Late Quaternary Paleolimnology in the Southern Hemisphere Tropics

LATE QUATERNARY PALEOLIMNOLOGY IN THE SOUTHERN HEMISPHERE TROPICS

Item Type text; Electronic Dissertation

Authors McGlue, Michael Matthew

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 03/10/2021 20:23:43

Link to Item http://hdl.handle.net/10150/204064 1

LATE QUATERNARY PALEOLIMNOLOGY IN THE

SOUTHERN HEMISPHERE TROPICS

Michael Matthew McGlue

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF GEOSCIENCES

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2011 2

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Dissertation Committee, we certify that we have read the dissertation prepared by Michael Matthew McGlue entitled ‚Late Quaternary Paleolimnology in the Southern Hemisphere Tropics‛ and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy

______Date: January 31, 2011 ANDREW COHEN

______Date: January 31, 2011 PETER DECELLES

______Date: January 31, 2011 VANCE HOLLIDAY

______Date: January 31, 2011 ROY JOHNSON

______Date: January 31, 2011 JAY QUADE

Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

______Date: January 31, 2011 Dissertation Director: ANDREW COHEN 3

STATEMENT BY AUTHOR

This document has been submitted in partial fulfillment of requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this document are allowable without special permission, provided that accurate acknowledgement of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the author.

SIGNED: Michael Matthew McGlue

ACKNOWLEDGEMENTS

Mom, Dad and Ken – I love you.

Friends make graduate school bearable. Christine Gans will forever hold a special place in my heart. Sarah Ivory always reminded me to keep looking up. Mark Trees is one of the finest men I have ever known. Andrew and Astrid Kowler made me feel like a part of their family. Kiram Lezzar helped me find some purpose. Anna Felton and Meg Blome always tried. Chris Dyson, Keith Rauch, Josh Carpenter, Mike Sandusky, Ted Dickson, Veronica Langhofer and Graig Avino have been there through it all and kept me laughing. Jules Omarini, Diego Munoz, Martin Plaza, Aguinaldo Silva, Hiran Zani, Tito Corrandini, Fred Gradella, Sidney Kuerten, Beatriz Lima de Paula, Preston Smith, Erin Abel, Erin Gleeson, John Mischler, Majie Fan, Xiaoyu Xiang, the staff at LacCore, and the staff at UA AMS are new friends I’ll not soon forget.

Money drives research. I am especially thankful to AAPG, GSA, Sigma Xi, UA- GPSC, Robert Bulter, Karchner Caverns, the Chevron Corporation and LacCore for providing small grants that aided my efforts. Larger grants from NSF, ExxonMobil, ACS-PRF and FAPESP made international fieldwork a possibility; Andy Cohen, Peter DeCelles and Mario Assine were instrumental in securing these funds and making them available to me. Chris Eastoe, David Dettman, Tim Jull and Peter Swarzenski helped keep analytical costs down and I am grateful to them. Geoff Ellis and Peter Rumelhart deserve special mention for all of their efforts on my behalf – both have my respect and thanks.

Teaching is really important. Owen Davis, Andy Cohen and Sarah Holmes gave me the opportunity to teach, and I am grateful to them for sharing their expertise.

The staff makes the Department work. To Eneida Guerra de Lima, Sharon Bouck, Kiriaki Xiluri, Norm Meader, Anne Chase, and Heather Alvarez – thank you for making my life much easier.

Eventually one must graduate. My committee (Andy Cohen, Jay Quade, Peter DeCelles, Vance Holliday and Roy Johnson) helped make that possible. Cheers!

DEDICATION

This dissertation is dedicated to Mark Trees, Kelly Wendt, Aguinaldo Silva and Fabrício Corrandini – without their courage and resolve – I am quite certain this dissertation never would have been completed. and to Molly McGlue (1992 – 2004).

TABLE OF CONTENTS

ABSTRACT ______8

INTRODUCTION ______10

PRESENT STUDY ______19

REFERENCES ______32

APPENDIX A: SEISMIC RECORDS OF LATE PLEISTOCENE ARIDITY IN LAKE TANGANYIKA, TROPICAL EAST AFRICA ______41 Permission to reprint from the copyright holder ______41 Abstract ______42 Introduction ______43 Description of study area ______47 Methods ______50 Results ______51 Discussion ______57 Conclusions ______74 Acknowledgements ______76 References ______77 Figures ______86 Tables ______98

APPENDIX B: ENVIRONMENTAL CONTROLS ON SHELL-RICH FACIES IN TROPICAL LACUSTRINE RIFTS: A VIEW FROM LAKE TANGANYIKA’S LITTORAL ______99 Permission to reprint from the copyright holder ______99 Abstract ______100 Introduction ______101 Background ______103 Methods ______106 Results ______108 Interpretations ______115 Discussion ______124 Conclusions ______128 Acknowledgements ______130

TABLE OF CONTENTS – Continued

References ______131 Figures ______139 Tables ______148

APPENDIX C: PLAYA-LAKE SEDIMENTATION AND ORGANIC MATTER ACCUMULATION IN AN ANDEAN PIGGYBACK BASIN: THE RECENT RECORD FROM THE CUENCA DE POZUELOS, NW ARGENTINA ______152 Abstract ______153 Introduction ______154 Regional Overview ______156 Methods and Materials ______159 Results and Interpretations ______162 Discussion ______166 Conclusions ______172 Acknowledgements ______175 References ______176 Figures ______184 Tables ______191

APPENDIX D: LIMNOGEOLOGY IN BRAZIL’S ‚FORGOTTEN WILDERNESS": A SYNTHESIS FROM THE GREAT LAKES OF THE PANTANAL ______197 Abstract ______198 Introduction ______199 Methods and Materials ______206 Results ______209 Discussion ______216 Conclusions ______228 Acknowledgements ______230 References ______232 Figures ______238 Tables ______246

ABSTRACT

Lake deposits are widespread throughout the Phanerozoic rock record and have long intrigued geologists and paleobiologists in search of natural resources or fossil biota. Low-energy lacustrine depositional environments, characterized by relatively rapid sediment influx rates and shallow zones of bioturbation, likewise produce highly-resolved archives of climate and ecosystems evolution.

This dissertation describes four studies that use lake sediments for Quaternary environmental analysis.

In East Africa, many decades of prior study provided the critical framework necessary for in-depth paleoenvironmental research at Lake

Tanganyika (3° – 9°S). Seismic stratigraphic analysis integrated with radiocarbon- dated sediment cores from the Kalya horst and platform document a dramatic lake level lowstand prior to ~106 ka and a minor, short-lived regression during the Last Glacial Maximum (32 – 14 ka). Paleobathymetric maps reveal that Lake

Tanganyika remains a large, connected water body even during episodes of extreme drought, which has implications for local and regional fauna. Over shorter timescales, geochronological, taphonomic and sedimentological analyses of shell beds around Kigoma (central Lake Tanganyika) document three distinct facies-types that are time-averaged over the latest Holocene. Lake level 9

fluctuations associated with the termination of the Little Ice Age (~ 16th century

CE) and subsequent encrustation played a key role in shell bed formation and persistence along high-energy littoral platforms, which has implications for structuring specialized communities of benthic fauna.

In central South America (18° – 22°S), we studied the limnogeology of small lakes in the Puna and the Pantanal. Analyses of these sites were undertaken to: 1) ascertain how the lakes act as depositional basins; 2) assess sedimentation rates; and 3) construct limnogeological databases to guide future interpretations of ancient sediment cores. At Laguna de los Pozuelos (Argentine

Puna), linear sedimentation rates approach 0.14 cm*y−1 in the playa-lake center, and litho- and organo-facies development are dominantly controlled by basin hydrology, climate and biological feedbacks (both nutrient cycling and bioturbation) from waterbirds. At Lagoas Gaíva, Mandioré and Vermelha

(Brazilian Pantanal), short-lived radioisotopes indicate uninterrupted depositional rates of 0.11 – 0.24 cm*y− 1, and hydrochemical and depositional patterns respond sensitively to changes in the seasonal flooding cycle of the Upper Paraguay River.

INTRODUCTION

Lake deposits are global in distribution and span the Phanerozoic. Lake sediments have been used to explain a number of important early themes in geology (e.g., the Ice Age hypothesis from lacustrine varves; Lyell, 1830) and they to continue provoke interest from oil and gas geologists, both as traditional source rocks and unconventional gas reservoirs (e.g., shale gas systems).

Pioneering work by Gilbert (1890) on the geology of Lake Bonneville and by

Forel (1901) on the limnology of Lake Geneva provided the motivation and theoretical foundation for more than a century of paleolimnological studies.

Paleolimnology is the study of geological, chemical and biological aspects of freshwater systems with a goal of determining environmental history and ecosystems evolution (e.g., Cohen, 2003). Climate change is perhaps the most important environmental issue facing our society, and developing a detailed understanding of the response of freshwater systems to past perturbations in temperature and precipitation is critical for: 1) assessing the impact of global climate events on local hydrologic networks; 2) calibrating regional climate models to accurately predict future change; and 3) sustainability planning and resource conservation efforts (Smol, 2002). Each of the dissertation projects described in the appendices use lake sediments to reconstruct paleoenvironments 11

in the southern hemisphere tropics. The tropics are vitally important to global atmospheric circulation (driving the Earth’s heat and moisture budgets) and a key region with respect to water resources and rapidly developing human populations

(e.g., McGregor and Nieuwolt, 1998; Yin and Battisti, 2000). Certain areas of the tropics have enjoyed a long history of scientific inquiry, particularly in limnology, geology and vegetation (e.g., East Africa; de Heinzelin, 1952; Bishop, 1960;

Livingstone, 1962; Hecky and Degens, 1973; Hay, 1976). In these regions, paleolimnology can provide quantitative insights on tropical climate dynamics.

My field locality at Lake Tanganyika (East Africa) provides an example (e.g.,

O’Reilly et al., 2003; Cohen et al., 2006; Tierney et al., 2010). By contrast, a number of expansive areas where deep atmospheric convection fundamentally influences global climate have escaped detailed limnogeological analyses. My studies in Brazil and Argentina attempt to follow the example set by early researchers in East Africa – to develop a broad understanding of the controls on depositional processes in order to use components in these lake sediments for rigorous assessments of Quaternary environmental history.

Lake Evolution

Two fundamental requirements exist for the formation of lacustrine environments of deposition: a source of water and the development of 12

topographic closure (e.g., Sladen, 1994; Carroll and Bohacs, 1999; Cohen, 2003).

For convenience, I define topographic closure as a depression capable of accommodating water and sediment over a geologically meaningful period of time (> 102 yrs). This definition is not, however, meant to convey stasis of the depression through time, nor does it imply complete hydrological closure.

Indeed, water levels for many of the world’s large tectonic lakes are at or above basin sill elevation (e.g., Lakes Tanganyika and Malawi; Scholz and Rosendahl,

1988).

The lakes studied during the course of my dissertation fall into two different categories: tectonic lakes (Lake Tanganyika and Laguna de los Pozuelos) and floodplain lakes (lakes of the Pantanal). In the broadest sense, topographic closure and lake formation in tectonic lake basins is strongly influenced by deformation of the lithosphere due to plate boundary processes. Topographic closure evolves non-linearly as a function of basin subsidence and is subsequently modulated by sediment and water influx. For example, active extension in tropical East Africa, resulting from mantle plumes and propagation of associated tensional stresses into Proterozoic mobile belts, drives thinning of the lithosphere and creation of rift basins by a combination of mechanical and thermal subsidence (Ebinger, 1989; Ebinger and Sleep, 1989; Nyblade and Brazier, 2002). 13

Lake formation accompanies this deformation as rift shoulders uplift and drainages re-organize (Gawthorpe and Leeder, 2000). Extant rift lakes in East

Africa (such as Lake Tanganyika) and Siberia (Lake Baikal) demonstrate the intimate relationship between half-graben geometry and lake shape, especially in areas where post-rift volcanism or anthropogenic activity has not altered basin shape or regional hydrologic networks. Shoreline development and bathymetry generally follow rift structure, resulting in long, narrow lakes with deep water focused adjacent to steeply subsided border faults and shoaling platforms common on flexural margins (e.g., Lambiase, 1990; Soreghan and Cohen, 1996).

Structural deformation associated with the overlapping or intersecting normal fault tips can produce substantial lake floor relief (‚accommodation zones‛ of

Rosendahl, 1987). Sedimentation atop these isolated structural highs is typically slow and dominated by suspension settling of autochonous organic matter, producing time-rich stratigraphic sequences favorable for paleoclimate studies

(e.g., Scholz et al., 2001; Felton et al., 2007). Numerous examples in the rock record (e.g, the Cretaceous Lagoa Feia Formation of Brazil; Abrahão and Warme,

1991) confirm observations made from modern lakes.

The evolution of topographic closure in foreland basins is more complex, due in part to multiple mechanisms of subsidence whose rates may vary both 14

spatially and in time, and the overprint of climate, which fundamentally alters weathering and the transport of sediment (e.g., Burbank, 1992). DeCelles and

Giles (1999) noted the existence of four depozones (wedgetop, foredeep, forebulge and backbulge) in retro-arc foreland basin systems, consistent with the geodynamic response of an elastic crust to shortening and loading in an adjacent thrust belt. However, tropical weathering of weakly resistant hinterland lithologies typically prevents topographic closure, as sedimentation rates exceed long-term subsidence rates (~ 10-1 mm/yr; Sinha and Friend, 1994; Aalto et al.,

2003). Rather, expansive low-gradient fluvial megafans typically dominate terrestrial depositional systems in continental foreland basin systems (Horton and

DeCelles, 2001; Leier et al., 2005). For example, modeling by Pelletier (2007) predicts higher sediment flux rates and strongly overfilled basins in the humid sub-tropical latitudes of the modern Andean foreland basin system. Large tectonic lakes are absent from these latitudes, except in endorheic wedgetop basins of the arid Puna plateau (McGlue and Cohen, 2006). The clearest examples of large lakes occupying foreland basins are associated with thick- skinned deformation, where thrust faulting leads to the formation of flexural basins fringed by basement-cored highlands and re-oganization of hydrologic networks. Some of the largest and best studied ancient lake deposits (e.g., the 15

Eocene Green River lakes of Western North America; Eugster and Hardie, 1975;

Pietras and Carroll, 2006; Smith et al., 2008) occupied foreland basins adjacent to basement-cored uplifts. Nevertheless, climatic modulation of erosion (persistent aridity or rain shadowing) or lithologies promoting a high dissolved load to bed load ratio created large foredeep lakes adjacent to the thin-skinned Sevier thrust belt (e.g., Cretaceous Lakes Draney and Peterson; Zaleha, 2006) and in South

America (e.g., Jordan et al., 2001). Expansive lakes and wetlands may also develop in low-gradient backbulge depressions, although detailed mapping and facies characterizations are lacking.

In floodplain lakes, fluvial processes are typically implicated in the development of topographic closure. Cohen (2003) noted that uni-directional flows create topographic closure by excavating depressions or depositing sedimentary barriers. Lake systems develop when: 1) bank full conditions are exceeded during seasonal floods; 2) groundwater infiltrates through hydromorphic soils; 3) crevasse splays breach extant river channels, flooding the proximal floodplain and 4) rainfall and runoff accumulate in these depressions.

The form and depth of floodplain lakes are typically controlled by the adjacent river (Cohen, 2003). This is especially true for oxbow lakes (neck cutoffs), which occupy abandoned channels and mark the wide floodplains of many lowland 16

meandering and anastamosing rivers. Both tectonic deformation (e.g., Holbrook and Schumm, 1999; Holbrook et al., 2006) and climate (Blum et al., 2000;

Rittenour et al., 2007) can cuase major channel belt re-organizations in alluvial rivers and dating different vintages of floodplain lakes has proven valuable for deciphering the environmental histories of large rivers.

Lacustrine Deposits

The sedimentary deposits of lacustrine systems are a valuable component of the Phanerozoic rock record. Tectonic lakes, unlike lakes formed by most other processes, commonly persist on the landscape for 105 – 107 yrs and can produce thick packages of strata containing archives of climatic, biological and surficial processes or natural resources (e.g. Olsen, 1986; Bohacs et al., 2000;

Johnson et al., 2002; Carroll et al., 2006). Modern rift lakes have received significant limnogeological study and their depositional processes are generally well understood (e.g., Johnson et al., 1987; Cohen, 1989; Scholz, 1995; Lezzar et al.,

1996; McGlue et al., 2006). Driven by the need for modern analogs for

(predominantly) Cretaceous hydrocarbon source rocks, these studies have greatly aided the interpretation of lacustrine strata in extensional basins (e.g. Lambiase,

1990). In contrast, modern foreland basin lakes present a data gap, as few limnogeological studies have been attempted (see Piovano et al., 2002 for an 17

exception). In this regard, accurate interpretations of lacustrine strata in the rock record are difficult, because the full range of controls on basin development and deposition in modern systems are not well understood. Issues related to preservation and exposure commonly confronting geologists and paleontologists in the field only exacerbate this problem.

Facies models for floodplain lakes evolved from critical analyses of modern oxbow lake stratigraphy in North America. Generally, three patterns of sedimentary infill are recognized for abandoned channels (oxbow lakes; Blum et al., 2000). The most common is a simple clay plug that forms following rapid neck-cutoff of a meander loop or channel segment in a system characterized by high suspended load. Clays and organic particles settle from suspension following individual flood events in these types of lakes. Or, oxbows may contain sands and silts interbedded with clays. Abandoned channels characterized by interbedded stratigraphy typically develop as an active fluvial channel migrates across the paleo-channel belt, depositing crevasse splay sands or levee facies.

Finally, oxbow infill is dominated by sand when channels are abandoned slowly

(typical of bedload dominated rivers); such is the case in channels marked by chute cutoffs. Oxbow fill fines upward in these systems, but the clay plug is generally thin. Paleoenvironmental inferences from floodplain lakes are 18

constrained by the depth of the channel, which controls the volume of the sedimentary infill, rapid depositional rates, and the potential for reworking by channel migration (Cohen, 2003). The lakes of the Pantanal do not appear to have developed from neck or chute cutoffs, however and their development on the Upper Paraguay floodplain may have been influenced by non-riverine processes. Their stratigraphy provides a new end member for floodplain lake systems.

PRESENT STUDY

The papers appended to this dissertation (A - D) provide complete details on the locality, methodology, results and conclusions for each study. What follows below are brief overviews of the individual projects.

Appendix A

At Lake Tanganyika, intermediate-resolution seismic reflection and long sediment core datasets were collected from the Kalya horst and platform region near the geographic center of the lake in 2004, during a research cruise associated with the Nyanza Project, an NSF-sponsored REU program hosted by the University of Arizona and my advisor Andrew Cohen. Collected with a marine sparker system adapted for a freshwater environment, the seismic reflection data were initially used to guide the coring program, in an effort to avoid shallow unconformities. These cores have formed the basis of a number of recent, important paleoclimate studies on the East African tropics (Felton et al.,

2007; Tierney and Russell, 2007; Tierney et al., 2008). Following completion of the coring program, the seismic data were set aside until I began my doctoral studies at the University. Given my background in stratigraphy and high- resolution geophysics, the data were provided to me as a research assistantship project during my first semester in Tucson by co-authors Andy Cohen and Kiram 20

Lezzar. Using laboratory facilities graciously provided by Roy Johnson, I processed the near channel of the seismic data using PROMAX © to enhance the signal to noise ratio, and built a project for their interpretation in SEISWORKS ©

(donated to the University by the Landmark Corporation) under the guidance of

Steve Sorensen and Roy Johnson. In most cases, the quality of the data was very high, especially in the upper 200 ms TWTT. Using basic techniques in lacustrine seismic sequence stratigraphy (e.g. Scholz, 2001), I developed a stratigraphic framework comprised of four sequences (A-D). On the Kalya platform, unconformities and stratal geometries were especially well expressed, and punctuated by stacked prograding clinoforms. These clinoforms were interpreted as deltaic sand bodies deposited during a major lake level lowstand, and the geomorphic expression of these features (specifically, the topset-to-foreset transition of the deepest clinoform) indicated paleowater depth was reduced by

~435 m compared to the modern. Paleobathymetric maps were developed from these and other geomorphic features on the platform, providing novel insights on the response of Lake Tanganyika to millennial-scale aridity (prolonged drought), which has implications for the lake's endemic biota. Likewise, modern bathymetry and lake volume estimates were revised, correcting older datasets widely cited in the literature. Importantly, integration of the seismic data with 21

well-dated sediment cores (age model development for cores KH3 and KH4 was completed by co-authors Anna Felton, Jim Russell, and Andy Cohen) provided age control for these major transgression-regression events, and correlation with regional records (e.g., Scholz et al., 2007) suggests water levels in Lake

Tanganyika respond to changes in insolation associated with orbital forcing, especially during MIS 5. This work was published in the Journal of

Paleolimnology in 2008. Because Lake Tanganyika may be a future paleoclimate drilling target due to its depth, great antiquity and bedrock ridges that slowly accumulate sediment, future work may include the extension of the seismic stratigraphic framework developed in this project into the southern basin of the lake.

Appendix B

Research on Lake Tanganyika’s littoral shell beds occurred over two field seasons of the Nyanza Project (2005 – 2006). I participated directly in the sampling during 2005, and helped to guide the collection of data in 2006 in collaboration with co-author and Nyanza mentor Mike Soreghan. My interest in this project was provoked by my own sedimentological research at Lake Edward,

East Africa, which revealed a series of pronounced lowstand deltaic deposits, capped by carpets of empty gastropod and clam shells. I interpreted these 22

accumulations as the product of winnowing, due in large part to the substantial bathymetric relief created by the paleo-deltas (conditions favorable for the development of contour currents), and the heavy, black, manganese coatings on the shells, indicating prior burial in a reducing and potentially anoxic environment (McGlue et al., 2006). Extensive shell beds had been reported from

Plio-Pleistocene outcrops adjacent to Lakes Albert and Turkana (e.g., Williamson,

1981; Feibel, 1987; Van Damme and Pickford, 1999) and new hydrocarbon exploration targeting similar accumulations in Cretaceous sub-salt rift basins along the Brazilian coast inspired me to dig deeper into the genesis and persistence of these nearshore freshwater carbonates. It was clear that the accumulations in Lake Edward shared some similarities to the more widespread coquinas in Lake Tanganyika, initially described by Cohen and Thouin (1987) and

Cohen (1989). For example, Cohen (1989) interpreted Lake Tanganyika’s nearshore molluscan coquinas as the products of winnowing, initiated through

Holocene regression events. My approach at Lake Tanganyika attempted to mesh sedimentological, geochronological, and paleontological techniques to: 1) characterize facies variability among shell beds encountered along the various littoral environments accessible from Kigoma and 2) assess the post-mortem survivability of Neothauma tanganyicense, as these large gastropod shells 23

provide crucial habitat for a number of specialized benthic organisms. Studies of marine shell beds, far more extensive in the literature, followed a similar tack

(e.g., Davies et al., 1989; Best and Kidwell, 2000). I recognized three distinct facies, based on clast composition, external accumulation geometry, shell packing, and thin section characteristics: gravel-rich mollusk hash, sandy and silty mollusk hash, and pure mollusk hash. Depositional style varied with lake- floor gradient and water depth, ranging from slope-front aprons and patches

(embayments and headlands, respectively) to beach ridges and expansive beds

(low-gradient deltaic platform environments). Taphonomic data indicate that N. tanganyicense are relatively damage resistant, except when transported into the shallow or supra-littoral zone by wave action or fish, as sand abrasion rapidly erodes the shell structure. Radiocarbon dating yielded a range of late Holocene ages, suggesting the deposits are time averaged over at least 1600 cal yrs BP. In concert with radiocarbon data, inferences from sediment cores suggested that lake level dyanmics (and associated changes in wave base) accompanying the termination of the Little Ice Age (16th – 19th centuries) may have concentrated the

N. tanganyicense into beds. The persistence of N. tanganyicense is likely assisted by rapid development of stromatolitic encrustations and early cement coatings, which help reduce post-mortem shell destruction by mechanical processes. The 24

results of this study are important for a number of reasons. First, the sedimentological analysis discovered substantial spatial variability in facies, which will be useful for interpreting the ancient rock record and in the search for new hydrocarbon reservoirs. Second, the genetic controls on pure gastropod hash development were tied to processes associated with both transgression and regression, which represents a departure from prior interpretations focusing exclusively on lowstand winnowing. Finally, shell survival in the littoral zone may have important implications for structuring specialized benthic communities within Lake Tanganyika, and if so provides a potentially unique example of lacustrine taphonomic feedback. This work was published in the journal Palaios in 2010. Future research on these shell beds will focus on expanding estimates of time averaging through new radiocarbon and amino acid dating, new sampling to address the cause of live:dead disagreement and a pilot study on the nitrogen isotopic composition of shell carbonate to track anthropogenic influence on N. tanganyicense ecology.

Appendix C

Geological and paleolimnological research in the Cuenca de Pozuelos

(Jujuy Province, northwestern Argentina) began in 2006, and two field seasons were spent collecting surface sediments, lake water samples and sediment cores 25

from this high-altitude (~3650 m a.s.l), arid piggyback basin. The initial goals of this research focused on gaining a better understanding of the sedimentology and shallow stratigraphy of Laguna de los Pozuelos, a playa-lake located near the center of the flat-floored, thrust fault bound basin. In spite of its recognition as a wetland of international importance (RAMSAR, 2009) and increasing pressure from mining activities in the region, little is known about the geology and limnology of this large lentic environment. Moreover, few studies have considered modern foreland basin lakes from the perspective of basin analysis.

In anticipation of a future study on late Quaternary stratigraphic evolution of the basin, I used a large suite of surface sediment samples to characterize modern sedimentary environments encountered in the field (specifically, playa-lake center, playa margin mudflats, fringing alluvial fans and axial deltas). Depending climate, watershed characteristics, and basin morphology, lake sediments are generally composed of a combination of terrigenous particles (sand, silt, and clay), carbonate, organic matter and biogenic silica (typically produced by diatoms, sponge spicules and plant phytoliths). These components were evaluated using standard techniques (e.g., Mortlock and Froelich, 1989; Johnson and

McCave, 2001). Basic information on sedimentation rate was obtained using a short (< 1.5 m) core collected from the playa-center environment and 26

radioisotopes (210Pb and 137Cs) by co-author Peter Swarzenski. Considerable time was spent assessing bulk organic matter geochemistry using stable isotopes (and

Rock Eval pyrolysis data, provided by co-author Geoff Ellis), as my intention was to develop similar indicator records on long cores. The focus on sedimentary organic matter was driven by the availability of facilities at the University, and the need for a better understanding of the carbon cycle in playa lakes situated on the Andean Puna. A number of active debates center on the late Quaternary climate history of the high Andes, and clearly modern calibration datasets are critical for meaningful paleoenvironmental reconstructions (e.g. Baker et al.,

2001; Plackzek et al., 2006; Fritz et al., 2007; Quade et al., 2008). I produced a series of concentration contour maps that depict the spatial distribution of major sedimentary components. These maps demonstrate that organic carbon and biogenic silica deposition are concentrated in the basin axis, whereas carbonate minerals are found in greater abundance along the playa’s evaporative (mud flat) margins. Axial and lateral margins show increases in particle size, consistent with active deltaic and alluvial fan deposition during the austral summer. Bulk sediment and organic mass accumulation rates approach 0.22 g cm- 2*y- 1 and 2.89 mg cm- 2*y- 1 respectively, indicating moderately rapid deposition with negligible deflation over historic time in the playa-lake center. Geochemical datasets 27

indicate that organic facies development is dominantly controlled by basin hydrology, climate and biological feedbacks (both nutrient cycling and

13 bioturbation) from waterbirds. Both δ COM and C:N data indicate a mixed provenance for the organic matter preserved in sediments. The carbon stable isotope data range from -16.4 to -24.6 ‰, probably in response to substantial

contributions from perennial C4-pathway aquatic plants that have mixed with algae in the playa-lake center. Samples from the playa margins and those

offshore from the deltas likely contain OM delivered from terrestrial (C3) vegetation. The C:N data indicate a substantial contribution from higher plants to the carbon cycle in LP, as well as early diagenetic losses of labile nitrogen from algae. This paper was submitted to the journal Sedimentology and was accepted pending revision in December, 2010. Future research at Laguna de los Pozuelos includes the completion of a late Quaternary stratigraphic evolution and paleoenvironmental analysis of several long cores collected in 2007 and the development of sedimentary records of recent pollution, including total mercury and lead isotopes.

Appendix D

The Pantanal (Portuguese for ‚swampland‛) is a vast tropical lowland wetland situated along Brazil’s western frontier (~ 18°S). A large Quaternary-aged 28

basin linked to Andean tectonism (Horton and DeCelles, 1997; Chase et al., 2010), the Pantanal forms the watershed of the Upper Paraguay River and is perhaps best known for the spectacular biodiversity of its fish, birds and reptiles

(Heckman 1998). Only a few Quaternary geological studies exist for the region

(e.g., Ussami et al., 1999; Assine and Soares, 2004). However, the commonplace earlier geological characterizations of the Pantanal as a simple, homogenous swamp are inaccurate for a number of reasons, including: 1) strong seasonal variability in hydrologic cycle of the Upper Paraguay River, which spans the basin longitudinally; 2) patterns of precipitation and temperature, which exhibit a strong north-south gradient; and 3) landscape morphology, with an increasing proportion of features indicative of drier paleoclimates to the south (Por, 1995;

Hamilton et al., 1997; Soares et al., 2004; Assine and Silva, 2009). The response of tropical wetlands to global environmental change is receiving increasing attention, especially because of concerns over greenhouse gas fluxes associated with enhanced methanogenesis and/or biomass burning in a warmer world (e.g.,

Shindell et al., 2004). Yet in spite of clear evidence of major shifts in regional paleohydrology (dunes, lunettes and paleosols) in the southern Pantanal, few in- depth paleolimnological analyses have been attempted in the region (see Oliveira

Bezerra and Mozeto, 2008 for an exception). The goal of this study was to assess 29

the limnogeology of several of the large floodplain lakes situated along the western margin of the Upper Paraguay River, with an eye cast towards using sediment cores from these lakes for future Quaternary paleoenvironmental reconstruction. To this end, I spent three field seasons in Brazil collecting lake water, surface sediments and short sediments cores from Lagoa Gaíva, Lagoa

Mandioré and Baia Vermelha. Initially, I produced bathymetric maps and water chemistry datasets for each lake, and discovered that the lakes were freshwater

(dilute), well-oxygenated, weakly basic systems with elongate basin shapes and shallow maximum water depths (< 5 m). In a pioneering early study, Junk et al

(1989) noted the importance of the arrival of seasonal riverine flood waters to wetland productivity and biodiversity in the tropics. It became clear after my initial field season (late in the austral spring, when the lakes are low and difficult to access at their sill channels) that certain sedimentary processes in the lakes were also dependant on arrival of the Upper Paraguay’s flood pulse. I constructed a series of lake-bottom contour maps in order to investigate modern facies relationships, to include patterns of organic matter and biogenic silica accumulation. In each of the lakes, basin sill points are marked by the apices of coarse sandy fans; such deposits are interpreted to reflect strong linkage to the

Upper Paraguay River flood pulse and traction flow processes. In contrast, 30

accumulation of fine-grained sediments is most pronounced in deep water areas and dominated by suspension settling of organic particles. Sediment biogeochemistry suggests that production and burial of biomass is highest in

Lagoa Mandioré and somewhat lower in Lagoa Gaíva and Baia Vermelha, likely due to morphometrics that favor strong mixing and intrabasinal nutrient recycling in the former. I developed carbon and nitrogen stable isotope datasets for the lake bottom samples, and used smear slide analysis to verify organic composition where possible. Geoff Ellis (USGS) provided Rock Eval pyrolysis data to help further discriminate the provenance of organic matter. In concert, the data indicate organic contributions from various lacustrine algae and vascular plants in each of the lakes. Peter Swarzenski attempted 210Pb analysis on each of the lakes, but only Lagoa Gaíva has measurable accumulations of unsupported 210Pb and therefore sedimentation rate data are only available for one lake (0.11 – 0.24 cm*y− 1). Mauro Parolin (Faculdade Estadual de Ciências e

Letras de Campo Mourão) lent his expertise in tropical sponge ecology to the project, and he noted the presence of gemmoscleres of C. secktii, as well as spicules of O. navicella and U. corallioides in the lakes. These sponges further implicate the role of the Upper Paraguay River flood pulse in the composition of biogenic sediments at the lakes. Sensitivity of the Paraguay River flood pulse to 31

climate change, neotectonics and fluvial autocyclic processes clearly has important implications for the limnogeology of Lagoa Gaíva, Lagoa Mandioré and Baia Vermelha. Accordingly their sediments are important sentinels of environmental change for the Pantanal, especially in Lagoa Gaíva, where water depths and morphometrics provide conditions favorable for continuous sedimentation. Actualistic datasets and space-for-time considerations provide a predictive framework for facies migrations with major changes to the Paraguay

River flood pulse. The rate and continuity of sedimentation will vary considerably as major hydrologic thresholds are crossed. Intensification of the flood pulse is likely to be represented by higher lake levels and concomitant decrease in grain size, an increase in the burial of organic matter due to elevated productivity, highstand deltas and lotic sponge assemblages. By contrast, weakening of the flood pulse promotes chemical sedimentation, reworking of the lake bottom by bioturbation, lowstand fan deposition and oxidation of organic matter as the lakes become isolated on the floodplain. The final version of this conceptual facies model was submitted for publication in the Journal of

Paleolimnology in February, 2011. Future work includes application of the model to the interpretation of Holocene-aged sediment cores from Lagoas Gaíva and Mandioré. 32

REFERENCES

Aalto, R., Maurice-Bourgoin, L., Dunne, T., Montgomery, D.R., Nittrouer, C.A., Guyot, J.L., 2003, Episodic sediment accumulation on Amazonian flood plains influenced by El Nino/Southern Oscillation: Nature v. 425, 493–497.

Abrahão, D., and Warme, J.E., 1990, Lacustrine and associated deposits in a rifted continental margin: Lower Cretaceous Lagoa Feia Formation, Campos Basin, offshore Brazil, in Katz, B.J., ed., Lacustrine Basin Exploration—Case Studies and Modern Analogs: American Association of Petroleum Geologists Memoir, v. 50, 287–305.

Assine, M,L. and Soares, P.C., 2004, Quaternary of the Pantanal, west-central Brazil: Quaternary International, v. 114, 23–34.

Assine, M.L., and Silva, A., 2009, Contrasting fluvial styles of the Paraguay river in the northwestern border of the Pantanal wetland, Brazil: Geomorphology, v. 113, 189-199.

Baker, P.A., Rigsby, C.A., Seltzer, G.O., Fritz, S.C., Lowenstein, T.K., Bacher, N. and Veliz, C., 2001, Tropical climate change at millennial and orbital timescales on the Bolivian Altiplano: Nature, v. 409, 698-701.

Best, M.M.R. and Kidwell, S.M., 2000, Bivalve taphonomy in tropical mixed siliciclastic–carbonate settings. I. Environmental variation in shell condition: Paleobiology, v. 26, 80–102.

Bishop,W.W., 1969, Pleistocene Stratigraphy of Uganda: Geological Survey of Uganda, Memoir X, 128 p.

Bohacs, K.M., Carroll, A.R., Neal, J., Mankiewicz, P., 2000, Lake Basin Type, Source Potential, and Hydrocarbon Character: An Integrated Sequence-stratigraphic- geochemical Framework. In Gierlowski-Kordesch E, Kelts K (eds) Lake Basins through Space and Time: AAPG Studies in Geology 46, 3–34.

Blum, M.D., Guccione, M.J., Wysocki, D.A., Robnett, P.C., Ruttledge, M., 2000, Late Pleistocene evolution of the lower Mississippi Valley, Southern Missouri to Arkansas: Geological Society of America Bulletin, v. 112, 221–235.

Burbank, D., 1992, Causes for recent uplift deduced from depositional patterns in the Ganges basin: Nature, v. 357, 48-50.

Carroll, A.R., and Bohacs, K.M., 1999, Stratigraphic classification of ancient lakes: Balancing tectonic and climatic controls: Geology, v. 27, 99–102.

Carroll, A.R., Chetel, L. M., and Smith, M. E., 2006, Feast to famine: Sediment supply control on Laramide basin fill: Geology, v. 34, 197-200.

Chase, C.G., Sussman, A.J., and Coblentz, D.D., 2010, Curved Andes: Geoid, forebulge, and flexure: Lithosphere, v. 1, 358-363.

Cohen, A.S., 1989, The taphonomy of gastropod shell accumulations in large lakes: An example from Lake Tanganyika: Paleobiology, v. 15, 26–45.

Cohen, A.S., 1989, Facies Relationships and Sedimentation in Large Rift Lakes: Examples from Lakes Turkana and Tanganyika: Paleogeography, Paleoclimatology, Paleoecology, v. 70, 65-80.

Cohen, A.S., Lezzar, K.E., Cole, J., Dettman, D., Ellis, G.S., Eagle, M., Plisnier, P.D., Langenberg, V., Blaauw, M. and Zilifi, D. 2006, Late Holocene linkages between decade-century scale climate variability and productivity at Lake Tanganyika, Africa: Journal of Paleolimnology

Cohen A.S., 2003, Paleolimnology: the history and evolution of lake systems. Oxford, UK: Oxford University Press.

Davies, D. J., Powell, E.N., and Stanton, R.J., 1989, Taphonomic signature as a function of environmental process: shells and shell beds in a hurricane- influenced inlet on the Texas coast: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 72, p. 317–356.

DeCelles, P.G. and Giles, K.A., 1996, Foreland basin systems. Basin Research, v. 8, 105-123. 34

Ebinger, C.J., 1989, Tectonic development of the western branch of the East African Rift System: Geological Society of America Bulletin, v. 101, 885–903.

Ebinger C.J. and Sleep N.H., 1998, Cenozoic magmatism throughout east Africa resulting from impact of a single plume: Nature, v. 395, 788–91.

Eugster, H. and Hardie, L.A., 1975, Sedimentation in an ancient playa-lake complex: The Wilkins Peak Member of the Green River Formation of Wyoming: Geological Society of America Bulletin, v. 86, 319-334.

Feibel, C.S., 1987, Fossil fish nests from the Koobi Fora Formation (Plio– Pleistocene) of northern Kenya: Journal of Paleontology, v. 61, 130–134.

Felton A.A., Russell J.M., Cohen, A.S., Baker, M.E., Chesley, J., Lezzar K-E, McGlue, M.M., Pigati, J.S., Quade, J., Stager, J.C., 2007, Paleolimnological evidence for the onset and termination of glacial aridity from Lake Tanganyika, East Tropical Africa: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 252, 405–423.

Forel, F.A., 1901, Handbuch der Seekunde. Allgemeine Limnologie, Stuttgart.

Fritz, S.C., Baker, G., Seltzer, G.O., Ballantyne, A., Tapia, M., Cheng, H. and Edwards, R.L., 2007, Quaternary glaciation and hydrologic variation in the South American tropics as reconstructed from the Lake Titicaca drilling project: Quaternary Research, v. 68, 410-420.

Hay, R. L. 1976, Geology of the Olduvai Gorge. Berkeley: University of California Press.

