GENETIC DIVERSITY, DIET, and HABITAT QUALITY of the AFRICAN MANATEE (Trichechus Senegalensis) in the DOWNSTREAM of the SANAGA RIVER WATERSHED, CAMEROON

Home , African manatee, Fresco (web browser), Manatee conservation

GENETIC DIVERSITY, DIET, AND HABITAT QUALITY OF THE AFRICAN MANATEE (Trichechus senegalensis) IN THE DOWNSTREAM OF THE SANAGA RIVER WATERSHED, CAMEROON

ARISTIDE TAKOUKAM KAMLA

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2019

To my parents for always believing and supporting me

ACKNOWLEDGMENTS

I am deeply grateful to the many people and institutions with whom this doctorate would never have been accomplished. I received various forms of support for this work ranging from moral support, mentoring, funding, logistic support, and more.

First, I would like to extend my gratitude to Drs. Lucy Keith-Diagne, Ruth Francis-

Floyd, Iske Larkin, and Robert Bonde who believed in me and supported my acceptance to the College of Veterinary Medicine at the University of Florida. I am also profoundly grateful to Sally O’Connell who facilitated and guided me through all the administrative and academic requirements for my program.

I was honored and lucky to have been surrounded by motivating, exceptional, well-experienced, and renowned environmental biologists in my doctoral committee. I started weakly yet, they made me strong; I was short-sighted, and they taught me how to see the big picture. They helped me believe in my capacity and challenged me through various scientific exercises. They were more than available when I needed them. Dr. Ruth Francis-Floyd you have always been supportive and even a fan of my work. Your words have always been motivating for me. Dr. Lucy Keith-Diagne, thank you for believing in me since the first time we met, almost 10 years ago. You gave me the chance to achieve my dream of becoming a senior manatee biologist through various forms of advice and support letters. Dr. Robert Bonde, thank you for serving with a lot of devotion as the Chair of my doctoral committee; you have always been a model for me in the field of manatee conservation and research. Despite your multiple responsibilities and busy agenda, you have always been available and prompt in providing me help and knowledge. Dr. Margaret Hunter, when I started my first manatee field observation and realized the African manatee was very cryptic and challenging to

capture, I felt discouraged, wondering how I could study what I cannot see or reach?

Through the genetic knowledge you provided me, you made it possible for the first time to remotely study the animal from their feces. My hope for the species to seek more knowledge will be unveiled trough application of that technique. Dr. Tom Frazer, each opportunity I had to meet with you was always rewarding. You always challenged my thinking process and the philosophy of science. Your advice makes me challenge myself. You taught me how to make science more alive and impactful by translating the findings into meaningful and concrete ecological problem-solving. The diversity of the background of my committee has contributed significantly to making me a better and more multi-skilled scientist.

I also thank Cathy Beck for her tremendous technical support in the diet analysis.

I knew nothing about microhistological analysis, and thanks to you, I am now skilled in that field, and I am already training other African scientists on the technique. Thanks to your support we know more about the African manatee diet in Cameroon.

I will never be grateful enough to Dr. Mark Hoyer and the LAKEWATCH program, to have provided me the skills, knowledge, and lab facilitation to characterize the water quality of Lake Ossa. Thanks to your support, I was able to demonstrate and understand the causes and consequences of the eutrophication of the lake. The knowledge you provided me is now being used to address the nutrient enrichment and

Salvinia proliferation in Lake Ossa.

The genetic lab team at U.S. Geological Survey (USGS) has played a fantastic role throughout my lab work. Dr. Jason Ferrante, you have followed me step by step through lab work and made sure I was learning the right way. You provided me with

laboratory rigor essential to conduct good genetic science. Gaia Meigs-Friend, I was particularly impressed by your devotion, availability, and promptness. You were always there to guide me through the lab and to advise me on the various lab protocols. I am also grateful to Caitlin Beaver for her help with the microsatellite analysis.

Dr. Iske Larkin, you have played a crucial and diverse role throughout the course of my Ph.D. program. I would like to especially thank you for the enriching teaching experience you provided, and which will undoubtedly contribute to the success of the next steps of my career. Thank you to always find a solution for me when I was falling short of funding.

Dr. Nicole Stacy, thank you very much for your continuous support and advice in my Ph.D. program. Your interest in my work has always motivated me.

I am also grateful to Debra Anderson, the Director of International Student

Services, and Matthew Mitterko, Associate Director of Graduate International Outreach at the University of Florida, for their continuous administrative support.

In Cameroon, I am grateful to my excellent team in Cameroon, Rodrigue

Ngafack, Eddy Yannick Nnanga Akono, Amandine Toumbou, and Lionel Yamb who provided me incredible assistance during field data collection. I am also very grateful to our boat pilot Joseph Nkembe. I would also like to give a special thanks to the fishermen who help us locate areas where we could easily find manatee feces.

Thank you to the formal conservators of Lake Ossa and Douala-Edea Wildlife

Reserves, Azangue Georges and late Rene Martin Itamouna, respectively who were both supportive of this work and permitted me to conduct my study in the reserves.

Special thanks to Dr. Damien Essono from the University of Yaounde, who joined me in the field and helped to identify plant specimens.

I am very grateful to Paul Thomson and Joyce Wang from the Wildlife

Conservation Network (WCN) who believed in me, coached me, and helped me to network with amazing partners. Through your guidance, I was accepted into the WCN scholarship program and received tremendous financial support for my Ph.D. program, and an opportunity to share my research and conservation work with donors.

I would to thank Kate Mastro, Senior Program Manager at Wildlife Conservation

Society (WCS) who facilitated the admistration of my Beinecke scholarship, and gave me the opportunity to participate to capacity building trainings that contributed to the success of my research program.

I would like to thank the Florida LAKEWATCH program for providing water sample collection tools and the analysis of the water samples. I am grateful to BioBase for processing the sonar data to generate the bathymetry and SAV data. The

Environmental Systems Research Institute (ESRI) and Conservation GIS provided me with the latest ArcGIS software and training that was essential to run the geospatial analysis of this study. SOFTGENETIC has provided me with a special discount on the

GeneMarker software, which was critical for genotyping the microsatellite data of this study. QIAGEN also gave me a great discount on the QIAmp Fast DNA Stool Kit and the microsatellite Type-IT master mix.

I am indebted to all the funders of my Ph.D. program; without them, I could not have afforded tuition, stipend, and research cost for my program.

First, I would like to thank the U.S. Government and U.S. Embassy in Cameroon for believing me and granted me unique support through the Fulbright Scholarship program. I will always be proud to be a Fulbright Alumni.

I would also like to thank the Wildlife Conservation Society (WCS) and the

Wildlife Conservation Network (WCN) who also granted me scholarships that helped cover part of my tuition and stipend. Without your support, I could not have made it through. Thank you also for the support you provide to emerging conservationists around the world and especially in Africa.

I am indebted to the Aquatic Animal Health (AAH) and the CVM intramural research grant award of the University of Florida for granting me a graduate assistantship and funding for my field research.

Thank you for the Save the Manatee Club for their fantastic support, and providing for my lab supplies.

I am immensely grateful to the U.S. Geological Survey (USGS) – Wetland and

Aquatic Research Center (WARC) of Gainesville for providing me with their laboratory facility where I performed all the lab work of this study. I am also grateful for all the staff at USGS, especially Susan Butler, James Reid, and Dr. Pamela Schofield.

I am thankful to the Ministry of Defense of Cameroon and Metabiota for providing access to their lab facility during the pre-processing of my genetic samples in

Cameroon. I am looking forward to a great collaboration with you.

My sincere gratitude to Matthew Lebreton and Nathan Wolfe for having always been an inspiration for me and provided with great advice and opportunities.

The support from my friends in Gainesville was a great catalyzer for me. Thank you, Karen Lopez, Jonathan Cowart, Hilda Chavez Perez, Anmari Alvarez Aleman,

Majo Robles, Martial Kiki, and Dr. Nelmarie Landrau Giovannetti.

I am also thankful to my friends in Cameroon, Dr. Ahmadou Nana, and Jean

Michel Takuo.

To my family, who have always believed and supported me through my education, I will always be grateful. They have been very patient and understanding with my busy life as there were times I traveled to Cameroon and returned to the U.S. without having the chance go visit them in the other part of the country. I can never thank enough my grandmother Mama Pauline; my mother, Sikandi Honorine; Sister and the husband who adopted me as their son since my childhood; and Mafeu Rose and

Takoundjou Philippe. My nephews Cyrille Takoukam, Fongang Veronique, Fotsing

Armel, Kengne Ariane, Laticia Ngoumgne, and Loic Feukam. My brothers Talla Victor,

Eric Fongang, Artis Nguabo, and Ghislain Fonkoua, and to my cousin Eddy Tambo.

Finally, to my late Father, Kamla Pierre who would have been proud of me and delighted to see the first Ph.D. of the family. More importantly, I would like to thank my wife, Ruth Nsona, who has been supportive of my work and has been patient through our distance relation. You provided me with moral support and motivation to achieve this work. Thank you for having always been there. I cannot thank enough my sister, mother and father-in-law, Grace Ndongo, Jolie Nguenda and Daniel Ndongo.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 14

LIST OF FIGURES ...... 16

ABSTRACT ...... 19

CHAPTER

1 INTRODUCTION ...... 21

Life History of the Manatee ...... 21 African Manatee ...... 23 Distribution ...... 24 Manatee Distribution in Cameroon ...... 26 African Manatee Diet ...... 26 Manatee Habitat ...... 29 Manatee Habitat Requirements ...... 29 General Manatee Habitat Description in the Downstream Sanaga River Watershed (DSRW) ...... 33 Douala-Edea Wildlife Reserve ...... 33 Lake Ossa Wildlife Reserve ...... 34 Introduction to Basic Limnology Concepts ...... 35 Nutrient Enrichment and Biological Productivity ...... 37 Macrophytes ...... 37 Lake Morphometry ...... 39 Limnology of Lake Ossa ...... 39 Water Chemistry and Eutrophication of The Lake Ossa Complex ...... 39 Morphology, Hydrology, and Geology ...... 40 Vegetation ...... 41 Aquatic Fauna ...... 42 Conservation Genetics and Noninvasive Studies in Manatee ...... 42 Molecular Genetics as a Tool for Inferring Manatee Movement ...... 47 Molecular Genetics and Management Unit Delineation ...... 50 Major African Manatee Studies in Cameroon ...... 55 Conservation Status of the African Manatee ...... 60

2 ASSESSMENT OF THE LAKE TROPHIC STATE MODELS IN PREDICTING SUBMERGED AQUATIC VEGETATION AND ITS IMPLICATIONS FOR MANATEE CONSERVATION AT LAKE OSSA, LITTORAL, CAMEROON ...... 66

Background ...... 66 Methods ...... 71

Study Area Description ...... 71 Data Collection ...... 73 Water chemistry ...... 73 Rainfall ...... 73 Bathymetry and submerged aquatic vegetation (SAV) ...... 74 Data Analyses ...... 75 Water chemistry and rainfall ...... 75 Spatial analysis and prediction ...... 76 Accuracy assessment ...... 78 Results ...... 79 Water Chemistry and Rainfall ...... 79 Spatial Analysis and Prediction ...... 81 Discussion ...... 82

3 DIET COMPOSITION OF THE AFRICAN MANATEE: SPATIAL AND TEMPORAL VARIATION WITHIN THE DOWNSTREAM OF SANAGA RIVER WATERSHED, CAMEROON ...... 104

Background ...... 104 Methods ...... 109 Study Area ...... 109 Sampling Design ...... 109 Fecal Sample Collection ...... 110 Habitat Characterization and Plant Library Collection ...... 110 Reference Slide preparation ...... 111 Photography ...... 111 Microhistological Analysis ...... 112 Quantification of Percent Food Plant Species Occurrence ...... 112 Data Analysis ...... 113 Habitat shoreline species characterization ...... 113 Diet Composition and diversity analysis ...... 115 Results ...... 115 Shoreline Vegetation Characterization ...... 115 Diet Composition of the African Manatee in the DRSW ...... 118 Manatee diet by location ...... 118 Manatee diet by season in Lake Ossa ...... 119 Manatee diet by feces size...... 120 Discussion ...... 120 Shoreline Vegetation ...... 120 Anatomy of the African Manatee Feces in the DRSW ...... 123 Shape ...... 123 Size ...... 124 Texture ...... 125 Color ...... 126 Component of the African Manatee Diet in DSRW ...... 127 Manatee diet by location ...... 131 Manatee diet by season in Lake Ossa ...... 133

Manatee diet by feces size...... 135 Inference on the Manatee Movement within DSWR ...... 135 Conservation Implications ...... 138

4 ASSESSMENT OF THE EFFECTIVENESS OF NONINVASIVE FREE- FLOATING FECAL SAMPLES OF THE AFRICAN MANATEE AS A SOURCE OF DNA FOR GENETIC ANALYSIS USING MITOCHONDRIAL, MICROSATELLITE, AND SEX IDENTIFICATION MARKERS ...... 153

Background ...... 153 African Manatee Conservation Status ...... 157 Study Justification and Objectives ...... 157 Methods ...... 159 Comparison of the Efficacy of the DNA Isolation Methods ...... 159 Sample Collection and storage ...... 159 DNA extraction ...... 160 Applying the optimized extraction method on the larger pool of samples ...... 161 DNA purification ...... 162 Mitochondrial DNA amplification ...... 162 Nuclear DNA Amplification ...... 163 Optimization of the nuclear fecal DNA amplification ...... 163 Applying the optimized nuclear PCR protocol ...... 163 Positive and Negative Controls ...... 165 Microsatellite Genotype Assignment ...... 166 Amplification Success and Genotyping Error Assessment ...... 166 Sex Identification ...... 167 Statistical Analysis ...... 168 Results ...... 168 Comparison of the Efficacy of DNA Isolation Methods ...... 168 Analysis of Full Data Set Using the QIAmp Fast DNA Stool Mini Kit ...... 169 Mitochondrial DNA PCR Success ...... 169 Microsatellite DNA PCR Success, Allelic Dropout, and False Allele Amplification ...... 170 Sex-specific Gene Amplification Success...... 171 Effect of Habitat on PCR Amplification ...... 171 Discussion ...... 171 Comparison of the Efficacy of the DNA Isolation Methods ...... 171 Fibrous vs non-fibrous fecal DNA extractions ...... 174 Analysis of full data set Using the QIAmp Fast DNA Stool Mini Kit ...... 174 Mitochondrial DNA PCR Amplification Assessment ...... 175 Microsatellite DNA PCR Success, Allelic Dropout and False Allele Amplification ...... 177 Effect of Habitat on PCR Amplification ...... 180 Conclusion ...... 181

5 GENETIC DIVERSITY AND CONNECTIVITY OF THE AFRICAN MANATEE IN THE DOWNSTREAM OF THE SANAGA RIVER WATERSHED ...... 194

Background ...... 194 Methods ...... 201 Study Area ...... 201 Fecal Sample Collection ...... 201 DNA Purification ...... 202 Mitochondrial DNA Amplification ...... 202 Microsatellite and Sex Marker Loci Amplification ...... 203 Positive and Negative Controls ...... 204 Sex Identification ...... 205 Genetic Analysis ...... 207 Mitochondrial analysis ...... 207 Discrimination of individuals ...... 208 Microsatellite DNA analysis ...... 209 Sex ratio ...... 211 Results ...... 211 Discrimination of Individuals ...... 211 Microsatellite DNA ...... 212 Sex Ratio ...... 214 Mitochondrial DNA ...... 214 Discussion ...... 216 There are at least 49 Manatees in the DSRW ...... 216 Male-biased Sex Ratio in the African Manatee? ...... 217 Manatees in the DSRW Constitute a Single Population...... 217 African Manatee Populations in Cameroon and Gabon are Genetically Distinct? ...... 219 Geographical Extend of the DSRW Manatee Population ...... 221 Inferring Manatee Population History in the DSRW ...... 223 Conclusions and Conservation Implications ...... 229

6 CONCLUSIONS AND FUTURE DIRECTION ...... 238

Summary of Study ...... 238 Future Directions ...... 242

APPENDIX

A SUPPORTING INFORMATION FOR CHAPTER 3 ...... 245

B SUPPORTING INFORMATION FOR CHAPTER 4 ...... 259

LIST OF REFERENCES ...... 269

BIOGRAPHICAL SKETCH ...... 296

LIST OF TABLES

Table page

2-1 Confusion matrix performance parameters values, formula, and abbreviations used for the SAV predictive model in Lake Ossa...... 93

2-2 Historical (by Wirrmann, 1992) and recent (from this study) trophic state parameters of Lake Ossa ...... 93

2-3 Surface water chemistry; TP is total phosphorus, TN is total nitrogen, TChl is total chlorophyll concentrations, and SD is Secchi depth...... 94

3-1 Diversity index of the plant species recorded in the African manatee feces ..... 144

3-2 Survey effort and diversity index of shoreline vegetation by location in the DSRW...... 144

3-3 Pair-wise coefficient of dissimilarity (Bray and Curtis distance) of shoreline plant communities ...... 144

3-4 Major plant species surveyed along the shoreline of the downstream Sanaga River watershed...... 145

3-5 List of plant type, family and species (with the common names) and their per cent frequency in the 113 African manatee feces ...... 146

4-1 Nuclear microsatellite PCR reaction mixtures used for the pre-amplification PCR ...... 187

4-2 Comparison of extraction method on African and Florida manatee fecal DNA quality and quantity ...... 187

4-3 Comparison of DNA quality and quantity by fecal sample types:...... 188

4-4 Number of PCR amplifications, PCR success, and genotyping errors ...... 189

4-5 Comparison of averages in total fecal DNA concentrations, amount of total DNA yield per mg of fecal material ...... 189

5-1 Manatee polymorphic microsatellite markers with population diversity information ...... 233

5-2 African manatee mitochondrial control region pairwise ɸST estimates using Tamura distance estimation...... 233

A-1 List of plant group, family and species by location, number of plots surveyed (in the brackets), and their relative abundance in percentage...... 245

A-2 Major identified diet plant species of African manatees surveyed by location of the downstream Sanaga River watershed...... 251

A-3 Major identified diet plant species of African manatees surveyed by seasons in Lake Ossa...... 252

A-4 African manatee major identified diet items in Lake Ossa by feces diameter size...... 252

B-1 List of polymorphic microsatellite markers used in this study...... 262

B-2 List of some published articles that used fecal DNA from wild population species and the characteristics of their DNA extraction and PCR methods. .... 266

LIST OF FIGURES

Figure page

1-1 Schematic illustration of the threats to the African manatee population and habitat in Cameroon...... 64

1-2 Manatee distribution in Cameroon...... 65

2-1 Map of Lake Ossa showing the 18 sampling stations, the weather station location and the arbitrary zones of the lake...... 95

2-2 Effects of phosphorous and nitrogen on chlorophyll concentrations...... 96

2-3 Effects of chlorophyll and distance to the outlet on Secchi depth...... 96

2-4 Monthly variations of rainfall, depth, Secchi depth and MDC...... 97

2-5 Bathymetry map of Lake Ossa showing the different depth gradient. The bathymetric points were recorded in September 2016 during the wet season. .. 98

2-6 Prediction of Secchi depth throughout Lake Ossa,...... 99

2-7 Prediction of percent of surface light available at the bottom throughout the Lake Ossa...... 100

2-8 Map of the distribution of the SAV in Lake Ossa measured using the Lowrance HDS 9 Gen 3...... 101

2-9 Confusion matrix parameter values...... 102

2-10 Prediction of manatee habitat suitability base on the seasonal water variation...... 103

3-1 Map of the downstream Sanaga River watershed showing the surveyed areas and the distribution of the surveyed plots...... 148

3-2 Shoreline vegetation composition profile by plant type across the four surveyed locations of the downstream Sanaga River watershed...... 149

3-3 Distribution of shoreline vegetation by species and family...... 150

3-4 Shoreline plant species composition profile across the surveyed locations of the downstream Sanaga River watershed...... 151

3-5 Relative abundance of top 15 plant species identified along the shorelines of Lake Ossa by season (water level) ...... 151

3-6 Per cent frequency distribution of the identified plant items from African manatee feces...... 152

3-7 Percent frequency distribution of the identified plant items from the African manatee feces collected in Lake Ossa by season ...... 152

4-1 Box plots depicting DNA concentration yields for each extraction method based on the 135 African manatee fecal subsamples...... 190

4-2 Frequency distribution of the total DNA concentrations of 235 African manatee free-floating fecal samples...... 190

4-3 Frequency distribution of A260/280 values...... 191

4-4 Example of chromatograms of the Control region mitochondrial DNA sequences generated from the DNA isolated from African manatee non- invasive fecal samples...... 191

4-5 Example of chromatograms showing microsatellite allelic dropout, causing false homozygotes...... 192

4-6 Proportion of positive PCRs (black diamonds) and PCRs with allelic dropouts (hollow circles) plotted against the total DNA concentration of the sample...... 192

4-7 Comparison of PCR success and allelic dropout rates by habitat type ...... 193

5-1 Map of control region mitochondrial DNA haplotypes identified in African manatee samples from Cameroon...... 234

5-2 Two-dimensional principal component analysis of microsatellite genotype data ...... 235

5-3 Plots of means (A) and standard deviations (B) of the posterior probabilities .. 235

5-4 African manatee control region haplotype maximum likelihood tree ...... 236

5-5 Hindcast annual mean significant wave height along the coast of West African between Cameroon and Gabon and Cameroon and Nigeria...... 237

A-1 Map of the downstream Sanaga River watershed showing the spatial distribution of the 113 feces ...... 253

A-2 African manatee feces. A) Shape and color, and B) measurement of the diameter using a caliper. The feces were collected in the Sanaga River...... 254

A-3 Cumulative number of plant species by the number of plots surveyed for each location within the downstream Sanaga River watershed...... 255

A-4 Venn diagram of the surveyed plant species in the four locations of the downstream Sanaga River watershed...... 256

A-5 African manatee identified plant diet composition profile by season in Lake Ossa...... 256

A-6 Histogram of the distribution of the diameters of 377 free-floating African manatee feces collected within the downstream Sanaga River watershed...... 257

A-7 Salvinia molesta proliferation in Lake Ossa...... 258

B-1 Protocol for comparing three DNA extraction protocols: 2CTAB/PCI (Vallet & al. 2008), NucleoSpin Soil Kit, and QIAamp DNA Fast Stool Kit ...... 259

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

GENETIC DIVERSITY, DIET, AND HABITAT QUALITY OF THE AFRICAN MANATEE (Trichechus senegalensis) IN THE DOWNSTREAM OF THE SANAGA RIVER WATERSHED, CAMEROON By

Aristide Takoukam Kamla

December 2019

Chair: Robert Bonde Cochair: Ruth Francis-Floyd Major: Veterinary Medical Sciences

The African manatee is a threatened and poorly known aquatic herbivorous mammal that inhabits the coastal and inland waters of the western and central Atlantic coast of Africa. The downstream of the Sanaga River watershed (DSRW) is an essential habitat for the species in Cameroon. However, it suffers alarming threats from poaching, accidental catch in fisheries, and habitat degradation that may jeopardize their survival.

This study is aimed at improving the conservation status of the African manatee in

Cameroon by generating scientific knowledge for its protection.

To assess the quality of the habitat of manatees in Lake Ossa, the physical, chemical, and biological parameters of the lake were monitored and the relationship between the parameters was established to predict the submerged aquatic vegetation

(SAV) surface and quantified areas with suitable depth for the species during the low- water season. Estimates indicated that almost no SAV in the lake (<5% of the lake surface) was due to low water transparency. Only 6% of the lake surface provided suitable water depth for the species during the low-water season.

To determine manatee feeding ecology in the DSRW, we surveyed the shoreline vegetation and identified plant fragments from 112 feces. We found that the shoreline vegetation was diverse (>160 plant species). A total of 36 food plants was documented in fecal samples with Echinochloa pyamidalis as the highest represented species

(53.5%). Location and season had a significant effect on diet composition.

To assess the reliability of African manatee fecal DNA for genetic analysis, we used noninvasive fecal samples to obtain DNA for PCR amplification mitochondrial and microsatellite markers. For the first time, the fecal DNA of a manatee species was used to successfully identify individuals and sex with high amplification success (80%) and moderate allelic dropout (24%).

We next assessed the level of diversity and connectivity of the manatee population within the DSRW using the DNA markers. Results showed high genetic diversity in the DSRW (He=0.66, Na=5), but low effective population size (Ne=45.5).

Manatees in the DRSW constitute a single population, while manatee populations in

Cameroon and Gabon were found to be genetically distinct (θST=0.374, P<0.0001). We concluded that manatee and their habitats within the DSRW are connected and this connectivity need to be preserved by merging its adjacent protected areas.

CHAPTER 1 INTRODUCTION

Life History of the Manatee

Manatees belong to the Order Sirenia, a taxonomic group composed of four extant fully aquatic and mostly herbivorous mammal species that inhabit the tropical and subtropical waters around the world. They are closely related to the elephant. The

Steller’s sea cow (Hydrodamalis gigas), a species that belonged to the same order, and inhabited the Bering Sea, went extinct only three decades after its discovery by the

Europeans in 1741. The Sirenia Order is composed of two families; the Dugongidae with a extant single species, the dugong (Dugong dugon), the closest modern relative of the Steller sea cow, occupies the warm coastlines of the western Pacific Ocean to the eastern coast of Africa. The species is fully marine and occupies the most extensive distribution range of all sirenians. The family Trichechidae comprises the three species of manatees; the West Indian manatee, subdivided into two subspecies including the

Florida manatee (Trichechus manatus latirostris) found mostly on both coasts of Florida and the Antillean manatee (Trichechus manatus manatus) that occupies the shallow coastal habitat and some rivers and estuaries of the Caribbean and the northwestern

Atlantic Ocean from Mexico to Brazil. Unlike the latter that can live in both marine and freshwater, the Amazonia manatee (Trichechus inunguis) lives exclusively in freshwater and is found in the Amazon river and tributaries to the basin. They are the smallest in size of all sirenians, and their white chest patch makes them unique among its congeners. The African manatee (Trichechus senegalensis), is the least known of all sirenians; like the West Indian manatee, they can live in fresh, brackish, and marine waters. The species distribution range is the broadest and most diverse of all the

manatees, extending over 6000km along the Atlantic coastline of Africa from South

Mauritania to Angola. They inhabit waters up to 2500km inland from the sea (Keith-

Diagne, 2014, 2015).

The sirenians have undeniable ecological, economic, aesthetic, and cultural importance. They are heavy grazers and seagrass cultivators that help maintain and recycle plant biomass and thus provide nutrients for the phytoplankton and fish. Some fish feed directly on manatee feces. In 2002, the annual average benefit of the removal of aquatic vegetation by manatee for Citrus County, Florida (USA)was estimated at

$300,000 (Solomon et al., 2004). The Florida manatee is gentle, charismatic, and loved by many enthusiasts and tourists will travel long distances to see them. Solomon and colleagues (2004) estimated the use-value of the Florida manatee to range between eight to nine million dollars per year in Citrus County, Florida (USA). Manatees can be seen as exceptional sentinels of harmful algal blooms because they are susceptible to it

(Bonde et al., 2004). Some communities are culturally very attached to manatees. For instance, in the Casamance region of Senegal, the Diola and the Mandingo communities venerate the African manatee and prohibit their hunting (Dodman et al.,

2008).

All the species of Sirenia are threatened and are classified as Vulnerable on the

International Union for Conservation of Nature (IUCN) Red List. The Antillean and the

Florida manatee subspecies are listed as Endangered. They are threatened by various anthropogenic factors, of which hunting is the most common among all the species except for the Florida manatee, whose major causes of mortality are from collisions with boats and exposure to harmful algal blooms. African and Amazonian manatees are

heavily impacted by incidental take and entanglement in fishing gear. Because of the importance of manatees to culture, and the threats that impact their survival, they deserve strong conservation efforts.

African Manatee

The life history of the African manatee is poorly understood. Most of the information on the species life history is inferred from literature available for the West

Indian manatee as they are close relatives. The gestation period maybe around 12 to 14 months and age of sexual maturity is suspected to be at four to five years based on guesstimate from the Florida manatee, and seasonal breeding at the start of the rising water has been documented by several studies (Akoi, 2004; Dodman et al., 2008;

Husar, 1978; Powell, 1996). No information exists on their life span.

African manatees are fusiform in shape with a spatula-like and horizontally flat tail and a pair of paddle-like forelimbs bearing nails. Almost no accurate weights for

African manatees exist, but they are believed to weigh less than 500kg. Average length measurements for African manatees is 249 cm (Powell 1996; Akoi 2004; Keith-Diagne

2014). Fine wrinkles and sparse hairs cover their grayish-brown skin. The eyes and ears are tiny, as the animal likely relies on their body hair and vibrissae to sense their environment (Reep et al., 2011). Like other sirenians, the skull and skeleton are heavy, and the snout has the lowest values of the rostral deflection spectrum of extant sirenians (15°- 40°) which is adapted for feeding on the emergent and natant vegetation

(Domning & Hayek, 1986). Each side of the jaw possesses five to seven teeth that are replaced as the anterior teeth wear, and the posterior teeth are move forward in a conveyor-like pattern (Husar, 1978). The wearing of teeth enamel results from the abrasive effect of the silica contained in most emergent plants they feed upon.

Distribution

African manatees are distributed in 21 countries. The following are the coastal countries of the species distribution from north to south along the Atlantic coast of

Africa: Mauritania, Senegal, Gambia, Guinea-Bissau, Guinea, Sierra Leone, Liberia,

Cote D’Ivoire, Ghana, Togo, Benin, Nigeria, Cameroon, Equatorial Guinea, Gabon,

Congo, Democratic Republic of the Congo, and Angola. The three inland countries of the range of the species include Mali, Niger, and Chad (Keith-Diagne, 2015). There are no population size estimates for any of the countries; tentative estimates of the population size are based on interviews with locals (Keith-Diagne, 2015; Nishiwaki et al., 1982). The construction of hydroelectric and agricultural dams in some of the major rivers within the species range prevents the latter from extending their habitat upstream of blocking dams. In other rivers, the construction of dams has isolated manatee populations. The Diama dam in Senegal and the Kayes dam in Mali have blocked manatees within 950km of habitat along the Senegal River. The Kainji dam in Nigeria and the Markala dam in Mali have isolated the manatee populations of the Niger River

(Keith-Diagne, 2015; Powell, 1996).

The African manatee is found in various habitat types across their distribution range, including coastal marine water, brackish estuaries and lagoons, rivers, and tributaries and lakes that connect to the sea. During the low water season, as water recedes away from the shoreline vegetation of major rivers, manatees tend to travel down to the estuaries or into adjacent lakes where they can have access to deep water.

During the high water season, as the water levels increase and inundates the adjacent vegetation and forest along the shoreline, manatees travel there to consume the accessible fresh vegetation (Powell, 1996).

All the 21 distribution countries of the African manatee rely heavily on hydroelectric energy. As the human demography increases, so will the need for energy, which may lead to more dam construction in the future and further fragmentation of the species’ habitat across its entire home range. At least three new hydroelectric dams are under construction along the Niger River alone including the Kandadji dam in Niger

(Construction Review Online, 2019), the Fomi dam in Guinea (Ferrini & Benavides,

2018), and the Kourouba dam in Mali (Korkosse, 2018). In Cameroon, the country had planned to build a total of 16 hydroelectrical dams by 2035 on major rivers with high hydrologic energy production potential. These dams would undoubtedly directly or indirectly affect some manatee populations in Cameroon.

African manatees live in a variety of habitats. Their habitat requirements are similar to those of the West Indian manatee; they need calm water, and access to food and freshwater (Marsh et al., 2011; Reep & Bonde, 2006). Manatees in Senegal, the

Gambia and Guinea-Bissau favor an extensive network of freshwater springs present in the nearshore marine habitats that they utilize for drinking (Powell, 1990; Keith-Diagne unpublish data). The species transits unsheltered coastlines but does not spend as much time there as in the sheltered areas. In some countries, the African manatee travels mostly at sundown, feeds mostly at night and rests during the day, while in other areas such as the N’dogo and N’gowe Lagoons in Gabon, manatees are seen traveling and feeding both during day and night (Keith-Diagne unpublished data). African manatees are sighted mostly during daylight hours in areas where there is less hunting pressure. Manatees rest during the day in water areas that are between 1-2m deep, or in the middle of a watercourse, and stay hidden in mangrove or under floating or

shoreline vegetation (Akoi, 2004). In central African countries (Gabon and Republic of

Congo), some manatees migrate to coastal habitats during the dry season and return upstream to access the food in flooded areas and lakes during the rainy season.

In extremely far inland manatee riverine habitat (Senegal, Niger, and Mali), during the dry season, manatees use specific and restricted areas where food is available year-round and will spread out to floodplain areas during the rainy season, when grasses sprout and vegetation is flooded (Keith Diagne 2014;Powell, 1996).

Manatee Distribution in Cameroon

In Cameroon, manatees are frequently sighted downstream (after the Edea hydroelectric dam) of the Sanaga River, (Dekeyser, 1955; Nishiwaki et al., 1982;

Powell, 1996). Manatees are also abundant in Lake Ossa (Figure 1-2), which may provide a sanctuary for manatees during low water periods. Along the Cameroonian coastline, there are suitable habitats for manatees (Hatt, 1934; Powell, 1996), and manatees are not rare there (Powell, 1996) as suspected by Allen (1942). Manatees are present in Akwayafe, Rio del Rey, Ngosso, Andokat, Meme Rivers, and in the extensive mangroves. African manatees are also present in the coastal creeks of southwestern

Cameroon near the Nigerian border (Mayaka et al. 2019; Powell, 1996). Manatees have also been found in the Munaya and Cross River, in Wouri River and lower estuary, in the downstream of Nyong, Dihende, Dipomba, and Ntem River (Grigione, 1996). In northern Cameroon, manatees are present in the Benue River, from the mouth of the

Faro to Lakes Léré and Trene.

African Manatee Diet

Little is known about the food plants of the African manatee in general or on a local scale. Keith-Diagne (2014) conducted a literature review of reported African

manatee food plants throughout the species range. The author reported about 70

African manatee food plant species. It is unclear whether these food plant species are representative of the true species diet composition over its distribution range. Most of the plants were documented from Senegal, Ivory Coast, and Gabon. Akoi (2004) conducted gross inspection on 35 African manatee fecal samples on the coast of the

Ivory Coast and found out that African manatee diet was composed predominantly of grasses. They also ate fruit, mud, and deposited organic material especially during dry seasons, when the decreased water level limited access to emergent vegetation. The same observation was made by Powell (1996) who reported anecdotal information of

African manatee eating fallen fruits, hippopotamus dung, cassava peels, rice plants, roots along the bottom, algae, fallen leaves, and fish caught in fishing nets (Marsh et al.,

2011). In addition to the fecal examination, Akoi made direct observations of manatee feeding and noted the species and the number of times manatees were observed feeding on each plant species. He recorded a total of 16 species, and seashore paspalum (Paspalum vagiantum) and Egyptian paspalidium (paspalidium germinatum) were the dominant plant species (estimated to represent about 47% of the manatee diet in the N’gni Lagoon). He further observed that leaves and stems were the most important plant parts manatees were observed eating, although, on some occasions, they were observed feeding on roots. Root feeding was also observed among African manatees in Lake Ossa (Takoukam Kamla, 2012).

Some African manatee populations are not exclusively herbivorous, as earlier suggested by the literature. Keith-Diagne (2014) conducted the first carbon (δ13C) and nitrogen (δ15N) isotope analysis on the ear bones from 24 African manatee carcasses

recovered in Senegal and Gabon to determine the lifetime diet composition of individuals. The stable isotope signatures recorded from samples collected in Gabon indicated that their diet was made up of 90 to 94% of plants and 6 to 7% of hermit crabs. The sampling of the ear bones from manatees from Senegal showed that their diet was composed of 46 to 57% of plants, 24 to 27% of fish, and 19 to 24% of mollusks. The author suggested that the difference in diet composition between manatees from Senegal and Gabon was mostly due to habitat differences. Mayaka and colleagues (2019) also documented fishermen reports of manatees stealing and eating fish from their fishing nets in the Lower Korup region in Cameroon. Carnivorous behavior has also been documented in the West Indian manatee (Courbis & Worthy,

2003; Powell, 1978) and has been anecdotally reported in the African manatee by local fishermen (Dodman et al., 2008; Powell, 1996; Takoukam Kamla, 2012). Carnivorous behavior has been considered opportunistic, suggesting that manatees feed on other food resources only when plants are not available. However, the isotope analysis by

Keith-Diagne (2014) demonstrated that manatees from Senegal have a high proportion of fish and mollusks in their average lifetime diets, indicating that the manatees regularly and abundantly have a diet of animal origin.

A fisherman reported through SIREN, an African marine mammal sighting network initiated by the African Marine Mammal Conservation Organization (AMMCO), a picture of fish remains that had been eaten by a manatee from his net

(www.ammco.org). In Lake Ossa and in the Korup region, it is common for fishermen to complain about manatees ‘’stealing’’ fish from their nets, thus reducing their fish yield.

In the lower reaches of Sanaga River, the local people who dive to collect clams

reported observing manatees diving with them to eat clams along the bottom

(Takoukam Kamla 2012; Mayaka et al., 2019).

Manatee Habitat

Manatee Habitat Requirements

The habitat of a species is a mix of components, including a geographical location and the set of biological and physical surroundings necessary for the population to survive and reproduce (Morrison et al., 2006). Habitat basic requirements for any given species include food, water, space, and cover (Leopold, 1933; Morrison et al.,

2006; USFWS, 1999). Thus, the distribution and the various form of movements of a species are highly controlled by the distribution of the overlap of their habitat requirements. Therefore, understanding a species habitat requirement can be paramount in predicting its distribution, and determining the spatiotemporal dynamics of the habitat can further allow researchers to make predictions on its movement patterns or to assess the probability it will become locally extinct.

Although each factor of a species habitat requirement is essential for its survival and reproduction, they do not all influence the later with the same strength. According to

Liebig’s law of minimum (Browne, 1942), growth is regulated by the scarcest resource rather than the sum of the resources currently available. Therefore, fluctuation in the factors that are not limiting has only minimal, marginal, or no effect on the survival success of a species (Morrison et al., 2006). Therefore, O’Connor (2002) recommended that when predicting or quantifying the habitat of a species, the accent should be placed on factors that are limiting to the species distribution. This conceptual approach for quantifying habitat is commonly used in the field of limnology, where for example, total

phosphorous, the common limiting factor of primary production in freshwater systems is used to predict the concentration of chlorophyll (Hoyer et al.,2015).

Manatee habitat requirements include, but are not limited to access to, warm, fresh, and deep waters of at least 2m (USFWS, 1999), food plant availability, access to cover, and space. Females with calves additionally require habitats with minimal ambient noise and water current (Gannon et al., 2007). Unlike Florida manatee habitat, where temperature is a limiting factor during the winter season (Marsh et al., 2011;

Reep & Bonde, 2006; USFWS, 2001), Amazonian, Antillean, and African manatees movements appear not to be influenced by temperature as their primary habitats are found in tropical areas where the water is warm year-round. Similarly, access to fresh water is not a limiting factor to the Amazonian manatee and the African manatees that inhabit freshwater systems. The Amazonian and African manatees have evolved in freshwater habitats with similar physical characteristics. The seasonal movements and the ecological activities of both species appear to be controlled mainly by water level, food plant accessibility, and cover (Akoi, 2004; Arraut et al., 2010; Best, 1982; Marsh et al., 2011). However, the water level appears to influence both cover and food plant accessibility when they are available. In fact, unlike the West Indian manatees and dugongs who feed primarily on submerged vegetation, Amazonian and African manatees have maintained a smaller rostral deflection angle that favors feeding on emergent vegetation (Domning, 2005). Emergent vegetation is mostly distributed along the shoreline, where water depth is the shallowest and very often too shallow to allow a large 300kg manatee to swim over to the bank to access it (Arraut et al., 2010;

Takoukam Kamla, 2012). Therefore, African and Amazonian manatees might have

greater access to emergent food plant species during the high-water season when shorelines, swamps, savannahs and forests are flooded, allowing them to reach the shallow growing plants that were once inaccessible.

Manatee cover includes areas where they can hide away from predation, human disturbance, and rough weather (currents and waves); this includes deep pools, floating vegetation beds (I observed two manatees that hid under floating vegetation beds as soon as they noticed our presence) or flooded forest or swamps (these areas are hardly accessible for fishermen). As with food plant access, cover increases with water level as the latter provides more water depth within the overall water system. In Lake Ossa, antelope grass (Echinochloa pyramidalis) is the most abundant macrophyte whose roots interweave to form a vegetation bed or mat along the shoreline. During the low water season, the mat is in very shallow water and is almost in direct contact with the bottom. However, when the water level rises during the high-water season it gradually inundates the floating Echinochloa bed. The rise of the floating bed creates a water space in the water column under the mat that provides an ample place to hide, as well as provides food accessibility to manatees and other species. Therefore, it appears that the water level in Lake Ossa might be the primary limiting factor that controls the distribution, behavior, and movement of manatees in Lake Ossa. Best (1982) found that unlike Antillean manatees that have a continuous breeding season, Amazonian manatees have a seasonal breeding season between December and July, which corresponds to the high-water season. Fishermen have reported similar seasonality in breeding activity in Lake Ossa (Takoukam Kamla, 2012) as they are observed more

frequently in herds between July and October, which corresponds to the higher water season there.

The water level in Lake Ossa depends on two main factors, rainfall across the lake watershed and the discharge rate of the nearby Sanaga River (Giresse et al., 2005;

Takoukam Kamla, 2012; Wirrmann & Elouga, 1998). Also, water infiltration, evaporation, and evapotranspiration negatively influence water levels; however, their impact remains relatively constant throughout the year. River Sanaga is the longest

(910km) in the country with a basin of about 133,000km2 (Ndam Ngoupayou et al.,

2007). Dubreuil and colleagures (1975), made a comprehensive monitoring study on the hydrology of Sanaga River in different regions of its basin, including in the DSRW for which the monitoring station was based at the level of the Edea bridge, 13km north from the southern end of Lake Ossa. The study presented flow data of the Sanaga River at the station of Edea for 27 years from 1943 to 1970. The maximum and minimum flows were 7570m3/s and 171m3/s respectively. The median flows across the same period during the low and the high-water season was 310m3/s and 7600m3/s, respectively. The maximum flow was often observed in October and the minimum in March. The monitoring period also included drought (1946-47) and wet years (1955-56) that had a dramatic influence on the minimum (280m3/s and 520m3/s respectively) and the maximum (5260m3/s and 7570m3/s) water flows. This data highlights the large monthly flow variability that is due to the rainfall seasonality and inter-annual variability influenced by drought. The maximum flow within that period is 45 times higher than the minimum. Moreover, the minimum flow appears to be half of the wet to drought years.

This enormous variability of the Sanaga River flow has a direct repercussion on the

water level in Lake Ossa, which decreases by half from the high to the low water season (Giresse et al., 2005).

General Manatee Habitat Description in the Downstream Sanaga River Watershed (DSRW)

The DSRW encompasses two important protected areas in the Littoral Region of

Cameroon: Douala-Edea Wildlife Reserve (in the Mouanko subdivision and the Lake

Ossa Wildlife Reserve in the Dizangué subdivision). Both protected areas lie within the division of Sanaga Maritime, in the Littoral Region. The DSRW encompasses five major types of habitats: lakes (Lake Ossa and Lake Tissongo), rivers (Sanaga, Kwa Kwa, and other tributaries), estuaries (Sanaga, Cameroon, and Nyong estuaries), mangroves, and coastal. This study did not include the coastal habitat because of the difficulty of acquiring manatee fecal samples from the turbulent, unsheltered sea.

Douala-Edea Wildlife Reserve

The formal Douala-Edea Wildlife Reserve (DEWR), which has been upgraded to a national park, is located in the Department of Sanaga Maritime. Its geographic coordinates lie between 3°14' and 3°50'N latitude and 9°34' and 10°03'E longitude. It has an area of about 1,600km2. Its limits extend from the Atlantic coast for a distance of

35km inland, with its eastern boundary along the river Dipombé. The reserve consists of two unequal parts: the larger in the south, which lies between the mouths of the Sanaga and Nyong rivers north to south; the smaller portion extends along the northern coast of the Sanaga up to Souelaba and extends along the Creek Kwa Kwa. The reserve, adjacent to the Atlantic Ocean, is a littoral forest dominated by the red ironwood

(Lophira alata) and the English bitter bark (Sacoglottis gabonensis). This type of vegetation covers most of the reserve. The abundance of both emergent species

characterizes the DEWR. Among the dominant species in the canopy is Coula edulis

(Oleaceae). Also, frequently present in the reserve is Ebenaceae (Diospyros spp.),

Guttiferae (especially Garcinia spp.) especially in the flooded areas, and Euphorbiaceae

(Protomegabaria stapfiana and Dichostemma glaucescens). However, there are several other types of vegetation according to elevation, drainage, topography, and soil type

(Newbery & Gartlan, 1996).

Lake Ossa Wildlife Reserve

Lake Ossa Wildlife Reserve (LOWR, Figure 1-2) is 13km from Edea between

3°45’ and 3°52’N latitude, and 9°45’ and 10°4’E longitude with approximately 300m elevation. LOWR covers an estimated area of 4,000ha of water; however, its limits are not well defined and change with high- and low-water conditions. The LOWR was created in 1968 and falls within the 3rd category of protected areas, according to

Cameroon’s waterbody classification system. The reserve was established specifically for the protection of the African manatee (Trichechus senegalensis). Emerged macrophytes dominate the aquatic vegetation.

The catchment area of the Lake Ossa complex is about 245,000ha and extends mostly on the northern part, drained by a network of small interconnected near- perennial streams. The south-western part of the watershed is very narrow with a steep slope. The lake itself is a lacustrine complex consisting of three lakes (Figure 1-1). Lake

Mévia in the north of the lake complex is 700ha; Ossa has the largest water surface area of the lakes with 37,000ha and Mwembé the smallest (300ha) is located at the most southern part of the lacustrine complex. The Lake Ossa complex is shallow, with a maximum depth of about seven meters during the wet season (Giresse et al., 2005).

The greatest width of the lake is about 7km.

(Echinochloa pyramidalis) and in some places with shrubs and rattan (Calamus sp).

Some of the shorelines have a discontinuity of sandbanks that are the preferred sites for the reproduction of Nile Softshell turtles (Trionyx triunguis). In the middle of the rainy season, a large proportion of the littoral forest becomes flooded, providing fresh plant food and refuge for the African manatee and other aquatic species.

About 17,000 human inhabitants live around the lake (BUCREP, 2010), and the majority of them depend directly or indirectly on the lake’s resources to survive. Over

300 fishers are active on the lake, especially during the low water season (Takoukam

Kamla, 2012). Women of the local community utilize the lands around the lake for agricultural purposes. Additionally, there is large agro-industry established on the south- west portion of the lake which has rubber and palm tree plantations covering an area as large as the lake itself.

Introduction to Basic Limnology Concepts

Limnology is the study of freshwater systems such as lakes, ponds, rivers, springs, and wetlands (Wetzel, 2001). It incorporates the biological, chemical, physical, and geological attributes of a freshwater system. The biological component refers to the set of living or dead organisms and their interaction with the other components of the system. The biological component is assessed using parameters such as productivity, species richness, biomass, or density. Depending on the level of productivity of the

system, it will be classified as oligotrophic (less productive), mesotrophic (intermediate level of productivity), eutrophic (high biological productivity), and hypereutrophic (very high productivity, LAKEWATCH, 2007). The chemical aspect of limnology describes and quantifies the essential chemical components of the water system. Concepts include consideration of limiting nutrients that impact plant or algal growth(e.g., nitrogen, phosphate, iron, silica, LAKEWATCH (2000). The physical aspect of limnology is related to the mechanical forces or structures such as light, wind, flow, lake structure, morphology, and other variables that affect both the chemical and the biological component of inland waters. Geology is an important aspect of limnology studies as characteristics of the bottom sediment strongly correlate with the chemical composition of the water (Alfredsson et al., 2013; Canfield & Hoyer, 1988b). Thus, life in inland water is a result of the interaction between the biological, chemical, physical, and geological components. They all interact following laws that limnologists can utilize to explain and predict outcomes in freshwater systems.

Understanding the biological, chemical, and physical components of Lake Ossa is paramount to comprehending how African manatees utilize their habitat. African manatees are mostly herbivorous; however, they also feed minimally on fish and mollusks (Keith-Diagne, 2014). Sirenians rely heavily on macrophytes as a food source and can eat an equivalent of 8% of their body weight of macrophytes daily (Marsh et al.,

2011). Manatees also use floating macrophyte beds as hiding places. Therefore, the aquatic plant productivity of a lake is important in determining the carrying capacity for manatees. Moreover, the distribution of the macrophytes and their seasonal availability within the lake can directly correlate with the spatiotemporal abundance of manatees.

Nutrient Enrichment and Biological Productivity

There is a strong relation between chlorophyll, depth, light penetration, total nitrogen (TN), and total phosphorus (TP) in the world north-temperate lakes (Brown et al., 2000). Phosphorus is the limiting nutrient for algae and plant growth in most lakes around the world (Hoyer et al. 2002). Determining whether TN or TP is limited in Lake

Ossa can be easily assessed by computing the ratio of TN to TP, if it is less than 10

(LAKEWATCH 2000). Hoyer and co-authors (2015) developed a chain of eutrophication models by establishing an empirical relationship between chlorophyll and

TP concentrations between chlorophyll concentrations from Secchi depth. Although the primary purpose of this model chain is to monitor eutrophication by controlling phosphorous loading, it can also be used to predict the areas of the lake with high fish productivity and species richness based only on Secchi depth.

Macrophytes

There are four broad categories of aquatic plants, including emergent (rooted, stem, and leaves above the water surface), rooted floating-leaves, free-floating

(unrooted), and submerged plants (LAKEWATCH, 2007). Several factors can affect the productivity and the distribution of aquatic plants including but not limited to light and nutrient availability, lake trophic state, lake morphology, depth, bottom, and sediment composition, wave action, and competition from other aquatic species (Alfredsson et al.,

2013). All these factors are somewhat interrelated and can have an independent or combined effect on each other. Nonetheless, light availability appears to be the most critical factor that affects submerged aquatic plant growth and distribution

(LAKEWATCH, 2007; Zimmerman et al., 1994). In streams, aquatic plant growth can be mostly influenced by water current (Butcher, 1993), but streams with a slow current

will accumulate enough nutrients and, therefore, would be limited primarily by light availability and openings in the forest canopy (Canfield & Hoyer, 1988) .

Photosynthetic plants require light to build their biomass from atmospheric or dissolved carbon dioxide. Owing to the high attenuation of irradiance through the water column, areas of the lake with greater depth might not receive enough light to support the growth of rooted plants Canfield and co-authors (1985) established a positive linear relationship between water transparency as indicated by Secchi depth and the maximum depth of plant colonization (MDC) based on data from 26, 27, and 55 lakes in

Florida, Finland, and Wisconsin, respectively. The author developed an empirical model that predicts MDC values from Secchi depth. He found that the 95% confidence interval ranged from 46% to 236% of the estimated MDC and hypothesized the large variability in the MDC-Secchi depth relationship model might be caused by the variation in light compensation points (intensity of light for which the photosynthetic and respiration rates are equal) among species. A more robust experiment by Caffrey and colleagues (2007) confirmed the MDC-Secchi depth relationship model by Canfield (1985). This empirical relationship has not yet been demonstrated in Lake Ossa. Therefore, the distribution of submerged vegetation in Lake Ossa can be predicted using MDC estimated from

Secchi depth under the assumption that light is the only limiting factor.

Understanding the distribution of phytoplankton productivity in Lake Ossa can be relevant for the management of its manatee population. The persisting fishermen- manatee conflicts in Lake Ossa result from the overlap of fishing areas and those highly frequented by manatees. Given that there is a correlation between fish and algae productivities (LAKEWATCH, 2007b), and that areas with high fish productivity will

attract more fishers, the distribution of fishing effort can, therefore, be predicted based on the distribution of chlorophyll concentration across the lake and at different seasons.

Lake Morphometry

Morphometric parameters play an important role in the functional and structural components of lake ecosystems (Håkanson, 2005) as they influence transport and movement processes in the lake, including resuspension, diffusion, mixing, burial, sedimentation, and outflow. These movements result from the exposure of the lake to wind, waves, and water currents (LAKEWATCH, 2001). Then, these movements will, in turn, affect water clarity, chemistry (phosphorous and nitrogen), and primary and secondary productivity. Various parameters are used to characterize a lake, including the hypsographic curve (graph relating the depth of a lake to its surface area) and the bathymetry map (contour lines joining points of the lake with equal depth). It also includes volume, maximum length, width and depth, and mean depth and width, shoreline length, and fetch (distance wind can travel over the water surface without intersecting a landmass).

Limnology of Lake Ossa

Water Chemistry and Eutrophication of The Lake Ossa Complex

Measurements performed by Nguetsop and co-authors ( 2004) In the Lake Ossa complex between 1993 and 1995 indicated that pH in the lake was low during the wet season (6.20 to 6.85) and higher during the dry season (6.29 to 8.99). The author suggested that the low pH during the low water season may be attributed to the acidic meteoric water in the region. The author noted that pH was neutral or moderately alkaline (7-8) in the western part of the lake. In Lake Mwembe and Lake Mevia, the pH was acidic (less than 7). The Secchi depth changes depending on the season. It is

higher during the dry season and lower during the rainy seasons as water from the

Sanaga River, containing a higher level of sediments, flows into the lake.

Wirrmann (1992) in 1985, measured trophic state parameters for Lake Ossa and found that total nitrogen, total phosphorus, and chlorophyll concentrations were 157,

12.9, and 8.4μg/l, respectively. Based on these values, the lake appeared to be mesotrophic based on Carlson’s (1977) classification. Three decades later, the previous parameters were reassessed by Njoya (2015) by measuring the concentration of the ammonium ion which alone was 1220μg/l and 12μg/l for nitrite. While the value of the orthophosphate (which is an underestimation of the total phosphate) was between 20 and 100μg/l. Secchi disk was between 16cm (0.5 feet) and 100cm (3.3 feet). More recent values of chlorophyll are not available. Njoya’s data suggested that the lake has become enriched, moving from a mesotrophic state (based on Carlson) in 1985 to a eutrophic state in 2015. The source of this nutrient enrichment was not clear. Thus, in this study, we reassessed the lake’s physical and biochemical properties to confirm enrichment of the lake and to determine potential sources of nutrients.

Morphology, Hydrology, and Geology

The catchment area of The Lake Ossa complex is about 165km2 and extends mostly on the northern part, drained by a network of small interconnected near- perennial streams. The south-western part of the watershed is very narrow and relatively hilly. The lake itself is a lacustrine complex of three lakes. Lake Mévia in the north of the lake complex is 7km2 large; Ossa has the largest water surface area of the lakes with 37km2, and Mwembé the smallest (3km2) is located at the most southern part of the lacustrine complex. The Lake Ossa complex is shallow with a maximum depth of

about seven meters during the wet season (Giresse et al., 2005). The maximum width of the lake is about 7km.

