of peninsular India showing the major rivers and adjacent oceans. The location of cores 19A and 05D are indicated in red whereas 16A (previous chapter) and SO93 (Hein et al., 2017) are shown in black.

Recent studies have shown that intensification of aridification of peninsular India, beginning ~4.2 ky BP (Zorzi et al., 2015; Dixit et al., 2014), was accompanied by human migration from the Indus region to the towards the Ganga plain (Gupta et al., 2006). The expansion of human settlements in central and south India, which occurred after the collapse of the Harappan civilization, led to adaptation of agriculture to a semi-arid climate in response to droughts in the northwestern part of the Indian subcontinent (Ponton et al., 2012; Giosan et al., 2017). This large-scale human adaptation was likely brought about by changes in monsoon intensity that regulates flood and drought cycles in peninsular India (Asouti and Fuller, 2008). Changes in

76 monsoon intensity during the Holocene have also affected the amount of sediment discharged from the continent (Goodbred and Kuehl, 2000) and the degree of OM degradation (Pattan et al., 2013). Siliciclastic sediment delivery exerts major controls on OM burial along continental margins by providing fine-grained materials with which OM is closely associated (Mayer, 1994; Keil and Mayer, 2014). OM sorption to mineral surfaces of clays has been shown to be a key factor in the ultimate preservation of OM (Mayer 1994; Kennedy et al., 2002). Such OM-mineral interaction is prevalent in the clay-silt size fractions and is believed to occur through a variety of mechanisms ranging from adsorption of OM onto individual clay particles to aggregation mechanisms in the silt size range (Keil and Mayer, 2014). Furthermore, the nature of the exported terrigenous OM has a strong influence on the composition of OM buried in adjacent continental margin sediments (Burdige, 2007; Blair et al., 2004). Such terrigenous OM derives from three primary sources: debris of recently synthesized biomass, soils, and sedimentary rocks (Hedges et al., 1986). Modern plant debris and soil OM are termed here as terrigenous “biospheric” OM, while carbon eroded from outcropping sedimentary rocks as “fossil/petrogenic” OM (cf. Galy and Eglinton, 2011). These components differ in their content of relatively refractory OM. The fractional abundance of each component exported to the oceans depends on a range of factors, including the nature and degree of upstream processing that modifies the OM during transfer through major bioactive reservoirs such as upland soils, floodplains, and channel infills (Blair et al., 2004). The degree of OM modification is also largely dependent on the terrestrial residence times of OM within these different reservoirs (Drenzek et al., 2009). Terrestrial residence times are controlled by multiple factors, including the geomorphic properties (e.g. topography), which are largely related to the geodynamic setting of the drainage margin (active versus passive; Leithold et al., 2006). In contrast to active continental margins, passive continental margins are characterized by low relief and long distance between the continental divide and the shoreline. Thus, fluvial OM transports in passive continental margins tend to have longer terrestrial residence times and are subjected to extensive reworking of biospheric and petrogenic OM within the continental reservoirs (e.g. Bouchez et al., 2010). As a consequence, the final composition of OM exported to the margins are modified and consist mostly of the highly refractive fractions of OM.

77

OM in continental margins contains a large fraction of carbon fixed from the atmosphere; therefore, its source and fate on millennial timescales is important for long-term carbon budgets. The Krishna-Godavari and Mahanadi river systems are considered regionally important in terms of OM export to the deep ocean (being two of the largest non-Himalayan rivers in India), but little is known on the long-term storage of organic carbon (OC) in sediments from these rivers, or about the relationships between terrestrial OC export/storage and monsoonal dynamics. In this study, a detailed history of OC discharge from these rivers since the Younger Dryas is presented. The overall objective of this study is to determine the provenance of OC and the primary mechanisms controlling its deposition and variability throughout the past 12 ka. Specific aims include: (i) the effects of Krishna-Godavari and Mahanadi fluvial discharge on the supply of terrestrial OM, and (ii) evaluating ISM impact on the regulation of terrestrial versus marine OC to the margins.

4.2. Materials and Methods

4.2.1. Oceanographic Setting

The present-day climate of the Indian peninsula and surface ocean circulation in the Bay of Bengal are primarily controlled by the ISM (Schott et al., 2009). Monsoonal rains induce a significant water and sediment discharge from the Krishna-Godavari and Mahanadi rivers. A positive precipitation over evaporation ratio as well as extensive fluvial discharge to the BoB result in water column stratification and the formation of a 15-20 m thick boundary layer that persists through and beyond the summer monsoon season (Madhu et al., 2006). Sediment trap data reveal that fluxes of lithogenic sediment, OC, and biogenic opal peak around late May, because of increased fluvial discharge that fuels primary production driven by diatoms (Stoll et al., 2007). The high flux of OC during this period has been attributed to increase in primary productivity and ballasting effects of lithogenic particles (Stoll, 2007). An oxygen minimum zone (OMZ) has been observed in the BoB (Singh et al., 2011), characterized by dissolved oxygen concentrations <10 µM with an upper (lower) boundary at 70-120 m (250-500 m), deepening towards the north and presenting maximum expansion of (~450 m) during the summer monsoon (Sarma et al., 2013). The occurrence of the OMZ has been attributed to the export of low oxygen

78 intermediate waters from the Arabian Sea into the BoB, strong stratification due to the fluvial freshwater plume, and increase in productivity due to injection of nutrient laden water into the photic zone (Sarma et al., 2013). Nonetheless, aerobic respiration rates within the BoB water column are lower than expected, which can be attributed to the rapid sinking of sediment ballasted OM resulting from increased terrestrial input (Naqvi et al., 1996).

4.2.2. Mahanadi Basin

The Mahanadi Basin is a pericratonic basin on the eastern flank of peninsular India formed as a result of the rifting of India and Australia during the Jurassic (Subrahmanyam et al., 2008). The Mahanadi River drains a total area of 8.8•104 km2 with an annual run-off and sediment discharge of 5.5•107 m3 and 3.1•107 tons, respectively (Subramanian, 1987). Suspended sediment discharged by the Mahanadi to the BoB is characterized by kaolinite, quartz, chlorite, and minor smectite and illite with more than 90% of the discharge occurring during the Indian summer monsoon (Chakrapani and Subramanian, 1990). The Mahanadi River drains the Precambrian rocks of the Eastern Ghats province as well as volcanic rocks of the eastern Deccan Plateau (Rickers et al., 2001). A 300m long sediment core, spanning Holocene to Late Pliocene, was recovered during the National Gas Hydrate Project (NGHP-01-19A, 18.97761°N, 85.65867°E, 1422 m water depth; hereafter referred to as core 19A; Fig. 4.1) and are comprised of nannofossil bearing hemipelagic clays, authigenic sulfides, volcanic glass, and plant debris (Collett et al., 2008).

4.2.3. Krishna-Godavari Basin

Like the Mahanadi Basin, the Krishna-Godavari Basin is pericratonic basin formed by Jurassic rifting of India and Australia (Gupta, 2006). The Krishna and Godavari Rivers drain a combined area of 5.65•105 km2 made up largely of Deccan Trap basalts and Precambrian metamorphic rocks (Rao and Kessarkar, 2001). The rivers deliver an estimated annual run-off and sediment discharge of 1.25•107 m3 and 1.74•107 tons, respectively, to the BoB (Subramanian, 1987). The sediments load is rich in smectite with minor feldspars, quartz, kaolinite and illite (Subramanian 1980; Usman et al.,

79

2018). Turbidites, slump deposits, and debris flows are common within the basin (Ramprasad et al., 2011). A sediment core NGHP-01-05D (16.02875°N, 82.04469°E, 925 m water depth; hereafter referred to as core 05D; Fig. 4.1) was recovered during the National Gas Hydrate Project, and are comprised of hemipelagic clays bearing foraminifera, silt/sand lamina beds, authigenic carbonates, and iron sulfides (Collett et al., 2008; Riedel et al., 2011). The portion of cores 19A and 05D sampled in this study (0-320 cm and 0-1500 cm, respectively) cover the entire Holocene, and lithostratigraphic characterizations show that mass transport events played no role in sedimentation at the two sites for the periods investigated in this study (Ponton et al., 2012; Phillips et al., 2014). The narrow continental shelf on the western part of BoB (Rao et al., 2012) reduced the effect of sea level changes during the Holocene (Zorzi et al., 2015), thus no large variations in the distance between the river mouths and the coring location were expected. Therefore, we consider that Mahanadi and Krishna-Godavari catchment remained identical during the Holocene.

4.3. Methods

4.3.1. Sampling

The sediment cores were stored refrigerated at the Woods Hole Oceanographic Institution (WHOI). A total of 70 (core 19A) and 90 (core 05D) samples were collected for sedimentological and bulk OC parameters. A subset of 11 and 14 targeted samples from cores 19A and 05D, respectively, were targeted for compound specific analysis and age model calibration.

4.3.2. Age Model

The age model calibration details are similar to those described for core 16A in the previous chapter. The ages at the bottom of cores 19A and 05D are 22.3 ky BP and 12.1 ky BP (Fig. 4.2), respectively. This corresponds to an average sampling resolution of ~360 and ~170 y for cores 19A and 05D, respectively. Linear sedimentation rate (LSR) values were calculated based on the calibrated ages.

80

Figure 4.2. Age-Depth model for cores 19A (a) and 05D (b). Uncertainties associated with age calibration are shown as a grey envelope enclosing the data points.

