Calibration and Interpretation of Holocene Paleoecological Records of Diversity from Lake Tanganyika, East Africa
Item Type text; Dissertation-Reproduction (electronic)
Authors Alin, Simone Rebecca
Publisher The University of Arizona.
Rights Copyright © is held by the author. Digital access to this material is made possible by the Antevs Library, Department of Geosciences, and 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 or the department.
Download date 05/10/2021 07:51:11
Link to Item http://hdl.handle.net/10150/231412 CALIBRATION AND INTERPRETATION OF HOLOCENE PALEOECOLOGICAL
RECORDS OF DIVERSITY FROM LAKE TANGANYIKA, EAST AFRICA
by
Simone Rebecca Alin
Copyright © Simone Rebecca Alin 2001
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
2001
r,y . .a. i`- { . +. -- /i THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE
As members of the Final Examination Committee, we certify that we have read the dissertation prepared bySimone Rebecca Alin entitled Calibration and Interpretation of Holocene Paleoecological
Records of Diversity from Lake Tanganyika, East Africa
and recommend that it be accepted as fulfilling the dissertation requieurent for t Degree of Doctor of Philosophy
'/57/Gi An re, Cohen Date C Q CA (r 2svF Karl Flessa Date l /C)/ Jonathan i1Ferpeck Date
Peter Rein th Date
Viz& I l Robert Robichaux Date
Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared undermy direct'on and recommend that it be accepted as fulfilling the dissertation requi t- ment . ,,-Th
l/ lL7 /6l Dissertation Director Andresv Cohen Dat'e 3
STATEMENT BY AUTHOR
This dissertation 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 dissertation are allowable without special permission, provided that accurate acknowledgment 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 copyright holder. 4
ACKNOWLEDGMENTS
This dissertation could not have been completed without the assistance and encouragement of many individuals and organizations. I was fortunate to have a supportive, accessible, and inspiring dissertation committee. I am grateful to Andrew Cohen for giving me the opportunity to work on a truly exceptional ecosystem and to live in Africa, a long -time dream of mine. His enthusiasm for Lake Tanganyika is contagious, and his breadth of experience there is impressive. Karl Flessa has consistently offered excellent advice and feedback; his tutelage in the art of scientific communication has been particularly valuable. Peter Reinthal and Rob Robichaux brought an important ecological perspective to my committee and have served as inspiring examples to me of people who have successfully melded their academic interests in biological conservation with effective field efforts. Jonathan Overpeck provided valuable perspective on paleoclimatic /global change issues. Les Kaufman consistently asked difficult, thought -provoking questions. I also thank Lisa Graumlich for serving as an excellent role model and supportive mentor early in my graduate career; Owen Davis for help on numerous occasions; John Lundberg and Lucinda McDade for rigorous classes, discussions, and field excursions; David Dettman for excellent guidance on all things carbon; and Julie Cole for stimulating discussions and seminars. Catherine O'Reilly's perseverance was essential in ensuring regular data collection in the field. In addition, Jeff Houser, Ruben Shapola, Kimambo, Chatta, Ibrahim, Chale, and Tony Thompson facilitated our safe diving. Mr. Chitamwebwa generously allowed me to use Tanzanian Fisheries Research Institute facilities in Kigoma. The Lake Tanganyika Biodiversity Project, particularly Kelly West, Graeme Patterson, and Andy Menz; Tanzanian Commission for Science and Technology; Tanzanian Immigration Agency; and Tanzania National Parks provided invaluable logistical assistance and necessary permits. I thank the Department of Geosciences staff and Paul Pronze and Dorian Voorhees at the Graduate College for logistical and bureaucratic assistance; Blackwell Publishing for permission to include Appendix A in my dissertation; and Vera Markgraf and Alex Wolfe for providing me laboratory space at INSTAAR (CU- Boulder). I received generous financial support for my dissertation research froman NSF Graduate Research Fellowship, Analysis of Biological Diversification RTG Fellowship, National Security Education Program Graduate International Fellowship, Graduate College Dean's Dissertation Fellowship, Wilson R. Thompson Scholarship, Geological Society of America, Graduate and Professional Student Council Travel Grant Fund, Graduate College Final Project Fund, Women in Science and Engineering Travel Grant Fund, organizers of several meetings, and NSF grants EAR -9627766 and ATM -9619458. Finally, I am tremendously grateful for the intellectual stimulation, moralsupport, and recreational opportunities provided by Annika Alin, Jonathan Fost, Carmie Garzione, Damian Hodkinson, Joy and Nick Bader, Sharon Hall, Stacie Gibbins, Dena Smith, Catherine O'Reilly, Mark Rollog, Steve Young, Dave Goodwin, Tim Shanahan,Kris Kerry, Eric Jensen, Elise Pendall, and Nathan English,among others. Finally, I am deeply grateful to Winston Wheeler; his support, intelligence, andsense of humor have gotten me through the past few months with a smileon my face. 5
DEDICATION
I dedicate this dissertation to my parents, Elizabeth Atly, Jack Alin, Susie Alin, and Carlos Reyes, for their bottomless reserves of support for me. Their lifelong encouragement for me to follow my dreams, their friendship, and often their material support have made this endeavor not just possible, but very rewarding. I am especially grateful that my graduate studies gave me the opportunity to travel in Tanzania with both my father and my mother. 6
TABLE OF CONTENTS
Page
LIST OF FIGURES 10 LIST OF TABLES 11 ABSTRACT 12
CHAPTER 1: INTRODUCTION 14 Background and Statement of the Research Problem .14 Format of the Dissertation 17
CHAPTER 2: PRESENT STUDY 20
CHAPTER 3: A HIGH -RESOLUTION RECORD OF NATURAL ENVIRONMENTAL CHANGE AND PALEOECOLOGICAL RESPONSE IN LAKE TANGANYIKA, EAST AFRICA, SINCE THE LATE GLACIAL .24 Abstract 24 Introduction 25 Methods 28 Core Collection and Sampling 28 Laboratory Analyses 29 Results .33 Geochronology and Sedimentology Data 33 Elemental and Stable Isotope Data 35 Diagenesis 38 Paleoecological Data 39 Discussion 41 Sedimentary Environment and Preservation Conditions 41 Environmental Variability and Paleoecological Response .42 Temporal Trend in SOM Preservation: Signal of Degradation or Primary Productivity ? 48 Charcoal and Lake Level Change 49 Conclusions 52
CHAPTER 4: CONCLUSIONS 77 Microinvertebrate Paleoecology and Conservation 77 Ostracods as Indicators of Environmental Change ..79
APPENDIX A: EFFECTS OF LANDSCAPE DISTURBANCE ON ANIMAL COMMUNITIES IN LAKE TANGANYIKA, EAST AFRICA 85 Abstract 87 Introduction 88 Methods 89 7
TABLE OF CONTENTS-Continued
Page
Study Sites and Data Collection 89 Evaluation of Water Quality 90 Faunal Censuses 90 Data Analysis 91 Results 91 Water Quality 91 Faunal Censuses and Rarefaction 93
Fish Trophic Analyses .93 Similarity Indices and Community Structure ..93 Discussion 94 Similarity Indices and Community Structure ..95 Ecological Tolerances .95 Lake Basin Parameters and Conservation 96 Caveats 96 Conclusions 96 Acknowledgments 97 Literature Cited 97 Appendix 1 ..99 Appendix 2 .101 Appendix 3 102
APPENDIX B: THE LIVE, THE DEAD, AND THE VERY DEAD: TAPHONOMIC CALIBRATION OF THE PALEOECOLOGICAL RECORD IN LAKE TANGANYIKA, EAST AFRICA 104 Abstract ..104 Introduction 105 Background on Study Area 107 Study Area and Methods 109 Sampling 109 Data Analysis 111 Results 115 Sedimentological and Radiocarbon Data for Cores 115 Characteristics of Ostracod Assemblages ..116 Fidelity of Death and Fossil Assemblages to the Living .120 Analysis of Core Interval Re- Sampling 122 Ordination 123 Discussion .126 Preservation of Life Assemblage Attributes in Death and Fossil Assemblages.....126 Fidelity Metrics 129 Sampling Efficiency 130 8
TABLE OF CONTENTS-Continued
Page
Scaling Issues and Equivalence of Fish and Ostracod Community Dynamics...... 132 Novel Contributions of Paleoecological Observations to Conservation Biology... 135 Conclusions 136 Acknowledgments 136 References ..160
APPENDIX C: PALEOENVIRONMENTAL RECORDS OF LANDSCAPE DISTURBANCE AT GOMBE STREAM NATIONAL PARK, AFRICA 165 Abstract ..165 Body of Paper 166
Acknowledgments . 171 References . 182
APPENDIX D: LIST OF ABBREVIATIONS USED FOR SPECIES NAMES IN ALL
OSTRACOD SPECIES ABUNDANCE DATA MATRICES . 184
APPENDIX E: RAW DATA FROM CORE LT97 -56V COLLECTED FROM 56 M WATER DEPTH NEAR MGONDOZI, TANZANIA 187 Species abundance data for fossil assemblages ..187 Taphonomic and ostracod abundance data for LT97 -56V fossil assemblages 190 Grain size distribution for samples from core LT97 -56V 191 Loss -on- ignition data for core LT97 -56V 193 Geochemical data for core LT97 -56V <63 µm sediment fraction ..194 Smear slide data for core LT97 -56V, based on 200 -point counts ..195
APPENDIX F: RAW DATA FOR LIFE AND DEATH ASSEMBLAGES COLLECTED FROM SURFACE SEDIMENTS AT MWAMGONGO, TANZANIA, BETWEEN OCTOBER 1997 AND JULY 1999 196
Sample abbreviations for life and death assemblages and sediment grain size .196
Water depths at quadrat locations throughout the sampling interval .197 Life assemblage species abundance data 198 Death assemblage species abundance data 204
Taphonomic and ostracod abundance data for death assemblages .210 Grain size distribution for life and death assemblages.... 211
APPENDIX G: RAW DATA FOR CORE MWA -1 COLLECTED FROM 10 M WATER DEPTH AT MWAMGONGO, TANZANIA 213 Fossil assemblage species abundance data 213 Fossil species abundance data for multiple samples from core interval 0 -1cm 216 Taphonomic and ostracod abundance data for MWA -1 fossil assemblages 218 9
TABLE OF CONTENTS - Continued
Page
Grain size distribution for core MWA -1 219
APPENDIX if RAW DATA FOR CORE MWA -2 COLLECTED FROM 5 M WATER DEPTH AT MWAMGONGO, TANZANIA 220 Fossil assemblage species abundance data ..220 Taphonomic and ostracod abundance data for MWA -2 fossil assemblages ..222 Grain size distribution for core MWA -2 223
APPENDIX I: RAW DATA FOR CORE MIT -1 COLLECTED FROM 15 M WATER DEPTH AT GOMBE STREAM NATIONAL PARK, TANZANIA .224 Species abundance data for fossil assemblages in MIT -1 ..224 Grain size distribution for samples from core MIT -1 .227
APPENDIX J: SMEAR SLIDE DATA FOR DEEPWATER CORES FROM GOMBE STREAM NATIONAL PARK AND MWAMGONGO, TANZANIA .228
REFERENCES 229 10
LIST OF FIGURES Page
Figure 3.1, Locations of all cores collected from Lake Tanganyika 55 Figure 3.2, Depth versus radiocarbon age profile for core LT97 -56V 57 Figure 3.3, Depth -calendar age profile for core LT97 -56V from Lake Tanganyika 59 Figure 3.4, Stratigraphy for cores LT97 -56V, LT97 -53V, and LT97 -35V ..61 Figure 3.5, Paleoenvironmental records from core LT97 -56V 63 Figure 3.6, Crossplots of elemental and stable isotopic data for LT97 -56V 65 Figure 3.7, Taphonomic data for ostracods and ostracod environmental indices 67 Figure 3.8, Species abundance, diversity, dominance, and ordination data for ostracods 69 Figure 3.9, Correspondence between the ostracod water depth index (OWDI) and the
stromatolitic lake level record .71 Figure 3.10, Compilation of paleoenvironmental interpretations and lake level curve... 73 Figure A.1, Map of study sites and distribution of rocky habitats 89 Figure A.2, Depth profiles of species density, species richness, and abundance ....93 Figure A.3, Depth profiles of fish trophic group abundance data 94
Figure A.4, Rank abundance curves for fish, mollusc, and ostracod communities .95 Figure B.1, Map of study sites and water depth curve for sampling period 138 Figure B.2, Species richness and accumulation curves for ostracod assemblages .140
Figure B.3, Species abundance histograms for all ostracod assemblages ..142 Figure B.4, Species occurrence frequency histograms for all ostracod assemblages.....144 Figure B.5, Species occurrence frequency histograms for ostracod life assemblages....146 Figure B.6, Rank abundance histograms for all ostracod assemblages and Spearman's rank order and Pearson's correlation test results for varying numbers of species .148 Figure B.7, Species accumulation curves for core MWA -1 150 Figure B.8, Ordination plots of ostracod life, death, and fossil assemblage samples.....152 Figure C.1, Geochronology, paleoenvironmental, and faunal data for deepwater cores.173 Figure C.2, Sedimentology and paleoecology of shallow -water cores 176 11
LIST OF TABLES Page,
Table 3.1, Radiocarbon data for core LT97 -56V from Lake Tanganyika 75 Table 3.2, Details of cores discussed in text 76 Table A.1, Analysis of variance for turbidity, species density, rarefied species richness, and abundance among study sites 92 Table A.2, Results of pairwise t tests of turbidity, species density, rarefied species richness, and abundance among study sites 92 Table A.3, Jaccard and Simpson indices of similarity for all taxonomic groups .94 Table B.1, Radiocarbon results for core MWA -1 from Mwamgongo, Tanzania 154 Table B.2, Fidelity of ostracod death and fossil assemblages to life assemblages .155 Table B.3, Ten most abundant species in life, death, and fossil assemblages 156 Table B.4, Average species abundance versus probability of detection 157 Table B.5, Occurrence frequency of species in resampled core interval 158 Table B.6, Predicted overlap in species composition within a core sample 159
Table C.1, Details of cores .178 Table C.2, Radiocarbon dates for Gombe Stream National Park and Mwamgongo 179 12
ABSTRACT
Lake Tanganyika is a complex, tropical ecosystem in East Africa, harboring an estimated 2,100 species. Extensive watershed deforestation threatens the biodiversityand ecological integrity of the lake. In this dissertation, ecological and paleoecological methods were employed to study the distribution of invertebrate biodiversity through space and time, with particular emphasis onlinkages between biodiversity and land -use patterns. Ecological surveys of fish, mollusc, and ostracod crustacean diversity at sitesin northern Lake Tanganyika representing different levels of watershed disturbance revealed
a negative correlation between biodiversity and intensity ofwatershed disturbance.
To elucidate the long -term relationship between disturbance and biodiversity,
paleoecological records of invertebrates offshore from watersheds experiencing different
degrees of anthropogenic disturbance were examined. Life, death, and fossil assemblages
of ostracod valves were compared to assess the reliability and natural variability inherent
to the paleoecological record. These comparisons indicated that paleoecological (i.e.
death and fossil) assemblages reliably preserve information on species richness,
abundance, and occurrence frequency at comparable -to- annual resolution. Unlike life
assemblages, species composition of paleoecological assemblages reflects input of
species from multiple habitat types. Ostracod paleoecological assemblages are
characterized by spatiotemporal averaging that renders them representative of larger areas
and longer time spans than life assemblages. Thus, paleoecological assemblages provide
an efficient means of characterizing longer -term, site -average conditions. 13
Natural variability in ostracod fossil assemblages from a sediment core representing the Late Glacial to the present indicates that abundance of individual ostracod species is highly variable. Ostracod assemblages were preserved in only the most recent 2,500 years of sediment. Species composition of ostracod assemblages reflects lake water depth. Core geochemical data indicate that the coring site may have been below the oxycline for 2,000 years, inhibiting ostracod survival and preservation.
Paleoecological, sedimentological, and stable isotope data revealed differences in biodiversity and watershed disturbance through time offshore from a pair of sites. The protected site is offshore from Gombe Stream National Park (Tanzania), the other offshore from a deforested watershed outside the park. Offshore from the deforested watershed, sedimentation rates increased, and turnover in ostracod species composition occurred during the past 50 years. Comparable changes were not observed offshore from the park. 14
CHAPTER 1: INTRODUCTION
Background and Statement of the Research Problem
Lake Tanganyika in the East African Rift is the oldest (>10 Ma), deepest (1,470 m), most biologically diverse, and one of the largest (32,600 km2) tropical lakes in the world.
The lake basin is composed of a series of alternating half grabens, such that benthic habitats in the lake are characterized by both structural diversity and isolation between adjacent habitat patches (Michel et al., 1992). A result of the lake's great antiquity, size, and complex habitat structure is that it has served as a "natural laboratory for evolution" and harbors a tremendous diversity of endemic species (cf. Brown, 1994; Coulter, 1991;
Martens, 1985; Michel et al., 1992; Park and Martens, 2001; Penney and Racek, 1968;
Poll, 1986; Rome, 1962; Wouters, 1988; Wouters and Martens, 1992; Wouters and
Martens, 1994; Wouters and Martens, 1999). A relatively recent tally indicates that Lake
Tanganyika houses at least 2000 species of animals, plants, and protozoa (Coulter, 1994).
Since its "discovery" by Europeans in the mid- 1800s, Lake Tanganyika has sparked worldwide scientific interest because it is a storehouse of potential informationon evolutionary patterns and processes, with ongoing projects applying techniquesas diverse as molecular biology, seismic imaging, behavioral ecology, morphometrics, and stable isotope ecology to problems in evolutionary biology (e.g. Cohen et al., 1997a;
Sturmbauer and Meyer, 1992).
In addition to being a scientific asset of global caliber, Lake Tanganyika is pivotalto the regional economy of East Africa, providing freshwater, food, andone of the most reliable means of transportation in a logistically- challengedcorner of the world. A 15 handful of open water fish species are extremely important to the regional economy both as important sources of animal protein and as an economic basis for local traditional fisheries. In addition, many species of benthic Tanganyikan cichlids are prized in the international aquarium trade.
Human populations in the four countries bordering the lake (Burundi, Tanzania,
Zambia, and Zaire) are among the most rapidly growing in the world. Anthropogenic threats to the long -term survival of the lake's evolutionarily unique and economically important species include the consequences of regional land- and lake -use patterns such as watershed deforestation and overfishing and the effects of global climate change on the lake ecosystem. Increased sediment influx from deforestation alters the species composition and productivity of aquatic ecosystems by altering habitat structure, availability of nutrients or toxins, and light penetration, which in turnmay change patterns of primary productivity (cf. Bootsma and Hecky, 1993; O'Reilly, 1999). Global climate change may disrupt current nutrient recycling processes by affecting the stability of the lake's thermal stratification (Plisnier, 2000).
Previous studies indicate that extent of watershed deforestation is negatively correlated with levels of biodiversity in adjacent lake habitats (Cohen et al., 1993; Alinet al., 1999; Appendix A). However, limited resources for monitoringa vast geographic area and the temporal variability of natural ecosystems make comprehensive assessments of the ecological health and integrity of communities within the lake intractable.Clearly, a longer -term perspective on the limnology and ecology of Lake Tanganyika is essential to unraveling its complex climatic, biological, and physicochemical historyat various time scales. 16
Using a suite of indicators to independently gauge the directions, magnitude, and
stimuli for natural and anthropogenic environmental change is a powerful approach to reconstructing environmental history from the sedimentary archives in lake basins, particularly in a place like Lake Tanganyika, where written records of limnological and
ecological conditions are sparse to the point of being anecdotal in many cases. Lake
environments, such as Tanganyika, provide nearly ideal circumstances for a retrospective
evaluation of changes in diversity and sedimentation rates through high- resolution
analysis of geochemical and paleoecological indicators in sediment cores. Sedimentation
rates in lakes are high, so the potential for preservation of a high quality fossil record is
relatively good. Therefore, lake sediments often provide an archive of environmental and
ecological change which is amenable to analysis on a variety of time scales, from human
to evolutionary.
However, insufficient taphonomic research addressing the preservational conditions
of lacustrine environments had been done to be able to interpret the sedimentary record of
biodiversity turnover in a quantitative context. In the course of my dissertation, I
attempted to fill this void by first calibrating the fidelity of the fossil record to living
communities, then by utilizing a variety of quantitative approaches to interpreting
paleoecological change in the Lake Tanganyika basin both over the time interval of
recent anthropogenic landscape alteration and over a somewhat longer interval to
establish baseline conditions of variability in invertebrate assemblages to compare with
paleoecological records from more densely populated and altered lakeshore habitats.
Further background information relevant to each study is detailed in Appendices A -D. 17
Format of the Dissertation
This dissertation consists of an article published in the journalConservation Biology
in 1999, which appears as Appendix Ato this dissertation; two prepublication
manuscripts, which appearas Appendices B and C; and an additional manuscript to be
submitted in the future for publication, whichappears as Chapter 3 of this dissertation.
For the paper entitled "Effects of landscape disturbanceon animal communities in
Lake Tanganyika, East Africa" (Appendix A), whichappeared in the journal
Conservation Biology in October 1999, I collaboratedwith researchers in the United
States, Africa, and Europe to investigate the relationshipbetween watershed disturbance
and species diversity in live communities of benthicfish, molluscs, and ostracod
crustaceans in Lake Tanganyika. Although I did not participate inthe field expedition
and data collection for this project,as it preceded my time in graduate school, I collated
all the data, performed all data analyses, andwrote the entire manuscript myself.
The manuscript in Appendix B is entitled "The live,the dead, and thevery dead:
taphonomic calibration of the recent record of paleoecologicalchange in Lake
Tanganyika, East Africa," which I will submit to the journalPaleobiology early in 2002.
I undertook this taphonomic calibration of the paleoecologicalrecord of biodiversity preservation by comparing life, death, and fossil assemblagesof ostracods froma site in northern Tanzania with the ultimate aim of utilizing thepaleoecological record to reconstruct the ecological history of various watersheds aroundthe lake experiencing different degrees of anthropogenic disturbance. Forthis project, I garnered themajority of my own funding for field work and analyses,spent nearly a year (1997 -98) in thefield 18
in Tanzania to collect monthly data (in cooperation with Catherine O'Reilly, also of the
Department of Geosciences), did all the laboratory work, performed all data analyses, and
wrote the manuscript myself. My major advisor, Andrew Cohen, and Catherine O'Reilly
collected the two short sediment cores discussed in this paper during a dive the following
summer field season. My co- author on this paper is Andrew Cohen.
After confirming the robustness of the lacustrine fossil record in the previous study, I
performed a high -resolution analysis of paleoenvironmental change and ecological
response in a core spanning the interval from the Mid Holocene to the present in Lake
Tanganyika (Chapter 3). This paper is entitled "A high -resolution record of natural
environmental change and paleoecological response in LakeTanganyika, East Africa,
since the Mid Holocene," and will be submitted to the journal Limnology and
Oceanography after addition of further substantiating data and refinement of the lake
level curve. For this project, the help of numerous people during the coring operation
was essential. After core collection and return to the United States, I sampled the core
and carried out the majority of analyses myself including radiocarbon sample preparation,
loss -on- ignition, stable isotope, elemental, grain size, and paleoecological analyses. Two
analyses, measurement of magnetic susceptibility and core x- radiography, were carried
out on a contractual basis by employees at the University of Rhode Island (laboratory of
John King) and the University of Miami (laboratory of Chris Scholz). These data are
auxiliary to the main body of the paper. My only co- author on this paper is my advisor.
Finally, I collaborated with several researchers from the Lake Tanganyika
Biodiversity Project to compare the ecological history for the past 250 years at a pair of 19
watersheds in northern Tanzania - one of them is protected in Gombe Stream National
Park, the other is outside the park and almost completely deforested. The resulting paper,
"Paleoenvironmental records of landscape disturbance at Gombe Stream National Park,
Africa," appears as a co- authored manuscript in Appendix C and will be submitted to
Nature in the near future. I collected all radiocarbon, sedimentology, and paleoecology
(ostracod assemblage) data for shallow -water cores (3 of 5 cores analyzed for this
project). Co- authors contributed nitrogen stable isotope and carbon:nitrogen ratio data
(Catherine O'Reilly), radiocarbon dates for deepwater cores (David Dettman),
paleoecology data for deepwater cores (Manuel Palacios -Fest), and21°Pbchronologies
for deepwater cores (Brent McKee). All data analyses were my own, except for
determination of 21°Pb sedimentation rates, and I wrote the entire manuscript. 20
CHAPTER 2: PRESENT STUDY
The methods, results, and conclusions of this study are presented in the papers
appended to this dissertation. The following is a summary of the most important findings
in these papers.
This dissertation represents research employing both ecological and paleoecological
methods aimed at studying distribution patterns of biodiversity in Lake Tanganyika, East
Africa, through various spatial and temporal scales. The first study examined the
relationship between watershed disturbance intensity and diversity of fish and
invertebrates living in rocky habitats offshore from the affected watershed. Study sites
representing low, moderate, and high watershed disturbance were located in the northern
basin of the lake, where gradients in human population density are steepest. Transects of
fish, mollusc, and ostracod (microscopic crustaceans) diversity were conducted at all sites
by scuba divers and a remotely operated vehicle at water depths between 1 and 80 meters.
Diversity in all three taxonomic groups generally correlated negatively with watershed
disturbance level, although the taxonomic groups appeared to have different response
thresholds to sediment inundation. Fishes and molluscs appeared to be more sensitive
than ostracods to sediment disturbance, thus ostracods may serve as conservative paleoecological indicators of benthic response to environmental change.
The second study of this dissertation focused on the reliability of the paleoecological record of ostracod diversity for reconstructing the ecological history of the living
assemblages that contribute skeletal remains to the sediment column through time.
Sediment samples were collected monthly for approximatelya year by scuba divers from 21 fixed rocky habitats at a site in northern Tanzania. Two short sediment cores were collected adjacent to the rocky habitats by divers operating hand coring devices. Patterns in species richness, abundance, and composition were compared among samples of live- collected ostracods, dead ostracod shells blanketing the surface of the rocky habitat
(death assemblages), and fossil ostracod remains contained in sediment core intervals.
The number of species present in death and fossil assemblages was generally comparable to the number of species of live individuals observed at each fixed site over the course of a year. Species abundance distributions in life assemblages were accurately represented in dead and fossil assemblages, although the dominant species differed from live assemblages. While rare species were not consistently detected with standard sample sizes of 500 individuals, it appeared that sampling sediment cores and death assemblages was an effective means of gauging the persistence of rare species in a community, which was not possible with live sampling only.
The third study of this dissertation used a long sediment core (LT97 -56V, 3.56 meters long, from 56 meters water depth) collected from a minimally disturbed region of the
Lake Tanganyika basin to study baseline (i.e., non -anthropogenic) variability in environmental and ecological conditions in the benthic community for the past6,300 years. Core samples were subjected to magnetic susceptibility measurements, x- radiography, grain size analysis, loss -on- ignition analysis for carbonate and sedimentary organic matter concentrations, carbon (C) and nitrogen (N) elemental abundance and stable isotope analyses, and reconstruction of paleoecological ostracod assemblages.
Smear slides were also examined to detect changes in the source or types of sediment through time. The chronology for core LT97 -56Vwas established by radiocarbon dating 22 of terrestrial plant fragments. Nitrogen stable isotope ratios, carbon to nitrogen ratios,
and sedimentary carbonate concentrations revealed a period (2,300 -4,300 years before present) of particularly stable and shallow lake stratification related to the cessation or
diminishment of the dominant regional southerly winds. With weak or absent southerly
winds, the lake's surface waters do not mix deeply, and recycling of nutrients from
nutrient -enriched deep waters is impeded. As a result, primary producers that can utilize
atmospheric nitrogen are favored, as was observed during the interval of stable
stratification indicated by core LT97 -56V. The major pattern observed in the ostracod
species abundance data for the core was that preservational conditions varied
substantially throughout the core, with no ostracod remains at all below 2,500 years
before present. Above this level, poor preservation correlated with high concentrations of
sedimentary organic matter. Across well -preserved intervals representing nearly all of
the 2,500 -year record of ostracod assemblage dynamics, no major turnover was observed
in the species richness, abundance, or composition of the ostracod community. The latter
result suggests that ostracod communities are relatively resilient in the face of natural
climatic and environmental variability, which has been documented in the Lake
Tanganyika basin over a range of time scales.
In the final study of this dissertation, I compared sediment cores collected offshore
from three watersheds in northern Tanzania. Two of the watersheds are encompassed in
Gombe Stream National Park, which achieved protected status and was depopulated in
the 1940s. Adjacent to the park is an extensively deforested watershed housing
Mwamgongo village, with high human population densities. Sediment cores were
collected at both sites in shallow ( <20 m) and deep (75 -100 m) water, by scuba divers 23 and a multi -coring device operated from a research vessel. Deepwater cores were dated with 21°Pb and 14C, on bulk sediment and terrestrial plant fragments, respectively.
Shallow -water core dates were based solely on terrestrial plant fragments. Grain size distributions, concentrations of carbonate and organic matter, elemental abundances and stable isotopes of C and N, smear slides, and ostracod assemblages were analyzed for changes through time. Sedimentation rates increased dramatically in cores off shore from the deforested site during the past few hundred years, but did not at the national park site.
In addition, stable N isotopes and ostracod assemblages changed progressively toward the present offshore from the deforested site in a manner not observed at the national park site.
Overall, the studies of this dissertation demonstrate the reliability and utility of using paleoecological records to reconstruct ecological history for a complex lacustrine ecosystem to determine the effects of human land use patterns on downstream ecosystems. The insights gained from such studies can be applied to conservation issues by describing the range and magnitude of natural environmental variability and the response of ecological communities to such change. Further, paleoecological studies can provide guidelines for restoring or protecting communities by illustrating the baseline conditions of the community and defining the response of the organisms to a range of disturbances. 24
CHAPTER 3: A HIGH -RESOLUTION RECORD OF NATURAL ENVIRONMENTAL CHANGE AND PALEOECOLOGICAL RESPONSE IN LAKE TANGANYIKA, EAST AFRICA, SINCE THE LATE GLACIAL
Abstract
We present here a high- resolution investigation of the relationship between natural environmental variability and paleoecological response in Lake Tanganyika from the
Late Glacial (-11-12 ka) to the present. Geochronological, sedimentological, carbon and nitrogen elemental and stable isotopic, and microinvertebrate paleoecological data were collected from a sediment core taken from the sparsely populated east central coastline of
Lake Tanganyika in relatively shallow water. Generally, the data support previous interpretations of regional paleoclimate and lake conditions, with wet and warm conditions during the interval from -r6.4 to 4.0 ka, and increasingly arid conditions since
-2.4 ka. However, for the interval from 4.0 to 2.4 ka, paleoenvironmental indicators suggest that the central part of Lake Tanganyika was stably stratified as a result of diminished southerly trade winds. Possible indications of changing primary productivity and nutrient recycling regimes occur throughout the core. Carbonate and ostracod crustacean preservation were low and nil, respectively, prior to -2.4 ka, indicating that the depth of the permanent thermocline and oxycline may have been shallower than the coring site (56 m) until then. Species abundance data for ostracods were used to define an ostracod water depth index (OWDI) for post -2.4 ka samples. OWDI successfully identified known intervals of lake level change and contributed new information on the timing of other lake level changes. Despite substantial environmental variability since
-2.4 ka, species richness and composition of ostracod assemblages remained relatively 25 stable through this interval. This long -term record of natural variability comesfrom an area that has experiencedless anthropogenic landscape alteration than most areas in East
Africa. It therefore serves as a baseline record benthic community dynamics in Lake
Tanganyika for comparison with recent conservation- related paleoecological reconstructions.
Introduction
In order to utilize paleoenvironmental reconstructions as a context for interpreting and predicting anthropogenically induced environmental changes, it is essential to have high -resolution records of changing environmental conditions and ecological response from an area experiencing minimal anthropogenic overprinting of the sedimentary record.
Many high- resolution, long -term records of environmental change have emerged from various lakes of the East African rift, although many of them report the influence of anthropogenic signals of landscape change during the past few hundred to few thousand years of sediment deposition (e.g., Beuning et al.,1997; Stager et al., 1997; Stager and
Johnson, 2000; Talbot and Lærdal, 2000; Taylor, 1990; Vincens, 1993). Here we present a high- resolution reconstruction ofpaleoenvironmental change since the Late Glacial in a sparsely populated region of Lake Tanganyika and examine the response of paleoecological invertebrate assemblages. The aim of this study was to provide context for paleolimnological interpretation of the effects of anthropogenic landscape alteration on lacustrine biodiversity and productivity (e.g. Alin and Cohen, 2001; Alin et al., 2000;
Wells et al., 1999). 26
Lake Tanganyika is a large, deep, meromictic tropical rift lake (32,600km2;Zmax =
1470 m; 3 -9 °S, 29-31°E) with permanently anoxic bottom waters (Fig. 3.1). Annual rainfall averages 900 -1000 mm yr 1 in the 231,000 km2 Tanganyika catchment ( Vincens,
1993). The two major rivers flowing into the lake are the Ruzizi River, which originates in Lake Kivu to the north, and the Malagarasi River, which drains a large area of the interior of Tanzania. The majority of water lost from the lake is through evaporation, although the lake currently has an outlet, the Lukuga River in the Democratic Republic of
Congo, at -775 m asl. Under modern climate conditions, the maximum depth of oxygenated waters in the lake varies from 50 -100 m at the north end of the lake to 240 m at the southern end (Coulter and Spigel, 1991). Relative depth of the oxycline is controlled by the strength of the southerly winds during Tanganyika's pronounced dry season from May to October (Coulter and Spigel, 1991). Absolute depth of the permanent thermocline fluctuates through time, depending on changes in wind strength and lake level (Haberyan and Hecky, 1987; Talbot, in press).
High -resolution paleoclimatic reconstructions using palynological and paleolimnological indicators have revealed that climate in East Africa became increasingly arid and seasonal during the past -2.5 ka (Beuning et al., 1997; Stager et al.,
1997; Stager and Johnson, 2000; Talbot and Lxrdal, 2000; Taylor, 1990; Vincens, 1993).
These studies of recent climate change in East Africa also show signals of anthropogenic landscape alteration. Many of these studies occurred in particularly densely populated regions, where separation of the discrete signatures of anthropogenic and natural environmental change on paleoenvironmental patterns and processes is difficult. The 27 continent -wide scale of the Late Holocene trend toward aridification generally supports the interpretation of a natural change in climatic conditions (cf. Talbot and Johannessen,
1992). However, none of the East African records are derived from areas with low human population density. Here we present one such sediment core record from the east central coastal region of Lake Tanganyika. High- resolution stable isotope data for the past 250 years from a nearby core confirm that anthropogenic landscape change has not yet affected paleoenvironmental records in this part of Lake Tanganyika (O'Reilly, in
prep).
Another impressive feature of the Lake Tanganyika ecosystem is the extensive evolutionary radiations of fish and invertebrates that inhabit it. In recent years, concern over the conservation status of the rich, endemic fauna of the lake has stimulated a number of studies utilizing paleoecological methods to reconstruct the ecological history of areas of the lake experiencing different degrees of anthropogenic disturbance (Wells et al., 1999; Alin et al., 2000; Alin et al., 2001; M. Palacios -Fest and A.S.C., in prep).
While these studies provide valuable insight into changes experienced by benthic ecological communities during the recent period of anthropogenic change, longer -term records of natural environmental variability and the paleoecological response of nearshore benthic communities in Lake Tanganyika's shallow -water habitats are lacking.
In this study, paleoecological and paleoenvironmental indicator data spanning -6400 years have been collected in order to elucidate the longer -term trajectory of natural environmental change in the lake basin and its catchment, and to examine changes in invertebrate assemblage composition during this interval. The data stem froma shallow- 28 water core representing Mid Holocene to modem environmental history in the central part of Lake Tanganyika. Collectively, these elemental, stable isotopic, and geochronological data refine the timeline and direction of environmental change for Lake
Tanganyika since the Mid Holocene. Furthermore, data from paleoecological invertebrate assemblages describe the ecological response in the shallow- water, benthic ecosystem to natural environmental fluctuations through the past 2400 years.
Methods
Core collection and sampling
In 1997 we collected a series of relatively shallow water ( <100 m) vibracores from the southern part of the northern basin of Lake Tanganyika. Coreswere cut into 1.5 m sections in the field, sealed, and shipped to the U.S. for further analysis. Magnetic susceptibility was measured at 2 cm intervals, using a Bartington Instruments susceptibility meter with a 100 mm loop sensor in the laboratory of J. King at the
University of Rhode Island. Subsequently, cores were split, x- rayed, and sampled in 1cm increments at 2 -4 cm intervals for radiometric, paleoenvironmental, and paleoecological analyses.
This paper will focus on results from core LT97 -56V, a 356 cm core collected in 56 m water depth about 700 m southeast of a small river outlet and near the village of
Mgondozi (5 °46.33' °S, 29 °56.03' °E), in an area of the Tanganyika basin with relatively low human population densities. Visual inspection of thecore at the time of collection indicated that it experienced minimal disturbance from the coringprocess, and the core 29 top appeared to be intact on recovery.The core was sampled every other centimeter for the upper 20 cm, and every fourth centimeter thereafter. Every other sample was analyzed for grain size distribution, sedimentary organic matter (SOM) and carbonate content, carbon and nitrogen elemental abundances and stable isotopes, and invertebrate assemblage composition. Ostracod crustaceans were used for paleoecological assemblage reconstruction because they are one of the most diverse taxonomic groups in the lake and are preserved abundantly and in sufficiently high quality to allow quantitative comparisons among species assemblages. Smear slides were also examined to look for gross changes in the source of sediment (i.e., in the balance between terrigenous and lacustrine inputs).
Laboratory Analyses
Ten radiocarbon dates were based on terrestrial plant fragments to avoid the problem of 14C reservoir age for samples produced within the lake (Table 1). Leaf and wood samples were prepared for radiometric analysis by acid -base -acid pretreatment to remove carbonates and diagenetic organic material. All radiocarbon analyses were made by the
Arizona AMS Laboratory and were calibrated using CALIB 4.3 (Stuiver et al., 1998a;
Stuiver et al., 1998b). The Southern Hemisphere correction was not applied, on the basis that the study sites are equatorial. Existing data on interhemispheric concentration differences in atmospheric 14C provide inconclusive evidence on the relevance of the
Southern Hemisphere correction to southern equatorial locations. It is possible that this relationship has changed with the southern migration of the intertropical convergence 30 zone throughthe Holocene, but the magnitude of any difference in the resulting age
model would be trivial (Haug et al., 2001).
For the purpose of establishing an age model for core LT97 -56V, midpoints of the
most probable calibrated age ranges (2g) were used, except for the samplefrom 2 -3 cm
where the next most probable age provided an age model more consistent with our
observation of an intact core top and was of essentially equivalent probability to the most
probable calibrated age. Ages were assigned to stratigraphic intervals based on a
polynomial fit of the radiocarbon data. The uppermost and lowermost radiocarbon dates
were excluded from the age model on the basis of apparentreworking and a depositional
hiatus, respectively. Sedimentation rates were estimated using linear fits of data points
falling along lines of similar slope. Calibrated radiocarbon ages are reported in ka
(thousands of calendar years before present).
Smear slides were prepared prior to sieving by suspending a small amount of wet
sediment in water, mixing thoroughly, then applying a drop of slurry to a glass slide with
a cover slip. Slides were examinedquantitatively for compositional variation under
transmitted light at 100x magnification, and sediment particles were classified into one of
four categories - siliceous (diatoms and sponge spicules), organic (flocculent organic
matter, charcoal, etc.), carbonate (primarily ostracod valves, some smallmollusc
fragments), and siliciclastic (dominantly quartz, some lithic fragments) - until 200 points
in a grid had been counted. The siliciclastic fraction represents exclusively terrigenous
sources of sediment. 31
Concentrations of sedimentary organic matter (SOM) and calcium carbonate were estimated by loss -on- ignition at 550 °C and 925 °C, respectively (Bengtsson andEnell,
1986). Remaining sediments were sieved using 1 mm, 106 µm, and 63 µm meshsieves, with the finest fraction of sediment ( <63 µm) retained on Whatman -1 qualitative filter paper.
The <63 µm sediment fraction was used for elemental and stable isotopic analyses.
Carbonate was removed from samples prior to analysis by reacting sediment with an excess of 1N HC1 for >24 hours. Samples werethen rinsed three times with distilled
water, dried at 60 °C, and pulverized. Elemental abundances and stable isotopic
composition of total organic carbon (TOC) and total nitrogen (TN) were determined at
the University of Arizona on a Costech Elemental Analyzer coupled to a Finnigan Delta -
plus XL continuous -flow mass spectrometer. Isotopic ratios are expressed in delta
notation (813C,815N)with respect to the standards Pee Dee belemnite and atmospheric
nitrogen. Precision for internal standards is 0.05 %o for813Cand 0.15 %o for815N(16).
Prior studies in African lakes have shown that a high proportion of coarser ( >63 gm)
organic matter (OM) is of terrestrial origin and is therefore of little importance for
understanding changes in autochthonous OM related to lake productivity (Talbot and
Johannessen, 1992). Accordingly, since the aim of this study was to understand the
effects of climate change on lacustrine productivity and ecological dynamics, we used the
<63 µm fraction for stable isotopic analyses. In contrast, the total concentration of SOM
is important in determining the preservational environment for carbonates, including
ostracod valves, so concentration of SOM from loss -on- ignition was used for such 32 comparisons. The overall shape of the SOM and TOC curves is very similar, although the amplitude of the curves differs.
Ostracods from the >1 mm sediment fraction were added to the 106 gm -1 mm fraction before splits of the 106 1,1m -1 mm sediment fraction were counted for ostracods.
For all samples containing sufficient ostracods, 500 individuals were identified to the species level following published descriptions whenever possible (Martens, 1985; Park and Martens, 2001; Rome, 1962; Wouters, 1988; Wouters and Martens, 1992; Wouters and Martens, 1994; Wouters and Martens, 1999). Roughly half of the estimated 200
Tanganyikan ostracod species remain to be described (Martens, 1994). Extensive reference collections at the University of Arizona were employed to assign identities to undescribed specimens. Taphonomic data were tallied for >100 specimens in each sample (percentages of adults vs. juveniles; disarticulated valves vs. carapaces; and cracked or broken, reduction stained, encrusted, corroded or abraded, and yellowed or opaque valves, suggesting incipient staining or encrustation). A few samples lacked sufficient ostracod abundance to perform standard 500 counts. In order to facilitate comparisons between samples of different sizes, Fisher's alpha (a) diversity index was computed (Magurran, 1988). Species abundance data for ostracod assemblages were analyzed by detrended correspondence analysis (DCA) using CANOCO 4 and detrended by 2nd order polynomials (Minchin, 1987; ter Braak and Prentice, 1988). Statistical tests were done in JMP IN 3.2.6 (Sall and Lehman, 1996).
Depth and substrate indexes were defined for ostracod species assemblages based on the output of a canonical correspondence analysis (CCA) of a database of live ostracod 33 species abundance data from various sites, substrate types, and water depths around Lake
Tanganyika (A. Cohen, unpublished data). The live database consists of 84 samples and
144 species. CCA Axis 1 was significantly and strongly correlated with substrate type
(rocks, sand, mud) and weakly with longitude. Axis 2 correlated significantly with water depth and latitude; Axis 3 with water depth; and Axis 4 with substrate type, disturbance level, vegetation type, and latitude. The ostracod water depth index (OWDI) was created by plotting samples in Axis 2 vs. 3 ordination space, where shallow ( <20 m, n =59) and deep ( >20 m, n =26) samples formed non -overlapping clusters. Species ordination scores were used to assign species to either the shallow or deep category. OWDI is defined as the ratio of individuals belonging to deep species to individuals in shallow species. The ostracod substrate index (OSI) was defined using Axis 1 species scores to assign species to rocky, sandy, or muddy categories. Separation of samples along Axis 1 with respect to substrate type was not as good as for depth, as many species are commonly collected live in more than one habitat type. OSI is equal to Nrocky (Nsandy +Nmuddy)-1, where N is the number of individuals in each category.
Results
Geochronology and Sedimentology Data
Ten radiocarbon dates for core LT97 -56V indicate two intervals of relatively constant sedimentation rates, with slower accumulation rates from 92 -302 cm (6.0 -0.8 ka, 0.4mm yr ') and a steep increase in rate above 92 cm (sedimentation rate = 1.0 mm yr ') (Fig.
3.2). The 813C profile of LT97 -56V closely resembles that for nearbycore LT97 -57V, 34 with an intact, 210Pb -dated core top, confirming the presence of core top sediments in
LT97 -56V (O'Reilly, in prep) (Fig. 3.1). Sedimentation rates for LT97 -56V indicate that sample resolution is on the order of 8 -10 years sampled at 32 -40 year intervals for the upper 20 cm, with resolution stable but gaps increasing to 56 -70 yearsfrom 20 to 92 cm.
Below 92 cm, sample resolution drops to 25 years sampled spaced at 200 year intervals.
Calibrated radiocarbon ages imply a depositional hiatus of 5 -6 ka near the base of core LT97 -56V, ending at -6.4 ka, (Table 2, Fig. 3.3). Grain size fines sharply across the hiatus at -x312 cm depth (-r6.4 ka), from domination by very fine -medium sands below to silts above, indicating that the overlying sediments were deposited at substantially greater water depth (Fig. 3.4). Overlying basal sands, x -ray stratigraphy of core LT97 -56V reveals that sediment structure alternated between massive or burrowed and thinly bedded or laminated (Fig. 3.4).
Smear slide data indicate that organic and siliciclastic particles were dominant and remained approximately equal in percentage throughout the core (organic: average =
47.1 ±12.2 %; siliciclastic: average = 51.6 ±12.9 %). Contributions of siliceous and carbonate particles were minimal, with averages of 0.7±0.9% and 0.6±0.9 %, respectively.
Despite the observation that organic matter makes up a relatively large proportion of the volume of each sample, as gauged by point counts, percentages of SOM by weight remain low because of the low density of SOM relative to siliciclastic material.
Magnetic susceptibility remains relatively low (average = 42.7 cgs) in the lower half of the core, but increases rapidly between 160 and 120 cm (2.0 -1.3 ka) to a new average of 95.8 cgs above 120 cm (Fig. 3.5). Peak values occur at 120 ±2 cm (1.3 -1.2 ka, high = 35
133.6 cgs) and 62 ±8 cm (0.5 -0.4 ka, high = 109.7 cgs). Minimum values abovethis level occur at 108±4 cm (1.1 -1.0 ka, low = 103.2 cgs) and 78 ±8 cm (0.7 -0.5 ka, low = 23.2 cgs). Magnetozones are illustrated as tie points among cores in Fig. 3.4.
