P HYLOGENETICS AND FEEDING ECOLOGY OF THE AFRICAN MANATEE , Trichechus senegalensis By LUCY WARD KEITH DIAGNE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUI REMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
Â© 2014 Lucy Ward Keith Diagne
To my father, Allan Reed Keith, for his love and for instilling in me a love of nature and life lon g learning, and to my husband, Tomas Diagne, for his unwavering love and support.
4 ACKNOWLEDGMENTS This doctorate never would have been accomplished without substantial support and collaboration from many people on both sides of the Atlantic Ocean. I am so grateful to everyone who has helped me and believed in me. First I would like to thank Dr. Ruth Francis Floyd and Dr. Charles Courtney, who recruited me and made it possib le for me to come to UF immense and always friendly for classes even when I was in the Sahara desert, and to the Aquatic Animal Health Program , which generously provided financial support . committee and I have very much appreciated all of their constant interest in, and enthusiasm for , my work. there are not enough words to express my gratitude to my chair, Dr. Roger Reep , for all his guidance, good advice , and thought provoking questions . Than k you Roger, and may you have a magnificent retirement! For the past 16 years, Dr. Robert Bonde has been a wonderful source of knowledge of all aspects of manatee biology for me, and he convinced me that I should use genetics to study the African manatee. He also was instrumental in helping me to import many of the samples used in this research. Dr. been a trusted advisor and a great boss for many years. I am honored to fol low his footsteps in Africa. genetics techniques, for sharing her immense knowledge of the subject, and for her friendship. Dr. Karen Bjorndal has been a patient advisor and I have greatly appr eciated her thought provoking questions . I thank Dr. Amanda Demopoulos for inspiring me to do
5 stable isotope work and for opening her lab to me , and for daily guidance in learning the analysis techniques. I also thank Dr. John Krigbaum for all his help sa mpling the manatee bone and for our good discussions about the study of stable isotopes . Dr. David Reed and Dr. Mike Miyamoto , thank you both for challenging me to think about genetics in new ways, and for everything I learned in both of you r classes, whic h were the two best I took during my time at UF. Thanks to Katie Brill and Dr. Miriam Marmontel for their collaboration on the aging study of manatee ear bones, and I look forward to publishing that work with you both. And a very special thanks to my two e leventh hour heroes, Dr. Shichao Chen and Richie Hodel, for their work on the divergence dating analysis for Chapter 3. I have been fortunate to have wonderful lab mates at USGS, and I thank Michelle Davis, Gaia Meigs Fried, John Butterfield, and Theresa Floyd for teaching and helping me every step of the way in the genetics lab, and Jennie McClain Counts for all her great help and adv ice in the stable isotope lab. Big t hanks to Mat Denton for patiently teaching me the SIAR program for my analyses, and to Michelle Quach for her help weighing some of my isotope samples. Southeast Science Center, who have welcomed me there and made me feel like part of the family. e been privileged to work with man y very dedicated and hard working collaborators , all of whom I also consider friends . Most of all I thank my dear friend Tim Collins for introducing me to Africa, for keeping me laughing, and for always being there for me. m greatly indebted to Aristide Kamla Takoukam , Heather Arrowood, Cyrille
6 Mvele, Jonathan Perez Rivera, Dr. Matt Shirley, Captain Abba Sonko, Soumaila Berthe, Colonel Soumana Timbo, Dr. Patrick Ofori Dansen, Josea Dossou Bodjr Ã© nou , Severin Tchibozo , Dr. Gianna Minton, A issa Regalla, and Dr. Kouame Djaha for sample collection and export. Dr. Angela Formia, Aimee Sanders Parnell , Ruth Starkey, Dr. Kath Jeffrey, Bas Huijbregts, Solange Ngouessono , Stephane Louembet, Dr. Hilde VanLeeuwe, Dr. Rich Parnell, Tomo Nishihara, Dr. Josie Demmer, Nia ga Boh, Haidar El Ali, Bolaji Dunsin, Dr. Uzoma Ejimadu, Dr. Ken Cameron, Dawda Saine , Betania Ferreira, Dr. Edem Eniang, Abdoulaye Bine Guindo, and Alfousseini Semega all greatly helped my fieldwork and logistics , but it i leave it at that, because every on e of these people went way above and beyond for me . I am so happy and thankful to have all of you in my life. I also thank Alain Seck of the IFAN Museum in Dakar for allowing me to destructively sample their mana tee skull collection, and Ni cole Auil Gomez and Dr. Tony Mignucci for providing samples for this study. To the many other people who work hard every day to study and conserve the African manatee, I thank you for all your efforts to understand and protect this unique species. At Sea camaraderie of Susan Kahraman, Cyndi Taylor , and Monica Ross , and for the best board of directors . Raising all the funds necessary to build the African manatee research and c onservation project in multiple countries over the past nine years , and doing all the field and office work in addition to my doctoral program has been challenging , but I feel honored to have the loyal support of my funders who have made it possible : Colum bus Zoo and Aquarium Conservation Fund, U.S. Fish and Wildlife Service Wildlife Without Borders Africa Program , U.S. Marine Mammal Commission,
7 Disney Worldwide Conservation Fund, Save Our Seas Foundation, and Save Our Species. I also grate ful ly acknowled ge past project funding from Save the Manatee Club, Sea World and Busch Gardens Conservation Fund , Idea Wild, Earthwatch Institute, and the Bay and Paul Foundations. I last five years with most of my sanity intact without the good humor and support of many excellent to Michelle Davis (who has been with me every step of the way and deserves extra special thanks for putting up with me in so many classes, the lab, and even providing me a place to l ive when I first came to Gainesville!), Susan Butler, Jim Reid, Cathy Beck (who, along with Bob Bonde , also gave me a place to live for several months ) , Susan Kahraman, Nicole Adimey, Suzanne Tarr Williams , Jackie Speake, Sabrina Davies, Vivian Dwyer, and Sheri Hall Proft . Whether near or far, thank you all for being t here for all the laughs, tears, and many glasses of wine . My family has always believed that I could accomplish anything I set my mind to , and their belief instilled in me a fearlessness and passion t hat has allowed me to achieve a remarkably rewarding life and career. I can never thank my parents Allan and Winkie, my sister Tess Keith , my s ister Coral Rabey , my brother in law Steve Rabey , and my nephew and niece Max and Sky , enough for their un waveri ng love and encouragement. Thanks also to my extended family , who has also always been so enthusiastic about everything I do , and to taken me into their hearts . Most importantly I thank my husband, Tom as Diagne, who not only is the love of my life, but who has gone to incredible lengths to help me succeed, including renting
8 cars and driving up to 16 hours through the night in Senegal to collect manatee samples for me while I was in school , and opening m any doors in the world of conservation in Africa for me. Thank you for always believing in me , and for being the best partner in
9 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 12 LIST OF FIGURES ................................ ................................ ................................ ........ 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 Sirenian Evolution ................................ ................................ ................................ ... 17 The Earliest Sirenians ................................ ................................ ...................... 18 Family Dugongidae ................................ ................................ .......................... 20 Family Trichechidae ................................ ................................ ......................... 22 African Manatee Evolution ................................ ................................ ................ 28 Current Status of the African Manatee ................................ ................................ .... 31 Mauritania ................................ ................................ ................................ ......... 34 Senegal ................................ ................................ ................................ ............ 34 The Gambia ................................ ................................ ................................ ...... 35 Guinea Bissau ................................ ................................ ................................ .. 36 Guinea ................................ ................................ ................................ .............. 37 Sierra Leone ................................ ................................ ................................ ..... 37 Liberia ................................ ................................ ................................ ............... 38 CÃ´te d'Ivoire (Ivory Coast) ................................ ................................ ................ 39 Ghana ................................ ................................ ................................ ............... 40 Togo ................................ ................................ ................................ ................. 41 Benin ................................ ................................ ................................ ................ 42 Nigeria ................................ ................................ ................................ .............. 43 Cameroon ................................ ................................ ................................ ......... 44 Equatorial Guinea ................................ ................................ ............................. 45 Gabon ................................ ................................ ................................ ............... 45 Republic of the Congo ................................ ................................ ...................... 46 Democratic Republic of Congo ................................ ................................ ......... 47 Angola ................................ ................................ ................................ .............. 47 Mali ................................ ................................ ................................ ................... 48 Niger ................................ ................................ ................................ ................. 49 Chad ................................ ................................ ................................ ................. 50 Burkina Faso ................................ ................................ ................................ .... 50 African Manatee Habitat and Ecology ................................ ................................ ..... 51 Sirenian Conse rvation Genetics ................................ ................................ ............. 53 Stable Isotope Analysis as a Tool to Study Foraging Ecology ................................ 76
10 Introduction to Stable Isotopes ................................ ................................ ......... 76 Stable Isotope Methodology ................................ ................................ ............. 84 Stable Isotope Research Studies on Manatees ................................ ................ 95 Ov erall Aims ................................ ................................ ................................ ......... 104 2 RANGE WIDE GENETIC DIVERSITY AND PHYLOGENETIC STRUCTURE OF THE AFRICAN MANATEE ................................ ................................ .................... 111 Study Rationale and Objectives ................................ ................................ ............ 111 Methods ................................ ................................ ................................ ................ 115 Sample Collection ................................ ................................ .......................... 115 Genetics Analysis ................................ ................................ ........................... 116 Statistical Analyses ................................ ................................ ........................ 118 Results ................................ ................................ ................................ .................. 121 Control Region ................................ ................................ ............................... 121 Cytochrome b ................................ ................................ ................................ . 125 Concatenated Control Region and Cytochrome b Sequences ....................... 127 Discussi on ................................ ................................ ................................ ............ 128 Control Region ................................ ................................ ............................... 128 Cytochrome b ................................ ................................ ................................ . 130 Phylogeography of the Af rican Manatee ................................ ........................ 130 Implications for Conservation of T. senegalensis ................................ ........... 135 3 PHYLOGENY OF THE AFRICAN MANATEE IN RELATION TO THE OTHER EXTANT TRICHECHID SIRENIANS ................................ ................................ .... 153 Study Rationale and Objectives ................................ ................................ ............ 153 Methods ................................ ................................ ................................ ................ 155 Results ................................ ................................ ................................ .................. 158 Discussion ................................ ................................ ................................ ............ 162 4 CARBON AND NITROGEN STABLE ISOTOPE ANALYSIS TO DETERMINE THE DIET OF THE AFRICAN MANATEE ................................ ............................ 175 Study Rationale and Objectives ................................ ................................ ............ 175 Methods ................................ ................................ ................................ ................ 183 Sample Collecti on ................................ ................................ .......................... 183 Laboratory Analysis ................................ ................................ ........................ 185 Bone Sample Mechanical Preparation ................................ ..................... 186 Bone Sample Chemical Preparation ................................ ........................ 188 Statistical Analyses ................................ ................................ ........................ 190 Results ................................ ................................ ................................ .................. 192 Food Source Isotope Analysis ................................ ................................ ........ 192 Manatee Bone Isotope Analysis ................................ ................................ ..... 193 Gabon Mixing Model Results ................................ ................................ .......... 195 Senegal Mixing Model Results ................................ ................................ ....... 196 Discussion ................................ ................................ ................................ ............ 198
11 Gabon Manatee Diet ................................ ................................ ...................... 200 Senegal Manatee Diet ................................ ................................ .................... 201 5 CONCLUSIONS ................................ ................................ ................................ ... 223 Next Steps for African Manatee Genetics an d Stable Isotope Research .............. 223 Other Research and Conservation Needs ................................ ............................ 225 APPENDIX: AFRICAN MANATEE FOOD RESOURCES ................................ ........... 230 LIST OF REFERENCES ................................ ................................ ............................. 239 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 261
12 LIST OF TABLES Table page 2 1 New African manatee control region and cytochrome b partial hapl otypes identified by this study . ................................ ................................ ........................ 137 2 2 African manatee control region haplotype div ersity indices and tests of neutrality . ................................ ................................ ................................ ............. 138 2 3 Control region sequence analysis of molecular variance (AMOVA) for African manatee populations ................................ ................................ ........................... 139 2 4 ST estimates and P values for exact tests of pairwise haplotype frequency comparisons. ................................ ........................ 140 2 5 Cytochrome b 1227 bp haplotype diversity indices ................................ ............. 141 2 6 Cytochrome b sequence analysis of molecular variance (AMOVA) African manatee populations ................................ ................................ ........................... 141 2 7 Cytochrome b pairwise ST estimates using Tamura distance estimation and P values for exact tests of pairwise haplotype frequency comparisons .................. 142 2 8 Concatenated control region and cytochrome b ST estima tes using distance estimation. ................................ ................................ ............................ 142 4 1 15 13 C Â±SD of plant species.. ................................ ......... 205 4 2 15 13 C Â± 1 SD of animal prey species. ........................... 206 4 3 13 C collagen and apatite value s for study samples by location . ................................ ................................ ................................ ............... 207 4 4 15 N collagen value s for study samples by location . ...... 208 4 5 13 15 N values for Senegal recent and historic samples by bone laye r.. ................................ ................................ ...................... 209 4 6 Gabon manatee mean die t proportions by ear bone layer and age category , based on SIAR analysis. ................................ ................................ ..................... 210 4 7 Descriptive s 13 15 N values for study samples by location, manatee bone sampling layer , and age class. ................................ .................... 211 4 8 Senegal River manatee mean diet proportions by ear bone layer and collectio n time , based on SIAR analysis. ................................ ................................ ............ 212 A 1 Documented food resources of the African manatee. ................................ ......... 230
13 LIST OF FIGURES Figure page 1 1 Th e range of the African manatee ................................ ................................ ....... 108 1 2 The eight skull morphometric variables to confir m species identity of the three extant manatees, T. manatus , T. senegalensis and T. inunguis .. ....................... 109 2 1 African manatee control region haplotype maximum likelihood tree. .................. 143 2 2 Map of con trol region mitochondrial DNA haplotypes identified in African manatee samples.. ................................ ................................ .............................. 144 2 3 Map of control region haplotypes identified in Senegal manatee samples.. ........ 145 2 4 Map of control region haplotypes identified in Gabon manatee samples. ........... 146 2 5 Un rooted neighbor joining tree of African manatee control region haplot ypes. .. 147 2 6 Map of cytochrome b mitochondrial DNA haplotypes identified in African manatee samples. ................................ ................................ ............................... 148 2 7 C ytochrome b maxim um likelihood tree for three previously published African manatee haplotypes and nine new haplotypes. ................................ .................. 149 2 8 Cytochrome b un rooted maximum likelihood tree for nine new Trichechus senegalensi s 1227 base pair haplotypes ................................ ............................ 150 2 9 Un rooted maximum likelihood tree for 14 Trichechus senegalensis concatenated control region and cytochrome b 1637 base pair sequences. ...... 151 2 10 Un rooted neighbor joining tree for African manatee concatenated control region and cytochrome b 1637 base pair sequences. ................................ ......... 152 3 1 M aximum likelihood tree for control region 410 bp haplotypes for Trichechus senegalensis , T. manatus clusters I,II,III , and T. inunguis . ................................ .. 169 3 2 Bayesian consensus tree of 76 trichechid control region haplotypes . ................. 170 3 3 Cytochrome b 605 bp haplotype maximum likelihood tree for Trichechus senegalensis , T. manatus , and T. inunguis rooted in the dugong. ...................... 171 3 4 Bayesian analysis of cytochrome b haplotypes visualized in a NJ tree rooted in the dugong. ................................ ................................ ................................ ......... 172 3 5 BEAST cytochrome b output tree of sirenian dates of diverge nce ...................... 173 3 6 Un rooted BEAST control region output tree of trichechid divergence. ............... 174
14 4 1 Photomicrograph and line drawings of wh ole section and detail from the central portion of the periotic dome of an age class 8 Florida manatee. ............. 213 4 2 Sampling regions of African manatee ear bones, including the type of bone found in each sample. ................................ ................................ ......................... 214 4 3 Gabon manatee samples by bone layer , and food resource species collected for this study. ................................ ................................ ................................ ....... 215 4 4 Senegal River manatee samples by bone layer , and food resource species collected for this study. . ................................ ................................ ....................... 216 4 5 13 15 N values for three aquatic plant species collected in both Senegal and Gabon. ................................ ................................ ................ 217 4 6 13 C collagen values 13 C apatite for Gabon and Senegal mana tees by periotic bone sampling layer.. ................................ ........................ 218 4 7 13 C and 15 N values for all Gabon and Senegal manatees by periotic bone sampling layer.. ................................ ................................ .............. 218 4 8 Senegal manatees shown by location and ear bone sampling layer.. ................. 219 4 9 SIAR output of Gabon manatees by ear bone sampling layer, shown in relation to food sou rces ................................ ................................ ................................ .... 219 4 10 13 15 N for Senegal m anatees including both historic and recent sample collection by location.. ................................ ..... 220 4 11 SIAR output of Senegal coast manatee samples by e ar bone sampling layer shown in relation to food sources. ................................ ................................ ..... 221 4 12 Boxplots of Senegal coast manatee diet sources by ear bone sampling layer:. . 221 4 13 SIAR output of Senegal River manatee samples by ear bone sampling layer shown in relation to food sources.. ................................ ................................ .... 222
15 Abstract of Dissertation Presented to the Graduate School of the University of Flori da in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PHYLO GENETICS AND FEEDING ECOLOGY OF THE AFRICAN MANATEE , Trichechus senegalensis By Lucy Ward Keith Diagne August 2014 Chair: Roger Reep Major: Veter inary Medical Sciences The African manatee is one of the least understoo d marine mammals in the world. It is believed to be in decline throughout its 21 country range, primarily due to illegal hunting and incidental bycatch. B aseline research is criticall y needed to inform conservation actions for the species . Over eight year s, African manatee genetic samples were collected from eight countries. P rimers were used to amplify a 410 base pair portion of the mitochondrial ( mtDNA ) control region, and a 1227 bas e pair port ion of the cytochrome b gene. Fourteen new control region and nine new cytochrome b haplotypes were identified for the species. These are the first African manatee genetic results from Senegal, Guinea, Mali, and Ivory Coast . New haplotypes were characterized by high haplotype diversity, but low nucleotide diversity, indicating expansion after a period of low effective population size . Phylogenetic trees identified two cl ade s aligned geographically, indicating regional population separation likely due to low recent dispersal. New African manatee and previously published m tDNA haplotypes for all extant t richechid species were analyzed to clarify the evolutionary placement of African man atees in relation to the other t richechids. Bayesian MCMC analy se s were
16 conducted to calculate divergence estimate s of t richechids from ancestral species, and to elucidate time of divergence of the Africa n manatee from the other t richechids. Cytochrome b Bayesian analyses indicated th e more basal Amazonian manatee spe cies diverged from the African and West Indian species 4.91 (CI 95%) million years ago (Mya) . Divergence analyses estimated that the African manatee diverged from the West Indian species 3.49 Mya (CI 95%) . Additionally, analyse s indicated a separation of A frican manatee North a nd South regional clades at 1.5 1 Mya (CI 95%) . Carbon and nitrogen stable isotope analyses were conducted to describ e African manatee diet for the first time. Average lifetime d iet was examined through comparison of periotic bone samp les and potentia l plant and animal food sources from Gabon and Senegal. Ranges 13 C placed Gabon manate es primarily in the range of C 3 plants, and most Senegal manate es in the C 4 plant range . Analyses of Senegal manatees also indicated that fish and mollusks are an importan t component of the diet of these pop ulation s .
17 CHAPTER 1 INTRODUCTION Sirenian Evolution Sirenians are a group of placent al (Eutherian) mamm al s whose origins can be traced back in evolutionary time to the Eocene Epoch, ~50 million years ago (Mya) (Domning 2001a). Within Eutheria, sirenians belong to the clade A frotheria, which, as the name suggests, are mamm als that originated on the African continent during the Cretaceous Period, and then are believed to have evolved on the Afro Arabian continent after the breakup of Gondwan ala nd approximately 65 Mya ( Simpson 1 945, Tabuce et al . 2007). Afrotheria includes a diverse group of mamm als including the elephants ( Proboscidea ), hyraxes ( Hyracoidea ), dugongs and manatees ( Sirenia ), tenrecs ( Tenrecidae ), golden moles ( Chrysochloridae ), elephant shrews ( Macroscelididae ), a nd aar dvarks ( Tubulidentata ) (McKenna 1975, Tabuce et al. 2007). Within Afrotheria, sirenians are classified in the superorder Paenungulata (meaning alm the elephants and hyraxes as well as two other extinct orders, the Embr ithopoda (rhinoceros like mamm als ) and the Desmostylia (hippopotamus like mamm als ) ( Simpson 1945 , Ozawa et al . 1997). Until recently , sirenians were placed in the clade Tethytheria, defined as mamm als that originated around the ancient Tethys Sea during th e P ale ogene Period, but subsequent genetic an aly ses have now confirmed that Paenungulata is within Afrotheria (Ozawa et al. 1997, Springer et al . 1999, Kellogg et al . 2007, Pardini et al . 2007). Some of the earliest sirenian fossils have been discovered ar ound the edges of the ancient Tethys Sea in present day France, Tunisia, Egypt, India, and Pakistan ( Domning and Gingerich 1994, Clementz et al. 2006, Benoit et al. 2013).