Lake Ossa is adjacent and connected to the Sanaga River through a 3km meandering and relatively deep channel (7 to 12 meters). Hence, the watercolor and its chemistry are influenced mainly by the seasonal incursion of Sanaga water (brownish and full of sediment) into the lake. Water transparency is not equal across all parts of the lakes. The water transparency in Lake Mevia is twofold higher than the transparency in the Lake Ossa complex and Lake Mwembe. The banks are mostly covered with weeds and shrubs and in some places with rattan. Some sections along the shoreline are sandy and preferred sites for the reproduction of soft-shell turtles. In the middle of the rainy season, a large proportion of the littoral forest becomes flooded. The Lake

Ossa complex lies on a yellow ferritic bed formed during the Palaeocene. The soil is abundant with iron-rich shales and black marlstones.

Vegetation

Emerged macrophytes dominate the aquatic vegetation. The antelope grass

(Echinochloa pyramidalis) is the dominant macrophyte followed by the seashore paspalum (Paspalum vaginatum) and bia (Sclerosperma manii). Other species include

Centrosoma pubescens, Hyptis lanceolate, Ethulia conyzoides, Fuirena umbrellata,

Odenlandia offinis, Pteriduim aquilinum, Ceraphyllum dermesum, Nymphaea lotus,

Cloeme afrospina, Dissotis rotundifolia, and Ludwigia abysinica (Njoya, 2015). The terrestrial and island forest is related to the climate, altitude, soil type, and topography

(Giresse et al., 2005) and is dominated by two species: Lophira alata (a tree also known as Azobé or red ironwood) and Saccoglottis gabonensis (bitter back tree). The natural forest was primarily reduced by the implantation of agro-industries like SAFACAM that

occupy the largest proportion of the reserve to grow rubber plants (Hevea sp.) and oil palm (Elaeis guneensis). These agricultural practices have adverse impacts on the drainage basin of the lake, which is exposed to erosion and landslides. The local population uses the rest of the land for subsistence agriculture, where they seasonally farm plantain (Musa sp), cassava (Manihot esculenta), cocoyam (Colocasia esculenta), and other lesser species.

Aquatic Fauna

A recent study by a master’s graduate student in the Lake Ossa complex

(Ndjassi, 2015) identified that there are about 33 fish species from 18 families (taxa) in the lake. The most common taxa are the Cichlidae family (seven species) representing

21% of total species in the lake, followed by Mormyridae (five species, 15%),

Claroteidae (four species, 12%), and Alestidae (three species, 9%). Other aquatic species include the freshwater softshell turtle (Trionyx triunguis), the dwarf crocodile

(Osteolaemus tetraspis), and the African manatee (Trichechus senegalensis). There are also some species of aquatic birds like ducks and herons.

Conservation Genetics and Noninvasive Studies in Manatee

With the emerging world of molecular biology, scientists are now able to efficiently ask research questions that were difficult or unanswerable using traditional ecological or morphological methods (Frankham et al., 2002). Molecular biology allows for the assessment of biodiversity with a finer resolution (down to the molecular or genetic level) and the understanding of the underpinning mechanisms at the origin of the visible macroscopic expression of diversity (e.g. species, genus, families, etc.), and how external factors such as the different anthropogenic and natural components of the environment shape the outcome of this diversity. The molecular understanding of

genetics now plays a crucial role in the effective management and conservation of species as it allows one to deduce connectivity, population structure, and origins of populations and taxonomic groups (Frankham et al., 2002; Schwartz et al., 2007).

Conservation genetics is an excellent example of the application of molecular biology to address conservation issues. Biologists now utilize genetic approaches in a variety of situations, including the use of genetic markers for forensic investigation in wildlife and endangered species, taxonomic affiliation, population history, and more (Allendorf &

Luikart, 2007). Genetic samples are the precious material from which molecular information can be extracted for analytical applications. However, obtaining high-quality samples for some wildlife species can be very challenging. Most studies rely on genetic isolation of tissues from live or dead individuals. However, such methods can be very invasive for live animals and carcasses do not always provide appropriate sample quality and quantity. Further, many wildlife species, especially marine wildlife, are very cryptic and elusive, making it challenging to capture them for sampling. Capturing wild animals is not always risk or pain-free; it may cause discomfort to the animal, and in some species, it may result in capture myopathy and even death if not handled properly

(Harthoorn et al., 1974). In developing countries, conducting capture of large marine species can be even more challenging because of limited logistical support and availability of trained personnel.

With the advent of the polymerase chain reaction (PCR), non-invasive sampling for DNA has become possible by requiring a minimal amount of DNA for the template

(Morin & Woodruff, 1996). Contrary to destructive sampling where the animal is killed and non-destructive sampling where the animal is live captured and biopsied to collect

tissues, non-invasive sampling relies on a tiny amount of DNA left behind by the animal either as feces, urine, hairs, scales, feathers, or shed epidermal cells (Taberlet et al.,

1999). Non-invasive samples have been exciting for field biologists since the 1990s as it allows for genetic sampling with no need to observe, handle, or capture the animal

(Höss et al.,1992; Ramón-Laca et al., 2015; Taberlet et al., 1999). The use of feces as a source of DNA has several advantages compared to other DNA sources. Fecal sampling is non-invasive as it does not require the collector to handle or even detect the target animal. Moreover, these samples are more available as all individuals frequently generate them. Finally, the feces of many species can be detected relatively quickly without the need for equipment or sophisticated logistics (Long et al., 2012).

The non-invasive fecal DNA sampling methods has been frequently applied in terrestrial and semi-aquatic species such as primates, bears, deer, and elephants

(Brazeal et al., 2017; Foote et al., 2012; Schwartz et al., 2007). However, its application in the aquatic environment has been limited due to the lesser accessibility of non- invasive samples such as feces, shed skin, or hair. For instance, feces of many species do not float, and even when they do, they becomes mechanically disintegrated quickly by waves and currents (Wotton & Malmqvist, 2001). Also, unlike feces from the terrestrial environment in which DNA is preserved as the sample dries up, feces in the aquatic environment are subject to faster environmental decomposition and possible contamination (Tikel et al., 1996; Muchett et al., 2019).

Despite the advantages of the non-invasive fecal DNA sampling, only a few genetic studies have applied this approach to a marine mammal species. In addition to the dugong, non-invasive fecal DNA sampling for genetical analysis was successfully

used in bottlenose dolphins (Tursiop truncates, Parsons, 2001; Parsons et al., 2003,

2006), Atlantic spotted dolphins (Stella frontalis, Green et al., 2007), right whales

(Eubalaena glacialis, Gillett et al., 2008, 2010; Rolland et al., 2006), Pacific walruses

(Odobenus rosmarus divergens, Bowles & Trites, 2013) and the West Indian manatee

(Trichechus manatus, Muschett et al., 2009; Díaz-Ferguson et al., 2017).

Environmental feces as a source of DNA was documented for the first time in a marine mammal in 1996, and more specifically in the dugong (Tikel et al., 1996). Before that, marine mammal biologists mainly relied on the remote sampling of skin biopsies from free-ranging animals in order to obtain DNA samples. Tikel and colleagues (1996) successfully isolated a 193bp fragment of the D-loop region of the mitochondrial DNA from fecal samples of dugongs collected in Borroloola, Australia. However, no statistical amplification success rate was reported.

Muschett and colleagues collected 35 free-ranging fecal DNA samples of the

Antillean manatee in Bocas del Toro in Panama that were preserved in 70% ethanol.

DNA was extracted from the fecal samples using the QIAmp DNA Stool Mini Kit

(QIAGEN). A segment of the control region of the mitochondrial DNA was successfully amplified in 18 samples using the primers L15926 and H16498 (Kocher et al., 1989).

However, five of the 18 sequences matched human mitochondrial DNA. The authors then developed a manatee specific forward primer (LTMC01) that was conjointly used with two reverse dugong primers (HDDCR01 and HDDCR02) that had been used to amplify mtDNA segments of 450 and 534bp respectively and yielded amplification success of 70% and 20% respectively. The authors suggested that the difference in amplification success between the two reverse primers might be due to the difference in

the DNA fragment length they generate. However, the Muschett and co-authors (2009) study did not provide any assessment of basic genetic parameters such as haplotype diversity or connectivity based on the successfully amplified sequences. Also, the author did not assign the sequences to a haplotype, nor did they record their findings in

GenBank. No qualitative measure of the amplicons was provided.

Díaz-Ferguson and collaborators (2017) were the second and most recent published study using non-invasive fecal DNA of manatees for genetic analysis. The goal of the study was to determine the genetic diversity and connective of the Antillean manatee in Panama. The author extracted DNA from 20 fresh free-floating samples of the Antillean manatee collected opportunistically in the Rio Negro River in 2013 and from samples from four soft tissue and four bone fragments collected from carcasses.

All samples were preserved in 95% ethanol. The DNA was extracted from feces using the Power Water Kit (Mo Bio, Carlsbad, California, USA). The control region primers

CR5 and CR4 (Garcia-Rodriguez et al., 1998) were used to amplify a segment of the mitochondrial DNA. However, only seven samples yield enough DNA for downstream analysis, and of the seven, only three fecal samples yielded reliable sequences. The three sequences were identical and sizeable to the haplotype J03, which was reported for the first time by the study.

The use of non-invasive fecal DNA for genetic assessment has not previously been explored date in the African manatee despite the availability of the fecal samples of the species that are frequently seen floating at the water surface. Recently, Hunter and colleagues (2018) developed Cytochrome b quantitative and droplet digital PCR eDNA essays to detect shed DNA of African manatee and the two subspecies of the

West Indian manatee from water samples collected in their respective habitats. The probe detected eDNA in only one of the three locations within Lake Ossa, where the samples were collected. Hunter and colleague’s eDNA results provide hope for advancement of scientific knowledge of the African manatee which suffers from poor management across most of its distribution due in part to the massive data gap that hinders effective management of the species. Given that fecal samples concentrate more DNA from the animal than water samples, it would not be surprising that feces from this species yield better quality and higher quantity of DNA.

To date, there is no published genetic study involving the use of either invasive or non-invasive African manatee nuclear or microsatellite DNA analyses. The use of the fecal DNA for genetic analysis may advance our knowledge of species diversity, abundance, structure, connectivity, and evolution. Such information is essential to inform management decisions and to design effective management strategies that could save the species from multiple threats.

Molecular Genetics as a Tool for Inferring Manatee Movement

There are different approaches in the use of population genetics to infer dispersal or migration between wildlife populations. However, they can be classified into two categories including inference from biparentally inherited markers (allozymes, autosomal microsatellites), and inference from unique-parentally inherited markers or sex-specific markers (mtDNA and Y-chromosome linked markers, Prugnolle & de

Meeus, 2002). Different microsatellite-based estimators are used to measure population structuring, including the F-, R-, and G- statistics. The F-statistics (Wright, 1965) is a widely used measure of genetic differentiation between subpopulations based on

autosomal nuclear markers. The basis of this estimator relies on the proportion of inbreeding as a result of the fragmentation of the population as individuals from the same subpopulation of a fragment will be more closely related than those from other subpopulations in a different fragment. Wright used a series of three inbreeding coefficients (FIS, FIT, and FST) to describe the distribution pattern of genetic variation within a species (Allendorf & Luikart, 2007; Frankham et al., 2002, 2004). While FIS and

FIT measure departure from Hardy-Weinberg proportions (the deficit of the proportion of heterozygotes) within subpopulations or fragments and in the entire base population respectively, FST quantifies the degree of difference in allele frequencies between populations or fragment. A lower FST implies a high level of similarity between two subpopulations assuming there is no influence from natural selection.

The R-statistic is an analog of the later assuming a stepwise mutation model

(Slatkin, 1995). RST can be computed from variance allele size, while FST is derived from the variance of allele frequency. The G-statistic (Nei, 1973), also called the coefficient of genetic differentiation, is a multi-allelic analog of F-statistic (Balloux & Lugon-Moulin,

2002). The interpretation of FST to infer population differentiation can be very dangerous; hence, FST, RST, and GST are usually reported together.

Other microsatellite-based methods to assess population differentiation include relatedness between individuals and the assignment probability (Prugnolle & de Meeus,

2002) . FST can also be computed from uniparental-inherited markers using Weir and

Cockerham's FST estimator, equivalent to ΦST. FST (mtDNA) values indicate the difference among female populations based on mitochondrial markers. FST(Y) is the differentiation in male populations base on Y-chromosome markers. These two

estimators can be used to estimate the female-to-male ratio of the effective number of migrants per generation. All these methods can be used to assess sex-biased dispersal by estimating population differentiation among females and males separately (Ségurel et al., 2008). Thus, if the females are more structured (more closely related) than are the males, it could be inferred that male dispersal is higher (Prugnolle & de Meeus,

2002).

Population fragments of the same species that have limited gene flow between the fragments tend to diverge regarding gene frequencies and become genetically structured over time. While some species will show little genetic structure over relatively large geographic distances like the Canada lynx population (over 3100km), others like the bull trout population within the Pacific Northwest of the United States can be highly structured over a small geographic scale of some few kilometers. Hunter and colleagues (2010) found a significant genetic differentiation between manatees in the

Cayes off Belize City with those in the Southern Lagoon system based on mitochondrial

(ΦST=0.078) and microsatellites markers (ΦST=0.029) with a p-value of 0.05. These two areas are host very different habitat types and are only about 10 miles apart. The unique genetic separation distribution pattern was in agreement with the telemetric study conducted by Auil and co-authors (2007) using VHF, GPS, and UHF transmitters that found reduced movement and strong site fidelity among the tagged manatees staying within 15 miles of their tagging location. Hunter suggested that the distinct habitat and the limited movement between the two sites might explain the observed genetic divergence. Therefore, it would not be surprising to find substructure among the

African manatee subpopulations in the small DRSW given the high contrast differences between habitats.

Genetic divergence between subpopulations under the forces of drift and gene flow only is dependent on the number of migrants irrespective of the population size

(Allendorf & Luikart, 2007; Wright, 1969). Therefore, the level of structuring that is measured through the FST will inform on the degree of manatee movements between the habitats of the DRSW. The information on the pattern and degree of genetic divergence among populations is necessary for designing effective conservation plans for protecting species (Allendorf & Luikart, 2007).

Molecular Genetics and Management Unit Delineation

Biodiversity is a critical concept in biological sciences that refers to the various forms of life and their variability on earth. Biodiversity is expressed through a continuum of the hierarchical dimension of life, including but not limited to genetic, population, species, and ecosystem. Biodiversity is vital for various reasons ranging from intrinsic through aesthetical, cultural, leisure, economic, and ecological values. However, anthropic and natural forces tend to decrease the amount of biodiversity at different dimensions. Therefore, conservationists, politicians, and scientists are concerned about biodiversity loss, which they endeavor to mitigate through legal and management actions. Although protecting each hierarchical level of biodiversity separately is necessary for efficient biodiversity conservation, management resources are always limiting to the efforts; therefore, prioritizing is imperative.

In order to consistently assess the status of species and manage their diversity efficiently, it is indispensable to define and delineate the assessment unit using objective and steady criteria (Green, 2005). A species might be assessed as “not

threatened” at the species level but might have subspecies or population segments that are “endangered” as it is the case of the West Indian manatee and its two subspecies.

While the IUCN conservation status assessment approach is limited to the subspecies level, the Endangered Species Act (ESA), as amended in 1978, includes provision for listing distinct population segments of vertebrates animals (Green, 2005). Although ESA provisions may be efficient, as it would protect a species at a resolution that matches the level of variability of the species, a lack of delineating a distinct population segment

(DPS) for the Antillean manatee in Puerto Rico is, however, politically controversial. In its recent ruling for the West Indian manatee the USFWS through the ESA did not provide criteria for designating a DPS.

Many authors have proposed various criteria for conservation unit designation, and they can be summarized into three categories: genetic, ecological, and threats. The importance of the delineation of a conservation unit is to maximize evolutionary potential while minimizing risks of extinction (Allendorf & Luikart, 2007). To help the ESA delineate a DPS, Waples (1991) proposed the concept of a significant evolutionary unit

(ESU) to identify a distinct population. The meaning of ESU proposed by Waples is different from its original meaning defined by Ryder (1986), which described an ESU as a population unit with a distinct adaptive evolutionary variation identifiable through the concordance of a set of data generated from different techniques. Unfortunately, these data were not always available for species. Unlike Ryder (1986), Waples, (1991) used only two criteria to assign ESU including a strong enough reproductive isolation from other conspecific units and that the population must represent a significant piece of the species evolutionary heritage (how much genetic diversity or ecological niche will the

species lose if that population goes extinct?). Moritz (1994) pointed out the limitation of using only mtDNA markers to assess historically isolated populations and proposed that in addition to the other criteria, ESUs should be reciprocally monophyletic for mtDNA, they should show a significant differentiation of allele frequency at nuclear loci. Crandall and co-authors (2000), considered the reciprocal monophyly ESU criterion proposed by

Moritz (1994) to be over stringent as it ignores adaptive differences between populations that are yet to show as a monophyletic relationship (very long isolation process than polyphyly and paraphyly). Crandall, therefore, proposed a diagnostic to distinguish a population based on ecological and genetic exchangeability and on whether the exchangeability is historical or recent. Finally, Green (2005) highlighted the limitations of the previous ESU definitions that focus on phylogeographical distinctions which are not always guaranteed in some species and on the use of neutral markers

(mtDNA) because of their more rapid mutation rate that might not reflect the level of adaptive variation in a population. To palliate the lack of consensus around the ESU concept, (Green, 2005) proposed the concept of designatable units (DUs), which in addition to the hierarchical phylogenetic criteria, included an independent conservation status (extinction risk) criterion. The author further developed a flowchart of the process discerning and using DUs. Green indicated that the DU should be considered mainly for assessment to support policies or listing agency requirements and should not be applied to the management unit.

(Moritz, 1994) also advised that ESUs should be mainly used for long-term management to set conservation strategies and priorities and that for short-term management, the management unit (MU) concept is more appropriate. The author

considered MUs as subsets of ESUs and defined them as a population showing no reciprocal monophyly for the mtDNA locus (unlike ESU) but presenting significantly divergent nuclear or mitochondrial allele frequencies (irrespective of their phylogeny) resulting from the low connectivity or gene flow between populations. The Moritz MU criteria are the most commonly used (Palsboll et al., 2007). Palsboll and co-authors.

(2007) criticized the fact that several studies that applied the Moritz MU criteria interpreted the term ‘’significant divergence in allele frequencies’’ to mean rejection of panmixia. The author, therefore, proposed that the delineating MU should be based on an observed estimate of genetic population divergence and that the divergence should be corroborated to the dispersal rate of individuals instead of relying solely on the historical degree of gene flow. Hence, the author encouraged the use of non-genetic demographic data such as capture-recapture efforts, telemetry, and others, along with the genetic data.

Overall, we should distinguish three types of conservation units that are all valuable for conservation: ESUs and DUs that are both concerned with long-term conservation needs and use historical divergence and conservation status (DU only), and MUs, also considered the logical unit for monitoring population demography and diversity that are used for short-term management and are defined based on allele frequencies and current population structure.

After analyzing and comparing the different conservation units and the criteria used to assign species in those units, the simplest unit of management for the African manatee across its range, given the current state of knowledge about the species, should be the ESU. The substantial historical phylogenetic divergence (about 2.03

million years before present) between the north and south populations identified by the range-wide phylogenetic study by Keith-Diagne (2014), suggests that African manatees should be delineated into two ESUs. The author used mitochondrial genetic markers to infer limited migration between African manatee populations in the northern and the southern part of their distribution range hypothetically due to a physical barrier near

Sierra Leone formed by the narrow continental shelf (measuring about 145km wide at the northern border of the country, to about 32km wide at the southern border), upwelling of cold water, and convergence of oceanic currents in that area. Although the study did not include representative genetic data from several countries located in the middle part of the species distribution, the strong level of differentiation observed so far could suggest that additional data may increase the significance of the divergence.

Because African manatees are still heavily hunted and bycaught across their distribution range, it would be logical to consider the entire species home range as one DU as the extinction rate appears to be similar across the distribution range of the species.

The study by Keith-Diagne further identified haplotype divergence within the north and the south African manatee populations. For example, the coastal Senegal manatee appears separated from the inland population, and the coastal Cameroon population seems isolated from the Gabon population. The lack of current genetic structure on dispersal among the manatee population makes it challenging to define management units for the species which would improve the efficiency of existing management strategies. The diversity of habitat across the species range most likely suggests a strong evolutionary adaptive variability that needs to be accounted for in the conservation strategies for the species. It is, therefore, urgent to conduct fine-scale

population structure studies along with population assignment, movement, telemetry, and mark capture-recapture studies to determine the current dispersal of the species that would enable identification of discrete populations or usable MUs.

Major African Manatee Studies in Cameroon

Nishiwaki and collaborators (1982) conducted one of the first (if not the first)

African manatee study in Cameroon. The study was carried out in the frame of an extensive field survey the authors achieved across the distribution range of the species

(except Chad) between July and August 1978 and January and March 1981. The survey consisted of gathering information on the African manatee through interviewing fishers, local habitats, biologists, and zoologists. Boat surveys were also conducted for direct observation of the animal. The authors concluded that the Sanaga River in

Cameroon, inhabits the second largest population of African manatees after the Niger

River. In Cameroon, the author found that there might not be manatees on the northwest coast of Cameroon, especially in rivers flowing along Mount Cameroon. The author attributed the absence of manatees in that area to the high rocky coast with volcanic lava shoreline. The southern coasts, on the contrary, were low swamp and sand and appeared to have abundant manatees. The author was informed that there were manatees in the upstream of the River Mungo, near the Cameroon estuary. The authors also interviewed residents at a fish landing site market in Edea near the Sanaga

River, who called the manatee “Maga” in the local language and reported that they are abundant in the river. Despite the Cameroon law protecting the manatees, they were reported being killed for the meat to entertain dignitaries. Finally, it was also reported that there were two manatee hunters near Mouanko (close to the Sanaga River estuary) who possessed a license to kill manatees, probably for local celebrations. Providing

continued education programs and promoting alternative livelihood in areas where communities kill manatees might contributes to changing their negative attitudes toward the species.

Grigione (1996) conducted the first comprehensive distribution study of the

African manatee in Cameroon. Between June and August 1989, the author conducted land and boat surveys in four regions of Cameroon, including Korup and Mamfe on the northwest coast, and Edea and Kribi on the south coast. The author used the frequency of reported sightings to estimate manatee density and found that in the four regions except for Kribi, manatee density was high. The author reported that manatee killing does not appear to be common and attributed this to the fearful attitudes and perceptions of the local community towards the manatee. However, the latter observation is in discrepancy with the finding of Nishiwaki et al. (1982), Powell (1996), and Takoukam Kamla (2012), who reported that manatees have been heavily hunted in

Cameroon. The author also reported that manatee abundance was high in the Akpa

Ndian, Cross, Munaya, and Sanaga rivers. The abundance was medium in Akpa Yafe,

Akpasang, Rio del Ray, Dihende, and Dipombe rivers. The abundance was reported low in Lake Ossa and Manyu River. The low abundance of manatees in Lake Ossa is in discrepancy with more recent literature (Takoukam Kamla, 2012). The manatees were reported absent in the Akegam and Awa rivers. Finally, within the Edea region, the same area where we conducted this study, the author found that the highest concentration of manatees was in the Sanaga River.

Between 1986 and 1995, Powell (1996) conducted the second large-scale survey of the African manatee throughout its entire distribution range. In this survey, the author

added beyond distribution and qualitative abundance, including habitat use and preference, seasonal movements, diet, threats, and local attitudes toward manatees. In contrast to the study of Nishiwaki (1982), Powell reported that manatees are not uncommon in the mangroves and coastal creeks of the southern coast of Cameroon despite its rocky shoreline. The author suggested that manatees may be using that coastline corridor to travel between the Rio del Rey area to the mangrove estuary near

Tiko. Powell also reported manatees in the Wouri River and two places there where manatee meat is frequently sold. To date, manatee meat is still sold in those two locations (one of them is called Marche Saker). The author found that manatees are common in Lake Ossa and that it may constitute a manatee refuge during the low-water season. In contradiction to the Grigione (1996) study, Powell reported that the lower reaches of the Nyong and Ntem rivers in the Kribi region are inhabited by manatees. He also pointed out that there are manatees in northern Cameroon more precisely in the

Benue and Faro rivers and in Lagdo Lake; It is now believed that manatees are no longer in Lagdo Lake since the construction of the Lagdo reservoir dam (Keith-Diagne pers. comm.).

Between 1995 and 2011, there appears to have been no dedicated manatee studies in Cameroon until the recent study conducted by Takoukam Kamla (2012). This author conducted interviews among fishers near the Douala-Edea Wildlife Reserve

(DEWR) and Lake Ossa Wildlife Reserve (LOWR) to confirm the presence of manatees in the Sanaga River, River Nyong, the coastal zone, Lake Ossa, and Lake Tissongo.

His studies revealed that the lower reaches of River Sanaga around Malimba village may have the highest density of manatees and the highest number of manatee carcass

report in the region. The leading cause of death is attributed to illegal hunting. Lake

Ossa and Sanaga estuary appear to provide a refuge for manatees in the Sanaga watershed during the dry season; while manatees are more abundant in the Sanaga

River during the high-water season where they take advantage of the flooded forest.

The author documented fisherman-manatee conflicts in DEWR and LOWR. Fishers proposed compensation for their torn nets as a solution to end the conflicts. The author also reported that fishers in DEWR and LOWR have an excellent and clear understanding of manatee life, their behavior, and habitat needs. Their perception might vary from one habitat to the other, reflecting the difference in the status of the manatee across the various regions. The author also conducted boat surveys in LOWR and indicated that the Lake Mevia site within the Lake Ossa complex provide the highest probability of sighting manatees during the dry season, with 05:00-10:00 hours being the best time of the day for manatee sightings at this site, and 18:00-22:00 hours the best time for sighting them at the other sites of the lake. He also noted that the

Lindema-Mevia canal might be the most crucial manatee feeding area in Lake Ossa during the dry season. Manatees are continuously hunted in Lake Ossa. Entanglement in nets and Chinese bamboo fish traps placed underwater by fishers is another threat to manatees in Lake Ossa (Mayaka et al., 2015).

In 2013, Mayaka and collaborators conducted an ethnobiological study of 174 local resource users in the Lower Sanaga Basin to determine the conservation status of the African manatee. The main finding of this study was that most interviewed users

(60%) sighted manatees at least once a month and perceived that the manatee population was either constant or increasing. They attributed this belief to increased

awareness. The interviewed users also reported that intentional and incidental catches, and habitat degradation by pollution, were the main threats to manatees. Accidental catches were reported to be more frequent in lakes. Boat collisions with manatees were reported as negligible in the entire study area.

Keith-Diagne (2014) conducted the first comprehensive range-wide genetic analyses of the African manatee in Cameroon. The result of this study is summarized in

Chapter 5.

The most recent study to date of the African manatee in Cameroon was published by Mayaka and collaborators (2019). This was the first extensive study of the species on the northwest coast of Cameroon. The authors conducted a semi-structured interview survey of 101 fishers in the southern Korup region to determine sightings, conflicts with humans, and local perception patterns of the species in the area. Contrary to the observation of Nishiwaki (1982), the authors discovered that manatees are present and abundant in the southern Korup area and that most sightings occurred in waterway intersections and at bends in rivers. Manatee sightings in the small upstream rivers was only during the wet season and only in the dry season in the coastal Ndian

River. The study also documents various forms of human-manatee conflicts, including crop-raiding, net destruction, and fish theft from fishing nets. In retaliation for the loss caused by manatees, fishers would kill manatees. Finally, the authors also described that the negative perceptions towards manatee decreases with the awareness of manatees’ legal protection and increases with the age and education level of the locals.

The authors highlighted the importance of raising awareness as a tool for manatee conservation.

Conservation Status of the African Manatee

The African manatee is the least studied of all sirenians including the West Indian manatee in the Caribbean, the Gulf of Mexico and the East Atlantic coast, the

Amazonian manatee in the Amazon River, and the dugong in the western Pacific Ocean and the eastern coast of African (Marsh et al., 2011). The wide distribution range of the

African manatee across the Atlantic coastal and inland waters of West and Central

Africa contrasts with the poor economic status of their countries of distribution and the minimal number of trained scientists studying them. The World Bank classifies all the 21 distribution countries of the African manatee except Gabon and Equatorial Guinee as low-income countries (World Development Report, 2016). Most low-income countries orient their limited financial resources to priorities other than wildlife research and conservation to meet the basic needs (food, shelter, and clothing) of humans (Nishiwaki et al., 1982), thus resulting in the low research effort on the African manatee.

Furthermore, aquatic ecological research poses challenging logistics and needs more financial resources when compared to land-based studies and also requires higher skill sets (Aragones et al., 1997; Munguia & Ojanguren, 2015), resulting in most ecological researchers in these areas orienting their interest on terrestrial species rather than marine or aquatic species, such as manatees.

Concomitantly, poverty in these countries has led to poaching of the African manatee by the local communities (Marsh et al., 2012). The species is hunted mostly for the meat, however, in some countries like Sierra Leone, Nigeria and Niger, manatee oil and the male genital organ is commercialized for medical and aphrodisiac uses (Powell,

1996). Although the contribution of poaching to African manatee mortality across its range is unknown, there is no doubt that it is the greatest threat and causes more

damage to the population than other threats, such as bycatch, habitat degradation, pollution, and climate change.

In Cameroon, Lake Ossa and the downstream of the Sanaga River provide a dry season sanctuary for manatees within the basin region (Powell, 1996). In these areas, they face constant and intensive anthropogenic pressure, including hunting, degradation of habitat via deforestation, pollution from industries and agriculture, accidental collision with boats, and entanglement in fishing nets (Powell, 1996). Recently, the species in

Lake Ossa started facing a new threat: the destruction of their feeding habitats by the proliferation of Salvinia molesta, an invasive aquatic plant from Brazil (AMMCO, 2019).

Figure 1-1 presents a schematic illustration of the threats that affect manatees in

Cameroon and the primary and secondary factors that fuel those threats.

The threats to manatees are persisting even though all the countries of its distribution have enacted legislative laws to protect the species (Powell 1996). African manatees are also under the protection of international laws. They are Red-listed by

IUCN (The International Union for Conservation of Nature) as ‘’Vulnerable’’; implying that there is a high probability of a 30% or more reduction in population size will result within a three-generation period (Keith-Diagne, 2015); thus the species should be provided a higher level of protection. They were recently upgraded to Appendix I of the

CITES (Convention on International Trade in Endangered Species) during the 2013

Conference of the Parties in Bangkok, Thailand; therefore, the commercial trade of the species is more restricted. In 2009, the CMS (Convention on Migratory Species) also upgraded the African manatee to Appendix I (endangered migratory species) as they were assessed as being in danger of extinction across a significant portion of their

distribution range (CMS, 2017). Because of this CMS status, countries that are parties of the CMS shall strictly protect the African manatee, especially by removing or mitigating obstacles to their migration as well as other factors threatening their populations.

Even though most of the countries within the distribution range of the African manatee have provided a legal protected status for the species, the implementation of these laws is weak, and management strategies intended to protect the species are insufficient. Nonetheless, in 2008, the CMS memorandum of understanding (MoU) concerning the conservation of the African manatee had been established and entered into effect. The MoU covers 29 range States, and as of August 2012, 17 countries had signed the agreement. The goal of the MoU is to improve the conservation status of the

African manatee throughout its distribution range through the application of strategic policy, research, conservation, and awareness actions (CMS, 2012). The related action plan included in the MoU comprises four objectives and themes including (1) improve policies and legislation for manatee protection, and strengthen their implementation, (2) improve understanding of the species and use information for its conservation management, (3) reduce pressures on them through the restoration and safeguarding of its habitats, and (4) promote a broad appreciation of the species and its ecological and cultural values (CMS, 2012a). This document guides the conservation efforts of the species across its distribution range and justifies the purpose of our study that aligns with the second objective of the action plan.

Overall, the species is believed to be declining across a significant portion of its range despite the lack of a systematic quantitative data on the mortality, threats, habitat,

or population size (Keith-Diagne, 2015). Effective conservation of the African manatee is unachievable without robust and accurate baseline information. Therefore, cost- effective research approaches (requiring only minimum funding and logistics, but generating relevant data), are most needed in Africa for the protection of manatees.

Figure 1-1. Schematic illustration of the threats to the African manatee population and habitat in Cameroon and the chain of factors causing the threats. The red boxes represent the threats, and the orange boxes represent the factors, and the arrows indicate the interaction between the factors, the threats, and the target (green circle).

Figure 1-2. Manatee distribution in Cameroon (map on the left, adapted from Grigione, 1966) and study site showing the Douala-Edea National Park and Lake Ossa Wildlife Reserve (map on the right). The area circled in red is the Downstream of the Sanaga River watershed. Map source: Google Map.

CHAPTER 2 ASSESSMENT OF THE LAKE TROPHIC STATE MODELS IN PREDICTING SUBMERGED AQUATIC VEGETATION AND ITS IMPLICATIONS FOR MANATEE CONSERVATION AT LAKE OSSA, LITTORAL, CAMEROON

Background

During the past two centuries, the level of anthropogenic input of nutrients to the

Earth’s surface waters has increased as a result of the industrial revolution and population growth(Galloway et al., 2004; Lee et al., 2016) .This has led to a profound change to ecosystem on the planet. The large increase in the of rate of land clearing, domestic animal husbandry, agro-industry, urbanization, dam construction, and other human activities has contributed to increased nutrient enrichment (eutrophication) in fresh and marine water systems, thus altering the natural hydrological cycles (Smith et al., 1999; Vitousek et al., 1997). Such increases can lead to uncontrolled proliferation of algal biomass, resulting in hypoxia and increasing light attenuation in the water column; thus, limiting the growth of submerged plants.

Data collected in 1985 (Wirrmann, 1992) suggested that Lake Ossa was mesotrophic based on the trophic state classifications by Carlson (1977). Values reported by Wirrmann (1992) for total nitrogen, total phosphorus, and chlorophyll concentrations in the lake were 157μg/l, 12.9μg/l and 8.4μg/l respectively. Recent data collected by Ndjoya (2015) indicated that the lake might be in a eutrophic state based on the Carlson classification. The concentration of the ammonium ion alone was

1220μg/l and 12μg/l for nitrite. The value of the orthophosphate (which is an underestimation of the total phosphate) was between 20μg/l and 100μg/l. Secchi depth measurements varied between 16cm (0.5 foot) and 100cm (3.3 feet). More recent values of chlorophyll after 1985 are not available.

Although nitrogen and phosphorus are essential nutrients for aquatic plant growth, their concentrations in water do not limit the growth of rooted macrophytes because they get nutrients from sediments (Canfield & Hoyer, 1992; Hoyer & Canfield,

1996; LAKEWATCH, 2007a). There are four broad categories of aquatic plants including rooted emergent (stem and leaves above the water surface), rooted floating- leaves, rooted submerged, and unrooted free-floating plants (LAKEWATCH, 2007a).

Rooted macrophytes mainly acquire nitrogen and phosphorous from sediments, but some emerged plant species can also capture atmospheric nitrogen. Thus, the productivity of these rooted plants depends on the lake bottom substrate, which in turn depends on the geology of the lake watershed (Canfield & Hoyer, 1988a). In Lake

Ossa, nutrients are mostly provided by run-off from the lake catchment basin or by the drainage from the basin of the Sanaga River that discharges highly turbid water into the lake (Wirrmann & Elouga, 1998). The sedimentation rate of Lake Ossa is relatively high, varying from 45cm to 192cm/1000 years (Giresse et al., 2005). Therefore, rooted submerged and emergent plants in Lake Ossa have large stores of phosphorous and nitrogen and their growth is probably not limited by nutrient availability but by the amount of light reaching the surface of their leaves.

Light availability at water bottom can be a limiting factor for submerged vegetation growth, as these plants need light for photosynthesis. Photosynthetic plants require light to build their biomass from atmospheric or dissolved carbon dioxide. Owing to the high attenuation of irradiance (flux of light penetration per unit area) through the water column, areas of the lake with greater depth might not receive enough light to grow. Canfield and collaborators (1985)established a positive linear relationship

between water transparency as indicated by Secchi depth and the maximum depth of plant colonization (MDC) based on data from 26 Florida, 27 Finland, and 55 Wisconsin lakes. The author developed an empirical model that predicts MDC values from Secchi depth. There they found that the 95% confidence interval ranged from 46% to 236% of the estimated MDC. The large variability in the MDC-Secchi depth relationship model was hypothesized to be caused by the variation in light compensation points (i.e., intensity of light for which the photosynthetic and respiration rates are equal) among species. A more robust experiment by Caffrey and co-authors (2007) confirmed the

MDC-Secchi depth relationship model of Canfield et al. (1985). This empirical relationship has not yet been demonstrated in Lake Ossa. Therefore, in this study, we examine whether the distribution of submerged vegetation in Lake Ossa can be predicted using MDC estimated from Secchi depth, with the assumption that light is the only limiting factor for biomass development.

Chlorophyll concentration (which reflects the productivity of algae), dissolved organic matter, and suspended sediments can independently influence transparency as measured by Secchi depth (Vant & Davies-Colley, 1984). Chlorophyll concentration also influences water transparency as indicated by Carlson (1977) and Canfield & Hodgson

(1983). Those authors demonstrated a significant relationship between chlorophyll concentration and water transparency in several water systems. Chlorophyll concentration, in turn, can be influenced by nutrient concentration especially phosphorus; a strong empirical relationship between chlorophyll concentrations and total phosphorus has been demonstrated in the world north-temperate lakes (Canfield,

1983) especially when phosphorous is limiting. It, therefore, appears from what

precedes that a chain of relationship models can be established between nutrient, chlorophyll concentration, and water clarity. The water clarity, in turn, can be used to predict MDC and therefore submerged vegetation biomass.

The predictive power of such a chain of models can vary greatly depending on the water system. For example, determinants other than chlorophyll concentration, such as suspended sediments and dissolved organic matter, can influence water transparency. If phosphorus is not limiting, then the relationship between nutrient concentrations and chlorophyll a will not be significant. Hoyer and co-authors (2015) examined this chain of empirical eutrophication models in a coastal environment

(Choctawhatchee Bay in Florida) to predict the impact of suspected eutrophication on submersed aquatic vegetation. The authors observed a significant relationship between total phosphorus and chlorophyll concentration and between chlorophyll concentration and Secchi depth. However, there was no significant relationship between the Secchi depth measurements and the seagrass distribution because light was not the limiting factor. Trophic state chain models have not been examined for Lake Ossa. Such an approach could be useful for managers of the Lake Ossa Wildlife Reserve to better understand the distribution of submerged vegetation which may impact the distribution of manatees.

Water quality and quantity of a waterboy may influence manatee habitat and their use of the habitat. Manatees are herbivorous, thus rely on aquatic plants for food. The presence and abundance of submerged aquatic vegetation (SAV) is strongly dependant on the physical and chemical properties of the the water body in consideration. Nutrient- enriched water bodies may experience an increased in biomass productivity reflected by

an increased chlorophyll concentration, which in turn affects the water clarity and limit the percentage of surface irradiance available at the bottom for the photosynthesis of

SAV (LAKEWATCH, 2007). Thus, less SAV would imply less food options for the manatees. Water depth can also influence manatee habitat use as the species preferred areas where water is atleast 2m deep (Reep & Bonde, 2006). In Florida, the nutrient enrichment of Lake Okeechobee was suspected to be at the origine of the harmful algal bloom events that have caused the death of hundreds of Florida manatees

(Broadwater et al., 2018). Therefore, the limonological knowledge of manatee habitats can be crutial for the management of the species.

Understanding the biological, chemical, and physical components of Lake Ossa is essential for understanding how African manatees utilize their habitat. African manatees are mostly herbivorous, although they have been reported feeding on fishes and mollusks (Keith-Diagne, 2014). Manatees rely heavily on macrophytes as a food source and can eat an equivalent of 10 to 15% of their body weight of plants daily (R.

Reep & Bonde, 2006). Manatees use floating macrophyte beds to hide when the risk of detection by fishers is high. Thus, the aquatic plant productivity of Lake Ossa appears to be a major limnological component that determines the carrying capacity and the spatiotemporal abundance of manatees in a water system.

The purpose of this study was to assess the trophic state and establish the bathymetry of Lake Ossa to determine their potential implications on SAV and on habitat use by the its population of African manatee.

The objectives of this study were: (1) to assess the current trophic state of Lake

Ossa, (2) to examine an empirical chain of trophic state models in Lake Ossa, and (3) to

determine the relationship between the current trophic state of the lake and the abundance of submerged vegetation. The latter is important because submerged vegetation could provide an alternate food source for manatees during the dry season when water surface recedes and becomes distant from the emergent bank vegetation, the primary source of available food for local manatees. Therefore, we will answer the following questions: (1) is there a significant relationship between nutrient concentrations and chlorophyll concentrations, (2) does water transparency in Lake

Ossa significantly correlate with chlorophyll concentrations, and (3) is water transparency a limiting factor for submerged vegetation viability in the lake?

Methods

Study Area Description

The Lake Ossa complex is located at 13km from Edea, Cameroon, between the

3°45’ and 3°52’N latitude, and 9°45’ and 10°4‘E longitude with approximately 300m elevation (Wirrmann & Elouga, 1998). The water surface is estimated to be 4000ha. The lake represents about 90% of the Lake Ossa Wildlife Reserve, created in 1968. The reserve was established to provide a refuge for the protection of the African manatee.

Mévia in the north of the lake complex is 700ha; Ossa has the largest water surface area of the lakes with 37,000ha, and Mwembé the smallest (300ha) is located at the most southern part of the lacustrine complex. The Lake Ossa complex is shallow with a

maximum depth of about seven meters during the wet season (Giresse et al., 2005).

The greatest width of the lake is about 7km.

Lake Ossa is adjacent and connected to the Sanaga River through a 3km meandering and relatively deep channel (5 to 9 meters). Hence, the seasonal incursion of Sanaga water (brownish and full of sediment) into the lake appears to influence its water color and chemistry. The water transparency of the lake is not homogeneous across its surface. The banks are mostly covered with weeds dominated by antelope grass (Echinochloa pyramidalis) and in some places with shrubs and rattan (Calamus sp). Some of the shorelines have a discontinuity of sandbanks that are the preferred sites for the reproduction of softshell turtles (Trionyx triunguis). In the middle of the rainy season, a large proportion of the littoral forest becomes flooded, providing fresh plant food and refuge for the African manatee and other aquatic species. The Lake Ossa complex lies on a yellow ferrallitic bed formed during the Palaeocene. The soil is abundant with iron-rich shales and black marlstones. Successive clay and sand cover shale and black marlstone layers (Giresse et al., 2005).

About 17,000 human inhabitants live around the lake (BUCREP, 2010), and the majority of them depend directly or indirectly on the lake’s resources to survive. Over

300 fishers are active on the lake, especially during the low water season (Takoukam

Kamla, 2012). Women of the local community utilize the lands around the lake for agricultural purposes. Additionally, there is a large agro-industry established on the south-west portion of the lake, which has rubber and palm tree plantations covering a surrounding area as large as the lake itself.

Data Collection

Water chemistry

A total of 18 sampling stations was established. They were distributed evenly across the lake (Figure 2-1). Water sampling was conducted at each station monthly from May through August 2016 and consisted of measuring the Secchi depth directly and total nitrogen (TN), total phosphorous (TP) and total chlorophyll (TChl) concentrations following the LAKEWATCH standard operation procedures (Canfield et al., 2002; Hoyeret al., 2012). A Secchi disc was used to measure the water clarity.

Chlorophyll was filtered immediately on the field from 300ml of the lake surface water using a handheld water pump and a Merck Millipore filter paper. Another volume (50ml) of surface water was collected for TP and TN concentration measurements. The filters and the 50ml water samples were frozen (-5°C) until shipped to the Fisheries and

Aquatic Sciences Laboratory of the University of Florida for analysis. The depth and

Secchi depth data of the lake has been monitored in the course of the ongoing Manatee

Monitoring Program (MMP) implemented by the African Marine Mammal Conservation

Organization (AMMCO). AMMCO established five permanent sampling stations (Figure

2-1) since January 2016 and has been recording Secchi depth, water depth, manatee presence, and other biophysical parameters monthly for four to five consecutive days.

These data were included in this study to assess the seasonal variation in these variables.

Rainfall

A weather station (Vantage Pro by Davis Instruments) was installed at the

AMMCO office in the south of Lake Ossa (Figure 2-1). The device was installed in

January 2016 and collected daily rainfall and uploaded to a cloud account from which monthly averages for rainfall were derived.

Bathymetry and submerged aquatic vegetation (SAV)

Lake Ossa, like the rest of the lakes of Cameroon, does not have any existing bathymetric data. Therefore, we conducted the first bathymetry measurements of Lake

Ossa. The bathymetry and SAV mapping of lakes has become easier with the advent of hydroacoustics global positioning systems (Abd Ellah, 2016; Valley et al., 2005). The hydroacoustic and GPS data were recorded simultaneously every two to three meters during the survey using the Lowrance Sonar (Lowrance HDS 9 Gen 3) with a built-in

GPS capability. The single-beam, 200-kHz transducer with a 20° beam angle of the

Lowrance sonar was mounted on the transom of a small boat that was used to follow predefined-parallel transects relatively spaced at 100-m intervals. The boat speed was maintained at approximately 7km/hr. The principle of the translation of acoustic signals to the depth and biovolume data is described in detail in Valley and co-authors (2015).

The biovolume refers to the average percentage of the water column taken up by vegetation. The lake was surveyed from 04 to 10 September 2016, during the highest water season of the year. The raw data (in SL2 format) was downloaded from the

Lowrance device and uploaded to the BioBase server for data processing. BiobBase is a cloud-based software platform that generates GIS data layers of depth, biovolume, and bottom hardness from an acoustic and GPS signal (Valley et al., 2015; Valley,

2016) collected. The processed data was delivered as an interpolated grid point in a comma-separated format (CSV). The data accuracy of the SAV data was accessed by ground-truthing at 50 randomly selected points including 25 with and 25 without vegetation as detected by the Lowrance sonar.

A weather station (Davis Instruments Vantage Vue Weather Box-6260) was placed at AMMCO’s Office located near the southern edge of the lake, in the town of

Dizangue. The station recorded rainfall, outdoor air temperature, pressure, and wind direction for at least two years starting in January 2016.

Data Analyses

The R-STUDIO 1.1.153 and ArcGIS Pro 2.1.3 software was used for all statistical and geoprocessing analyses.

Water chemistry and rainfall

The monthly values of the water chemistry variables (Secchi depth, TP, TN,

TChl) were averaged by the three lake zones delineated perpendicularly to the water flow in the lake (south, middle, and north as illustrated in Figure 2-1) and for the area as a whole (Table 2-3). Extreme outliers (greater than 3 standard deviations) were discarded. The average TN/TP ratio was computed to determine whether nitrogen or phosphorus is a limiting factor to algae growth. A simple linear regression model was assessed between TN and TChl, TP and TChl, Secchi depth and TChl, and Secchi depth and distance of the water station at the outlet of the lake. These variable values were also compared to those collected in May 1985 by Wirrmann (1992) in the same lake to assess the dynamics of the lake trophic state.

The Secchi depth and water depth data from the MMP, as well as the rainfall data recorded in 2016 and 2017, were averaged by month. The maximum depth of plant colonization (MDC) values at each of the water stations were predicted based on the

Lambert-Beer equation (2-1):

퐼푧 −ln⁡( ) (2-1) 푀퐷퐶 = 퐼표 K

Where Iz is the light intensity at depth z, I0 is the light intensity at the surface,

(Iz/I0) is the percentage of light at MDC of the submerged aquatic plant of interest.

(Chambers & Kalff, 1985) assessed the maximum depth of colonization (MDC) for charophytes and angiosperms and estimated that on average, they occur at 11% and

21% of the surface incident irradiance respectively. Because angiosperm plants dominate Lake Ossa, a maximum target depth of SAV colonization was set at 20% (i.e.,

Iz/I0 = 0.2). K is the light attenuation coefficient usually measured by a light meter. A lightmeter was not available for use in this study, so we used an alternative estimator based on Secchi depth (SD) values following the equation by Poole and Atkins (1929) which is applied worldwide:

1.7 (2-2) 퐾 = SD

The predicted MDC values along with the Secchi depth and depth were plotted on the same date scale graph to graphically visualize how likely SAV will occur over the year (Figure 2-4).

Spatial analysis and prediction

The bathymetry grid point data were plotted on ArcGIS and converted into a 16m cell size raster layer using the conversion tool of the software. Then the cells were classified by equal interval depth ranges of one meter (Figure 2-5).

The light attenuation coefficient was used to predict the percentage of surface irradiance (I0/Iz) at the bottom (of depth Z) of each of the raster cell based on the

Lambert-Beer equation

퐼푧 (2-3) = 푒−퐾푍 퐼표

Where Iz and Io are the light intensities at the bottom (depth z) and the water surface respectively. Given the K proxy in equation 2-2, the percent irradiance at a bottom z (Equation 2-3) can be rewritten as:

퐼푧 (2-4) = 푒−1.7푍/푆퐷 퐼표

Thus, the percentage irradiance depends only on the depth z (bathymetry data) and the SD values. Because there was a strong relationship between SD values (Y) at each sampling station and the Euclidean distance (X) of the station to the outlet of the lake, the linear regression (Equation 2-5) was used to predict the SD value at each raster cells using the raster calculator. The Euclidean distance is the straight-line distance between two points.

푌 = 0.073⁡푋 + 0.69 (2-5)

The straight linear distance (X) of each raster cell to a source cell situated at the outlet of the lake was calculated using the Euclidean Distance function of the Spatial

Analysis tool of ArcGIS. Then Equation 2-5 was applied to the Euclidean Distance raster layer to generate a predicted SD layer (Figure 2-6) using the Raster Calculator function of the Spatial Analysis Tool. The predicted SD and the bathymetry raster layers were used to generate the predicted percent irradiance at bottom (Iz/I0) layer by applying Equation 2-4 using the Raster Calculator function. Finally, the latter layer was classified (Figure 2-7) as follows:

• Class 1: Raster cells predicted to receive <15% of surface irradiance at bottom; and therefore, assumed to bear no SAV.

• Class 2: Raster cells predicted to receive between 15% and <25% of surface irradiance at bottom and expected to bear moderate SAV if only light availability was limiting.

• Class 3: Raster cells predicted to receive ≥25% of surface irradiance at bottom and therefore expected to bear abundant SAV if only light availability was limiting.

The observed SAV grid point data collected through the Lowrance Sonar was processed using the same procedure as with the bathymetry data described above.

Cells with biovolume <10% were classified as noise or insignificant (Class 1). Cells with biovolume between 10% and <25% were classified as moderate SAV (class 2) and abundant SAV (Class 3) were cells with biovolume ≥25% (Figure 2-8).

Accuracy assessment

To assess the correlation between our predictive model and the observed SAV distribution, a confusion matrix (Boughorbel et al., 2017) was constructed from a random subsample representing 20% (35708 cells) of the total cells across the lake. For each of the sub-sample cells, corresponding values from both the predicted SAV layer

(Figure 2-7) and the observed SAV (Figure 2-8) as detected through the Lowrance

Sonar were extracted using the Extract by Point function of the Spatial Analysis Tool of

ArcGIS. The extracted data was then transformed to a binomial data where class 1 of both layers were assigned to “absence,” and class 2 and 3 of both layers were assigned to “presence.”

The number of true and false positive cells, as well as the number of true and false negative cells, were counted. True positive (Tp) and true negative (Tn) are cells for which both the observed and the predicted values correspond to “presence” and

“absence” respectively. False-positive (Fp) are cells that observed SAV value is

“absence” and predicted value is “presence.” False-negative (Fn) are cells that

observed SAV value is “presence” and predicted value is “absence” (Figure.2-9). The confusing matrix performance metrics (Fawcett, 2006; Matthews, 1975) were calculated including the error rate, accuracy, sensitivity, specificity, precision, false positive rate, and Matthew correlation coefficient (Table 2-1).

Results

Water Chemistry and Rainfall

All water chemistry variables indicated a level of gradient from the south (around the outlet of the lake) towards the north (farthest areas from the outlet). The average values of TN, TP, and TChl were highest in the south and were lower in the middle and the northern zones of the lake. The overall average values (TN, TP, and TChl) for the lake were more than double the values recorded by Wirrmann (1992) in May 1985

(Table 2-2). The average Secchi depth and the ratio of TN to TP concentrations (TN/TP) were lower in the south and middle and highest in the north zone of the lake (Table 2-3).

The whole lake TN/TP ratio grand mean was greater than 17 (19.3), suggesting that

Lake Ossa chlorophyll growth is most likely phosphorus limited at least from May through September (corresponding to the sampling period of this study; Sakamoto,

1966; Val H. Smith, 1982).

There were significant but weak positive linear relationships between TP and Chl concentrations (Figure 2-2A), between TN and Chl concentrations (2-2B) with phosphorus and nitrogen accounting respectively for 16% and 8% of the variance in Chl concentrations of Lake Ossa. There was also a significant, but weak negative linear relationship between Secchi depth and TChl concentration with the later accounting for

24% of the variance in the observed Secchi depth (Figure 2-3A). The Euclidean distance of the sampling points to the outlet of the lake surprisingly showed a significant

relationship with the average Secchi depth at each sampling point (Figure 2-3B). The distance of the sampling point to the outlet accounted for 76% of the variance in the observed Secchi depth (r=0.76, p<0.05). Further assessment, using multivariate linear regression analyses with Secchi depth as the dependent variable and TChl concentration and distance of the sampling point to the lake outlet as the explanatory variables, showed that only the distance of the sampling was significant (P=0.001). After accounting for the distance of the sampling point to the outlet, TChl concentration accounted for only another 2.8% of the variance in Secchi depth (R2= 0.78, p<0.05).

The monthly mean rainfall recorded in Lake Ossa over the two years (2016 and

2017) identified two seasons (Figure 2-4A); the rainy season that goes from April to

October with a short peak (about 250mm) in May and the highest peak in August or

September (about 500mm), then a dry season from November to March. The monthly variation of the average water level in the lake follows approximately the same variation pattern of that of the rainfall with a delayed peak at a month from the main rainfall peak

(Figure 2-4B). The water level increased by about 2.5m from the dry to the rainy season.

The monthly mean Secchi depth measured in the five permanent stations during the two years and the predicted MDC (using equation 1) follow the same variation pattern with the monthly mean water depth (Figure 2-4B). Monthly Secchi depth varied from 0.57m to 1.39m and MDC from 0.54m to 1.31m. The MDC yearly curve (Figure 2-

4B) appears to be way above the lake bottom throughout the year. This gap between

MDC and the bottom is larger (up to 3m) during the high-water season and smaller during the low-water season (at least 1m).

Spatial Analysis and Prediction

The bathymetric map showed that Lake Ossa is a shallow water system with depths less than 5m within the lake basin and up to 9m in the channel connecting the lake to the Sanaga River (Figure 2-5). An average increase of 2.5m of the water depth from the low dry to the high rainy season implies that water depth in the entire lake basin is less than 2.5m during the dry or low water season, with most of the lake having a depth under 1.5m. The water surface areas during the high-water season were estimated to be at least 4900ha and the volume to be at least 161 million cubic meters.

The spatial distribution map of the Secchi depth (Figure 2-6) was predicted based on the linear regression model equation (Y=0.073 X+0.69) explaining the relationship between the Secchi depth and the distance of the sampling point to the outlet of the lake (Figure 2-3B) because it explained about 76% of the variation in the

Secchi depth values. The predicted Secchi depth map showed a strong, increasing gradient moving from the south to the north of the lake with the value varying from 0.7m to 1.65m.