81

4.3.3. Grain size and mineral surface area

Sample preparation procedures for grain size and surface area analyses followed the same protocol as those previously reported in Usman et al (2018).

4.3.4. Bulk elemental and isotopic measurements

The sample preparation and analysis were conducted as previously described in McIntyre et al (2016) and Usman et al (2018). The analytical errors were 0.02% and 0.03 for total organic carbon (TOC) and 13C, respectively. The absolute error on the 14C measurements range from 0.0078 to 0.0343 (Fm), which corresponds to 105 to 1600 in radiocarbon years. The majority (>87%) of samples exhibited an absolute error of less than 0.009 (Fm)

4.3.5. Plant wax fatty acid extraction

Fatty acid extraction and quantification procedures have been described in Usman et al (2018) and in the previous chapter.

4.3.6. Compound-specific 13C and 14C measurements

A ca. 5% aliquot of purified fatty acid was set aside for 13C determination. Stable carbon isotopic analysis was performed using a gas chromatograph (GC, Thermo Scientific Trace GC Ultra, equipped with an Agilent VF-1ms column, 60 m x 250 μm ID, film thickness: 0.25 μm) coupled to an isotope ratio mass spectrometer (IRMS, Thermo Scientific Delta-V Plus). The 13C values were corrected for the added methyl group during methylation. Units are report ‰ notation relative to VPDB. The remainder of the purified FAMEs was isolated by preparative capillary gas chromatography (PCGC; Eglinton et al., 1996). Each sample was injected 30-50 times to ensure sufficient amounts of carbon for 14C measurements. 10% of the injected sample went into the detector for retention time and purity control while the remainder was transferred to a fraction collector and individual fatty acid n-C16, n-C24, n-C26, n-

C28, and combined n-C18+20+22, n-C30+32 were collected in single glass traps. The traps were eluted with 1 mL DCM and passed over 2 cm silica column to remove potential column bleed. The purity and quantification of the isolated compounds was performed on a GC-FID. The samples were then transferred to DCM-rinsed and dried tin capsules optimized for measuring liquid samples. The solvent was dried over a 40°C hot plate

82 and neatly wrapped for 14C analysis. 14C was measured on MICADAS at the ETH- Zurich laboratory of Ion Beam Physics. 14C values were corrected for the added methyl group during methylation. Blank assessments, error propagations and analysis are described in details in the previous chapter and in Tao et al (2015).

4.4. Results

4.4.1. Linear sedimentation rates

Cores 19A and 05D encompass ~12 and ~23 ky of deposition, respectively, with long- term average sedimentation rates of 14 and 125 cm/ky, respectively. Variable linear sedimentation rates can be observed in both cores during the Holocene (Fig. 4.3). The LSR values in the Mahanadi core range from ~10 cm/kyr prior to mid-Holocene (~5 ky BP) to ~30 cm/kyr, with peak LSR values (reaching up to ~80 cm/kyr) observed in the last 2000 years (Fig. 4.3a). LSR values in the Krishna-Godavari are higher than the Mahanadi with values ranging from ~100 cm/kyr in the early Holocene to ~260 cm/kyr in the later part of the Holocene. Similar to the Mahanadi, the Krishna-Godavari also exhibit peak LSR value (up to ~900 cm/kyr) in the last 2000 years (Fig. 4.3b). These values represent an order of magnitude difference in sedimentation between the Mahanadi and the Krishna-Godavari, as well as between the early and late Holocene.

83

Figure 4.3. Linear sedimentation rates (LSR) for cores 19A (a) and 05D (b). 84

4.4.2. Grain size distribution and mineral-specific surface area.

Grain-size distributions of siliciclastic particles in the Mahanadi and Krishna-Godavari cores range from 0.2 to 24 μm and the median grain size (MGS) varies from 2.5 to 6 μm (mean = 4 μm; Fig. 4.3a). The Mahanadi exhibits a higher MGS compared to the Krishna-Godavari. In general, the median grain size does not exhibit variations that are systematically correlated with other sedimentological and geochemical parameters. Application of an inversion algorithm shows a two end-member model that accounts for more than 80% of the variance, with the silt end member with size range 2-63 μm (modal grain size of 5.4 μm) and a clay end member with size range 0.2-2 μm (modal grain size of 1.36 μm). Proportionally, the silt and clay end member contribute 75 and 23 % to total distribution, indicating that the median grain size is strongly influenced by the proportion of the silt fraction. Similar to grain size, mineral surface area (MSA) data show significant variability throughout the Holocene with values ranging from 46 to 75 m2/g (average= 59 m2/g). However, unlike grain size characteristics, surface area values display systematic temporal variability with relatively high values during the early Holocene that decrease gradually through the later part of the Holocene. The MSA values in the Krishna- Godavari is generally higher than those of the Mahanadi (Fig. 4.3b).

4.4.3. Bulk organic and biomarker characteristics

The organic geochemical characteristics of sediments from both core locations reveal notable spatial and temporal variations. In core 05D, bulk TOC contents range from 0.88 to 2.25% with relatively higher values in the early Holocene (average, 1.92%; n = 19) than the late Holocene (average, 1.29% n = 71). In contrast to 05D, core 19A exhibits a narrower range of variability in TOC contents, varying from 1.03 to 1.29% with no systematic variation between the early and late Holocene (Fig. 4.3c). 13 13 The δ C values of bulk OC (δ COC) vary between -20.78 to -17.17 ‰ for Krishna- Godavari (average = -19.12 ‰). For the Mahanadi, the values range from -21.09 to - 13 19.11 ‰ (average = -20.63 ‰). Both records display remarkable similarity in δ COC 13 through the Holocene until the last 1,000 years. There is a gradual increase in δ COC beginning from the early Holocene (~10,000 yr BP) punctuated by a reversal around the mid Holocene (4,800 and 4,500 years for Mahanadi and Krishna-Godavari,

85 respectively). However, in the last 1,000 yr BP core 19A recorded a sudden decrease 13 in δ COC values towards the present while core 05D show a sudden decrease 13 followed by an increase in δ COC towards the present (Fig. 4.3d). 14 Bulk uncorrected radiocarbon compositions (Δ COC) increase from early to late Holocene (-744 to -204‰ and -736 to -125‰ for cores 19A and 05D, respectively) as would be expected for sediments with limited to no post-depositional reworking. When 14 expressed as age-corrected Δ COC (i.e. Δ values, Stuiver and Polach, 1977), the Δ values in core 19A and 05D decrease through the Holocene until ~ 2000 yr BP. This progressive decrease in Δ was punctuated by intermittent Δ increases at ~7,600 and ~ 4,500 yr BP (earlier in 19A than 05D, Fig. 4.3e). The latter coincides with a reversal 13 in δ COC composition. Core 05D exhibit identical Δ pattern over the Holocene with reversals occurring at ~7,600, ~ 5,000, and 3,800 yr BP. The last ~2000 yr BP is characterized by slight increase in Δ values for both cores.

86

Figure 4.4. Bulk sedimentological and geochemical profiles for cores 19A (blue) and 05D (red) showing (a) median grain size (MGS), (b) Mineral surface area (MSA), (c) 13 total organic carbon (TOC) contents, (d) stable carbon isotopic (δ COC) composition, 14 and (e) the decay-corrected Δ COC (Δ) values.

87

4.4.4. Plant wax fatty acids

The abundance of short (C16-20) and long-chain (C26-32) saturated fatty acid (n-FAs) displays significant and systematic variability among the measured samples. n-C16 was the most abundant homologue, with average concentrations of 100 µg/g OC and 88 µg/g OC for core 19A and 05D, respectively. The FAs display a bimodal distribution with carbon maxima centred on n-C16 and n-C26 for the short and long-chained homologues, respectively. Long-chain fatty acids (LCFAs) exhibit a strong even-over- odd carbon preference with concentrations ranging from 25 to 126 (average = 50 μg/g OC) for core 05D and 68 to 144 μg/g OC (average 95 μg/g OC) for core 19A. There is no discernible trend in LCFA abundance through the entire span of both cores (Fig. 4.4). Relatively small differences in isotopic composition were observed among long-chain homologues. Thus, isotopic compositions of fatty acid are expressed as abundance 13 13 13 weighted average δ C26-32 (δ CLCFA). δ CLCFA compositions show a good positive 13 2 correlation with δ COC (R = 0.69 and 0.94 for cores 19A and 05D, respectively, Fig. 4.5). The early Holocene part of both records (~10,000 to 5,000 yr BP) is marked by

13 low δ CLCFA values that gradually increase through the later part of the record. Similar to bulk OC, LCFA 14C age increases down-core from 556 to 22,356 and 514 13,836 14C years for cores 19A and 05D, respectively indication no post-depositional reworking of OC.