18 The Earliest Sirenians The earliest sirenian Family is the Prorastomidae, aquatic q uadrupeds that evolved approximately 50 million years ago in the Eocene ( Domning 2001a). It is hypothesized that Prorastomids origin al ly evolved the Old World and expanded their ranges alo ng ancient shorelines to reach the loc ale that is the present day Ca ribbean Sea ( Domning 2001, Clementz et al . 2009 b ). Until recently , the only known Prorastomid fossils were discovered at an archeologic al site near Seven Hills, Jamaica that dated to the early to middle Eocene (Domning 2001a). A Prorastomus sirenoides skul l dated to approximately 48 50 Mya is the most ancient sirenian fossil currently known (Savage et al. 1994). A fossil skeleton from another branch of the Prorastomid family, Pezosiren portelli , was discovered at the same site and dated to 47 49 Mya (Domnin g 2001a). It is thought to be morphologically advanced in comparison to P . sirenoides (Domning 2001a). However, no fossil remains of the earliest sirenians were found on the eastern side of the Atlantic Ocean until 2012, when Hautier et al. published their finding of a Prorastomid fossil vertebra from phosphate mine in northern Seneg al. This fossil is significant because it lends support to the hypothesis that the earliest sirenians evolved in this region alo ng with their closest relatives (Hautier et al. 2 012). The Seneg al fossil is 40% larger than any of the P. portelli fossil vertebrae and has been designated as indet et al. 2012). In 2013 Benoit et al. reported the find of a sirenian fossil petros al (ear) bone at Djebel Chambi, Tunisia , in a fossil bed believed to date to the same time period in the Eocene as the Jamaican site. This is the oldest fossil sirenian found in Africa to date , and features of this bone appear to indicate that it is smaller and more primitive than Proras tomus , but the species has yet to be named (Benoit et al. 2013).
19 Pezosiren portelli was a quadruped al , pig sized mamm al that was capable of living on land, as indicated by anatomic al features including well developed legs and pelvic bones, and a multi ver tebr al sacrum (indicating that the hind limbs were still used for swimming, rather than the tail), but it al so had adaptations to the aquatic environment including enlarged nas al openings located at the top of the snout, a lack of nas al sinuses, and numero us heavy, swollen and dense (pachyosteosclerotic) rib bones that would have he lped it to submerge (Domning 2000, Domning 2001a, Reep and Bonde 2006). These aquatic adaptations indicate that it was a true sirenian that probably spent much of its life in the water (Domning 2001a). However, Pezosiren al so had a straight (undeflected) rostrum, indicating that it had not yet evolved for e ating submerged plants (Domning 2001a). The Protosirenidae also evolved in the Eocene and existed concurrently with Prorasto midae, but some species in this family existed later into the Eocene and exhibited greater adaptations to the aquatic lifestyle, such as a more flattened tail that was the main propulsion for swimming, a shorter neck, and a single sacr al vertebra with a we aker connection to the pelvis, indicating reduced use of the hind limbs f or swimming or walking (Domning 2000). Addition al ly, the Protosirens had alr eady developed a rostr al deflection similar to that of modern manatees, indicating that they were definitel y feeding off the bottom of the waterways they lived in (Domning and Gi ngerich 1994 , Domning 2000). Fossils from Protosirenidae have been found at the Seven Hills site in Jamaica alo ng with the Prorastomidae fossils (Domning 1982, Domning 2001a). Domning ( 2002) al so reported that the fossil evidence indicates that the transition of sirenians from a terrestri al to a fully aquatic lifestyle probably occurred
20 rapidly over a period of only a few million years during the early and middle Eocene. Protosirens al so are believed to have origin al ly evolved in the Old World, Tethys Sea region and dispersed to the present day Caribbean and South America (Domning 1982). The earliest Protosiren fossils have been found in Egypt, Libya, and Pakistan, and included Protosiren smithae and P. fraasi (Domning et al. 1982, Domning and Gingerich 1994). To date only a few fossil fragments have been found in the Western Atlantic; these were discovered in Florida and dated to the mid Eocene (Domning et al. 1982, Domning, 1982). Famil y Dugongidae The earliest members of the sirenian Family Dugongidae coexisted with Protosiren and exhibited an even more advanced stage of evolution (Domning 2000). These anim als were fully aquatic and no longer supported their weight on their limbs ; the f ront limbs had become flippers to aid with steering while swimming, their necks became shorter, and their bodies were now oriented more horizont al ly in the water in order to better facilitate swimming and hydrostatic control (D omning 2000, Clementz et al. 2006). The reduction and eventu al loss of the hind limbs helped to move the center of gravity further forward in the dugongs, which in turn moved the body to be horizont al in the water (D omning 2000, Reep and Bonde 2006). Dugongidae has been the most proli fic sirenian family since their appearance in the Eocene ( Domning 2000). The earliest members of Dugongidae were the subfamily H ali theriinae, with two more subfamilies (Dugonginae and Hydroam ali nae) appearing in the Oligocene (Barnes et al . 1985). Subfamil y H ali theriinae was a gener ali zed group with more species (including Eosiren, Eotheroides , and H ali therium ) that foraged on seagrass leaves and sm all er rhizomes (Clementz et al . 2009 b ). Metaxytherium , also a member of this subfamily, is
21 believed to have ev olved from H ali therium in the Oligocene and then given rise to subfamily Hydrom ali nae in the Miocene (Domning et al. 2007). Subfamily Dugonginae consisted primarily of large tusked speci ali sts that fed on large seagrass rhizomes and had fewer species (and includes one species still extant, Dugon g dugon ), while Hydrom ali nae was an exclusively Pacific Ocean group that eventu al ly evolved to ow ( Hydrodam al is gigas ), the largest sirenian to ever live and the only one to live in cold northern water s (Domning et al . 2007, Clementz et al. 2009 a ). The Dugongidae family became very diverse in the Miocene, with multiple dugong species co occurring in sever al ocean basins over three epochs, for example in the middle to late Eocene in Egypt, the Oligocene in Florida, the early Miocene in India, and the early Pliocene in Mexico (Domning 1982, V Ã© lez Juarbe et al . 2012). The Miocene was a very warm Epoch glob all y, and this is believed to have played a role in providing many suitable habitats for sirenians wor ldwide (Domning 2001b). The high amount of sirenian diversity exhibited by fossils from this time (12 genera) is thought to have occurred due to niche partitioning and unique morphologic al speci ali zations within each species, particularly rostr al deflectio ns, tusk morp hology, and body sizes (Domning 1982, Domning 2001b, Reep and Bonde 2006, V Ã© lez Juarbe et al. 2012). During the late Pliocene, a period of glob al cooling began that resulted in the formation of an arctic ice cap and increased continent al glaci ation ( Domning 2001b ). This in turn lowered sea levels, and resulted in degradation and loss of species diversity of seagrass beds due to erosion and runoff, which is hypothesized to have added stress to the dugong species that had evolved to fit very spec ific seagrass feeding niches (Domning 1982). Addition al ly, the closure of the Centr al American Seaway by the
22 formation of the Isthmus of Panama in the middle Pliocene caused large sc ale oceanic changes and extinctions in the Caribbean which are hypothesize d to have heavily impacted both seagrasses and the species t hat depended upon them (Domning 2001b, Marsh et al . 2012). Fin al ly, the dugongids also had low crowned (brachyodont) teeth which Domning (1982) hypothesized led to increased tooth wear and possibl y increased mort ali ty. By this time sirenians in the family Trichechidae had evolved, and the species Ribodon limbatus of the Amazon Basin exhibited both horizont al tooth replacement (supernumerary molars) which eliminated the problem of tooth wear, and ha d higher crowned (hypsodont) teeth that were better suited to feeding on plants with higher silica content, such as grasses and seagrasses (Domning 1982, Domning 2001b). It is thought that trichechids eventu al ly replaced the dugongids in the Caribbean (Dom ning 1982). D ugongids in the Pacific Ocean also developed hypsodont teeth and were able to evolve through the problems of increased tooth wear, even if their Caribbean relatives did not (Domning 19 82 ). Family Trichechi dae The dispers al of Protosirenians t o South America and their subsequent isolation in this region is believed to be the origin of the Family Trichechi dae during the Miocene (Domning 1982). However, this is still a gap in the fossil record, as a directly linked ancestr al species has not yet b een discovered (Barnes et al . 1985). Similar to their protosirenid ancestors, Trichechidae in the Tertiary Period had slight rostr al deflections that indicated a diet of floating or emergent aquatic plants (Domning 1982). In the Pliocene, trichechids moved into the interior of the Amazon basin, which at that time was a large inland sea with four outlets to the north, south, east, and west (Domning 1982). About 4 million years ago (early Pliocene) uplift of the Andes mountain range
23 began, which closed off th e outlets forming an inland sea, which then re opened to the west in the late Pliocene, forming the current draina ge of the Amazon River (Domning 1982). Some t richechids were isolated in the Amazon Basin when it was closed off from the west coast, and it i s hypothesized that they adapted to feeding on newly abundant freshwater plants which are believed to have increased significantly due to nutrient runoff from the Andes (Domning 1982). Other trichechid species were coast al and lived in the Caribbean and Fl orida (Domning 2005, Marsh et al . 2012). Ribodon is thought to be the direct ancestor of modern trichechids, but there are still gaps in knowledge about where Trichechus origin al ly evolved (Marsh et al . 2012). One scenario hypothesizes that they evolved in the Amazon Basin during the transition from the Pliocene to the Pleistocene (Domning 2005, Marsh et al . 2012). Trichechus fossils from this time and place are reported to be more similar in morphology to the West Indian manatee ( Trichechus manatus ) than t o the Amazonian manatee ( T. inunguis ) that is found in the Amazon Basin today (Domning 1982). Another part of the theory is that the Amazonian manatee is the descendent of the ancestr al form that was isolated in the Amazon Basin, while the West Indian mana tee evolved in the Orinoco Basin in Colombia and coast al ly along the Caribbean and northern South America ( Vianna et al . 2006, Marsh et al. 2012). The African manatee ( Trichechus seneg ale nsis ) is believed to be most closely related to the West Indian manat ee ( Domning and Hayek 1986, Vianna et al . 2006 ), and divergence will be discussed in subsequent chapters . To understand the finer sc ale evolution of the three extant t richechids, it is of interest to examine the attributes that distinguish th em from each other. Domning and Hayek (1986) wrote an exhaustive description of morphologic al features and variation
24 between the three extant t richechids, which described the key features used to differentiate the species. This study used 273 manatee skull s ( 150 T. manatus , 86 T. inunguis, and 37 T. seneg ale nsis ) from museum and private collections around the world (Domning and Hayek 1986). Attributes used to differentiate the species included extern al characters such as body size and shape, snout morpholog y, and nails, as well as intern al characteristics such as ribs, dentition, and 110 individu al crani al measurements grouped into categories of 36 variables (Domning and Hayek 1986). Using a stepwise discriminant an aly sis of all 36 variables, Domning and Hay ek demonstrated that a grouping of eight specific variables, taken together, correctly determined the species of their sampled adult skul ls 100% of the time (Table 1). Trichechus inunguis was found to be the most derived of the extant species, while T. se neg ale nsis was found to have retained the most primitive features, but the authors noted that all three species have speci ali zed in differen t directions (Domning and Hayek 1986). This study reported that there are few derived characters shared by any two o f the species of Trichechus , but it did define sever al shared characters (synapomorphies) of T. manatus and T. seneg ale nsis and concluded that these two species likely shared a more recent common ancestor with each other than either does with T. inunguis ( Domning and Hayek 1986). However, the study further indicated that the time of separation of T. manatus and T. seneg ale nsis was probably within a relatively short time frame after the two species separated from T. inunguis (Domning and Hayek 1986) . For T. s eneg ale nsis , unique anatomic al features ( autapomorphies) include a shortened rostrum, the presence of longitudin al crests on the floor of the mesorostr al
25 fossa and p ala ta l surface, narrow bases of the supraorbit al processes, and a broadening of the zygomat ic arch and coronoid pro cess (Domning and Hayek 1986, Domning 1994) . African manatees also have more teeth than the other two extant species; Domning and Hayek (1986 ) hypothesize that this may be due to the fact that the tooth row typic al ly sits farther fo rward in African manatees than in the others, and features of T. seneg ale nsis include a decreased rostr al deflection, the tempor al crests, and incomplete division o f the foramen incisivum (hole in the anterior of the p ala te) compared to the other extant trichechids, the presence of a crista orbitotempor al is (an arch at the top of a recess in the tempor al w all , see V34 in Table 1) which was common in Eocene sirenians but is very uncommon in the other extant trichechids, a lack of postorbit al apophyses (bony outgrowths on the tempor al w all immediately behind the supraorbit al processes), a strongly curved frontopariet al suture, a smooth anterior front al al s (the bones at the top of the skull between the pariet al and exoccipit al bones) (Domning and Hayek 1986). These features are considered primitive because they are all either more similar to an ancestr al sirenian than another extant s pecies, or they are dissimilar to T. inunguis , which is considered the most derived of the three living trichechids (Domning and Hayek 1986). The primitive nature of T. seneg ale nsis was also illustrated in Domning and Hayek (1986) when their discriminant a n aly sis assigned a fossil Trichechus bakerorum skull dated to the Pleistocene of South Carolina as 0.99 probability of being T. seneg ale nsis . T. bakerorum is actu al ly thought to be ancestr al to T. seneg ale nsis , T. manatus and possibly to T. inunguis ( Domni ng and Hayek 1986).
26 A separate an aly sis of T. seneg ale nsis skulls within the Domning and Hayek 1986 study determined that there was no basis for prior designations of sub species within T. seneg ale nsis (including Manatus vogelii, Manatus nasutus , and Mana tus oweni ) made by sever al explorers in the 1800 s (Balkie 1857, Domning and Hayek 1986). Domning and Hayek reported that their classification an aly sis assigned all of these skulls to T. seneg ale nsis with probabilities greater than 99%. However, they did fi nd evidence to support that the two subspecies of T. manatus ( T.m. manatus and T. m. latirostris ) were distinguishable both due to the separate geographic locations of populations as well as distinctive crani al characters ( Domning and Hayek 1986). Each of the three Trichechid species possesses numerous autapomorphies that defined it in this study ( Domning and Hayek 1986). One of the most obvious differences between species is rostr al deflection, which is defined as ranging from 29 o 52 o in T. manatus , 25 o 41 o in T. inunguis , and 15 o 40 o in T. seneg ale nsis (Domning 1982, Domning and Hayek 1986). These differences in the angle of the snout are adaptations that all ow each species to exploit food plants in the different habitats in which they live (Do mning 1982, Reep and Bonde 2006). A greater degree of rostr al deflection positions the mouth closer to the substrate for bottom feeders, while the rest of the body is able to stay horizont al in the water column, facilitating horizont al swimming while feeding (Do mning 2000). The West Indian manatee has the most strongly deflected snout of any fossil or living trichechid, which is believed to be an adaptation to a diet based primarily on seagrasses and other submerged plants ( Domning 1982) . At the other end of the trichechid spectrum is the African manatee, which lives pr edominant ly in murky and muddy rivers and lagoons with less submerged vegetation, and feeding primarily on
27 floating and emergent vegetation, therefore it has a snout with the least deflecte d rostrum (Domning 1982, Powell 1996). Amazonian manatees also live in an ecosystem with dark water that does not promote submerged plant growth, so this species also feeds primarily on floating and emergent veget ation along riverbanks (Domning 1982). Sever al stud ies have used fossil sirenians to assess ancient seagrass ecology and diversity while simultaneously studying the diet of extinct sirenians (Domning 1981, Domning 2001b, Clementz et al . 2006 , Velez Juarbe et al . 2012). Evidence of seagrasses are sparse in the fossil record, but they first appeared in the mid Cretaceous Period and have been found in the same loc ations as sirenian fossils (Eva 1980, Domning 2001b, Clementz et al . 2009 a ). Domning (2001b) hypothesized that dugongids in the Miocene, and particul arly the dugongines with large tusks, speci ali zed on the largest seagrasses, which were considered climax species. The tusks were likely used to uproot the rhizomes of these seagrasses, just as m odern dugongs do today (Domning 2001b). In this way, it is hy pothesized that the dugongines kept the larger seagrasses from reaching climax level, which all owed other seagrass species to thrive as well, which then became a niche for other dugongids that fed on sm all er rhizomes ( Domning 2001b ). Clementz et al. (2006 ) used tooth enamel bioapatite from Protosirenidae ( Protosiren smithae ) and sever al species of Dugongidae ( Eosiren libyca , Eotheroides sp. , and H ali therium taulannense ) from fossil beds representing the ancient Tethys Sea to determine diet and habitat type (freshwater vs. marine). Results of carbon and oxygen stable isotope an aly ses indicated that these species were all primarily aquatic, and that early in their evolution, sirenians adapted to a predominantly marine, seagrass -
28 based diet without evidence of an intermediate connection to freshwater habitats (Clementz et al . 2006). Domning (1981) had previously described fossil sirenians associated with seagrass beds in both the Tethys region and Florida dating from the middle Eocene, which further supports thi s conclusion. Clementz et al. (2006) reported that the low variation in carbon stable isotope v alu es for sirenian fossils sampled in that study suggested that their dietary preferences were highly focused on seagrasses, and that sirenians did not evolve to eat other aquatic resources such as freshwater plants or kelp until the evolution of the Family Trichechidae and the Subfamily Hydrodam ali nae. African Manatee Evolution Most fossil remains of sirenians have been found in the Mediterranean, Flor ida, and t he Caribbean (Domning 1981, Domning et al . 1982, Domning and Gingerich 1994, Domning 2001 a , Domning 2001b, Domning 2002, Clementz et al . 2006) but archeologic al exploration has unearthed very few sirenian fossils on the western side of the African continen t , and to date they have all been Prorastomidae fossils as discussed above (Hautier et al. 2012, Benoit et al. 2013) . Therefore , no T. seneg ale nsis fossils have yet been found. The little that is currently known about the origins of T. seneg ale nsis has bee n determined through morphologic al studies of skeletons and genetic an aly sis of mitochondri al DNA ( mtDNA; Husar 1978, Domning and Hayek 1986, Parr 2000, Vianna et al. 200 6 ). Domning (2005) reported that African manatees probably diverged from T. manatus th rough dispers al from South America across the Atlantic Ocean during the Pliocene (5 2.5 Mya ) or Pleistocene ( 2.5 0.01 Mya ). It is highly unlikely that the species evolved in the Old World since all other trichechids are known to have evolved in South A merica (Domning 2005). Vianna et al. (200 6 ) performed molecular clock an aly ses using nucleotide divergence to date back to
29 ancestr al nodes for all three extant t richechids. This study included genetic samples from six T. seneg ale nsis , 189 T. manatus , and 9 3 T. inunguis , and estimated time of divergence of T. seneg ale nsis to approximately 309,000 years ago, which , if accurate, would indicate that dispers al occurred during the Pleistocene (Vianna et al . 200 6 ). Domning (2005) wrote a thoughtful and thorough s ynopsis of the all accumulated evidence that indicates African manatees evolved in South America and dispersed to Africa no earlier than the late Pliocene (not before approximately 3.6 million years ago). The lines of evidence he summarized included that t he fossil record provides a direct line back to South America as the center of evo lution of Trichechidae (Domning 1982) , that the closest relative of the African manatee is the West Indian manatee ( T. manatus ) as shown by morphology ( Domning 1982, Domning and Hayek 1986) , that manatees traveling across the ocean could have survived withou t food for several months (Best 1983), and that all three extant trichechids exclusively share a genus of nematode ( Heterocheilus ) in their stomach and intestines, with the same species ( Heterocheilus tunicatus ) found in T. manatus and T. inunguis , and a more derived species ( Heterocheilus domningi ) in T. seneg ale nsis (Sprent 1983). All of these factors together provide reasonable proof that manatees in Africa disperse d from South America, and that it was physic al ly possible for them to do so (Domnin g 2005). But how did t hey get there? One hypothesis that has been proposed is that the tremendous outflow of water from the Amazon River has been shown to carry a freshwater lens alm ost entirely across the Atlantic Ocean on the South Equatori al Countercurrent, which manatees Karger et al. 1988 , Domning 20 05). In this scenario manatees c ould have followed floating mats of
30 vegetation out of the Amazon River and then been swept up in the current ( Bonde 2009). This would make the most sense if the direct ancestors of the African manatee came from the Amazon, but genetic and fossil evidence both indicate that they are more c losely related to T. manatus , which does occur at the mouth of the Amazon and alo ng the coast of Brazil, but also is thought to have its center of origin in the Orinoco Basin in Colombia ( Domning and Hayek 1986 , Vianna et al. 2006 ). In the same study Domni ng (2005) also cites a study by Boekschoten and Best (1988) that mentions that manatees could have been carried across to Africa on an Atlantic current approximately 18,000, years ago, but given the previous molecular evidence that T. seneg ale nsis may have arrived in Africa between 309,000 and over three million years ago, this time frame is likely much too recent ( Vianna et al . 2006, Domning 2005). Addition al ly, Domning (2005) cites records of manatee sightings on the northeastern seaboard of the USA as fa r north as Massachusetts ( Reid 1995), and in Greenlan d and Europe from the late 1700s and 1800 s, implying that manatees could have traveled to Africa using a North Atlantic current (Domning 2005). However, the theory of manatees traveling to Africa via the North Atlantic gyre seems dubious . Sightings from the 1700s and 1800 s were not verified by the co llection of specimens (Domning 2005) and could have easily been misidentified. More importantly, manatees exposed to water less than 20 o C for even sever al wee ks have been shown to die of cold stress ( Bossart 2001), and without fresh water to drink, an oceanic journey of sever al months seems implausible (Ortiz 1998). Therefore, the most likely hypothesis is that the African manatee dispersed by an ocean current from the centr al Atlantic Ocean.