Hecky, R.E., and Degens, E.T., 1973, Late Pleistocene–Holocene chemical stratigraphy and paleolimnology of the Rift Valley lakes of Central Africa: Woods Hole Oceanographic Institution, Technical Report, v. 73–28, 1–93. de Heinzelin, J., and Verniers, J., 1996, Realm of the Upper Semliki (Eastern Zaı¨re). An Essay on Historical Geology: Muse´e Royal de l’Afrique Centrale– Tervuren, Annales, Sciences Géologiques, v. 102, 1–83.

Gawthorpe, R.L. and Leeder, M.R., 2000, Tectono- sedimentary evolution of active extensional basins: Basin Research, v. 12, 195-218. 35

Gilbert G.K., 1890, Lake Bonneville. U.S. Geological Survey Monograph, 1:1–275

Hamilton S.K., Sippel S.J., Calheiros D.F., Melack J.M., 1997, An anoxic event and other biogeochemical effects of the Pantanal wetland on the Paraguay River: Limnology and Oceanography, v. 42, 257–272.

Heckman C.W., 1998, The Pantanal of Poconé. Kluwer, Den Haag

Holbrook, J.M. and Schumm, S.A., 1999, Geomorphic and sedimentary response of rivers to tectonic deformation: a brief review and critique of a tool for recognizing subtle epeirogenic deformation in modern and ancient settings: Tectonophysics, v. 305, 287–306.

Holbrook, J. M., Autin, W.J., Rittenour, T.M., Marshak, S. and Goble, R.J. 2006, Stratigraphic evidence for millennial-scale temporal clustering of earthquakes on a continental-interior fault: Holocene Mississippi River floodplain deposits, New Madrid seismic zone, USA: Tectonophysics, v. 420, 431–454.

Horton, B.K. and DeCelles, P.G., 2001, Modern and ancient fluvial megafans in the central Andean foreland basin system, southern Bolivia: Basin Research, v. 13, 43–63.

Johnson, T.C., Halfman, J.D., Rosendahl, B.R., Lister, G.S., 1987, Climatic and tectonic effects on sedimentation in a rift-valley lake—evidence from high- resolution seismic profiles, Lake Turkana, Kenya: Geological Society of America Bulletin, v. 98, 439–447.

Johnson, T.C., Scholz, C.A., Talbot, M.R., Kelts, K., Rickettts, R.D., Ngobi, G., Beuning, K., Ssemmanda, I., McGill, J.W., 1996, Late Pleistocene desiccation of Lake Victoria and rapid evolution of Cichlid Fishes: Science, v. 276, 786-788.

Jordan, T. E., Schlunegger, F., and Cardozo, N., 2001, Unsteady and spatially variable evolution of the Neogene Andean Bermejo foreland basin, Argentina: Journal of South America Earth Sciences, v. 14, 775 – 798.

Junk, W.J., Bayley, P.B., Sparks, R.E., 1989, The flood pulse concept in river- floodplain systems: Canadian Special Publication of Fisheries and Aquatic Sciences, v. 106, 110-127. 36

Lambiase. J.J., 1990, A model for tectonic control of lacustrine stratigraphic sequences in continental rift basins. In Lacustrine Basin Exploration: Case Studies and Modern Analogs, ed. B Katz, pp. 265–76. Am. Assoc. Pet. Geol. Mem. 50. 340 pp.

Leier, A. L., DeCelles, P. G., and Pelletier, J. D., 2005, Mountains, monsoons, and megafans: Geology, v. 33, 289–292.

Lezzar, K.E., Tiercelin, J.-J., De Batist, M., Cohen, A.S., Bandora, T., Van Rensbergen, P., Le Turdu, C., Wafula, M., and Klerkx, J., 1996, New seismic stratigraphy and Late Tertiary history of the North Tanganyika Basin, East African Rift system, deduced from ultifold reflection and high resolution seismic data and piston core evidence: Basin Research, v. 8, 1–28.

Livingstone, D.A., 1962. Age of deglaciation in the Ruwenzori range, Uganda: Nature v. 194, 859-860.

Lyell, C., 1830, Principles of Geology, V. 1. J. Murray, London.

McGlue MM, Scholz CA, Karp T, Ongodia B, Lezzar K-E., 2006, Facies architecture of flexural margin lowstand delta deposits in Lake Edward, East African Rift: Constraints from seismic reflection imaging: Journal of Sedimentary Research, v. 76, 942–958.

McGlue, M.M., and Cohen, A.S., 2006, The Evolution Of Topographic Closure And Foreland Basin Lakes: A Modern Time-Slice Perspective From The Andes: Geological Society of America Annual Meeting Abstracts with Program

McGregor, G.R., and Nieuwolt, S., 1998, Tropical Climatology. John Wiley and Sons, Chichester, 339 pp.

Nyblade, A.A., and Brazier, R.A., 2002, Precambrian lithospheric controls on the development of the East African rift system: Geology, v. 30, 755–758. de Oliveira Bezerra, M.A. and Mozeto, A.A., 2008, Deposição de carbono orgânico na planície de inundação do Rio Paraguai durante o Holoceno médio: Oecolgia Brasiliera, v. 12(1), 155-171.

Olsen, P.E., 1986, A 40-million-year lake record of Early Mesozoic orbital climatic forcing: Science, v. 234, 842-848.

O’Reilly, C.M., Alin, S.R., Plisnier, P.D., Cohen, A.S., and Mckee, B.A., 2003, Climate change decreases aquatic ecosystem productivity of Lake Tanganyika, Africa: Nature, v. 424, 766–768.

Pelletier, J.D., 2007, Erosion rate determination from foreland basin geometry: Geology, v. 35:1, 5-8.

Pietras, J.T. and Carroll, A.R., 2006, High-resolution stratigraphy of an underfilled lake basin: Wilkins Peak Member, Eocene Green River Formation, Wyoming, USA: Journal of Sedimentary Research, v. 76, 1197- 1214.

Piovano, E.L., Ariztegui, D. and Damatto Moreira, S., 2002, Recent environmental changes in Laguna Mar Chiquita (central Argentina): a sedimentary model for a highly variable saline lake: Sedimentology, v. 49, 1371-1384.

Placzek, C., Quade, J. and Patchett, J., 2006, Geochronology and stratigraphy of late Pleistocene lake cycles on the southern Bolivian Altiplano: Implications for causes of tropical climate change: Geological Society of America Bulletin, v. 118, 515-532.

Por, F.D., 1995, The Pantanal of Mato Grosso (Brazil). Kluwer, Dordrecht, Germany

Quade, J., Rech, J.A., Betancourt, J.L., Latorre, C., Quade, B., Rylander K.A. and Fisher, T., 2008, Paleowetlands and regional climate change in the central Atacama Desert, northern Chile: Quaternary Research, v. 69, 343-360.

RAMSAR (2009) The Annotated RAMSAR List.

Rittenour, T.M., Blum, M.D. and Goble, R.J. 2007, Fluvial evolution of the lower Mississippi River valley during the last 100-kyr glacial cycle: Response to glaciation and sea-level change: Geological Society of America Bulletin, v.119, 586–608.

Rosendahl, B.R., 1987, Architecture of continental rifts with special reference to East Africa: Annual Review of Earth and Planetary Sciences, v. 15, 445–503.

Scholz, C.A., and Rosendahl, B.R., 1988, Low lake stands in Lakes Malawi and Tanganyika, East Africa, delineated with multifold seismic data: Science, v. 240, 1645–1648.

Scholz, C.A., King, J.W., Ellis, G.S., Swart, P.K., Stager, J.C., and Colman, S.M., 2003, Paleolimnology of Lake Tanganyika, East Africa, over the past 100 kyr: Journal of Paleolimnology, v. 30, 139–150.

Scholz, C.A., 1995, Deltas of the Lake Malawi Rift, East Africa: Seismic expression and exploration implications: American Association of Petroleum Geologists Bulletin, v. 79, 1679–1697.

Scholz, C.A., Johnson, T.C., Cohen, A.S., King, J.W., Peck, J., Overpeck, J.T., Talbot, M.R., Brown, E.T., Kalindekafe, L., Amoako, P.Y.O, Lyons, R.P, Shanahan, T.M., Castaneda, I.S., Heil, C.W., Forman, S.L., McHargue, L.R., Beuning, K.R., Gomez, J., and Pierson, J., 2007, East African megadroughts between 135-75 kyr ago and bearing on early-modern human origins: Proceedings of the National Academy of Sciences, v. 104, 16416-16421.

Shindell, D. T., Walter, B.P., and Faluvegi, G., 2004, Impacts of climate change on methane emissions from wetlands: Geophysical Research Letters, v. 31, L21202, doi:10.1029/2004GL021009

Sinha, R., and Friend, P.F., 1994, River systems and their sediment flux, Indo- Gangetic plains, northern Bihar, India: Sedimentology, v. 41, 825–845.

Sladen, C.P., 1994, Key elements during the search for hydrocarbons in lake systems, in Gierlowski-Kordesch, E., and Kelts, K., Global geological record of lake basins, vol. 1: Cambridge, Cambridge University Press, 3–17.

Smith M.E., Carroll A.R., Singer B.S., 2008, Synoptic reconstruction of a major ancient lake system: Eocene Green River Formation, western United States: Geological Society of America Bulletin, v. 120, 54–84.

Smol, J.P.. 2002 Pollution of lakes and rivers: a paleoenvironmental perspective, 2nd edn. Oxford, UK: Blackwell Publishing.

Soares A.P., Soares P.C. and Assine M.L., 2004, Areiais e lagoas do Pantanal, Brasil: herança paleoclimática?: Revista Brasiliara Geologia, v. 33(2), 211-224.

Soreghan, M.J., and Cohen, A.S., 1996, Textural and compositional variability across littoral segments of Lake Tanganyika: The effect of asymmetric basin structure on sedimentation in large rift lakes: American Association of Petroleum Geologists Bulletin, v. 80, 382–409.

Tierney, J.E., Mayes, M.T., Meyer, N., Johnson, C., Swarzenski, P.W., Cohen, A.S. and Russell, J.M., 2010, Late twentieth-century warming in Lake Tanganyika unprecedented since AD 500. Nature Geoscience, v. 3, 422-425.

Tierney, J.E., Russell, J.M., Huang, Y., Sinninghe Damsté, J.S., Hopmans, E.C. and Cohen, A.S., 2008, Northern Hemisphere controls on tropical southeast African climate during the past 60,000 years: Science, v. 322, 252-255.

Tierney J.E. and Russell J.M., 2007, Abrupt Climate change in Southeast Tropical Africa influenced by Indian monsoon variability and ITCZ migration: Geophysical Research Letters, v. 34, 1–6.

Ussami N., Shiraiwa S., Dominguez J.M.L., 1999, Basement reactivation in a sub- Andean foreland flexural bulge: the Pantanal wetland, SW Brazil: Tectonics v. 18(1), 25–39.

Van Damme, D., and Pickford, M., 1999, The late Cenozoic Viviparidae (Mollusca, Gastropoda) of the Albertine Rift Valley (Uganda-Congo): Hydrobiologia, v. 390, 171–217.

Williamson, P.G., 1981, Palaeontological documentation of speciation in Cenozoic molluscs from Turkana Basin: Nature, v. 293, p. 437–443.

Yin, J.H., Battisti, D.S., 2000, The importance of tropical sea surface temperature patterns in simulations of Last Glacial Maximum climate: Journal of Climate, v. 14, 565-581.

Zaleha, M. J., 2006, Sevier orogenesis and nonmarine basin filling: Implications of new stratigraphic correlations of Lower Cretaceous strata throughout Wyoming, USA. Geological Society of America Bulletin, v. 118, 886-896.

APPENDIX A:

SEISMIC RECORDS OF LATE PLEISTOCENE ARIDITY IN LAKE TANGANYIKA, TROPICAL EAST AFRICA

Michael M. McGlue1, Kiram E. Lezzar1, Andrew S. Cohen1, James M. Russell2, Jean-Jacques Tiercelin3, Anna A. Felton1, Evelyne Mbede4, Hudson H. Nkotagu4

1Dept. of Geosciences, University of Arizona, Tucson, AZ 2Dept. of Geological Sciences, Brown University, Providence, RI 3UMR 6118 Géosciences-Rennes CNRS, Université de Rennes1, Rennes France; 4Dept. of Geology, University of Dar es Salaam, Dar es Salaam, Tanzania

Permission to reprint from copyright holder This work is reprinted here with permission from the Journal of Paleolimnology. Correspondence with Nel van der Werf, Springer Publications Specialist, follows:

Dear Sir, With reference to your request (copy herewith) to reprint material on which Springer Science and Business Media controls the copyright, our permission is granted, free of charge, for the use indicated in your enquiry.

This permission:  allows you non-exclusive reproduction rights throughout the World.  permission includes use in an electronic form, provided content is password protected and at internet  excludes use in any other electronic form. Should you have a specific project in mind, please reapply for permission.  requires a full credit (Springer/Kluwer Academic Publishers book/journal title, volume, year of publication, page, chapter/article title, name(s) of author(s), figure number(s), original copyright notice) to the publication in which the material was originally published, by adding: with kind permission of Springer Science and Business Media.

The material can only be used for the purpose of defending your dissertation, and with a maximum of 100 extra copies in paper. Permission free of charge on this occasion does not prejudice any rights we might have to charge for reproduction of our copyrighted material in the future.

Best regards, Nel van der Werf ( [email protected])

Abstract

New intermediate resolution, normal incidence seismic reflection profiles from

Lake Tanganyika’s central basin capture dramatic evidence of base-level change during two intervals of the late Pleistocene. Four seismically defined stratigraphic sequences (A-D) tied to radiocarbon-dated sediment cores provide a chronology for fluctuating environmental conditions along the Kalya Platform.

Stacked, oblique clinoforms in Sequence C are interpreted as prograding siliciclastic deltas deposited during a major regression that shifted the paleo-lake shore ~ 21 km towards the west prior to ~ 106 ka. The topset-to-foreset transitions in these deltas suggest lake level was reduced by ~ 435 m during the period of deposition. Mounded reflections in the overlying sequence are interpreted as the backstepping remnants of the delta system, deposited during the termination of the lowstand and the onset of transgressive conditions in the basin. The youngest depositional sequence reflects the onset of profundal sedimentation during lake level highstand. High amplitude reflections and deeply incised channels suggest a short-lived desiccation event that reduced lake level by ~ 260 m, interpreted as a product of Last Glacial Maximum (32 - 14 ka) aridity. Paleobathymetric maps constructed for the two interpreted regressions reveal that despite the positive lake-floor topography created by the Kavala Island Ridge Accommodation Zone, 43

Lake Tanganyika remained a large, mostly connected water body throughout the late Pleistocene. The results of this analysis further imply that Lake Tanganyika is the most drought resistant water body in the East African tropics, and may have acted as a refuge for local and migrating fauna during periods of prolonged aridity.

Introduction

Over the past several decades, paleolake level data from the Great Lakes of

East Africa have been used to infer variability in the tropical climate system, serving as indicators for trends in effective precipitation (e.g. Street and Grove

1979; Kutzbach and Street-Perrott 1985; Johnson et al. 1996; Farrera et al. 1999).

Large tectonic lake basins are a particularly important resource for understanding the response of the tropics to climatic flux, principally because of the unmatched spatial and temporal resolution of their sedimentary deposits.

Cenozoic extension of the East African crust spurred the fortuitous evolution of deeply-subsided lake basins fed by watersheds that, in some instances, span several degrees of latitude. These conditions are ideal for developing long (104 –

106 yr) sedimentary records of both local and regional climate change for the

African tropics. Moreover, profundal depositional environments found in the deepest of the rift lakes are rarely impacted by erosion (e.g. Johnson et al. 2003), 44

thus avoiding the clipping of paleoenvironmental records during protracted periods of aridity. The key relationship that makes paleoenvironmental studies using data from East African rift lakes feasible is the interrelationship between effective precipitation, lake surface elevation, and basin sill height. Lake hydrology exerts a fundamental control on patterns of sediment accumulation and facies architecture in large lakes (Carroll and Bohacs 1999). Over relatively short intervals of geologic time (e.g. orbital frequencies; < 105 yrs), climate change appears to be a more persistent forcing mechanism on rift-lake basin hydrology than basin geodynamics or volcanism.

Recent breakthroughs in lake drilling notwithstanding (Koeberl et al.

2005; Scholz et al. 2006), most studies of African paleolake levels in extant lakes have been based on inferences made from relatively short sediment cores.

Although a great wealth of knowledge has been derived from sediment core- based studies, certain limitations arise when evaluating base level dynamics exclusively from core data, especially on record duration imposed by the length of the coring device, and the indirect nature of lake level inferences derived from microfossils or geochemical data. In the absence of other corroborating data, establishing the magnitude of lake level change in large lakes using indirect records from sediment cores remains a challenge. The response of Lake 45

Tanganyika to arid climatic conditions prevalent during the Last Glacial

Maximum (LGM; ~ 32 – 14 ka) provides a pertinent example. Despite numerous investigations, the magnitude of lake level fall during the LGM has remained equivocal for decades. No fewer than thirteen sediment-core based analyses have been published on the subject, with interpreted estimates of lake level decline ranging from 150 – 600 m (e.g. Livingstone 1965; Hecky and Degens 1973;

Haberyan and Hecky 1987; Tiercelin et al. 1988; Gasse et al. 1989; Tiercelin and

Mondeguer 1991; Williamson et al. 1991; Vincens et al. 1993; Chalie 1995; Lezzar et al. 1996; Bergonzini et al. 1997; Cohen et al. 1997; Scholz et al. 2003).

Quantifying the magnitude of regression events is important not only for quantitative paleoclimatology (e.g. Hastenrath and Kutzbach 1983), but also when considering the impact of bathymetric variability on ecological interactions and the tempo of speciation (e.g. Sturmbauer and Meyer 1992; Johnson et al.

1996; Verheyen et al. 2003).

One approach that can be adopted to aid lake level reconstructions is to combine the results of sediment core analyses with independent geomorphic evidence of paleo-shorelines. For example, hanging strandlines and wave cut terraces can be used to assess the extent of highstand conditions in a lake basin’s history, when preserved materials permit age-dating. This technique has been 46

widely employed to reconstruct lake level histories for sites around the world (e.g.

Talbot and Delibrias 1980; Thompson 1992; Plackzek et al. 2005). In the absence of subaerial evidence, sub-lacustrine geomorphologic features imaged with high frequency, marine-type seismic sources can also substantially improve lake level reconstruction efforts (e.g. Scholz 2001; D’Agostino et al. 2002; Anselmetti et al.

2006). Synthesizing seismically-imaged evidence of paleo-shoreline position with sediment core analyses provides one of the most robust methods available to paleolimnologists for characterizing lake level history.

The goals of this study are to provide new insights on late Quaternary lake level change in Lake Tanganyika, central East Africa, using a grid of seismic profiles correlated with dated sediment cores. Several prior studies on

Tanganyikan lake levels have used sediment core data in concert with observations from seismic profiles (Tiercelin et al. 1988; Mondeguer 1991; Scholz et al. 2003). Where seismic data has been extensively used (e.g. Lezzar et al. 1996;

Cohen et al. 1997), low line density, poor resolution and wide error bars associated with the reflection seismic radiocarbon method (RSRM) have permitted only minimum age estimates for Quaternary lake level fluctuations. As a result, the full potential of seismic profiling for lake level reconstruction has not yet been realized on Lake Tanganyika. In this study, we collected a tight grid of 47

intermediate-resolution seismic data in a region favorable for assessing the impact of climate on the stratigraphic record. We focused on developing a detailed seismic stratigraphic framework that is used to frame new arguments about the bathymetric impacts of two major regression events that affected Lake

Tanganyika during the late Pleistocene. We place chronological constraints on our seismic interpretation through correlation with a growing database of well- dated sediment cores that were collected along the seismic grid, and to the north of our study area (Cohen et al. 2004; Felton et al. 2007; Scholz et al. 2007).

Description of Study Area

Lake Tanganyika and the Western Rift Valley

The formation of Western Rift Valley of the East African Rift System

(EARS) and its extant lakes has been traced to the middle Miocene (Cohen et al.

1993; Nyblade and Brazier 2002). Extensional deformation of the upper crust followed deep crustal heterogeneities and helped form a series of elongate rift basins, marked by high angle, basin bounding border faults (Morely 1989;

Versfelt and Rosendahl 1989). Lake Tanganyika occupies several opposite polarity rift basins, and it is centrally located within the Western Rift Valley. The lake spans the international borders of four different nations: Tanzania, Zambia,

Democratic Republic of the Congo and Burundi (Figure 1). 48

Lake Tanganyika is the second largest lake on Earth by volume, reaching a maximum depth of ~ 1470 m in its southern basin (Capart 1952; Rosendahl et al.

1988). Due to its position relative to the equator, the lake and its surrounding environments experience a moist tropical climate, receiving an average of ~ 1200 mm of precipitation annually (Coulter and Spigel 1991). Patterns of rainfall in the region are affected by monsoon winds and the yearly migration of the Inter- tropical Convergence Zone (Nicholson 1996). Vegetation surrounding the lake is primarily Zambezian (seasonally-arid) woodlands, with Afromontane belts at higher elevations, reflecting the regional climate and local orographic affects

(White 1983).

Lake Tanganyika is meromictic, with a seasonally well-mixed eplimnion and an anoxic hypolimnion below ~150 m (Hecky and Degens 1973). The lake is hydrologically open, draining to the west into the greater Congo basin through the Lukuga River (Figure 1). The basin’s sill depth is relatively shallow (< 15 m), allowing the lake to become hydrologically closed with only minor drops in effective precipitation. At present, enriched surface water oxygen isotope values suggest that evaporation is the dominant process governing water loss from the lake (Craig et al. 1974; Dettman et al. 2005). Surface water chemistry and carbonate precipitation today is strongly influenced by exposures of Cenozoic 49

volcanic rocks and hydrothermal inputs from Lake Kivu to the north (Haberyan and Hecky 1987; Felton et al. 2007).

The Kalya Platform

Our study was focused on the Kalya Platform, a westward dipping flexural margin of the East Marungu Half Graben, located in the central part of Lake

Tanganyika, south of the Mahale Mountains (Figure 2). The Kalya Platform is structurally linked to the Kalya Horst. Movement on normal faults has created an asymmetric, along-strike platform morphology, characterized by a steeper gradient in the north, which gradually diminishes to the south. In the dip direction, the platform morphology is that of a homoclinal ramp (Figure 3).

Several minor normal faults dissect the platform, but large faults creating significant stratigraphic separation were not imaged. Subsidence on the platform is relatively minor, due to flexural uplift associated with subsidence on the basin- bounding border fault (e.g. Lezzar et al. 2002). Data collection efforts were focused at this site because of its potential for an intact stratal record of paleoshoreline position unobscured by steep canyons, deepwater bypass or pervasive mass wasting. In addition, the position of the seismic grid up-dip of the Kalya

Horst allows for correlation with high-resolution sediment core records of late 50

Pleistocene and Holocene climate change (Felton et al. 2007; Tierney and Russell

2007).

Methods

Acquisition and processing of seismic reflection profiles

Seismic reflections profiles used in this study were collected on Lake

Tanganyika in 2004 in conjunction with the Nyanza Project research-training program. Kullenberg piston cores, described in detail elsewhere (Cohen et al.,

2004; Felton et al. 2007) were collected along the seismic grid during the same field season (Figure 2). Details of the field acquisition program, including source and receiver parameters, are summarized in Table 1. In all, more than 400 line km of digital, normal-incidence seismic data were available for the analysis. Data were collected in a grid pattern, with a mean line spacing of ~ 6 km (Figure 2).

The position of the seismic grid was oriented so that dip lines were orthogonal to the NW-SE axis of the half graben. Landmark Promax 2D ® seismic processing software was used to enhance the signal-to-noise ratio of the data (Table 1).

Stratigraphic and paleobathymetric interpretations

Following processing, data were transferred to Landmark Seisworks 2D® for interpretation. Our interpretation strategy focused on developing a detailed stratigraphic framework for the Kalya Platform using the principles of seismic 51

sequence stratigraphy, especially as applied to lacustrine strata (Bohacs et al.

2000; Scholz 2001). Depositional geometries and unconformities are well expressed on the Kalya Platform, whereas stratal relationships in deepwater equivalents are dominantly conformable. On our seismic profiles, sequence boundaries were mapped as surfaces displaying onlap or truncation of reflections; these surfaces may be bound by flooding surfaces characterized by downlap.

Interpretations from seismic reflection profiles were correlated to two sediment cores collected along our seismic grid: core KH3 and core KH4 (Figure 2).

Previously published age models for these cores, developed from AMS 14C dates corrected for Lake Tanganyika’s hard water effect, help place chronological constraints on the youngest sequence interpreted in the basin (Cohen et al. 2004;

Felton et 2007). Paleo-bathymetric reconstructions were facilitated by compiling and digitizing modern water depth information (Capart 1952; Rosendahl et al.

1988) into a geographic information system. Volumetric calculations followed the geometric method discussed by Wetzel (2001).

Results

Seismic sequences

We identified four depositional sequences that can be mapped across the

Kalya Platform, labeled Sequence D, C, B and A (from oldest to youngest, see 52

Figures 4 and 5). The depositional geometries in the oldest sequence are difficult to map with a high degree of confidence due to a decreased signal-to-noise ratio, caused in part by the inability of the high-frequency seismic source to penetrate great thicknesses of sediment.

Sequence D

Sequence D is the oldest stratigraphic sequence that can be interpreted in the dataset. This depositional package is best expressed in the center of the seismic grid, as acoustic penetration is lost as the section thickens approaching a major border fault to the southwest. In the center of the grid, Sequence D reaches a maximum thickness of ~ 90 ms two way travel time (TWTT).

Sequence D has a wedge-shaped external geometry and it is comprised of low to moderate amplitude, semi-continuous to discontinuous reflections (Figure 4).

Most of the internal reflections that make up Sequence D are tilted and dip towards the southwest. These reflections terminate against overlying reflections near the base of Sequence C (Figure 5). Along strike, reflections in Sequence D are locally chaotic or poorly defined.

Sequence C

Sequence C overlies Sequence D and displays highly variable reflection geometries across the Kalya Platform. The basal contact for Sequence C is defined by a high angle stratigraphic discordance marked by reflection 53

terminations (Figure 5). The internal character of Sequence C varies with position on the platform. Down-dip, near the distal margin of the platform, a series of well-developed, westward prograding clinoforms mark the depositional sequence. Several discrete oblique clinoform bodies can be mapped, and they appear to stack within Sequence C (Figure 4). The external form of these deposits varies from north to south on the Kalya Platform. To the north, foreset and bottom-set beds are steeply dipping, whereas equivalent reflections to the south are more gently inclined. Although clinoform morphology varies along the platform, in all cases dipping foreset reflections terminate against topset reflections, forming a toplap unconformity (Figure 6). In addition, onlapping reflections are often encountered along the toes of distal foreset beds. Internally, clinoforms are made up of low amplitude, semi-continuous reflections. The deepest clinoform in Sequence C occurs near the southern margin of the seismic grid, ~ 602 ms TWTT (~ 435 m) below the modern lake surface (Figure 7). Other clinoforms in this sequence, located near the northern end of the platform, occur between 480 and 518 ms TWTT below the modern lake surface. Up-dip of the clinoforms, Sequence C exhibits several long, steeply dipping reflections that terminate against overlying Sequence B reflections in an angular unconformity near 370 ms TWTT (Figure 5). The region between the clinoforms and the 54

angular discordance is characterized by low amplitude, discontinuous seismic facies and by reflection-free seismic facies. In some instances, discontinuous reflections are concave up, producing local unconformities through the truncation of overlying reflections.

Sequence B

Sequence B can be traced throughout the study area, and it occupies the stratigraphic interval directly beneath Sequence A. The basal contact of

Sequence B is readily identified by a high-angle stratigraphic discordance, where underlying, dipping reflections terminate against a set of overlying, flat-lying reflections (Figure 5). Sequence B is commonly thicker than Sequence A, reaching an average thickness of ~ 35 ms TWTT. A strike line collected across the platform suggests a thickening of the sequence from north to south, as reflections diverge and locally onlap (Figure 4). Sequence B is dominated by high amplitude, semi-continuous, parallel seismic facies. Although this facies typifies the majority of the reflections in Sequence B, important variations do exist. In the center of the seismic grid, mound-shaped (concave down) reflections occur just above the basal contact, and appear to progressively back-step to a higher stratigraphic level within the sequence. Numerous reflection truncations characterize the internal structure of the mounds. In the strike direction, these 55

mounds thicken the section, and form discrete lens-shaped packages of reflections (Figure 8). These lenses are characterized by bi-directional downlap and internal reflection terminations. The local relief created by the mounds is minor, approaching 22 ms TWTT. These mounded reflections occur over an area of ~ 10 km2, and are presently found ~ 410 ms TWTT below the modern lake surface.

Sequence A

Sequence A is the uppermost depositional package in the study area. The upper contact of Sequence A is the modern sediment water interface. The basal contact of Sequence A is marked by onlapping reflections near the distal

(western) margin of the platform, where the lake floor morphology changes from a homoclinal ramp to a steep slope at the platform-to-deep basin transition

(Figure 6). Sequence A reflections progressively terminate onto inclined underlying Sequence B reflections, forming an unconformable contact (Figure 6).

In areas where onlapping reflections are not found, Sequence A reflections display a drape geometry. The thickness of Sequence A typically varies between

16 – 24 ms (~12 to 17 m) across the Kalya Platform. The maximum observed thickness (~ 52 ms TWTT) occurs near the eastern limit of the seismic survey grid, where a series of pronounced channels create additional accommodation 56

space near the lake bottom (Figure 5). Sequence A is dominated by low amplitude, continuous, parallel seismic facies. These reflections extend over great distances, routinely exceeding 15 km. Whereas low amplitude continuous seismic facies characterizes the majority of Sequence A, a significant departure occurs on seismic lines that extend to within a few kilometers of the eastern shoreline. In this more proximal setting, low amplitude reflections are interrupted by a high amplitude, high frequency, two-cycle, reflection event

(HARE) above 350 ms TWTT (Figure 9). These continuous, parallel reflections grade down-dip into several channels, and they do not appear in the section deeper than ~ 360 ms TWTT. The geometry of Sequence A reflections is likewise altered by the channels themselves. The channels create minor negative relief on the lake bottom and they incise into the underlying reflections of Sequences B and C (Figure 10). Reflections within the channels are truncated and variable.

Seismic facies in the channels ranges from low amplitude to high-amplitude continuous to chaotic.

Sediment cores from Sequence A

Two Kullenberg cores were retrieved from Sequence A. Core KH3 was collected in ~ 600 m of water on the western side of the Kalya Horst. The core

(7.75 m long) penetrated to ~ 6 m above the base of Sequence A. The core is 57

dominated by profundal lithofacies (organic-rich ooze and clay) and has a basal age of ~ 60 ka (Figure 11). A full account of the core’s geochronology and sedimentology has been presented elsewhere (Felton et al. 2007). Sedimentation rates over the time period encompassed by the core range between 0.085-0.224 mm/yr. Core KH4 was collected in ~ 330 m of water on the distal margin of the

Kalya Platform (Cohen et al. 2004). This core is 7.29 m long, is likewise dominated by profundal lithofacies, with a basal age of ~ 41 ka (Figure 11). The sedimentation rates for core KH4 vary between 0.175-0.231 mm/yr based on available radiocarbon data.

Discussion

Seismic evidence of lake level change

The sequence stratigraphic framework (Sequences D through A) we have developed for the Kalya platform provides a relative chronological context for interpreting the region’s paleo-environmental history. Depositional processes control reflection geometries and seismic facies characteristics, with stratal continuity, impedance contrast and bed spacing contributing to a lesser extent

(Mitchum et al. 1977). As such, the vertical variability observed on our seismic profiles implies significant changes in the mode of deposition over the late

Pleistocene along the Kalya Platform. 58

An accurate interpretation of the paleo-environment during Sequence D time is challenging given the limited resolution of this depositional package in our dataset. The wedge-shaped external form of Sequence D is common for hanging wall stratigraphic sequences in extensional basins, due to increased accommodation spaced created adjacent to major structures (Morley 1989;

Schlische and Olsen 1990). The most defining characteristic of Sequence D is its angular, truncated reflections and low-to-moderate amplitude seismic facies.

Lake level change is a routinely invoked mechanism for the origin of angular unconformities such as those that mark Sequence D (e.g. Scholz and Rosendahl

1988). The angular stratigraphic discordance in Sequence D, coupled with highly variable seismic facies characteristics, suggest this package may have been deposited during a lake level lowstand that exposed the entirety of the Kalya platform. However, because the tilting and truncation of Sequence D reflections may reflect the influence of tectonics, the paleo-environment during the deposition of Sequence D remains equivocal. Growth strata are generally under- reported in lake-basin studies, but Soreghan and others (1999) noted the importance of uplift and rotation on footwall blocks on the development of canyons and acoustically complex coarse-grained deposits in Lake Malawi. In addition, a growing body of research suggests that syn-deformational 59

unconformities may be pervasive in tectonically active basins (Riba 1976; Anadón et al. 1986; Hardy and McClay 1999). Gawthorpe and others (1997), using examples from the Suez Rift, demonstrated that the geometry of rift-basin sequences can be strongly impacted by the vertical movement on buried normal faults. Whereas our seismic system is not suited to the imaging of deep structures, previous research using multichannel seismic has led to the development of detailed fault maps for Lake Tanganyika (e.g. Rosendahl et al.

1988; Versfelt and Rosendahl 1989). Versfelt and Rosendahl (1989: Fig. 2) present a structural map that shows a large, down-to-the-southwest normal fault that occupies a position east of our seismic grid. Given the presence of this normal fault, we cannot rule out the possibility of a purely tectonic origin for the reflection configuration in Sequence D.

Sequence C reflects deposition during a period of significantly reduced lake level. The strongest evidence for lowstand conditions in this sequence are the platform-margin clinoforms coupled with a pronounced up-dip angular unconformity (Figure 5). During the interval when these clinoforms developed the lake’s shoreline would have been shifted by ~ 21 km to the west of its current location. Clinoform development in rift basins often correlates with a basinward shift of flexural margin river systems during subaerial exposure of the platform 60

(Scholz 1995). Consequently, we interpret the platform-margin clinoforms in

Sequence C as westward prograding, siliciclastic lowstand delta deposits. There is a rich body of literature from lacustrine settings that documents clinoform- shaped deltas, extending back to the work of Gilbert (1890) on paleo Lake

Bonneville. In rift-lake basins, seismic studies have identified flexural margin clinoforms similar to those we observe in our dataset in both Lakes Malawi and

Edward (Scholz 1995; McGlue et al. 2006).

Oblique clinoforms on the Kalya Platform are typified by shingled internal reflections and top truncation of foreset reflections (Figure 6). Inclined foreset reflections mark the construction of the delta front at the lakeshore. Toplap forms through active sediment bypass indicative of a high-energy depositional environment without vertical accommodation space. Such an environment would have existed in Lake Tanganyika if lake level was much reduced; exposure of the platform to subaerial processes would allow for appreciable erosion and sediment bypass down-dip. The topset-to-foreset transition of the clinoforms is a useful geomorphic indicator that marks the relative position of lake level during

Sequence C time (Burbank and Anderson 2001). Clinoforms on the Kalya

Platform are stacked (a common phenomena typically attributed to delta-lobe switching), such that the oldest deposits are found at the lowest stratal level. In 61

our dataset, the deepest clinoform occurs near the southern terminus of the platform (Figure 7). Although several minor normal faults impact the stratigraphic section, overall this region of the Kalya Platform displays a low gradient. As a consequence, lake level interpretations made from these deposits can be made with greater confidence than those to the north, which are being uplifted in the footwalls of several N-NNE trending platform faults. Using these criteria, we interpret lake level during Sequence C time to have been at least 436 m below the modern lake surface. The deepest clinoform appears to have been partially eroded by a channel that truncated the top of the feature, but nevertheless the offlap break is clear on our seismic profiles (Figure 7). East of the clinoforms, a large angular unconformity reflects exposure of the platform updip of the lowstand delta. Reflections between the updip unconformity and the downdip clinoforms are discontinuous and in some locales, concave up. We interpret this seismic response as distributary channels along a delta plain environment.

We interpret Sequence B as a terminal lowstand to early transgressive period in the basin following the major regression in Sequence C. Sequence B reflections are high amplitude and semi-continuous, suggesting a different depositional environment than that which prevailed during Sequence C time. We 62

interpret Sequence B reflections as sediments deposited in a variety of near-shore environments, ranging from exposed lake margin to the submerged littoral zone.

Perhaps the most striking evidence of the onset of transgressive conditions during Sequence B time comes from three mound-shaped reflection sets that back step to progressively higher stratigraphic levels within the sequence. These mounds contain numerous internal reflection terminations, and onlapping reflections from the west (Figure 8).

Numerous mechanisms exist for the formation of mound-shaped seismic reflections, ranging from contrasting depositional processes to differential compaction (e.g. Mitchum et al. 1977), but two explanations are most likely in this case. The position of the mounds directly up-dip of the Sequence C clinoforms suggests that these reflections may record retrogradational deltaic sedimentation. Transgressive deltaic deposits have been encountered in marine shelf sequences (e.g. Yoo and Park 2000; Porebski and Steel 2006) but they have not previously been reported from East Africa. The apparent lack of transgressive deltaic deposits in African rift lakes can be explained by a number of factors, including: 1) rapid rise of lake level following major regressions, trapping the fluvial system close to the hinterland; 2) high wave energy during periods of lake level rise, leading to poor preservation of the backstepping delta 63

or 3) spatial aliasing in wide seismic grids. An intriguing aspect of these deposits is their contracted area, which strongly contrasts the expansive deltaic clinoforms of Sequence C. One possible explanation for the limited spatial extent of these features is that they reflect a reorganization of onshore vegetation during the transgression. As the region emerged from arid conditions in Sequence C time, fluvial sediment yield and rates of overland flow may have declined substantially as dense lowland vegetation returned and stabilized the landscape (e.g. Langbein and Schumm 1958). Evidence from pollen records in Lake Malawi and dust records from the Gulf of Aden seem to support this interpretation, as removal of lowland vegetation has apparently accompanied regional aridity over several instances in the Pleistocene and early Pliocene (Vincens 1991; deMenocal 1995;

Cohen et al. 2007).

An equally plausible origin for the mounded reflections in Sequence B lies in shallow water carbonate production. Given the proper water depth and chemistry, vast stromatolite banks or ooid shoals can develop in lakes, as is the case in Lake Tanganyika today (Cohen and Thouin 1987; Soreghan and Cohen

1996). In addition, previous studies on other large lakes have seismically imaged bioherms with a characteristic mound-shaped external form (Colman et al. 2002).

Modern stromatolites in Lake Tanganyika are present down to ~ 40 m below the 64

lake surface, whereas ooid shoals in both modern Lake Tanganyika and paleolake examples form at littoral depths (Swirydczuk et al. 1979; Cohen and Thouin

1987). Regardless of a siliciclastic or carbonate origin, the presence of these deposits implies a shallow sub-aqueous paleoenvironment during Sequence B time, at least down-dip of the mounds.

We interpret Sequence A to reflect the onset of hemipelagic sedimentation in a profundal depositional environment. The low amplitude, continuous seismic facies that distinguishes Sequence A from underlying reflections is typical of deepwater deposits in Lake Tanganyika (Tiercelin et al. 1989; Lezzar et al. 1996).