Elemental and Stable Isotope Data
Concurrent with the abrupt fining of sediment grain size at 312 cm and resumption of sediment deposition, percentages of SOM and carbonate preserved in the sediment doubled, from 1.4% to 3.2% SOM and from 0.3% to 0.6% carbonate (Fig. 3.5, Appendix
E). SOM percentages continue to increase almost continuously upcore, with peaks in abundance at 126 -127 cm (-1.4 ka) and 62 -71 cm (0.6 -0.5 ka). Concentration of SOM is negatively correlated with both depth (rz = 0.68, p < 0.0001) and age(r2= 0.61, p <
0.0001), indicating a progressive increase in SOM toward the present, which may be a function of either changes in primary productivity or preservation. Since the SOM -depth relationship is stronger, residuals of the SOM -depth relationship (SOM,sd_depth) are included in correlation tests among variables to determine whether an increase in productivity or gradual decay of SOM better explains the distribution of other data.
Carbonate concentrations increase to a broad peak of 2.2% from 175 to 140 cm (2.3-
1.6 ka) with an additional sharp peak in carbonate concentration (4.2 %) at 54 -55 cm
(- 1580 AD) representing shell hash in the sediment (Fig. 3.4).
Elemental abundance of carbon and nitrogen in the fine sediment fraction (i.e. TOC and TN from elemental analysis) are essentially identical and mirror the shape of the
SOM curve in major features, although the elemental analyzer data show greater fine- 36 scale lability (Fig. 3.5, Appendix E). Elemental abundance, stable isotope, and grain size data defined four environmental zones in the core.
In sandy zone 1 sediments preceding the depositional hiatus, SOM and carbonate concentrations were lowest, whereas C:N, S 15N, and 813C ratios were relatively high
(Fig. 3.5). In zone 2, S 15N and C:N values decline gradually through the period -6.4 -4.2 ka. The transition from zone 2 to zone 3 is delineated by sudden shifts in b 15N and C:N values to core -wide low values which persist from 4.0 to -2.3 ka. The steep increase in carbonate concentration at 175 cm occurs roughly contemporaneously with shifts in b 15N and C:N and helps define the period 2.4 to 2.2 ka as environmentally transitional. In zone 4, S15N and C:N values rise abruptly and maintain high levels from -2.2 ka to the present. Through this interval,813Crose gradually until -0.3 ka, when the trend reversed, accelerating from - 1900 AD to the present. This trend has been ascribed to the
Suess effect in tandem with recent climate changes affecting lacustrine productivity
(O'Reilly, in prep). The uppermost core sample had anomalously high percentages of C and N relative to all other samples in the core (shown as outlying values in Fig. 3.6A,C insets).
Regression indicated that TOC and TN concentrations are highly correlated (r 2 =
0.93, p < 0.0001), suggesting that the majority of N is organically bound (Talbot and
Johannessen, 1992). TOC was also regressed against TN for each zone separately to determine the amount of exchangeable N (i.e., the fraction of N bound to inorganic surfaces) in sediment from each zone (Fig. 3.6A). For regressions with positivey- intercepts (zone 2: yint = 0.0137, r 2 = 0.83; zone 4: yint= 0.0148, r2= 0.96), the value of 37 the y- intercept was subtracted from TN to give a corrected, solely organic TN (Talbot, in
press).
Corrected TN values were used to calculate C:N ratios and were used for all
subsequent comparisons of TN and other variables. C:N ratios close to or below 10 are
indicative of lacustrine sources of organic matter, whereas terrestrial organic matter
characteristically has C:N ratios »20. Throughout core LT97 -56V, C:N values fall
between 10 and 30 (raw values in zones 2 and 4 were lower before correction).
Sedimentary organic matter in zones 3 and 4 is dominated by lacustrine primary
productivity, whereas zone 2 shows greater influence of terrestrial organic matter sources
(Fig. 3.5).
The N isotope signal in lake sediments is largely a consequence of the815Nof algal
OM because of the much higher N content of lacustrine versus terrestrial OM (Talbot, in
press). No significant relationship exists between815N and C:N (r 2= 0.07, p = 0.0693),
although the data do form three discrete clusters: 1) terrestrially- dominated,15N- enriched
samples (zone 2); 2) lacustrine- dominated,15N- depleted samples (zone 3); and 3)
lacustrine- dominated, 15N- enriched samples (zone 4). Zone 1 samples are intermediate in
value between those of zones 2 and 4, indicating a mix of terrestrial and lacustrine OM.
In contrast, C:N and813Care weakly correlated (Fig. 3.6G). Data points from each
zone cluster together, again with zone 1 valuesintermediate to zones 2 and 4. Data
clusters correspond in this case to: 1) terrestrially- dominated,13C- depleted samples,
representing C3 vegetation (zone 2); 2) lacustrine- dominated, 13C- depleted samples,
consistent with either lacustrine or C3 primary productivity (zone 3); and 3) lacustrine- 38
dominated, relatively 13C- enriched samples, representing some contribution of C4 organic
matter (zone 4).
Diagenesis
Both organic nitrogen and carbon percentages decline monotonically down the core
(Fig. 3.5), suggesting that diagenesis and selective degradation are important factors to
consider in analyzing the stable isotope data. Following Talbot and Johannessen (1992),
we regressed TOC and TN against813Cand815N,respectively, to test for the imprint of
diagenetic processes on the stable isotopic signatures of the core sediment samples (Fig.
3.6B,C). Various bacteria have high fractionation values, preferentially scavenging or releasing isotopes of C or N during degradation, thus affecting residual stable isotope ratios. A diagenetic signature should appear as a correlation between elemental
abundances and stable isotope ratios (Talbot and Johannessen, 1992). There is a
significant correlation between TOC and813C(r 2 = 0.41, p < 0.0001), although the correlation disappears when only samples with low to moderate TOC (< 1.25 wt. %) are
included, when the correlation should be strongest if diagenesis is responsible for the relationship (r 2 = 0.07, p = 0.1124). Corrected TN values do not correlate with b 15N,
lending support to the interpretation that diagenetic processes do not play a substantial role in establishing stable isotopic signatures (full data set: r 2 = 0.04, p = 0.1143; TN <
0.9 %: r 2 = 0.02, p = 0.4225). Finally, the nitrogen stable isotopic record shifts over short
stratigraphic intervals and in both directions, suggesting no systematic diagenetic bias in the record. Thus, while degradation may have contributed to long -term changes in SOM 39
and TN, it does not appear that diagenetic processes have introduced any systematic bias
to the stable isotopic sedimentary record.
Paleoecological Data
Ostracod remains were absent below 179 cm (--2.4 ka) (Fig. 3.7). Above this core
depth, ostracod abundance ranges from 7 to 5570 valves g -1, with an average of 1360
valves g -1. Three anomalous low values correspond to the peaks in SOM at 62 -71 cm
(0.6 -0.5 ka) and 126 -127 cm (-1.4 ka), although overall ostracod abundance is not
correlated with concentrations of carbonate (r 2 = 0.0003, p = 0.9345), SOM (r 2 =
0.0001, p = 0.9658), or SOMdepth -rsd (r2= 0.0063, p = 0.6994).
Species richness of assemblages remains relatively stable throughout core LT97 -56V,
with an average of 34.3 ±7.7 species sample -1 (Fig. 3.7). Samples with the lowest species
richness occur simultaneously with peaks in SOM concentration and low values in
ostracod abundance (62 -71 cm, 126 -127 cm). Insufficient ostracods were present in these
samples to count an entire sample of 500 (samples contained 52, 12, and 102
individuals). If low species richness values are excluded, average species richness of
core samples increases, and the standard deviation decreases substantially (36.8 ±2.4
species sample -1). Values of Fisher's cc are more stable throughout the core, with the
only markedly lower value corresponding to the sample with only 12 individuals, which
is probably too small a sample for Fisher's cc to correct (Fig. 3.8).
Taphonomic data indicate that alteration of the ostracod assemblageswas most severe
in the two intervals with low ostracod abundance and high concentrations of SOM. 40
Valves in these altered, depauperate assemblages were characterized by high proportions of corroded, encrusted, reduction stained, and yellowed valves, and were dominated by adults and more heavily calcified species (Fig. 3.7). Reduction staining and encrustation show multiple peaks in the upper few meters of core LT97 -56V.
Composition and abundance of dominant species fluctuates fairly randomly throughout the ostracod -bearing section of LT97 -56V (0 -173 cm), which represents -2.4 ka of apparently continuous deposition (Appendix E). All dominant species persist through the ostracod -bearing interval of the core. While some are more abundant at the top and some at the bottom, it appears that ostracod assemblages in this undisturbed region of Lake Tanganyika experienced great fine -scale variability through time, but no long -term or concerted turnover in species composition, as seen in cores from disturbed areas of the lake (Fig. 3.8). Detrended correspondence analysis also revealed no striking upcore trend in ostracod species assemblage composition, which may be seen as contiguous samples forming clusters or lines (Fig. 3.8). However, DCA Axis 1 is positively correlated with taphonomic variables (% abrasion: r 2 = 0.50, p = 0.0001; % adults: r 2 = 0.76, p < 0.0001; % reduction -stained: r 2 = 0.42, p = 0.0004; % encrusted: r2= 0.49, p = 0.0001; % yellowed/opaque: r2= 0.49, p = 0.0001) and OWDI (r2= 0.45, p = 0.0002), and negatively correlated with species abundance (r2= 0.35, p = 0.0014), richness (r 2 = 0.65, p < 0.0001), and Fisher's a (r 2 = 0.37, p = 0.0009). Outliers consisted of samples with low ostracod abundance (samples labeled 21, 23, and 37 in Fig.
3.8) and the first sample postdating the ostracod -barren section of the core (labeled 50). 41
Despite the lack of apparent changes in dominant species composition, ostracod assemblages appear to sensitively reflect lake level. OWDI accurately replicated known lake level changes in historical data since -1850 AD (Birkett et al., 1999; Evert, 1980) and stromatolitic records (Cohen et al., 1997b) (Fig. 3.9), despite the fact that not all species in core samples were represented in the live database and that the depth range of samples in the live database lies entirely above the water depth of core LT97 -56V.
Discussion
Sedimentary Environment and Preservation Conditions
Preservation of ostracod assemblages appears to have been strongly influenced by
SOM percentage. The worst preservation quality, lowest abundance of ostracod valves, and greatest representation of adults and heavily calcified species occurred concurrently with the highest residual SOM, suggesting that high SOM intervals result in selective dissolution of weakly calcified ostracod species and juveniles, thus biasing the resulting species richness, abundance, and composition of ostracod assemblages. Encrustation also peaked in this section of the core, probably as a result of an abundance of available carbonate from the shelly hash at 54 -55 cm adjacent to high SOM levels. SOM contributes to the dissolution of carbonate by forming organic acids and lowering the pH of interstitial waters (Dean, 1999). When excess calcium and carbonate in solution diffuse into adjacent stratigraphic levels, favorable conditions are created for deposition of diagenetic carbonate crusts on sedimentary particles 42
Apart from the zones clearly affected by taphonomic alteration (shaded in Fig. 3.7),
ostracod species assemblages have approximately equal representation of species thatare
weakly calcified (e.g., Mecynocyprian.sp. 37, M. emaciata, Cypridopsis n.sp. 6A, C.
n.sp. 21, and C. n.sp. 23 in Appendix E) relative to those that are heavily calcified
(Romecytheridea ampia, R. longior, Mecynocypriaopaca, Mesocyprideis n.sp. 2B,
Cyprideis n.sp. 24, C. n.sp. 25, and Gomphocythere woutersi).
Lack of temporal clustering of ostracod assemblages in ordinationspace suggests no
significant turnover in the ostracod community in the face of natural environmental
variability at this site. DCA Axis 1 reflects the influence of high SOM valuesand effects
of selective dissolution of ostracod valveson species assemblage preservation and
composition. Indeed, the shape of the curve of DCA Axis 1scores plotted through time
resembles that of the SOM curve (Fig. 3.7). Simultaneity of peaks in DCAAxis 1, SOM,
and OWDI can be explained by covariation in water depth and SOMpreservation, with
deeper water intervals having higher SOM (cf. Verschurenet al., 2000)
Environmental Variability and Paleoecological Response
Previous paleolimnologic studies provide paleoclimatic interpretations forthe Lake
Tanganyika basin that are somewhat at odds with interpretations suggestedby sedimentological and geochemical data from the cores discussed herein (cf.Cohen et al.,
1997b; Haberyan and Hecky, 1987; Talbot, in press; Vincens, 1993). Thediscussion that follows serves as a preliminary review of previous paleoclimaticreconstructions. In addition, a crude lake levelcurve from the Late Glacial to the present was constructed 43
based on a preliminary synthesis of our core data with data from several other studies in
preparation for publication. A more thorough synthesis and revision will precede the
submission of this manuscript for publication.
Pollen -based paleoclimatic reconstruction of rainfall and temperature for the northern
basin of Lake Tanganyika by Vincens (1993) documented a warm, wet early -mid
Holocene 10.0 -2.6 14C ka (11.4 -2.7 ka), with peaks in the extent and diversity of humid
tropical African forests (Fig. 3.10). These Holocene warm, wet "optimum" conditions
were preceded by a gradual post -LGM amelioration of climate beginning around 15.0 14C
ka (18.0 ka), although conditions remained generally colder and drier, and the vegetation
more cold- and drought -tolerant and representative of higher elevations, throughout this
interval than at present (Vincens, 1993). Finally, from 2.6 14C ka (2.7 ka) to the present,
the pollen record suggests that climate gradually became more arid, with reduction in
forest cover and extension of grasslands. The pollen record from core LT98 -2M indicates that gradual or step -wise replacement of arboreal pollen bygrass pollen occurred between 2.3 and 1.1 ka, refining the timeline of this C3 to C4 transitionnear core
LT97 -56V (Cohen et al., 1999).
Talbot's (in press) stable isotopic record of paleoenvironmental variability from the southern basin of Lake Tanganyika indicates a sharp peak in phytoplankton productivity around 11.3 14C ka (13.3 ka) and a prolonged period of stable stratification when productivity was dominated by N- fixing cyanobacteria during the mid -Holocene, 5.3-3.4
14C ka (6.3 -3.7 ka) (Fig. 3.10). Unfortunately, Talbot'score record terminates at 3.4 14C ka. Haberyan and Hecky's (1987) paleolimnologicalrecord of 25 -0 ka from the southern 44
basin of Lake Tanganyika suggests that the lake maintainedopen lake conditions
throughout the interval from 11.4 to 3 ka.
Our stable isotopic and elemental data generally reinforce andcomplement
paleoclimatic and paleolimnological interpretations of Talbot (in press),Haberyan and
Hecky (1987), and Vincens (1993), although grain size and geochronologicaldata
collated from cores in Table 3.2 imply that Lake Tanganyika didnot experience open
basin conditions until the Mid Holocene ( -6.4 ka) incontrast to the conclusions of
Haberyan and Hecky (1987). Moreover,our data overlap with and extend Talbot's (in
press) record of lacustrine productivity to thepresent. First, the peak in zone 1 of b 15N
and concurrent moderate -high values of C:Nmay coincide with Talbot's sharp peak in
phytoplankton productivity at 11.3 ka (Fig. 3.5).
With the highest C:N ratios and low813Cand b 15N values, zone 2 productivity is
strongly influenced by dominantly C3 terrestrial inputs from 6.1to 4.3 ka. Lack of
sedimentary structures other than burrows supportan interpretation of well -mixed and
oxygenated waters extending to at least the water depth ofcore LT97 -56V (-56 m at
modem lake level). Based on data from this study, lake levelwas probably close to
modern levels throughout deposition of zone 2 sediments, whichsupports the conclusions
of Haberyan and Hecky (1987).
Zone 3 paleoproductivity indicators (C:N,815N)suggest a dominance of
cyanobacterial production from 4.0 to 2.4 ka and refine the timing ofchanges in the
strength of southerly trade winds, extending the period of lax windsthrough the interval
4.0 -2.4 ka (- r3.7 -2.4 14C ka) for central Lake Tanganyika (cf. Talbotin press] : 5.3 -3.4 45
14C ka; Haberyan and Hecky [1987]: -4.1 -3.0 14C ka for southern basin cores).
Cyanobacteria are capable of fixing atmospheric nitrogen and characteristically have
815N values approximating 0 %o. Throughout zone 3, 815N values hover between -0.5 and
1 %0. Low C:N ratios characterize cyanobacteria relative to other phytoplankton(Talbot, in press), and the lowest C:N ratios in core LT97 -56V occur in zone 3 (Fig. 3.5).813C values for zone 3 are also consistent with cyanobacterial production in Lake Tanganyika
(C. O'Reilly, pers. comm). Finally, according to Talbot (in press), positive correlation between C:N and 815N and negative correlation between C:N and TOC confirm that cyanobacteria are responsible for low 815N and C:N zone 3 values (Fig. 3.6E,F).
Dominance of cyanobacterial production in zone 3 supports the interpretation of stable
thermal stratification of Lake Tanganyika during this interval, with reduced upwelling
and recycling of nutrients from the monimolimnion, conditions which favor nitrogen -
fixing producers.
Absence of ostracods and low carbonate levels prior to 2.4 ka suggest that permanent
thermocline depth was shallower than the core site (56 m) during the 4.0 -2.4 ka interval
of stable stratification. Monimolimnetic waters in meromictic lakes are generally
corrosive to carbonate because of lower pH and higher CO2 concentrations than in
mixolimnetic waters (Finney and Johnson, 1991), and preservation of carbonate in
deepwater cores from Lake Tanganyika is typically poor. However, Haberyan and Hecky
(1987) report increased preservation of carbonate in a deepwater (440 m) Lake
Tanganyika core at -3.0 14C ka (3.2 ka), indicating greatly increased precipitation of
calcium carbonate in the lake's surface waters at this time. Other relatively shallow- 46 water cores from central Lake Tanganyika (core LT98 -2M, 110 m water depth; core
LT97-61V, 67 m water depth) also show sharp increases in carbonate preservation at approximately 2.4 ka (A.S.C., unpub. data; Cohen et al., 1999). Initial appearance of ostracods in core LT97 -56V is contemporaneous with increases in carbonate preservation and resumption of deeper mixing of lake surface waters, based on the815Nprofile (Fig.
3.5), although the first appearance of ostracods in core LT98 -2M occurs at 1.5 ka, -900 years after increases in carbonate preservation, demonstrating that increases in carbonate preservation are not necessarily linked to location of coring sites with respect to permanent thermocline depth. Preservation of laminae and/or thin bedding structures in core LT97 -56V zones 2 and 3 supports the interpretation of deposition under anoxic conditions, such that bioturbation did not disrupt these fine -scale features of the core, until -2.4 ka. Between 2.4 and 1.5 ka, when ostracods first appear in LT98 -2M, the permanent thermocline depth was probably between 110 and 56 m below current lake level.
Vincens (1993) argued that the Late Holocene (our zone 4) was a period of increasing aridity in the northern basin of Lake Tanganyika. While anthropogenic alteration of her palynological record may obfuscate the accurate quantification of some paleoclimatic signals, she argues that the continent -wide distribution of indicators of increasing aridity during the Late Holocene confirms a climatic change in addition to changing land-use patterns. 815N and C:N data from core LT97 -56V suggest a sudden resumption of deeper mixing between 2.4 and 2.2 ka, suggesting rapid intensification of the southerly trade winds (Fig. 3.5). Substantial variation in lake level is indicated by OWDI inzone 4 47
(Figs. 3.7, 3.8). Peak carbonate abundance corresponding to the shelly hash layer at -54-
55 cm coincided temporally with the Little Ice Age (-1300-1900 AD; shell hash age
-1580 AD). Presence of shell hash in water of this depth suggests lake level >30 m lower than modern lake surface elevation. Shell hash and OWDI indicators delineate the timing of a known, but temporally unconstrained, lake lowstand from Cohen et al. (1997b) to the interval 1730 -1800 AD, previously only known to have occurred between the late 16th and early 19th centuries. Drought chronologies from East Africa indicate that several extreme and prolonged droughts occurred during the 1700s and 1800s AD (Nicholson, 1998b). In addition, OWDI further substantiates Late Holocene high stands reported by Cohen et al.
(1997b) (Cohen et al [1997]: 1250 -1550 AD and 430 ±110 AD; OWDI: 1300 -1500 AD,
-550 AD). Reconstructed OWDI values from core LT98 -2M also successfully reconstruct the 1250 -1550 AD and 430 ±110 AD highstands from Cohen et al. (1997b).
Finally, the zone 3 rise in magnetic susceptibility is corroborated by coherent changes in the magnetic stratigraphy of nearby cores (LT97 -35V, LT97 -53V, LT97 -57V; Figs.
3.1,3.4). A steep rise in magnetic susceptibility occurred between 2.0 and 1.3 ka in
LT97 -56V, -53V, and -57V, suggesting a substantial change in the extent or efficiency of watershed weathering related to aridification of regional climate. Changes in magnetic susceptibility have also been ascribed to signals of shoreline proximity, lake level, or fire intensity (Negrini et al., 2000; Thompson and Oldfield, 1986). As our records of magnetic susceptibility are difficult to correlate with other indicators of lake level, their main utility was as an alternate means of stratigraphic correlation and corroboration of radiocarbon results. However, by constraining the timing of grain size shifts, 48 depositional hiatuses, and lake level changes, magnetic susceptibilityrecords contributed to a composite lake level curve (Fig. 3.10).
Temporal Trend SOM Preservation: Signal of Degradation or PrimaryProductivity?
Several lines of evidence suggest that both gradual long -term degradationand strong
temporal changes in lacustrine productivity contribute to the patterns seen in SOM,
carbonate, and813Crecords. Strong correlation of SOM versus depth or age strongly
suggests that sufficient degradation of OM has occurredthrough time to make SOM
residuals the better explanatory variable for, e.g., observed taphonomic damage,despite
the fact that overall loss to degradation is slight (slope of SOM -depth linear regression =
0.01439, r 2 = 0.68) and does not appear to have overprinted stable isotopic records, as
discussed previously. Some degradation of OM in cores from shallow -water, oxygenated
environments is expected and probably explains the shallow monotonic decline of SOM
downcore in LT97 -56V.
Other observations point to increased productivity in the Late Holocene as an
explanation for steeper slope of the SOM curve in zone 4. Peaks in SOM preservation
occur at -1.4 and0.6 -0.5 ka in core LT97 -56V, however the coincidence of increased
water depth with more concentrated SOM intervalsconfounds the interpretation of these
high SOM values as signals of increased productivity (cf. Verschuren et al., 2000). Data
from other, deeper -water cores shows that SOM preservation increased steeply between
2.4 and 2.1 ka (core LT98 -2M, 110 m water depth; core LT00 -03, 608 m) (Fig. 1),
contemporaneously with the inferred resumption of southerly trade winds and increased 49 mixing of surface waters (Cohen et al., 1999; Zilifi and Eagle, 2000). SOM preservation subsequently decreased somewhat in core LT00 -03 at 0.8 ka but remained higher throughout the Late Holocene than prior to -2.4 ka. Presumably these changes are related to increased nutrient recycling efficiency driving increased primary production in the mixolimnion. Carbonate peaks at 2.3 -1.6 ka in core LT97 -56V, - 2.4 -0.7 ka in LT97-
61V, and -2.4 -0.6 ka in LT98 -2M support an interpretation of increased Late Holocene productivity, as photosynthetic drawdown of aqueous CO2 stimulated precipitation of
CaCO3 (Wetzel, 1983). Finally, steadily increasing813Cvalues from -'2.4 to 0.6 -0.5 ka
are consistent with increasing lacustrine productivity. However, the temporal trend in
813C is also consistent with paleoclimatic and paleobotanical data indicating an increased
presence of C4 vegetation in the landscape surrounding Lake Tanganyika (Vincens,
1993). Increased productivity and C4 input probably both contributed to rising 813C
values. Palynological data from core LT98 -2M indicate that the most rapid shift in the
transition to modern vegetation composition occurred ca. 1.1 ka, which implies that the
earlier rise in813Cvalues may be attributable to lacustrine productivity, whereas C4
terrestrial production may explain sustained high813Cwhen other indicators suggest
declining lacustrine productivity.
Charcoal and Lake Level Change
In the course of writing the manuscript for Appendix C, I discovered that the
sedimentary charcoal abundance records gathered by Cohen et al. (1999) showed promise
as indirect indicators of lake level change, with high pre -anthropogenic charcoal 50
abundance reflecting increased fire frequency under drought conditions. In order to
accurately constrain the timing of drought and flood events, the charcoal records much be
anchored by independently established, reliable sediment chronologies, e.g. radiocarbon
ages from material that is not prone to the considerable14C reservoir effect in Lake
Tanganyika.
Geochronological and charcoal abundance data from deepwater cores collected
offshore from Gombe Stream National Park and Mwamgongo village in Tanzania may
contribute new information constraining the timing of a lake lowstand known to have
occurred in Lake Tanganyika between the late 16th and early 19th centuries from
winnowed shell deposits in relatively deep water dating from this interval (Cohen et al.,
1997b). Core LT98 -58M from Gombe shows a strong peak in charcoal abundance from
-1810 to 1835 AD. This charcoal peak matches the time frame (1790 -1835 AD) of a
particularly widespread and severe drought in East Africa, known from oral histories and
Nile River flow records (Nicholson, 1998b). During the drought, fire frequency was
reportedly much higher, and many lakes experienced lowstands (Victoria, Malawi, and
Naivasha, among others) (Nicholson, 1998a; Nicholson, 1998b; Verschuren et al., 2000).
A concurrent peak in charcoal abundance occurs in core LT98 -37M from Mwamgongo,
although its overall magnitude is attenuated relative to the peak in LT98 -58M, perhaps
due to preservational conditions or core location. The- 1810 -1835 AD charcoal peak in these cores may represent increased regional fire frequency related to severe drought conditions, and it suggests that Lake Tanganyika also experienced a lowstand during the
-1790-1835 AD interval. Interestingly, a peak in the siliciclastic fraction of sediment 51 increased shortly after the charcoal peak, which suggests that watershed erosion may have increased upon resumption of rainy conditions because of slope denudation byfires.
If rainfall and lake level were extremely low in the first decades of the19th century,
apparently contradictory lake level indicators from the lowermost samples of shallow -
water core MIT -1 from Gombe Stream National Park can be reconciled.The samples are
composed of very coarse sand (Fig. 2b in Appendix C), characteristically deposited in
very shallow water ( <1 m), and contain a relativelyhigh abundance of the exclusively
deepwater ostracod, Gomphocythere downingi (Park and Martens, 2001). The presence
of G. downingi strongly suggests a lake level >10 m higher than present -day levels
(average= 774.2 ±0.8 m asl, 1885 -1999 AD (Birkett et al., 1999; Evert, 1980)), while the
coarse sand indicates a lake level -15 m lower than present. Historical dataplace lake
level 4 -10 m above present lake level from the late 1840s through the early 1880s (Evert,
1980). I propose that Lake Tanganyika was -15 m lower than present through the first
few decades of the 19th century, explaining the presence of beach sands in the bottom of
core MIT -1, then rose by -25 m quickly enough through the late 1830s and early 1840s
that sediments characteristic of deeper water G. downingi habitats were not preserved
concurrently with the remains of the deepwater ostracod species. The sudden fining of
sediment and disappearance of G. downingi at -17 cm may correspond temporally to the
lake level fall of >10 m that occurred 1874 -82 AD after an alluvial dam at the lake's
outlet, the Lukuga River, was breached. Both the rapid rise in the early to mid -1800s and
the precipitous decline in the late 1800s are consistent with data on lake levels and
rainfall levels throughout equatorial East Africa at the time (Nicholson, 1998a,ó). The 52
ostracod water depth index confirms that the bottom few samples of core MIT -1 are
characterized by deepwater ostracod assemblages. If this lake level interpretation is
correct, it constrains the selection of CALIB- assigned radiocarbon ages for the dated
intervals higher in the core to the low probability estimates for the period 1914 -1959 AD
(Table 2 in Appendix C).
Another, possibly larger peak in charcoal abundance at the bottom of core LT98 -37M
(- 1564 -1585 AD) occurred contemporaneously with an even more severe period of
drought in East Africa.Nile flow records suggest extreme drought conditions in Lake
Victoria from the mid -1500s through the first few decades of the 1600s (Nicholson,
1998b).
The potential robustness and utility of using charcoal as a drought/fire frequency /lake
level indicator in Lake Tanganyika needs to be tested. However, the correlations outlined
above suggest that it is worth examining other existing records from the lake to determine
whether new information on the timing of droughts and lowstands may be gleaned from
them.
Conclusions
Based on the paleoenvironmental and paleoecological data presented here,we conclude that the interval from 4.0 to 2.4 ka was a period of diminished southerly trade
winds, persistent, shallow stratification, and reduced upwelling and nutrientrecycling. 53
Furthermore, stratigraphic, loss -on- ignition, and ostracod abundance data indicatethat the permanent thermocline was shallower than the current depth of core LT97-56V (56 m)
(Fig. 3.10). Dominance by cyanobacteria, as indicated by stable nitrogen isotopesand
815N C:N ratios, provides the primary evidence for this change. Values of C:N and were
remarkably stable from 4.0 to 2.4 ka, suggesting that the lassitude of winds was a
persistent condition throughout this interval. These data extend the duration of Mid
Holocene water column stability described by Haberyan and Hecky (1987) and Talbot (in
press).
While no major or persistent turnover was apparent in ostracod assemblages in
response to the environmental transitions during zone 3 (2.2 ka- present), thespecies
richness and abundance of assemblages in a few intervals (62 -71 cm, 154 -55 cm) were
strongly diminished by higher than normal concentrations of SOM. Despite selective
preservation of adults and heavily calcified species in high SOM intervals, OWDI values
correctly identify contemporaneous high stands (cf. Cohen et al., 1997b). The majority
of reconstructed assemblages showed no signs of taphonomic bias. The fact that no
major ecological turnover resulted from substantial, documented environmental
fluctuations such as lake level changes of the Little Ice Age (0.6 -0.2 ka) suggests that
benthic communities are relatively resilient to natural disturbance. This snapshot of
baseline variability in benthic invertebrate assemblages bolsters the conclusions of Wells
et al. (1999), Alin et al. (2001; 2000), and Palacios -Fest and Cohen (in prep) that
anthropogenic landscape disturbance explains the sudden and dramatic turnover seen in 54 ostracod assemblages in sedimentcores offshore from highly disturbed watersheds in the densely populated northern basin of Lake Tanganyika. 55
Figure 3.1. Map of Lake Tanganyika with locations of allcores discussed in the text indicated. Inset box shows location of Lake Tanganyika within Africa. 56
t t,-nj.
Burundi
4 °S
R. Mala arasi LT97-35V LT97-56V, 53V Lukuga R. LT97-57V _ 6°S LT97-61V LT98-2M Dem. LT00-03 Rep. Congo Tanzania
-8 °S
Zambia ' 29°E ,, 31°E . _ 1 57
Figure 3.2. Depth- radiocarbon age profile for core LT97-56V from Lake Tanganyika. Error bars correspond to 2614Cage ranges. Shaded areawith wavy erosional surface depicts the stratigraphic position of sands deposited at a lowerlake level. 50 2,000 Radiocarbon4,000 6,000age (14C years BP) 8,000 10,000 12,000 150100 250200 350300 59
Figure 3.3. Depth- calendar age profile forcore LT97 -56V from Lake Tanganyika. Error bars correspond to 26 calibratedage ranges. The polynomial line and equation represent the age model used in this paper. Dashed lines correspond to the depth and timing of the end of the depositional hiatus. 60
14,000
12,000
fi:4 A410,000
a 8,000 U to4) at 6,000
-ci a) 4,000 U
2,000
0 0 50 100 150 200 250 300 350 Depth in core (cm) 61
Figure 3.4. Stratigraphy for cores LT97 -56V, LT97 -53V, and LT97 -35Vincluding radiocarbon ages, sedimentary structures revealed by x- radiography, and grain size distribution. A gas gap of 13 cm has been excised from the stratigraphic column for LT97 -56V between 56 and 69 cm. Other profiles from LT97 -56V exclude this interval. Capital letters indicate correlated magnetic susceptibility zones. 134139 14C ages 315±45323±46 depth(cm) 20 0 LT97-56V 14C ages 20 o LT97-53V 14C ages 20 0 LT97-35V 343±34 408060 Ii216111111. _':-.-:-:-:-: .1 390±50430±90 408060 /1/4 608040 ....wwww.. wwwww2.. 667±61 140120100 430±65 140120100 140120100 11111111111...... 2846±611970±79 200180160 DMAA 4. 200180160 200180160 40141-93 260240220 1035±45 260240220 260240220 4910±140 320300280 .0»2 ON~1111 1365 ±70 320300280 320300280 / / / / / /////, 10,210±140 340 0 °lo 100 0 70 140 2088 ±77 340 ±\111111 380360340 ,,,,,,,/,/,/iii, // / /i, Sedimentary structure key: VtV brown mud with black flecks (charcoal ?) cgs 420400380 12,590±95 420400 r`1111 alternating1/'r'IV'/ / / /muds/laminae / and thin sands massive/featurelessmottled/burrowedthin beds and/or wispy beds laminae 480440460 480460440 ;' ;' ;!; sandmissingalternatingshell bed x -rays wispy beds and massive ; Grain size key: 540520500 520500 rhythmite bundles > 1 mm 560 ... , , - sandygravellyplantisolated gravel mattersand gastropods <63-106 63106 Nm µm-1µm mm 580 0 cgs 60 120 63
Figure 3.5. Paleoenvironmental records fromcore LT97 -56V. From left: Calibrated age profiles of magnetic susceptibility, and concentration of carbonate and sedimentary organic matter from loss -on- ignition (from unsieved samples). Fine sediment fraction ( <63 µm) profiles of total nitrogen (TN), stable isotopes of nitrogen and carbon (815N, 813C), and carbon:nitrogen ratio (C:N)of <63 pm sediment fraction. TN -axis is truncated (core top TN = 0.30 %). At right,core zones used in this paper. zone 20001000 4 40003000 -z- 3 60005000 2 -11-12,000{ 0 mag. sus.) (cgs) 140 0 carbonate (wt. %) 5 0 (wt. %) SOM 10 0 (wt. %) TN 0.2 0 615N ( %o) 3 -25 813C ( %o) -19 10 C:N 30 1 65
Figure 3.6. Crossplots of elemental and stable isotopic data for the <63µm sediment fraction from core LT97 -56V. Symbolsare the same for all plots: diamonds represent zone 1 data (-11-12 ka), circles zone 2 (- 6.4 -4.0 ka), squares zone 3 (- 4.0 -2.4 ka), and triangles zone 4 (-2.2 ka- present). Regression lines in (A)are shown for data from zones 2 and 4. (A), (C), and (E) scaled to maximize dispersion of the majority of data points; inset boxes show distribution of all data, includingone outlying value representing the core top sample. Both inset boxes have the same scaling for the y -axis as in the larger version, with a maximum x -axis value of 6. 66
TOC, wt % 815N
0 1.25 2.5 -1 0 1 2 3 . 0.4 0.4 . zone 4 A \I A /' °a °° z A _ 0.2 r p AA . r:, A Í -;A ° p O p!; 0 zone 2 B 0 A 0 -19 -19 A ° ;': A 4ka, A A A -21f A A IA A -21 U A AA A 8O A Fob OA ® Ap LA -23 Op p.-1.10 O -23 :. p o o ® o o C D -25 30 1 30 o . o . o o ® : o 0 o 000 zÜ20 o o 20 4:ck>o A A .,S 9s.,. ,.-.c A O °' 41 4 O ñ' t p ° E F lo " 0 '1.25 2.5-1 0 I 3 TOC, wt % 815N 30 o o
Ü 20 ooO 66.. A° O O g A A ° °° G 10 -25 -23 -21 -19 8130 67
Figure 3.7. From left: Number of species and valves per gram for all samples bearing ostracods. Middle panels: Taphonomic data vs. calendar age profiles-left to right: % carapaces (vs. disarticulated valves), % cracked or broken, % corroded or abraded, % adults (vs. juveniles), % reduction stained, % encrusted, and % yellowed or opaque, indicating incipient reduction staining or encrustation. At right: Loss -on- ignition SOM concentration, DCA axis 1 ordination scores, and ostracod water depth and substrate indexes (OWDI, OSI). Shaded bars are aligned with SOM peaks. Dashed lines indicate samples containing <500 individuals for calculating OWDI and OSI. S4cc .1cS eSS Fcc ,G Q,S `Gc {°, c° Oc b e,a Sca e,11° \o4 SO4' qe CI"' ' S1 o 500 o b c4 c a c c° Scc 1 aa4 >,o 15001000 i 25002000 0 # spp. 50 10° #/g 104 0 I 20 0 800 % of individuals500 40 0 40 0 40 0 40 0 I wt. % 10 -0.5 scores I lo levellake hi crs substrate fn 69
Figure 3.8. Species abundance -calendar age profiles for two shallow (R. ampia, M. emaciata) and two deep (M. n.sp. 2B, G. coheni) species in core LT97 -56V. Center: Calendar age profile for Fisher's a diversity index. Shaded bars represent deepwater intervals where SOM was highest. At right: Ordination (DCA) plot of core LT97 -56V ostracod species abundance data. Numbers correspond to sample numbers discussed in the text. 1 1000 500 20001500 50 2500 0 200 % individuals 400 150 60 a 15 10 levellake hi -1 -1 DCA Axis 1 71
Figure 3.9. Correspondence between the ostracod water depth index (OWDI) and the stromatolitic lake level record of Cohen et al. (1997) for the past 2,500 calendar years. Solid arrows indicate matches between documented highstands from Cohen et al. (1997). The dashed arrow indicates a known lake lowstand of previously unconstrained timing. 2000 AD low OWDI high 2000 760 AD (m aboveLAKE msl) LEVEL770 780 1500 AD 1500 * ? ------: ' ' ' ' - - - 1000 AD .. ALLUVIUMBREACHINGLAKE OPENS N 1000 AD o 500 INCR.PRECIP. b cizt 500 AD z OPENS?L. KIVU U , NO .' (a) 500 BC 500BC CURRENT OUTLET 768ms1 LEVEL 1000 BC 1000 BC Cohen et al. (1997) i 73
Figure 3.10. Compilation of major conclusions of recent paleoclimaticand paleolimnological reconstructions for the Lake Tanganyika basin and a composite lake level curve based on data in this paper and other references. Blacklines indicate lake level, with dashed sections representing intervals with no direct evidence oflake level. Light gray line indicates inferred depth of permanent thermocline, with dashes representing sparseness of direct indicators. Water depths and locations for cores in references: Vincens (1993) - 915 m, northern basin; Haberyan and Hecky (1987) -440 m, southern basin; Talbot (in press) -140 m, southern basin; Cohen et al. (1997) - stromatolites at various depths, throughout basin. References or cores used to generate composite lake level curve indicated at far right, with reference numbers indicatedin Table 3.2 (ref. 5 = Cohen et al. [1997]). Age (ka BP) 14C (cal) 0 (0) Vincens(1993) Haberyan & Heckybasin closed for (1987) (in press)Talbot Cohen et al. (1997) this paper - Composite lake level curve + Refs. 1 (0.9) increasing(forests aridity decline, muchweaker of thermal period, no data, core fluctuated+12 between and -40 m of lake level fluctuatesm), (always oxycline >_ -40 deepens,zone 4: lake level la, 3, 5 32 (3.2)(1.9) grasses increase) CaCO3 deposition stratification, highest top lost P current outlet level ( -775 m asl)* m, zoneoxycline 3: lake abovetrade level corewinds > -40 stronger , °. 4 (4.5) thermal stratificationopen basin, strong water column no data prior site, weak trade windszone 2: lake level i 1 i 1 la, 3 5 (5.7) highest Holocene temperaturerainfall and thermoclineopen basin, established permanent southerly trade windsstability, reduced to J000 BC terrestrial organic > -40 m, high input 1 ' , 76 (7.8)(6.8) (maximumand extent diversity of forests) annualupwelling mixing at core ofsite, completeopen basin, P no sediment, lake level la, lc, Id, 3 9 (10.2) 8 (8.8) water column upwelling, productivityrelatively constant conditions below core site? 1110 (13.0)(11.4) throughout, butdry and cool stratification ofpermanent watercolumn, thermallake level closed basin, peak in phytoplanktonproductivity at 11.3 (13.3) between -50 and -65 m zone 1: lake level la, Id 1312 (15.4) (14.0) gradually increasing(open woodlands) temperaturerainfall and lake closed, rising rising 4- 1550 AD, 430±110* highstands: AD; 1250- no core record 1514 (16.8)(18.0) lowest lake level(- 200 -300lake m) closed, *core site transgressed at 15.2 (18.2) century (dating unclear)lowstands:late 500 16'h AD, -early 194 m below present level-100 -50 0 75
Table 3.1. Radiocarbon dates for core LT97 -56V from Lake Tanganyika.
Sample Material Depth in Fraction 14C Calibrated year BPArea under number dated core (cm) modern 14C age (2e) probability curve*
Core LT97 -56V
AA -43467 single leaf 0 -1 .9606 323 ± 46 390 (483 -297) 1.000 AA -43468 single leaf 2 -3 .9834 134 ± 39 225 (279 -171) .423 102 (152 -51) .405 25 (45 -6) .153 AA-28408plant fragments 17 -18 .9615 315 ±45 385 (477 -293) 1.000 AA-43385single leaf 50 -51 .9582 343 ± 34 395 (480 -310) 1.000 AA-27665plant fragments92 .9203 667± 61 617 (693 -540) 1.000 AA-43469single leaf 162 -63 .7825 1,970 ± 79 1922 (2117 -1726) 1.000
AA-43470single leaf 198 -99 .7017 2,846 ± 61 2961 (3082 -2840) .869
AA-43471single leaf 242-43 .6067 4,014 ± 93 4488 (4735 -4240) .904 AA-43472single leaf 298 -99 .54254,910 ±140 5683 (5928 -5438) .916 AA-43473single leaf 322 -23 .283610,120 ± 14011,827 (12,392- 11,261).938
*All calibrated ages representing >0.100 relative area under probability distribution reported. 76
Table 3.2. Details of cores discussed in text
Water depthLength Latitude Longitude Reference Core (m) (cm) (S) (E) [ #]
LT97 -56V 56.0 342 5 °46.33' 29 °56.03'this paper [ la]
LT97 -57V 75.8 563 5 °47.17' 29 °55.95'this paper [ 1b], O'Reilly (ms.) [2]
LT97 -53V 70.4 584 5 °46.29' 29 °55.89'this paper [ lc]
LT97 -35V 58.8 528 5 °23.05' 29 °44.96'this paper [ ld]
LT97 -61V 67.4 365 5 °58.13' 29 °49.44'this paper [le]
LT98 -2M 110.0 49 6.165° 29.706° Cohen et al. (1999) [3]
LT00 -03 608.0 170 6.70° 29.86° Zilifi & Eagle (2000) [4] 77
CHAPTER 4: CONCLUSIONS
The research projects in this dissertation are thematically unified by my underlying motivation to contribute to biological conservation efforts through a combination of ecological, paleoecological, and paleoenvironmental approaches. In this section, I will first synthesize the most important conservation- related results that arose from papers in this dissertation. Secondly, I will discuss the utility of ostracods as environmental indicators in Lake Tanganyika. Finally, I will briefly describe some evidence, not detailed elsewhere in this dissertation, describing a possible correlation of paleoenvironmental indicators in sediment cores with published drought chronologies in
East Africa.
Microinvertebrate Paleoecology and Conservation
In my first dissertation project (Appendix A), I suggested that ostracods may serve as conservative indicators of anthropogenic watershed disturbance, as they appear to have a higher response threshold to sediment inundation than do either fish or molluscs (Alin et al., 1999). If they are to be useful indicators of benthic lacustrine community integrity as a whole, it is important to establish that their paleoecological record is both robust and representative of a habitat area larger than that physically represented by the cross- sectional area of a coring device.
Ostracods are preserved abundantly in sediment cores from Lake Tanganyika
(typically hundreds to thousands of individuals per gram of sediment), and taphonomic alteration of assemblages such as corrosion is easily detected. Given the generally good 78 quality of the specimens, it is relatively straightforward to collect statistically robust sample sizes for finely sampled core intervals (sub- centimeter sampling would in most cases yield sufficient material, although time -averaging of assemblages or physical/biological mixing of sediments may render sampling at such high resolution pointless (cf. Anderson, 1993; Anderson and Battarbee, 1994)). Thus, the major question of how much habitat area and what length of time an assemblage represents is crucial in determining the validity of using ostracods as indicators for benthic communities.
My taphonomic comparison of life, death, and fossil assemblages indicated that death and fossil assemblages are remarkably consistent in species composition and abundance patterns compared to life assemblages, suggesting that the spatial and temporal integration of post- mortem assemblages makes them more representative of the habitat or benthic community as a whole than life assemblages of ostracods could be. In Appendix
B, I discussed in detail my idea that spatiotemporally averaged assemblages of dead and fossil ostracods are more comparable to fish communities in Tanganyika, which are characterized by extremely low variability in membership on timescales of years to a few decades, than are life assemblages. Low variability among death assemblages
(representing one to two years accumulation) and fossil assemblages (representing deposition over the past few decades) suggests that the species pool at the study site has been relatively constant, although indications of anthropogenic change at this site have been described in detail in Appendix C and are apparent in the upper few samples of both shallow -water cores. During the course of this study, I also observed thatmany rare species are among the persistent members of a site's species pool. This observation 79 could not have been made based on sampling of live collections and does not appear to be based on selective preservation or degradation of species related to shell thickness and calcification. I suggested in Appendix B that this observation may provide a predictive tool for identifying the component of the ostracod fauna that is resilient and may be more likely to persist through or recolonize after a disturbance event. In contrast, another component of the fauna can be identified as appearing ephemerally in the fossil record.
These species may be comprised of shifting populations rather than representing
consistent members of the local /regional species pool.
I believe that the evidence amassed in this dissertation demonstrates that ostracods are
useful and reliable indicators of benthic community conditions. Their conservative
nature as indicators gives me confidence that changes reflected in ostracod assemblages
through time most likely correspond to more substantial transitions in the macrofauna at
the same site.
Ostracods as Indicators of Environmental Change
Martens (1994) estimated that possibly as many as 200 species and 25 genera of
ostracod inhabit Lake Tanganyika, although only about half that number of species and
two thirds that number of genera have been taxonomically described. Among described
taxa, endemism is 38% at the genus level, and 94% at the species level for ostracods.
High endemism is likely both a reflection and consequence of the habitat isolation
between adjacent patches of similar habitat that characterizes Lake Tanganyika's lake
benthic ecosystem. Tanganyikan ostracod death and fossil assemblages are characterized 80 by having a handful of species that are moderately abundant(range: 1- 15% usually) and
appear in most samples, complemented by asubstantial number of rare species that occur
in far fewer samples and in lower abundance generally. In deepwater,it is typical for
50% of species account for -90% of individuals, while in shallow water,only 25 -30% of
species account for the same percentage of individuals. Thus, most species areextremely
rare and are likely to elude detection in routineecological sampling. As a result of the
rarity of most species and the taxonomic uncertainty of many Tanganyikan ostracod
species, little is known about the ecological preferences of individual species, making the
use of individual ostracod species as environmentalindicators difficult.