31 Vianna et al. (200 6 ) published the first five mtDNA control region and three cytochrome b haplotypes for the African ma natee. The six samples from that study resulted in five control region haplotypes in two clusters, one for the samples from Guinea Bissau (Y01 and Y02), and a second for th re e other samples from Ghana, Niger and Chad (Y03, Y04, Y05) ( Vianna et al. 2006 ). No other samples from Africa we re available for analysis at that time ( Vianna et al. 2006 ), but c omparis on of increased numbers of African manatee mt DNA haplotypes with T. manatus and T. inunguis haplotypes , as well as more rigorous divergence dating, are important analyse s that are need ed to better enable us to determine the point s of divergence between the s e three species. Current Status of t he African Manatee The African manatee is one of the least understood marine mamm als in the world, and has recently been identified as the least studied large mamm al in Africa (Trimble and Van Aarde, 2010). The range o f the species includes 21 countries (the African Atlantic coast from Mauritania to Angola, and the interior countries of M ali , Niger and Chad, Figure 1 1 ) and is larger than the width of the United States, yet to date there have been only a handful of stud ies, most of which were short term field surveys with no follow up or loc al capacity building. There are no estimates of abundance for the species anywhere in its range and the impact of hunting and habitat destruction are poorly documented, but the trade in manatee bushmeat is well known throughout Africa. In recent times hydroelectric and agricultur al dams have also isolated manatee populations in many major rivers, including in Lake Volta, Ghana (Akosombo Dam, constructed in 1965), the Niger River (Kainj i Dam in Nigeria, constructed in 1968, and the Maka la Dam in M ali , completed in 1945), and the Seneg al River (Diama Dam,
32 constructed in 1983 and the Felou Dam in Kayes, Mali). Since 2008, the Niger River Basin Authority has authorized the construction of t hree addition al multi purpose dams in the Niger River: at Fomi in Guinea, Taoussa in M ali , and Kandjadji in Niger (Diarra 2011). All of these are likely to further restrict manatee habitat in these countries and lead to genetic isolation of populations (Fi gure 1 1) . Dam structures have also killed manatees both through entrapment and death in dam gates (Powell 1996, L. Keith Diagne unpublished data). Natur al isolation is also believed to occur in inland river systems 1000 km and more from the coast in count ries such as Chad. Furthermore, prior to the current study, only 21 genetic samples had ever been an aly zed for the entire species, all of which came from only five of the twenty one range countries: Ghana, Cameroon, Niger, Chad, and Guinea Bissau (Parr 200 0, Vianna et al. 2006). Conservation efforts are greatly hindered by a lack of basic information about the species, and most behavior al and physiologic al attributes are currently inferred from the Florida manatee ( T . m . latirostris ) . The IUCN Red List cat egory and criteria classifies the African manatee as Vulnerable. Criteria for placement of the species in this category are based upon threats, particularly hunting and incident al capture in fishing gear, that appear to be continuing to increase throughout the range, with loc al ly high rates and near extirpation in some regions. Lack of protein and continued poverty for human populations , and limited enforcement of nation al laws are believed to be driving increasing hunting levels ( Keith Diagne and Powell 20 14 ). The Red List also acknowledges that dams are increasingly and permanently fragmenting populations in many areas, including Ghana, M ali , Niger, Nigeria, and Seneg al , and that populations segmented into sm all er habitat
33 areas are also susceptible to incr eased loc al hunting pressure. It is estimated that there is a high probability of a 30% or greater reduction in population size will result within a 90 year, three generation period ( Keith Diagne and Powell 20 14 ). The Convention of Migratory Species (CMS) successfully created an Action Plan and MOU for the African Manatee in 2008 , and raised the species from Annex II to Annex I in 2009, a more endangered status. Addition al ly, the Convention on Internation al Trade in Endangered Species (CITES) voted at their Conference of Parties in March 2013 to raise the species from Appendix II to Appendix I, which bans all leg al internation al trade, but leaves enforcement of illeg al trade up to the range countries. Over all , the trajectory for the African manatee is believ ed to be in serious decline without drastic and wide ranging conservation efforts on their beh alf ( Keith Diagne and Powell 20 14 ). There are no African manatee population estimates based on quantitative information. Nishiwaki et al . (1982) and Nishiwaki (1 984) made a number of intuitive estimates for 15 countries, based primarily on interviews rather than direct surveys. Roth and Waitkuwait (1986) and Akoi (2004) estimated 750 800 manatees for Cote he manatee population in Benin at 125 individu al s. There are no population number estimates for other countries at this time. One population that may no longer exist is that of the Lake Chad basin it is reported that manatees were gone from Lake Chad by 1 929 (Hatt 193 4 , S alk ind 1998), and alt hough reports of them occurred in the Chari, Bamingui, Bahr Kieta, and Logone Rivers, S alk ind (1998) reported that they were no longer found in these waterways. T oday the drying and desertification of Lake Chad no long er provides much suitable habitat. Available population information by country follows:
34 Mauritania The species occurs in the Seneg al River and its tributaries (Powell, 1996); this river forms the border between Mauritania and Seneg al . It is an inf requent inhabitant of the Diawli ng Nation al Park, a wetland reserve of interconnecting streams, lakes and ponds adjacent to the Seneg al River near the Diama dam (Perrin 2001). Manatees also likely inhabit the Gorgol Marigot, a large season al ly flooded tributary ea st of Kaed i. Senegal In Seneg al , the manatee was reported as close to extinction by Navaza and Burnham (1998) but more recent work indicate s that this is not the case ( Diop 2006 , Ba et al. 2008 , Keith Diagne and Oceanium Dakar, unpublished data). In the S eneg al River manatees are permanently isolated from the coast by the Diama Dam, located in the Seneg al River east of St. Louis, and upstream by the Felou Dam in Kayes , M ali which prevents them from moving further eastward. Manatees are no longer reported w est of the Diama dam in the Seneg al River delta near St. Louis or coast al ly in this region (Keith Diagne unpublished data ). However, manatees are sighted throughout the Seneg al River east of the dam. Manatees are frequently seen in Guiers Lake year round, and in 2005 Tocc Tocc Community Reserve was created for their protection alo ng the northwestern shore of Lake Guiers. In February 2014, Tocc Tocc Reserve was designated as Seneg s fifth Ramsar site. During the rainy season when the Seneg al River floods, manatees are regularly seen near Kanel and Matam in eastern Seneg al (Oceanium Dakar and Keith Diagne, unpublished data). As flood waters recede during the dry season, manatees become stranded in season al lakes at Wendou Kanel and the loc al people, alo ng w ith assistance from the Seneg ale se NGO Oceanium Dakar, have been rescuing and releasing them back into the main Seneg al River since
35 1984. Beginning in 2008, manatees became trapped behind a new agricultur al dam at Navel, near Matam, on the Seneg al River. F ive manatees died as a result of entrapment behind the dam, and dam grates were removed in 2010. Seven live manatees were rescued in early 2009 and satellite tags were attached to three adult manatees that were rescued and released back into the Seneg al Ri ver (Keith Diagne et al. unpublished data). These manatees used over 265 km of the Seneg al River to the north and south of the release site and made directed trips to feeding areas (Keith Diagne et al . unpublished data). Manatees are also frequently sighte d at freshwater springs that occur alo ng coast al Seneg al from the Delta S alo um region to the southern border of the country and beyond into Guinea Bissau. In the Seneg al portion of the Gambia River, manatees are reported as far inland as Niokolo Koba Natio n al Park during the rainy season ( Hill et al. 1998, Ba et al. 2008, P. Brillant pers. comm. ), and they are also reported in this river across the border in Guinea ( ThÃ©ophile et al. 2008 ). In the Casamance River manatees are reported as far east as Kolda, a lso at Basse Casamance Nation al Park, and near the mouth of the river they can be seen daily drinking from a freshwater spring at Point St. George ( Ba et al. 2008 , Keith Diagne, unpublished data). The only reports of recent manatee hunting in Seneg al are f rom the Delta S alo um region ( IUCN and Parc Nation al du Delta du S alo um 2005 , Kei th Diagne, unpublished data). The Gambia In T he Gambia, numbers are thought to have declined, but as of 1993 the manatee was still numerous in the Gambia River . They have been fully protected for many years but are still believed to be hunted extensively (Perrin 2001, Jallow 2008 ). Manatees are reported throughout the Gambia River and its tributaries, and in Niumi Nation al Park,
36 Bao Bolong Nation al Park, Kiang West Nation al Par k, and the Tanbi Wetlands Complex. There are no estimates of population numbers and little current research or conservation activities. In 2014 manatee hunting was reported in Jareng in the Niamina East, and reports of incident al ly caught manatees were rep orted from Jareng as well as Tendaba in the Lower River Region (D. Saine pers. comm.). Manatees are also affected by the destruction of their habitat due to deforestation of mangroves ( Jallow 2008). Guinea Bissau Guinea Bissau at one time was considered t o be one of the last sanctuaries of the manatee, because of the relatively undisturbed state of its mangroves, wetlands and river systems ( Schumann 1995 , Powell 1996). Silva and AraÃºjo (2001 ) found that manatees occupied a wide variety of habitats and were most abundant around the Bijagos Archipelago. Based on interviews, a tot al of 256 sightings involving 439 individu al s interviewed were reported. Powell (1996) reported se eing about 20 individual sterna from manatees hunted in the Bijagos Archipelago by a single Seneg ale se fisherma n who season al ly fished there. In 1997, the government signed an agreement with IUCN to develop a Nation al Plan for Conservation of the West African Manatee in Guinea Bissau, and some training and survey work started, but the work stopped when the war started in 1998 and has never resumed (Silva and AraÃºjo 2001). The major source of mort ali ty before the war was accident al capture in fishing nets; they were not extensively hunted. Silva and AraÃºjo (2001) reported 209 manatees killed between 1990 1998, approximately 23 per y ea r. More recently, manatees have been advertised by a private company for export on the internet, and two were exported to the Toba Aquarium in Japan in 1996 ( Asano and Sakamoto 1997, Kataoka et al. 2000).
37 Guinea Little information is available on the manatee in Guinea. The country has extensive suitable habitat (Powell 1996), but no systematic studies have been carried out ( Barnett and Prangley 1997). M anatees occur in the Soumba, KonkourÃ© and Fat al a estuaries, S angareyah Bay, Rio Komponi, Rio Nunez and the border area of the Guinean southern coast, notably in the Benty estuary ( ThÃ©ophile et al . 2008 , Camara 2011). In the eastern side of Guinea, manatees are reported in the Niger, Tinkisso and Gambia Rivers and at Haute Niger Nation al Park ( ThÃ©ophile et al. 2008). In Boffa and BokÃ©, manatees are more easily observed during the rainy season, in channels when they are feeding ( ThÃ©ophile et al. 2008). Manatees are reported in the estuary at DubrÃ©ka, and a new research project is about to begin studying them there (O. Camara, pers. comm.) In 2009 a manatee carcass was recovered in KanfarandÃ© (C. Kpoghomou pers. comm.) , and two manatee c al and Benty in 2011 an d 2013 respectively (Camara 2011, O. Camara pers. comm.). According to CITES records, six live manatees were exported from Guinea to China in 2008. Sierra Leone The presence of manatee s has long been recorded throughout Sierra Leone. Early work confirmed i ts occurrence throughout the coast al areas of Bonthe and Pujehun districts and the Sierra Leone River estuary as well as in all rivers ( Lowes 1970, Robinson 1971 ). Reeves et al . (1988) reported their presence widely in Sierra Leone, including the Great Sca rcies, Little Scarcies, Sierra Leone, Sherbro, Wanje and Sewa river systems and Lake Mape. Manatees migrate between the M al em, Shenge, and Wange River s and probably move upstream near Teboh, whilst their presence is confirmed at
38 Yeluba Island ( Reeves et al . 1988). The southern side of the coast al area (Shenge to Sulima) is believed to contain the most important numbers of manatees, but there are no data on population numbers ( Siaffa and J al loh 2008 ). Recent work by Siaffa and J al loh (2006) reported that the ir occurrence in many wetlands is season al , with high numbers seen up rivers during the rainy season when the water level is high. They are also frequently reported devastating rice fields during the rainy season, and with the increase of the water level o f water there is also an increase in accident al captures in fishing nets ( Siaffa and J all oh 2008). As of the late 1980s the species was still widely distributed in the country, but the catches at that time were thought to be unsustainable. The manatee s are trapped, netted and harpooned (Powell 1996). Alt hough the manatee is considered sacred in northern Sierra Leone , and therefore not directly hunted, when incident al ly caught it is still eaten ( Siaffa and J all oh 2008). Hunting is most prev ale nt in the wetla nds of southern Sierra Leone, where manatee meat and oil are relatively common household food items, selling for approximately 2,500 3,000 Leone (less than 70 cents in USD) per h alf kg piece (Siaffa and J all oh 2006). Heavy hunting activity is also reported in river mouths, particularly in mangrove areas ( van der Winden and Siaka 2005 ). Liberia Manatees have been recorded from the St. Paul River, Mesurado River, the lower Moro, St. John River, as well as the Cestos and Sankwen Rivers and in the Piso Lake re gion (Powell 1996). Information collected from fishermen and riparian communities in 2006 indicated that manatees occur in the Cav al la River estuary in the southeast ( Wiles and Makor 2008). The main threats to manatees in Liberia are reported as accident al
39 entanglement in fishing nets, direct hunting, boat collision (propeller injuries) , and habitat destruction ( Wiles and Makor 2008 ). No information is available on status. CÃ´te d'Ivoire (Ivory Coast) Information about manatee distribution and relative abun dance in Ivory Coast was collected for the first time between 1978 and 1983 by the German Technic al Assistance Mission (MATA), which surveyed 107 sites and estimated that the number of manatees living in Ivory Coast was well below 850 individu al s ( Roth and Waitkuwait 1986). Akoi (2004) reported that the manatee occurs alo ng the entire coast al region of Ivory Coast, as well as in all main rivers and their estuaries, and alm ost all lagoons. According to findings of surveys and the movements of individu al s mon itored by VHF radio telemetry from 1986 1988 and 2000 2002 (Powell 1996, Akoi 2004), six main areas of occurrence were identified: the Aby Tendo Ehy lagoon complex with the estuaries of River Tanoh and River Bia; the EbriÃ© lagoon complex with the mouth of River ComoÃ©; the west EbriÃ© lagoon complex with the mouth of River Agneby; the Tagba Makey Tadio Niouzoumou lagoon complex with the mouths of River Bandama and tributaries mo uths of the Sassandra, San Pedro and Cav al ly Rivers. Groups of 10 to 20 individu al s were observed sever al 2004). Human development is believed to be the reason why manatees have completely di sappeared from waters around Abidjan (Akoi 2004). A program of research and education led by K. Akoi from 1986 until his death in 2009 was the longest running manatee research program in Africa to date. The population was tentatively estimated at 750 800 m anatees in five to six sm all , isolated populations (Akoi 2004). Illeg al hunting is still a problem, as is habitat destruction including development and dams ( the
40 Kossou dam in the Bandama River and the Buyo dam in the Sassandra River) . Hunting is illeg al , but continues, with traps, harpoons, hook lines, baited hooks and nets all used (Roth and Waitkuwait 1986, Nicole et al. 1994, Powell 1996, Akoi 2004). There was some success in educating potenti al hunters and in enforcing the hunting ban in some areas, wi th the aid of Wildlife Conservation Internation al ( Akoi 2000 ), but unfortunately many programs ended after the death of Akoi in 2009 , and there are currently no research or conservation activities taking place. According to CITES records, four live manatee s were exported to Taiwan in 2004. At least two of these were sent to Farglory Ocean Park; one of these died in 2010 and the second is still in captivity at the park (T. Mignucci, pers. comm.). Ghana The African manatee occurs in coast al and inland waterw ays in Ghana, particularly in the Afram arm of Lake Volta, and in the rivers Dayi, Asukawkaw, Obusum, Sene, Digya and Oti (Ofori Danson 1995 ). Manatees occur in Abi, Tano and Ehy lagoons in the southwest of Ghana (Roth and Waitkuwait 1986).They are also fo und in the River Tano, the lagoons and swamps associated with the lower Volta and in Lake Volta (formed by construction of the Akosombo dam). There are reports of sightings in the tributaries of the river Tordzie, such as Lolo, Al tra, and H ortor in the sou thern area of Tongu. However, there is no report of its presence in the White or Black Volta Rivers. Ofori Danson and Agbogah (1995) concluded that the confluence of Oti could define t he upper limit of the distribution area of the manatee in Ghana. The Afr ican manatee has become very elusive in Lake Volta under stress from declining water levels and hunting (Ofori Dansen pers. comm.). In addition, environment al degradation in the Lake Volta is on the increase. The situation continues to worsen as the human populat ion of Ghana
41 continues to grow, urbanization and tourist development increase and human activities become diversified (Ofori Dansen pers. comm.). As a result, there is growing concern to educate the populace and also to facilitate the Ghana Wildlife Division charged with the al and aquatic wildlife resources to take action . Leg al protection has been established for manatees in Ghana, and the Wildlife Division prohibits trade and hunting of manatees. H owever, the enforcement of wildlife laws protecting the manatee is frustrated by a lack of resources, manpower and limited awareness of existing regulations (Ofori Dansen pers. comm.). In 2007, the Earthwatch ng conservation of the W est African this project, no manatees were encountered , suggesting that the stock could be too depleted to support further remov al and there is urgent need for enforcement and conservation measures (Ofori D ansen pers. comm.). Togo Manatees were previously reported in Kpessi, Agbodrafo, AbatÃ©kopÃ© (near AnÃ©ho), and in Lake Togo at Ekpui, Togoville, KÃ©ta, Akoda and KouÃ©nou, but this is no longer the case ( Segniagbeto et al. 2008 ). The manatee is observed in Lak e Togo with its tributaries the Zio and Haho, and in the Mono River. According to the findings of surveys undertaken in 2006 among riparian communities and field observations, the manatee population on Lake Togo is important ( Segniagbeto et al. 2008 ) . Ther e are two concentration areas in Lake Togo: one in the south of the lake, and the other at the junction between the lake and the River Haho. The existence of these two concentration areas would justify the number of skulls observed in the villages of AmÃ©dÃ© hoÃ©vÃ© and Dekpo, which are the closest to those areas. It seems that the confluence between Lake Togo and the river Haho gener al ly supports more manatees than in the Lake Togo Zio
42 area ( Segniagbeto et al. 2008). All interviewed fishermen exploiting Lake To go have confirmed this. The population in the Mono River is very sparse according to fishermen, but there is a reported area of population concentration at AdamÃ©. The northern distribution limit of the species in the Mono is in the village of AgomÃ© Glozou . Young manatees are regularly observed by fishermen, and in Lake Togo, mating herds are observed especi al ly when individu al s are sighted in groups during the flood period of the Haho and Zio Rivers. It is necessary to design a significant map of former an d current manatee areas to establish an efficient conservation status, and to determine ( Segniagbeto et al. 2008). Benin In Benin, the manatee had been thought to be extinct (Perrin 20 01). However, this is not the case (Powell 1996, Tchibozo 2002, Dossou Bodjr Ã© nou 2003, Dossou Bodjr Ã© nou et al. 2006 ). The manatee is sporadic al ly distributed across the whole country ( I chola and T chibozo 2008 ). It occurs in coast al areas, including estuari es and coast al lagoons, in the great rivers in both brackish and fresh water, and in freshwater lakes, with reported sightings from both northern and southern wetlands ( Affomasse 1999, ABE 1999, Guedegbe 2000). In the Niger v all ey, manatees may be encounte red in areas of medium depth at the confluences of the MÃ©krou Niger and the Al ibori Niger, and in the branches of Bello Tounga and Kompa Gourou. Manatees are found in Bonou, WÃ©bossou, SÃ¨kodji GomÃ¨, Ouinhi, Agonli HouÃ©gbo (Kpoto) and Lac SÃ©lÃ© ( Adjakpa 2002, Tchibozo 2002). They are frequently seen stranded in season al lakes o ff the Oueme River during the dry season (Tchibozo 2002, Dossou BodjrÃ© nou et al. 2006 , Chabi YaourÃ© 2012 ). In the Mono River, manatees are seen migrating during high water