88

-20.00

-22.00 y = 1.6007x + 5.7378 R² = 0.6856

-24.00 (‰) LCFA C

13 -26.00 δ

y = 1.6187x + 4.1493 -28.00 R² = 0.9376

-30.00 -22.00 -21.00 -20.00 -19.00 -18.00 -17.00

13 δ COC (‰)

Figure 4.5. Relationship between bulk OC and LCFA δ13C (blue – Mahanadi, red – Krishna-Godavari)

4.5. Discussion

4.5.1. Sedimentological control on OC loadings

Terrestrial organic matter in continental margin sediments adjacent to river mouths are the results of complex interactions between organic matter produced and altered within the basin and mineral particles derived from erosion. Thus, physical and chemical characteristics of sediment exert an influence on the OM loading of sediments (Galy et al., 2008a). The comparison between the Mahanadi and Krishna- Godavari provides further insights into mineral surface area controls on OC loading. While both basins have similar physiographic and vegetation characteristics, they differ slightly in underlying bedrock as mineral surface area characteristics. This contrast in surface area is primarily driven by the clay mineral compositions of the two basins. While the total clay mineral abundances of both basins vary between 60 to 80%, clay mineral compositions differ markedly with ~20 to 30% smectite in the Mahanadi, and up to 60% in the Krishna-Godavari (Phillips et al., 2014). As a consequence, the MSA values of Krishna-Godavari sediments that are 1.5-2 times

89 higher than those of the Mahanadi (Fig. 4.4b) for comparable TOC and grain size (Fig. 4.4a-c). Sediments from the Mahanadi and Krishna-Godavari exhibit OC:MSA ratios that are in the range defined as “typical deltaic/deep-sea sediments” (e.g. Blair and Aller, 2012; Hedges and Oades, 1997), suggesting that surface area exerts a first order control on OC loading. However, while OC exhibits a relatively sharp and continuous decline in Krishna-Godavari sediments younger than ~5 ky BP, MSA remains fairly constant until 2 ka. This suggests that in addition to surface area, other intrinsic and extrinsic sedimentary parameters (e.g., grain size, mineralogy, and primary productivity) may influence OC loading in these continental margin sediments. This observation is similar to those in the Huanghe Delta where sediments are characterized by low OC:MSA (Tao et al., 2016). This is mainly due to erosion of the clay-rich/OC-lean sediments from the Loess Plateau that also has a significantly high proportion of smectite (Zhang et al., 1995). The early Holocene portion of the Krishna-Godavari core is characterized by relatively high OC loadings that progressively decrease through the later part of the record (Fig. 4.6). The Mahanadi core varies widely through the entire record with intervals of high and low loadings and, in contrast to the Krishna-Godavari, the Mahanadi displays no systematic variation from early to late Holocene (Fig. 4.6). This can be attributed to configurations of each drainage basin. While the Krishna-Godavari drains the Deccan Plateau and the Indian craton (in the west-central and eastern part of its drainage basin), the Mahanadi drains the Indian Craton almost exclusively. These basin characteristics are manifested in the mineral compositions, where the Krishna- Godavari drains smectite-rich clay mineral in its arid to semi-arid upper catchment and kaolinite-rich clay minerals in its lower reaches. Meanwhile, the Mahanadi drains predominantly kaolinite-rich basement rocks. The increasing contribution of smectite- rich minerals results in very high surface area that is disproportionate to the bulk OM, leading to an overall decrease in the OC loadings.

90

0.50

0.45

0.40 ) 2 0.35

0.30

0.25

0.20

0.15

Bulk OC Loadings (mg OC/m 0.10

0.05

0.00 0 2,000 4,000 6,000 8,000 10,000 12,000 Calibrated Age (ky BP)

Figure 4.6. Bulk OC loadings in the Holocene (blue – Mahanadi, red – Krishna- Godavari).

To further assess surface area control on loading, molecular biomarkers loading scheme of Freymond et al. (2018), that focuses exclusively on organic compounds of terrestrial origin, was adopted to minimize the effects of the extrinsic factors that confounds OC:MSA relationships. Similar to the bulk OC loadings, there is a large variation in the LCFA loading in both locations and the Mahanadi sediments have LCFA loading values that are twice as high as those in the Krishna-Godavari (Fig. 4.7). However, in sharp contrast to bulk OC, there is generally a slight increase in LCFA loadings values from early to late Holocene, especially in the last 5 ky. This difference can be attrbuted to the nature of OM that is associated with the mineral phase. Specifically, a larger relative contribution of terrestrial OC to the overall OC pool will generally lead to higher LCFA loading while the bulk OC remains unchanged (as is the case in the Mahanadi sediments). Increases in sedimentation rate (i.e. increased burial rate and decreased oxygen exposure time; Galy et al., 2007) may induce higher burial efficiency of LCFA while the lower bulk OC loading could be a consequence of less opportunity for marine OC to sorb to the mineral surface and as a result terrestrial OC input are less “diluted” by marine OC contributions.

91

5.000

4.500

4.000 )

2 3.500

3.000

g LCFA/m 2.500 �

2.000

1.500

1.000 LCFA Loadings ( 0.500

0.000 0 2,000 4,000 6,000 8,000 10,000 12,000

Calibrated Age (ky BP)

Figure 4.7. Biomarker (LCFA) loadings in the Holocene (blue – Mahanadi, red – Krishna-Godavari).

In order to examine the spatial variations in organic geochemical characteristics of the BoB sediments, the relationship between isotopes of carbon and OC loadings we assessed (Fig. 4.8). The large variability in δ13C signature and OC loading pattern (especially in the Krishna-Godavari sediments) reveal changes that would be expected from variable contributions of terrestrial (C3 versus C4-plants) plants and 13 marine OC. Notably, there appears to be a positive correlation between δ COC and OC loading in the Mahanadi sediments, suggesting increasing contribution of isotopically enriched OC (marine OC or terrestrial C4-plants) with increasing OC loading. Also, a sediment core collected at the mouth of the Godavari correlates well with bulk OC loadings (r2 = 0.55; see Fig. 2.7 in chapter 2). The latter likely indicates that the variation in the Krishna-Godavari can be attributed to mixing and dilution of purely Godavari derived OC by OC derived from the catchment of the Krishna.

92

Figure 4.8. Bulk OC δ13C (top panel) and Δ (bottom panels) as a function of OC loadings (blue – Mahanadi, red – Krishna-Godavari). The size of the circles denotes the TOC contents.

Plots of the LCFA δ13C versus the OC loadings show strong correlations between the parameters (p-value < 0.0005; Fig. 4.9), indicating that the OC is likely derived terrestrial rather marine OC. Furthermore, the positive relationship between Δ and OC loading, especially in the Krishna-Godavari, suggests the sediments tends to host younger OC (higher 14C values) at higher OC loading similar to findings in the Godavari proximal core (Usman et al.,2018). It can be hypothesized that this is a manifestation of within basin processing of organic matter. For instance, at the early stage OM- mineral interaction, a large portion of the available mineral surfaces is effectively

93 coated with OM comprising of old and young OC. However, as the OM transits through the basin, the younger OC gets oxidized leaving the more recalcitrant older OC and a higher mineral surface area. This leads to an overall decrease in the total OC per unit surface area and hence lower OC loading. Alternatively, sediment entrainment in bottom boundary layer and repeated resuspension-deposition cycles may lead to aging of OC (e.g. Bao et al., 2016). However, the high-resolution age model renders the latter scenario unlikely as a resuspension-deposition loop would have produced large down-core variations in 14C ages. In addition, a general decrease in OC loading from early to late Holocene lends credence to the former scenario of OC loss/replacement. While there are several studies on the provenance of sediments and OC in the Indian peninsula (Giosan et al., 2017), the observed OC loading seems to be strongly linked 13 14 to OC age and C, as evidenced by the higher positive correlation between Δ COC, 13 δ COC, and OC loading. This suggests that the source and nature of terrestrial OC play a more important role in the fate of OC that is eventually buried in Mahanadi and Krishna-Godavari margins.

94

-20.00 -21.00 -22.00 y = 23.991x - 33.535 R² = 0.64 -23.00 -24.00

(‰) -25.00

LCFA -26.00 C 13

δ -27.00 -28.00 -29.00 -30.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 Bulk OC Loadings (mg OC/m2)

-20.00 -21.00 -22.00 -23.00

-24.00 (‰) -25.00 LCFA C

13 -26.00 δ -27.00 y = -26.008x - 19.974 -28.00 R² = 0.3771 -29.00 -30.00 0 0.1 0.2 0.3 0.4 0.5 0.6 2 Bulk OC Loadings (mg OC/m )

Figure 4.9. LCFA as a function of Bulk OC loadings (blue – Mahanadi, red – Krishna- Godavari).