Mean K estimated values for each of the 18 water sampling sites during the four sampling months varied from 1.04 to 2.74m-1 with a grand mean of 1.71m-1. The monthly mean K estimated values for the five permanent sampling sites together for two years (2016 and 2017) varied from 1.2 to 3.0m-1 with an annual mean of 2.1 and 1.8m-1 for the years 2016 and 2017 respectively. The monthly direct normal irradiance recorded in Lake Ossa in 2016 varied from 42473Wm-2 in Aug to 112918Wm-2 in

January (ARMINES, 2018).

The model predicted that only about 3.7% of the lake surface bottom received more than 15% of the surface light irradiance and therefore was more likely to sustain

submerged vegetation. The spatial model predicted that the SAV is distributed only on the edges of the lake and more abundant toward the middle and in the north zone of the lake (Figure 2-7). Similarly, the observed SAV data indicated that only about 2% of the bottom is occupied with SAV which has a similar marginal distribution pattern of a predicted distribution (Figure 2-8). Both the predicted and the observed values indicated that SAV is almost entirely absent in Lake Ossa.

The confusion matrix analysis revealed that the prediction model performed strongly (Table 2-1, Figure 2-9). The error and the false-positive rates were only 4% and

3% respectively, while the accuracy and the specificity were 96% and 97%, respectively. The sensitivity (29%) and the precision (14%) of the model were both moderate.

Discussion

This is the first comprehensive limnological study conducted for Lake Ossa since

1985. Updating the information about the trophic state of the lake was necessary for a retrospective understanding of the major natural and anthropogenic driving factors responsible for the dynamics of lake chemistry. This update was also important to make hypothesis and predicton on how water quality and quantity could affect African manatee habitat suitability in Lake Ossa during the low and the high water season. The trophic state of a lake influences the plant and wildlife diversity and abundance of its system through the trophic network that interconnects the elements of the ecosystem. In the current case, we applied our knowledge of water chemistry to appreciate the abundance and distribution of potential submerged manatee food plants. Such information is necessary to develop hypotheses on the movement of the manatee during the low water season when the emergent food plants become limited, and will

ultimately inform management plans and conservation strategies designed to protect the species.

The trophic state represents the total amount of biomass sustained in a water body or its level of biological productivity as a response to its nutrient load

(LAKEWATCH, 2007b). Lake Ossa trophic states parameters (TN, TP, and TChl) doubled in three decades, moving from a mesotrophic to a eutrophic state based on

Carlson’s trophic state classification (Carlson, 1977), which is the more commonly used trophic state classification index (Devi Prasad & Siddaraju, 2012). Thus, the lake system went from a moderate to a high level of biological productivity. However, the lake appears to sustain only a limited level of fish and submerged plant productivity because of its poor water clarity that limits the amount of light available for phytoplankton growth, which in turn reduces fish biomass and diversity. Fishers in Lake Ossa reported that fish have become scarce compared to the time of their fathers who spent a less amount of effort on a greater amount of catch (pers. comm.).

The average value of TP (28.15μg/l) in Lake Ossa measured in this study was almost half of the mean TP (49μg/l) of Florida lakes based on data set from 508 Florida lakes (Canfield et al., 2018); and mean TN (340.9μg/l ) in Lake Ossa was three times lower than the mean value of Florida lakes (984μg/l). However, the mean TChl

(19.8μg/l) and Secchi depth (1.17m) in Lake Ossa were just slightly lower than the average values for the Florida lakes (25μg/l and 1.7m). Thus, despite the nutrient enrichment in Lake Ossa, the trophic state of the lake remains at the lower level compared to the average of Florida lakes.

The decreasing nutrient gradient (TP and TN) moving from the outlet of the lake

(in the south) to the north indicates that the flow of the Sanaga River provides most of the nutrients for the lake. The Sanaga River is the major river of Cameroon, measuring

918 km long with a drainage basin close to 140,000km2, representing about 30% of the national territory (Van der Waarde, 2007). Thus, the Lake Ossa drainage basin, which is only 245km2 wide (Wirrmann & Elouga, 1998) may only contribute marginally to the nutrient load when compared to the large Sanaga catchment area; also the agro- industry activities around the lake were already there several years ago before the lake was classified mesotrophic based on Wirrmann (1992) measurements, and there have not been major increases in the Lake Ossa watershed use that would explain the doubling of nutrient concentration. Therefore, to have a better understanding of the factors that may have contributed to nutrient increases in Lake Ossa in the past three decades, it is necessary to investigate the major anthropogenic and biological transformations that have occurred in the Sanaga catchment area during that period.

Three main anthropogenic factors could be responsible for the nutrient load increase in the Sanaga River. These include forest clearings, farming, and urbanization.

Cameroon loses approximately 220Kha of forest every year (FAO, 2010), which is about 0.5% of the territory. This loss is mostly driven by clearing for agriculture and urbanization (Epule, 2014). Agricultural land in Cameroon has increased from 16% to

21% from 1965 to 2015 (World Bank 2018), and the urbanization rate is 3.6%/year (The

World Bank, 2019). The forest clearing in Cameroon exposes soils, and accelerates erosion, sediment loading, and organic matter deposit in the rivers, leading to increased nutrient loading. The increased urbanization coupled with poor domestic and industrial

waste management in most cities in Cameroon may further contribute to the nutrient enrichment of its rivers. Epule (2014) indicated that the increase in agricultural land in

Cameroon is due to poor soil fertility causing people to either seek additional land or use chemical fertilizers. Fertilizers used in domestic and industrial agriculture are rich in phosphorous and nitrogen. Those fertilizers would likely drain into rivers during the rainy season and contribute to the nutrient load. It is also important to note that the number of hydroelectrical dams on the Sanaga River has increased to three including the Song

Loulou, Edea, and more recently, Lom-Pangar dams. Hydroelectric dams can cause accumulation of sediments and nutrients over long periods of time; however, once flushed, large amounts of nutrients can be released downstream. Therefore, the change in Lake Ossa water chemistry might be a reflection of the dynamic of the anthropogenic activities occurring in the Sanaga catchment area.

Chlorophyll concentrations and Secchi depth values followed an opposite gradient in Lake Ossa. As chlorophyll decreased from the south (near the outlet) to the north, Secchi depth increased (Table 2-3). This observation supports two hypothetical scenarios: (1) the increased concentration of chlorophyll generated in Lake Ossa is responsible for the decrease in water clarity; (2) the bulk of Lake Ossa chlorophyll is brought in by the Sanaga flow, and therefore, water clarity is influenced mainly by suspended sediment loading from the Sanaga River rather than the increased chlorophyll concentration. The prevailing scenario might also depend on the season of the year as discussed below.

Lake Ossa has two seasonal flows (Takoukam Kamla, 2012). During the high- water season, the Sanaga River effluent drains its suspended sediments and

chlorophyll into the lake, which in turn settles down as the water flows away from the lake inlet. As the suspended sediments settle down, water clarity increases. This scenario matches the scenario encountered in the current study. Water sampling was done during the high-water season. The distance of the sampling at the point accounted for 76% of the variation in Secchi depth, while chlorophyll concentrations accounted for only 24% (Figure 2-3) when assessed independently. The multivariate analysis further demonstrated that chlorophyll concentration did not contribute significantly to the water clarity and that the distance to the main lake inlet is the major control of water clarity. In a similar study conducted in Choctawhatchee Bay in Florida, USA showed a different result indicating a strong relationship between water clarity and chlorophyll concentration with the later accounting for almost half (47%) and color for 39% of the variation in Secchi depth (Hoyer et al., 2015). This difference might be due to the very low water clarity in Lake Ossa compared to Choctawhatchee Bay. Therefore, during the high-water season, scenario 1 is likely dominant.

During the low water season, the lake water flows back into the Sanaga River.

The clarity during that season is therefore expected to increase as there are less suspended sediments flowing from the river into the lake. However, our results clearly show that clarity instead decreased during the low water season (Figure 2-4B). The main explanation for this decrease lies in the fact that during the dry season, the low water level (less than 2.5m) exposes the water bottom to the strong forces of the dominant Harmattan winds that may displace sediment from the bottom. Unfortunately, we did not measure the chlorophyll concentrations in the lake during the low water season. However, given the lower water clarity at that season, we suspect that the

chlorophyll concentration should be lower compared to the rainy season and that there would be no north-south gradient as observed during the rainy season. Nonetheless, the high-water clarity will lead to increased chlorophyll concentration in areas of the lake basin that are less exposed to the winds in Lake Mevia (Figure 2-1), which is surrounded by hills and high trees. Thus, scenario 1 would likely occur in those areas sheltered from the direct wind and neither scenario 1 nor 2 would occur in areas readily exposed to the wind.

This study established the first high-resolution bathymetric map of Lake Ossa.

Local fishers believed that there are many manatee holes (deep pockets) in the lake in which they take shelter to avoid detection (Takoukam Kamla 2012). Our bathymetric map does not seem to agree with the fisher’s beliefs. Lake Ossa’s main basin bottom relief seems very regular, flat, and shallow. The only relatively deep part of the lake is located along the creek connecting the lake to the Sanaga River (Figure 2-5). The creek is almost two times deeper than the main lake. Simulation on our bathymetric map using a 2.5m water depth decrease from the high to the low water season indicates that during the low water period, 38% and 94% of the lake area has a depth less than 1m and 2m respectively (Figure 2-10). Generally, manatee suitable habitat includes areas of at least 2m deep (Reep & Bonde, 2006; USFWS, 1999) and only 6% of the lake bottom meets that criteria; thus based only on water depth, Lake Ossa may not be an ideal refuge for the African manatees during the low water season as the lower reaches of the Sanaga River provides greater depths at that time of the year.

The surface light at the bottom model (assuming a light attenuation coefficients k=1.7/SD) predicted that only 4% of Lake Ossa received sufficient light at the bottom

(>15% of surface irradiance) through at least four months corresponding to the data collection period of this study. Because the data collection was conducted only during the rainy season, which is the time of year with the lowest direct normal irradiance and lowest water depth, we might expect a slightly different result during the dry/low-water seasons despite the offset of the decreased Secchi depth during that season (Figure 2-

4). Our model prediction did not change significantly when using different K values

(K=1.6/SD, K=1.8/SD, K=2.0/SD), probably because of the low values obtained from the

Secchi depth.

Our model prediction of the distribution of SAV in Lake Ossa was in strong agreement with the observed SAV distribution (Figure 2-8); both indicating that SAV is absent across almost the entire lake bottom (96% and 98% respectively). The confusion matrix analysis (Figure 2-9, Table 2-1) further confirmed the high performance of our predictive model as indicated by the low error rate (4%) and high accuracy (96%). Our predictive model relied on only two main variables including Secchi depth spatial distribution and bathymetry. There, we demonstrated that the spatial distribution of

Secchi depth in the lake could be extrapolated from a single point at the outlet using equation 5. The bathymetry map can be updated by applying the change of water level at one point to the entire water surface depth data. Therefore, we demonstrated that we could monitor the distribution of SAV in Lake Ossa by monitoring the Secchi depth value and the water depth from a single permanent sampling point, placed ideally near the outlet of the lake. However, our model needs to be improved by including data from the low water season and expanding data collection for several years to account for seasonal and annual variability.

The results of this study contribute to the knowledge of the ecology of the African manatee in Lake Ossa. The bathymetry data suggest that only very few areas of the lake remain suitable for the species during the low-water season. The SAV distribution in Lake Ossa shows that African manatees in Lake Ossa do not have SAV as a food source alternative to the emergent vegetation that, although still available during the low water season, is disconnected from the water column as the water levels recede. This observation further supports the hypothesis that manatees migrate out of the lake during the low-water season and return during the rising-water season when the emergent vegetation becomes flooded. Such a migratory pattern in relation to high and low water has been documented in other African manatee populations in many countries throughout Africa such as Sierra Leone, Gambia, and Ivory Coast (Akoi,

2004; Powell, 1996; Reeves et al., 1988) whereby the species moves upstream during the rainy season (as water level increases and fresh vegetation become inundated) and would migrate downstream during the dry season (as water level recedes). Migration in response to water level variations has also been observed in manatee species living in tropical habitats such as the Amazonian manatee (Arraut et al., 2010; Deutsch et al.,

2003; Marmontel et al., 2002);and the Antillean manatees in many countries of its distribution including Mexico, Costa Rica, Honduras, and Trinidad (Colmenero-Rolón,

1986; Reynolds III & Odell, 1991; Reynolds et al., 1995). The Florida manatee, on the other hand, makes seasonal migratory movements but rather in relation to change in water temperature (Deutsch et al., 2003).

The low quality of manatee habitat in Lake Ossa during the low water season demonstrated by the current study contrasts with the higher frequency of manatee

sightings during the same season as documented by the MMP. With the decrease in water depth and food access during the low water season, it is expected that manatee abundance would be lower compared to the abundance during the high-water season.

The higher manatee sighting frequency during the low water season does not necessarily imply an increase in the abundance of the species during that season. The more plausible explanation is that manatees in Lake Ossa during the low water season are less cryptic, therefore easily detectable as their habitat is restricted to only deep areas off-shore in the lake; and during the higher water season sightings are rare as access by biologists is limited when the manatees migrate into the flooded forest and swamp. The majority (78%) of the manatee sightings reported by the MMP during the low water season are from only two specific areas of the lake. Thus, using survey approaches that will estimate abundance rather than just sighting frequencies will be crucial to determine the seasonal dynamic of the manatee population in Lake Ossa and corroborate it to the seasonal change of the habitat.

The current study revealed major environmental and conservation concerns to be addressed to sustain a healthy habitat for the aquatic wildlife in Lake Ossa and especially for the African manatee, a keystone species occupying the Sanaga River watershed. The eutrophication of Lake Ossa is most likely common to the other aquatic habitats in Cameroon. Like many developed countries, Cameroon has strong environmental laws. However, the difference lies in the level of enforcement and implementation of the law which is apparently very weak in Cameroon and therefore is in need of more stringent enforcement. The enforcement should go along with the implementation of a water quality monitoring program that can provide indicators to

assess the impact of the regulations and therefore improve development of future conservation initiatives.

It is important to reinforce manatee patrols and fishery management in Lake

Ossa, especially during the dry season as the individual manatees that did not migrate out of the lake system may be more vulnerable to hunting and bycatch. As mentioned above, access to food and deep-water areas are limited during the low water season.

Therefore, resident manatees may tend to be confined to small geographic areas of the lake where they could be easily be spotted. In fact, previous research shows that during the low water season, manatee sightings tend to be concentrated in a particular location called OHE, in Lake Mevia (Figure 2-1), but when the water level increases, they become rare at that location (Takoukam Kamla 2012). Because food accessibility is limited, manatees would likely take the risk of grazing in very shallow areas where they can easily be detected and trapped by potential hunters. The author once surprised a manatee during the dry season feeding in an area less than 1m deep. As soon as the manatee noticed our presence, it suddenly started “walking” on its flippers and dragging its body through the shallow area until it reached a deeper area where it submerged and was no longer detectable.

Fishing activities in Lake Ossa are seasonal and peak during the lowest water season, increasing the risk of encounters of manatees with gillnets, which often result in manatee-fishermen conflicts in Lake Ossa. Fishers complain that manatees destroy their nets and are more likely to hunt or not release any bycaught manatee as a form of compensation. Therefore, it is very important to design and implement fishing rules that will include establishing no-fishing zones in vulnerable manatee areas and promote a

more sustainable fishing practice. A study should be conducted to determine the spatial distribution of vulnerable manatee areas. The various gillnet fishing techniques employed in Lake Ossa should be assessed, and those that maximize fish catch while minimizing manatee bycatch should be promoted.

Very little has been done regarding the limnological study of Lake Ossa and other lakes, rivers, and coastal waters of Cameroon. Limnological knowledge in

Cameroon, as in many developing countries, is still very basic (Wetzel & Gopal, 1999).

There is a need to encourage local academic institutions, multi-disciplinary, integrative, and extensive basic limnological research and promote training programs to help the conservation and education to ensure sustainable habitat development. The Florida

LAKEWATCH program (Canfield et al., 2002) is a good example of how the benefit of such extensive basic and collaborative limnological research and training programs support conservation management. Establishing a collaborative and participative limnological research program in Cameroon similar to that of LAKEWATCH might be an important step to address data gap on the water systems of Cameroon.

Table 2-1. Confusion matrix performance parameters values, formula, and abbreviations used for the SAV predictive model in Lake Ossa. Tp is true positive, Fp is false positive, Tn is true negative, Fn is false negative, Pp is total positive predicted, Pn is total negative predicted, Op is total positive observed, On is total negative observed, and T is the total number of cells Performance metric Abbreviation Formula Value Tp + Fn

Error rate ERR T 4% Tp + Tn

Accuracy ACC T 96% Tp Sensitivity (True positive rate) SN Op 29%

Specificity (True Tn

negative rate SP On 97% Tp Precision (positive predictive value PREC Pp 14% Fp

False-positive rate FPR On 3% Tp ∗ Tn − Fp ∗ Fn Matthew correlation coefficient MCC √(Pp ∗ Pn ∗ Op ∗ On) 0.18

Table 2-2. Historical (by Wirrmann, 1992) and recent (from this study) trophic state parameters of Lake Ossa indicating rapid eutrophication in three decades. TP is total phosphorus, TN is total nitrogen, and TChl is total chlorophyll concentrations. SD is Secchi depth. Trophic state was based on Carlson (1977) TN TChl SD Trophic Month/Year Sources TP(μg/l) (μg/l) (μg/l) (m) state May 1985 Wirrmann,1992 157 12.9 8.4 NA Mesotrophic July-Sept 2016 Current study 348.9 28.15 19.8 1.17 Eutrophic

Table 2-3. Surface water chemistry; TP is total phosphorus, TN is total nitrogen, TChl is total chlorophyll concentrations, and SD is Secchi depth. Monthly values for all 18 stations sampled from May through August 2016 were averaged by zone (South, Middle, and North) of the Lake Ossa and combined for the entire lake. Values in the parentheses represent monthly minimum and maximum. Parameter South (outlet) Middle North Combined TN (μg/l) 398.9 (110-680) 329.4 (70-930) 327.5 (210-550) 348.9 (70-930) TP (μg/l) 40.6 (8-209) 21.2 (7-46) 25.1 (7-150) 28.2 (7-209) TN/TP 15.9 (2.4-42) 15.9 (7.8-34.3) 23.4 (3-50) 19.3 (2.4-50) TChl (μg/l) 29.9 (10-53) 17.4 (9-28) 15.2 (9-27) 19.8 (8-53) SD (m) 0.8 (0.4-1.2) 1.1 (0.8-1.8) 1.5 (0.8-2.8) 1.2 (0.4-2.8)

Figure 2-1. Map of Lake Ossa showing the 18 sampling stations, the weather station location and the arbitrary zones of the lake. The red circle indicates the outlet.

Figure 2-2. Effects of phosphorous and nitrogen on chlorophyll concentrations. A) Relationships between monthly TP (µg/l) and monthly Tchl (µg/l) concentrations; B), and between monthly TN (µg/l) and total chlorophyll concentrations for all 18 stations sampled in Lake Ossa between May and August 2016.

Figure 2-3. Effects of chlorophyll and distance to the outlet on Secchi depth. A) Relationships between monthly Secchi depth (m) and monthly Tchl (µg/l) concentrations, and B) between station average Secchi depth (m) and distance to the outlet of each station (km) for all 18 stations sampled in Lake Ossa between May and August 2016

Figure 2-4. Monthly variations of rainfall, depth, Secchi depth and MDC. Comparison of monthly rainfall: A) with monthly averages of water depth, Secchi depth and a maximum depth of plant colonization (MDC), and B) measured in the five permanent sampling stations in Lake Ossa in 2016 and 2017.

Figure 2-5. Bathymetry map of Lake Ossa showing the different depth gradient. The bathymetric points were recorded in September 2016 during the wet season.

Figure 2-6. Prediction of Secchi depth throughout Lake Ossa, based on the established linear relationship between the Euclidean distance of a sampling point for the outlet of the lake and Secchi depth at that point.

Figure 2-7. Prediction of percent of surface light available at the bottom throughout the Lake Ossa. The areas in yellow are those where no SAV is expected while the areas in green are those where SAV presence is expected.

100

Figure 2-8. Map of the distribution of the SAV in Lake Ossa measured using the Lowrance HDS 9 Gen 3. Values were measured on September 2016. The areas in yellow are those where no SAV is expected while the areas in green are those where SAV presence is expected.

101

Predicted

Positive Negative Positive Tp=184 Fn=450 Op=634 Observed Negative Fp=1133 Tn=33941 On=35074 Pp=1317 Pn=34391 T=35708

Figure 2-9. Confusion matrix parameter values. Tp is true positive, Fp is false positive, Tn is true negative, Fn is false negative, Pp is total positive predicted, Pn is total negative predicted, Op is total positive observed, On is total negative observed, and T is the total number of cells assessed.

102

Figure 2-10 Prediction of manatee habitat suitability base on the seasonal water variation in Lake Ossa. The spatio-temporal modelling of the bathymetry was based on the average monthly water depth data recorded at one point of Lake Ossa between December 2015 and December 2017. The initial bathymetry map was conducted when water level was at highest (Lmax, top second frame) during the raining season, then simulations were performed for series of decrease in water level by 1, 2, 2.5, and 3m consecutively. Areas in blue, are those with a depth of at least 2m and assumed to be suitable for manatees; areas in green are those less than 2m and less suitable for the manatees; areas in red are those that become dried as water level recedes.

103

CHAPTER 3 DIET COMPOSITION OF THE AFRICAN MANATEE: SPATIAL AND TEMPORAL VARIATION WITHIN THE DOWNSTREAM OF SANAGA RIVER WATERSHED, CAMEROON

Background

The availability of food resources is an essential component of the ecological niche of species and contributes to the species’ distribution patterns (Cox & Moore,

2000). The more available and diverse food resources are, the greater the survivorship of species due to reduced competition within an overlapping ecological niche. On the other hand, a shift in food availability may cause a species to locally decline or to migrate temporally or permanently. The optimal foraging theory by MacArthur & Pianka

(1966) and Emlen (1966) states that natural selection advantages individuals that harvest the maximum amount of energy per feeding effort. The theory provides a logical framework and can explain and predict choices made by wildlife species on certain foraging strategies and within different habitats (Marsh et al., 2011). Understanding the influences on a species’ distribution and its seasonal diet can be crucial for effectively managing wildlife populations.

The Order Sirenia, also known as sea cows, consists of four extant fully aquatic and hind-gut fermenter herbivorous species. The West Indian manatee resides in the

Caribbean, the Gulf of Mexico, and the US Atlantic coast; the Amazonian manatee is restricted to the Amazon River; the African manatee is distributed across rivers, lakes, lagoons, estuaries, and continental shelves along the west and central Atlantic coast of

Africa (Husar, 1978; Keith-Diagne, 2015; Powell, 1996); and the dugong in the western

Pacific Ocean and along the eastern coast of Africa (Marsh et al., 2011).

104

The African manatee is the least studied of all sirenians (Marsh et al., 2011).

They are threatened mainly by poaching, accidental catches in fishing nets, and habitat degradation, despite legal protection in all their 21 range countries. African manatees are also under the protection of international laws (Powell, 1996). They are Red listed by IUCN (The International Union for Conservation of Nature) as ‘’Vulnerable’’ (IUCN

2019) and belong to Appendix I of the Convention on International Trade in Endangered

Species (CITES) and of the Convention on Migratory Species (CMS).

In Cameroon, manatees are present in the lower reaches of most rivers and lakes with a direct connection to the sea. Some of these rivers include the Sanaga,

Akwayafe, Rio del Rey, Ngosso, Andokat, Meme, Munaya, Cross, Wouri, Nyong,

Dihende, Dipomba, and Ntem Rivers (Nishiwaki et al., 1982, Powell, 1996; Grigione

1996). The African manatee is also present in Lake Tissongo and Lake Ossa. In northern Cameroon, manatees are present in the Benue River, from the mouth of the

Faro to Lake Léré.

Sirenians are the only herbivorous marine mammals, feeding on a variety of plant species (Bengtson, 1981). The generalist diet of sirenians has been well documented overall (Allen et al., 2018; Best, 1981; Castelblanco-Martínez et al., 2011;

Keith-Diagne, 2014; Marsh et al., 1982; Mignucci-Giannoni & Beck, 1998), however very little is known about the diet of the African manatee. Best (1981) conducted a comprehensive review of the diet of sirenians, both quantitatively and qualitatively. The review did not include information on the diet of the African manatee. The author reported that manatees are exclusively herbivorous, consume about 9% of their body mass in plants daily, and only 45 to 70% of the consumed aquatic plants are digestible.

105

Florida manatees are known to consume over 60 food plant species (Ames, et al. 1996; Bengtson, 1981; Hartman, 1979; Hurst & Beck, 1988). Mignucci-Giannoni &

Beck (1998) reported ten aquatic plant species consumed by the Antillean manatee in

Puerto-Rico, while in Belize, for the same species, 30 food items were reported by Allen et al. (2018). Colares & Colares (2002) documented 24 aquatic macrophytes species consumed by the Amazonian manatee and Guterres-Pazin et al. (2014), added 30 new aquatic plant species to the food plant list for the same species.

Plant species composition and feeding strategies of the African manatee differs widely across habitats (Keith-Diagne, 2014; Powell, 1996). Keith-Diagne (2014) conducted a comprehensive review of reported African manatee food plants across their distribution range. Keith-Diagne documented 70 suspected and confirmed African manatee food plant species reported through self-observations and various available scientific papers (Akoi, 2004; Bessac & Villiers, 1948; Husar, 1978; Ogogo et al., 2013;

Powell, 1996; Reeves et al.1988).

Akoi (2004) conducted gross inspection of 35 African manatee fecal samples from the coast of Ivory Coast and corroborated the findings of Powell (1978) that the

African manatee diet is composed predominantly of grasses but that they also consume fruits, mud, and deposited organic material particularly during the dry season, when the decreased water level limits access to emergent vegetation. Best (1983) also reported

Amazonian manatees feeding on mud and organic material as a result of low water levels and limited food access.

Keith-Diagne (2014) conducted the first carbon (δ13C) and nitrogen (δ15N) isotope analysis on ear bones from 24 African manatee carcasses recovered in Senegal

106

and Gabon to determine the lifetime diet composition of the individuals. The stable isotope signatures recorded from samples collected in Gabon indicated that their diet was made up of 90 to 94% of plants and 6 to 7% of hermit crabs. The sampling of the ear bones of manatees from Senegal showed that their diet was composed of 46 to

57% of plants, 24 to 27% of fish, and 19 to 24% of mollusks. The author suggested that the difference in diet composition between manatees from Senegal and Gabon was mostly due to habitat differences. Carnivorous behavior has also been documented in the West Indian manatee (Courbis & Worthy, 2003; Powell, 1978) and has been anecdotally reported in the African manatee by local fishermen (Dodman et al.,2008;

Powell, 1996; Takoukam Kamla, 2012). Carnivorous behavior has been considered opportunistic, suggesting that manatees feed on other food resources only when plants are not available. The isotope analysis by Keith-Diagne showed that manatees from

Senegal have a high proportion of fish and mollusks in their average lifetime diets, indicating that the manatees regularly and abundantly have a diet of animal origin.

Lake Ossa and the downstream reaches of the River Sanaga provide a dry season sanctuary for manatees within the Sanaga River Basin (Powell,1996), and they are present year-round. These two areas encompass two adjacent protected areas including Lake Ossa Wildlife Reserve and the National Park of Douala-Edea, respectively. Even within these protected areas, manatees are hunted and accidentally caught in fishing nets because of weak law enforcement and the lack of an adapted management strategy. Thus, to establish an efficient management plan, key information on the use of habitat by the species is needed for the two protected areas. Therefore, identifying, documenting, and explaining the geographical and temporal variation of the

107

available food plant species will help identify important manatee ecological niches. The latter will help prioritize protection and conservation efforts to ensure the long-term survival of manatees in this region. Furthermore, the discrimination between manatee diets across habitats may allow for a better understanding of manatee movement patterns within the DRSW.

The African manatee is an elusive and cryptic species, living in the turbid waters of the riverine, lacustrine, and estuarine habitats of the DSWR were water transparency in most places is less than a meter (Chapter 2). These conditions result in poor visibility, and consequently, it is challenging to conduct direct observational studies on the feeding behavior and diet of the African manatee. The collection of free-floating fecal material is a non-invasive and informative means of developing data on the diet of local animals. Micro-analyzing, the histological structure of the plant fragments found within fecal material, is an ideal method for studying the diet of the species. Microhistological analysis has been widely and successfully used to study the diets of the West Indian manatee and the Amazonian manatee (Allen et al., 2018a; Beck & Clementz, 2012;

Colares & Colares, 2002; Guterres-Pazin et al., 2014; Hurst & Beck, 1988; Mignucci-

Giannoni & Beck, 1998).

The purpose of this study is to contribute to the conservation of the African manatee in the DSRW by determining manatee diet composition and characterizing the aquatic vegetation present in the area. More specifically, the goals of this study were to

(1) document aquatic plants present in the DSRW; (2) document the diet composition of

African manatees in the DSRW; and (3) assess differences and similarities in diet

108

composition by locations or habitats, water levels (seasonal variation), and feces bolus size.

Methods

Study Area

The DSRW encompasses two protected areas in the Littoral Region of

Cameroon (Figure 3-1). The Douala-Edea National Park, formerly known as the Douala-

Edea Wildlife Reserve, was created in 1932 and covered about 160,000 hectares of land and water (Blaikie & Simo, 2000). The park stretches along both sides of the lower reaches of the Sanaga River and 100km along the Atlantic coastline of Cameroon

(Latitude 3°14’ 3°50′N and longitude 9°34′-10°03’E) (Feka et al., 2009). The park surface is covered by about 6.4% of mangrove dominated by Rhizophora racemose.

The rest of vegetation (80%) is predominantly a tropical lowland equatorial forest. The mangrove is seriously threatened by deforestation. The local community utilizes the mangrove woods to smoke fish. Fishing is the major economic activity in the area.

The Lake Ossa Wildlife Reserve is a complex of lakes located at 13km from

Edea, Cameroon, between the 3°45’ and 3°52’N latitude, and 9°45’ and 10°4‘E longitude with approximately a 300m elevation (Wirrmann & Elouga, 1998). The water surface is estimated to be 4000ha. The lake represents about 90% of the Lake Ossa

Wildlife Reserve, created in 1968 and falls within the 3rd category of protected areas, according to classification criteria for Cameroon. The reserve was established to provide a refuge for the protection of the African manatee.

Sampling Design

The shorelines of four areas within the DSRW (Figure 3-1) were surveyed twice between June and September 2016 to characterize the vegetation. They were also

109

frequently visited between June 2015 and November 2017 in search of free-floating fecal samples. These areas represented three habitat types: lake (Lakes Tissongo and

Ossa), river (Sanaga River), and estuary (Sanaga Estuary) as presented in Figure 3-1.

Plant surveys and fecal collection were only conducted during the high-water season

(average depth >2m) in Lake Tissongo, the Sanaga River and the Estuary, but were collected during both the high and the low-water seasons (average depth <2m) in Lake

Ossa. The low-water season extends from November to May or June, and the high- water season extends from July to October or November (see Figure 2-4).

Fecal Sample Collection

A total of 112 free-floating manatee boluses and one fecal sample taken from the lower intestine of a stranded calf carcass were examined using microhistological analyses techniques, described below. The distribution of fecal collection sites is presented in Table 3-1 and Appendix A, Figure A-1. In Lake Ossa, 29 fecal samples were collected during the high-water season and 31 during the low-water season. The

GPS location of each fecal sample was recorded, and when not disintegrated, the diameter of the feces was measured using a caliper (Appendix A, Figure A-2). Fecal and gut samples were preserved in 70% ethanol until ready for examination (Allen et al.,

2018; Hurst & Beck, 1988).

Habitat Characterization and Plant Library Collection

In order to understand the variety of plant species accessible to the African manatee in each location, the shoreline vegetation was surveyed. A total of 238 100m- transects were evenly spaced along the shoreline of the study area. Each transect consisted of two to six 1 x 1m-plots for a total of 958 plots distributed across the four study locations (Table 3-2); the GPS coordinates of each plot was recorded (Figure 3-1)

110

and the plots were placed within two meters of the shoreline in areas that were considered accessible by manatees. Plant samples were collected from each plot and identified by an expert plant taxonomist. The dried plants were preserved in newspapers. Finally, the percent coverage by plants in the plot was estimated.

Reference slide preparation

The reference slide preparation was done following the protocol described in

Hurst and Beck (1988) with some modifications. The epidermis of the leaf and sometimes stems of frequently encountered plant species within the survey areas were mounted on a microscope slide for identification under 100X and 400X magnification.

The leaf or the stem was scraped carefully with a scalpel, and Visikol® for Plant

Biology™ solution was used to clear out the chlorophyll to improve the optical quality of the specimen. No staining was necessary as most of the microhistological characters were visible. Specimen material was permanently mounted on a Visikol® MOUNT™ solution. The slides were dried for two to four days before the evaluation with a microscope. The Visikol MOUNT was added between the slide and the coverslip as necessary.

Photography

A 14 Megapixel OMAX (Model: A35140U) microscope USB digital camera was used to take photomicrographs of plant material. The digital camera was mounted on the microscope through an OMAX fixed microscope adapter (Model: A3RDF50) with a

0.5X reduction lens that increased the field of view. Photomicrographs were captured using both 10X and 40X objective magnifications and under a standard bright field illumination and polarized light that revealed glowing microhistological characters. A

111

digital catalog of the photomicrographs was organized by plant family, and priority was given to the most abundant plant species and plants suspected of being manatee food.

Microhistological Analysis

Fecal samples were examined using microhistotological features of undigested plant fragments and were analyzed using the techniques described by Hurst and Beck

(1988). The digesta samples were washed free of dirt and fine particles with tap water over a 30-mesh (0.52mm) screen. A small quantity (about 100mg) of washed sample was spread uniformly over a 2x3inch slide to which drops of Hertwig’s solution or

Visikol® for Plant Biology™ optical clearing solution was added and heated over an alcohol flame when necessary to clear pigments from the plant cells, providing a better view of cellular structures. Five slide replicates were performed for each fecal sample.

The slides were observed under a microscope at 100X magnification. Plant fragments were identified by comparing their histological structures with voucher microscope slides, photomicrographs, or illustrations from references. For fragments that could not be identified but with distinctive microscopic features, a generic name was assigned, e.g., “Unidentified 1”. The same generic name was given to fragments with the same features.

Quantification of Percent Food Plant Species Occurrence

Each of the five slides for each fecal sample was examined using 20 different fields of view indicated by the vertical and horizontal graduation of the mechanical stage as described by Hurst & Beck (1988). Each of the fields of view was observed within a gridded frame of the microscope eyepiece whereby vertical lines represent randomly numbered (1 to 11) transect lines. Transect line 1 is at the center of the grid; even transect lines on the left and odd lines on the right. Plant fragments at the interception of

112

a transect line with the horizontal lines were recorded starting from transect line 1.

Intercepts on a transect line were examined subsequently until five plant fragment records were achieved. When the latter was not achieved on transect line 1, the observer continued to the next transect line number. Plant species were recorded on a data collection sheet and later entered into an Excel spreadsheet.

The occurrence of each plant species on a slide was counted; then, the percent occurrence in a fecal sample was obtained by tallying occurrence for each species from each of the five slides. Thus, each fecal sample was examined at 500 intercept points (5 slides x 20 fields of view x 5 intercepts). The percent occurrence of each food plant species was averaged by location, season, and feces size.

Data Analysis

Habitat shoreline species characterization

All the collected data were recorded in an Excel spreadsheet and analyzed using both EXCEL Data Analysis tools or software XLSTAT-Ecology and RStudio 1.2. Plot locations were mapped using ArcGIS 10.6. For each location, the number of unique plant species encountered, and the relative abundance of each of them, was computed by averaging the percent occurrence per plot by the total number of plots for that location. The relative abundance of the plant families and types was also computed.

The relative abundance of dominant plant species was plotted on a chart bar for a visual representation of the species composition by location. The species diversity of the shoreline vegetation was estimated using the Shannon diversity index H (Shannon,

1948).

113

푛 3.1 퐻 = − ∑ 푃푖⁡ln푃푖 푖=1

Where, n is the total number of species, Pi the proportion, or relative abundance of the species i.

The difference in species composition between locations was measured using the Bray-Curtis dissimilarity matrix (Bray & Curtis, 1957). In order to buffer the influence of strongly dominate species, the relative abundance of each species was standardized by first taking the logarithm of the relative abundance before calculating the distance matrix (Kindt & Coe, 2005).

The analysis of similarities (ANOSIM) was performed on the PAST 3.24 software

(Hammer et al., 2001) to test for the significance of the difference in species composition between location and sites. ANOSIM is a non-parametric statistical test similar to an ANOVA test which uses a permutation and randomization methods from a ranked similarity matrix to generate the R-statistics that determines whether the similarity between groups is greater than or equal to the similarity within groups (Clarke,

1993). All ANOSIM tests in this study were performed with 10,000 permutations.

The similarity of percentages (SIMPER) computed the contribution of each species in the dissimilarities between locations and seasons using the Bray-Curtis similarity index of the most frequent species. The Bonferroni correction was applied to compensate for errors due to the multiple comparisons test (Clarke, 1993).

Shoreline species composition and diversity were also compared between seasons for Lake Ossa as it was the only location surveyed during both the low and high-water season.

114

Diet composition and diversity analysis

The same metrics (relative abundance, number of species, Shannon diversity index) used to describe the shoreline vegetation were also used to characterize and compare diet species composition between locations, seasons (Lake Ossa only), and feces bolus size. Feces were categorized as small or large depending on their diameter

(≤4cm, and >4cm respectively). The cutoff value was determined by building a frequency distribution histogram of the diameter of 377 fecal samples collected during this study. Each of the top five dominant identified plant food species was compared between locations, seasons, and fecal sizes using the Kruskal-Wallis test (a non- parametric equivalence of ANOVA) since the distribution of relative abundance failed the test for normality (Kruskal & Wallis, 1952).

Results

Shoreline Vegetation Characterization

Among the 958 plots surveyed during this study, 160 plant species, 122 genera, and 43 families were recorded along shorelines of the four locations (Lake Ossa, Lake

Tissongo, Sanaga Estuary, and Upper Sanaga). The highest number of species was recorded in Lake Ossa (111), followed by Upper Sanaga (96), Sanaga Estuary (51), and

Lake Tissongo (16) (see Table. 3-2). The plateau on the species accumulation curves for each location (Figure A-3) indicates that our surveyed effort captured nearly the maximum number of species present in each location except in Lake Tissongo. The

Shannon diversity index (H) varied across locations with the highest values recorded in

Upper Sanaga (H=3.55), Sanaga Estuary (H=3.09), and the lowest value in Lake

Tissongo (H=1.86).

115

In the comparison of the surveyed shoreline plant species composition using the

ANOSIM, there was a significant difference between locations (R=0.35; P=0.00009); indicating a strong effect of the habitat type on the plant species composition. The pairwise comparison of species composition by location indicated a significant difference between each location pair (p=0.0006 for each pair). The Bray and Curtis pair-wise coefficients of dissimilarity also indicated a high difference in species composition and relative abundance between locations. The Bray and Curtis distance was greatest between Lake Ossa and Sanaga Estuary (87% of dissimilarity) and lowest between the Upper Sanaga and Sanaga Estuary (66% of dissimilarity) (see Table 3-3).

A posteriori comparison using the SIMPER method allowed estimation of the contribution of each plant species to the significant differences, as indicated in Table 3-

4. Echinochloa pyramidalis had the highest contribution (34.45%) in the difference of species composition between locations. The overall average dissimilarity computed by the SIMPER method between locations was 90.36%.

Overall, emergent macrophytes were the highly dominant plant type along shorelines (70.7%), trees represented 13.0%, shrubs (11.9%), and the less abundant plant types were the free-floating macrophytes, representing only 4.3% (Table A-1;

Figure 3-2). The most abundant plant species across locations were E. pyramidalis

(20.7%), Dissotis erecta (9.2%) and Eremospatha macrocarpa (8.8%) (see Figure 3-

3A). The relative abundance of emergent macrophytes was highest in Lake Tissongo

(90.6%) and Lake Ossa (85.5%) and was lowest in Sanaga Estuary (37.4%) and Upper

Sanaga (69.5%). The Sanaga Estuary shoreline has the highest abundance of shrubs

(22.6%), trees (31.7%) and free-floating macrophytes (8.3%). The most abundant plant

116

families along shorelines were Poaceae (27.5%), Arecaceae (12.0%), Euphorbiaceae

(10.6%) Melastomataceae (10.7%), and Rhizophoraceae (6.3%) (see Figure 3-3B). The abundance of Poaceae along shorelines varied by location and was highest in Lake

Ossa where it was50.5% of the total represented species and in Upper Sanaga

(30.2%); while it was lowest at the Sanaga Estuary (10.6%) and Lake Tissongo

(18.7%). The Lake Tissongo shoreline was highly dominated by Arecaceae (38.0%) and

Euphorbiaceae (27.1%); while the Sanaga Estuary shoreline was dominated by

Rhyzophoraceae (19.9%) and Euphorbiaceae (11.6%) (see Table A-1).

The dominant plant species in Lake Ossa and the Upper Sanaga was E. pyramidalis, while the dominant plant species in Lake Tissongo and the Sanaga Estuary were E. macrocarpa and Rhizophora racemosa respectively (Figure 3-4). The Venn diagram in Figure A-4 shows the number of shared and unique plant genera between locations. Five genera were found ubiquitous to all locations, including Echinochloa,

Dissotis, Commelina, Uapaca, and Commelina. Lake Ossa had the highest number of unique plant genera (26). Lake Ossa and Upper Sanaga had the highest number of shared genera (50).

Although there was no significant difference in shoreline plant species composition between seasons (based on the ANOSIM test) in Lake Ossa (R=-0.009; p=0.86), the number of shoreline plant species encountered during the high-water season (101) was 2.6 times higher than in the low-water season (39). E. pyramidalis, D. erecta, and D. falcipila were more abundant in the low water than in the high-water season (Figure 3-5). The Shannon diversity index was higher (H=2.74) during the high water than during the low water (H=2.01) season.

117

Diet Composition of the African Manatee in the DRSW

The number of unique plant species recorded in each fecal sample ranged from

1 to 9, with an average of 4. Most plant fragments (89.2%) were identified to at least the genus level. Among the 113 fecal samples analyzed, 31 plant species (from nine plant families) consumed by the African manatee were recorded (Table 3-1). Fruits from five tree species (from four plant families) were also recorded. The greatest number of food plant species were found in Lake Ossa (24) and the Sanaga Estuary (24) and the lowest in the Upper Sanaga (15) and Lake Tissongo (16).

The most represented plant type was grasses of emergent habit, which were present in all fecal samples and constituted 96.1% of the samples. Vascular plants constituted only 0.8% of samples. Three free-floating plant species (Nymphaea lotus,

Salvinia molesta, and Eichhornia crassipes) were found but were poorly represented

(0.23%, all three species combined) and were present in only one fecal sample (Table

3-5). No submerged aquatic plants were found.

The three most abundant families were Poaceae representing 71.1% of the overall diet, Cyperaceae (20.1%) and Arecaceae (4.8%). The five most abundant species were E. pyramidalis representing 53.5% of the fecal fragments, followed by

Cyperus sp. (12.9%), Rynchospora corymbosa (5.15%), E. macrocarpa (4.7%), and

Leersia hexandra (4.08%) (Figure 3-6,3-7). These species were also the most frequent and occurred in 98 (80.5%), 26 (23.0%), 38 (33.6%), 31 (27.4%), and 32 (28.3%) of the

113 fecal samples respectively (Table 3-4).

Manatee diet by location

The analysis of food plant species composition from manatee feces showed that there was a moderate but significant difference between location (R=0.18, p=0.0003).

118

Posteriori pairwise comparison showed that only the difference in species composition between Lake Ossa and Sanaga Estuary were significant (p=0.0006); with E. pyramidalis, Cyperus sp., and Unidentified13 showing the greatest contribution to the dissimilarity between the two locations (41.5%, 27.2%, and 10.8% respectively). The overall average dissimilarity between locations was 66.9%.

The Kruskal-Wallis analysis showed significant differences in the abundance of the five dominant plant species taken separately across locations (Table A-2).

Echinochloa pyramidalis was the dominant species in all locations but was more abundant in feces from Lake Ossa (mean=63.2%, SD=33.34) and Upper Sanaga

(mean=59.7%, SD=20.0); while E. macrocarpa was abundant only in Lake Tissongo

(mean=29.7%, SD=35.3); Cyperus sp. was very abundant in the Sanaga Estuary

(mean=29.9%, SD=44.12). Unidentified13 was only present and abundant in the

Sanaga Estuary (mean=12.1%, SD=27.9) and Rynchospora corymbosa was most abundant in Lake Ossa (mean=8.3%, SD=12.9).

Manatee diet by season in Lake Ossa

A total of 60 fecal samples collected in Lake Ossa were analyzed, including 29 collected during the high-water season and 31 during the low water season. The analysis of the food plant species composition in manatee feces showed that there was a moderate but significant difference between seasons in Lake Ossa (R=0.18, p=0.00001). A total of 22 and 18 plant species was recorded in the fecal samples collected in Lake Ossa during the high- and low-water seasons, respectively. The major contributors to the dissimilarity between seasons in Lake Ossa were E. pyramidalis

(47.9%), Cyperus sp. (15.8%), and R. corymbosa (12.6%). The average dissimilarity between seasons was 53.2%.

119

The comparison of the abundance of individual dominant species by season shows that only E. pyramidalis (p=0.0001) and Cyperus sp. (p=0.007) were significantly different across seasons (Table A-3). Echinochloa pyramidalis fragments were two tiimes more abundant in manatee feces during the low water season (mean=81.1%,

SD=21.6) than the high water season (mean = 44.2%, SD = 33.1). Cyperus sp. was 40 times more abundant in manatee feces during the high water season (mean=15.3%,

SD=24.8) than during the low water season (mean=0.4%, SD=1.48), see Figure 6.

Figure A-5 gives a clear visualization of the manatee diet composition profile between the low- and high-water season.

Manatee diet by feces size

For both small (diameter ≤4cm) and large (diameter >4cm) fecal bolus samples

(N=29) within Lake Ossa, the diet was similar to E. pyramidalisE. pyramidalis, E. macrocarpa, L. hexandra, R. corymbosa, and Cyperus sp. as the most dominant species. Although the ANOSIM comparison test showed no significant difference in diet composition between feces sizes (R=0.002, p=0.38), small size feces seemed to have less E. macrocarpa (mean=1.91%, SD=2.8) and Cyperus sp. (mean=5.7%, SD=12.37) than large size feces (mean=6.8% and 20% respectively), see Figure A-6.

Discussion

Shoreline Vegetation

The high plant diversity along the shorelines (H=3.52, 160 species) of the DSRW is very characteristic of tropical areas of southern Cameroon that are considered to have the richest flora in continental tropical Africa (Myers, 1988). This high diversity of aquatic plant species provides a broad array of food options for the African manatee.

Plant species composition was highly variable among the four study locations (R=0.35;

120

P=0.00009), reflecting the difference in habitat type and water quality among those locations.

Lake Ossa and the Sanaga Estuary were the locations with the highest Bray and

Curtis pair-wise coefficient of dissimilarity while Upper Sanaga and Sanaga Estuary had the lowest. This result is unsurprising because the spatial distance between Lake Ossa and the Sanaga Estuary is the longest (40km) pair-wise distance amongst the locations while the spatial distance between the Upper Sanaga and the Sanaga Estuary is the shortest. Also, the water quality in Lake Ossa and the Sanaga Estuary are at the two extremes. Lake Ossa is purely freshwater (with an average specific conductivity of

15µS/cm), but the Sanaga Estuary is brackish water. Finally, Sanaga Estuary is under the influence of tide while Lake Ossa is not. Thus, Lake Ossa, which is a fresh and stagnant water habitat was dominated by plants of the family Poaceae (50.5%), such as

E. pyramidalis, whereas the Sanaga Estuary was dominated by Rhyzophoraceae

(19.9%) and Euphorbiaceae (11.6%) that are characteristic of brackish water habitat.

Although Lakes Ossa and Tissongo are both lacustrine habitats, the Bray and Curtis distance between the two locations was high (69% of dissimilarity, Table 3-

3). This is not surprising because although the lakes are both connected to the Sanaga

River, they are in two different micro-climates. Lake Tissongo is closer to the sea

(17km), is at a lower elevation (6m) and may be under the influence of the tide, while

Lake Ossa is about 50km from the sea, is situated at a greater elevation (10m) and is not influenced by tides. Furthermore, Lake Tissongo is surrounded by a dense tropical lowland forest (Ngea, 2010) that supports the growth of E. macrocarpa as this plant favors forest margins (Ed, 2015), which could explain why the E. macrocarpa is the

121

dominant shoreline plant species of Lake Tissongo. In contrast, Lake Ossa is surrounded mostly by agro-industry and secondary forest with areas cleared by the local community for agriculture.

The absence of submerged aquatic plants in the entire study area could be associated with the high turbidity and low transparency of water with a Secchi depth of less than 2m (Takoukam Kamla, unpublished data), preventing light from reaching the bottom to sustain plant growth. Thus, emergent macrophytes were the highly dominant shoreline plant type in all locations (about 70.7%). However, recently in 2017 (after the field data collection for this study), Lake Ossa has experienced a massive proliferation of the free-floating aquatic fern Salvinia molesta (water moss), which has become the dominant macrophyte plant species of the lake (Takoukam Kamla, unpublished data). In

2016 as we were conducting this survey, S. molesta had a relative abundance of only about 0.5% (Table A-1). Also, this plant seems to compete with E. pyramidalis as we observed that the later was dead or dried in areas colonized by S. molesta. During the high-water season, as the water level rises, S. molesta carpets will float over E. pyramidalis (a semi-floating plant), preventing the latter from getting access to light, eventually resulting in death. The proliferation of S. molesta is partly associated with the recent increase in nutrient enrichment of the lake that might have been caused by the construction of the largest reservoir dam of the country (Lom-Pangar), upstream on the

Sanaga River, which feeds Lake Ossa during the high-water season.

A similar situation of nutrient enrichment has been documented in Lake

Okeechobee in southern Florida (Canfield & Hoyer, 1988b). The lake functions as a reservoir and its flow has been altered for flood control and irrigation. Thus, during the

122

high-rain season, the lake accumulates nutrients (nitrogen and phosphorus) from the discharges of water from its anthropized and fertilized vast watershed that favors the development of algal blooms. When the reservoir is opened or overflooded, its fresh and nutrient-rich water, as well as many cyanobacteria from the bloom, are discharged into the Caloosahatchee and St. Lucie Rivers, triggering intense algal blooms at the estuaries and nearby coasts (Krimsky et al.2018). Cyanobacteria and some dinoflagellates like Kerenia brevis produce toxins that are harmful to humans and animals. Brevetoxins produced in harmful algal bloom events have been implicated in the deaths of hundreds of manatees in Florida. (Broadwater et al., 2018).

Water-level seasonality appeared to have no significant influence on the shoreline plant species composition in Lake Ossa although the aquatic vegetation species composition in Lake Ossa during the high-water season was 2.6 times higher than during the low-water season (Figure 5). A significant difference probably would have been observed if a representative portion of the vegetation of the inundated forest was surveyed. The vegetation in the flooded forest are understory plants that can tolerate low light availability, are usually seasonal, and deposit seeds that germinate during the receding flow and release seeds during the high and peak flows (Reet et al.

2011).

Anatomy of the African Manatee Feces in the DRSW

Shape

The feces of the African manatee collected were free floating on the water surface. They were cylindrical in shape with flat extremities like a bar or rod (Figure A-

2). Chame (2003) made a review of the morphometric characteristics of terrestrial mammal feces based on data obtained from scientific journals and field guides. Based

123

on the morphometric characteristics of the feces, the author identified nine similar fecal groups and accorded with Seton (1925) that the shape and content of feces, which are also a reflection of the animal’s anatomy, are an excellent method of identifying mammals at the level of the order and sometimes at the species level. Chame did not include marine mammals in the grouping; however, when comparing the African manatee feces’ morphometric characteristics against the groups identified by the author, it appears they matched well with Group VIII, which characterizes large ungulates such as the elephants, hippopotamus, and rhinoceros. The closest terrestrial relative of manatees are elephants and hyrax (Kellogg et al., 2007; Murphy et al., 2001;

Pardini et al., 2007). Therefore, it is unsurprising that manatee and elephant feces have a similar shape, especially when both feed on a fibrous plant diet.

Size

African manatee feces have a diameter varying from 2 to 8.5cm and average between 4 and 6cm (Table A-4). The frequency distribution of the diameters of the 377 fecal samples measured followed a bell-shape curve of a normal distribution. The

Shapiro-Wilk test also supported the hypothesis that the diameter of the samples belonged to a normal distribution (w=0.99, P=0.08). Hema and collaborators (2017) conducted the first assessment of the age structure of the West African Savannah elephant by using fecal circumference analysis. The circumference of 624 elephant feces measured varied between 15 to 50cm (i.e., 4.7 to 16cm diameter), and the frequency distribution curve of the feces sizes was similar (bell-shape) to that of the

African manatee obtained in this study. A similar fecal size distribution was also reported in the Savannah elephants in Tanzania (Nowak et al., 2009) where the feces diameter varied between 5.45 to 17.2cm.

124

Hema and co-authors (2017) coupled the feces size of 51 elephants in the wild with their presumed age and found that there was a strong positive relationship between the presumed age of the elephants and the size of the feces (r2:0.925, F:605.27, n= 51, p<0.0001). A similar relationship between age class and feces size was reported in the

African elephant's population in Amboseli National Park in Kenya (Morrison et al., 2005) and those from the Kasungu National Park in Malawi (Jachmann & Bell, 1984). Hema and co-authors (2017), therefore, used the age class-feces size relationship to assign free ranging fecal samples of elephants to an age class. No study has investigated the relationship between feces size (diameter or circumference) in sirenians to date. In the current study, we could not associate the free-ranging fecal samples to an individual nor measure the size of the individual. However, given that elephants and sirenians are ancestrally related and the similarity in the shape of their feces, a strong relationship between feces size and animal size (age class) in manatees would be expected.

Defining this relationship would be useful for non-invasive studies involving parameters of population age structure in manatees, especially in Africa, where direct or photogrammetric measurements of the individuals are logistically difficult. When establishing such relationships with sex and diet, the type of individuals may account for specific variables that may affect the feces size (Jachmann & Bell, 1984; Morrison et al.,

2005).

Texture

African manatee feces obtained from the DSWR were highly fibrous when compared to Florida manatee feces from Crystal River and those of the African manatees sampled along the coast of Senegal. Fibrous plant fragments of up to 10cm long were observed. The fibrous nature of the African manatee diet in the DSWR

125

reflects the type of macrophyte plants present within their prospective habitats. Manatee in the DSWR can only feed on emergent and the few existing floating species of vegetation as there is no sea grass in the area because of the low water transparency and lack of salinity (Takoukam Kamla, Chapter 2). In Crystal River (FL) and on the

Senegal coast, manatees have access to sea grasses and other submerged vegetation.

Emergent macrophyte vegetation is known to be more fibrous than seagrasses and other submerged aquatic plants (Best, 1981; Marsh et al., 2011), which may explain why the manatee feces in DSRW are more fibrous.