43 periods, and a re reportedly concentrated near the village of HÃªvÃª and near the beaches of Avlo in the Boucle du Roy in Benin (Dossou BodjrÃ©nou, 2003, I chola and T chibozo 2008). Manatees are hunted throughout Benin for food and tradition al medicines ( I chola and T chibozo 2008). Nigeria The manatee is distribute d throughout Nigeria including the Niger, Benue and Cross Rivers and their tributaries, as well as coast al ly ( Sykes 1974, Powell 1996, Awobamise 2008). Manatees are present in Lake Kainji above the dam and this popul ation is isolated from other manatees both upriver (by the Kandjadji dam under construction in northwestern Niger), and downriver by the Kainji dam (constructed in 1968). Manatees occur alo ng the length of the Benue River and most of its tributaries, inclu ding the Gongola, Taraba, Donga Rivers, the Pie River as far as Yankari, the Katsena Al a River and the Deb River, which drains Lake Pandam, an important dry season refuge (Powell 1996, Oboto 2002, Sykes 2010 ). The manatee is believed to be depleted through out Nigeria due to hunting and incident al capture during fishing operations, including the use of explosives in rivers (Powell 1996, Oboto 2002, E. Eniang, pers. comm.). It is hunted for its meat, oil and for organs used in tradition al medicine ( Oboto 2002 , Awobamise 2008 ) . There is no effective enforcement of protection laws. Another substanti al threat is habitat destruction due to development and pollution of the Niger Delta by oil development. A fisherman of Sapele district reported that manatees of diff erent ages and sizes were found dead and floating after the Jesse petroleum pipeline fire incident (Oboto 2002). Sever al new studies of manatees in Nigeria have begun in the last five years, including work by University of Uyo in the Cross River region (En iang et al . unpublished data), a project training former manatee hunters in
44 aquaculture as an alt ernative livelihood in Lekki Lagoon near Lagos (B. Dunsin, unpublished data), and a new study of manatees in Lake Pandam ( R. Gbegbaje pers. comm. ) Cameroon In Cameroon, manatees occur throughout the coast al region, from the extensive mangroves and estuarine waters of the Ndian Delta and Bakassi area in the west to the Wouri and Mungo Rivers and the Cameroon Estuary, and south to the Sanaga River and the lower se ctions of the Nyong and Ntem Rivers ( Nishiwaki 1982 , Powell 1996, Noupa 2008 ). Manatees are frequently observed in the Dou ala Edea Wildlife Reserve, the lower Sanaga River, and Lake Tissongo ( Takoukam 2012). Manatees occur in the Sanaga River as far as Ede a, where there is a dam and rapids, and in Lake Ossa (connected to the Sanaga), which has been documented as an important dry season sanctuary (Powell 1996, Takoukam 2012). Inland, manatees are found in the upper Cross River, especi al ly around the Munaya C ross confluence, and they also occur in northern Cameroon in the Benue River, from the Faro River and the Mayo Kebbi to Lakes TrÃ©nÃ© and LÃ©rÃ© in Chad (Powell 1996). Manatees cannot descend the Benue into Cameroon south of the Lagdo dam. Alt hough Grigione (1 996 ) reported that illeg al hunting was very limited in Cameroon, more recent reports have shown otherwise: manatees are regularly hunted in the lower Sanaga region and Lake Ossa, and manatee meat can be found in markets in Dou ala on a regular basis (Powell 1996, Takoukam 2012) . Alo ng the coast, incident al capture in fishing nets and habitat destruction are also problems ( Takoukam et al. 2013 ). According to CITES records, two live manatees were exported from Cameroon to South Korea in 2008, and four more wer e exported to China in 2010. A new marine mamm al stranding network set up in
45 2013 that covers coast al Cameroon, the Wouri and lower Sanaga Rivers and Lake Ossa documented 157 manatee sightings (comprising 349 individu al s) and thirteen incident al ly caught m anatees between February and October 2013 ( Takoukam et al. 2013). Equatorial Guinea The African manatee is present in the coast al areas of Equatori al Guinea on mainland Africa, but is absent from Bioko and AnnobÃ³n islands ( Dodman 2008 c ). The main areas of occurrence are in the Muni and Cogo estuaries, and it most likely occurs in the Rio Woro estuary and the Rio Ecucu near Bata, where one was captured in 1988 ( RÃ©publica de Guinea Ecuatori al 2005). Machado (1998) considered that the Rio Muni area contained prime habitat for the manatee. Bolobo (2001) reported that the Rio Muni supported an appreciable population of manatees. They likely occur in the lower reaches of the MitÃ©mÃ©lÃ© River on the mainland (Powell 1996). In the southernmost part of Equatori al Guin ea, manatees are occasion al ly sighted in Corisco Bay and around Corisco Island where sparse seagrass beds of H alo dule wrightii have been documented (Keith and Coll ins 2007). There are no current manatee research or conservation activities in Equatori al Gui nea . Gabon Gabon may have one of the highest densities of manatees remaining in Africa (Powell 1996, Keith et al. 2006, Keith and Collins 2007, Keith Diagne 2011). Reports of particularl y at the northern ends and in associated rivers (Keith et al. 2006, Keith and Collins 2007). During the rainy season manatees are also frequently sighted in season al ly flooded
46 Longa, Mafoume, Mouaga , and Marimossi Lakes (Keith Diagne unpublished data). Louembet (2008) reported high manatee season al use of the Abanga River and associated lakes, as well as hunting of 133 manatees in this region from 2005 2008. Manatees are reporte d, but are sighted much less frequently in Mondah and Corisco Bays, the Libreville Estuary and Komo River, the lakes of the Ogooue region, and Banio and Fernan Vaz Lagoons ( Mbina 2001, Endamne 2007, Keith Diagne 2011, Keith Diagne unpublished data). Bycatc h occurs in gillnets and directed hunting is reported, primarily in the lower Ogooue River and associated tributaries and lakes in the region of LambarÃ©nÃ©, as well as occasion al ( Louembet 2008, Mvele and Arrowood 2013, Keith Diagne unpub lished data , G. Minton pers. comm.). Republic of the Congo Manatees occur throughout coast al rivers, lakes and lagoons of the Republic of the Congo, and have been documented in Conkouati Nation al Park, and the Kouilou and LoÃ©mÃ© Rivers ( Akoi 1994, Kaya 200 5, B al and BrÃ©heret 2007 , Kaya 2008 ). Surveys of Conkouati Lagoon and adjacent rivers and lakes indicated that manatees are seen primarily in the lakes and at the mouth of Conkouati Lagoon near sandbanks (Akoi 1994, Dodman et al. 2006, Keith Diagne unpubli shed data). Surveys of the Kouilou River region conducted by the NGO Renatura (B al and BrÃ©heret 2007) confirmed manatee presence in Lakes Nanga, Ndinga, and Katina, as well as the following tributaries of the Kouilou: the Mboukou Massi, the Loundji, and th e Midounvo rivers. However, the same study concluded that manatees were not present in Lake Youbi and presence could not be confirmed in Lake K oubambi ( Bal and BrÃ©heret 2007) . In the LoÃ©mÃ© River, manatees are reported primarily in the lower river and its e stuarine areas,
47 and fishermen al so reported that manatees occur in Lake Kayo, south of the LoÃ©mÃ©, in the dry season (Kaya 2005). Democratic Republic of Congo Hatt (193 4 ), All en (1942 ) and Powell (1996) reported that manatees were common in the lower reac hes of the Congo River. Manatees are regularly reported alo ng the coast at Parc Marin des Mangroves and a hunted manatee was observed at Boma in 2013 (S. Mbungu pers. comm.) Status is unknown, but given the frequency of sightings reported across the river on the Angola side ( Collins et al . 2011 ) it is possible there is a sizeable population in the lower Congo River. Powell (1996) reported that a loc al name for the species exists in the upper reaches of the Congo River so it may occur there as well, alt hough other reports for this region ( All en 1942, Dodman 2008b, Hart pers. comm.) do not believe manatees exist above the rapids, and today it would have to be an extremely isolated population since it is unlikely manatees can traverse the extreme gorges in the river west of Kinshasa. Angola Manatees have been reported alo ng the coast from the Congo River in the north, to the Longa River in centr al Angola, but little information is available on abundance or status (Powell 1996, Dodman 2008 a , Collins et al. 2011 ) . Morais (2006 ) found manatees in the Longa River , extending their southernmost range to that river from the Cuanza. Manatees have been documented the length of the Cuanza River and its tributaries and lagoons from the river mouth to Cambambe dam, which, p rior to construction of the dam, was characterized by steep rapids that also presented a natur al barrier (Morais 2006). The manatee s distribution also extends up the Luc al a River (a main tributary on the north bank) for at least 30km upstream from the Cua nza River. Key sites of the river
48 basin include areas at and around Cauigia Lagoon, Cabemba Lagoon, the TÃ´a Lagoons, Quissungo Lagoon and a sm all extension of the Ngolome Lagoon, as well as in the Caua River and in the Massangano region (Morais 2006). Base d upon boat and interview surveys conducted from 2007 2009 in the lower Congo River from the mouth to 40 km upriver, as well as numerous tributaries on the Angola side of the river, manatees are commonly sighted in this region (Collins et al. 2011). One ma natee was sighted by the survey team in the Nzadi Caca tributary of the Congo River in 2007 and it stayed near the boat for 1.5 hours, surfacing approximately every 8 10 minutes (Collins et al. 2011). Only one loc al hunter was identified in all interviews of villages from the mouth to 40 km upriver, and he was interviewed by Keith Diagne before he passed away in 2008. The hunter estimated he had killed approximately three manatees a week (using harpoons and nets) for the past 30 years (Keith Diagne unpublis hed data). Manatees are hunted on the Cuanza River, manatee meat has been seen for s ale in the capit al of Luanda, and a fisherman from a village at a lagoon on the Bengo River reported 77 manatees killed during 1998 ( Ron 1998, Morais 2006). Xavier (2011) documented a manatee hunted by net capture in Lake CaÃºmba. Mali Manatees are found throughout the entire Niger River system of M ali and in the Bani River but may have been reduced by hunting (Powell 1996, Berthe 2011). The Maka la dam at Segou (constructed in 1945), another planned dam at Taoussa, as well as addition al dams in Guinea and Niger, have chopped M ali populations into sm all er habitat areas, and is of great concern for the conservation of the species in that country. There are no estimates of population numbers. In 2010 a manatee carcass was recovered in Koulikoro ( Timbo 2010), c onstituting the furthest
49 inland record for the species, alt hough there are reports of sightings further west from Bamako ( Kienta et al. 2008 ). Manatees are frequently sighted in the Bani River near DjennÃ© and Mopti, and have been documented regularly drinking from a freshwater spring near the village of Sokon (Berthe 2011). Manatees are also frequently reported in Lake Debo and the surrounding area in th e inland Niger delta ( Kienta 1982, Kombo and Toko 1991 , Kienta et al. 2008). In western M ali manatees occur in the Seneg al River up to the Felou dam at Kayes , which prevents them from moving further upriver (Kei th Diagne, unpublished data). Niger Poche (1 973) d study by Ciofolo and Sadou (1996 ) found manatees throughout the Niger River in Niger. In the northern Ayorou region (extending from the border with M ali to Tounga Faire), mana tees were sighted throughout the year and a large number of skulls collected by hunters indicated heavy poaching (Ciofolo and Sadou 1996). In the region of Park W, manatees were sighted in the main channel of the Niger River and a fem ale was accident al ly c aptured in 2002 (Ciofolo and Sadou 1996, Issa 2008 ). In 2010 a new manatee study was initiated in this region (Boubacar 2010). In the southern Niger Boumba Gaya section of the river, manatees have been observed during low water periods, which may indicate migration elsewhere during other times of the year (Ciofolo and Sadou 1996). A m ale manatee was captured in this area in 2004 but only survived in captivity for two weeks ( Issa 2008 ). In June 2012 , sever al manatees were observed together in the Niger River in downtown Niamey near a major bridge ( a possible mating herd). One fem ale from this group was harpooned and killed by a hunter; the carcass was confiscated and biologic al samples were collected (Boubacar 2012). Manatees are
50 heavily hunted throughout Nig er for meat and organs (which are believed to have he ali ng powers), and alt hough the species is fully protected under national laws , there is little if any enforcement ( Issa 2008). Chad Manatees were once abundant in the Chad basin including the Chari, Ba m ingui, Bahr Kieta, and Logone R ivers, but became rare by 1924 and are now believed to be extinct in this region ( Hatt 1934, Powell 1996, S al kind 1998 ). Lake Chad is greatly reduced in size and desertification is a big problem there. Today manatees occur on ly in Lakes LÃ©re and TrÃ©ne , as well as the Mayo Kebbi River which feeds the lakes from the west ( Idriss 2008 ). In a survey in 1995, manatees were found to be less abundant than formerly , but not uncommon in both lakes (S al kind 1998). Hunting continued on t he rivers and lakes, despite enforcement efforts. The anim als were sought mainly for their oil, which is shipped with dried meat to Cameroon (Perrin 2001). As of 2011 it has been reported that there are few manatees left in Lakes LÃ©re and TrÃ©ne, and that t he loc al come across the border to kill them at night (J. Hart and A. Wachoum pers. comms.) . Burkina Faso Manatees inhabit all of the nations that surround Burkina Faso (M ali , Ivory Coast, Ghana, Togo, Benin and Niger) , but have never been reported within this country . They are present in Volta Lake above the dam (see Ghana above). However, Perrin (2001) could find no mention of its occurrence in the upper tributaries of the Vo lta (White Volta, Red Volta and Black Volta) or in the Mekrou River, which forms the boundary between Burkina Faso and Togo/Benin and drains the wetlands of the Parc Nation al de l'Arly.
51 Pending directed surveys, its occurrence there must be considered poss ible (Perrin 2001). African Manatee Habitat and Ecology African manatees inhabit a wide range of habitats across their enormous distribution, from lagoons w ithin equatorial rainforests to rivers at the edge of the Sahara Desert, around coast al islands in t he Atlantic Ocean, and in many other habitats in between. The speci es lives in large and sm all rivers, coast al bays and estuaries, freshwater and s alt water lagoons, lakes, reservoirs , and flooded forests and savannahs . They have been recorded in freshwater , brackish and saltwater, and in general their habitat requirements seem to be similar to T. manatus ; they require calm water with access to food and fresh water ( Reep and Bonde 2006 , Marsh et al . 2012). I n Seneg al , The Gambia, and Guinea Bissau, manatees are attracted to a large network of freshwater seeps or springs that are found in nearshore marine habitats , which they are frequently observed drinking ( Powell 1990, Keith Diagne unpublished data ). They may transit areas of unsheltered coast, but they are usually rare in these areas. Optim al coast al habitats for manatees, based on the movements of radio tagged manatees in Ivory Coast and a number of reported sightings from other areas are: a. coast al lagoons with abundant growth of mangrove, aquatic plant s and/ or emergent herbaceous growth; b. estuarine areas of larger rivers with abundant mangrove ( Rhizophora racemosa, R. harrisonii, R. mangle ) in the lower reaches and lined with grasses further upriver; c. sh all ow (<3 m depth) and protected coast al areas with fringing mangroves or marine macrophytes, particularly the seagrasses Ruppia sp., H alo dule wrightii or Cymodocea nodosa (Powell 1996, Keith and Collins 2007, Keith Diagne and Powell 2014 ) . Powell (1996) reported that in riverine habitats that have
52 ma jor seasonal fluctuations in flow rates and water levels , manatees seem to prefer areas that have access to deep pools or connecting lakes for refuge during the dry season , and season all y flood into swamps, savannahs or forests during the rainy season . In C entr al African countries including Gabon and Republic of the Congo, manatees use coast al lagoons during the dry season and then migrate up rivers to feed in flooded forests and lakes during the rainy season (Keith and Collins 2007 , Keith Diagne unpublishe d data ) . Manatees that live extremely far inland in rivers in countries such as Seneg al and M ali use specific feeding areas where year round aquatic and shoreline plants occur , and likely also feed on freshwater mussels during the dry season. They spread o ut onto flood ed plains during the rainy season to feed on emergent vegetation. African manatees feed primarily on vegetation, and 70 species of plants have been documented as manatee food to date (Appendix ) throughout their range (Villiers and Bessac 1948, Powell 1996, Reeves et al. 1998, Akoi 2004, Ogogo et al. 2013, Keith Diagne this study ). In countries including Seneg al (Powell unpublished data, Keith Diagne unpublished data) and Sierra Leone (Reeves et al. 1988), manatees are also reported to eat sm all I n many countries, including Seneg al , The Gambia, M ali , Gabon, and Angola, m anatees are also reported to eat freshwater and estuarine mollusks , and shell remains have been found in manatee stomachs (Powell 1996, Collins et al. 2011, Appendix ). African manatees are mostly solitary, with mothers and c alv es forming the princip al soci al unit. Manatees will often rest together in loose, sm all groups of two to six individu al s and mating herds have been observed in Gabon, Seneg al , Nigeria and
53 Sierra Leone (E. Eniang pers. comm., Keith Diagne unpublished data). In some African countries such as Ivory Coast , manatees feed princip al ly at night , and travel in the late afternoon and at night (Powell 1996), but in other countries such as in al they can be seen traveling and feeding during all hours of day and night (Keith et al. 2006, Keith and Collins 2007, Keith Diagne unpublished data). They usu al ly rest during th e day in water that is 1 2 m deep , sometimes i n the middle of a watercourse, hidden in mangrove or other plant roots , or under floating or shoreline vegetation (Powell 1996) . They make little disturbance while swimming. These latter behaviors may be due to hunting pressures, and it appears that manatees that are easily sighted during daylight hours gener al ly occur in places where little hunting occurs. Sirenian Conservation Genetics All four extant sirenian species (three t richechids and one dugong) have be en listed as either Endangered or Threatened under the U.S. Endangered Species Act of 1973, all are listed as Vulner able on the IUCN Red List (IUCN 20 14 ), and all are protected under the Marine Mamm al Protection Act of 1972. However, many populations of al l four species still face serious threats to their long term surviv al . Genetic an aly ses are a useful tool that has been increasingly used over the past 2 5 years to assist biologists and managers in understanding manatee population structure, genetic divers ity, evolution, and phylogeography. However, in comparison to many other species, genetics research on sirenians is still in its infancy, and alt hough a majority of previous work has focused on the West Indian manatee ( T. manatus ), in recent years genetics studies have been undertaken with all extant sirenians.
54 In the 1970 s two studies determined the chromosome numbers for the Amazonian manatee (56 chromosomes; Loughman et al. 1970) and the West Indian manatee (48 chromosomes; White et al. 1976 ). In both s tudies chromosome counts were determined by karyotyping from isolated cells (Loughman et al . 1970, White et al. 1976) . It is of interest to note that the chromosome number has still not been determined for the African manatee , primarily due to the difficul ty in obtaining a blood sample that can be transported to a laboratory rapidly enough. Chromosome number is important not only in that it can differentiate between species, but within a species chromosomes can have variation between individu al s such as inv ersions, deletions or mutations that may result in genetic diversity (Frankham et al. 2009) . The first manatee genetics research was a study of Florida manatee ( Trichechus manatus latirostris ) tissue from 59 individu al s from both coasts of peninsular Flori da by technique had only recently been invented in 1983 (Mullis and F aloona 1987), and was deter min e allozyme (protein) phenotypes and identified 24 loci, of which 10 ( 41.67%) were polymorphic . However, the results also indicate d that there was little gene diversity between regions throughout Florida, and therefore gave the first indication that the species may have low genetic di 1988). In 1993, Bradley et al. conducted the first an aly sis of m t DNA using PCR to examine cytochrome b from liver and blood samples from three Florida manatees. Mitochondria are organelles fou nd in the cells of all eukaryotic anim als which contain circular, double stranded DNA that is matern al ly inherited (Frankham et al. 2010).