The dominance of diatomaceous ooze and clay in the sediment cores collected from this sequence confirms that sedimentation was dominated by suspension fall-out of organic-rich material from the lake’s epilimnion. Laterally continuous reflections, coupled with a long mean reflection length (> 15 km), support the notion that widespread deepwater conditions prevailed during most of Sequence

A time. Given the asymmetric morphology of the basin, the Kalya Platform lagged the Kalya Horst environment in its return to a profundal environment, as evidenced by the progressive onlap of LAC reflections from the west onto the toe of the distal platform (Figure 6). 65

The only remarkable internal variability within Sequence A occurs near the eastern margin of the study area, where we observe a prominent HARE embedded within more typical low amplitude reflections ~ 24 ms TWTT below the lake floor (Figure 9). We interpret this HARE as evidence for a smaller fluctuation in lake level during Sequence A time than what is observed earlier in the record. High frequency base level fluctuations induced by climate change are common in tropical lakes and Lake Tanganyika is no exception. Numerous analyses of sedimentary indicator materials have suggested that changes in effective moisture have altered the lake’s surface elevation and forced hydrologic closure in the recent past (e.g. Haberyan and Hecky 1987; Alin and Cohen 2003).

The strong impedance contrast between the low amplitude, continuous facies and the HARE suggests a density change in the sediments that comprise them; desiccated lake sediments, hardened by subaerial exposure and evaporation of interstitial pore-fluids, is one possible means of creating this seismic response.

Because the HARE is only observed above 350 ms TWTT, we interpret that lake level was reduced by a maximum of ~ 262 m during this short-lived event.

Through correlation to previously published sediment core records, we suggest this event corresponds to the LGM, an interval of known tropical aridity caused by high latitude glaciation (Gasse 2001). Trace elements and sedimentary 66

organic matter analysis suggests that climatic conditions consistent with tropical aridity associated with the LGM persisted in Lake Tanganyika between about 32-

14 ka (Felton et al. 2007). Notably, the LGM interval spans ~ 0.8 m in the KH3 sediment core record. For that reason, resolving the LGM in seismic records in regions that remained sub-aqueous is probably not possible due to the tuning effect produced by a source frequency at or below ~ 1 kHz. Consequently, a seismic stratigraphic manifestation of this global climate event can only be found where the lake bottom was exposed, which we interpret to be in areas where the modern water depth is less than ~ 262 m. An interesting feature in our seismic dataset is the spatial relationship between the incised channels that transverse the platform and the HARE we interpret as evidence of the LGM. The law of cross cutting relationships demands that the channels we observe are young features, formed late in Sequence A time based on the displacement of reflections by channel incision near the modern water bottom (Figure 10). Channel formation can occur during lake level highstands through localized faulting, large scale tilting that increases stream power, or through the activity of turbidity currents

(e.g. Soreghan et al. 1999). Likewise, channels can incise during lake level lowstands due to exposure and down-cutting. A number of minor normal faults flank the channel margins in our data, suggesting a tectonic origin (Figure 9). 67

However, the stratigraphic separation on these faults is relatively minor, whereas the channels incise an average of ~ 40 ms into underlying strata. Therefore, we interpret the channels to be dominantly climatic in origin, likely associated with the LGM.

Geochronology of lowstand events

The transition to Sequence A

We can trace the timing of the onset of relative highstand conditions using sedimentation rates derived from our sediment cores. Felton and others (2007) report a basal age for core KH3 of ~ 60 ka. The age model developed by these authors suggests that the sedimentation rate over this time interval varies, with an average approaching 0.13 mm/yr. By extrapolating this sedimentation rate to the basal contact of Sequence A near the Kalya Horst and assuming a velocity of

1450 m/s in the upper most sediments, we estimate an age of ~ 106 ka for the onset of deposition. At the KH4 site, the age model presented in Cohen and others (2004) implies an average sedimentation rate approaching 0.20 mm/yr.

Extrapolation of this rate from the lake bottom the sequence boundary suggests profundal sedimentation commenced ~ 55 ka at this site. Due to the variable nature of sedimentation in large lakes, age determinations developed through extrapolation must be made with caution. In the case of core KH3, the 68

extrapolated age is bolstered by the similarity in seismic response both above and below the cored interval (Felton et al. 2007). Because depositional processes principally control seismic facies, we can be confident that similar sediments prevail beneath the cored interval of Sequence A at the KH3 site.

The chronology from core KH4 is somewhat more challenging to interpret, given the location of the coring site at the edge of the Kalya Platform.

The seismic stratigraphic section beneath the core appears to be incomplete, due to the complex topography created by an underlying, uplifted delta lobe (Figure

6). Moving up-dip from platform edge, the section thickens to 18 – 21 ms

(TWTT) along most of the seismic grid; in these locales, the range of time encompassed by Sequence A expands to ~89 – 103 ka on the platform. This suggests that the age-extrapolation made from the KH4 core site probably underestimates the timing of the stratigraphic transition from Sequence B to

Sequence A. Likewise, sub-seismic scale hiatuses in the KH4 record are suggested by the core’s lithostratigraphy and may further contribute to underestimating the onset of Sequence A sedimentation. However, given the bathymetric contrast between the KH3 (~ 600 m) and KH4 (~ 330 m) sites, we can be certain of a time lag between transgression in the deepwater horst and flexural platform environments. This inference is confirmed by onlapping reflections 69

moving up the platform margin from the west (Figures 5 and 6). As a result, we suggest that the major transgression associated with the deposition of Sequence A triggered sedimentation at the KH3 site by ~ 106 ka, and followed thereafter at the KH4 site.

Regional evidence from the Kavala Island Ridge (Scholz et al. 2003), Lake

Malawi (Scholz et al. 2007) and northern Lake Tanganyika (Lezzar et al. 1996) helps address the timing of lowstand sedimentation along the Kalya Platform.

Scholz et al. (2007) report a major transgression near the crest of the Kavala

Island Ridge at ~ 97 ka, based on estimates from optically stimulated luminescence dating. These authors suggest that the lowstand preceding this transgression (- 393 m below modern) was an orbitally-forced mega-drought interval that severely impacted the African tropics. In Lake Malawi, a series of mega-droughts occurred between 135 – 70 ka (Scholz et al. 2007). We suggest that maximum lowstand conditions on the Kalya Platform occurred during this same interval. While we cannot directly determine the amount of time encompassed by Sequences B and C, the KH3 chronology suggests that lowstand deltaic deposition on the Kalya Platform occurred prior to 106 ka. This chronology is broadly consistent with the observations from both the Kavala

Island Ridge and Lake Malawi. Correlation with prior seismic results from the 70

Bujumbura and Mpulungu Basins suggests that the ‚a‛ discontinuity (Lezzar et al.

1996; Cohen et al. 1997) and the ‚A‛ event (e.g. Tiercelin et al. 1989) in northern and southern Tanganyika, respectively are coeval with the base of Sequence A.

Using the more precise KH3 chronology, we suggest that the time content of

Sequence A is appreciably greater than the ~ 35.3 – 39.7 ka inferred from RSRM calculations of Lezzar and colleagues (1996). The reasoning behind this mismatch likely lies in assumptions inherent to the RSRM method regarding the relationship between sedimentation rates and acoustic facies characteristics.

The deltaic clinoforms in Sequence C allow for the first quantitative estimate of paleo-lake shore position for Lake Tanganyika. Prior research has suggested a sub-aerial exposure surface indicative of a regression event that lowered lake level by more than 600 m (Scholz and Rosendahl 1988; Cohen et al.

1997). However, the non-unique origin of angular unconformities in extensional basins, used to infer precise low lake stands in these earlier studies, and present in

Sequence D, precludes a definitive identification of lake levels from Sequence D.

Evidence of the LGM in Sequence A

The HARE occurs, on average, 24 ms TWTT below the modern lake bottom near the eastern shore (Figure 9). Assuming a velocity of 1450 m/s in the uppermost sediments of Sequence A, we estimate the accumulation of ~ 17.4 m of 71

sediment since the termination of the LGM. Felton and others (2007) suggest that glacial aridity ceased in the region around 14 ka. Therefore, we estimate that sedimentation rates close to shore since the termination of the LGM have been appreciably higher than those near the distal platform margin. Assuming the cessation of LGM aridity occurred at 14 ka, we derive a sedimentation rate for proximal, nearshore component of Sequence A above the HARE of 0.12 cm/yr.

This estimate agrees well with littoral/deltaic sedimentation rate measurements presented by McKee and others (2005), and provides a tentative explanation for the increased thickness of Sequence A near along the eastern margin of the seismic grid.

The magnitude of lake level lowering in Tanganyika during the LGM has been a subject of considerable debate for several decades. Seismic evidence from the present study bears a strong resemblance to high amplitude reflections discovered in echosounder records from the Bujumbura basin (Bouroullec et al.

1991; Lezzar et al. 1996). Sediment cores that penetrated the high amplitude ‚C2‛ reflections, dated at ~ 23 ka, were composed of fine-grained sand to clayey-sand with mud balls (Lezzar et al. 1996). Cohen and others (1997) suggested that the lowstand implied by these features reduced lake level by ~ 160 m, desiccating the

Bay of Burton. Our new estimates confirm this paleogeographic scenario, but 72

expand the magnitude of the regression by nearly 40%. Other studies have attempted to quantify water levels during the LGM using diatoms (see Gasse

2001; Scholz et al. 2003), as direct sedimentological evidence of desiccation has not been recovered. Our estimate for the magnitude of the LGM regression falls closest to the range reported in Gasse (< 300 m; 2001) developed from planktonic/epiphytic diatoms.

Paleobathymetric reconstruction and implications

Based on our estimated lake levels for the key sequences discussed here we have constructed a series of paleobathymetric maps for Lake Tanganyika (Figure

12). A geometric analysis of available bathymetric data reveals that the modern volume of Lake Tanganyika is ~19,690 km3, about 3.0% larger than values regularly encountered in the literature. During the LGM (late Sequence A), the volume of Lake Tanganyika was reduced to ~ 12,800 km3. During the African mega-drought interval (Sequence C), the volume of Lake Tanganyika was reduced to ~ 11,240 km3, ~ 43% less than the modern.

Our seismic evidence suggests that the lake did not split into smaller, disconnected pools during the major Early Late Pleistocene regression event of

Sequence C. Close examination of multichannel seismic data collected in the

1980’s reveals that the most likely locus for a significant separation occurs at the 73

Kavala Island Ridge, in the center of the lake (Figure 1). The crest of the Kavala

Island Ridge creates a pronounced bathymetric high ~ 360 m below the lake surface (Scholz et al. 2003). Evidence from top-truncated seismic reflections and sediment bulk density profiles supports the interpretation that the crest of the

KIR is exposed during major regression events (Scholz et al. 2003; Scholz et al.

2007). However, the KIR is an asymmetric feature and the region west of the structural crest lies in water exceeding ~ 580 m. Paleobathymetric reconstructions created from this study show the lake would have remained connected during both the LGM and the African mega-drought interval. Unlike many other rift lakes, most of Lake Tanganyika’s volume is contained below the

500 m isobath, perhaps due to the lake’s great antiquity and a longer time period of subsidence on half-graben bounding border faults. As a result, Lake

Tanganyika is the most drought-resistant water body in the Western Rift Valley, and likely the only standing water body in East Africa that would have persisted as a very deep lake during the mega-drought interval. Consequently, the lake may have been an important refuge for drought-intolerant fauna during intense periods of Quaternary aridity, and served as an important watering hole for local and migrating fauna throughout these climatic crises.

Conclusions

The results of this integrated assessment of the Kalya Platform’s stratigraphy yield new insights on the magnitude of two major regressions that affected Lake Tanganyika during the late Quaternary. Whereas both the LGM and African mega-drought regressions forced significant lake level decline, only the earlier event strongly impacted the Platform’s stratigraphic record. We suggest this reflects both the magnitude and the duration of Early Late

Pleistocene tropical aridity (~106 ka).

1. We interpret four stratigraphic sequences on the Kalya Platform. The origin

of the oldest sequence (Sequence D) remains equivocal due to a lack of

penetration and resolution in the deeper stratigraphic section.

2. Stacked, oblique clinoforms within Sequence C are interpreted as prograding

lowstand delta deposits. Using the topset-to-foreset transition of the deepest

delta lobe at the southern end of the Kalya Platform, we interpret that lake

level was reduced by at least 435 m during the deposition of this deposit.

Delta lobes at the northern end of the Kalya Platform within this sequence

underestimate the magnitude of lake level change due to footwall uplift on

active normal faults. 75

3. Sequence B records the end of lowstand conditions in the basin and the onset

of rising water levels following the Sequence C drought. Mounded reflections

within this sequence are interpreted as retrogradational backstepping deltaic

deposits, marking the littoral – supra-littoral boundary during the time

interval of deposition.

4. Sequence A records the return of the basin to highstand conditions, where

profundal depositional processes dominant the section. The onset of

profundal sedimentation, based on extrapolations from sediment cores,

commenced by ~106 ka in the deep basin and shortly thereafter on the

platform. A major departure from the low amplitude continuous seismic

facies that characterizes the majority of Sequence A occurs near the eastern

lake shore, where a high amplitude, two-cycle reflection marks the section

above 350 ms TWTT, up-dip of highly incised channels. We interpret this

facies change to mark the short-lived LGM climatic event. The stratigraphic

position of this high amplitude reflection implies that lake level was reduced

by ~ 260 m during this regression.

5. Paleo-bathymetric reconstructions for the Pan-African ‚mega-drought‛ period

and the LGM (32 – 14 ka) suggest that Lake Tanganyika remained a large,

integrated water body despite significant, basinwide lake level decline. Due to 76

its age and border fault configuration, it seems likely that Lake Tanganyika

was the only large water body marking the East African landscape during the

African megadrought interval. If correct, this interpretation has implications

for the survival of local and regional fauna during episodes of extreme

drought.

Acknowledgements

This study was an outgrowth of the Nyanza Project, a Research Experience for Undergraduates site program supported by NSF ATM 02239020. We are grateful to UMR CNRS/UBO 6538 "Domaines Océaniques", European Institute of

Marine Studies, Plouzané, France for the financial and logistical support provided for the seismic survey, and to Jacques Bégot for his technical field

assistance. We thank all the Nyanza 2004 student participants for their efforts in the field and in the lab, especially Christine Gans, Winston Wheeler, Louis

Helfrich and Marla Torrado. Roy Johnson, Trey Wagner, and Steve Sorensen provided lab access and the Landmark Corporation is thanked for the donation of interpretation software. Peter Cattaneo provided helpful advice with seismic processing issues. We thank Bob Lyons for helpful comments on an earlier version of the manuscript.

References

Alin SR, Cohen AS (2003) Lake-level history of Lake Tanganyika, East Africa, for the past 2500 years based on ostracode-inferred water-depth reconstruction. Palaeogeogr Palaeoclimatol Palaeoecol 199:31-49

Anadon P, Cabrera L, Colombo F, Marzo M, Riba O (1986) Syntectonic intraformational unconformities in alluvial fan deposits, eastern Ebro Basin margins (SE Spain). In Allen PA and Homewood P (eds) Foreland basins: International Association of Sedimentologists Special Publication 8, pp 259–271

Anselmetti FS, Ariztegui D, Hodell DA, Hillesheim M, Brenner M, Gilli A, KcKenzie JA, Mueller AD (2006) Late Quaternary climate-induced lake level variations in Lake Petén-Itzá, Guatemala, inferred from seismic stratigraphic analysis. Palaeogeogr Palaeoclimatol Palaeoecol 230:52-69

Bergonzini L, Chalie F, Gasse F (1997) Paleoevaporation and paleoprecipitation in the Tanganyika Basin at 18,000 yr B.P. inferred from hydrologic and vegetation proxies. Quat Res 47:295-305

Beuning KRM, Talbot MR, Kelts K (1997) A revised 30,000-year paleoclimatic and paleohydrologic histroy of Lake Albert, East Africa. Palaeogeogr Palaeoclimatol Palaeoecol 136: 259-279

Bohacs KM, Carroll AR, Neal J, Mankiewicz P (2000) Lake Basin Type, Source Potential, and Hydrocarbon Character: An Integrated Sequence-stratigraphic- geochemical Framework. In Gierlowski-Kordesch E and Kelts K (eds) Lake Basins through Space and Time: AAPG Studies in Geology 46, pp 3-34

Bouroullec JL, Rehault JP, Rolet J, Tiercelin JJ, Mondequer A (1991) Quaternary sedimentary processes and dynamics in the northern part of the Lake Tanganyika trough, East Africa Rift System. Evidence of lacustrine eustatism?. Bulletin Centres de Recherche et Exploration. Elf-Aquitaine 15:343–368 78

Burbank DW, Anderson RS (2001) Tectonic Geomorphology: A Frontier in Earth Science Blackwell Publishing

Capart A (1952) Le milieu géographique et géophysique, Résultats scientifiques de l’exploration hydrobiologique du Lac Tanganika (1946-1947). Institut Royal des Sciences Naturelles de Belgique 1:3–27

Carroll AR, Bohacs KM (1999) Stratigraphic classification of ancient lakes: Balancing tectonic and climatic controls. Geology 27:99 -102

Chalié F (1995) Paléoclimats du bassin Tanganyika Sud au cours des 25 derniers mille ans: reconstitution quantitative par le traitement statistique des données polliniques. CR Acad Sci Paris 320:205–210

Cohen AS, Thouin C (1987) Near shore carbonate deposits in Lake Tanganyika. Geology 15:414–418

Cohen AS, Lezzar KE, Russell J, Scholz CA, Tiercelin JJ, Gans C, Helfrich CL 2004 The Lake Tanganyika Accomodation Zone Structural Highs: Probable Archive of Continuous Miocene to Recent Paleoenvironmental and Paleoclimatic Information for East Africa. AGU Ann Mtg Abstr w/ Prog

Cohen AS, Lezzar KE, Tiercelin JJ, Soreghan M (1997) New paleogeographic and lake level reconstructions of Lake Tanganyika: implications for tectonic, climatic and biological evolution in a rift lake. Basin Res 9:107–132

Cohen AS, Soreghan MJ, Scholz CA (1993) Estimating the age formation of lakes: An example from Lake Tanganyika, East African Rift system. Geology 21:511-514

Cohen AS, Stone JR, Beuning KRM, Park LE, Reinthal PN, Dettman D, Scholz CA, Johnson TC, King JW, Talbot MR, Brown ET, Ivory SJ (2007) Ecological Consequences of Early Late-Pleistocene Megadroughts in Tropical Africa. PROC NATL ACAD SCI U S A 79

Colman SM, Kelts K, Dinter D (2002) Depositional history and neotectonics in Great Salt Lake, Utah from high-resolution seismic stratigraphy. Sed Geol 148:61- 78

Coulter GW, Spigel RH (1991) Hydrodynamics. In: Coulter, GW (ed) Lake Tanganyika and its Life. Oxford University Press, London, pp 49-75

Craig H, Dixon F, Craig V, Edmond J, Coulter G (1974) Lake Tanganyika geochemical and hydrographic study: 1973 expedition. Scripps Institution of Oceanography, Publication 75:1-83

D'Agostino K, Seltzer GO, Baker PA, Fritz S, Dunbar R (2002) Late-Quaternary Lowstands of Lake Titicaca: Evidence from High-Resolution Seismic Data. Palaeogeogr Palaeoclimatol Palaeoecol 179:97-111 deMenocal PB (1995) Plio-Pleistocene African climate. Science 270: 53-59

Dettman D, Palacios-Fest MR, Nkotagu HH, Cohen AS (2005) Paleolimnological investigations of anthropogenic environmental change in Lake Tanganyika: VII. Carbonate isotope geochemistry as a record of riverine runoff. J Paleolimnol 34:93-105

Felton AA, Russell JM, Cohen AS, Baker ME, Chesley J, Lezzar K-E, McGlue MM, Pigati JS, Quade J, Stager JC (2007) Paleolimnological evidence for the onset and termination of glacial aridity from Lake Tanganyika, East Tropical Africa. Palaeogeogr Palaeoclimatol Palaeoecol doi:10.1016/j.palaeo.2007.04.003

Farrera I, Harrison SP, Prentice IC, Ramstein G, Guiot J, Bartlein PJ, Bonnefille R, Bush M, Cramer W, von Grafenstein U, Holmgren K, Hooghiemstra H, Hope G, Jolly D, Lauritzen S-E, Ono Y, Pinot S, Stute M, Yu G (1999) Tropical climates at the Last Glacial Maximum: a new synthesis of terrestrial paleoclimate data. I. vegetation, lake-levels and geochemistry. Clim Dyn 15: 823-856

Gasse F (2000) Hydrological changes in the African tropics since the Last Glacial Maximum. Quat Sci Rev 19:189-211

Gasse F, Ledee V, Massault M, Fontes J-C (1989) Water-level fluctuations of Lake Tanganyika in phase with oceanic changes during the last glaciation and deglaciation. Nature 342:57-59

Gawthorpe RL, Sharp IR, Underhill JR, Gupta S (1997) Linked sequence stratigraphic and structural evolution of propagating normal faults. Geology 25:795-798

Haberyan KA, Hecky, RE (1987) The late Pleistocene and Holocene stratigraphy and paleolimnology of Lakes Kivu and Tanganyika. Palaeogeogr Palaeoclimatol Palaeoecol 61:169-197

Hardy S, McClay KR (1999) Kinematic modeling of extensional fault-propagation folding. J Struct Geol 21:695–702

Hastenrath SL, Kutzback JE (1983) Paleoclimatic estimates for water and energy balance budgets of East African Lakes. Quat Res 19:141-153

Hecky RE, Degens ET (1973) Late Pleistocene-Holocene chemical stratigraphy and paleolimnology of the Rift Valley lakes of Central Africa. Technical Report, Woods Hole Oceanographic Institute, pp 1-144

Johnson TC, Brown ET, McManus J, Barry S, Barker P, Gasse F (2002) A high- resolution paleoclimate record spanning the past 25,000 years in Southern East Africa. Science 296:113-115

Johnson TC, Scholz CA, Talbot MR, Kelts K, Ricketts RD, Ngobi G, Buening K, Ssemmanda I, Mcgill JW (1996) Late Pleistocene desiccation of Lake Victoria and rapid evolution of cichlid fishes. Science 273:1091–1093

Koeberl C, Peck J, King J, Milkereit B, Overpeck J, Scholz CA (2005) The ICDP Lake Bosumtwi drilling project: A first report. Scientific Drilling 1:23 – 27

Kutzbach JE , Street-Perrott FA (1985) Milankovitch forcing of fuctuations in the level of tropical lakes from 18 to 0 kyr BP. Nature 317:130-134

Langbein WB, Schumm SA (1958) Yield of sediment in relation to mean annual precipitation. Trans Am Geophys Union 39:1076–1084

Lezzar KE, Tiercelin J-J, De Batist M, Cohen AS, Bandora T, Van Rensbergen P, Le Turdu C, Wafula M, Klerkx J (1996) New seismic stratigraphy and Late Tertiary history of the North Tanganyika Basin, East African Rift system, deduced from multifold reflection and high resolution seismic data and piston core evidence. Basin Res 8:1–28

Lezzar KE, Tiercelin J-J, Le Turdu C, Cohen AS, Reynolds DJ, Le Gall B, Scholz CA (2002) Control of normal fault interaction on the distribution of major Neogene sedimentary depocenters, Lake Tanganyika, East African rift. AAPG Bull 86:1027–1059

Livingstone DA 1965 Sedimentation and the history of water level change in lake Tanganyika. Limnol Oceanogr 10:607–610

McKee BA, Cohen AS, Dettman DL, Palacios-Fest MR, Alin SA, Ntungumburangye G (2005) Paleolimnological investigations of anthropogenic environmental change in Lake Tanganyika: II: Geochronologies and mass sedimentation rates based on 14C and 210Pb data. J Paleolimnol 34:19-29

McGlue MM, Scholz CA, Karp T, Ongodia B, Lezzar K-E (2006) Facies architecture of flexural margin lowstand delta deposits in Lake Edward, East African Rift: Constraints from seismic reflection imaging. J Sed Res 76:942 – 958

Mitchum RM, Vail PR, Sangree JB (1977) Seismic stratigraphy and global changes of sea level, part 6: stratigraphic interpretation of seismic reflection patterns in depositional sequences, in Payton, CE (ed) Seismic Stratigraphy- applications to hydrocarbon exploration. AAPG Memoir 26, pp. 117-133

Mondeguer A (1991) Thesis, The Univesity of Western Brittany

Morley CK (1989) Extension, Detachments and Sedimentation in Continental Rifts (With Particular Reference to East Africa). Tectonics 8:1175 – 1192

Nicholson SE (1996) A review of climate dynamics and climate variability in eastern Africa. In: Johnson, TC, Odada, EO (eds) The Limnology, Climatology, and Paleoclimatology of the East African Lakes. Gordon and Breach, Amsterdam, pp 25-56.

Nyblade AA, Brazier RA (2002) Precambrian lithospheric controls on the development of the East African rift system. Geology 30:755–758

Placzek C, Quade J Patchett PJ (2006) Geochronology and stratigraphy of plate Pleistocene lake cycles on the southern Bolivian Altiplano: implications for causes of tropical climate change. Geol Soc Am Bull 118:515-532

Porebski SJ, Steel RJ (2006) Deltas and Sea Level Change. J Sed Res 76: 390 – 403

Riba O (1976) Syntectonic unconformities of the Alto Cardener, Spanish Pyrenees: A genetic interpretation. Sed Geol 15:213–233

Rosendahl BR (1988), Seismic Atlas of Lake Tanganyika, East Africa: Project Probe Geophysical Atlas Series. Duke University, Durham

Schlische RW, Olsen PE (1990) Quantitative filling model for continental extensional basins with application to early Mesozoic rifts of Eastern North America. J Geol 98: 135–154 83

Scholz CA (1995) Deltas of the Lake Malawi Rift, East Africa: Seismic expression and exploration implications: AAPG Bull 79:1679–1697

Scholz CA (2001) Applications of Seismic Sequence Stratigraphy in Lacustrine Basins. In Last, WM and Smol, JP (eds) Tracking Environmental Changes in Lake Sediments: Physical and Chemical Techniques. Kluwer Academic Publishers, Dordrecht, The Netherlands

Scholz CA, Rosendahl BR (1988) Low lake stands in Lakes Malawi and Tanganyika, East Africa, delineated with multifold seismic data. Science 240:1645–1648

Scholz CA, Cohen AS, Johnson TC,King JW (2006) The 2005 Lake Malawi Scientific Drilling Project. Scientific Drilling 2:17-19

Scholz CA, Johnson TC, Cohen AS, King JW, Peck J, Overpeck JT, Talbot MR, Brown ET, Kalindekafe L, Amoako PYO, Lyons RP, Shanahan TM, Castenada IS, Heil CW, Forman SL, McHargue LR, Beuning KR, Gomez J, Pierson J (2007) East African megadroughts between 135-75 kyr ago and implications for early human history. PROC NATL ACAD SCI U S A

Scholz CA, King JW, Ellis GS, Swart PK, Stager C, Colman, SM (2003) Paleolimnology of Lake Tanganyika, East Africa, over the past 100 k yr. J Paleolimnol 30:139-150

Soreghan MJ, Scholz CA, Wells JT (1999) Coarse-grained deep-water sedimentation along a border fault margin of Lake Malawi, Africa: seismic stratigraphic analysis. J Sed Res 69:832–846

Street FA, Grove AT (1979) Global maps of lake level fluctuations since 30,000 BP. Quat Res 12:83 –118

Sturmbauer C, Meyer A (1992) Genetic divergence, speciation and morphological stasis in a lineage of African cichlid fishes. Nature 358:578 – 581

Swirydczuk K, Wilkinson BH, Smith GR (1980) The Pliocene Glenn’s Ferry Oolite- II: Sedmentology of Oolitic Lacustrine Terrace Deposits. J Sed Petrol 50:1237 – 1248

Talbot MR, Delibrias G (1980) A new late Pleistocene-Holocene water-level curve for Lake Bosumtwi, Ghana. Earth Planet Sci Lett 47:336-344

Thompson TA (1992) Beach ridge development and lake level variation in southern Lake Michigan. Sed Geol 80:305 – 318

Tiercelin, J-J, Mondeguer, A (1991) The geology of the Tanganyika trough. In Coulter, GW (ed) Lake Tanganyika and its life. Oxford University Press, London

Tiercelin J-J, Scholz CA, Mondeguer A, Rosendahl BR, Ravenne C (1989) Discontinuites sismiques et sedimentaires dans la serie de remplissage du fosse du Tanganyika, Rift Est Africain, [Seismic and sedimentary discontinuities in the Lake Tanganyika Rift, East Africa]. CR Acad Sci Paris 309:1599-1606

Tierney J, Russell, J M (2007) Abrupt Climate change in Southeast Tropical Africa influenced by Indian monsoon variability and ITCZ migration. Geophys Res Lett

Verheyen E, Salzburger W, Snoeks J, Meyer A (2003) Origin of the Superflock of Cichlid Fishes from Lake Victoria, East Africa. Science 300:325-329

Versfelt J, Rosendahl BR (1989) Relationship Between Pre-Rift Structure and Rift Architecture in Lakes Tanganyika and Malawi, East Africa. Nature 337:3-356

Vincens A (1991) Late Quaternary vegetation history of South-Tanganyika Basin. Climatic implications in South Central Africa. Palaeogeogr Palaeoclimatol Palaeoecol 86:207-226 85

Vincens A, Chalié F, Bonnefille R, Guiot J, Tiercelin J-J (1993) Pollen-Derived Rainfall and Temperature Estimates from Lake Tanganyika and Their Implication for Late Pleistocene Water Levels. Quat Res 40:343-350

Wetzel RG (2001) Limnology: Lake and River Ecosystems. Academic Press, San Diego

White F (1983) The Vegetation of Africa. A Descriptive Memoir to Accompany the UNESCO/AETFAT/UNSO Vegetation Map of Africa. UNESCO, Paris, 356 pp.

Williamson D, Thouveny N, Hillaire-Marcel C, Mondeguer A, Taieb M, Tiercelin J-J, Vincens, A (1991) Chronological potential of palaeomagnetic oscillations recorded in late quaternary sediments from Lake Tanganyika. Quat Sci Rev 10:351-361

Yoo DG, Park SC (2000) High-resolution seismic study as a tool for sequence stratigraphic evidence of high-frequency sea-level changes: latest Pleistocene– Holocene example from the Korea Strait. J Sed Res 70:296–307

Figures

Figure A.1: Overview map of the EARS. An arrow marks the location of Lake Tanganyika, northwest of Lake Malawi. Inset map is a digital elevation model showing the full extent of the lake. Note the location of the Lukuga River (RL), the sole hydrological outlet for Lake Tanganyika. The region inside the box is the focus area of this study, presented in Figure 2. The positions of the Kavala Island Ridge (KIR) and the Kalya Horst (KH) marked by ovals. MR = Malagarasi River. LR = Lake Rukwa

Figure A.2: Seismic trackline map for sparker profiles collected in 2004. Stars mark core locations discussed in the text. Major basement ridges (heavy lines with XXX’s) and faults appear as light grey lines; cherries mark the down- dropped hangingwall of the faults. Note the position of the seismic grid southwest of the Mahale Mountains (MM). Major rivers (Lufubu and Luginezi Rivers) enter the lake both orthogonally and obliquely. Bathymetric contour interval is ~ 100 m. See text for details

Figure A.3: A typical, southwest-northeast oriented seismic dip profile illustrating the gently sloping, ramp morphology of the Kalya Platform. Note that the vertical scale is in two way travel time (milliseconds). The basement-involved Kalya Horst marks the center of the line. An inset map (upper left corner) shows the relative position of the line on the seismic grid. The box defines the data shown in Figure 10 and a dashed line marks the position of the crossing strike line in Fig. 4

Figure A.4: Seismic dip profile (southwest-northeast) from the Kalya Platform (A) with stratigraphic interpretation (B). S-A = Sequence A. S-B = Sequence B. S-C = Sequence C. S-D = Sequence D. Line location marked on inset trackline map in upper panel; box marks data described in detail in Figure 7. Platform margin oblique clinoform and pronounced up-dip angular unconformity are well expressed within Sequence C, indicative of low lake level conditions during the African mega-drought interval

Figure A.5: Seismic strike profile (northwest-southeast) from the Kalya Platform (A) with stratigraphic interpretation (B). Line location marked on inset trackline map (upper panel). The wedge-shaped external form of Sequence D is evident, despite a loss of seismic signal deep in the section. Note the acoustically complex, stacked clinoforms within the overlying Sequence C. Stratigraphic sequences expand to the southwest increased accommodation space created by the Kalya Horst.

Figure A.6: Seismic detail (distal Kalya Platform) illustrating the clinoform morphology present in Sequence C and the location of sediment core KH4 (A) with interpretation (B). The facies contrast between the sequences interpreted in this study is pronounced in this locale. The stratigraphic section beneath the core site appears incomplete and Sequence A expands to the east. This reflection configuration implies age extrapolations made beneath the core site may underestimate the true age of the onset of profundal sedimentation following the African mega-drought interval. See text for details.

Figure A.7: Seismic detail (see Figure 3 for line location) of the buried clinoform used to define the magnitude of the lake level lowstand the occurred during Sequence C time. Note that several topset reflections are eroded by an overlying channel, but the key topset-foreset transition is preserved (marked by a star on the lower panel). Based on this reflection configuration, we suggest lake level was reduced by at least 436 m prior to 106 ka.

Figure A.8: Evidence of backstepping mounds in seismic sequence B. Seismic panels cross at the marked location, illustrating the cross-sectional geometry of a single mound. The backstepping mounds are spatially restricted, and thicken the section where they are present. Line drawings (right) illustrate the internal complexity of these deposits, which are interpreted as the transgressive remnants of the major lowstand delta system in Sequence C.

Figure A.9: Seismic detail (see Figure 4 for line location) of the high amplitude reflection event (HARE) observed along the proximal (nearshore) portion of the Kalya Platform. Note the strong seismic facies contrast between these reflections and the low amplitude continuous reflections typical of Sequence A. The line also illustrates the additional accommodation space created by a set of young channels that traverse the platform from northwest to southwest. The law of cross cutting relationships suggests the channels have been recently active, and may reflect increased incision during the LGM low lake stand.

Figure A.10: Typical sub-lacustrine channels encountered on the Kalya Platform. Channels are shallow and narrow near the northern margin of the study area, and tend to deepen and laterally amalgamate near the southern terminus of the seismic grid. Offset of the lake bottom reflection, dominance of low amplitude continuous fill and cross cutting relationships indicate a recent period of channel downcutting. See text for discussion.

Figure A.11: Lithostratigraphy and geochronology (AMS 14C; see Cohen et al 2004 and Felton et al 2007) for sediment cores KH3 (A) and KH4 (B). Profundal sediments (organic rich silt and clay) dominate in the cores, suggestion deposition primarily by suspension fall-out. Sedimentation rates from these cores were used to estimate the onset of Sequence A sedimentation, marked by the return to a deep lake environment following a major regression prior to 106 ka

Figure A.12: Bathymetric maps for Lake Tanganyika, tropical East Africa for the African mega-drought period, prior to 106 ka; the Last Glacial Maximum (32 – 14 ka) and modern. Note that the lake remains a large, mostly connected waterbody in spite of these Pleistocene regression events. Most of Lake Tanganyika’s volume is held beneath the 500 m isobath, suggesting it is a strongly drought resistant feature and therefore may have been an important refuge for drought- intolerant species during the pan-African mega-drought interval.

Tables

Table A.1: Seismic Acquisition and Processing Parameters, Kalya Region, Lake Tanganyika

Source SIG Sparker

Source Depth 1 m

Power 1000 J

Frequency 0 – 1500 Hz (producing vertical resolution < 1m)

Sampling Rate 3000 Hz

Shot Rate 2000 ms

Receiver SIG six-channel streamer (towed 60 m from source)

Acquisition DELPH digital acquisition system (1700 ms record length)

Navigation Garmin GPS II

Processing Bandpass filter, horizontal stack, top mute, AGC

APPENDIX B:

ENVIRONMENTAL CONTROLS ON SHELL-RICH FACIES IN TROPICAL LACUSTRINE RIFTS: A VIEW FROM LAKE TANGANYIKA’S LITTORAL

Michael M. McGlue1, Michael J. Soreghan2, Ellinor Michel3, Jonathan A. Todd3, Andrew S. Cohen1, John Mischler4, Christine S. O’Connell5, Oceana S. Castaneda6, Richard J. Hartwell7, Kiram E. Lezzar1 and Hudson H. Nkotagu8

1Dept. of Geosciences, The University of Arizona, Tucson AZ 2School of Geology and Geophysics, University of Oklahoma, Norman OK 3Dept. of Palaeontology, The Natural History Museum, London UK 4Dept. of Geosciences, Penn State University, University Park PA 5Earth Systems Program, Stanford University, Stanford CA 5Dept. of Earth Sciences, Dartmouth College, Hanover NH 6Earth Science Program, Fayetteville Manlius High School, Manlius NY 8Dept. of Geology, University of Dar es Salaam, Dar es Salaam, Tanzania

Permission to reprint from copyright holder This work is reprinted here with permission from Palaios. Correspondence with Jill Hardesy, SEPM Journals Editor, follows:

Dear Mr. McGlue, I am writing in reply to your e-mail that was sent to SEPM concerning copyright permission to reproduce the publication listed below. We understand that this article will be reproduced in your dissertation for the University of Arizona.

McGlue, M.M., et al., 2010, Environmental Controls on Shell-Rich Facies in Tropical Lacustrine Rifts: A View From Lake Tanganyika’s Littoral: PALAIOS, v. 25, 426–438.

This letter is to inform you that nonexclusive world rights are hereby granted for both print and electronic media on the condition that proper credit is provided. A standard credit line for ‚SEPM (Society for Sedimentary Geology)‛ is acceptable.

Sincerely yours,

Jill Hardesty Managing Editor, PALAIOS Editorial Office

100

Abstract

Lake Tanganyika, the world’s largest tropical rift lake, is unique among its counterparts in East Africa for the remarkable diversity of mollusk-rich sediments in its littoral zone. Molluscan shell beds are, however, a common feature of ancient lacustrine rift deposits and thus a better understanding of their spatial and temporal development is important. Targeted surveys across the littoral region of the Kigoma Basin reveal three surficial shell-rich facies that differ widely in depositional style and geometry. A unifying characteristic of these deposits is the volume of shells of Neothauma tanganyicense, a large, viviparous gastropod endemic to the lake. Reservoir-corrected radiocarbon dating indicates that Neothauma deposits in these surficial sediments are time averaged over at least the last ~1600 calendar years BP. Preservation of fossil Neothauma shells in the littoral zone depends on both environmental conditions and on post- mortem shell modifications. Interaction between shells and mobile siliciclastic grains, facilitated by wave action and storms, represents a particularly destructive taphonomic process in the study area. Rank scoring of damage to Neothauma suggests that stromatolitic encrustations or early calcite coatings may help mitigate shell destruction caused by hydraulic fragmentation and abrasion.

Persistence of Neothauma in littoral beds has important implications for the 101

structuring of specialized communities of shallow water benthos, as well as for improving analog models for hydrocarbon reservoirs in lacustrine carbonates.