The fecal sample taken from the intestine during the necropsy of a stranded manatee calf in Lake Ossa contained only mud and clay. Although manatees have been reported feeding on mud as a nutritive complement during periods of food scarcity

(Best, 1981; O’Shea, 1986; Colares & Colares, 2002), the mud in the calf feces was most likely an indication that since the animal was caught in a fishing net, it could not get access to vegetation or nursing and, therefore ingested mud in an effort to survive.

Some feces contained abundant pieces of fruits of about 0.5cm size with a texture as hard as that of dried corn. These were mostly ripe fruits of E. macrocarpa,

Calophyllum inophyllum, Canthium ciliatum, Macaranga sp., Ficus copreifolia, Ficus rubiginosa, and Paullinia pinnata. Akoi (2004) also documented manatees in the Ivory

Coast, eating fruits from Macaranga heudoletti, Dalbergia ecastaphyllum,

Chrysophyllum delevoyi, and Drepanocartus lunatus. A fruitivorous diet has been reported in the Amazonian manatee as well (Guterres-Pazin et al., 2014).

Color

The color of the feces collected in this study was mostly green but also varied from black, brown, or dark green depending on the type of plants the manatee had fed

126

on or the age of the feces. Very fresh fecal samples appeared to have the same green color on the surface of the feces as the interior; whereas older feces were greener inside than on the surface, while very old samples were dark green both at the surface and in the interior. Feces of manatees that fed more on rhizomes than leaves appeared to have a lighter green color than the feces samples that contained mostly leaves or stems. It is unclear how long manatee feces can float at the surface before being disintegrated. The change in the feces color over time may be a good proxy to estimate the age of floating feces. This is very important, as in some studies such as genetic and fecal hormone analysis, using fresh fecal samples is crucial to yield high-quality DNA and hormones. Therefore, an experiment designed to correlate the change in fecal color over time with the composition of the types of vegetation ingested by the manatee, and the degradation process of the feces themselves would be valuable.

Component of the African Manatee Diet in DSRW

The number of unique plant species identified in each sampled varied between one to nine, with the majority of feces (55%) having between three and five unique plant species. A similar range of plant species per sample was observed by Allen et al. (2018) in Antillean manatee from Belize (1 to 6), Colares & Colares (2002) in the Amazonian manatee in Central Amazon rivers and lakes (1 to 7), and Guterres-Pazin et al. (2014) in the Amazonian manatee from the Solimões and Negro rivers (1 to 10). This confirms that like the Amazonian and the West Indian manatee (Hartman, 1979; Marsh et al.,

2011), the African manatee is likelyl opportunistic in their diet.. This generalist ability by selection of multiple numbers of species per individual might also reflect the plant diversity of the feeding ground, as manatees do not usually feed at the same location

127

for an extended time. Foraging manatees often visit multiple sites in a single feeding bout (Akoi, 2004).

Among the 36 plants recorded from the 113 fecal samples analyzed in this study,

19, 9, and 8 were identified at the level of the species, genus and family taxa respectively. Among these plants, 14 genera were previously reported or documented in the list of African manatee food resources by Keith-Diagne (2014). Those previously reported species include E. pyramidalis, Acroceras zizanioides, Cyperus papyrus,

Eichhornia crassipes, Dalbergia sp, L. hexandra, Nymphaea lotus, Pennisetum purpureum, Rhynchospora corymbosa, S. molesta, Leptochla sp., Panicum sp., Ficus sp., and Macaranga sp. (Akoi, 2004; Husar, 1978; Keith-Diagne, 2014; Ogogo et al.,

2013; Powell, 1996; Reeves et al., 1988). In this study, we recorded 15 new plant species that were not previously reported in the manatee literature. The newly reported species include Cynodon sp., Hypperhenia, Pycreus lanceolata, Remirea maritima,

Eremospatha macrocarpa, Centrosema pubescens, Milletia mannii, Albizia sp.,

Platostoma sp., Acanthospermum sp., Calophyllum inophyllum, Canthium ciliatum,

Ficus copreifolia, Ficus rubiginosa, and Ascolepis. With these new species, the total number of documented food plant species for the African manatee now exceeds 90.

This high diversity of the African manatee diet is not surprising given the wide home range of the species that extends along approximately 6000km of coastline (measure estimated from Google Earth Pro) and to inland waters located over 2000km from the coastline. Also, the species’ home range encompasses a broad spectrum of habitat types from the dense equatorial rainforest to the semi-arid grassland of the Sahel, and from the fresh inland waters down to the coastal marine systems. Given the

128

opportunistic feeding behavior of the African manatee (Marsh et al., 2011), the vast area of occupancy, and the high plant diversity within their home-range, it’s expected that the

85 plant species documented to date might represent just a small fraction of the range- wide total number of food plants consumed by African manatees.

The manatee diet in DSRW was 96% made up of emergent grasses and occurred in all fecal samples analyzed. This observation is congruent with Akoi's (2004) manatee diet study in the Fresco Lagoon Complex in the Ivory Coast that concluded that the diet of the African manatee consisted predominately of leaves of emergent plant species along the shorelines. A similar proportion of emergent macrophytes has been reported among the Amazonian manatee population in the Mamirauá and Amanã

Sustainable Development Reserves in the Amazon River (Guterres-Pazin et al., 2014).

The larger proportion of emergent plants found in the diet of manatees in this study and that of the Amazonian manatee supports Domning & Hayek's (1986) hypothesis on the rostral deflection of the species and its implication on their diet niche. The authors found that the African and the Amazonian manatees with the lowest values in the rostral deflection spectrum of extent sirenians (15°-40° and 25°-41° respectively) is adapted for feeding on the emergent and natant vegetation. Conversely, dugongs have the highest rostral deflection (about 70°) which is more suited for feeding on bottom vegetation or rhizomes. The West Indian manatee with an intermediate rostral deflection (29°- 52°), between these two extremes, resonates well with their generalist foraging niche

(seagrass, natant and emergent vegetation).

Unlike the Amazonian manatee, floating plant species represented only an insignificant fraction of the African manatee diet in the DSRW, and no submerged plant

129

fragments were detected. These proportions reflect the plant composition of the habitat of the species in the DSRW which was made up of 70.7% of emergent plants, only

4.3% of floating plants and no submerged vegetation. This result further demonstrates the opportunistic feeding habit of manatees.

The high-frequency occurrence of plants from the family Poaceae in this study is similar to the results reported by Guterres-Pazin and co-authors (2014) and Colares &

Colares (2002), which showed that the family Poaceae was found in 91.5% and 96% of samples respectively. Plants belonging to the Poaceae family have also been reported in the diet of the West Indian manatee (Best, 1981). The consumption of plants from the

Poaceae family by all the manatee species highlights the importance of this family in the diet of Trichechids. Echinochloa pyramidalis appeared to be the most consumed manatee plant species of the DSRW, which occurred in 86.7% of the 113 fecal samples analyzed and contributed to about 53.5% of the species diet in the study area. Again, this result is unsurprising given that E. pyramidalis was the dominant shoreline plant species (47%) in the DSRW. A species of the same genera, Echinochloa polystachya, was reported by Colares & Colares (2002) and Guterres-Pazin and co-authors (2014) to occur in the Amazonian manatee feces (13.8% and 22.4% respectively), although the frequency of occurrence was lower when compared to the value obtained in this study.

Despite the abundance of fish and mollusk species in the DSRW, no fragment of fish or mollusk was present in any fecal sample examined in this study. This result does not corroborate with the local fishermen's reports of manatee eating fish and mollusk

(Takoukam Kamla, 2012). It neither corroborates with Keith-Diagne (2014) findings indicating that 6 to 7% of the diet of African manatee in Gabon (which is very close to

130

Cameroon and has a similar habitat as that in the DSRW) was made up of hermit crabs.

This difference could be due to fact that the diet analysis method used in this study did not have the capacity to detect fish or mollusk fragments present in the manatee feces because they were completely digested or the cellular structure of the fragment was not discernable at the magnification of 100X; which is a limit of the microhistological analysis. Also, all ingesta samples in this study were feces only. Stomach contents are less digested and allow for better detection of those food items (Hurst & Beck, 1988)but necropsy samples and stomach contents were not available for the present study.

Therefore, the diet result reported here may be biased in favor of the highly fibrous plants. Additional stable isotope studies from the DSWR will be crucial to confirm or deny the omnivorous diet of the African manatee there.

Manatee diet by location

The significant difference in manatee diet composition between the locations

(R=0.18, p=0.0003) of the DSRW is unsurprising as the shoreline species composition in those locations had a similar distribution. The greatest difference in diet composition was observed between Lake Ossa and the Sanaga Estuary (with an overall average dissimilarity of 66.9%). These two locations also showed the greatest Bray and Curtis distance (87% of dissimilarity). This is additional evidence of the opportunistic feeding behavior of manatees as the species tend to feed on the most available plants in each location. It is important to note that the Bray and Curtis dissimilarity matrix is a statistic that uses the frequency of each element at each site to assess the compositional dissimilarity between two sites. As a few species dominated most of the shoreline plants identified, the distance of species composition between locations may only reflect the differences of location for this small subset of subspecies. Similar observation on the

131

influence of the local vegetation on manatee diet has been reported amongst African manatee populations in the Ivory Coast (Akoi, 2004), Amazonian manatees (Guterres-

Pazin et al., 2014), Antillean manatee in Belize (Allen et al., 2018) and the Florida manatee (Alves-Stanley et al., 2010).

Echinochloa pyramidalis alone contributed to about 42% of the dissimilarity in diet composition between the two locations. E. pyramidalis represented only 3.5% of the shoreline vegetation of the Sanaga Estuary whereas, in Lake Ossa, it is the dominant plant species representing about 47% of its shoreline vegetation. In the Sanaga

Estuary, both E. pyramidalis and Cyperus sp. were the most consumed plants (37% and

30% respectively) even though E. pyramidalis was rare in the estuary. It is possible that feces from the Sanaga Estuary that had a higher occurrence of E. pyramidalis were transported from the Upper Sanaga River by the water flow. It could also be that manatee fed in the Upper Sanaga River where E. pyramidalis is more abundant before moving down to the estuary where they have deeper water to rest. Finally, it may be possible that manatees in the DSRW have a stronger preference for E. pyramidalis; therefore, they would feed on the limited available amount before feeding on the most available plant as a complemental food. The later hypothesis is very plausible as E. pyramidalis was the most represented food plant at the two locations (Lake Tissongo and Sanaga Estuary), even where it was not the dominant shoreline plant species. Also, in Lake Tissongo, the second most frequent and occurring fecal fragment plant after E. pyramidalis was E. macrocarpa, the dominant shoreline plant species of the lake. This further highlightg the selective habits of the manatee to a specific food plant, and its opportunistic behavior when the latter plant is not available.

132

Despite the abundance of mangrove in the Sanaga Estuary, no Rhizophora sp. fragment was recorded in the feces; whereas, the plant has been reported as an African manatee diet item in several countries including Cameroon (Husar, 1978; Keith-Diagne,

2014; Powell, 1996). The absence of Rhizophora sp. in the diet of manatee in this study could be due to the abundance of grasses in the mangrove area which are more accessible and maybe more nutritive for the manatees than the Rhizophora leaves hanging over the estuary edges. Manatees are, in general, are more grazers than browsers (Marsh et al., 2011).

Manatee diet by season in Lake Ossa

The seasonal effect on the manatee diet was explored only in Lake Ossa because of the low fecal sample size for the low-water season in the other locations.

There was a significant difference in the manatee diet composition between the low- and the high-water seasons in Lake Ossa. This result is unsurprising as the shoreline plant survey showed that the number of species encountered during the high season

(101 plant species) was nearly three-fold greater than that of the low-water season (39) even though the number of plant species was under-surveyed due to limited access by boat into the flooded shrubs. The difference in seasonal plant availability was reflected in the diet as only 18 plant species were recorded in the feces during the low water season and more (22) were observed during the high-water season. Also, the number of plant species per sample varied seasonally.

During the low-water season, the majority (68%) of fecal samples contained between one to three unique plant species and only 26% contained between four to six species. Meanwhile, during the high-water season, only 21% of fecal samples contained between one and three species, and the majority contained a greater number of unique

133

plant species (72%). These results contrast with those obtained for the Amazonian manatee showing that the species is more selective of their food plants during the high- water when they have greater access to forage and consume a more diverse number of plant species during the low-water season (Colares & Colares, 2002; Guterres-Pazin et al., 2014). The seasonal variation of the diet reported in this study is consistent with

Best's (1981) observations on the Amazonian manatee in the Amazon Basin. As water levels rise and flood the grass plains of the lake, which are inaccessible during the low water season, manatees have access to a more diverse and abundant plant species; which might justify the greater number of plant species recorded in the fecal samples during that season. When the water level drops and water surfaces recede during the dry season, manatee food availability is reduced and limited to floating or semi-floating species. Floating plant species were scarce in Lake Ossa, and E. pyramidalis was the major semi-floating species. Therefore, it is not surprising that the manatee diet in Lake

Ossa was highly dominated by E. pyramidalis (81.1%) during the low-water season whereas it only represented 44.2% of the diet during the high-water season (Figure 6,

Figure 7).

The abundance of food plants during the high-water season may be indirectly responsible for the seasonal manatee breeding activity. Fishermen in Lake Ossa reported seeing and hearing manatee mating herds mostly during the rising water season starting in June/July. This time period of mating of the African manatee (start of the rainy season) is also reported from many other countries (Keith-Diagne, pers. comm.). Mating usually takes place before dawn between 3 to 5am. I once heard a manatee mating herd in Lake Ossa splashing water around 4:00am. The fishermen

134

reported often seeing manatee placentas floating at the surface of the water around the same season. Like the Amazonian manatee (Best, 1981, 1982), the African manatee may have adapted to synchronize their breeding period with the seasonal variation in the availability of food so that calves are born during the rising water season when females have access to a greater amount and better quality of forage necessary to store enough fat for the gestation and lactation.

Manatee diet by feces size

Assuming that manatees from a location have similar diet and that feces bolus size are correlated to the body size of the manatee and thus to the age class, and that feces diameter in the lower range (<4cm) represents calves and juveniles, and those in the upper range (>4cm) are from adult manatees, diet of the African manatee in Lake

Ossa would not vary significantly between manatee size classes. This finding would be congruent with previous studies in Belize and Mexico (Allen et al., 2018; Castelblanco-

Martínez et al., 2009). The smaller occurrence of E. macrocarpa and Cyperus sp. in small size feces than in large size feces might be due to the spinier and higher rigidity of their leaves as compared to the relatively soft leaves of the frequently consumed plant,

E. pyramidalis. The E. macrocarpa and Cyperus sp. leaves might be too hard for a juvenile to masticate or digest.

Inference on the Manatee Movement within DSWR

Food plant availability is one of the main Florida manatee habitat requirements besides access to warm, fresh and deep waters of at least 2m (R. Reep & Bonde, 2006;

US Fish and Wildlife Service (USFWS), 1999). Any change in their water requirements may trigger seasonal or permanent migration. Unlike Florida manatee habitat, where the temperature is the limiting factor during the winter season (Marsh et al., 2011; Reep &

135

Bonde, 2006; USFWS, 2001), African, Amazonian, and Antillean manatees’ movements seem not to be influenced by temperature as they live in tropical areas where the water is warm year-round; though temperature may influence movements in the southern range of Brazil for Antillean manatees, and possibly north and south range extensions in

Africa (Bonde, pers. comm.).

Similarly, access to freshwater may not be a limiting factor to the African and the

Amazonian manatees that inhabit freshwater systems. Thus, the seasonal movements and the ecological activities of both species appear to be controlled mainly by behavior, forage availability and water level (Akoi, 2004; Best, 1982; Marsh et al., 2011).

However, food availability and accessibility are dependent on the water level, which in turn depends on the seasonal rainfall.

The African manatees in the DSRW feed mainly on the emergent vegetation mostly distributed along the shoreline where water depth is the lowest and very often too shallow to allow a 300-kg manatee to swim over to the bank vegetation (Arraut et al., 2010; Takoukam Kamla, 2012). During the low-water season, E. pyramidalis is the main diet of the African manatee but it is only accessible in the few areas of the lake where the shoreline is deep enough to accommodate manatees. Thus, during the low- water season (between January and April), because of the lower food availability relative to the high-water season, the manatee population in Lake Ossa may leave the lake and move to the Sanaga Estuary where they can take advantage of the high tide for access to shoreline plants. The interview survey conducted by Takoukam Kamla

(2012) among 144 fishermen in the DSRW indicated that manatees are more frequently seen in the Sanaga River estuary during the dry season than during the rainy season.

136

In addition to the identification of plant species in fecal specimens, correlations were made between plant material identified in fecal samples and the presence or absence of that plant in the area where the fecal sample was collected. A Venn diagram was built through the Heberle et al. (2015) web-based software to analyze the inclusivity and exclusivity between diet and potential food plants by location. Inference on manatee movement was based on the presence of unique plant species (present exclusively in one location) in feces from areas where the plant is not found. In order to understand manatee movement between Lake Ossa and other locations, their identified diet plant species were matched against potential food plant species that were present in any other location excluding Lake Ossa. The presence of a plant species exclusive to the other location in feces collected in Lake Ossa could be an indication that the manatee fed in the location where the plant is present before migrating into Lake Ossa

One fecal sample from Tissongo Lake showed plant fragments of Eichhornia crassipes and two with Pennisetum purpureum. These two plants, according to our shoreline plant survey, were only present in the lower portion of the Sanaga River.

Neither was observed in the plant community of Lake Tissongo. This suggests that manatees move between Lake Tissongo and the lower part of the Sanaga River. Lake

Tissongo and the Sanaga River are connected through a 7km narrow channel. Unlike

Lake Ossa, water flows in one direction throughout the year, from the lake to the river.

Therefore, it is most likely that the feces collected in Lake Tissongo that contained a fragment of E. crassipes and P. purpureum was from a manatee that fed in the lower part of the river before entering the lake. This result is expected as the two locations are hydrologically connected and are spatially proximate.

137

Three tagged manatees in the Senegal River were reported migrating long distances up to 281km, presumably in search of food (Diagne et al. unpublished data,

Keith-Diagne, 2014). In contrast, other African manatee populations seem to remain in a small home range when all their habitat requirements are met. For instance,12 tagged manatees in the Fresco Lagoon complex showed a very small home range of less than

5km (Akoi, 2004). Also, 17 radio-tagged manatees in the lagoon system in the Ivory

Coast migrated only short distances within about 10km and on rare occasions would migrate up to 42km. The long-distance dry season migration in the Senegal River might be due to the scarcity of food during the dry season given the Sahelian climate of the area. It is also possible that this long migration was unusual as the three manatees were rescued and starving individuals trapped behind a dam in the Senegal River.

Whereas, with the abundance of macrophyte vegetation within the lagoon complex of

Ivory Coast, long-distance migration of the manatee there might be unnecessary.

Like the lagoon complex of Ivory Coast, the DSRW has abundant macrophyte vegetation, suggesting they would make only short migrations. However, in some areas, during the low-water season, water recedes from the shoreline, making this vegetation less accessible. Therefore, manatees may migrate to other areas with better food accessibility; hence, manatees from Lake Tissongo may migrate to the lower part of the

Sanaga River.

Conservation Implications

With the recent massive proliferation of Salvinia molesta that has killed an important local plant population of E. pyramidalis, there has been a drastic decrease in the relative abundance of manatees in Lake Ossa, as documented by the manatee monitoring program of the African Marine Mammal Conservation Organization

138

(AMMCO), which has conducted monthly surveys of the population in the lake since

2015. It is likely that the manatees have migrated to other habitats out of Lake Ossa where there is better food availability.

It appears that the proliferation of Salvinia has profoundly modified the plant species composition of the Lake during the low and the high-water seasons. As mentioned earlier, the manatee diet during the low-water season in Lake Ossa was highly dominated by E. pyramidalis (81.1%). However, a large surface area of the plant mat has been killed by the invasion of S. molesta. Moreover, it was noticed that the manatees do not feed on E. pyramidalis when it is surrounded by S. molesta (Figure A-

7). Thus, habitat suitability for the African manatee in Lake Ossa during the low-water season may have declined significantly with the advent of the S. molesta proliferation.

Our plant surveys showed that during the high-water season, manatees used to have access to more diverse food plant species from the flooded shrub forest.

Unfortunately, the current S. molesta proliferation has also invaded and extended its devastating floating bed well into the flooded area, killing most plant species and turning the area into almost a monospecific vegetation of S. molesta, with dead plant detritus beneath the bed (Figure A-7). Therefore, the flooded areas of Lake Ossa that used to be an important feeding area for the local manatee population during the high-water season appears to have significantly lost its ecological value for the manatee, given that they do not feed on S. molesta.

Only one fecal sampled in Lake Ossa contained some fragments of S. molesta which may have been ingested accidentally, suggesting that S. molesta is not an important plant for the manatee diet. Although scientists have not yet established clear

139

determinants of the nutritional quality this plant may have for sirenians (Aragones et al.,

2012), it is suggested that despite their aquatic life, that manatees have dietary requirements similar to that of large terrestrial herbivores which are well known and can, therefore, be used for the interpretation of the diet analysis in manatees (Marsh et al.,

2011). Moozhiyil & Pallauf (1986) studied the chemical composition of S. molesta and its potential as a source of food for terrestrial ruminants. The authors found that the plant has a relatively high crude protein (CP) concentration (12.4% of dry matter) comparable with CP concentration (15.3% in DM) of E. pyramidalis in Cameroon (Pare et al., 2011). However, S. molesta contains a high concentration of crude ash (17.3% in

DM), lignin (13.7% in DM) and a considerable amount of tannins (0.93% in DM) that may lower acceptance and digestibility by ruminants. Although the concentration of tannin in S. molesta is well under the range (2-3%) of tannin concentrations of most manatee food plant species (Best, 1981), it is expected that it will have a negative effect on the digestibility of ingested nutrients (Moozhiyil & Pallauf, 1986). Lignin reduces the nutritional availability of plant fiber as it constitutes a physical barrier to microbial enzymes and prevents the latter from digesting the cell-wall polysaccharides of the plant. Because of this, S. molesta is known as an “anti-quality component in forages”

(Moore & Jung, 2001). Therefore, it is not surprising that the African manatee in Lake

Ossa would not choose to feed on S. molesta and would tend to avoid the plant. The

Florida manatees have also been reported avoiding plant species with a high concentration of toxic secondary components (Bengtson, 1981; Reynolds III, 1981).

Salvinia molesta proliferation is a serious threat to the plant diversity of Lake

Ossa, including manatee food plants. The plant doubles its biomass every four to ten

140

days when conditions are favorable (Mitchell & Tur, 1975). In Lake Ossa, the conditions are very favorable for the Salvinia proliferation: water is stagnant, it has become eutrophic, nutrient enrichment is increasing (Takoukam Kamla et al., unpublished data), and there is enough light throughout the calendar year. So, if nothing is done, the lake surface may be completely covered by S. molesta in a few months or years. Therefore, to save the African manatee and the biodiversity in Lake Ossa, it is urgent to mitigate the proliferation of S. molesta.

Three mitigation options exist to control S. molesta, including mechanical removal by hand or using harvester machines; chemical removal using herbicides and a biological solution that uses an insect called Salvinia weevil (Cyrtobagous salviniae) that feeds exclusively on the S. molesta (Cozad, 2017; Flores & Carlson, 2006). The latter solution appears to be better adapted for the African context as it has been successfully applied to several countries on the continent such as Senegal, Namibia, and South-Africa (Martin et al., 2018; Pieterse et al., 2003). Therefore, the biological control of the S. molesta could help temporally restore the vegetation of the lake and may create favorable conditions for the recovery of its manatee population.

Salvinia molesta proliferation in Lake Ossa is a symptom of a more profound problem of nutrient enrichment of the lake; and, addressing only Salvinia proliferation is equivalent to treating the symptoms without addressing the disease itself. It is possible that the removal of S. molesta could give way to the proliferation of other invasive aquatic plants like Eichhornia crassipes (water hyacinth). S. molesta thrives best at higher temperatures (Cary & Weerts, 1983) and can survive extreme temperatures between -3° and 43°C (Whiteman & Room, 1991). With climate change and expected

141

increasing temperature and rainfall around lake Ossa, S. molesta may proliferate faster because of the resulting increasing evaporation and run-offs that may increase nutrient concentration in the lake (Short et al., 2016). Therefore, for a long-lasting and sustainable solution, it is crucial to mitigate nutrient enrichment of the lake and the

Sanaga River. An important step towards that end will be to develop and implement a water management plan for the country.

It is essential to also establish no-fishing and no-disturbance zones in areas of the lake where beds of E. pyramidalis have not yet been invaded by S. molesta, as those plots might be valuable feeding areas during restoration efforts for the few manatees still present in the lake. Therefore, a complete aerial mapping of the lake macrophyte vegetation should be performed to identify such areas for adaptive management efforts.

Overall, this study has documented for the first time most of the macrophytes and other shoreline vegetation of the DSRW and identified 160 species. A microscope slide library of 116 of 160 plant species has been prepared and is securely kept at AMMCO’s headquarter in Dizangue, Cameroon where it can be freely accessed and referenced for future manatee diet studies in Cameroon or other regions of Africa.

Some of the plants encountered have been uploaded in iNaturalist and will be uploaded in the GBIF (Global Biodiversity Information Facility) database for archival and future references.

This study also documented the African manatee diet in Cameroon for the first time and revealed the most common plant species consumed by the species. This information can play a critical role in the management strategy of the species.

142

Finally, the current study documents the proliferation of the Salvinia molesta in

Lake Ossa. The shoreline plant surveys conducted in the framework of this study in

2016 before the expansion of Salvinia will serve a baseline to monitor the species and its impact on E. pyramidalis (the major manatee food plant in Lake Ossa) and potential other water quality issues.

The shoreline vegetation and diet survey in the Sanaga River, Estuary and Lake

Tissongo was only focused on the high-water season. Sampling during the low-water season in these areas will provide a better understanding of the seasonal diet dynamic.

Because of time constraints and logistic limitations, most of the flooded forest was not surveyed. Future macrophyte and manatee diet studies in the areas should investigate the variety of plants present in the flooded regions.

The current study was also limited by the lack of stomach content samples, which are very difficult to obtain. Stomach contents from stranded manatees could be very valuable in future studies as they will help to estimate and determine the foraged species that are missing or were undetected when running microhistological analysis from fecal samples. It will also be interesting to investigate new molecular techniques such as genetic analysis whereby the mitochondrial DNA of all the fecal plant fragments could be extracted and screened against a library using primers specific to each of the voucher plant species encountered in the areas where the feces were collected.

143

Table 3-1. Diversity index of the plant species recorded in the African manatee feces for each surveyed location within the downstream Sanaga River watershed. Sanaga Upper Lake Lake All

Estuary Sanaga Tissongo Ossa sites Sample size 33 9 11 60 113 Minimum Number of 24 15 16 24 32 plant species Species diversity 1.81 1.53 1.58 1.45 1.83 (Shannon index H)

Table 3-2. Survey effort and diversity index of shoreline vegetation by location in the DSRW. Sanaga Upper Lake Lake All Estuary Sanaga Tissongo Ossa sites Number of plots surveyed 80 146 64 668 958 Number of species 68 96 16 111 160 Species diversity (Shannon index H) 3.09 3.55 1.86 2.61 3.52

Table 3-3. Pair-wise coefficient of dissimilarity (Bray and Curtis distance) of shoreline plant communities between locations in the downstream Sanaga River watershed. Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa Sanaga estuary 0.00 Upper Sanaga 0.66 0.00 Lake Tissongo 0.82 0.76 0.00 Lake Ossa 0.87 0.68 0.69 0.00

144

Table 3-4. Major plant species surveyed along the shoreline of the downstream Sanaga River watershed and their relative contribution to the Bray-Curtis similarity index between Sanaga Estuary, Upper Sanaga, Lake Tissongo and Lake Ossa. Values were calculated using the SIMPER method. Average Abundance % Sanaga Upper Lake Lake Contribution Cumulative Species Estuary Sanaga Tissongo Ossa % % Echinochloa pyramidalis 3.25 15.8 16.8 46 34.45 34.45 Dissotis erecta 2.44 1.89 27.1 5.15 9.331 43.78 Eremospatha macrocarpa 0.625 0 32.3 2.07 7.155 50.93 Alchornea cordifolia 10.6 4.32 6.72 0.531 5.626 56.56 Ipomoea alba 0 9.35 0 1.59 4.792 61.35 Rhizophora racemosa 19.9 0 0 0 4.562 65.91 Pennisetum purpureum 5.81 6.61 0 0.322 4.177 70.09 Dissotis falcipila 0 0.137 5.57 3.926 74.01 Ficus capreifolia 3.25 5.6 0 0.132 3.258 77.27 Eichornia crassipes 8.31 3.8 0 0 3.105 80.38 Ficus capreifolia 0 5.14 0 1.03 2.83 83.21 Canthium ciliatum 0 0.685 0 3.82 2.813 86.02 Ludwigia stolonifera 0 0 0 3.24 2.553 88.57 Laccosperma secundiflorum 0 0 5.63 1.89 2.213 90.79 Acroceras zizanioides 1.25 1.71 0 1.13 1.837 92.62 Polygonum lanigerum 0.375 2.33 0 0.892 1.827 94.45 Salvinia molesta 0 0 0 2.15 1.564 96.01 Fuirena umbellata 0 0 1.56 1.89 1.535 97.55 Laccosperma robustum 0 0 0 2 1.287 98.84 Cyperus haspan 0.5 0 0.781 1.29 1.164 100 Overall Average dissimilarity = 90.36%

145

Table 3-5 List of plant type, family and species (with the common names) and their per cent frequency in the 113 African manatee feces collected in the downstream Sanaga River watershed. Plant types Common English names n Per cent (%) Grasses 96.11 Poaceae Grass 71.08 Echinochloa pyramidalis Antelope grass 98 53.51 Southern cutgrass, swim Leersia hexandra rice grass 32 4.08 Unidentified 13 8 3.53 Poaceae stem 43 3.48 Pennisetum purpureum Elephant grass 28 2.62 Acroceras zizanioides Oat grass 17 1.88 Bermuda grass, devil Cynodon sp.* grass 8 1.23 Jaragua grass, Giant Hyparrhenia sp.* thatching grass 12 0.52 Panicum maximum Guinea grass 4 0.13 Unidentified 6 1 0.07 Unidentified 15 1 0.03 Leptochloa sp. Sprangletops 1 0.01 Cyperaceae Flatsedge 20.08 Cyperus sp. Nut sedge, nut grass 26 12.85 Rynchospora corymbosa Matamat 38 5.15 Pycreus lanceolatus* Epiphytic flatsedge 9 0.87 Remirea maritima* Beachstar 3 0.43 Ascolepis sp.* 5 0.34 Other Cyperaceae 10 0.21 Unidentified 2 3 0.09 Unidentified 8 2 0.08 Unidentified 4 2 0.06 Arecaceae Palm 4.72 Eremospatha macrocarpa* Small rattan 31 4.72 Pontederiaceae Pickerelweed 0.08 Eichornia crassipes Water-hyacinth 2 0.08 Vascular Plants 0.80 Asteraceae Aster, Daisy, Sunflower 0.39 Acanthospermum sp.* Starburr 7 0.22 Other Asteraceae 2 0.17 Fabaceae Legume, pea, bean 0.24 Centrosema pubescens* Butterfly pea 4 0.17 Rosewood, Bombay, Dalbergia sp. black wood 1 0.05 Millettia mannii* Ndu Ezi (Nigera, Igbo) 1 0.02 Albizia sp.* Silk tree 1 0.00

146

Table 3-5. Continued Plant types Common English names n Per cent (%) Lamiaceae pure, deadnettle, sage 0.02 Platostoma sp.* Chinese mesona 1 0.02 Nymphaeaceae Water lilies 0.12 White Egyptian lotus, Nymphaea lotus tiger lotus 1 0.12 Salviniaceae Watermoss 0.03 Giant Salvinia, Kariba Salvinia molesta weed, water fern 1 0.03 Other plants species 3.24 Fruits Eremospatha macrocarpa Small rattan Calophyllum inophyllum* Beach touriga Canthium ciliatum* Hairy turkey berry Ficus copreifolia* Fig shrub Ficus rubiginosa* Rusty fig Macaranga sp. Parasol leaf tree Clay 1 0.71 *Newly reported African manatee food plants

147

Figure 3-1. Map of the downstream Sanaga River watershed showing the surveyed areas and the distribution of the surveyed plots.

148

Figure 3-2. Shoreline vegetation composition profile by plant type across the four surveyed locations of the downstream Sanaga River watershed.

149

Figure 3-3. Distribution of shoreline vegetation by species and family. A) Relative abundance of the top 20 plant species and B) families identified along the shorelines of the downstream Sanaga River watershed.

150

Figure 3-4. Shoreline plant species composition profile across the surveyed locations of the downstream Sanaga River watershed.

Figure 3-5. Relative abundance of top 15 plant species identified along the shorelines of Lake Ossa by season (water level)

151

Figure 3-6. Per cent frequency distribution of the identified plant items from African manatee feces collected from the four locations within the downstream Sanaga River watershed.

Figure 3-7. Percent frequency distribution of the identified plant items from the African manatee feces collected in Lake Ossa by season

152

CHAPTER 4 ASSESSMENT OF THE EFFECTIVENESS OF NONINVASIVE FREE-FLOATING FECAL SAMPLES OF THE AFRICAN MANATEE AS A SOURCE OF DNA FOR GENETIC ANALYSIS USING MITOCHONDRIAL, MICROSATELLITE, AND SEX IDENTIFICATION MARKERS

Background

With the advent of the polymerase chain reaction (PCR), non-invasive sampling for DNA has become possible by requiring a very small amount of DNA for the template

(Morin & Woodruff, 1996). Contrary to destructive sampling where the animal is killed and non-destructive sampling where the animal is live captured and biopsied to collect tissues, non-invasive sampling relies on a tiny amount of DNA left behind by the animal either as feces, urine, hairs, scales, feathers or shed epidermal cells (Taberlet et al.,

1999). Non-invasive samples have been very attractive for field biologists since the

1990s as it allows for genetic sampling with no need to observe, handle, or capture the animal (Höss et al.,1992; Ramón-Laca et al., 2015; Taberlet et al., 1999). The use of feces as a source of DNA can also provide several benefits over other DNA sources.

Fecal sampling is non-invasive as it does not require the collector to handle or even detect the target animal. Moreover, these samples are more available as they are frequently generated by all individuals. Finally, the feces of many species can be detected relatively easily without the need for sophisticated logistical equipment of

(Long et al., 2012).

Non-invasive Genetic Studies in Manatees

The non-invasive sampling approach has been widely used in terrestrial and semi-aquatic species such as primates, bears, deer, and elephants (Brazeal et al.,

2017; Foote et al., 2012; Schwartz et al., 2007). However, its application in the aquatic environment has been limited due to the lesser accessibility of non-invasive samples

153

such as feces, shed skin, or hair. For instance, feces of many species do not float in water, and when they do, they may become rapidly disintegrated by mechanical waves, current patterns, and degraded by fish and microbes (Wotton & Malmqvist, 2001). Also, unlike feces in the terrestrial environment whose DNA is easily preserved as the sample dries up, feces in the aquatic environment are subject to faster decomposition and contamination (Tikel et al., 1996; Muchett et al., 2009).

The recent development of the environmental DNA (eDNA) approach has now made it possible to genetically monitor and even quantify cryptic aquatic species from collected water samples by detecting the DNA markers of shed cells (Foote et al., 2012;

Thomsen et al., 2012). The application of eDNA is still in development as there are some challenges to be addressed, such as accurately translating the detection or non- detection of species-specific fragments. For instance, the non-detection of species- specific DNA fragment does not always imply the absence of the species and the presence of the later could have been the result of the transportation of the species

DNA from another location through water current dispersal (Roussel et al., 2014).

Environmental feces as a source of DNA was documented for the first time in a marine mammals in 1996, specifically for the dugong (Tikel et al., 1996). Before then, marine mammal biologists mainly relied on the remote sampling of skin biopsies from free-ranging animals to obtain DNA samples. Tikel and colleagues (1996) successfully isolated a 193 base pairs (bp) fragment of the D-loop region of the mtDNA from fecal samples of dugongs. Muschett and colleagues collected 35 free-ranging fecal DNA samples of the Antillean manatee in Bocas del Toro. The authors developed a manatee specific forward primer (LTMC01) that was conjointly used with two reverse dugong

154

primers (HDDCR01 and HDDCR02) that had been used to amplify mtDNA segments of

450 and 534bp respectively and yielded amplification success of 70% and 20% respectively. The authors suggested that the difference in amplification success between the two reverse primers might be due to the difference in the DNA fragment length they generated. Díaz-Ferguson and collaborators (2017) was the second and most recent published study using non-invasive fecal DNA of manatees for genetic analysis. The authors extracted DNA from 20 fresh, free-floating samples of the

Antillean manatee collected opportunistically in the Rio Negro River in Panama. The author employed the control region primers CR5 and CR4 to amplify a segment of the mtDNA. However, only seven samples yielded enough DNA for downstream analysis and of the seven, only three fecal samples yielded reliable sequences.

The use of non-invasive fecal DNA for genetic assessment has not previously been explored for the African manatee, though feces from the species at are frequently observed floating on the water’s surface. Recently, Hunter et al. (2018) developed

Cytochrome b quantitative and droplet digital PCR eDNA assays to detect shed eDNA of African manatees, Amazonian manatee and the two subspecies of the West Indian manatee from water samples collected in their respective habitats. The probe detected eDNA in only one of the three locations within Lake Ossa in Cameroon where the samples were collected. Hunter and colleagues eDNA results are the first of remote detection of African manatees in the region. This remote approach will facilitate determination of the distribution of this elusive species. Given that fecal samples retain cells from the animal as opposed to free floating cells in water samples, we would

155

expect that feces of the African manatee could yield even better quality and higher quantities of DNA.

No published study to date has assessed the use of environmental fecal samples to isolate nuclear DNA in sirenians. Previous fecal DNA studies were limited only to the isolation and amplification of mtDNA fragments and more precisely, the control region

(Díaz-Ferguson et al., 2017; Muschett et al., 2009). Isolating longer length and multi-loci nuclear DNA fragments from non-invasive samples is more challenging than mtDNA, because the latter is more resistant to degradation and is in greater amount (number of copies) than nuclear or microsatellite DNA. Analyzing nuclear DNA fragments from fecal samples allows researchers to determine more genetic parameters with finer resolution than that offered by mtDNA. In manatees, for instance, nuclear DNA isolated from tissue samples has allowed for sex determination (Lanyon et alet al., 2009; McHale et al., 2008; Tringali et al., 2008a), individual identification, relatedness, inbreeding, and kinship (Davis, 2014); genetic population structure, and connectivity (Hunter et al.,

2010a, 2012; Nourisson et al., 2011; Tucker et al., 2012).

To date, there is no published genetic study involving the use of either invasive or non-invasive collected African manatee nuclear or microsatellite DNA for analyses.

Keith-Diagne (unpublished data) has isolated and genotyped sex and microsatellite

DNA from 59 African manatee tissue samples, across several countries of the species distribution range, mostly from carcasses. Keith-Diagne used 35 nuclear DNA microsatellite markers developed by Garcia-Rodriguez (2000) Hunter et al. (2010b) and

Tringali et al. (2008b), and three sex markers developed by Tringali et al., 2008a.

156

However, statistical or genetic parameters are not available from the genotyped samples at this time.

African Manatee Conservation Status

The African manatee is the least studied of all sirenians including the West Indian manatee in the Caribbean, the Gulf of Mexico and the eastern Atlantic coast, the

Amazonian manatee in the Amazon River, and the dugong in the western Pacific Ocean and the eastern coast of Africa (Marsh et al., 2011). The dugong populations of the eastern coast of Africa is also understudied. The African manatee is distributed across rivers, lakes, lagoons, estuaries, or continental shelves along the west and central

Atlantic coast of Africa (Husar, 1978; Powell, 1996). In Africa, they are threatened by poaching, accidental catches in fishing nets, and habitat degradation (Powell, 1996) despite the legal protection provided by all 21 range countries (Keith-Diagne, 2015).

African manatees are also under the protection provided by of international laws. They are Red Listed by IUCN (The International Union for Conservation of Nature) as

‘’Vulnerable’’; (IUCN 2019); belong to Appendix I of the CITES (Convention on

International Trade in Endangered Species) and of the CMS (Convention on Migratory

Species). In Cameroon, the African manatee is legally protected and belongs to Class A of integrally protected species.

Study Justification and Objectives

The use of the fecal DNA for genetic analysis may contribute to improved understanding of species diversity, abundance, structure, connectivity, and evolution.

Such information is paramount to inform management decisions and design effective management strategies that could save the species from multiple threats. The African manatee is elusive and secretive. It is very difficult to collect tissue samples from the

157

African manatee. Even the carcass of the animal is usually secretly consumed by the local human community. Therefore, relying on African manatee tissues as a source of

DNA for genetic analysis will undoubtedly extend the length for the completion of a study. Therefore, the use of non-invasive DNA could address the limitations imposed on future genetic studies of the species by the scarcity of adequate and available tissue samples.

African manatee feces in Cameroon are highly fibrous compared to Florida manatee feces that were collected from various locations in Florida. This is because the former feed mostly on emergent fibrous vegetation like antelope grass (Echinochloa pyramidalis); whereas the Florida manatees feed mostly on seagrass that is lower in fiber content (Best, 1981).

In this study, we investigated the reliability of African manatee free-floating fecal samples to provide acceptable DNA quantity and quality for genetic analysis using mtDNA, microsatellite, and specific sex identification markers. We first determined the optimized fecal DNA isolation protocol by comparing the efficacy of three extraction protocols and determining which part of the fecal samples yielded higher quality manatee DNA. Secondly, we used the optimized method on a larger sample size to assess the quantity and quality of the isolated DNA. Thirdly, we determined the reliability of the isolated DNA as a template for amplifying mtDNA, microsatellite, and sex identification markers. And finally, we evaluated the effect of habitat on the quantity and quality of the obtained fecal DNA.

158

Methods

Comparison of the Efficacy of the DNA Isolation Methods

Sample collection and storage

Five fresh African and five Florida manatee free-floating fecal samples were collected in Lake Ossa, Cameroon and in Florida Power and Light Interim Warm Water

Refuge, Cape Canaveral on June 6th, 2017 and February 3rd, 2017 respectively. The samples were preserved in 50ml sterile tubes half-filled with 95% ethanol and stored at the U.S. Geological Survey laboratory in Gainesville, Florida. Because African manatee fecal samples contained a high percentage of fibrous plant material, we wanted to determine whether manatee DNA would be more abundant on the fibrous or non-fibrous component of the fecal samples. Thus, for each of the African manatee fecal sample, the fibrous and the non-fibrous material of the samples were separately extracted. We removed all the plant fibers and fragments from the 50ml tube containing the preserved samples. The remaining finer-sized fecal material (non-fibrous) was separated from the ethanol by centrifuging (at 4500rpm for 20min), the supernatant (ethanol) was discarded and the pellet (non-fibrous fecal material) retained. Thus, three types of fecal samples were used for this study, including non-fibrous Florida manatee feces, non-fibrous

African manatee fecal material, and fibrous African manatee feces. Each of the five

Florida manatee feces was subsampled in nine aliquots of 300mg each. The fibrous and non-fibrous part of each African manatee feces were also subsampled in nine fecal aliquots of 300mg each. Thus, each of the three feces types had a total of 45 subsamples for a grand total of 135 subsamples consisting of 300mg that were separately introduced into 5ml centrifuge tubes, ready for extraction. A detailed description of the fecal sample preparation is presented in Appendix B.

159

DNA extraction

Total DNA, considered as the sum of both exogenous (non-target DNA), and endogenous DNA, yielded values during the isolation process. We used 2μl of DNA extract for each of the 135 aliquots to analyze on the Epoch™ Microplate

Spectrophotometer, and the DNA concentration (ng/μl) and purity (A260/A280) values were recorded. The measures were repeated twice for each aliquot, and the average value was recorded. Each sample had three aliquots per extraction method and feces type (Appendix B). The aliquot for each sample with the A260/A280 value closest to 1.8 and having the highest DNA concentration was retained for the PCR amplification.

Thus, a total of 45 aliquots was retained for the downstream analysis.

PCR amplification

In order to assess the quality of pure manatee DNA in each sample, the primer pair CR-4 (5’-AGATGTCTTATTTAAGAGGAA-3’) and CR-5 (5’-

TCACCATCAACACCCAAAGC-3’) developed by Garcia-Rodriguez et al. (1998) was used to amplify a 410bp segment of the control region displacement loop of the species mtDNA. Because DNA extracts from the 2CTAB/PCI methods had a very high DNA concentration (more than 100ng/μl), they were diluted down to 50ng/μl before the PCR reaction. Each PCR reaction included: 1μl of DNA extract (with total DNA concentration varying between 5 and 50ng/μl), sterile PCR water, 1 x PCR buffer (10mM Tris-HCl, pH

8.3, 50mM KCl, 0.001% gelatin; Sigma-Aldrich, Inc., St. Louis MO), 0.8mM dNTP,

3.0mM MgCl2, 0.25mM BSA, 0.25μM of each primer, and 0.04 units of Sigma Jump

Start Taq DNA polymerase. All PCR amplifications were carried out on a SimpliAmp thermal cycler (Applied Biosystems, Thermo Fisher Scientific Inc). The PCR conditions were as follows: Initial denaturation at 94˚C for 5 minutes, then 34 cycles of

160

denaturation for 1 minute at 94˚C, annealing for 1 minute at 53˚C, follow by extension for 1 minute at 72˚C, and final extension for 10 minutes at 72°C. The amplification was completed with a 4°C hold.

PCR products were examined on a gel electrophoresis using 1% agarose and a mixture of 5μl of the PCR product and 2μl of 1x load dye. PCR products that indicated a band on the gel were cleaned using ExoSap purification (Affymetrix, Santa Clara, CA).

ExoSap was conducted as follows: 5μl of PCR product from each sample was mixed with 2μl of ExoSap solution, and the mix was heated on a SimpliAmp thermal cycler to

37°C for fifteen minutes, followed by heating to 80°C for 15 minutes to inactivate the enzyme. Clean products were submitted to MCLab (San Francisco, California, USA) to run MC Easy Format™ DNA sequencing. The quality of the sequence was accessed using GENEIOUS Version 11.1.5. The same software was used to align the sequences along with an African manatee reference sequence AY963895 published by Vianna et al. (2006). Then the sequences were trimmed to the size of the reference sequence

(410bp). The values of the high-quality percentage (HQ%) of the trimmed sequences were recorded.

Applying the optimized extraction method on the larger pool of samples

The extraction method that yielded the highest HQ% was then used to isolate

DNA from 235 African manatee fecal samples collected in Lake Ossa (93), Lake

Tissongo (60) and the Sanaga River (82) between May 2016 and August 2017. The amount of fecal material used varied between 400 and 1260 mg depending on the dryness of the sample; this variation in water content was likely due to the type of plant material on which the animal fed.

161

Of the 235 extracted fecal samples, 110 with a total DNA concentration greater than 10ng/μl were selected for PCR amplification of the mtDNA and nuclear DNA. A second round of extraction of the highest yielding 91 fecal samples using a larger amount of fecal material (400 and 1260mg) was performed to yield improved DNA yield and minimize possible sample and extraction cross-contamination. The second-round extraction was also useful to generate enough DNA template volume to accommodate the multiples microsatellite PCR replicates performed in this study.

DNA purification

Before DNA amplification of the 91 re-extracted samples, the OneStep™ PCR

Inhibitor Removal Kit (Zymo Research) was used to purify DNA extracts further, and then the A260/280 was remeasured on an Epoch™ Microplate Spectrophotometer to assess the change in the DNA purity. The Zymo OneStep inhibitor removal procedure allows efficient removal of contaminants from the DNA preparation that can inhibit the

PCR reaction. More importantly, it is designed to remove polyphenolic compounds, humic acids, and tannic acids contained the environmental water and plant fragments within the fecal sample.

Mitochondrial DNA amplification

The control region of the mtDNA of each of the 110 samples was amplified in a

25μl reaction volume containing 12.5μl AmpliTaq Gold™ 360 Master Mix (ThermoFisher

Scientific), 1μl of 360 GC Enhancer, 0.5μl (10uM) of each primer (forward and reverse),

8μl water and 2.5μl DNA. All PCR amplifications were carried out on a SimpliAmp thermal cycler. The PCR conditions were as follows: Initial denaturation at 94˚C for 5 minutes, then 34 cycles of denaturation for 1 minute at 94˚C, annealing for 1 minute at

162

53˚C, follow by extension for 1 minute at 72˚C, and final extension for 10 minutes at

72°C. The amplification is completed with a 4°C hold.

Nuclear DNA Amplification

Optimization of the nuclear fecal DNA amplification

Most microsatellite genetic studies of manatees have used a PCR amplification protocol as described by Pause et al. (2007) or Garcia‐Rodriguez et al., (2000). We tested the above PCR protocol on a few randomly selected manatee fecal DNA samples, but it resulted in poor microsatellite amplifications due to the low quantity and poor quality of isolated fecal DNA. Therefore, we performed a series of optimization processes. First, we tested a multiplex pre-amplification protocol by Piggott and co- authors, (2004) which improved amplification but resulted in abundant non-specific amplification (Figure B-2). Second, the PCR amplification was further improved by using a pre-made commercial master mix (Type-it Multiplex PCR Master Mix). Third, to minimize spurious priming during PCR, we applied a touchdown procedure (Don et al.,

1991), which decreased considerably the number of non-specific peaks (Figure B-2).

These series of optimizations were finally combined into a single PCR protocol described in detail below.

Applying the optimized nuclear PCR protocol

The two-step multiplex pre-amplification PCR method was used to amplify 15 microsatellite and three sex marker loci from the same 110 African manatee free- floating feces samples. The PCR reactions were performed as described by Piggott et al. (2004) with some modifications described herein. The two-step multiplex pre- amplification PCR method was used to increase the quantity of the targeted DNA template and reduce errors that can be introduced during the PCR (Piggott et al., 2004).

163

The method consisted of carrying out three separate initial high-volume (38μl) PCRs containing non-labeled primers and 9μl of the template DNA. Each of the initial PCRs contained multiplexes of primers with four to nine loci within each (See Table B-1). Then

2 to 3.5μl of the PCR product from the initial amplification were used as the template in a second step PCR that ran in multiplex. PCR mixtures for each step of the pre- amplification PCR method are described in Table 4-1. The PCR thermocycling conditions were as follow: 15 minutes initial pre-denaturation at 95°C, followed by a touch-down procedure consisting of 1 minutes at 95°C, annealing for 1 minute (see

Table 4-1 for annealing temperature (Tm)), decreasing from 5°C above Tm to Tm-0.5°C during the first 10 cycles (with 0.5°C decremental steps in cycles 2 to 10) and then maintaining annealing at Tm for the remaining cycles (14 for the pre-amplification step and 24 cycles for the second step), and ending with an extension step at 72°C for 10 cycles. The annealing temperature of the pre-amplification step was that of the multiplex with the lowest annealing temperature (Table B-1). A total of 24 cycles and 34 cycles was performed for the pre-amplification and the second-step amplification, respectively.

Each fecal sample was genotyped for 13 microsatellite loci: Tma-E01, Tma-E04,

Tma-E14, Tma-K01, Tma-SC05, Tma-SC13 (Pause et al., 2007), Tma-FWC01, Tma-

FWC04, Tma-FWC08, Tma-FWC09, Tma-FWC15, Tma-FWC17, andTma-FWC18

(Tringali et al., 2008b). The 15 markers were divided into six multiplex groups containing three markers each except for two (M09 and M11) that contained only two markers

(Table B-1). Each sample was also genotyped for three sex-specific loci: TML-SMCX2,

TML-SMCY (Tringali et al., 2008a), and DSRY (Mchale et al., 2008). In order to minimize genotyping errors due to sample contamination, most samples were extracted

164

twice. Each of the extracts was used to run two to three replicates of each of the three pre-amplification PCRs (A, B, C). Each of pre-amplification replicates was then used as the template of one to two second-step multiplex PCR replicates. Thus, each sample was amplified at each locus in four to seven second-step PCR replicates (Figure B-3).

All PCR products were sequenced on an ABI3730xl 96-capillary electrophoresis genetic analyzer (Applied Biosystems, Foster City, California) with the GeneScan500 size standard at the DNA Analysis Facility on Science Hill (Yale University, New Haven,

Connecticut). Fragment data were scored using GeneMarker, version 2.7.4 (Soft

Genetics, LLC, State College, Pennsylvania).

Positive and negative controls

The fecal DNA of a known Florida manatee (Snooty) was used as a positive control for the microsatellite and sex marker PCR amplification. The sample was collected in captivity on June 17th, 2013 and was preserved in 95% Ethanol and stored at the USGS, Wetland and Aquatic Research Center. We extracted DNA from Snooty fecal material using the same protocol described above. The microsatellite genotypes generated from Snooty fecal DNA in this study were matched with the genotype of the same animal generated in 2011 using DNA isolated its blood samples.

DNA isolated from a male African manatee tissue samples (MCAD1401) collected from a carcass in Cameroon by Keith-Diagne (unpublished data) was also used as a positive control. The microsatellite and sex genotypes of MCAD1401 were known and were obtained from Keith-Diagne database (unpublished data) to match the genotype generated using the current protocol. RNase-free water was used as a negative control.

165

Microsatellite Genotype Assignment

The genotypes of all replicates for each sample and each locus were combined to construct a consensus genotype. Samples that failed to amplify for more than 30% of the 13 microsatellite loci were excluded from the final data set. The consensus genotype was decided using the criteria as described in Hedmark & Ellegren (2006): A heterozygote was accepted only if each of the two alleles occurred at least twice across the replicates. A homozygote was accepted only if three or more of its replicates were unambiguously homozygotes. When a homozygote or heterozygote could not be confirmed, the sample was classified as ambiguous and was treated as missing data in the further analysis. These were samples showing a homozygote profile at only one or two replicates and failed to amplify in other samples or samples for which one or two replicates were homozygotes, and one was a heterozygote. Samples that had third alleles (false alleles) in more than one replicate were also classified as ambiguous.

Allelic dropout was confirmed when one or more replicates for a sample were homozygous while others were heterozygous. False alleles were calculated as the number of peaks (bearing the same morphology as that of alleles at a locus) having an allele size within the allele size ranges in the population for that locus but were not confirmed in the consensus genotype.

Amplification Success and Genotyping Error Assessment

Amplification success was calculated by tallying the total number of positive PCR replicates across samples and loci divided by the total number of PCR replicates attempted. We considered positive PCR one that generated a product that yielded at least one visible and scorable peak within the allele-size range of the given locus.

Scorable peaks were those falling within the allele range for the locus, with a minimum

166

height of 100, and bearing the characteristic morphology of alleles at that locus. The allelic dropout rate was calculated as the proportion of the total number of positive replicates at heterozygous loci showing a homozygous profile while the consensus was heterozygous. Allelic dropout was estimated only for amplifications at loci for which a consensus genotype could be constructed.