These considerations make it extremely difficult for one to collect sufficiently large
live samples of the ostracod species that might be most useful as indicator species, i.e.
those with narrow ecological tolerances, in order to sufficiently delineate those
tolerances. While it has been convincingly argued that paleoecological research may
effectively and, more importantly, non -invasively contribute important information on
the ecological requirements of endangered species (Steadman, 1995), such an approach is
unlikely to be useful for illuminating ostracod ecology. Unlike marine
macroinvertebrates, ostracods are small enough that post -mortem transport out of life
habitats is more likely than not, making inferences about life habitat from death
assemblages unreliable (cf. Kidwell and Flessa, 1995). Furthermore, Lake Tanganyika is
of sufficiently large volume to be buffered from the dramatic fluctuations in salinity that
characterize many smaller lakes. In many cases, it is possible to develop quantitative
inferences on environmental tolerances, e.g. salinity, based on collections of 81 cosmopolitan ostracod species from large numbers of lakes with differing environmental conditions (e.g. Cohen et al., 1983). Lake Tanganyika offers different challenges than most lakes in this respect, both because of its relatively buffered environmental fluctuations and the extremely high levels of species endemism. Finally, it is likely that, because of the tremendous differences in organismal scale between ostracods and their observers, i.e. us, we are not able to sample the environment at a sufficiently fine -scale to detect or quantify the environmental variables that control the distribution of ostracods both individually and as species (sensu Levin, 1992).
Despite these substantial barriers to identifying fine -scale ecological tolerances for particular Tanganyikan ostracod species, there are a few species which provide unambiguous information about their environment. For instance, Gomphocythere downingi is a deepwater ( >25 m) species. Finding G. downingi fossils in shallow -water cores is a clear indication that water depth has been substantially deeper in the past, as upslope transport and contamination of assemblages seems unlikely. However, few species appear to have such well -defined tolerances with respect to depth (or substrate, etc.), making paleoenvironmental interpretation based on their presence or absence difficult. This is particularly the case because regional endemism of ostracod species within the lake adds a potentially confounding biogeographic signal to environmental interpretations as well.
In order to avoid the problem associated with relying on one or a few Tanganyikan ostracod species for paleoenvironmental reconstruction, I used the live ostracod occurrence database for Lake Tanganyika created and maintained by Andrew Cohen as a 82 basis for a canonical correspondence analysis (CCA). The database comprises 85 ostracod life assemblages collected from sites around Lake Tanganyika with various water depths ( <1 to >50 m), substrate types, subaquatic vegetation, and disturbance levels
(subjectively assigned to low, moderate, and high disturbance categories on the basis of sedimentation impact related to watershed deforestation). On the basis of the CCA, it was clear that ostracods are cleanly segregated by depth (Axes 2 and 3) at the assemblage level into shallow ( <20 m) and deep ( >20 m) categories, and somewhat less well differentiated on the basis of substrate affinity into three categories: muddy, sandy, and rocky. Primary axes were also significantly correlated with latitude, longitude, and disturbance. Using species scores, I assigned species to primary depth and substrate affinity categories, recognizing that some plasticity in ecological preferences occurs in most of the species.
By calculating simple ratios of numbers of individuals in all shallow species versus those in all deep species for all samples, I generated an ostracod water depth index
(OWDI-discussed further in Chapter 3) that accurately reconstructed known lake level changes during the last millenium, despite the fact that not all species in the core examined occurred in the live database (the training data set). The OWDI curve for an additional core also correctly identified known lake highstands. These results suggest that OWDI generates reproducible results, despite changes in the particular assemblages of species in each cores.
I also used results from the CCA to generate a substrate index based on ostracods
(OSI) by comparing the abundance of all fine -grained (sand and mud) species with the 83 abundance of coarse- grained species. OSI clearly differentiated life, death, and fossil assemblages on the basis of the bulk substrate affinity of the assemblages (Chapter 3).
For the deepwater core examined in Chapter 3, the substrate index was very similar to the depth index, a reflection of the correlation between water depth and substrate grain.
Although I did not expressly examine the distribution of samples with respect to disturbance, it would be interesting to determine whether a disturbance index could be defined in a similar manner. In cores collected offshore from heavily disturbed watersheds, mostly in Burundi, faunal changes were reflected mainly in the disappearance of rare species, increasing dominance in the remaining species assemblage, and declining species richness. Such obvious patterns are straightforward to interpret, particularly when they correlate well with changes in sedimentation rate or substrate. It is possible that a disturbance index based on the collective disturbance tolerances of ostracod assemblages may be able to detect more subtle changes at less extensively disturbed locales. It would be interesting to reevaluate the results of ostracod assemblage reconstruction from the many cores collected from various parts of Lake Tanganyika with such a method.
The OWDI and OSI results suggest that the assemblage -scale ecological tolerances reflected in a database like this can be extremely useful in paleoecological reconstruction, despite the fact that current knowledge on the ecology of individual species is rudimentary. An increase in the taxonomic, geographic, and depth rangecoverage of the database is desirable to improve the ability of indexes based on the database to 84
discriminate and resolve environmental conditions. However, even with current coverage
levels, the database yields useful, accurate environmental indicators.
I can foresee many future applications of the results of the CCA/environmental index
approach in Lake Tanganyika, which could be tailored to the particular research question
at hand. For instance, my choice of cutoff between shallow and deepwater species (20
m) was somewhat arbitrary. If an investigator was interested in recreating lake level
fluctuations from a shallow -water core, it might be useful to replace a shallower threshold
for the 20 -m cutoff. There may also be interesting applications to biogeographic
questions, as primary CCA axes were also strongly correlated with both latitude and
longitude, which probably reflects the influence of local endemics in the Tanganyikan
ostracod fauna. Finally, my results suggest that an ecosystem does not necessarily have
to be well known from an ecological or taxonomic standpoint to yield reliable
paleoecological and paleoenvironmental information. For instance, ata lake like Issyk
Kul in Kyrgyzstan, virtually unknown in the English- language scientific literature (cf.
Liu, 1999), it may be possible to employ ostracods as meaningful paleoenvironmental
indicators with a relatively minimal field calibration effort. 85
APPENDIX A: EFFECTS OF LANDSCAPE DISTURBANCE ONANIMAL COMMUNITIES IN LAKE TANGANYIKA, EAST AFRICA
Conservation Biology
The Journal of the Society for Conservation Biology Volume 13 Number 5.October 1999
Cover. Twoimmature sea turtlesof the genus Cbelonia, caught in the Bismarck Sea near Manus Island, Papua New Guinea. Differences in color, size, and shell morphology of the "black turtle' °eft) and 'green turtle' (right) and their sympatry in some localities suggest they are distinct species. New genetic data (this issue), however, show no major divergence between them, setting up a debate on evolutionarily versus geopolitically signifi- cant units. Photo by Peter C. H. Pritchard. Sec pages 990 -1016. Contents
Editorial 953Conservation Medicine GARY L MEFFE Letter 955Graduate Conservation Education LENA ERIKSSON Issues in International Conservation 956Endangered Species Legislation beyond the Borders of the United States JUSTINA C. RAY AND JOSHUA R_ GINSBERG 959The Australian Endangered Species Protection Act 1992 J. C. Z. WOINARSJ AND ALkRIC FISHER 963 Endangered Species Protection in Canada G. G. E. SCUDDER 966 The Role of Legislation in Conserving Europe's Threatened Species IAN F. G. Md. FAN, ANDREW D. WIGHT, AND GWYN WILLIAMS Essay 970Reciprocal Model for Meeting Ecological and Human Needs in Restoration Projects CATHY GEIST AND SUSAN M. GALATOWTTSCH Conservation in Practice 980Conservation Endocrinology: a Noninvasive Tool to Understand Relationships between Carnivore Colonization and Ecological Carrying Capacity JOEL BERGER, J. WARD TESTA, TOM ROFFE, AND STEVEN L MONFORT Conservation Forum 990Evolutionary Significant Units versus Geopolitical Taxonomy: Molecular Systematics of an Endangered Sea Turtle (genusCbelonia) STEPHEN A. KARL AND BRIAN W. BOWEN 1000Status of the Black Turtle PETER C. H. PRITCHARD 1004Using Character Concordance to Define Taxonomic and Conservation Units JAMES M. GRADY AND JOSEPH M. QUATTRO 1008Molecular Systematics, Ethics, and Biological Decision Making under Uncertainty KRISTIN SHRADER- FRECHETTE AND EARL D. MCCOY 1013In War, Truth Is the First Casualty BRIAN W. BOWEN AND STEPHEN A. KARL Contributed Papers 1017Effects of Landscape Disturbance on Animal Communities in Lake Tanganyika, East Africa SIMONE R ALIN, ANDREW S. COHEN, ROGER BILIS, MASTA MUKWAYA GASHAGAZA, ELLINOR MICHEL, JEAN-JACQUES TIERCELIN, KOEN MARTENS, PETER COVEUERS, SIMA KEITA MBOKO, KELLY WEST, MICHAEL SOREGHAN, SONA KIMBADI, AND GASPARD NTAKIMAZI 1034Statistical Power of Presence -Absence Data to Detect Population Declines DAVID L STRAYER
Punted on recycled paper. continued on next page 86
From: "Doyle Lindsay"
Dear Dr Alin,
In response to your request to includeyour article in your thesis. I am happy for you to do this subject to the usual acknowledgements, and that UMI sell on demand single copies only of the article for scholarly purposes. If you wish to distribute or publish your thesis commercially please reapply for permission.
Best wishes, Lindsay
Lindsay Doyle Permissions Controller Blackwell Publishing
Contact details: Blackwell Publishing Osney Mead Oxford OX2 OEL T: +44 (0) 1865 206430 F: +44 (0) 1865 206096 E: [email protected] 87
Contributed Papers
Effects of Landscape Disturbance on Animal Communities in Lake Tanganyika, East Africa
SIMONE R. ALIN,' § §§ ANDREW S. COHQV; ROGER BILIS,f MASTA MUKWAYA GASHAGAZA,t ELLINOR MICHEL,§ JEAN-JACQUES TIERCELIN," KOEN MARTENS,tt PETER COVELIERS,tt SIMA KEITA MBOKO, §§ KELLY WEST,`" MICHAEL SOREGHAN,ttt SONA KIMBADI, §§ AND GASPARD NTAIIMAZItt* 'Department of Geosciences, University of Arizona, Tucson, AZ 85721, U.S.A. tJ.L.B. Smith Institute of Ichthyology, Somerset Street, Private Bag 1015, Grahamstown 6140, South Africa *Faculty of Apiculture, National University of Rwanda, B.P. 117, Bucare, Rwanda S institute for Systematics and Population Biology, University of Amsterdam, P.O. Box 94766, 1090 GT Amsterdam, The Netherlands ~Département des Sciences de la Terre, Université de Bretagne Occidentale, 6, Avenue Le Gorgeu, 29287 Brest Cedex, France ttRoyal Belgian Institute of Natural Sciences, Vautierstraat 29, 1040 Brussels, Belgium **Rue du Prince Royal 89, 1050 Brussels, Belgium SSCentre de Recherche en Hydrobiologie/Uvira Station, Democratic Republic of Congo, B.P. 254, Bujumbura, Burundi '~Lake Tanganyika Biodiversity Project, B.P. 1119, Bujumbura, Burundi tttSchool of Geology and Geophysics, University of Oklahoma, Norman, OK 73109, U.S.A. ***Université du Burundi, Département de Biologie, B.P. 2700, Bujumbura, Burundi
Abstract: Watershed deforestation, road building, and other antbmpogenic activities result in sediment in- undation of lacustrine babitats. In Lake Tanganyika, this threatens the survival of many rock- dwelling spe- cies by altering the structure and quality of rocky babitats. We investigated the relationship between babitat quality, as related to watershed disturbance intensity, and the biodiversity of faunal communities at three rocky littoral sites of low, moderate, and bigb disturbance. Turbidity measurements and other environmen- tal observations confirmed that our lake sites represented a gradient of disturbance conditions. We docu- mented differences in species density (number of species per constant area or time), species richness, abun- dance, and tropbic ecology for fishes, molluscs, and ostracods Fish censuses were performed by scuba divers at 1 -20 m and by remotely operated vehicle (ROV) at 40 -80 m. in the ffisb survey, abundance, species den- sity and richness, and berbivory reached their maxima at intermediate water depths. The depth range of ber- bivores, however, was restricted at higher- disturbance sites. The ROV fish surveys at the bigb-disturbance site showed bigb species richness despite low species density and abundance, and piscitvres were proportionally more prevalent than in all other surveys Molluscs censused by diver quadrats and sieve samples showed de- creasing species richness and species density (sieve samples only) with increasing disturbance and no signifi- cant abundance trend Ostracod species richness was similar between low- and moderate - disturbance sites but was markedly lower at the high - disturbance site (species density and abundance data were not avail- able). Our faunal analyses suggest that all tbree taxonomic groups are negatively affected by sediment inun- dation but may bave varying response thresholds to disturbance. Further, Ibis study emphasized the utility of using complementary survey tecbniques to monitor and ultimately manage biodiversity in complex fresbwa- ter ecosystems.
SS S email alineu.arizona. edu Paper submitted December 23. 1996: revised manuscript accepted February 10. 1999. 1017
Conservation Biology. Pages 1017-1033 volwnc 13. No. 5. October 1999 88
1018 Contributed Papers Alin et ai
Efectos de la Perturbación del Ambiente sobre las Comunidades animales en el Lago Tanganyica, Africa Oriental Resumen: La deforestación de las cuencas, construcción de caminos y otras actividades antropogénicas re- sultan en la inundación de sedimento en los habitats lacustres. La sobrevivencia de muchas especies está en peligro en el Lago de Tanganyica debido al cambio de la estructura y calidad de habitats rocosos. En este estudio investigamos la relación entre la calidad del habitat, en función de la Intensidad de la perturbación de la cuenca, y la biodiversidad de las comunidades faunísticas en tres áreas litorales rocosas caracterizadas por perturbación baja, moderada, y alta. Inicialmente, seleccionamos estos sitios basados en la extensión de la deforestación de las cuencas. Mediciones de turbidez y otras observaciones del medio ambiente confirma- ron que estos sitios del lago representan un gradiente de las condiciones de perturbación. Documentamos las diferencias de densidad de especies (número de especies en una área o tiempo constante), riqueza de espe- cies, abundancia y ecología trófica de peces, moluscos y ostracodos. El censo de peces fué becbo por buzos en los primeros 20 m de profundidad y por un vebículo operado remotamente (ROV) entre los 40 y 80 m. En el muestreo de peces, la abundancia, densidad y riqueza de especies y bábitos herbívoros alcanzaron sus máxi- mos en profundidades intermedias. Sin embargo, el rango de profundidad de berbivoros, fue restringido en sitios de alta degradación. Las muestras de ROV en los sitios de alta degradación mostraron altas riquezas de especies a pesar de la baja densidad y abundancia, y los pscfvoros fueron proporcionalmente más comunes que en ¡as demás muestras. Los censos de moluscos obtenidas en cuadrantes mediante buceo y muestras tam- izadas mostraron un decremento de riqueza y densidad de especies (muestras de cribado únicamente) con un incremento de la perturbación, sin ninguna tendencia de abundancia significativa. La riqueza de espe- cies de ostrácodos fue similar entre los sitios de perturbación baja y moderada pero notablemente menor en los sitios de alta degradación (no bubo datos de densidad y abundancia específicas). Los análisis de fauna sugieren que los tres grupos taxonómicos están afectados negativamente por la inundación de sedimentos pero puede baben variantes en la respuesta a la perturbación Además, este estudio da énfasis en la utilidad de usar técnicas complementarias de muestreo para observar y eventualmente manejar la biodiversidadde ecosistemas complejos de agua duke.
Introduction creasing levels and maximum depths of benthic primary productivity in the littoral zone. In undisturbed areas of Lake Tanganyika lies in the western branch of the Afri- Lake Tanganyika, the rocky littoral zone normally expe- can Rift. It is one of the world's largest, deepest, and old- riences exceptional water clarity; the mean euphotic est lakes, and harbors over 1400 species of animals, zone extends to 28 m (Hecky 1991), although benthic plants, and protists, many endemic to the lake (Coulter algae often grow well below this depth. Incoming sedi- 1994). The ecosystem is characterized by species -poor ments may also bind or release nutrients or toxins, alter- pelagic communities but diverse littoral- sublittoral com- ing energy flows through communities. Cracks and crev- munities. The latter include largely endemic species ices fill with sediments, reducing habitat heterogeneity. flocks of cichlid and noncichlid fishes, molluscs, and This reduces the range of potential habitat types and crustaceans that are renowned for their morphological, predation refuges for many highly specialized, steno- ecological, and behavioral diversity (Coulter 1991). Ex- topic, or juvenile fish and invertebrate species. Burial of tant biodiversity in Lake Tanganyika is threatened by a rocky substrates also decreases the surface area available variety of human activities. for epilithic algal growth and invertebrate colonization. Sediment inundation resulting from watershed defores- Concern about the potential effects of these anthropo- tation and other activities (e.g., municipal and industrial genic disturbances is high in the riparian region because discharges, road building) is among the most immediate of the economic importance of the lake's biotic re- and important for littoral -sublittoral communities (Cohen sources. Given this concern, how should the conserva- et al. 1993; A. Vandelannoote, personal communication). tion biologist or manager monitor the health of such a Cohen et al. (1993) found low species richness among complex freshwater ecosystem? Lake Tanganyika pre- fishes and ostracods associated with adjacent watershed sents a formidable challenge because its littoral commu- deforestation. Unfortunately, the responses of different nities rival some marine environments in ecological com- taxonomic groups were not directly comparable because plexity, and natural variability is poorly understood. We of differences in sampling methodologies. attempted to ascertain the consequences of landscape Increases in the suspended sediment loads that are disturbance for lake biota by integrating environmental carried by influent water masses can alter rocky benthic and faunal surveys to determine whether species density, habitats in several ways (Cohen et al. 1993). Increased species richness, and abundance change systematically turbidity may reduce light penetration and result in de- along a disturbance gradient. In particular, we were in-
conservation Biology Voiumc 13, No. S. Octobcr 1999 89
An et ac Contributed Papers 1019
terested in determining whether different taxonomic tershed adjacent to our study site is 0.23 km2. Some defor- groups responded in concert to environmental stress and estation has occurred along the eastern side of the penin- whether we could detect changes in trophic structure sula since the late 1980s, including areas adjacent to our among fish communities. study site, but the watershed remains largely unfilled. Our primary aim was to assess the effects of watershed Our moderate-disturbance study site was at Luhanga, disturbance on the unique fauna of lake Tanganyika's approximately 11 km south of Uvira in the Democratic nearshore rocky habitats. In addition, we were interested Republic of Congo (Tat 3 °30.35'S, long 29 °9.43'E), where in developing protocols for monitoring biodiversity in human population density is 20 -50 people/km2 (Répub- this lake. Extensive knowledge of the diversity and distri- lique du Zaire 1988). The watersheds draining this area bution of Tanganyikan organisms (e.g., Brichard 1989; are considerably larger than those at Cape Banza: the Coulter 1991) and the immediacy of human population area of the watershed adjoining this study site is 6.96 pressures threatening the lake's ecological dynamics km2. Deforestation has accelerated in this area during the (Caljon 1992; Bootsma & Hecky 1993; Cohen et al. 1993; past decade, and much of the watershed has been con- Coulter & Mubamba 1993; Lowe- McConnell 1993) made verted to cassava and banana cultivation. Nearby road Tanganyika an appropriate subject for this study. work resulted in rubble accumulation in the shallower portions of the study location in 1992. Our high- disturbance site was located approximately Methods 3 km south of Gitaza in northern Burundi, offshore of the 28.9 -km highway marker (lat 3 °37.45'S, long 29 °20.93'E), Study Sites and Data Collection where human population density is 200 -399 people/km2 (République du Burundi Bureau Central de Récensement We collected census data for fishes, molluscs, and ostra- 1990). Watersheds in this region are also larger than on cods from rocky habitats at three sites in order to com- the Ubwari Peninsula, with the watershed onshore from pare community diversity and sediment inundation. Our this study site having an area of 2.33 km2. Deforestation sites (Fig. 1) were located in the northern basin of the and conversion to cassava and banana cultivation is nearly lake, where human population densities are most vari- complete. Other disturbances include road resurfacing able and a range of disturbance conditions exists. Sites with rubble input to shallow waters in 1992, fish were initially selected based on visually obvious differ- tion for the international aquarium trade, and artisanal ences in the extent of watershed deforestation. The ex- fishing activity (gill netting and purse seining). tent of deforestation was confirmed qualitatively by At all sites, rocky habitats were interspersed with ground surveys because recent LANDSAT images without sandy or muddy habitats, which serve as sources of non- cloud cover were not available for all three drainages. rocky habitat species. Bedrock at the three study sites is Our low-disturbance site was at Cape Banza on the Ub- similar in chemical composition, so qualitative differ- wari Peninsula in the Democratic Republic of Congo (lat ences in incoming sediment stem from area, gradient, 4°3.07'S, long 29°14.64'E). Human population density is and land -use differences among the watersheds. lower here than at our other sites ( <5 people/km2; Répub- All turbidity measurements and faunal census data lique du Zaire 1988), and there are no roads. Watersheds were collected during October - December 1992, coin- are small, limiting sediment discharge. The area of the wa- ciding with the rainy season at Lake Tanganyika, when
b) 29° ;
Figure 1. Map of east Africa, with the largest of the African Great Lakes depicted, from north to south: Lake Victoria, Lake Tanganyika, and Lake Malawi (D.R. Congo, Democratic Republic of Congo) (a) and location of biodiversity study sites (b). The area of continuous rocky littoral habitat near each study site is indicated by black shading (adapted from Brichard 1989).
cmserration Biology Vellum 13. No. 5. October 1999 90
1020 ContributedPapers Alin et al
lake and river turbidity and sediment discharge were all Ribbink et al.1983; taxonomy following Poll 1956, at their annual maxima. Given our limited window of op- 1986; Liem & Stewart 1976; Travers 1988; Daget et al. portunity to use the remotely operated vehicle (ROV), 1991). At least two replicate samples were collected at we chose to sample during the wet season in order to in- each depth, each taking 20 -30 minutes. corporate direct measurements of environmental differ- For deep -water fish surveys, we employed a Phantom ences among our sites. II ROV operated from the mother ship via a 300-m tether. The ROV was equipped with a PISCES remote conductiv- ity-temperature-depth (CID) probe with pH and dis- Evaluation of Water Quality solved oxygen sensors and a video camera with laser te- Turbidity measurements were collected from lake wa- lemetry for size determinations and telephoto capability ters and influent rivers near each site with a Hach for close -up identification work. Constant area transects (Model 2100P) turbidimeter. These values serve as a proved impractical for the ROV fish surveys. Therefore, short -term proxy for measures of sedimentation rate ROV census samples were time transects, with all fishes over longer intervals. Turbidity measurements are on a observed in 10- minute intervals of video footage at each logarithmic scale and were thus log -transformed before study depth being counted and identified using only seg- statistical analysis. Lake and river measurements were ments of videotape in which the ROV was moving along treated separately. Measurements taken within the lake the transect (field of view approximately 2 m wide at 1 m reflected background turbidity levels in the water col- in front of the camera). Three to five replicate samples umn but were strongly affected by time of day and wind were tallied at each depth. Because the pilot maintained speed, which influence standing primary productivity a fairly constant ROV speed of approximately 0.1 knot levels. Productivity levels also depend on nutrient con- (185 m/h), each 10- minute interval corresponds to ap- centrations, which may be affected by incoming sedi- proximately 30 m of travel. The ROV typically cruises ap- ment loads. River measurements were more important proximately 1 m off the bottom. for assessing influent river sediment inputs; these values The scuba and ROV transect data are not directly com- fluctuated with discharge pulses following rainfall and parable because of differences in transect area and time reflect actual sediment loads in influent rivers. (Bortone et al. 1986), acuity of the human eye compared The maximum depth of benthic algal growth was re- to ROV lenses (Greene & Alevizon 1989), variable re- corded by scuba divers and a Phantom II ROV. The sam- sponses of fish species to divers versus the ROV, and the pling arm of the ROV was used to excavate small effects of decreasing light levels with depth on identifi- trenches to determine whether sediments at each site cation accuracy and fish reactions to diver or ROV pres- were laminated or bioturbated. Laminated, or undis- ence. The scuba and ROV survey methods are both turbed, layers of sediment indicate depositional condi- prone to certain biases. Diurnal transects underestimate tions (e.g., high sedimentation rate or anoxia) unsuitable nocturnal as well as cryptic species counts and abun- for faunal inhabitation. Denser influent river waters fol- dances. Because our transect areas were large, small spe- low lake bottom topography and deposit more sediment cies were also underestimated. Although we tried to se- on rocky slopes than would be evident solely from obser- lect sites with comparable habitat complexity, our sites vation of lake surface water turbidity (Craig 1974; differed in this respect because of sediment inundation Wombwell 1986). Although surface turbidity measure- at the more disturbed sites. With fewer refuges at these ments reflect individual pulses of sediment influx, obser- sites, we might expect to detect a higher percentage of vations of sediment quality and algal growth were useful all species and individuals living within the transect for inferring relative sediment deposition and light pene- area. This tends to inflate diversity estimates at the more tration conditions over the longer term among sites. disturbed sites, giving a conservative estimate of diver- sity differences among sites. Many researchers exclude Faunal Censuses wandering species -those that do not permanently re- side in the target habitat -because they can bias species Quantitative fish census data were collected at each site counts toward an overestimation of within- habitat spe- by scuba divers at depths of 1, 5, 10, and 20 m and by cies diversity. For this reason we have followed other analyzing ROV videotapes taken at depths of 40, 60, and workers (e.g., Luckhurst & Luckhurst 1977) and ex- 80 m. Bottom time available to divers at 40 m was too cluded wandering species from all analyses (they are in- short to collect quantitative data, but qualitative data dicated in Appendix 1), although their observed abun- were collected at 40 m at all sites for comparison of the dances are reported in the fish species list. two census methods. Other potential censusing biases, which we believe The scuba transects involved laying out two parallel were not major concerns in this study, are as follows. 25-m transect lines 2 m apart at each depth, allowing Visibility at all sites was 2 -3 m, so we do not consider sufficient time for fishes to resettle, and then counting this a likely source of error. Consistency of species iden- and identifying all fishes between the transect lines (cf. tifications among divers was not a problem either, be-
Con.rcrration Biology Volume 13. No. 5. October 1999 91
Alin et a1 Contributed Papers 1021
cause divers discussed survey observations after each size among compared samples served as the standard, and dive and agreed upon consistent names for species in extremely small samples were discarded. We report both question. One individual was responsible for the identifi- species density from raw species counts and species rich- cation of all fishes on the ROV transects, so those data ness estimates based on rarefaction. We use the term di- were also internally consistent. versity to encompass measures of both species density Fish species were assigned to six trophic groups based and species richness. on available gut content and behavioral data: benthic al- Faunal similarity among sites and across depths was givores, phytoplanktivores , benthic invertivores, zooplank- calculated according to both Jaccard and Simpson simi- tivores, piscivores, and omnivores /unknown species (Poll larity indices (Simpson 1960; Magurran 1988). Compari- 1953, 1956; Gashagaza & Nagoshi 1985; Mbomba 1983; sons of faunal similarity were made between pairs of Nshombo 1983; Yamaoka 1983; Brichard 1989; M. Hori, sites, with species lists pooled across depths for each personal communication) (Appendix 1). These groups site, and between adjacent depths within each site. Both were intended to represent the dominant food type con- indices are biased when species counts are highly dis- sumed by each species but do not necessarily reflect the similar; the Jaccard index is biased toward greater differ- dietary breadth of most species. ences, whereas the Simpson index is biased toward Mollusc diversity data were collected by scuba divers in greater similarity. the form of visual surveys (for larger taxa) and sieved sed- We computed means for species density, rarefied spe- iment samples (taxonomy following von Martens 1897; cies richness, and abundance for each site at each depth Leloup 1953; Brown 1994; West 1997). Visual surveys and for each site with depths pooled. Generally, repli- consisted of tallies by species of all molluscs within 1-m2 cate sampling was inadequate for comparisons among sampling quadrats. Quadrats were placed haphazardly on sites involving individual depths, but analysis of variance rocky substrate at each study depth, and smaller rocks with replicates combined among depths provided con- were turned over to count all live molluscs. There were servative tests for between -site differences because pool- one to three replicates at each depth. Sediment samples ing multiple depths increased within -site variability. Anal- were collected from 1-m2 areas so that census coverage ysis of variance was performed, and the null hypothesis could include smaller and infaunal species. Sediments was that species density or richness, abundance, or tur- were coarsely sieved with a 2.0-mm screen. Both live and bidity varied more within sites (i.e., across depths) than recently deceased individuals were counted (Appendix among localities. When the null hypothesis could be re- 2), and the latter were identified with the aid of Rose Ben- jected, pairwise t tests were used to compare individual galstain. sites. To correct for errors associated with multiple com- Ostracods were extracted from sediment samples parisons, the Bonferroni method was employed (Sokal & taken from rocky crevices, ledges, or sandy patches adja- Rohlf 1995), giving an alpha criterion of 1.67% (=5%/3 cent to rocky habitats. We collected 250 -cm3 sediment possible comparisons). samples from the upper 1 cm of sediment. Samples were fine- sieved (100 p.m sieve) to remove fine debris, and 350 individuals in each sample were identified to species Results with a stereomicroscope (following Rome 1962; Martens 1985; Wouters & Martens 1992, 1994) (Appendix 3). Water Quality One sample was collected and tallied for each depth at all three sites. Of the estimated 200 ostracod species in Cape Banza, the low-disturbance site, had significantly Lake Tanganyika, the majority remain undescribed, but lower lake turbidity levels than the moderate (Luhanga) extensive collections of reference material at the Univer- and high (Burundi) disturbance sites, which did not differ sity of Arizona (with A.S.C.) and the Royal Belgian Insti- significantly from each other (Tables 1 & 2). River turbid- tute of Natural Sciences (with K.M.) made consistent ity measurements reflected consistently increasing turbid- identifications possible. ity with disturbance, although the low- and moderate - disturbance sites were not significantly different from Data Analysis each other. Maximum depths (z,) of benthic algal growth and Sample sizes for fish and mollusc surveys varied. Raw spe- bioturbation indicated that the observed turbidity gradi- cies counts in each transect or quadrat give an estimate of ent among our sites probably accurately reflected longer species density (number of species per constant area or term average conditions. At the low -, moderate -, and time). It is also useful to be able to compare species rich- high- disturbance sites, the zmax values for benthic algal ness in samples of constant size to account for variation in growth were 63, 50, and 35 m, and the z for sampling intensity. To estimate the number of species ex- bioturbation were 95, 92, and 75 m, respectively. pected with constant sample size, we used Hurlburt's rar- Divers noted that the rocky slope at Cape Banza was efaction method (Magurran 1988). The smallest sample not inundated with sediments; the thin veneer of sedi-
conservation Biology Volume 13, No. 5, Octobcr 1999 92
1022 Contributed Papers Alinet al
Table 1.Results of analysis of variance for differences in turbidity, species density, rarefied species richness, and abundance among study sites in lake Tanganyika. site Cape Banza Lubanga Burundi Variable (low dist.) (moderate dist.) (bigb disc.) F,df,p Turbidityb lake -0.38(0.17) -0.066 (0.17) -0.084 (0.12) 5.48;2,27; 0.005 river 0.63(0.36) 0.88 (0.62) 1.87 (0.72) 36.12;2,136; 0.005 Species density fishes (scuba) 23.1(3.45) 19.6(6.52) 17.8(3.49) 3.05;2,27;0.05 fishes (ROV) 8.2(3.51) 5.4(1.79) 5.1(3.18) 4.56;2,35,0.025 molluscs (quadrat) 3.3(1.52) 2.7(1.35) 2.5(0.76) 1.21;2,41; ns molluscs (sieve) 14.8(3.83) 10.6(4.72) 6.4(3.51) 13.97;2,44;0.025 ostracods 28.8(5.40) 31.0(4.74) 20.2(4.02) 6.96;2,12;0.025 Rarefied species richness fishes (scuba) 17.7(3.42) 18.8(4.00) 12.3(3.34) 7.76;2,24;0.005 fishes (ROV) 5.0(0.82) 4.0(0.86) 6.6(1.81) 11.00;2,25;0.001 molluscs (quadrat) 5.0(1.28) 3.1(1.04) 1.8(0.51) 8.12;2,10;0.025 molluscs (sieve) 8.3(1.36) 5.9(0.80) 3.8(1.35) 19.91;2,17;0.005 Abundance fishes (scuba) 524.9(476.0) 223.4(159.6) 1015.0(753.7) 5.98;2,27;0.01 fishes (ROV) 164.3(122.9) 83.3(45.0) 22.0(17.9) 10.50;2,35;0.001 molluscs (quadrat) 12.8(23.9) 20.7(18.8) 27.8(22.2) 1,48;2,41; ns molluscs (sieve) 79.4(44.8) 160.3(162) 86.4(98.1) 2.77;2,44; (IS 'Means (and standard deviations) are reported for each data set in the site columns. bTurbidlty measurements are given in log(MU) wbere MU is normalized turbidity units.
ment present at this locality was composed of locally de- were observed at Cape Bann in the deepest ROV obser- rived shell fragments and other autochthonous organic vations at 96 m. Finally, shallow trenches dug by the detritus. In contrast, rocky slopes at Luhanga and Bu- ROV arm showed that the sediments blanketing the rundi were inundated by terrigenous sediments and cov- rocky slopes at Luhanga and Burundi were laminated, ered with extensive bacterial mats starting at around 100 m black, and rich in organic material, indicating rapid at Luhanga and at 80 m at Burundi. No bacterial mats burial and accumulation of anoxic sediments.
Table 2. Results of Bonferronl-corrected pairwise t test comparisons of turbidity, species density, rarefied species richness, and abundance between study sites in Lake Tanganyika.
Site Cape Banza-Lubanga Lubanga-Burundi Burundi -Cape Banza (low-moderate dist.) (moderate-bigb dist.) (bigb low disc.) Variable t, df, p t,df,p t, df,p Turbidity lake -4.34, 21, 0.001 0.23, 17, ns 3.99, 16, 0.01 river -0.79, 53, ns -8.11, 133. 0.001 3.41,86,0.01 Species density fishes (scuba) 1.55, 20, ns 0.74, 17, ns -3.31, 17, 0.01 fishes (ROV) 2.56, 24, ns 0.35, 24, ns -2.26, 22, ns molluscs (sieve) 3.03, 36, 0.01 2.34, 24, ns -5.64, 28, 0.001 ostracods 0.68, 8, ns 3.89, 8, 0.01 2.86, 8, ns Rarefied species richness fishes (scuba) -0.66, 17, ris 3.53, 14, 0.01 -3.41, 17, 0.01 fishes (ROY) 2.59, 21, ns -4.21, 16, 0.001 2.53, 13, ns molluscs (quadrat) 2.24, 7, ns 2.38, 9. ns -4.85, 4, 0.01 molluscs (sieve) 4.44, 15, 0.001 3.43, 10, 0.01 -4.91, 9, 0.001 Abundance fishes (scuba) 1.99, 20, ns -3.41, 17, 0.01 1.74, 17, ns fishes (ROV) 2.30, 24, ns 4.42, 24, 0.001 -3.9-, 22, 0.001
c;u<,servatio, 81o1ogy Volumc 13, No. S. October 1999 93
Alin et al Contributed Papers 1023
Faunal Censuses and Rarefaction species density was significantly higher at Cape Banza than at Burundi. Average abundance data showed no ob- Across taxonomic groups, mean species density, species vious trend relating to sediment effects, with significantly richness, and abundance varied with depth (Fig. 2). Fish fewer individuals at the moderate - than at the high- distur- and mollusc species richness peaked at shallow to inter- bance site. The ROV transects revealed different diversity mediate depths and declined at greater depths, whereas trends in deeper waters (Table 1). Average rarefied spe- ostracod species richness increased discontinuously up cies richness was significantly higher at Burundi than at to 40 m. Luhanga. Species density was lower at both Burundi and In shallow (scuba) transects, mean rarefied fish species Luhanga than at Cape Banza in the analysis of variance, richness was significantly greater at Cape Banza (low dis- and the Bonferroni-corrected, pairwise t tests lacked ade- turbance) and Luhanga (moderate disturbance) than at quate power to pinpoint the source of this difference. the Burundi (high disturbance) site (Table 2), and mean Abundance of fish in the ROV transects declined with in- creasing disturbance across sites; both Cape Banza and Luhanga had significantly more fishes than Burundi. d) Cape Banza - Luhanga Mean rarefied mollusc species richness decreased with 50 Burundi increasing disturbance regardless of whether quadrat or 40 sieve data were considered, although not all pairwise comparisons were significant (Tables 1 & 2). Species 30 density decreased with increasing disturbance only in 20 the sieve data, with Cape Banza and Luhanga having sig- nificantly higher species density than Burundi. Mollusc 10 abundance appeared to increase with disturbance in the quadrat data and showed no pattern in the sieve data. b) e) The variance, however, was too large to establish statisti- 16 15 cal differences. 12 Observed ostracod species richness was significantly >3 10 higher at Luhanga than at Burundi. Because a constant 8 number of individuals were counted for each sample, com- 4t 5 4 parative abundance and density data were not available.
Oi 510 20 40 60 80 0 5 10 2040 Fish Trophic Analyses e) 4 2400 300 The proportional importance of herbivores (benthic algi- vores and phytoplanktivores) declined at shallower depths .1600 200 with increasing disturbance (Fig. 3). The percentage of in- dividuals feeding on benthic invertebrates and zooplank- ' 800 100 ton increased with increasing depth at all sites, but did so more steeply at the moderate- and high-disturbance sites, 0 compensating for the relative decline in herbivory. Pis 510 20 40 60 80 civory was a relatively minor feeding mode at all sites and Depth (m) depths, except in the ROV survey of high-disturbance Figure 2. Profiles of species density (number of species sites, where the highest percentage of piscivores was seen per time or area transect), species richness (number of despite the low absolute numbers of individuals. species in a constant sample size), and abundance Numerical abundance of herbivores was also highest (number of individuals per sample) against depth: at Cape Banza and persisted to greater depths than at the fishes, species density (a); fishes, mean rarefied species more disturbed sites. Microcarnivores (benthic inverti- richness at each depth (sample sizes used for rarefac- vores and zooplanktivores) showed their largest peaks at tion were 29 individuals for 0 -20 m and 20 individu- Luhanga and Burundi, with zooplanktivores being espe- als for 40 -80 m) (b); fishes, mean abundance per cially abundant at the Burundi site. Piscivores made up sample for scuba and ROV surveys (c); ostracods, spe- only a small number of the individuals surveyed at all cies counts for each sample (d); molluscs, single rare - sites and depths. fled estimate for each site with all samples (sieve + quadrat) at each depth pooled (e); molluscs, mean Similarity Indices and Community Structure abundance per sample for sieve and quadrat data to- gether (j). Error bars indicate the standard error for Jaccard and Simpson indices show that fish faunal simi- points representing multiple data points. larity was high across depths in the shallower portion of
Conservation tkokily Volume 13. No. S. October 1999 94
1024 Contributed Papers Alin et al
a) Cape Banta - low disturbance b) Luhanga - moderate disturbance e) Burundi - high disturbance
4000 I3OIher 1000 5000 Pnci voga Zoopianktivores 3000 750 3750 Benthic invvÓvores ¢Phyioptakùvores Bmdhu dpvores 500 12000c = 1000 250
10 20 40 60 80 Figure 3. Depth profiles of total
100 100 numbers of individuals and per- centages of individuals belonging 75 75 to cach tropbic group at each site: 50 50 Cape Banza, low-disturbance site 25 25 (a), Luhanga, moderate- distur- bance site (b), and Burundi, high- 5 10 20 40 60 60 ° 1 s 10 20 40 60 SO Depth (m) Depth (m) disturbance site (c).
each survey and declined along with species density and turnover and instability over the last two decades (cf. richness starting between 20 and 40 m, and that faunal Goldschmidt et al. 1993). lakes Victoria and Tanganyika similarity among sites was higher in scuba transects and appear to be experiencing somewhat different forms of lower in ROV transects (Table 3). Faunal similarity for environmental degradation, however, and may manifest molluscs and ostracods was generally high as well, both faunal responses quite differently as well. Cichlid species among depths within a site and across sites. in Lake Victoria may find refuge from fishing and preda- Fish species abundances observed by scuba at Cape tion pressures in rocky habitats and marginal ponds and Banza and Luhanga were more equitably distributed and swamps (Ogutu -Ohwayo 1993), but reproductive barri- included more uncommon species than the Burundi fish ers among cichlid species there are easily breached un- fauna (Fig. 4). Similar trends were apparent for molluscs der the present eutrophic conditions, thereby reversing and, to a lesser extent, ostracods, but no clear differ- speciation -in- progress for many nascent species (See - ences were apparent in ROV fish surveys. hausen et al. 1997). In contrast, rocky habitat communi- ties in Lake Tanganyika are most threatened by sediment inundation along heavily deforested shorelines, eliminat- Discussion ing habitat and refuge alike for rock -dwelling fish, mol- lusc, and ostracod species. Turbidity measurements and Regional concern about the potential for serious human other environmental observations clearly differentiated effects on the Lake Tanganyika ecosystem was spurred our low- and high -disturbance sites from each other, with by observations of Lake Victoria's dramatic ecological the moderate -disturbance site being more similar to one
Table 3.Jaccard and Simpson indices of similarity for all taxonomic groups.'
Site Cape Banza Luhanga Burundi Averages across depths (low dlsL) (moderate dirt.) (high dist.) Across depths fishes (scuba) 0.50-0.72 0.56-0.79 0.54-0.81 fishes (ROV) 0.33-0.66 0.29-0.50 0.22-0.36 molluscs (all) 0.62-0.84 0.52-0.74 0.43-0.71 ostracods 0.45-0.70 0.52-0.73 0.48-0.72
Site
Cape Battra- Lubanga Luhanga- Burundi Burundi -Cape Banza (lour moderate dist.) (moderate -high dist.) (high -loo' dist.) Across sites fishes (scuba) 0.63-0.87 0.53-0.82 0.44-0.82 fishes (ROV) 0.31-0.52 0.41-0.59 0.28-0.50 molluscs (all) 0.62-0.90 0.40-0.67 0.45-0.87 ostracods 0.50-0.68 0.39-0.68 0.52-0.85 'Similarity values between all pairs of adjacent depths within each site were averaged for each data set (across depths). Similarity values be- tween pairs of sites are based on species lists pooled across depths for each site (across sites). Values are in the form ofJaccard- Simpson.
Cunxmtion Biology Volume 13. No. 5. October 1999 95
Alin et al. Contributed Papers 1025
Declining similarity values for fishes at depths >20 m (both among and within sites) possibly reflected the dis- Cape Buna Lump appearance of numerically dominant and widespread Burundi herbivorous species with increasing depth. The effect of decreasing light levels on visual fish identifications may also may have contributed to lower similarity values at greater depths.