Introduction

The inland waters of tropical East Africa have been a subject of curiosity since the mid-19th century, when explorers commissioned by the Royal Geographical Society

(RGS) in London began searching for the source of the Nile River. The second RGS expedition, famously led by R.F. Burton and J.H. Speke, introduced the western world to several of the Great Rift Valley lakes, including Lake Tanganyika (LT). Since that time,

LT has been the subject of many decades of rich scientific inquiry, spanning such diverse fields as evolutionary biology, limnogeology, and paleoclimatology. Today, the lake is recognized as one of the most biologically diverse freshwater ecosystems found anywhere on Earth (Kawanabe et al., 1997). LT is especially noted for its endemic fauna, which includes species flocks of cichlid fish, crabs, and gastropods (Coulter, 1994;

Cumberlidge et al., 1999; West et al., 2003).

Due to their sediment-producing potential, organisms with carbonate hardparts lie at an important interface between bio- and geosystems in lake basins. In LT’s littoral zone, endemic mollusks produce conspicuous biogenic carbonate accumulations. Cohen and Thouin (1987) first documented the presence of these nearshore coquinas, while subsequent expeditions mapped similar surficial deposits in water depths up to ~70 m

(Tiercelin et al., 1992; Soreghan and Cohen, 1996). In most cases, these deposits are dominated in biomass by shells of the gastropod Neothauma tanganyicense Smith, 1880. 102

Hereafter, we will refer to this snail by its generic name alone for convenience. The species is relatively well known to ichthyologists, as numerous fish utilize these gastropod shells as their breeding substrate (Sato and Gashagaza, 1997; Gordon and Bills,

1999; Koblmüller et al., 2007). From a geologic perspective, however, many important questions regarding Neothauma and the accumulations they help form remain unanswered. For instance, detailed facies and taphonomic analyses have not been attempted. Studies focusing on geochronology are likewise absent. As a consequence, little is known about the accumulation history of these deposits. This knowledge gap is striking considering the importance of lacustrine shell beds in the rock record, both as archives of paleobiologic information and as hydrocarbon reservoir rocks in continental rifts (Williamson, 1981; Abrahão and Warme, 1990; Chang et al., 1992; van Damme and

Pickford, 1999).

The goal of this study is to provide a thorough analysis of Neothauma shell accumulations found in the littoral zone of the Kigoma Basin, western

Tanzania. Our approach has been threefold. First, field data were used to refine facies characterizations of these deposits. Second, radiocarbon dating was done on

Neothauma shells from each newly defined facies to provide preliminary constraints on the extent of time averaging within accumulations. Finally, we performed taphonomic analyses on Neothauma in order to evaluate post-mortem damage across different littoral environments. Whereas numerous taphonomic 103

studies have been conducted on invertebrates in marine environments (e.g.,

Davies et al., 1989; Meldahl and Flessa, 1990; Best and Kidwell, 2000), similar studies on lacustrine systems are comparatively rare. Our integrated assessment brings new focus on the mechanisms of shell-bed development and preservation in the littoral zones of continental rift lakes. Post-mortem Neothauma shell survival appears to influence, and potentially structure, a number of specialized benthic communities within LT—a far-reaching implication for future biodiversity, conservation, and paleobiologic studies in the basin.

Background

Geologic Setting

Lake Tanganyika, situated between 3° S and 9° S in East Africa, is the world’s largest tropical rift lake (Fig. 1). Most geologic evidence suggests that LT formed during the middle Miocene (9–12 Ma; Cohen et al., 1993; Nyblade and

Brazier, 2002). At present, the lake occupies several linked half-graben basins and acts as a mixed carbonate-siliciclastic depositional system strongly influenced by both tectonics and climate (e.g., Soreghan and Cohen, 1996). Lake Tanganyika has a volume of ~19,690 km3 and reaches a maximum depth of ~1470 m in its southern basin (Rosendahl, 1988; McGlue et al., 2008). The lake is hydrologically open, draining to the west via the Lukuga River (Fig. 1). Lake waters are slightly 104

conductive due to the warm monsoonal climate (20–24°C mean annual temperature [MAT]) and significant yearly losses due to evaporation. Lake

Tanganyika is saturated with respect to carbonate due to cation-rich input from the Ruzizi and Malagarasi Rivers (Haberyan and Hecky, 1987; Casanova and

Hillaire-Marcel, 1992). Small hydrothermal vents in northern LT indicate the potential importance of groundwater to the lake’s ionic composition and alkalinity, but extensive datasets from other coastal regions are lacking (Tiercelin et al., 1993). Southerly wind-driven waves impact the lake’s littoral zone during the dry season (May-September) and wind speeds over historic times have ranged up to ~11 m/s (O’Reilly et al., 2003).

Study Area

The study area is situated along the shoaling margin of the lake’s central basin near the town of Kigoma, Tanzania (Fig. 2) and along the flexural margin of the Kigoma basin. Kigoma’s nearshore region displays four geomorphic environments: (1) fault-controlled headlands, (2) embayments, (3) beaches, and

(4) a delta. Headlands are characterized by steep outcrops of Proterozoic quartzite and cobble-dominated beaches. Embayments are typically narrow (<2 km in breadth) cuspate environments with sand to cobble beaches. Commonly, bays are separated from one another by promontories with significant (>50 m) topographic 105

relief. The lake floor associated with headlands and embayments exhibits stepped or ramp-like morphologies with high gradients (Fig. 2). Conversely, the bathymetry of the Luiche River delta resembles a platform; the lake floor has a low gradient over ~3 km from the river mouth, but abruptly ends at a westward dipping slope. Beaches located northwest of the Luiche delta are moderately wide

(~50–150 m), sandy, and lack significant vegetation.

Target Taxon

Neothauma tanganyicense, the most common large bioclast in littoral shell carbonates, is a species of viviparous gastropod endemic to LT (Fig. 3). Van

Damme and Pickford (1999) have proposed that the genus evolved early in the

Miocene, based on fossil occurrences in the Lake Albert Basin. Neothauma is considered a relict endemic of a more widespread, diverse clade, but it is closely related to Bellamya, which includes a number of extant African species (van

Damme and Pickford, 1999, Sengupta et al., 2009). Neothauma is thought to comprise a single living species and thus contrasts with the highly diverse cerithioid snail clades in LT, some of which occur in shell-rich facies but generally account for a trivial proportion of the total biomass.

Neothauma are recognizable by their smooth, globose shells with rounded whorls, sinuous growth lines, dextral coiling, and obtuse protoconch (Brown, 106

1980). Shells display determinate growth and typical adult heights range between

35–65 mm (Leloup, 1953). Shells of live animals range in color from white to light brown to gray (Fig. 3A). Neothauma distribution is patchy but lakewide, with known shell beds indicated in Fig. 1. Although few published data exist on

Neothauma ecology, early dredge surveys and recent SCUBA sampling have encountered live Neothauma in water depths up to 50 m on a wide range of sandy, muddy, and shelly substrates (Leloup, 1953; E. Michel and G. Kazumbe, personal observations, 2006). Neothauma-rich facies are absent beneath LT’s chemocline (80–150 m water depth), as anoxia and alkalinity limit snail habitat and enhance corrosion of shell carbonate, respectively.

Methods

Dead Neothauma shells were collected from lake floor substrates using

SCUBA and a variety of hand-operated devices, including scoop samplers and box cores. Divers described deposits in the field following the methods outlined in

Kidwell et al. (1986) and Kidwell and Holland (1991). Modern facies nomenclature follows the recommendations of Schnurrenberger et al. (2003) for lake sediments. In some cases, lithified Neothauma deposits occur just beneath unconsolidated lake-floor deposits or in older outcrops on the lake plain. Samples of this kind were sectioned for microfacies analysis and examined using a Leitz 107

Ortholux petrographic microscope with a Luminoscope ELM2A cathodoluminescence (CL) adapter.

At each site, shell collections were made from the upper 10 cm of the lake floor in order to complete a comparative biostratinomic analysis (Table 1).

Sampling along the Luiche River delta was accomplished by swimming depth transects orthogonal to the shoreline (NE to SW; Fig. 2). Depth transects provided the opportunity both to evaluate the spatial extent of shell-rich accumulations, as well as to collect shells from representative sites in both shallow and deeper water. Presence-absence counts of erect sponges, which commonly grow on shell beds, were also collected during these transects. Because Neothauma-rich accumulations along headlands and bays are spatially restricted, shells were collected along lines that crossed several isobaths at these sites. At beach sites, shells were retrieved from representative deposits. In all cases, shell samples were labeled and securely packaged to minimize damage during transport. Prior to analysis, shells were gently washed with deionized water and dried at 40°C. Eight nicely preserved Neothauma shells from surficial collections were selected at random for radiocarbon dating. Handling and pretreatment followed standard procedures for shell carbonate. Radiocarbon dates were corrected for LT’s old 108

carbon reservoir using the curve developed by Felton et al. (2007) and converted to calendar years using CalPal07 (Table 2; Weninger et al., 2008).

For the biostratinomic analysis, all shells (n = 1547) were examined under

10x stereoscopic magnification. Only shells and identifiable shell fragments >4 mm from lake floor collections were considered in the analysis. Each Neothauma shell was compared to a reference set and scored for four taphonomic variables:

(1) fragmentation, (2) abrasion, (3) encrustation, and (4) oxidation patinas (Table

3). The scoring system recognized three damage states per variable: no damage

(score = 0), low damage (score = 1), and high damage (score = 2). A single evaluator (McGlue) was employed for the analysis in order to maintain internal consistency. Following Kidwell et al. (2001), results were presented as high- threshold damage profiles with 95% confidence intervals and ternary taphograms

(Kowalewski et al., 1995) of full-frequency data.

Results

Sedimentology

Embayments (Sites HT-1-05 and KB-1-05).—Accumulations encountered in bays are parautochthonous, poorly-sorted beds of gravelly mollusk hash. These deposits are restricted to 15–22 m water depth and form wedge-shaped aprons along slope fronts (Fig. 4A). Compositionally, the deposits consist of whole and 109

fragmented bioclasts and coarse lithoclasts. The bioclasts include multiple species of mollusks, but Neothauma and the small (<25 mm long) unionid clam

Coelatura burtoni Woodward are dominant in the size fraction that is >4 mm. Box cores and trenching reveal densely packed sediments lacking preferential arrangement of skeletal material in cross section. Beds are internally complex and up to 20 cm thick; modern sediments overlie well-indurated equivalents. In thin section, large well-preserved fragments of Neothauma aragonite are conspicuous, marked by green CL. These fragments are cemented with smaller mollusks and coarse clasts of sandstone, quartzite, and quartz (Fig. 5A). Calcite cement is dominant, marked by bright orange-red CL.

Headland (Site MB-1-05).—Accumulations encountered at the headland site are parautochthonous, well-sorted patches of gravelly mollusk hash.

Individual patches are restricted to the hollows created between stromatolite pillars and typically cover <5 m2 (Fig. 4B). Compositionally, the deposits consist of

Neothauma with shells of Pleiodon spekii Woodward (a robust, endemic unionid mussel) and gravel- to cobble-sized sandstone lithoclasts present in lesser abundances. Patches are densely packed with highly variable thicknesses, in some instances exceeding 0.5 m. Thin sections from indurated deposits display 110

microbial carbonate (patchy pink CL) with a massive to crudely laminated microstructure (Fig. 5D).

Beaches (Sites UB-1-05 and UB-2-05).— Accumulations encountered along beaches are allochthonous, well-sorted ridges of sandy mollusk hash. Individual ridges are elongate and narrow, typically <2 m wide and tens to hundreds of meters long with multiple semi-parallel ridges occurring in certain locations (Fig.

4C). The ridges are polytypic, consisting of Neothauma shells with other larger mollusks, including Coelatura burtoni, Pila ovata (Olivier), and Melanoides admirabilis (Smith). Sediments are loosely packed and lack preferential arrangement of skeletal material along bedding planes and in cross section. Beds are internally simple, grading from sand to pure shell material; individual ridges can reach up to 10–15 cm thick.

Deltaic Platform.—This environment includes sites of varying water depth: sites NLP-1S-06 and NLP-2S-06 (both 8 m water depth), sites NLP-1D-06 and NLP-

2D-06 (both 20 m deep), and site ULB-1-05, with 10 m water depth.

Accumulations studied along the Luiche River delta cover at least 8 km2, although spot grab samples suggest the deposit is likely more extensive. Our transect-based observations allow us to differentiate two subenvironments based on water depth. Deposits in <10 m of water are allochthonous beds of bioturbated, 111

sandy-silty mollusk hash (Fig. 4D). In contrast, accumulations in deeper water

(~20 m) are parautochthonous beds of pure mollusk hash. Beds in both subenvironments are polytypic, dominated by Neothauma, with varying abundances of Coelatura shells and other smaller bioclasts. The bioclasts do not display any discernable preferred orientation. Beds in shallow water are loosely packed and mixed up to 50% by weight with sand and silt, whereas the accumulations in deeper water display dense packing of shells, less siliciclastic material, and many erect sponges (Fig. 4E). Trenching along the depth transects indicates that beds thicken with increasing water depth up to a maximum thickness of ~15–20 cm.

An ancient example of the shallow water hash was retrieved from outcrops north of the lake, along the plains of the Ruzizi River (Fig. 1). In thin section, large fragments of heavily pitted, recrystallized Neothauma (dull CL) are conspicuous, cemented with calcite to sand-sized grains of angular to sub- rounded quartz (Fig. 5E). A lithified equivalent of the deep water hash was discovered in association with the modern facies on the Luiche platform. In thin section, high-Mg calcite exhibiting dark red CL thickly coats well-preserved gastropod aragonite (Fig. 5H).

Geochronology 112

Results of radiocarbon dating are summarized in Table 2. Age data reveal an admixture of late Holocene shell material across the study area. Neothauma from the embayment and headland sites (n = 4) are similar in age, clustering around the early 19th century (median age of 1807 + 105 CE [common era] reservoir-corrected calendar years). Radiocarbon ages on shells collected from the

Luiche delta (n = 4) exhibit a wider variation. Two shells collected along the ~20 m isobath date to 337 + 60 CE and the present, respectively. A shell collected from the ~8 m isobath yielded an age of 1809 + 98 CE, whereas a shell collected up-dip along the beach returned an age of 1004 + 21 CE. These data, together with a reservoir corrected shell age from a previous study (Cohen et al., 1997), demonstrate a mixed temporal range spanning more than 1600 calendar years at the two-sigma level.

Taphonomy

The rank-order importance of the four taphonomic variables evaluated in this study varies with littoral subenvironment (Fig. 6). Qualitative evaluation of high-threshold damage profiles suggests three groups among the sites in terms of their overall taphonomic signature. Shells accumulating along beaches and those in shallow water on the deltaic platform exhibit a characteristic pattern wherein abrasion > fragmentation > oxidation patina > encrustation damage (Fig. 113

6). Neothauma collections from beds at depths >10 m along the deltaic platform and at the headland site exhibit a different taphonomic signature in which encrustation > oxidation patina > fragmentation > abrasion damage. Encrustations at the headland site are heavy and stromatolitic (Fig. 3D), whereas oxidation patinas are generally thin stains and cover <30% of the shell exterior. In contrast, oxidation patinas on samples collected along the 20 m isobath on the deltaic platform commonly cover >50% of the exterior surface (Fig 3E). Encrustations at these sites are patchy and probably algal in origin. The final group, where

Neothauma shells exhibit damage with a pattern of encrustation > fragmentation

> oxidation patina > abrasion, typify accumulations in embayments (Fig. 6).

Encrustations on shells from embayments are commonly stromatolitic.

Pair-wise tests using the confidence limits presented in Figure 6 provide a means for quantitative comparison among the sites (Table 4). In many cases, sites with dissimilar bathymetry exhibit statistically significant differences in damage state. For example, shells along shallow to deep transects on the deltaic platform are commonly statistically different from each other for all taphonomic variables evaluated. Shallow water samples on the delta are statistically similar to shells collected along the platform beach; only site NLP-2S-06 exhibits statistically different abrasion patterns than the other three locales. Shallow water delta and 114

beach samples are strongly dissimilar from the headland site and mostly dissimilar in comparison to the two bay sites (Table 4). In these cases, similarity among sites occurs most commonly for the fragmentation variable. Samples from all sites in >10 m of water, regardless of lake floor gradient, are statistically similar for abrasion.

Ternary taphograms provide a full-frequency perspective on damage states among sites. Fragmentation frequency is inversely related to water depth (Fig.

7A). Shells situated in deep water plot towards the no damage and low damage poles, whereas shells situated <10 m sub-bottom are almost invariably heavily fragmented. A strong spread is likewise evident on both the abrasion and encrustation taphograms. Abrasion data also exhibit an inverse relationship with water depth; samples at >10 m water depth typically group along the no-damage pole (Fig. 7B). Encrustation patterns exhibit a positive relationship with water depth (Fig. 7C). Neothauma shells collected from sites in <10 m of water depth rarely display encrustation on more than 10% of the shell exterior. Frequency data for oxidation patina coverage are more complicated, especially on the deltaic platform (Fig. 7D). Shells located in ~20 m water depth group near the high- damage pole, whereas samples from shallow water group towards the low- damage pole. Along the platform beach, shells commonly show little evidence of 115

oxidation patinas and plot near the no-damage pole. The bay sites likewise group near the no-damage pole, as does site ULB-1-05. Damage patterns from ULB-1-05 are intriguing because they commonly plot relatively close to the embayment sites for all variables. This site is a bay south of the Luiche River delta, but the lake-bottom gradient at this site more closely resembles other sites on the deltaic platform. Pair-wise comparisons of damage indicate that ULB-1-05 is statistically indistinguishable from HT-1-05 for abrasion and KB-1-05 for fragmentation and abrasion (Table 4).

Interpretations

Embayment and Headland Sites

We interpret the Neothauma shell beds encountered at the embayment

(HT-1-05 and KB-1-05) and headland (MB-1-05) sites as relict parautochthonous assemblages (sensu Kidwell et al., 1986) due to: (1) the absence of live animals, despite extensive SCUBA surveys; (2) repeated radiocarbon ages centered on the early 19th century; and (3) shell erosion patterns implying minimal long-distance transport (Fig. 6; Table 2). Neothauma shells in these locales probably reflect populations that existed at a time when environmental conditions allowed for the accumulation of fine-grained sedimentary organic matter, providing the snails with a readily available food source. Today, net sedimentation in embayments 116

and along headlands is low, and dead Neothauma shells commonly far outnumber living specimens (live-dead disagreement). Living populations of

Neothauma are known to inhabit clear shallow waters marked by sandy and silty substrates (E. Michel, personnel observations, 2006). Following this analog, the extant lake bottom at embayments and headlands near Kigoma likely does not support large populations of live Neothauma. Sedimentation along the flexural margins of rift lakes, however, is inherently subject to change, especially considering the frequency of base-level variations associated with climate change in the tropics. Remarkably consistent radiocarbon data on Neothauma from embayments and headlands, coupled with low abrasion damage, provide compelling evidence for the concentration of a local population of animals sometime between the early 18th and early 20th centuries. During this interval, modest lake-level fluctuations were common due to regional shifts in precipitation associated with the Little Ice Age (~1550–1850 CE) as well as a ~10 m regression associated with a breach of LT’s sediment-filled outlet channel in the late 1800s (Alin and Cohen, 2003; Cohen et al., 2005). We conclude that relative shifts in the paleoshoreline during this interval periodically allowed more hospitable conditions for Neothauma in these environments. 117

We interpret shell bed formation in embayments to be linked with mass transport processes. Structureless bedding and chaotic arrangement of bioclasts suggest slope-front, gravelly mollusk hash beds may be the remains of shallow sublacustrine slides. Slides were probably common along bathymetric slopes during the Little Ice Age, as gravitational instabilities and changes in interstitial pore pressures along slope crests likely resulted from regressions >20 m (Cohen et al., 2005). Massively bedded lithified sediments underlying the modern hash beds are bioclastic hybrid arenites. In thin section, these rocks exhibit poor size sorting, consistent with a mass transport origin (Fig. 5A). Calcite cement (Fig. 5B) observed in bioclastic hybrid arenites may result from input of calcium and bicarbonate-rich fluids along coastal faults, but extensive data on the geohydrology of LT are absent. Relict Neothauma shells in these deposits appear to be most affected by processes resulting in fragmentation and encrustation (Fig.

6). Fragmentation of mollusk shells is often difficult to interpret, given the vast number of processes that result in shell breakage (e.g., Zuschin et al., 2003). Our data suggest mechanical processes outweigh biologic processes as the primary cause of shell fragmentation in the bays we studied. In embayments, the maximum distance shells can be transported by currents is limited by lake- bottom physiography. Nevertheless, reworking of shells is common, given the 118

high instantaneous current velocities achieved in each half-wave cycle during storms (e.g., Allen, 1985). Saltation of Neothauma shells along the lake floor during storms may provide an effective means of producing cracking along weak points on the shell. Alternatively, fragmentation may result from biologic processes, but we consider such damage less likely. Molluscivorous crabs abound in LT, but experimental studies demonstrate that shell microstructure and size makes adult Neothauma strongly resistant to predation (e.g., West and Cohen,

1996). Transport of dead Neothauma by nesting cichlids seems unlikely to contribute to significant fragmentation, but maintenance of shell middens does keep Neothauma shells at the lake bottom and thus subject to more continuous physical destruction. In order for these processes to effectively operate,

Neothauma shells must remain exposed on the lake floor; heavy coverage by encrusting sponges and bryozoans suggests this is indeed the case (Fig. 9).

Gravelly patches of mollusk hash discovered at the headland site are interpreted as mixed origin (sedimentologic and extrinsic biogenic) concentrations. Accumulations at MB-1-05 are similar in some aspects to cichlid nests described by Sato and Gashagaza (1997) but a number of differences suggest concentration is not accomplished solely through biogenic means.

Observations of Lamprologous callipterus show that these cichlids build 119

substantial so-called clump nests along the fringes of rocky substrates at many locales in Lake Tanganyika (Sato and Gashagaza, 1997). Although empirical data on transport distances are lacking, observations suggest these fish do not ordinarily move shells more than 10 m when constructing nests (M. Taborsky, personal communication, 2008). Since stromatolitic reefs along headlands span several tens of meters in the dip direction, hash patches may represent abandoned nests that were subsequently reworked into the inter-reef accumulations by wave and storm-driven currents. Such reworking explains the admixture of Neothauma with materials not directly useful for brooding, such as sandstone cobbles and disarticulated valves of the unionid mussel Pleiodon spekii (Fig. 4B). Ultimately, these sediments are cemented together to form gastropod-rich stromatolite boundstones (Fig. 5C).

Taphonomic processes resulting in encrustation and the development of oxidation patinas dominate along headlands (Fig. 7). Cryptic irregular laminations revealed in CL images and XRD-based mineralogy indicate encrustations are aragonitic with minor biogenic calcite and probably stromatolitic in origin (after Mazzoleni et al., 1995). We suggest these features contribute to shell preservation in this environment. Wilson (1975) noted that encrustations reduce post-mortem destruction of chambered hollow skeletons 120

along high-energy marine shelves. Headland environments in rift lakes are also high-energy environments where the potential for shell preservation through burial in fine-grained sediment is low. Rapid encrustation of Neothauma shells likely abates the mechanical processes that would otherwise lead to shell loss due to fragmentation or abrasion. As noted above, extensive data on groundwater flows into LT are lacking, but Rosen et al. (2004) noted the importance of concentrated seeps for the accelerated growth of lake-margin microbial carbonates and a similar process may be influencing shell beds situated near coastal faults near Kigoma.

Delta Sites

Mollusk hash deposits encountered along the deltaic platform, from ~20 m below the lake surface to the ridges encountered along beaches, are interpreted as sedimentological concentrations with heavy extrinsic biogenic modification at shallow-water sites. Damage patterns suggest that at present fossil and subfossil shells from ~ 20 m of water are transported by storm currents into ~

8 m of water and form sandy and silty mollusk hash beds. Radiocarbon ages on shells from the 8 m isobath and the platform beach (1809 + 98 CE and 1004 + 21

CE, respectively) demonstrate temporal mixing of older Neothauma populations in the pure mollusk hash beds situated in deeper water. Bioturbation by fish 121

strongly controls depositional style along the ~ 8 m isobath, as evidenced by numerous mounds, depressions, and scattered nests observed during diver surveys (Fig. 4D). Ultimately, many of these shells are transported onto the platform beach to form ridges. Shells accumulating in beach ridges and at ~ 8 m below the lake surface on the platform exhibit pronounced abrasion and fragmentation damage (Fig. 7). Pitting from impacts of mobile sand grains and the oscillatory rolling of shells along the lake bottom due to waves produce characteristic damage to Neothauma, resulting in complete loss of external growth lines, luster, and shell density (Fig. 3C). Although shell survival in the foreshore area is difficult, preservation in paleo-highstand deposits does occur.

Allochemical sandstones encountered along the lake plain attest to the preservation of such deposits, albeit with alteration from meteoric diagenesis commonly observed (Fig. 5F).

We interpret the pure mollusk hash encountered along the ~20 m isobath to have accumulated by sedimentologic processes. Radiocarbon data on shells from this locale are time averaged and include the youngest (modern) and oldest

(337 + 60 CE) shells in the study area. These data are important because they confirm that samples of Neothauma are parautochthonous in these locales, and demand a concentration mechanism that allows for temporal mixing. Presently, 122

sedimentation rates northwest of the Luiche River appear to be low, probably as the result of sub-lacustrine channels that divert deltaic sediments to the south

(e.g., Soreghan et al., 1999). Modern Neothauma are not limited by food resources in this environment, but our surveys and initial radiocarbon data indicate that ancient shells are more abundant than subfossil shells in these deposits.

Taphonomic data reveal a dominance of damage associated with prolonged lake- floor exposure, expressed as encrustation and oxidation patina coverage on shell surfaces (Fig. 7). In contrast, heavy shell fragmentation is low to moderate and abrasion damage is virtually absent. Well-preserved gastropod aragonite observed in hardground grainstones suggests that these deposits have remained submerged, as dissolution associated with meteoric diagenesis is absent (Fig. 5H).

In concert with inferences from sediment core studies, we interpret these data to reflect a complex accumulation history that is probably typical of low gradient littoral zones in tropical rift basins. In ocean basins, shell lags develop as sea level rises and wave ravinement reworks nearshore sediments (e.g., Cattaneo and Steel,

2003). Commonly, shell lags develop as thin but regionally extensive units overlying sediments with evidence of shoreface erosion and inundation (Van

Waggoner et al., 1990; Cattaneo and Steel, 2003). Figure 8 illustrates the recent stratigraphy of Neothauma hash beds north of the Malagarasi River delta, ~30 123

km south of the Luiche River delta (Parson, 2001). At this locale, pure mollusk hash beds overlie sandy units with abundant shell fragments and a muddy transgressive surface. We interpret this stacking pattern as evidence for recent landward migration of the shoreline. Radiocarbon data, though limited, place live

Neothauma along the Malagarasi delta near the beginning of the Little Ice Age

(~ 1535 + 73 CE) and thus local death assemblages would have been subject to concentration by wave action. Given their proximity, we conclude the same process acted to concentrate mollusk hash beds offshore of the Luiche River delta. The role of hydraulic and biogenic processes associated with lake-level lowstands, however, cannot be discounted in the development of the deposits we observed. For example, Cohen (1989) noted that phenotypic deviance among

Paramelania damoni (Smith) shells could be reconciled by time averaging of populations solely through winnowing of vertically stacked shell beds.

Winnowing does not seem to be the dominant process affecting mollusk-hash bed formation along the Luiche delta, as the majority of our samples lack black reduction patinas characteristic of prior burial in oxygen-deficient sediments (e.g.,

Owen et al., 1996). Reworking by a change in wave base associated with the late

Little Ice Age transgression (mid-19th century) provides an alternate mechanism for the admixture of diachronous P. damoni described by Cohen (1989) and fits 124

more robustly with data from this study. We suggest that winnowing and biogenic activities probably play secondary roles in concentrating Neothauma during periods of relative lake-level lowstand.

Discussion

Examples of Neothauma-rich facies in the recent rock record at LT confirm preservation across a variety of littoral subenvironments (Fig. 9). Water depth clearly influences depositional style, patterns of damage, and in some instances, processes of preservation. Our analysis indicates shell survival in Lake

Tanganyika’s littoral zone hinges on post-mortem shell modifications that abate mechanical taphonomic processes, as adult Neothauma shells show little indication of biologically induced damage. The most destructive taphonomic zone in LT, and probably of most mixed carbonate-siliciclastic rift lakes, is the foreshore of low-gradient deltaic platforms (Fig. 9). Here, mobile sand grains induce heavy taphonomic damage as bioclasts are abraded and fragmented.

Mitigation of damage in this environment is difficult, but at greater water depths, shell preservation is possible in spite of prolonged exposure on the lake bottom.

Shells at sites NLP-1D-06 and NLP-2D-06 are currently developing into hardgrounds due to heavy post-emplacement calcite cementation (Fig. 3D). Early diagenetic cements increase both shell diameter and density, which in turn 125

increase the environmental energy required for shoreward transport. These processes create a preservation feedback loop that extends taphonomic half life

(sensu Cummins et al., 1986) along deltaic platforms, as residence in deeper water below wave base ultimately delays transport into the taphonomically active foreshore. The origins of early diagenetic cements in LT are speculative, given the dearth of data on coastal hydrology, but mixing of lake water with shallow groundwater aquifers may provide the conditions necessary for early cementation. It is important to note that the climatic sensitivity of lake systems may also play a role in littoral zone shell preservation. Frequent deltaic avulsions and lobe switching prompted by changes in effective precipitation can promote preservation through rapid burial of shell-rich accumulations.

Neothauma-rich accumulations detailed in this study suggest a wider diversity of littoral zone carbonate facies than has previously been documented for rift lakes. The rock record in East Africa exhibits shell-rich accumulations mostly in low gradient littoral environments. For example, Betzler and Ring

(1995) reported the existence of bivalve-rich accumulations from the Chiwondo beds of Lake Malawi that are similar to the hash deposits encountered on the

Luiche River delta. Lake Turkana’s famous mollusk-rich facies of the Plio–

Pleistocene Koobi Fora Formation (Brown and Feibel, 1986; Lepre et al., 2007) 126

and shore-zone accumulations from Lake Bogoria (Renaut and Owen, 1991) are broadly similar to the sandy mollusk hash deposits found at shallow depths on the Luiche platform. Intriguingly, few studies have noted the potentially important contributions of fish towards shaping the depositional style of shell- rich accumulations. Cichlids play a clear role in concentrating shells at headland and embayment sites as well as at 8-m-water depth on the Luiche Platform, where fish and nest counts were high along our transects (Castañeda and O’Connell,

2006). Feibel (1987) noted the existence of fossil fish nests in the bioclastic sandstones and packstones in the Koobi Fora, providing supporting evidence that extrinsic biogenic processes play at least an accessory role in the development of some rift-lake nearshore carbonates.

From a hydrocarbon systems perspective, rift lakes hold significant value for their potential to develop organic-rich source intervals (e.g., Talbot, 1988).

Reservoir intervals in rift lakes have received less attention from industry, most likely because of size limitations and poor reservoir quality associated with flexural margin sand bodies (Katz, 2001). Cretaceous coquina reservoirs in rift lakes are an important exception to this trend, as prolific fields offshore of Brazil and the Congo demonstrate (e.g., Abrahão and Warme, 1990; Harris et al., 1994;

Carvalho et al., 2000). Our study highlights at least two new rift-lake reservoir 127

facies: bioclastic hybrid arenites and gastropod-rich stromatolitic boundstones.

Vertical stacking juxtaposes both facies types with stromatolite reefs, which together may represent significant prospective reservoir intervals over geologic time. Both facies develop along high-gradient depositional surfaces down dip of significant paleorelief (faulted promontories), which could aid identification in subsurface datasets.

The implications of shell survival in long-lived lacustrine systems are intriguing. In terms of rift lakes, studies of cichlid fish bear on this topic. Ribbink

(1990) noted that the species richness of fish communities inhabiting hard lake- floor substrates (i.e., boulders, cobbles) exceeds those associated with fine-grained substrates. The preservation of shells in LT’s littoral zone helps create a unique suite of intermediate-hardness substrates that helps to structure several specialized benthic communities. Ecological surveys across LT’s littoral zone have noted certain taxa living in association with Neothauma shell beds. Cichlid utilization of empty Neothauma shells is well documented, but other fish including Auchenoglanis occidentalis (Valenciennes), a bagrid catfish, appear to utilize shells in nest-building activities as well (Ochi et al., 2001). Certain crab species, including Platythelphusa maculata Cunnington, also appear to be restricted to Neothauma-rich substrates (Marijnissen et al., 2008). The small, 128

rounded bodies of P. maculata and their marked substrate preference suggest a potential evolutionary adaptation reliant on the presence of Neothauma shell beds. Additionally, Michel (2004) encountered the endemic cerithioid snail

Vinundu westae Michel living on shell beds, as well as their better-documented rocky habitats. Census work conducted during our sampling transects noted an abundance of freshwater erect sponges living on the shell beds, particularly at ~

20 m water depth on the deltaic platform (Fig. 4E). Mean sponge abundances at sites NLP-1D-06 and NLP-2D-06 exceeded several hundred individuals of varying morphotypes (Castañeda and O’Connell, 2006). Although research on LT’s sponges is in its infancy, the limited data available suggest that some species exclusively utilize Neothauma shell beds as habitat (Weier, 2005). Taken in concert, these observations suggest that the preservation of dead Neothauma may directly contribute to the structuring of shallow benthic communities in LT’s littoral zone.

Conclusions

1. Actualistic analyses of Neothauma-rich accumulations in Lake

Tanganyika’s littoral zone reveal three modern facies types: gravel-rich mollusk hash, sandy and silty mollusk hash, and pure mollusk hash. Depositional style varies with lake-floor gradient and water depth, ranging from slope-front aprons 129

and patches (embayments and headlands, respectively) to beach ridges and expansive beds (low gradient deltaic platform). Vertical stacking of gravel-rich, mollusk hash beds with stromatolitic reefs constitutes a prospective reservoir facies in ancient lacustrine rift basins.

2. Reservoir-corrected radiocarbon data on Neothauma from the study area suggest that mollusk hash beds on the Luiche River delta are time averaged over at least the latest Holocene. Pure mollusk-shell beds on the Luiche delta are interpreted as transgressive lags, consistent with stratigraphic inferences from the

Malagarasi River delta. Radiocarbon data on Neothauma shells from embayment and headland sites center on the early 19th century and are consistent with fluctuating paleoenvironments of the Little Ice Age.

3. Evaluation of four taphonomic variables suggests relatively high preservation potential for Neothauma shells across the littoral subenvironments encountered near Kigoma. Preservation is lowest at shallow (<10 m) water depths on deltaic platforms, where abrasion and fragmentation damage from interaction with waves and sand grains are pronounced. Rapid development of stromatolitic encrustations and early cement coatings may help reduce post-mortem shell destruction by mechanical processes. 130

4. Shell survival in the littoral zone may have important implications for structuring specialized benthic communities within Lake Tanganyika. Gastropod- rich shell-hash beds constitute a key intermediate hardness substrate that seems to be exclusively used by a number of specialized organisms, including several species of fish, crabs, and sponges. Biogenic feedback of this kind is uncommon in freshwater systems and is probably limited to long-lived lakes where the processes of evolution and diagenesis have time to operate.

Acknowledgments

This study was an outgrowth of the Nyanza Project, a Research Experience for

Undergraduates site program supported by the National Science Foundation (ATM-

0223902). We thank all the Nyanza Project students from the 2005 and 2006 field seasons for their contributions leading to the completion of this work; special thanks to C. Gans and L. Powers for their tireless efforts in the field and lab. G. Kazumbe provided unique

SCUBA expertise and guidance that was vital for scientific diving in Lake Tanganyika. I.

Petit and M. Mukli provided key logistical support and D. Dettman generously provided access to the CL microscope. Constructive reviews from E. Gierlowski-Kordesch and M.

Zuschin substantially improved the text.

131

References

ABRAHÃO, D., and WARME, J.E., 1990, Lacustrine and associated deposits in a rifted continental margin: Lower Cretaceous Lagoa Feia Formation, Campos Basin, offshore Brazil, in Katz, B.J., ed., Lacustrine Basin Exploration—Case Studies and Modern Analogs: AAPG Memoir, v. 50, p. 287–305.

ALIN, S.R., and COHEN, A.S., 2003, Lake-level history of Lake Tanganyika, East Africa, for the past 2500 years based on ostracode-inferred water-depth reconstruction: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 199, p. 31– 49.

ALLEN, J.R.L., 1985, Principles of Physical Sedimentology: The Blackburn Press, Caldwell, NJ, 272 p.

BETZLER, C., and RING, U., 1995, Sedimentology of the Malawi-Rift: facies and stratigraphy of the Chiwondo-Beds, northern Malawi: Journal of Human Evolution, v. 28, p. 23–36.

BEST, M. M. R., and KIDWELL, S. M., 2000, Bivalve taphonomy in tropical mixed siliciclastic–carbonate settings. I. Environmental variation in shell condition: Paleobiology, v. 26, p. 80–102.

BROWN, D., 1980, Freshwater Snails of Africa and their Medical Importance: Taylor and Francis, London, 617 p.

BROWN, F.H., and FEIBEL, C.S., 1986, Revision of lithostratigraphic nomenclature in the Koobi Fora region, Kenya: Journal of the Geological Society, London, v. 143, p. 297–310.

CARVALHO, M.D., PRACA, U.M., DA SILVA-TELLES, A.C., JAHNERT, R.J., and DIAS, J.L., 2000, Bioclastic carbonate lacustrine facies models in the Campos Basin (Lower Cretaceous), Brazil, in Gierlowski-Kordesch, E.H., and Kelts, K.R., eds., Lake Basins through Space and Time: AAPG Studies in Geology 46, p. 245–256.

CASANOVA, J., and HILLAIRE-MARCEL, C., 1992, Late Holocene hydrological history of Lake Tanganyika, East Africa, from isotopic data on fossil stromatolites: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 91, p. 35–48. 132

CASTAÑEDA, O., and O’CONNELL, C., 2006, Characterizing Molluscan Shell Accumulations: Ecological, Depositional and Taphonomic Variations on the Northern Luiche River Platform: Nyanza Project Annual Report (unpublished)

CATTANEO, A., and STEEL, R.J., 2003, Transgressive deposits: a review of their variability: Earth-Science Reviews, v. 62, p. 187–228.

CHANG, H.K., KOWSMANN, R.O., FIGUEIREDO, A.M.F., and BENDER, A.A., 1992, Tectonics and stratigraphy of the East Brazil Rift system: An overview: Tectonophysics, v. 213, p. 97–138.

COHEN, A.S., 1989, The taphonomy of gastropod shell accumulations in large lakes: An example from Lake Tanganyika: Paleobiology, v. 15, p. 26–45.

COHEN, A.S., and THOUIN, C., 1987, Nearshore carbonate deposits in Lake Tanganyika: Geology, v. 15, p. 414–418.

COHEN, A.S., SOREGHAN, M.J., and SCHOLZ, C.A., 1993, Estimating the age formation of lakes: an example from Lake Tanganyika, East African Rift system: Geology, v. 21, p. 511–514.

COHEN, A.S., TALBOT, M.R., AWRAMIK, S., DETTMAN, D.L., and ABELL, P., 1997, Lake level and paleoenvironmental history of Lake Tanganyika, Africa as inferred from late Holocene and modern stromatolites: Geological Society of America Bulletin, v. 109, p. 444–460.