55 Mitochondri al DNA has a high mutation rate and is highly variable in anim als , which makes it useful in defining popul ations within species and for determining taxonomy (Frankham et al. 2010). Within mtDNA are specific regions known as loci, which p rimarily code for proteins and t ransfer RNA (tRNA). The control region is a non coding segment of the mtDNA which is useful i n determining genetic diversity between populations because it typic al ly has a higher rate of mutation than co ding regions such as cytochrome b (Frankham et al . 2010). The mtDNA cytochrome b gene has been used in numerous anim al studies because it has beco me known for its utility both in studying population level relationships as well as recent divergence levels (Farias et al. 2001). In Bradley et al. (1993) , DNA was isolated using phenol extraction followed by an ethanol precipitation. Univers al primers f or human cytochrome b developed by Kocher and White (MVZ3 and MVZ4; 1989) were used because they had been previously shown to amplify homologous genes in numerous other anim al species (Bradley et al. 1993). A tot al of 225 bases composing 75 codon triplets were examined for each sample (Bradley et al. 1993). No statistic al an aly ses are reported by this study, and the sequence comparisons to each other and to those of other species may have been done by hand. This study reported that all the manatee sequences were identic al , and while there were only three samples in this study, the authors speculated that, given the Florida manatee may lack polymorphism and that all manatees in Florida constituted a single population (Bradley et al. 1993). A separate phylogenetic result of this study was the identification of two codons as unique to the manatee and the African elephant
56 ( Loxodonta africana ), which were different from codons in the same positions for 20 other anim als (including humans) lending credibility to the previous morphologic al placement of these two s pecies as subungulates (Novacek 1992, Bradley et al . 1993). Garcia Rodriguez et al. (1998) conducted the first genetics st udy that an aly zed the mtDNA control region displacement loop (D loop) of T. manatus and T. inunguis (Amazonian manatee) samples. D loop is a non coding region of the mtDNA and is useful in determining genetic diversity between populations because it has a higher rate of mutation than coding regions such as cytochrome b (Frankham et al. 2010). This was also the first phylogeographic al study of any sirenian (Garcia Rodriguez et al. 1998). DNA was isolated from eighty six manatee tissue, blood, and bone sample s from eight locations in Florida, the Caribbean (Puerto Rico, Dominican Republic, and the Caribbean coast of Mexico), and South America (Brazil, Venezuela, Colombia, and Guyana) using a phenol chloroform methodology devised by Hillis et al. and then ampli fied using PCR (Hillis et al. 1996, Garcia Rodriguez et al. 1998). Southern et al. (1988 ) developed t he CR 4 primer from a conserved segment of the control region, and CR 5 was developed from the tRNA Pro consensus sequences for dolphin, cow, and mouse . Ga rcia Rodriguez et al. ( 1998 ) were the first to use CR 4 and CR 5 primers for manatees, and they have been u sed in many subsequent studies . Statistic al tests used for data an aly nd haplotype diversities (Nei 1987) both within regions and for all samples, a Kimura 2 parameter alg orithm (Kimura 1980) to determine sequence divergence within and between populations, and a Chi square test of haplotype frequencies between populations (Garcia Rodriguez et al. 1998). An an aly si s of molecular variance ( AMOVA) was used to determine genetic diversity
57 between populations and estimates of gene flow, with the following equation used to determine gene flow due to migration: Nm = 1/2(1/ F ST 1) (1 1) where Nm is the number of m igrants per generation, and F ST is the heterozygosity of the sub populations ( Garcia Rodriguez et al. 1998, Frankham et al. 2010). P hylogenetic A n aly sis U sing P arsimony ( PAUP; Swofford 1993 ) software was used to determine the number of site differences bet ween haplotypes (Garcia Rodriguez et al. 1998). Haplotypes are a combination of all eles on adjacent loci that are inherited together, which are indicators of genetic diversity, and can tell us about relatedness between populations . In Garcia Rodriguez et al. ( 1998) sixteen haplotypes were reported for T. manatus and of these, 12 were found only in a single location, which indicated that they were endemic haplotypes. The significance of this result is that it indicates definite population boundaries between sites and low dispers al , and thus low genetic mixing (Garcia Rodriguez et al. 1998). Migration events are believed to occur, but not frequently enough to homogenize the different populations (Garcia Rodriguez et al. 1998). Eight haplotypes were reported for T. inunguis , all from one population (Garcia Rodriguez et al. 1998). T. manatus haplotype diversity ( HD = 0.839) and nucleotide diversities ( = 0.040) were both reported to be high, and sequence divergence was also considered high in comparison to mtD NA control regions of other marine mamm als (Garcia Rodriguez et al. 1998). In the study, only one T. manatus haplotype each was discovered for the Florida samples (n = 23, collected from both east and west coasts), as well as the samples from coast al Brazi l (n = 6) and Mexico (n = 6), alt hough Brazil and Mexico had low sample numbers and later research by Vianna et al .
58 (200 6 , discussed below) identified addition al haplotypes for these locations (Garcia Rodriguez et al. 1998). T he single Florida manatee hapl otype could be explained by the species experiencing a bottleneck caused by a founder effect as the result of a recent re colonization of Florida within the past 12,000 years at the end of the last ice age (Garcia Rodriguez et al. 1998). Other regions stud ied showed high haplotype diversity, such as samples from Colombia and Guyana, which had over 20 polymorphisms (Garcia Rodriguez et al . 1998). Gene flow estimates ( Nm ) ranged between 0.0 3.5 (4 is the highest v alu e and represents panmixia), but estimates of mov ement between most locations were low, with exceptions being between Puerto Rico / Dominican Republic, Colombia / Mexico, and Guyana / Brazil (Garcia Rodriguez et al. 1998). The phylogeographic results in this study were unexpected in that they ind icated three distinct haplotype clusters that did not ali gn with the previous taxonomic delineations of the two subspecies, T. m . l atirostris (the Florida manatee) and T. m . manatus (the Antillean manatee ) (Domning and Hayek 1986, Garcia Rodriguez et al. 199 8). These three clusters included Florida, Puerto Rico, the Dominican Republic, and five haplotypes from Colombia (cluster I), Mexico, Venezuela, and 17 haplotypes from Colombia (cluster II), and Guyana and Brazil (cluster III), all of which mostly showed distinct geographic separation but also indicated long distance dispers al (Garcia Rodriguez et al. 1998). Time of divergence of the three clusters was estimated based on control region an aly ses and fossils using other species, and the estimates were highl y variable: 0.5 1 Mya based upon terrestri al mamm al rates of chang e (~8 15% / million years), or 3.5 7 Mya based on marine mamm al rates of change (1 2% / million years) (Garcia Rodriguez et al. 1998). An addition al unexpected result of the study was
59 that t hree T. manatus samples from Guyana also had a T. inunguis haplotype (Garcia Rodriguez et al . 1998). These were hypothesized to be either the result of an ancestr al T. inunguis haplotype that was retained by T. manatus , a range expansion into Guyana by T. inunguis , the result of a possible mixing of the two species in a captive situation in the 1960 s, or hybridization between the two species (Garcia Rodriguez et al. 1998). This study was v alu able not only as the first large sc ale phylogeographic study of T . manatus , but also from a conservation perspective, because the clusters could be used to redefine management units (Garcia Rodriguez et al. 1998). In 2000, as part of her doctor al dissertation, Garcia Rodriguez developed the first eight nuclear DNA micr osatellite markers for T. manatus latirostris using DNA isolation and PCR methodology reported in Garcia Rodriguez et al. (1998). Unlike mitochondri al DNA, nuclear DNA is inherited from both parents, so it can be used to determine pedigrees, and it mutates more rapidly so it can be used to identify individu al s (Pause et al. 2007 , Frankham et al. 2010). Short tandem repeats of base pairs of nuclear DNA, known as microsatellites, are gener al ly used as molecular markers (Frankham et al. 2010). Skin samples (n= 223) from manatee carcasses from both coasts of Florida were used and populations were assumed to be the four management units within the state (Northwest, Southwest, St. Johns River, and Atlantic coast) (Ga rcia Rodriguez 2000). Statistic al an aly ses includ all elic distribution between F statistics to examine genotype distribution, and Rho st , which used allele size estimates in a measure of F st , all of which were run using GENEPOP softwar e version 1.2 (Garcia Rodriguez 2000). Results indicated that microsatellite variation (both genotypic and all elic frequency distributions)
60 did occur between east and west coasts in the Florida manatee, alt hough high levels of gene flow did occur be tween coasts an d populations (Garcia Rodriguez 2000). Therefore , this study recommended that the Florida manatee should be considered a single Evolutionarily Significant Unit (ESU) consisting of two management units based upon the east and the west coast m anatee populations (Garcia Rodriguez 2000). Low levels of all elic diversity again indicat ed that Florida manatees had likely experienced a bottleneck due to founder effect, but this work also confirmed that differentiation between populations was possible using the finer sc ale an aly ses of mi crosatellites (Garcia Rodriguez 2000). In 200 6 , the mo st extensive sirenian genetics study to date was conducted by Vianna et al . It focused primarily on T. manatus , and included 189 T. manatus samples from ten locations throughout Florida, the Caribbean, and the Atlantic coast of South America; 93 T. inunguis samples from three countries within the Amazon River basin (Brazil, Colombia, and Peru); and 6 T. seneg ale nsis samples from four countries (Guinea Bissau, Ghana, Ni ger, and Chad) (Vianna et al. 200 6 ). Sequences (n = 42) from Garcia Rodriguez et al. (1998) were also included, as well as three dugong ( Dugong dugon ) samples used as an outgroup for phylogenetics an aly ses (Vianna et al. 200 6 ). This population level resear ch examined the phylogeny of the three trichechid species as well as phylogeography, primarily of T. manatus . Mitochondrial DNA was isolated from skin, musc le or bone samples using phenol chloroform methodology or QIAGEN DNeasy kits ( Qiagen Inc., V ale ncia, CA) , then the control region (D loop) and cytochrome b portions were amplified and sequenced (Vianna et al . 200 6 ). PCR procedures were modified slightly from Garcia Rodriguez et al . (1998) by the use of
61 different primers for D loop ( L15926 and H16498; Koc her et al. 1989) and modification of some cycles and anne ali ng times (Vianna et al. 200 6 ). Statistic al tests included population pairwise F ST c alc ulated using ARLEQUIN software, c alc ulation of Nm and a Mantel Test to determine gene flow and significance of distance between populations, nucleotide evolution models were tested using MODELTEST software (version 3.06; Posada and Crand all 1998), and use of F ST and geographic coordinates of populations to test for historic al barriers to gene flow (using BARRIER s oftware version 2.2, Manni et al. 2004, Vianna et al. 200 6 ). One limitation of using mtDNA was that statistic al an aly ses were only a measure of matern al gene flow since there is no patern al contribution to mtDNA (Vianna et al. 200 6 ). AMOVA was used to c alc ulate differences in genetic structure between different populations, and for rates of divergence this s tudy used a rate of 2% per million years based on the rate previously c alc ulated for the dugong based on fossils (Vianna et al. 200 6 ). West Indian mana tee mtDNA an aly sis in Vianna et al. (2006) identifi ed three similar clusters to those in Garcia Rodriguez et al. (1998), but with slight differences based upon samples included from two addition al locations (Belize and French Guiana) , and with the identifi cation of haplotype A in Mexico, which linked it to both clusters I and II (Vianna et al. 200 6 ). The clusters included the following locations: cluster I (FL, Puerto Rico, Dominican Republic, Mexico, and Colombia), II (Mexico, Belize, Colombia , and Venezue la) and III (Brazil , French Guiana, and Guyana). Twenty haplotypes were identified over all and Col o mbia again had the highest haplotype diversity, indicating it could either be a mixing zone or the origin of the species (Vianna et al. 200 6 ). Across the spe cies range it was observed that genetic diversity decreased in the extremes of
62 latitudin al distribution as in a linear stepping stone dispers al model, with Fl orida and Brazil presenting less diverse populations (Vianna et al. 200 6 ). An association test bet ween genetic and straight line geographic distances between populations found no correlation, but the test between genetic distances and coastline geographic distances did correlate, which supported the hypothesis that manatees migrate alo ng coasts rather than in the deeper open ocean (Vianna et al. 200 6 ). As in Garcia Rodriguez et al. , this study reported this is likely the result of a founder effect, which may be due to a process of re colonization by few individu al s from lower latitudes to the more north ern and southern regions following glaciation (Vianna et al. 200 6 ). Addition al ly, the barrier an aly ses indicated that two barriers to gene flow had occurred in the past: one that isolated Puerto Rico and the Dominican Republic, and the other that separated both of the Guyanas and Brazil from rest of the north coast of South America (Vianna et al. 200 6 ). The study hypothesized that during lower sea levels during the Pleistocene ice age the Greater and Lesser Antilles had both been peninsular landmasses that cut off manatee movement between them in the north and south Caribbean (Vianna et al. 200 6 ). This study was also the most extensive to date for Amazonian manatees, for which 31 haplotypes were identified (Vianna et al. 200 6 ). These haplotypes are all close ly related but many are rare, which is a pattern that can be typic al after a bottleneck event. Using pairwise differences, the study estimated the bottleneck was 129,216 years ago (Vianna et al . 2006). One of the most interesting results was the identifica tion of a further seven Antillean manatee samples (from Brazil, French Guiana and Guyana) that had Amazonian mtDNA haplotypes, and one Amazonian manatee
63 sample that had an Antillean haplotype (Vianna et al. 200 6 ). This builds upon the data reported in Garc ia Rodriguez et al. (1998) and led to the conclusion that hybridization was occurring (Vianna et al. 200 6 ). Based on the location of the samples, the hybrid zone appears to be the delta and mouth of the Amazon River, north alo ng the coast to Guyana, which mirrors a strong current of water exiting the Amazon River (Vianna et al. 200 6 ). In terms of conservation management of West Indian manatees from Brazil, whose numbers are estimated at less than 500 individu al s, it may be important not to translocate indiv idu als into the hybrid zone, which could further impact their low haplotype diversity (Vianna et al. 200 6 ). Despite the sm all number of samples for African manatees compared to the other species, this study identified five control region haplotypes with hi ghest haplotype diversity (h=0.9333) of all three manatee species (Vianna et al . 200 6 ) . Two clusters were seen, one comprising samples from Guinea Bissau (Y01 and Y02), and the other comprising samples from Ghana, Chad and Niger (Y03, Y04, Y05), three coun tries that are geographic al ly close to each other (Vianna et al. 200 6 ). The phylogenetic distance between the clusters also indicated high diversity within the species (Vianna et al. 200 6 ) . This observed diversity in African manatee samples indicated a str onger population structure than exists in T. inunguis , and the study concluded with the recommendation that a larger sc ale genetics study of African manatees from throughout their range would be informative (Vianna et al. 200 6 ). Results of both control re gion and cytochrome b an aly ses reveal ed large genetic differences between all three species, sign ali ng large timeframes since divergence, on the order of 300,000 to 620,000 years ago (Vianna et al. 200 6 ). The phylogeny results
64 differed between the control region and cytochrome b , however. The mtDNA control region phylogeny suggested that T. inunguis is a sister group to the T. manatus cluster I, indicating that T. manatus is a paraphyletic species, since some cluster I haplotypes appear more related to the Amazonian manatee than the other T. manatus clusters (Vianna et al. 200 6 ). However, the cytochrome b an aly sis suggested that T. inunguis is the most bas al species to both T. manatus and T. seneg ale nsis , and that these two species are more closely related t o each other and had a common ancestor with a long separation time from the Amazonian species (Vianna et al. 200 6 ). interpretation of the results agreed with the cytochrome b data based on genetic an aly ses of all three species, which also agre es with previ ous morphological data (Domning 1994). In many studies , cytochrome b data is proving to be more reliable than the control region because it is less prone to homoplasy, defined as the similarity of genes for reasons other than co ancestry (Vian na et al. 200 6 ). Trichechus manatus = 0.038648) reflecting the three clusters (Vianna et al. 200 6 ). In 2005 a second population genetics and phylogeography study was also published, this one by Cantanhede et al . which focused on T. inunguis . The go als of this research were to use mtDNA to examine species phylogeny and habitat use throughout the Amazon Basin, and then the observed genetic di versity could be used to inform conservation and management measures for the species (Cantanhede et al . 2005). This study used 68 samples of manatee skin primarily collected from captive orphan c alv es that originated from six locations within the Amazon (C antanhede et al . 2005 ).
65 M itochondrial DNA was is olated using phenol chloroform methodology and the control region was amplified by PCR using the same primers as in Garcia Rodriguez et al. (1998). Phylogenetic an aly ses were conducted using PAUP software, an d previously banked (Gen Bank) mtDNA control region sequences for the dugong and African elephant were used as outgroups (Cantanhede et al. 2005). Population pairwise F ST was c alc and haplotype diversity and genetic diversity within populations, and a Kimura 2 parameter model was used to c alc ulate divergence estimates (Cantanhede et al . 2005). Nucleotide evolution was an aly zed using MODELTEST software (version 3.06), a nested haplotype an aly sis was performed to estimate differences between haplotypes and c alc ulate relative haplotype ages, and a nested clade an aly sis was employed to compare haplotype frequencies and geographic al distances relative to time (Cantanhede et al. 2005). Many o f these tests were chosen because they had previously been used in Garcia Rodriguez et al. (1998) and results could then be used to make direct comparisons between the T. manatus results in that study with T. inunguis in this study (Cantanhede et al. 2005) . Thirty one unique haplotypes were identified in Cantanhede et al. (2005), as well as two haplotypes that were identified in T. manatus (H27 and HP) which is considered to be further evidence of hybridization between the two species (Garcia Rodriguez et al. 1998, Cantanhede et al. 2005). Only 24% of haplotypes were identified in more than one individu al , indicating a high rate of endemism (Cantanhede et al . 2005). Nested clade an aly sis showed that the most ancestr al haplotype was H2 and also that there w as not a significant correlation between genetic diversity and geographic distance
66 (Cantanhede et al. 2005). Results of an AMOVA indicated that genetic variability between populations was 94.6% and genetic variability within populations was 5.4%, yet after correction factors were applied no significant differences were found between loc ali ties as measured by F ST (Cantanhede et al. 2005). Over all , both nucleotide and while gene flow was restricted, there was evidence of dispers al throughout the Amazon populations sampled, and both factors combined indicated that the T. inunguis populations were panmictic (Cantanhede et al. 2005). These results were interesting given the hig h numbers of manatees hunted historic al ly in the Amazon, and indicated that genetic diversity had not been severely reduced by hunting (Cantanhede et al. 2005). Also , as commerci al hunting had ended 30 40 years before this study, it was hypothesized that t he genetic results indicated that the species was experiencing populati on expansion after a bottleneck (Cantanhede et al . 2005) . T his result was also hypothesized in Vianna et al. (200 6 ) and illustrated by the median joining network showing a single star s haped cluster with sever al bas al haplotypes at its center and numerous more recent (and less prev ale nt) haplotypes radiating outward. In terms of conservation measures, this study proved that the species could be considered a single population for manageme nt purposes, as well as for captive manatee releases (Cantanhede et al. 2005). MODELTEST tested 56 evolutionary models and determined the best fit for the observed sequences was an HKY85 model with gamma distribution, which resulted in four distinct cluste rs, one for T. inunguis as well as the three T. manatus clusters previously reported by Garcia Rodriguez et al. in 1998 (Cantanhede et al. 2005).