Sex Identification

Sex was determined by coamplifying the TML-SMCX2, TML-SMCY (Tringali et al., 2008a), and DSRY (Mchale et al., 2008). Females were expected to exhibit a single band (TML-SMCX2) of approximately 86bp, while males exhibit two bands of approximately (TML-SMCY) 108bp and (DSRY) 127bp. The three markers were amplified simultaneously in the same multiplex PCR reaction. Each sample was PCR amplified for each of the three loci three times. Male sex was assigned to a sample when the SCMX2 locus amplified in at least two of the three replicates and that three or more of the combined six PCR replicates of the two male-specific markers were positive. Female sex was assigned to a sample when the SCMX2 locus amplified in at least two of the three replicates, and none (or just one) of the combined six PCR replicates of two male-specific markers were positive. Otherwise, the sex was recorded as undetermined.

The accuracy of our sex determination protocol was assessed by using three positive controls. Two of the positive controls were from a DNA template isolated from tissue samples from a known female and male African manatee. The other control positive was DNA template that we isolated from the feces of the famous captive male

Florida manatee, Snooty. The sex of these positive controls was confirmed using our protocol.

167

Statistical Analysis

The Two-way ANOVA was used to test for the significant difference in average

DNA yield and purity ratio among extraction methods and by fecal sample types.

Friedman’s test was used to test for the significant difference in the average HQ% for D- loop among extraction methods. The Student’s t-test was used to assess for the difference in DNA yield, A260/A280, HQ%, PCR success rate, and allelic dropout between samples collected from the lake and those from the river. All tests were run on

R-Studio software with the significant value threshold set at 0.05.

Results

Comparison of the Efficacy of DNA Isolation Methods

The 2CTAB/PCI protocol yielded a significantly (P<0001) higher DNA concentration (average = 234.1 ng/µl) compared to the NucleoSpin® Soil Kit

(NucleoSpin) and QIAmp Fast DNA Stool Mini Kit (QIAmp) protocols (Table 4-2, Figure

4-1). There was a significant difference in the DNA purity level among extraction methods with QIAmp bearing the value further away from the ideal value 1.8 (average

A260/280=2.28), whereas the 2CTAB/PCI provided the closest A260/280 value to 1.8

(Table 4-2). There was no significant difference in average DNA concentration among the three fecal sample groups (P=0.5); however, non-fibrous African manatee feces had the poorest average DNA purity (2.11) and lowest DNA yield (82.7ng/µl) (Table 4-3).

The non-fibrous Florida manatee samples yielded the highest average DNA concentration (97.1ng/µl).

There was no significant difference in the HQ% of trimmed D-loop sequence among extraction methods; however, QIAmp average HQ% value was the highest

(95.4%), and its values were more consistent across sample replicates as indicated by

168

the lowest standard error (Table 4-2). The NucleoSpin yielded the highest PCR success

(100%); however, the quality of the sequence yield was the lowest (80.04%) compared to those obtained from the other extraction methods. The 2CTAB/PCI method yielded the lowest PCR success (80% of the samples amplified). In summary, the QIAmp protocol was selected as the most robust isolation type as it yielded the highest D-loop

HQ% and a high amplification success rate of the D-loop.

Analysis of Full Data Set Using the QIAmp Fast DNA Stool Mini Kit

We applied the QIAmp protocol to assess our full data set of African manatee free-floating fecal samples (n=235). Due to a A260/280 value above the ideal 1.8 value, an inhibitor removal procedure was applied on a subset of 91 samples as described in the methods section. The mean of total DNA concentration of the 235 fecal samples extracted using the QIAmp protocol was 15.3ng/μl, and the values ranged from 0 to

101ng/µl. Most samples (55%) yielded DNA isolates with a total concentration between

5 and 15ng/µl, and only a few samples (2%) yielded a total DNA concentration greater than 50ng/µl (Figure 4-2). The second round of extraction yielded an even higher DNA concentration ranging between 13.25 and 112.5ng/μl with an average of 39.6ng/μl. The mean A260/280 value of the 91 samples improved from 2.1 to 1.8 (Figure 4-3) after the subset of 91 re-extracted isolates were further purified using the Zymo OneStep

Inhibitor Removal Kit.

Mitochondrial DNA PCR Success

From the 110 fecal DNA isolates selected for the mtDNA PCR-amplification, 101

(92%) yielded a sequence that successfully aligned with the reference sequence; the remaining nine samples yielded poor-quality sequences with ambiguous chromatogram peaks. The trimmed-to-reference sequence (410pb) of the 101 samples had a high

169

HQ% value averaging around 94.6%. The quality of the chromatograms was clear and very discriminative between base peaks (Figure 4-4). Only 12 bases (out of 41410) among all the sequences needed manual editing of the sequence.

Microsatellite DNA PCR Success, Allelic Dropout, and False Allele Amplification

Across the 13 loci that were tested, two loci, Tma-K01 and Tma-E01, were excluded from the final data set because more than 30% of the samples failed to amplify at that locus. A total of 15 (13.6%) samples failed to amplify at more than four loci, and those were removed from the final data set to be used for the assessment of the amplification success and genotyping errors. An additional 20 (18.2%) samples that yielded a combination of ambiguous genotypes or amplification failure at more than four loci, were further removed from the final data set (75 samples) for the downstream genetic analysis. Successful PCRs yielded high quality and scorable chromatograms

(Figure 4-5). Across 110 samples, we ran a total of 5528 microsatellite PCR reactions on the fecal DNA of the African manatee which yielded high levels of PCR success

(80%) across loci, replicates, and samples. The PCR success across loci ranged between 61% for locus Tma-FWC17 and 89% for locus Tma-FWC08. The allelic dropout was moderate, with an overall average of 24% and ranged across loci from

11% (Tma-FWC04) and 36% (Tma-FWC09). The number of false alleles was generally low, with a total of 57 allelic dropouts across all loci which ranged from 0 (loci Tma-

FWC04 and Tma-FWC15) and 16 (Tma-FWC09) (Table 4-4). On average, the PCR success rate only slightly increased, and allelic dropout slowly decreased with the increase of DNA extract concentration (Figure 4-6).

170

Sex-specific Gene Amplification Success

All but three samples yielded a positive PCR amplification at locus Tml-SMCX2 for all three replicates. We were able to assign a gender to 95 out of the 110 fecal samples (86.4%). The sex genotypes from a total of 15 fecal samples were ambiguous and removed from further analyses. The amplification success of TML-SMCY and

DSRY for male determined as male were 74.4% and 68.1% respectively. The sex assignment for the three primers was consistent for the 95 samples.

Effect of Habitat on PCR Amplification

The average total DNA concentration and the amount of DNA yield per mg of fecal material were two-fold higher for fecal samples collected in the river (21.9 ng/µl and 4.5µg/mg) than in lakes (11.8ng/µl and 2.1µg/mg), P<0.0001 (Table 4-5). The DNA purity level and the HQ% of the control region sequences were not significantly different between the two habitat types. PCR success and allelic dropout rates were significantly different between habitats, with PCR success higher in the river (87.5% in the river versus 79.4% in lakes), and allelic dropout was two times lower for samples collected in the river (18.5% versus 29.1% in lakes) (Figure 4-7).

Discussion

Comparison of the Efficacy of the DNA Isolation Methods

Overall, the QIAmp protocol yielded the highest DNA quality as indicated by the average sequence HQ% (95.4%) and the number of PCR successes (93%). Although the 2CTAB/PCI method generated DNA quantities that were one order of magnitude higher than that yielded by QIAmp and Nucleospin, it provided the lowest number of

PCR success. This result corroborates that of the Espinosa et al. (2015) study which compared the efficacy of the 2CTAB/PCI and the Qiagen DNA Stool Kit (an older

171

version of QIAmp Fast DNA Stool Mini Kit used in this study) and found that the later yielded on average higher PCR amplification success of the Control region (77.08%), and microsatellite loci (80.56%) of the fecal DNA of guanaco (Lama guanicoe) compared to the former (29.17% and 68.06% respectively) when the Chelex resin treatment was not applied. The high total DNA quantity yielded with 2CTAB might be an indication that this protocol is not selective on the type of organism and therefore, it isolates DNA from other cellular organisms such as bacteria and plants in addition to the target manatee DNA. A higher concentration of the non-target DNA in the final isolate would increase competition for the PCR reagents at the detriment of the target

DNA during the PCR amplification. Indeed, when PCR amplifications were run using the initial DNA concentration of DNA extract isolated using the 2CTAB/PCI protocol, all the reactions failed and only worked best when this initial volume was diluted to 50ng/μl before being used in the PCR reactions.

A different result was obtained in the Vallet et al. (2008) study which also compared the efficacy of the 2CTAB/PCI and the Qiagen DNA Stool Kit in isolating fecal

DNA of two herbivorous mammals, the Barbary macaque (Macaca sylvanus) and the

Western lowland gorilla (Gorilla g. gorilla). The amplification success of a nuclear gene of samples extracted using the 2CTAB/PCI on average was higher (97.1% and 94.3%) than those extracted using the Qiagen DNA Stool Kit (54.3% and 65.7%) for both species respectively. The initial DNA concentrations of the extract were not reported for this study. The discordant result between the current study and that of Vallet et al.

(2008) could be due to the difference in statistical power, target species, and the version of DNA Stool Kit used. The latest QIAmp was used in this study while the older Qiagen

172

DNA Stool Mini Kit was used in Vallet et al. (2008) study. Qiagen discontinued the earlier kit and replaced it with the former which uses a novel liquid InhibitEX Buffer that is more efficient at depletion of PCR inhibitors compared to the inhibitor removal tablets of the older Qiagen DNA Stool Kit. Our results indicate that even with the new InhibtEX

Buffer of the QIAmp, the purity of the DNA was the lowest (as indicated by a high

A260/280=2.28) and the 2CTAB method appears to be more efficient at removing PCR inhibitors generating a A260/280 value closer to 1.8. This further reinforces the suggestion that the low amplification success with the 2CTAB method is more likely due to the higher proportion of the background DNA (non-target DNA) in the final DNA extract compared to our desired manatee DNA. Unlike terrestrial animals, African manatee feces are deposited in an aquatic environment where water favors the degradation of DNA and the other organic material of the feces, which therefore results in the proliferation of bacteria and maggots. Indeed, we found some African manatee fecal samples containing maggots when grossly examined during collection. It is very likely that the DNA extract using the 2CTAB method includes a higher proportion of the

DNA originating from non-target organisms. The QIAmp provides two selective options in its protocol that allow removal of either target bacterial DNA or human-like cells. This selectivity would likely decrease the amount of background DNA caused by bacteria and possible contamination.

Our results suggest that the QIAmp is more efficient at isolating better quality

DNA from manatee fecal samples. The QIAmp protocol has been used in several studies to isolate DNA from noninvasive fecal samples of wild species populations such as brown bears (De Barba & Waits, 2009; Skrbinšek et al., 2010), chimpanzees and

173

gorillas (Arandjelovic et al., 2009, 2011; Mitchell et al., 2015; Morin et al., 2001). It had also been used to successfully isolate mtDNA from the West Indian manatee (Muschett et al., 2009).

Fibrous vs non-fibrous fecal DNA extractions

Although the difference in DNA concentration yield by the three extraction methods was not significantly different, the fibrous part of the African manatee feces resulted in generation of DNA concentrations (90.2ng/μl) that were slightly higher than those obtained from the non-fibrous part of the feces (82.7ng/μl). Moreover, DNA from the fibrous part of the feces had a better purity as the A260/280 ratio was closer to the ideal spectrum value 1.8. This may suggest that the plant fragments in the feces retain a greater amount of sloughed epithelial cells of the manatee intestine than those cells that are washed off in the ethanol preservative. Further investigation should determine the proportion of true manatee DNA from the total DNA yield from each part of the feces. Florida manatee fecal samples yielded on average slightly higher DNA quantity

(97.1ng/μl) than the African manatee feces. The Florida manatee feces were collected during the winter season from brackish water; the lower temperature and higher salinity may have contributed to the preservation of the DNA (Eichmiller et al., 2016; Schulz &

Childers, 2011; Seymour et al., 2018). Conversely, the African manatee feces were collected in a tropical and freshwater habitat which is conducive to DNA degradation.

Analysis of full data set Using the QIAmp Fast DNA Stool Mini Kit

The QIAmp protocol was successful in isolating DNA from 235 African manatee feces. The average total DNA concentration of 2.9ng/mg was slightly higher than that reported by Muschett et al. (2009) for the Antillean manatee feces collected in Panama

(13ng/μl) using an older version of the QIAmp. Díaz-Ferguson and collaborators (2017)

174

obtained lower total DNA concentration (5-10ng/μl) for Antillean manatee fecal samples collected in some different locations than those reported in Muschett et al. (2009), likely due to the older age of the sample.

The purity of DNA samples is usually measured using the A260/280 absorbance and should be interpreted as an indication of purity rather than absolute purity of our desired manatee DNA. Theoretically, pure DNA would have an A260/A280 ratio of about 1.8 (Matlock, 2015). The average A260/280 absorbance ratio of the DNA isolates in this study were four units above the generally accepted value. However, after applying the Zymo One Step Inhibitor Removal protocol to the re-extracted samples, we observed a left shift in the distribution of the A260/280 spectrum with an average of 1.8.

The higher value of the A260/280 ratio is not a concern; however, a test we conducted on four samples (Figure B-1) to assess the influence of the high A260/280 on microsatellite PCR success, indicated that samples with high purity ratio values (above

2.0) yielded PCR chromatogram peaks of a poor resolution (more stutters) and increased allelic dropout. After purifying the same samples with the Zymo One Step

Inhibitor Removal protocol, the quality of allele peaks was improved, and allelic dropout reduced. This result suggests that the Zymo One-Step PCR Inhibitor Removal Kit may be efficient at improving the purity of the fecal DNA and making them more optimal for downstream PCR applications.

Mitochondrial DNA PCR Amplification Assessment

This is the first mtDNA genetic analysis of the African manatee using fecal DNA samples; we successfully amplified a 410bp control region of the mtDNA isolated from the African manatee samples. Overall, the mtDNA amplification success in this study was high (92%). It was even higher than the PCR amplification success reported in the

175

Keith-Diagne (2014) study (32/49, 65.3%) that used DNA from tissue and blood samples from opportunistically collected African manatee carcasses. The author attributed the PCR failures to the highly degraded condition of some samples. In regard of this difference in PCR success, African manatee feces may be a cost-effective alternative DNA source, especially for the mtDNA genetic applications. Manatee carcasses in Africa are rare and difficult to acquire. It took more than six years for Dr.

Keith-Diagne to collect 65 opportunistic manatee tissue and blood samples across seven countries throughout much of the species distributional range. In this study, 65 fecal samples were collected in only four months. Moreover, collecting and preserving fecal samples requires only basic logistics and skills compared to tissue and blood sampling.

The disadvantage of using fecal DNA in mitochondrial studies is the risk of duplicate sampling, i.e., unknowingly collecting fecal samples from the same individual.

However, this risk could be greatly minimized by collecting on the same day a set of two to three fresh fecal sample from multiple locations while ensuring that the locations are geographically far apart to avoid an overlap of an individual defecating in two locations.

The two or three samples collected from the same location could be used alternatively in case of the failure of one sample producing useable DNA. The risk of duplicate samples could also be eliminated by genotyping a set of microsatellite DNA markers for individual identification. However, this method can be very expensive and labor intensive as it requires at least four PCRs repetitions. Another drawback is that tissue and blood samples provide more biological information on the donor than fecal samples

176

do. Therefore, tissue or blood should always be prioritized when accessible and fecal samples should only be used as an alternative or complementary sample.

The PCR amplification success of the mtDNA in this study was higher than those reported in the Muschett et al. (2009) and Díaz-Ferguson (2017) studies (79% and 42% respectively) using the same marker (control region). The higher performance of our protocol may be attributed to the various optimization steps we applied during the extraction process including determining which between the fibrous or the non-fibrous part of the fecal material yielded higher quality DNA, using a more advanced version of the QIAmp DNA Stool Kit, and further purifying the fecal DNA with the Zymo OneStep

PCR Inhibitor Removal Kit prior to the PCR amplification. The high quality of the DNA sequences (HQ%=95%) generated has further demonstrated the high performance and the reliability of our isolation protocol (Figure 4-4).

Microsatellite DNA PCR Success, Allelic Dropout and False Allele Amplification

In this study, we successfully amplified 13 manatee microsatellite markers and three sex markers from African manatee fecal samples. To the best of our knowledge, these are the first results of microsatellite genetic analysis of manatees using DNA from free-floating fecal samples. Previous genetic analysis of manatees using fecal DNA focused only on mitochondrial markers. Environmental nuclear DNA is more challenging to isolate compared to the mtDNA. There are only two copies of the nuclear genome per somatic cell, whereas in animals each of the several mitochondria per cell contain two to 10 copies of mtDNA for a total of 220 to 1720 mtDNA molecules per cell (Robin &

Wong, 1988; Wiesner et al.,1992). The shorter size (about 16000pb) and the circular structure of the mtDNA provide more stability to the molecule when compared to the linear nuclear DNA, which is prone to rapid fragmentation. Finally, mitochondrial

177

organelle membranes provide higher protection to the mtDNA against degradation

(Foran, 2006). Thus, microsatellite DNA is more prone to degradation and amplification failure during PCR reactions; making it difficult to successfully isolate and amplify samples.

The overall PCR amplification success rate for microsatellite markers in this study was high (80%) and comparable with success rates obtained from similar studies that used fecal DNA of other species including the brown bear, wolverine, chimpanzees, and the right whale (Table B-2). The allelic dropout across all replicates was however very high compared to values obtained in studies that used non-invasive DNA but was similar to the values reported by Morin et al. (2001) and Gillett and collaborators (2008) for the Tai chimpanzee (24%) and the right whale (27%) respectively.

Figure 4-6 indicates that the initial total DNA concentration above 30ng/μl DNA had only a slightly positive effect on the PCR amplification success and the allelic dropout. For example, samples with initial total DNA concentrations around 30 and

60ng/μl yielded a similar PCR amplification and allelic dropout rate. This result may indicate that total DNA concentration does not always reflect the quality and the quantity of the target DNA; thus, emphasizing the importance of quantifying the proportion of true target DNA to select the best sample for the PCR amplifications. This quantification could be done by running gel electrophoresis on the product of the mtDNA PCR amplification and comparing band brightness with that of reference samples with known

DNA template concentration (Gillett et al., 2008). Also, when the concentration of the target DNA template is known, it is possible to reduce the number of amplification replicates for samples that show high target-DNA concentrations and maximize the

178

replicate on samples with lower concentrations (Arandjelovic et al., 2009; Morin et al.,

2001). In the absence of the quantification of the target DNA concentration, we recommend prioritizing fecal DNA with an initial total DNA concentration of at least

30ng/μl to optimize the microsatellite amplification success and minimize error rates.

The concern of the high allelic dropout was addressed by repeating each extraction twice to reduce dropout due to stochastic fecal subsampling and potential contamination of the feces by the environmental DNA from a different individual. Each extract duplicate was pre-amplified at least twice in a high volume PCR reaction in which 9μl of the template was used to boost the number of copies of the target DNA

(Piggott et al., 2004) and then employed the resulting PCR product of the preamplification as the template for the final standard PCR. Thus, each sample was processed with PCR between four and eight times for each microsatellite marker, as recommended by Taberlet and collaborators (1996). This robust approach allowed us to address the concern of allelic dropout by constructing a consensus genotype from the independent genotype replicates. Thus, the utility of the two-step multiplex protocol at increasing amplification success and reducing allelic dropout in PCR using poor quality

DNA (Arandjelovic et al., 2009; Piggott et al., 2004) is further demonstrated in this study.

The genotypes that resulted from the PCR replicates of the three sex markers

(TML-SMCX2, TML-SMCY, and DSRY) also allowed us to construct a consensus sex genotype and assign sex to about 86% of the fecal samples analyzed. It is important to note that Tringali et al. (2008a) developed TML-SMCX2, TML-SMCY, and DBY7 primers to be used simultaneously to determine the sex of manatees. However, during

179

the testing phase of our study, amplification of the DBY7 failed for all the samples except for the positive controls. We suspected that the amplification failure of the DBY7 was due to a mutation at the flanking region of the DBY7 locus in the African manatee.

We also suspected that the larger size of the sequence of DBY7 (331bp) marker to be the cause of the amplification failure. Because fecal DNA are low quantity and quality nuclear DNA, amplifying a DNA segment of that size could be more challenging. In contrast to the DBY7, SMCX2 locus of the X chromosome, which has a shorter sequence length of only 84bp, amplified successfully in all samples. This result suggested that the segment-length hypothesis was likely the cause of the failure of the

DBY7 locus. Therefore, we replaced the DBY7 locus with a shorter Y-specific (DSRY,

153bp) marker developed by McHale et al. (2008), which successfully amplified in most male samples.

Effect of Habitat on PCR Amplification

Habitat type has a strong influence on the quantity of the total DNA yield, as indicated by the result of this study. Fecal samples from the river yielded total DNA quantities that were double of that yielded from lakes. Moreover, the average PCR success rate on samples collected in the river was significantly higher than those from the lake (Table 4-5, Figure 4-7) and the former yielded an average allelic dropout rate two times lower than the latter. Such a difference in DNA quantity and PCR success between two freshwater habitats that are less than 30km apart was not expected. An explanation of this may be found in the difference of the hydrodynamic characteristic of the two habitats. In lakes, water is more stagnant, and manatee boluses floating at the surface of the water are undisturbed and may persist for several days. On the contrary, the current in rivers does not now allow for feces to stay in one general area as in the

180

lake system. Moreover, the mechanical actions of the current would quickly disintegrate the feces. Therefore, fecal samples found in the river are more likely fresh and would yield greater quantity DNA, whereas feces from the lake can be old and would yield less

DNA. Indeed, during our manatee fecal survey, the frequency of manatee feces encounter was at least ten times higher in the lake than in the river.

Another possible reason for the difference in DNA quality and quantity between the two habitat types could be the different microbial components and activities between the two habitat types. Barnes and co-authors (2014), recently conducted a literature review and experiments to assess the effect of indicators of microbial communities on eDNA degradation rate of Common Carp (Cyprinus carpio). Those indicators included biochemical oxygen demand (BOD) used as a proxy for community metabolism, the concentration of chlorophyll a (Chl) used as a proxy of primary production, and total eDNA used as a proxy for microbial density and pH. The authors surprisingly found that eDNA degradation rates decreased with increased BOD, Chl and total eDNA. In Chapter 2, we demonstrated that Chl and water turbidity in Lake Ossa followed an increasing gradient from the outlet (where the adjacent Sanaga River connects with the lake) to the apex of the lake; which indicated that Chl was higher in

River Sanaga than in the lake. Therefore, our results corroborate with Barnes observation as Sanaga River with a higher Chl showed lower DNA degradation than fecal DNA collected from lakes Ossa and Tissongo.

Conclusion

This study demonstrated that manatee free-floating fecal samples can be used as a reliable source of DNA for both mtDNA and microsatellite genetic applications.

DNA from fecal samples can be isolated using the 2CTAB/PCI, the NucleoSpin, and the

181

QIAmp protocols; however, the latter appears to be more efficient. Because manatee fecal content is dominated by plant fragments, which are the source of various PCR inhibitors such as tannin, polysaccharides, and polyphenols (Schrader et al., 2012) these inhibitors are likely to be coextracted with the target DNA. Conducting an extra purification step may significantly improve the quality of the DNA and the success of the downstream reactions. For the first time, microsatellite DNA was successfully isolated from manatee environmental feces and used for genotypic profiling.

Successful amplification of nuclear loci from low quantity and quality DNA can be very challenging. However, the two-step multiplex protocol has allowed us to amplify microsatellite loci with high PCR amplification success. The replication of the PCR reactions has highly contributed to correcting for the high allelic dropout by allowing the construction of a consensus genotype. The consensus genotype can be reliably used in downstream genetic analysis such as individual identification, fine-scale population structure, size and connectivity, kinship, and more; while mtDNA data may be reliably used in phylogenetic analysis, assessing genetic diversity, evolution, population connectivity, and structure at a larger scale (Bonde et al., 2008). The protocol developed in this study allowed for sex determination using non-invasive manatee fecal

DNA for the first time. Data on the species sex can be crucial to determining population sex ratio and sex-biased dispersal and habitat use.

Although difficult to locate, fecal samples from the river remained likely fresh and yielded higher DNA quantity than fecal samples obtained from the lake. Fecal sample encounter rate in the river could be improved by using scent detection dogs trained to detect manatee feces rather than opportunistic visual detection. Scent dogs have been

182

successfully used to detect right whale feces in the open ocean. Rolland and collaborators (2006) reported that fecal sample collection rates were more than four- times higher when using a scent detection dog than when using the opportunistic visual detection method.

It is important to note that all the primers used in this study were designed for the

West Indian manatee. Although this species and the African manatee are related, a mutation or single nucleotide polymorphism (SNP) or homoplasy may occur at the priming region of some loci, making them less sensitive and complementary to the designed primers. Such a mutation might explain why some markers like TML-K01 and

TML-E01 failed to amplify in most of the African manatee samples. Designing new microsatellite and sex-specific primers for the African manatee genome may further improve the amplification success and relieve allelic dropout.

The conclusions of this study may be applicable to the other sirenian species, including the West Indian manatee, the Amazonian manatee, and the dugong. Since the advent of molecular ecology, genetic analysis using fecal DNA has only been used scantly in the marine mammal field and in only two studies to date in sirenians. Although the West Indian manatee and more specifically the Florida manatee have been extensively studied using population genetic tools and DNA obtained from tissues and blood (Davis, 2014; Garcia-Rodriguez, 2000; Hunter et al., 2010a; Hunter et al., 2012;

Nourisson et al., 2011; Vianna et al., 2006), fecal DNA may be a useful alternative to the more invasively acquired DNA samples when live capture is not possible. Manatee fecal DNA may be used to determine site fidelity especially in areas where poor water transparency does not allow for photo-identification. This could be done by collecting

183

fecal samples from the same locations and gradually building a catalog of their microsatellite genetic profile. The microsatellite genotype data generated from this study

(Chapter 5) revealed a female African manatee whose feces were collected from the same location multiple times within a month. Manatee fecal DNA may also help estimate the population size of manatees in areas where visibility does not allow useful aerial synoptic survey techniques. Lukacs and Burnham (2005) provided a comprehensive review of how noninvasive genetic sampling can be used in capture- recapture methods to estimate the population size of some species.

The African manatee is the least known of all sirenian species. Progress on the scientific knowledge about the species has been hampered by four major limitations including skills, logistics, funding, and the cryptic nature of the species. As a consequence, there is only one comprehensive genetic study on the species (Keith-

Diagne, 2014). If the skill limitation is likely to be addressed in the next few decades with the emergence of local researchers dedicated to African manatee studies, the three other limitations are however, not expected to be solved soon. All the distribution areas of the African manatee encompass low-income countries where wildlife research is not likely to become a priority in the near future. The cost of acquisition of marine research logistics is higher than most local research institutions can afford. The protocol developed in this study opens a new window that can assist in addressing the issues of logistics, funding, and species accessibility.

Acquiring manatee fecal samples can be very inexpensive and easy as it only requires a locally available boat and a hand net to scoop floating feces at the surface of the water. It does require some basic skills for sterile collection of fecal samples.

184

Preserving the samples can also be very affordable locally. In Cameroon, for instance, the 95% ethanol used to preserve the fecal samples is available for sale in biomedical stores at an affordable price. Moreover, this protocol allows the ability to sample individuals without having to see or touch them, which addresses the issue of the cryptic nature of the species. More interestingly, the feces collected may be used for several other research applications including diet, hormonal analysis, toxicology, and fecal microbial assessment (Burgess et al., 2012; Frank, 2015; Hurst & Beck, 1988;

Tsukinowa et al., 2008; Wyrosdick et al, 2018)

The other advantage of using fecal samples is that unlike tissue samples, the international shipping of the samples at this time does not require a CITES permit; which can be very difficult and expensive to obtain, often taking a very long time to acquire. Additionally, genetic analysis can also be done in local in-country laboratories if they implement a strong lab policy against contamination and provide deep freezing preservation of the DNA. The isolation and PCR amplification from fecal samples may not require an advanced laboratory setting when using commercial extraction kits like the QIAmp or Nucleospin kits which uses low-level bio-hazardous reagents. The PCR amplification may be easily achieved locally, if there is a thermocycler and stable source of electricity. Sequencing the PCR product might be the step that would require international shipping. However, shipping PCR products from Africa may be easier than shipping tissue samples and does not require a CITES permit. The amplification of the mtDNA from feces is relatively easy and by far cheaper than microsatellite markers; this is because the former does not require multiple PCR replicates and usually only involves two markers for the control region and cytochrome b. The sex-markers will also

185

be relatively easy and cheap to analyze as they involve only three markers that can be simultaneously run in a multiplex and require a fewer number of PCR replications compared to the microsatellite markers. The product of the multiplexed sex marker PCR could be also visualized on a gel, however, this approach may lead to misassignment of sex, because fecal DNA PCR often generates non-specific amplifications. Thus, sequencing the PCR product on an ABI capillary electrophoresis genetic analyzer and visualizing it using an appropriate software is the best option to determine amplified segments correspond to the target sex-specific gene. Therefore, future genetic research on the African manatee should consider focusing first on mtDNA and sex-marker analysis before moving into microsatellite analysis when more funding is available.

Several unanswered research questions can be addressed using the mitochondrial and sex marker information from the feces of the African manatee, and progressively bring light to the understanding of this poorly known species. There are still many countries where African manatee genetics are yet to be explored. Even in countries where genetic samples have been collected, there are still many regional or local habitats that have not yet been explored genetically. The use of fecal DNA might help increase the spatial resolution of the genetic knowledge of the African manatee across its entire home range.

186

Table 4-1. Nuclear microsatellite PCR reaction mixtures used for the pre-amplification PCR and the second multiplexed PCR of African manatees. Method modified from Piggott et al. (2004). Multiplex PCR: first step Multiplex PCR: second PCR mixture (pre-amplification) step 2μl to 3.5μl (of pre- DNA template 9μl amplification product) 2x Type-it Multiplex PCR Master Mix 19μl 6.7μl Labeled primer with variable concentrations 25uM each primer 0.01uM of each primers (unlabeled) (see Table B-1)

Rnase-free water Variable Variable That of the primer with the lowest Tm (Table Tm of the multiplex (Table Annealing temperature B-1) B-1)

Start annealing temperature 5°C above lowest Tm and decrease by 0.5°C/cycle for Touch-down 10 cycles and at the lowest Tm for 14cycles Reaction volume 38μl 13.4μl 34 (including 10 touch- Number of cycles 25 (including 10 touch-down cycles) down cycles)

Table 4-2. Comparison of extraction method on African and Florida manatee fecal DNA quality and quantity: HQ%=high quality sequence percentage value, TNS= Total number of sub-samples. Amplification success were samples that yielded control region mitochondrial sequences that could be completely aligned with the 410bp reference sequence. Each of the five Florida manatee feces and the fibrous and non-fibrous parts of five African manatee feces were subsampled in three replicates of 300mg fecal material for each of the three extraction methods. Extraction methods included the 2CTAB/PCI by Vallet et al. (2008), NucleoSpin® Soil kit (Macherey-Nagel) and the QIAmp Fast DNA Stool Mini kit (QIAGEN) protocols. Average DNA Average DNA Number of Average D- concentration purity amplification loop HQ% Extraction method TNS (ng/μl ) (A260/280) success 2CTAB/PCI 15 234.1 ± 16.9 1.87 ± 0.03 12/15 89.9 ± 5.8 NucleoSpin Soil 15 17.3 ± 1.6 1.82 ± 0.01 15/15 80.4 ± 9.6 QIAmp 15 18.6 ± 1.7 2.28 ± 0.05 14/15 95.4 ± 3.5

187

Table 4-3. Comparison of DNA quality and quantity by fecal sample types: fibrous and non-fibrous parts of African manatee feces and non-fibrous feces of the Florida manatee, averaged estimated DNA concentration, and average estimated DNA purity. Ts = Trichechus senegalensis, Tml = Trichechus manatus latirostris, TNS= Total number of sub-samples Fecal Average DNA Average DNA texture concentration purity Species type TNS (ng/μl) (A260/280) Ts Fibrous 15 90.2 ± 16.8 1.94 ± 0.03 Ts Non-fibrous 15 82.7 ± 16.6 2.11* ± 0.06 Tml Non-fibrous 15 97.1 ± 20.9 1.92 ± 0.02 * P-value purity 0.0001

188

Table 4-4. Number of PCR amplifications, PCR success, and genotyping errors (allelic dropout and false alleles) across 11 microsatellite polymorphic markers used in this study. L01 L18 L9 L17 L4 L8 E14 SC13 E04 L15 SC05 Total Number of PCRs attempted 507 520 602 447 447 401 520 520 508 441 615 5528 Number of amplification success 390 444 483 273 353 355 405 443 409 354 524 4433 PCR amplification success rate 77% 85% 80% 61% 79% 89% 78% 85% 81% 80% 85% 80% Number of amplification success at heterozygous loci 150 224 264 168 301 289 317 221 166 144 300 2544 Number of allelic dropouts 25 68 95 52 32 49 65 62 44 42 72 606 Allelic dropout rate 17% 30% 36% 31% 11% 17% 21% 28% 27% 29% 24% 24% Number of false alleles 6 4 16 4 2 6 2 6 0 0 11 57 *Microsatellite name corresponding to each locus code are found in Table B-1

Table 4-5. Comparison of averages in total fecal DNA concentrations, amount of total DNA yield per mg of fecal material, DNA purity, high quality sequence percentage (HQ%), PCR success and allelic dropout between African manatee fecal samples collected from a lake and a river habitat in Cameroon. Average Total Average DNA yield Average PCR Average Allelic Habitat DNA (ng/μl) per mg of feces A260/280 HQ% value success rate dropout Lakes 11.8 ± 0.6 2.1 ± 0.1 2.3 ± 0.04 93.53 ± 1.6 79.4±1.3 29.1 ± 2.3 (n) (152) (152) (152) (70) (63) (63) River 21.9 ± 1.9 4.5 ± 0.4 2.2 ± 0.03 97.0 ± 1.12 87.5 ± 1.5 18.5 ± 2.6 (n) (n=82) (82) (82) (31) (32) (32) P-value <0.0001 <0.0001 0.38 0.2 0.0002 0.006

189

Figure 4-1. Box plots depicting DNA concentration yields for each extraction method based on the 135 African manatee fecal subsamples. Each of the five Florida manatee feces and the fibrous and non-fibrous parts of five African manatee feces were subsampled in three replicates of 300mg fecal material for each of the three extraction methods. Extraction methods included the 2CTAB/PCI by Vallet et al. (2008), NucleoSpin® Soil kit (Macherey-Nagel) and the QIAmp Fast DNA Stool Mini kit (QIAGEN) protocols

Figure 4-2. Frequency distribution of the total DNA concentrations of 235 African manatee free-floating fecal samples collected in the downstream of the Sanaga River watershed between May 2016 and August 2017.

190

Figure 4-3. Frequency distribution of A260/280 values. Data generated from 91 African manatee fecal DNA extracts measured on Epoch™ Microplate Spectrophotometer before and after applying the inhibitor removal using the Zymo OneStep PCR Inhibitor Removal Kit. The samples were collected in the downstream of Sanaga River watershed.

Figure 4-4. Example of chromatograms of the Control region mitochondrial DNA sequences generated from the DNA isolated from African manatee non- invasive fecal samples. The segments presented on the screenshot are site numbered 162-210 (of the 410bp sequence) when aligned with the reference haplotype sequence AY963895.1 published by Vianna et al. (2006). The chromatograms were generated in Geneious Version 11.1.5.

191

Figure 4-5. Example of chromatograms showing microsatellite allelic dropout, causing false homozygotes. Results of four independent genotyping replicates of one African manatee fecal DNA using the pre-amplification method at microsatellite locus Tma E14. This manatee is a heterozygote with alleles 238 and 248. While the first three PCRs detected both alleles, the last PCR detected only one of the alleles. The chromatograms were generated in GeneMarker, version 2.7.4.

Figure 4-6. Proportion of positive PCRs (black diamonds) and PCRs with allelic dropouts (hollow circles) plotted against the total DNA concentration of the sample. Each point corresponds to one African manatee fecal DNA extract. On average, the PCR success rate only slightly increased, and allelic dropout slowly decreased with the increase of DNA extract concentration.

192

Figure 4-7. Comparison of PCR success and allelic dropout rates by habitat type generated from 91 African manatee free-floating fecal samples collected in the downstream of the Sanaga River watershed.

193

CHAPTER 5 GENETIC DIVERSITY AND CONNECTIVITY OF THE AFRICAN MANATEE IN THE DOWNSTREAM OF THE SANAGA RIVER WATERSHED

Background

The population structure and migratory patterns of the African manatee within the

Downstream of the Sanaga River Watershed (DSRW) are poorly understood, and no previous study has been conducted. Powell (1996) suggested from fisher’s reports that manatees are not uncommon in the DSRW. He added that during the dry season, the water level in the Sanaga River drops drastically and manatees may migrate to deep pools that Lake Ossa provides during the low water periods. Although water level is shallow upstream during the dry season, the estuary and Lake Tissongo also remain relatively deep during the low-water season, and the influence of tide provide even more water depth (Takoukam pers. comm.); therefore, manatees might also take refuge in the

Sanaga Estuary and tributaries during the low water season. The traditional ecological knowledge study by Takoukam Kamla (2012) shows that fishers reported higher manatee observation frequency in the lakes (Lake Ossa, Tissongo) and the estuary during low water level periods, while manatee sighting upstream in the Sanaga River is higher during the rainy season. However, there is no existing scientific documentation to date on the population structure and movement patterns of the African manatee in the DRSW; this is mainly because of the elusiveness of the species and the high cost associated with relatedness and telemetry studies.

African manatee hunting and accidental catch have been abundantly documented, especially in the DSRW (Keith-Diagne, 2015; Mayaka et al., 2013;

Mayaka et al. 2015; Nishiwaki et al., 1982; Powell, 1996; Takoukam, 2012).

Unfortunately, the impact of these threats on population health and viability is poorly

194

understood. Because of the cryptic nature of the species, it is difficult to determine how many manatee individuals are currently in the system and what the population trends are. More importantly, the genetic make-up of the population is barely known. DNA is the raw and variable material that allows organisms to adapt to environmental challenges such as diseases and climate change (Frankham et al., 2002). The level of genetic diversity in a species reflects its resilience to environmental challenges. The level of diversity of the manatee in the DSRW is still unclear. It is worth addressing these knowledge gaps to assist with the long-term survival of the DSRW manatee population.

With the advent of molecular biology, scientists are now able to more easily and efficiently address research questions that are difficult or unanswerable using traditional ecological or morphological methods (Frankham et al., 2002). Conservation genetics provides an excellent example of the application of molecular biology to address conservation issues. Biologists now utilize genetic approaches in a variety of situations, including the use of genetic markers for forensic investigation in wildlife and endangered species, non-invasive estimation of population dynamics and diversity, taxonomic affiliation and population history, and more (Allendorf et al., 2007).

Various genetic markers are used in the field of conservation genetics to determine taxonomic status at different levels. Those markers include allozymes, mtDNA, microsatellites, single nucleotide polymorphisms (SNPs), gene expression, genomics, and chromosomes (Bonde et al., 2008; Frankham et al., 2002; Raza, Shoaib,

& Mubeen, 2016). Bonde and collaborators (2008) discussed the usefulness of each of these markers and provided guidelines on how they could be used to assist in the

195

conservation of manatees. The two commonly used markers to date in the manatee are mtDNA and microsatellite. Mitochondrial DNA is circular, double-stranded DNA located in the mitochondria and is responsible for cellular respiration in eukaryotic organisms.

Its inheritance is uniparental, meaning that unlike nuclear DNA that is inherited from both parents, mtDNA is only provided by the mother (Frankham et al., 2004). This inheritance pattern of the mtDNA, coupled with its relatively high mutation rate and variability, makes it a useful genetic material to characterize populations and trace evolutionary and taxonomic groups. The control region of the displacement loop and cytochrome-b are the mtDNA markers that have been consistently used to study the phylogenetic, phylogeography, and evolution of manatees (Cantanhede et al., 2004;

Garcia-Rodriguez et al., 1998; Hunter et al., 2010a; Hunter et al., 2012; Keith-Diagne,

2014; Luna, 2013; McClenaghan & O’Shea, 1988; McDonald, 2005; Vianna et al.,

2006). Compared to cytochrome-b, the control region is non-coding and as a result, typically exhibits a higher mutation rate and higher diversity than cytochrome-b, which makes it useful to determine variation at a lower hierarchical scale of genetic diversity

(Frankham et al., 2002). Whereas cytochrome-b has a slower mutation rate and is more useful for assessing diversity at the regional level and to determine the evolution process of the taxa (Bonde et al., 2008; Vianna et al., 2006)

Microsatellites or simple sequence repeats (SSRs) are neutral (non-coding) nuclear DNA that are inherited biparentally. This process is also called “fingerprinting.”

Microsatellites can also be used to define fine-scale population structure and well as estimate population size and connectivity. An allele here is defined by the number of repeats of a given motif, which are usually di-, tri- or tetra- nucleotides. Because they

196

are biparentally inherited, the genotype of a locus gene is reported as a pair of numbers that indicate the allele size at each of the two homologous loci. An individual is said to be homozygous when the allele size is the same for both locus site; otherwise, they are said to be heterozygotes. Because of their high mutation rate and codominance (Raza et al., 2016), microsatellites measure diversity even at the individual level by genotyping multiple loci that will generate a combination of alleles that can be viewed as a bar code that distinguishes between individuals.

The early genetic study of manatees was conducted by McClenaghan & O’Shea

(1988) using allozyme phenotypes. This study provided signals indicating a low genetic diversity in the species. It was not until 1993 and 1998 that mtDNA markers

(cytochrome-b and control region respectively ) were used for the first time to assess diversity in manatees (Florida manatee) (Bradley et al., 1993; Garcia-Rodriguez et al.,

1998). Overall, manatees exhibit high mtDNA control region haplotype diversity estimated at h=0.86, 0.88 and 0.91 for the West Indian, Amazonian, and African manatees, respectively (Keith-Diagne, 2014; Vianna et al., 2006). The nucleotide diversity was relatively low and estimated at π=0.039, 0.005, 0.0175 respectively. The

Florida manatee, a subspecies of the West Indian manatee, has the lowest mtDNA diversity exhibiting only a single control region haplotype diversity (Garcia-Rodriguez et al., 1998; Vianna et al., 2006).

Only one comprehensive mtDNA genetic study of the African manatee has been conducted to date (Keith-Diagne, 2014). The author used DNA isolates from 49 manatee tissue and one blood samples collected from seven different countries including Senegal, Guinea, Ghana, Ivory Coast, Mali, Cameroon, and Gabon. Keith-

197

Diagne identified 14 new control region and nine cytochrome-b haplotypes. Five control region and the three cytochrome-b haplotypes were previously identified (Vianna et al.,

2006). This high number of haplotypes make the African manatee the sirenian with the highest control region haplotype diversity (HD=0.909); such high diversity is crucial as it reflects the capacity of the species to continuously evolve, adapt and become resilient to the ever changing environment caused by factors such as climate change (Frankham et al., 2002). However, Keith-Diagne found only a low mtDNA nucleotide diversity

(π=0.0175) indicative of a possible reduction in effective population sizes.

The AMOVA and Pairwise ɸST differences test performed by Keith-Diagne suggested significant differentiation within and between regions (North and South regional groups in Africa) and also between coastal and inland manatee populations in

Senegal. The highest numbers of the control region (n=08) and cytochrome-b (n=05) haplotypes were found in Gabon; whereas in Cameroon, only two control regions (TS-

CR06 and TS-CR07) and two cytochrome-b (TS-CYTB03 and TS-CYTB04) haplotypes were identified from seven and four sequences, respectively. The sequences from

Cameroon were obtained from tissue samples collected from hunted manatees in the

DRSW and the Douala Estuary. All the control region and the cytochrome-b haplotypes found in Cameroon were endemic. The author suggested isolation of the Cameroon coastal population from the other coastal populations potentially due to the rough waves and the scarcity of food along the coast, which might prevent migration between coastal habitats or populations in that region. The maximum likelihood and the BEAST phylogenetic analyses of control region evolutionary trajectories placed the African manatee as a sister species of the West Indian manatee, with the Amazonian manatee

198

being the basal species (Keith-Diagne, 2014). This corroborated with the morphologically based phylogeny of sirenians by Domning (1994).

The use of nuclear genetic markers such as microsatellites in manatee genetic studies is very recent and was applied for the first time in Florida manatee in 2000

(Garcia-Rodriguez, 2000). This study developed the first eight nuclear DNA microsatellite markers for manatees. Ten others were later designed by Pause and colleagues (2007), 18 by Tringali and colleagues (2008b), and five others by Hunter and collaborators (2010b). Although these microsatellites have not been used as intensively as mtDNA markers in manatee genetic studies, they have, however, demonstrated their usefulness in the few fine-scale manatee genetic diversity studies published to date.

Microsatellite studies of the manatee included various aspects including population differentiation, population history, individual relatedness, determination of management units, hybridization and inbreeding among Amazonian and West Indian manatee populations (Davis, 2014; Hunter et al., 2010a; Hunter et al., 2012; Lima et al., 2019;

Nourisson et al., 2011; Tucker et al., 2012) . Overall, the Florida manatee subspecies

(He=0.48, Na=4.8) yielded similar microsatellite genetic diversity as those reported for

Belize (He=0.46, Na=3.4), Puerto-Rico (He=0.45, Na=3.9), and Mexico Chetumal Bay

(He=0.46, Na=3.0), Ascension Bay (He=0.43, Na=2.0), and the Gulf of Mexico

(He=0.41, Na=2.62) (Hunter et al., 2010a; Hunter et al., 2012; Nourisson et al., 2011).

Microsatellite genetic analysis using Wright's statistics (1949) has allowed authors to establish divergence between different manatee populations including Puerto Rico and

Florida (FST=0.16, RST=0.12, P<0.001, using 15 loci, Hunter et al., 2012), Belize and

Florida (FST=0.14, P<0.001, using 16 loci, Hunter et al., 2010b), Chetumal Bay and the

199

Gulf of Mexico in Mexico (FST=0.10, P<0.05, 13 loci, Nourisson et al., 2011). The study by Tucker and collaborators (2012) indicated a low level of differentiation between the four management units of the Florida manatee on the east and the west coasts (GST

=0.018, 11 loci, Tucker et al., 2012). The study also found that the Florida manatee is related (r=0.115) at the level equivalent to the first cousin (Davis, 2014); they also showed a considerable inbreeding coefficient (Fx=0.108). High level of inbreeding

(FIS=0.456, 0.282) has also been documented in the West Indian and Amazonian manatee populations within the Amazon basin respectively (Satizábal et al., 2012).

No microsatellite genetic study on the African manatee has been published to date. In 2014, Keith-Diagne (unpublished data) conducted the first microsatellite genetic analysis of the African manatee. Keith-Diagne used a panel of 34 microsatellites and three sex-specific primers to successfully genotype DNA isolated from 42 African manatee tissue samples collected from eight countries of the species distribution:

Senegal, Guinea, Mali, Niger, Ivory Coast, Benin, Cameroon, and Gabon. However, the results of this study have not been published to date. This study, for the first time, will investigate microsatellite marker diversity among African manatees using fecal DNA.

This current study aimed at contributing to the conservation status of the African manatee by generating key genetic information that will guide decision making and management of the species. The specific objectives of this study were to (1) assess the microsatellite and mtDNA genetic diversities of the African manatee in the DSRW, (2) determine the population structure of the species within the DSRW based on mtDNA and microsatellite DNA, and (3) measure the level of mtDNA connectivity between manatee populations in Cameroon and Gabon.

200

Methods

Study Area

The DSRW encompasses two protected areas in the Littoral Region of

Cameroon (Figure 5-1). The Douala-Edea National Park, formerly known as the Douala-

Edea Wildlife Reserve, was created in 1932 and covered about 160.000 ha of land and water (Blaikie & Simo, 2000). The park stretches along both sides of the lower reaches of the Sanaga River and 100km along the Atlantic coastline of Cameroon (Latitude 3°

14’ 3°50′N and longitude 9°34′-10°03’E, Feka et al., 2009). The park surface is covered by about 6.4% of mangrove dominated by Rhizophora racemosa. A tropical lowland equatorial forest dominates the rest of vegetation (80%). The mangrove is seriously threatened by deforestation (Feka et al., 2009). The local community utilizes the cut mangrove wood to smoke fish. Fishing is the dominant economic activity in the area.

The Lake Ossa Wildlife Reserve is a complex of lakes located at 13km from

Edea, Cameroon, between the 3°45’ and 3°52’N latitude, and 9°45’ and 10°4‘E longitude with approximately 300m elevation (Wirrmann and Elouga, 1998). The water surface is estimated to be 4000ha. The lake represents about 90% of the Lake Ossa

Fecal Sample Collection

The shorelines of four areas within the DSRW (Figure 3-1) were frequently visited between June 2015 and November 2017 in search of free-floating fecal samples.

These areas represented three habitat types: lakes (Lakes Tissongo and Ossa), the river (Sanaga River), and an estuary (Sanaga Estuary), as presented in Figures 3-1 and

201

5-1. Fecal samples were preserved in 50ml sterile tubes half-filled with 95% ethanol and stored at the U.S. Geological Survey laboratory in Gainesville, Florida. Of the 235 fecal samples extracted, 110 with a total DNA concentration higher than 10ng/μl were selected for PCR amplification of the mtDNA and nuclear DNA. The majority of 110 fecal samples were extracted twice.

DNA Purification

Before the DNA amplification, the OneStep™ PCR Inhibitor Removal Kit (Zymo

Research) was used to purify DNA extracts further, and then the A260/280 was remeasured on a NanoDrop spectrophotometer to assess the change in the DNA purity.

The Zymo OneStep inhibitor removal procedure allows efficient removal of contaminants from the DNA preparation that can inhibit the PCR reaction. More importantly, it is designed to remove polyphenolic compounds, humic acids, tannin contained in water, and plant fragments from the feces sample.

Mitochondrial DNA Amplification

The control region of the mtDNA of each of the 110 samples were amplified in a

25μl reaction volume containing 12.5μl AmpliTaq Gold™ 360 Master Mix (ThermoFisher

Scientific), 1μl of 360 GC Enhancer, 0.5μl (10uM) of each primer (forward and reverse),

8μl water and 2.5μl DNA. All PCR amplifications were carried out on a SimpliAmp thermal cycler (Applied Biosystems, Thermo Fisher Scientific Inc). The PCR conditions were as follows: Initial denaturation at 94˚C for 5 minutes, then 34 cycles of denaturation for 1 minute at 94˚C, annealing for 1 minute at 53˚C, follow by extension for 1 minute at 72˚C, and final extension for 10 minutes at 72°C. The amplification is completed with a 4°C hold.

202

Microsatellite and Sex Marker Loci Amplification

The two-step multiplex pre-amplification PCR method was used to amplify 15 microsatellite and three sex marker loci from the same 110 African manatee free- floating feces. The PCR reactions were performed as described by Piggott et al. (2004) with some modifications described herein. The two-step multiplex pre-amplification PCR method is used to increase the quantity of the targeted DNA template and reduce errors that can be introduced during PCR (give a few more refs). The method consisted of carrying out three separate initial high-volume (38μl) PCRs containing non-labeled primers and 9 µl of the template DNA. Each of the initial PCRs contained multiplexes of primers with four to nine loci within each (See Table B-1). Then 2.0 to 3.5μl of the PCR product from the initial amplification were used as the template in a second step PCR that ran in multiplex. PCR mixtures for each step of the pre-amplification PCR method are described in Table 4-1. The PCR thermocycling conditions were as follow: 15 minutes initial pre-denaturation at 95°C, followed by a touch-down procedure consisting of 1 minute at 95°C, annealing for 1 minute (see Table 4-1 for Tm), decreasing from 5°C above Tm to Tm-0.5°C during the first 10 cycles (with 0.5°C decremental steps in cycles

2 to 10) and then maintaining annealing at Tm for the remaining cycles (14 for the pre- amplification step and 24 cycles for the second step), and ending with an extension step at 72°C for 10 cycles. The annealing temperature of the pre-amplification step was that of the multiplex with the lowest annealing temperature (Appendix B, Table B-1). A total of 24 cycles and 34 cycles was performed for the pre-amplification and the second-step amplification, respectively.

Each fecal sample was genotyped for 13 microsatellite loci: Tma-E01, Tma-E04,

Tma-E14, Tma-K01, Tma-SC05, Tma-SC13 (Pause et al., 2007), Tma-FWC01, Tma-

203

FWC04, Tma-FWC08, Tma-FWC09, Tma-FWC15, Tma-FWC17, and Tma-FWC18

(Tringali et al., 2008b). The 15 markers were divided into six multiplex groups containing three markers each except for two (M09 and M11) that contained only two markers

(Appendix B, Table B-1). Each sample was also genotyped for three sex-specific loci:

TML-SMCX2, TML-SMCY (Tringali et al., 2008a), and DSRY (Mchale et al., 2008). In order to minimize genotyping errors due to sample contamination, most samples were extracted twice. Each of the extracts was used to run two to three replicates of each of the three pre-amplification PCRs (A, B, C). Each of pre-amplification replicates was then used as the template of one to two second-step multiplex PCR replicates. Thus, each sample was amplified at each locus in four to seven second-step PCR replicates

(Appendix B, Figure B-3).

All PCR products were sequenced on an ABI3730xl 96-capillary electrophoresis genetic analyzer (Applied Biosystems, Foster City, California) with the GeneScan500 size standard at DNA Analysis Facility on Science Hill (Yale University, New Haven,

Connecticut). Fragment data were scored using GeneMarker, version 2.7.4 (Soft

Genetics, LLC, State College, Pennsylvania).

Positive and Negative controls

The fecal DNA of a known Florida manatee (Snooty) was used as a positive control for the microsatellite and sex marker PCR amplification. The sample was collected from Snooty while in captivity on June 17th, 2013 and was preserved in 95%

Ethanol and stored at the U.S. Geological Survey laboratory in Gainesville, Florida. We extracted DNA from Snooty’s fecal material using the same protocol as described above. The microsatellite genotypes generated from the Snooty fecal DNA in this study were matched with the genotype of the same animal generated in 2011 using DNA

204

isolated from blood samples. The snooty genotypes generated from tissue DNA and from fecal DNA matched; the sex genotype using fecal DNA also confirmed the male gender of Snooty. This result indicates the reliability of our non-invasive fecal DNA genotyping approach.

Sex Identification

Sex was determined by coamplifying the TML-SMCX2, TML-SMCY (Tringali et al., 2008a), and DSRY (Mchale et al., 2008) primers. Females were expected to exhibit

205

a single band (TML-SMCX2) of approximately 86 base pairs (bp), while males exhibit two bands of approximately (TML-SMCY) 108bp and (DSRY) 127bp. The three markers were amplified simultaneously in the same multiplex PCR reaction. Each sample was

PCR amplified for each of the three loci three times. Male sex was assigned to a sample when the SCMX2 locus amplified in at least two of the three replicates and that three or more of the combined six PCR replicates of the two male-specific markers were positive. Female sex was assigned to a sample when the SCMX2 locus amplified in at least two of the three replicates, and none (or just one) of the combined six PCR replicates of two male-specific markers were positive. Otherwise, the sex was recorded as undetermined.