COHEN A.S., PALACIOS-FEST, M.R., MSAKY, E.S., ALIN, S.R., MCKEE, B., O’REILLY, C.M., DETTMAN, D.L., NKOTAGU, H.H., and LEZZAR, K.E., 2005, Paleolimnological investigations of anthropogenic environmental change in Lake Tanganyika: IX. Summary of paleorecords of environmental change and catchment deforestation at Lake Tanganyika and impacts on the Lake Tanganyika ecosystem: Journal of Paleolimnology, v. 34, p. 125–145.

COULTER, G.W., 1994, Lake Tanganyika in Martens, K., Gooderis, B. and Coulter, G.W., eds., Speciation in Ancient Lakes: Archiv fur Hydrobiologie–Beihefte Ergebnisse der Limnologie, v. 44, p. 127–137.

133

CUMBERLIDGE, N., VON STERNBERG, R., BILLS, I.R., and MARTIN, H., 1999, A revision of the genus Platythelphusa A. Milne-Edwards, 1887 from Lake Tanganyika, East Africa (Decapoda: Potamoidea: Platyhelphusidae): Journal of Natural History, v. 33, p. 1487–1512.

CUMMINS, H., E., POWELL, N., STANTON, R.J., and STAFF, G.M., 1986, The rate of taphonomic loss in modern benthic habitats: how much of the potentially preservable community is preserved?: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 52, p. 291–320.

DAVIES, D. J., POWELL, E.N., and STANTON, R.J., 1989, Taphonomic signature as a function of environmental process: shells and shell beds in a hurricane- influenced inlet on the Texas coast: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 72, p. 317–356.

FEIBEL, C.S., 1987, Fossil fish nests from the Koobi Fora Formation (Plio– Pleistocene) of northern Kenya: Journal of Paleontology, v. 61, p. 130–134.

FELTON, A.A., RUSSELL, J.M., COHEN, A.S., BAKER, M.E., CHESLEY, J., LEZZAR, K-E., MCGLUE, M.M., PIGATI, J.S., QUADE, J., and STAGER, J.C., 2007, Paleolimnological evidence for the onset and termination of glacial aridity from Lake Tanganyika, East Tropical Africa: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 252, p. 405–423.

GORDON, A.K., and BILLS, I.R., 1999, Aspects of the feeding and reproductive biology of the Lake Tanganyikan cichlid, Lamprologus ornatipinnis (Pisces, Cichlidae): Environmental Biology of Fishes, v. 55, p. 431–441.

HARRIS, N.B., SORRIAUX, P., and TOOMEY, D.F., 1994, Geology of the Lower Cretaceous Viodo Carbonate: A lacustrine carbonate in the South Atlantic rift, in Lomando, A.J., Schreiber, B.C., and Harris, P.M., eds., Lacustrine Reservoirs and Depositional Systems: SEPM (Society for Sedimentary Geology) Core Workshop 19, p. 143–172.

HABERYAN, K.A., and HECKY, R.E., 1987, The late Pleistocene and Holocene stratigraphy and paleolimnology of Lakes Kivu and Tanganyika: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 61, p.169–197.

134

KATZ, B.J., 2001, Lacustrine basin hydrocarbon exploration—current thoughts: Journal of Paleolimnology, v. 26, p. 161–179.

KAWANABE, H., HORI, M., and NAGOSHI, M., 1997, Fish Communities in Lake Tanganyika: Kyoto University Press, Kyoto, 298 p.

KIDWELL, S.M., and HOLLAND, S.M., 1991, Field description of coarse bioclastic fabrics: PALAIOS, v. 6, p. 426–434.

KIDWELL, S.M., FÜRSICH, F.T., and AIGNER, T., 1986, Conceptual framework for the analysis and classification of fossil concentrations: PALAIOS, v. 1, p. 228–238.

KIDWELL, S.M., ROTHFUS, T.A., and BEST, M.M.R., 2001, Sensitivity of taphonomic signatures to sample size, sieve size, damage scoring system, and target taxa: PALAIOS v. 16, p. 26–52.

KOBLMÜLLER, S., DUFTNER, N., SEFC, K.M., AIBARA, M., STIPACEK, M., BLANC, M., EGGER, B., and STURMBAUER, C., 2007, Reticulate phylogeny of gastropod–shell- breeding cichlids from Lake Tanganyika—the result of repeated introgressive hybridization: BMC Evolutionary Biology, v. 7, no. 7, doi: 10.1186/1471-2148-7-7.

KOWALEWSKI, M., FLESSA, K.W., and HALLMAN, D.P., 1995, Ternary taphograms: Triangular diagrams applied to taphonomic analysis: PALAIOS, v. 10, p. 478–483.

LELOUP, E., 1953, Exploration hydrobiologique du lac Tanganyika (1946–47). Gastéropodes. Résultats Scientifiques de l’Exploration Hydrobiologique du Lac Tanganyika (1946–1947), v. 3, no. 4, p. 1–272.

LEPRE, C.J., QUINN, R.L., JOORDENS, J.C.A., SWISHER, C.C., and FEIBEL, C.S., 2007, Plio– Pleistocene facies environments from the KBS Member, Koobi Fora Formation: Implications for climate controls on the development of lake-margin hominin habitats in the northeast Turkana Basin (northwest Kenya): Journal of Human Evolution, v. 53, p. 504–514.

MARIJNISSEN, S.A.E., MICHEL, E., KAMERMANS, M., OLAYA-BOSCH, K., KARS, M., CLEARY, D.F.R., VAN LOON, E.E., RACHELLO DOLMEN, P.G., and MENKEN, S.B.J., 2008, Ecological correlates of species differences in the Lake Tanganyika crab radiation. Hydrobiologia, v. 615, p. 81–94, doi: 10.1007/s10750-008-9566-y. 135

MAZZOLENI, A.G., BONE, Y., and GOSTIN, V.A., 1995, Cathodoluminescence of aragonitic gastropods and cement in Old Man Lake thrombolites, southeastern South Australia: Australian Journal of Earth Sciences, v. 42, p. 497–500.

MCGLUE, M.M., LEZZAR, K-E., COHEN, A.S., RUSSELL, J.M., TIERCELIN, J-J., FELTON, A.A., MBEDE, E., and NKOTAGU, H.H., 2008, Seismic records of late Pleistocene aridity in Lake Tanganyika, tropical East Africa: Journal of Paleolimnology, v. 40, p. 635–653.

MELDAHL, K.H., and FLESSA, K.W., 1990, Taphonomic pathways and comparative ecologic and taphonomic facies in a modern intertidal/shallow shelf environment: Lethaia, v. 23, p. 43–60.

MICHEL, E., 2004. Vinundu, a new genus of gastropod (Cerithioidea: ‘Thiaridae’) with two species from Lake Tanganyika, East Africa, and its molecular phylogenetic relationships: Journal of Molluscan Studies, v. 70, p. 1–19.

NYBLADE, A.A., and BRAZIER, R.A., 2002, Precambrian lithospheric controls on the development of the East African rift system: Geology, v. 30, p. 755–758.

OCHI, H., KANDA, T., and YANAGISAWA, Y., 2001, Nest building and brooding behavior of the bagrid catfish, Auchenoglanis occidentalis (Valenciennes, 1840), in Lake Tanganyika: Copeia, v. 2001, p. 566–570.

O’REILLY, C.M., ALIN, S.R., PLISNIER, P.D., COHEN, A.S., and MCKEE, B.A., 2003, Climate change decreases aquatic ecosystem productivity of Lake Tanganyika, Africa: Nature, v. 424, p. 766–768.

OWEN, R.B., RENAUT, R.W., and WILLIAMS, T.W., 1996, Characteristics and origins of laminated ferromanganese nodules from Lake Malawi, Central Africa, in Johnson, T.C., and Odada, E.O., eds., The Limnology, Climatology and Paleoclimatology of the East African Lakes: Gordon and Breach, Amsterdam, p. 461–474.

PARSON, K.L., 2001, Controls on shallow-water carbonates of Lake Tanganyika: Unpublished M.S. thesis, University of Oklahoma, Norman, 77 p.

136

RENAUT, R.W., and OWEN, R.B., 1991, Shore-zone sedimentation and facies in a closed rift lake: the Holocene beach deposits of Lake Bogoria, Kenya, in Anadón, P., Cabrera., L. and Kelts, K., eds., Lacustrine Facies Analysis (Special Publication of the International Association of Sedimentologists 13): Blackwell, Oxford, p. 175–195.

RIBBINK, A.J., 1990, Alternative life-history styles of some African cichlid fishes: Environmental Biology of Fishes, v. 28, p. 87–100.

ROSEN, M.R., AREHART, G.B., LICO, M.S., 2004, Exceptionally fast growth rate of a 100-year-old microbially-mediated tufa, Big Soda Lake, Nevada: Geology, v. 32, p. 409–412.

ROSENDAHL B.R., 1988, Seismic Atlas of Lake Tanganyika, East Africa: Project PROBE Geophysical Atlas Series Vol. 1, Duke University, Durham, North Carolina, 82 p.

SATO, T., and GASHAGAZA, M.M., 1997, Shell brooding cichlid fishes of Lake Tanganyika: Their habitats and mating systems, in Kawanabe, H., Hori, M., and Nagoshi, M., eds., Fish Communities in Lake Tanganyika: Kyoto University Press, Kyoto, p. 219–238.

SCHNURRENBERGER, D.W., RUSSELL, J.M., and KELTS, K., 2003, Classification of lacustrine sediments based on sedimentary components: Journal of Paleolimnology, v. 29, p. 141–154.

SENGUPTA, M.E., KRISTENSEN, T.K., MADSEN, H., and JØRGENSEN, A., 2009, Molecular phylogenetic investigations of the Viviparidae (Gastropoda: Caenogastropoda) in the lakes of the Rift Valley area of Africa: Molecular Phylogenetics and Evolution, v. 52, p. 797–805.

SOREGHAN, M.J., and COHEN, A.S., 1996, Textural and compositional variability across littoral segments of Lake Tanganyika: The effect of asymmetric basin structure on sedimentation in large rift lakes. AAPG Bulletin, v. 80, p. 382–409.

SOREGHAN, M.J., SCHOLZ, C.A., and WELLS, J.T., 1999, Coarse-grained deep-water sedimentation along a border fault margin of Lake Malawi, Africa: seismic stratigraphic analysis: Journal of Sedimentary Research, v. 69, p. 832–846. 137

TALBOT, M. R., 1988, The origins of lacustrine source rocks: evidence from the lakes of tropical Africa, in Fleet, A. J., Kelts, K., and Talbot, M.R., eds., Lacustrine Petroleum Source Rocks: Geological Society Special Publication 40, London, p. 29–43.

TIERCELIN, J.-J., SOREGHAN, M.J., COHEN, A.S., LEZZAR, K.E., and BOUROULLEC, J.-L., 1992, Sedimentation in large rift lakes: example from the middle Pleistocene– modern deposits of the Tanganyika trough, East African Rift System: Bulletin des Centres de Recherches Exploration-Production Elf Aquitaine, v. 16, p. 83–111.

TIERCELIN, J.-J., PFLUMIO, C., CASTREC, M., BOULEGUE, J., GENTE, P., ROLET, J., COUSSEMENT, C., STETTER, K.O., HUBER, R., BUKU, S., and MIFUNDU, W., 1993, Hydrothermal vents in Lake Tanganyika, East African Rift system: Geology, v. 21, p. 499–502.

VAN DAMME, D., and PICKFORD, M., 1999, The late Cenozoic Viviparidae (Mollusca, Gastropoda) of the Albertine Rift Valley (Uganda-Congo): Hydrobiologia, v. 390, p. 171–217.

VAN WAGGONER, J.C., MITCHUM, R.M., CAMPION, K.M., and RAHMANIAN, V.D., 1990, Siliciclastic sequence stratigraphy in well logs, cores, and outcrops: AAPG Methods in Exploration Series 7, 55 p.

WEIER, T., 2005, Shedding light on the cryptic underbelly of Lake Tanganyika’s benthos: New environmental indicators and potential paleoenvironmental indicators among sponges. Nyanza Project Report Annual Report Online, Checked November 2009.

WENINGER, B., JÖRIS, O., AND DANZEGLOCKE, U., 2008, CalPal-2007: Cologne Radiocarbon Calibration and Palaeoclimate Research Package. http://www.calpal.de/, accessed 2008-10-01.

WEST, K., and COHEN, A.S., 1996, Shell microstructure of gastropods from Lake Tanganyika, Africa: adaptation, convergent evolution and escalation: Evolution, v. 50, p. 672– 681.

WEST, K., MICHEL, E., TODD, J.A., BROWN, D., and CLABAUGH, J., 2003, The gastropods of Lake Tanganyika: Diagnostic key, classification and notes on the 138

fauna: Occasional Publications, International Association of Theoretical and Applied Limnology [Societas Internationalis Limnologiae, SIL]), no. 2, 130 p., 101 figs.

WILLIAMSON, P.G., 1981, Palaeontological documentation of speciation in Cenozoic molluscs from Turkana Basin: Nature, v. 293, p. 437–443.

WILSON, J. L., 1975, Carbonate Facies in Geologic Time: Springer-Verlag, New York, 471 p.

ZUSCHIN, M., STACHOWITSCH, M., and STANTON, R.J., 2000, Patterns and processes of shell fragmentation in modern and ancient marine environments: Earth- Science Reviews, v. 63, p. 33–82.

139

Figures

Figure B.1: Overview map of Lake Tanganyika, tropical East Africa. The study area (outlined in box) is located near the town of Kigoma, Tanzania. Known shell beds are combined from the authors’ research experiences and records from Leloup (1953). Allochemical sandstone samples (see Fig. 5 and text for details) were collected from paleo-highstand deposits on the Ruzizi River plains (dot at top). Triangle in study area is the location of core T97-1VC (see Fig. 8).

140

Figure B.2: Study area enlarged from box in Figure 1. Symbols used for sites are the same as those used for ternary taphograms (Fig. 7). Bathymetric profiles A, B, and C illustrate the difference in lake-floor gradient across the littoral zone. Pale gray areas denote approximate known extents of Neothauma-rich facies. Asterisks denote locations where samples of well-indurated equivalents of modern unconsolidated facies were discovered. w.d. = water depth.

141

Figure B.3: Examples of Neothauma and different damage states. A) Undamaged, dead. B) Heavy mechanical fragmentation damage (score = 2). C) Heavy abrasion damage (score =2) typical of samples encountered in beach ridges. D) Shell sample from the headland site with heavy stromatolitic encrustations (score = 2). E) Oxidation patina on sample collected along the 20 m isobath on the Luiche River deltaic platform (score = 2).

142

Figure B.4: Field photographs of representative facies. Neothauma shell collections used in biostratinomic analyses were made from the upper 10 cm of these lake floor substrates. A) Slope-front gravelly mollusk hash from embayment sites. B) Gravelly mollusk hash from the headland site. C) Beach ridges composed of sandy gastropod hash. D) Sand and silt-rich mollusk hash from ~8 m below lake surface on the deltaic platform. Note cratered appearance due to fish bioturbation. E) Pure gastropod hash found along the 20 m isobath on the deltaic platform. Branching sponges are common in this locale and appear to utilize shell beds exclusively as their substrate.

143

Figure B.5: Photomicrographs of representative facies under plane light (left images: A, C, E, G) and cathodoluminescence (CL; right images: B, D, F, H). Primary aragonite produces apple-green CL, whereas biogenic calcite ranges from pink to red. Quartz and quartz-rich sediments commonly exhibit blue or purple CL. A-B) Bioclastic hybrid arenite sample from an embayment site. C-D) Gastropod-rich boundstone sample from headland site. E-F) Allochemical sandstone sample from paleo-foreshore deposits, Ruzizi plains. G-H) Gastropod grainstone from ~20 m below the lake surface on the Luiche River deltaic platform. Scale bar = ~3 mm. Nt = fragments of Neothauma shell.

144

Figure B.6: High-frequency damage threshold diagrams for all sites evaluated in this study. Error bars represent 95% confidence limits for the data. See text for details.

145

Figure B.7: Ternary taphograms for all sites evaluated in this study. Symbols representing sites are the same as in Figure 2. Dark symbols signify shell collections made in water <10 m deep and light symbols those from deeper (>10 m) water. A) Fragmentation. B) Abrasion. C) Encrustation. D) Oxidation patina.

146

Figure B.8: Stratigraphic log of vibracore T97-1VC, collected at 26 m water depth north of the Malagarasi River delta (modified after Parson, 2001). Note the presence of a Neothauma shell lag near the top of the core, capping a paleodelta complex and a muddy transgressive unit. Neothauma shells from this lag are heavily fragmented, consistent with erosion through wave ravinement.

147

Figure B.9: Summary diagram of taphonomic processes and Neothauma-rich carbonates near Kigoma, Tanzania. Gravelly mollusk hash beds developing along high gradient lake floors are commonly juxtaposed with stromatolite reefs, helping to form a potential reservoir facies. Shell hash beds on the deltaic platform support abundant benthic life, especially sponges, crabs and fish along the ~20 m isobath.

148

Tables

Table B.1: Description of sites studied on Lake Tanganyika’s littoral margin near Kigoma, Tanzania. Sample size includes the number of Neothauma shells scored for taphonomic variables.

Site name Environment Water Lake-floor Sample depth gradient size

HT-1-05 Bay; base of slope 17–22 m 8% 148

KB-1-05 Bay; base of slope 13–21 m 11% 150

MB-1-05 Wave-exposed 12–15 m 16% 150 headland

UB-1-05 Platform beach Swash zone <3% 168

UB-2-05 Platform beach Swash zone <3% 151

NLP-1S- Deltaic platform 8 m <3% 152 06

NLP-1D- Deltaic platform 20 m <3% 168 06

NLP-2S- Deltaic platform 8 m <3% 150 06

NLP-2D- Deltaic platform 20 m <3% 160 06

ULB-1-05 Bay on deltaic 10 m <3% 150 platform

149

Table B.2: Radiocarbon data from Neothauma shells collected near Kigoma, Tanzania. Ages are adjusted for Lake Tanganyika's old carbon reservoir using a correction curve developed in Felton et al. (2007). Shells from the headland and embayment sites (HT-1-05, MB-1-05) are centered on the late 19th century, whereas samples from the delta region are time averaged over ~1600 calendar years.

Reservoir- Calibrated 2 sigma

14 Lab C Error corrected age range Calendar accession Site age 14C 14C age (yrs BP) (yrs BP) years CE number yrs

AA84619 HT-1-05 1045 45 110 139 ± 100 38–239 1811 ± 100

AA84620 HT-1-05 1078 44 148 143 ± 105 37–248 1807 ± 105

AA84621 MB-1-05 1043 36 110 141 ± 98 42–239 1809 ± 98

AA84622 MB-1-05 1092 36 167 147 ± 115 32–262 1803 ± 115

AA84623 NLP-1D-06 2296 44 1686 1613 ± 60 1553– 337 ± 60 1673

AA84624 NLP-2D-06 986 35 36 modern modern modern

AA84625 NLP-1S-06 1045 36 110 141 ± 98 42–239 1809 ± 98

AA84626 UB-1-05 1768 36 1023 946 ± 21 925–967 1004 ± 21

AA4772a Malagarasi 1260 70 375 415 ± 73 341–488 1535 ± 73 Delta a Data point from Cohen et al. (1997)

150

Table B.3: Taphonomic variables and scoring system used in this study. See text for details.

Variable No damage Low damage High damage Remarks (0) (1) (2)

Fragmentation None Minor chips Large angular Environmental to apertural fragments energy indicator lip

Abrasion None; Dull luster Heavy pitting, Environmental original chalky, eroded energy indicator luster

Encrustation None Covering Covering >10% of Coverage by <10% of shell shell encrusting organisms

Oxidation patina None; Covering Covering >10% of Proxy for lake- original <10% shell bottom exposure color (white- of shell cream-gray)

151

Table B.4: Site-by-site comparison of taphonomic data using 95% confidence intervals. In each cell, the results are listed (in order) as a string for the four test variables: fragmentation, abrasion, encrustation, and oxidation patinas. n = statistical similarity at the 95% level; * = statistically different damage.

Site KB-1-05 MB-1- UB-1- UB-2- NLP-1S- NLP-2S- NLP-1D- NLP-2D- ULB-1- 05 05 05 06 06 06 06 05

HT-1-05 nnn* nnn* n*** n**n n**n **** *n** *nn* *n**

KB-1-05 nnn* ***n **** **** **** *n** nnn* nn**

MB-1-05 n*** **** **** ***n *n** *nnn *n**

UB-1-05 n*n* n*n* n*** **** n*** ****

UB-2-05 nnnn nn** **** n*** ****

NLP-1S-06 n*** **** **** ****

NLP-2S-06 **** ***n ****

NLP-1D-06 *nn* nn*n

NLP-2D-06 *n**

152

APPENDIX C:

PLAYA-LAKE SEDIMENTATION AND ORGANIC MATTER ACCUMULATION IN AN ANDEAN PIGGYBACK BASIN: THE RECENT RECORD FROM THE CUENCA DE POZUELOS, NW ARGENTINA

Michael M. McGluea, Geoffrey S. Ellisb, Andrew S. Cohena, Peter W. Swarzenskic

aDept. of Geosciences, The University of Arizona, Tucson AZ bEnergy Resources Program, U. S. Geological Survey, Denver CO cPacific Science Center, U. S. Geological Survey, Santa Cruz CA

Pending submission to Sedimentology

153

Abstract

Expansive playa-lake systems situated in high-altitude piggyback basins are important and conspicuous components of both modern and ancient cordilleran orogenic systems. Extant playa lakes provide vital habitat for numerous endemic species, whereas sediments from these deposystems may record signals of climate change or develop into natural resources over geologic time. The Cuenca de Pozuelos, a piggyback basin in NW Argentina, provides the opportunity for an actualistic assessment of playa-lake sedimentation and organic matter accumulation in the Andean Puna. The locus of deposition is Laguna de los Pozuelos, a shallow, mildly alkaline playa-lake fed axially by small intermittent rivers and laterally by ephemeral alluvial-fan complexes. Bulk sediment and organic mass accumulation rates in the playa-lake center approach

0.22 g cm − 2 y − 1 and 2.89 mg cm − 2 y − 1 respectively, indicating moderately rapid deposition with negligible deflation over historic time. Surficial facies maps reveal axial concentrations of total organic carbon (up to 3.3 wt %), whereas playa margins exhibit higher percentages of carbonate minerals. The concentration of biogenic silica reaches maximum values near deltas and represents areas of enhanced diatom productivity and waterbird habitat. Elemental and stable isotopic analyses point to a mixed organic matter composition in the playa-lake 154

center, with a substantial contribution from algae and C4-pathway aquatic macrophytes. Preservation of organic matter appears to be highest in the playa- lake center, likely related to the perennial persistence of the lake water column.

Hydrogen index values reflect terrestrial contributions and oxidation at the playa margins and deltas, the latter primarily a function of seasonal desiccation and heavy avian bioturbation. Modern analog data from Laguna de los Pozuelos provide key insights for: (1) environmental reconstructions of ancient lake sequences, and (2) organic facies models for intracontinental piggyback basins.

Introduction

Cordilleran orogenic systems, typified by eastward subduction of oceanic crust and the development of contractional mountain belts, produce a number of different basin types in which thick packages of sediment have accumulated (e.g.

Jordan, 1995; DeCelles & Giles, 1996). Whereas continental scale, partitioned foreland basins typify lowland retro-arc positions, smaller thrust fault-bounded basins commonly develop high atop the orogenic wedge itself (‚piggyback basins‛; Ori & Friend, 1984; Beer et al., 1990; Talling et al., 1995; Horton &

DeCelles, 1997; Sobel et al., 2003; Zaleha, 2006). The environmental conditions that prevail in these small, high-altitude basins are typically extreme, with low effective precipitation and wide annual temperature fluctuations. Yet, a number 155

of striking terrestrial ecosystems mark these basins, including large lakes and wetland complexes. In the modern Andes, for example, playa-lakes and their associated environments provide essential habitat and food resources for a variety of endemic fauna, including camelids (e.g. Vicugna vicugna), flamingos (e.g.

Phoenicoparrus andinus, Phoenicoparrus jamesi) and shorebirds (e.g.

Recurvirostra andina, Fulica cornuta). Additionally, playa-lake deposits are of considerable interest to geologists and paleolimnologists (Rosen, 1994).

Throughout the Americas alone, playa-lake sediments have been used: (1) to reconstruct local and regional paleohydrology (e.g. Eugster & Hardie, 1975;

Eugster & Hardie, 1978; Dargam & Depetris, 1996; González & Maidana, 1998);

(2) to investigate ancient environments and their peoples (Markgraf et al., 1984;

Waters, 1989; Castiglia & Fawcett, 2006; Pietras & Carroll, 2006); and (3) for economic purposes, such as in the development of trona and oil shale (e.g. Dyni,

2006).

The purpose of this study is to explore organic sedimentation in the

Cuenca de Pozuelos, an internally-drained piggyback basin in the Jujuy Province of northwest Argentina (Fig. 1). The most striking hydrologic feature of the basin is Laguna de los Pozuelos (LP), a shallow, polymictic playa-lake situated near the center of the basin. Geological studies focusing on late Quaternary aspects of LP 156

are absent (see Gangui, 1998, for one notable exception). This knowledge gap on

LP is remarkable, especially considering: (1) the lake’s international reputation as an environmentally-sensitive wintering site for Andean and Nearctic waterbirds

(RAMSAR, 2009); and (2) the growing recognition of the importance of small lakes to the global carbon budget (Dean & Gorham, 1998; Duarte et al., 2008). As one of the most expansive lacustrine ecosystems in the Puna, LP provides a unique opportunity to investigate the patterns, provenance, and preservation of organic sedimentation in a high-altitude piggyback basin, where climatic and tectonic conditions are relatively well constrained. Our focus on recent sedimentary organic matter in LP was driven by the need for a better understanding of the carbon cycle in playa lakes situated on the Andean Puna. A number of active debates center on the late Quaternary climate history of the high Andes, and modern calibration datasets are vital for meaningful paleoenvironmental reconstructions over longer time scales (e.g. Baker et al.,

2001; Plackzek et al., 2006; Fritz et al., 2007; Quade et al., 2008). In addition, our results have implications for improving facies models of some ancient intracontinental foreland basins.

Regional Overview 157

The study site occupies the floor of the Cuenca de Pozuelos (~3650 m above sea level), a NNE-SSW-oriented piggyback basin in the Puna morphotectonic province of the Andean cordillera (Allmendinger et al., 1997;

Gangui, 1998; Fig. 1). The Cuenca de Pozuelos is ~ 110 km long and ~ 25 km wide. Westward-verging thrust sheets carrying Ordovician marine siliciclastic and volcanic rocks bound the flat-floored basin. Small exposures of Cretaceous continental sediments crop out near the southern end of the basin, whereas

Neogene ignimbrites are widespread along the eastern flank of the basin (Page,

1996). Relief between the basin floor and the crests of marginal mountain ranges exceeds 450 m (Fig. 2).

Laguna de los Pozuelos is a large, seasonally fluctuating playa-lake.

Satellite images and field observations indicate that LP’s surface area can exceed

135 km2 during years with above average precipitation (e.g. Mirande & Tracanna,

2009). Conversely, intervals of prolonged drought commonly lead to desiccation of the lake and complete exposure of the playa floor. Bathymetric variability is low due to the negligible basin-floor gradient. When completely filled, LP reaches a maximum depth of < 2 m near its center and progressively shoals towards its margins. Lake waters are brackish (4-29 ppt) and slightly basic (mean pH of 8.65), based on two field seasons of measurements. Lake water chemistry 158

appears to be spatially homogenous due to wind mixing. High concentrations of

Na+ and Ca2+ (553 and 129 mg/L, respectively) characterize LP’s cation

- - composition, and Cl and HCO3 dominate the anions (761 and 198 mg/L).

Shallow aquifers provide the most important source of water to LP (Igarzabal,

1978). In addition to groundwater, the playa-lake is fed by three axial, intermittent river systems: (1) the Rio Santa Catalina, which enters from the north and drains the Sierras de Rinconada to the northwest; (2) the Rio Cincel, which enters from the southeast and drains the Sierras de Quichagua to the south; and (3) the Rio Chico, which enters from the southwest and drains the southern extent of the Sierras de Rinconada (Fig. 1). Numerous ephemeral streams emerge from adjacent highlands and build alluvial fans laterally along both sides of the playa-lake (Fig. 2).

Total precipitation in the Cuenca de Pozuelos approaches 320 mm/year and annual air temperatures range between 3-13°C (Legates & Willmott, 1990a;

Legates & Willmott, 1990b). Climate in the region is nominally controlled by the South American Summer Monsoon system (e.g. Zhou & Lau, 1998; Garreaud et al., 2009). Rainfall is strongly seasonal, with ~ 70% of the yearly total falling during the austral summer. As with many regions on the Andean plateau, the El

Niño-Southern Oscillation (ENSO) phenomenon modulates patterns of 159

precipitation over the Cuenca de Pozuelos. This is especially true during cold- phase La Niña intervals, which bring enhanced total rainfall and higher lake levels to LP. Southeasterly summer winds are generally weak, whereas westerly winter winds are substantially stronger. Due to its persistence through many dry seasons, LP provides important habitat for numerous waterbirds, including all three species of New World flamingos: Phoenicopterus chilensis, Phoenicoparrus andinus and Phoenicoparrus jamesi (e.g. Caziani et al., 2001; Mascitti, 2001;

Moschione & Sureda, 2008).

A vegetation survey conducted in the basin indicates that plant assemblages generally follow topographic and soil-moisture gradients

(Bonaventura et al., 1995). C3- pathway shrubs including Fabiana densa and

Baccharis boliviensis are abundant on alluvial slopes, with minor CAM pathway succulents of varying species (including Maihueniopsis glomerata and Opuntia soehrensii) interspersed (Fig. 3). Grasses, including Stipa ichu, Distichlis spicata, and Festuca sp., carpet the basin floor. Within LP, meadows of Ruppia sp. are ubiquitous in shallow water around the southern river deltas (Fig. 3).

Methods and Materials

Surface sediments (n = 79; Table 1) were collected from the floor of the

Cuenca de Pozuelos and LP using a standard scoop sampler. The strategy for 160

sediment retrieval focused on sampling each of the basin’s different depositional environments, including the playa-lake, playa margins (wet and dry mud flats or lateral distal alluvial fans) and axial deltas. Sediment smear slides were inspected under a petrographic microscope to estimate major mineralogical and biogenic components. A sediment core (LP07-1A) was collected near the playa-lake center during the dry season of 2007, when water level was much reduced (Fig. 2).

Sediment geochronologies, in terms of both mass accumulation (g cm - 2 y - 1) and linear sedimentation rates (cm y- 1), were derived from this core using multiple

210 137 radioactive tracers, including excess Pb (t1/2 = 22.3 yr) and Cs (t1/2 = 30.1 yr) at the USGS Pacific Science Center in Santa Cruz, CA, following the methodology presented in Swarzenski et al. (2006).

Elemental and stable isotopic analyses of sedimentary organic matter

(OM) were conducted at the University of Arizona. Total organic carbon (TOC),

13 total nitrogen, and δ Com were measured on a continuous-flow gas-ratio mass spectrometer (Finnigan Delta PlusXL) coupled to a Costech elemental analyzer.

Samples were combusted in the elemental analyzer. Standardization is based on

13 acetanilide for elemental concentration, NBS-22 and USGS-24 for δ Com.

13 Precision is better than ± 0.09 for δ Com based on repeated internal standards.

Atomic C:N ratios presented in Table 1 were corrected for contributions of 161

inorganic nitrogen following the procedure outlined by Talbot (2001). Select decalcified samples from across LP were analyzed by Rock-Eval pyrolysis at the

University of Houston in order to further discriminate the provenance of OM

(Table 2; Espitalie et al., 1977; Katz, 1983).

Particle-size analyses conducted at the University of Arizona utilized a

Malvern laser-diffraction particle size analyzer coupled to a Hydro 2000S sample dispersion bench. Sample pre-treatment followed the recommendations of

Johnson & McCave (2008) for the isolation of the terrigenous fraction of lake sediments. Samples for particle-size analysis were also treated at 80°C with 1M

HCl for 24 hours to remove carbonate minerals. Following completion of the digestions, samples were placed on a mechanical shaker in a dilute solution of sodium hexametaphosphate for several hours to disaggregate clays. Prior to analysis, sediments were examined optically to ensure that all the organic matter, biogenic silica and carbonate had been removed.

Total inorganic carbon (TIC) and biogenic silica (BiSi) analyses were conducted at the Limnological Research Center at the University of Minnesota.

Weight percent of total inorganic carbon per sample was determined using a UIC

Inc. total carbon coulometer. Analytical precision associated with this technique was typically better than 0.2%. Biogenic silica analyses utilized multiple 162

extractions of hot alkaline digestions following a modified protocol designed by

DeMaster (1979). Reported values have an analytical precision of ~ 1.0%.

Results and Interpretations

Table 1 lists the results of sedimentological and geochemical analyses for all surface samples in the study area. Figure 4A illustrates the spatial distribution of particle sizes across the basin floor. Sediment particle sizes are in the silt and clay fractions for most of LP (mean = ~26 µm) and probably reflect a combination of suspension rainout and aeolian deposition. Coarse silt and sand-sized particles

(> 40 µm) are largely restricted to deltaic channels and plains along the axial margins of the basin. At present, the Rio Cincel builds a coarse sandy fan near its mouth, whereas sand-sized particles are less extensive from the other deltas.

Along LP’s western margin, restricted pockets of coarse silt and fine sand correlate with distal alluvial fan surfaces (Fig. 4A).

Values of BiSi range between 0.5 and 3.9 wt %, with a mean of 2.0 wt %.

Biogenic silica concentrations are highest along LP’s axial margins, with samples exceeding 2.5 wt % near the Rio Santa Catalina and Rio Chico deltas (Fig. 4B).

Intermittent mats of benthic diatoms were encountered offshore of the Rio Chico delta. Moderate BiSi concentrations characterize the playa-lake center, whereas

LP’s margins, typified by wet and dry mud flats, generally exhibit low (< 1.0 wt %) 163

values (Fig. 4B). Phytoplankton, particularly the diatoms Cocconeis placentula,

Nitzschia hungarica, and Navicula sp., represent the dominant source of biogenic silica within LP (Maidana et al., 1998).

Concentrations of TIC follow a much different pattern than BiSi, with maximum values (> 1.0 wt %) in mud-flat environments along LP’s northwestern margin (Fig. 4C). Across the study area, values of TIC range between 0.01 and 1.6 wt %, with a mean of 0.4 wt %. Smear slide analysis indicates that LP’s TIC is composed of biogenic carbonate (ostracodes and Chara stems) and fine authigenic calcite. At present, elevated TIC concentrations are associated exclusively with LP’s margins, where standing water evaporates rapidly and efflorescent crusts develop on mud-cracked plains in the austral winter. In contrast, in the basin center and near the deltas, values of TIC are generally below 0.4 wt % (Fig. 4C).

Values of TOC range between 0.1 and 3.3 wt %, with a mean of 1.4 wt %

(Fig. 4D). Samples exhibiting TOC values exceeding 1.5 wt % were collected from all environments, but they are always fine grained (< 20 µm; Fig. 5).

Organic carbon concentrations are richest along the axis of the playa-lake, < 2 km from distal reaches of the Rio Santa Catalina delta (Fig. 4). Contours illustrate elevated values of TOC (> 1.6 wt %) in the playa-lake center, whereas samples 164

from the southern deltas and playa margins routinely exhibit TOC values <1.0 wt

% (Fig. 4D). Samples from LP typically contain low total nitrogen (Table 1 and

Fig. 6A). C:N values range from 2-79 in the study area, with a mean of ~ 22 (Fig.

6B). Samples collected north of the Rio Santa Catalina delta and within the mouth of the Rio Cincel delta exhibit low total nitrogen, leading to C:N values greater than 30. Most of the playa-lake sediments have C:N values < 20, which is typical for lakes with mixtures of algal and higher plant matter (e.g. Meyers &

Teranes, 2001). An exception to this pattern is near the Rio Chico delta, where values of total nitrogen substantially increase and C:N values fall to < 5.

Flamingo activity is high in this locale, probably due to the availability of food resources and favorable water depths (M. McGlue, personal observations, 2006-

08). Seasonal foraging likely impacts both the carbon and nitrogen cycles across similar microhabitats in LP. Peterson & Fry (1987) noted that avian waste products, especially urea, are rich in organic nitrogen and ammonia. Local eutrophication from these waste products may result in phytoplankton blooms, as evidenced by elevated concentrations of BiSi in the distal Rio Chico environment

(Fig. 4B).

Carbon stable isotope data range from -16.4 to -24.6 ‰, with a mean of -

20.2‰ (Table 1). Sedimentary organic matter derived exclusively from 165

lacustrine phytoplankton typically exhibits lighter isotopic values (-25.0 to -

13 30.0‰; Meyers & Ishiwatari, 1993).  Com values from our samples indicate a substantial contribution of isotopically heavy OM to the playa-lake environment

(Fig. 6B). Rock Eval pyrolysis data confirm a mixed organic-matter provenance for LP surface sediments (Table 2). Hydrogen index (HI) values range from 74 –

190 mg HC/gm TOC. Hydrogen index values increase with distance from the playa margins, and reach maximum values in the playa-lake center (Fig. 7).

Samples with values <100 mg HC/gm TOC reflect higher contributions of oxidized terrestrial vegetation, consistent with their proximity to the deltas and playa margins (Talbot & Livingstone, 1989). Taken together, carbon isotope and

HI data indicate that sedimentary OM in LP is derived from both autochthonous and allochthonous sources, including: (1) annual algal blooms and planktonic

primary productivity, using the C3 photosynthetic pathway; (2) perennial C4- pathway aquatic macrophytes, especially Ruppia sp. meadows situated in shallow

(< 20 cm) water depths, and (3) watershed vegetation, dominated by C3-pathway steppe shrubs and grasses with minor contributions from succulents (CAM pathway). Published carbon isotope data on Andean steppe taxa are rare, but

Quade et al. (2007) measured values exceeding -21.0‰ on Fabiana and Baccharis specimens at elevations similar to those in the Cuenca de Pozuelos. 166

Figure 8 illustrates sedimentation and mass accumulation rates inferred from core LP07-1A, collected from the playa-lake center. Encompassing the upper ~ 14 cm of brown (5YR 3/4) silty clay in core LP07-1A, decay of excess Pb-

210 implies sedimentation with negligible deflation at a mean rate of 0.14 cm y- 1, consistent with many lake records over periods of short duration (e.g. Sadler,

1981). Cesium-137 data are consistent with the Pb-210 results and show a pronounced peak related to atmospheric weapons testing in 1963 (Fig. 8). Carbon

13 and nitrogen concentrations increase towards the top of the core, and  Com trends toward more positive values. These geochemical patterns indicate a relative rise in water level at the core site. Bulk sediment and organic-carbon mass accumulation rates at the core site average 0.22 g cm-2 y-1 and 2.89 mg cm-2 y-1, respectively. These rates of organic carbon accumulation are broadly comparable to larger endorheic basins, such as Lake Titicaca and lowland

Andean foreland basin lakes during regressive intervals (Binford et al., 1992;

Piovano et al., 2002). Mass accumulation rates of organic carbon in LP over the historic period exceed those of smaller, high-altitude basins in the Andes (Abbott et al., 1997).