Ecological Tolerances
1 713 19253137131955 61 67 Depth profiles for fishes show that herbivores such as benthic algal feeders and phytoplanktivores were typi- cally prevalent at shallower depths and were gradually replaced by microcarnivores (benthic invertivores and zooplanktivores) with increasing depth as algae became less abundant (Brichard 1989). The depth range of herbi- vore domination was restricted to shallower depths at more disturbed sites, with microcarnivory increasing at shallower depths compared to the low- impact site. A likely cause of this restriction is decreased water clarity or substrate inundation associated with increasing sedi- 7 13 19 25 31 37 4349 55 ment influx (Table 1). The huge peak seen in zooplankt - Species rank yore abundance at Burundi supports the hypothesis of Figure 4. Rank abundance curves for surveyed com- McKaye and Gray (1984) that zooplanktivores, given munities fish (scuba) (a), fish (ROV) (b), mollusc (c), their mobility, resettle disturbed areas more rapidly than and ostracod (d). benthic feeders. Increased sediment loading may contribute to in- or the other, depending on the variable. Our faunal data creased bacterial production by providing surplus nutri- show that species richness and density correlated nega- ent influx and biological oxygen demand, leading to an- tively with sediment disturbance level for fishes, mol- oxic conditions near the substrate -water interface at luscs, and ostracods in rocky Tanganyikan habitats, al- depths much shallower than the oxycline. laminated though the statistical comparisons, particularly between sediments, indicating little bioturbation and anoxic con- the moderate-disturbance site and either disturbance gra- ditions, as shallow as 40 m at Burundi, and the presence dient end member, were not always robust. An excep- of bacterial mats at Luhanga and Burundi both suggested tion to this pattern of declining species richness with in- a decline in oxygen levels near the sediment -water inter- creasing disturbance was that rarefied species richness in face at our more disturbed sites. the ROV fish surveys was highest overall at the high -dis- Such conditions may help explain patterns of diversity turbance site. The only significant abundance trends and abundance observed at greater depths in our sur- were seen in the fish survey data. In shallow transects, veys. Fish abundance was high in all scuba transects but the moderate -disturbance site had significantly fewer declined sharply with depth, particularly in the Burundi fishes than the high- disturbance site.In deep -water ROV census. In the mollusc surveys, abundance ap- transects, fish abundance declined with increasing distur- peared to increase dramatically at the 40-m collection bance level. point of the high -impact site, although this peak was not statistically significant. Differences in oxygen metabo- Similarity Indices and Community Structure lism between fishes and invertebrates could lead to dif- ferential tolerance to low -oxygen conditions. Michel To infer environmental quality from species assemblage (1994) noted that some gastropod species are more tol- data, it is important that species distributions be broad erant of hypoxic conditions than others, and Coulter enough that a common species pool is shared across (1967) and Verheyen et al. (1994) made similar obser- sites and that environmental tolerances of the constitu- vations among species of Tanganyikan fish. Alternately, ent species are known. Further, it is important not to the concurrent increase in mollusc abundance and de- confound geographically circumscribed species distribu- crease in fish abundance may reflect changes in trophic tions with interpretations of environmental tolerances competition pressure between fishes and molluscs. or quality. Many of the numerically dominant species in Differences in life history may also partially explain all taxa were common to all three sites, as were a large the community-level responses of fishes, molluscs, and number of the less dominant species (Appendices 1 -3). ostracods to the same environmental stress. Long life
Conservation Biology Volume 13. No. 5. octobcr 1999 96
1026 Contributed Papers NM er ac
spans and low reproductive replacement rates may ren- part of the lake and 250 m in the southern basin), result- der fishes and molluscs more susceptible to localized ex- ing in a narrow ring of benthic habitat that encircles the tinction following individual disturbance events or with lake's perimeter (Cohen 1995). The patchy distribution increasing levels of disturbance. Tanganyikan ostracods of substrates on the lake bottom (rocky, sandy, muddy) are not ecologically well known but probably have sig- may have stimulated speciation through evolutionary time nificantly shorter life spans than fishes and molluscs and by making demographic and genetic exchange between may reproduce several times per year (Martens 1994). adjacent populations of stenotopic species improbable Ostracods, being primarily detritivores, may respond or impossible (Michel et al. 1992; Sturmbauer & Meyer positively in terms of both diversity and abundance with 1993). Local endemicity in fishes and ostracods is high, increased sediment input up to some threshold of inun- with species ranges often limited to small stretches of dation (Table 1). Because of their different life- history shoreline (Brichard 1989; Cohen 1994). The same patchy traits, molluscan and ichthyofaunas may continue to re- habitat distribution and the degree of habitat specificity flect disturbance longer after environmental perturba- and stenotopy that may have fostered the origination of tion abates than do ostracod communities. species may also render them more vulnerable to extinc- Finally, changes in sediment influx resulting from land - tion, because habitat destruction and fragmentation lead use changes may result in qualitative structural differ- to greater distances between neighboring populations ences in lake habitats, selectively eliminating some habitats and diminish their ability to recover from both natural and increasing the area of others. Many cichlids distin- and anthropogenic environmental perturbations by re- guish appropriate habitat based on the particle size, as- colonization (Bruton & Merron 1990). Although we might pect, and water depth of rocky habitats (Brichard 1989). expect lake Tanganyika to be initially more resilient than Sediment loading may lead to greater habitat homogene- Lake Victoria to environmental pressures because of Tan- ity in the littoral-sublittoral zones of Lake Tanganyika by ganyika's greater volume and lower human population burying relatively rare rocky outcrop patches. Several density, large -scale changes in habitat or water quality interrelated effects of changing habitat complexity may may be even more difficult to remediate in lake Tangan- affect standing diversity or abundance levels. Simplifica- yika because of the geographic distribution of habitat tion of habitat structure by infilling of cracks and crev- types and long flushing time, respectively (cf. Bootsma & ices results in fewer refugia from predation for many spe- Hecky 1993). cies or their juveniles and less overall habitat area for cryptic and nocturnal species. Many Tanganyikan cichlids Caveats are substrate spawners, unlike their rocky- habitat coun- terparts in Lakes Victoria and Malawi, which rely on fe- One inherent difficulty in monitoring biodiversity is that male mouth - brooding for reproduction (Poll 1986; See - surveys represent isolated time points. Thus, our data hausen etal.1998). Inundation of habitat substrate lack temporal perspective on faunal stability. Although therefore may have selective and dire consequences for we lack long -term data sets for our sites, previous re- reproductive success for these fishes. Because benthic search by Hon et al. (1983) and Sato et al. (1988) pro- productivity on rocks greatly exceeds that on sandy sub- vides fish species lists for two of our sites for compari- strates (C. O'Reilly, personal communication), a reduc- son. Although sampling protocols, areas surveyed, and tion in rocky substrate area may have magnified effects total species richness differed between surveys in both on diversity and abundance at higher trophic levels. This cases, their lists allow rudimentary comparisons. Of the may be reflected in a higher susceptibility of stenotopic 37 species observed by Hon et al. (1983) at Luhanga, rock -dwelling fish species than of sand-dwellers to ex- 86% were resampled in our scuba transects. At Cape tinction, which relates both to the degree of specializa- Banza, 80% of the 35 species tallied by Sato et al. (1988) tion in rock- dwelling cichlids and to the relative rarity of were observed by our scuba divers. Data from Nakai and and distance between rocky habitat patches (cf. Bruton Yuma (1988) also allowed comparison of molluscan fau- & Merron 1990; Ribbink 1990; Lowe- McConnell 1993). nal stability through time at Cape Banza. Of 11 species Increasing disturbance regimes may select for less steno- observed by them, 73% were also observed by our topic species that are able to cross barriers and recolo- divers. These comparisons suggest that spatial and tem- nize disturbed habitats. poral variability in littoral habitat communities are com- parable among sites, and that this variability is unlikely to affect the qualitative trends in our results. Lake Basin Parameters and Conservation Lake basin morphology has important implications for conservation planning because the habitable portion of Conclusions the bottom of Lake Tanganyika is delimited by the inter- section of the steep lake bottom morphology with the Our preliminary results using scuba and ROV transects oxycline (approximately 100 m depth in the northern and sieve sampling to survey environmental conditions
Conservation Biology Votumc 13. No. 5, October 1999 97
Alin et al Contributed Papers 1027
and biodiversity in Lake Tanganyika appear promising. Bruton, M. N., and G. S. Meson. 1990. The proportion of different eco- Fish, mollusc, and ostracod diversity generally corre- ethological sections of reproductive guilds of fishes in some Afri- can inland waters. Environmental Biology of Fishes 28:179 -187. lated negatively with disturbance level, although each Caijon, A. G. 1992. Water quality in the Bay of Bujumbura (lake Tangan- taxonomic group may have a different response thresh- yika) and its influence on phytoplankton composition. Mitteilun- old. Our analyses suggest that censuses of both fishes gen Internationale Vereinigung fur Theoretische und Angewandte and invertebrates, particularly those groups whose envi- Limnologie 23:55 -65. ronmental tolerances are known, could provide useful Cohen, A. S. 1994. Extinction in ancient lakes: biodiversity crises and conservation 40 years after J. L Brooks. Pages 451 -479 in IC Mar- information about the specific impetus of environmental tens, B. Goddecris, and G. Coulter, editors. Speciation in ancient pressures (e.g., oxygen levels). Fishes and molluscs may lakes. Advances in Limnology 44. Schwcizerbart'sche Verlagsbuch- be more sensitive than ostracods in the early phases of handlung, Stuttgart, Germany. sedimentation impact. Fish data collected by scuba and Cohen, A. S. 1995. Paleoecological approaches to the conservation biol- ROV were not directly comparable because of method- ogy of benthos in ancient lakes: a case study from Lake Tanganyika. ological differences, but both data sets were necessary Journal of the North American Benthological Society 14:654 -668. Cohen, A. S., R. Bills, C. Z. Cocquyt, and A. G. Galion. 1993. The im- for generating meaningful trophic depth profiles. The pact of sediment pollution on biodiversity in Lake Tanganyika. ROV was critical for recording data on environmental Conservation Biology 7:667 -677. thresholds, such as maximum depths of algal growth Coulter, G. W. 1967. Low apparent oxygen requirements of deep - and oxygenation, beyond depths accessible by scuba. Al- water fishes in Lake Tanganyika. Nature 215:317 -318. though its deployment came at considerable financial Coulter, G. W. 1991. lake Tanganyika and its life. Oxford University Press, Oxford, United Kingdom. and logistical cost, it revealed patterns that would other- Coulter, G. W. 1994. Lake Tanganyika. Pages 13 -18 in K. Martens, B. wise have been missed in our surveys, such as the pre- Goddeeris, and G. Coulter, editors. Speciation in ancient lakes. Ad- cipitous drop in fish abundance at depth at the high-dis- vances in Limnology 44. Schweizerbart'sche Verlagsbuchhand- turbance site. lung, Stuttgart, Germany. Coulter, G. W., and IL Mubamba. 1993. Conservation in Lake Tangan- yika, with special reference to underwater parks. Conservation Bi- ology, 7:678 -685. Acknowledgments Craig, H. 1974. lake Tanganyika geochemical and hydrographie study: 1973 expedition. Reference series 75-5. Scripps Institution of We thank S. LaRosa, D. Lee, S. Smith, the University of Bu- Oceanography, San Diego, California. rundi, Metalusa, and the Food and Agriculture Organiza- DaSet, J., J. P. Gosse, G. G. Teugels, D. F. E. Thys van den Audenacrde. 1991. Checklist of freshwater fishes of Africa. Volume 4. ORSTOM, tion Lake Tanganyika Research Project for technical field Paris. assistance; the Ministry of Research and Higher Education Gashagaza, M. M., and M. Nagoshi. 1985. Comparative study on the in Burundi and the National Center for Scientific Research food habits of five species of Lamprologus in Lake Tanganyika in the Ministry of Higher Education and Scientific Re- (Cichlidac). Pages 32 -33 in H. Kawanabe, editor. Ecological and search in Zaire for research permits; J. Fost and M. Pala - Limnological Study on Lake Tanganyika and Its Adjacent Regions 3. Department of Zoology, Faculty of Science, Kyoto University. cios-Fest for Spanish translation; S. Connolly and J. IaPeyre Kyoto, Japan. for help with data analysis; and E. Allison, J. Alroy, S. Con- Goldschmidt, T., F. Witte, and J. Wanink. 1993. Cascading effects of nolly, C. Frissell, R. Hecky, R. Lowe- McConnell, P. Rein - the introduced Nile perch on the detritivorous /phytoplanktivorous thal, M. Rosenzweig, P. Verburg, and two anonymous re- species in the sublittoral arras of Lake Victoria. Conservation Biol- viewers for helpful discussions and comments on the ogy 7:686 -700. Greene, L E., and W. S. Alevizon. 1989. Comparative accuracies of vi- earlier versions of the manuscript. This work was sup- sual assessment methods for coral reef fishes. Bulletin of Marine ported by NOAA -NURC -UCAP Grant UCAP -92 -04 (A.C.), Science 44:899 -912. a National Science Foundation (NSF) Graduate Research Hecky, R. E. 1991. The pelagic ecosystem. Pages 90-110 in G. W. Fellowship (SA.), and an award from the NSF - funded Re- Coulter, editor. Lake Tanganyika and its life. Oxford University search Training Group in the Analysis of Biological Diver- Press, Oxford, United Kingdom. sification at the University of Arizona (SA.). Hort, M., K. Yamaoka, and K. Takamura. 1983. Abundance and micro- distribution of cichlid fishes on a rocky shore of lake Tanganyika. African Study Monographs 3 :25 -38. Lcloup, E. 1953. Gastéropodes. Résultats scientifiques de l'exploration Literature Cited hydrobiologique du lac Tanganyika (1946 47). Institut Royal des Sciences Naturelles de Belgique 3(4):1-273. Bootsma, H. A., and R. E. Hecky. 1993. Conservation of the African Great licm, K., and D. J. Stewart. 1976. Evolution of the scale-eating cichlids of Lakes: a limnological perspective. Conservation Biology 7:644 -656. Lake Tanganyika. a generic revision with a description of a new spe- Bortone, S. A., R. W. Hastings, and J. L. Ogleby. 1986. Quantification of cies. Bulletin of the Museum of Comparative Zoology 147:319 -350. rccf fish assemblages: a comparison of several in situ methods. Lowe -McConnell, R. H. 1993. Fish faunas of the African Great Lakes: ori- Northeast Gulf Science 8:1 -22. gins, diversity, and vulnerability. Conservation Biology 7:63 4-643. Brichard, P. 1989. Pierre Brichard's book of cichlids and all the other Luckhurst, B. E., and K. Luckhurst. 1977. Analysis of the influence of fishes of Lake Tanganyika. T. F. H. Publications, Neptune City, New substrate variables on coral reef fish communities. Marine Biology Jersey. 49:317 -323. Brown, D. 1994. Freshwater snails of Africa and their medical impor- Magurran, A. E. 1988. Ecological diversity and its measure. Princeton unce. Taylor & Francis, London. University Press, Princeton, New Jersey.
Conservation Biology Volume 13. No. 5, October 1999 98
1028 Contributed Papers Alin et al
Martens, K. 1985.TanganyikacypridopsLsgen.n. (Crustacca, Ostra- Ribbink, A. J.1990. Alternative life- history styles of some African coda) from Lake Tanganyika. Zoologica Scripta 14:221 -230. cichlid fishes. Environmental Biology of Fishes 28:87 -100. Martens, K. 1994. Ostracod speciation in ancient lakes: a review. Pages Ribbink, A. J., A. C. Marsh, A. C. Ribbink, and B. J. Sharp. 1983. A pre- 203 -222 in K. Martens, B. Goddeeris, and G. Coulter, editors. Spe- liminary survey of the cichlid fishes of rocky habitats in Lake ciation in ancient lakes. Advances in Limnology 44. Schweizer - Malawi. South African Journal of Zoology 18:149 -310. baresche Verlagsbuchhandlung, Stuttgart, Germany. Rome, R 1962. Ostracodes. Résultats scientifiques dc l'exploration hy- Mbomba, N. B. 1983. Comparative ecology of algal feeding cichlids in drobiologique du lac Tanganyika (1946 -47). Institut Royal des Sci- relation to their developmental stages. Pages 27 -28 in H. Kawanabc, ences Naturelles de Belgique 3(8):1 -305. editor. Ecological and Limno logical Study on Lake Tanganyika and Sato, T., M. Yuma, Y. Niimura, K. Nakai, N. Abc, M. Nishida, M. Its Adjacent Regions 3. Department of Zoology, Faculty of Science, Nshoinbo, and S. Yamagishi. 1988. Fish fauna around Ubwari Pen- Kyoto University. Kyoto, Japan. insula. Pages 14 -15 in H. Kawanabe and M. K. Kwentuenda, edi- McKaye, K. R., and W. N. Gray. 1984. Extrinsic barriers to gene flow in tors. Ecological and Limnological Study on Lake Tanganyika and Its rockdwelling cichlids of Lake Malawi: macrohabitat heterogeneity Adjacent Regions 5. Department of Zoology, Faculty of Science, and reef colonization. Pages 169 -184 in A. A. EcheUe and 1. Korn- Kyoto University. Kyoto, Japan. field, editors. Evolution of species flocks. University of Maine at Seehausen, O., J. J. M. van Alphen, and F. Witte. 1997. Clchlid fish di- Orono Press, Orono. versity threatened by eutrophication that curbs sexual selection. Michel, E. 1994. Why snails radiate: a review of gastropod evolution in Science 277:1808 -1811. long -lived lakes, both recent and fossil. Pages 285 -317 in K. Mar- Sechausen, O., E Lippitsch, N. Bouton, and H. Zwennes. 1998. Mbipi, tens, B. Goddeeris, and G. Coulter, editors. Speciation in ancient the rock-dwelling cichlids of lake Victoria: description of three lakes. Advances in Limnology 44. Schweizerbart'sche Verlagsbuch- new genera and fifteen new species (Telcostei). Ichthyological Ex- handlung, Stuttgart, Germany. ploration of Freshwaters 9:129 -228. Michel, A. E., A. S. Cohen, K. West, M. R. Johnston, and P. W. Kai Simpson, G. G. 1960. Notes on the measurement of faunal resem- 1992. Large African lakes as natural laboratories for evolution: ex- blance. American Journal of Science 258a:300 -311. amples from the endemic gastropod fauna of Lake Tanganyika. Mit - Sokal, R., and F. J. Rohlf. 1995. Biometry. W.H. Freeman, New York. teilungen Internationale Vereinigung für Theoretische und Ange - Sturmbauer, C., and A. Meyer. 1993. Mitochondria) phyiogeny of the en- wandte Limnologie 23:85 -99. demic mouthbrooding lineages of cichlid fishes from Lake Tangany- Nakai, K., and M. Yuma. 1988. Molluscan fauna around Ubwari Penin- ika in Eastern Africa. Molecular Biology and Evolution 10:751 -768. sula Pages 63 -63 in H. Kawanabc and M. K. Kwetuenda, editors. Travers, R A. 1988. Diagnosis of a new African Mastacembelid spiny - Ecological and limnological study on Lake Tanganyika and its adja- eel genus Acthiomasucembelus gen. nov. (Masrtcembeloidei: Syn- cent regions 5. Department of Zoology, Faculty of Science, Kyoto branchiformes). Cybium 12:255 -257. University. Kyoto, Japan. Verheyen, E., R. Binai and W. Decleir. 1994. Metabolic rate, hypoxia Nshombo, M. 1983. Change of food habits of Perissodus mtcrolepts tolerance and aquatic surface respiration of some lacustrine and with its development (Cichlidac). Pages 36 -37 in H. Kawanabc, ed- riverinc African cichlid fishes ( Pisces: Cichlidac). Comparative Bio- itor. Ecological and Limnological Study on Lake Tanganyika and Its chemistry and Physiology 107A:403-411. Adjacent Regions 3. Department of Zoology, Faculty of Science, von Martens, E. 1897. Beschalte wdchtiere Deutsch Ost- Afrika. K. Kyoto University. Kyoto, Japan. Möbius, Deutsch Ost- Afrika, IV, Berlin. Ogutu- Ohwayo, R. 1993. The effects of predation by Nile perch, Lates West, K. A. 1997. Perspectives on the diversification of species flocks: niloticusL, on the fish of Lake Nabugabo, with suggestions for systematics and evolutionary mechanisms of the gastropods (Prosy conservation of endangered endemic cichlids. Conservation Biol- branchia: Thiaridae) of Lake Tanganyika, East Africa Ph.D. disserta- ogy 7:701 -711. tion. University of California, Los Angeles. Poll, M. 1953. Poissons non Cichlidac. Resultats scientifiques de l'ex- Wombwell, V. 1986. The chemical timnology of Lake Tanganyika. Se- ploration hydrobiologique du lac Tanganyika (1946 -47). Institut nior honors thesis. Colorado College, Colorado Springs. Royal des Sciences Naturelles dc Belgique 3(5A ):1 -251. Woutcrs, K., and K. Martens. 1992. Contribution to the knowledge of Poll, M. 1956. Poissons Cichlidae. Résultats scientifiques dc l'explora- Tanganyikan cythcraccans, with the description of Mesocyprideis tion hydrobiologique du tac Tanganyika (1946 -47). Institut Royal nom. nov. (Crustacea, Ostracoda). Bulletin de l'Institut Royal des des Sciences Naturelles de Belgique 3(5B):1 -619. Sciences Naturelles de Belgique, Biologic 62:159 -166. Poll, M. 1986. Ctacsification des Cichlidae du lac Tanganyika: tribus, Wouters, K, and K. Martens. 1994. Contribution to the knowledge of genres et espèces. Memoires de l'Academie Royale de Belgique, the Cyprfdets species flock (Crustacea:Ostracoda) of Lake Tangan- Classc des Sciences 45(2):1 -163. yika, with the description of three new species. Bulletin de l'Institut République du Burundi Bureau Central dc Récensemcnt. 1990. Récen- Royal des Sciences Naturelles dc Belgique, Biologic 64:1II -128. sement general dc la population et dc l'habitation: resultats provi- Yamaoka, K. 1983. Relation between structure and ecology of algal soires. Bureau Central dc Recensement. Gitega, Burundi. feeding cichlids in Lake Tanganyika. Pages 24 -26 in H. Kawanabe, République du Zaire. 1988. Aménagement du territoire: schéma na- editor. Ecological and Limnological Study on Lake Tanganyika ant' tional population, 1986. Département des Travaux Publics et dc Its Adjacent Regions 3. Department of Zoology, Faculty of Science l'Aménagement du Territoire, Republic du Zaire, Kinshasa. Kyoto University. Kyoto, Japan.
Con..crvation Biology Volume 13, No. 5, October 1999 99
Alin et al Contributed Papers 1029
Appendix 1 Distribution and relative abundance of fish species across study sites in Lake Tanganyika'
Scuba (1 -20 m) ROVb (40 -80 m) Cape Banza Lubanga Burundi Cape Banza Lubanga Burundi (low dist)(moderate dist.) (high dist.)(low dirt.)(moderate dist.) (high dist.) Benthic algivores Asprotilapia leptura + + - - - - Cunningtonla longiventralis - + - - - - Cyathopbarynx furdfer + + + - - - Eretmodus cyanostictus + + + + - - - Limnotilapia dardennli + + - - - - Opbtbalmotilapia nasutus + + + + - - - Opbtbalmotilapia ventralis + + - - - - Petrocbromis epbipplum + + - + - - Petrocbromis famula + + - - - - Petrocbromts fascYolatus + + + - - - Petrocbromis ortbognatbus + + + - - - Petrocbromis polyodon + + + - - - Pseudosimocbromis curvifrons + + - - - - Slmocbromis babaulti - + + - - - Stmocbromis diagramma - + - - - - Slmocbromis marginatus + + - - - - Simoebromis sp. - + - - - - Spatbodus marlieri + - - - - - Tanganicodus irsacae - + - - - - Telmatocbromis dbonti + - - - - - Telmatocbromis temporalis + ++ + + + + + - - - Tropbeus mooril + + + + - - - Tropheus polli + - - - - - Benthic invertivores Aetbiomastacembelus cf. albomaculatus - - + + - - Aetbiomastacembelus platysoma - + - - - - Altolamprologus compressiceps + + + + - + Aucbenoglanis ocddentalis - - - + + - Aulonocranus dewlndtt + + + + - - - Cballnocbromis bricbardl + - - - - - Cbryslcbtbys sp. + - - + - - Gnatbocbromis permaxillaris - - - - - + Gnatbochromis pfeffert - + + - - - Julidocbromis marlleri + + + - - + Julidocbromis transcriptus + + - - - - Lamprologus calllpterus + + + - - - Lamprologus cf. flnallmus - - - + + + ++++ + + Lobocbtlotes labiatus + + + - - - Neolamprologus furdfer + + + + - - Neolamprologus leleupi + + - - - - Neolamprologus modestus - - - + - - Neolamprologus mondabu + + + - - - Neolamprologus toae + + + + + - - - - Neolamprologus tretocepbalus + + + - - - Neolamprologus n.sp. "bifrcnaau" + - - - - - Neolamprologus n.sp. "orange" + - - - - - Reganocbromis calliurus - - - - - + Synodontis dbonti + - - - - - Synodontis multipunctatus + - + + - - Synodontis petricola + - - - - - Synodontis sp. + - - + - + Trematocara sp. - - - - +++ +++ Xenotilapta cf. caudafasciata - - - - + + Xenottlapla flavipinnis + + + - - - Xenotilapla fluorescens - - - + - - Xenotilapla sima + + + - - - Xenotilapia n.sp. "large" - - - - - + Xenntilapla n.sp. "white-hp" - - - + + + ++++ Xenotilapla sp. - - + + + + + + + + continued
Conservation Biolog Volume 13. No. 5. October 1999 loo
1030 Contributed Papers Alin et al
Appendix 1 (continued)
Scuba (1 -20 m) ROVb (40 -80 m) Cape Banza Lubanga Burundi Cape Banza Lubanga Burundi (low dist.)(moderate dist.) (high dist.)(low dist)(moderate dist.) (bigb dist.) Phytoplanktivores Lampricbtbys tanganicanus + + + - - - Neolamprologus bricbardi ++++ ++++ ++++ ++++ ++ Neolamprologus gracilis + - - - - - Neolamprologus multifasciatus + - - - - - Neolamprologus savoryl ++ ++ ++ +++ - - Telmatocbromis bifrenatus ++++ ++++ + ++ - - - Zooplanktivores Cyprlcbrnmis microlepidotus +++ ++ ++++ ++++ + ++++ Cypricbromis sp. - - - - - + Microdontocbromis tenuidentatus + + - - - - - Neolamprologus calliurus + +++ - ++++ + + - Paracypricbromis brieni + + + + + + ++ - - - Pararypricbromis nigripinnis + + - ++++ ++++ ++++ Tangacbromis dbanisi - - - - - + Piscivores (including scale eaters) Aetbiomastacembelus cunningtonl - - - - - + Aetbiomastacembelus ellipsifer + + - - - - Aetbiomastacembelus moorii + + + + - - Aetbiomastacembelus sp. + - - - + - Batbybates sp.` - - + - - + Boulengerocbromis microlepist + - - + + + Ctenocbromis bentbicola - - - - + ++ Cypbotilapia frontasa + + + + + + ++++ Greenwoodocbromis cbrLstyi - - - - + + + Haplotaxodon micr+olepis` + - + - - - Hemibates stenosoma` - - - - - + Lamprologus lemairil + + + + + + Lates angusNfrons` - - + - - - Lates mariae` - + - - - - Lates micTolepis` + - - - - - Lates sp.` - - - + - - Lepidiotamprologus attenuatus + + - - - - Lepidiolamprologus cunningtoni + - - - + - Lepidiolamprologus elongatus + ++ + + + ++ Lepidiolamprologus profundlcola + + + - + - Lepidiolamprologus sp. - - - - + - Malapterurus electricus - - - - + - Neolamprologus fascYatus + + - - - - Perissodus microlepis + + + + + - - Perissodus paradoxtts - - + - - - Perissodus straelini + - - + - - Perissodus sp. - - - + + + Omnivores/unknown Lamprologus sp. - - + - - - Neolamprologus cf. buescberi + - - - - - Neolamprologus cf. wautbioni - - + - - - Neolamprologus n.sp. "black" - - - - + - Neolamprologus n.sp. "small orange" + - - - - - Neolamprologus n.sp. 2 - - - + + + - Neolamprologus n.sp. 3 - - - + - - Neolamprologus sp. - - - + - - Total species censused 64 52 38 28 23 22 Total species including wanderers 67 53 41 30 26 25 Total individuals tallied 5774 2457 8120 1883 1035 231 Rarefaction results 56.1 52" 34.2 16.0 14.0 22° "Abundance key: - absence of species in the surveys at this site; + species represents <2% of population at site (all deptbs); + +, speciesrepre- sents 2 -5% of population; + + +. species represents 5-10% of population; + + + +, species represents >10% of population. bROV, remotely operated vehicle. `Species considered - wanderers' and excluded from all quantitative analyses. °Sample used as a standard for rarefaction.
Conservation Biology volume 13. No. 5. Octobcr 1999 101
Alin et al Contributed Papers 1031
Appendix 2 Distribution and relative abundance of mollusc species across study sites in Lake Tanganyika'
Site Cape Banza Lubanga Burundi (low dist.) (moderate dist) (big!, dist.) Bivalves Family Unionidae Caelatura burtoni + + + + Family Mutelidae Mutela spekei + - - Gastropods Family Thiaridac Anceya giraudi + + ++ + + ++ + + ++ Bridouxia giraudi ++ ++ ++ Bridouxia ponsonbyi + + - Bridouxia sp. + ++ + + ++ - Lavigeria n. sp. "fine striped" ++++ + + - Lavigeria n. sp. `wide band" - + - Lavigeria grandis + + + + + - Lavigerla cf. nassa "fine ribbed" + + + + + ++++ Lavigeria cf. nassa `small fine" + + + + - Lavigeria cf. paucicostata "coarse ribbed" + + - Lavigeria cf. paucicostata `sand lay" + ++ + - Lavigeria cf. paucicostata `spiny" + - - Mysorelloides multisulcata + - - New genus n.sp. "guillemei" + - - New genus n.sp. + ++ + ++ + ++ Paramelania damoni forme: crassigranulata + - + Paramelania damoni forme: imperialts + - + Reymondia borel + + ++ + + ++ + Reymondia n.sp. 2 + + + - Reymondia n.sp. 3 ++ + + + Spekia n.sp. `cohere" ++ + ++ Spekia zonata - - + Stanleya neritinoides + - + Stormsia minima + + + + Syrnolopsis minuta + - - Synolopsis sp. + + + Tanganyk is rufofilosa - - + Total number of species (quadrat + sieve) 27 19 15 Total number of individuals 653 1132.5 654 Rarefied species richness (site -wide) 27b 18.8 15.0 'Abundance key is the same as in Appendix 1. 'Sample used as a standard for rarefaction.
Conservation Biology. Volume 13, No. 5. October 1999 102
1032 Contributed Papers Alin et al
Appendix 3 Distribution of ostracod species across study sites In Lake Tangamilta.
Site Cape Banza Lubanga Burundi (lou dist.) (moderate dist) (bigb dist) Superfamily Cypridoidea Family Candonidae Candonopsis depressa + + - Candonopsis n.sp. 7 - - + Candonopsis n.sp. 8 + + - Candonopsis n.sp. 11 - - + Candonopsis n.sp. 12 + - - Candonopsis n.sp. 13 + - - Candonopsis n.sp. 15 - + - Family Cyclocyprididae
Allocypria aberrans . + - + Allocypria ctavijormtsgroup + + + + + Allorypria bumllis + + + Allocypria cf. inclinata - + - Allocypria n.sp. '5 + - + Allocypria n.sp. 8 + - - Allocypria n.sp. 10 + - + Allocypria n.sp. 11 - + - Allocypria n.sp. 16 + - + Allocypria n.sp. 17 + + - Allocypria n.sp. 18 - + - Mecynocypria complanata - + - Mecynocypria cf. conoidea + + + Mecynocypria deflexa + - + Mecynocypria emaciata + + + Mecynocypria opaca + + + Mecynocypria subangulata + + - Mecynocypria n.sp. 8 + + + Mecynocypria n.sp. 9 + + - Mecynocypria n.sp. 14 + + - Mecynocypria asp. 17 + - + Mecynocypria n.sp. 19 + - - Mecynocypria n.sp. 20 + + - Mecynocypria n.sp. 22 + - - Mecynocypria n.sp. 29 (opaca group) + + + Mecynocypria n.sp. 30 + - - Mecynocyprfa n.sp. 31 + + - Mecynocypria n.sp. 32 + - - Mecynorypria n.sp. 33 - + - Mecynocypria n.sp. 36 - + - Mecynocypria sp. - + - Family Cyprididae Cypridopsis bidentata - - + Cypridopsis obliquata + - - Cypridopsis serrata + + + - Cypridopses n.sp. 5 + + + + ++ Cypridopsis n.sp. 6 (species group) + ++++ + Cyprldopsis n.sp. 8 + - + Cypridopsis n.sp. 13 + + + - Cypridopsts n.sp. 15 - + - Cypridopsis n.sp. 16 + - - Cypridopsis n.sp. 17 - - + Cypridopsis n.sp. 18 + + + continued
Conservation Biology Volumc 13, No.5,October 1999 103
Alin et aL Contributed Papers 1033
Appendix 3 (continued)
Site Cape Banza Lubanga Burundi (low disc.) (moderate dist.) (bigb dist.) Cypridopsis n.sp. 22 - + - Cypridopsts n.sp. 23 - + - Tanganyikacypridopsts acantbodes + + + Tanganyikarypridopsts calcarata + + + + Tanganyikarypridopsts depressa + + + + Tanganyikacypridopsis n.sp. 3 + + + + -Tanganyikacypridopsis n.sp. 4 + - + Tanganyikacypridopsis n.sp. 5 + + + Tanganyikacypridopsis n.sp. 8 + + + + Superfamily Cytheroidea Family Cytherideidae Arcbaeorypridets tuberculata + + + Cypridets sp. (bolletje group) - + - Cypridets n.sp. 1 + + + Cyprtdeis n.sp. 24 - + - Mesorypridets trsacae ++++ ++ ++++ Mesoryprideis n.sp. 2b + + + Romecytberidea ampla +++ ++ +++ Romecytbertdea tenuisculpta +++ ++++ +++ Romecytberidea n.sp. 13 ++++ ++++ +++ Romecytberidea n.sp. 15 - + - Romerytberidea n.sp. 18 + - - Tanganyikacytbere burtonensts + + + Tanganyikacytbere calfoni + + - - Family Limnocytheridae Gompborytbere alata ++ + + + + ++++ Gompbocytbere cristata + + + + Gompbocytbere curta ++++ +++ ++++ Gompbocytbere n.sp. "downingi" - - + Gompbocytbere n.sp. "woutersi" + + - Gompbocytbere n.sp. 11 - + - Gompborytbere n.sp. + + - Cytheroidea indet. 1 + - - Superfamily Darwinuloidea Family DarwinuGdae Darwinula stevensoni - + + New genus, new species 1 + + - Total number of species: 60 56 42 'Abundance key Is the same as in Appendix 1.
ConscnzIion Biology Volumc 13, No. 5. October 1999 104
APPENDIX B: THE LIVE, THE DEAD, AND THE VERY DEAD: TAPHONOMIC CALIBRATION OF THE RECENT RECORD OF PALEOECOLOGICAL CHANGE IN LAKE TANGANYIKA, EAST AFRICA
Formatted for submission to Paleobiology.
Simone R. Alin* and Andrew S. Cohen
Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA. *Present address: Large Lakes Observatory, University of Minnesota Duluth, 10 University Drive, Duluth, MN 55812 (email: simone .alin @stanfordalumni.org).
Abstract
High- resolution (annual to decadal) paleoecological records of biodiversity turnover can contribute a long -term perspective to conservationbiology on baseline ecological variability and the response of communities to environmental change. We present here a detailed comparison of species assemblage characteristics (species richness, abundance, composition, and occurrence frequency) for live, dead, and recent fossil ostracod samples from Lake Tanganyika, East Africa. This study calibrates the fidelity of paleoecological samples to live diversity patterns for the purpose of reconstructing community dynamics through time.
Species richness in death and fossil assemblages is comparable to that in a year's accumulation of life assemblages. The temporal resolution of the fossil samples in Lake
Tanganyika could be as short as one year. Species abundance distributions were statistically indistinguishable among all data sets. The majority of species had matched ranks among live, dead, and fossil data sets, but dominant species were different in live samples. Species occurrence frequencies in dead and fossil data identified ecologically persistent species and may be useful for delimiting local species pools. Rare species 105 collectively dominated all three data sets, and large numbers of rare species characterize
Tanganyikan ostracod communities on both ecological and paleoecological time scales.
Analysis of sampling efficiency indicates that approximately 28% of species in each
death or fossil assemblage are unique. Ordination reveals that life assemblages of
ostracods are characterized by high spatiotemporal heterogeneity. Variability in species
diversity and composition was reduced in death and fossil assemblages as a result of
spatial and temporal averaging.
Death and fossil assemblages appear to preserve characteristics of benthic ostracod
assemblages with high fidelity and at a spatiotemporal resolution that renders them
appropriate indicator taxa for the benthic community as a whole. Sampling surficial shell
assemblages in highly resolved archives such as lake deposits represents a more efficient
way of assessing the average ecological conditions at a site than repeated live sampling.
Furthermore, paleoecological analyses can generate novel insights into long -term community variability and membership with direct relevance to conservation
applications.
Introduction
Paleoecological reconstruction is playing an increasingly important role in addressing conservation biological problems (e.g., Binford et al., 1987; Brenner et al., 1999; Davis et
al., 2000; Finney et al., 2000; MacPhee, 1999; Miller et al., 1999; Pandolfi, 1996;
Rodriguez et al., 2001; Steadman, 1995). One of the perennial problems encountered in
paleoecological reconstruction and interpretation is assessing the fidelityof the 106
sedimentary archive of ecological and environmental change to conditions that existed at
the time of deposition. Two transitions are inherent to the formation of the fossil record:
transition from life to death assemblage and from death to fossil assemblage. Numerous
studies on the fidelity of death assemblages to living communities have been performed
(references in Kidwell, 2001; Kidwell and Bosence, 1991; Kidwell and Flessa, 1995),
although few investigators have had the opportunity to compare death assemblages with
recent fossil assemblages formed in the same environment (Russell, 1991), much less the
opportunity to contrast living communities with comparable death and fossil assemblages
(Fürsich and Flessa, 1987; Wolfe, 1996).
Paleoecological baseline studies are increasingly being employed for conservation
purposes to reconstruct environmental conditions prior to human intervention (Brenner et
al., 1993; Kowalewski et al., 2000; Rodriguez et al., 2001). Microfossils like ostracods
allow investigators to collect statistically robust sample sizes for analysis at low cost.
However, the taphonomy of aquatic microfossils, especially lacustrine ones, has received
less attention than marine macrofossils (cf. Martin and Liddell, 1988; Martin et al., 1996;
Wolfe, 1996). Because of their small size and susceptibility to transport, microfossil distributions within a site might not map their life habitat with the same fidelity seen in
marine molluscs (Kidwell, 2001; Kidwell and Bosence, 1991). Although high habitat
fidelity has been observed in marine foraminiferal assemblages (Martin and Liddell,
1988), Wolfe (1996) observed low spatial fidelity in his lacustrine diatom assemblages.
Such paleolimnological studies raise the question of howrepresentative fossil
assemblages are of the once -living communitiesthat contributed to them in terms of 107
species richness, abundance, composition, and occurrence frequency. Differences in the
spatial and temporal scales of analyses in ecology and paleoecology may affect the
conclusions that can be drawn from life versus death or fossil assemblages (cf. Anderson,
1993; Cohen, 2000; Levin, 1992; Pandolfi, 1996).
In this study, we compared assemblage structure and composition of ostracod life,
death, and fossil assemblages from Lake Tanganyika in order to calibrate the sedimentary
record of biodiversity turnover with respect to living communities. Working in an extant
ecosystem with a continuously accumulating sedimentary record allowed us to
simultaneously examine the biases in the transition from life to death assemblage and
those intrinsic to the translation from death assemblages into the fossil record. The
quality of preservation of diversity patterns determines the extent to which
paleoecological insights can be applied to problems in conservation biology. Herewe
examine changes in variability associated with sampling at different temporal and spatial
scales (i.e., ecological vs. paleoecological). We further show how some paleoecological
data may be better suited for some conservation applications than data based on live collections alone.
Background on Study Area
Lake Tanganyika is a tropical rift lake housing extensive radiations of fish and invertebrate species (Fig. B.1). In addition to serving as a natural laboratory for the study of evolution (e.g., Michel et al., 1992), Lake Tanganyika and its biologicalresources represent an important source of nutrition and an economic basis for the region's human 108 populations. Thus, much attention has been focused on the conservation of the
Tanganyikan ecosystem (Alin et al., 1999; Cohen et al., 1993). Using paleoecological
reconstruction, Wells et al. (Wells et al., 1999) showed that areas of the lake that
experienced intensive deforestation of their watersheds also saw substantial declines in
the diversity of their ostracod faunas in recent decades or centuries. To place these
observations in a long -term, natural context, it is necessary to calibrate the fidelity of the
fossil record to living communities and to estimate the resolution of lake sediment record.
Lake basins can provide extensive and highly resolved paleorecords for
reconstructing past environmental and ecological changes in terrestrial milieus. Lakes are
also excellent settings for studying the responses of ecosystems to natural and
anthropogenic environmental change on human time scales. High lacustrine
sedimentation rates often result in sedimentary records of annual to decadal resolution,
allowing high resolution reconstruction of environmental and ecological change. Many
lakes contain annually laminated sediments, reflecting seasonal cycles of productivity
and/or stratification. Because of the accumulation of annual laminations, combined with
high -resolution dating techniques (21°Pb, 14C), it is possible to estimate the resolution of
the sedimentary record based on sediment accumulation rates, sampling resolution, and
the depth of the taphonomically active zone (TAZ: the post -burial zone through which
biotic and physicochemical processes continue to alter death assemblages prior to their
ascension to the fossil record - Davies et al., 1989). Lake -wide average sedimentation
rates in large lakes are typically on the order of 1.0 mm/yr when measured over tens to
thousands of years (Cohen, 2000; Johnson, 1984). Subseasonal sampling resolution is 109 therefore potentially possible (Davidson, 1988). Maximum depths of bioturbation and the TAZ tend to be shallower (2 -5 cm in Lake Tanganyika: Cohen, 2000) than in nearshore marine settings (up to -1.4 m in rare cases: Kidwell and Bosence, 1991), as a result of the lower densities and burrowing depths of bioturbating organisms found in
lakes. These factors combine to make lacustrine sediments amenable to very high -
resolution paleoenvironmental and paleoecological reconstruction.
Methods
Sampling
Ostracod life and death assemblages were collected in surface sediment samples
offshore from Mwamgongo, Tanzania, in Lake Tanganyika (Fig. B.1A,B). The shallow
benthic habitat at this site was composed dominantly of silty sands with patches of rocky
habitat. Colored bolts were affixed with underwater epoxy to four rocks, two each at 5
and 10 m water depth, to identify sampling locations. Each month, eight surface
sediment samples were collected using a diver -operated suction sampler modified from
Gulliksen and Delis (1975). A quadrat (25 x 25 cm) was used to collect surface
sediments from two adjacent patches on each bolted rock. Quadrat series were numbered
1A -4A (10 m) and 5A -8A (5 m). Sampled rocks at each depth were a few meters apart.
Sediment samples were collected monthly October -December 1997, February -July 1998,
and July 1999. This period spanned the transition from a very dry year to a very wet El
Niño year. During the sampling interval, lake level rose rapidly nearly 2.5 meters and
then declined again by 1.0 -1.5 m (Fig. B.1 C). Water depth data were recorded using a 110
dive computer and were calibrated using NASA satellite altimetry data (C. Birkett,
personal communication).
Ostracod fossil assemblageswere obtained from two short sediment cores collected
using a hand -coring device in July 1999. Core MWA -1 (16cm length -0 -9 cm
discussed here) was collected between 10m sampling sites, and core MWA -2 (11 cm
length) was taken adjacent to one of the 5 m sampling sites. Coreswere sectioned into 1
cm intervals in the field.
Six radiocarbon dates were obtained for core MWA -1 through the NSF Arizona AMS
Facility. All dates were derived from single terrestrial leaf fragments to avoid the dual
problems of mixing carbon sources of varying ages and the 14C reservoir effect of Lake
Tanganyika. Based on global atmospheric post -bomb 14C decaycurves, single leaf
fragments allow the assigment of calendar ages to within a fewyears of leaf production
(following Levin and Kromer, 1997; Nydal and Lövseth, 1983). Pre -bomb radiocarbon
dates were assigned using CALIB 4.3 (Stuiver et al., 1998a; Stuiver et al., 1998b). No
Southern hemisphere correction was applied because the study site is equatorial.
All surface sediments and core intervals were sieved using 1mm, 106 gm, and 63 µm
sieves, dried at 60 °C, and weighed. All ostracod individuals in the >1mm and 106 µm -1 mm sediment fractions that remained articulated and retained soft parts and coloration typical of live specimens were included in life assemblages andwere identified to the species level. All 80 surface sediment samples were tallied for live diversity.
For death and fossil assemblages, ostracods from the >1mm size fraction were added to the 106 µ m -1 mm size fraction before subsampleswere removed for counting and 111
identification of individuals. Standard sample sizes of 500 were used for death and fossil
assemblages. Thirty -seven of the 80 possible death assemblages were counted. Samples
from November 1997, July 1998, and July 1999 were counted for all quadrat series (1A-
8A) in order to sample changes in death assemblages at the beginning, middle, and end of
the collection period. In addition, all remaining monthly samples were counted for two
(3A, 7A) of the eight samples takenso that higher frequency changes in death
assemblages at both depths could be examined. For cores, all 1 -cm intervals were
counted. In addition, multiple ostracod counts were done for one core interval (0 -1 cm of
core MWA -1) in order to assess the reliability of a given sample in representing the full
ostracod assemblage present in that interval.
Ostracods were identified following Rome (1962), Martens (1985), Wouters (1988),
Wouters and Martens (1992), DuCasse and Carbonel (1994), Wouters and Martens
(1994), Wouters and Martens (1999), and Park and Martens (2001). For the many
Tanganyikan ostracod species not yet described, extensive reference collections at the
University of Arizona were used to identify individuals to the level of genus, with a
numbered species designation.
Data Analysis
Ranges of species richness values in live, dead, and fossil samples were compared in box -and -whisker plots. Individual samples of live species richness datawere
successively pooled both across space and through time to estimate the minimumamount
of spatial and temporal averaging represented by death and fossilassemblage data. 112
Kruskal- Wallis nonparametric analysis of variance was used to test for significant differences in species richness among data sets, because the Shapiro -Wilk test of normality rejected the hypothesis that live data were normally distributed (Sall and
Lehman, 1996). Dunn's test for multiple comparisons with samples of different sizes was employed to localize the difference amongsamples (Zar, 1984).
Species abundance data were calculated as the average abundance for each species in samples where it occurred. Species abundance distributions were compared among data sets using paired Kolmogorov-Smimov tests for goodness of fit (Zar, 1984).Life assemblages varied in number of individuals (range: 26 -1229, median: 139) and thus in their relative species abundances. Live abundance data were scaled to have comparable percentage intervals to those in the standard dead and fossil sample of500 individuals for the purposes of binning them into uniform abundance categories. This was done by multiplying the percent abundance for each species in a sample by the number of individuals in that sample and dividing the product by 500.
Occurrence frequencies were computed for all species in each data set. Live, dead, and fossil data sets contained different numbers of samples (80, 37, and 19, respectively).
Occurrence frequency bins varied in size such that each bin represented 10% of samples in a data set (resulting in bin sizes of 8, 4, and 2 samples for live, dead, and fossil data, respectively). Species occurrence frequency distributions for live, dead, and fossil data sets were compared using paired Kolmogorov- Smirnov tests, using one data set for expected values, the other for observed values. 113
For rank abundance tests, species in all data sets were ordered based on their abundance in the live data set. Species that did not occur in the live data set were excluded. Rank abundance data were compared using Spearman's coefficient of rank correlation (Sall and Lehman, 1996). Abundance data were also compared by pairwise computation of Pearson's product- moment correlation coefficient, which does not explicitly address species rank.
Fidelity of species composition among live, dead, and fossil data sets was compared using a variety of methods. Percentages of live species found dead and vice versa were used as fidelity metrics (Kidwell, 2001; Kidwell and Bosence, 1991). Comparison of species composition between life and death assemblages were also extended to assess the fidelity of the fossil data to both. We also determined percentages of dead individuals from species found live as a means of gauging live:dead fidelity (Kidwell and Bosence,
1991).
Additional sediment subsamples tallied for ostracods from the 0 -1cm interval of core
MWA -1 were used to generate a species sampling curve for the entire core interval. Six samples of 100 individuals each, three samples of 500, and an additional sample of 610 were counted. For one of the samples of 100, a running tally was kept of each new species occurrence. A logarithmic curve was fitted to the sampling curve in order to determine whether our sample size of 500 was sufficient to pass the inflection point of the diversity curve. In addition, we tallied numbers of occurrences for all species in four samples of 500 (five of six samples of 100 were pooled for this comparison) and calculated detection probabilities for species in differentaverage abundance classes. 114
Another means of judging the adequacy of sample sizes is to calculate the predicted percentage of unique species in each sample based on the value of Fisher's a for the observed species distribution from the same core interval (following Koch, 1987). To estimate the values of Fisher's a and x needed to generate the expected number of species in each occurrence category, code from Rosenzweig (Rosenzweig, 1995: p. 194 modified by M. Rosenzweig) was used. The expected number of species in an additional sample of
500 was calculated as a x, a x2/2, a x3/3, 43- Eaf/n, with 43 being the average species
richness for four counted samples, in four observed occurrence categories (Koch, 1987;
Magurran, 1988). Probabilities of occurrence (pa) for four samples of 500 were then used
to calculate the predicted similarity in species composition for one additional sample of
500 (Koch, 1987: Table 3 & Appendix).
Detrended correspondence analysis (DCA) was employed using CANOCO 4
software to explore the combined live, dead, and fossil database for differences in
community structure among sample types (ter Braak and Smilauer, 1998). In order to
avoid some of the gradient distortions reported for DCA (Minchin, 1987; Pielou, 1984),
detrending was executed using polynomials rather than segments (ter Braak and Prentice,
1988). Species relative abundance data failed the Shapiro -Wilk test of normality, hence
all data were log- transformed. In addition, the CANOCO option to downweight rare
species was employed in the DCA analysis. Explicit measurements of environmental
variables other than water depth were lacking for this study.