67 Interestingly, the phylogeny indicated that T. manatus cluster I could be sister to T. inunguis , which would m ean T. manatus is paraphyletic, but this was also not significantly more likely than T. manatus being monophyletic (Cantanhede et al. 2005). Divergence estimates were c alc ulated assuming a divergence rate of 2% substitution / site / million years, and yiel ded a result of genetic divergence between T. inunguis and T. manatus ranging from 6.3 8.2%, which is consistent with a time period ranging from 3.1 4 million years since divergence (Cantanhede et al . 2005). This is plausible as it corresponds with the time period of Andes uplift in South America that Domning (1982) hypothesized caused the isolation of the ancestors of T. inunguis within an enclosed Amazon inland sea. Doctor al dissertation research on the population genetics of dugongs in Austr ali a was published by McDon ald in 2005 and is the mo st extensive study conducted for this species to date. The complete dugong mitochondri al genome had been previously published as part of other studies focusing on Eutherian and Afrotherian phylogenies by Arnason et al. (2002) and Murata et al. (2003), but no research had previously examined mtDNA or nuclear DNA within specific dugong populations (McDonald 2005). This study examined the mtDNA control region of 115 dugong samples to determine population level geneti cs and phylogeography (McDonald 2005). Similar to previous studies, statistic al an aly ses included population pairwise F ST c alc ulated from ARLEQUIN, AMOVA, nested clade an aly sis, and isolation by distance (IBD software). Results of this study indicated that mtDNA control region sequences from dugongs in Asia were significantly different from Austr ali an sampled populations, and within Austr ali a there was significant difference between populations on the western and
68 eastern sides of the continent. It was hypot hesized that closure of the Torres Strait by a peninsula during the last glaciation period would have separated these populations and likely account for the observe d genetic differences (McDonald 2005). Nuclear DNA microsatellite loci that had been previou sly reported for T. manatus (Garcia Rodriguez et al. 2000) was used to test for dugong population markers which indicated much higher all elic diversity and level of gene flow than those of T. manatus (McDonald 2005). Dugong isolation by distance results in dicated significant effects across the continent, but identifi ed genetic mixing of the two populations where they overl ap on the north coast (McDonald 2005). A comparison of mtDNA with nDNA showed m ale biased gene flow, and over all high levels of gene flow between populations over the huge species range in Austr ali a made it difficult to de fine management units (McDonald 2005). In 2007 and 2008, two publications describing the development of addition al nuclear DNA microsatellite markers to study T.m. latiro stris were published by Pause et al. (10 microsatellite markers) and Tring ali et al. (18 addition al microsatellite ma rkers) . These markers are now being utiliz ed in current studies of T.m. latirostris from northwest Florida that have not yet been published (M. Hunter, pers. comm. ). Nuclear DNA will be particularly useful to study the four management units of the Florida manatee, since all individu als of the sub species share the same single mtDNA haplotype. The Florida manatee microsatellites indicate great er heterozygosity than other T. manatus populations, indicating that they are recovering from the bottleneck due to a founder effect occurring approximately 12,000 years ago (Tring ali et al . 2008 b ). Broderick et al. (2007) also developed 26 microsatellite markers for the dugong. Hunter
69 et al. (2010b) expanded the number and versatility of microsatellite markers through the identification of additional new primers, as well as a cross species amplification of microsatellite markers for the Florida manatee and the dugong , which all owed for the developments of a series of markers for each species which has great potenti al for studying individu al genetic variability within populations, and for population level conservation efforts . In a separate publication Trin g ali et al. (2008a) also identified sex specific DNA genes for T. manatus using X and Y chromosome genes. These were based upon Y primer pairs ( DBY 7 and SMCY 17) that had been previously identified for the African elephant, a closely related species to sire nians in the superorder Paen ungulata (Hellborg and Ellegren 2003, Tring ali et al. 2008a). Using tissue samples from 60 Florida manatees of known sex (30 m al es, 30 fem al es) sampled at necropsy, this study isolated nuclear / genomic DNA using packaged kits ( Qiagen DNeasy or PUREGENE DNA), and PCR was performed for DBY 7 and SMCY 17 markers using primers from the previous study ( Hellborg and Ellegren 2003, Tring ali et al. 2008a). SMCY 17 PCR yielded fragments in both fem ale and m ale samples, but since fem al es do not have this gene, it was hypothesized to be its X chromosome an alo g, SMCX, which also differed in size and nucleotide sequence from SMCY 17 (Tring ali et al. 2008a). DBY 7 only yielded a fragment in m ale samples, so its presence sign ale d m ale gender (Tring a li et al. 2008a). All of these sequences matched an alo gous conserved genes ( DBY, SMCY and SMCX ) in other mamm als including humans (Tring ali et al. 2008a). The authors were able to design primers specific to the chromosomes for use in PCR multiplex reaction s as well as for agarose gel electrophoresis (Tring ali et al. 2008a). This study has obvious
70 v alu e in confirming manatee gender when it cannot be determined through visu al or other means (such as inspection of the p elvic bones during necropsy). Hunter et al. (2010a) examined the Antillean manatee ( T. m. manatus ) population in Belize using both mtDNA and nDNA an aly ses, and compared results to the Florida sub species. This study examined population genetics but was also of particular interest for conservatio n purposes, sinc e manatees in Belize were heavily hunted for sever al hundred years until protection was put in place in 1936 (Hunter et al. 2010a). The Belize manatee population is now estimated at approximately 1000 individu al s and is believed to be the l argest breeding population of the Antillean sub species, but genetic diversity was unknown until this study (Hunter et al. 2010a). Tissue and blood samples were collected from 118 manatees in three locations in Belize (Belize City Cays, Southern Lagoon, an d Placentia Lagoon), and 96 Florida manatee genotypes were randomly chosen from the four Florida population management units for comparison to the Belize DNA (Hunter et al. 2010a). DNA was isolated using Qiagen DNeasy kits, and for mitochondri al DNA an aly s is primers CR 4 and CR 5 were used to amplify a portion of the mtDNA control region for 113 individu al s (Hunter et al. 2010a). Statistic al tests for mtDNA an aly sis included examining the degree of differentiation (F ST ST between sites within Belize, and between Belize and Florida samples) c alc ulated using ARLEQUIN version 3.1, estimates of sequence divergence used the Kimura two D of selective neutr ali ty, genetic dive rsity, nucleotide diversity, number of polymorphic sites, and number of nucleotide substitutions were c alc ulated (Hunter et al . 2010a). For microsatellite an aly sis , 16 polymorphic microsatellite primers were PCR amplified using 118 individu als , and all
71 ind ividu al s were amplified at 14 or more loci (Hunter et al . 2010a). Statistic al an aly sis for nuclear DNA was performed using Gen Al Ex software version 6.2 to estimate the level of polymorphism by observed and expected heterozygosity, the average number of all eles per locus , and to test genetic distance and isolation between Belize and Florida populations (Hunter et al. 2010a). Other Gen Al Ex 6.2 programs used to assess population structure and relatedness included GENEPOP 4.0, GENECAP, KINGROUP, BOTTLENECK, and STRUCTURE 2.3.2 (Hunter et al. 2010a). Over all results of the Belize study reported low haplotype diversity, with microsatellite and all elic variation characteristic of a sm all , isolated population enduring a bottleneck (Hunter et al. 2010a). Three mtDNA haplotypes were identified, one of which (A03) was only found at Southern Lagoon, and endemic all eles were found at both Southern Lagoon and Belize City (Hunter et al . 2010a). The 16 nuclear microsatellite markers had lower levels of variation than the Flo rida population over 14 loci (Hunter et al . 2010a). In two microsatellite markers (TmaKb60 and TmaK01) linkage D indicated it was not significant (Hunter et al. 2010a). A lt hough distribution of pairwise relatedness was not significantly different from Hardy Weinberg Equilibrium (defined as all eles and genotype frequencies being at equilibrium in a large, randomly mating population over multiple generations, with no disturb ance from migration, mutation , and selection; Frankham et al. 2010), full sibling pairwise relatedness was 19.9% of pairs, sign ali ng some inbreeding , which suggests high representation by sever al families (Hunter et al. 2010a). Bayesian an aly ses strongly s eparated the Belize and Florida populations indicating definitively that no breeding is occurring between these sub -
72 species (Hunter et al. 2010a). Florida manatees were shown to have greater genetic diversity than the Belize population and appear to be rec overing more rapidly, possibly due to a shorter bottleneck (Hunter et al. 2010a). When the Belize data was an aly zed separately, Bayesian an aly sis placed it as one population, while the LOCPRIOR model divided it into two populations: southern Lagoon and Bel ize City (Hunter et al. 2010a). These data show that the Belize population is increasing in genetic difference from the Florida sub species, but also within Belize sub populations are becoming isolated from each other, and it is unusu al to see manatee popu lations in such close proximity not breeding (Hunter et al. 2010a). The authors conclude d that it is possible that a combination of factors (low population numbers due to prolonged exploitation, isolation, some inbreeding, and genetic drift) are responsibl e for the differences observed, and that the population could be at risk if current trends continue (Hunter et al . 2010a). Recommendations we re made that the population should be monitored and that management intervention to increase gene flow might be nee ded in the future (Hunter et al . 2010a). In 2011 Nourisson et al. examined population genetics and phylogeography of T. manatus in Mexico using microsatellites. The go al of the study was to use a finer sc ale genetics methodology to assess this sm all popul ation (estimated at 700 1700 individu al s) for population structure, in order to make conservation recommendations (Nourisson et al. 2011). Previously reported mtDNA data indicated three haplotypes for Mexico, which were separated by location: only one ha plotype (J) is found in the Gulf of Mexico population, while all three (J, A, A4) are found in the Caribbean coast population (Vianna et al. 200 6 ). As in Belize, manatees in Mexico were heavily hunted over sever al
73 hundred years and it was hypothesized that the population likely underwent a bottleneck (Nourisson et al. 2011). Ninety four skin and blood samples were collected from throughout Mexico, DNA was isolated using either phenol chloroform or DNeasy techniques, and PCR amplification was performed using 13 microsatellite loci that had previously been designed for manatees by Garcia Rodriguez et al. (2000) and Pause et al. (2007) (Nourisson et al. 2011). Florida manatee genotypes (n=95) from all four management units were also included in this study for p opulation structure resolution (Nourisson et al. 2011). Statistic al tools used in this study were the same used in the Belize an aly ses and included GEN AL EX 6.2, GENEPOP 3.4 and ARLEQUIN 3.1 programs for determination and comparison of genetic diversity inc luding observed and expected heterozygosity, degree of differentiation (F ST ST between sites within Mexico, and between Mexico and Florida samples) and inbreeding (Nourisson et al. 2011). ARLEQUIN was used to check for deviation from Hardy Weinberg equilibrium, Bayesian an aly ses within the STRUCTURE program assigned all indi vidu al s to sub populations by genotype, and BOTTLENECK, as the name implies, was used to estimate a population bottleneck (Nourisson et al . 2011). Results of Nourisson et al. (2011) indicated three distinct genetic clusters within Mexico: the Gulf of Mexi co, Chetum al Bay and Ascension Bay. The latter two sites are both on the Caribbean coast, and Ascension Bay samples showed a mix of genotypes from the other two locations, indicating that this may be a mixing zone (Nourisson et al. 2011). The Gulf of Mexic o samples had the lowest genetic diversity (based on heterozygosity and number of all eles) and this was attributed to a bottleneck due to founder effect (Nourisson et al. 2011). Results indicated that while migration occurs
74 from the Gulf of Mexico populati on to the Caribbean coast, there is no migration in the opposite direction (Nourisson et al . 2011). Four samples from Chetum al Bay had high percentages of Florida population ancestry based upon the STRUCTURE an aly sis, and it was hypothesized that there may have been recent movement of a few individu al s from Florida across the Caribbean (Nourisson et al. 2011 ). The study results indicated that the Chetum al Bay population is distinctly separate from Belize populations, even though geographic al ly the countries are adjacent and share a border at Chetum al Bay (Nourisson et al. 2011). Over all the Mexican Caribbean coast had the highest genetic diversity but the sm all er population size, which raises important conservation considerations (Nourisson et al . 2011). The study recommended conservation actions in Ascension Bay as high priority since this area is a mixing zone between the two other genetic clusters, maintaining manatee habitat between clusters to facilitate continued mixing, and care in releasing captive ma natees back into the areas they were origin al ly rescued from (Nourisson et al. 2011). This study, like the Belize manatee research before it, exemplifies the precision with which genetics an aly ses can be used to target conservation recommendations to speci fic populations and locations. Aside from studying extant manatees and dugongs, phylogenetics research has also sought to understand the deeper evolutionary heritage of sirenians and other related species. As discussed above, sirenians belong to the supe rorder Afrotheria, and within Afrotheria, sirenians are classified in the superorder Paenungulata based upon genetic an aly ses utilizing both mitochondri al and nuclear DNA, amino acid sequences, and chromosome painting (Ozawa et al. 1997, Springer et al. 19 99, Kellogg et al. 2007, Pardini et al. 2007). In one recent study, Kuntner et al. (2011) examined the phylogeny
75 of all members of the superorder Afrotheria in order to test a model that orders species into hierarch al lists by categories of evolutionary di stinctiveness (ED) and phylogenetic diversity (PD) to illustrate a method of ranking conservation priorities. This study used nine previously published markers (four mtDNA and five nuclear DNA) for all Afrotherian species as well as those for 14 other plac ent al mamm als (used as outgroups) and combined both types of markers to create data matrixes (Kuntner et al. 2011). Bay e sian an aly sis was performed using M RBAYES software and conservation priorities were assessed using a Mesquite software package, Tuatara Module 1.01 (EDGE and HEDGE methods) which considers factors such as termin al branch length, extinction probability and the estimation of contribution of a species to over all diversity (Kuntner et al . 2011). Results ranked the top 20 species of conservati on priority based upon ED and PD, and all three Trichechids and the dugong were ranked near the top (Kuntner et al. 2011). This an aly sis used criteria primarily limited to genetics and did not consider other factors important for consideration in conservat ion planning. However, one interesting result of this study was that the phylogeny supported T. inunguis as sister taxon to T. seneg ale nsis , rather than T. manatus , as in Vianna et al. (2006 ). In summary, throughout manatee genetics research over the past 24 years, studies have become more sophisticated, moving from basic mitochondri al DNA techniques to finer sc ale nuclear DNA an aly ses and using increasingly complex computer programs to an aly ze data more efficiently and precisely. Most an aly ses have used m anatee blood and tissue samples, but with the development of ancient DNA techniques , hopefully more reliable results will be able to be obtained from bone as well. This will all ow for sampling from older specimens such as those in museums and
76 hunted manate es, both of which can bring insight into past population structure. Sirenian studies have alm ost all been a mix of phylogeny, population level genetics, and phylogeography. For all sirenians there has been a need to both understand their evolutionary histo ry as well as contemporary population trends and distributions, with the over arching go al of conserving these unique creatures. Stable Isotope Analysis as a Tool to Study Foraging Ecology Introduction to Stable Isotopes All elements have isotopes, which a re different forms of the same element defined by varying numbers of neutrons in the nucleus of the atom (Fry 2006). The number of neutrons in relation to the number of protons is that which determines the stability of the isotope; gener al ly those having a n equ al or greater number of neutrons relative to protons are more stable, alt hough hydrogen and helium have stable isotopes with f ewer protons than neutrons (Fry 2006). Stable isotopes comprise less than 10% of all known isotopes, they do not decay, they are not radioactive, and they are abundant throughout all matter (Fry 2006, Crawford et al. 2008). The lightest isotopes, defined by having a low mass, account for more than 95% of all isotopes in the elements hydrogen (H), carbon (C), nitrogen (N), oxygen (O), and sulfur (S), but these elements also have sm all percentages of high mass isotopes (Fry 2006). For example, carbon has a low mass isotope ( 12 C) with a fundament al abundance on Earth of 98.89%, and a high mass isotope ( 13 C) with a fundament al abunda nce of 1.11% (Fry 2006). The abundance ratios of light and heavy isotopes vary minutely in nature, and it is the relative differences in ratios that can be measured against an internation al standard reference (Hobson and Wassenaar 2008). Standard isotopic references come from PeeDee Belemnite limestone (PDB) for carbon, Vienna Standard Mean Ocean Water (VSMOW) for oxygen
77 and hydrogen, atmospheric N 2 (air) for nitrogen, and Canyon Diablo Triolite (CDT) for sulfur (Hobson and Wassenaar 2008). By definition th e v alu e of these standards is 0 0 / 00 (Newsome et al. 2010). Since the elements H, C, N, O and S comprise most of the mass of organic materi al , these properties make them v alu able for studying ecologic al processes at many levels, from microbes to entire ecos ystems (Fry 2006). A good an alo gy provided by Fry (2006) is to think of stable isotopes as a dye or tracer that follows a substance through an ecologic al cycle, all owing us to detect its presence. In nature, carbon, nitrogen and sulfur (CNS) are the eleme nts primarily linked to the cycling of organic matter, and hydrogen and oxygen (HO) are primarily linked to hydrologic al cycles (Fry 2006). Large pools of stable isotopes exist in the atmosphere, soil, and the oceans, and these produce regular, characteris tic distributions as well as predictable patterns of cycling (Fry 2006). Two important processes, mixing (combining ) and fractionation (separation ) affect abundances of stable isotopes, and will be discussed further below. Stable isotope v alu es are denote negative, expressed as parts per thousand ( 0 / 00 ) relative difference to the standard for the element of interest (Hobson and Wassenaar 2008) . The delta v alu e is determined by the standard delta equation: 0 / 00 ) = (R sample / R standard 1) x 1000 (1 2) 13 15 N), R v alu es are the isotopic ratios of the sample and the standard reference comparatively measured by the mass spectrometer (light to heavy i sotopes), and the equation is multiplied by 1000 to transform v alu es into primarily whole numbers (Hobson and Wassenaar 2008) . For
78 natur al alu es range between 100 and +50 0 / 00 (Fry, 2006). A sample that has more of the heavier isotope in r elation to the standard, is referred to as being al ly less of the heavy isotope is referred to as et al . 2008). The carbon cycle is based upon active exchange of carbon dioxide (CO 2 ) between the atmos phere, the ocean surface and terrestri al ecosystems (Fry 2006). Freshwater 13 C v alu es vary widely based upon the amount of CO 2 dissolved within the specific water body as well as other inorganic factors, and therefore can have a distinct loc al signature (F ry 2006). Carbon uptake by plants differs between C 3 plants (which include which involves a net fractionation of 20 0 / 00 between atmospheric CO 2 and plant biomass, and that of C 4 plants ( al plants and s al t grasses) which involves a net fractionation of approximately 5 0 / 00 ( MacFadden et al. 2004, Fry 2006, Cerling et al. 2009). Grasses that live alo ng the land / water interface photosynthesize carbon by using either the C 3 or C 4 pathway, depending upon species ( MacFadden et al. 13 C v alu es that range from 30 to 15 N v alu es that range from Al ves 2007). The difference between C 3 and C 4 plants is based upo n the biochemic al pathways they use during photosynthesis: carbon atoms are incorporated as either 3 3 ) or 4 carbon (PEP carboxylase, C 4 ) sugars (Hobson and Wassenaar 2008 , Nelson and Cox 2008 ). Rubisco strongly discriminates aga inst 13 CO 2 , which gives C 3 plants strongly 13 C depleted isotope ratios (Hobson and Wassenaar 2008). C 4 plants, which evolved more recently and can tolerate lower CO 2 conditions, have much less 13 CO 2 discrimination
79 due to PEP carboxylase (Fry 2006, Hobson a nd Wassenaar 2008 ). Most fully marine plants (including seagrasses) fractionate carbon using the C 3 pathway, alt hough because the carbon source in the oceans can be derived from bicarbonate as well as atmospheric CO 2 13 C v alu es of C 3 marine plants ca n be more enriched than terrestri al C 3 plants ( MacFadden et al. 2004). Thus an aly 13 C in diet can easily discriminate between C 3 and C 4 plants. There is a third plant category, Crassulacean Acid Metabolism plants (CAM, which includes cactuses and eu phorbia), which use a photosynthetic pathway similar to C 4 plants, but are speci al ly adapted to live in arid conditions ( MacFadden et al. 2004 ). Since CAM plants are not relevant to sirenian research, they will not be disc ussed further here. In the nitrog en cycle, most nitrogen is present in the atmosphere as N 2 gas, which is gener al ly constant at 0 0 / 00 (Fry 2006) . Since there are multiple processes and reactions within soils (including sedimentation, assimilation of decomposed organic matter, etc.), there is currently no gener al theory describing 15 N in soil (Hobson and Wassenaar 2008). Nitrogen isotope ratios can be used to discriminate between freshwater and marine systems through their baseline values: freshwater ranges from approximately and saltwater ranges from approximately 16 to 12 (Hobson 1999). Signals for 15 N in aquatic systems can be affected by pollution such as fertilizer runoff, sewage, or anim al waste runoff from agricultur al regions (Fry 2006). Plants that fix N 2 from th e atmosphere have 15 N v alu es of about 2 to +2 0 / 00 , which is relatively close to the 0 0 / 00 v alu e of atmospheric N 2 (Peterson and Fry 1987). Plant 15 N is determined by isotope fractionations associated with nitrogen uptake and metabolism by the plant ( Hob son and Wassenaar 2008).
80 In the oxygen cycle there are three isotopes that are used as tracers for processes in the biosphere: 16 O, 17 O, and 18 O (Fry 2006). Most oxygen isotope dynamics are involved in water cycles and these isotopes can be used to identi fy specific water sources, particularly in anim al studies (Fry 2006, Cerling et al. 2009). Oxygen isotopes can also be used to differentiate between aquatic and terrestri al habitats, as shown in a study by Clementz et al . (2001) that investigated marine ma mm al habitat and foraging ecology using stable isotopes. They reported that 18 O v alu es than those of terrestri al 18 O v alu es for aquatic taxa were significantly differen t between different freshwater, estuarine and marine habitats (Clementz et al. 2001). Hydrogen isotopes are also tied to water cycles and can be used to trace water in dietary sources and specific water sources. Because of these specificities to water sour ces, both hydrogen and oxygen isotopes are used to trace water intake and movement/migration patterns of animals (Fry 2006). Stable isotopes are a very useful tool in animal studies, particularly for wildlife that is difficult to observe and/or may travel over large distances ( Hobson and Wassenaar 2008) . Differences in stable isotope ratios of carbon ( 13 C/ 12 C) and nitrogen ( 15 N/ 14 N) between dietary sources and trophic levels allows the source of the diet to be traced and the trophic level of an animal dete rmined (Peterson and Fry 1987, Hobson 1999). For example, the standard stepwise enrichment ratio from a food source to the 13 C is ~1 3 0 / 00 , 15 N it is ~3 5 0 / 00 ( DeNiro and Epstein 1978, Schoeninger and DeNiro 1984) . Carbon isotopes in animal tissues can indicate differences bet ween C 3 and C 4 plants, and differentiate between terrestrial, freshwater
81 and marine ecosystems (Fry 2006, Alves Stanley and Worthy 2009). For example, the mean carbon isotope compositions of freshwater vegetation ( es ( 10 to consumer tissues appropriately so that their relative contributions to diet can be determined (Clementz and Koch 2001). Stable nitrogen isotopes in animal tissue basically trace protein pa thways, because nitrogen is mostly absent in lipids and carbohydrates (Hobson and Wassenaar 2008). Animals with lower protein diets will likely exhibit greater overall diet tissue discrimination for 15 N than those species with higher protein diets (Hobson and Wassenaar 2008) . 15 N value differences indicate changing trophic levels, food web isotopic signature, and nutritional stress or fasting 15 N is high because the ratio of excreted to assimilated nitrogen is infinitely high during periods of starvation (McCutchan et al. 2003). Therefore, a starved animal will exhibit higher 15 N values than an animal that is in good body condition (McCutchan et al. 2003). Nitrogen isotopic values increase by 10 t o 15 0 / 00 in many food webs; for example, increases of 3 to 5 successive trophic transfers each increases the 15 N content by 3 5 0 / 00 (Peterson and Fry 1987). H owever , DeNiro and Epstein (1981) reported trophic level enrichments ranging from 1 6 . In additi on to trophic level shifts, 15 N value changes can be attributed to factors such as abiotic factors in plants of the same species, levels of stress and even digestive anatomy in consumers: in herbivores , for example, foregut fermenters are functionally at a higher trophic level and h ave relatively higher 15 N values than hindgut fermenters due to the location of the microflora in their digestive tracts (Sponheimer et al . 2003). Nitrogen is also excreted as
82 15 N values, and also must be taken int o consideration for species adapted to conserving water or fasting (Hobson and Wassenaar 2008). In the study of stable isotopes, there are two important processes that play a critical role, mixing and fractionation. Mixing is the combining of two or more substances, and stable isotopes can both trace mixing processes to show which sources contribute to the mixture, and identify which sources make up the largest percentage of the mixture (Fry 2006 ). Mixing models are used in scientific research and are desi gned to tease apart the components of the mixture as well as the percentage that each c omponent contributes to it (Fry 2006). Linear mass b ala nce mixing models were designed to quantify the contribution from each food source to a consumer's diet (Phillips and Gregg 2001). A two source, mass b ala nce mixing model equation for a stable isotope was developed (Phillips and Koch 2002). The standard equation is shown here for carbon isotopes: 13 C m = X ( 13 C X 13 C tissue X ) + Y ( 13 C Y 13 C tissue Y ) (1 3) X Y where the subscripts X, Y, and m represent two food sources and the mixture (the consumer tissue), respectively, refers to the fraction al contribution of C from each food 13 C tissue X is the trophic fractionation (the change in 13 C from the food source during assimilation into the tissue) (Phillips and Koch 2002) . However, scientists also needed a mixing model equation for two stab le isotopes and three food sources. A duel element (X, Y), three source, mass -
83 b ala nce, linear mixing model was also developed and is described by the following equations (Phillips and Gregg 2001): m = A A B B C C m A A B B C C (1 4) A B C where X and Y refer to stable isotopes such as 13 C and 15 fraction al contribution of each isotope from each food source (mass) to tot al diet, A, B, and C refer to three food sources, and the subscript m indicates the mixture of sources (Phillips and Gregg 2001). This model has been used for numerous species and requires that the topic values for the consumer f all within a polygon outlined by the potenti al food sources (Phillips and Gregg 2001). Sever al stable isotope mixing model computer programs have been developed since 2001 (IsoError, IsoConc, IsoSource and sever al Bayesian models such as IsotopeR and SIAR ), each with increasing parameters that can be adjusted by the user, such as une qu al concentrations of dietary items, various sources of uncertainty, and the ability to apply the models to individu als or populations (Hopkins and Ferguson 2012). Bayesian models all ow biologists to fit probability models to isotopic data, include prior information, and efficiently estimate numerous parameters (Hopkins and Ferguson 2012), all of which makes the model more re ali stic for applications to re al world circumstances. Fractionation is the opposite of mixing, and separates isotopes in a predictab le manner in the environment as a result of chemic al , biologic al or physic al processes (Reich and Worthy 2006). Similarly, diet tissue discrimination describ es the difference in isotopic ratios between diet and consumer tissue ( Alv es Stanley and Worthy 200 9).