Discussion 167

As with many tectonically active lake systems, patterns of OM accumulation in LP appear to be linked with depositional environment and regional hydrology (e.g. Huc et al., 1990; Fig. 9). Particle size is inversely correlated with carbon richness in LP, suggesting dilution of OM by terrigenous sediments. Figure 4A illustrates this relation well, especially near the Rio Cincel delta and along playa margins influenced by distal alluvial-fan sedimentation.

Allochthonous OM prevails in these environments, exhibiting characteristically

13 depleted  Com and C:N values exceeding 25, indicating OM that is derived from steppe vegetation covering the adjacent highlands and from basin-floor grasses

(Fig. 6). In distal deltaic and central playa-lake environments, TOC, HI, and BiSi values are higher, whereas C:N values typically decline, suggesting greater contributions from algae to the OM pool (Fig. 9). Additionally, the playa-lake center and distal deltaic plains are fine-grained environments, which promote better preservation of autochthonous OM as interstitial oxygen is limited (Fig.

4A; Gray, 1981). At playa margins, TOC richness declines and the concentration of carbonate minerals increases substantially, indicating auto-dilution in this environment.

Key factors affecting the development of autochthonous OM in LP include the availability of nutrients, rates of primary productivity, and basin hydrology. 168

Nutrient loading has not been well studied at LP, but important contributions likely originate from: (1) groundwater and river-borne solutes, (2) imported N from migratory waterbird wastes, and (3) mineral aerosols. Surficial sediment

13 geochemistry, particularly δ Com, indicates that the OM pool in the playa-lake center is likely comprised of a mixture of algae and aquatic macrophytes (Fig. 6).

Ecological studies of LP demonstrate that its algal biomass, dominated by chlorophytes (e.g. Chlamydomonas tremulans) and diatoms, exceeds that of other high-altitude wetlands in the region (Mirande & Tracanna, 2009). The preservation of this biomass, however, is limited by LP’s mixing regime and shallow bathymetry. Prevailing winds mix LP throughout the year, as ice cover is commonly restricted to shoreline and marginal ribbons in the austral winter.

Low molecular weight organic components are especially labile in polymictic lakes, as wind mixing rapidly resuspends OM following sedimentation, subjecting it to further degradation by aerobic respiration. Similar processes in marine environments are known to account for substantial diagenetic losses of OM (e.g.

Henrichs, 1992). Seasonal desiccation produces a similar effect and serves to reduce TOC richness along LP’s margins (Fig. 4D). Wind mixing is also likely responsible for the transport of isotopically enriched plant matter from shallow margins and deltas into deeper environments (Fig. 9). Alternatively, enriched 169

carbon isotope values could reflect OM degradation (Talbot & Livingstone, 1989).

However, selective loss of 12C during oxidative fractionation typically leads to a 1-

13 2‰ positive shift in  Com, which alone cannot explain the mean values observed in LP (Fig. 6). As a result, a mixed provenance for OM in the playa-lake environment appears most likely.

As nitrogen is a key limiting nutrient in many subtropical and arid lake systems, the impact of waterbirds on biogeochemical cycles at LP is important

(e.g. Cowell & Dawes, 1991; Lebo et al., 1992). Similar to many high-altitude

Andean lakes, flamingoes are abundant at LP in the austral summer. However,

LP is unique in that large flocks of waterbirds persist on site throughout the austral winter as well (Caziani et al., 2001; Mascitti, 2001). The concentration of mineral nitrogen in LP waters is below the detection limit of standard ion chromatography analysis, thus indicating that allochthonous organic nitrogen delivery may be critical for phytoplankton productivity. As a result, organic facies development may be linked to year-round nitrogen delivery and algal fertilization from avian waste (Fig. 9).

A particularly intriguing aspect of the carbon cycle at LP is the role of waterbirds in bioturbation. Hurlbert & Chang (1983) were among the first to note the impact of flamingoes on lacustrine microbenthos in the Andes. 170

Flamingoes feed by filtering algae and zooplankton through their bills – a process that requires inverting their heads and stirring up lake-bottom sediment.

Data on foraging microhabitats for New World flamingoes suggests a preference for playa margin and shallow playa-lake environments (Mascitti, 2001). At LP, flamingo bioturbation in these environments reduces the preservation potential of OM due to prolonged persistence in near-surface, well-oxygenated sediments.

A number of authors have posited a causal link between bioturbation and declines in both sedimentary TOC and HI values (Pratt, 1984; Ariztegui et al.,

1996). Data from LP supports this link, as Rock Eval data indicate that only samples in the playa-lake center have hydrocarbon-source potential (S1+S2 > 2.0; after Tissot & Welte, 1978). Water depths in the playa-lake center routinely exceed ~ 40 cm, thus limiting bioturbation by wading birds (e.g. Mascitti, 2001).

Eocene Fossil Basin, a piggyback basin in the Western Cordillera of North

America, shares similarities with the Cuenca de Pozuelos, including basin morphometry and tectonic origin. The geologic record of Fossil Basin, noted for its kerogen-rich lacustrine deposits, exhibits facies patterns that resemble LP, such as basin-centered concentrations of TOC and massive, carbonate-rich, TOC- poor sediments along margins (Buchheim & Eugster, 1998). Leggitt et al. (1998) noted the intimate association of waterbirds and lake environments in Fossil 171

Basin. Although under-reported in the literature (see Genise et al., 2009 for an exception), the results of our study indicate that foraging behavior of gregarious waterbirds likely influences patterns of OM preservation in piggyback lake systems.

The fundamental requirements for the evolution of a playa-lake system include hydrologic closure and climatic conditions where evaporation exceeds precipitation on a yearly basis (e.g. Smoot & Lowenstein, 1991; Rosen, 1994). In piggyback basins, thrust faulting creates hangingwall topographic barriers that focus fluvial and groundwater flow paths toward footwall flats, where hydrologic closure, playa-lake environments and in some cases, basin-overfilling conditions prevail (Carroll & Bohacs, 1999; Garcia-Castellanos et al., 2003). As such, processes such as sediment starvation do not likely influence the accumulation of

OM in piggyback basins. Over short time scales (101 – 104 yrs), we suggest that climate exerts a primary control on both hydrology and organic sedimentation in piggyback-basin playa lakes. The persistence of a high-altitude playa-lake in the

Andes necessitates enough precipitation to maintain the local groundwater system through the austral winter. Drought-related depression of the groundwater table exposes playa-lake systems to aeolian re-working and removal

(Rosen, 1994). Consequently, OM preservation in LP hinges on maintenance of 172

the water table elevation at or near the ground surface. Excess 210Pb data demonstrate that over the last ~ 100 years, deflation has not impacted LP at the coring site in a significant manner. In fact, elemental and stable isotopic profiles indicate rising water levels over this period, because concentrations of TOC rise

13 and  Com values become more enriched (Fig. 8). Eugster & Hardie (1975) surmised that seasonal oxygen minima in strongly-evaporative playa lakes may occur following seasons with heavy precipitation, as freshwater lenses overlie remnant brines, enhancing OM preservation through the development of pycnoclines (i.e. temporal meromixis). Pycnocline formation is probably a highly transient process in modern LP, as soluble evaporite minerals are absent and the sediment color and TOC richness of the 210Pb interval indicate the maintenance of an oxidizing environment. Longer sediment cores encompassing the late

Pleistocene, however, indicate that pycnocline development may contribute to reducing conditions at the lake bottom and greater OM preservation (McGlue et al., 2009).

Conclusions

The results of this integrated assessment of the recent sedimentology and geochemistry in Laguna de los Pozuelos provide novel insights on the processes that affect the deposition and preservation of organic matter in high-altitude, arid 173

piggyback basins. Sediment geochemistry indicates that organic facies development is dominantly controlled by basin hydrology, climate and biological feedbacks (both nutrient cycling and bioturbation) from waterbirds.

1. Modern facies maps show axial deposition of organic carbon, reaching a

maximum value of 3.3 wt %. Playa margin sediments, prone to seasonal

evaporation, are characterized by lower TOC and higher concentrations of

carbonate minerals. At most sample sites, the mean particle size exhibits

an inverse relationship with TOC, suggesting that the environment of

deposition exerts a control on OM accumulation. Sediments rich in

biogenic silica (> 3.0 wt %) are principally along the axial margins of the

basin, near sites where gregarious wading birds are active.

2. Basin hydrology is a fundamental control on organic preservation, as

sediments in the playa-lake center often remain subaqueous even during

the austral winter. Although wind mixing inhibits the preservation of

organic matter, the center of the playa lake escapes additional

resuspension of sediment from flamingo bioturbation because of its depth.

Lead-210 data imply rapid sediment mass accumulation in the playa-lake

center, and minimal deflation during the historic period at the core site. 174

The persistence of the playa lake throughout the dry season enhances OM

preservation, because aeolian erosion is limited.

13 3. Both δ Com and C:N data indicate a mixed provenance for the organic

matter preserved in sediments. The carbon stable isotope data range from

-16.4 to -24.6 ‰, probably in response to substantial contributions from

perennial C4-pathway aquatic plants that have mixed with algae in the

playa-lake center. Samples from the playa margins and those offshore

from the deltas likely contain OM delivered from terrestrial (C3)

vegetation. The C:N data indicate a substantial contribution from higher

plants to the carbon cycle in LP, as well as early diagenetic losses of labile

nitrogen from algae.

4. The Rock-Eval data indicate a spatial relation with respect to the deposition

of organic matter, with HI values below 110 mg HC gm-1 TOC focused

along delta and playa margin mud flats, consistent with our model of

degradation of terrestrial organic matter by oxidation. Samples from the

playa-lake center contain better preserved mixtures of algal and aquatic

plant remains that have only modest hydrocarbon source-potential from

Type II kerogens. 175

5. Waterbird activity strongly impacts the biogeochemical cycles at LP.

Bioturbation of playa margin and shallow playa-lake sediments from

waterbird foraging reduces the amount organic matter preserved in

sediments because of the oxidation that accompanies re-suspension. Waste

products from birds are a valuable source of allochthonous nitrogen to the

lake that fertilizes various autotrophic communities. Data from ancient

analogs imply that such processes are an under-appreciated component of

carbon and nitrogen cycles in high-altitude lakes.

Acknowledgements

This research was sponsored by grants-in-aid of research from AAPG, GSA,

Sigma Xi and the University of Arizona-Chevron Corporation Summer Research fund to MM. The ACS/PRF (45910-AC8) and ExxonMobil Upstream Research also provided financial support. The authors gratefully acknowledge the logistical assistance and scientific expertise of E. Piovano and A. Kirschbaum during their involvement in the project. The staff at Monumento Naturale

Laguna de los Pozuelos provided vital access during our field campaigns. We deeply appreciate the assistance of C. Gans, J. Omarini, F. Cordoba, M.

Barrionuevo and E. Gleeson in the field, and M. Fan, X. Zhang, J. Ash, and M.

Trees in the laboratory. Conversations with S. Ivory, A. Kowler and C. Johnson 176

substantially improved aspects of the text. The staff at the Limnological

Research Center at the University of Minnesota, R. Schow, H. Steblay and M.

Murphy graciously hosted MM while laboratory work for this project was completed.

References

Abbott, M.B., Seltzer, G.O., Kelts, K.R. and Southon, J. (1997) Holocene Paleohydrology of the tropical Andes from lake records. Quatern. Res., 47, 70-80.

Allmendinger, R.W., Jordan, T.E., Kay, S.M. and Isacks, B.L. (1997) The evolution of the Altiplano-Puna Plateau of the central Andes. Annu. Rev. Earth Planet. Sci., 25, 139-174.

Ariztegui, D., Farrimond, P. and McKenzie, J.A. (1996) Compositional variations in sedimentary lacustrine organic matter and their implications for high Alpine Holocene environmental changes: Lake St. Moritz, Switzerland. Org. Geochem., 24, 453-461.

Baker, P.A., Rigsby, C.A., Seltzer, G.O., Fritz, S.C., Lowenstein, T.K., Bacher, N. and Veliz, C. (2001) Tropical climate change at millennial and orbital timescales on the Bolivian Altiplano. Nature, 409, 698-701.

Beer, J.A., Allmendinger, R.W., Figueroa, D.A. and Jordan, T.E. (1990) Seismic stratigraphy of a neogene piggy-back basin, Argentina. AAPG Bull., 74, 1183- 1202.

Binford, M.W., Brenner, M., and Engstrom, D.E. (1992) Temporal sedimentation patterns in the nearshore littoral of Lago Huinaimarca. In: Lake Titicaca: A Synthesis of Limnological Knowledge (Eds C. Dejoux and A. Iltis), 29-37.

Buchheim, H.P. and Eugster, H.P. (1998) Eocene Fossil Lake: The Green River Formation of Fossil Basin, southwestern Wyoming. In: Modern and ancient lake systems: New problems and perspectives (Eds J.K. Pitman and A.R. Carroll), Utah Geological Association Publication 26, 191-208. 177

Carroll, A.R. and Bohacs, K.M. (1999) Stratigraphic classification of ancient lakes: Balancing tectonic and climatic controls. Geology, 27, 99-102.

Castiglia, P.J. and Fawcett, P.J. (2006) Large Holocene lakes and climate change in the Chihuahan Desert. Geology, 34, 113-116.

Caziani, S.M. and Derlindati, E.J. (2000) Abundance and habitat of high Andes flamingos in Northwestern Argentina. Waterbirds, 23, 121-133.

Caziani, S.M., Derlindati, E.J., Talamo, A., Suredam A.L., Trucco, C.E. and Nicolossi, G. (2001) Waterbird richness in Altiplano wetlands of Northwestern Argentina. Waterbirds, 24, 103-117.

Cowell, B.C. and Dawes, C.J. (1991) Nutrient enrichment experiments in three central Florida lakes of different trophic status. Hydrobiologia, 220, 217-231.

Dargam, R.M. and Depetris, J. (1996) Geochemistry of waters and brines from the Salinas Grandes basin, Córdoba, Argentina. II. Gypsum dissolution-calcite precipitation, and brine evolution. Int . J. Salt Lake Res., 5, 81–101.

Dean, W.E. and Gorham, E. (1998) Magnitude and significance of carbon burial in lakes, reservoirs, and peatlands. Geology, 26, 535-538.

DeCelles, P.G. and Giles, K.A. (1996) Foreland basin systems. Basin Res., 8, 105- 123.

DeMaster, D.J. (1979) The Marine Budgets of Silica and Si32. PhD Thesis, Yale University, New Haven, CT, 308 pp.

Duarte, C. M., Prairie, Y. T., Montes, C., Cole, J.J., Striegl, R., Melack, J. and

Downing, J.A. (2008) CO2 emissions from saline lakes: A global estimate of a surprisingly large flux. J. Geophys. Res., 113, 1-7.

Dyni, J.R. (2006) Geology and resources of some world oil-shale deposits: U.S. Geological Survey Scientific Investigations Report 2005-5294, 42 pp.

178

Espitalie, J., Laporte, J.L., Madec, M., Marquis, F., Leplat, , Paulet, J. and Boutefeu, A. (1977) Methode rapide de caracterisation des roches meres, de leur potential petrolier et de leur degre d’evolution. Rev. Inst. Fr. Pétrol., 32, 23-42.

Eugster, H. and Hardie, L.A. (1975) Sedimentation in an ancient playa-lake complex: The Wilkins Peak Member of the Green River Formation of Wyoming. Geol. Soc. Am. Bull., 86, 319-334.

Eugster, H. and Hardie, L.A. (1978) Saline lakes. In Lakes: Chemistry, Geology and Physics (Ed. A. Lerman), Springer-Verlag, New York, 237-293.

Fritz, S.C., Baker, G., Seltzer, G.O., Ballantyne, A., Tapia, M., Cheng, H. and Edwards, R.L. (2007) Quaternary glaciation and hydrologic variation in the South American tropics as reconstructed from the Lake Titicaca drilling project. Quatern. Res., 68, 410-420.

Gangui, A.H. (1998) A combined structural interpretation based on seismic data and 3D gravity modeling in the northern Puna/Eastern Cordillera, Argentina. PhD Thesis, Freien Universität, Berlin, 176 pp.

Garcia-Castellanos, D., Vergès, J., Gaspar-Escribano, J. and Cloetingh, S. (2003) Interplay between tectonics, climate and fluvial transport during the Cenozoic evolution of the Ebro Basin (NE Iberia). J. Geophys. Res., 208, 23-47.

Garreaud, R.D., Vuille, M., Compagnucci, R. and Marengo, J. (2009) Present-day South American climate. Palaeogeogr. Palaeoclimatol. Palaeoecol., 281, 180-195.

Genise, J.F., Melchor, R.N., Archangelsky, M., Bala, L.O., Straneck, R. and de Valais, S. (2009) Application of neoichnological studies to behavioural and taphonomic interpretation of fossil bird-like tracks from lacustrine settings: the Late Triassic-Early Jurassic? Santo Domingo Formation, Argentina. Palaeogeogr. Palaeoclimatol. Palaeoecol., 272, 143-161.

González, M.A. and Maidana, N.I. (1998) Post-Wisconsinian paleoenvironments at Salinas del Bebedero basin, San Luis, Argentina. Journal of Paleolimnology, 20, 353-368.

179

Gray, J.S. (1981) The Ecology of Marine Sediments: An Introduction to the Structure and Function of Benthic Communities. Cambridge, Cambridge University Press, 185 pp.

Henrichs, S.M. (1992) Early diagenesis of organic matter in marine sediments: progress and perplexity. Mar. Chem., 39, 119-149.

Horton, B. K. and DeCelles, P.G. (1997) The modern foreland basin system adjacent to the central Andes. Geology, 25, 895-898.

Huc, A.Y., LeFournier, J., Vandenbroucke, M. and Bessereau, G. (1990) Northern Lake Tanganyika; an example of organic sedimentation in an anoxic rift lake. In: Lacustrine basin exploration; case studies and modern analogs (Ed. B.J. Katz), AAPG Memoir 50: Tulsa, OK, 169-185.

Hurlbert, S. H. and Chang, C.C.Y. (1983) Omitholimnology: effects of grazing by the Andean flamingo (Phoenicoparrus andinus). Proc. Natl Acad. Sci. USA, 801, 4766-4769.

Igarzábal, A.P. (1978) La Laguna de Pozuelos y su Ambiente Salino. Acta Geológica Lilloana (Argentina), 15, 80-103.

Johnson, T.C. and McCave, I.N. (2008) Transport mechanism and paleoclimatic significance of terrigenous silt deposited in varved sediments of an African rift lake. Limnol. Oceanogr., 53, 1622-1632.

Jordan, T.E. (1995) Retroarc foreland and related basins. In: Tectonics of Sedimentary Basins (Eds C. J. Busby and R. V. Ingersoll), Blackwell Science, Cambridge, MA, 331-362.

Katz, B.J. (1983) Limitations of ‚Rock-Eval‛ pyrolysis for typing organic matter. Org. Geochem., 4, 195-199.

Lebo, M.E., Reuter, J.E., Rhodes, C.L. and Goldman, C.R. (1992) Nutrient cycling and productivity in a desert saline lake: observations from a dry, low productivity year. Hydrobiologia, 246, 213-229.

180

Legates, D.R. and Willmott, C.J. (1990a) Mean Seasonal and Spatial Variability in Gauge-Corrected, Global Precipitation. International Journal of Climatology, 10, 111-127.

Legates, D.R. and Willmott, C.J. (1990b) Mean Seasonal and Spatial Variability in Global Surface Air Temperature. Theoretical and Applied Climatology, 41, 11-21.

Leggitt, V.L., Buchheim, H. and Biaggi, R.E. (1998) The Stratigraphic Setting of Three Presbyornis Nesting Sites: Eocene Fossil Lake, Lincoln County, Wyoming. National Park Service Paleontological Research, Volume 3.

Maidana, N., Vigna, M. and Mascitti, V. (1998) Ficoflora de la Laguna de Pozuelos (Jujuy, Argentina), I: Bacillariophyceae. Boletín de la Sociedad Argentina de Botánica, 33, 171-179.

Markgraf, V., Bradbury, J., Forester, R.M., Singh, G. and Sternberg, R.S. (1984) San Agustin Plains, New Mexico: age and paleoenvironmental potential reassessed. Quatern. Res., 22, 336-343.

Mascittti, V. (2001) Habitat Changes in Laguna de Pozuelos, Jujuy, Argentina: Implications for South American Flamingo Populations. Waterbirds, 24, 16-21.

McGlue, M.M., Cohen, A.S., Ellis, G., Swarzenski, P.W., Kirschbaum, A.K., and Lupo, L., (2009) Nonmarine depositional patterns in an arid piggyback basin: Sedimentology, stratigraphy and paleoenvironments from Monumento Naturale Laguna de Los Pozuelos, Argentina: AAPG Annual Meeting Abstracts with Programs.

Meyers, P.A. and Ishiwatari, R. (1993), Lacustrine organic geochemistry; an overview of indicators of organic matter sources and diagenesis in lake sediments. Org. Geochem., 20, 867-900.

Meyers, P.A. and Teranes J. L. (2001) Sediment organic matter. In: Tracking Environmental Change Using Lake Sediments, Volume 2, Physical and Geochemical Methods, (Eds W.M. Last and J.P. Smol), Springer, New York, 239- 269.

181

Mirande, V. and Tracanna, B.C. (2009) Estructura y controles abióticos del fitoplancton en humedales de altura. Ecología Austral, 19, 119-128.

Moschione, F.N. and Sureda, A.L. (2008) Monitoring high-Andes flamingos at Laguna de los Pozuelos National Monument, Argentina: preliminary results. Flamingo, 16, 48-50.

Ori, G.G. and Friend, F. (1984) Sedimentary basins formed and carried piggyback on active thrust sheets. Geology, 12, 475-478.

Page, R. (1996) Mapa Geologico de la Provincia de Jujuy, Republica Argentina, Escala 1:500000. Servicio Geológico Minero de Argentina, Buenos Aires.

Peterson, B.J. and Fry, B. (1987) Stable Isotope in Ecosystem Studies. Annual Reviews in Ecological Systems, 18, 293-320.

Pietras, J.T. and Carroll, A.R. (2006) High-resolution stratigraphy of an underfilled lake basin: Wilkins Peak Member, Eocene Green River Formation, Wyoming, USA. J. Sed. Res., 76, 1197-1214.

Piovano, E.L., Ariztegui, D. and Damatto Moreira, S. (2002) Recent environmental changes in Laguna Mar Chiquita (central Argentina): a sedimentary model for a highly variable saline lake. Sedimentology, 49, 1371-1384.

Placzek, C., Quade, J. and Patchett, J. (2006) Geochronology and stratigraphy of late Pleistocene lake cycles on the southern Bolivian Altiplano: Implications for causes of tropical climate change. Geol. Soc. Am. Bull., 118, 515-532.

Pratt, L.M. (1984) Influence of paleoenvironmental factors on preservation of organic matter in Middle Cretaceous Greenhorn Formation, Pueblo, Colorado. AAPG Bull., 68, 1146-1159.

Quade, J., Rech, J.A., Latorre, C., Betancourt, J.L., Gleason, E., Kalin-Arroyo, M. (2007) Soils at the hyperarid margin: the isotopic composition of soil carbonate from the Atacama Desert. Geochimica et Cosmochima Acta, 71(15), 3772-3795.

182

Quade, J., Rech, J.A., Betancourt, J.L., Latorre, C., Quade, B., Rylander K.A. and Fisher, T. (2008) Paleowetlands and regional climate change in the central Atacama Desert, northern Chile. Quatern. Res., 69, 343-360.

RAMSAR (2009) The Annotated RAMSAR List.

Rosen, M.R. (1994) The importance of groundwater in playas: a review of playa classifications and sedimentology and hydrology of playas. In: Paleoclimate and Basin Evolution of Playa Systems (Ed. M.R. Rosen), Geol. Soc. Am. Spec. Paper 289, 1-18.

Sadler, M. (1981) Sediment accumulation rates and the completeness of stratigraphic sections. J.Geol., 89, 569-584.

Smoot, J.P. and Lowenstein, T.K. (1991) Depositional environments of nonmarine evaporites. In: Evaporites, Petroleum, and Mineral Resources (Ed. J.L. Melvin), Developments in Sedimentology, p. 189-347.

Sobel, E.R., Hilley, G.E. and Strecker, M.R. (2000) Formation of internally drained contractional basins by aridity-limited bedrock incision. J. Geophys. Res., 107, 2-23.

Swarzenski, P.W., Baskaran, M., Orem,W.H. and Rosenbauer, R.J. (2006) Historical reconstruction of contaminant inputs in Mississippi River delta sediments. Estuar. Coast. Mar. Sci., 29, 1094-1107.

Talbot, M.R. (2001) Nitrogen Isotopes in Palaeolimnology. In: Tracking Environmental Change Using Lake Sediments, Volume 2, Physical and Geochemical Methods (Eds W.M. Last and J. Smol), Springer, New York, 401-439.

Talbot, M.R. and Livingstone, D.A. (1989) Hydrogen index and carbon isotopes of lacustrine organic-matter as lake level indicators. Palaeogeogr. Palaeoclimatol. Palaeoecol., 70, 121-137.

Talling, J., Lawton, T.F., Burbank, D.W. and Hobbs, R.S. (1995) Evolution of latest Cretaceous–Eocene nonmarine deposystems in the Axhandle piggyback basin of central Utah. Geol. Soc. Am. Bull., 107, 297-315.

183

Tissot, B. and Welte, D.H. (1978) Petroleum formation and occurrence, second revised and enlarged edition: Springer-Verlag, New York, 699 pp.

Waters, M.R. (1989) Late Quaternary lacustrine history and paleoclimatic significance of pluvial Lake Cochise, southeastern Arizona. Quatern. Res., 32, 1– 12.

Zaleha, M. J. (2006) Sevier orogenesis and nonmarine basin filling: Implications of new stratigraphic correlations of Lower Cretaceous strata throughout Wyoming, USA. Geol. Soc. Am. Bull., 118, 886-896.

Zhou, J. and Lau, K-M. (1998) Does a monsoon climate exist over South America? J. Climate, 11, 1020–1040.

184

Figures

Figure C.1: Geologic map of the Cuenca de Pozuelos (simplified Page, 1996). Cross section A-A’ (simplified from Gangui, 1998) illustrates the subsurface structure of the basin and the syn-orogenic stratigraphic packages (black and grey horizons). Westward-verging thrust faults control subsidence in the basin and provide accommodation space for Quaternary deposits. Laguna de los Pozuelos occupies an axial location in the basin, and is seasonally fed by three ephemeral rivers. (LP = Laguna de los Pozuelos, SR = Sierras de Rinconada, SQ = Sierras de Quichagua, T = Tertiary, Q = Quaternary, M = Miocene, K = Cretaceous, O = Ordovician).

185

Figure C.2: (A) SRTM 90 digital elevation model of Cuenca de Pozuelos. Inset profile along cross section A-A’ (B) shows the relief created by hangingwall topographic barriers with the basin floor. Inset map (C) shows locations of surface sediment samples and the core LP07-1A. Surface sediments (n=79) were collected by hand from the playa-lake center, margin, and deltaic environments.

186

Figure C.3: Vegetation in the Cuenca de Pozuelos. Bonaventura et al. (1995) noted that the vegetation density exceeds other basins in the Puna, because mean annual rainfall approaches 300 mm. (A) C4 aquatic macrophytes dominated by

Ruppia sp. (B) Basin floor C3 vegetation, dominated by Stipa and Festuca spp. grasses. (C) Semiarid steppe shrubs that cover alluvial slopes in the basin, dominated by Fabiana densa and Baccharis boliviensis. (D) Succulent vegetation on alluvial slopes.

187

Figure C.4: Maps of surface sample physical sedimentology and geochemistry representing dominantly benthic diatoms, (C) total inorganic carbon (wt %), and (D) total organic carbon (wt %).

188

Figure C.5: Plot illustrating the inverse relation between mean grain size and total organic carbon. Samples < 20 μm are richest in organic carbon, especially those collected from the playa-lake center. Coarse silt and sandy samples, focused in playa margin and deltaic environments, have TOC values < 1.0 wt %.

Figure C.6: (A) Total organic carbon (wt %) vs. total organic nitrogen (wt %). Nitrogen values have been corrected for contributions from inorganic nitrogen. 13 (B) C:N vs. δ Com. Grey boxes spanning general fields of known values for lacustrine algae, C3 plants, and C4 plants from Meyers & Teranes, 2001.

189

Figure C.7: Modified van Krevelen diagram for a select transect of surface samples that span LP both axially and laterally. Inset map shows spatial distribution of samples. Note increasing HI values with increasing distance from playa margins and deltas. See text for details.

Figure C.8: Lead-210 and Cs-137 data from sediment core LP07-1A, with elemental, stable isotopic and organic mass accumulation rate (MAR) profiles. Systematic decay of excess PB-210 indicates negligible deflation and sediment reworking at the core site. Organic carbon richness (mean = 1.31 wt %) and - 2 - 1 MARoc (mean = 2.89 mg cm y ) increase toward the top of the dated interval.

190

Figure C.9: Synthesis of modern-recent (~ 100 years) sedimentation at LP. See text for details.

191

Tables

Table C.1: Surface sediment geochemistry and physical properties, Laguna de los Pozuelos (M = Playa margin; D = Delta; L = Playa-lake).

13 Sample Name Latitude Longitude Mean TIC TOC TN BiSi δ C C/N D (°S) (°W) (wt % ) (wt %) (wt %) (wt %) (‰) (atm) (µm)

LP06-01 (M) 22.2701 65.9858 13.52 0.29 0.93 0.03 2.42 -19.22 36

LP06-02 (M) 22.2697 65.9817 9.70 0.33 0.95 0.05 2.36 -19.49 22

LP06-03 (M) 22.2767 65.9898 59.02 0.23 0.63 0.01 2.14 -17.70 74

LP06-04 (M) 22.2755 65.9836 30.84 0.32 0.88 0.05 2.22 -19.86 21

LP06-05 (M) 22.2757 65.9788 13.98 0.35 1.04 0.07 2.29 -19.57 17

LP06-06 (D) 22.2910 65.9875 59.06 0.40 0.79 0.02 2.07 -21.73 46

LP06-07 (D) 22.2982 65.9893 30.99 0.04 0.83 0.02 1.36 -22.38 48

LP06-08 (M) 22.3062 66.0332 54.16 1.55 1.38 0.06 1.77 -17.12 27

LP06-09 (M) 22.3058 66.0212 15.05 0.40 1.06 0.05 1.71 -19.82 25

LP06-10 (M) 22.3044 66.0188 11.42 0.22 1.42 0.11 2.19 -20.29 16

LP06-11 (D) 22.3036 66.0020 12.81 0.14 1.19 0.06 3.08 -21.38 23

LP06-12 (D) 22.3036 65.9941 12.49 0.01 1.60 0.07 3.58 -23.46 25

LP06-13 (D) 22.3049 65.9493 14.77 0.10 1.07 0.03 2.05 -22.82 48

LP06-14 (L) 22.3048 65.9849 19.85 0.01 3.30 0.14 3.04 -24.59 27

LP06-15 (M) 22.3074 65.9753 22.79 0.48 1.19 0.05 1.26 -20.45 28

LP06-16 (M) 22.3068 65.9707 15.52 0.46 1.38 0.06 1.31 -19.74 29

LP06-17 (M) 22.3129 66.0376 26.69 0.94 1.10 0.05 1.96 -17.72 26

LP06-18 (L) 22.3132 66.0241 14.31 0.73 1.70 0.13 2.08 -18.60 15 192

LP06-19 (L) 22.3136 66.0133 12.88 0.41 1.30 0.05 1.64 -19.43 32

LP06-20 (L) 22.3149 66.0041 12.68 0.27 1.07 0.05 2.07 -20.26 21

LP06-21 (L) 22.3179 65.9954 6.42 0.20 1.36 0.08 2.22 -20.14 20

LP06-22 (L) 22.3199 65.9858 11.06 0.16 1.24 0.09 1.85 -21.07 17

LP06-23 (M) 22.3237 65.9691 14.36 0.57 2.00 0.69 1.35 -19.49 3

LP06-24 (M) 22.3255 65.9568 11.74 0.13 1.24 0.09 0.49 -19.10 16

LP06-25 (M) 22.3411 65.9606 10.15 0.26 1.32 0.11 1.78 -19.90 14

LP06-26 (L) 22.3345 65.9831 9.79 0.30 1.37 0.10 2.35 -20.34 16

LP06-27 (M) 22.3554 65.9641 14.75 0.49 1.18 0.09 1.06 -19.40 15

LP06-28 (M) 22.3539 65.9677 8.84 0.29 1.65 0.13 1.37 -19.90 15

LP06-29 (L) 22.3318 65.9966 8.19 0.35 1.44 0.12 1.61 -20.43 14

LP06-30 (L) 22.3313 66.0078 12.27 0.25 2.02 0.12 2.16 -21.24 21

LP06-31 (L) 22.3308 66.0181 11.59 0.73 1.49 0.10 1.25 -19.83 17

LP06-32 (M) 22.3302 66.0249 11.06 0.81 1.38 0.09 1.15 -18.96 17

LP06-33 (M) 22.3452 66.0412 36.86 0.97 1.26 0.07 1.45 -17.61 21

LP06-34 (M) 22.3459 66.0341 10.10 1.25 1.97 0.10 1.09 -17.25 23

LP06-35 (L) 22.3472 66.0286 7.49 0.97 1.83 0.11 2.25 -18.92 20

LP06-36 (L) 22.3494 66.0229 11.25 0.40 1.30 0.09 1.45 -20.40 17

LP06-37 (L) 22.3515 66.0133 8.06 0.43 1.81 0.10 1.21 -18.64 21

LP06-39 (M) 22.3636 66.0328 7.75 0.45 1.41 0.10 1.56 -20.13 17

LP06-40 (L) 22.3635 66.0279 7.77 0.42 1.40 0.10 2.87 -20.34 16

LP06-41 (L) 22.3635 66.0231 9.71 0.34 1.42 0.09 2.56 -20.31 18

LP06-42 (L) 22.3642 66.0144 8.07 0.35 1.48 0.08 1.86 -19.09 21

LP06-43 (L) 22.3633 65.9886 18.58 0.30 1.34 0.08 2.26 -20.00 19 193

LP06-44 (L) 22.3634 65.9848 9.05 0.38 1.82 0.09 1.62 -19.16 23

LP06-45 (L) 22.3635 65.9773 10.23 0.54 1.58 0.10 1.27 -19.41 19

LP06-46 (M) 22.3632 65.9659 16.44 0.52 1.29 0.08 1.71 -19.76 20

LP06-47 (M) 22.3812 65.9668 90.92 0.13 0.76 0.04 0.83 -23.76 22

LP06-48 (M) 22.3795 65.9713 19.71 0.73 1.55 0.08 1.84 -19.60 23

LP06-49 (L) 22.3787 65.9748 18.82 0.85 2.30 0.08 2.68 -16.38 35

LP06-50 (L) 22.3787 65.9795 10.45 0.65 1.52 0.11 2.46 -20.29 16

LP06-51 (L) 22.3789 65.9835 14.03 0.49 1.44 0.10 2.63 -20.34 16

LP06-52 (L) 22.3792 65.9899 11.70 0.39 2.26 0.12 2.26 -17.34 22

LP06-53 (L) 22.3794 66.0192 5.17 0.34 1.36 0.11 1.56 -20.61 15

LP06-54 (L) 22.3786 66.0225 7.88 0.35 2.35 0.13 2.69 -18.40 21

LP06-55 (L) 22.3782 66.0272 11.37 0.38 1.93 0.12 3.37 -18.83 19

LP06-56 (L) 22.3779 66.0299 12.71 0.54 1.48 0.12 2.62 -20.27 14

LP06-57 (L) 22.3778 66.0317 10.03 0.68 2.04 0.12 1.97 -18.48 21

LP06-58 (M) 22.3901 66.0426 156.3 0.93 0.93 0.01 1.41 -18.42 79

LP06-59 (M) 22.3913 66.0365 13.38 0.50 1.44 0.12 0.97 -19.46 14

LP06-60 (D) 22.3978 65.9965 9.57 0.21 1.64 0.11 2.45 -22.33 17

LP06-61 (D) 22.3995 65.9915 15.61 0.32 1.43 0.09 1.75 -21.62 19

LP06-62 (D) 22.4017 65.9875 120.0 0.14 0.88 0.02 1.00 -22.85 51

LP06-63 (D) 22.4042 65.9912 181.0 0.04 1.07 0.03 0.70 -24.60 42

LP06-64 (D) 22.4010 65.9980 12.05 0.15 1.83 0.10 3.71 -22.30 21

LP06-65 (M) 22.3913 66.0365 18.22 0.93 1.49 0.09 3.07 -18.74 19

LP06-66 (D) 22.3901 66.0426 101.8 0.40 0.50 0.02 0.71 -19.21 29

LP06-67 (D) 22.4068 65.9937 154.7 0.04 0.14 0.02 1.03 -23.80 8 194

LP06-68 (D) 22.4039 65.9992 13.64 0.31 1.44 0.09 2.88 -21.92 18

LP06-69 (D) 22.3996 66.0200 11.25 0.25 1.29 0.11 3.53 -20.95 14

LP06-70 (D) 22.3987 66.0239 7.56 0.32 1.37 0.11 2.70 -20.42 15

LP06-71 (D) 22.4004 66.0241 6.61 0.34 1.59 0.11 1.46 -19.93 17

LP06-72 (D) 22.4021 66.0193 11.69 0.23 1.30 0.10 2.99 -20.71 15

LP06-73 (D) 22.4036 66.0158 7.45 0.29 1.40 0.10 1.84 -21.15 16

LP06-74 (D) 22.4040 66.0141 6.93 0.25 1.45 0.09 3.25 -18.46 20

LP06-75 (D) 22.4076 66.0152 7.77 0.31 1.53 0.11 2.52 -21.25 16

LP06-76 (D) 22.4065 66.0216 9.90 0.25 1.45 0.10 3.85 -21.21 17

LP06-77 (D) 22.4060 66.0276 10.15 0.80 2.06 0.12 2.68 -20.16 20

LP06-78 (D) 22.4104 66.0276 128.7 0.02 0.23 0.13 1.13 -23.27 2

LP06-79 (D) 22.4119 66.0276 36.23 0.06 0.83 0.25 2.54 -21.40 4

LP06-80 (D) 22.4066 66.0254 21.24 0.32 1.40 0.58 3.25 -21.01 3

195

Table C.2: Surface sediment Rock-Eval pyrolysis data, Laguna de los Pozuelos. Coordinates for samples are presented in Table 1.