In order to examine the effects of spatial averaging across substrate types, an ostracod
substrate index (OSI) was defined for ostracod species assemblages basedon the output 115 of a canonical correspondence analysis (CCA) of a database of live ostracod species abundance data from various sites, substrate types, and water depths around Lake
Tanganyika. CCA Axis 1 was significantly and strongly correlated with substrate type
(rocks, sand, and mud). OSI was defined using Axis 1 species scores to assign species to rocky, sandy, or muddy categories. Separation of samples along Axis 1 with respect to substrate type was good although not complete, as many species are commonly collected live in more than one habitat type. OSI is equal to (Nsandy +Nmuddy) N irocky, where N is the number of individuals in each category, so that smaller values correspond to a greater proportion of rocky species, and larger values to more sandy or muddy species.
Results
Sedimentological and Radiocarbon Data for Cores
Visual inspection revealed three zones differing in organic content and particle size in core MWA -1. Between 3.5 and 7.5 cm there was a transition between the reddish brown silty sand at the core top (0 -3.5 cm) and the darker brown silty sand with organic fragments and some pebbles in the lower core (7.5 -16 cm). Grain size data showed fining of particles upwards of 8 cm in the core, with average weight percents of particles
<1061.1m increasing from 20% below 8 cm to 32% above 8 cm.
Visual inspection of core MWA -2 suggested a possibility of finer sediments above 5 cm and higher organic content below, with reddish brown sand throughout, although the core showed no overall trend in grain size based on granulometry. Mean weight percents of particles <_ 106 µm were comparable to those at the top of core MWA -1 at 31%. 116
Granulometric data for surface sediments showed dramatic month -to -month
fluctuations in quantity (ranging from <1 to -40 g dry sediment per quadrat) and particle
size distribution. Variations in sediment particle size were not correlated with
fluctuations in ostracod diversity and abundance. Mean weight percentages of particles
<106 pm were 38% and 65% at 5 and 10 m, respectively.
Radiocarbon dates obtained from single leaf fragments in core MWA -1 are shown in
Table 1, and suggest a midcore depositional hiatus of -300 years. Based on the jump in
radiocarbon ages, the hiatus lies between 9 and 10 cm. In this paper we present ostracod
data only from the upper 9 cm of core MWA -1 because our aim was to calibrate the
currently accumulating paleoecological record with the extant living and death
assemblages. Post -bomb radiocarbon dates indicate that the upper 9 cm of core MWA -1
represent approximately the last three decades of deposition (Table 1). Material suitable
for radiocarbon dating was unavailable from core MWA -2. We assume that sediment
accumulation rates, and thus sample resolution, were comparable for both cores.
Estimated ages in upper MWA -1 suggest recent sediment accumulation rates between 0.6
and >4 mm/yr in the nearshore zone at our site.
Characteristics of Ostracod Assemblages
The live data set, composed of 80 samples, consisted of 15,765 individuals and 64
species (Appendix F). Total death assemblage individuals tallied were 18,175, comprising 87 species (Appendix F). Fossil assemblages contained 8,851 individualsand
79 species (Appendices G, H). 117
Ranges of values in species richness raw data are shown for ostracod life, death, and fossil assemblages in Figure 2A. Kruskal -Wallis tests for analysis of variance soundly rejected the hypothesis that the live, dead, and fossil data sets shared a common range of species richness values (Hc =99.2, p «0.001). Dunn's multiple comparison test showed significant differences between live species richness data and both death (Ql;,,e_dead= 9.103, p <0.001) and fossil assemblage data (Qlve_fossil= 6.158, p <0.001), with no difference between species richness values of death and fossil assemblage data (Qaeaa-fossil=0.793, p >0.20). However, after species richness data were pooled across all samples through time or space (Fig. B.2B), species richness values were comparable among live data pooled through space, live data pooled through time, and raw death and fossil assemblage data (Fig. B.2C). Kruskal- Wallis tests detected no difference among these data sets for species richness values (Hc= 5.804, p >0.10). Interestingly, pooling samples through either space or time resulted in equivalent numbers of species, although the spatial accumulation curve (circles in Fig. B.2B) ascended more steeply initially indicating high spatial heterogeneity in ostracod life assemblages.
Species abundance histograms are shown in Figure 3. All plots show a predominance of species represented by fewer than 1% of individuals (average) in a sample. Paired
Kolmogorov -Smimov tests for goodness -of -fit indicated no difference in the species abundance structure of the three data sets (p >0.50 in all comparisons).
Occurrence frequencies tallied for species in all data sets are shown in Figure 4. Live species occurrence frequencies showed a unimodal distribution, with a prominent peak representing species that occurred in <10% of samples. In contrast, death and fossil 118 assemblage occurrence frequencies were bimodally distributed, with large peaks at both ends of the distribution representing species present in <_ 10% and >90% of samples. The distribution of species in the lowest live occurrence category across death and fossil assemblages (Fig. B.4) shows that most species remain in the lowest occurrence categories, but several appear in the highest occurrence category, representing those species that are rare but persistent. In contrast, species comprising the >90% bins for both death and fossil assemblages are somewhat more evenly distributed across the live data set (insets in Fig. B.4), indicating that persistent species occur in all live occurrence classes. Interestingly, when live data were pooled through either time or space, a bimodal distribution similar to those for the death and fossil assemblages resulted (Fig.
B.5). Thus, pooled live data again display comparable patterns to death and fossil assemblages.
Kolmogorov- Smirnov tests confirmed that the shapes of the dead and fossil occurrence frequency distributions were indistinguishable (dmax( 1 0,79)=3 .8, p >0.50), whereas live data were distributed significantly differently from both (live -dead: dmax(10,64) =13.4, p <0.01; live -fossil: dmax(10,64) =15.0, p<0.005). One possible caveat for interpreting the shape of occurrence frequency histograms is that when values in the expected data set differ markedly from equality (as ours did), the robustness of this test may become unreliable (Pettitt and Stephens, 1977). However, based on the variability in test p- values reported by Pettitt and Stephens (1977) related to inequality of expected values and the strength of our results, it is highly unlikely that the direction of these 119
relationships has been misidentified by this test. In other words, the variability of p-
values is smaller than the offset required to change the significance of our results.
Species ranked in order of their live abundance follow the expected hollow curve for
live data (Fig. B.6). Ranked species abundance data appear to differ substantially
between life and both death and fossil assemblage data sets (Fig. B.6). However,
Spearman's test of rank correlation reveals significant,albeit weak, correlation for all
three comparisons (rlive -dead= 0.450, p= 0.0002; rlive-fossil=0.329, p= 0.008;rdead- fossil = 0.795, p <0.0001). In contrast, Pearson's correlation coefficient indicates significant correlation
only between live:dead and dead:fossil abundances (alive-dead=0.257, p= 0.041;rlive-
fossil=0.065, p =0.61; rdead-fossil =0.895, p <0.0001). Both results are strongly influenced by
the number of species in the comparison. To test the effect of number of species
compared, we plotted p- values for Pearson's and Spearman's coefficients for various
subsets of the ranked species abundance data (Fig. B.6). In the ten species comparison,
only the first ten species in order of live rank were retained, and so on. It is apparent that
only the dead:fossil abundance correlation was consistently and highly (p <0.0001)
significant using either coefficient in sample sizes up to 40. With 60+ species, all
correlations were significant except for live:fossil abundance with Pearson's coefficient.
Together, these results indicate that the majority of species have correlated abundances
among data sets, despite the presence of a few strong outliers. It is worth noting,
however, that the results were largely determined by the size of the data set considered.
Spearman's coefficient in particular seemed to lack statisticalpower to discriminate
among data sets with apparently very different distributions (Kendall's rank correlation 120 coefficient performed no better). Pearson's product- moment correlation coefficient appeared to be slightly less susceptible to variations in numbers of species compared.
Thus, it may be useful to adopt more stringent criteria (e.g., p <0.0001) for rejecting the null hypothesis of zero correlation among data sets containing many species (e.g., >50).
Alternately, if rank location of the most abundant species is of most interest, results of
Spearman's test of rank correlation may be better evaluated on the basis of an rz criterion, e.g.,r2 must be >0.50 for rank order correlation to be deemed significant. The r2 values for our live:dead:fossil comparisons are ¿live-aead =0.20,r21i,,e_fossi1=0.11, and r2dead- fossil=0.63. Level of similarity in ranked abundance diagrams is more clearly conveyed by r2 values, which indicate the strength, rather than the significance, of the correlation.
Fidelity of Death and Fossil Assemblages to the Living
Raw fidelity measures for species composition among data sets were generally high
(Table 2; range: 53 -90 %, median: 78.5). Agreement was closest in percentages of live species found dead, live species found as fossils, and fossil species found dead.
Appearance of disagreement in percentages of dead species found live, dead species found as fossils, and fossil species found live is largely an artifact of differences in species richness between samples being compared. For this reason, percentages of maximum possible agreement were also calculated for the three cases where the fauna in the denominator (i.e., in % live found dead, number of live species is the denominator, number found dead the numerator) was more species rich than its comparison data set
(following Kidwell, 2001). After this adjustment, fidelity in species compositionamong 121 all lists was quite high (range: 77 -90 %, median: 88 %). Interestingly, agreement among species lists decreased when the lists were truncated to include only the more abundant species, again indicating that some rare species are among the persistent species pool at this site.
Percentages of dead individuals found live at the same depth were 98% at 5 m and
87% at 10 m. When the percentage of dead individuals at 10 m which are only found live at 5 m was added to the number of dead individuals found live at 10 m, the agreement increased from 87% to 99 %. This implies a role for down -slope transport in determining the species composition of death and fossil assemblages, at least in shallow water.
Fidelity can also be examined by comparing numbers of dominant taxa shared among the data sets (Table 3) (cf. Kidwell and Bosence, 1991). Fidelity among dominant taxa was highest between the dead and fossil data sets, with eight of ten dominant speciesin common at 5 m and seven of ten shared at 10 m. Live:dead agreement was substantially lower, with four of ten dominants shared at 5 m and only two of ten at 10 m. Finally, the fewest matches occurred between live and fossil species lists, with three of ten matching at 5 m and only one at 10 m. However, agreement between 5 and 10 m within each data set was quite good. In the live data set, seven of ten dominant taxa were shared. Dead and fossil data sets had nine and eight species of ten dominants in common between depths. 122
Analysis of Core Interval Re- Sampling
The species accumulation curve resulting from resampling a single core interval is
shown in Figure 7. A total of 2,710 individuals were counted, yielding 61 species. The
order in which samples were added affected the regression equation minimally, and allr2
values were >0.95. Extrapolation of the logarithmic curve to 10,000 individuals yielded
an estimate of -66 species, suggesting that >90% of all species in this core interval had
been sampled. However, if the curve is extrapolated to the total number of individuals
contained in this core interval (88,000), the estimated total species richness for the
sample is 85, suggesting that our sampling could be as poor as -50 %. Eighty -five
species is not an unreasonable number for this site, as 101 species were tallied in the live,
dead, and fossil data sets together, although it is unclear that extrapolation to such sample
sizes would be robust. In any case, our sample size of 500 was sufficient to have crossed
the inflection point on the sampling curve. Total species richness of the resampled core
interval is approximately 1.5 times as high as that observed in a similar analysis on
another core from Lake Tanganyika (Wells et al., 1999) and can probably be explained
by diversity differences between water depths of the cores (40 m in Wells et al. vs. 10 m
here).
Species composition comparisons for four samples containing 500 individuals
revealed that almost half the species occurred in all four samples (Table 4). Another third
of species appeared in two to three samples. Table 5 shows the observed probabilities of
detection for species based on their average abundance in samples of 500. Except for
>0.65% category, groups contained species with varying numbers ofoccurrences, as they 123 were grouped by abundance rather than occurrences. All species present in >0.65% abundance fell into the 100% detection probability category, and only species with
<0.25% abundance had <50% probability of detection.
For estimating numbers of unique species, Fisher's a and x were determined to be
11.2 and 0.994, respectively, for a total species richness of 58 in 2000 individuals (four samples x 500 /sample). The log series distribution fit our data well, as the values of a and x changed minimally when calculated with various subsets of the data. Table 6 shows the values for expected number of species in each category, probability of species in each category being present in one additional sample, and predicted number of species from that category to appear in the additional sample. Our results indicate that 28% of the species in each of our samples of 500 may be unique, i.e., they are not likely to be present in an additional sample taken from the same core interval. This is essentially equivalent to the 26% of species present in <0.20% average abundance in Table 5.
Comparison of these two results suggests that, on average, only those species represented by a single individual in a sample of 500 are unlikely to be resampled in an additional tally from the same sample.
Ordination
Ordination plots based on DCA of the live -dead -fossil database reveal fairly good separation for the majority of live species assemblages at 5 and 10 m, although some overlap is apparent (Fig. B.8A). In contrast, close association of death and fossil assemblages was seen at the same site (Fig. B.8B), with both groups offset from live 124 samples. Indirect gradient analysis indicated that DCA Axis 1 was strongly correlated with species richness of samples (more negative Axis 1 loadings = higher diversity) and with the abundance of dominant taxa (Mesocyprideis irsacae and Romecytheridea ampia for dead and fossil samples [negative loadings on Axis 1], Allocypria mucronata for live
samples at 5 m, and Romecytheridea tenuisculpta for 10 m live samples [positive
loadings for both on Axis 1]). Abundance of these species varies with substrate, with A.
mucronata being the only "rocky" species among them. Based on the ecological
correlates of Axis 1, death and fossil assemblages are offset from life assemblages by
virtue of being richer in species and having species compositions reflecting a degree of
spatial averaging across habitat type. We interpret Axis 1 as representing dominantly
substrate texture /grain size (positive Axis 1 loadings corresponding to coarse -grained
[rocky] habitats, negative loadings to fine- grained [sandy, muddy] substrate). DCA Axis
2 was correlated with both depth (higher Axis 2 loadings = greater depth) and abundance
of dominant taxa.
Ostracod substrate index (OSI) values support the interpretation of Axis 1 as related
to substrate grain. Average OSI values for death and fossil assemblages (3.1 ±0.9, and
4.2 ±1.0, respectively) are both substantially higher than for life assemblage OSI values
(1.7 ±1.6), confirming that fine -grained substrate species comprise a greater percentage of
individuals in death and fossil assemblages.
The pattern within the live ostracod data was not simple to interpret. What overlap
did occur may be somewhat attributable to lake level fluctuations and depth preferences
of the samples' constituent ostracod species. Samples from 5 and 10 m in closest 125 proximity were those from the 5 -m locations during high water months (March -May
1998, Fig. B.1C) and from 10 -m quadrats in relatively low water months (October 1997,
July 1999), although this pattern was not consistent throughout the remaining samples.
Samples collected from adjacent quadrats (i.e., on the same rock) tend to plot closer together in ordination space, but samples collected from different rocks in the same month are sometimes more similar. Finally, sample series from single quadrat locations tended to follow complex trajectories that frequently ended at a point in ordination space closer to the origination point than many of the intervening samples (Fig. B.8C,D).
These patterns simply confirm the high degree of spatiotemporal heterogeneity in living ostracod assemblages reported by Cohen (1995; 2000). The heterogeneity is probably caused by numerous factors such as changes in lake level, species population levels in preceding months, and seasonal to inter -annual climate cycles.
Dispersion in ordination space of death and fossil assemblages was substantially lower than among live samples. Samples were distributed along both axes, indicating the importance of at least two environmental gradients in determining their species composition (Fig. B.8B). Many of the death assemblage samples are so similar to assemblages from core MWA -1 that they are superimposed in ordination space. In general, dead samples had higher loadings on Axis 2 than fossil samples. In both data sets, samples from 10 m have higher loadings on Axis 2 than samples from 5 m, a pattern that matched the results for live samples. Thus, death and fossil assemblages retained relationships observed among the living assemblages with respect to water depth to some 126 extent, although sample variability was muted in death and fossil assemblages relative to live samples.
Discussion
Preservation of Life Assemblage Attributes in Death and Fossil Assemblages
Numerous lines of evidence suggest that both death and fossil assemblages accurately preserve the community structure and composition attributes of the living ostracod fauna at high resolution, despite the fact that spatial and temporal averaging reduce the variability seen in life assemblages. Although species richness of death and fossil assemblages exceeds that of raw live diversity in quadrats by two- to three -fold, comparable and statistically indistinguishable numbers of species were present in pooled life assemblages. Based on pooled live data, minimum estimates for time -averaging of death and fossil samples were on the order of one year. Alternately, minimum spatial integration seen in death and fossil assemblages was roughly equivalent to several square meters of habitat area. Thus, it appears that death and fossil assemblages accurately represents the diversity of living ostracod assemblages at annual and habitat -scale resolution.
Species abundance distributions are not significantly different among live, dead, and fossil data sets (Fig. B.3). All three are dominated by rare species, with 29 of 64 live species, 64 of 87 dead species, and 57 of 79 fossil species having average abundances of less than one percent. Translation of ostracod life assemblages into death and fossil 127 assemblages does not appear to have introduced significant bias to species abundance distributions.
Although the rank abundance diagrams appear quite different, ranks of the majority of species were significantly correlated according to Spearman's rank correlation test.
Pearson's correlation coefficient was more influenced by outliers in the abundance data.
Obvious outliers in histograms for death and fossil assemblages (Fig. B.5) can be accounted for by considering habitat preference and varying preservation potential of species related to shell thickness. Based on the live ostracod CCA, the outlying species
Mesocyprideis irsacae, Romecytheridea ampia, Mesocyprideis pila, and
Tanganyikacypridopsis depressa reach their highest abundance in sandy sediment.
Mecynocypria emaciata is most abundant in rocky habitats, although it is commonly collected in sandy habitats as well. All five taxa, with the possible exception of M. emaciata, have well -calcified valves relative to many of the cypridoidean ostracod species in Lake Tanganyika. Of the six most abundant live species (with depths pooled, in order: Romecytheridea tenuisculpta, Allocypria mucronata, Cypridopsis sp. 18,
Romecytheridea longior, Allocypria sp. 11, and Allocypria inclinata), R. longior, A. mucronata, and A. inclinata are known primarily from rocky habitats. R. longior is a shallow -water sandy species, most often observed in poorly- calcified juvenile instars. Of the six most abundant live species, only A. mucronata and C. sp. 18 are relatively poorly calcified.
While it remains possible that there is slight taxonomic bias against the most poorly calcified cypridoidean species, overall correlationamong the assemblages suggests 128
minimal skew in the ranked abundances of death and fossil assemblages with respect to
life assemblages. However, death and fossil assemblages bear the signature of spatial
mixing across habitat types, as reflected by the dominance of M. irsacae and R. ampia in
both. Life and death assemblages for this study were collected only from rocky habitat
patches at our site. Death assemblages clearly comprised individuals from both rocky
and silty -sand habitats, which were more extensive at this location. Cores were collected
in the silty -sand facies, and ordination indicated that they contained assemblages very
similar to the death assemblages. Thus, ostracod death assemblages appear to be
spatially integrated to an extent that renders them more representative of fauna at an
entire site than are individual live samples. The similarity between death and recent
fossil assemblages suggests that this fidelity is carried through to the paleoecological
record as well.
The fidelity of microinvertebrate death and fossil assemblages to life assemblages
reported here differs in several respects from results of comparable studieson marine
macroinvertebrate death assemblages. Kidwell and Flessa (1995) conclude,on the basis
of a compilation of taphonomic data, that transport out of the immediate life habitat is rare. We argue here that death and fossil assemblages are spatially integrated within sites
across substrate or habitat types. For the purposes of paleoecological reconstruction, we deem this a positive outcome of taphonomic processes, in that it renders the averaged
samples more representative of a larger area of habitat. Kidwell and Flessa (1995) make
a similar argument for the virtues of time -averaging. In marine macroinvertebrate
assemblages, "most species with preservable hardpartsare...represented in the local 129 death assemblage, commonly in the correct rank order abundance" (Kidwell and Flessa,
1995). Ostracod death and fossil assemblages in Lake Tanganyika are statistically indistinguishable in rank correlation tests from life assemblages, although the shape of their rank abundance curves appears markedly different. It is unclear to what extent rank order results for micro- and macroinvertebrates are comparable (cf. Kidwell 2001). Rank order differences among ostracod life, death, and fossil assemblages may stem from within -site transport and homogenization of spatially heterogeneous ostracod population distributions.
Fidelity Metrics
Values obtained for the fidelity metrics of Kidwell and Bosence (1991: Table 1) lend further support to the interpretation that fidelity of death assemblages to contemporaneously collected life assemblages is quite high. Agreement ranged between
77 and 90 percent for all comparisons of percent of live found dead, percent of live found as fossils, and percent of fossil species found dead. Agreement was weaker for the complementary comparisons (% dead found live, % fossil species found live, % dead found as fossils), ranging between 53 and 83 percent, because of differences in the total species richness among the data sets. After applying the "maximum possible agreement" correction suggested by Kidwell (2001) to account for this problem, the agreement values obtained were equal to those obtained in the first set of comparisons (i.e., ranging between 77 and 90 %). This is because the correctionto "maximum possible agreement" values involves a conversion from values of "% deadfound live" back to "% live found 130 dead." These "maximum possible agreement" metrics, which ultimately use the total number of taxa in the smaller fauna as the denominator, are equivalent to Simpson's index of similarity (Simpson, 1960). Simpson's index provides a maximum estimate for similarity between the two samples being compared. Any estimate of similarity will be influenced by the total species richness of one sample or the other. Using the larger fauna in the denominator only serves as an indirect indicator of the discrepancy of species richness between the two samples. Thus, it seems reasonable to report only the Simpson - equivalent (i.e., maximum) fidelity value, as the disparity in species richness can be assessed more efficiently by simply comparing numbers of species between samples.
Sampling Efficiency
Counting multiple samples from the same core interval allows assessment of the adequacy of sampling for our death and fossil assemblages. Analysis of the data from this core interval indicated that approximately 28% of species in a sample of 500 can be expected to be unique. This corresponds closely with the percentage of species represented by fewer than 0.20% of individuals on average. Therefore, downweighting rare species in ordination analyses should also be done on the basis of their inconsistent detection in death or fossil assemblages. However, the dominance of rare species is a trademark of Tanganyikan ostracod assemblages. We would not want to exclude rare species entirely from paleoecological diversity analyses on the basis of incomplete sampling. Indeed, previous studies have suggested that disappearance of rare species from ostracod assemblages can be an important indicator ofsevere anthropogenic 131 disturbance in adjacent watersheds (Wells et al., 1999). Furthermore,ecological studies have demonstrated the importance of rare species in assessments ofecological integrity at sites (e.g., Cao et al., 1998), and it makes sense to extend this perspective to the paleoecological record wherever possible. Although inclusion of rare species is desirable, it would not serve to place much importance on the particular identities of the rarest species, as their exact composition is likely to changewith additional sampling from the same core interval or surface sediment sample. Maximum similarity metrics such as Simpson's should not surpass 70% on average if none of the predicted unique species were shared among samples from the same interval. Despite caveats about the unreliability of detecting rare taxa, observed agreement in pairwise comparisons of samples from interval 0 -1 exceeds this value (79% on average), demonstrating greater actual species similarity than predicted.
When cumulative species richness from core MWA -1 is superimposed on data from the uppermost core interval (0 -1), the trends are indistinguishable (Fig. B.7). This suggests local stability in the species pool for dominant taxa through the period represented by the core (-30 yr). Rare species appeared at the same rate throughout the core as in the resampled interval,reflecting sampling intensity and the abundance distribution of rare species. This suggests continuity in numbers and abundance of rare taxa through the core. 132
Scaling Issues and Equivalence of Fish and Ostracod Community Dynamics
Levin (1992) discussed the importance of crossing scales in ecology to facilitate the prediction of the effects of global environmental change on communities and ecosystems.
In this context, the two aspects of scale most frequently addressed are spatial and temporal scales of sampling and analysis. Levin (1992) also emphasized that the scales inherent to the observer of the environment are important, with the observer here being an individual of any species in its environment. Thus, in addition to consideration of objective scales of time and space, investigators should also consider the scale perspective of the organisms under study, as each species' response to environmental change will be influenced by its life history characteristics, resource requirements, and disturbance responses (Levin, 1992).
Ideally, ostracods could be employed as paleoecological indicators of the entire benthic community. However, body size differences between ostracods, larger invertebrates, and fish suggest that organismal scaling issues must be considered. Cohen
(2000) discussed the implications of the disparity between fish and ostracod community dynamics in Lake Tanganyika for the application of paleoecological insights to conservation. Nakai et al. (1994) demonstrated the stability of fish assemblages at a site in Lake Tanganyika for over a decade and attributed this stability to deterministic and highly coevolved species interactions. In contrast, Cohen (2000) described ostracod community dynamics as highly patchy in space and time, at the scale of hundreds of meters, and invoked metapopulation dynamics as the mechanism for the maintenance of ostracod diversity. In our study, species richness rose more steeply when live diversity 133 data were pooled through space than when pooled through time (Fig.B.2B). This confirms that spatially heterogeneous patterns in ostracod life assemblages arealso borne out at the scale of meters. The shallower rise in thetemporally pooled curve implies some degree of stability in the speciespool, despite the patchiness of ostracod distributions, and may represent a seasonal succession of species.
Cohen (2000) concluded that "the contrast between cichlid and ostracod diversity structure in space and time suggests that no one taxonomic groupis likely to serve as a robust model for how diversity is maintained." Thus, in order to used ostracod paleoecology as a general indicator of benthic community integrity, it is necessary to understand the relationship between fish and ostracod diversity dynamics. Apparent differences in stability between fish and ostracod communities were equalized to some extent when ostracod death and fossil assemblages were studiedinstead of life assemblages. Our ordination results indicated that, while overall variability among live samples was quite high, death and fossil assemblage variability was fax lower. Death assemblages, which we have shown to preserve characteristics of live communities with good fidelity, were thus better indicators of the summation of ecological conditions at our site than any single live -collected samples were. In addition, samples in both sediment cores were very similar to death assemblages, suggestingminimal change in the ostracod fauna at this locality during the past few decades.
Inter -sample relationships in ordination space were largely determined by the abundance of common taxa. Identity of rare species was less consistent among samples in all data sets, but rare taxa were downweighted and played relatively minor roles in the 134 outcome of the ordination. The reduced variability seen in death and fossil assemblages is thus largely attributable to the influence of common taxa. Lower variation in death and fossil samples supports the notion of stability in the dominant component of the local species pool, despite high spatiotemporal heterogeneity of live populations and the instability of the large proportion of rare taxa in ostracod assemblages. When viewed through death and fossil assemblages, with their inherent spatiotemporal averaging, ostracod community dynamics no longer appear so disparate from those of fish.
After thorough consideration of spatiotemporal scaling issues, the question again arises: can ostracods safely be used as paleoecological indicators of the integrity and function of benthic communities as a whole? Dominant taxa in ostracod assemblages show similar stability to fish assemblages when viewed in time- averaged death or fossil assemblages. One caveat to the utility of ostracods as indicator taxa for the entire benthos is that ostracods may have a different response threshold than either fish or molluscs to sediment inundation, which is the dominant anthropogenic threat to
Tanganyikan habitats and species (Alin et al., 1999). This makes sense, as small -scale influx of sediment may constitute a disruption of habitat or food quality for fish, but may simply represent additional nutrients to ostracods. Larger -scale sedimentation changes may render habitat unsuitable for both taxa. Therefore, ostracods may provide a conservative estimator of change for the benthic community. 135
Novel Contributions of Paleoecological Observations to Conservation Biology
Two observations suggest that sampling death and fossil assemblages provide a more efficient means of gauging the response of ecological communities to natural or anthropogenic environmental change than live sampling alone. First, ordination of live, dead, and fossil data sets showed lower variability among dead and fossil samples than among live samples. Post -mortem mixing of ostracod assemblages leaves its signature on the species composition and richness of death and fossil assemblages by integrating sample membership across habitat types and through time. Spatiotemporal averaging allows death and fossil assemblages to simultaneously retain high -resolution information about the ostracod life assemblages that contributed to them and render information about the average composition of these communities averaged over short time scales.
Second, occurrence frequencies of species indicate which species are ecologically persistent. Some persistent species are rare and would not be identified as persistent based on live sampling. Species occurrence frequencies give us a means of identifying species, using paleoecological data, that may be more likely to recolonize a habitat after a disturbance, based on their tendency to recur through time. Wells et al. (1999) observed the disappearance of rare species at a highly disturbed site following watershed deforestation. Examining the occurrence frequency of species in paleoecological assemblages may provide a predictive tool for identifying components of an invertebrate community that are relatively extinction -prone versus those that are extinction -resistant. 136
Conclusions
Death and fossil assemblages of ostracods in Lake Tanganyika accurately record attributes of living ostracod communities such as species richness, abundance, and composition. Spatial and temporal averaging of death and fossil assemblages, on the scale of meters and months to years, reduces variability in species diversity and composition relative to life assemblages. Because death and fossil assemblages represent communities integrated across habitat type and through the vicissitudes of fine -scale ostracod population dynamics, these assemblages are more representative of community dynamics at the whole habitat scale than individual live -collected samples can be.
Paleoecological sampling of ostracod assemblages therefore provides a high -resolution tool for analyzing benthic community dynamics through time.
Furthermore, paleoecology can generate insights useful for conservation that are not amenable to analysis based on live communities alone unless one samples the live communities for sufficiently long intervals. Paleoecological analyses of diversity changes lend perspective on the ecological persistence of individual species whose abundance and stability may not be correlated. Ordination of ostracod fossil assemblage data shows promise as a reliable, informative means of assessing trends in diversity and composition of the benthic community through time.
Acknowledgments
We gratefully acknowledge assistance with field collections by K. Fadhili, J. Houser,
R. Shapola, T. Thompson, and especially C. O'Reilly. We thank the Tanzanian 137
Commission for Science and Technology for research permits, the Tanzanian Fisheries
Research Institute and the United Nations Development Program/Global Environmental
Facility's Lake Tanganyika Biodiversity Project for logistical assistance. We thank the
National Science Foundation/University of Arizona AMS Facility for radiocarbon dates and D. Dettman and O.K. Davis for assistance in preparing radiocarbon samples. We also thank K. Flessa, P. Reinthal, D. Dettman, L. Kaufman, J. Overpeck, R. Robichaux, D.
Goodwin, and D. Harris for helpful conversations; K. Flessa, R. Robichaux, and P.
Reinthal for detailed comments on previous versions of this manuscript; and C. Birkett for NASA satellite altimetry lake level data used to calibrate sampling location water depths. S.R.A. is thankful for financial support of this project from a University of
Arizona Graduate College Dean's Fellowship, a National Science Foundation Graduate
Research Fellowship, a National Security Education Program Graduate International
Fellowship, a Geological Society of America Graduate Student Research Grant, the
Wilson R. Thompson Scholarship (Department of Geosciences, University of Arizona), and the University of Arizona Graduate Student Final Project Fund. This study was also supported by NSF grants EAR -9627766 and ATM -9619458. This is a contribution of
IDEAL (International Decade of East African Lakes). 138
Figure B.1. A, Map of Africa with Lake Tanganyika inset. B, Location of study area. C, Water depth curve for quadrat series lA and 2A throughout the sampling period. Sampled months are indicated by the black bars under the curve. Water depths for other quadrat series are 0.6 m deeper for series 3A -4A and 5.3 m shallower for series 5A -8A. B 29° r Burundi 30° RepublicDemocratic of i...of __ Mwamgongo 4° ow Congo Tanzania 6° C E 10- 0- J A J1 O -1998 J A ji 1N .b al. 11- - 100km 8° 13-12- -- - IN Zambia 140
Figure B.2. A, Box plots of species richness values for all samples in the live (median =15), dead (median =37), and fossil data sets (median =36). B, Species accumulation curves resulting from pooling successive life assemblage samples within each quadrat location through the duration of the sampling period (10 months) and from pooling successive quadrats within each month up to the total number (8) of quadrats collected each month. C, Box plots of species richness for life assemblages pooled through space (median =36), life assemblages pooled through time (median=34.5), and raw death and fossil assemblages. 141
50
cA 40 .,-,a) Isi 8 30 * 20 10 10 A o live dead core # quadrats pooled (9) 1 2 3 4 5 6 7 8
1 2 3 4 5 67 8 910 # months pooled (u) 50 c40 a) '30 + 01 * *20 10 C 0 1 live live dead core (space) (time) 142
Figure B.3. Species abundance histograms for life, death, and fossil assemblage data. 143
45 LIVE
c.) 30 a) r::1 c.) * 15
0
60 DEAD a .0 40 * 20
0 60
1 3 5 7 9 >10 Average abundance ( %) 144
Figure B.4. Species occurrence frequency histograms for life, death, and fossil assemblage data. Life assemblage inset: Histogram shows distribution of live species in the <10% bin across the dead (black) and fossil (white) data sets. Occurrence frequency bins are the same as for large histograms. Death assemblage inset: Histogram shows the distribution of dead species in the >90% bin across the live data set. Fossil assemblage inset: Histogram shows the distribution of fossil species in the >90% bin across the live data set. 145
30 is LIVE
20
cA
0 30 6 DEAD
o --
0
1020 30 40 50 60 70 80 90 100 % of samples species occurs in 146
Figure B.5. Speciesoccurrence frequency histograms for life assemblage data pooled through space and through time. 147
20
2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 # of samples species occurs in 148
Figure B.6. Above, Rank abundance histograms based on live species abundance for life, death, and fossil assemblage data. Numbered species in panels B and C correspond to: 1= Romecytheridea ampia, 2= Mesocyprideis irsacae, 3= Mesocyprideis pila, 4= Tanganyikacypridopsis depressa, and 5= Mecynocypria emaciata (see text for discussion). Bottom, Distribution of p- values for Spearman's (S) and Pearson's (P) correlation coefficients for different numbers of ranked species (symbols: live vs. dead [S, crosses; P, open triangles], live vs. fossil [S, diamonds; P, solid triangles], dead vs. fossil [S, circles; P, squares]). Circles and squares overlap throughout. The line across the lower part of the graph represents = 0.05. 149
30 LIVE
15
2 1 DEAD
34 5
2 FOSSIL
20 - 1
34 5 0 11 21 31 41 51 61 Live species rank
10 20 30 4050 60 64 # species 150
Figure B.7. Species accumulation curve for core MWA -1. Open circles represent species tallied in subsamples from interval 0 -1 (cm). Solid circles indicate cumulative diversity from core interval 7 -8 cm through interval 0 -1 cm. The logarithmic curve is fit to the open circles. 151
75
o 60 o
S = 8.4 ln(N)-11.1 15 r2= 0.96
0 1000 2000 3000 4000 # individuals (N) 152
Figure B.B. Ordination plot of life (solid circles, 10 m; solid squares, 5 m), death (open circles, 10 m; open squares, 5 m), and fossil (open diamonds, 10 m; crosses, 5 m) ostracod assemblages. A, All life, death, and fossil assemblages plotted together. Mean values for life assemblages at 5 and 10m indicated with enlarged gray square and circle, respectively. B, Enlarged view of death and fossil assemblage distribution. Mean death assemblage values for 5 and 10 m indicated with enlarged, graysquare and circle, respectively. Core top samples from 5 and 10 m indicated by large,gray triangle and diamond. C, Temporal trajectory of monthly quadrat life assemblages at 10m in ordination space (triangles, dashed lines: 1A; diamonds, solid lines: 3A). First sample in each series is filled. D, Temporal trajectories at 5 m (circles, solid line: 5A;squares, dashed line: 8A). SI
i- S' VDU sixV I Z'I-Z VDU sixV I S'0- S'Z L'0
L'0- Z'Z S'I
L'0- t7'0 VDU sixV I I E'05' VD0 stxV I 154
Table B.1. Radiocarbon dates for core MWA -1 from 10 m water depth at Mwamgongo, Tanzania. Calendar ages reported for pre -bomb dates are all 2a age ranges with >0.1 relative probability.
Sample Depth in Fraction 14C Estimated number core (cm) modern 14C age calendar age range (AD)
AA -38063 2 -3 1.0934 post -bomb 1997 AA -38064 4 -5 1.1254 post -bomb 1992 AA -41870 7 -8 1.1729 post -bomb 1987 AA -41871 8 -9 1.5133 post -bomb 1971 -2 AA -38065 10 -11 0.9689 254 ±39 1632 -1670 1527 -1553 1780-1797 AA-38066 12 -13 0.9590 336 ±41 1466 -1644 155
Table B.2. Fidelity of ostracod death and fossil assemblages to life assemblages: presence- absence data.
5m 10m 5 & 10
% of live species found dead 88% 80% 89% (50/57 spp.) (39/49 spp.) (57/64 spp.)
% of live species found as fossils 77% 90% 89% (44/57 spp.) (44/49 spp.) (57/64 spp.)
% of dead species found live 65% 53% 66% (50/77 spp.) (39/74 spp.) (57/87 spp.) % of max. possible 88% 80% 90%
% of dead individuals found live 98% 87% 94%
% of dead species found as fossils 74% 76% 83% (57/77 spp.) (56/74 spp.) (72/87 spp.) % of max. possible 88% 86% 90%
% of fossil species found live 68% 68% 71% (44/65 spp.) (44/65 spp.) (57/80 spp.) of max. possible 78% 90% 89%
% of fossil species found dead 88% 86% 90% (57/65 spp.) (56/65 spp.) (72/80 spp.) average 76.7% 75.5% 81.3%
% found in all three 47% 44% 56% (41/88 spp.) (38/86 spp.) (55/99 spp.) 156
Table B.3. Ten most abundant species in life, death, and fossil assemblages at both depths in order of abundance. Numbers in parentheses after species names represent species rank in life assemblages at the same depth.
5 m: live dead fossil
Allocypria mucronata Mesocyprideis irsacae Mesocyprideis irsacae Cypridopsis n.sp. 6C Romecytheridea ampia (8) Romecytheridea ampia (8) Allocypria n.sp. 10 Mecynocypria emaciata Cypridopsis n.sp. 6A (6) Cypridopsis n.sp. 18 Cypridopsis n.sp. 18 (4) Tanganyikacypridopsis depressa Allocypria inclinata Cypridopsis n.sp. 6A (6) Mesocyprideis n.sp. 4 Cypridopsis n.sp. 6A Romecytheridea tenuisculpta (7) Mesocyprideis pila Romecytheridea tenuisculpta Mesocyprideis pila Gomphocythere curta Romecytheridea ampia Tanganyikacypridopsis depressa Cypridopsis n.sp. 18 (4) Allocypria n.sp. 11 Mesocyprideis n.sp. 4 Mecynocypria emaciata Allocypria cf. inclinata Cypridopsis n.sp. 23 Gomphocythere alata
10 m: live dead fossil
Romecytheridea tenuisculpta Mesocyprideis irsacae Mesocyprideis irsacae Romecytheridea long ior Romecytheridea ampia (4) Romecytheridea ampia (4) Allocypria n.sp. 11 Cypridopsis n.sp. 6A Mesocyprideis n.sp. 4 Romecytheridea ampia Romecytheridea tenuisculpta (1) Tanganyikacypridopsis depressa Allocypria inclinata Cypridopsis n.sp. 18 Cypridopsis n.sp. 6A Allocypria mucronata Mesocyprideis n.sp. 4 Gomphocythere curta Cypridopsis n.sp. 6C Tanganyikacypridopsis depressa Mecynocypria emaciata Cypridopsis n.sp. 25 Mecynocypria emaciata Cyprideis n.sp. 25 Tanganyikacypridopsis n.sp8 Mesocyprideis pila Cypridopsis n.sp. 23 Mecynocypria n.sp. 19 Mecynocypria n.sp. 20 Mesocyprideis pila 157
Table B .4. Average species abundance versus probability of detection based on resampling a fossil assemblage. average abundance # species probability of detection
> 0.65% 24/58 (41%) 100% >0.20- 0.64% 19/58 (33 %) 75% <0.20% 15/58 (26 %) 33% 158
Table B.S. Occurrence frequency of species in resampled core interval.
# occurrences # s ecies abundance ran ' e median abundance
4 28/58 (48 %) 0.30 -21.0% 2.00% 3 12/58 (21%) 0.25 -0.60% 0.28% 2 7/58 (12 %) 0.10-0.25% 0.20% 1 11/58 (19 %) 0.05 -0.25% 0.05% 159
Table B.6. Calculation of predicted overlap in species composition among samples from the interval 0 -1 cm in core MWA -1.
Observed occurrences Number of s ecies ex ected i Number of s ecies in new samle
1 11.1 0.25 2.8 2 5.5 0.50 2.8 3 3.7 0.75 2.8 4 22.7 >0.99 22.7
Average total S: 43 (100 %) Predicted shared S:31.1 (72 %) 160
References
Alin, S. R., Cohen, A. S., Bills, R., Gashagaza, M. M., Michel, E., Tiercelin, J. J., Martens, K., Coveliers, P., Mboko, S. K., West, K., Soreghan, M., Kimbadi, S., and Ntakimazi, G., 1999, Effects of landscape disturbance on animal communities in Lake Tanganyika, East Africa: Conservation Biology. Oct., v. 13, no. 5, p. 1017- 1033.
Anderson, N. J., 1993, Natural versus anthropogenic change in lakes: the role of the sediment record: Trends in Ecology and Evolution, v. 8, p. 356 -361.
Binford, M. W., Brenner, M., Whitmore, T. J., Higuera, G. A., Deevey, E. S., and Leyden, B., 1987, Ecosystems, paleoecology and human disturbance in subtropical and tropical America: Quaternary Science Reviews, v. 6, no. 2, p. 115 -128.
Brenner, M., Whitmore, T. J., Flannery, M. S., and Binford, M. W., 1993, Paleolimnological methods for defining target conditions in lake restoration: Florida case studies: Lake and Reservoir Management, v. 7, no. 2, p. 209 -217.
Brenner, M., Whitmore, T. J., Lasi, M. A., Cable, J. E., and Cable, P. H., 1999, A multi- proxy trophic state reconstruction for shallow Orange Lake, Florida, USA; possible influence of macrophytes on limnetic nutrient concentrations: Journal of Paleolimnology, v. 21, no. 2, p. 215 -233.
Cao, Y., Williams, D. D., and Williams, N. E., 1998, How important are rare species in aquatic community ecology and bioassessment ?: Limnology and Oceanography, v. 43, no. 7, p. 1403 -1409.
Cohen, A. S., 1995, Paleoecological approaches to the conservation biology of benthos in ancient lakes: a case study from Lake Tanganyika: Journal of the North American Benthological Society, v. 14, no. 4, p. 654 -668.
Cohen, A. S., 2000, Linking spatial and temporal change in the diversity structure of ancient lakes: examples from the ecology and palaeoecology of the Tanganyikan ostracods: Advances in Ecological Research, v. 31, p. 521 -537.
Cohen, A. S., Bills, R., Cocquyt, C. Z., and Caljon, A. G., 1993, The impact of sediment pollution on biodiversity in Lake Tanganyika: Conservation Biology,v. 7, p. 667- 677.
Davidson, G. A., 1988, A modified tape -peel technique for preparing permanent qualitative microfossil slides: Journal of Paleolimnology,v. 1, no. 3, p. 229 -234. 161
Davies, D. J., Powell, E. N., and Stanton, R. J., 1989, Relative rates of shell dissolution and net accumulation- a commentary: canshell beds form by the gradual accumulation of biogenic debris on the sea floor ?: Lethaia, v. 22, p. 207 -212.
Davis, M., Douglas, C., Calcote, R., Cole, K. L., Winkler, M. G., and Flakne, R., 2000, Holocene climate in the Western Great Lakes national parks and lakeshores: implications for future climate change: Conservation Biology, v. 14, no. 4, p. 968- 983.
DuCasse, O., and Carbonel, P., 1994, Recent Cytherideinae (Crustacea, Ostracoda) from Lake Tanganyika. Archaeocyprideis tuberculata n. gen. n. sp.: systematics, distribution, ecology: Revue de Micropaleontologie, v. 37, no. 2, p. 97 -112.
Finney, B. P., Gregory- Eaves, I., Sweetman, J., Douglas, M. S. V., and Smol, J. P., 2000, Impacts of climatic change and fishing on Pacific salmon abundance over the past 300 years: Science, v. 290, p. 795 -799.
Fürsich, F. T., and Flessa, K. W., 1987, Taphonomy of tidal flat molluscs in the northern Gulf of California: paleoenvironmental analysis despite the perils of preservation: Palaios, v. 2, p. 543 -559.
Gulliksen, B., and Derás, K. M., 1975, A diver -operated suction sampler for fauna on rocky bottoms: Oikos, v. 26, no. 2, p. 246 -249.
Johnson, T. C., 1984, Sedimentation in large lakes: Annual Review of Earth and Planetary Sciences, v. 12, p. 179 -204.
Kidwell, S. M., 2001, Preservation of species abundance in marine death assemblages: Science, v. 294, p. 1091 -1094.
Kidwell, S. M., 2001, Ecological fidelity of molluscan death assemblages, in Aller, R. C., ed., Organism- Sediment Interactions: Papers in Honour of Donald C. Rhoads, University of South Carolina Press.
Kidwell, S. M., and Bosence, D. W. J., 1991, Taphonomy and time -averaging of marine shelly faunas, in Allison, P. A., and Briggs, D. E. G., eds., Taphonomy: releasing the data locked in the fossil record: Topics in Geobiology: New York, Plenum, p. 116- 209.
Kidwell, S. M., and Flessa, K. W., 1995, The quality of the fossil record: populations species, and communities: Annual Review of Ecology and Systematics,v. 26, p. 269- 299. 162
Koch, C. F., 1987, Prediction of sample size effects on the measured temporal and geographic distribution patterns of species: Paleobiology, v. 13, no. 1, p. 100 -107.
Kowalewski, M., Avila Serrano, G. E., Flessa, K. W., and Goodfriend, G. A., 2000, Dead delta's former productivity: two trillion shells at the mouth of the Colorado River: Geology, v. 28, p. 1059 -1062.
14C0 Levin, I., and Kromer, B., 1997, Twenty years of atmospheric zobservations at Schauinsland station, Germany: Radiocarbon, v. 39, no. 2, p. 205 -218.
Levin, S. A., 1992, The problem of pattern and scale in ecology: Ecology, v. 73, no. 6, p. 1943 -1967.
MacPhee, R. D. E., 1999, Extinctions in Near Time: Causes, Contexts, and Consequences, in MacPhee, R. D. E., and Sues, H. -D., eds., Advances in Vertebrate Paleobiology: New York, Kluwer Academic.
Magurran, A. E., 1988, Ecological Diversity and Its Measurement: Princeton, Princeton University Press.
Martens, 1985, Tanganyikacypridopsis gen.n. (Crustacea, Ostracoda) from Lake Tanganyika: Zoologica Scripta, v. 14, no. 3, p. 221 -230.
Martin, R. E., and Liddell, W. D., 1988, Foraminiferal biofacies on a north coast fringing reef (1 -75 m), Discovery Bay, Jamaica: Palaios, v. 3, p. 298 -314.
Martin, R. E., Wehmiller, J. F., Harris, M. S., and Liddell, W. D., 1996, Comparative taphonomy of bivalves and foraminifera from Holocene tidal flat sediments, Bahia la Choya, Sonora, Mexico (Northern Gulf of California): taphonomic grades and temporal resolution: Paleobiology, v. 22, no. 1, p. 80 -90.
Michel, A. E., Cohen, A. S., West, K., Johnston, M. R., and Kat, P. W., 1992, Large African lakes as natural laboratories for evolution: examples from the endemic gastropod fauna of Lake Tanganyika: Mitt. Internat. Verein. Limnol., v. 23, p. 85 -99.