Sex was determined by coamplifying three primers. Females were expected to exhibit a single band (TML-SMCX2) of approximately 86bp, while males exhibit two bands of approximately (TML-SMCY) 108bp and (DSRY) 127bp. The three markers were amplified simultaneously in the same multiplex PCR reaction. Each sample was

206

Genetic Analysis

Mitochondrial analysis

The mtDNA control region sequences of the unique genotypes identified in this study were aligned with a reference sequence (AY963894.1, Vianna et al., 2006) using the MUSCLE algorithm (Edgar, 2004) in GENEIOUS V11.1. (Biomatters Ltd., Auckland,

New Zealand). Then the sequences were trimmed to 410bp and examined for error using the same software. The summary statistics including haplotype (HD), nucleotide

(π) and sequence (k) diversities, the number of polymorphic sites (S), and the number of haplotypes (h), were calculated using the DNAsp software V6.12 (Librado & Rozas,

2009). Tajima’s D and FU’s Fs neutrality tests (Fu, 1997; Tajima, 1989) were conducted using ARLEQUIN 3.5 (Excoffier & Lischer, 2010). These two tests use different approaches to determine whether a population is at drift-mutation equilibrium. While

Tajima’s D compares nucleotide diversity with segregating sites, Fu’s Fs, allow to detect population expansion and genetic hitchhiking by using the number of unique mutations to estimate the mutation rate. The statistical significance of both tests was set at

P<0.05.

The analysis of molecular variance (AMOVA) was computed using ARLEQUIN version 3.5 software (Excoffier & Lischer, 2010), and the tests of differentiation between manatees from the three locations within DRSW. Divergence between Cameroon and

Gabon populations were estimated using pairwise FST and ɸST statistics. The ɸST test assesses differentiation between populations by comparing haplotype frequencies and molecular distances, while the FST exact test is based on the distribution pattern of the haplotype frequencies. All tests were conducted using 10,000 permutations and a P- value <0.05. To account for the differences between transition and transversion rates,

207

the Tamura genetic distance model was used to estimate sequence distance (Tamura,

1992). The haplotypes of the manatee population in Gabon used for this comparison were those identified by Keith-Diagne (2014) and included a total of nine haplotypes from 21 individuals. The Cameroon haplotypes used in this study included those found by Keith-Diagne (2014) and new haplotypes identified in this study.

The new control region haplotype sequences found in this study were combined with the published (n=5, Vianna et al., 2006) and non-published sequences (n=18,

Keith-Diagne 2014) to update the species most-likely phylogenetic tree using MEGA software 7.0 (Tamura et al., 2013). The Trichechus manatus haplotype Tm M01 (Gen

Bank accession AY963856) published by Vianna et al. (2006) was used as the outgroup haplotype on which the tree was rooted. The maximum likelihood (based on the General

Time Reversible model with invariant sites and gamma shape parameters, GTR+I+G) was determined by Keith-Diagne (2014) to be the best criterion of the evolution model explaining the history substitution of the African manatee control region haplotypes.

Therefore, the later model was used to build the most-likely phylogenetic tree using a bootstrap analysis of 1,000 replicates.

Discrimination of individuals

In order to ensure that the microsatellite markers were variable enough to discriminate individuals (Waits et al., 2001), we calculated the unbiased probability of identity (PID) using GENECAP (Wilberg & Dreher, 2004). PID is a measure of the probability that two individuals randomly picked from a population have the same genotypic profile at the studied loci (Paetkau & Strobeck, 1994), while P(ID)sib is a probability of identity of siblings. The shadow effect, which is the probability that two genotypes are identical by chance among a given number of samples was estimated

208

using GenAlEx v6.41) (Hunter et al., 2010b). The software GenAlEx v6.41 was used to identify matching genotypes as well as genotypes that mismatched at only one or two loci; Those genotypes were excluded from the final data as they were likely duplicate samples from the same individual.

Microsatellite DNA analysis

Evidence of null alleles and allele stutter in loci was examined using MICRO-

CHECKER version 2.2 (Van Oosterhout et al., 2004). Departures from Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium (LD) were assessed in GENEPOP version

4.7 (Raymond & Rousset, 1995) using the Markov chain method with the following parameters: dememorization 1,000, batches 100, iterations per batch 1,000). The LD test checks for non-random associations between alleles of the different loci assessed.

The P-values were adjusted by applying the Bonferroni sequential correction for the multiple comparison tests (Rice, 1989). The average degree of relatedness (R) within the overall samples was estimated in GenAlEx 6.5 using the Queller & Goodnight

(1989) mean estimator. We used this estimator because it corrects for bias due to small sample sizes (as it is the case in this study); it also generates relatedness value for a single group of individuals. The level of genetic polymorphism at the nuclear level, including observed (Ho) and expected heterozygosity (He), average number of alleles per locus (Na), average number of effective alleles (Ne), and the polymorphism information content (PIC) were estimated using GenAlEx 6.5 (Peakall & Smouse, 2006).

The software was also used to estimate FIS, the inbreeding coefficient.

An AMOVA applied to the matrix of pairwise genetic dissimilarity between samples in GenAlEx 6.5 was used to compute the global FST and RST for the DSRW.

These two parameters estimate the level of differentiation within the population, based

209

on different mutation models; FST uses the infinite alleles model (IAM), while RST uses the stepwise mutation model (SMM). The same matrix was exported in R-STUDIO to conduct a Principal Component Analysis (PCA) in order to graphically visualize the genetic distance and relatedness among manatees in the three DSRW locations, including Lake Ossa (n=35), Sanaga River (n=9), and Lake Tissongo (n=5). The eigenvalue analysis was used to group the variation between samples into a reduced number of dimensions. The program STRUCTURE version 2.3 (Pritchard et al., 2010) was used to identify eventual subpopulation structuring within the DSRW without a priori population assignment. The genotype data were simulated using the admixture model, and the value K was set between 1 and 7 with a burn-in period of 100,000 iterations and

100,000 Monte Carlo Markov Chain iterations. The most likely number of populations,

K, was obtained by assessing the value K for which the likelihood of the posterior probability LnP(K) was the smallest. An ad hoc statistic of the LnP(K) values was performed in STRUCTURE HARVESTER (Earl & vonHoldt, 2012) to calculate and plot the ΔK values, which measure the rate of change of LnP(K) between two consecutive K as described by Evanno and colleagues (2005).

The NeESTIMATOR version 2.1 (Do et al., 2014) was used to estimate the effective population size over a generation (NE) based on molecular ancestry, and linkage disequilibrium methods (Peel et al., 2004; Waples et al., 2016; Waples, 1989;

Waples, 2006). The M_P_Val.exe program was used to perform the M-Ratio (r/k) test

(Garza & Williamson, 2001) that detects recent reductions in effective population size based on the ratio between the number of alleles (r) and the range in allele size (k). The following parameters were used: mutation rate µ=5 x 10-4, θ=4Neµ=0.11, the

210

recommended value of Δg=3.5 and Ps=0.9. Recent reduction in effective population size was assumed if M<0.68 (Hunter et al., 2010a). The existence of a possible bottleneck was determined by assessing the heterozygosity excess using

BOTTLENECK Version 1.2 (Cornuet & Luikart, 1996). Only the non-parametric

Wilcoxon test was considered because the number of markers was less than 30.

BOTTLENECK test for drift-mutation equilibrium using three mutation models including the IAM, TPM, and the SMM.

Sex ratio

The sex ratio was computed by dividing the number of unique genotypes identified as males by the number of unique genotypes identified as females. The sex ratio was computed for Lake Ossa and the entire DSRW. We could not estimate the sex ratio for the other locations (Lake Tissongo, and Sanaga River) due to the small sample size.

Results

Discrimination of Individuals

Of the total of 110 African manatee fecal DNA samples, 75 samples were retained for subsequent analysis as they each yielded consensus genotypes. From the set to meet a minimum of 75% of the 12 loci panel assessed, one locus (locus Tma-

K01) was removed from the panel because more than 30% of the samples failed to amplify at that locus. Of the 75 samples, 26 (34%) were removed from the final data set because they either matched at all loci (n=14) or mismatched by only one (n=09) or two loci (n=3) with other sample genotypes. Thus, 49 samples with unique genotypes were retained in the final data set. The samples were geographically distributed as follows:

Lake Ossa (n=35), Sanaga River (n=9), and Lake Tissongo (n=5), Figure 5-1.

211

The unbiased P(ID) was estimated to be 7.2E-11 and P(ID)sib was 5.8E-05 and was in a similar order of magnitude to the observed P(ID) and P(ID)sib (4.5E-10 and 9.5E-05 respectively). P(ID) = 44.5E-10 implies that a population size of up to 1/ P(ID) (2.2 E8) manatees can be assessed to unambiguously discriminate individuals. This population size is far over the estimated population size of the species throughout its home range

(10,000, Keith-Diagne 2015); thus, our microsatellite markers provide enough variability for individual discrimination. The shadow effect was extremely low (3.53E-09). Both the unbiased and observed P(ID) and P(ID)sib values were within the values recommended

(<0.001, Waits et al., 2001) for reliable identification of individuals. This information could suggest that the 12 loci used in this study had enough variability for individual identification. The average relatedness was R=-0.023 ± 0.007, which could suggest that our sampled individuals were not closely related; and, there was no evidence of sampling bias resulting from a large proportion of closely related individuals.

Microsatellite DNA

TmaE01 and Tma-FWC17 showed evidence of null alleles. Thirteen locus pairs showed evidence of linkage disequilibrium (P<0.05); however, after a Bonferroni sequential correction (P<0.0008), none of the 66 locus pairs were in linkage disequilibrium. TmaE01, Tma-FWC17, and Tma-FWC18 showed evidence of departure from Hardy-Weinberg equilibrium (P<0.05); after a Bonferroni sequential correction

(P<0.004), only TmaE01 and Tma-FWC18 were still not in agreement with HWE.

Therefore, the two loci were excluded from the final locus panel.

The 12 nuclear microsatellite markers for the DSRW manatee population showed a higher level of genetic diversity (He=0.66 ± 0.03, PIC=0.6, Na=5 ± 0.5 and Ne=3.1 ±

212

0.2, Table 5-1) than Puerto Rico, Belize, Florida and Mexico (Davis, 2014; Hunter et al.,

2010a; 2012; Nourisson et al., 2011).

The overall inbreeding coefficient of the DSRW manatee population was low

(FIS=0.046) but not significant (P=0.16); wich could suggest a low level of inbreeding within the population. The global FST was very low (0.009) and not significant (P=0.22), which could suggest an absence of genetic differentiation within DRSW; however, RST was moderate (0.177) and significant (P=0.005) suggesting a slight genetic differentiation. The PCA did not identify any genetic distance between manatees in

Ossa and Tissongo lakes and the Sanaga River (Figure 5-2). The Bayesian and resulting Evanno et al. (2005) methods using log-likelihood and ΔK analysis also suggested no genetic clustering between the three locations as “K=1” yielded the lowest

LnP(K) and ΔK values were very low for K=2-7 (Figure 5-3A, B).

The Wilcoxon test suggested that the manatee population in DSRW is at mutation-drift equilibrium as a bottleneck signature were not detected for the TPM

(P=0.19) and SMM (P=0.5); also, the frequency distribution test remained in a normal L- shape distribution. However, the bottleneck signature was found for the IAM

(P=0.0005). The M-ratio was 0.606 and was slightly lower than the threshold value of

0.68 (Garza & Williamson, 2001); which could suggest a possible recent reduction in effective population size. Wright (1931) defines NE as the ideal Wright-Fisher population size that would show the same level of genetic drift and inbreeding as the considered population. The effective population size estimates were 45.5 (29.4-82.5, 95% confidence interval) using the linkage equilibrium approach and for allele frequency

213

≥0.05. The molecular coancestry method yielded an effective population size estimate of 53.2 (0.1 – 267).

Sex Ratio

The overall male to female sex ratio estimate in the DSRW was 1.8:1 including

30 males, 17 females, and two undetermined sex. The sex ratio estimate (1.8:1, including 22 male, 12 females, and one unknown sex) in Lake Ossa was the same as that of the overall study area, DSRW. Of the 35 unique samples collected in Lake Ossa,

24 were collected during the low-water season and showed a sex ratio estimate of 3:1

(18 males and six females). There were only 11 unique samples collected during the high-water season in Lake Ossa; they yielded a lower sex ratio (0.66:1) and included four males, six females, and one unknown.

Mitochondrial DNA

A total of four control region haplotypes was found from the 47 unique samples identified in this study; two samples failed to amplify. One sample that failed to be included in the final nuclear data set because of poor microsatellite amplification was added to the final data set for the mtDNA analysis because the sample showed a unique haplotype for Lake Ossa. Among the four haplotypes, three are being reported for the first-time including TS-CR17 (n=1), TS-CR18 (n=1), and TS-CR19 (n=3) (Figure

5-1). The fourth haplotype (TS-CR06; Keith-Diagne, 2014) was the most frequent

(n=42). Figure 5-1 shows the geographical distribution and frequency of the six control region haplotypes identified to date in Cameroon.

Mitochondrial genetic diversity estimates (π=0.003, HD=0.294, and k=1.3) of manatees in Cameroon were lower than those previously reported in Keith-Diagne

(2014) study that used a smaller sample size (n=7). Our estimates were also lower than

214

those reported for manatee populations in Gabon (π=0.004, HD=0.813, and k=1.56; n=19) and Senegal (π=0.007, HD=0.543, and k=2.8; n=18) by the same author. The current number of haplotype sites (S=9, including one transversion and eight transitions) for the manatee populations in Cameroon was higher than the value previously reported

(S=5; Keith-Diagne, 2014). The pairwise mtDNA genetic comparison among the three locations within DSRW indicated a high and significant FST=0.336 (P=0.038 ± 0.002) and ΦST =0.509 (P=0.031±0.002) between Lake Ossa and Lake Tissongo (Table 5-2); which could suggest a considerable genetic difference between the two locations.

Mitochondrial DNA genetic differentiation between Lake Ossa and Sanaga River and between Sanaga River and Lake Tissongo were low and not significant (Table 5-2). The genetic differentiation estimates between Cameroon and Gabon were FST=0.486

(P≤0.00001 ± 0.0) and ΦST (Tamura) = 0.374 (P≤0.00001 ± 0.0). Within Cameroon,

Tajima’s D was large and negative (D=-0.913), while the more sensitive test Fu’s FS was low and negative (-0.107); therefore, the null hypotheses of selective neutrality could not be rejected and suggested a recent population expansion after a recent bottleneck; however both tests were not significant (P>0.05). Tajima’s D and Fu’s FS values reported in this study were very low compared to those shown in Keith-Diagne

(2014) for Cameroon.

The evolutionary analysis performed in MEGA6 allowed for the generation of the most likely phylogenetic tree (Log-likelihood=-0.886.99) with 1,000 bootstrap replicates

(Figure 5-4). The resulting control region consensus tree with the new haplotypes indicated two genetically and geographically separated main groups, the North and

South regional groups (Figure 5-4).

215

An additional separation was also observed within the South regional group; the two distinct subpopulations are shown in the consensus tree generated in this study.

The first subpopulation, represented by the haplotypes primarily from the Niger River of the South region group as described by in Keith-Diagne (2014), included haplotypes from, Chad, Niger, Mali, and Cameroon. The second subpopulation, the coastal haplotypes of the South region group, included control region haplotypes from Ghana,

Ivory Coast, Cameroon, and Gabon. Two of the new haplotypes (TS-CR17 and TS-

CR18) identified in this study grouped with the coastal subpopulation of the south regional group; while TS-CR19 (another new haplotype) instead grouped with the further inland subpopulation of the south regional group (Figure 5-4). Haplotypes TS-

CR17 and TS-CR18 were only found in Lake Ossa (Figure 5-1) and only one sample for each. TS-CR19 was found both in Lake Ossa and Lake Tissongo.

Discussion

There are at least 49 Manatees in the DSRW

The high microsatellite polymorphism (PIC=0.6) of the African manatee in the

DSRW has allowed discrimination of individuals based on DNA isolated from free fecal samples and using only ten microsatellite markers. A total of 49 unique genotypes were derived from the 75 fecal DNA isolates that were successfully genotyped, suggesting that at least 49 African manatees were living in the DSRW, with 35 collected in Lake

Ossa, nine collected in the Sanaga River and five collected in Lake Tissongo. The higher minimum number in Lake Ossa supports the belief that Lake Ossa is an important refuge for the African manatee in the DSRW (Powell, 1996; Takoukam Kamla,

2012). However, the sampling effort was higher in Lake Ossa because of logistical and

216

proximity conveniences which may have contributed to the higher minimum number of individuals in Lake Ossa.

Male-biased Sex Ratio in the African Manatee?

The overall male to female sex ratio (1.8:1) in the DSRW was far higher than the

1:1 values reported for the Florida manatee (Marsh et al., 2011; O’Shea & Hartley,

1995; Rathbun et al.,, 1995), the Antillean manatee in north-eastern Brazil (de Meirelles,

2008), and manatees in Puerto-Rico (Mignucci-Giannoni et al., 2000). A similar sex ratio

(1.9:1) was, however, reported in dugongs less than five years old (Kwan, 2002). These results could suggest that female African manatees may be more vulnerable to threats such as hunting and accidental catch than males. Nevertheless, this interpretation should be considered with caution as the high sex ratio could also be due to a seasonal sex-biased dispersal whereby the more adventurous males had a broader use of the habitat and their feces were more likely available for collection than female feces. For example, females with calves would likely stay in hidden areas to protect them, while males may move around more for food, warm refuge or searching for receptive females.

Manatees in the DSRW Constitute a Single Population.

No fine-scale genetic population study has been published in the African manatee to date. The weak and non-significant genetic differentiation (Global

FST=0.009) within DSRW indicates that there is no barrier to gene flow between the three study locations (Lake Ossa, Tissongo, and Sanaga River) and that DSRW constitutes a single panmictic unit (Frankham et al., 2002; Wright, 1978). Although the global RST was significant, it appears to not reflect the biological significance of the genetic connectivity within DSRW as it disagreed with the other analysis of genetic differentiation, including PCA and Bayesian approaches that indicated that manatees in

217

DSRW are panmictic. It is also possible that the RST value was biased due to the difference in sampling size between locations (Balloux et al., 2002; Goodman, 1997).

Indeed, the three locations are physically connected and are geographically close enough (within 50km). Manatee movement patterns and migration are variable with the type of habitat and seasonal rainfall and temperature (Marsh et al., 2011). Florida manatees can migrate distances up to 2360km along the Atlantic Coast of the USA in search of seagrasses (Deutsch et al., 2003), and the African manatee has been documented migrating distances up 265km (Keith-Diagne, 2014) in certain areas.

Therefore, it is not surprising that manatees in the DSRW within a 50km radius indicated a high level of gene flow between locations. Our initial hypothesis that manatees in DSRW exhibit only restricted movement due to the abundance of food resources was then not necessarily justified. Such restricted movement has been documented by the African manatee in the N’gni Lagoon complex in Ivory Coast, where some tagged manatee movement was limited within less than five km due to the abundance of accessible food and freshwater (Akoi, 2004).

There was minimal nuclear structuring among manatee within DSRW, Lake Ossa and Lake Tissongo. The mtDNA analysis showed a high and significant differentiation among those locations. This could suggest that manatee in the DSRW may exhibit a male-bias sex dispersal whereby females are philopatric, with males dispersing more

(Prugnolle & de Meeus, 2002). However, this result should be interpreted with caution as the sample size in Lake Tissongo was low (n=5). The higher mean pairwise relatedness further supported the male-biased dispersal among female manatees (-0.10

± 0.03) in Lake Ossa compared to males (-0.05 ± 0.02) of the same location. The

218

difference in mean pairwise relatedness between sexes could not be estimated for Lake

Tissongo because of the small sample size. The difference in sex ratio between seasons in Lake Ossa (3:0 low-water and 0.66:1 high-water season) further emphasizes the hypothesis of sex-biased dispersal within DSRW (Prugnolle & de Meeus, 2002).

This finding could suggest that during the high-water season when water level increases in the lake and the adjacent Sanaga River, more habitat become available, therefore more males than females disperse out of lake, while some females remain in the lake to rear their calves. Manatee accidental catch reported in the past five years in Lake Ossa has been mostly calves (AMMCO unpublish data). AMMCO also reported once sonar images of a manatee and calf in Lake Ossa; which could suggest that Lake Ossa might be a calving site. Similar male-driven dispersal has been genetically documented in the

Amazonian manatee (Satizábal et al., 2012). The male Florida manatee is also known to exhibit longer-distance migration than philopatric females (Deutsch et al., 2003;

Marsh et al., 2011).

African Manatee Populations in Cameroon and Gabon are Genetically Distinct?

The manatee population in DSRW was compared only to Gabon because only for this country there was enough manatee mtDNA sample size to adequately measure genetic differentiation using mtDNA. The mtDNA value θST=0.374 between Cameroon and Gabon was significant (P≤0.00001 ± 0.0) and slightly higher than the value reported by Keith-Diagne (2014, θST=0.322) for the two countries. This difference might be attributed to the discovery of three additional rare haplotypes (TS-CR17, TS-CR18, TS-

CR18) for Cameroon (Hunter et al., 2012). The difference could also be due to the difference in sample sizes between the two locations. Nevertheless, the result from both studies indicated a large genetic differentiation between Cameroon and Gabon manatee

219

populations. The high number of private (found in only one region) control region haplotypes between the two countries further supports the genetic divergence between the two countries.

This mtDNA genetic divergence between the two countries could be due to isolation by distance; future microsatellite data comparison between the two countries may help confirm this. The water along the coast corridor between Gabon and

Cameroon is very rough throughout most of the year as it was reported to us by the local fishers who named the coast adjacent to the south of Sanaga Estuary “God No

Dey” (meaning God is not there) in reference to the roughness of the sea there. The

MetaoOceanView (2019) historical map of the sea state along the Central Atlantic coast also indicated that the mean significant wave height of the coast south to DSRW is relatively high (about 1m) compare to the northern coast to DSRW (0.5m) (Figure 5-5).

This strong current might constitute a vicariant barrier that may be limiting gene flow between DSRW and Gabon; and therefore, leading progressively to their genetic separation overtime. Finally, considering the substantial mtDNA divergence between

Cameroon and Gabon manatees, and the “Stepping Stone” population model of the species dispersal along the coast of Africa, the two countries could be designated as two distinct management units (Palsboll et al., 2007). Cameroon and Gabon shared only one common control region haplotype (TS-CR06), which could also suggest the existence of limited migratory movements between the two locations; it could also indicate that the manatee populations from the two countries were founded from a similar ancestral population.

220

Geographical Extend of the DSRW Manatee Population

In the previous sections, we demonstrated that manatee in the DSRW is one single population and they are genetically separated with the manatees from Gabon; however, some geographical boundaries of this population is not clearly defined yet.

The water fall cascade near the hydroelectrical dam of Edea (Figure 5-1) constitute the north eastern boundary of the population as no manatee has been reported upstream the dam and waterfall. There is no inland physical aquatic connection between the

DSRW and the closest Lokoundje River, 43km south (Figure 1-2), where manatee are also reported (Nishiwaki et al., 1982; Powell, 1996; Takoukam Kamla, 2012). Thus, movement of manatee toward the south from the Sanaga River can only be coastal.

Because of the roughness of the sea along the south coast from the Sanaga Estuary

(Figure 5-5) and the lack of sea grasses (due to the murky water along the coast in that area), southern movements from Sanaga River to Lokoundje, Kienke, Lobe or Ntem

Rivers (Figure 1-2), although possible, may be less likely compared to the likelihood of movement towards the north. Northern movements of manatees from the Sanaga

Estuary are likely favored because of both the inland and coastal pathways. The north coast from Sanaga Estuary to Douala Estuary and up to Porth Harcourt in Nigeria appears to be less rough than the south coast (Figure 5-5). Alternatively, manatees from the Sanaga River can easily connect to the Douala Estuary and tributaries (Wouri,

Nkam, and Dibamba Rivers) using the northern mangrove creeks such as the Kwakwa

(Figure 1-2). Therefore, we would expect manatees in the northern adjacent habitat to be more closely genetically related to the DSRW manatee than manatees in the southern adjacent habitats. The fact that two haplotypes TS-CR06 and TS-CR07 have been documented in both DSRW and Wouri River (Figure 5-1) further support the

221

hypothesis of a high genetic connectivity between the two locations. It would not be surprising if future microsatellite data analysis from the Wouri River indicate that manatees from the Wouri Estuary, tributaries and the DSRW belong to the same population. It is possible that the DSRW population extends up to Ivory coast or Ghana as the new haplotype TS-CR18 identified in this study in Lake Ossa were closely related to haplotype TS-Y03 documented in Ghana, Ivory-Coast and Gabon by Vianna et al.

(2006) and Keith-Diagne (2014). As TS-CR07 (Keith-Diagne, 2014), the new haplotype

TS-CR19 we found in lakes Tissongo and Ossa also grouped with the Niger River haplotypes (Mali, Niger and Chad, Figure 5-4), further indicative of northern manatee movements from DSRW to the Niger River. However, it appears that movements of manatees between the DSRW and the Niger River or Ivory coast or Ghana might be rare as the frequency of their closely related haplotype is very low. No genetical data of manatee living the habitats between the DSRW and Gabon has been reported so far. It will be crucial to bridge this phylogeographical gap to determine the southern limit of the

DSRW manatee population.

While awaiting additional fine scale population structure data over the adjacent habitats to DSRW to determine with more certainty the limit of its population, the current mtDNA and microsatellite genetic information can be used to suggest management units. Crandall et al. (2000) proposed a delineation of management units within species based on the genetic and ecological exchangeability and on whether the latters occurred in a recent or historical time frame. Genetic exchangeability is accepted when there is limited gene flow between populations (Nm<1, Crandall et al., 2000; Frankham et al., 2002). Ecological exchangeability is considered when there is no evidence of

222

population differentiation as a result of selection or drift; which could be indicated by the lack of morphological and habitat difference (Frankham et al., 2002). Nm between

DSRW and Gabon based on Nei (1973) GST estimation was 0.68; therefore, genetic exchangeability between the two population is rejected (+). The limited geneflow between the two population is both historical (0.85 million years ago, as suggested by the control region output tree of trichechid divergence constructed by Keith-Diagne,

2014) and recent. The manatee habitats in Gabon and Cameroon are very similar and are located in the tropical forest; thus, we accept ecological exchangeability between the two populations (-). Considering the precedents, manatee population in the DSRW and Gabon should fall in category 7 according to the categorization scheme of Crandall et al. (2000); which implies that manatees in both locations should be treated as a single population which allows geneflow consistent with the current population structure.

We would expect manatee populations between DSRW and Niger River and may be between DSRW and Ivory-coast and Ghana to fall either in category 7 or 8 as we demonstrated that manatee from DSRW are more likely to migrate northward than southward, and therefore might indicate a lower genetic differentiation compared to that with the Gabon population. Therefore, while awaiting additional genetic information, we would suggest that manatees from Gabon up to at least Ghana be treated as single population or management unit. Because this management unit encompasses several countries, “political” management unit should be considered to account for the difference in policy and culture between countries.

Inferring Manatee Population History in the DSRW

The level of diversity observed in manatees in DSRW using the microsatellite

DNA was higher than values reported for manatee populations in Florida (He=0.48,

223

Na=4.8, Tucker et al., 2012), Mexico [Chetumal Bay (He=0.46, Na=3.0), Ascension Bay

(He=0.43, Na=2.0), and Gulf of Mexico (He=0.41, Na=2.62, Nourisson et al., 2011)] and

Puerto-Rico (He=0.45, Na=3.9, Hunter et al. 2012); but was slightly lower than the microsatellite DNA diversity estimated for the Amazonian manatee in the Colombian and Peruvian Amazon basin [He (0.609-0.749), Na (3.66-5.73); n (5-12, Satizábal et al.,

2012)]. The high genetic diversity of manatee in the DSRW could be either due to population expansion or migration. The first hypothesis is less likely when considering the hunting pressure that persists in the DSRW even though that pressure has been reduced since the establishment of the two protected areas. The migration hypothesis is more plausible as DSRW is physically connected to adjacent habitat and the mtDNA data indicated that manatee might exchange genes with manatees in Niger River and

Gabon (Keith-Diagne, 2014).

With the three additional haplotypes identified in this study, and two other new non-published haplotypes identified by Keith-Diagne (TS-CR14 and TS-CR15), the

African manatee contains 24 control region haplotypes, the second highest number among sirenian species after the Amazonian manatee with 34 control region haplotype reported so far (Garcia-Rodriguez et al., 1998; Keith-Diagne, 2014; Vianna et al., 2006).

Within Africa, Gabon (h=8) exhibits the highest number of haplotypes, followed by

Cameroon (h=6, of which 5 were found in the DSRW). However, nucleotide diversity remained very low (π=0.003) in Cameroon. There are two types of genetic mutations, transversion and transition (Guo et al., 2017). Among the nine haplotype polymorphic sites identified in Cameroon, only one resulted from a transversion which is a DNA point substitution mutation whereby the nucleotide base is changed from a purine (Adenine or

224

Guanine) to a pyrimidine (cytosine and thymine) and vice-versa (Futuyma, 2013; Guo et al., 2017). In transition mutation, a purine base is exchange with another purine or pyrimidine with another pyrimidine, maintaining the same number of the rings on the nucleotide base. Transversion mutations occur much less frequently than transition, which might explain why only 11% of the nine substation sites of the haplotypes in

Cameroon were transversion. Also, transition are likely to persist as a silent mutation as they are less likely result into an amino acid substitution (Futuyma, 2013).

In the DSRW, the high number of haplotypes and low mtDNA nucleotide variation could suggest a rapid expansion from a reduced effective population size (Frankham et al., 2002). Indeed, historically effective population size estimates based on nuclear DNA were low when using either the linkage disequilibrium approach (NE=45.5) or the molecular coancestry method (NE=53.2), further suggesting that the population may have experienced an historical bottleneck. NE is an important genetic and demographic information for the management of a wildlife population as it reflects the rate of loss of genetic variability as well as the rate of increase in inbreeding in that population

(Ferchaud et al., 2016). Based on evidences documented since 1980, Frankham and colleagues(2014) indicated that an NE=50 is too small for preventing inbreeding depression over five generations in the wild and that an NE of at least 100 is needed to reduce loss in total fitness to less than 10%. Thus, an NE around 50 could suggest that manatees in the DSRW are under a high risk of extinction and that the current

“Vulnerable” IUCN Red List status may be underestimating the likelihood of extinction of the species assuming that other African manatee populations are experiencing such a low effective population size. There is a correlation between between NE and the adult

225

census population size (NC) (Ferchaud et al., 2016; Richard Frankham et al., 2014).

Using the effective to census population ratio average of 10 to 20% (Frankham et al.,

2014), we can extrapolate the adult manatee census population size of DSRW to be between 250 and 500 manatees. The NE values in this study were higher than those reported for the Chetumal Bay (NE=32.4), Ascencion Bay (NE=4.9), and Gulf of Mexico

(NE=27.6) manatee population in Mexico (Nourisson et al., 2011). The effective population size for the Florida manatee was estimated at 197.2 for the east coast, 429.4 for Gulf Coast, and 1260 for the total Florida population (Tucker et al., 2012). Other possible causes of the small NE in the DSRW manatee population could include the recent reduction in population size due to the higher hunting pressure the population had experienced before they became legally protected. It could also be due to a founder effect whereby a few individual manatees migrated along the coast, colonized the

DSRW and expanded to form the current population. This hypothesis has also been suggested by Keith-Diagne (2014) based on her mtDNA genetic results. Such a stepping stone population model has been documented in the West Indian manatee

(Nourisson et al., 2011; Vianna et al., 2006). The small NE could also be due to the combination of the two possible causes mentioned above. Finally, the small sampling sizes may have also biased NE towards a smaller value. Therefore, it is essential to collect additional samples, especially from Lake Tissongo and Sanaga River.

The M-ratio value (<0.68) also supported the hypothesis that manatee in DSRW have suffered a recent reduction in population size; while the negative values of

Tajima’s D values of both Tajima’s D and Fu’s Fs tests agreed with population expansion hypothesis suggested by the high haplotype number and low nucleotide

226

diversity mentioned earlier. The difference between the Tajima’s D and Fu’s Fs is likely due to that the later test is more sensitive as it accounts for the polarity of mutations.

The Wilcoxon test failed to reject the null hypothesis for heterozygosity excess under the TPM and the SMM model. However, the test under the IAM model was very significant and could suggest that the population experienced a recent bottleneck. The conclusion from the TPM and SMM models more likely reflect the biological reality experienced by the DSWR manatee. This is because mutation of microsatellite fragments likely follows a stepwise pattern by the addition and subtraction of one repeat unit at the time rather the infinite allele model whereby alleles could mutate to any allele size state with the same probability. Small size populations often experience heterozygosity deficit due to inbreeding and drift (Frankham et al., 2004). The absence of heterozygosity deficit in the DSRW manatee population despite its small effective population size could be due to the compensation effect of migration that may continuously fuel in new alleles in the population and causes a migration-drift equilibrium. Indeed, the DSRW may be exchanging genes with neighbor habitats, including Wouri estuary, Dibamba, Nkam, Nyong, Ndian, and Niger Rivers. It is important to sample these locations to understand the level of gene flow between these adjacent habitats.

Evidence of this migration is supported by the phylogeny results (Figure 5-3) and the spatial distribution of control region haplotypes in Cameroon (Figure 5-1). One haplotype (TS-CR06) occurred ubiquitously and was the most abundant haplotype in

Cameroon (84.5%; n=58). The DSRW and the Wouri River shared two haplotypes in common, indicating that manatees from these two areas are connected. This

227

information further emphasizes the need to sample this adjacent habitat to determine the level of gene flow and connectivity between the two locations. TS-CR06 haplotype has been identified in one Gabon sample (4.8%, n=21) and grouped with the coastal population (Gabon, Cameroon, Ghana, Ivory Coast) of the South regional clade (Figure

5-4). The greater frequency of this haplotype in Cameroon compared to Gabon might have been the result of a small population of manatees from Cameroon that migrated to

Gabon or vice versa; however, the formal hypothesis is likely because current along the coast between and Gabon are southward (MetOceanView, 2019).

Only haplotype (TS-CR06) was found in all studied locations in Cameroon. This haplotype is the most frequent in Cameroon; it was previously identified in Cameroon by

Keith-Diagne (2014, n=5) along with two other haplotypes TS-CR07 (n=2) and TS-CR16

(n=1, Keith-Diagne, unpublished). The updated control region consensus tree built with the new haplotypes was similar to that constructed by Keith-Diagne (2014). The author identified a genetic separation of the African manatee populations into two major groups or evolutionary significant unit (ESU) based on the control region haplotype; the North

(Senegal, Guinea-Bissau and Guinea) and South regional clades. Keith-Diagne indicated that the South regional group was further divided into two distinct subpopulations which included Ghana, Mali, Ivory Coast, Chad, Niger, Cameroon and

Gabon appeared to be further divided in two subpopulations; the coastal subpopulation which comprised haplotypes from Gabon, Cameroon, Ghana and Ivory coast; and the inland subpopulation including haplotypes from Mali, Niger and one haplotype from

Cameroon. The three new haplotypes identified in this study grouped with the South regional group. TS-CR17 and TS-CR18, found in Lake Ossa in a single individual,

228

grouped with the coastal subpopulation of the South regional group along with the previously identified TS-CR07 (Figure 5-4). It is possible that these two new haplotypes came from a few migrants from Gabon or had emerged from the same founder population that colonized Gabon. The other new haplotype (TS-CR19) which was identified in Lake Ossa and Tissongo grouped with the further inland subpopulation

(Niger, Mali, Chad) of the South regional group. This result is not surprising as the Niger

River is located less than 400km apart from the Niger Delta in Nigeria, and Figure 5-5 indicates that the sea along this corridor is relatively quiet throughout the year.

Therefore, it is more likely that manatee in the DSRW are not isolated and may exhibit short distance migration between neighbor habit and long-distance migration to habitats in other neighboring countries.

Conclusions and Conservation Implications

The microsatellite genetic structure of manatees in the DSRW demonstrated that it is composed of a single population. DSRW encompasses two separated protected areas including Lake Ossa Wildlife Reserve (LOWR) upstream DSRW and the Douala-

Edea National Park (DENP, Figure 5-1). These two protected areas play an important role in the protection of the manatee and other wildlife species because of the extra conservation efforts. However, the presence of the geographical gap (about 16km along the Sanaga River) between these two protected areas may be causing a source-sink dynamic whereby manatee source populations from the protected areas when moving into the non-protected areas are more easily killed (sink). Therefore, we highly suggest extending and merging the two protected areas to completely encompass the entire distribution of the species within the DSRW.

229

Considering the small effective population size of manatee in DSRW, the population may be at the brink of its viable population size, and additional hunting pressure may become severely detrimental for the long-term survival of the population.

Moreover, the male-biased sex ratio may slow the growth rate of the population. It is not clear what the cause is of the male-biased sex ratio; in the event it resulted from a selective anthropic pressure whereby female manatee are more vulnerable to hunting or bycatch, then additional hunting would cause a triple-negative effects: (1) on the population dynamic through the reduction of individuals, (2) which would also result in the loss of rare alleles, thus loss of diversity, (3) finally on the growth rate whereby there would be only a few female available to reproduce. Therefore, it is crucial to increase manatee conservation efforts within DSRW through law-enforcement, awareness, alternative livelihood and the establishment of no-fishing zones.

The coastal geographical limits of the DSRW manatee population is not known.

Defining these limits is crucial to ensure that the entire population is well accounted for in management plans and the integrity of the population is preserved. Therefore, it is imperative to conduct genetic sampling of the adjacent and nearby habitats to measure the level of connectivity between these locations and DSRW.

At the regional level, manatees in Cameroon and Gabon are genetically distinct, with Gabon exhibiting a higher mtDNA diversity than Cameroon. The two countries form two separate management units. This separation should be taken into consideration when designing regional manatee management plans for Central Africa in order to preserve the diversity of these two populations. Increased population diversity is either slowly fueled by mutation or more rapidly by even a few immigrants per generation from

230

a diverse population (Frankham et al., 2004, 2002). Therefore, even though distance and rough sea conditions may hinder manatee movements along the coastal migratory corridor between the two countries, our data indicated a low level of historical movement between the locations. The additional presence of a physical barrier may compromise this connectivity. As suggested earlier, African manatees may exhibit a

“stepping stone” population model along the coast (Keith-Diagne 2014). Therefore, preserving the coastal connectivity among manatee habitat is crucial to maintain or increase their genetic diversity. One of these potential physical barriers is deep seaports constructed along the coast. There are three deep seaports between DSRW and

Gabon, including Kribi deep seaport in Cameroon, Port of Bata in Equatorial Guinea, and Port Libreville in Gabon (Kingsleychenikwi et al., 2018; Ngueguim et al., 2017).

Kribi deep seaport extends over 30km along the coastline and dredged channels several kilometers across the already narrow continental shelf; added to the intense maritime traffic may further strengthen the barrier. Although the impact of seaports on

African manatee coastal movement has not been assessed (such a study is highly needed), regional conservation strategies of the African manatee should encourage coastal development structures that are compatible with the coastal migration of manatees and other aquatic species.

This study demonstrated that the elusive and cryptic African manatee could be studied genetically at the mitochondrial and nuclear level using noninvasive fecal DNA.

Hopefully, this will facilitate future African manatee genetic studies and provide the opportunity to sample the species in locations where they are not directly accessible.

Many manatee habitats in Cameroon and across the species distribution range have not

231

been explored genetically to date. We suggest training manatee researchers in each of the distribution countries of the species on how to collect and preserve quality fecal samples for genetical analysis. Priority should be given to countries where no or few genetic samples have been collected including Guinea-Bissau, Guinea, Liberia, Cote d’Ivoire, Ghana, Mali, Togo, Niger, Chad, Togo, Benin, Nigeria, Equatorial Guinea,

DRC, and Angola. The collected genetic samples should be examined first to explore the mtDNA haplotype diversity, evolutionary lineage and management units, before progressing to microsatellite analysis to define fine-scale population structure and estimating population status and size. These analyses would provide clearer insight into the conservation status of the species at various scales include range-wide, regional, national, and local; therefore, noninvasive fecal DNA analysis may be an excellent asset for understanding the conservational needs and protection of this imperiled species.

232

Table 5-1. Manatee polymorphic microsatellite markers with population diversity information within the downstream of Sanaga River watershed. Pause et al. (2007) (P), Tringali et al. (2008b) (T), number of alleles (Na), effective number of allele (Ne), heterozygosity observed (Ho), heterozygosity expected (He), polymorphic information content (PIC), null alleles (*), departed from HWE (§),dropped from analysis (#). Observed Locus name size range Na Ne Ne/Na Ho He PIC TmaE01P 276-284 4 2.91 0.73 0.36§*# 0.66 0.59 TmaE04 P 233-245 5 3.75 0.75 0.72 0.73 0.69 TmaE14 P 236-252 4 3.87 0.97 0.64 0.74 0.69 Tma-FWC01T 132-140 3 2.05 0.68 0.52 0.51 0.40 Tma-FWC04 T 187-195 5 3.98 0.80 0.87 0.75 0.71 Tma-FWC08 T 149-186 8 3.41 0.43 0.82 0.71 0.66 Tma-FWC09 T 187-193 4 2.87 0.72 0.71 0.65 0.58 Tma-FWC15 T 218-240 5 2.57 0.52 0.56 0.61 0.57 Tma-FWC17 T 211-22 8 4.18 0.52 0.57* 0.76 0.73 Tma-FWC18 T 178-182 3 2.06 0.69 0.39§# 0.52 0.40 TmaSC05 P 135-161 5 3.23 0.65 0.82 0.69 0.63 TmaSC13 P 99-134 6 2.50 0.42 0.66 0.60 0.56 Mean 5 3.12 0.66 0.64 0.66 0.60

Table 5-2. African manatee mitochondrial control region pairwise ɸST estimates using Tamura distance estimation (below the diagonal) and P-values for exact tests of pairwise haplotype frequency comparisons (above the diagonal) for manatees in three locations within the downstream of the Sanaga River watershed. Sanaga River Lake Ossa Lake Tissongo Sanaga River 0 0.121 ±0.003 0.301±0.005 Lake Ossa 0.092 0 0.0312±0.002 Lake Tissongo 0.082 0.509 0

233

Figure 5-1. Map of control region mitochondrial DNA haplotypes identified in African manatee samples from Cameroon. Three new (solid colors) and three previously published haplotypes (patterns; Keith-Diagne 2014 and Keith- Diagne unpublished data) from 56 sequences are indicated by pie charts. The red points represent locations where fecal samples used in this study were collected. Circle size corresponds to the total number of samples per location and slices are proportional to haplotypes found (see inset table). Asterisks (*) indicate previously published haplotypes identified in Cameroon.

234

Figure 5-2. Two-dimensional principal component analysis of microsatellite genotype data (N=10 loci) for Lake Ossa (green circles), Lake Tissongo (red circles), and Sanaga River (blue circles) fecal samples (N=49 individuals). Ellipses are 95 % confidence intervals for each set.

A B

Figure 5-3 Plots of means (A) and standard deviations (B) of the posterior probabilities, among STRUCTURE 2.3 runs for each value of K (1–7, N=10 loci) for the downstream of the Sanaga River Watershed.

235

Figure 5-4. African manatee control region haplotype maximum likelihood tree inferred by using a General Time Reversible model (GTR+I+G, Nei and Kumar 2000) with invariant sites ([+I], 21.4585% sites) and a discrete Gamma distribution (5 categories (+G, parameter = 0.1025)). The tree is rooted in Trichechus manatus control region haplotype M01 (Gen Bank accession AY963856) and is drawn to scale, with branch lengths measured in the number of substitutions per site and bootstrap values at nodes. The tree was updated from Keith-Diagne (2014), the new haplotypes identified in this study are those in red rectangles.

236

Figure 5-5. Hindcast annual mean significant wave height along the coast of West African between Cameroon and Gabon and Cameroon and Nigeria. The data was screen captured from MetOceanView website (https://app.metoceanview.com/hindcast/)

237

CHAPTER 6 CONCLUSIONS AND FUTURE DIRECTION

Summary of Study

The Downstream Sanaga River watershed (DSRW) is essential habitat for the

African manatee (Trichechus senegalensis) in Cameroon. The species faces severe threats, including poaching, accidental catch, and habitat degradation. Improved understanding of the species’ habitat characteristics, feeding behavior, and genetic structures can provide better insight into their movements and habitat use, which are essential for the development of consequential management strategies. Lake Ossa,

Lake Tissongo, and the lower reaches of the Sanaga River contain the most important population of manatees in Cameroon. The three areas are located in two adjacent protected areas, Lake Ossa Wildlife Reserve and the Douala-Edea National Park, which are two of the four protected areas of the country hosting this imperiled species.

Understanding the status of the species and their interaction between the two areas is crucial for proper decision making and management of the two protected areas for the best interest of the African manatee and other species protected under their “umbrella.”

Therefore, in this study, we aimed at improving the conservation status of the

African manatee in Cameroon by generating sound scientific information to be used by managers and other stakeholders for the protection of the species. In Chapter 2, we assessed the physical and chemical characteristics of the African manatee habitat in

Lake Ossa and made inferences about habitat use and movement of the species. We built an empirical chain of models that can help predict the distribution of submerged aquatic plants with the change in water clarity. We also developed the first bathymetry

238

map of the lake that we used to model and compare the surface of suitable habitats based on water depth between the low- and high-water seasons.

In Chapter 3, we characterized the shoreline vegetation of the DSRW, determined its influences on the diet of the local manatee population, and, therefore, on their movements. We also determined the effect of season on the diet composition of the species in Lake Ossa. In Chapter 4, we assessed the reliability of using noninvasive fecal DNA for successfully genotyping the African manatee based on mitochondrial, microsatellite, and sex-specific markers. We established optimized fecal DNA isolation and PCR amplification protocols to yield DNA and genetic information comparable to those generated from fresh tissues. Finally, in Chapter 5, we estimated the genetic haplotype and allelic diversity of the manatee population in the DSRW using noninvasive fecal DNA. We also assessed the connectivity between manatee populations within the DSRW and those located in the neighboring habitats. The level of genetic connectivity between Lake Ossa, Lake Tissongo, and the Sanaga River was also investigated; which allowed us to infer the movement of the species within the

DSRW.

The intended goals of this study were mostly achieved. In Chapter 2, we demonstrated a nutrient enrichment of Lake Ossa by the waters from the Sanaga River that has resulted in eutrophication of the lake. There has recently been a massive proliferation of an invasive aquatic plant called Salvinia molesta. We also demonstrated that Lake Ossa now only sustains a tiny (<5%) amount of submerged aquatic vegetation due to the low transparency of the water, implying that manatees there must only rely on emergent shoreline vegetation. Our bathymetric model showed that during the low-

239

water season, as the water recedes, most of this shoreline vegetation is exposed and less accessible to the manatees due to the lowered water level. The same model also indicated that during that season, only about 6% of the lake has a suitable depth (>2m) for manatee resting and movements. From these findings, we concluded that Lake

Ossa may not be a suitable habitat for manatees during the low-water season, and at that time of year the species may migrate to the lower reaches of the Sanaga River where they have access to deeper water levels created by the tide.

In Chapter 3, we found that the shoreline vegetation habitat was very diverse

(more than 160 species of macrophytes) and differed between locations. This diversity and difference were also reflected in the manatees’ diet and the antelope grass

(Echinochloa pyramidalis) appeared to be their primary food plant, representing more than 50% of their diet. In Lake Ossa, during the low-water season, we noticed that the manatees relied almost exclusively on E. pyramidalis as it is the most abundant plant at the forefront of the shoreline vegetation and remains until the water recedes. Thus, we suggested that manatees have limited access to food resources during the low-water season compared to the high-water season when they have access to fresh and abundant vegetation originating from the adjacent swamp and flooded forest regions.

This finding supports the conclusion from Chapter 2 that the lake is less suitable for the species during the low-water season and that manatees may migrate down to the estuary where the high tide might give them better access to the shoreline vegetation. In the future, the destruction of E. pyramidalis by the invasion of Salvinia molesta may further decrease the suitability of the lake for the manatees during both seasons.

240

In Chapter 4, we demonstrated that noninvasive fecal DNA is reliable for genetic analysis based on both mitochondrial and nuclear markers. The quality of the mitochondrial genetic data obtained from the feces of the African manatee appeared to be comparable to those obtained from similar studies using tissue samples. For the first time, reliable microsatellite and sex-specific genotypes were generated from non- invasive DNA; creating a new venue for studying the elusive and cryptic manatee without having to capture or see them. The genetic data obtained from African manatee feces has allowed us to remotely explore their genetic make-up and compare connectivity within the DSRW.

In Chapter 5, significant results indicated that manatees in the DSRW constitute a single population and that there is high connectivity between its habitats. This finding also supports the conclusion of Chapter 2, where we demonstrated the drastic seasonal change of water depth, and the lack of submerged aquatic vegetation in Lake Ossa, may predispose the species to migrate in and out of the system. This hypothesis was also supported in Chapter 3, where we have shown that the seasonal availability of food plants influenced the manatee diet in Lake Ossa due to the change in water level.

Moreover, the restricted food availability during the low-water season may also predispose the species to migrate seasonally between the lake and other nearby habitats of DSRW. Therefore, this study provided sufficient evidence that manatees between the habitats of DSRW are interconnected and that such connectivity should be preserved. Although this information on the low genetic divergence within DSRW was not surprising, the scientific demonstration of this connectivity was crucial because there has been no tangible evidence of manatee movement between the two protected areas

241

because the species is cryptic and elusive. With this scientific-based information, managers of the LOWR and DENP would have additional evidence to make consequential and effective decisions for the protection of the species.

Future Directions

In this study, we demonstrated that E. pyramidalis is an essential component of the manatee diet in the DSRW. Therefore, mapping the seasonal distribution and abundance of this plant might be an important step to designate of critical habitat for the manatees for the manatees that need priority conservation efforts implemented. We have suggested that Salvinia molesta may negatively impact manatee food plants. It is imperative to quantify the impact of Salvinia on available manatee food plants as this may constitute a new, but potentially severe, threat to their survival in Lake Ossa.

Salvinia molesta proliferation in Lake Ossa is a symptom of a more profound issue of nutrient enrichment of the lake, and, addressing only Salvinia proliferation is equivalent to treating the symptoms without addressing the disease itself. Therefore, for a long- lasting and sustainable solution, it is crucial to mitigate nutrient enrichment of the lake and the Sanaga River. An essential step towards that end will be to develop and implement a water management plan for the country.

It is essential to also establish no-fishing and no-disturbance zones in areas of the lake where beds of E. pyramidalis have not yet been invaded by S. molesta, as those plots might be valuable feeding areas during restoration efforts for the few manatees still present in the lake. A complete aerial mapping of the lake’s macrophyte vegetation should be completed to identify these areas for adaptive management efforts.

242

We found that feces from the river, although difficult to locate, yielded better quality and higher quantity of DNA when compared to DNA obtained from the lake water. Scent detection dogs may improve the detection rate of fecal samples in rivers and lakes, as has been demonstrated with the detection of right whale feces in open water (Rolland et al., 2006). Investigating the efficiency of using scent detection dogs in manatee fecal survey scouting and subsequent recovery would help improve the amount and quality of DNA isolated from manatees.

All the primers used in this study were designed originally for the West Indian manatee. Designing new microsatellite and sex-specific primers for the African manatee genome may further improve the amplification success and relieve allelic dropout.

The African manatee is the least known of all sirenian species. Progress on the scientific knowledge about the species has been hampered by four major limitations, including low study skill levels, difficult logistics, low funding, and the overall cryptic nature of the species. Given the opportunities offered by acquiring noninvasive fecal

DNA and its relatively low cost, it is convenient to use the technique to study African manatee genetics with a finer resolution at a larger scale. This will allow scientists to define management units for the species across its entire home range. The same approach could help provide reliable estimates of the population size at various locations using individual identification and capture-recapture approaches.

It is also important to initiate manatee movement tracking within DSRW using the satellite tagging approach in order to further understand the movements of the manatee within and out of the system. If possible, information from tracking individuals could be corroborated with the genetic data, and used to estimate the correlations between the

243

two approaches. This correlation would also allow scientists to build a model that would help to accurately interpret the biological meaning of genetic results generated from

African manatee populations.

The take home message for this study is that manatees within DSRW migrate within and out of the system in response to the seasonal availability and suitability of its habitats. We suggest that this connectivity should be preserved by extending and merging the two protected areas (LOWR and DENP) to encompass the entire distribution of the species within DSRW and increasing law-enforcement, awareness, alternative livelihoods, and the establishment of no-fishing zones.