Sample S1 S2 HI OI Tmax

Name (mg HC / gm (mg HC / gm (mg HC / gm (mg CO2 / gm (°C) rk) rk) TOC) TOC)

LP06- 02 0.14 0.85 125 69 349

LP06- 04 0.10 0.62 83 51 377

LP06- 06 0.25 0.67 88 45 398

LP06- 12 0.50 1.67 115 64 408

LP06- 17 0.19 0.93 92 49 396

LP06- 18 0.27 1.43 111 62 424

LP06- 19 0.25 1.15 104 57 420

LP06- 20 0.34 1.03 99 50 374

LP06- 21 0.22 0.84 73 30 401

LP06- 22 0.30 1.09 107 53 402

LP06- 23 0.11 1.03 57 32 425

LP06- 24 0.16 0.85 96 52 407 196

LP06- 29 0.30 1.21 88 50 413

LP06- 37 0.26 1.83 145 160 427

LP06- 39 0.17 1.33 119 65 422

LP06- 40 0.26 1.93 151 98 427

LP06- 41 0.21 1.67 144 82 430

LP06- 42 0.22 1.72 156 105 427

LP06- 43 0.19 1.32 120 72 427

LP06- 44 0.20 1.77 126 73 429

LP06- 45 0.34 2.34 190 155 427

LP06- 46 0.09 1.02 99 48 427

LP06- 53 0.21 1.45 123 68 416

LP06- 70 0.13 1.13 98 55 420

LP06- 79 0.21 0.67 74 35 391

LP06- 80 0.14 1.27 94 55 421

197

APPENDIX D:

LIMNOGEOLOGY IN BRAZIL’S ‚FORGOTTEN WILDERNESS‛: A SYNTHESIS FROM THE LARGE LAKES OF THE PANTANAL

Michael M. McGlue1, Aguinaldo Silva2, Fabrício A. Corradini3, Hiran Zani4, Mark A. Trees1, Geoffrey S. Ellis5, Mauro Parolin6, Peter W. Swarzenski7, Andrew S. Cohen1, Mario L. Assine8

1Dept. of Geosciences, The University of Arizona, Tucson AZ 2Dept. de Ciências, Universidade Federal de Mato Grosso do Sul, Corumbá MS 3Faculdade de Geografia, Universidade Federal do Pará, Marabá PA 4Instituto Nacional de Pesquisas Espaciais, São José dos Campos SP 5Energy Resources Program, U.S. Geological Survey, Denver CO 6Faculdade Estadual de Ciências e Letras de Campo Mourão, Campo Mourão PR 7U.S. Geological Survey, Santa Cruz CA 8Dept. de Geologia Aplicada, Universidade Estadual Paulista, Rio Claro SP

Pending submission to Journal of Paleolimnology

198

Abstract

Sedimentary records from floodplain lakes have a large and commonly untapped potential to record wetland response to global change. The Brazilian Pantanal is a vast seasonally-inundated savanna floodplain system controlled by the flood pulse of the Upper Paraguay River. Little is known, however, about how floodplain lakes within the Pantanal act as sedimentary basins, or what impacts hydroclimatic variables have on limnogeological processes. This knowledge gap was addressed through an actualistic analysis of three large, shallow (< 5 m) floodplain lakes in the western Pantanal: Lagoa Gaíva, Lagoa Mandioré and Baia

2- 4+ 2+ Vermelha. The lakes are dilute (CO3 > Si > Ca ), mildly alkaline, freshwater systems, the chemistries and morphometrics of which evolve with seasonal flooding. Lake sills are bathymetric shoals marked by siliciclastic fans and marsh vegetation; flows at the sills likely undergo seasonal reversals with the changing stage of the Upper Paraguay River. Deposition in deeper waters, typically encountered in proximity to margin-coincident topography, is dominated by reduced silty-clays with abundant siliceous microfossils and organic matter.

Stable isotopes of carbon and nitrogen plus hydrogen index measured on bulk organic matter suggest that contributions from algae (including cyanobacteria)

and other C3-vegetation dominate in these environments. The presence of lotic 199

sponge spicules, coupled with patterns of terrigenous sand deposition and geochemical indicators of productivity, points to the importance of the flood pulse for sediment and nutrient delivery to the lakes. Flood-pulse plumes and waves likewise affect the continuity of sedimentation; short-lived radioisotopes indicate rates of 0.11 – 0.24 cm*y− 1 at sites of uninterrupted deposition. A conceptual facies model, developed from insights gained from modern seasonal processes, can be used to predict limnogeological change when the lakes become isolated on the floodplain or during intervals associated with a strengthened flood pulse. Amplification of the seasonal cycle over longer time scales suggests carbonate, sandy lowstand fan and terrestrial organic matter deposition and bioturbation during arid periods, whereas deposition of lotic sponges, aquatic organic matter, and highstand deltas characterize wet intervals. The results hold substantial value for interpreting perturbations to the regional hydrologic cycle, including alteration of effective precipitation in sub-tropical South America.

Introduction

Global estimates indicate that the total land surface area occupied by wetlands in the humid tropics and subtropics exceeds 3.3 x 106 km2 (Maltby and

Turner 1983). Prior to the 20th century, attitudes towards such wetlands were largely negative, fueled by the desire to reclaim perennially-inundated lands for 200

agriculture or industry (Mitsch and Gosselink 2000). More recently, international efforts have demonstrated the value of wetlands conservation, especially for the protection of numerous aquatic species and for maintaining hydrologic resources critical to downstream human populations. Perhaps most importantly, vast tropical wetlands are now recognized for the unique role they play in global biogeochemical cycles, particularly in the production and sequestration of greenhouse gases (Bartlett and Harriss 1993; Cao et al. 1998; Mitsch et al. 2009).

Global changes in air temperature or increased variability in the water cycle, such as those predicted by the IPCC AR4, are thought to place wetland ecotones at substantial risk for terrestrialisation (Chauhan and Gopal 2001; Bates et al. 2008).

Few in depth paleo-records from unaltered, tropical wetlands are available, however, to assess the response of these sensitive systems to well-known climatic perturbations of the Quaternary (see Donders et al. 2005 for an exception).

Stratigraphic sequences from large lakes in the Pantanal may provide a unique lens through which the response of tropical wetlands to environmental change can be viewed (Figure 1). The Pantanal is the world’s largest freshwater wetland and the hydrologic basin of the Upper Paraguay River (PR; Heckman

1998). Scientific inquiry into the region was delayed until the 20th century, as early attempts at exploration were deterred by harsh environmental conditions 201

along the Brazilian frontier (Por 1995). Sixteenth-century maps depicted the

Pantanal as a single large lake, purportedly because of reports given to Spanish conquistadors by lake-shore dwelling Indians (Por 1995). Surveying conducted centuries later has revealed that thousands of lakes dot the Pantanal landscape

(Ab’Saber 1988; Soares et al. 2004). However, limnogeological datasets from the

Pantanal are generally absent, constraining our understanding of the lakes and their linkages to the PR floodplain.

The purpose of this study is to provide the first synthesis of modern time- slice limnogeological knowledge on three lakes found along the floodplain of the

PR: Lagoa Gaíva, Lagoa Mandioré and Baia Vermelha (Figure 1). Several research groups have noted the potentially valuable archive of paleoclimate information stored within deposits from these and nearby floodplain lakes (Mayle et al. 2004; de Oliveira Bezerra and Mozeto 2008). Inferences from lake sediment may likewise be vital for understanding more recent environmental dynamics in the greater Pantanal. As tropical wetlands are ecologically sensitive, such information holds considerable value for conservation and sustainability planning. Before longer sediment core-based reconstructions can be fully realized, a number of key questions need to be addressed, including: How do these lakes act as depositional basins? What are the dominant limnological 202

processes that influence patterns of sedimentation? What baseline variability exists in the geochemistry of organic matter in modern lake sediments? What relationships exist between the lakes and the greater Pantanal wetland system?

Our focus with this study was to address these questions using observations and data acquired from the lakes from 2007-2009.

Site Description

The Pantanal (16-20°S, 55-58°W) straddles the borders of Brazil, Bolivia and

Paraguay (Figure 1; Alho et al 1988; Por 1995). Estimates from remote sensing techniques indicate that the inundated surface area of the entire Pantanal exceeds

130,000 km2 at maximum flood conditions (Hamilton et al. 2002). The Pantanal is particularly renowned for the biodiversity of its flora and for its role as a natural wildlife sanctuary, especially for numerous species of fish and aquatic birds (Heckman 1998; Junk et al. 2006; Lopes et al. 2007). Vegetation in the basin is a mosaic of cerrado (tropical savanna), Amazon-derived semi-deciduous forest,

Chaco-derived seasonal dry forest and aquatic plants (Cole 1960; Prance and

Schaller 1982). The spatial distribution of plant communities is heterogeneous and controlled dominantly by patterns of seasonal flooding, topography and soil type (Pinder and Rosso 1998). Lagoa Gaíva (LG), Lagoa Mandioré (LM) and Baia

Vermelha (BV) are located along the western margin of the PR (Figure 1). The 203

lakes are amongst the largest perennial standing bodies of water in the Pantanal and its Paraguai–Corixo Grande sub-region (Supplemental Table 1; de Magalhães

1992). The lakes are interspersed within the Serra do Amolar, a mountain range locally creating > 800 m of relief.

Geology

The Pantanal sits in a low-altitude (mean elevation < 200 m a.s.l.), elliptically-shaped, seismically-active basin (Figure 1; Assine and Soares 2004).

Andean tectonics are often invoked to explain the presence of the Pantanal basin, although its position within the foreland remains equivocal. Horton and

DeCelles (1997) argued that the Pantanal forms the backbulge depozone of the

Modern Andean foreland basin system. More recent research by Chase et al.

(2010) supports this hypothesis, as geoid anomalies indicate the presence of a flexural forebulge > 200 km west of the Pantanal. In contrast, Ussami et al. (1999) used geophysical data to suggest that the Pantanal depression formed through extension, set in motion by the interaction of the Andean forebulge with the

Neoproterozoic Paraguai fold-thrust belt. Regardless of its tectonic origin, at least

500 m of Quaternary siliciclastic and lateritic sediments fill the Pantanal basin

(Ussami et al. 1999). Although a number of fault trends are apparent in the basin, the NE-SW trending Transbrasiliano Lineament is most significant, as the 204

course of the PR is strongly altered by its presence in a number of locations

(Assine and Soares 2004). Several large rivers draining basin-margin highlands enter the Pantanal, forming expansive, low-gradient fluvial megafans north and east of the PR (Assine and Silva 2009). A wide variety of lacustrine environments are present in association with distal megafan surfaces, including thousands of saline and fresh ponds in the Nhecolândia region of the southern Pantanal

(Tricart 1982; Barbiéro et al. 2002).

Climate

Tropical latitude and seasonal migration of the Intertropical Convergence

Zone (ITCZ) controls climate patterns in the Pantanal. Mean annual air temperature in the region is ~ 25°C. Precipitation is spatially variable and strongly seasonal. Near the study area, annual precipitation typically exceeds 1000 mm, whereas precipitation to the north is greater, but more seasonally variable

(Supplemental Figure 1). The majority of precipitation (> 70%) falls during a prolonged wet season that lasts from late October to early April. Evaporation exceeds precipitation during most of the year in the Pantanal, with some areas experiencing a net annual loss approaching 300 mm (Alfonsi and Paes de

Camargo 1986; Por 1995). A shift in wind direction accompanies the change in seasons, as austral summer winds are from the northwest, whereas winds in the 205

austral winter are dominantly from the east-northeast. Over the past 40 years,

850 mb wind speed during the dry season is 5-6 m/s, whereas wet season velocities are comparatively slower.

Hydrology and Limnology

The hydrology of the Pantanal is controlled by the flooding cycle of the

PR, precipitation and evapo-transpiration. A particularly notable aspect of PR’s annual flooding cycle is the delayed passage of its flood pulse moving from north to south, altering stage in the main channel by more than 5 m over the course of several months (Hamilton et al. 1997). Small tributary channels connect each of the study lakes to the PR (Figure 1). Exchange between the lakes and the PR is strongest following the passage of the flood wave, which commonly occurs several months after peak austral summer precipitation. Seasonal retention of the flood waters in the northern Pantanal results from the different mechanisms of floodplain inundation, which include: (1) overbank flow of the PR and smaller tributaries, commonly in association with heavy seasonal rainfall; (2) a backwater effect, where overland flow on floodplains becomes impeded by the elevated stage of the PR; and (3) local ponding of rainfall on heavily-vegetated floodplain soils (Hamilton 1999).

Approach 206

Tropical wetlands typically house a wide variety of depositional environments that are linked by prevailing climate and the hydrology of large rivers. Inundation and passage of the PR flood pulse are key mechanisms controlling floodplain ecological interactions and productivity in the Pantanal

(Junk et al. 1989; de Oliveira and Calheiros 2000). However, the implications of flood pulse dynamics for limnogeological processes represent a key knowledge gap for the region. In order to redress this gap, we undertook an assessment of modern floodplain lake water chemistry, bathymetry, morphometrics, physical sedimentology and sedimentary biogeochemistry. Integrated analyses of limnological processes and their influence on geological products are vital, as accurate interpretations of paleoenvironments developed from sediment cores hinges on robust calibration datasets that validate the meaning and fidelity of sedimentary proxy data.

Materials and Methods

We conducted reconnaissance bathymetric and water sampling surveys in

2007 and 2008, using a hand-held fathometer. Water chemistry was determined in situ using a YSI model 85 multi-meter for dissolved oxygen, conductivity and temperature whereas a Hach Sension1 meter was used to measure pH 207

(Supplemental Table 2). Major ions were determined using a Perkin-Elmer

Optima 5300 ICP-OES at the University of Arizona.

The upper 2-3 cm of the modern lake floors (n = 40, 69 and 45 for LG, LM and BV, respectively) were collected using a ponar grab-sampler in order to assess the physical and geochemical attributes of sediments that have accumulated over several seasonal flooding cycles. Sample grids were constructed in order to provide maximum coverage of the different environments encountered in the field. The spacing between individual sampling stations varied, but in all cases was < 2.5 km. Short (< 1.5 m) sediment cores were collected in order to evaluate rates of sedimentation. Cores were collected from BV and LM in 2008 and LG in

2009 using a Livingstone-style square rod corer or an aluminum-barrel vibracorer system. Sediment geochronologies and linear sedimentation rates (cm*y− 1), were

210 derived for LG cores using the decay of excess Pb (t1/2 = 22.3 yr) at the USGS in

Santa Cruz, CA, following the calculations presented in Swarzenski et al. (2006).

Precision in the activity of excess 210Pb was typically better than 10%.

Smear slides were inspected under a petrographic microscope to estimate major sedimentary components. Concentrations of biogenic silica (BiSi), total organic carbon (TOC), total nitrogen (TN) and stable isotopes of C and N were measured to assess productivity and the composition of organic matter (OM) in 208

the lake sediments. Biogenic silica analyses were completed at the University of

Minnesota utilizing multiple extractions of hot alkaline digestions following the modified protocol of DeMaster (1979). Reported values have an analytical precision of ~ 1.0%. Elemental and stable isotopic analyses of sedimentary OM

13 were conducted at the University of Arizona. Total organic carbon, TN, δ COM

15 and δ NOM were measured on a continuous-flow gas-ratio mass spectrometer

(Finnigan Delta PlusXL) coupled to a Costech elemental analyzer. Samples were combusted in the elemental analyzer. Standardization is based on acetanilide for

13 elemental concentration, NBS-22 and USGS-24 for δ Com. Precision was better

13 15 than ± 0.09 and 0.20 for δ Com and δ Nom, respectively, based on repeated internal

13 15 standards. δ COM values are given with respect to the PDB standard and δ NOM values are reported with respect to air. A select number of decarbonated samples were analyzed by Rock-Eval pyrolysis at the University of Houston in order to further discriminate the provenance of sedimentary OM (Espitalie et al. 1977).

Particle size analyses were conducted at the University of Arizona using a

Malvern laser-diffraction analyzer coupled to a Hydro 2000S sample dispersion bench. Particle size provides key insights on environmental energy and sedimentary processes (traction flow vs. suspension settling). Samples were pre- treated according to Johnson and McCave (2008) to remove BiSi and OM, whereas 209

authigenic minerals were removed by excess hot 1M HCl digestion for 24 hours.

Following completion of the digestions, samples were disaggregated using a solution of sodium hexametaphosphate and a mechanical shaker. Prior to analysis, sediments were examined optically to ensure the removal of non- terrigenous particles.

A subset of samples from each lake was processed for sponge microfossils at the Faculdade Estadual de Ciências e Letras de Campo Mourão following the procedure outlined in Volkmer-Ribeiro and Turcq (1996). Sample aliquots were suspended in water, dried on microscopic slides, Entellan mounted and permanently sealed with cover slips. Species identification and nomenclature followed Volkmer-Ribeiro and Pauls (2000) and included: (1) megascleres, which are spicules that integrate the sponge skeletal network; (2) microscleres, a smaller spicule within the sponge skeleton; and (3) gemmoscleres, spicules that cover the gemmules and which ultimately define families, genera and species in freshwater sponges.

Results

Morphometrics, Bathymetry, and Water Chemistry

Figure 1 and Supplemental Tables 1 and 2 present morphometrics, bathymetry and water chemistry data for LG, LM and BV. The lakes share a 210

number of limnological characteristics in common after the passage of the PR

2- 4+ 2+ flood pulse, including: (1) dominant CO3 > Si > Ca hydrochemistries; (2) well- oxygenated water columns; (3) low (freshwater) conductivity values; (4) weakly- basic pH values; (5) elongate basin shapes with fetch distances < 25 km; (6) shallow average water depths (< 3.5 m); and (7) maximum water depths proximal to high relief lake-margins. Complete thermal stratification was not detected in any of the lakes, but confirmation of mixing state awaits future studies.

Recent satellite imagery indicates that LG has a surface area of ~ 100 km2 during passage of the flood. The basin sill is located along the eastern margin of the northern sub-basin of the lake, at an altitude of ~ 95 m a.s.l. Proximal to the sill, LG is very shallow, with water depths < 2.5 m encountered ~ 2 km from shore

(Figure 1). A bathymetric low exists < 3 km from the eastern high-relief margin of the southern sub-basin; water depths were routinely > 4.5 m in this area. Lagoa

Mandioré has a surface area > 150 km2 during the passage of the flood pulse. The basin sill is found on the southeastern lake margin at an altitude of ~ 93 m a.s.l.

Water depths generally increase with distance from the shoreline in LM. An exception to this trend occurs < 1 km from the basin’s northeastern high-relief margin, where measurements of water depth were > 4.0 m (Figure 1). Baia

Vermelha exhibits a highly variable surface area (< 150 km2) resulting from an 211

expansive, low-gradient plain on the northern lake margin (Figure 1). Several intermittent channels (~ 91-92 m a.s.l.) connect BV with the PR on the eastern side of the basin. The bathymetry of BV is complicated by the presence of a circular headland ~ 3.5 km offshore from the northwestern axial lake margin. West of the headland, bathymetric contours are approximately concentric and reach a maximum depth of ~ 3.1 m (Figure 1). Baia Vermelha is deepest (~ 3.5 m) east of the headland, ~ 2 km from its southern, high-relief lake margin.

Sedimentology and Geochemistry

Figure 2 presents the spatial distribution of mean siliciclastic particle size and the concentrations of BiSi, TOC, and TN in the lake floor sediments of LG,

LM and BV. In LG and LM, dark green silts and clays are commonly associated with the deepest areas of the basins, whereas similar sediments in BV were found only west of the headland, where water depths exceeded 2.5 m.

Photomicrographs of sedimentary components in these environments are presented in Supplemental Figure 2. Terrigenous siliciclastic grains in the sand fraction (> 63 μm) are common in the lakes, especially proximal to basin sill points. In LG, a broad, sand-rich fan emanating from the basin sill occupies the shallow northern sub-basin (Figure 2). Likewise, lake floor substrates at the northern end of the southern sub-basin, where the lake receives weak overflow 212

from Lagoa Uberaba, are dominantly sandy. In LM, axial margins and cuspate bays are commonly sand-rich environments. Inflow from the PR at the basin sill forms a sandy deltaic fan with a surface area > 40 km2, whereas a much smaller sand-rich delta exists along the northern axis of LM (Figure 2). Approximately 3 km offshore from the northern delta, a 1.5 km-long, partially-vegetated sand bar creates a pronounced bathymetric shoal; a similar shoreline-parallel bar exists off a cuspate bay on the southwestern margin of the lake. Whereas coarse-grained sediment west of the headland in BV is confined near the shoreline, lake floor sediment east of the headland is commonly sandy. A northeast oriented, ~ 30 km2, sand-rich fan occupies much of the center of the basin east of the headland.

Along the eastern lake margin, sandy sediments mark flood-pulse connection points with the PR.

Values of BiSi in LG range between 0.0 and 3.5 wt %, with a mean of 1.5 wt

% (Supplemental Table 3). Biogenic silica concentrations are highest on the eastern side of southern LG, where samples exceed 2.5 wt % (Figure 2).

Elsewhere in the basin, lake floor sediment is more depleted in BiSi, with samples generally exhibiting values below 2.0 wt %. Total nitrogen ranges from 0.0 to 0.4 wt %, with a mean of 0.2 wt %, whereas values of TOC range from 0.0 to 3.3 wt %, with a mean of 1.5 wt %. Contours of TN and TOC are simple and concentric in 213

the southern sub-basin. Geochemical analyses of OM from LG indicate relatively

13 15 narrow ranges of C/N, δ COM, and δ NOM values (Figure 3). Values of C/N and

13 δ COM typically plot within known fields for lacustrine algae mixed with C3-

15 pathway terrestrial plant matter (Meyers and Ishiwatari 1993). Values of δ NOM range from +1.8 to +3.3 ‰, with a mean of +2.5‰. Hydrogen index values for LG sediments are < 200 mg HC/ gm TOC and plot near the Type II-Type III kerogen boundary (Figure 3).

Lake floor sediments in LM exhibit BiSi values between 0.0 and 12.9 wt %, with a mean of 4.3 wt % (Supplemental Table 3). Axial concentrations of BiSi are high (> 7.0 wt %) in LM, especially < 2.5 km from the western lake shore (Figure

2). Sediments deposited near the basin sill have less BiSi (< 2.0 wt %). Total nitrogen ranges from 0.0 to 1.0 wt %, with a mean of 0.5 wt %, whereas values of

TOC are particularly high, ranging from 0.0 to 8.5 wt %, with a mean of 4.9 wt %.

Contours of TN and TOC indicate elevated concentrations on the western side of the basin axis, away from point sources of fluvial sediment (Figure 2). Total nitrogen and TOC demonstrate statistically significant correlations with water depth in LM (r2 = 0.51 and 0.46). Sedimentary OM in LM exhibits the broadest range of δ13C and δ15N values encountered, whereas the range of C/N values (10.0

13 to 13.6) is comparable to the other lakes (Figure 3). The mean δ COM isotopic 214

composition (-24.0 ‰) is at least 2.5‰ more positive than the means of LG and

15 BV. In contrast, values of δ NOM are the most depleted in the study area, with a range of -0.1 to +3.2 ‰ and a mean of +1.4 ‰ (Supplemental Table 3). Hydrogen index values for LM surface sediments range from 206 – 357 mg HC/ gm TOC

and fall within the Type II kerogen field (Figure 3).

Lake floor sediments in BV exhibit BiSi values between 0.0 and 4.3 wt %, with a mean of 1.6 wt % (Supplemental Table 3). The highest concentrations of

BiSi are located west of the headland and near the southeastern lake margin; most of the lake floor substrates between the axial margins are < 2.0 wt % BiSi

(Figure 2). Total nitrogen ranges from 0.0 to 0.4 wt %, with a mean of 0.2 wt %, whereas values of TOC range from 0.0 to 4.4 wt %, with a mean of 2.0 wt %. The patterns of TN and TOC contours are very similar to BiSi, with elevated concentrations along the axial margins (Figure 2). C/N values on BV samples are slightly higher (mean = 12.5) than those measured in the other lakes, whereas

13 δ COM sample means are comparable to LG (-26.5 and -27.0 ‰, respectively).

15 Sedimentary OM in BV exhibits δ NOM values that range from +0.3 to +4.4 ‰, with a mean of +3.1‰. Hydrogen index values for BV sediments range from 123

– 287 mg HC/ gm TOC and plot near the Type II-Type III kerogen boundary

(Figure 3). 215

Figure 4 presents 210Pb-derived sedimentation rates for two cores collected from LG. The upper 20-30 cm of both cores contain massive, dark-green, silty clay with OM, diatoms and sponge spicules commonly present. Linear sedimentation rates vary spatially within the southern sub-basin. Recent sedimentation has been relatively slow (0.11cm*y− 1) at core site GVA09-3A, which is situated near the northern end of the southern sub-basin (Figure 1). At the southern core site (GVA09-2A) recent sedimentation has progressed more rapidly, at a rate of 0.24 cm*y− 1. A 137Cs-derived gecohronology for GVA09-3A of

0.1 cm yr-1 supports the sedimentation rate obtained from the excess 210Pb distribution.

Sponge Microfossils

Figure 5 and Supplemental Table 4 illustrate the variety of sponge microfossils detected in the modern sediments of LG, LM and BV. Megascleres and gemmoscleres of two lentic species were common to all three lakes:

Radiospongilla amazonensis Volkmer-Ribeiro and Maciel 1983 and

Trochospongilla variabilis Bonetto and Ezcurra de Drago 1973. In BV, spicules of the lentic sponge Corvoheteromeyenia sp. ranged from common to abundant in lake floor sediments west of the headland, but were much less common proximal to connections points with the PR. Modern sediments in the southern sub-basin 216

of LG contained spicules of Metania spinata Carter 1881 and Heteromeyenia sp. sponges; both are commonly associated with lakes and wetlands in the cerrado biome (Volkmer-Ribeiro 1992)

Notably, all three lakes contained the robust megascleres and gemmules of Corvospongilla secktii Bonetto and Ezcurra de Drago 1966, a sponge known to colonize rocky substrates and channel-margin vegetation in large neotropical rivers (Figure 5; Batista et al. 2003). These spicules were found in sediments across LG and BV, whereas they were more localized in LM. Lagoa Mandioré also contained abundant spicules of the lotic sponge Oncosclera navicella Carter 1881 in sediments deposited near the basin sill; such spicules were not detected in the center and northern axis of the basin. In LG, spicules of lotic sponges were found across the basin, and included gemmoscleres of Uruguaya corallioides

Bowerbank 1863.

Discussion

Any discussion of limnogeological processes in the floodplains of the

Pantanal must consider the downstream variability exhibited in the passage of the PR flood pulse. A key characteristic distinguishing LG, LM and BV from floodplain lakes in many other tropical watersheds is the near anti-phasing of the dominant rainy season with the arrival of riverine flood waters (Veloso 1972; 217

Heckman 1998). The two-to-four month time lag associated with the passage of the PR flood pulse from the north to the central Pantanal is important to lake hydrochemistry, morphometrics and ultimately, sedimentary processes. The arrival of the PR flood pulse following the austral summer causes lake surface areas, fetch distances and bathymetric gradients to reach their maxima as wind speeds begin to reach peak velocity (Supplemental Figure 1). This has implications for wave development, sediment re-suspension and the temporal continuity of stratigraphic records. Because of its tapered shape and short effective fetch, the deepest regions of LG likely escape the influence of waves

(critical depth = 4.1 m), whereas the potential for wave re-working and sediment re-suspension is much higher in LM and BV due to their morphometrics. This provides a plausible explanation for the well-behaved decay of excess 210Pb in the uppermost sediments in LG (Figure 4). Stratigraphic records from the deepest regions of LG are likely to be continuous, whereas those from LM and BV appear to be somewhat more punctuated. In LM, unsupported 210Pb is confined to the upper 2-3 cm of sediment cores, interpreted to result from recent wave-driven re- suspension of sediments at the core site (McGlue unpublished). The impact of waves on LM is further illustrated by shoreline-parallel sand bars and cuspate sandy beaches; such features are well known from lakes with high energy 218

coastlines (Adams and Wesnousky 1998). In BV, erratic downcore concentrations of excess 210Pb suggest bioturbation may impact the continuity of the stratigraphy of this basin (McGlue unpublished). Feral cows and gregarious water birds in the central Pantanal make the potential for top-down bioturbation high, especially at shallow sub-lacustrine sites. Highly-variable sedimentation rates could produce a similar downcore 210Pb decay pattern, but massive lithologic units bound by irregular bedding planes in the BV sediment core suggest this alternative is less likely.

Another key difference between the Pantanal lakes and other floodplain lakes is the mechanism of riverine inundation. The water cycles of many tropical floodplain lakes rely on groundwater seepage through hydromorphic soils or from overland flow when bankfull conditions are exceeded (Cohen 2003).

In contrast, LG, LM and BV are silled basins linked to the PR by perennial channels. Tributaries connecting the lakes to the PR not only influence water cycling, but they also impact lake sedimentology. Channels impart a first-order control on flow velocity and the routing of bed-load into the lake basins, whereas unconfined overland flows are more diffuse, allowing vegetated levees and floodplain soils to act as sediment traps. Terrigenous particle size contours show the impact of fluvial traction-flows on the lakes (Figure 2). Fan-shaped sand 219

bodies occupy broad areas of the lake floors and are key components of the modern facies architecture that are genetically linked to the passage of the flood pulse (Figure 6). The orientation of fan apices confirms that flows originating at basin sills are the source of these deposits. Another hallmark of ‚fill-phase‛ sedimentology is the presence of lotic sponge microfossils, which reflect at least seasonal connectivity between the lakes and the PR. The ecological characteristics of readily-identifiable sponge remains are well known, and the spicules of lotic species are typically well preserved, albeit with some taphonomic overprint from transport (Batista et al. 2003; Volkmer-Ribeiro and Pauls 2000). Prior research in

Brazil has demonstrated the utility of spicule analysis to discriminate between fluvial and lacustrine environments in sediment cores (Parolin et al. 2007).

Results from the present study show that down-core variability in lotic vs. lentic assemblages could provide an important ecological perspective on fluvial- lacustrine connectivity for the floodplain lakes of the Pantanal.

Sand-sized particles are common in all modern lake floor samples except those collected from relatively deep water adjacent to high-relief lake margins, where suspension settling and organic sedimentation dominate. Although subsurface datasets from the lakes are unavailable, the correlation of bathymetric lows and high-relief rocky margins suggests the potential for neotectonic controls 220

on accommodation space in the basins (Figure 1). In the Pantanal, structural influences on fluvial channel patterns are not unusual and shallow earthquakes have been recorded in the region (Assine and Soares 2004). Por (1995) suggested that karst plays an intimate role in large lake formation in the Pantanal, but our data suggest that this is not likely for LG, LM and BV. Hydrochemical datasets show that the lakes are extremely dilute, whereas karst basin waters are typically rich in divalent metals due to dissolution of carbonate aquifers (Supplemental

Table 2). During sampling, the lake waters broadly reflected the chemical composition of the PR, especially for conservative cations; these waters are under- saturated with respect to calcite (Hamilton et al. 1997).

Figures 6 and illustrate our interpretation of a three-phase seasonal evolution that links lacustrine hydrochemistry, morphometrics and sedimentology to stage of the PR, which is strongly modulated by regional climate. Seasonal evolution of the large lakes of the Pantanal shares a number of similarities to lakes on the floodplain of the Orinoco River described by Hamilton and Lewis (1987). Channelized linkages to the PR may serve as conduits for reversing flows following the passage of the flood pulse, during a transient

‚draining-phase‛ leading to full isolation of the lakes from the river

(Supplemental Figure 3). In the central Pantanal, the stage height of the PR 221

drops by > 5 m in the austral summer, when the flood pulse is retained in the north and evaporation increases due to elevated air temperatures. Theoretical estimates of lake level decline due to evaporation alone indicate lake surfaces drop by at least 0.6 m annually. At peak inundation, lake-surface altitudes for LG,

LM and BV are typically 5, 4, and 6 m below their sill heights, respectively. As a result, short-lived reversed flows are probable for LG and LM, and possible (but likely rare) for BV. Complete isolation of the lakes generally occurs in the austral summer, suggesting direct rainfall and groundwater inputs are less important to morphometrics and bathymetry than the PR flood pulse.

Both fluvial and lacustrine processes control organic facies development in the large lakes of the Pantanal. River-borne solutes probably provide a key source of nutrients that help control algal productivity in LG, LM and BV. Junk et al. (1989) were among the first to recognize the importance of flood pulse dynamics for nutrient delivery and productivity in the tropics, whereas Hamilton et al. (1997) noted that the floodplains of the central Pantanal are sinks for flood pulse-derived OM and sites of inorganic nutrient transformation. However, concentrations of N, P and suspended solids are lower in the PR than in other rivers draining tropical savannas (Hamilton et al. 1997). As a consequence, local allochthonous sources of nutrients may likewise be important. Macrophyte 222

decomposition and internal recycling are two possible local sources of phytoplankton-limiting nutrients. Floating macrophytes and marsh vegetation were commonly encountered near the sills and along the margins of each of the lakes, and dry-season decomposition of such vegetation is a known source of dissolved nutrients in other floodplain lakes (Furch et al. 1983). Likewise, mixing-driven re-suspension of lake bottom sediments can lead to the release of nutrients, but this process also prolongs the exposure of labile organic compounds to degradation by aerobic respiration (Forsberg et al. 1988; Henrichs

1992). Ultimately, OM-rich facies development across broad regions of the lakes is mediated by siliciclastic dilution during ‚fill-phase‛ sedimentation.

Biogeochemical signals in LM samples are complex and appear to reflect elevated productivity and subsequent degradation (Figure 2). Organic sediments from LM are relatively rich in 13C (Figure 3). The factors that can

13 influence δ COM values include: 1) the isotopic composition of the carbon source

(commonly lake water DIC); 2) phytoplankton assemblage and metabolic rates;

3) the proportion of C3 and C4 terrestrial vegetation within particulate organic matter; and 4) diagenesis (Meyers and Teranes 2001; Talbot and Johannessen

13 1992). At present, δ CDIC measurements on lake or river water samples are

unavailable, but we assume that the concentration of aqueous CO2 is relatively 223

- high due to LM’s pH (8.3). Therefore, utilization of HCO3 during photosynthesis is not a likely explanation for the enriched carbon isotopic signature observed in these samples (Figure 3). Hamilton et al. (1997) noted that C/N values for particulate organic matter (POM) in the PR averages 8.5 but algae comprised only a portion of this OM. Wantzen et al. (2002) reported a range of -26.0 to -

32.0‰ for δ13C values in POM in the northern Pantanal. These values are

consistent with geochemical characteristics of both C3 terrestrial plants and algae.

Smear slides show that LM sediments contain abundant green algae and diatoms, with cyanobacteria, including Anabaena sp., common in some samples

(Supplemental Figure 2). Algae blooms were observed during our sampling and the relatively high δ13C values can be at least partially accounting for by the preferential uptake of 12C during photosynthesis, thus depleting the residual DIC reservoir in the lake (McKenzie 1985; Amorim et al. 2009). Given its long fetch

(~ 22 km), wind driven internal nutrient fertilization provides a compelling explanation for algal blooms and elevated carbon burial in LM. Intriguingly, samples with very high δ13C values generally exhibit nitrogen isotope ratios below 2.0 ‰ (Figure 3). The presence of nitrogen-fixing cyanobacteria in these samples provides one viable explanation for these δ15N values, as well as the high

TN concentrations (Talbot 2001). However, HI values from LM are below 400 224

mg HC / gm TOC, suggesting cyanobacteria do not dominate the phytoplankton assemblage and local allochthonous sources may also contribute to the lake’s carbon cycle (Talbot and Livingston 1989). Importantly, oxygen index values

exceeding 200 mg CO2/gm TOC indicates that diagenesis also affects the geochemistry of OM in LM. Re-suspension and oxidation of buried sediments can alter primary geochemical signals, due to selective loss of N and 12C (in the latter,

13 due to oxidative fractionation that may increase δ COM values by 1-2 ‰).

13 Therefore, elevated C/N and δ COM may in part reflect the influence of oxidation in some samples.

Biogeochemical data from LG and BV lack the variability of samples from

LM and suggest lower productivity. Sedimentary OM samples from LG and BV have low δ13C values and C/N values < 20, commonly thought to reflect admixtures of lacustrine algae and vascular plants (Meyers and Ishiwatari 1993).

Similar to LM, values of lake water pH suggest that dissolved CO2 in isotopic equilibrium with the atmosphere is likely the DIC source for primary producers in these lakes. Hydrogen index values below 300 mg HC / gm TOC for both LG and BV support the interpretation of a mixed organic provenance. Examination of smear slides confirms the presence of algae and terrestrial plant fragments in both lakes (Supplemental Figure 2). In general, the nitrogen isotope composition 225

of OM from LG and BV is consistent with δ15N values associated with lake sediments composed of both algae, which commonly preserve the isotopic signature of lake water dissolved inorganic nitrogen, and terrestrial vegetation,

which utilize atmospheric N2. The data are also consistent with common aquatic macrophytes and flood-pulse derived POM in the Pantanal. Several BV samples, collected west of the headland, exhibit high δ13C values and low δ15N values (<

2.6‰); lake-margin marshes, dominated by C4-pathway Paspalum repens likely contribute to the isotopic composition of these samples (Fellerhoff et al. 2003).

Rock Eval data suggest that post-depositional alteration of OM is less pronounced in LG and BV when compared with LM, likely the result of differences in morphometrics, wave development and sediment re-suspension.

Figure 6 presents a conceptual model of sedimentation derived from the observed modern seasonal sedimentary cycle and its linkages to the PR flood pulse. Data described in this study provide constraints on the processes that influence modern lithofacies patterns and geochemical baselines. With caution, these inferences can be used to predict facies migrations during intervals of major environmental change, such as those that might be preserved in sediment core records. Evidence of meaningful alteration of climate is already well established by relict landforms (dunes, lunettes, and deflation pans) in the 226

southern Pantanal (Soares et al. 2003). Thus, over short intervals of time, climate change in the PR headwaters is clearly an important mechanism that could lead to significant change in sedimentation in LG, LM and BV. Climatic conditions that promote a stronger flood pulse and deeper lakes, such as sustained southerly advance of the ITCZ, are likely to: (1) create additional fluvial entry points into the lakes, leading to ephemeral marginal fans; (2) back-flood existing incipient river channels, creating highstand deltas; (3) increase primary productivity and algal blooms due to enhanced riverine nutrient supply; (4) enhance the accumulation of sponge spicules from lotic species assemblages; and (5) develop fine-grained organic facies where suspension fall-out processes dominate, perhaps with higher temporal continuity due to increased water depths and stable water column stratification. Clearly, changes in sedimentation accumulation rates can be expected as major hydrologic thresholds are crossed. In contrast, sustained lake level lowstands associated with reduced effective precipitation in the PR headwaters will serve to magnify ‚isolation-phase‛ processes and sedimentation, including: (1) curtailed deposition of lotic sponge spicules; (2) incision of incipient river channels, with the potential for lowstand fan deposition and potential reworking of the lake floor; (3) possible precipitation of reduced metals and chemical (carbonate) sediments as isolated lake waters concentrate and mix; 227

(4) reworking of sub-aerially exposed lake bottom deposits, through deflation or bioturbation; and (5) oxidation of labile biogenic sediments and diagenetic evolution of elemental and stable isotopic values on OM (Figure 6). A space-for- time substitution with large lakes in the southern Pantanal (e.g., Lagoa Cáceres;

Figure 1) provides insights on chemical sedimentation during lowstands. This region experiences lower yearly precipitation and delayed passage of the flood pulse; minor calcite precipitation occurs in these lakes today (Supplemental

Table 5).

The processes described herein are common to many floodplain lake systems, often resulting in low-to-intermediate resolution records with highly variable sedimentation rates (Cohen 2003). Importantly, neotectonic alterations of drainage patterns are an alternative mechanism that could substantially alter the PR and its connection to floodplain lakes in the Pantanal. For example, faulting of the shallow crust have been implicated in the capture and redirection of large rivers in tropical Africa, which in some cases have influenced sedimentary processes in wetlands and lakes (Gumbricht et al 2001; McGlue et al

2006). Similar tectonic processes could affect the position of the PR, cutting the lakes off from the flood pulse and leading to ‚isolation-phase‛ sedimentation.