Miller, G. H., Magee, J. W., Johnson, B. J., Fogel, M. L., Spooner, N. A., McCulloch, M. T., and Ayliffe, L. K., 1999, Pleistocene extinction of Genyornis newtoni: human impact on Australian megafauna: Science, v. 283, p. 205 -208.
Minchin, P. R., 1987, An evaluation of the relative robustness of techniques for ecological ordination: Vegetatio, v. 69, p. 89 -107. 163
Nakai, K., Kawanabe, H., and Gashagaza, M. M., 1994, Ecological studies on the littoral cichlid communities of Lake Tanganyika: the coexistence of many endemic species, Pp. 373 -389 in Martens et al. 1994.
Nydal, R., and Lövseth, K., 1983, Tracing bomb 14C in the atmosphere 1962 -1980: Journal of Geophysical Research, v. 88, no. C6, p. 3621 -3642.
Pandolfi, J. M., 1996, Limited membership in Pleistocene reef coral assemblages from the Huon Peninsula, Papua New Guinea: constancy during global change: Paleobiology, v. 22, no. 2, p. 152 -176.
Park, L. E., and Martens, K., 2001, Four new species of Gomphocythere from Lake Tanganyika, East Africa (Crustacea, Ostracoda): Hydrobiologia, v. 450, p. 129 -147.
Pettitt, A. N., and Stephens, M. A., 1977, The Kolmogorov- Smirnov goodness -of -fit statistic with discrete and grouped data: Technometrics, v. 19, no. 2, p. 205 -210.
Pielou, E. C., 1984, The Interpretation of Ecological Data: New York, John Wiley & Sons, 263 p.
Rodriguez, C. A., Flessa, K. W., and Dettman, D. L., 2001, Effects of upstream diversion of Colorado River water on the estuarine bivalve mollusc Mulinia coloradoensis: Conservation Biology, v. 15, p. 249 -258.
Rome, D. R., 1962, Exploration Hydrobiologique du Lac Tanganyika: Brussels, Belgium, Institut Royal des Sciences Naturelles de Belgique, 305 p.
Rosenzweig, M. L., 1995, Species Diversity in Space and Time: New York, Cambridge University Press, 436 p.
Russell, M. P., 1991, Modern death assemblages and Pleistocene fossil assemblages in open coast high energy environments, San Nicolas Island, California: Palaios, v. 6, p. 179 -191.
Sall, J., and Lehman, A., 1996, JMP Start Statistics: A Guide to Statistics and Data Analysis Using JMP and JMP IN Software: Belmont, California, Duxbury Press, 521 p.
Simpson, G. G., 1960, Notes on the measurement of faunal resemblance: American Journal of Science, v. 258a, p. 300 -311.
Steadman, D. W., 1995, Prehistoric extinctions of Pacific Island birds: biodiversitymeets zooarchaeology: Science, v. 267,p. 1123 -1131. 164
Stuiver, M., Reimer, P. J., Bard, E., Beck, J. W., Burr, G. S., Hughen, K. A., Kromer, B., McCormac, F. G., van der Plicht, J., and Spurk, M., 1998a, INTCAL98 Radiocarbon age calibration 24,000 - 0 cal BP:Radiocarbon, v. 40, p. 1041 -1083.
Stuiver, M., Reimer, P. J., and Braziunas, T. F., 1998b, High- precision radiocarbon age calibration for terrestrial and marine samples.: Radiocarbon, v. 40, p. 1127 -1151. ter Braak, C. J. F., and Prentice, I. C., 1988, Atheory of gradient analysis: Advances in Ecological Research, v. 18, p. 271 -317. ter Braak, C. J. F., and Smilauer, P., 1998, CANOCO Reference Manual andUser's Guide to Canoco for Windows: Software for Canonical Community Ordination (version 4): Ithaca, NY, Microcomputer Power, 352 p.
Wells, T. M., Cohen, A. S., Park, L. E., Dettman, D. L., and McKee, B. A., 1999, Ostracode stratigraphy and paleoecology from surficial sediments of Lake Tanganyika, Africa: Journal of Paleolimnology, v. 22, p. 259 -276.
Wolfe, A. P., 1996, Spatial patterns of modern diatom distribution and multiple paleolimnological records from a small arctic lake on Baffin Island, Arctic Canada: Canadian Journal of Botany, v. 74, p. 435 -449.
Wouters, K., 1988, On Romecytheridea ampia Wouters sp. nov.: Stereo -Atlas of Ostracod Shells, v. 15, no. 23, p. 101 -106.
Wouters, K., and Martens, K., 1992, Contribution to the knowledge of Tanganyikan cytheraceans, with the description of Mesocyprideis nom. nov. ( Crustacea, Ostracoda): Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 62, p. 159 -166.
Wouters, K., and Martens, K., 1994, Contribution to the knowledge of the Cyprideis species flock (Crustacea : Ostracoda) of Lake Tanganyika, with the description of three new species: Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 64, p. 111 -128.
Wouters, K., and Martens, K., 1999, Four new species of the Cyprideis species flock ( Crustacea: Ostracoda) of Lake Tanganyika: Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 69, p. 67 -82.
Zar, J., 1984, Biostatistical Analysis: Englewood Cliffs, N.J, Prentice -Hall, 718 p. 165
APPENDIX C: PALEOENVIRONMENTAL RECORDS OF LANDSCAPE DISTURBANCE AT GOMBE STREAM NATIONAL PARK, AFRICA
Formatted for submission to Nature.
Simone R. Alin *t, Catherine M. O'Reilly * *, Andrew S. Cohen*, David L. Dettman *, Manuel R. Palacios- Fest §, & Brent A.McKeep
*Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA. Terra Nostra Earth Science Research, 3220 W. Ina Rd. #8105, Tucson, AZ 85721, USA. ('Departmentof Geology, Tulane University, New Orleans, LA 70118, USA. tPresent Address: Large Lakes Observatory, University of Minnesota, Duluth, 10 University Drive, 215 RLB, Duluth, MN 55812 USA. *Present Address: Department of Hydrology and Water Resources, University of Arizona, Tucson, AZ 85721, USA.
Abstract
Northwestern Tanzania harbors globally significant vertebrate and invertebrate biodiversity in forest habitat of Gombe Stream National Park (GSNP) and in Lake
Tanganyika (Coulter, 1994; Goodall, 1971). Rapid population growth and watershed deforestation in recent decades threaten the diversity and ecological integrity of terrestrial and lacustrine ecosystems there through habitat degradation and erosion (Alin et al., 1999; Cohen et al., 1993; Goodall, 1971). Here we present a decadally- resolved,
250 -year paleoenvironmental reconstruction of the history of land use change in the
GSNP area and the effects of these changes on the Lake Tanganyika ecosystem, based on sediment cores collected offshore from undisturbed watersheds within the park and a deforested watershed outside park boundaries. Geochronological, sedimentological, stable isotopic, and paleoecological data reveal dramatic recent increases in sedimentation rate, shifts in nitrogen stable isotope profiles, and transitions in 166 invertebrate assemblages offshore from the deforested watershed. The sedimentrecord reveals that past climate events have amplified the effects of slope denudation, triggering massive erosion. Future integrity of ecosystems near GSNP depends on our ability to predict and mitigate potentially devastating synergistic effects of land use and climate change in this area of critical interest for biological conservation. Furthermore, contrasting ecological histories of these watersheds emphasize the importance of protecting landscapes to conserve aquatic ecosystems.
Body of Paper
Many watersheds in the densely populated northern Lake Tanganyika basin are vulnerable to extensive erosion and sediment deposition in the lake as steep watersheds bordering the lake are deforested and converted to agriculture. Northwestern Tanzania, encompassing two Gombe Stream National Park (GSNP) and Mwamgongo (MWA) watersheds is characterized by steep, rift escarpment topography. The watersheds are comparable in area (8.5 km2 for MWA, 2.6 km2 each for GSNP watersheds), maximum elevation (1400 -1460 m), rainfall (- 1600 mm/yr), and bedrock geology (middle
Proterozoic metasediments). However, MWA watershed currently has a human population density of -800 people km-2, whereas the population density of GSNP is <5 people km -2 (Cohen et al., 1999). All watersheds experienced similar land use and population densities until 1943 AD, when Gombe Stream became a game reserve to protect chimpanzees and their habitat, and local inhabitants were relocated to 167
Mwamgongo village (Bygott, 1992; Cohen et al., 1999). Another pulse of population growth occurred at MWA in the early 1970s when the Tanzanian government consolidated rural populations into existing villages to improve infrastructure. Since
Gombe Stream achieved national park status in 1968, park inhabitants have consisted of park staff and researchers affiliated with the Jane Goodall Institute.
Sedimentation rates offshore from watersheds in Lake Tanganyika reflect rates of watershed soil erosion consistent with natural and anthropogenic watershed processes.
Deepwater sedimentation rates offshore from GSNP were essentially constant from
1700 to 1998 AD (GSNP -D sed. rt. =1.6 mm yr', r 2 = 0.98) (Table 2, Fig. la). In contrast, sedimentation rates increased substantially offshore from MWA during the past
150 years. From the base of the core (-1450 AD) to 22 cm depth (- 1865 AD), the sedimentation rate was on the order of 0.5 mm yr1 (r 2 = 0.89). Between 1865 and 1965
AD, sedimentation rates increased to 1.4 mm yr' (r 2= 0.96), with an additional increase to 2.5 mm yr' since 1965 AD at MWA (r2= 0.99). These data suggest a five -fold increase in sediment deposition rate offshore from the deforested watershed not seen offshore from the national park.
Nearshore sedimentation is prone to more severe impacts of watershed disturbance as a result of proximity to sediment and waste sources. The shallow -water sedimentation rate in the upper part of core GSNP -S was 2.5 -4 mm yr' (Tables 1 & 2, Fig. 2a). The three bottom14Cages in GSNP -S give essentially identical calibrated dates, such that sedimentation in the interval - 12 -17cm may representan event bed, with near instantaneous deposition of >5 cm of sediment around either 1770AD or the early 1930s. 168
The abrupt decrease in sediment grain size immediately below the lowermost date along with the identical age scenarios suggests event bed deposition (Fig. 2b). Alternately, this section of core may represent - 100 years of deposition (1667 -1770 AD), giving a sedimentation rate of 0.4 mm yr', which is low for a nearshore, shallow -water environment such as this. In a third, low- probability scenario, lower GSNP -S dates may all have been deposited in the early 1900s (1914 -50 AD), though not as an event bed, with a sedimentation rate of -1.1 mmyr'.Because of the uncertainties associated with the age model for GSNP -S, it is difficult to assesswhether a change in sedimentation rate occurred in the upper core. Shallow -water cores from MWA suggest a steadily increasing sedimentation rate at this site, from -'0.6 mm yr' in the 1970s and 1980s to 4-
12 mm yr' during the 1990s (quadratic fit, r 2 = 0.85). In contrast, sedimentation rates in the lower portion of the core were probably much lower, estimated at -0.2 mmyr"'
(Table 2, Fig. 2a). However, the 2a age ranges are quite long; deposition may have occurred during a much shorter interval at higher rates. Thus, the radiocarbon data are suggestive, but inconclusive, about a substantial increase in sedimentation rate between the 17th and 20th centuries in this core, due to the constraints of calibrating radiocarbon ages from the past 400 years.
Concentration of sedimentary organic matter (SOM) in sediment cores reflects changes in input from primary productivity stemming from either lake or watershed processes. SOM in core GSNP -D fluctuated very little through time (Fig. lb). In contrast, input of SOM to core MWA -D increased rapidly during the period when relocation from GSNP doubled the population size of MWA watershed. Peak SOM 169 concentrations in MWA -D coincided with anomalously high rainfall in East Africa in
1961 -62 AD (Nicholson, 1999), which suggests that watershed soils were rendered vulnerable to erosion during particularly wet years by deforestation.
Soil and waste sources of nitrogen are characterized by higher stable isotope ratios
(815N) than lacustrine primary producers in Lake Tanganyika (Macko and Ostrom, 1994;
Talbot, in press), and changing nutrient sources are reflected in sediment cores. The b
15Nrecord in GSNP -D shows that an enrichment in 815N values occurred between 1840 and 1900 AD (Fig. 1d), after which point they decreased slightly and hovered around a new mean (average 1763 -1826 AD: 0.4±0.2 %o; 1844 -1970 AD: 1.1±0.3 %0). Enrichment of
815N values suggests increased importance of terrestrial nitrogensources to the sediment.
The 815N curve for MWA -D does not show the small mid -1800s shift toward enriched b
' 5N values, buta dramatic enrichment in815N began around 1950 AD and peaked simultaneously with SOM in the late 1960s. Together, the SOM and 815N records suggest that the 1961 -62 AD high rainfall event in East Africa triggered massive erosion of soil to the lake. C:N ratios should also reflect this impulse of terrestrial nitrogen to the lake. However, because algal organic matter (OM) contains a higher proportional abundance of nitrogen than terrestrial OM, any increase in lacustrine productivity stimulated by the increased influx of nutrients may mask the terrestrial C:N signature associated with such a large pulse of sediment.
Deforestation -related sediment inundation alters habitat structure and productivity of lake habitats, thereby affecting fish and invertebrate species richness and abundance
(Alin et al., 1999; Cohen et al., 1993; O'Reilly, 1999).Effects of landscape disturbance 170 on species composition of benthic communities were manifested in paleoecological records from the deforested site. Marked faunal transitions in ostracod assemblages at
MWA were contemporaneous with indications of environmental change. In deepwater cores, contrasts between sites were most apparent in ordination plots (Fig. 1g). MWA -D assemblages form a cluster, representing deposition prior to -1930 AD, and a more diffuse group of the five samples deposited since1930 AD, which do not overlap with earlier samples in ordination space. Among the five most recent core samples, MWA -D ostracod assemblages show a trend toward more positive DCA axis 2 values, indicating a progressive change in species composition. Core GSNP -D ostracod assemblages did not manifest either of these patterns.
Proximity to shoreline disturbance renders shallow -water faunas especially susceptible to anthropogenic erosion. The ostracod fauna in MWA -S 10 underwent a profound transition in its dominant species between 16th-17th and 20th centurycore intervals (Fig. 2c). At the bottom of core MWA -S 10, two species of Gomphocythere were numerically dominant. They declined in abundance midcore, and three other species replaced them as dominants. The timing of the transition in deepwater ostracod assemblages suggests a mid -20th century turnover. No comparable shift in dominant species occurred in GSNP -S. In DCA plots of shallow -water core samples, GSNP -S assemblages appear to be distributed randomly through ordination space with respect to time of deposition, with the exception of the bottommost few samples representing a higher lake level fauna. In contrast, the most recent assemblages in cores MWA-S10 and
MWA -S5 were increasingly divergent from older samples. 171
Collectively, these paleoenvironmental and paleoecological records from the Gombe
Stream National Park area illustrate recent history of land use change, climate events, and their combined influence on the ecology of Lake Tanganyika. Paleoenvironmental indicators support an interpretation of greatly increased soil erosion in the deforested
Mwamgongo watershed. Despite the physical proximity of the two sites, the benthic environment offshore from GSNP appears to have been largely buffered from the dramatic changes observed offshore from MWA. The signature of natural climate drought and flood impacts is evident in the pre -anthropogenic sedimentary charcoal and siliciclastic records at both sites. However, population growth and land use change render landscapes, and the lake habitat in turn, especially vulnerable to dramatic change when natural environmental events, such as the 1961 -62 AD East Africa rainfall event, trigger them. The challenge to the future sustainability of the Lake Tanganyika ecosystem and its watersheds lies in our ability to predict and mitigate the coupledeffects of direct alterations to the lake's catchment, such as deforestation, with indirect anthropogenic effects, such as global climate change. Paleoenvironmental contrasts between protected and deforested watersheds further illustrates the vital role of protecting terrestrial landscapes in the preservation of aquatic ecosystems.
Acknowledgments
We thank the Nyanza Project, UN /GEF Lake Tanganyika Biodiversity Project,
Tanzania National Fisheries (Kigoma), Gombe Stream National Park staff, Jane Goodall 172
Institute researchers, and crews of the RN Explorer and the R/V Echo for logistical and field assistance. We are grateful to the Tanzanian Commission for Science and
Technology, Tanzania National Parks, and Tanzanian Immigration Agency for permission to conduct research at Lake Tanganyika and Gombe Stream National Park.
We thank O.K. Davis for assistance identifying radiocarbon samples. This work was supported by the National Science Foundation, UN /GEF Lake Tanganyika Biodiversity
Project, National Security Education Program, University of Arizona, Geological Society of America, and Sigma Xi. In addition, we thank R. Robichaux for detailed comments on previous versions of the manuscript. 173
Figure C.1. Geochronology, paleoenvironmental, and faunal data for deepwater cores.
Sediment cores were collected moderately deep (75 -100 m) water offshore from
Mwamgongo and Gombe Stream National Park (Table 1). Data from GSNP -D in triangles and dashed lines, data from MWA -D in squares and solid lines. a, 210Pb (half - life 22.3 years) data were collected at 1 cm intervals using the polonium method (McKee et al., 1983; Nittrouer et al., 1979) to depths where supported210Pb levelswere reached.
Sedimentation rates were determined according to an advection- diffusion model based on the slope and intercept of a linear regression of the excess 210Pb data (Guinasso and
Schink, 1975). Filled symbols and white regression line accompany post -1965 AD data, open symbols and black line the 1865 -1965 AD regression for Mwamgongo. The slope of the post -1965 AD regression line is significantly higher than the 1865 -1965 AD regression slope (p= 0.0339).14C dates augmented 210Pb chronologies in the lowercores. Midpoints of the most probable 26 age range were used for all 14C dates except where noted. b,
Sedimentary organic matter from loss -on- ignition (Bengtsson and Enell, 1986). c,815N ratios were determined using an Isochrom Continuous Flow Stable Isotope Mass
Spectrometer coupled to the elemental analyzer and are reported in the delta notation relative to atmospheric nitrogen. d, Detrended correspondence analysis (DCA, detrended by 2nd order polynomials (Minchin, 1987; ter Braak and Prentice, 1988)) performed on ostracod species abundance data using CANOCO 4 (ter Braak and Smilauer, 1998), with rare speciesdownweighted. Despite downweightingrare species, the distribution of outlying samples was determined by rare species in samples, with species occurring in only a few samples loading heavily on the primary DCAaxes. Filled symbols represent 174 samples deposited since 1930 AD, open samples pre -1930 AD. Bold numbers next to filled symbols indicate core sample number, with 1 corresponding to the core top. 0 1997Year (AD)Calendar 2000 0.7 20 10.- -1864-1897-1930-1964 1950- lA 0 '1977 1997 1900- A -I CitO , LIl 20 10,- -1879-1915-1951 3A A O 5 -0.7 0.1 a Excess 210Pb (dpm/g) 1 10 100 (%b SOM LOI) 20 0 c 815N MO 3 -0.7 d Ostracod assemblages DCA Axis 1 4 0.7 176
Figure C.2. Sedimentology and paleoecology of shallow -water cores, collected in 5 -15 m water offshore from Gombe Stream National Park and Mwamgongo. a, Chronologies for shallow -water cores were based solely on radiocarbon dates on terrestrial leaf samples. Midpoints of 26 radiocarbon age ranges are listed in order of decreasing probability when multiple dates with relatively high probabilities exist (Stuiver et al.,
1998a; Stuiver et al., 1998b). b, Sedimentation rates for dated shallow -water cores.
Because both cores appeared to have mid -core depositional hiatuses, data from upper and lower core age series were regressed separately to determine sedimentation rates. Dashed boundaries between regression zones indicate uncertainty in boundary location. e, Grain size distributions for shallow -water cores. Asterisk indicates sample containing the ostracod Gomphocythere downingi. d, Relative species abundance profiles for five of the ten most abundant ostracod species at both sites. Five hundred individuals were identified to the species level (Park and Martens, 2001; Rome, 1962; Wouters and
Martens, 1999) for all 1 -cm core samples. Eight of the ten most abundant species are shared by cores GSNP -S and MWA -S 10. Species displayed here are ranked by core- wide average abundance, from left to right, 2, 3, 9, 10, and 1 for GSNP -S and 2, 3, 7, 5, and 1 for MWA -S 10. NationalGombe(GSNP Stream -S)Park 1986 2.5 -4.0 1 1769 or 1932 1971 1 Mwamgongo 1769, 1667, or 1934 1770 or 1934 t 18 0 (MWA-SGrain 10)size key: 198719921997 4-120.6 y a 63-106 >1mm106 µm-1 mm µm 155516481971 0.2 1 .: < 63 µm a Calendar year (AD) b Sedimentation rate (nu /yr) 1 15 c Grain0 size (% wt.) 100 0 d Dominant ostracod species 30 0 (% abundance) 20 0 10 0 15 0 50 Table C.1. Core details. Water Core Depositional Latitude Longitude GSNP-DGombeCore* Stream (LT98-58M) National Park, Tanzania depth (m) length (cm) 76 42 interval (yrs) 250 4°41'18" (S) 29°37'00" (E) MWA-DMwamgongo,GSNP-S (LT98-37M) (MIT-1) Tanzania 9515 4519 100-250t 500 4°37'21.6"29°37'39.6"4°36'04" 29°37'57" MWA-S5MWA-S 10 (MWA-2) (MWA-1) 105 1116 450t-r30 4°36'34" 29°38'34" usingcores).and* Core shallowa OfficialHedrick monikers -water -Marrsnames core of (watercores are depth appended to distinguish between are a combination of site abbreviation Multicorer. Coordinates established by GPS. Shallow -water in parentheses. Deepwater cores were collected with -D or -S indicating deep MWA -S thesecorestand DepositionA depositional12coreswere cm. collectedwere may estimated hiatus be by sporadic, using a map. of -300scuba years divers exists using at aapproximately hand -coring apparatus. 9 cm. Coordinates for with a possible hiatus of -100-200 between 9 TablenumberSample C.2. Radiocarbon dates for Gombe Stream National Park and Mwamgongo cores*. Material dated core Depth(cm) modernin Fraction 14C age14C Calendar (2a range) year AD probability curve Area under AA-41870AA-38064AA-38063Core MWA-1 singlesingle leaf leaf 7-84-52-3 1.12541.17291.0934 post-bomb 198719921997 AA-3AA-41871 8065 single leaf 10-11 8-9 1.5133 .9689 post-bomb254 ± 39 178015581648 (1755-1805)(1516-1599)(1616-1679) 1971-2 .188. 290478 149617471941 (1494-1497)(1739-1754)(1935-1947) 1.000 .003..025015 CoreAA-38066 MIT-1 single leaf 12-13 2-3 1.1889 .9590 336 ± 41 1555 (1466-1644) 1985-6 AA-41874AA-41873AA-41872 singleleaf fragments leaf 8-9 12-13 1.5255 .9782 post-bomb177 ± 35 1769 (1722-1815) 1970-1 ..542 1678 (1654-1702)t1932 (1914-1950) .174. 088197 AA-43719 single leaf 13-14 .9660 278 ± 75 1568185717701856 (1442-1694)t1934 (1849-1864)(1726-1813)(1830-1881) (1918-1949) .008.159..041792 AA -41875 single leaf 16 -17 .9749 204 ± 35 1769 (1726 -1812) .293.566 Core LT98 -37M 185619341667 (1853(1919(1640-1693) -1858)-1949) .003. 138 AA32724AA -32720-30560 leafleafsingle fragments fragments leaf 43 30 -44 -31 3 -4 1.1450 .9469. 9616 post -bomb440315 ± ±5560 1557 (1450 -1663) 1990 -1 1.000 .772 Core LT98 -58M 133615901465 (1334 (1546(1400 -1337) -1634)-1530) .004. 224 AA -30561 1.0343 . post -bomb AA -32719 leaf fragmentsfragments 18 2 -3-19 9647 290 ± 70t 17701563 (1734(1444 -1806)-1682) early 1950s? ..870 109 AA -32728 leaf fragments 37 -38 .9744 210 ± 60 193017591939 (1908(1626(1931 -1951)-1891)-1947) ..801021129 variable* All radiocarbon radiocarbon dates reservoir age of Lake Tanganyika's surface waters. Radiocarbon were established on terrestrial plant material to avoid the substantial and 1551 (1521 -1580) ages were .069 theagesdatesmeasured basis towere that at assignedbased the the Arizona studyon by single referencesites AMS leaf are Laboratory fragments,essentially to post -bomb and soequatorial thewere atmospheric atmospheric calibrated ( <5° S).decay using decayAs curves26'27.the CALIB curve Intertropical allows 4.324'25.within Most assignment Convergence posta Post few -bomb -bombyears of of production. The Southern Hemisphere correction was not applied, on effectivelyZonestratigraphicallyt These passes dates Northern over have thedisordered been hemispheresites eliminatedtwice or conflicteach atmosphere fromyear, with considerationplant as in samples a on are as likely to have grown in an 21°Pb data for the Southern one. the basis that they either are same stratigraphic level. 182
References
Alin, S. R., Cohen, A. S., Bills, R., Gashagaza, M. M., Michel, E., Tiercelin, J. J., Martens, K., Coveliers, P., Mboko, S. K., West, K., Soreghan, M., Kimbadi, S., and Ntakimazi, G., 1999, Effects of landscape disturbance on animal communities in Lake Tanganyika, East Africa: Conservation Biology, v. 13, no. 5, p. 1017 -1033.
Bengtsson, L., and Enell, M., 1986, Chemical analysis, in Berglund, B. E., ed., Handbook of Holocene Palaeoecology and Palaeohydrology: Chichester, UK, John Wiley and Sons, p. 423 -450.
Bygott, D., 1992, Gombe Stream National Park: Dar es Salaam, Tanzania National Parks, 71 p.
Cohen, A. S., Bills, R., Cocquyt, C. Z., and Caljon, A. G., 1993, The impact of sediment pollution on biodiversity in Lake Tanganyika: Conservation Biology, v. 7, p. 667- 677.
Cohen, A. S., Palacios -Fest, M. R., Dettman, D., Msaky, E., Livingstone, D., and McKee, B., 1999, Lake Tanganyika Biodiversity Project Special Study on Sediment Discharge and its Consequences: Paleolimnological Investigations, Lake Tanganyika Biodiversity Project.
Coulter, G. W., 1994, Lake Tanganyika: Ergebnisse der Limnologie, v. 44, p. 13 -18.
Goodall, J., 1971, In the Shadow of Man: Boston, Houghton Mifflin.
Guinasso, N. L., and Schink, D. R., 1975, Quantitative estimates of biological mixing rates in abyssal sediment: Journal of Geophysical Research, v. 80, no. 3032 -3043.
Macko, S. A., and Ostrom, N. E., 1994, Pollution studies using stable isotopes, in Lajtha, K., and Michener, R. H., eds., Stable Isotopes in Ecology and Environmental Science: Oxford, Blackwell Scientific Publications.
McKee, B. A., Nittrouer, C. A., and DeMaster, D. J., 1983, The concepts of sediment deposition and accumulation applied to the continental shelf near the mouth of the Yangtze River: Geology, v. 11, p. 631 -633.
Minchin, P. R., 1987, An evaluation of the relative robustness of techniques for ecological ordination: Vegetatio, v. 69, p. 89 -107.
Nicholson, S. E., 1999, Historical and modern fluctuations of lakes Tanganyika and Rukwa and their relationship to rainfall variability: Climatic Change, v. 41, p. 53 -71. 183
Nittrouer, C. A., Sternberg, R. W., Carpenter, R., and T, B. J., 1979, The use of 210Pb geochronology as a sedimentological tool: applications to the Washington continental shelf: Marine Geology, v. 31, p. 297 -316.
O'Reilly, C. M., 1999, Benthic algal productivity in Lake Tanganyika and the effects of deforestation: UN /GEF Lake Tanganyika Biodiversity Project (RAF /92/G32).
Park, L. E., and Martens, K., 2001, Four new species of Gomphocythere from Lake Tanganyika, East Africa (Crustacea, Ostracoda): Hydrobiologia, v. 450, p. 129 -147.
Rome, D. R., 1962, Exploration Hydrobiologique du Lac Tanganyika: Brussels, Belgium, Institut Royal des Sciences Naturelles de Belgique, 305 p.
Stuiver, M., Reimer, P. J., Bard, E., Beck, J. W., Burr, G. S., Hughen, K. A., Kromer, B., McCormac, F. G., van der Plicht, J., and Spurk, M., 1998a, INTCAL98 Radiocarbon age calibration 24,000 - 0 cal BP: Radiocarbon, v. 40, p. 1041 -1083.
Stuiver, M., Reimer, P. J., and Braziunas, T. F., 1998b, High- precision radiocarbon age calibration for terrestrial and marine samples.: Radiocarbon, v. 40, p. 1127 -1151.
Talbot, M. R., in press, Nitrogen isotopes in palaeolimnology, in Last, W., and Smol, J., eds., Tracking Environmental Change Using Lake Sediments: Physical and Chemical Techniques, Kluwer Press. ter Braak, C. J. F., and Prentice, I. C., 1988, A theory of gradient analysis: Advances in Ecological Research, v. 18, p. 271 -317. ter Braak, C. J. F., and Smilauer, P., 1998, CANOCO Reference Manual and User's Guide to Canoco for Windows: Software for Canonical Community Ordination (version 4): Ithaca, NY, Microcomputer Power, 352 p.
Wouters, K., and Martens, K., 1999, Four new species of the Cyprideis species flock (Crustacea: Ostracoda) of Lake Tanganyika: Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 69, p. 67 -82. 184
Appendix D: List of abbreviations used for species names in all ostracod species abundance data matrices. Listed alphabetically by species abbreviation, except where species is conferred to another species, when they are listed together.
Abbreviation Full name Al ab Allocypria aberrans Al cf abAllocypria cf. aberrans Al cl Allocypria claviformis (striated) , Al hu Allocypria cf. humilis Al in Allocypria inclinata Al cf in Allocypria cf. inclinata Al mu Allocypria mucronata Al 5 Allocypria n.sp. 5 Al 10 Allocypria n.sp. 10 Al cf 10Allocypria cf. n.sp. 10 Al 11 Allocypria n.sp. 11 Al 17 Allocypria n.sp. 17 Al 18 Allocypria n.sp. 18 Al 20 Allocypria n.sp. 20 Ar tu Archaeocyprideis tuberculata Ar 2 Archaeocyprideis n.sp. 2 Ar 10 Archaeocyprideis n.sp. 10 Cd ca Cyprideis caljoni Cd ma Cyprideis mastai Cd 1 Cyprideis n.sp. 1 Cd 3 Cyprideis n.sp. 3 Cd 18 Cyprideis n.sp. 18 Cd 23 Cyprideis n.sp. 23
Cd 24 Cyprideis n.sp. 24 , Cd 25 Cyprideis n.sp. 25 Cn de Candonopsis depressa Cn cf deCandonopsis cf. depressa Cn 2 Candonopsis n.sp. 2 Cn 7 Candonopsis n.sp. 7 Cn 8 Candonopsis n.sp. 8 Cn 9 Candonopsis n.sp. 9 Cn 12 Candonopsis n.sp. 12 Cn 15 Candonopsis n.sp. 15 Cn 16 Candonopsis n.sp. 16 Cp bi Cypridopsis bidentata Cp co Cyprid gpsis colorata Cp la Cypridopsis lacustris Cp cf la Cypridopsis cf. lacustris . Cp ob Cypridopsis obliquata Cp se ,Cypridopsis serrata Cp 5 Cypridopsis n.sp. 5 185
Abbreviation Full name
C 6A C rido /sis n. s.6A Cp 6B Cypridopsis n.sp. 6B Cp 6C Cypridopsis n.sp. 6C Cp 15 Cypridopsis n.sp. 15 Cp 17 Cypridopsis n.sp. 17 Cp 18 .Cypridopsisn.sp. 18 Cp 21 Cypridopsis n.sp. 21 Cp 22 Cypridopsis n.sp. 22 Cp 23 Cypridopsis n.sp. 23 Cp 25 Cypridopsis n.sp. 25 Cp sp Cypridopsis n.sp. Da st Darwinula stevensoni cf El cf. Elpidium Go al Gomphocythere alata Go co Gomphocythere coheni Go cr Gomphocythere cristata Go cu Gomphocythere curta Go do Gomphocythere downingi Go wi Gomphocythere wilsoni Go wo Gomphocythere woutersi Go 11 Gomphocythere n.sp. 11 Ka br Kavalacythereis braconensis Ka hy Kavalacythereis hystrix Li 8 Limnocythere n.sp. 8 Mc cd Mecynocypria cf. declivis Mc cm Mecynocypria complanata Mc cf cmMecynocypria cf. complanata Mc cn Mecynocypria cf. conoidea Mc dc Mecynocypria declivis Mc df Mecynocypria deflexa Mc em Mecynocypria emaciata Mc ob Mecynocypria obtusa (cf. n.sp. 13) Mc op Mecynocypria opaca Mc pa Mecynocypria cf. parvula Mc su Mecynocypria subangulata Mc cf suMecynocypria cf. subangulata Mc 9 Mecynocypria n.sp. 9 Mc cf 11Mecynocypria cf. n.sp. 11 Mc 17 Mecynocypria n.sp. 17 Mc 19 Mecynocypria n.sp. 19 Mc 20 Mecynocypria n.sp. 20 Mc cf 20Mecynocypria cf. n.sp. 20 Mc 21 Mecynocypria n.sp. 21 Mc 22 Mecynocypria n.sp. 22 Mc 29 Mecynocypria n.sp. 29 Mc 30 Mecynocypria n.sp. 30 Mc 31 Mecynocypria n.sp. 31 186
Abbreviation Full name Mc cf 32Mecynocypria cf. n.sp. 32 Mc 33 Mecynocypria n.sp. 33 Mc cf 34Mecynocypria cf. n.sp. 34 Mc 36 Mecynocypria n.sp. 36 Mc 37 Mecynocypria n.sp. 37 Mc 39 Mecynocypria n.sp. 39 Mc 40 Mecynocypria n.sp. 40 Ms ir Mesocyprideis irsacae Ms pi Mesocyprideis pila Ms 2B Mesocyprideis n.sp. 2B Ms 4 Mesocyprideis n.sp. 4 Ms 9 Mesocyprideis n.sp. 9 Pr 1 .Proparacytheridea n.sp. 1 Ro am Romecytheridea ampia Ro lo Romecytheridea longior Ro te Romecytheridea tenuisculpta Tc ma Tanganyikacypris matthesi Tc 1 Tanganyikacypris n.sp. 1 Tp ac Tanganyikacypridopsis acanthodes Tp ca Tanganyikacypridopsis calcarata Tp de Tanganyikacypridopsis depressa Tp Tanganyikacypridopsis n.sp. 1 Tp 3 Tanganyikacypridopsis n.sp. 3 Tp 4 Tanganyikacypridopsis n.sp. 4 Tp 5 Tanganyikacypridopsis n.sp. 5 Tp 6 Tanganyikacypridopsis n.sp. 6 Tp 8 Tanganyikacypridopsis n.sp. 8 Tt bu Tanganyikacythere burtonensis Appendix E: Raw data from core LT97 -56V collected from 56 m water depth near Mgondozi, Tanzania. Speciesdepth abundance in Al Al data Al Al for fossil assemblages. See Appendix D for a list of species abbreviations. Al Al Al Al Al Ar Ar Ar Cd Cd Cd Cd Cd Cn Cn Cn Cn Cn Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp Da Go Go core (cm) cl hu in cf in mu 50-1 0 5 0 3 0 1 1 10 11 200 tu 0 2 1 0 2 0 1 10 ma 30 0 0 18 24 25 de0 15 5 0 2 5 7 0 8 0 15 0 la cf la se0 6A 6C 15 18 21 22 23 25 0 0 5 0 0 2 7 0 34 2 st 0 al 3 co 7 4-5 0 3 _ 0 30 0 0 0 4 0 0 0 0 13 58 0 0 0 0 01 0 0 1 0 9 3 0 0 71 0 26 0 0 3 6 , 16-1712-13 8-9 0 12 54 00 0 06 0 0 02 0 1 43 0 0 0 1 0 1 0 1410 8 0 0 0 20 00 00 20 0 32 48 0 0 2 0 1614 2 021 02 694 34 30-3122-23 0 0 0 00 50 0 1 0 0 21 2 0 0 0 1 0 0 1413 18 3 40 0 0 00 01 0 0 0 0 291210 4`37 15 0 04 1618 2 0 83 0 01 8 113 2 38-39 0 4 0 2 0 0 0 0 0 41 0 0 0 1 0 0 5 23 0 0 0 0 0 0 0 0 12 0 0 9 01 0 0 , 6 4 54-5546-47 0 75 0 0 1 20 03 0 01 0 0 0 0 0 1 0 0 16 48 0 - 0 0 0 0 0 0 1 0 82 43 0 0 13 5 0 11 2 40 0 12 14 9 7 1 70-7162-63 0 0 0 0 00 01 0 00 0 01 0 0 01 0 0 12 33 06 91 0 0 0 0 0 1 0 0 0 0 01 0 0 0 0 00 0 01 02 1 0 86-8778-79 0 4 0 01 0 2 0 0 0 1 0 0 0 0 0 0 1933 2899 01 0 0 0 0 0 0 0 114 3 41 0 0 16 5 0 12 6 01 01 3 20 2 94-95 0 87 0 _ 1 0 3 0 0 23 , 0 , 0 0 0 0 12 13 _ 0 0 0 0 0 1 01 1 1011 6 0 0 111 14 0 12 0 26 5 102-103 . 0 1 0 0 1 0 0 0 0 0 0 01 0 14 41 _ , . 0 0 0 . 0 0 , 1 . 0 0 16 0 01 110-111 6 0 2 4 0 , 0 , 0 0 0 0 0 0 14 24 0 0 0 0 2 2 0 15 10 , 0 0 124 0 17 01 5 4 126-127118-119 0 5 1 0 041 0 . 0 0 0 0 0 1 0 01 80 . 0 0 1822 1119 _ . Ò0 0 0 0 0 _ 0 0 0 13 1 04 0 0 . 11 0 0 21 0 01 0 08 461 134-135 0 . 9 0 41 0 1 0 0 20 0 , 0 2 0 0 22 25 5 _ 8 2 0 20 0 0 0 01 20 14 7 4 0 _ 0 6 0 16 . 0 14 5 1 4 0 0 , 0 142-143 0 4 0 0 0 0 0 0 0 1 , 0 , 0 1 0 0 . 4 0 0 10 0 15 0 0 150-151 0 8 0 0 0 31 . 0 0 0 30 _ 0 0 1 0 0 26 20 0 0 . 0 , 0 0 . 32 0 . 6 . 3 0 0 8 0 , 6 0 , 2 13 0 1 158-159 0 12 , 0 5 0 5 . 0 0 01 1 . 0 . 0 , 0 58 11 2 . 0 0 0 0 , 0 0 1 0 11 7 0 0 0 16 8 0 10 9 01 4 6 1 166-167 0 11 0 2 0 . 0 0 . 0 0 0 0 . 0 0 0 0 0 0 0 0 0 0 0 0 0 4 178-179174-175 00 35 0 - 56 - 00 0 1 0 - 0 20 0 0 - 0 - 02 0 1 02 10 - 1 74 _ 0 - 0 0 - 0 0 - 0 -_ 0 - 0 - 8 - 0 - 0 - 0 20 0 1216 - 28 1 120 2 58 Go Go Go Go Go Go Ka Ka Li Mc Mc Mc Mc Mc Mc Mc Mc Mc Mc Mc Mc Mc Ms Ms Ms Ms Ms Pr Ro Ro core (cm)depth in , cr cu do wi wo 7 11 br 3 by 8 7 cn em ob op pa 12 9 2 20 9 21 29 7 30 2 s 4 36 1 10937 40 0 ir 4 pi14 2B 17 4 9 0 1 0 am 83 lo 35 4-50-1 . 0 1 0 1 34 0 2 00 3 0 4 0 1 2926 00 4227 3 4 1 11 0 4 1 2 0 139 80 34 10 33 00 0 1 48 78 12-13 8-9 0 1 4 1 32 0 15 7 00 30 0 1 473 0 1 5115 0 2428 30 0 13 2 03 10 0 11 5 1 307 0 161 69 0 7 1 11 6 22 7 01 0 1 0 407815 147 44 22-2316-17 0 1 5 0 32 8 00 13 7 1 0 11 2 1 18 0 2146 24 21 11 1 0 52 0 58 01 8789 02 0 12 74 1010 8 0 0 0 4135 143166 38-3930-31 0 6 1 41 0 11 4 0 21 0 . 1 0 1 6313 00 4438 4 1 . 5 0 4 0 4 0 114116 0 0 1 16 6 0 1 0 21 28 142 64 46-47 0 1 11 2 1 33 0 4 0 2 4 5 0 62 2 , 0 4 0 11 3 2 0 0 77 0 4 22 28 0 0 1 65 54 62-6354-55 0 0 41 0 10 4 0 11 0 0 07 0 1 40 01 00 35 27 0 002 06 0 0 04 040 0 124 23 0 0 061 14 5 1 0 0 0 33 03 64 65 70-7186-8778-79 20 063 14 60 01 25 50 0 10 70 0 0 0 41 5 0 4116 302 02 765 0 10 0 50 1 11 60 0 118 54 03 0 1 381 441017 0 0 0 3649 9072 3 7 9 91 4 6 0 79 , 4 0 0 0 24 97 9 0 29 0 , , 0 0 , 0 0 102-10394-95 0 1 31 . 0 5 . 0 11 0 6 1 37 0 1116 2 31 28 0 47 4 1 35 0 71 1 2 4 1 17 8 . 0 0 01 2016 169154 110-111 0 4 0 7 . 0 . 6 0 76 0 1 . 4113 , 0 , 2 . 1 1 . 0 3 . 0 70 0 0 1 14 18 0 0 175 , 4 65 0 10 6 , 0 0 0 42 2 . 0 0 118-119 , 0 4 0 0 0 , 1 1 20 1 . 1 126-127 01 0 1 43 0 3 . 0 80 0 05 . 0 3 0 3 0 0 2 0 0 03 . 5 0 11 0 47 11 4 0 0 30 . 4 138 15 2 , 6 9 0 0 0 0 , 60 6 0 0 46 134-135 , 0 ,. 0 0 0 47 0 , _ 142-143 0 6 5 0 11 8 0 . 13 0 3 . 0 1 . 20 0 11 0 _ 31 8 - 0 4 0 0 . 0 73 12 6 56 12 5 12 0 - 0 0 50 181 . 3 159 158-159150-151 06 . 3 1 042 0 1 0 671 0 0 1 1 187 7725 0 76 01 31 10 5 01 473 1 20 1 0 2543 3410 0 41 3418 3 0 0 0 602819 127193 178-179174-175166-167 0 86 31 00 17 80 0 1 21 0 2 1 0 1 . 164 22 . 0 2219 20 43 45 0 31 93 72 0 4462 27 2 04 08 5 1 0 0 0 24 5 222112 189
000-40000m000^-1^400 0^100-4000-40 C C C C CC C C C 0--. C C C C Ea 00 0 00 0 -+000 000 O0000^IOOOOOOONO^-i0000000000 OOOOOOOOOOOOOOOOOOOOOOOOO HA.. ba.) "cN00NknO-4knNOcnOONNMd-NNON00--i--1M[- ß, c t s C"CDCDCOOOOOOOOOOOOOOOOOOOON E--1 U
- 4ONN---4 -+ -+ -+ .--+ Ñr-, O-1N00000 OO -1-4 OOO OOOOOO00^00000000-40000'00-40 ! (I) 000 0000C00000000000-'CD 00000 (240 - E M--I ONr--knM^IOh[knO .~ MNM.--1 ONNknM-"CTNinOr--{ --1NM InknONN ,E .--.1 l¡)O\.-1 ....1N CM M.7t-Le)O 0001-r.--ir..1.1.--1t...-1 .--1 .---1 --i....1 N1 1 [[ ß. O4oo .O(-1Oo0 D4NO00,O4NO00.D4NOo0 400 .--rNMM'VO[[00ONO^-+r-,N b O .-1 .-4 .--1 --1 M--1.-.4.-4.-1,--I--1 Taphonomic and ostracod abundance data for LT97 -56V fossil assemblages. # # # # core (cm)depth in 40 -1-5 carapaces # 14.0 2.0 # valves 98.086.0 # adults 2.06.0 cracked/broken 52.042.0 reductionstained 2.01.0 corroded/abraded 1.0 0 opaqueyellow/ 11.0 7.0 encrusted # 6.01.0 # cods/ gram2,3723,313 221612 -23 -17-13 8 -9 12.2 5.06.02.0 95.087.894.098.0 7.06.06.13.0 70.068.439.0 5.01.01.0 13.3 5.02.0 0 31.014.011.211.0 5.04.0 0 5,569 459474403 54463830 -55-47 -31-39 10.0 4.08.0 96.090.092.0 4.05.08.0 53.047.055.0 34.0 4.02.05.0 14.0 6.01.0 0 15.017.0 9.05.0 16.0 2.06.08.0 2,4041,038 280504 86787062 -71-87-63-79 4.78.0 0 100.0100.0 92.095.3 37.211.012.012.5 57.061.075.069.8 27.918.0 8.0 0 46.020.937.5 6.0 28.037.525.6 9.0 32.037.212.5 7.0 1,040 9931 7 118110102 -111-119-103 94 -95 10.0 1.0 0 100.0 99.090.0 4.06.06.0 67.061.048.051.0 29.015.0 3.02.0 2.03.08.0 0 4.03.01.0 6.02.03.0 2,5651,1411,313 583 150-151142134126 -143 -135-127 10.015.9 6.02.0 r 94.090.098.084.1 36.5 4.0 59.051.040.058.7 34.912.0 2.04.0 23.8 0 23.814.0 8.09.0 9.51.0 0 2,8441,804 674 16 178174166158 -175-167-179 -159 4.08.0 92.096.0 8.06.01.0 0 71.073.049.067.0 3.02.01.0 0 0 28.0 4.01.0 0 2,8381,4331,902 274 191
Grain size distribution for samples from core LT97 -56V. All weights in grams.