244

APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 3

Table A-1. List of plant group, family and species by location, number of plots surveyed (in the brackets), and their relative abundance in percentage. Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Emergent macrophytes 37.44 69.53 90.63 85.49 70.77 Amaranthaceae 0.00 0.17 0.00 0.01 0.05 Amaranthus esculentus 0.00 0.07 0.00 0.00 0.02 Cyathula prostrata 0.00 0.10 0.00 0.01 0.03 Amaryllidaceae 0.00 0.00 0.00 0.01 0.00 Allium sp. 0.00 0.00 0.00 0.01 0.00 Araceae 1.38 0.55 0.78 0.16 0.72 Colocassia esculentus 0.00 0.55 0.00 0.00 0.14 Cyrtosperma merkusii 0.00 0.00 0.00 0.03 0.01 Cyrtosperma senegalense 1.38 0.00 0.78 0.13 0.57 Arecaceae 3.38 0.00 37.97 6.77 12.03 Calamus acidus 0.00 0.00 0.00 0.81 0.20 Eremospatha macrocarpa 0.63 0.00 32.34 2.07 8.76 Laccosperma robustum 0.00 0.00 0.00 2.00 0.50 Laccosperma secundiflorum 0.00 0.00 5.63 1.89 1.88 Nypa fruticans 2.50 0.00 0.00 0.00 0.63 Raphia sp. 0.25 0.00 0.00 0.00 0.06 Asteraceae 0.38 3.18 0.00 0.37 0.98 Ageratum conyzoides 0.00 1.37 0.00 0.21 0.39 Chromolaena odorata 0.00 0.05 0.00 0.00 0.01 Eclipta prostrata 0.13 0.89 0.00 0.12 0.28 Melanthera scandens 0.25 0.21 0.00 0.03 0.12 Mikania micrantha 0.00 0.66 0.00 0.00 0.17 Unknown Asteraceae 0.00 0.00 0.00 0.01 0.00 Athyriaceae 0.00 0.74 0.00 0.76 0.37 Diplazium sammatii 0.00 0.74 0.00 0.76 0.37 Boraginaceae 0.00 1.03 0.00 0.06 0.27 Heliotropium indicum 0.00 1.03 0.00 0.06 0.27 Calophyllaceae 1.23 0.00 0.00 0.00 0.31 Calophyllum inophyllum 1.23 0.00 0.00 0.00 0.31 Cleomaceae 0.00 1.14 0.00 0.13 0.32 Cleome ciliata 0.00 0.03 0.00 0.00 0.01 Cleome rudidosperma 0.00 0.14 0.00 0.03 0.04 Cleome spinosa 0.00 0.97 0.00 0.10 0.27 Commelinaceae 1.25 2.14 1.88 0.47 1.43

245

Table A-1. Continued Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Commelina benghalensis 1.25 2.14 1.88 0.47 1.43 Convolvulaceae 1.30 14.78 0.00 1.80 4.47 Hewittia sublobata 0.00 0.12 0.00 0.00 0.03 Ipomoea alba 0.00 9.35 0.00 1.59 2.73 Ipomoea batatas 0.00 0.14 0.00 0.00 0.03 Ipomoea involucrata 0.69 2.47 0.00 0.01 0.79 Ipomoea mauritiana 0.13 0.07 0.00 0.00 0.05 Ipomoea preussii 0.00 0.14 0.00 0.00 0.03 Ipomoea quamoclit 0.00 0.07 0.00 0.01 0.02 Ipomoea sp. 0.25 1.82 0.00 0.15 0.55 Ipomoea triloba 0.24 0.62 0.00 0.03 0.22 Costaceae 0.00 0.23 0.00 0.01 0.06 Costus afer 0.00 0.23 0.00 0.01 0.06 Cucurbitaceae 0.00 1.21 0.00 0.00 0.30 Cucumis melo 0.00 0.55 0.00 0.00 0.14 Cucumis moschata 0.00 0.17 0.00 0.00 0.04 Luffa aegyptiaca 0.00 0.34 0.00 0.00 0.09 Momordica charantia 0.00 0.14 0.00 0.00 0.04 Cyperaceae 1.64 0.00 2.34 5.04 2.26 Cyperus haspan 0.50 0.00 0.78 1.29 0.64 Cyperus papyrus 1.14 0.00 0.00 0.00 0.28 Fuirena umbellata 0.00 0.00 1.56 1.89 0.86 Pycreus lanceolatus 0.00 0.00 0.00 1.86 0.47 Rynchospora corymbosa* na na na na na Dioscoreaceae 0.00 0.10 0.00 0.00 0.03 Dioscorea cayenensis 0.00 0.10 0.00 0.00 0.03 Fabaceae 1.38 3.33 0.16 0.59 1.36 Aechynomene sensitiva 0.00 0.00 0.00 0.03 0.01 Aeschynomene indica 0.63 0.00 0.00 0.09 0.18 Aeschynomene sensitiva 0.00 0.00 0.00 0.07 0.02 Calopogonium mucunoides 0.00 0.03 0.16 0.03 0.06 Canavalia rosea 0.06 0.00 0.00 0.00 0.02 Centrosema pubescens 0.00 0.24 0.00 0.01 0.06 Pueraria phaseoloides 0.38 2.64 0.00 0.04 0.76 Vigna lutea 0.00 0.00 0.00 0.10 0.03 Vigna radiata 0.31 0.41 0.00 0.22 0.24 Hydroleaceae 0.00 0.00 0.00 0.03 0.01 Hydrolea sp. 0.00 0.00 0.00 0.03 0.01 Lamiaceae 0.00 1.64 0.00 0.37 0.50 hyptis lanceolata 0.00 1.47 0.00 0.31 0.45

246

Table A-1. Continued Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Leonotis sp. 0.00 0.07 0.00 0.04 0.03 Solenostemon monostachyus 0.00 0.10 0.00 0.01 0.03 Liliaceae 0.00 0.00 1.56 0.00 0.39 Unknown Liliaceae 0.00 0.00 1.56 0.00 0.39 Malvaceae 1.50 1.34 0.00 0.06 0.72 Abutilon sp. 0.25 0.68 0.00 0.00 0.23 Clappertonia sp. 0.00 0.00 0.00 0.03 0.01 Corchorus olitorius 0.00 0.07 0.00 0.00 0.02 Melochia corchorifolia 0.00 0.27 0.00 0.01 0.07 Triumfetta cordifolia 0.00 0.17 0.00 0.00 0.04 Unknown Malvaceae 1.25 0.00 0.00 0.00 0.31 Urena lobata 0.00 0.14 0.00 0.01 0.04 Marantaceae 1.19 0.48 0.00 0.01 0.42 Haumania danckelmaniana 0.00 0.00 0.00 0.01 0.00 Marantochloa sp. 1.19 0.48 0.00 0.00 0.42 Melastomataceae 2.44 2.05 27.11 10.72 10.58 Dissotis erecta 2.44 1.91 27.11 5.13 9.15 Dissotis falcipila 0.00 0.14 0.00 5.57 1.43 Dissotis rontondifolia 0.00 0.00 0.00 0.01 0.00 Nephrolepidaceae 0.00 0.21 0.00 0.00 0.05 Nephrolepis biserrata 0.00 0.21 0.00 0.00 0.05 Onagraceae 0.13 0.63 0.00 4.44 1.30 Ludwigia decurrens 0.13 0.17 0.00 0.04 0.08 Ludwigia hyssopifolia 0.00 0.46 0.00 1.17 0.41 Ludwigia Stolonifera 0.00 0.00 0.00 3.23 0.81 Orchidaceae 0.00 0.07 0.00 0.01 0.02 Ansellia africana 0.00 0.07 0.00 0.01 0.02 Oxalidaceae 0.00 0.00 0.00 0.04 0.01 Oxalis barrelieri 0.00 0.00 0.00 0.04 0.01 Pandanaceae 6.25 0.00 0.00 0.00 1.56 Pandanus candelabrum 6.25 0.00 0.00 0.00 1.56 Passifloraceae 0.00 0.31 0.00 0.00 0.08 Passiflora foetida 0.00 0.31 0.00 0.00 0.08 Poaceae 10.56 30.18 18.67 50.46 27.47 Acroceras zizanioides 1.25 1.71 0.00 1.12 1.02 Bambousa vulgaris 0.00 0.03 0.00 0.01 0.01 Cynodon dactylon 0.00 0.07 0.00 0.10 0.04 Echinochloa pyramidalis 3.25 15.88 16.80 46.95 20.72 Leersia hexandra 0.25 0.27 0.00 0.45 0.24 Melinis sp. 0.00 0.00 1.88 0.18 0.51

247

Table A-1. Continued Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Panicum maximum 0.00 0.89 0.00 0.04 0.23 Paspalum conjugatum 0.00 0.14 0.00 0.00 0.03 Paspalum dilatatum 0.00 0.00 0.00 0.11 0.03 Paspalum polystachyum 0.00 0.96 0.00 0.61 0.39 Paspalum purpureum 0.00 1.71 0.00 0.37 0.52 Paspalum sp. 0.00 0.10 0.00 0.04 0.04 Pennisetum purpureum 5.81 6.61 0.00 0.32 3.19 Sacciolepsis africana 0.00 0.00 0.00 0.07 0.02 Setaria barbata 0.00 0.32 0.00 0.01 0.08 Sorghum arundinaceum 0.00 1.47 0.00 0.06 0.38 Polygonaceae 0.75 3.13 0.00 2.11 1.50 Polygonum lanceolata 0.00 0.00 0.00 0.61 0.15 Polygonum lanigerum 0.38 2.33 0.00 0.90 0.90 polygonum salicifolium 0.38 0.80 0.00 0.61 0.45 Polypodiidae 0.00 0.00 0.00 0.12 0.03 Leptosporangiate fern 0.00 0.00 0.00 0.12 0.03 Proteaceae 0.00 0.01 0.00 0.00 0.00 Leucospermum sp. 0.00 0.01 0.00 0.00 0.00 Pteridaceae 2.50 0.00 0.00 0.00 0.63 Acrostichum aureum 2.50 0.00 0.00 0.00 0.63 Rubiaceae 0.06 0.22 0.16 0.90 0.34 Galium sp 0.00 0.00 0.00 0.01 0.00 Oldenlandia corymbosa 0.00 0.00 0.16 0.06 0.05 Oldenlandia diffusa 0.00 0.02 0.00 0.83 0.21 Pentodon pentandrus 0.00 0.20 0.00 0.00 0.05 Stipularia africana 0.06 0.00 0.00 0.00 0.02 Solanaceae 0.00 0.10 0.00 0.00 0.03 Physalis angulata 0.00 0.07 0.00 0.00 0.02 Solanum nigrum 0.00 0.03 0.00 0.00 0.01 Verbenaceae 0.00 0.34 0.00 0.01 0.09 Stachytarpheta jamaicensis 0.00 0.34 0.00 0.01 0.09 Vitaceae 0.15 0.23 0.00 0.01 0.10 Cayratia ibuensis 0.16 0.23 0.00 0.01 0.09 Free-floating macrophytes 8.31 4.21 0.00 4.58 4.28 Araceae 0.00 0.21 0.00 0.15 0.09 Pistia stratoides 0.00 0.21 0.00 0.15 0.09 Azollaceae 0.00 0.00 0.00 0.36 0.09 Azolla africana 0.00 0.00 0.00 0.36 0.09 Lemnaceae 0.00 0.21 0.00 0.18 0.10 Lemna minor 0.00 0.21 0.00 0.18 0.10

248

Table A-1. Continued Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Lentibulariaceae 0.00 0.00 0.00 0.70 0.17 Utricularia vulgaris 0.00 0.00 0.00 0.70 0.17 Nymphaeaceae 0.00 0.00 0.00 1.04 0.26 Nymphaea lotus 0.00 0.00 0.00 1.04 0.26 Pontederiaceae 8.31 3.80 0.00 0.00 3.03 Eichornia crassipes 8.31 3.80 0.00 0.00 3.03 Salviniaceae 0.00 0.00 0.00 2.15 0.54 Salvinia molesta 0.00 0.00 0.00 2.15 0.54 Shrubs 22.56 11.47 8.28 5.38 11.92 Apocynaceae 0.00 0.00 0.00 0.27 0.07 Allamanda cathartica 0.00 0.00 0.00 0.01 0.00 Landolphia senegalensis 0.00 0.00 0.00 0.25 0.06 Arecaceae 0.00 0.00 0.00 0.01 0.00 Elaeis guineensis 0.00 0.00 0.00 0.01 0.00 Dilleniaceae 0.00 0.00 0.00 0.07 0.02 Tetracera macrophylla 0.00 0.00 0.00 0.07 0.02 Euphorbiaceae 11.50 6.44 6.72 0.53 6.30 Alchornea cordifolia 10.56 4.32 6.72 0.53 5.53 Macaranga sp. 0.94 0.00 0.00 0.00 0.23 Mallotus oppositifolius 0.00 0.58 0.00 0.00 0.15 Manihot esculenta 0.00 1.54 0.00 0.00 0.39 Fabaceae 0.00 1.61 0.00 0.10 0.43 Millettia macrophylla 0.00 0.96 0.00 0.10 0.27 Millettia sanagana 0.00 0.65 0.00 0.00 0.16 Malvaceae 0.50 0.82 0.00 0.16 0.37 Glyphaea brevis 0.50 0.82 0.00 0.16 0.37 Phyllanthaceae 5.13 0.48 1.56 0.34 1.88 Phyllanthus amarus 0.00 0.38 0.00 0.06 0.11 Phyllanthus reticulatus 0.00 0.03 0.00 0.08 0.03 Uapaca guineensis 5.00 0.07 1.56 0.16 1.70 Uapaca mole 0.13 0.00 0.00 0.03 0.04 Rubiaceae 1.25 0.68 0.00 3.85 1.45 Canthium angustifolium 1.25 0.00 0.00 0.01 0.31 Canthium ciliatum 0.00 0.68 0.00 3.82 1.13 Nauclea pobeguinii 0.00 0.00 0.00 0.03 0.01 Sapindaceae 4.19 1.44 0.00 0.03 1.41 Allophylus africanus 0.31 0.00 0.00 0.00 0.08 Allophylus bullatus 0.00 0.14 0.00 0.00 0.03 Paullinia pinnata 3.88 1.30 0.00 0.03 1.30 Trees 31.69 14.78 1.09 4.55 13.03 Annonaceae 1.19 0.00 0.00 0.58 0.44

249

Table A-1. Continued Sanaga Upper Lake Lake Estuary Sanaga Tissongo Ossa All sites Plant type (n=80) (n=146) (n=64) (n=668) (n=958) Cleistopholis patens 1.19 0.00 0.00 0.00 0.30 Xylopia sp. 0.00 0.00 0.00 0.55 0.14 Xylopia sp.2 0.00 0.00 0.00 0.03 0.01 Clusiaceae 0.00 0.00 0.00 0.09 0.02 Symphonia globulifera 0.00 0.00 0.00 0.09 0.02 Combretaceae 0.94 0.48 0.00 0.00 0.35 Combretum zenkeri 0.31 0.48 0.00 0.00 0.20 Terminalia catappa 0.63 0.00 0.00 0.00 0.16 Fabaceae 0.00 0.79 0.47 1.32 0.64 Acacia sp. 0.00 0.79 0.00 0.15 0.23 Dalbergia sp. 0.00 0.00 0.00 1.17 0.29 Guibourtia demeusei 0.00 0.00 0.47 0.00 0.12 Gentianaceae 1.25 0.00 0.63 0.01 0.47 Anthocleista djalonensis 0.00 0.00 0.63 0.01 0.16 Anthocleista liebrechtsiana 1.25 0.00 0.00 0.00 0.31 Meliaceae 2.88 0.68 0.00 0.15 0.93 Trichilia emetica 2.88 0.68 0.00 0.15 0.93 Moraceae 3.25 10.88 0.00 1.55 3.92 Ficus benghalensis 0.00 0.00 0.00 0.37 0.09 Ficus capreifolia 0.00 5.21 0.00 1.03 1.56 Ficus mucuso 0.00 0.07 0.00 0.01 0.02 Ficus capreifolia 3.25 5.60 0.00 0.13 2.25 Rhizophoraceae 19.88 0.00 0.00 0.00 4.97 Rhizophora racemosa 19.88 0.00 0.00 0.00 4.97 Urticaceae 2.31 1.95 0.00 0.85 1.28 Cecropia obtusifolia 0.00 0.00 0.00 0.43 0.11 Myrianthus serratus 2.31 1.95 0.00 0.41 1.17

250

Table A-2. Major identified diet plant species of African manatees surveyed by location of the downstream Sanaga River watershed. The stars represent the level of significance (determined using the Kruskal-Wallis test) in the frequency of each plant species between each pair-wise location. The second column indicates the locations for which significance was observed. Min = minimum, Max = maximum, SD = Standard deviation, Var = variance, SE = Standard error and CV = Coefficient of variation Pair-wise Location significance Mean Min Max SD Var SE CV a Lake Ossa (n=60) Echinochloa pyramidalis** c 63.23 0 100.00 33.34 1111.65 4.34 0.53 Eremospatha b,c,d macrocarpa*** 3.40 0 28.00 6.37 40.53 0.83 1.87 Rynchospora b,c,d corymbosa*** 8.27 0 65.00 12.94 167.46 1.68 1.56 Cyperus sp.* c 7.61 0 74.00 18.80 353.28 2.45 2.47 c Unidentified 13*** 0 0 0 0 0 0 b Lake Tissongo (n=11) Echinochloa pyramidalis** - 44.60 2.0 76.00 26.17 684.65 8.27 0.59 Eremospatha a,c,d macrocarpa*** 29.69 0 95.00 35.26 1243.10 11.15 1.19 Rynchospora a corymbosa*** 0.09 0 1.00 0.29 0.08 0.09 3.16 Cyperus sp c 0.55 0 6.00 1.72 2.98 0.55 3.16 c Unidentified 13*** 0 0 0 0 0 0 c Estuary Sanaga (n=33) Echinochloa pyramidalis** a 37.14 0 100.00 40.06 1604.87 7.08 1.08 Eremospatha a,b macrocarpa*** 0.09 0 3.00 0.51 0.26 0.09 5.66 Rynchospora a corymbosa*** 2.58 0 33.00 6.49 42.06 1.15 2.52 Cyperus sp* a,b 29.94 0 100.00 44.12 1946.18 7.80 1.47 Unidentified 13*** a,b,d 12.08 0 84.00 27.85 775.49 4.92 2.31 d Upper Sanaga (n=09) Echinochloa pyramidalis** - 59.67 30.20 88.60 20.02 400.61 7.08 0.34 Eremospatha a,b macrocarpa*** 0 0 0 0 0 0 Rynchospora a corymbosa*** 0 0 0 0 0 0 Cyperus sp* - 0.22 0 2.00 0.63 0.40 0.22 2.83 c Unidentified 13*** 0 0 0.0 0.0 0.0 0.0

251

Table A-3. Major identified diet plant species of African manatees surveyed by seasons in Lake Ossa. The stars represent the level of significance (determined using the Kruskal-Wallis test) in the frequency of each plant species between the low- and the high-water seasons. Min = minimum, Max = maximum, SD = Standard deviation, Var = variance, SE = Standard error and CV = Coefficient of variation. Season (water level) Mean Min Max SD Var SE CV High water (n=29) Echinochloa pyramidalis*** 44.17 0 100.00 33.11 1096.02 6.26 0.75 Eremospatha macrocarpa 5.12 0 28.00 7.79 60.74 1.47 1.52 Leersia hexandra 6.97 0 71.00 15.38 236.57 2.91 2.21 Rynchospora corymbosa 10.01 0 65.00 14.58 212.66 2.76 1.46 Cyperus sp** 15.34 0 74.00 24.75 612.73 4.68 1.61 Low water (n=31) Echinochloa pyramidalis*** 81.07 3 100.00 21.64 468.08 3.95 0.27 Eremospatha macrocarpa 1.79 0 16.00 4.03 16.27 0.74 2.25 Leersia hexandra 2.26 0 31.40 6.84 46.82 1.25 3.03 Rynchospora corymbosa 6.65 0 47.60 10.94 119.73 2.00 1.65 Cyperus sp** 0.37 0 7.50 1.48 2.19 0.27 3.99

Table A-4. African manatee major identified diet items in Lake Ossa by feces diameter size. Min = minimum, Max = maximum, SD = Standard deviation, Var = variance, SE = Standard error and CV = Coefficient of variation Feces size Mean Min Max SD Var SE CV Small (diameter ≤ 4cm, n=11) Echinochloa pyramidalis 49.24 0 100 33.24 1105.01 10.51 0.68 Eremospatha macrocarpa 1.91 0 7 2.78 7.72 0.88 1.46 Leersia hexandra 11.95 0 71 21.08 444.58 6.67 1.77 Rynchospora corymbosa 13.91 0 65 18.17 329.98 5.74 1.31 Cyperus sp 5.73 0 37 12.37 153.11 3.91 2.16 Large (diameter > 4cm, n=18) Echinochloa pyramidalis 48.93 1 97 33.78 1140.82 8.19 0.69 Eremospatha macrocarpa 6.77 0 28 9.11 82.98 2.21 1.35 Leersia hexandra 6.09 0 38 11.52 132.73 2.79 1.89 Rynchospora corymbosa 5.74 0 39.6 10.04 100.85 2.44 1.75 Cyperus sp 20.03 0 74 28.83 830.89 6.99 1.44

252

Figure A-1. Map of the downstream Sanaga River watershed showing the spatial distribution of the 113 feces collected and used in this study. Collection sites are yellow circles.

253

Figure A-2 African manatee feces. A) Shape and color, and B) measurement of the diameter using a caliper. The feces were collected in the Sanaga River.

254

Figure A-3. Cumulative number of plant species by the number of plots surveyed for each location within the downstream Sanaga River watersh

255

Figure A-4. Venn diagram of the surveyed plant species in the four locations of the downstream Sanaga River watershed and showing shared and private species each location and combination of locations.

Figure A-5. African manatee identified plant diet composition profile by season in Lake Ossa.

256

Figure A-6. Histogram of the distribution of the diameters of 377 free-floating African manatee feces collected within the downstream Sanaga River watershed. Left to the blue line are feces presumably generated from juvenile manatees; while the frequency distribution right to the blue line represents feces from adults.

257

Figure A-7. Salvinia molesta proliferation in Lake Ossa. In 2016, Salvinia molesta was practically absent in the lake and manatees were feeding almost everywhere as indicated by the feeding sign in the upper left picture. Then by 2017, the invasive plant, S. molesta, started proliferating moderately along the shoreline and no more feeding signs were observed in those areas (upper right photo). In 2018, the proliferation intensified and invaded even flooded shrub forests. By 2019, the plant started invading the open water of the lake.

258

APPENDIX B SUPPORTING INFORMATION FOR CHAPTER 4

Protocol B-1. Protocol for comparing three DNA extraction protocols:

2CTAB/PCI (Vallet et al., 2008), NucleoSpin Soil Kit, and QIAamp DNA Fast

Stool Kit

Extraction methods: • 2CTAB/PCI (Vallet et al. 2008) • NucleoSpin Soil Kit • QIAamp DNA Fast Stool Kit Three sample Treatment groups • Group 1: Regular Florida free-floating manatee fecal samples (usually not fibrous) • Group 2: Regular African manatee fecal samples (mostly fibrous) • Group 3: Processed African manatee fecal samples (without the fibers).

2CTAB/PCI NucleoSpin QIAamp Total (Vallet et al. Soil Kit DNA fast 2008) stool kit Group 1 5 samples X 3 5 samples X 3 5 samples X 45 replicates replicates 3 replicates Group 2 5 samples X 3 5 samples X 3 5 samples X 45 replicates replicates 3 replicates Group 3 5 samples X 3 5 samples X 3 5 samples X 45 replicates replicates 3 replicates Total 45 45 45 45 x 3replicates= 135

A- Fecal sample collection on the field

• Half-fill 25 50ml centrifuge tubes with 95% ethanol before going to the field. • Secure the lid tightly to avoid any evaporation and store at room temperature. Avoid direct contact with sunlight. • Using a dip net with fine mesh to scoop the entire free-floating manatee bolus from the surface of the water.

259

• Using sterile gloves and forceps collect fecal material from the interior of the collected bolus. Complete the volume of half-filled 50-ml tubes with solid fecal material. Ensure that the fecal is well submerged into ethanol. • Label the sample tube.

B- Preparing and aliquoting the fecal samples for the extraction

B1- Fibrous fecal samples • To remove large plant fragments or fibers from the African manatee samples: • Take a 50ml tube containing Ts fecal sample (Tube A), vortex and decant into an empty sterile 50ml tube (Tube B) containing an ethanol solution mixed with suspended fine-size fecal material. Tube A now containing only fibrous material, should be dried for 5 minutes to remove the remaining ethanol. • Centrifuge Tube B at 4500 rpm for 20 minutes, discard the supernatant (ethanol) and keep the pellet (non-fibrous fecal material). Let the pellet dry for 5 minutes to remove any remaining ethanol. • Use a sterile spatula to collect 300mg of pellet from Tube B and transfer it into nine 5ml centrifuge tubes corresponding to the nine aliquots for that fecal sample (Tip: zero the balance with an empty 5ml tube, then add pellet to the tube and adjust until reaching 300mg or your desired fecal material weight). • Aliquot the fibrous material from Tube A into nine 5ml centrifuge tubes using sterile forceps. Use the same method described above to ensure each tube contains 300mg of fecal material. B2- Non-fibrous Florida manatee fecal. • With a spatula collect an amount of feces out of the initial tube containing the feces and let it dry for 5 minutes to remove the remaining ethanol. • Use the same method as above to aliquot 300mg of feces into each of the nine 5ml tubes for each fecal sample.

C- Start DNA extraction (for all sample groups)

C1- For 2CTAB/PCI extraction • Add 2000µl of CTAB to each of the nine 5ml centrifuge tubes containing the fecal pellet or plant material fragments. • Follow the rest of the 2CTAB/PCI protocol without further modification as described in Vallet et al. (2008). • Store the DNA isolate at -20°C for use within a week or at -80°C for longer-time storage

260

C2- Nucleospin Soil • Transfer the beads that come with the NucleoSpin Soil Kit into the nine 5 ml tubes containing the fecal aliquot. Resuspend the fecal material by adding 700µl of lysis buffer SL1 or SL2. • Add 150μl Enhancer SX, close tight the cap pf the 5ml tube, and wrap parafilm around the top before vortexing to ensure the tube does not open or and contents leak out. • Follow the rest of the steps of the NucleoSpin Soil protocol until the elution step without further modification. • Use 100ml of elution buffer SE to elute the DNA from the spin column into a 1.5ml Eppendorf LoBind microcentrifuge tubes. • Store the DNA isolate at -20°C for use within a week or at -80°C for longer-time storage

C3- QIAamp • Add 1ml InhibitEX buffer to each of the nine 5mL aliquot tubes containing the fecal material. Vortex continuously for 1 minute or until the solid material is thoroughly homogenized. • Centrifuge the sample for 1 minute to pellet stool particle. • Pipet 25µl Proteinase K into a new 1.5ml microcentrifuge tube. • Pipet 600µl supernatant from the homogenate into the 1.5 ml microcentrifuge tube containing Proteinase K. • Add 600µl Buffer AL and vortex for 15 s. • Incubate at 70°C for 10 minutes • Add 600µl of ethanol 95% to the lysate and mix by vortexing. • Carefully apply 600µl lysate to the QIAamp spin column. • Follow the rest of the steps of QIAamp protocol until the elution step without further modification • Use 100ml of TE buffer to elute the DNA from the spin column into a 1.5 mL Eppendorf LoBind microcentrifuge tubes. • Store the DNA isolate at -20°C for use within a week or at -80°C for longer-time storage

261

Table B-1. List of polymorphic microsatellite markers used in this study. Pause et al. (2007) (P), Tringali et al. (2008b) (Tb), Tringali et al. (2008a) (Ta), McHale et al. (2008) (M), annealing temperature (Tm). Primer Primer Pre-amp short Tm concentration multiplex Primer code (°C) (uM) Multiplex TmaE01P E01 54 0.134 M09 A TmaE04 P E04 57 0.181 Mm B Tma E14 P E14 56 0.103 M11 A Tma-FWC01Tb L01 58 0.079 M05 C Tma-FWC04 Tb L04 58 0.098 M07 B Tma-FWC08 Tb L08 58 0.13 M07 B Tma-FWC09 Tb L09 58 0.059 M05 C Tma-FWC15 Tb L15 57 0.037 Mm B Tma-FWC17 Tb L17 58 0.055 M07 B Tma-FWC18 Tb L18 58 0.095 M05 C TmaSC05 P SC05 60 0.118 Mm B TmaSC13 P SC13 56 0.118 M11 A TmaK01 P K01 54 0.107 M09 A TML-SMCY Ta SMCY 59.5 0.1 Ms C TML-SMCX2 Ta SMCX2 59.5 0.1 Ms C DSRYM SRY 59.5 0.1 Ms C

262

Figure B-1. Effect of the Zymo OneStep PCR Inhibitor Removal Kit treatment on PCR amplification. A) chromatograms of the PCR product from four African manatee fecal DNA isolated using the QIAmp Fast DNA Stool Mini Kit without any further PCR inhibitor removal treatment of the DNA isolates; B) same DNA isolates as in A except that the later was further treated with the Zymo OneStep PCR Inhibitor Removal to remove before PCR amplification of the microsatellite locus TmaSC13. The chromatograms were generated in GeneMarker, version 2.7.4.

263

Figure B-2. Effect of Touch-down protocol on the PCR amplification of the microsatellite locus TmaE26 of three African manatee fecal DNA samples (A, B, and C): TD=Touch-down whereby annealing temperature was set 5°C above primer Tm and decreased by 0.5°C after every cycle until reaching Tm. PCR1=Pre- amplification step, and PCR2=Second step PCR amplification. In the top lane chromatograms no TD was applied in either PCR1 or PCR2, in the second lane, TD was applied only at PCR1, in the third lane, TD was applied only at PCR2 and in the bottom lane, TD was applied at both PCR1 and PCR2. The three African manatee fecal DNA samples were isolated using the QIAmp Fast DNA Stool Mini Kit without any further PCR inhibitor removal treatment of the DNA isolates. The chromatograms were generated in GeneMarker, version 2.7.4.

264

Figure B-3. Diagram showing the experimental design of DNA isolation and nuclear PCR amplification of fecal DNA applied in this study.

265

Table B-2. List of some published articles that used fecal DNA from wild population species and the characteristics of their DNA extraction and PCR methods. Number of Fecal Amplifica- Number Templa- Number Amplifica- Allelic False amplifica- sample Extraction tion of loci te of tion dropout allele tion Authors Study objective Species size methods method typed volume cycles success rate rate repeats Effect of storage type and length and microsatellite repeat motif on amplification 9 40 (15 Zhu et al., success and Giant Phenol- microsa- touch 2017 ADO panda 255 chloroform tellites 10ng down) >80% 3% <8% 3

Wild QIAmp conventio- 0.1-1ng Mitchell et Population Chimpan- DNA stool nal and up al., 2015 structure zees 247 Mini kit multiplex 21 to 8μl 40 3-6 times Pre- amplifica- 16% Mimi Wild tion with first step Arandjelov Population central QIAmp nested 8 and 2% ic et al., estimate and chimpan- DNA stool primer in microsa- 88.6% 2nd 2011 structure zees 444 Mini kit 2nd step tellites (179/222) step 1 3 single step nested Skrbinšek, Optimization of QIAmp primer 13 et al., microsatellite Brown DNA stool amplifica- microsa- 2010 multiplex Bears 1023 Mini kit tion tellites 2μl 39 88% 16.10% 0.3 1 41% 25 for stan- 5μl the 58% for dard 5% (initial) initial both PCR standar Evaluate the and 3μl step standard and d PCR performance of of the and 35 and 34% for and 8% De Barba multiple pre- QIAmp 8 initial in in the preamplifi pre- for pre- & Waits, amplification to Brown DNA stool microsa- the 2nd 2nd cation amplifi- amplifi- 2009 standard PCR bear 33 Mini kit tellites step step PCR cation cation 3

266

Table B-2. Continued Number of Fecal Number Templa- Amplifica Allelic False amplifica- sampl Extraction Amplifica- of loci te Number -tion dropout allele tion Authors Study objective Species e size methods tion method typed volume of cycles success rate rate repeats 16,15, Western 343 19 and 30 cycle chimpanzee MG + 15 all (WC), western 229W respecti- primer gorilla (WG), G+ vely together Test the mountain 190W MG, and 30 Arandjelov efficiency of gorilla (MG), C + QIAmp WG, cycles ic et al., the two-step black&white 165B DNA stool Two-step WC and 5μl single 90%- 0.5- 2009) multiplex PCR colobus (BWC) WC Mini kit amplification BWC (initial) primers 94% 4-9% 1.72% 4 to 5 Standard and pre- 12.5% amplification 81% for stan- approach, the dard 3 to 7 for Test the 25 for the standard and standard Hedmark efficiency of initial of the 37 for 91% for 2.4% <1% PCR and & the pre- QIAmp pre- 18 the the pre- pre- for both 3 for the Ellegren, amplification DNA stool amplification microsa- 2μl and standard amplifica- amplifi- approa- second 2006 multiplex PCR Wolverine 48 Mini kit step tellites 12ul PCR tion cation ches step

multiple tube approach 12 Prigioni et Pop size (Taberlet et microsa- al., 2006 estimate European otter 187 Ns al., 1996) tellites Ns ns 41.20% ns ns at least 4 average <1% 81% for <1% for for preampli- Pre- Pream fication amplifi- plifica- vs cation tion vs Surface 25 initial 60.08% vs 1.4% 4.6% test the wash 12μl and then for for for efficiency of Three species protocol 6 initial 40 in the conventio conven- conven Piggott et preamplifica- of Australian (Banks et Two-step microsa- and second nal pcr tional tional al.,2004 tion method marsupials 10 al., 2002) amplification tellites then 2μl step method PCR PCR

267

Table B-2. Continued

Fecal Number Templa- Amplifica Allelic False Number of sample Extraction Amplifica- of loci te Number -tion dropout allele amplifica- Authors Study objective Species size methods tion method typed volume of cycles success rate rate tion repeats Optimization of microsat multiplex 107 Conventio- amplification using Tai (only QIAmp nal 9 Morin et single-step PCR chimpan- 90 DNA stool singleplex microsa- al., 2001 and sequencing zee used) Mini kit PCr tellites 2μl 45 84% 24% 2 to 7 Conventio- nal PCR 5 Gillett et DNA quantity and Right with touch microsa- al., 2008 quality whale 28 down tellites 5μl 52 80% 27% 6 to 9 100% (mtDNA), 69% (sex Twice for 5 loci), sex gene microsa- 54% to and up to 7 Conventiona tellite 90% times for Gillett, et Individual Right Qiagen l PCR with and one (microsa- microsatelli al. 2010 identification whale 118 DNeasy touch down mtDNA 5μl 52 tellite) -te Conventio- Muschett West QIAmp nal PCR one et al., Methodology Indian DNA stool with touch mtDNA 2009 testing manatee 35 Mini kit down (410bp) 4μl 56 79% 1 Díaz- Ferguson, Hunter, & Genetic Power one Guzman, composition and Antillean Water, Conventio- mtDNA 2017 connectivity manatee 20 MO BIO nal PCR (410bp) 10ng 35 15% 1

268

LIST OF REFERENCES

Abd Ellah, R. G. E. (2016). Bathymetric study of Wadi El-Rayan Lakes, Egypt. Lakes, Reservoirs and Ponds, 10(2), 110–125.

Akoi, K. (2004). Fishers and the West African manatee in the Fresco lagoon complex, CoÌ‚ted’Ivoire : common property, conflict and conservation. University of Kent at Canterbury.

Alfredsson, H. A., Oelkers, E. H., Hardarsson, B. S., Franzson, H., Gunnlaugsson, E., & Gislason, S. R. (2013). The geology and water chemistry of the Hellisheidi, SW- Iceland carbon storage site. International Journal of Greenhouse Gas Control, 12, 399–418. https://doi.org/10.1016/j.ijggc.2012.11.019

Allen, A. C., Beck, C. A., Bonde, R. K., Powell, J. A., & Gomez, N. A. (2018). Diet of the Antillean manatee (Trichechus manatus manatus) in Belize, Central America. Journal of the Marine Biological Association of the United Kingdom, 98(7), 1831– 1840. https://doi.org/10.1017/s0025315417000182

Allen, G. M. (1942). Extinct and vanishing mammals of the Western Hemisphere: with the marine species of all the oceans. American Committee for International Wild Life Protection, 11, 1–620.

Allendorf, F., & Luikart, G. (2007). Conservation and the genetics of populations. In Mammalia. Retrieved from https://www.degruyter.com/view/j/mamm.2007.71.issue- 4/mamm.2007.038/mamm.2007.038.xml

Alves-Stanley, C., Worthy, G., & Bonde, R. (2010). Feeding preferences of West Indian manatees in Florida, Belize, and Puerto Rico as indicated by stable isotope analysis. Marine Ecology Progress Series, 402, 255–267. https://doi.org/10.3354/meps08450

Ames, A. L., Van Vleet, E. S., & Sackett, W. M. (1996). The use of stable carbon isotope analysis for determining the dietary habits of the Florida manatee, Trichechus manatus latirostris. Marine Mammal Science, 12(4), 555–563.

AMMCO. (2019). Aquatic ferns: The silent invader of Lake Ossa. Retrieved September 24, 2019, from African Marine Mammal Conservation Organization website: http://www.ammco.org/index.php?rub=17&id=6

Aragones, Lemnuel, Lawler, I. R., Marsh, H., & Domning, D. P. (2012). The role of Sirenians in aquatic ecosystems. Sirenian Conservation: Issues and Strategies in Developing Countries, pp. 4–11. Retrieved from https://www.researchgate.net/publication/232275888

Aragones, L. V., Jefferson, T. A., & Marsh, H. (1997). Marine mammal survey techniques applicable in developing countries. Asian Marine Biology, 14, 15–39.

269

Arandjelovic, M., Guschanski, K., Schubert, G., Harris, T. R., Thalmann, O., Siedel, H., & Vigilant, L. (2009). Two-step multiplex polymerase chain reaction improves the speed and accuracy of genotyping using DNA from noninvasive and museum samples. Molecular Ecology Resources, 9(1), 28–36. https://doi.org/10.1111/j.1755-0998.2008.02387.x

Arandjelovic, Mimi, Head, J., Rabanal, L. I., Schubert, G., Mettke, E., Boesch, C., … Vigilant, L. (2011). Non-invasive genetic monitoring of wild central chimpanzees. PLoS ONE, 6(3), e14761. https://doi.org/10.1371/journal.pone.0014761

ARMINES / MINES ParisTech / TRANSVALOR / ECMWF - COPERNICUS. (2018). Atmosphere Monitoring Service – Monthly and yearly average Direct Normal Irradiation maps from CAMS Radiation Service over Africa. Retrieved August 18, 2019, from JADE website: http://www.soda-pro.com/help/cams-services/cams- radiation-service/download-africa-volume

Arraut, E. M., Marmontel, M., Mantovani, J. E., Novo, E. M. L. M., Macdonald, D. W., & Kenward, R. E. (2010). The lesser of two evils: seasonal migrations of Amazonian manatees in the Western Amazon. Journal of Zoology, 280(3), 247–256. https://doi.org/10.1111/j.1469-7998.2009.00655.x

Auil, N. E., Powell, J. A., Bonde, R. K., Andrewin, K., & Galves, J. (2007). Belize Conservation Programme 10 year summary to Liz Claiborne Art Ortenberg Foundation. New York.

Balloux, F., & Lugon-Moulin, N. (2002). The estimation of population differentiation with microsatellite markers. Molecular Ecology, 11(2), 155–165. https://doi.org/10.1046/j.0962-1083.2001.01436.x

Barnes, M. A., Turner, C. R., Jerde, C. L., Renshaw, M. A., Chadderton, W. L., & Lodge, D. M. (2014). Environmental conditions influence eDNA persistence in aquatic systems. Environmental Science and Technology, 48(3), 1819–1827. https://doi.org/10.1021/es404734p

Beck, C. A. (1988). Microhistological Characteristics of Selected Aquatic Plants of Florida , With Techniques for the Study of Manatee Food Habits CHARACTERISTICS OF SELECTED with Techniques for the Study of cn Fish and Wildlife Service U . S . Department of the Interior. (February).

Beck, C. A., & Clementz, M. T. (2012). Techniques for determining the food habits of sirenians (pp. 126–132). pp. 126–132. Retrieved from https://pubs.er.usgs.gov/publication/70038540

Bengtson, J. L. (1981). Ecology of manatees (Trichechus manatus) in the St. Johns River, Florida. University of Minnesota.

Bessac, H., & Villiers, A. (1948). Le lamantin du Sénégal. Nature (Paris), 3158, 188– 189.

270

Best, R. C. (1981). Foods and feeding habits of wild and captive Sirenia. Mammal Review, 11(1), 3–29. https://doi.org/10.1111/j.1365-2907.1981.tb00243.x

Best, R. C. (1982). Seasonal breeding in the Amazonian manatee, Trichechus inunguis (Mammalia: Sirenia). Biotropica, 14(1), 76–78. Retrieved from https://www.jstor.org/stable/pdf/2387764.pdf?refreqid=excelsior%3Aca61619d407a d4967fb4c3261b0418aa

Best, R. C. (1983). Apparent dry-season fasting in Amazonian manatees (Mammalia: Sirenia). BIOTROPICA, 15(1), 61–64.

Blaikie, P., & Simo, M. J. (2000). Cameroon’s environmental accords: signed, sealed, but undelivered. In E. B. Weiss & H. K. (Eds. . Jacobson (Eds.), Engaging Countries: Strengthening Compliance with International (pp. 435–448). Cambridge: MIT press.

Bonde, R. K., Aguirre, A. A., & Powell, J. (2004). Manatees as Sentinels of Marine Ecosystem Health: Are They the 2000-pound Canaries? EcoHealth, 1(3), 255–262. https://doi.org/10.1007/s10393-004-0095-5

Bonde, R. K., McGuire, P. M., & Hunter, M. E. (2008). A review of the key genetic tools to assist imperiled species conservation: analyzing West Indian manatee populations. Journal of Marine Animals and Their Ecology, 5(1), 8–19.

Boughorbel, S., Jarray, F., & El-Anbari, M. (2017). Optimal classifier for imbalanced data using Matthews Correlation Coefficient metric. PLoS ONE, 12(6). https://doi.org/10.1371/journal.pone.0177678

Bowles, E., & Trites, A. W. (2013). Faecal DNA amplification in Pacific walruses (Odobenus rosmarus divergens). Polar Biology, 36(5), 755–759. https://doi.org/10.1007/s00300-013-1296-6

Bradley, J. L., Wright, S. D., & McGuire, P. M. (1993). The Florida manatee: cytochrome b DNA sequence. Marine Mammal Science, 9(2), 197–202. https://doi.org/10.1111/j.1748-7692.1993.tb00445.x

Bray, J. R., & Curtis, J. T. (1957). An ordination of the upland forest communities of Southern Wisconsin. Ecological Monographs, 27(4), 325–349. https://doi.org/10.2307/1942268

Brazeal, J. L., Weist, T., & Sacks, B. N. (2017). Noninvasive genetic spatial capture- recapture for estimating deer population abundance. The Journal of Wildlife Management, 81(4), 629–640. https://doi.org/10.1002/jwmg.21236

Broadwater, M. H., Van Dolah, F. M., & Fire, S. E. (2018). Vulnerabilities of Marine Mammals to Harmful Algal Blooms. In Sandra E. Shumway JoAnn M. Burkholder Steve L. Morton (Ed.), Harmful Algal Blooms: A Compendium Desk Reference (1st ed., pp. 191–222). Retrieved from

271

https://books.google.com/books?hl=en&lr=&id=PXddDwAAQBAJ&oi=fnd&pg=PA19 1&dq=Vulnerabilities+of+Marine+Mammals+to+Harmful+Algal+Blooms&ots=4LjVT n9bLr&sig=fILivNNZgMImGOtPNvQqrRIGqtY

Brown, C. D., Hoyer, M. V., Bachmann, R. W., & Canfield, J. (2000). Nutrient- chlorophyll relationships: an evaluation of empirical nutrient-chlorophyll models using Florida and north-temperate lake data. Canadian Journal of Fisheries and Aquatic Sciences, 57(8), 1574–1583. https://doi.org/10.1139/f00-090

Browne, C. A. (1942). Liebig and the law of the minimum. Liebig and after Liebig. Publication, 16, 71–82.

BUCREP. (2010). Rapport final 3ème recensement général de la population et de l’habitat au Cameroun. Répartition de la population résidant dans la province du Littoral par Département et par Arrondissement/District. Yaoundé, Cameroun. Yaounde, Cameroon.

Burgess, E. A., Lanyon, J. M., Brown, J. L., Blyde, D., & Keeley, T. (2012). Diagnosing pregnancy in free-ranging dugongs using fecal progesterone metabolite concentrations and body morphometrics: A population application. General and Comparative Endocrinology, 177(1), 82–92. https://doi.org/10.1016/j.ygcen.2012.02.008

Butcher, R. W. (1993). Studies on the ecology of rivers: I. On the distribution of macrophytic vegetation in the rivers of Britain. The Journal of Ecology, 58–91.

Caffrey, A. J., Hoyer, M. V., & Canfield, D. E. (2007). Factors affecting the maximum depth of colonization by submersed macrophytes in Florida lakes. Lake and Reservoir Management, 23(3), 287–297. https://doi.org/10.1080/07438140709354017

Canfield, D. E., Langeland, K. A., Linda, S. B., & Haller, W. T. (1985). Relations between water transparency and maximum depth of macrophyte colonization in lakes. Journal of Aquatic Plant Management, 23(1), 25–28. Retrieved from http://www.apms.org/japm/vol23/v23p23.pdf

Canfield, D.E., & Hoyer, M. V. (1992). Aquatic macrophytes and their relation to the limnology of Florida lakes. University of Florida Library, SP115.

Canfield, Daniel E. (1983). Prediction of chlorophyll a concentrations in Florida Lakes: the importance of phosphorus and nitrogen. JAWRA Journal of the American Water Resources Association, 19(2), 255–262. https://doi.org/10.1111/j.1752- 1688.1983.tb05323.x

Canfield, Daniel E., Brown, C. D., Bachmann, R. W., & Hoyer, M. V. (2002). Volunteer lake monitoring: testing the reliability of data collected by the Florida LAKEWATCH Program. Lake and Reservoir Management, 18(1), 1–9. https://doi.org/10.1080/07438140209353924

272

Canfield, Daniel E., & Hodgson, L. M. (1983). Prediction of Secchi disc depths in Florida lakes: impact of algal biomass and organic color. Hydrobiologia, 99(1), 51–60. https://doi.org/10.1007/BF00013717

Canfield, Daniel E., & Hoyer, M. V. (1988a). Regional geology and the chemical and trophic state characteristics of Florida lakes. Lake and Reservoir Management, 4(1), 21–31. https://doi.org/10.1080/07438148809354375

Canfield, Daniel E., & Hoyer, M. V. (1988b). The eutrophication of Lake Okeechobee. Lake and Reservoir Management, 4(2), 91–99. https://doi.org/10.1080/07438148809354817

Canfield, Daniel E, Bachmann, R. W., & Hoyer, M. V. (2018). Long-term chlorophyll trends in Florida lakes. J. Aquat. Plant Manage, 56, 47–56. Retrieved from http://water.epa.gov/lawsregs/rulesregs/upload/

Cantanhede, A. M., Da Silva, V. M. F., Farias, I. P., Hrbek, T., Lazzarini, S. M., & Alves- Gomes, J. (2004). Phylogeography and population genetics of the endangered Amazonian manatee, Trichechus inunguis Natterer, 1883 (Mammalia, Sirenia). Molecular Ecology, 14(2), 401–413. https://doi.org/10.1111/j.1365- 294X.2004.02413.x

Carlson, R. E. (1977). A trophic state index for lakes. Limnology and Oceanography, 22(2), 361–369. https://doi.org/10.4319/lo.1977.22.2.0361

Cary, P. R., & Weerts, P. G. J. (1983). Growth of Salvinia molesta as affected by water temperature and nutrition I. Effects of nitrogen level and nitrogen compounds. Aquatic Botany, 16(2), 163–172. https://doi.org/10.1016/0304-3770(83)90091-8

Castelblanco-Martínez, D. N., Morales-Vela, B., Hernandez-Arana, H. A., & Padilla- Saldivar, J. A. (2009). Diet of the manatees (Trichechus manatus manatus) in Chetumal Bay, Mexico. Latin American Journal of Aquatic Mammals, (7), 39–46. Retrieved from http://www.lajamjournal.org/index.php/lajam/article/view/286

Chambers, P. A., & Kalff, J. (1985). Depth distribution and biomass of submersed aquatic macrophyte communities in relation to Secchi depth. Canadian Journal of Fisheries and Aquatic Sciences, 42(4), 701–709. https://doi.org/10.1139/f85-090

Chame, M. (2003). Terrestrial mammal feces: a morphometric summary and description. Memórias Do Instituto Oswaldo Cruz, 98(suppl 1), 71–94. https://doi.org/10.1590/S0074-02762003000900014

Clarke, K. R. (1993). Non-parametric multivariate analyses of changes in community structure. Austral Ecology, 18(1), 117–143. https://doi.org/10.1111/j.1442- 9993.1993.tb00438.x

Colares, I. G., & Colares, E. P. (2002a). Food Plants Eaten by Amazonian Manatees (Trichechus inunguis, Mammalia : Sirenia). Brazilian Archives of Biology and

273

Technology, 45(1), 67–72. https://doi.org/10.1590/S1516-89132002000100011

Colares, I. G., & Colares, E. P. (2002b). Food plants eaten by Amazonian manatees (Trichechus inunguis, Mammalia: Sirenia). Brazilian Archives of Biology and Technology, 45(1), 67–72. https://doi.org/10.1590/S1516-89132002000100011

Colmenero-Rolón, L. C. (1986). Aspectos de la ecología y comportamiento de una colonia de manatíes (Trichechus manatus) en el municipio de Emiliano Zapata, Tabasco. In Anales del Instituto de Biología. UNAM, Serie Zoología, 56, 589–602.

Construction Review Online. (2019). Republic of Niger signs US $129m deal for support Kandadji dam project. Retrieved October 30, 2019, from https://constructionreviewonline.com/2019/06/republic-of-niger-signs-us-129m-deal- for-support-kandadji-dam-project/

Convention on Migratory Species (CMS). (2012a). Action Plan for the Conservation of the West African Manatee. Retrieved from www.iucnredlist.org

Convention on Migratory Species (CMS). (2012b). Conserving Cetaceans and Manatees in the Western African Region. Retrieved from www.ketosecology.co.uk

Cornuet, J. M., & Luikart, G. (1996). Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics, 144(4), 2001–2014. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8978083

Courbis, S., & Worthy, J. A. J. (2003). Opportunistic carnivory by Florida manatees (Trichechus manatus latirostris). Aquatic Mammals, 29, 104–107. https://doi.org/10.1578/016754203101023942

Cox, C. B., & Moore, P. D. (2000). Biogeography : An Ecological and Evolutionary Approach (6th ed.). Blackwell Science.

Cozad, L. W. (2017). Giant salvinia, Salvinia molesta (Salviniaceae): evaluation of sub- optimum temperatures on survival of the giant salvinia weevil, Cyrtobagous salviniae (Coleoptera: Curculionidae) and integration of management practices with aquatic herbicid (Louisiana State University). Retrieved from https://digitalcommons.lsu.edu/gradschool_theses

Crandall, K. A., Bininda-Emonds, O. R. R., Mace, G. M., & Wayne, R. K. (2000, July 1). Considering evolutionary processes in conservation biology. Trends in Ecology and Evolution, Vol. 15, pp. 290–295. https://doi.org/10.1016/S0169-5347(00)01876-0

Davis, M. C. (2014). Relatedness, inbreeding, and kinship of the endangered Florida manatee (Trichechus manatus latirostris). University of Florida.

De Barba, M., & Waits, L. P. (2009). Multiplex pre-amplification for noninvasive genetic sampling: is the extra effort worth it? Molecular Ecology Resources, 10(4), 659– 665. https://doi.org/10.1111/j.1755-0998.2009.02818.x

274

de Meirelles, A. C. O. (2008). Mortality of the Antillean manatee, Trichechus manatus manatus, in Ceará State, north-eastern Brazil. Journal of the Marine Biological Association of the United Kingdom, 88(6), 1133–1137. https://doi.org/10.1017/S0025315408000817

Dekeyser, P. L. (1955). Les mammifères de l’Afrique noire française. Dakar, Senegal: IFAN.

Deutsch, C. J., Reid, J. P., Bonde, R. K., Easton, D. E., Kochman, H. I., & O’Shea, T. J. (2003). Seasonal movements, migratory, behavior, and site fidelity of West Indian manatees along the Atlantic coast of the United States. The Wildlife Society, 1, 1– 77. Retrieved from https://www.jstor.org/stable/pdf/3830830.pdf?refreqid=excelsior%3A36d6301966d9 c37a0f8534278a962801

Devi Prasad, A. G., & Siddaraju, P. (2012). Carlson’s trophic state index for the assessment of trophic status of two lakes in Mandya district. Advances in Applied Science Research, 3(5), 2992–2996. Retrieved from www.pelagiaresearchlibrary.com

Díaz-Ferguson, E., Hunter, M., Guzman, H. M., & Guzmán, H. M. (2017). Genetic composition and connectivity of the Antillean manatee (Trichechus manatus manatus) in Panama. Aquatic Mammals, 43(4), 378–386. https://doi.org/10.1578/AM.43.4.2017.378

Do, C., Waples, R. S., Peel, D., Macbeth, G. M., Tillett, B. J., & Ovenden, J. R. (2014). NeEstimator v2: re-implementation of software for the estimation of contemporary effective population size (Ne) from genetic data. Molecular Ecology Resources, 14(1), 209–214. https://doi.org/10.1111/1755-0998.12157

Dodman, T., Ndiaye, M. D. D., & Sarr, K. (eds. . (eds). (2008). Conservation Strategy for the West African Manatee. Retrieved from http://archive.wetlands.org/Portals/0/Africa docs/Conservation strategy for the west african manatee_optimized.pdf

Domning, D. P. (1994). A Phylogenetic analysis of the Sirenia. In D. T. eds Whitmore FC Jr, Berta A (Ed.), Contributions in Marine Mammal Paleontology Honoring Proceedings of San Diego Society of Natural History (pp. 177–189.).

Domning, D. P., & Hayek, L. A. C. (1986). Interspecific and intraspecific morphological variation in manatees (Sirenia: Trichechus). Marine Mammal Science, 2(2), 87– 144. Retrieved from https://uflib.illiad.oclc.org/illiad/FUG/illiad.dll?Action=10&Form=75&Value=1079308

Domning, Daryl P. (2005). Fossil sirenia of the west Atlantic and Caribbean region. VII. Pleistocene Trichechus manatus Linnaeus, 1758. Journal of Vertebrate Paleontology, 25(3), 685–701. https://doi.org/10.1671/0272- 4634(2005)025[0685:FSOTWA]2.0.CO;2

275

Domning, Daryl P., & Hayek, L.-A. C. (1986). Interspecific and Intraspecific Morphological Variation in Manatees (Sirenia: Trichechus). Marine Mammal Science, 2(2), 87–144. https://doi.org/10.1111/j.1748-7692.1986.tb00034.x

Don, R. H., Cox, P. T., Wainwright, B. J., Baker, K., & Mattick, J. S. (1991). “Touchdown” PCR to circumvent spurious priming during gene amplification. Nucleic Acids Research, 19(14), 4008. https://doi.org/10.1093/nar/19.14.4008

Dubreuil, P., Guiscafré, J., Nouvelot, J. F., & Olivry, J. C. (1975). Le bassin de la riviere Sanaga.

Earl, D. A., & vonHoldt, B. M. (2012). STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources, 4(2), 359–361. https://doi.org/10.1007/s12686- 011-9548-7

Ed, V. (2015). Eremospatha macrocarpa. Retrieved June 8, 2019, from http://www.palmpedia.net/wiki/Eremospatha_macrocarpa

Edgar, R. C. (2004). MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research, 32(5), 1792–1797. https://doi.org/10.1093/nar/gkh340

Eichmiller, J. J., Best, S. E., & Sorensen, P. W. (2016). Effects of temperature and trophic state on degradation of environmental DNA in lake water. Environmental Science & Technology, 50(4), 1859–1867. https://doi.org/10.1021/acs.est.5b05672

Emlen, J. M. (1966). The role of time and energy in food preference. The American Naturalist, 100(916), 611–617. https://doi.org/10.1086/282455

Epule, T. E. (2014). Deforestation in Cameroon : causes and effects on cereal yields, population vulnerability and response through reforestation and other policies.

Espinosa, M. I., Bertin, A., Squeo, F.Cortés, A., Gouin, N., Squeo, F. A., Cortés, A., & Gouin, N. (2015). Comparison of DNA extraction methods for polymerase chain reaction amplification of guanaco (Lama guanicoe) fecal DNA samples. Genetics and Molecular Research, 14(1), 400–406. https://doi.org/10.4238/2015.January.23.13

Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of clusters of individuals using the software structure: a simulation study. Molecular Ecology, 14(8), 2611–2620. https://doi.org/10.1111/j.1365-294X.2005.02553.x

Excoffier, L., & Lischer, H. E. L. (2010). Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources, 10(3), 564–567. https://doi.org/10.1111/j.1755- 0998.2010.02847.x

276

Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters , 27(8), 861–874.