84 Both mixing and fractionation processes obviously operate simultaneously throughout the biosphere. Within the biosphere, plants and microbes take in nutrients with stable isotope signatures originating from the environment, and fractionation results in a stable isotope signature within the organism, primarily carbon, nitrogen and sulfur (Fry 2006). This occurs due to differing rates of heavy vs. light isotope incorporation during biogeochemic al reactions within primary sources of production, such as pla nts ( Alv es 2007). Stable isotopes record source information about their origins, and this sets an isotopic baseline that can subsequently be shifted by isotopic fractionation (Peterson and Fry 1987). At the next level of the food web, consumers of primary producers have different ratios of stable isotopes from their diet ( diet tissue discrimination) due to biochemic al processes such as preferenti al loss of 12 CO 2 during respiration, preferenti al uptake of 13 C during digestion, and/or metabolic discrimination during the synthesis of new tissues (DeNiro and Epstein 1978). Stable Isotope Methodology Stable isotope an aly sis has the potenti al to reve al both long term and short term assimilation histories as well as anim al distribution, movement and migration patt erns, by examining both slow and fast turnover tissues and measuring isotopic v alu es that potenti al ly represent the diet incorporated into the tissue (Peterson and Fry 1987, Hobson and Wassenaar 2008). In addition to the tissues themselves, the time of col lection of samples is important: a sample collected today would give information about an anim s recent or lifetime diet and the ecosystem(s) it lives in, but museum specimens can give historic al information about the species diet and environment in the past (Hobson and Wassenaar 2008) . This in turn could have implications for conservation and management of the species today. Stable isotope an aly ses can also
85 be performed on extinct species, which gives insight into both historic al environments and species evolution ( Clementz and Koch 2001, MacFadden et al. 2004). The information researchers can obtain varies depending upon the type of anim al tissue used to measure stable isotopes. There are two gener al categories of samples: fixed tissues that are metaboli c al ly and isotopic al ly inert (such as hair, claws / fingernails, or feathers), and dynamic tissues which are the opposite, those tissues that are metabolic al ly active and continu al blood, intern al organs, an d bone collagen) (Hobson and Wassenaar 2008) . Turnover rates, defined as the time required to incorporate stable isotopes from diet into tissue, are highly variable between tissue types and between taxa, but in gener al they range from days (as in liver and blood) to weeks (muscle, skin) to a lifetime (bone collagen) ( Alv es Stanley et al. 2010, Hobson and Wassenaar 2008) . Stable isotope values from tissues with a faster turnover rate therefore will reflect a more recent diet than those with a slower rate. T urnover rates in anim als are best determined through captive anim al studies where diet can be changed and the length of time for tissue turnover can be recorded in a controlled setting (Fry 2006, Alv es Stanley and Worthy 2009). Hobson and Clark (1992) c alc ulated turnover rate as: y = a + be ct (1 5) where y = a is the v alu e approached asymptotic al ly, b is tot al change in v alu e after a diet switch, c is turnover rate, and t is time since the diet switch. Turnover is expre ssed in terms of h alf life, the time it takes the isotopic composition of the tissue to reach the midpoint between initi al and fin al v alu es: X=(ln0.5) / c ( Alv es Stanley and Worthy 2009).
86 Diet tissue discrimination is the difference in isotopic ratios bet ween consumer tissue and diet ( Alv es Stanley and Worthy 2009). In trophic studies, diet tissue discrimination is often described using the equation: h X A B = h X A h X B (1 6) where h X is the stable isotope, A is the consumer, and B is the diet (McCutchan et al . 2003, Newsome et al. 2010). Since consumers are typic al ly enriched in the heavy isotope relative to diet, A B v alu es are defined as positive (Newsome et al. 2010). In early stable isotope research studies disparities in digestibility and digestion times that might lead to differences in diet tissue discrimination were n ot taken into consideration (Peterson and Fry 1987) , but more recent research is starting to examine these parameters. In a study by Codron et al. (2011) domestic goats were used to test the digestibility and subsequent assimilation of 13 C isotopes into bl ood (feces were also tested). Study results showed that the C 3 plant components of the diet were over represented in the 13 C blood and feces v alu es due to faster 13 C incorporation rates, while the C 4 plant components were under represented (Codron et al. 2011). Alt hough diet was controlled in this study, the results 13 C v alu es do not reflect C 3 and C 4 diet components equ al ly (Codron et al. 2011). This study recommended that diet tissue discrimination factors should be considered carefully, since different anim als have different digestion rates, and an anim s tissue must be in isotopic equilibrium before an accurate estimate of dietary enrichment is able to be obtained (Codron et al. 2011).
87 Many anim al stable isotope studies to date have focused on identi fying and quantifying dietary characteristics. D al erum and AngerbjÃ¶rn (2005) describe three approaches to studying tempor al variation in dietary stable isotopes: 1) a comparison of the same tissue type sampled over time; 2) a comparison of different tissue s with differing turnover rates; or 3) a comparison of different parts of the same tissue that grows progressively to study isotopic v alu es in chronologic al order. The first two methods can address questions about both short and long term variation, while the third method will address only a shorter time frame (D ale rum and AngerbjÃ¶rn 2005). These same techniques can also be applied to other types of stable isotope studies, such as determining habitat use, migratory patterns, or food webs between different s pecies. In designing a study the researcher needs to consider which stable isotope(s) and which tissue type(s) will address their questions. Considerable isotopic variation still exists among different tissues and metabolites within individu al anim als (P et erson and Fry 1987 ). Bone has two components that can be used in stable isotope an aly ses. Bioapatite is the miner ali zed (inorganic) component of mamm ali an bone, while collagen is composed of proteins (organic portion) that are the main constituent of anim al connective tissues (Clementz et al. 2007, Nelson and Cox 2008). The inorganic and organic fractions of bone specific al ly incorporate carbon from different components of diet: the carbon in the protein part of the diet is more strongly labeled in collage n, and carbon in blood (produced through respiration and bulk dietary carbon, including carbohydrates, lipids, and proteins) is more strongly labeled in bioapatite (Clementz et al. 2009 a ). Bone collagen has a slow turnover rate, all owing for the detection of sever al
88 years of dietary history, and it is also considered to reflect an average lifetime diet (Hobson and Clark 1992, Clementz et al. 2007 ). Both the inorganic and organic components of bone are constantly being remodeled during growth throughout an a nim s lifetime (Clementz et al . 2007). Turnover rates for bone from terrestri al mamm als have been estimated at ~4 years for bone bioapatite and ~7 years for bone collagen (Clementz et al . 2007). Bone is easily preserved and can also tell us about the die t and life histories of anim als from the past, since very old and fossilized bones can also be used for stable isotope an aly ses (Peterson and Fry 1987). Mammalian b one collagen is 2 6 0 / 00 13 C while bioapatite is gener al ly enriched by 6 15 0 / 0 0 relative to diet (Peterson and Fry 1987, Cerling and Harris 1999, Passey et al. 2005). Ungulate herbivores have been shown to have a bone collagen 13 C enrichment of 12 14 0 / 00 relative to diet (Cerling and Harris 1999). Both the inorganic and organic com ponents of bone are remodeled during growth et al . 2007). Bone remodeling is accomplished through the removal of mineralized bone by osteoclasts, followed by the formation of bone matrix through the osteoblasts tha t subsequently become mineralized ( Hadjidakis and Androulakis 2006) . Hadjidakis and Androulakis (2006) describe t he bone remodeling cycle as consisting of three consecutive phases: resorption, during which osteoclasts digest old bone; reversal, when mononu clear cells appear on the bone surface; and formation, when osteoblasts lay down new bone until the resorbed bone is completely replaced. In a study of Florida manatees using 1213 periosteal bones, resorption was shown to occur in the growth layer groups a fter 15 years of age, et al . 1996).
89 Although not yet studied for manatees, because all bone is remodeled during an also remodeled. Marmontel et al . (1996) also analyzed ear bones from Antillean manatees (n=18) and found results similar to Florida manatees after 20 years of age. However, ear bones of female manatees were determined to experience greater bone resorption than males when the data were adjusted for differences in length and age, and this due to reproduction (Marmontel et al . 1996). The Marmontel study classified levels of ear bone GLG resorption into categories (none, light, moderate, or heavy) based upon the number and extent of primary osteons (also known as Haversian systems ), and secondary osteons that occurred within the bone due to remodeling as manatees age. These f eatures disrupt or obliterate annual adhesion lines used to age manatees, but some of the original bone persists, allowing researchers to read annual lines. Thus, as the bone is remodeled, new tissue forms within it, in the form of primary and secondary os teons, but some of the original bone persists, and stable isotope sampling will include In contrast to bone, the enamel cap of mamm al teeth forms prior to eruption over a relat ively short time and, once deposited, is no longer remodeled by the body (Clementz et al . 2007). As a consequence, differences in the 13 C v alu es of teeth and bone are heavily influenced by development al and season al changes in diet (Clementz et al. 2007). Clementz et al . (2007) also reported that foraging on dietary resources that 13 C v alu es (for example freshwater and marine vegetation) or macronutrient composition (such as proteins vs. lipids) at different times of the year
90 would create significant offsets between tooth and bone isotope records. Long term diets can be studied using an entire tooth, and short term diets can be examined using subsets of growth layer groups within a tooth (W alk er and Macko 1999). Mean carbon isotopic enrich ment of tooth enamel relative to diet is reported to be approximately 12 ali an herbivores ( MacFadden et al. 2004). Dynamic tissues such as blood, intern al organs, and blubber/fat have much faster turnover rates and can be strong ly influenced by the most recent dietary activities ( D ale rum and AngerbjÃ¶rn 2005). Blood plasma studies of American black bears, three species of se als , and two species of birds reported 13 C h alf life v alu es ranging from 0.4 4 days (reviewed in D ale rum a nd AngerbjÃ¶rn 2005). Stable isotopes of whole blood, which has been studied in American black bears and sever al species of bats and birds had 13 C h alf life v alu es ranging from 4 14 days for birds, 40 days for the bears and 113 126 days in the bats, so in s ome cases these an aly ses could record sever al months of diet and habitat use (reviewed in D ale rum and AngerbjÃ¶rn 2005). Lipids are depleted for 13 C compared with other types of molecules, and can mislead the an aly ses by 13 C sign al (DeNiro and Epstein 1978). In both higher and lower organisms it has been found that lipids in fat reserves are gener al ly 2 8 0 / 00 depleted in l3 C (Peterson and Fry 1987). There have been very few studies of stable isotopes in anim al skin, with the exception of m anatee skin which is discussed in further detail below. Alt hough claws and fingernails can give a chronologic al record of diet and habitat use over a longer time frame, there are few studies to date that have used this tissue for stable isotope an aly ses. One study of human fingernails examined the diet of Inuit
91 als (se als , wh ale , and fish) was significantly enriched in 13 C, 15 N, and 34 S compared to Danish citizens who ate primarily the meat o f domesticated anim als (Buchardt et al. 2007). The study also reported that human fingernails grow approximately 0.10 0.12 mm/day, and that nail clippings in the study represented food intake four to six months earlier (Buchardt et al. 2007). A recent stud y of Ringed se al ( Pusa hispida ) claws indicated that the part of the claw representing the most recent growth (c all ed an annulus and defined as 4 8 months 13 C v alu 13 C, which supports the us e of claws to monitor viscer al 13 C (Ferreira et al. 2011). Addition al ly this study determined that claw growth layers could be used to determine se s ages and measure the level of mercury in the se als , so together with the stable isotope an aly ses these claws provide a chronologic al record of se al diet, contaminant load, and life history (Ferreira et al. 2011). Many stable isotope studies of wildlife have used hair to determine 13 C and 15 N dietary and habitat v alu es. Hair is often easy to collect and pr ovides a chronologic al reference that can also elucidate migration/movement patterns (Hobson and 13 C and 15 N (Cerling et al. 2009). One example of a wildlife study that used hair w as by Cerling et al . (2009), who used hair from individu al African elephants within the same family group over 6 years to study diet and habitat use. Growth rates of hairs were determined by comparing overlapping stable isotope patterns of 13 C and 15 N, a nd were found to be between 0.73 and 1.04 mm/day (Cerling et al. 2009). Portions of eight different hairs from each individu al elephant were used, hair samples were up to 580 mm long
92 (median= 460 mm), and hairs had chronologies recording up to 20 months (m edian= 18 months) (Cerling et al. 2009). Samples for 13 C and 15 N an aly sis were collected from each 5 mm interv al (which corresponded to ~6 days in these samples) the results showed very high correlation of stable isotope chronologies between family members (Cerling et al. 2009). This study was able to show season al changes in diet between C 3 and C 4 plants was strongly correlated to wet and dry seasons, and most importantly, in an instance where elephants and cows were using the same habitat, the cows out competed the elephants for the grasses they would nor m al ly have eaten during the rainy season, so the elephants continued to eat C 4 plants, a dry season forage (Cerling et al. 2009). This type of data shows a species adaptability but also a conservation ch all enge that was only available due to stable isotope an aly sis (Cerling et al. 2009). Once samples are collected, they often need to be preserved for varying amounts of time before laboratory an aly sis can occur. In the case of museum specimens, these may have been stored for long periods of time and/or prese rved by methods that can impede obtaining accurate stable isotope results (Hobson and Wassenaar 2008) . It is important that preservation methods do not degrade or alt er the stable isotope compositions of samples, otherwise accurate results will not be obta ined (Barrow et al . 2008, Hobson and Wassenaar 2008). Preservation methods for samples have included drying, freezing, and storage in 70% ethanol, form al in, saturated sodium chloride (NaCl) or dimethyl sulfide (DMSO) buffer solutions. Barrow et al . (2008) used turtle epidermis from three species of marine and freshwater turtles to test the effects of five preservation methods on stable carbon and nitrogen isotope an aly ses. Samples were preserved over varying lengths of time with a range of 1 60 days (Barrow et al .
93 13 15 N v alu es for tissues preserved in 70% ethanol and NaCl solution were not significantly different from the controls (samples dried at 60 o Celsius), but that the lipid content of samples should be taken into consideration before using a NaCl solution (Barrow et al. 2008). Preservation by NaCl solution can vary with 13 15 N (Xu et al. 2011), and lipids are known to be depleted for 13 C compare d with other molecules misleading an aly 13 C sign al (DeNiro and Epstein 1978, Barrow et al. 2008). In the Barrow et al. study, samples preserved in DMSO were significantly 13 15 N v alu es were severely alt ered, makin g it impossible to interpret these samples (Barrow et al. 2008). In the same study, preservation by freezing had a significant effect on samples at 60 days, with the result that they were 13 15 N compared with to contro ls (Barrow et al. 2008). Barrow et al. also summarized 16 other studies that used 20 different preservative methods and their effects on different tissue samples, and found that freezing typic al ly showed no effect, but some methods (including shock freezin g and 15 N, other methods (form al in, form al 13 C and 15 N for alm ost every tissue examined, and ethanol solutions greater than 70% had mixed resu lts (Barrow et al. 2008). A 2011 study by Xu et al. examined the preservation effects of form ali n, ethanol, 13 15 N of fish muscle tissues (3 freshwater species) and found that all preservation methods, except drying, had si 13 C and 15 N v alu es. Form al 13 C v alu es for all samples were
94 depleted 1 0 / 00 15 N v alu es increased (Xu et al. 2011). This study noted that effects of preservation appear to be highly taxa specific, but that duration of preservation did not significantly affect the difference in stable isotope v alu es between preserved samples and controls (Xu et al 13 C and 15 N v alu es were gener al ly changed dramatic a l ly within the early stage of preservation and became stable over relatively long term preservation, furthermore this study recommended the development of a preservation correction factor to prevent the mis estimation of energetic pathways (Xu et al. 2011) . Preservation methods should be carefully considered and preserved samples should, if possible, be tested against controls to ensure v al ues are not al tered by preservation. Stable isotopes are measured using continuous flow isotope ratio mass spectromete r (CF IRMS) which converts samples to pure CO 2 , H 2 O, SO 2 or N 2 gases (Hobson and Wassenaar 2008). In order for this to occur, samples must first be chopped, ground or powdered and then combusted to convert them to ultrapure gaseous an al ytes that can then h ave their isotopic ratios measured in relation to a reference gas sample of the same type (Fry et al. 1992). Techniques for preparing samples for combustion vary depending upon sample type (hair, bone, tissue, etc . ). In gener al , samples must be dried and l ipids are removed to eliminate their effect on carbon isotope ratios (Reich and Worthy 2006, Al ves Stanley and Worthy 2009). Samples are then dried again to remove any solvent, ground up and placed in tin capsules for mass spectrometry an aly sis (Reich and Worthy 2006, Alves Stanley and Worthy 2009).