depth in 106 µm-63 -106 total dry core (cm)>1 µm 1 mm µm <63 µm weight 0 -1 0.018 0.517 1.500 4.778 6.813 2 -3 0.032 1.535 2.600 7.694 11.861 4 -5 0.022 0.789 2.001 3.651 6.463 6 -7 0.046 1.4401.743 8.176 11.405 8 -9 0.028 0.640 0.935 3.239 4.842 12 -13 0.008 0.630 0.835 3.970 5.443 16 -17 0.014 1.543 1.015 3.679 6.251 22 -23 0.001 0.322 0.475 2.599 3.397 30 -31 0.006 0.387 0.865 4.019 5.277 38 -39 0.009 0.316 0.710 3.990 5.025 42 -43 0.094 0.557 0.685 7.882 9.218 46 -47 0.002 0.245 1.084 3.207 4.538 50 -51 0.051 0.690 2.776 10.683 14.200 54 -55 0.019 0.664 0.672 3.732 5.087 58 -59 0.001 0.319 1.014 7.329 8.663 62 -63 0.001 0.258_0.634 2.468 3.361 70 -71 0.002 0.524 0.419 2.223 3.168 78.-79 0.001 0.154 0.486 5.053 5.694 82 -83 0.003 0.361 0.879 9.002 10.245 86 -87 0.004 0.497 0.840 4.027 5.368 90 -91 0.013 0.496 0.791 10.462 11.762 94 -95 0.184 1.188 1.158 6.584 9.114 102 -103 0.004 0.332 0.527 5.032 5.895 110 -111 0.004 0.376 1.292 6.455 8.127 _118-119 0.001 0.304 0.720 5.154 6.179 126 -127 0.001 0.415 0.361 4.395 5.172 130-131 0.007 0.525 1.098 10.131 11.761 134 -135 0.001 0.283 1.617 4.054 5.955 138 -139 0.002 0.524 2.081 10.753 13.360 142 -143 0.000 0.263 0.513 4.556 5.332 146 -147 0.000 0.167 0.351 10.129 10.647 150 -151 0.001 0.167 0.427 6.627 7.222 154 -155 0.002 0.363 0.389 11.769 12.523 158-159 0.000 0.419 0.945 6.841 8.205 162 -163 0.011 0.491 1.382 11.248 13.132 166 -167 0.001 0.828 1.812 5.154 7.795 170 -171 0.000 0.485 0.998 10.002 11.485 174 -175 0.007 0.258 1.185 8.337 9.787 178 -179 0.003 0.3321 11.690 13.156 182 -183 0.010 0.407 1.132 8.070 9.619 186 -187 0.007 0.222 0.342 15.636 16.207 190 -191 0.003 1.110 1.582 7.975 10.670 192
depth in 106 µm-63 -106 total dry core (cm)>1 µm 1 mm µm <63 pm weight 198 -199 0.006 0.445 1.821 8.093 10.365 206 -207 0.002 0.217 1.547 6.815 8.581 214 -215 0.001 0.208 1.164 2.403 3.776 222 -223 0.005 0.718 2.346 5.711 8.780 230 -231 0.001 0.370 1.051 6.235 7.657 238 -239 0.002 0.308 1.114 6.623 8.047 242 -243 0.001 0.428 0.001 11.956 12.386 246 -247 0.003 0.313 1.021 5.944 7.281 250 -251 0.006 0.222 0.001 12.228 12.457 254 -255 0.007 0.309 0.820 6.076 7.212 258 -259 0.006 0.359 0.000 14.933 15.298 262 -263 0.000 0.300 1.995 6.068 8.363 266 -267 0.008 0.312 1.607 11.665 13.592 270 -271 0.004 0.355 0.816 6.383 7.558 274 -275 -0.001 0.356 1.856 9.544 11.755 278 -279 0.001 0.331 0.327 7.365 8.024 286 -287 0.005 0.496 1.169 7.781 9.451 294 -295 0.009 0.742 1.607 5.445 7.803 298 -299 0.024 0.737 2.044 8.785 11.590 302 -303 0.004 0.774 0.804 6.525 8.107 306 -307 0.007 1.607 2.160 9.084 12.858 310 -311 0.001 0.846 0.016 5.933 6.796 314 -315 0.023 5.511 2.946 7.223 15.703 318 -319 0.002 2.750 2.115 4.871 9.738 322 -323 0.028 5.216 3.795 6.412 15.451 326 -327 0.006 2.071 4.153 4.084 10.314 330 -331 0.006 7.513 2.277 4.586 14.382 334 -335 0.009 4.359 2.326 2.948 9.642 338 -339 0.002 9.117 1.700 3.577 14.396 342 -343 0.001 9.685 0.980 0.952 11.618 Loss -on- ignition data for core LT97 -56V. Whole sediment % % samples were measured. core (cm)depth in 40 -1-5 organic 6.767.34 carbonate 2.001.30 core (cm)174depth166 -175 -167in organic % 3.334.28 carbonate % 2.142.46 22 -231612 -17 -13 8 -9 6.937.207.81 0.970.001.00 198190182 -191-183-199 3.093.713.99 0.520.580.81 30 -31 5.396.84 1.341.03 214206 -215-207 r 4.513.32 0.760.58 62544638 -63-55-47 -39 7.205.285.587.08 0.984.190.841.07 246238230222 -231-247-239 -223 4.283.403.914.17 0.660.740.750.73 94867870 -95 -71-87-79 5.395.896.058.71 0.871.301.111.29 278270262254 -279 -271-263-255 2.963.163.573.21 0.510.580.570.61 126118110102 -111-127-119 -103 6.895.354.494.85 1.101.161.211.25 310302294286 -311-303-295-287 3.183.523.142.98 0.640.630.55 158150142134 -151-143-159 -135 4.143.644.464.80 2.302.142.071.66 342334326318 -343-335-327 -319 0.581.601.881.59 0.140.340.350.32 GeochemicalCorrection of data %N for values core LT97described -56V in <63 Chapter µm sediment 3. fraction. core (cm)depth in 0.5 4.80%C (raw)%N 0.31 (corn) %N 0.30 (raw)15.34C/N (con)C/N16.10 _ -22.213C S 15N S 1.8 core (cm)depth in 174.5 %C 0.90 (raw)%N 0.08 (corr.) %N0.08 (raw)11.89C/N (con)C/N11.89 S 13C -22.5 b 15N 0.2 12.54.58.5 1.691.451.15 0.150.120.09 0.140.100.07 11.1212.2013.07 . 12.3213.9315.71 -21.7-20.5-21.5 2.22.31.8 186.5182.5178.5 0.620.901.22 0.040.110.08 0.040.110.08 14.4011.5712.05 14.4011.5712.05 -23.2-23.0-22.8 -0.3 0.10.4 16.5 11.47 , 12.52 2.2 190.5 1.08 , 11.98 11.98 0.6 2.02 0.18 , 0.16 -20.8 0.09 0.09 -23.0 22.5 1.111.41 0.12 0.10 , 11.84 . 13.52 -20.4 2.01.3 198.5 0.73 0.06 0.06 12.82 12.82 -23.1 0.3 46.538.530.5 2.101.40 0.110.160.08 0.090.150.07 13.2013.1313.75 , 15.3414.4616.83 -20.0-20.5 2.3 222.5214.5206.5 0.911.081.03 0.090.080.07 0.090.080.07 13.2213.0012.52 12.5213.2213.00 -23.2-23.3-23.1 0.40.30.5 62.554.5 1.611.12 0.130.09 0.110.07 12.8113.22 14.5116.01 -19.9-19.6 2.41.41.6 , 238.5230.5 1.07'1.18 0.080.09 0.080.09 13.2813.8613.16 18.3513.1613.28 -23.6-23.0 0.70.31.4 78.570.586.5 2.061.201.60 0.090.120.16 0.070.110.14 , 13.3213.9813.01 16.8814.7914.72 -20.6-19.7-19.6 2.31.8 254.5246.5262.5 0.830.630.78 0.060.060.04 0.040.03 14.7914.39 20.8919.57 -23.4-23.7 1.11.0 118.5110.5102.594.5 0.870.921.111.45 0.080.060.070.11 0.070.050.09 13.4313.7513.6613.59 16.8317.7617.3715.56 -21.5-21.1-20.9-20.3 2.41.31.51.4 294.5286.5278.5270.5 0.660.700.561.11 0.030.040.060.05 0.020.030.05 16.5916.5517.8514.96 25.1722.9221.1127.78 -23.6-24.2-23.9-24.0 1.41.21.91.7 150.5142.5134.5126.5 0.941.071.62 0.080.070.11 0.060.050.09 13.7313.5814.5015.30 17.7916.9517.2918.13 -22.0-21.4-21.0 2.71.61.9 326.5318.5310.5302.5 0.510.720.901.08 0.030.050.06 0.030.050.04 18.2115.26_17.9516.31 23.2621.7218.2115.26 -22.8-23.0-24.0-23.7 0.71.31.41.9 170.5166.5162.5158.5 0.901.281.161.25 0.120.070.090.10 0.080.120.060.08 11.1512.1412.4912.73 11.15_15.1714.8615.00 -22.5-21.9-22.6-22.5 0.52.81.41.2 338.5334.5 0.450.430.59 0.030.02 0.030.02 17.8417.9617.76 17.8417.9617.76 -21.7-22.1-21.9 1.61.7 Smear slidedepth data in for core LT97 -56V, based on 200 point counts. % % % % depth in % % % % core (cm) 40 -1-58 -9 siliciclastic 51.550.025.0 organic 46.048.571.5 carbonate 0.53.01.0 siliceous 0.51.51.0 core (cm)182174166 -183-175-167 siliciclastic 49.554.554.0 organic 44.545.047.0 carbonate 0.51.0 0 siliceous 0.52.51.0 3022 -31-231612 -17-13 55.048.041.5 9.5 44.552.058.586.0 0.54.5 0 00 214206 198190-215-207 -191-199 46.565.061.550.5 52.549.034.538.5 0.5 0 0.5 0 62544638 -63-55-47 -39 56.049.555.0 44.049.043.043.5 0.5 0 2.00.51.0 0 246238230222 -231-247-223-239 60.043.520.052.5 56.578.046.038.5 0.01.51.0 0.5 0 94787086 -71-95-79 -87 67.051.036.031.0 31.548.061.565.5 0.50.51.0 0 2.53.01.0 0 278270262254 -271-263-255-279 69.563.553.0 44.030.035.536.5 0.5 0 3.01.0 00 126118110102 -111-127-103-119 55.050.554.033.0 43.548.046.064.0 0.52.01.0 0 0.51.0 0 302294286310 -303-295-287 -311 65.553.069.563.5 35.033.546.030.5 0.5 0 0.51.5 0 158150142134 -151-135-143-159 49.050.059.038.5 50.048.558.539.5 0.51.51.0 0.52.5 0 342334326318 -343-335-327-319 73.560.061.550.0 26.040.038.548.5 00 0.51.5 00 196
Appendix F: Raw data for life and death assemblages collected from surface sediments at Mwamgongo, Tanzania, between October 1997 and July 1999.
Sample abbreviations for life and death assemblages and surface sediment grain size.
Abbreviations consist of a combination of quadrat number (1 -8); A for sediment sample, L for life assemblage, and D for death assemblage; and a two -letter code for month of collection, as illustrated in the table below.
Two pairs of 25x25 cm quadrats were collected each month at both 5 and 10 m water depth. Quadrat pairs at 10 m were numbered 1 -2 and 3 -4, pairs at 5 m have the numbers 5 -6 and 7 -8.
date sediment life death collected samples assemblagesassemblages 8- Oct -97 1AOc 2LOc 3DOc 8- Nov -97 1 ANo 2LNo 3DNo 1- Dec -97 lADe 2LDe 3DDe 5- Feb -98 1AFe 2LFe 3DFe 8- Mar -98 1 AMr 2LMr 3DMr 16- Apr -98 1 AAp 2LAp 3DAp 13- May -98 1 AMy 2LMy 3DMy 17- Jun -98 1 AJn 2LJn 3DJn 23- Jul -98 1 AJ8 2LJ8 3DJ8 25- Jul -99 1 AJ9 2119 3DJ9 wereWaterprovidedMissing measured depths months by atC. using quadrat Birkett of water scuba locationsat NASA.depth depth data throughoutgauges necessitated in Oct. the '97,sampling calibration Mar. '98,interval using Apr. (Oct. satellite'98, Jun. 1997 altimetry'98, -Jul. Jul. 1999). '98, data and Watergraciously Jul. depths'99. quadrats collected month water depth (m) quadrats collected month water depth (m) 1A-2A NovFebDecOct -97 -98-97 -97 11.210.410.310.2 5A -6A NovFebDecOct -97 -98-97 -97 4.95.95.15.0 1A-2A MayMarApr -98Jun -98 -98 -98 12.212.312.111.5 5A -6A MayMarApr -98 Jun-98 -98 -98 6.97.06.86.2 3A3A1A-2A -4A -4A NovOct -97Jul Jul-97 -99 -98 10.810.711.411.9 7A5A -8A-6A NovOct -97 -97Jul -98-99 5.15.06.16.6 3A3A -4A -4A MarDecAprFeb -97-98 -98 12.612.011.710.9 7A -8A MarAprFebDec -98 -98-97 6.96.36.05.2 3A -4A May -98JunJul -98 -99-98 11.912.412.712.8 7A -8A May -98JunJul -98Jul -98 -99 6.26.77.07.1 . 198
Life assemblage species abundance data. See Appendix D for species abbreviations.
Al cfAlAlAlAlAlAlAlAlAlArCdCdCdCnCnCpCpCpCpCpCpCpCp sampleabciclhuinmu10111820tuca24252 8 bicose56A6C1517
1LNo 0 000 5 1 0 1 00000 0 0 0 00 0,0 1 3 00 , . 0 0 000 0 0000 2 15 00 1LDe 0032100200000 - - 1LFe 0 000 1 00 3 00000 0 0 0 0000 0 2 00 1LMr 0 0 0016 0 000002000 0 0 000 0 1 00 1LAp 0 0 00 3 0 017000 3 000 0 0000 0 0 00
1LMy 0 0 00 5 0040000 1 000 0 0000 0 7 0 0 1LJn 0 0 0021 2 038000 1 0 1 0 0 0 3 00 0 1 00 1LJ8 0 0 00 8 3 000000000 0 0000 0 2 00 , . 1LJ9 0 0 02 12 9 00000 1 000 0 0 2 00 2 2 00 , 2LOc 0 0 00 1835040000 1 2 0 0 0200 5 7 00 2LNo 0 0 0012 4 0,1000,000 1 0 0 3 00 0 0 00 2LDe 0 0 0040 120700000 1 0 0 0000 0 0 00
0 0 1 0 000000000 0 0000 0 2 00 2LFe 0 0 . 2LMr 0 0 0016 1 020000000 0 0000 0 0 00 2LAp 0 0 00 5 0 070000000 0 0000 0 1 00 2LMy 0 000 1 1 070000000 0 0000 0 3 00 2LJn 0 0 0019 0 01300000 1 0 0 0000 0 1 00 _ _ . . 2118 0 00210 13 0200000000 0 0000 0 2 00 . , , , . 2LJ9 0 000 7 15 '000000000 0 0 1 1 0 0 5 00 3LOc 0 0 0010780160000 1 1 0 0 2900 2 17 00 _ _ , , _ 3LNo 0 0 0Or5 0 0000000 3 0 0 0000 0 1 00 , . , . . 3LDe 0 0 0 1 15 6 07000 1 0 2 0 0 0 1 00 1 3 00 , , . . - 3LFe 0 0 00 1 0 0 1 00000 2 0 0 0 1 00 0 5 00 . . . , , 3LMr 0 0 02 3 0020000000 0 0000 1 2 00 . . . , , _ _ , _ . , , 3LAp 0 0 00 2 2 0280000000 0 00i3O0 13 00 3LMy 0 0 00 0_002000 0 0 1 0 0 0 0 0 0 3 0 0 3LJn 0 0 0 1 3 3 0830000 1 2 0 0 0 200 0 1600
3LJ8 1 0 0 00 3 390870000000 0 04 00 1 3700 , . . , . . .. , 3LJ9 0 0 00 1295060002 0 1 0 0 1 400253000 , 4LOc 0 0 02 8 1005 000202 00 0 000 1 0 2000 _ . _ , . , . 4LNo 0 0 00 1 2 0 1 0000000 0 0 1 00 0 3 00 V ' V 4LDe 0 0, 00 4 220170000000 0 0000 2 5 00
4LFe 0 0 00 0 3 012000 1 000 0 0000 2 1200 . , _ _ . . , 4LMr 0 0 00 2 0 05 000 1 000 0 002 0 3 1 00 l i _ , . y
4LAp 1 0 0 0 1 0 0 013000 5 0 1 0 ' 0 0 0000 2 5 00 4LMy 0 0 00 1 0 0.240000000 0 0000 0 4_00
4LJn 0 0 00 0 0 023000 1 0 1 0 0 0 0 0 0 1 3 0 0 , , , 4LJ8 0 0 00 9 8 1 170000 1 2 0 0 0 1 0 2 3 7 00 - . , , 4LJ9 0 0_00 2 38000000000 0 04 00 0 1600 _,_... , . , . SLOc 04000 0 41_00000000059 5 3 2 0 0 56 1 0
5LNo 0 2 00 9 20-00000000032 1 1 00288400 , . , , . 5LDe 0_ 3 00 0 4300000000 0 2 0 3 1 0 028200
5LFe 0 0 00 0 360 1 0000000 0 1 8 00 0 8 00 I 0 0 0 0 0 0 0 0 0 0 0 0 0I i S O I 00 8 0 S a3"IS , 0 0 0 0 0 0 0 0 0 0 0 000 I 00 I 00 60Lt aQ"IS 0 0 0 0 0 t7 0 0 0 0 Z 000000000 00 L£ °N"IS 0 0 I 0 0 0 £ i 0 0 00000000000 001S , .0£ 0 0 0 0 0 0 0 0 0 0 0 i 000000_00 Z S 0 611t7 0 0 0 0 I 00 Z 0 0 0 00 i 000000 Z Z 0 81-117 0 0 0 0 0 0 0 0 0 0 0 0000000000 0 0 uf'1t 0 0 0 0 0 0 0 0 0 0 0 00000 I 000 I 0 0 '11I117
I 0 0 0 0 0 0 0 0 0 ¡ O T T O 000000 0 0 dVrlt7 0 0 0 0 0 0 0 0 0 0 0 00000 I 000 i Z 0 IIAIlt 0 0 0 0 0 0 0 0 0 0 0 00 I 000 I 000 0 i adZt7 0 0 0 0 I 0 0 Z 0 0 0 00 I 000 Z 000 Z 0 acr-17
0 0 0 I 0 0 0 i 0 0 0 000000 1 0000 0 °N1t7
0 0 0 0 0 0 0 0 0 0 0 1 7 0 T O O O I O O T Z O 001t7 . . . 0 0 0 0 i 0 0 0 0 0 .f/ 000000 I 009I0 0 6f1£ . . f 0 o o I 0 i 000o Z Z o0OZ 0 Bfrl£ 00 0 0 . 0 0 0 0 0 0' 0 0 0 0 0 00000 1 i 00b I 0 ufZ£ 0 0 0 0 0 0 0 0 0 0 0 i 0000000000 0 XY1i'I£ , 0 0 0 0 0 0 0 0 0 0 0 000 1 O i I O O Z O 0 dV'I£
0 0 0 1 0 0 0 1 0 0 0 1 00 1 0 1 1 0000 0 IL1IrI£ 0 0 0 0 0 0 0 0 0 0 0 0 í 000 I I 0000 0 037£ 0 0 0 0 0 0 0 0 0 0 0 000 I 0 I 000 i I 0 aQri£ 0 0 0 0 0 i 0 0 0 0 01700000000 00 0 °N'I£ 0 0 0 0 0 0 0 0 0 0 C0' 000000 I 00 I Z L I DOrI£ , . , , , . . 0 0 0 0 0 0 0 0 0 0 0 Z 000CO000 Z 0 0 6f"IZ , 0 0 0 0 0 0 0 0 0 I 0 0 I 0000000 í Z 0 BfZZ 0 0 0 0 0 0 0 0 0 0 0000000 Z 00 00 0 uVIZ 0 0 0 0 0 0 0 0 0 0 0000000 I 00 00 0 XIAIZZ 0 0 0 0 0 0 0 0 0 0 0 000000 i 00 00 0 dV7Z 0 0 0 0 0 0 0 0 0 0 0000000 1 00 00 0 IL1I1Z 0 i 0 0 0 0 0 0 0 0 0 000000000 00 0 03-IZ , 0 0 0 0 0 0 io0 i 00 S 000 I 0 aQHZ 0 0 0 0 0 , 0 0 0 0 £ 0 0 0 0 0 0 0000000000£ 0 °II'IZ 0 0 0 0 0 0 0 I 0 0 0000000 Z 006ZZ 0 001Z Z , 0 0 0 0 0 0 0 Z 0 0 0 000000000 1 S 0 6f71 0 0 0 0 0 0 0 0 0 0 0 00000 í 000 i 0 0 81-II 0 0 0 0 0 0 0 0 0 0 0 000 000 1 0000 0 urn 0 0 0 0 0 0 0 0 00 0 0000 1 0000 00 O XL1I7I 0 0 0 0 0 0 0 0 0 0 00 I 0000 í 0000 0 AMt . . , . . , 000 0 0000000 0 Iy1IrI i 0 0 0 0 0 0 0 0 0.0 0 0 0 0 0 0 i 0 0 0 0 0 000 0 00 Z 0000 0 od-II - - 0 0 0 0 ' 0 0 0 0 0000000000 L 0 aQ'I i ,0-9-0. . , . ,
0 0 0 0 £ 0 ' 0 0 0 0 0 00000000000 0 °NrI t i£0£6ZOZ61 6nsdocuauwo. qoM!AA n oai sdsSZ£Z8ia Idures owNATowowowNATowov1iowowowux°J°!J°rJ°IJ°rJ°guQdpdpdpdj
661 200
McMcMsMsMsPrRoRoRoTcTcTpTpTpTpTpTt sample3740irpi2B 1 amlo temaiacde3 4 8bu
1LNo 0 0 0 0 003 4 15 0 1 000002 . 1LDe 0 0 2 0 007 7 33 000000 3 0 1LFe 0 0 0 0 008 29440000002 1 1LMr 0 0 0 0 00 8 1037000000 1 1 1LAp 0 0 0 0 0 1 644840.00,0009 1 1LMy 0 0 0 0 005 9 43 0000004 1 1LJn 0 0 0 0 00 8 5082000000395
1 LJ 8 0 0 0 1 0062442000000 72 1LJ9 0 0 0 0 001117 61 0000000 1 2LOc 0 0 7 0 005 106500000000 2LNo 0 0 0 0 00 3 3 2400000000
1 00 1 1 2LDe 0 0 2 009 18 4000.00 2LFe 0 0 0 0 00 8 103000000000 2LMr 0 0 1 0 005 20 3 8 0 1 000020 2LAp 0 0 0 1 006236000000040 , 2LMy 0 0 00 00 1 15 3700000040 2LJn 0 0 0 0 0019244800000092
2LJ8 1 0 0 0 00181842000000 8 4 2LJ9 0 1 00 004 11 550 1 0000 1 5 3LOc 0 0 5 0 1 07 26117000000 1 6 3LNo 0 0 1 1 001032 14 0000000 8 , 3LDe 0 0 1 0 009 18 280 1 000002 , - 3LFe 0 0 0 0 00146 13 00000000 3LMr 0 0 1 1 1-02310370 1 0000 1 4 3LAp 0 1 0 0 0011 5 64000000 3 1 3LMy 0 0 0 0 1 011 4 350 3 000002 , 3LJn 0 0 1 0 2 02010 680 2 0000 1 3 _ _ , , 3LJ8 0 0 0 2 1 01114870 1 000024 3LJ9 0 2 00 1 04 7 1450 1 00000 1
4LOc 0 0 7 1 0 1.,0321611370 3 000 0016 . , 4LNo 0 0 1 0 00 1 6 6 0000000 1 , , . 4LDe 0 0 1 1 001010440000000 1
4LFe 0 0 0 1 004 11 5500000040 , 4LMr 0 0 0 00 0 1109 280 2000002 . 4LAp 0 0 0 0 201127890200 1 0 1 4 4LMy 0 2 0 0 002 6 4500000000
4LJn 0 0 2 0 1 0127 380600000 5
4LJ8 0 0 0 0 00191948 1 0000 1 0 8
4LJ9 0 3 1 0 1 0 3 2 430000000 3
5LOc 0 0 1 000 1 0 1 00000000
5LNo 0 0 0 0 00 1 0 2 0 1 1 00000
5LDe 0 0_0_0 000 0 0 0002 0000 5LFe 0 0 0 000 1 0 3 00000000 201
Al cfAlAlAlAlAlAlAlAlAlArCdCdCdCnCnCpCpCpCpCpCpCpCp sampleabciclhuinmu10111820tuca24252 8 bicose56A6C1517 5LMr 0 0008481 03600000 00 0 0262 0 8 7900 5LAp 0 00029410540000000 0 019 1 0 0 8200 5LMy 0 00014,3701800000 00 0 1 7 00 0 6 00 5LJn 0 6 00 06302100000 00 0 01100 6 3 00 , 0 0 00 01470 0000 00 1 01300 415000 5LJ8 60 , , . . , 5119 0 4 00 0 8 0000000 0 0 1 000 0 9 1 00 , 6LOc 088001321530 30000000217142 0 02280 00 6LNo 0 0 00 1 11 0000000 0 0 7 06 001360 0 1 , . 6LDe 0 1 00 4 35000000000 7 040 01890 00 6LFe 0 0 00 1 330 1 0000000 0 02500 1 1 00 6LMr 0 0 0046620290000000 0 02400 4 5200 6LAp 0 0 0020540350000000 0 02500 6_4400 6LMy 0 0 0026320300000000 0 1 5 00 1 2300 6LJn 0 1 00 1 3105 00000 1 0 1 0 1 00 1 1 00 , 6LJ8 0 0 00 416602000000 02801500113 1 00 6LJ9 0 4 00 3 4 000000000 0 0000 12 0 00 7LOc 012900 1 700000002 3 023 1 400551900 7LNo 0 4 00 1 1000000000 0 0 0000 0 0 0 . . . , 7LDe 0 1 00 142160 5 0 1 0002 0 4 1 260 051 5 00 . , . 7LFe 0 0 00612501400000 1 0 0 0 3 00 181300 7LMr 0 0 1 086 9 01300000 1 0 0 00 00 9 2400 7LAp 1 0 002621 03400 1 1 0 3 0 0 0 2 00 3 3400 , 7LMy 0 0 00 5 9 015_0,0000 1 0 0 0 1 00 9 8 00 7LJn 0 2 00 2 280000 1 00 3 0 0 0 1 00 0 8 00 , . . , . , . 7LJ8 0 1000 3 700 1 0000 1 40 '3 0 040 18 13 00 . . , . , 7LJ9 0 1 00 3 42000000 0 0 0 0 0000 3 0 00 _ . , , _ . 8LOc 0 15 00456200000004 0 1 3 5 0 0 3 2900 8LNo 0 0 00 3 660 1 1 00002 0 1 040 0 1 1 00 . . . 8LDe 0 5 0 1 261770 1 0000000 0 1 2 0 0 9 5 00 . , 8LFe 0 0 00592101200000 00 0 00 00 6 2000 , 8LMr 0 00089160290000000 0 09 1 0 7 2000 , , . , 8LAp 0 0 00 13 1201000000100 0 0 0 1 0 02400 8LMy 0 0 00 3 9100000 1 0 00 0 0 3 00 2 11 00
8LJn 0 1 02 0 14000000 1 2 0 0 0 1 00 0 1200 , . 8LJ8 0 1200 0 1110000000 3 0 0 0 2 2 0 16 5 00 8LJ9 0 13 00 1 3290 1 0000 1 1 0 5 1 200200 00 ZOZ
dp dp d3 dp usa oJ oJ oJ oJ oJ orJ uN Ow Ow 0w 0w 0w 0w 0w 0w 0w 0w 0w ajdulrs 81 £Z SZ ds ls p o0 n no IM onn ici wo u0 wo do ns 6 61 OZ 6Z 0£ IL
S JIAIZ T T Z I 0 t O t 0 0 0 I 0 0 0 0 0 0 0 0£ 0 0 . , _ . L£ - dV75 Z O O O L O E O O O O O O £ 0 0 0 0 0 0 0
b L Z T O 0 0 0 0 0 0 0 0 0 4\I-IS 0 0 0 ' 0 0 t 0 0 0 Mg SI 0 5 0 0 8 0 0 Z O O 0 0 0 0 T 6 0 0 0 0 0 0 , , , , . , 8f-I5 IL 0 LI 0 0 Z O O O O O O O T 0 i 0 0 0 0 0 0 0 6f-IS 017Z 0 0 0 0 0 0 0 0 0 0 i 0 0 0 0 0 0 0 0 0 0 0O'19 SZZ I £ tfi 0 0 0 0 0 5 0 0 I 0 0 0 Z £ 0 a' 0 I S , 00 0NI'19 T L Z 9 0 0 0 0 0 Z 0 0 0 0 0 I 0 0 0 O'S 0 0 0 0Q'19 0£ 0 0-1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Z 0 0 0 I , _ 0379 9 0 I 0 0 0 b i 0 0 0 0 0 0 0 0 £ 0 0 0 0 0£ , . 0 , . JIAIZ9 I L O T 9 0 0 0 0 0 L . O O I 0 0 0 Z 00 T 0 0 0 0 I 0 I 0 0 9 0 9 L O t 0 0 0 0 0 0 0 0 0 0 0 0 dVrl9 , 4\I19 09 I 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 I 0 0 0 0 uf-19 9 0 9 0 0 I 0 I 0 0 O 0 0 0 0 I 0 0 't 0 0 0 0 8f79 Z£Z 0 LZ 0 0 0 0 0 0 0 Z 0 0 0 I 0 0 0 0 0 Z 0 0 , . 6f79 LZ 0 0 0 0 0 0 0 0 0 0 0 o o 0 0 0 . 0 0 0 0 0 0 00-IL 8 Z 1 6 1 0 0 0 T £ 0 0 0 0 L 0 00 8 0 0 0 T T 00 oIIIL 0 T 0 0 0 i T 0 0 0 I 0 0 0 0 0 0 0 T 0 0 0 0
OEIL I 9 I £ 6Z O T T Z£ 9 5 0 0 0 0 0 0 I O O Z O O O Z O4ZL 0 0 t 0 0 I O I Z O T 0 0 0 0 0 0 0 0 0 0 0 0 IIAIrIL I 0 0 0 0 I I 0 0 0 I 0 0 0 0 0 0 0 0 00, . . WO d`d'ZL 0'0 1 0 0 6 £ I Z O I 0 0 0 0 I 0 0 0 0 0 0 0 ignIlL T £ i 0 0 T 0 Z 0 0 0 0 0 0 0 0 0 0 Z 0 0 0 0 -1L uf L 0 S 0 0 Z 0£ 0 0 0 T 0 0 0 £ 0 0 0 0 0 0 0 8f-IL 06 0 9 0 0 t I O Z O O Z O 0 0 T 0 0 0 0 0 0 0
6f'1L L 0 T 0 0 0 0 0 0 0 0 0 £ 0 0 0 0 0 0 0 0 0 0 0078 6 I i Z 0 0 0 I 0 0 0 I Z 0 0 0 i 00 T 0 0 0 0 0N'18 Z Z T 0 0 Z Z 0 Z 0 0 0 0 I 0 0 L o o o o .. . . _ I£ , , 0Q'18 L£ 0 0 0 0 0 Z 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0£ , 0318 00 8 0 0 0 0 0 I 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1IA1-I8 0 T £ 0 0_ I I 0 i 0 0 0 0 0 0 0 Z 0 0 0 0 0£ , dárl8 0 0 0 0 0 Z_ Z 0 0 0 I 0 0- Loo 0 0 0 0 0 0 0 /IA178 I 0 T 0 0 8 T 0 0 0 0 I 0 0 0 0 0 0 0 0 0 0 0 uf-18 Z T I 0 0 0 I T T i 0 0 0 0 0 0 T 0 0 0 0 0 0 0 . . . . , . _ .
81'18 OL i 6 0 0 Z Z 0 0 0 0 £ 0 0 0 T 0 T 0 0 0 0 0 6rI8 91 0 5 0 0 0£ 0 0 0 0 0 0 0 0 0 0 0 I 0 0 0 i 203
McMcMsMsMsPrRoRoRoTcTcTpTpTpTpTpTt
sample3740irpi2B 1 amlo tema 1 acde3 4 8bu . . . . 5LMr 0 0 2 0 00132 2000 1 00000 5LAp 0 0 7 0 009 0 2300200000 5LMy 0 0 0 0 000 0 2200200000
5LJn 0 0 0 0 00 1 0 17 00000000
5LJ8 0 5 0 0 00 1 0 3 0 1 000000 5119 0 0 0 0 000 0 0 000000 1 0
6LOc 0 18 9 0 009 1 4 00000000
6LNo 0 0 1 0 00 1 0 0 00000000 , 6LDe 0 0 0 0 000 0 0 00'000000 0 0 0 00 00 6LFe 0 0 00 0 2 0.0 0 0 6LMr 0 2 1 0 004 0 12 00000000 , 6LAp 0 0 2 1 004 0 2400000000 6LMy 0 1 1 1 0 003 1 2300000002 6LJn 0 1 0 1 004 0 1 00000000 6LJ8 0 3 0 0 00 3 0 3 0 1 000000 , 6LJ9 0 0 0 0 000 0 1 0000000'0 , 7LOc 0 1 11 1 0065 1 11 0000000 1 7LNo 0 0 4 0 0034 1 6 00000000 , 7LDe 024 5 2 00100 9 00000000 7LFe 0 0 1 2 002 0 3 0000000 1 , 7LMr 0 1 3 1 0018 1 8 0 1 00000 1 . - 7LAp 0 1 6 1 00292 52-0 1 000002 7LMy 0 2 0 1 00 3 1 4000 1 0000 1
7LJn 0 5 2 0 0024 1 260 1 00000 . . . 7LJ8 0 12 4 1 0031 3 240 1 00000 3 7LJ9 0 1 0 0 000 0 1 0000000 1 8LOc 0 j0 3 1 00342 13 1 0000000 8LNo 0 0 6 0 0040 5 17 0000000 1
8LDe 0 0 3 0 00 3 02000 1 00000 8LFe 0 0 1 0 00 5 0 5 00000000 8LMr 0 0 6 1 0020-0 100 1 000000 8LAp 0 0 2 1 1 0494 220100000 8LMy 0 0 00 002 2 2600000002 , , 8LJn 0 0 3 0 0025 1 280 1 0 1 0000 , - . . 8LJ8 0 0 3 0 2 021 1 300 1 0_0000.2 8LJ9 0 0 1 0 002 0 6 1 0000000 Death assemblage species abundance data. See Appendix D for species abbreviations. Al Cn Cp i Al Al Al Al Al Al Al Al Cn Cn Cn Cn Cn Da Ar Cd Cd Cd Cd Cd 1 Cn I cf I cf Cp i Cp sample hu in incf mu 5 3 11 17 18 20 r tu 3 ca 1 1 3 2411 2519 de de 2 . 8 1 12 15 Cpbi Cpco 1 la Cpob Cpse 2 Cp 5 6ACp 6B 6CCp 14 Cp15 Cp17 Cp 18 . Cp2321 2511 st 11DNo DJ8 r 20 20 9 36 _ 10 2 0 0 0 0 0 0 16 13 0 1 0 02 0 0 0 0 1 0 0 3 0 1 2136 0 7 20 0 26 16 00 1DJ9 2 52 f 431 57 40 2 0 0 031 20 0 0 0 12 20 0 001 2 0 031 0 r 0 0 3 , 0 1 0 0 1 0 _ 22 0 11 0 0 2033 . 1914 , 2015 5 0 2DNo 21 1 2 3 0 92 0 1 00` 1 , 01 0 0 0 r 1110 20 0 0 2 0 0 0 1 0 0 00 0 0 0 0 2334 0 7 0 0 1 2923 19 11 0 2DJ92DJ8 2 1 4 1 15 8 1 26 0 1 0 3 0 0 0 14 8 1719, 0 0 1 0 0 1 0 0 0 0 , 0 0 0 0 24 0 11 8 31 0 2518 13 12 5 0 3DNo3DOc 2 0 21 31 0 0 0 0 0 02 1 0 0 0 1013 17 0 1 0 01 0 1 . 1 . 0 01 20 01 0 0 0 1 0 4029 0 13 1 0 28 291316 1112 01 , 0 0 0 0 0 16 2 7 3DDe 2 0 1 . 0 0 0 0 0 0 01 0 0 0 0 39 0 0 0 1 30 3DMr3DFe 2 1 0 21 0 03 31 0 0 0 0 1 0 0 0 20 9 2019 0 1 0 23 0 32 00 0 0 0 0 0 20 0 2334 0 54 03 0 2329 1511 1312 00 3DMy3DAp 31 0 08 3 1 04 5 0 0 1 0 1 0 0 0 10 16 6 0 0 40 00 031 0 0 0 0 00 00 31 06280 35 00 47 1 00 20 3723 1817 1110 9 0 3DJ83DJn 3 0 356 43 40 1 269 0 0 20 , 3 1 0 0 0 141112 1716 0 0 0 0 20 0 0 1 0 0 0 0 01 0 303938 0 24 0 0 252019 1110 1811 0 16 0 r 0 3DJ9 , 0 , 0 ' 0 r 0 0 . 3 2 12 2 ' 10 1 4DJ84DNo 0 1 21 35 10 78 031 06 0 0 1 03 20 0 0 00 1416 1416 0 0 0 21 4 1 0 0 0 0 0 1 0 4 1 0 1 . 222031 0 1211 0 1 0 233630 242313 1014 02 5DNo4DJ9 1 0 651 8 0 1 41 0 0 04 1 06 0 0 0 57 1213_ 8 0 0 0 20 830 0 0 0 0 1_ 0 1 0 23 0 1 4632 0 1 15 6 0 1 49 3116 1814 0 0 0 0 0 0 0 49 , 1 r 0 0 0 0 , , 0 0 0 5DJ8 0 0 . 0 0 . _ 0 . . 0, , 7 , 1 1 1 3 11 51 19 6 7 r 0 0 0 0 49 0 0 28 0 0 0 0 0 0 0 0 0 0 ' . 6DJ86DNo 101 2 80 _ 2411 - 0 1 r 20 . 20 0 1 02 3 1 , 0 00 0 78 16 6 0 . 0 1 0 1 0 1 0 0_0 r 01 00 2 1 0 0 0 1 3517 01 12 2 01 00 2441 21 8 19 1 0 1 7DOc6DJ9 O0 O0 341 12 4 1 0 0 0 1 00 0 2 1 0 1 0 0 r 15 2215 0 0 21 0_2 2 1 0 0 0 50_ 0 00 01 0 1413 0 67 03 0 21 12 9 47 0 7DFe7DDe7DNo 0 1 0 1 21 2210 02 01 0 00 0_0 2_ 01 0 00_ 0 2__4 14 6_ 0_ 0 0_0 40 0_ 1 0_3 0_ 0 0_0 0_0 0_ 0 01 0_0 01_ 0 1 42,44 0_0 12_ 0_ 0 12 1 0 44_46 17 25 16 8 0 Mc Mc Mc Mc Mc sample Elcf Goco 4 Goal 8 Gocr Gocu 13 Gowi woGo Go 11 Kabr 3 Mccm cmcf Mcdc dccf Mcdf Mcem Mcop Mcpa 1 Mc 9 Mc19 Mc20 20cf Mc21 1 Mc22 Mc29 Mc30 Mc31 32cf Mc33 34cf Mc36 Mc39 1DNo 0 , 2 . 0` 0 4 0 3 0 0 0 0 0 2211 6 0 0 9 0 0 9 0 0 0 0 0 0 0 0 . 7 0 10 2 3 3 17 1DJ91DJ8 0 1 9 0 18 5 0 3 0 1 23 0 0 0 0 0 1 . 28 9 03 0 0 12 0 0 0 374 0 0 00 0 0 0 00 2DNo 0 0 1 6 01 0 41 , 0 _.,0 0 0 0 25 10 0 0 19 2 0 0 1 0 01 0 0 1 0 0 0 2DJ92DJ8 0 3 10 48 1 10 0 2 0 03 0 0 0 0 0 2422 11 4 0 0 0 1910 02 01 0 29 0 0 0 0 0 01 0 3DNo3DOc 0 0 1 11 8 0 1 2014 8 0 765 0 20 0 0 0 0 03 2326 11 2 231 0 0 111418 20 001 0 793 0 0 0 0 0 0 0 3DFe3DDe 01 2 1 7 20 24 8 0 3 20 , 0 1 0 0 0 0 00 262518 12 84 0 0 0 20 0 1 0 1 6 00 0 0 01 0 0 0 3DAp3DMr 0 0 6 1 13 0 4 0 0 0 0 0 00 031 29 9 , 0 0 0 2518 0 0 0 1 11 6 0 0 1 0 0 0 0 1 0 3DMy 0 24 39 1 17 8 0 2 00` 21 0 0 0 0 0 2615 92 3 0 0 . 13 _ 0 0 0 57 0 0 02 0 0 0 . 0 3DJ83DJn 01 31 87 0 2011 0 0 0 0 0 0 0 0 0 13 7 0 0 0 . 20 0 0 0 57 0 0 0 0 0 2 0 3DJ9 1 _ 0 11 0 , 26 0 4 0 0 0 . 0 0 25 86 20 0 0 18 30 00 0 3 00 . 0 0 0 01 0 0 4DJ84DNo 00 31 79 20 12 5 0 4 0 . 23 0 01 0 0 0 1 2415 7 0 0 0 2019 1 1 _ 0 37 0 0 0 0 0 0 0 , 7 2 0 21 , 21 2 0 0 , i 0 0 0 0 0 0 0 0 0 5DNo4DJ9 00 1 9 30 1 6 0 4 0 0 0 0 3 0 . 0 9 43 0 2 0 . 7 21 , 30 0 3 0 0 1 0 0 0 0 1 0 6DNo5DJ8 0 03 56 7 15 74 0 . 20 00 01 001 0 0 0 1 0 2348 48 0 1 20 70 2217 8 21 1 0 63 01 0 0 0 00 1 0 1 6DJ8 0 1_ 0 , 0 0 0 0 0 0 0 47 1 0 0 1 1 0 0 0 0 0 0 0 7DOc6DJ9 0 031 63 20 84 0 70 0 0 0 1 0 0 0 0 3112 12 0 01 0 1311 0 0 0 _ 11 8 0 0 0 0 0 , 0 . 0 7DFe7DDe7DNo 0 2 1 246 21 19 46 20 34 1 0 221 03 0 0 0 01 572920 1314 9 0 02 0 2021 20 021 0 72 0 0 00 0 0_0 00 00 206
Ñ2oMoNOOo00.000Ooo"00^+¡OOOo00O O"I N-4-100N00-400000000000000000000 Ñ 000000000 E In 0000000O0000000000 Ñc,OONOOOOOO N-4ONO+cn-O-+OOONOO CL. a)--1ON0\-4M--INl/')Nin00N\OONNNd-d-NinvOcnd-d-Z4N F-+ T 3NN-4-4NN(VMNNNN--1N-IN-4"4NN-+"I--+ 1+N--+M L L c d -4O-4e tMcrl00-4NONNO-4-4inOONOM000d-O E-4 U IZOOOOOOOOONOOOO-4 H od 00000000000000 ,_,ÑoO00N-NO---+00oNNOON00^i00 4- , o00--i o00oO-O`Oo000000000000ocio . _ . N . o yOO\--.OOMNNentin NVOON Q\O\OO 00O\--+Q\ +-+ tt 4~+ MMNNMMNN-4N--4NN'ct^'NNen*1- . ^it -iN--1 . . a OM O QttNM[--OOOOooinkn 00--1OOv1M----+ o .--.4 t .--I t --.1 ,_.., .--I ,N ---, (y-~Ir.i,--r,--i ~1O 4--4 --1 MO00NOre)V-) ce)ti-) O\00ON 0\Mtno000 0000 00 00tt t tO p NO\ 'nOvO NO\ OOtt-,OnnvNO\00Ot 4:4_0-00000000000.00000000000000
_ , EA O 00O --4o0 O\ 00 ".4 71. tÑÑÑNNMO MM ÑNÑ NÑM4^4 -I Ñ 4y o00oON000000--1"4-'r+O-+N0000-400o0 4.,-^4NM00NwtOMO\NNO00ONztni-Md-N000\l/'lMO\VD(Y)M CLN--ien-4N--1-4Nr1NN--44"- 4NNN'--4-"cf.)--(--+"I -+NNN ct) Ot`00NcM cM cYl 00N-1 NNN ztNNM'ztNNNrl NOMd-NN ...,I- O\N00N00N000O\'-,N00NO00NO\ NN0\VO"-1-4NO\O
OC 0000---471'-"4ONOO--(OMOOOOONNOOOOO\O '7I- 4) _ c)oa,a, o 0 0 oa) Eoo ZtitiZtitiOZA1Qti°Óo,\Z°ÓOti\Z°°Z°OZAú,a Cid AA AAAAAAAAA AAAAAA AA AAAAA (I) ,--4--+--4N enM c r ) NN MMMr1MM Mnrz L e ) V1\O\p V O NNNN Al Al Alcf Al Al 5 Al11 Al17 Al18 Al Ar Cd Cd 1 Cd 3 Cd Cd25 Cn Cncf Cn 2 Cn 7 Cn 8 Cn12 Cn15 Cp Cp Cpcfla Cp Cpse Cp 5 Cp r 6BCp Cp6C Cp 15 Cp 17 Cp 18 Cp23 Cp25 Da st 7DMrsample hu 01 in 1 in 2 mu 3 3 6 0 1 0 20 0 tu 1 ca 0 0 0 24 5 18 de 0 de 1 1 1 1 0 0 bi 0 co 0 2 ob 0 21 2 6A63 0 8 01 0 , 24 23 10 0 7DMy7DAp 1 20 20, 45 31 46 0 0 0,0 32 03 0 08 26 31 5 0 0 0 1 0 1 1 0 0 0 0 01 0 0 01 3833 01 10 7 01 0 2533 111519 787 0 3 8 . , 19 9 6 r 0 0 0 0 0 29 0 7DJn 2 0 , 1 0 1 0 0 0 4 0 0 0 0 0 1 0 1 0 0 0 1 1 0 29 0 1 7DJ97DJ8 320 01 25 1 _ 11 4 1 3 0 0 21 13 3 0 0 0 1810 2016 0 0 3 02 4 1 0 0 0 01 0 0 1 411 241916 0 1 11 28 40 0 282522 151914 10 78 00 8DJ98DJ88DNo 0 0 1 03 25 59 320 20 0 0 023 630 0 0 0 12'1617 6 2016 Or0 0 60 00 r 4 0 01 0 0 0 1 0 0 21 2523 0 11 6 00 0 2919 1918 12 6 0 cf Go Go Go Go Go Go Go Ka Mc Mc cf Mc Mc cf Mc Mc Mc Mc Mc Mc Mc Mc cf Mc Mc Mc Mc Mc Mc cf Mc Mc cf Mc Mc 7DMrsample El 0 co 1 al 9 cr 03 cu 5_ wi 0 wo 1 11 0 br 2 cm 0 cm 0 dc 0 dc 0 df 0 em 45 op 10 pa 1 9 0 19 0 20 20 20 3 21 0 22 0 29 4 30 0 31 0 32 0 33 0 34 1 36 0 39 0 7DMy7D A p 0 02 661 3 76 0 76 0 20 01 0Ó 00 0 0 5327 21~ 00 0 0 3526 0 , 0 , 00 43 0 01 0 0 0 0 0 , 7 , 18 7 7DJ87DJn 0 . 2_1 7 0 10 4 0 361 0 0 1 20 _ 0 0 , 0 0 1 4341 6 201 0 0 11 02 0 0 4 0 01 0 001 021 0 0 8DNo7DJ9 0 221 12 3 0 1 10 5 0 8 , 0 . 03 0 0 0 0 0 2634 56 0 02 0 19 6 4 1 0 0 52 0 0 0 0 0 1 0 0 8DJ98DJ8 0 - 2 -- 79 40 - 35 0 - 51 - 0 - 52 - 0 - 0 0 0 00 3237 - 37 - 0 - 0 0 2214 - 24 - 0 0 68 - 0 0 0 - 00 0 - 0 - 0 , sample Mc40 Ms ir Ms pi 2BMs Ms 4 Pr 1 amRo Rolo Rote maTc Tc 1 Tpac Tpca Tpde Tp 3 Tp 5 Tp 8 buTt 7DMr 51 8 r 01 13 0, . 77 2 27 1 01 0 5 23 0 2 0 2 7DMy7DAp 1 ,204 7583 - 1519 - 1 2012 0 100 62 37 3815 00 1 0 . 0 2216 20 02 0 0 7DJn 01 63 16 - 0 12 , 0-118' 85 39 0 1 01 2 21 0 1 0 0 0 8DNo7DJ97DJ8 04 143 4695 252624 . 00- 291713 00 788173 , 08 215224 0 . 0 0 521 281612 01 0 0 01 8DJ98DJ8 0 7288 1513 0 2117 0 109 91 10 6 2216 01 0 0 0 1 1718 0 0 0 0 1 210
Taphonomic and ostracod abundance data for death assemblages.