Feka, N. Z., Chuyong, G. B., & Ajonina, G. N. (2009). Sustainable utilization of mangroves using improved fish-smoking systems: a management perspective from the Douala-Edea Wildlife Reserve, Cameroon. Tropical Conservation Science, 2(4), 450–468. https://doi.org/10.1177/194008290900200406

Ferchaud, A. L., Perrier, C., April, J., Hernandez, C., Dionne, M., & Bernatchez, L. (2016). Making sense of the relationships between Ne, Nb and Nc towards defining conservation thresholds in Atlantic salmon (Salmo salar). Heredity, 117(4), 268– 278. https://doi.org/10.1038/hdy.2016.62

Ferrini, L., & Benavides, L. (2018). The Fomi Dam // Potential for a Project of Shared Interest | News | Nexus - The Water, Energy & Food Security Resource Platform. Retrieved October 30, 2019, from https://www.water-energy- food.org/resources/resources-detail/the-fomi-dam-potential-for-a-project-of-shared- interest/

Flores, D., & Carlson, J. W. (2006). Biological control of giant salvinia in east Texas waterways and the impact on dissolved oxygen levels. J. Aquat. Plant Manage, 44, 115–121. Retrieved from http://www.apms.org/japm/vol44/v44p115.pdf

Food and Agriculture Organization of the United Nations (FAO). (2010). Global forest resources assessment 2010: main report. Food and Agriculture Organization of the United Nations. In Forestry Paper (Vol. 147). https://doi.org/ISBN 978-92-5-106654- 6

Foote, A. D., Thomsen, P. F., Sveegaard, S., Wahlberg, M., Kielgast, J., Kyhn, L. A., … Gilbert, M. T. P. (2012). Investigating the potential use of environmental DNA (eDNA) for genetic monitoring of marine mammals. PLoS ONE, 7(8), e41781. https://doi.org/10.1371/journal.pone.0041781

Foran, D. R. (2006). Relative degradation of nuclear and mitochondrial DNA: an experimental approach. Journal of Forensic Sciences, 51(4), 766–770. https://doi.org/10.1111/j.1556-4029.2006.00176.x

Frank, J. J. (2015). Preliminary Examination of Brevetoxin Exposure in West Indian Manatees (Trichechus manatus) Stranded in the Northern Gulf Of Mexico (University of Florida). Retrieved from https://search.proquest.com/docview/1747453509?pq-origsite=gscholar

Frankham, R, Ballou, J., & Briscoe, D. (2004). A Primer of Conservation Genetics. Cambridge University Press.

Frankham, R, Briscoe, D., & Ballou, J. (2002). Introduction to conservation genetics. Cambridge university press.

277

Frankham, Richard, Bradshaw, C. J., & Brook, B. W. (2014). Genetics in conservation management: revised recommendations for the 50/500 rules, Red List criteria and population viability analyses. https://doi.org/10.1016/j.biocon.2013.12.036

Fu, Y. X. (1997). Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics, 147(2), 915–925. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9335623

Futuyma, D. J. (2013). Evolution (3rd editio). Sunderland, MA.

Galloway, J. N., Dentener, F. J., Capone, D. G., Boyer, E. W., Howarth, R. W., Seitzinger, S. P., … Vörösmarty, C. J. (2004). Nitrogen cycles: past, present, and future. Biogeochemistry, 70(2), 153–226. https://doi.org/10.1007/s10533-004-0370- 0

Gannon, J. G., Scolardi, K. M., Reynolds, J. E., Koelsch, J. K., & Kessenich, T. J. (2007). Habitat selection by manatees in Sarasota Bay, Florida. Marine Mammal Science, 23(1), 133–143. https://doi.org/10.1111/j.1748-7692.2006.00096.x

Garcia-rodriguez, A. I. (2000a). Genetic studies of the West Indian manatee (Trichechus manatus). University of Florida.

Garcia-rodriguez, A. I. (2000b). Genetic studies of the West Indian manatee (Trichechus manatus). University of Florida.

Garcia-Rodriguez, A. I., Bowen, B. W., Domning, D., Mignucci-Giannoni, A. A., Marmontel, M., Montoya-Ospina, R. A., … Mcguire, P. M. (1998). Phylogeography of the West Indian manatee (Trichechus manatus): how many populations and how many taxa? Molecular Ecology, 7(9), 1137–1149. https://doi.org/10.1046/j.1365- 294x.1998.00430.x

Garcia‐Rodriguez, A. I., Moraga‐Amador, D., Farmerie, W. M., & King, T. L. (2000). Isolation and characterization of microsatellite DNA markers in the Florida manatee (Trichechus manatus latirostris) and their application in selected sirenian species. Molecular Ecology, 9(12), 2161-2163. Retrieved from https://onlinelibrary.wiley.com/doi/pdf/10.1046/j.1365-294X.2000.10534.x

Garza, J. C., & Williamson, E. G. (2001). Detection of reduction in population size using data from microsatellite loci. Molecular Ecology, 10(2), 305–318. https://doi.org/10.1046/j.1365-294x.2001.01190.x

Gillett, R. M., White, B. N., & Rolland, R. M. (2008). Quantification and genetic profiling of DNA isolated from free-floating feces of the North Atlantic right whale (Eubalaena glacialis). Marine Mammal Science, 24(2), 341–355. https://doi.org/10.1111/j.1748-7692.2008.00192.x

Gillett, Roxanne M., Frasier, T. R., Rolland, R. M., & White, B. N. (2010). Molecular identification of individual North Atlantic right whales (Eubalaena glacialis) using

278

free-floating feces. Marine Mammal Science, 26(4), 917–936. https://doi.org/10.1111/j.1748-7692.2010.00380.x

Giresse, P., Maley, J., & Kossoni, A. (2005). Sedimentary environmental changes and millennial climatic variability in a tropical shallow lake (Lake Ossa, Cameroon) during the Holocene. Palaeogeography, Palaeoclimatology, Palaeoecology, 218(3– 4), 257–285. https://doi.org/10.1016/j.palaeo.2004.12.018

Goodman, S. J. (1997). RST Calc: a collection of computer programs for calculating estimates of genetic differentiation from microsatellite data and determining their significance. Molecular Ecology, 6(9), 881–885. https://doi.org/10.1046/j.1365- 294X.1997.00260.x

Green, D. M. (2005, December). Designatable units for status assessment of endangered species. Conservation Biology, Vol. 19, pp. 1813–1820. https://doi.org/10.1111/j.1523-1739.2005.00284.x

Green, M. L., Herzing, D. L., & Baldwin, J. (2007). Noninvasive methodology for the sampling and extraction of DNA from free-ranging Atlantic spotted dolphins (Stenella frontalis). Molecular Ecology Notes, 7(1287–1292). https://doi.org/10.1111/j.1471-8286.2007.01858.x

Grigione, M. M. (1996). Observations on the status and distribution of the West African manatee in Cameroon. African Journal of Ecology, 34(2), 189–195. https://doi.org/10.1111/j.1365-2028.1996.tb00612.x

Guo, C., McDowell, I. C., Nodzenski, M., Scholtens, D. M., Allen, A. S., Lowe, W. L., & Reddy, T. E. (2017). Transversions have larger regulatory effects than transitions. BMC Genomics, 18(1), 1. https://doi.org/10.1186/s12864-017-3785-4

Guterres-Pazin, M. G., Marmontel, M., Rosas, F. C. W., Pazin, V. F. V., & Venticinque, E. M. (2014). Feeding ecology of the Amazonian manatee (Trichechus inunguis) in the mamirauá and amanã sustainable development reserves, Brazil. Aquatic Mammals, 40(2), 139–149. https://doi.org/10.1578/AM.40.2.2014.139

Håkanson, L. (2005). The importance of lake morphometry for the structure and function of lakes. International Review of Hydrobiology, 90(4), 433–461. https://doi.org/10.1002/iroh.200410775

Hammer, O., Harper, D. A. T., & Ryan, P. D. (2001). PAST:Paleontological Statistics software package for education and data analysis. Palaeontologia Electronica, 4(1), 9pp. Retrieved from https://folk.uio.no/ohammer/past/

Harthoorn, A. M., Van Der Walt, K., & Young, E. (1974). Possible therapy for capture myopathy in captured wild animals. Nature, 247(5442), 577–577. https://doi.org/10.1038/247577a0

Hartman, D. (1979). Ecology and Behavior of the Manatee (Trichechus manatus) in

279

Florida. American Society of Mammalogists, Special Pu, 28–35.

Hatt, R. T. (1934). A manatee collected by the American Museum Congo Expedition, with observations on the recent manatees. Bulletin of the American Museum of Natural History, 66, 533–566.

Heberle, H., Meirelles, G. V., da Silva, F. R., Telles, G. P., & Minghim, R. (2015). InteractiVenn: a web-based tool for the analysis of sets through Venn diagrams. BMC Bioinformatics.

Hedmark, E., & Ellegren, H. (2006). A test of the multiplex pre-amplification approach in microsatellite genotyping of wolverine faecal DNA. Conservation Genetics, 7(2), 289–293. https://doi.org/10.1007/s10592-005-9000-1

Hema, E. M. E. M., Di Vittorio, M., Luiselli, L., Guenda, W., Petrozzi, F., Luiselli, L., & Guenda, W. (2017). First assessment of age and sex structures of elephants by using dung size analysis in a West African savannah. European Journal of Ecology, 3(1), 1–8. https://doi.org/10.1515/eje-2017-0001

Höss, M., Kohn, M., Pääbo, S., Knauer, F., & Schroder, W. (1992). Excrement analysis by PCR. Nature, 359(6392), 199.

Hoyer, M. V., & Canfield, D. E. J. (1996). Largemouth bass abundance and aquatic vegetation in Florida lakes: an empirical analysis. J. Aquat. Plant Manage., 34, 23– 32.

Hoyer, M. V., Frazer, T. K., Notestein, S. K., & Canfield, D. E. (2002). Nutrient, chlorophyll, and water clarity relationships in Florida’s nearshore coastal waters with comparisons to freshwater lakes. Canadian Journal of Fisheries and Aquatic Sciences, 59(6), 1024–1031. https://doi.org/10.1139/f02-077

Hoyer, M. V., Terrell, J. B., & Canfield, D. E. J. (2015). Chain of eutrophication models for assessing the potential impact of nutrient enrichment on Choctawhatchee Bay, Florida, USA. Florida Scientist, 78, 20–36. https://doi.org/10.2307/24321840

Hoyer, M. V., Wellendorf, N., Frydenborg, R., Bartlett, D., & Canfield, D. E. (2012). A comparison between professionally (Florida department of environmental protection) and volunteer (Florida LAKEWATCH) collected trophic state chemistry data in Florida. Lake and Reservoir Management, 28(4), 277–281. https://doi.org/10.1080/07438141.2012.736016

Hunter, M. E., Auil-Gomez, N. E., Tucker, K. P., Bonde, R. K., Powell, J., & McGuire, P. M. (2010). Low genetic variation and evidence of limited dispersal in the regionally important Belize manatee. Animal Conservation, 13(6), 592–602. https://doi.org/10.1111/j.1469-1795.2010.00383.x

Hunter, M. K., Broderick, D., Ovenden, J. R., Tucker, K. P., Bonde, R. K., Mcguire, P. M., & Lanyon, J. M. (2010). Characterization of highly informative cross-species

280

microsatellite panels for the Australian dugong (Dugong dugon) and Florida manatee (Trichechus manatus latirostris) including five novel primers. Molecular Ecology Resources, 10(2), 368–377. https://doi.org/10.1111/j.1755- 0998.2009.02761.x

Hunter, M E, Meigs-Friend, G., Ferrante, J. A., Takoukam Kamla, A., Dorazio, R. M., Keith-Diagne, L., … Reid, J. P. (2018). Surveys of environmental DNA (eDNA): a new approach to estimate occurrence in Vulnerable manatee populations. Endangered Species Research, 35, 101–111. https://doi.org/10.3354/esr00880

Hunter, Margaret E., Mignucci-Giannoni, A. A., Tucker, K. P., King, T. L., Bonde, R. K., Gray, B. A., & McGuire, P. M. (2012). Puerto Rico and Florida manatees represent genetically distinct groups. Conservation Genetics, 13(6), 1623–1635. https://doi.org/10.1007/s10592-012-0414-2

Hurst, L. A., & Beck, C. A. (1988). Microhistological characteristics of selected aquatic plants of Florida, with techniques for the study of manatee food habits. In U.S. Fish and Wildlife Service Biological Report 88(18). Retrieved from https://apps.dtic.mil/docs/citations/ADA323036

Husar, S. L. (1978). Trichechus senegalensis. Mammalian Species, (89), 1. https://doi.org/10.2307/3503790

Jachmann, H., & Bell, R. H. V. (1984). The use of elephant droppings in assessing numbers, occupance and age structure: a refinement of the method. African Journal of Ecology, 22(2), 127–141. https://doi.org/10.1111/j.1365- 2028.1984.tb00686.x

Keith-Diagne, L. (2014). Phylogenetics and feeding ecology of the African manatee, Trichechus senegalensis. University of Florida.

Keith-Diagne, L. (2015). Trichechus senegalensis (errata version published in 2016). Retrieved July 5, 2019, from The IUCN Red List of Threatened Species 2015: e.T22104A97168578 website: http://dx.doi.org/10.2305/IUCN.UK.2015- 4.RLTS.T22104A81904980.en

Kellogg, M. E., Burkett, S., Dennis, T. R., Stone, G., Gray, B. A., McGuire, P. M., … Stanyon, R. (2007). Chromosome painting in the manatee supports Afrotheria and Paenungulata. BMC Evolutionary Biology, 7(1), 6. https://doi.org/10.1186/1471- 2148-7-6

Kindt, R., & Coe, R. (2005). Tree diversity analysis. A manual and software for common statistical methods for ecological and biodiversity studies. In World Agroforestry Centre. https://doi.org/10.1198/tas.2008.s264

Kingsleychenikwi, C., Wang, X., Kingsleychenikwi, C., & Wang, X. (2018). The Hub Status and Transportation Network of Kribi Port. Open Journal of Applied Sciences, 08(06), 226–270. https://doi.org/10.4236/ojapps.2018.86019

281

Kocher, T. D., Thomas, W. K., Meyer, A., Edwards, S. V, Pääbo, S., Villablanca, F. X., & Wilson, A. C. (1989). Dynamics of mitochondrial DNA evolution in animals: amplification and sequencing with conserved primers. Proceedings of the National Academy of Sciences of the United States of America, 86(16), 6196–6200. https://doi.org/10.1073/pnas.86.16.6196

Korkosse, O. (2018). Seuil de Kourouba : Fonctions et avantages – MALI 24 INFO. Retrieved October 31, 2019, from http://mali24.info/seuil-de-kourouba-fonctions-et- avantages/

Krimsky, L., Phlips, E., & Havens, K. (2018). A Response to frequently asked questions about the 2018 algae blooms in Lake Okeechobee, the Caloosahatchee, and St. Lucie Estuaries. Retrieved from http://edis.ifas.ufl.edu.

Kruskal, W. H., & Wallis, W. A. (1952). Use of ranks in one-criterion variance analysis. Journal of the American Statistical Association, 47(260), 583. https://doi.org/10.2307/2280779

Kwan, D. (2002). Towards a sustainable indigenous fishery for dugongs in Torres Strait : a contribution of empirical data analysis and process. (James Cook University, Townsville, Australia). Retrieved from https://researchonline.jcu.edu.au/1376/

LAKEWATCH. (2000). A Beginner’s Guide to Water Management: Nutrients. Florida.

LAKEWATCH. (2007a). A Beginner’s Guide to Water Management: Aquatic Plants in Florida. Florida.

LAKEWATCH. (2007b). A Beginner’s Guide to Water Management: Fish Communities and Trophic State in Florida Lakes. Florida.

Lanyon, J. M., Sneath, H. L., Ovenden, J. R., Broderick, D., & Bonde, R. K. (2009). Sexing sirenians: validation of visual and molecular sex determination in both wild dugongs (Dugong dugon) and Florida manatees (Trichechus manatus latirostris). Aquatic Mammals, 35(2), 187–192. https://doi.org/10.1578/AM.35.2.2009.187

Lee, R. Y., Seitzinger, S., & Mayorga, E. (2016). Land-based nutrient loading to LMEs: a global watershed perspective on magnitudes and sources. Environmental Development, 17, 220–229. https://doi.org/10.1016/j.envdev.2015.09.006

Leopold, A. (1933). Game Management. Wisconsin, Madison: The University of Wisconsin.

Librado, P., & Rozas, J. (2009). DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics, 25(11), 1451–1452. https://doi.org/10.1093/bioinformatics/btp187

Lima, C. S., Magalhães, R. F., Marmontel, M., Meirelles, A. C., Carvalho, V. L., Lavergne, A., … Santos, F. R. (2019). A hybrid swarm of manatees along the

282

Guianas coastline, a peculiar environment under the influence of the Amazon River plume. Anais Da Academia Brasileira de Ciências, 91(suppl 3). https://doi.org/10.1590/0001-3765201920190325

Loaiciga, H. A., Valdes, J. B., Vogel, R., Garvey, J., & Schwarz, H. (1996). Global warming and the hydrologic cycle. Journal of Hydrology, 174(1–2), 83–127. https://doi.org/10.1016/0022-1694(95)02753-X

Long, R., MacKay, P., Ray, J., & Zielinski, W. (2012). Noninvasive survey methods for carnivores (Island Press, Ed.). Retrieved from https://books.google.com/books?hl=en&lr=&id=BYnTI87OwAAC&oi=fnd&pg=PR7& dq=Noninvasive+survey+methods+for+North+American+carnivores&ots=GXuA5P7 fI0&sig=qn24jWnZ122XqYsV-grjduN1sR4

Lukacs, P. M., & Burnham, K. P. (2005). Review of capture-recapture methods applicable to noninvasive genetic sampling. Molecular Ecology, 14(13), 3909–3919. https://doi.org/10.1111/j.1365-294X.2005.02717.x

Luna, F. de O. (2013). Population genetics and conservation strategies for the West Indian manatee (Trichechus manatus Linnaeus, 1758) in Brazil. Universidade Federal De Pernambuco Centro De Tecnologia E Geociências.

MacArthur, R. H., & Pianka, E. R. (1966). On optimal use of a patchy environment. The American Naturalist, 100(916), 603–609. https://doi.org/10.1086/282454

Marmontel, M., Guterres, M. G., Meirelles, A. C. O., Calvimontes, J., & Rosas, F. C. W. (2002). Lago Amanã: destino estival de manaties Amazónicos en la Amazônia Occidental Brasileña. In X Reunión de Trabajo de Especialistas En Mamíferos Acuáticos y IV Congreso de La Sociedad Latinoamericana de Especialistas Em Mamíferos Acuáticos.

Marsh, H., Channells, P. W., Heinsohn, G. E., & Morrissey, J. (1982). Analysis of Stomach Contents of Dugongs from Queensland. In Wildl. Res (Vol. 9). Retrieved from http://www.publish.csiro.au/wr/pdf/WR9820055

Marsh, H., O’Shea, T. J., & Reynolds III, J. E. (2011). Ecology and Conservation of the Sirenia: Dugongs and Manatees. New York: Cambridge University Press.

Martin, G. D., Coetzee, J. ., Weyl, P. S. R., Parkinson, M. C., & Hill, M. P. (2018). Biological control of Salvinia molesta in South Africa revisited. Biological Control, 125, 74–80. https://doi.org/10.1016/J.BIOCONTROL.2018.06.011

Matlock, B. (2015). Assessment of Nucleic Acid Purity. Retrieved September 19, 2019, from Thermoscientific website: https://assets.thermofisher.com/TFS- Assets/CAD/Product-Bulletins/TN52646-E-0215M-NucleicAcid.pdf

Matthews, B. W. (1975). Comparison of the predicted and observed secondary structure of T4 phage lysozyme. BBA - Protein Structure, 405(2), 442–451.

283

https://doi.org/10.1016/0005-2795(75)90109-9

Mayaka, T. B., Awah, H. C., & Ajonina, G. (2013). Conservation status of manatee (Trichechus Senegalensis Link 1795) in Lower Sanaga basin, Cameroon: an ethnobiological assessment. Tropical Conservation Science, 6(4), 521–538. https://doi.org/10.1177/194008291300600406

Mayaka, T. B., Koh‐Dimbot, J. P., & Keith‐Diagne, L. W. (2019). Occurrence patterns of African manatees, conflicts with humans, and local perception in the Southern Korup Area, Cameroon. Aquatic Conservation: Marine and Freshwater Ecosystems, 1–13. https://doi.org/10.1002/aqc.3210

Mayaka, T. B., Takoukam Kamla, A., & Self-Sullivan, C. (2015). Using pooled local expert opinions (PLEO) to discern patterns in sightings of live and dead Manatees (Trichechus senegalensis, Link 1785) in Lower Sanaga Basin, Cameroon. PLOS ONE, 10(7), e0128579. https://doi.org/10.1371/journal.pone.0128579

McClenaghan, L. R., & O’Shea, T. J. (1988). Genetic variability in the Florida manatee (Trichechus manatus). Journal of Mammalogy, 69(3), 481–488. https://doi.org/10.2307/1381339

McDonald, B. J. (2005). Population genetics of dugongs around Australia : implications of gene flow and migration (PhD thesis, James Cook University.). Retrieved from https://researchonline.jcu.edu.au/1327/

Mchale, M., Broderick, D., Ovenden, J. R., & Lanyon, J. M. (2008). A PCR assay for gender assignment in dugong (Dugong dugon) and West Indian manatee (Trichechus manatus). Molecular Ecology Resources, 8(3), 669–670. https://doi.org/10.1111/j.1471-8286.2007.02041.x

MetOceanView. (2019). Hindcast-West Africa. Retrieved November 2, 2019, from https://app.metoceanview.com/hindcast/

Mignucci-Giannoni, A. A., & Beck, C. A. (1998). The diet of the manatee (Trichechus manatus) in Puerto Rico. Marine Mammal Science, 14(2), 394–397. https://doi.org/10.1111/j.1748-7692.1998.tb00733.x

Mignucci-Giannoni, A. A., Montoya-Ospina, R. A., Jimenez-Marrero, N. M., Rodríguez- López, M. A., Williams, J. E. H., & Bonde, R. K. (2000). Manatee mortality in Puerto Rico. Environmental Management, 25(2), 189–198. https://doi.org/10.1007/s002679910015

Mitchell, D. S., & Tur, N. M. (1975). The rate of growth of Salvinia molesta (S. Auriculata Auct.) in laboratory and natural conditions. Journal of Applied Ecology, 12(1), 213– 225. Retrieved from https://www.jstor.org/stable/pdf/2401730.pdf?refreqid=excelsior%3A6a4ada2b2feb 9d2d857a89585ba2f5ac

284

Mitchell, M. W., Locatelli, S., Ghobrial, L., Pokempner, A. A., Sesink Clee, P. R., Abwe, E. E., … Gonder, M. (2015). The population genetics of wild chimpanzees in Cameroon and Nigeria suggests a positive role for selection in the evolution of chimpanzee subspecies. BMC Evolutionary Biology, 15(1), 3. https://doi.org/10.1186/s12862-014-0276-y

Moore, K. J., & Jung, H.-J. G. (2001). Lignin and fiber digestion. Journal of Range Management, 54(4), 420–430. https://doi.org/10.2458/azu_jrm_v54i4_moore

Moozhiyil, M., & Pallauf, J. (1986). Chemical composition of the water fern, Salvinia molesta, and its potential as feed source for ruminants. Economic Botany, 40(3), 375–383. https://doi.org/10.1007/BF02858995

Morin, P. A., Chambers, K. E., Boesch, C., & Vigilant, L. (2001). Quantitative polymerase chain reaction analysis of DNA from noninvasive samples for accurate microsatellite genotyping of wild chimpanzees (Pan troglodytes verus). Molecular Ecology, 10(7), 1835–1844. https://doi.org/10.1046/j.0962-1083.2001.01308.x

Morin, Phillip A., & Woodruff, D. S. (1996). Non-invasive Genotyping for Vertebrate Conservation. In R. K. Wayne & T. B. Smith (Eds.), Molecular Genetic Approaches in Conservation (pp. 298–313). Retrieved from http://labs.biology.ucsd.edu/woodruff/pubs/155.pdf

Moritz, C. (1994). Defining “Evolutionarily Significant Units” for conservation. Trends in Ecology and Evolution, 9(10), 373–375. https://doi.org/10.1016/0169- 5347(94)90057-4

Morrison, M. L., Marcot, B., & Mannan, W. (2006). Wildlife-Habitat Relationships: Concepts and Applications (3rd ed.). Connecticut, Washington D.C.: Island Press.

Morrison, T. A., Chiyo, P. I., Moss, C. J., & Alberts, S. C. (2005). Measures of dung bolus size for known-age African elephants (Loxodonta africana): implications for age estimation. Journal of Zoology, 266(1), 89–94. https://doi.org/10.1017/S0952836905006631

Munguia, P., & Ojanguren, A. F. (2015). Bridging the gap in marine and terrestrial studies. Ecosphere, 6(2). https://doi.org/10.1890/ES14-00231.1

Murphy, W. J., Eizirik, E., Johnson, W. E., Zhang, Y. P., Ryder, O. A., & O’Brien, S. J. (2001). Molecular phylogenetics and the origins of placental mammals. Nature, 409(6820), 614–618. https://doi.org/10.1038/35054550

Muschett, G., Bonacic, C., & Vianna, J. A. (2009). A noninvasive sampling method for genetic analysis of the West Indian manatee (Trichechus manatus). Marine Mammal Science, 25(4), 955–963. https://doi.org/10.1111/j.1748- 7692.2009.00310.x

Myers, N. (1988). Threatened biotas: “hot spots” in tropical forests. Environmentalist,

285

8(3), 187–208. Retrieved from https://link.springer.com/content/pdf/10.1007%2FBF02240252.pdf

Ndam Ngoupayou, J., Kpoumie, A., Boeglin, J., Lienou, G., Nfocgo, A., & Ekodeck, G. (2007). Transports solides et érosion mécanique dans un écosystème tropical d’Afrique : exemple du bassin versant de la Sanaga au Sud-Cameroun. 6–9.

Ndjassi, C. G. (2015). Identification des poissons consommes et etat des destruction des filets de pêche par le lamantin dans la Reserve de Faune du Lac Ossa. Institut des Sciences Halieutiques de Yabassi, Cameroon.

Nei, M. (1973). Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences of the United States of America, 70(12). https://doi.org/10.1073/pnas.70.12.3321

Newbery, D. M. C., & Gartlan, J. S. (1996). A structural analysis of rain forest at Korup and Douala-Edea, Cameroon. Proceedings of the Royal Society of Edinburgh Section B: Biological Sciences, 104, 177–224. https://doi.org/10.1017/s0269727000006138

Ngea, P. (2010). Douala – Edea Reserve: the last refuge of hope for man and nature (Part 2). Retrieved June 8, 2019, from Enviro*mend website: http://enviromend.blogspot.com/

Ngueguim, R. J., Betti, L. J., Dicka Kwambe, E., Momo, S. M. C., & Temgoua, F. L. (2017). Deforestation, biodiversity and biomass losses in Kribi deep sea port area (Cameroon): some mitigating measures. Journal of Ecology and The Natural Environment, 9(6), 87–98. https://doi.org/10.5897/jene2016.0617

Nguetsop, V. F., Servant-Vildary, S., & Servant, M. (2004). Late Holocene climatic changes in West Africa, a high resolution diatom record from equatorial Cameroon. Quaternary Science Reviews, 23(5–6), 591–609. https://doi.org/10.1016/j.quascirev.2003.10.007

Nishiwaki, M., Yamaguchi, M., Shokita, S., Uchida, S., & Kataoka, T. (1982). Recent survey on the distribution of the African manatee. Scientific Report of the Whale Research Institute, 34, 137–147. Retrieved from https://www.icrwhale.org/pdf/SC034137-147.pdf

Njoya, K. M. R. (2015). Qualité de l’eau et structure des peuplements macrophytiques du lac Ossa à Dizangue-Cameroun. Institut des Sciences Halieutiques de Yabassi.

Nourisson, C., Morales-Vela, B., Padilla-Saldívar, J., Tucker, K. P., Clark, A., Olivera- Gómez, L. D., … McGuire, P. (2011). Evidence of two genetic clusters of manatees with low genetic diversity in Mexico and implications for their conservation. Genetica, 139(7), 833–842. https://doi.org/10.1007/s10709-011-9583-z

Nowak, K., Jones, T., & Lee, P. C. (2009). Using dung bolus diameter for age

286

estimation in an unstudied elephant population in Udzungwa mountains, Tanzania. Pachyderm, 46(1), 47–52. Retrieved from http://www.pachydermjournal.org/index.php/pachy/article/view/138/0

O’Connor, R. J. (2002). The conceptual basis of species distribution modeling: time for a paradigm shift. Predicting species occurrences: issues of accuracy and scale. Washington, DC, USA: Island Press.

O’shea, T. J. (1986). Mast foraging by West Indian manatees (Trichechus manatus). Journal of Mammalogy, 67(1), 183–185.

O’Shea, T. J., & Hartley, W. C. (1995). Reproduction and early-age survival of manatees at Blue Spring, Upper St. Johns River, Florida. In O. A. and P. (eds.) (Ed.), Population Biology of the Florida Manatee (Trichechus manatus latirostris) (pp. 157–176). Retrieved from https://pubs.er.usgs.gov/publication/70164317

Ogogo, A. U., Eniang, E. A., & Nkamenyin, O. O. (2013). Ecology and conservation of the West African manatee (Trichechus senegalensis) in Eniong Creek, South Nigeria. International Journal of Research in Applied. Natural and Social Sciences, 1(1), 19–24.

Paetkau, D., & Strobeck, C. (1994). Microsatellite analysis of genetic variation in black bear populations. Molecular Ecology, 3(5), 489–495. https://doi.org/10.1111/j.1365- 294X.1994.tb00127.x

Palsboll, P., Berube, M., & Allendorf, F. (2007). Identification of management units using population genetic data. Trends in Ecology & Evolution, 22(1), 11–16. https://doi.org/10.1016/j.tree.2006.09.003

Pardini, A. ., O’Brien, P. C. ., Fu, B., Bonde, R. ., Elder, F. F. ., Ferguson-Smith, M. ., … Robinson, T. . (2007). Chromosome painting among Proboscidea, Hyracoidea and Sirenia: support for Paenungulata (Afrotheria, Mammalia) but not Tethytheria. Proceedings of the Royal Society B: Biological Sciences, 274(1615), 1333–1340. https://doi.org/10.1098/rspb.2007.0088

Pare, M. N., Koné, D., Kengne, I. M., Dongo, K., & Foundation, M. G. (2011). Nutritional potential of Echinochloa pyramidalis (Lam.) Hitchc . & chase , a forage plant used in constructed wetlands treatment of faecal sludge and wastewater. African Journal of Agricultural Research, 6(18), 4397–4408. https://doi.org/10.5897/AJAR11.166

Parsons, K. M., Durban, J. W., & Claridge, D. E. (2003). Comparing two alternative methods for sampling small cetaceans for molecular analysis. Marine Mammal Science, 19(1), 224–231. https://doi.org/10.1111/j.1748-7692.2003.tb01104.x

Parsons, Kim M. (2001). Reliable microsatellite genotyping of dolphin DNA from faeces. Molecular Ecology Notes, 1(4), 341–344. https://doi.org/10.1046/j.1471- 8278.2001.00098.x

287

Parsons, Kim M., Durban, J. W., Claridge, D. E., Herzing, D. L., Balcomb, K. C., & Noble, L. R. (2006). Population genetic structure of coastal bottlenose dolphins (Tursiops truncatus) in the northern Bahamas. Marine Mammal Science, 22(2), 276–298. https://doi.org/10.1111/j.1748-7692.2006.00019.x

Pause, K. C., Nourisson, C., Clark, A., Kellogg, M. E., Bonde, R. K., & Mcguire, P. M. (2007). Polymorphic microsatellite DNA markers for the Florida manatee (Trichechus manatus latirostris). Molecular Ecology Notes, 7(6), 1073–1076. https://doi.org/10.1111/j.1471-8286.2007.01781.x

Peakall, R., & Smouse, P. E. (2006). Genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6(1), 288– 295. https://doi.org/10.1111/j.1471-8286.2005.01155.x

Peel, D., Ovenden, J. R., & Peel, S. L. (2004). NeEstimator: software for estimating effective population size. Queensland Government, Department of Primary Industries and Fisheries.

Pieterse, A. H., Kettunen, M., Diouf, S., Ndao, I., Sarr, K., Tarvainen, A., … Hellsten, S. (2003). Effective biological control of salvinia molesta in the Senegal River by means of the weevil Cyrtobagous salviniae. AMBIO: A Journal of the Human Environment, 32(7), 458–462. https://doi.org/10.1579/0044-7447-32.7.458

Piggott, M. P., Bellemain, E., Taberlet, P., & Taylor, A. C. (2004). A multiplex pre- amplification method that significantly improves microsatellite amplification and error rates for faecal DNA in limiting conditions. Conservation Genetics, 5(3), 417– 420. https://doi.org/10.1023/B:COGE.0000031138.67958.44

Powell, J. A. (2006). Evidence of carnivory in manatees (Trichechus manatus). Journal of Mammalogy. https://doi.org/10.2307/1379938

Powell, James A. (1978). Evidence of carnivory in manatees (Trichechus manatus). Journal of Mammalogy, 59(2), 442–442. https://doi.org/10.2307/1379938

Powell, James A. (1990). Manatees in the Bijagos Archipelago: recommendations for their conservation.

Powell, James A. (1996). The distribution and biology of the West African manatee (Trichechus senegalensis Link, 1795). Nairobi, Kenya.

Prigioni, C., Remonti, L., Balestrieri, A., Sgrosso, S., Priore, G., Mucci, N., & Randi, E. (2006). Estimation of European otter (Lutra lutra) population size by fecal DNA typing in southern Italy. Journal of Mammalogy, 87(5), 855–858. https://doi.org/10.1644/05-MAMM-A-294R1.1

Pritchard, J. K., Wen, X., & Falush, D. (2010). Documentation for Structure software: Version 2.3. Retrieved October 5, 2019, from http://burfordreiskind.com/wp- content/uploads/Structure_Manual_doc.pdf

288

Prugnolle, F., & de Meeus, T. (2002). Inferring sex-biased dispersal from population genetic tools: a review. Heredity, 88(3), 161–165. https://doi.org/10.1038/sj.hdy.6800060

Queller, D. C., & Goodnight, K. F. (1989). Estimating relatedness using genetic markers. Evolution, 43(2), 258–275. https://doi.org/10.1111/j.1558- 5646.1989.tb04226.x

Ramón-Laca, A., Soriano, L., Gleeson, D., & Godoy, J. A. (2015). A simple and effective method for obtaining mammal DNA from faeces. Wildlife Biology, 21(4), 195–203. https://doi.org/10.2981/wlb.00096

Rathbun, G. B., Reid, J. P., Bonde, R. K., & Powell, J. A. (1995). Reproduction in free- ranging Florida manatees (Trichechus manatus latirostris). In A. and P. O’Shea (Ed.), Population Biology of the Florida Manatee (Trichechus manatus latirostris) (pp. 135–156). Retrieved from https://pubs.er.usgs.gov/publication/2002239

Raymond, M., & Rousset, F. (1995). GENEPOP (Version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity, 86(3), 248–249. https://doi.org/10.1093/oxfordjournals.jhered.a111573

Raza, S., Shoaib, M. W., & Mubeen, H. (2016). Genetic markers: importance, uses and applications. International Journal of Scientific and Research Publications, 6(3), 221–223. Retrieved from https://pdfs.semanticscholar.org/7391/563bf1781b989ad1db987ad528a61f952173. pdf

Reep, R., & Bonde, R. (2006). The Florida Manatee: Biology and Conservation. Gainesville, Fl.: University Press of Florida.

Reep, R. L., Gaspard, J. C., Sarko, D., Rice, F. L., Mann, D. A., & Bauer, G. B. (2011). Manatee vibrissae: evidence for a “lateral line” function. Annals of the New York Academy of Sciences, 1225(1), 101–109. https://doi.org/10.1111/j.1749- 6632.2011.05992.x

Reet, J. G., Webb, J. A., Cousens, R. D., & Greet, J. (2011). The importance of seasonal flow timing for riparian vegetation dynamics: a systematic review using causal criteria analysis. Freshwater Biology, 56, 1231–1247. https://doi.org/10.1111/j.1365-2427.2011.02564.x

Reeves, R. R., Tuboku-Metzger, D., & Kapindi, R. A. (1988). Distribution and exploitation of manatees in Sierra Leone. Oryx, 22(2), 75–84. https://doi.org/10.1017/S0030605300027538

Reynolds III, J. E., & Odell, D. K. (1991). Manatees and Dugongs (p. 192). p. 192. New York: Facts on file. Inc.

Reynolds III, John E. (1981). Aspects of the social behaviour and herd structure of a

289

semi-isolated colony of West Indian manatees Trichechus manatus. Mammalia, 45(4), 431–452. https://doi.org/10.1515/mamm.1981.45.4.431

Reynolds, J. E., Szelistowski, W. A., & León, M. A. (1995). Status and conservation of manatees Trichechus manatus manatus in Costa Rica. Biological Conservation, 71(2), 193–196. https://doi.org/10.1016/0006-3207(94)00046-S

Rice, W. R. (1989). Analyzing tables of statistical tests. Evolution, 43(1), 223. https://doi.org/10.2307/2409177

Robin, E. D., & Wong, R. (1988). Mitochondrial DNA molecules and virtual number of mitochondria per cell in mammalian cells. Journal of Cellular Physiology, 136(3), 507–513. https://doi.org/10.1002/jcp.1041360316

Rolland, R. M., Hamilton, P. K., Kraus, S. D., Davenport, B., Gillett, R. M., Wasser, S. K., … Kraus, S. (2006). Faecal sampling using detection dogs to study reproduction and health in North Atlantic right whales (Eubalaena glacialis). J. Cetacean Res. Manage, 8(2), 121–125.

Roussel, J.-M., Paillisson, J.-M., Tr Eguier, A., & Petit, E. (2014). The downside of eDNA as a survey tool in water bodies. Journal of Applied Ecology, 52, 823–826. https://doi.org/10.1111/1365-2664.12428

Ryder, O. A. (1986). Species conservation and systematics : the dilemma of subspecies. Trends Ecol. Evol., 1, 9–10.

Sakamoto, M. (1966). Primary production by phytoplankton community in some Japanese lakes and its dependence on lake depth. Arch Hydrobiol, 62, 1–28.

Satizábal, P., Mignucci-Giannoni, A. A., Duchêne, S., Caicedo-Herrera, D., Perea- Sicchar, C. M., García-Dávila, C. R., … Caballero, S. J. (2012). Phylogeography and sex-biased dispersal across riverine manatee populations (Trichechus inunguis and Trichechus manatus) in South America. PLoS ONE, 7(12). https://doi.org/10.1371/journal.pone.0052468

Schrader, C., Schielke, A., Ellerbroek, L., & Johne, R. (2012). PCR inhibitors - occurrence, properties and removal. Journal of Applied Microbiology, 113(5), 1014– 1026. https://doi.org/10.1111/j.1365-2672.2012.05384.x

Schulz, C. J., & Childers, G. W. (2011). Fecal bacteroidales diversity and decay in response to variations in temperature and salinity. Applied and Environmental Microbiology, 77(8), 2563–2572. https://doi.org/10.1128/AEM.01473-10

Schwartz, M. K., Luikart, G., & Waples, R. S. (2007). Genetic monitoring as a promising tool for conservation and management. Trends in Ecology & Evolution, 22(1), 25– 33. https://doi.org/10.1016/J.TREE.2006.08.009

Ségurel, L., Martínez-Cruz, B., Quintana-Murci, L., Balaresque, P., Georges, M., Hegay,

290

T., … Vitalis, R. (2008). Sex-specific genetic structure and social organization in Central Asia: insights from a multi-locus study. PLoS Genetics, 4(9). https://doi.org/10.1371/journal.pgen.1000200

Seton, E. T. (1925). On the study of scatology. Journal of Mammalogy, 6(1), 47. https://doi.org/10.2307/1373469

Seymour, M., Durance, I., Cosby, B. J., Ransom-Jones, E., Deiner, K., Ormerod, S. J., … Creer, S. (2018). Acidity promotes degradation of multi-species environmental DNA in lotic mesocosms. Communications Biology, 1(1), 4. https://doi.org/10.1038/s42003-017-0005-3

Shannon, C. E. (1948). A mathematical theory of communication. The Bell System Technical Journal, 27, 379–423.

Short, F. T., Kosten, S., Morgan, P. A., Malone, S., & Moore, G. E. (2016). Impacts of climate change on submerged and emergent wetland plants. Aquatic Botany, 135, 3–17. https://doi.org/10.1016/J.AQUABOT.2016.06.006

Skrbinšek, T., Jelenčič, M., Waits, L., Kos, I., & Trontelj, P. (2010). Highly efficient multiplex PCR of noninvasive DNA does not require pre-amplification. Molecular Ecology Resources, 10(3), 495–501. https://doi.org/10.1111/j.1755- 0998.2009.02780.x

Slatkin, M. (1995). A measure of population subdivision based on microsatellite allele frequencies. Genetics, 139(1).

Smith, V. H., Tilman, G. D., & Nekola, J. C. (1999). Eutrophication: impacts of excess nutrient inputs on freshwater, marine, and terrestrial ecosystems. Environmental Pollution, 100(1–3), 179–196. https://doi.org/10.1016/S0269-7491(99)00091-3

Smith, Val H. (1982). The nitrogen and phosphorus dependence of algal biomass in lakes: an empirical and theoretical analysis. Limnology and Oceanography, 27(6), 1101–1111. https://doi.org/10.4319/lo.1982.27.6.1101

Solomon, B. D., Corey-Luse, C. M., & Halvorsen, K. E. (2004). The Florida manatee and eco-tourism: toward a safe minimum standard. Ecological Economics, 50(1–2), 101–115. https://doi.org/10.1016/J.ECOLECON.2004.03.025

Taberlet, P, Griffin, S., Goossens, B., Questiau, S., Manceau, V., Escaravage, N., … Bouvet, J. (1996). Reliable genotyping of samples with very low DNA quantities using PCR. Nucleic Acids Research, 24(16), 3189–3194. https://doi.org/10.1093/nar/24.16.3189

Taberlet, Pierre, Waits, L. P., & Luikart, G. (1999). Noninvasive genetic sampling: look before you leap. Trends in Ecology & Evolution, 14(8), 323–327. https://doi.org/10.1016/S0169-5347(99)01637-7

291

Tajima, F. (1989). Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics, 123(3), 585–595. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2513255

Takoukam Kamla, A. (2012). Activity center , habitat use and conservation of the West African manatee (Trichechus senegalensis Link , 1795). University of Dschang.

Tamura, K. (1992). Estimation of the number of nucleotide substitutions when there are strong transition-transversion and G+C-content biases. Molecular Biology and Evolution, 9(4), 678–687. https://doi.org/10.1093/oxfordjournals.molbev.a040752

Tamura, Koichiro, Stecher, G., Peterson, D., Filipski, A., & Kumar, S. (2013). MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and Evolution, 30(12), 2725–2729. https://doi.org/10.1093/molbev/mst197

The World Bank. (2019). Agricultural land (% of land area) - Cameroon | Data. Retrieved October 21, 2019, from https://data.worldbank.org/indicator/AG.LND.AGRI.ZS?locations=CM

Thomsen, P. F., Kielgast, J., Iversen, L. L., Wiuf, C., Rasmussen, M., Gilbert, M. T. P., … Willerslev, E. (2012). Monitoring endangered freshwater biodiversity using environmental DNA. Molecular Ecology, 21(11), 2565–2573. https://doi.org/10.1111/j.1365-294X.2011.05418.x

Tikel, D., Blair, D., & Marsh, H. D. (1996). Marine mammal faeces as a source of DNA. Molecular Ecology, 5(3), 456–457. https://doi.org/10.1046/j.1365- 294X.1996.00107.x

Tringali, M. D., Davis, M. C., Rodriguez-Lopez, M. A., Bolen, E. E., Sullivan, J. G., & Haubold, E. M. (2008). Simultaneous use of the X- and Y-chromosome genes SMCX, SMCY, and DBY for sex determination in the Florida manatee (Trichechus manatus latirostris). Marine Mammal Science, 24(1), 218–224. https://doi.org/10.1111/j.1748-7692.2007.00147.x

Tringali, M. D., Seyoum, S., Carney, S. L., Davis, M. C., Rodriguez-Lopez, M. A., Reynolds Iii, J. E., & Haubold, E. (2008). Eighteen new polymorphic microsatellite markers for the endangered Florida manatee, Trichechus manatus latirostris. Molecular Ecology Resources. Retrieved from http://agris.fao.org/agris- search/search.do?recordID=US201300856566

Tsukinowa, E., Karita, S., Asano, S., Wakai, Y., Oka, Y., Furuta, M., & Goto, M. (2008). Fecal microbiota of a dugong (Dugong dugong) in captivity at Toba Aquarium. Journal of General and Applied Microbiology, 54(1), 25–38. https://doi.org/10.2323/jgam.54.25

Tucker, K. P., Hunter, M. E., Bonde, R. K., Austin, J. D., Clark, A. M., Beck, C. A., … Oli, M. K. (2012). Low genetic diversity and minimal population substructure in the endangered Florida manatee: implications for conservation. Journal of Mammalogy,

292

93(6), 1504–1511. https://doi.org/10.1644/12-MAMM-A-048.1

U.S. Fish and Wildlife Service (USFWS). (1999). Multi-species recovery plan for South Florida: West Indian Manatee (Trichechus manatus). Vero Beach, Florida.

U.S. Fish and Wildlife Service (USFWS). (2001). Florida manatee recovery plan (Trichechus manatus latirostris), Third revision. Atlanta, Geogria.

US Fish and Wildlife Service (USFWS). (1999). Multi-species recovery plan for the threatened and endangered species of South Florida. Retrieved from https://www.fws.gov/verobeach/MSRPPDFs/ExecSum.pdf

US Fish and Wildlife Service (USFWS). (2001). Florida Manatee Recovery Plan, Third Revision. Retrieved from https://www.fws.gov/northflorida/manatee/Documents/Recovery Plan/MRP-start.pdf

Vallet, D., Petit, E. J., Gatti, S., Levréro, F., & Ménard, N. (2008). A new 2CTAB/PCI method improves DNA amplification success from faeces of Mediterranean (Barbary macaques) and tropical (lowland gorillas) primates. Conservation Genetics, 9(3), 677–680. https://doi.org/10.1007/s10592-007-9361-8

Valley, R. D., Johnson, M. B., Dustin, D. L., Jones, K. D., & Lauenstein, M. R. Nawrocki, J. (2015). Combining hydroacoustic and point-intercept survey methods to assess aquatic plant species abundance patterns and community dominance. Journal of Aquatic Plant Management, 53, 121–129.

Valley, Ray D. (2016). Spatial and temporal variation of aquatic plant abundance: quantifying change. J. Aquat. Plant Manage., 54, 95–101.

Valley, Ray D., Drake, M. T., & Anderson, C. S. (2005). Evaluation of alternative interpolation techniques for the mapping of remotely-sensed submersed vegetation abundance. Aquatic Botany, 81(1), 13–25. https://doi.org/10.1016/j.aquabot.2004.09.002

Van der Waarde, J. (2007). Integrated river basin management of the Sanaga River, Cameroon. Benefits and challenges of decentralised water management. Delft, The Netherlands.

Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M., & Shipley, P. (2004). Micro- Checker: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes, 4(3), 535–538. https://doi.org/10.1111/j.1471- 8286.2004.00684.x

Vant, V. N., & Davies-Colley, R. J. (1984). Factors affecting clarity of New Zealand lakes. New Zealand Journal of Marine and Freshwater Research, 18(3), 367–377. https://doi.org/10.1080/00288330.1984.9516057

Vianna, J. A., Bonde, R. K., Caballero, S., Giraldo, J. P., Lima, R. P., Clark, A., …

293

Santos, F. R. (2006). Phylogeography, phylogeny and hybridization in trichechid sirenians: Implications for manatee conservation. Molecular Ecology, 15(2), 433– 447. https://doi.org/10.1111/j.1365-294X.2005.02771.x

Vitousek, P., Mooney, H. A., Lubchenco, J., & Mellilo, J. M. (1997). Human domination of earth. Science, 227(July), 494–499. https://doi.org/10.1126/science.277.5325.494

Waits, L. P., Luikart, G., & Taberlet, P. (2001). Estimating the probability of identity among genotypes in natural populations: cautions and guidelines. Molecular Ecology, 10(1), 249–256. https://doi.org/10.1046/j.1365-294X.2001.01185.x

Waples, R. K., Larson, W. A., & Waples, R. S. (2016). Estimating contemporary effective population size in non-model species using linkage disequilibrium across thousands of loci. Heredity, 117(4), 233–240. https://doi.org/10.1038/hdy.2016.60

Waples, R. S. (1991). Genetic interactions between hatchery and wild salmonids: lessons from the Pacific Northwest. Canadian Journal of Fisheries and Aquatic Sciences, 48(Suppl.1), 124–133. https://doi.org/10.1139/f91-311

Waples, R S. (1989). A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics, 121(2), 379–391. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2731727

Waples, Robin S. (2006). A bias correction for estimates of effective population size based on linkage disequilibrium at unlinked gene loci. Conservation Genetics, 7(2), 167–184. https://doi.org/10.1007/s10592-005-9100-y

Wetzel, R. G. (2001). Limnology: Lake and River Ecosystems (Third Edit). London: Academic Press.

Wetzel, R. G., & Gopal, B. (1999). Limnology in developing countries. Volume 2. Limnology in Developing Countries. Volume 2.

Whiteman, J. B., & Room, P. M. (1991). Temperatures lethal to Salvinia molesta Mitchell. Aquatic Botany, 40(1), 27–35. https://doi.org/10.1016/0304- 3770(91)90071-C

Wiesner, R. J., Rüegg, J. C., & Morano, I. (1992). Counting target molecules by exponential polymerase chain reaction: copy number of mitochondrial DNA in rat tissues. Biochemical and Biophysical Research Communications, 183(2), 553–559. https://doi.org/10.1016/0006-291X(92)90517-O

Wilberg, M. J., & Dreher, B. P. (2004). genecap: a program for analysis of multilocus genotype data for non-invasive sampling and capture-recapture population estimation. Molecular Ecology Notes, 4(4), 783–785. https://doi.org/10.1111/j.1471- 8286.2004.00797.x

294

Wirrmann, D. (1992). Le Lac Ossa: une monographie préliminaire. Revue Géographique du Cameroun, XI.

Wirrmann, D., & Elouga, M. (1998). Lake Ossa: a new Iron Age site in the Cameroonian littoral province. Nyame Akuma. 16-26.

World Development Report 2016: Digital Dividends. (2016). Retrieved September 24, 2019, from The World Bank website: https://www.worldbank.org/en/publication/wdr2016

Wotton, R. S., & Malmqvist, B. (2001). Feces in aquatic ecosystems: feeding animals transform organic matter into fecal pellets, which sink or are transported horizontally by currents; these fluxes relocate organic matter in aquatic ecosystems. BioScience, 51(7), 537–544. https://doi.org/10.1641/0006- 3568(2001)051[0537:fiae]2.0.co;2

Wright, S. (1931). Evolution in Mendelian Populations. Genetics, 16(2).

Wright, S. (1949). The genetical structure of populations. Annals of Eugenics, 15(1), 323–354. https://doi.org/10.1111/j.1469-1809.1949.tb02451.x

Wright, S. (1965). The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution, 19(3), 395–420. https://doi.org/10.1111/j.1558-5646.1965.tb01731.x

Wright, S. (1969). Evolution and the Genetics of Populations: The Theory of Gene Frequencies (Vol. 2). Univ of Chicago Press.

Wright, S. (1978). Evolution and the Genetics of Populations. Volume 4. Variability within and among natural populations. The University of Chicago Press.

Wyrosdick, H., Chapman, A., Mignucci-Giannoni, A. A., Rivera-Pérez, C. I., & Bonde, R. K. (2018). Internal parasites of the two subspecies of the West Indian manatee Trichechus manatus. Diseases of Aquatic Organisms, 130(2), 145–152. https://doi.org/10.3354/dao03270

Zhu, Y., Liu, H.-Y., Yang, H.-Q., Li, Y.-D., & Zhang, H.-M. (2017). Factors affecting genotyping success in giant panda fecal samples. PeerJ, 5, e3358. https://doi.org/10.7717/peerj.3358

Zimmerman, R. C., Cabello-Pasini, A., & Alberte, R. S. (1994). Modeling daily production of aquatic macrophytes from irradiance measurements: a comparative analysis. Marine Ecology Progress Series, Vol. 114, pp. 185–196. https://doi.org/10.2307/24849689

295

BIOGRAPHICAL SKETCH

Aristide Takoukam Kamla was born in Douala, Cameroon, Central Africa. He grew up in the West Region of Cameroon, in a small town called Dschang where he completed his elementary, secondary, and higher education levels. Aristide earned his

Bachelor of Science in animal biology at the University of Dschang in 2007. He graduated with his Master’s degree in animal biology in 2011 from the same university.

During his Master’s program, Aristide was introduced to ecological concepts and got to hear about the African manatee for the first time. He soon became curious about the species and decided to study the animal as part of his graduate research. Aristide’s initial research goal was to estimate the population size of this poorly known species that he was studying for the first time. He soon discovered that manatee was very elusive and cryptic and that he could not achieve his goal by relying on simple direct observational methods. Therefore, he reoriented his study goal to focus on the distribution, habitat use, and threats of the African manatee in the two protected areas in Cameroon including Lake Ossa Wildlife Reserves and the Douala-Edea National

Park. Aristide realized that in addition to being elusive, the African manatee was very threatened by hunting and accidental catches. He then became passionate about the species and made its protection his career goal. For four years after his Master’s study, while applying for doctoral scholarships, Aristide worked as an ecology assistant for a research institute named METABIOTA based in the capital city of Cameroon. During that time, he would spend his weekends and holidays travelling to Lake Ossa, 300km away from the capital city, to implement community-based awareness conservation projects. His engagement for the manatee led him to established in 2012 a non-profit organization (African Marine Mammal Conservation Organization, AMMCO) dedicated

296

to protecting the African manatee in Cameroon. Through his struggle trying to save the manatee, Aristide met Dr. Lucy Keith Diagne (a senior biologist devoted to the conservation of the African manatee) who agreed to mentor and supported his work.

Aristide then decided to embark on a long journey abroad, and went in the United

States to acquire the skills that would help him protect the imperiled manatee. In 2014,

Aristide received a scholarship from the Fulbright program to pursue his Ph.D. at the

University of Florida. He received other complementary scholarships, including the

Wildlife Conservation Network and the Wildlife Conservation Society scholarships.

There could not be a better place in the world to study manatee than at the

University of Florida and U.S. Geological Survey in Gainesville, where Aristide volunteered and carried out most of his laboratory investigations. Through his Ph.D. program, he acquired various essential skills, including microhistological analysis techniques to determine manatee diet from feces, noninvasive genetic analysis, live capture and health assessments, limnology as a tool to assess manatee habitat quality, and remote sensing using a drone. Aristide did not keep all those skills for himself; each time he had the occasion to traveled to Cameroon for fieldwork, he would use that opportunity to share the acquired skills with young emerging researchers. He has now trained more than 30 African biologists and mentored 12 Master’s students from his country and hired seven of them who work for his non-profit organization.

During his Ph.D. program, Aristide finally found the answer to his initial puzzling question. How do we count an animal that we cannot even see? Aristide found out that he could isolate the DNA from floating feces of the species and use it to identify them.

Aristide, therefore, became the first biologist to identify manatees using their feces. For

297

him, it was a breakthrough for the needed studies of the elusive African manatee, who is least know of all sirenians. Aristide also established the first bathymetric map of Lake

Ossa and evidenced the eutrophication of the lake and the invasion of an invasive aquatic species (Salvinia molesta), where he has devoted his career to addressing the issue.

Aristide's current goal is to use the tools he acquired through his Ph.D. program to advance the scientific knowledge of the African manatee and other aquatic wildlife in order to preserve nature for future generations to enjoy.

298