4.5.2. Carbon isotopic composition of Krishna-Godavari and Mahanadi System: implication for the provenance and fate of OC

13 14 The d COC and Δ C values of sedimentary OM from both cores are found within the range of particulate organic matter from continental margin sediments (e.g. Bao et al.,

13 2016; Galy et al., 2008b). However, the core-top d COC values (-21.8‰ and -18.8‰ for 19A and 05D, respectively) are also similar to those reported in riverbed sediments of the modern Godavari riverine sediments (Usman et al., 2018; Cui et al., 2017) and

95 are within the range of values reported for the Holocene paleo-soils from the Indian plains (Sarkar et al., 2009). Although, the contribution of marine OC to the core locations cannot be fully discounted, terrestrial OC contributions dominate the sedimentary record (Cui et al., 2017; Giosan et al., 2017). Moreover, the terrestrial input appears to be dominated by compounds derived from higher plants. Our average chain length (ACL) of LCFA are consistently higher than 28 suggesting a higher plant dominance. A high proportion of terrestrial higher plants contribution during the

13 13 Holocene is also inferred from the correspondence between d COC and d CLCFA measured on the same sample, the latter exclusively derived from terrestrial plants. Terrestrial OC deposited in marine sediments is generally much older than the depositional age of the sediments (e.g. Drenzek et al., 2007). This results from (i) a mixture of different OC pools with contrasting age, such as fossil OC, pre-aged soil OC, and modern plant remains; and (ii) OC aging during transport from land to ocean. Using terrestrial OC deposited in marine sediments as a paleo-vegetation proxy thus requires estimating the lag time between OC ages and depositional ages (Galy et al., 2008b). The measured 14C age of OC, LCFA and foraminifera on targeted levels in both cores were measured. The difference between foraminifera age and the bulk OC (i.e., OC-forams) and LCFA (i.e. LCFA-forams) gives the OC and LCFA age offsets, respectively (an indication of the apparent age of OC and LCFA). The results show a progressive increase in the apparent ages of OC and LCFA throughout the Holocene (Fig. 4.10). In the early Holocene portion of the Mahanadi cores, the LCFA appears to be consistently older than the bulk OC while the reverse is the case in later part of the record (< 4 ky BP; Fig. 4.10a). On the other hand, the Krishna-Godavari OC appears to be consistently older than the LCFA through the most part of the Holocene, except ~500 yr BP when LCFA is about 1600 years older than bulk OC (Fig. 4.10b). The consistent increase in the age offset of bulk OC and LCFA in the late Holocene suggests that the OC pool is dominated by components characterized by long residence times within the basin. An increase in storage time would imply that the system became slower and OC transport from source to the sedimentary sink became longer, by prolonging OC residence in intermediate terrestrial storage pools. 14C depth profiles of soils collected within the modern Godavari shows 14C ages that are as old or older than the apparent age offset. This suggests that deep mineral soil OC are likely the source of aged OC delivered to the adjacent margin.

96

3500

3000

2500

2000

1500

1000

500 Age Offset (yr) 0

-500 Bulk OC -1000 LCFA

-1500 0 2000 4000 6000 8000 10000 12000

Calibrated Age (yr BP)

3500

3000

2500 Bulk OC 2000 LCFA 1500

1000

500 Age offset (yr)

0

-500

-1000

-1500 0 2000 4000 6000 8000 10000 12000 Calibrated Age (yr BP)

Figure 4.10. Apparent ages of bulk OC and LCFA through the Holocene (blue – Mahanadi, red – Krishna-Godavari).

97

The Mahanadi segment is characterized by a humid climate with a mean annual precipitation (MAP) >1200 mm/yr and up to 3200 mm/yr (CWI), and has a positive annual water balance, i.e., MAP is larger than annual potential evapotranspiration (e.g., Prentice et al., 2011). Modelled paleo-precipitation reconstructions in the LGM suggest a reduction of MAP (~ 420 mm/yr) in our study area (Tharammal et al., 2013). It is very likely that MAP dropped below 1000 mm/yr in the Godavari and even below 800 mm/yr in the upper catchment of the Godavari (See Fig. 2.1c ii). Thus, moisture content may have been an important driver of the positive shift in d13C of terrestrial plant (e.g., Diefendorf et al., 2010). As a result, the change in record are thought to be closely linked to shifts in the relative abundance between different plant functional types and C3/C4 plant species relative contributions. Atmospheric pCO2 in combination with local air temperature has been observed to exert an influence on isotopic composition of plant tissues (e.g. Ehleringer and Cerling, 2002). However, the lack of

13 significant correlations in between Holocene pCO2 variability and n-alkanes d C, suggests that pCO2 played minor role as driver of vegetation composition during the Holocene. Long-term marine pollen records from the Krishna-Godavari shows evidence for the predominance of humid communities, mangrove and coastal forest in the early Holocene; interpreted as reflecting the wettest period in the Holocene, with vegetation

13 type dominantly C3 species (Zorzi et al., 2015). In the same vein, our d CLCFA record

(Fig. 11) indicates an overall predominance of C3 vegetation during the early Holocene. The lacustrine pollen record from Lake Lonar, located in the central part of the Godavari catchment, show high moisture conditions during the early Holocene between 9 – 5 ky BP accompanied by expansion of forest ecosystem and arboreal

13 plants (Prasad et al., 2014). The low d CLCFA during the mid-Holocene is likely linked to the continuous expansion of tropical and sub-tropical evergreen forests in the region (Zorzi et al., 2015), as evergreen angiosperms performs produce strong carbon

13 fractionation (Diefendorf et al., 2010). The late Holocene exhibits higher d CLCFA values in response to reduced proportion of trees versus herbs observed in the pollen 13 record (Zorzi et al., 2015; Prasad et al., 2014). The relatively high CLCFA values and abundance of Poaceae observed in the later part of the marine pollen record (Zorzi et al., 2015), could be possibly linked to the expansion of corn farming and its expansion henceforth (e.g. Zhang and Hung, 2010). Evidence of enhanced agricultural activity

98 during the past ~4 ky has been proposed based on analysis of terrestrial biomarker compositions in sediments from the Godavari River (Ponton et al., 2012).

4.5.3. Comparison with other leaf-wax record

13 Two major d C end members can be identified with respect to plant ecosystems namely (a) C3 woody plants, prevalent in (semi-) moist environments; and (b) C4 non- wood plants, dominant in (semi-) arid environments (Collister et al., 1994; Rommerskirchen et al., 2006; Vogts et al., 2012). In this regard, vegetation shift

13 deduced from d CLCFA record implies a change from C3-plant dominated ecosystem during the early Holocene to a C4-plant vegetation during the late Holocene. Although

13 significant enrichment of d C in C3 plants resulting from water stress (e.g. Diefendorf et al., 2010; Prentice et al., 2011), differential wax production between plant groups

13 (e.g. Bush and McInerney, 2013), or the wide range of d C of C3 plants alone (e.g.

Ehleringer and Cerling, 2002) can lead to under/over estimation of C3 plants contribution, thus affecting the accuracy of plant mixing models. Unfortunately, the impacts of these factors are difficult to quantify. Nevertheless, the high positive correlation between d13COC and d13CLCFA likely imply that the d13C (both OC and

LCFA) changes represent a shift from a dominant C3 to C4-vegetation in the region because of better adaption of C4-plants to arid conditions (e.g. Ehleringer et al., 1997). During the late Holocene, significant aridification in the Core Monsoon Zone (CMZ) beginning at 4.7 ky BP (Ponton et al., 2012) coincides with a reduction in water availability and biome contractions, indicative of relatively weak monsoon conditions (Zorzi et al., 2015). Comparisons with our results from the Mahanadi and Krishna-

Godavari indicate stronger C4-vegetation expansion in the Krishna-Godavari

13 13 13 Comparison of our d COC and d CLCFA records of paleo-vegetation (d C) trends compare favourably with those obtained from a sediment core at the mouth of the Ganges-Brahmaputra River (Hein et al., 2017 and Contreras-Rosales et al., 2014).

13 The results show a lower degree of variability than do the d CLCFA of Hein et al. (2017)

13 and are more isotopically enriched than the d Calkanes of Contreras-Rosales et al. (2014). This is likely to be a consequence of the differences in carbon isotopic fractionation of alkanes and fatty acids. Many of the differences between datasets can be explained by the differences of sedimentary and OC provenance in the core locations. All the afore-mentioned studies were carried out at the mouth of the Ganges-

99

Brahmaputra Rivers with a predominantly Himalayan source while our study focuses exclusively on the tropical Mahanadi and Krishna-Godavari Rivers. Published Holocene records from our locations (e.g. Ponton et al., 2012; Giosan et al., 2017; Cui et al., 2017; Usman et al., 2018) document a major shift in sediments supply during the mid-Holocene whereas no evidence of changes in sediment supply was observed in the Ganges-Brahmaputra (Contreras-Rosales et a., 2014; Hein 2017).The apparent temporal offsets between the records can be attributed to the resolution of the records as well as a manifestation of the north-south migration of the intertropical convergence zone (ITCZ) in response to ISM variations in the core monsoon zone.

4.5.4. Regional Synthesis of Holocene monsoon records.

Contrasting our record with regionally relevant records related to the ISM has yielded close correlation with regional events in the Bay of Bengal. Despite the numerous data available for this part of the world ocean, only selected high-resolution, high-quality (AMS 14C-dated, corrected and calibrated) records that focus exclusively on the time intervals panning the last 12 ky and are regionally representative (Fig. 4.11). These records are mostly in the Northern BoB (see Fig. 2.1) and includes: Hein et al. (20017), Thamban et al. (2007), Contreras-Rosales et al. (2014).

100

Figure 4.11. Synthesis of Holocene changes in the summer monsoon recorded in the 13 BoB. (a) sedimentation rate (this study), (b) d COC (this study), (c) precipitation record 13 (Thamban et al., 2007), dDalkanes (Contreras-Rosales et al., 2014), d CLCFA (this study), 13 and d CLCFA (Hein et al., 2017). This study: blue – Mahanadi, red – Krishna-Godavari.

101

A summary of the regionally consistent climate events identified in comparison with other BoB records is as follows:

• The early Holocene intensification of summer monsoon occurred between 10

and 8 ky BP with relatively high dDalkanes and precipitation values. • A mid-Holocene decline in monsoon intensity beginning between 5 - 4 ky BP, accompanied by increase in d13C values of bulk OC and LCFA. • Around 2 ky BP, deterioration in monsoon intensity reached its peak. This is evident in the highest d13C of bulk OC and LCFA recorded in the entire Holocene record. This is also reflected in the decrease in precipitation. • The last 1 ky BP witnessed an enhancement in monsoon intensity as this period recorded low d13C values and increase in precipitation. This is also the period with peak sediment accumulation.