# # # # # # # cracked/reductioncorroded/yellow/ # # cods/ samplecarapacesvalvesadultsbroken stained abradedopaqueencrusted gram 1DNo 6.493.6 4.5 4.5 0 0 12.2 3.8 11,805 1DJ8 10.3 89.7 8.8 10.8 0 0 12.4 2.1 7,441 1DJ9 11.8 88.2 8.0 10.2 0 0 8.6 4.3 9,913 2DNo 9.390.7 6.7 7.3 0 0 10.7 2.7 6,818 2DJ8 6.393.7 7.0 6.3 0 0 7.6 0.6 12,621 2DJ9 10.8 89.2 7.2 6.0 0 0 7.2 3.0 7,901 3DOc 16.0 84.0 10.0 14.0 0 0.5 8.5 2.0 11,493 3DNo 10.2 89.8 6.8 8.0 0 0 21.6 4.5 7,978 3DDe 11.9 88.1 8.9 12.6 0 0 13.3 2.2 5,955 3DFe 6.094.0 11.2 7.5 0 0 10.4 3.0 3,886 3DMr 8.491.6 11.9 4.2 0 0.7 21.7 7.0 4,778 3DAp 7.892.2 8.7 11.7 0 0 16.5 3.4 8,849 3DMy 7.492.6 9.2 9.8 0 0 17.8 4.3 6,961 3DJn 14.7 85.3 10.7 11.3 0 0 16.0 2.7 8,401 3DJ8 6.893.2 6.8 10.8 0 0 16.9 2.7 7,400 3DJ9 5.994.1 8.1 10.4 0 0 14.8 3.0 7,647 4DNo 4.1 95.9 7.5 8.8 0 0 12.2 0.7 8,664 4DJ8 9.690.4 6.6 10.8 0 0 7.8 0.6 5,590 4DJ9 11.8 88.2 7.2 11.8 0 0 9.8 3.9 7,853 5DNo 11.688.4 5.8 22.6 0 0.6 3.2 0.6 7,459
5DJ8 8.6r 91.4 7.5 18.7 0 0 5.9 2.1 0 6DNo 21.378.7 12.5 22.8 0 0.7 22.8 8.8 2,475 6DJ8 14.7 85.3 4.7 12.0, 0 0 9.3 6.0 5,695 6DJ9 20.479.6 12.1 26.1 0 0 3.2 0.6 612 7DOc 14.0 86.0 10.5 11.9 0 0 16.1 4.9 3,028 7DNo 22.7 77.3 12.1 6.8 0 2.3 23.5 0.0 1,668
7DDe 17.7 82.3 6.8 i 8.2 0 0 7.5 6.8 5,789 7DFe 5.594.5 5.5 12.4 0 0 3.4 6.2 1,688 7DMr 14.3 85.7 13.0 7.1 0 0 8.4 6.5 2,751 7DAp 8.1 91.9 11.4 16.1 0 0 12.8 8.7 1,710 7DMy 17.1 82.91 13.2 10.5 0 0 17.8 8.6 1,291 7DJn 5.894.2 4.4 11.2 0 0 7.3 3.9 3,964 7DJ8 7.892.2 5.9 14.4 0 1.3 13.1 8.5 3,109 7DJ9 10.4 89.6 5.9 17.0 0 0 28.9 8.1 9,558 8DNo 9.091.0 12.4 16.4 0 0 16.9 9.0 2,400 8DJ8 12.5 87.5 6.9 11.8 0 0 11.1 9.7 3,043 8DJ9 7.892.2 4.6 11.8 0 0 7.2 0.7 5,939 211
Grain size distribution for samples from core LT97 -56V. All weights in grams.
depth in 106 µm-63 -106 total dry core (cm)>1 µm 1 mm µm <63 µm weight 0 -1 0.018 0.517 1.500 4.778 6.813 2 -3 0.032 1.535 2.600 7.694 11.861 4 -5 0.022 0.789 2.001 3.651 6.463 6 -7 0.046 1.440 1.743 8.176 11.405 8 -9 0.028 0.640 0.935 3.239 4.842 12 -13 0.008 0.630 0.835 3.970 5.443 16 -17 0.014 1.543 1.015 3.679 6.251 22 -23 0.001 0.322 0.475 2.599 3.397 30 -31 0.006 0.387 0.865 4.019 5.277 38 -39 0.009 0.316 0.710 3.990 5.025 42 -43 0.094 0.557 0.685 7.882 9.218 46 -47 0.002 0.245 1.084 3.207 4.538 50 -51 0.051 0.690 2.776 10.683 14.200 54 -55 0.019 0.664 0.672 3.732 5.087 58 -59 0.001 0.319 1.014 7.329 8.663 62 -63 0.001 0.258 0.634 2.468 3.361 70 -71 0.002 0.524 0.419 2.223 3.168 78 -79 0.001 0.154 0.486 5.053 5.694 82 -83 0.003 0.361 0.879 9.002 10.245 86 -87 0.004 0.497 0.840 4.027 5.368 90 -91 0.013 0.496 0.791 10.462 11.762 94 -95 0.184 1.188 1.158 6.584 9.114 102 -103 0.004 0.332 0.527 5.032 5.895 110 -111 0.004 0.376 1.292 6.455 8.127 118 -119 0.001 0.304 0.720 5.154 6.179 126 -127 0.001 0.415 0.361 4.395 5.172 130 -131 0.007 0.525 1.098 10.131 11.761 134 -135 0.001 0.283 1.617 4.054 5.955 138 -139 0.002 0.524 2.081 10.753 13.360 142 -143 0.000 0.263 0.513 4.556 5.332 146 -147 0.000 0.167 0.351 10.129 10.647 150 -151 0.001 0.167 0.427 6.627 7.222 154 -155 0.002 0.363 0.389 11.769 12.523 158-159 0.000 0.419 0.945 6.841 8.205 162 -163 0.011 0.491 1.382 11.248 13.132 166 -167 0.001 0.828 1.812 5.154 7.795
170 -171 0.000 0.485 r 0.998 10.002 11.485 174 -175 0.007 0.258 1.185 8.337 9.787 178 -179 0.003 0.332 1.131 11.690 13.156 182 -183 0.010 0.407 1.132 8.070 9.619 186 -187 0.007 0.222 0.342 15.636 16.207 190 -191 0.003 1.110 1.582 7.975 10.670 212
depth in 106 pm.63 -106 total dry core (cm)>1 µm 1 mm µm <63 µm weight 198 -199 0.006 0.445 1.821 8.093 10.365 206 -207 0.002 0.217 1.547 6.815 8.581 214 -215 0.001 0.208 1.164 2.403 3.776 222 -223 0.005 0.718 2.346 5.711 8.780 230 -231 0.001 0.370 1.051 6.235 7.657 238 -239 0.002 0.308 1.114 6.623 8.047 242 -243 0.001 0.428 0.001 11.956 12.386 246 -247 0.003 0.313 1.021 5.944 7.281 250 -251 0.006 0.222 0.001 12.228 12.457 254 -255 0.007 0.309 0.820 6.076 7.212 258 -259 0.006 0.359 0.000 14.933 15.298 262 -263 0.000 0.300 1.995 6.068 8.363 266 -267 0.008 0.312 1.607 11.665 13.592 270 -271 0.004 0.355 0.816 6.383 7.558 274 -275 -0.001 0.356 1.856 9.544 11.755 278 -279 0.001 0.331 0.327 7.365 8.024 286 -287 0.005 0.496 1.169 7.781 9.451 294 -295 0.009 0.742 1.607 5.445 7.803 298 -299 0.024 0.737 2.044 8.785 11.590 302 -303 0.004 0.774 0.804 6.525 8.107 306 -307 0.007 1.607 2.160 9.084 12.858 310 -311 0.001 0.846 0.016 5.933 6.796 314 -315 0.023 5.511 2.946 7.223 15.703 318 -319 0.002 2.750 2.115 4.871 9.738 322 -323 0.028 5.216 3.795 6.412 15.451 326 -327 0.006 2.071 4.153 4.084 10.314 330 -331 0.006 7.513 2.277 4.586 14.382 334 -335 0.009 4.359 2.326 2.948 9.642 338 -339 0.002 9.117 1.700 3.577 14.396 342 -343 0.001 9.685 0.980 0.952 11.618 Appendix G: Raw data for fossil assemblages in core MWA -1 collected from 10 m water depth at Mwamgongo, Tanzania. ofFossil intervaldepth assemblage 0 -1 cm shownspecies here. abundance data. See Appendix D for a list of species abbreviations. Sample #1 from resampling Al in core(cm) abAl Alci Alcl huAl muAl 1 Al 5 10cf Al11 3 , Al18 20Al 1 Artu Ar 2 1 Cd24 Cd2513 Cn 2 Cn 8 3 Cn 9 Cn15 Cpbi Cpco Cpse Cp 5 3 6ACp 6CCp 8 Cp15 Cp 18 Cp23 Cp25 Da st Elcf Go 14al Goco Gocr 3 Gocu 16 Gowi woGo 7 0-1 (1) 1-2 0 06 0 2 1 0 0 0 0 0 1 2 02 0 1119 28 01 1 00 0 , 201 60 2 0 2730 0 1 0 28 7 2210 14 9 0 0 8 4 1 , 3 21 , 0 8 3-42-3 0 0 0 1 21 0 0 1 2 0 1 0 22 0 13 9 2319 01 0 0 0 1 0 42 0 2934 20 031 1213 2215 9 1 0 0 171413 543 2 2217 02 42 5-64-5 0 0 0 03 01 0 0 0 1 0 0 1 22 2 1 48 202215 0 01 01 0 0 0 1 421 0 222616 1613 7 6 282024 281115 1013 6 0 0 1316 63 20 262822 0 1014 7 7-86-7 0 0 , 0 02 0 _ 0 1 0 2 0 0 1 0,0 16_ 5 16 0 0 0 00 , 0 1 - 01 4 00 13 11 06 4 22 8 1 0 26 4 1 38 0 8 8-9 0 0 0 1 31 0 2 0 r 0 2 1 0 5 12 0 2 0 0 4 1 0 6 0 23 14 2 14 31 7 0 0 30 2 2 67 0 5 10-119-10 20 0 0 31 0 0 0 25 0 0 0 0 42 17 21 0 01 0 021 0 1 0 43 0 2327 1416 35 1317 8 18 20 0 0 0,42,0 30 21 241 110162 0 10 6 12-1311-12 0 0 0 1 20 0 0 Or31 020 001 00 0 401 1 0 1 0 0 0 0 031_ 05 0 202533 211312 2 13 161812 ,15 13 3 00 201 414932 23 4 1 ,213 206197 0 16 59 15-1614-1513-14 0 0 01 001, - 00, 0 0 0 - 22, - 1 31 11, - 0 - 81 0 1 - 0 1 01 0 - 0 0 1 - 3 2 1 0 - 2722 - 2023 - 32-12 - 1310 1211 10 9 01 - 0 - 2446 -- 2 1 05 - 109110 0 - 10 5 , depth Mc in core (cm) Go11 Kabr Mccm Mccn Mc dc Mc df Mcem Mcop Mcpa Mc 9 Mc20 20cf Mc21 Mc22 Mc29 Mc30 Mc31 Mc 33 Mc36 Mc40 Ms ir Ms pi 2BMs Ms 4 Pr 1 amRo Ro lo Ro te maTc Tc 1 Tpac 0-1 (1) 1-2 0 4 1 0 1 0 00 31 34 8 12 5 0 01 12 6 05 30 0 11 3 20 01 03 0 111010 109 29 7 23 2926 00 5958 1315 15 9 0 2 1 0 1 3-42-3 0 0 2 02 - 0 1 01 2226 1011 0 0 69 0 2 00 12 8 02 0 0 0 02 120 92 2613 31 3532 0 7061 11 6 17 8 01 25 02 5-64-5 0 01 0 0 0 31 3224 - 10 7 0 1 0 12 7 0 0 1 0 14 9 0 0 0 0 01 104134 2217 01 33 0 1 6855 92 12 9 2 1 4 1 0 7-86-7 0 0 0 0 0 03 2410 8 00 0 69 0 23 0 1 17 8 0 0 1 0 0 01 9194 1417 42 3934 0 5074 68 18 7 20 2 1 20 8-9 1 0 0 , 0 0 3 19 14 0 0 2 0 0 0 . 4 . 0 1 0 . 0 0 54 20 0 30 , 2 59 8 9 0 2 0 10-119-10 0 0 1 0 1 0 01 2413 15 8 0 0 01 0 0 0 10 5 0 0 1 01 01 21 3956 10 7 0 2410 0 1 2943 3 12 2 0 0 1 0 11-12 0 0 0 0 0 1 6 . 1 0 0 3 0 0 0 _ 14 0 0 0 0 0 43 6 0 9 0 19 0 1 0 0 2 13-1412-13 0 0 0 0 0 0 96 12 9 0 1 0 23 0 3 0 8 0 0 1 0 1 0 01 3733 1312 0 10 8 0 10 8 01 20 0 00 01 15-1614-15 0 0 0 00 0 20 1614 1011 0 1 0 96 0 0 1 0 2019 0 0 1 01 0 30 7165 1417 0 1112 00 2824 20 13 6 0 0 0 in coredepth(cm) Tpca Tpde Tp 3 Tp 5 Tp 8 buTt 0-1 (1) 2-31-2 042 232218 01 0 20 263 4-53-4 20 232629 31 0 0 06 7-86-75-6 24 342T , 0 0 0 480 9-108-9 04 3028 31 01 0 0 1 12-1311-1210-11 042 445531 01 0 1 0 01 15-1614-1513-14 340 435137 00 21 0 421 Fossil species abundance data for multiple samples from interval 0 -1 cm in core MWA -1. See Appendix# of E for a list of species abbreviations. Al Alcf Al Alcf Al Al Al Ar Ar Cd Cd Cn Cn Cn Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp Da Go Go Go Go Go Go Ka count 1 hu 2 in 6 mu 1 10 0 11 3 18 0 20 1 tu 0 2 1 24 19 25 13 2 0 1 8 3 1 9 0 bi 0 co 0 se 2 5 3 6A 30 6C 8 15 0 1828 23 10 25 9 st 0 al14 8 co 41 cr 3 cu 16 wi 0 wo 78 br 4 1 32 21 0 201 01 20 0 1 02 2 0 1311 9 2819 0 1 0 0 2 1 06 42 0 2734 01 03 1213 7 2215 14 9 0 1413 54 2 2117 02 24 0 4 0 , 2 , 0 1 . 0 2 0 8 23 0 0 0 2 0 29 20 1 22 9 17 3 22 10 5 3 0 , 16 7 6 0 0 22 0 0 0 0 1 0 , 0 0 2 2 1 22 0 0 0 0 1 2 , 0 . 24 28 _ 1 876 20 0 0 1 0 20 1 . 00 01 2 1 0 16 45 201615 00 0 1 01 001 0 1 41 00 222613 111613 60 2820 4 221511 1013 8 0 1 261316 463 201 382628 0 14 78 0 # of Mc Mc Mc Mc Mc Mc Mc Mc Mc Mc cf Mc Mc Mc Mc Mc Mc Mc Ms Ms Ms Ms Pr Ro Ro Ro Tc Tc Tp Tp Tp Tp Tt count 1 cm 1 cn 0 dc 0 df 1 em 8 op 5 pa 9 20 20 5 21 0 22 0 29 3 30 0 31 1 33 3 40 1 101 ir pi29 2B 3 264 1 0 am 58 lo 15 te 15 ma 0 1 1 ac 1 ca 2 de 18 3 0 Tp 8 0 bu 2 2 0 , 0 0 3 . 34 . 12 . &0120 1 . 6 0 . 3 0 11 2 0 0 0 109 7 2 29 0 59 13 9 0 2 , 0 4 22 0 0 3 3 2 2 0 1 26 . 11 0 0 9 0 2 0 8 2 0 0 2 120 13 3 . 32 0 61 6 8 0 5 2 0 23 1 2 6 1 , 4 2 0 0 22 10 0 0 6 0 2 0 12 0 0 0 0 92 26 1 35 0 70 11 17 1 2 0 0 29 1 0 , 6 65 0 0_0 00 31 . 3224 10 7 0 1 00 12 7 00 0 1 0 14 9 0 0 0 0 1 104134 2217 01 3333 0 1 6855 92 12 9 21 41 0 42 2326 31 00 40 87 00 0 0 , 03 2410 8 0 0 69 0 23 01 17 8 0 01 00 01 9194 1417 42 3934 0 5074 86 18 7 02 21 20 22 3427 0 1 0 08 Taphonomic and ostracod abundance data for MWA -1 fossil assemblages. Sample #1 from recount used for 0 -1 cm. core (cm)depth 0in -1 (1) carapaces # 10.4 # valves 89.6 # adults 11.4 cracked/broken # 13.5 reduction stained # 0 corroded/abraded # 1.0 # yellow/opaque 16.1 encrusted # 5.7 # cods/ gram 4,494 32 -3-41 -2 15.213.311.5 84.886.788.5 20.415.913.2 r 12.9 4.67.4 0 0.4 0 21.621.714.9 4.25.41.4 2,8463,5402,630 7 -8654 -7-5-6 13.216.911.2 6.0 86.883.194.088.8 21.015.812.3 8.3 13.210.4 8.89.3 0 0.40.7 0 22.422.119.517.2 5.94.62.96.3 2,4503,8233,4683,609 1110 -12 -119 -10 8 -9 16.712.710.1 8.6 91.487.389.983.3 13.011.0 9.28.8 11.714.210.6 8.2 0 1.0 0 31.327.016.618.2 2.33.32.41.9 5,7685,8653,3905,231 13121514 -13-14 -15-16 15.610.110.2 8.8 91.384.489.989.8 21.413.811.4 8.7 14.116.216.1 9.0 0.3 0 0.6 0 29.728.638.934.9 5.63.62.71.9 2,9917,0145,4085,014 Grain size distribution for core MWA -1. All weights in grams. in coredepth(cm) >lm m 106 µm- lmm 63 -106 µm <63µm weighttotaldry 320 -1-3-41 -2 0.530.880.550.21 4.9535.3935.8487.218 2.4324.5222.3421.849 0.570.440.680.58 13.1898.2289.0758.494 7654 -7-5-8-6 0.940.721.62 0.7 6.6346.9245.3716.472 2.1592.7493.0822.207 0.680.870.670.72 11.24510.12211.092 10.06 1110 -11-129 -108 -9 0.653.151.461.23 10.05911.05510.789 9.813 2.2812.9882.3941.915 0.530.360.510.66 14.02515.63714.964 15.22 15141312 -15-13-14-16 0.720.841.121.02 10.3466.2519.8458.632 2.7931.6821.9941.974 0.810.440.350.29 13.21313.44813.3559.091 Appendix H: Raw data for fossil assemblages in core MWA -2 collected from 5 m water depth at Mwamgongo, Tanzania. Fossil assemblage species abundance data. See Appendix D for a list of species abbreviations.Al Al Al Al Al Al Al Ar Ar Cd Cd Cd Cd Cn Cn Cn Go Go Go Go Go Go depth in cl incf 5 10 tu 2 3 2 8 15 Cp Cp Cp Cp Cp 5 Cp Cp Cp15 Cp18 Cp Cp al , core (cm) 0-11-2 ab 0 1 0 1 hu 00 0 mu 52 0 0 1 2 1 0 ca 0 0 24 10 3 25 1718 01 03 0 bi 01 co 02 . ob 0 1 se 0 1 . 03 6A4530 . 6C 12 0 _ 41 20 1 23 15 7 25 79 10 8 co 1 cr 0 cu 1510 wi 0 wo 03 3-42-3 0 0 01 0 1 0 01 0 2 1 20 0 0 5 1 14 7 01 0 0 0 0 0 0 1 0 36 21 5 01 1613 20 5 68 1911 64 031 1713 0 60 1 5-64-5 0 0 . 0 1 0 0 1 - 0 0 0 1 20 0 1 0 1 54 16 9 0 0 1 03 01 0 0 0 00.28 35 78 2 1619 89 3 12 7 2 1 1 2511 0 4 1 7-86-7 0 0 01 0 0 01 0 03 0 02 0 546 1314 7 0 0 02 30 00 0 0 0 292628 09 15 1 252010 1310 9 421 1813 23 001 2915 0 3 10-119-108-9 01 0 0 . 0 02 01 0 0 0 0 1 0 . 43 15 6 0 _ 0 1 0 01 . 021 0 0 0 2827 72 01 1417 8 53 15 9 23 0 1817 0 1 6 depth in Go 11 Ka Mc Mccf Mc Mc Mc Mc Mc17 Mc19 Mc Mc21 Mc Mc Mc Mc Ms Ms Ms Ro Ro Ro Tc Tc 1 Tpac Tp Tp Tp 3 Tp 5 Tp 8 Tt , 33 36 ir 4 am te ma de bu core (cm) 0-1 0 br 1 cm 3 cm 0 . df 0 em 11 op 3 pa 0 1 1 20 19 0 29 5 0 0 , 40 0 178 pi13 6 89 lo 14 24 4 0 1 ca 0 , 27 0 0 _ 0 0 2-31-2 0 0 1 0 0 0 1 _ 15 98 0 04 0 11 7 01 58 01 01 02 150151 2125 4024 6263 27 12 5 0 1 0 1 0 0 1 3430 0 01 0 02 4-53-4 0 2 1 201 0 023 2113 3 1 0 0 0 1011 7 2 10 8 03 0 0 177192 2216 8 2024 4154 5 2315 7 0 0 1 4 1 20 2623 00 00 00 05 6-75-6 0 1 0 01 2 2314 62 0 0 0 8 2 1 1012 0 20 0 162189 32 3024 5256 . 9 9 0 0 02 0 1 2833 001 0 0 30 8-97-8 0 1 0 30 . 0 03 96 10 9 0 1 0 0 1316 0 1 13 7 0 02 0 1 198178 2413 2631 6140 2 45 0 1 , 0 0 1 , 2132 r 2 00 01 2 10-119-10 - 0 0 0 - 0 - 03 9 - 27 - 0 - 0 - 0 -- 89 - 0 14 5 - 0 - 0 0 - 223228 3123 18 - 4250 - 32 37 0 0 1 - 0 20 2220 - 0 - 0 00 - 30 , Taphonomic and ostracod abundance data for MWA -2 fossil assemblages. * denotes missing data. depth in # # # cracked/ # reduction # corroded/ # # yellow/ # # cods/ core (cm) 20 -1-31 -2 carapaces 11.410.8, valves 88.689.2 adults 9.28.9 broken 12.2 8.5 * stained 0 * abraded 0 * opaque 13.817.1 * encrusted 5.17.3 * gram 2,6341,0301,130 6543 -7-5-4-6 6.54.67.96.0 93.595.492.194.0 13.716.8 9.97.5 13.711.510.914.2 1.6 0 0 23.820.918.518.3 11.315.315.8 6.7 581823660596 10 -119 -10 87 -9-8 13.810.0 9.78.5 90.390.091.586.2 15.111.4 9.56.4 20.4 9.36.97.1 2.10.90.0 0 0 38.629.325.534.4 12.912.114.017.9 661557744729 Grain size distributiondepth in for core MWA -2. >1 106 µm- 63 -106 <63 total dry core (cm) 20 -1-31 -2 0.0330.0730.166mm 1 mm 7.6668.0685.547 2.8882.7383.411µm 0.8300.8430.803µm weight11.41711.7229.927 6543 -7-5-4-6 0.1900.0990.1740.081 8.6049.5629.4599.586 2.6343.0722.1043.944 0.6100.6620.6630.805 12.09113.53812.34714.273 10 -119 -107 -88 -9 0.1410.1380.2020.181 7.1156.9146.7188.438 2.8322.7082.6023.207 0.6560.6200.7220.698 12.52410.34710.58110.440 Appendix I: Raw data for core MIT -1 collected in 15 m water depth at Gombe Stream National Park, Tanzania. Species abundance data for fossil assemblages in MIT -1. See Appendix D for a list of species abbreviations.Al Al Al Al Al Al Al Al Al Ar Ar Cd Cd Cd Cd Cn Cn Cn Cn Cn Cn Go depth in ab cl hu in mu 5 11 18 20 2 23 25 cf 2 7 8 16 , Cp , Cp Cp , Cp Cp Cp Cp15 , Cp18 Cp Cp25 Da cf Go al Go core (cm) 0-1 0 0 1 21 0 3 1 2 0 31 tu 7 21 ca 1 0 , 2414 7 de 0 de 0 0 0 2 1 0 bi 0 co 0 la 0 se 6 6A 17 6C 8 3 19 2315 14 st 0 El 0 8 co 4 cr 3 2-31-2 0 0 3 60 234 0 1 320 0 0 53 0 01 0 1318 10 5 0 0 0 1 0 2 0 0 0 0 24 2930 0 0 2018 1414 324 01 0 11 8 1514 9 0 4-53-4 0 0 02 1 30 3 0 01 0 0 1 23 00 0 0 18 69 0 02 31 0 0 1 0 02 0 0 , 32 3335 74 1 301 2431 1115 7 6 , 0 0 1110 8 11 8 0 5-6 0 0 1 2 0 0 1 0 01 0 89 4 0 0 0 1 0 0 0 0 0 34 43 , 11 14 0 21 0 5 2 5 18 3 1 14 6-7 0 0 0 2 0 0 0 3 0 , 84 , 0 0 0 1 , 0 0 0 0 1 0 3 58 2 21 0 9 2 7-8 0 0 0 0 4 5 0 0 0 2 . 0 . 0 12 0 0 0 2 1 0 01 0 30 0 27 16 7 0 0 10 10 0 3 3 3 . 8-9 0 0 5 32 0 0 0 0 , . 0 4 0 0 0 0 0 1 0 , 0 01 0 2 31 56 21 18 18 1110 0 0 20 11 7 0 10-119-10 0 0 1 60 5 32 , 0 _ 0 1 0 30 30 01 02 1 . 0 536 231 01 0 21 0 30 0 30 0 2 574 332632 31 23 4239 211418 11 00 0 212013 78 20,1 11-12 0 4 2 . 0 0 , 0 - 2 . 1 0 0 1 0 1 0 0 01 34 9 0 0 1 13-1412-13 0`0 _ 0 02 . 30 5,2 02 0 0 , 0 . 01 02 0 0 06 02 1 0 01 0 0 01 01 01_ 0 0 2 3128 83 21 293914 2320 7 0 0 1713 98 01 2 2 . 9 0 0 , 0 28 6 , 9 14-15 0 0 2 1 0 1 0 0 1 0 1 1 0 , 0 0 0 0 0 0 0 0 1 20 0 15-16 ' 1 0 1 0 . 1 0 1 1 _ 0 0 0 1 0 50 _ 0 0 0 . 0 . 1 0 0 0 0 0 . 4130 9 0 1914 85 . 10 8 0 16 1218 01 18-1917-1816-17 0 . 0 0 _ 00 . 0 _ 0 0 . 0 1 30 1 02 - 0 - 0 31 301 , 43 0 02 001 0 0 00 0 0 0 031 5046 079 40 2515 1817 69 0 -- 0 1 262825 10 8 0 Mc Mc Mc core (cm)depth in Gocu Godo Gowi woGo 15 Go 11 1 Kabr Mccn Mcdf Mcem 9 Mcop Mcpa sucf 11cf Mc 17 1 Mc20 10 20cf Mc21 1 Mc29 Mc31 Mc33 1 Mc36 Ms ir Mspi Ms2B 1 Ms 412 Ms 9 Pr 1 amRo Ro 14lo Rote maTc Tc 1 0-11-2 5329 0 0 12 1 0 0 0 11 36 , 021 0 0 0 4 0 1 2617 0 . 0 02 6755 4457 4 13 02 201 . 4547 10 62 02 0 1 . 3-42-3 4630 0 0 1114 0 0 1 0 0 201613 4235 , 31 0 0 02 6 0 0 21 0 21 0 1 6264 4544 4 1 1311 0 0 5364 9 21 0 0 5-64-5 41 0 0 1116 0 1 0 0 7 , 3335 2 0 0 0 12 32 01 28 0 24 0 20 0 0 6053 4452 36 15 0 02 5068 15 63 24 1 0 0 9 , 6-7 45 0 0 0 0 0 0 42 0 1 0 0 0 0 0 , 29 0 0 1 2 53 26 26 0 0 55 0 0 7-88-9 4365 01 0 161117 0 0 02 0 1 _ 13 9 2831 0 00 0 0 13 5 321 04 1918 5 0 0 0 1 5455 2938 82 2016 0 2 1 5060 67 31 0 1 0 10-119-10 9978 0 0 , 10 0 0 0 0 89 2725 01 0 0 0 79 0 1 0 12 21 0 0 4834 3534 34 24 0 0 1 4452 4 4 0 0 11-12 90 0 0 15 0 01 . 0 , 0 5 13 0 0 0 4 0 7 . 0 0 52 28 16 0 47 . 4 0 0 0 12-13 79 0 . 0 1814 0 , 01 0 12 , 1719 0 0 0 0 12 50 0 10 5 0 0 48 39 3 2617 0 0 40 11 3 2 0 0 14-1513-14 8986 00 0 16 0 0 1 0 _ 65 17 . 0 01 0 20 97 0 02 19 6 0 0 04 4939 5136 0 1 29 0 01 5738 - 1 21 0 0 15-16 81 0 . 0 15 2 0 2 0 7 15 0 0 0 0 11 0 2 15 0 0 _ 0 67 51 , . 20 0 4 . 26 5 0 0 18-1917-1816-17 364350 50 230 151214 _ 00_ , 0 30 001 1412 0 251417 0 1 0 0 02 2510 02 1 45 1 311716 0 0 0 554936 283750 021 191817 01 53-2 4537 530 - 24 01 0 core (cm)depth in 0-1 Tpac 0 Tpca 7 Tpde19 Tp 3 1 Tp 4 0 Tp 5 1 Tp 6 0 Tp 8 0 buTt 5 2-31-2 01 272 232615 201 r 0 0 . 0 1 21 679 4-53-4 0 5 17 0 1 0 0 00 0 41 , 7-86-75-6 03 1 10 57 262216 11 0 0 002 020 51 9-108-9 11 203 2329 01 0 00 0 0 018 10-11 24 0 0 0 1 1 1 13-1412-1311-12 021 526 382530 0 1 0 00 1 01 46 . 15-1614-15 1 20 4439 2 1 20 00 1 0 95 18-1917-1816-17 2 1 46 516065 043 0 00 01 0 70 1 Grain size distribution for samples from core MIT -1. All weights in grams. depth in 106 µm- 63 -106 <63 total dry core (cm) 20 -1-31 -2 >1 mm 0.3680.3180.074 12.84010.2461 mm14.023 2.4632.3092.382µm , 0.5230.3310.436µm weight 16.19416.98113.138 654 -7-5-63 -4 0.4160.3450.2501.067 13.14315.87113.62816.044 2.7982.9443.4671.707 0.4090.3100.4720.535 20.12316.81116.90419.568 10 -119 -10 87 -8-9 0.5830.9110.8161.884 10.74417.95410.198 8.546 2.5881.9281.6471.357 0.3110.2290.5590.206 22.01211.36814.50412.577 14131211 -15-13-14 -12 2.0631.3851.0241.137 13.55511.31815.081 8.512 2.0902.3121.6151.078 0.3440.2290.5230.358 17.19215.34011.20418.888 18171615 -18-17-19-16 11.604 8.3983.2703.360 10.261 5.0443.9506.585 0.6891.4261.4701.596 0.2610.1780.4870.248 18.33513.21511.93815.339 228
Appendix J: Smear slide data from deepwater cores from Gombe Stream National Park and Mwamgongo, Tanzania.
Core LT98 -58M (Gombe)
depth in % % % % core (cm)siliciclasticorganiccarbonatesiliceous 3 -4 49.0 50.5 0.5 0 6 -7 51.5 47.5 1 0 9 -10 53.5 46.0 0.5 0 12 -13 64.5 35.5 0 0 15 -16 48.0 52.0 0 0 18 -19 55.0 44.5 0.5 0 21 -22 52.0 47.5 0 0.5 24 -25 48.5 51.0 0.5 0 27-28 64.5 34.5 0.5 0.5
30 -31 57.5 41.0 1 0.5 33 -34 52.5 44.0 2 1.5 36 -37 61.0 37.5 1.5 0
37 -38 62.5 36.0 1 0.5
Core LT98 -37M (Mwamgongo)
depth in % % % % core (cm)siliciclasticorganiccarbonatesiliceous 3 -4 61.5 37.5 0.5 0.5 9 -10 54 45.5 0 0.5 12 -13 59.5 40.5 0 0 15 -16 50.5 49 0 0.5 18 -19 47.5 52.5 0 0 21 -22 61 39 0 0 24 -25 59 41 0 0 27 -28 66.5 33.5 0 0 30 -31 75 25 0 0 33 -34 62.5 37 0.5 0 36 -37 66.5 33.5 0 0 39 -40 67 33 0 0
42 -43 57.5 41 0.5 1 43 -44 54.5 44.5 0.5 0.5 229
REFERENCES
Alin, S. R., and Cohen, A. S., 2001, Late Holocene records ofinvertebrate diversity and turnover from Lake Tanganyika, East Africa, and theirconservation implications: North American Paleontological Convention 2001 Program and Abstract, v. 21, p. 22.
Alin, S. R., Cohen, A. S., Bills, R., Gashagaza, M. M., Michel, E., Tiercelin, J. J., Martens, K., Coveliers, P., Mboko, S. K., West, K., Soreghan, M., Kimbadi, S., and Ntakimazi, G., 1999, Effects of landscape disturbance on animal communities in Lake Tanganyika, East Africa: Conservation Biology, v. 13, no. 5, p. 1017 -1033.
Alin, S. R., Cohen, A. S., Palacios -Fest, M., Msaky, E., Dettman, D., and McKee, B., 2000, A multi -indicator study of watershed disturbance and paleoecological change in Lake Tanganyika, East Africa, through c. 500 years: Eos, Transactions, American Geophysical Union 2000 Fall Meeting, v. 81, no. 48, p. F263.
Anderson, N. J., 1993, Natural versus anthropogenic change in lakes: the role of the sediment record: Trends in Ecology and Evolution, v. 8, p. 356 -361.
Anderson, N. J., and Battarbee, R. W., 1994, Aquatic community persistence and variability: a palaeolimnological perspective, in Giller, P. S., Hildrew, A. G., and Raffaelli, D. G., eds., Aquatic Ecology: Oxford, Blackwell Science.
Bengtsson, L., and Enell, M., 1986, Chemical analysis, in Berglund, B. E., ed., Handbook of Holocene Palaeoecology and Palaeohydrology: Chichester, UK, John Wiley and Sons, p. 423 -450.
Beuning, K. R. M., Talbot, M. R., and Kelts, K., 1997, A revised 30,000 -year paleoclimatic and paleohydrologic history of Lake Albert, East Africa: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 136, p. 259 -279.
Birkett, C., Murtugudde, R., and Allan, T., 1999, Indian Ocean climate event brings floods to East Africa's lakes and the Sudd Marsh: Geophysical Research Letters, v. 26, no. 8, p. 1031 -1034.
Bootsma, H. A., and Hecky, R. E., 1993, Conservation of the African Great Lakes: a limnological perspective: Conservation Biology, v. 7, p. 644 -656.
Brown, D., 1994, Freshwater Snails of Africa and their Medical Importance: London, Taylor and Francis.
Cohen, A.S., Bills, R., Cocquyt, C.Z., and Caljon, A.G., 1993, The impact of sediment pollution on biodiversity in Lake Tanganyika: Conservation Biology, v. 7, p. 667- 677. 230
Cohen, A. S., Dussinger, R., and Richardson, J., 1983, Lacustrine paleochemical interpretations based on eastern and southern African ostracods: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 43, p. 129 -151.
Cohen, A. S., Lezzar, K. E., Tiercelin, J. J., and Soreghan, M., 1997a, New palaeogeographic and lake -level reconstructions of Lake Tanganyika: implications for tectonic, climatic and biological evolution in a rift lake: Basin Research, v. 9, p. 107- 132.
Cohen, A. S., Palacios -Fest, M. R., Dettman, D., Msaky, E., Livingstone, D., and McKee, B., 1999, Lake Tanganyika Biodiversity Project Special Study on Sediment Discharge and its Consequences: Paleolimnological Investigations, Lake Tanganyika Biodiversity Project.
Cohen, A. S., Talbot, M. R., Awramik, S. M., Dettman, D. L., and Abell, P., 1997b, Lake level and paleoenvironmental history of Lake Tanganyika, Africa, as inferred from late Holocene and modern stromatolites: Geological Society of America Bulletin, v. 109, no. 4, p. 444 -460.
Coulter, G. W., 1991, Lake Tanganyika and its Life: Oxford, United Kingdom, Oxford University Press.
Coulter, G. W., 1994, Lake Tanganyika: Ergebnisse der Limnologie, v. 44, p. 13 -18.
Coulter, G. W., and Spigel, R. H., 1991, Hydrodynamics, in Coulter, G. W., ed., Lake Tanganyika and its Life: Oxford, Oxford University Press, p. 49 -75.
Dean, W. E., 1999, The carbon cycle and biogeochemical dynamics in lake sediments: Journal of Paleolimnology, v. 21, p. 375 -393.
Evert, M. J., 1980, Le lac Tanganyika, sa faune, et la peche au Burudi: Bujumbura, Burundi, Universite du Burundi.
Finney, B. P., and Johnson, T. C., 1991, Sedimentation in Lake Malawi (East Africa) during the past 10,000 years: a continuous paleoclimatic record from the southern tropics: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 85, p. 351 -366.
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, no. 169 -197. 231
Haug, G. H., Hughen, K. A., Sigman, D. M., Peterson, L. C., and Röhl, U., 2001, Southward migration of the Intertropical Convergence Zone through the Holocene: Science, v. 293, p. 1304 -1308.
Kidwell, S. M., and Flessa, K. W., 1995, The quality of the fossil record: populations species, and communities: Annual Review of Ecology and Systematics, v. 26, p. 269- 299.
Levin, S. A., 1992, The problem of pattern and scale in ecology: Ecology, v. 73, no. 6, p. 1943 -1967.
Liu, S. X., 1999, Ostracode abundance analysis of Lake Issyk -Kul, Kyrgyzstan, Central Asia [Master's thesis]: University of Minnesota, 80 p.
Magurran, A. E., 1988, Ecological Diversity and Its Measurement: Princeton, Princeton University Press.
Martens, K., 1985, Tanganyikacypridopsis gen.n. (Crustacea, Ostracoda) from Lake Tanganyika: Zoologica Scripta, v. 14, no. 3, p. 221 -230.
Martens, K., 1994, Ostracod speciation in ancient lakes: a review: Ergebnisse der Limnologie, v. 44, p. 203-222.
Michel, A. E., Cohen, A. S., West, K., Johnston, M. R., and Kat, P. W., 1992, Large African lakes as natural laboratories for evolution: examples from the endemic gastropod fauna of Lake Tanganyika: Mitt. Internat. Verein. Limnol.,v. 23, p. 85 -99.
Minchin, P. R., 1987, An evaluation of the relative robustness of techniques for ecological ordination: Vegetatio, v. 69, p. 89 -107.
Negrini, R. M., Erbes, D. B., Faber, K., Herrera, A. M., Roberts, A. P., Cohen, A. S., Wigand, P. E., and Foit, F. F., 2000, A paleoclimate record for the past 250,000years from Summer Lake, Oregon, USA: I. Chronology and magnetic proxies for lake level: Journal of Paleolimnology, v. 24, p. 125 -149.
Nicholson, S. E., 1998a, Fluctuations of rift valley Lakes Malawi and Chilwa during historical times: a synthesis of geological, archeological and historical information, in Lehman, J. T., ed., Environmental Change and Response in East African Lakes: Dordrecht, Kluwer Academic.
Nicholson, S. E., 1998b, Historical fluctuations of Lake Victoria and other lakes in the northern rift valley of East Africa, in Lehman, J. T., ed., Environmental Changeand Response in East African Lakes: Dordrecht, Kluwer Academic,p. 7 -36. 232
O'Reilly, C., in prep, Using carbon and nitrogen isotopes in lake sediments todetect land use change.
O'Reilly, C. M., 1999, Benthic algal productivity in Lake Tanganyika and the effects of deforestation: UN /GEF Lake Tanganyika Biodiversity Project (RAF /92/G32).
Park, L. E., and Martens, K., 2001, Four new species of Gomphocythere from Lake Tanganyika, East Africa (Crustacea, Ostracoda): Hydrobiologia, v. 450, p. 129 -147.
Penney, J. T., and Racek, A. A., 1968, Comprehensive Revision of a Worldwide Collection of Freshwater Sponges (Porifera: Spongillidae): Washington, D.C., Smithsonian Institution Press.
Plisnier, P. -D., 2000, Recent climate and limnology changes in Lake Tanganyika: Verh. Internat. Verein. Limnol., v. 27, p. 1 -4.
Poll, M., 1986, Classification des Cichlidae du lac Tanganika: Tribus, genres et espèces: Brussels, Académie Royale de Belgique.
Rome, D. R., 1962, Exploration Hydrobiologique du Lac Tanganyika: Brussels, Belgium, Institut Royal des Sciences Naturelles de Belgique, 305 p.
Sall, J., and Lehman, A., 1996, JMP Start Statistics: A Guide to Statistics and Data Analysis Using JMP and JMP IN Software: Belmont, California, Duxbury Press, 521 p.
Stager, J. C., Cumming, B., and Meeker, L., 1997, A high- resolution 11,400 -yr diatom record from Lake Victoria, East Africa: Quaternary Research, v. 47, p. 81 -89.
Stager, J. C., and Johnson, T. C., 2000, A 12,400 14C yr offshore diatom record from east central Lake Victoria, East Africa: Journal of Paleolimnology, v. 23, p. 373 -383.
Steadman, D. W., 1995, Prehistoric extinctions of Pacific Island birds: biodiversity meets zooarchaeology: Science, v. 267, p. 1123 -1131.
Stuiver, M., Reimer, P. J., Bard, E., Beck, J. W., Burr, G. S., Hughen, K. A., Kromer, B., McCormac, F. G., van der Plicht, J., and Spurk, M., 1998a, INTCAL98 Radiocarbon age calibration 24,000 - 0 cal BP: Radiocarbon, v. 40, p. 1041 -1083.
Stuiver, M., Reimer, P. J., and Braziunas, T. F., 1998b, High -precision radiocarbon age calibration for terrestrial and marine samples.: Radiocarbon, v. 40, p. 1127 -1151.
Sturmbauer, C., and Meyer, A., 1992, Genetic divergence, speciation and morphological stasis in a lineage of African cichlid fishes: Nature, v. 358, p. 578 -581. 233
Talbot, M. R., in press, Nitrogen isotopes in palaeolimnology, in Last, W., and Smol, J., eds., Tracking Environmental Change Using Lake Sediments: Physical and Chemical Techniques, Kluwer Press.
Talbot, M. R., 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 and Planetary Science Letters, v. 110, p. 23 -37.
Talbot, M. R., and Laerdal, T., 2000, The Late Pleistocene -Holocene palaeolimnology of Lake Victoria, East Africa, based upon elemental and isotopic analyses of sedimentary organic matter: Journal of Paleolimnology, v. 23, p. 141 -164.
Taylor, D. M., 1990, Late Quaternary pollen records from two Ugandan mires: evidence for environmental change in the Rukiga Highlands of southwest Uganda: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 80, p. 283 -300.
ter Braak, C. J. F., and Prentice, I. C., 1988, A theory of gradient analysis: Advances in Ecological Research, v. 18, p. 271 -317.
Thompson, R., and Oldfield, F., 1986, Environmental Magnetism: London, Allen and Unwin.
Verschuren, D., Laird, K. R., and Cumming, B. F., 2000, Rainfall and drought in equatorial east Africa during the past 1100 years: Nature, v. 403, p. 410 -414.
Vincens, A., 1993, Nouvelle séquence pollinique du Lac Tanganyika: 30,000 ans d'histoire botanique et climatique du Bassin Nord: Review of Palaeobotany and Palynology, v. 78, p. 381 -394.
Wells, T. M., Cohen, A. S., Park, L. E., Dettman, D. L., and McKee, B. A., 1999, Ostracode stratigraphy and paleoecology from surficial sediments of Lake Tanganyika, Africa: Journal of Paleolimnology, v. 22, p. 259 -276.
Wetzel, R. G., 1983, Limnology: Fort Worth, Saunders College Publishing.
Wouters, K., 1988, On Romecytheridea ampia Wouters sp. nov.: Stereo -Atlas of Ostracod Shells, v. 15, no. 23, p. 101 -106.
Wouters, K., and Martens, K., 1992, Contribution to the knowledge of Tanganyikan cytheraceans, with the description of Mesocyprideis nom. nov. (Crustacea, Ostracoda): Bulletin de L'Institut Royal des Sciences Naturelles de Belgique,v. 62, p. 159 -166. 234
Wouters, K., and Martens, K., 1994, Contribution to the knowledge of the Cyprideis species flock (Crustacea : Ostracoda) of Lake Tanganyika, with the description of three new species: Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 64, p. 111 -128.
Wouters, K., and Martens, K., 1999, Four new species of the Cyprideis species flock (Crustacea: Ostracoda) of Lake Tanganyika: Bulletin de L'Institut Royal des Sciences Naturelles de Belgique, v. 69, p. 67 -82.
Zilifi, D., and Eagle, M., 2000, Sediment cores from the Kalya platform, slope and horst: high resolution and long -time span paleolimnology and paleoclimate records, in Cohen, A. S., ed., The Nyanza Project 2000 Annual Report: Tucson.