Likewise, crevasse splay development and channel migrations have been 228

documented in the central Pantanal over the past several decades (Assine 2005).

Avulsion of the PR could spur meaningful changes in the magnitude of the food pulse on LG, LM and BV without coeval climate change. Independent constraints on fluvial geomorphic evolution are therefore necessary context for interpreting paleoenvironments in the Pantanal and similar wetlands.

Conclusions

1. Situated along the western margin of the PR and strongly influenced by its

flood pulse, LG, LM and BV are freshwater, well-oxygenated, weakly basic,

dilute systems with elongate basin shapes and shallow maximum water

depths (< 5 m). Correlation of bathymetric maxima and coincident margins

suggests a neotectonic control on lake formation, likely modulated by

channel migrations of the PR and its tributaries.

2. The flood pulse of the PR impacts the sedimentology of the lakes through

delivery of terrigenous sand particles, sponge spicules, POM and productivity-

limiting nutrients. In each of the lakes, basin sill points are marked by the

apices of sandy fans; such deposits are characteristic of ‚fill-phase‛

sedimentation. The presence of gemmoscleres of C. secktii, as well as spicules

of O. navicella and U. corallioides, further implicates the role of the PR flood

pulse in the composition of biogenic sediments at LG, LM and BV. 229

Accumulation of fine-grained sediments is most pronounced in deep water

areas, where suspension fall out and settling of organic particles dominates.

3. Sediment biogeochemistry (BiSi, TN and TOC) suggests that production and

burial of biomass is highest in LM and somewhat lower in LG and BV. Stable

isotopes of carbon and nitrogen and HI values indicate a mixed (various

lacustrine algae and vascular plants) organic provenance for the lakes.

Cyanobacteria and post-depositional re-suspension of lake bottom sediments

strongly influence geochemical signals in LM, whereas bulk OM

geochemistry is less variable in LG and BV. Both intrabasinal (decayed lake-

margin vegetation, aquatic macrophytes and sediment recycling) and

extrabasinal (PR flood pulse) sources of nutrients appear to be important for

productivity in the lakes. The range of concentration values encountered for

modern lake floor sediments places important constraints for the

interpretation organic indicators from sediment cores.

4. Sites of un-interrupted sedimentation were only encountered in the deepest

regions of LG. Decay of short-lived radioisotopes in the upper ~ 30 cm

sediments of LG suggests sedimentation rates range between 0.11 – 0.24

cm*y− 1. In LM and BV, critical wave depths typically exceed maximum water

depths, suggesting sediment re-suspension from waves precludes the steady 230

accumulation of excess 210Pb. Bioturbation by water birds or feral cows also

limits depositional continuity at shallow depths in all of the lakes.

5. Sensitivity of the PR flood pulse to climate change, neotectonics and fluvial

autocyclic processes has important implications for the limnogeology of the

large floodplain lakes and accordingly their sediments are important sentinels

of environmental change for the Pantanal. Actualistic datasets and space-for-

time considerations provide a predictive framework for facies migrations with

major changes to the PR flood pulse. The rate and continuity of

sedimentation will vary considerably as major hydrologic thresholds are

crossed. Intensification of the flood pulse is likely to be represented by higher

lake levels and concomitant decrease in grain size, an increase in the burial of

OM due to elevated productivity, highstand deltas and lotic sponge

assemblages. By contrast, weakening of the flood pulse promotes chemical

sedimentation, reworking of the lake bottom by bioturbation, lowstand fan

deposition and oxidation of OM as the lakes become isolated on the

floodplain.

Acknowledgements

The title for this contribution is adapted from: The Pantanal – Brazil’s

Forgotten Wilderness by Vic Banks (1991). The research presented in this paper 231

was supported by the American Chemical Society (PRF program grant 45910-

AC8), the São Paulo Research Foundation (FAPESP grant 2007/55987-3) and the

UA-Exxon Mobil COSA project. Generous grants to MM from Laccore, the

Chevron Corporation, Kartchner Caverns, and PAGES assisted in the completion of this project. Research in Brazil would not have been possible without logistical and scientific support from ECOA-Brazil, EMBRAPA-Brazil, UFMS – Campus do

Pantanal, the Fazenda Santa Teresa and the citizens of Amolar (MS – Brazil). We are extremely grateful to K.Wendt, F. dos Santos Gradella, S. Kuerten, B. Lima de

Paula, D. Calheiros, R. Lins, A. Lins, and T. Matsushima for their assistance. C.

Gans and E. Guerra de Lima provided vital support at UA during our 2009 field season. C. Helfrich, C. Landowski, X. Zhang, C. Eastoe, J. Ash and the staff of the

Limnological Research Center at the University of Minnesota provided support in the laboratory. Critical reviews by S. Harris, C. Turner, S. Ivory, and M. Blome substantially improved the quality of the text.

232

References

Ab’Saber AN (1988) O Pantanal Mato-Grossense e a teoria dos refúgios. Revista Brasileira de Geografia 50 especial: 9-57.

Adams KD and Wesnousky SG (1998) Shoreline processes and age of the Lake Lahontan high-stand in the Jessup embayment, Nevada. Geol Soc Am Bull 110:1318–1332.

Alfonsi RR and Paes de Camargo MB (1986) Condições Climáticas para a Região do Pantanal Mato-Grossense. In: Boock A Anais do 1 Simpósio sobre Recursos Naturais e Sócio-Econômico do Pantanal. EMBRAPA-CPAP 5:29-42.

Alho CJR, Lacher TE, Goncalves HC (1988) Environmental degradation in the Pantanal ecosystem. Bioscience 38 3:164-171.

Amorim MA, Turcq PFM, Turcq BJ, Cordeiro RC (2009) Origem e dinâmica da deposição dos sedimentos superficiais na Várzea do Lago Grande de Curuai, Pará, Brasil. Acta Amazônica 39 1:165-172.

Assine ML and Soares PC (2004) Quaternary of the Pantanal, west-central Brazil. Quat Int 114: 23–34.

Assine ML (2005) River avulsions on the Taquari megafan, Pantanal wetland, Brazil. Geomorphology 70:357–371.

Assine ML and Silva A (2009) Contrasting fluvial styles of the Paraguay river in the northwestern border of the Pantanal wetland, Brazil. Geomorphology 113:189- 199.

Barbiéro L, de Querioz Neto JP, Ciomai G, Sakamoto AY, Capellari B, Fernandes E, Valles V (2002) Geochemistry of water and ground water in the Nhecolandia, Pantanal of Mato Grosso, Brazil: Variability and associated processes. Wetlands 22:528–540.

Bartlett KB and Harriss RC (1993) Review and assessment of methane emissions from wetlands. Chemosphere 26:261-320.

233

Bates BC, Kundzewicz ZW, Wu S, Palutikof JP, Eds., (2008): Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change, IPCC Secretariat, Geneva, 210 pp.

Batista TCA, Volkmer-Ribeiro C, Darwich C, Alves LF (2003) Freshwater sponges as indicators of floodplain lake environments and of river rocky bottom in Central Amazônia. Amazoniana 18:525-549.

Cao M, Gregson K, Marshall S (1998) Global methane emission from wetlands and its sensitivity to climate change. Atmos Environ 32:3293–3299.

Chase CG, Sussman AJ, Coblentz DD (2010) Curved Andes: Geoid, forebulge, and flexure. Lithosphere 1:358-363.

Chauhan M and Gopal B (2001) Biodiversity and management of Keoladeo National Park (India): a wetland of international importance. Biodiversity in Wetlands: Assessment, Function and Conservation Volume 2. Backhuys Publishers, Leiden

Cohen AS (2003) Paleolimnology: the history and evolution of lake systems. Oxford UniversityPress, Oxford, UK

Cole MM (1960) Cerrado, caatinga and pantanal: the distribution and origin of the savanna vegetation of Brazil. Geogr Jour 126:168-179.

DeMaster DJ (1979) The Marine Budgets of Silica and Si32. PhD Thesis, Yale University,New Haven, CT, 308 pp.

Donders TH, Wagner F, Dilcher DL, Visscher H (2005) Mid- to late-Holocene El Niño-Southern Oscillation dynamics reflected in the subtropical terrestrial realm. PROC NATL ACAD SCI USA 102:10904–10908.

Espitalie J, Laporte JL, Madec M, Marquis F, Leplat G, Paulet J, Boutefeu A (1977) Methode rapide de caracterisation des roches meres, de leur potential petrolier et de leur degre d’evolution. Rev Inst Fr Pétrol 32:23-42.

Fellerhoff C, Voss M, Wantzen KM (2003) Stable carbon and nitrogen isotope signatures of decomposing tropical macrophytes. Aquat Ecol 37:361 – 375. 234

Forsberg BR, Devol AH, Richey JE, Martinelli LA, Dos Santos H (1988) Factors controlling nutrient concentrations in Amazon floodplain lakes. Limnol Oceanogr 33:41–56.

Furch K, Junk WJ, Dieterich J, Kochert N (1983) Seasonal variation in the major cation (Na, K, Mg, and Ca) content of the water of Lago Camaleao, an Amazonian floodplain lake near Manaus, Brazil. Amazoniana 8:75-89.

Gumbricht T, McCarthy TS, Merry CL (2001) The topography of the Okavango Delta, Botswana, and its tectonic and sedimentological implications. S Afr J Geol 104:243-264.

Hamilton SK, Sippel SJ, Calheiros DF, Melack JM (1997) An anoxic event and other biogeochemical effects of the Pantanal wetland on the Paraguay River. Limnol Oceanogr 42: 257–272.

Hamilton SK, Sippel SJ, Melack JM (2002), Comparison of inundation patterns among major South American floodplains. J Geophys Res 107(D20) 8038.

Hamilton SK (1999) Potential effects of a major navigation project (Paraguay- Parana Hidrovia) on inundation in the Pantanal floodplains. Regulated Rivers: Research and Manangement 15: 289–299.

Hamilton SK and Lewis WM (1987) Causes of seasonality in the chemistry of a lake on the Orinoco River floodplain, Venezuela. Limnol Oceanogr 32:1277–1290.

Heckman CW (1998) The Pantanal of Poconé. Kluwer, Den Haag

Henrichs SM (1992) Early diagenesis of organic matter in marine sediments: progress and perplexity. Mar Chem 39:119-149.

Horton BK and DeCelles PG (1997) The modern foreland basin system adjacent to the central Andes. Geology 25:895– 898.

Johnson TC and McCave IN (2008) Transport mechanism and paleoclimatic significance of terrigenous silt deposited in varved sediments of an African rift lake. Limnol Oceanogr 53:1622-1632. 235

Junk WJ, Bayley PB, Sparks RE (1989) The flood pulse concept in river-floodplain systems. Canadian Special Publication of Fisheries and Aquatic Sciences 106:110- 127.

Junk WJ, Nunes Da Cunha K, Wantzen KM, Petermann P, Strussmann C, Marques MI, Adis J (2006) Biodiversity and its conservation in the Pantanal of Mato Grosso, Brazil. Aquat Sci 68(3):278-309.

Lopes I, Minõ C, Del Lama S (2007) Genetic diversity and evidence of recent demographic expansion in waterbird populations from the Brazilian Pantanal. J Biol 67(4):849–857. de Magalhaes NW (1992) Conheca o Pantanal. Terragraph Artes e Informatica Sao Paulo.

Maltby E and Turner RE (1983) Wetlands of the world. Geog Mag 55:12-17.

Mayle FE, Beerling DJ, Gosling WD, Bush MB (2004) Responses of Amazonian ecosystems to climatic and atmospheric carbon dioxide changes since the last glacial maximum. Philos Trans R Soc Lond B Biol Sci 359:499-514.

McKenzie JA (1985) Carbon isotopes and productivity in the lacustrine and marine environment. In: Strumm W Chemical processes in lakes. 99–118

Meyers PA and Ishiwatari R (1993) Lacustrine organic geochemistry: an overview of indicators of organic matter sources and diagenesis in lake sediments. Org Geochem 20:867-900.

Meyers PA and Teranes JL (2001) In: Last WM and Smol JP (eds) Tracking Environmental Change Using Lake Sediments: Physical and Geochemical Methods. Springer, New York 2:239-269.

Mitsch WJ and Gosselink JG (2000) Wetlands, 3rd ed. John Wiley & Sons, Inc., New York, USA 236

Mitsch WJ, Nahlik A, Wolski P, Bernal B, Zhang L, Ramberg L (2009) Tropical wetlands: seasonal hydrologic pulsing, carbon sequestration, and methane emissions. Wetl Ecol Manag DOI 10.1007/s11273-009-9164-4.

Moreira-Turcq P, Jouanneau JM, Turcq B, Seyler P, Weber O, Guyot JL (2004) Carbon sedimentation at Lago Grande de Curuai, a floodplain lake in the low Amazon region: insights into sedimentation rates. Palaeogeograp Palaeoclimatol Palaeoecol 214(1-2):27–40. de Oliveira Bezerra MA and Mozeto AA (2008) Deposição de carbono orgânico na planície de inundação do Rio Paraguai durante o Holoceno médio. Oecol Bras 12(1):155-171. de Oliveira MD and Calheiros DF (2000) Flood pulse influence on phytoplankton communities of the southern Pantanal floodplain, Brazil. Hydrobiologia 427: 101- 112.

Parolin M, Volkmer-Ribeiro C, Stevaux JC (2007) Sponge spicules in peaty sediments as paleoenvironmental indicators of the Holocene in the upper Parana river, Brazil. Rev Bras Paleo 10:17-26.

Pinder L and Rosso S (1998) Classification and ordination of plant formations in the Pantanal of Brazil. Plant Ecol 136:151-165.

Por FD (1995) The Pantanal of Mato Grosso (Brazil). Kluwer, Dordrecht, Germany

Prance GT and Schaller GB (1982) Preliminary study of some vegetation types of the Pantanal, Mato Grosso, Brazil. Brittonia 34:228–251.

Soares AP, Soares PC, Assine ML (2004) Areiais e lagoas do Pantanal, Brasil: herança paleoclimática?. Rev Bras Geo 33(2):211-224.

Swarzenski PW, Baskaran M, Orem WH, Rosenbauer RJ (2006) Historical reconstruction of contaminant inputs in Mississippi River delta sediments. Estuar Coast Mar Sci 29:1094-1107.

237

Talbot MR (2001) Nitrogen Isotopes in Palaeolimnology. In: Last WM and Smol JP (eds) Tracking Environmental Change Using Lake Sediments: Physical and Geochemical Methods. Springer, New York 2:401-439.

Talbot MR and Johannessen T (1992) A high resolution palaeoclimatic record for the last 27,500 years in tropical West Africa from the carbon and nitrogen isotopic composition of lacustrine organic matter. Earth Planet Sci Lett 110:23-37.

Talbot MR and Livingstone DA (1989) Hydrogen index and carbon isotopes of lacustrine organic-matter as lake level indicators. Palaeogeogr Palaeoclimatol Palaeoecol 70:121-137.

Tricart J (1982) El Pantanal: Un ejemplo del impacto de la geomorfologia sobre el medio ambiente. Geografia 7(13-14): 37-50.

Ussami N, Shiraiwa S, Dominguez JML (1999) Basement reactivation in a sub- Andean foreland flexural bulge: the Pantanal wetland, SW Brazil. Tectonics 18(1):25–39.

Veloso HP (1972) Aspectos fito-ecológicos da Bacia do Alto Rio Paraguai. Instituto de Geografia São Paulo (USP), Brazil

Volkmer-Ribeiro C and Turcq B (1996) SEM analysis of siliceous spicules of a freshwater sponge indicate paleoenvironmental changes. Acta Microscópica 5:186-187.

Volkmer-Ribeiro C and Pauls SM (2000) Esponjas de água dulce (Porifera, Demospongiae) de Venezuela. Acta Biol Venez 20(1):1-28.

Volkmer-Ribeiro C (1992) The freshwater sponges in some peatbog ponds in Brazil. Amazoniana 12:317-335.

Wantzen KM, Machado FD, Voss M, Boriss H, Junk WJ (2002) Seasonal isotopic shifts in fish of the Pantanal wetland, Brazil. Aquat Sci 64:239-251.

238

Figures

Figure D.1: A = Regional map of the Brazilian Pantanal, the world’s largest neotropical wetland and the hydrological basin of the Upper Paraguay River (adapted from Assine and Silva 2009). Ten large perennial lakes mark the western floodplain of this river, straddling the border of Brazil and Bolivia. The rectangle outlines the extent of B. B = LANDSAT (2000) image of LG, LM, and BV, illustrating the position of the lakes with respect to the main channel of the PR and the highlands of the Serras do Amolar. C = bathymetric map and sample grid for LG. Sediment core locations are marked with stars. D = bathymetric map and sample grid for LM. Heavy red lines indicate the location of high-relief lake margins and rocky shorelines; orange lines mark sand bars. E = bathymetric map and sample grid for BV.

239

Figure D.2: Concentration maps of mean particle size, biogenic silica, total nitrogen and total organic carbon for LG, LM, and BV. Top row = LG. Middle row = LM. Bottom row = BV. The contour interval is the same for each of the lakes, illustrating the elevated productivity and burial of OM in LM versus the other lakes.

240

Figure D.3: Cross plots of key biogeochemical indicators in the surface sediments of LG, LM, and BV. A = TN vs. TOC. Sediments in LM are particular rich in carbon and nitrogen, suggesting elevated rates of primary productivity. B = δ13C vs. C/N. Note the elevated C/N values exhibited by BV sediments, probably due to an increased role of vascular plant material in the basin’s carbon cycle.

Sediments from LG fall within well-defined ranges for mixed algal and C3- terrestrial vegetation. C = δ15N vs. δ13C. Low δ15N values in some LM sediments likely results from the presence of nitrogen-fixing cyanobacteria in the basin. D = HI vs. OI. Rock Eval pyrolysis data support a mixed organic provenance interpretation for lakes in the Pantanal.

241

Figure D.4: 210Pb profiles for sediment cores from LG. Location of cores presented in Figure 1. Sedimentation rates among floodplain lakes are highly variable, even among the large lakes of the Pantanal. Well-behaved decay of excess 210Pb in LG is likely due to the lake’s water depth, which at many locales exceeds the critical wave depth.

242

Figure D.5: Common fluvial and lacustrine sponge spicules encountered in lake bottom sediments from the study lakes. A = gemmosclere of the lotic sponge Corvospongilla secktii. B = gemmosclere of the lotic sponge Oncosclera navicella. C = gemmosclere of the lentic sponge Radiospongilla amazonensis. D = gemmosclere of the lentic sponge Trochospongilla variabilis. E = gemmosclere of the lotic sponge Uruguaya corallioides. F = megasclere of the lotic sponge Corvospongilla secktii. G = megasclere of the lentic sponge Heteromeyenia sp. H = megasclere of the lentic sponge Corvoheteromeyenia sp. I = megasclere of the lentic sponge Metania spinata.

243

Figure D.6: Conceptual model of sedimentation in the floodplain lakes of Pantanal, based on modern inferences. See text for details.

Supplemental Figure D.1: A = composite mean surface precipitation rate (mm/day), DJF for tropical South America. B = composite mean surface precipitation rate (mm/day), JJA. C = Average monthly precipitation and temperature values from northern Pantanal station data (Cuiabá, Brazil). D = Average monthly precipitation and temperature values from central Pantanal station data (Corumbá, Brazil). Locations of Cuiabá and Corumbá are marked in Figure 1.

244

Supplemental Figure D.2: Smear slides showing physical and biogenic components of lake bottom sediments from LG, LM and BV. Images A through F were collected using the 40x objective. A, B = surface sediments from LG. C,D = surface sediments from LM. E,F = surface sediments from BV. G,H = (100x) images of sediments from LM, illustrating green algae types, phytoliths and cyanobacteria present in the samples.

245

Supplemental Figure D.3: Seasonal cycle of the Upper Paraguay River in the northern and central Pantanal, illustrating the time lag associated with passage of the flood pulse (modified after data presented in Heckman 1998). The study lakes track the curve for the central Pantanal, experiencing a ‚fill-phase‛ during the austral winter, not during the austral summer when precipitation in the region reaches its maximum.

246

Tables

Supplemental Table D.1: Morphological aspects of the large permanent floodplain lakes adjacent to the PR. Shoreline parameters calculated from GeoCover 2000 image data (https://zulu.ssc.nasa.gov/mrsid). Depth data for L. Castelo and L. Negra from de Oliveira Bezerra and Mozeto 2008. Depth data for L. Cáceras and L. Jacadigo were collected by our group in 2003.

Lake Lat. Lon. Inundated Inundated Fetch Shoreline Maximum name (°S) (°W) Area Perimeter (km; development depth (m) (km2) (km) direction) L.Orion 17.049 58.387 126 60 18; NW- 1.51 n/a SE L.Piranhas 17.360 58.160 84 60 11 NW- 1.85 n/a SE L.Uberaba 17.511 57.808 332 92 30; NW- 1.42 n/a SE L.Gaíva 17.788 57.714 97 55 11; NE- 1.57 4.7 SW L.Mandioré 18.138 57.550 154 74 22; NE- 1.68 4.0 SW B.Vermelha 18.392 57.508 147 69 18; NW- 1.61 3.5 SE L. Castelo 18.550 57.574 9 30 6; N-S 2.82 7.3 L. Cáceres 18.958 57.771 27 32 8; NW- 1.74 3.4 SE L. Negra 19.074 57.531 10 14 4; N-S 1.78 2.3 L. Jacadigo 19.215 57.773 96 70 11; NW- 2.02 0.3 SE

247

Supplemental Table D.2: Hydrochemistry of the study lakes. Data for the PR from Hamilton et al. 1997. Units for major ions are given in mg/L, except where noted.

2+ + + 4+ - 2- - 3- - Name pH DO Cond. Ca Na K Si Cl SO4 NO3 PO4 CO3 (mg/L) (ppt) (mg/L) LG 8.3 9.9 0.05 5.3 3.7 2.6 9.4 0.9 na na 0.1 35.1 LM 8.2 9.3 0.04 4.5 3.3 2.9 23.7 0.7 4.3 4.4 0.3 25.5 BV 7.9 9.5 0.04 4.6 3.9 2.5 23.9 0.6 1.0 6.8 0.2 30.5 PR 6.9 4.5 0.04 5.0 1.6 1.7 7.5 0.4 0.1 na na 30.9

248

Supplemental Table D.3: Location, sedimentology and geochemistry of surface sediment samples from LG, LM, BV.

13 15 Sample Lat. Lon. Depth Mean BiSi TOC TN δ Com δ Nom name (°S) (°W) (m) D (wt. %) (wt. %) (wt. (‰) (‰) (μm) %) LG-1 17.677 57.691 1.6 55 1.2 0.9 0.1 -27.1 2.8 LG-2 17.677 57.706 1.9 76 1.8 1.1 0.1 -27.2 2.9 LG-3 17.691 57.705 2.4 37 0.7 1.1 0.1 -27.4 3.3 LG-6 17.706 57.706 2.9 48 1.3 1.1 0.1 -27.1 3.2 LG-7 17.720 57.707 2.4 102 1.0 1.9 0.2 -27.0 2.7 LG-9 17.733 57.689 2.8 45 2.3 2.1 0.2 -27.2 3.2 LG-10 17.736 57.704 3.3 24 2.2 0.6 0.1 -26.6 3.0 LG-11 17.737 57.721 3.6 67 1.6 1.3 0.1 -27.3 3.1 LG-12 17.737 57.737 3.2 50 0.8 0.8 0.1 -27.5 2.9 LG-13 17.742 57.746 3.2 55 0.8 0.8 0.1 -27.6 3.0 LG-14 17.754 57.752 2.9 46 1.9 0.6 0.1 -25.2 3.1 LG-15 17.755 57.739 4.4 53 0.8 0.8 0.1 -27.1 3.4 LG-16 17.753 57.722 4.4 88 1.3 1.0 0.1 -27.4 3.6 LG-17 17.754 57.705 4.3 43 2.3 2.5 0.3 -27.0 3.8 LG-21 17.768 57.571 2.5 22 1.0 1.1 0.1 -25.7 4.4 LG-22 17.768 57.737 4.5 35 1.1 1.6 0.2 -27.2 3.5 LG-23 17.768 57.722 4.4 18 2.6 2.7 0.3 -27.5 3.3 LG-24 17.769 57.706 4.5 11 2.0 3.2 0.4 -27.3 3.2 LG-25 17.785 57.689 3.9 18 2.1 2.6 0.2 -26.8 3.0 LG-26 17.783 57.706 4.7 9 1.4 2.9 0.3 -27.2 3.0 LG-27 17.784 57.722 4.4 12 1.4 2.6 0.3 -27.5 3.1 LG-28 17.784 57.737 4.3 24 1.0 2.0 0.3 -27.4 3.1 LG-29 17.784 57.754 3.9 44 1.5 0.8 0.1 -27.4 3.7 LG-30 17.798 57.753 3.4 25 2.0 1.5 0.2 -27.0 3.2 LG-31 17.798 57.739 4.2 33 1.9 1.5 0.2 -27.3 3.6 LG-32 17.799 57.721 4.4 23 2.1 1.9 0.2 -27.3 3.3 LG-33 17.798 57.705 4.3 18 1.9 2.4 0.3 -27.3 3.3 LG-35 17.807 57.688 3.8 45 0.5 1.9 0.2 -26.7 3.6 LG-36 17.814 57.705 3.6 54 0.6 0.9 0.1 -27.9 2.3 LG-40 17.710 57.713 1.6 25 1.9 1.4 0.1 -27.1 2.9 LG-41 17.690 57.713 1.8 32 2.5 1.8 0.2 -26.9 2.9 LG-42 17.666 57.699 1.7 40 1.7 1.5 0.1 -27.4 2.8 LG-43 17.768 57.694 3.6 13 2.7 2.8 0.3 -27.2 3.7 LG-44 17.752 57.696 3.4 54 2.2 2.8 0.2 -23.6 0.3 LM-1 18.042 57.608 2.1 35 5.2 4.0 0.3 -24.0 2.7 LM-2 18.042 57.600 2.3 38 2.7 2.6 0.2 -24.5 2.7 249

LM-5 18.048 57.570 2.5 69 2.7 1.8 0.2 -25.1 2.5 LM-6 18.049 57.580 2.9 34 3.8 5.4 0.5 -24.4 1.5 LM-7 18.052 57.588 2.9 29 4.8 6.4 0.6 -24.1 1.2 LM-8 18.054 57.597 2.6 40 4.3 3.4 0.3 -24.2 1.8 LM-9 18.057 57.606 2.5 31 5.6 3.4 0.3 -24.3 2.8 LM-10 18.075 57.601 2.8 32 3.8 4.9 0.4 -24.2 1.6 LM-11 18.071 57.590 3.3 36 4.4 7.0 0.7 -24.3 1.4 LM-12 18.069 57.581 3.4 28 4.0 6.5 0.7 -23.1 1.2 LM-13 18.067 57.572 3.4 47 4.7 5.5 0.6 -22.9 0.8 LM-14 18.064 57.562 4.0 5 6.5 6.0 0.6 -24.5 2.4 LM-15 18.080 57.553 3.7 22 6.1 6.1 0.7 -24.7 2.0 LM-16 18.083 57.563 3.4 26 4.0 6.0 0.7 -22.7 1.2 LM-17 18.087 57.573 3.3 47 5.2 5.5 0.6 -23.2 1.0 LM-18 18.089 57.584 3.3 28 5.2 7.4 0.8 -22.9 0.4 LM-19 18.093 57.595 2.8 33 4.6 6.8 0.6 -24.4 1.7 LM-20 18.110 57.590 2.7 25 4.8 4.3 0.4 -23.4 1.4 LM-21 18.104 57.578 3.4 33 3.0 8.5 1.0 -23.2 0.3 LM-22 18.101 57.567 3.5 31 5.2 6.7 0.7 -23.4 0.9 LM-23 18.099 57.557 3.4 25 4.9 6.2 0.7 -23.6 1.2 LM-24 18.096 57.546 3.4 33 3.7 6.6 0.8 -23.2 1.2 LM-25 18.090 57.535 3.0 24 4.9 6.3 0.6 -25.0 2.5 LM-26 18.085 57.528 2.4 21 5.8 4.2 0.4 -24.0 2.4 LM-27 18.109 57.528 3.1 26 3.8 5.7 0.6 -24.3 1.9 LM-28 18.113 57.540 3.4 26 4.0 6.3 0.7 -23.2 1.0 LM-29 18.115 57.541 3.4 27 5.7 6.6 0.7 -24.1 0.4 LM-30 18.118 57.562 3.5 26 4.7 6.7 0.7 -23.1 0.9 LM-31 18.122 57.572 3.6 20 8.2 6.6 0.6 -22.8 0.3 LM-32 18.126 57.580 3.3 18 7.7 6.5 0.6 -24.3 2.2 LM-33 18.144 57.579 3.2 29 4.3 8.3 0.8 -23.5 0.7 LM-34 18.141 57.565 3.5 16 8.9 5.8 0.6 -22.3 0.2 LM-35 18.136 57.553 3.7 23 4.5 8.4 0.8 -23.1 0.6 LM-36 18.133 57.543 3.4 46 4.2 5.4 0.6 -23.2 0.0 LM-37 18.129 57.533 3.4 42 2.2 4.5 0.5 -22.5 0.4 LM-38 18.124 57.521 3.2 42 2.3 3.0 0.3 -24.4 1.4 LM-39 18.148 57.523 3.2 78 2.0 1.8 0.2 -24.0 1.0 LM-40 18.151 57.534 3.5 77 1.7 3.0 0.4 -22.8 0.4 LM-41 18.155 57.546 3.6 30 3.5 6.8 0.8 -23.5 1.1 LM-42 18.159 57.560 3.5 29 7.4 6.9 0.7 -22.7 0.1 LM-43 18.163 57.574 3.8 21 7.9 6.7 0.6 -24.7 2.3 LM-44 18.168 57.586 3.0 29 5.6 6.6 0.7 -26.9 3.3 LM-45 18.187 57.581 3.4 16 6.3 6.8 0.7 -24.5 1.8 LM-45 18.187 57.580 3.4 15 6.8 6.4 0.7 -23.8 0.9 250

LM-46 18.182 57.565 3.6 22 7.9 6.0 0.6 -22.8 0.1 LM-47 18.179 57.552 3.6 21 6.0 6.1 0.7 -23.0 0.1 LM-48 18.175 57.537 3.5 63 3.2 4.6 0.5 -23.6 0.4 LM-49 18.172 57.524 3.2 77 1.5 4.1 0.5 -24.2 -0.1 LM-50 18.169 57.513 3.0 103 1.4 1.5 0.2 -24.7 0.7 LM-51 18.184 57.502 2.4 71 1.2 0.7 0.1 -24.9 2.2 LM-52 18.187 57.509 2.6 81 1.2 1.7 0.2 -24.9 1.3 LM-53 18.195 57.508 2.3 72 1.2 1.0 0.1 -24.0 1.9 LM-54 18.190 57.520 2.7 84 1.0 1.1 0.1 -24.5 1.5 LM-55 18.193 57.531 3.0 44 1.3 3.4 0.4 -24.6 1.3 LM-56 18.196 57.547 3.4 16 6.2 5.7 0.6 -23.5 0.8 LM-57 18.200 57.560 3.4 17 5.6 6.2 0.6 -24.0 1.3 LM-58 18.205 57.575 3.3 13 5.9 6.4 0.7 -24.3 1.3 LM-59 18.208 57.587 2.9 20 5.6 7.3 0.8 -25.0 0.9 LM-60 18.221 57.589 2.0 45 3.4 2.1 0.2 -25.0 3.2 LM-62 18.219 57.553 3.0 28 3.3 5.8 0.6 -25.1 1.3 LM-63 18.214 57.540 2.8 24 2.0 3.3 0.3 -24.8 2.3 LM-64 18.229 57.554 2.6 36 5.3 4.7 0.5 -25.5 2.3 LM-65 18.229 57.568 3.4 102 7.4 2.7 0.2 -26.0 2.6 LM-66 18.232 57.581 2.6 38 12.9 5.5 0.5 -25.6 2.5 LM-67 18.196 57.515 2.6 39 2.1 3.4 0.3 -25.4 2.2 LM-68 18.188 57.497 1.8 77 1.0 0.8 0.1 -23.0 1.9 LM-69 18.199 57.530 2.8 40 1.5 3.2 0.3 -25.4 2.4 BV-1 18.350 57.585 2.7 14 3.5 3.0 0.3 -24.5 2.6 BV-2 18.341 57.580 3.0 9 3.3 3.9 0.4 -25.2 2.1 BV-3 18.331 57.575 2.5 15 2.6 3.5 0.4 -24.0 2.5 BV-4 18.337 57.555 2.4 18 3.4 3.2 0.3 -24.8 2.2 BV-5 18.342 57.558 2.9 19 2.3 3.3 0.3 -25.9 2.9 BV-6 18.351 57.567 3.0 19 1.7 2.4 0.2 -26.3 3.0 BV-7 18.362 57.572 3.1 19 2.1 2.9 0.3 -26.6 3.0 BV-10 18.352 57.543 2.5 41 1.9 1.9 0.2 -26.3 2.2 BV-11 18.343 57.537 2.3 32 3.0 4.4 0.4 -26.5 2.1 BV-12 18.355 57.519 2.1 44 3.8 2.5 0.2 -27.4 1.5 BV-13 18.362 57.527 2.4 69 1.3 1.4 0.1 -26.5 2.3 BV-14 18.371 57.535 2.9 101 1.1 1.0 0.1 -26.9 2.7 BV-15 18.381 57.543 3.4 76 2.6 1.7 0.2 -26.9 3.2 BV-17 18.393 57.525 3.4 59 1.4 1.2 0.1 -27.0 2.7 BV-18 18.383 57.518 3.3 70 1.3 1.5 0.2 -27.1 3.2 BV-19 18.373 57.511 2.7 104 1.4 0.7 0.1 -26.8 3.2 BV-20 18.363 57.504 2.3 108 1.4 0.8 0.1 -25.8 2.9 BV-21 18.372 57.488 2.1 73 2.7 1.8 0.2 -25.9 2.1 BV-22 18.393 57.500 3.2 131 1.2 1.5 0.1 -27.2 2.5 251

BV-23 18.382 57.493 2.7 93 1.5 1.4 0.1 -26.9 3.1 BV-24 18.404 57.507 3.5 73 1.2 1.1 0.1 -26.8 2.9 BV-25 18.414 57.513 3.4 75 1.3 1.5 0.2 -27.1 3.3 BV-26 18.426 57.495 3.2 120 0.6 0.9 0.1 -27.4 2.9 BV-27 18.416 57.489 3.3 102 0.8 1.1 0.1 -27.3 2.7 BV-28 18.406 57.483 3.0 112 0.6 1.9 0.2 -27.1 2.3 BV-29 18.395 57.477 3.1 92 0.5 1.0 0.1 -27.0 2.4 BV-30 18.384 57.470 2.7 70 0.9 1.4 0.1 -27.5 2.3 BV-31 18.394 57.454 2.4 73 1.5 1.1 0.1 -25.6 2.3 BV-32 18.405 57.459 2.4 65 1.2 1.2 0.1 -26.7 2.3 BV-33 18.427 57.472 2.8 88 0.7 3.6 0.3 -24.9 2.0 BV-34 18.437 57.478 3.1 96 0.6 1.6 0.2 -27.6 2.7 BV-35 18.456 57.473 4.4 113 0.8 1.8 0.2 -28.0 2.3 BV-36 18.447 57.464 2.7 64 1.0 1.8 0.2 -27.9 2.1 BV-38 18.427 57.450 2.8 60 0.5 2.6 0.3 -27.3 2.2 BV-39 18.452 57.455 2.8 95 1.6 1.3 0.1 -27.8 2.1 BV-40 18.433 57.439 2.4 55 4.3 5.5 0.5 -27.6 1.8 BV-41 18.419 57.435 2.0 114 1.9 1.6 0.1 -25.4 2.3 BV-42 18.420 57.470 2.1 129 1.1 2.2 0.2 -24.1 2.8 BV-44 18.357 57.582 1.9 255 0.3 3.9 0.2 -24.7 1.8 BV-46 18.349 57.570 3.1 20 0.5 2.3 0.2 -26.2 2.8 BV-47 18.359 57.524 3.2 43 0.7 2.0 0.2 -26.8 3.0 BV-49 18.363 57.531 2.4 79 0.7 1.6 0.2 -25.9 2.3

252

Supplemental Table D.4: Sponge microfossil analysis for select sub-samples from LG, LM and BV. +++ = abundant. ++ = common. + = minor. - = rare. Ø = absent.

Sample Sponge Megascleres Megascleres Microscleres Gemmoscleres Diatoms Phytoliths Species whole fragment LG 07 C. seckti ++ +++ Ø + + + R. amazonensis LG 10 M. spinata ++ +++ Ø ++ + + U. corallioides Heteromeyenia sp. T. variabilis C. seckti LG 29 T. variabilis +++ +++ Ø ++ Ø Ø U. coralioides LG 30 T. variabilis +++ +++ Ø ++ - ++ R. amazonensis LG 36 U. corallioides ++ ++ Ø - Ø Ø

LG 44 U. corallioides ++ ++ Ø + Ø Ø

LM 06 None - ++ Ø Ø +++ ++

LM 18 T. variabilis - ++ Ø - ++ +

LM 31 None Ø ++ Ø Ø +++ +

LM 40 None - ++ Ø Ø ++ -

LM 49 None - - Ø Ø +++ ++

LM 53 R. amazonensis - ++ Ø + +++ ++ O. navicella LM 60 R. amazonensis ++ ++ Ø + ++ + O. navicella LM 62 C. seckti +++ +++ Ø - Ø + O. navicella BV 02 T. variabilis Ø ++ Ø ++ - ++ Corvoheteromeyenia sp. BV 04 Corvoheteromeyenia + +++ Ø ++ + + sp. R. amazonensis C. seckti BV 06 C. seckti ++ ++ Ø Ø Ø Ø

BV 12 C. seckti ++ ++ Ø - - + Corvoheteromeyenia sp. BV 29 C. seckti ++ ++ Ø - Ø + Corvoheteromeyenia sp. (rare) BV 33 R. amazonensis ++ ++ Ø + - - C. seckti

253

Supplemental Table D.5: Surface sediment samples collected from L. Cáceras and L. Jacadigo, southern Pantanal. Loss-on-ignition analyses indicate the presence of minor calcite.

Lake Name Sample ID Lat (°S) Lon (°W) LOIoc LOIic

L. Cáceres CA-03-01 18.979 57.751 11.14 0.43

L. Cáceres CA-03-02 18.968 57.754 10.71 0.42

L. Cáceres CA-03-03 18.970 57.768 9.46 0.39

L. Cáceres CA-03-04 18.965 57.776 9.84 0.18

L. Cáceres CA-03-05 18.954 57.772 7.44 0.27

L. Cáceres CA-03-06 18.949 57.758 10.87 0.37

L. Cáceres CA-03-08 18.960 57.778 9.72 0.29

L. Cáceres CA-03-09 18.946 57.776 14.08 0.45

L. Jacadigo JCD-03-02 19.212 57.820 15.62 0.26

L. Jacadigo JCD-03-03 19.208 57.816 18.32 0.04

L. Jacadigo JCD-03-04 19.204 57.813 16.84 0.48

L. Jacadigo JCD-03-05 19.202 57.808 16.99 0.81