The monsoon events described here seems to be synchronous with the Holocene “Rapid Climate Change” events observed in several globally distributed, multi-proxy records (Mayewski et al., 2004). Barring any significant uncertainty associated with dating, the temporal offset in the proxy response can be attributed to different response times of proxies to same forcing mechanism.

4.5.5. Potential relationship between climate (i.e. ISM intensity) and anthropogenic activities.

During the peak Harappan and Mohenjaro civilizations (~4.6 ky BP), most of agricultural expansion occurred within the Indus valley and these civilizations started declining at ~4 ky BP, leaving abandoned settlements between 3.6 to 3.4 ky BP (Kenoyer 1998). The decline of the Indus civilization has been largely attributed to inferred changes in the course of the Ghaggar-Hakra River. The change in river course eventually led to the drying out of the Ghaggar-Hakra and forced the migration of settlements towards the south of the Indus valley (Madella and Fuller, 2006). Archaeological evidence reveals that the onset of sedentary agriculture occurred in the northern and central Deccan Plateau, in response to ISM weakening in the Core Monsoon Zone (CMZ) (Fuller, 2011). The development of human settlement in the

102 northern Deccan is dated at 3.9 ky BP (Possehl 1999) and the majority of dated sites are close to rivers indicating a need for a reliable source of water (Fuller 2011). The link between climate change, or more precisely changes in ISM strength and the disappearance of Indus civilization has rarely been explored or the role of the ISM has been interpreted as only secondary (Madella and Fuller, 2006). Our geochemical records (especially those of pure terrestrial origin) are too coarse to fully resolve the precise timing of the ISM weakening. Our age model from the BoB as well as continental records of Fleitmann et al. (2007) suggest the onset of weak monsoon events at ~5 – 4 ky BP. This interval coincides with the decline of African Humid Period (Schulz et al., 1998) and an abrupt increase of dust in Tibetan Plateau (Thompson et al., 2006) suggesting the onset of dry arid climate. Most precipitation over the Indian subcontinent falls during the period of intensified ISM (Gadgil, 2003). Archaeological evidence for the adoption of low rainfall crop pattern beginning last 2 ky (Deotare, 2006) would seem to corroborate the desiccation of river channels in the late Holocene (e.g. Sharma et al., 2004).

4.6. Conclusions

This study presents a high-resolution Holocene paleo-vegetation reconstruction from the continental margin of the Bay of Bengal adjacent to the Mahanadi and Krishna- Godavari Rivers. The bulk mineralogical and isotopic composition of the margin sediments suggests that terrestrial OC is predominant through the Holocene. MSA values in the sediment record correlate with the dominant bedrock geology of the river catchment and a shift in MSA values, in tandem with other geochemical data, likely reflect a shift in sediment source region. Bulk OC and biomarker composition suggest that terrestrial sediment sources and distribution are similar for both river basins until ~5 ky BP, during which geochemical data from these cores can be considered to comprise of continuous regional paleo-environmental record for the Holocene within

13 13 the CMZ. The d COC and d COC records indicate an overall predominance of C3 vegetation under moist conditions that pervades the early Holocene and gradual shift towards C4-plant predominance as a result of the weakening of the ISM towards the late Holocene. Radiocarbon measurements on bulk OC and LCFA reveal a systematic offset that increases concomitantly with increasing sedimentation rate, suggesting that

103 decrease in monsoon intensity may have led to decrease in OC transport rates and/or alteration in the relative proportion of fresh versus pre-aged biospheric OC. This indicates a causal link between ISM dynamics and carbon cycling with the drainage basin. Our records compare favourably with other regional paleo-environmental records and capture large scale variations in precipitation and vegetation associated with for instance the early Holocene climatic optimum “EHCO” (Contreras-Rosales et al., 2014) and Holocene Rapid Climate Change “RCC (Mayewski et al., 2004). Interpreting the record within the context of past cultural evolution provides important insights into the effect of monsoon variation on human settlements and the likely impacts of anthropogenic perturbations on terrestrial OC cycling. Taken together, the results reveal a tight coupling between Holocene ISM variations and OC cycling within and out of river basins, sometimes mediated by human activities, and further highlights the importance of climate (e.g. ISM variations) and anthropogenic activities (e.g. agriculture) within the broader context of global climate change.

References

Asouti, E. and Fuller, D. (2008) Trees and Woodlands in South India: Archaeological Perspectives. Left Coast Press, Walnut Creek, California. Bao, R., McIntyre, C., Zhao, M.X., Zhu, C., Kao, S.J. and Eglinton, T.I. (2016) Widespread dispersal and aging of organic carbon in shallow marginal seas. Geology 44, 791-794. Bendle, J.A., Weijers, J.W.H., Maslin, M.A., Damste, J.S.S., Schouten, S., Hopmans, E.C., Boot, C.S. and Pancost, R.D. (2010) Major changes in glacial and Holocene terrestrial temperatures and sources of organic carbon recorded in the Amazon fan by tetraether lipids. Geochemistry Geophysics Geosystems 11, 1-13. Blair, N.E. and Aller, R.C. (2012) The fate of terrestrial organic carbon in the marine environment. Ann Rev Mar Sci 4, 401-423. Blair, N.E., Leithold, E.L. and Aller, R.C. (2004) From bedrock to burial: the evolution of particulate organic carbon across coupled watershed-continental margin systems. Marine Chemistry 92, 141-156. Bouchez, J., Galy, V., Hilton, R.G., Gaillardet, J., Moreira-Turcq, P., Pérez, M.A., France-Lanord, C. and Maurice, L. (2014) Source, transport and fluxes of Amazon

104

River particulate organic carbon: Insights from river sediment depth-profiles. Geochimica et Cosmochimica Acta 133, 280-298. Burdige, D.J. (2005) Burial of terrestrial organic matter in marine sediments: A re- assessment. Global Biogeochemical Cycles 19. GB4011. Burdige, D.J. (2007) Preservation of Organic Matter in Marine Sediments: Controls, Mechanisms, and an Imbalance in Sediment Organic Carbon Budgets? Chemical Review 107, 467- 485. Bush, R.T. and McInerney, F.A. (2013) Leaf wax n-alkane distributions in and across modern plants: Implications for paleoecology and chemotaxonomy. Geochimica et Cosmochimica Acta 117, 161-179. Chakrapani, G. and Subramanian, V. (1990) Preliminary studies on the geochemistry of the Mahanadi River basin, India. Chemical Geology 81, 241-253. Charles, C.D., Cobb, K., Moore, M.D. and Fairbanks, R.G. (2003) Monsoon–tropical ocean interaction in a network of coral records spanning the 20th century. Marine Geology 201, 207-222. Collett, T., Boswell, R., Cochran, J., Kumar, P., Lall, M., Mazumdar, A., Raman, M., Ramprasad, T., Riedel, M., Sain, K., Sathe, A., Vishwanath, K. and Party, N.E. (2014) Geologic implications of gas hydrates in the offshore of India: Results of the National Gas Hydrate Program Expedition 01. MArine and Petroleum Geology 58, 3-28. Collister, J., Rieley, G., Stern, B., Eglinton, G. and Fry, B. (1994) Compound-specific d13C analyses of leaf lipids from plants with differing carbon dioxide metabolisms. Organic Geochemistry 21, 619-627. Contreras-Rosales, L.A., Jennerjahn, T., Tharammal, T., Meyer, V., Lückge, A., Paul, A. and Schefuß, E. (2014) Evolution of the Indian Summer Monsoon and terrestrial vegetation in the Bengal region during the past 18 ka. Quaternary Science Reviews 102, 133-148. Diefendorf, A.F., Freeman, K.H., Wing, S.L. and Graham, H.V. (2011) Production of n-alkyl lipids in living plants and implications for the geologic past. Geochimica et Cosmochimica Acta 75, 7472-7485. Diefendorf, A.F., Mueller, K.E., Wing, S.L., Koch, P.L. and Freeman, K.H. (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc Natl Acad Sci U S A 107, 5738-5743. 105

Cui, M., Wang, Z., Nageswara Rao, K., Sangode, S.J., Saito, Y., Chen, T., Kulkarni, Y.R., Naga Kumar, K.C.V. and Demudu, G. (2017) A mid- to late-Holocene record of vegetation decline and erosion triggered by monsoon weakening and human adaptations in the south-east Indian Peninsula. The Holocene, 1-12. Deotare, B. (2006) Late Holocene climatic change: Archaeological evidence from the Purna Basin, Maharashtra. Journal of the Geological Society of India 68, 517-526. Dixit, Y., Hodell, D.A. and Petrie, C.A. (2014) Abrupt weakening of the summer monsoon in northwest India 4100 yr ago. Geology 42, 339-342. Drenzek, N.J., Montluçon, D.B., Yunker, M.B., Macdonald, R.W. and Eglinton, T.I. (2007) Constraints on the origin of sedimentary organic carbon in the Beaufort Sea from coupled molecular 13C and 14C measurements. Marine Chemistry 103, 146- 162. Drenzek, N.J., Hughen, K.A., Montlucon, D.B., Southon, J.R., dos Santos, G.M., Druffel, E.R.M., Giosan, L. and Eglinton, T.I. (2009) A new look at old carbon in active margin sediments. Geology 37, 239-242. Eglinton, T.I. and Eglinton, G. (2008) Molecular proxies for paleoclimatology. Earth and Planetary Science Letters 275, 1-16. Eglinton, T., Aluwihare, L., Bauer, J., Druffel, E. and McNichol, A. (1996) Gas Chromatographic Isolation of Individual Compounds from Complex Matrices for Radiocarbon Dating. Analytical Chemistry 68, 904-912. Ehleringer, J. and Cerling, T. (1997) C4 photosynthesis, atmospheric CO2, and climate. Oecologia 112, 285-299. Ehleringer, J. and Cerling, T. (2002) The Earth system: biological and ecological dimensions of global environmental change, in: Mooney, H., Canadell, J. (Eds.), Encyclopedia of Global Environmental Change. John Wiley & Sons, Ltd, Chichester, pp. 186-190. Enzel, Y., Ely, L., Mishra, S., Ramesh, R., Amit, R., Lazar, B., Rajaguru, S., Baker, V. and Sandler, A. (1999) High-Resolution Holocene Environmental Changes in the Thar Desert, Northwestern India. Science 284, 125-128. Feng, X., Cui, H., Tang, K. and Conkey, L. (1999) Tree-Ring dD as an Indicator of Asian Monsoon Intensity. Quaternary Research 51, 262-266. Fleitmann, D., Burns, S.J., Mangini, A., Mudelsee, M., Kramers, J., Villa, I., Neff, U., Al-Subbary, A.A., Buettner, A., Hippler, D. and Matter, A. (2007) Holocene ITCZ and

106

Indian monsoon dynamics recorded in stalagmites from Oman and Yemen (Socotra). Quaternary Science Reviews 26, 170-188. Freymond, C.V., Kündig, N., Stark, C., Peterse, F., Buggle, B., Lupker, M., Plötze, M., Filip, F., Giosan, L. and Eglinton, T. (Submitted) Evolution of biomolecular loadings along a major river system. Fuller, D.Q. (2011) Finding Plant Domestication in the Indian Subcontinent. Current Anthropology 52, S347-S362. Gadgil, S. (2003) The Indian monsoon and its variability. Annual Review of Earth and Planetary Sciences 31, 429-467. Galy, V. and Eglinton, T. (2011) Protracted storage of biospheric carbon in the Ganges-Brahmaputra basin. Nature Geoscience 4, 843-847. Galy, V., France-Lanord, C. and Lartiges, B. (2008a) Loading and fate of particulate organic carbon from the Himalaya to the Ganga-Brahmaputra delta. Geochimica et Cosmochimica Acta 72, 1767-1787. Galy, V., François, L., France-Lanord, C., Faure, P., Kudrass, H., Palhol, F. and Singh, S.K. (2008b) C4 plants decline in the Himalayan basin since the Last Glacial Maximum. Quaternary Science Reviews 27, 1396-1409. Galy, V., France-Lanord, C., Beyssac, O., Faure, P., Kudrass, H. and Palhol, F. (2007) Efficient organic carbon burial in the Bengal fan sustained by the Himalayan erosional system. Nature 450, 407-410 Giosan, L., Ponton, C., Usman, M., Blusztajn, J., Fuller, D.Q., Galy, V., Haghipour, N., Johnson, J.E., McIntyre, C., Wacker, L. and Eglinton, T.I. (2017) Short communication: Massive erosion in monsoonal central India linked to late Holocene land cover degradation. Earth Surface Dynamics 5, 781-789. Goodbred, S. and Kuehl, S. (2000) Enormous Ganges-Brahmaputra sediment discharge during strengthened early Holocene monsoon. Geology 28, 1083-1086. Gunnell, Y., Anupama, K. and Sultan, B. (2007) Response of the South Indian runoff- harvesting civilization to northeast monsoon rainfall variability during the last 2000 years: instrumental records and indirect evidence. Holocene 17, 207-215. Gupta, L.P., Subramanian, V. and Ittekkot, V. (1997) Biogeochemistry of particulate organic matter transported by the Godavari river, India. Biogeochemistry 38, 103-128 Hedges, J.I., Clark, W.A., Quay, P.D., Richey, J.E., Devol, A.H. and Santos, U.D. (1986) Compositions and Fluxes of Particulate Organic Material in the Amazon River. Limnology and Oceanography 31, 717-738. 107

Hedges, J. and Oades, J. (1997) Comparative organic geochemistries of soils and marine sediments. Organic Geochemistry 27, 319-361. Hein, C.J., Galy, V., Galy, A., France-Lanord, C., Kudrass, H. and Schwenk, T. (2017) Post-glacial climate forcing of surface processes in the Ganges–Brahmaputra river basin and implications for carbon sequestration. Earth and Planetary Science Letters 478, 89-101. Hughes, M., Xiangding, W., Xuemei, S. and Garfin, G. (1994) A preliminary reconstruction of rainfall in North-Central China since AD 1600 from tree-ring density and width. Quaternary Research 42, 88-99. Keil, R.G. and Mayer, L.M. (2014) Mineral Matrices and Organic Matter, Treatise on Geochemistry, pp. 337-359. Kennedy, M.J. and Wagner, T. (2011) Clay mineral continental amplifier for marine carbon sequestration in a greenhouse ocean. Proc Natl Acad Sci U S A 108, 9776- 9781. Kenoyer, J. (1998) Ancient Cities of the Indus Valley Civilization. Oxford University Press, Karachi. Leithold, E.L., Blair, N.E. and Perkey, D.W. (2006) Geomorphologic controls on the age of particulate organic carbon from small mountainous and upland rivers. Global Biogeochemical Cycles 20 Liu, K.-b., Yao, Z. and Thompson, L.G. (1998) A pollen record of Holocene climatic changes from the Dunde ice cap, Qinghai-Tibetan Plateau. Geology 26. Madella, M. and Fuller, D.Q. (2006) Palaeoecology and the Harappan Civilisation of South Asia: a reconsideration. Quaternary Science Reviews 25, 1283-1301. Madhu, N.V., Jyothibabu, R., Maheswaran, P.A., John Gerson, V., Gopalakrishnan, T.C. and Nair, K.K.C. (2006) Lack of seasonality in phytoplankton standing stock (chlorophyll a) and production in the western Bay of Bengal. Continental Shelf Research 26, 1868-1883. Mayer, L.M. (1994) Surface-Area Control of Organic-Carbon Accumulation in Continental-Shelf Sediments. Geochimica Et Cosmochimica Acta 58, 1271-1284. Mayewski, P.A., Rohling, E.E., Curt Stager, J., Karlén, W., Maasch, K.A., Meeker, L.D., Meyerson, E.A., Gasse, F., van Kreveld, S., Holmgren, K., Lee-Thorp, J., Rosqvist, G., Rack, F., Staubwasser, M., Schneider, R.R. and Steig, E.J. (2004) Holocene Climate Variability. Quaternary Research 62, 243-255.

108

Milliman, J. and Farnsworth, K. (2011) River Discharge to the global ocean:A global synthesis. Cambridge University Press, New York. Milliman, J. and Meade, H. (1983) World-wide delivery of river sediments to the ocean. Journal of Geology 91, 1-21. Naqvi, S., Shailaja, M., Kumar, M. and Gupta, R. (1996) Respiration rates in subsurface waters of the northern Indian Ocean: Evidence for low deomposition rates of organic matter within the water column in the Bay of Bengal. Deep Sea Research II 43, 73-81. Pancost, R.D. and Boot, C.S. (2004) The palaeoclimatic utility of terrestrial biomarkers in marine sediments. Marine Chemistry 92, 239-261. Pattan, J.N., Mir, I.A., Parthiban, G., Karapurkar, S.G., Matta, V.M., Naidu, P.D. and Naqvi, S.W.A. (2013) Coupling between suboxic condition in sediments of the western Bay of Bengal and southwest monsoon intensification: A geochemical study. Chemical Geology 343, 55-66. Phillips, S.C., Johnson, J.E., Underwood, M.B., Guo, J., Giosan, L. and Rose, K. (2014) Long-timescale variation in bulk and clay mineral composition of Indian continental margin sediments in the Bay of Bengal, Arabian Sea, and Andaman Sea. Marine and Petroleum Geology 58, 117-138 Ponton, C., Giosan, L., Eglinton, T.I., Fuller, D.Q., Johnson, J.E., Kumar, P. and Collett, T.S. (2012) Holocene aridification of India. Geophysical Research Letters 39, 6. Prasad, S., Anoop, A., Riedel, N., Sarkar, S., Menzel, P., Basavaiah, N., Krishnan, R., Fuller, D., Plessen, B., Gaye, B., Rohl, U., Wilkes, H., Sachse, D., Sawant, R., Wiesner, M.G. and Stebich, M. (2014) Prolonged monsoon droughts and links to Indo- Pacific warm pool: A Holocene record from Lonar Lake, central India. Earth and Planetary Science Letters 391, 171-182. Prentice, I.C., Meng, T., Wang, H., Harrison, S.P., Ni, J. and Wang, G. (2011) Evidence of a universal scaling relationship for leaf CO2 drawdown along an aridity gradient. New Phytol 190, 169-180. Ramprasad, T., Dewangan, P., Ramana, M.V., Mazumdar, A., Karisiddaiah, S.M., Ramya, E.R. and Sriram, G. (2011) Evidence of slumping/sliding in Krishna–Godavari offshore basin due to gas/fluid movements. Marine and Petroleum Geology 28, 1806- 1816.

109

Rao, V. and Kessarkar, P. (2001) Geomorphology and geology of the Bay of Bengal and Andaman Sea. In: Sengupta, R., Desa, E. (Eds.), The Indian Ocean - a perspective Part II, Oxford, pp. 817-868. Rao, N.K., Saito, Y., Nagakumar, K.C.V., Demudu, G., Basavaiah, N., Rajawat, A.S., Tokanai, F., Kato, K. and Nakashima, R. (2012) Holocene environmental changes of the Godavari Delta, east coast of India, inferred from sediment core analyses and AMS 14C dating. Geomorphology 175-176, 163-175. Rickers, K., Mezger, K. and Raith, M. (2001) Evolution of the Continental Crust in the Proterozoic Eastern Ghats Belt, India and new constraints for Rodinia reconstruction: implications from Sm–Nd, Rb-Sr and Pb–Pb isotopes. Precambrian Research 112, 183-210. Riedel, M., Collett, T.S. and Shankar, U. (2011) Documenting channel features associated with gas hydrates in the Krishna–Godavari Basin, offshore India. Marine Geology 279, 1-11. Rommerskirchen, F., Plader, A., Eglinton, G., Chikaraishi, Y. and Rullkötter, J. (2006) Chemotaxonomic significance of distribution and stable carbon isotopic composition of long-chain alkanes and alkan-1-ols in C4 grass waxes. Organic Geochemistry 37, 1303-1332. Sarkar, A., Sengupta, S., McArthur, J., Ravenscroft, P., Bera, M., Bhushan, R., Samanta, A. and Agrawal, S. (2009) Evolution of Ganges–Brahmaputra western delta plain: clues from sedimentology and carbon isotopes. Quaternary Science Reviews 28, 2564-2581. Sarma, V.V.S.S., Krishna, M.S., Viswanadham, R., Rao, G.D., Rao, V.D., Sridevi, B., Kumar, B.S.K., Prasad, V.R., Subbaiah, C.V., Acharyya, T. and Bandopadhyay, D. (2012) Intensified oxygen minimum zone on the western shelf of Bay of Bengal during summer monsoon: influence of river discharge. Journal of Oceanography 69, 45-55. Schott, F.A., Xie, S.-P. and McCreary, J.P. (2009) Indian Ocean circulation and climate variability. Reviews of Geophysics 47. Schulz, H., von Rad, U. and Erlenkeuser, H. (1998) Correlation between Arabian Sea and Greenland climate oscillations of the past 110,000 years. Nature 393, 54-57. Sharma, S., Joachimski, M., Sharma, M., Tobschall, H., Singh, I., Sharma, C., Chauhan, M. and Morgenroth, G. (2004) Lateglacial and Holocene environmental changes in Ganga plain, Northern India. Quaternary Science Reviews 23, 145-159.

110

Singh, S.P., Singh, S.K. and Bhushan, R. (2011) Behavior of dissolved redox sensitive elements (U, Mo and Re) in the water column of the Bay of Bengal. Marine Chemistry 126, 76-88. Stoll, H.M., Arevalos, A., Burke, A., Ziveri, P., Mortyn, G., Shimizu, N. and Unger, D. (2007) Seasonal cycles in biogenic production and export in Northern Bay of Bengal sediment traps. Deep Sea Research Part II: Topical Studies in Oceanography 54, 558- 580. Subrahmanyam, V., Subrahmanyam, A.S., Murty, G.P.S. and Murthy, K.S.R. (2008) Morphology and tectonics of Mahanadi Basin, northeastern continental margin of India from geophysical studies. Marine Geology 253, 63-72. Subramanian, V. (1980) Mineralogical input of suspended matter by Indian Rivers into the adjacent areas of the Indian Ocean. Marine Geology 36, 29-34. Subramanian, V. (1987) Environmental Geochemistry of Indian river basin - a review. Geological Society of India 29, 205-220 Tao, S., Eglinton, T.I., Montluçon, D.B., McIntyre, C. and Zhao, M. (2016) Diverse origins and pre-depositional histories of organic matter in contemporary Chinese marginal sea sediments. Geochimica et Cosmochimica Acta 191, 70-88. Tao, S.Q., Eglinton, T.I., Montlucon, D.B., McIntyre, C. and Zhao, M.X. (2015) Pre- aged soil organic carbon as a major component of the Yellow River suspended load: Regional significance and global relevance. Earth and Planetary Science Letters 414, 77-86. Thamban, M., Kawahata, H. and Rao, V. (2007) Indian Summer Monsoon Variability during the Holocene as Recorded in Sediments of the Arabian Sea: Timing and Implications. Journal of Oceanography 63, 1009-1020. Tharammal, T., Paul, A., Merkel, U. and Noone, D. (2013) Influence of Last Glacial Maximum boundary conditions on the global water isotope distribution in an atmospheric general circulation model. Climate of the Past 9, 789-809. Thompson, L., Mosley-Thompson, E., Brecher, H., Davis, M., Léon, B., Les, D., Lin, P.-N., Mashiota, T. and Mountain, K. (2006) Abrupt tropical climate change: Past and present. Proceedings of the National Academy of Science 103, 10536-10543. Thompson, L., Yao, T., Mosley-Thompson, E., Davis, M., Henderson, K. and Lin, P.- N. (2000) A High-Resolution Millennial Record of the South Asian Monsoon from Himalayan Ice Cores. Science 289, 1916-1919. Traverse, A. (1989) Paleopalynology. Unwin Hyman, Boston. 111

Tudhope, A., Lea, D., Shimmeld, G., Chilcott, C. and Head, S. (1996) Monsoon Climate and Arabian Sea Coastal Upwelling Record in Massive Corals from Southern Oman. Palaios 11, 347-361. Usman, M.O., Kirkels, F.M.S.A., Zwart, H.M., Basu, S., Ponton, C., Blattmann, T.M., Ploetze, M., Haghipour, N., McIntyre, C., Peterse, F., Lupker, M., Giosan, L. and Eglinton, T.I. (2018) Reconciling drainage and receiving basin signatures of the Godavari River system. Biogeosciences 15, 3357-3375. Vogts, A., Schefuß, E., Badewien, T. and Rullkötter, J. (2012) n-Alkane parameters from a deep-sea sediment transect off southwest Africa reflect continental vegetation and climate conditions. Organic Geochemistry 47, 109-119. Wang, B., Wu, R. and Lau, K.-M. (2001) Interannual Variability of the Asian Summer Monsoon: Contrasts between the Indian and the Western North Pacific–East Asian Monsoons. Journal of Climate 14, 4073-4090. Weber, M., Wiedicke, M., Kudrass, H., Hübscher, C. and Erlenkeuser, H. (1997) Active growth of the Bengal Fan during sea-level rise and highstand. Geology 25, 315-318. Yadava, M. and Ramesh, R. (2005) Monsoon reconstruction from radiocarbon dated tropical Indian speleothems. The Holocene 15, 48-59. Zhang, J., Huang, W., Létolle, R. and Jusserand, C. (1995) Major element chemistry of the Huanghe (Yellow River), China - weathering processes and chemical fluxes. Journal of Hydrology 168, 173-203. Zorzi, C., Goni, M.F.S., Anupama, K., Prasad, S., Hanquiez, V., Johnson, J. and Giosan, L. (2015) Indian monsoon variations during three contrasting climatic periods: The Holocene, Heinrich Stadial 2 and the last interglacial-glacial transition. Quaternary Science Reviews 125, 50-60.

112

Chapter 5: Conclusions and outlook

113

6.1. Conclusions

Charles Lyell once wrote that “the present is the key to the past”. This has held true for several centuries after that pronouncement. Recent advancements in technology have led to high precision measurements that allow probing into the past with remarkable accuracy. One of the fundamental objectives of Earth science research is to understand how the global carbon cycle has adapted to external perturbations in the past and to quantitatively predict how it will respond to future disturbances. In the context of anthropogenic climate change, this becomes even more imperative in the context of anthropogenic climate change (i.e., how will the Earth’s natural systems respond to massive human forcing), because a thorough understanding of the past may offer us a window into the future. The studies conducted in the scope of this thesis developed high-resolution records of Indian summer monsoon variability spanning the entire Holocene by using sediment cores collected from the continental margins in the Bay of Bengal adjacent that are to rivers characterized by high fluvial discharge. This high sedimentation regime allowed centennial to millennial scale resolutions of past variability in vegetation and hydrology in response to changing monsoon intensity in subcontinental India. In addition, riverine sampling from the modern river basin enabled the attribution of OM sources and its evolution in the modern Godavari river basin as well as reconciliation of terrestrial organic matter signatures with those recorded in continental margin adjacent to the river mouth.

The major findings of this thesis are itemized in the following subsections:

6.1.1. Provenance and evolution of organic matter in the Godavari River Basin and linkage to the marine sedimentary archive

The large lithological and vegetation gradients within the modern Godavari system provide important end-members that allowed source-fingerprinting of fluvially transported sediments and organic matter. Long-chain fatty acids exhibited a strong relationship with the bulk OC, implying a terrestrial plant origin for the riverine OC. The positive linear correlation between riverine organic carbon (OC) and mineral surface area (MSA) indicate that sediment mineralogy (largely controlled by basin lithology) exerts a significant control on OC characteristics. Stable carbon isotopic composition 114 revealed that C4-plants are dominant in the upper part of the catchment and decrease in contribution towards the sea. This seaward reduction in the contribution by C4-plants has been largely attributed to the vegetation gradients within the basin and/or in-river processing of OC. The remarkable correspondence between the characteristics of continental margin sediments and those from the river basin allowed a straightforward attribution of signals recorded in the marine archive to specific regions of the basin. Bulk and biomarker characteristics of present-day Godavari basin and adjacent continental margin sediments revealed a mid-Holocene shift in sedimentological and geochemical signature that is consistent with OC and sediment provenance change towards higher contributions of Deccan Plateau derived materials from the upper basin. While signals recorded in the offshore Godavari sedimentary archive can be attributed to specific parts of the basin, the major mechanism of OC transport and the extent of in-river processing are necessary to understand the factors controlling the nature and fate of terrestrial OC preserved in sedimentary sequences. This become more pressing in the face of increasing anthropogenic activities since human perturbation of the landscape has induced significant alteration of basin hydrological characteristics that has impacted export flux of terrestrial organic matter.

6.1.2. Timescale of terrestrial OC export

This study represents the first high-resolution (centennial scale) bulk and biomarker reconstruction of the Holocene paleoclimate on the Indian peninsula. Stable carbon isotopic measurements of terrestrial plant waxes provided an integrative record of the paleo-vegetation in central India. This revealed a progressive increase in aridity- adapted vegetation beginning in the mid Holocene (~ 4.5 ky BP). Radiocarbon compositions of bulk and long-chain fatty acids showed the temporal dynamics of vascular plant OC transport to the ocean. Comparison of bulk OC and LCFA to foraminifera revealed an increase in the apparent age of OM delivered to the continental margin. This increasing age offset suggests that a progressive increase in aridity impacted OM composition either by changing the source to pre-aged soil OC or by increasing the terrestrial residence time of OC. A coupled isotopic mixing model was employed to estimate the proportion of different OC types. Results show that the OC pool is dominated by fresh biospheric OC fraction, indicating an almost

115 instantaneous transfer of terrestrial from source to the sedimentary sink. In contrast, pre-aged terrestrial OC predominates in the late Holocene, suggesting aging of OM during riverine transport and/or enhanced erosion of old soil OC. The latter is further exacerbated during the last 2 kys by changes in anthropogenic land use. Soil age- depth profiles were used to constrain the minimum age and fractional contribution of soil OC required to produce the age offsets observed in the sedimentary records. Results indicate that more than 65% of pre-aged soil OC (with minimum 14C age of 3,400 yrs) is required to explain the apparent age offset. This would imply that extensive erosion of deep soils (down to the bedrock) within the basin must have taken place during the late Holocene. While modern flux estimates indicate an order of magnitude decrease in sediment export to the Godavari continental margin as a result of dam constructions (e.g. Pradhan et al., 2014), seismic profiles of the Godavari shelf indicate a rapid buildup of thick freshly deposited sediment packages (Forsberg et al., 2007). Reconciling this apparent contradiction will provide better constraints on soil OC export budgets.

6.1.3. Long-term Indian monsoon variability: impacts on anthropogenic activity and carbon cycling

The high-resolution record of the Indian summer monsoon variability in subcontinental India provided an opportunity to investigate the link between continental climate and the dynamics of land-to-sea transfer of terrestrial OC. In order to provide a more holistic perspective of ISM impact of river-dominated continental margin, this thesis also presented a Holocene record of the Mahanadi River system and Krishna- Godavari system. The latter is closer to the mouth of the Krishna than the exclusively Godavari core presented in chapter 3 (see Fig. 4.1. for reference). By contrasting records from this study with paleo-environmental datasets from the northern BoB, a strong correlation with regional events in Indian subcontinent was found. Similar to the Godavari core, bulk OC and biomarker compositions indicate a shift in OM provenance ~5 ky BP. This coincides with large-scale change in regional precipitation and vegetation implying that ISM variation exert a significant control on OC composition. Juxtaposing these regional climate changes with anthropogenic activity revealed a causal link between ISM deterioration in the later part of the Holocene and human

116 migration to central and southern India with its attendant perturbation of the terrestrial carbon cycle.

In general, this thesis presents high-resolution paleoclimatic reconstructions of the ISM from sedimentary archives in river-dominated continental margins that integrate signals to produce a regionally extensive records. It provides useful insights into land- to-sea transport of terrestrial OC. The results presented here highlights the importance of climate (i.e. ISM) impact on sedimentation, vegetation, regional anthropogenic migration patterns, and terrestrial OC cycle.

6.2. Outlook

Based on the results and conclusions from this work, we have identified future research that could prove valuable to our understanding of climate and anthropogenic influence on carbon cycle dynamics. Total riverine transported OC represents a mixture of multiple sources but is generally dominated by soil input (Hedges and Oades, 1997). Under natural conditions, soil erosion is approximately balanced by soil production through weathering, however the shift from natural to agricultural land use removes the natural vegetation and typically increases soil erosion by one to two orders of magnitude (Montgomery, 2007). The Krishna-Godavari and Mahanadi river drain some of the most populated areas of India where agriculture represents the mainstay of the local economy (IMHA, 2011). Agricultural practices have exacerbated soil erosion in India, thereby upsetting the natural balance and accelerating soil loss beyond equilibrium levels. Despite the impediments to sediment resulting from dam constructions, a copious amount of sediments is exported to the margin as evidenced by increasing slope instability and failure on the Godavari Delta, resulting from rapid deposition on the narrow shelf (e.g. Forsberg et al., 2007). Also, satellite images show a large plume of suspended river sediments from the mouth of the Godavari to the BoB (Sridhar et al., 2008). Using seismic and well log data to quantify the sediment (and associated OC) deposited on the margin and comparing the results to export flux of OC will provide a tight constraint on the net balance between sources and sink of carbon in this region. The nature and type of soil OC plays a crucial role in ascertaining if soils OC acts as a carbon source or sink in the coupled land-atmosphere-ocean carbon cycle. Findings

117 from this work show that human perturbation of the terrestrial ecosystem has led to the exhumation of “pre-aged” soil mineral-associated OC. However, the nature of this OC (i.e. pre-aged or aging during transport) is still largely unknown. To this end, a ramped temperature oxidation analysis (Hemingway et al., 2018) would provide information on the age-spectrum of the bulk OC that could prove vital in our understanding of carbon balance between the biospheric and geologic carbon cycle. For instance, if the mobilized soil OC is dominated by the petrogenic fraction, remineralization of this OC during riverine transport would constitute a net source of carbon. This is because petrogenic OC would have been presumably locked in the geologic carbon cycle for millions of years. The present work demonstrates that variations in monsoonal intensity markedly influence the dynamics of terrestrial carbon transport to the continental margin. While climate effects on the terrestrial carbon flow through the drainage basin can manifest in changes in terrestrial residence times or relative proportions fresh versus aged OC, disentangling one from the other requires development of robust analytical techniques. In this regard, ramp pyrolysis might offer significant insights. More comprehensive studies on organic matter recycling and mobilization by river systems (e.g. Galy and Eglinton, 2011) are necessary for a thorough understanding of the underlying processes involved in distinct types and sizes of river basins. OC fluxes and inventories across different continuum of rivers, estuaries, and river- dominated continental margins should be performed in order to identify major sites of remineralization, recycling, and exchange of OC (e.g. Bao et al., 2016) This will ultimately improve our understanding of export and exchange processes between terrestrial and marine pools.

References

Bao, R., McIntyre, C., Zhao, M.X., Zhu, C., Kao, S.J. and Eglinton, T.I. (2016) Widespread dispersal and aging of organic carbon in shallow marginal seas. Geology 44, 791-794. Forsberg, C.F., Solheim, A., Kvalstad, T.J., Vaidya, R., Mohanty, S. (2007) Slope instability and mass transport deposits on the Godavari River Delta, East Indian Margin from a regional geological perspective. In: Lykousis, V., Sakellaious, D., Locat, J. (eds.) Submarine Mass Movements and Their Consequences, 19-27.

118

Galy, V. and Eglinton, T. (2011). Protracted storage of biospheric carbon in the Ganges-Brahmaputra basin. Nature Geoscience 4, 843-847. Hedges, J. and Oades, J. (1997) Comparative organic geochemistries of soils and marine sediments. Organic Geochemistry 27, 319-361. Montgomery, D.R. (2007) Soil erosion and agricultural sustainability. Proceedings of the National Academy of Sciences 104, 13268-13272 Hemingway, J.D., Hilton, R.G., Hovius, N., Eglinton, T.I., Haghipour, N., Wacker, L., Chen, M.-C., Galy, V.V. (2018) Microbial oxidation of lithospheric organic carbon in rapidly eroding tropical mountain soils. Science 360, 209-212. Indian Ministry of Home Affairs (2011). Census Digital Library. Pradhan, U.K., Wu, Y., Shirodkar, P.V., Zhang, J. and Zhang, G.S. (2014) Multi-proxy evidence for compositional change of organic matter in the largest tropical (peninsular) river basin of India. Journal of Hydrology 519, 999-1009. Sridhar, P. N., Ali, M. M., Vethamony, P., Babu, M. T., Ramana, I. V., Jayakumar, S. (2008). Seasonal Occurrence of Unique Sediment Plume in the Bay of Bengal, Eos Transactions AGU, 89, 22–23.

119

Data Archiving/Access.

All data associated with this work has been archived in the ETH Zurich research collections and can be accessed via the following DOI:

doi:10.3929/ethz-b-000304504

120