95 Stable Isotope Research Studies on M anatees The first stable isotope research on manatees was conducted by Ames et al. (1996). This study examin 13 C v alu es in sloughed skin collected over a year from three live, captive Florida manatees ( Trichechus manatus latirostris ), and compared 13 C v alu es from samples of the food they ate (98% lettuce, but also carrots, wheat sprouts, Hyd rilla verticillata , Eichhornia crassipes , primate chow biscuits, and for one manatee, a prepared milk formula). In a second part of the same study, Ames et al. an aly zed tissue samples (liver, kidney, blubber and skin) from dead, wild Florida manatees (n=20 13 C v alu es to v alu es reported in the literature for aquatic vegetation known to be consumed by wild Florida manatees. The results 13 C v alu es for skin of the captive manatees was higher an average of +4. 1 0 / 00 over the lettuc e that comprised 98% of their diet. This is higher than the 1 2 0 / 00 reported for terrestrial mammals, and the higher offset may be due in part to the 13 C v alu es for the remaining 2% of food sources in the study (DeNiro and Epstein 1976, Schoeninger and DeNiro 1984 ). For the one manatee that was switched from milk 13 C v alu es of sloughed skin decreased over seven months to the same v alu es as for the other manatees consuming lettuce, showing that the skin reflected the carbon signature of the new diet in that time frame (Ames et al. 1996). In comparison, the skin from the dead, wild manatee samples was more enriched than that of the captive manatees, and was consistent with the published 13 C v alu es of seagrasses (Am es et al . 1996). Interestingly, alt 13 C v alu es for freshwater plants eaten by wild manatees were included in this study, it 13 C sign als in their tissues. For the other tissue sample s, most of the blubber samples were depleted in 13 C
96 compared to the kidney, liver, and skin tissues. Peterson and Fry (1987) had previously found that lipids in fat reserves (such as the blubber in manatees) are gener al ly 2 0 / 00 8 0 / 00 depleted in l3 C in r elation to the diet. In this study plants were not collected from the same loc ali ties that the dead manatees had come from (difficult because the carcasses had been recovered from 14 different counties in Florida) and turnover rates of all the tissues were not considered, but it was a good first step in determining the 13 C v alu es in foods consumed by both captive and wild Florida manatees and in some of their tissues. Captive studies , where diet can be controlled , are important for determining turnover rates and species specific trophic enrichment factors for specific tissues (DeNiro and Epstein 1978, Hobson and Wassenaar 2008 ). A 1999 study by W alk er and Macko an aly zed the 13 C and 15 N v alu es from teeth of eight marine mamm al spe cies representing all trophic levels, including the Florida manatee. This was the first stable isotope study of manatee teeth, and all manatee samples came from the east coast of Florida. At the beginning of the study the authors an aly zed front and back mo lars from three two year old manatees in order to examine isotopic heterogeneity (W alk er and Macko 1999). They found that both 13 C and 15 N sign als were less than 1 0 / 00 different between front and back teeth, but that these values were more heterogeneous than of any of the seven other species sampled , which had values varying from 0.1 0.4 0 / 00 (W alk er and Macko 1999). In the di scussion the authors concluded that since the results were alm ost homogeneous, a representative sample can document the dent al isotopic signatures for manatees. However, this may be an inaccurate conclusion. The sampled manatees were probably
97 young enough not to have lost and replaced any teeth yet, so it is possible the sampled teeth all developed at the same time, and that isotopic differences still might be seen between different teeth in older anim als that develop new back teeth later in life after shed ding their origin al teeth. Also , Florida manatees are gener al ly known to nurse for nursing to a plant diet, we would not expect to see differences in 13 C v alu es. F or the 13 C and 15 N v alu es of teeth between species, teeth from ten stranded manatees recovered from the east coast of Florida were an aly zed. Values for 13 C (mean 9.9 Â± 0.8 0 / 00 ) of the Florida manatee were higher compared to all other spec ies, likely reflecting a seagrass diet , and 15 N v alu es (mean 7.8 Â± 0.8 0 / 00 ) were lower than all other species, reflecting an herbivorous rather than a carnivorous diet (W alk er and Macko 1999). Because the sampled teeth were collected from wild manatees, these v alu es are a good baseline for comparison to other manatee samples from the east coast of Florida. This study also illustrated that long term diets could be studied using an entire tooth, and short term diets can be examined using subsets of growth l ayer groups within a tooth (W alk er and Macko 1999). In 2006 Reich and Worthy published the first stable isotope study to examine 15 N v alu 13 C in skin of wild Florida manatees, which was also the s hesis. As in Ames et al. (1996), this study also used skin from carcasses. Reich (2001) origin al ly sampled epiderm al , derm al, and muscle layers of the skin separately, but found no statistic al ly significant difference for 13 C v alu es between tissue types when they were compared using ANOVA, so only derm al tissues were an aly zed for the 2006 publication. Addition al ly, Reich and Worthy
98 (2006) collected 25 species of marine and aquatic plants from four different region of Florida (consistent with locations of carcass recovery) and an aly 13 C and 15 N v alu 13 15 N v alu es for plants or manatee samples in the same habitat within regions, but stable isotope values varied among regions (Reich and Worthy 2006). Data w ere an aly zed using a du al isotope 13 15 N ), three source (based upon freshwater, estuarine and marine plants) mixing model based upon mass b ala nce equations (as shown previously above, Phillips and Gregg 2001). This was the first study to estimate the proportional contribution of food resources t o manatee diets using a mixing model. To achieve results that fell within v al id v alu es, diet tissue enrichment v alu es for all anim als were assumed to be 3 for 15 N and 1 13 C based upon previously published studies ( Schoeninger & DeNiro 1984, H obson and Welch 1992, Reich and Worthy 2006 ). These enrichment values fell within the accepted range for mammals in gener al , as well as for manatees specific al ly (Ames et al . 1996). This all owed the estimation of the proportional contributions of each plan t category to manatee diet s and to further categorize diet by region. Alt hough the numbers of manatee samples for each region were sm all (range= 2 11 samples per region), the mixing model results indicated that isotopic v alu es of manatee epiderm al layers v aried significantly by region, indicating important differences in foraging strategies (Reich and Worthy 2006). For example, freshwater plants comprised h alf 10% of the diet in the so utheast, and none on the centr al east coast of Florida. This showed not only that diet varied region al ly, but it has implications for managing critic al habitat for the species (Reich and Worthy 2006). At the time of this study, the turnover
99 rate for manate e skin had not been determined, so a time component was not able to be included in the an aly ses. Addition al ly, potenti al season al differences in 13 C and 15 N v alu es for the plants were not considered. Carbon and nitrogen isotopic turnover rates for Flori da manatee skin was first researched and published by Alv es (Master s thesis 2007, Alv es Stanley and Worthy 2009). Skin from rescued manatees was collected over more than one year while manatees were in captivity and transitioned from a wild to captive die t ( Alv es Stanley and Worthy 2009). Alv es (2007) reported the mean h alf life for 13 C turnover in manatee epidermis as 55 days, but in later an aly ses the manatee data separated by habitat type . The mean h alf life for 13 C turnover was reported as 53 days for manatees rescued from coast al regions and 59 days for manatee skin from riverine regions ( Alv es Stanley and Worthy 2009). Mean h alf life for 15 N turnover was reported as 42 days ( Alv es, 2007) but in Alv es Stanley and Worthy (2009) the data was similarly d ivided between coast al and riverine habitats and reported as mean h alf life for 15 N turnover at 27 and 58 days, respectively . The study reported that carbon h alf life c alc ulated for manatee epidermis was very slow compared with previous turnover studies on other species, and that this may be caused by the manatees very slow metabolism ( Alv es Stanley and Worthy 2009). Because of these slow turnover rates, 13 C and 15 N stable isotope an aly sis in manatee epidermis can be useful in summarizing average dietar y intake over sever al months. Addition al ly, the study reported a diet tissue discrimination 13 C v alu for long term captive manatees on a lettuce diet, but 15 N v alu es did not differ between manatee skin and diet ( Alv es Stanley and Worthy 2009).
100 In 2010, Alv es Stanley et al. published the first stable isotope research on skin from Antillean manatees ( Trichechus manatus manatus ) from Belize and Puerto Rico, and the same study included skin samples from wild Florida manatees sampled on the west co as t of Florida and the St. John s River. This study hypothesized that manatees 13 15 N signatures based upon the different types of vegetation in their diet, and across different regions. Results indicated that manatees sampled in Belize and Puerto Rico had a d iet composed primarily of seagrasses, whereas Florida manatees exhibited greater region al variation, with some samples indicating manatee diet s were composed of only freshwater plants, vs. other locations where diet only indicated seagrass ( Alv es Stanley e t al 13 15 N v alu es differed based upon location in F lorida and Puerto Rico, but not for sex or age class or combinations of factors ( Alv es Stanley et al . 2010). One interesting aspect of this study examined epiphytes on seagrass blades for 13 15 N enrichment, but 13 15 N v alu es of seagrass ( Alv es Stanley et al. 2010). Results of plant an aly ses showed that freshwater plants were the most depleted for 13 C, estuarine plants were inter mediate, and seagrasses were the most enriched in 13C 13 C v alu es did not vary among seagrass species ( Alv es Stanley et al. 2010). Not surprisingly, 13 C v alu es followed the same trend as the plant s, and were lower in f reshwater manatees and higher for coast al environments ( Alv es Stanley et al. 2010). 15 N v alu es, all seagrasses were lower than other plants analyzed, but there was no difference between freshwater and estuarine species, and seagr asses did not differ 15 N v alu es ( Alv es Stanley et al.
101 2010). Values for 15 N for plants also differed by season in one location (St. Johns River) where they were collected season al ly ( Alv es Stanley et al . 2010). For manatee 15 N v alu es were lower in manatees from coast al regions and more higher from freshwater areas . Belize manatee samples were even more specific, 15 N v alu es between specific lagoons ( Alv es Stanley et al. 2010). Alt hough this study contributed v alu able data for wild West Indian manatees, the diet tissue discrimination v alu es were taken from stranded manatees, which are often compromised and cannot be considered he alt al ally , no Because the east coast has two of five management units for Florida manatees, it may not be prudent to extrapolate results to all Florida manatees without including any samples from the ea st coast in the study. Over all , this study showed that the 13 15 N v alu es of manatee food resources, i.e., plants, can change by region and season. Stable isotopes have also been used to compare food preferences and habitat use between fossil sireni ans and the Florida manatee. Oxygen stable isotope studies in anim als are less common than the use of carbon and nitrogen, but 18 O can be used to determine specific water sources used by anim als (Fry 2006). The 18 O v alu e of an aquatic mamm s body water 18 O v alu e of the water in which it lives, 18 O v alu es should exhibit 18 O v alu es than populations that remain in a single, loc ali zed water body (Clementz and Ko ch 2001). The stable isotopes 18 O and 13 C were used to an aly ze tooth enamel from fossil Florida sirenians (Protosirenidae and Dugongidae) from the Eocene to late Miocene epochs, and from Pleistocene and Recent manatees
102 (Trichechidae), also from Florida ( MacFadden et al. 2004). Results indicated that Protosirenidae and extinct 13 C v alu es ranging from 1.9 4.1 (mean= 3.4) , which they interpreted to represent a speci ali zed diet of predominantly 18 O v alu e of 30 hich indicated a princip al ly marine habitat ( MacFadden et al. 2004). These v alu es differed significantly from results for 13 C ranging from 13.9 0.5 (mean = 5.7 ) , interpreted as a more gener ali zed die t ranging from C 3 plants to 18 O v alu e of 2 9 a more diverse preference for freshwater and marine habitats ( MacFadden et al. 2004). As teeth develop in mamm als , the enamel gener al ly miner a li zes over a period of sever al months (depending on factors such as size of the anim al ), and MacFadden et al . (2004) hypothesize d that if sirenian tooth eruption time is also taken into consideration, this process could take up to a year in sirenians. Ther efore it can serve 13 C v alu es of food consumed as well as fluctuations in body water 18 O v alu es during miner ali zation ( MacFadden et al. 2004). This paper is v alu able in describing the technique and results for stable isotope an aly ses of sirenian tooth enamel, and for the comparison to the feeding ecology of extinct sirenians. For extant manatees this study used only 6 teeth from three manatees that have spent their entire 18 O and 13 C v alu es of wild Florida manatees. A 2007 study by Clementz et al. 13 C v alu es in diet and three types of tissue (bone collagen, bone and tooth bioapatite) from four species of sirenians: the West Indian manatee, the Amazonian manatee ( Trichechus inunguis ), the
103 Dugong ( Dugong dugon ow ( Hydrodam al is gigas ). Mean 13 C v alu es were then compared to those of six marine mamm al species, as well as terrestri al herbivores and carnivores (Clementz et al . 20 07). Stable carbon isotope v alu es for dugong samples were extremely high for all tissue types, and consistently higher than those for all other marine mamm als including seagrass foraging West Indian manatees (Clementz et al. 2007). Interesting results for 13 C tissue diet 13 C tooth diet v alu es for captive manatees and those that fed on freshwater vegetation (12.5 all species in the study and were similar to v alu es for terrestri al ungulates, but v alu es for du gongs and manatees that fed on seagrass were significantly lower (10.8 et al . 2007). 13 C tooth bone v alu foraging manatees, and tooth bioapatite was consis tently 13 C enriched relative to bone (Clementz et al 13 C apatite collagen v alu es were als ( all et al . 2007). In a later study Clementz et al. (2009 a 13 C apatite collagen v alu es between marine and freshwater foraging Florida manatees: while freshwater foraging manatees 13 C bioapatite 13 C collag en v alu es as terrestri al herbivores (and fell within a 95% confidence interv al with terrestri al herbivores), 13 C apatite collagen v alu es and did not f all within the 95% confidence interv al (Clementz et al. 2009 a ). The authors concluded that these results indicate the nutrition al qu ali ties of freshwater and marine plants are markedly different, and that given all manatees have the same digestive system
104 13 C apatite collagen v alu es reflect nutrition al and composition al differences between plant types (Clementz et al. 2009). These two studies were important because they highlighted both the importance of determining a species specific and tissue specific diet discrimination v alu 13 C, a nd the v alu e of combined an aly ses of multiple tissues to determine the qu ality of manatee diet. The studies reviewed here give a gener al overview of most of the stable isotope research of Trichechids to date. Very little stable isotope research has been c onducted for the Amazon ian manatee and the dugong, and none prior to this study (Chapter 4) has been performed for the African manatee. Stable isotope research is a non invasive way to collect valuable manatee foraging data that can greatly benefit manager s in knowing what types of habitat to protect. Samples that can be collected for African manatees include bone and teeth from hunted animals, those that died naturally, and museum collections. Bone analyses can indicate a lifetime average diet to provide a n accurate baseline for manatee foraging in different habitats . Overall Aims The African manatee has been the subject of the least research of any sirenian species, and in particular, the species has been the subject of very little genetics research . The two mtDNA genetics studies that hav e previously occurred had only a few samples available for an aly sis (Parr 2000, Vianna et al . 2006). Only five control region haplotypes and three cytochrome b haplotypes have previously been published for the species, a nd these originated from only f rom four of 21 range countries. However, African manatee haplotype diversity has been shown to be highest among the three manatee species (Vianna et al. 200 6) which hints at high over all genetic diversity. Basic information s uch as the number of distinct populations or evolutionary significant
105 units within the , and over all genetic diversity are all still unknown. Additionally, t here are no accurate estimates of abundance for the species , the effects of hunting and habitat destruction on the species are poorly documented, and the high level of trade in manatee bus hmeat throughout western Africa is surely impacting many populations (IUCN 2014). D ams have isolated populations in many countries, a nd natur al isolation likely occurs in inland river systems 1000 km or more from the coast in countries such as Chad (Powell 1996, Keith Diagne unpublished data). Conservation efforts are hindered by a lack of basic information about the species. Detailed reports have been published on the phylogeographic al distribution of manatees throughout their ranges, with the exception of Africa (Garcia Rodriguez et al. 1998, Cantanhede et al. 2005, Vianna et al. 2006, Hunter et al. 2010a, Nourisson et al . 2011 ). Base d upon the immense size of their range, which spans over 5930 km of African coastal habitat and thousands of kilometers more of inland riverine habitat, it is unlikely that there is much genetic mixing between individu al s from the northern end of the spec ies range with those from the southern end. Additionally, movement pattern data from the first satellite tracking study for African manatees (Keith Diagne et al . unpublished data), as well as studies of the closely related Florida manatee ( Deutsch et al. 2003) , also indicate that it is improbable that African manatees have home ranges greater than sever al hundred kilometers. Therefore it is unlikely that this species is panmictic. Genetics has the ability to shed light on population structure, both within specific countries and waterways as well as in the larger regional and range wide context.
106 Additionally mtDNA, particularly cytochrome b , is useful for estimating species and population divergence times ( Vianna et al. 2006 , Sun et al . 2011) . A few previou s estimates have been made for trichechid species divergence times ( Garcia Rodriguez et al. 1998, Cantanhede et al. 2005, Vianna et al. 2006), but sample sizes were very sm all for T. seneg ale nsis in previous estimates, so more robust sample sizes as well a s an updated estimate of rate of evolutionary change are needed for more accurate divergence estimates. A comparison of genetic variation between African manatees and the other trichechid and dugong species will increase our understanding of sirenian evolu tion. We also know little about the ecology of the African manatee. Stable isotope research is a valuable tool used to understand manatee feeding ecology , which can also have important conservation implications in managers in determining which types of fo od resources and habitat are important to protect. Carbon and nitrogen stable isotope analyses will identify the types of food resources utilized by manatees between the varied ecosystems in which they live, and can identify diversities in diets between di fferent populations and habitat types. Bone samples from African manatees can be collected from hunted animals, those that died naturally, and museum collections , and can indicate a lifetime average diet to provide an accurate baseline for manatee foraging in different habitats. Additionally, comparisons of recently collected bone with historic or museum samples allows us to investigate differences in food resources utilized today with those of the past, which in turn sheds light on changes to the ecosystem over time.
107 Building upon genetics and stable isotope research that has begun for West Indian and Amazonian manatees, analyses from their African cousins will provide a fascinating comparison that will continue to increase our understanding of the life hi story and evolution of sirenians. Therefore, the aims of this doctoral research were: 1. to increase the mitochondrial genetic information available for the African manatee 2. to conduct a phylogeographical analysis of mitochondrial genetic data for the African manatee from countries across the species range. 3. to conduct phylogenetic analysis of all extant sirenians in order to determine the evolutionary placement of the African manatee in relation to all other extant sirenians 4. to conduct carbon and nitrogen stab le isotope analysis to describ e African manatee diet for the first time . 5. to examine whether African manatee lifetime average diet v aried based upon the habitat the individual lived in and due to the environmental context of differing plan t and animal speci es it consumed. 6. to use stable isotope analyses to compare historic and present day African manatee samples to investigate possible differences in diet due to changes in habitat and environment over time .
108 Figure 1 1 . The range of the African manatee, s howing only major rivers. The locations of major dams are indicated by red bars , and the location of new major dams under construction are shown by orange bars . Base map courtesy of Ellen McIlhinny.
109 A B C D E F G H Figure 1 2. The e ight skull morphometric variables used by Domning and Hayek (1986) to confirm species identity of the three extant manatees, T. manatus , T. senegalensis and T. inunguis . Photos V30 V76 courtesy of Daryl Domning and Caryn Self Sullivan , Photos V91 V98 by Lu cy Keith Diagne.
110 Figure 1 2 . C ontinued A) V30: Extensions of posterolateral borders of supraorbital processes intersect: anterior to the posterior end of the mesorostral fossa, or about at its posterior end or less than 2 cm behind it or 2 cm or more behi nd it. B) V34: posteroventrally from lateral side of base of supraorbital process. C) V37: Portion of frontal parietal suture crossing skull roof straight, or strongly curved . D) V47: Dorsolateral border of exoccipital (adjacent to mastoid foramen) relatively smooth and rounded or forms a prominent posteriorly overhanging flange . E) V60: Transverse valleys of lower molars obstructed by distinct cristae obliquae extending anterolingually from hypoconids, or not obstructed . F) V76: Angle between occlusal plane of molars and masticating surface of mandibular symphysis (in degrees). G) V91: Shape of palatal surface of rostrum . H) V98: Relative se paration of mandibular angles.
111 CHAPTER 2 RAN GE WIDE GENETIC DIVERSITY AND PHYLOGENETIC STRUCTURE OF THE AFRICAN MANATEE Study Rationale and Objectives The African manatee ( Trichechus senegalensis ) is one of the least underst ood marine mammals in the world. The range of the species is larger than th e width of the United States and includes 21 countries along the Atlantic coast of Africa from Mauritania to Angola, and the interior co untries of Mali, Niger and Chad. T o date there have been very few studies on the species , most of which were short term field surveys with no follow up or local capacity building (Nishiwaki et al . 1982, Reeves et al . 1988, Grig i one 1996, Silva and AraÃºjo 2001) . There are no accurate estimates of abundance for the species anywhere in its range and the impact of hunting and h abitat destruction are poorly documented, but the trade in manatee bushmeat is well known throughout Africa. In recent times hydroelectric and agricultural dams have also isolated manatee populations in many major rivers, including the Senegal, Niger and V olta Rivers. Dam structures have also killed manatees both through entrapment and death in water control gates (Powell 1996, L. Keith Diagne unpublished data). Natural isolation is also believed to occur in inland river systems 1000 km and more from the co ast in countries such as Chad. Prior to the current study, less than 25 genetic samples had been analyzed for the species, all of which came from only six of the twenty one range countries: Ghana, Cameroon, Gabon, Niger, Chad, and Guinea Bissau (Parr 2000, Vianna et al . 2006). The lack of genetic information is not only of academic importance , but is necessary for conservation. I t is believed that threats to the species, including illegal hunting, isolation of populations due to dams, and other anthropogeni c causes may be resulting in serious population declines in many countries where the species
112 occurs. Conservation and management efforts are greatly hindered by a lack of basic information about the species, including phylogeographic structure and genetic diversity. The International Union for the Conservation of Nature ( IUCN ) Red List category and criteria classifies t he African manatee as Vulnerable and estimates that there is a high probability of a 30% or greater reduction in population size that will r esult within a 90 year, three generation period ( Keith Diagne and Powell 20 14 ). The Convention of Migratory Species (CMS) raised the species f rom Annex II to Annex I in 2009 in recognition of the high levels of threats to the species . Additionally, the Con vention on International Trade in Endangered Species (CITES) voted at their Conference of Parties in March 2013 to raise the species from Appendix II to Appendix I, which bans all legal international trade, but leaves enforcement of illegal trade up to the range countries. Overall, the trajectory for the African manatee is believed to be in serious decline without drastic and wide ranging conservation efforts on their behalf ( Keith Diagne and Powell 20 14 ). The African manatee has been the subject of the le ast genetics research of any sirenian species . Genetic analysis is a useful tool that has been in creasingly appli ed over the past 25 years to assist biologists and managers in understanding manatee population structure, genetic diversity, evolution, and ph ylogeography. Basic information, such as the number of distinct populations or evolutionary significant units genetic diversity is still unknown. Therefore , range wide gene tic analyses could assist in address ing the paucity of baseline data by determining distinct populations, relatedness
113 between populations, effective population sizes, and indicators of population changes such as bottlenecks, which are effective in assessin g the true impact of threats and informing conservation and management decisions. Only two mitochondrial DNA ( mtDNA ) genetic studies have previously occurred for African manatees: Parr (2000) analyzed 21 samples from five countries (Cameroon, Chad, Gabon, Ghana, and Guinea Bissau), and Vianna et al . (2006) used six of the same samples from three of the countries ( Chad , Ghana and Guinea Bissau) and one new sample from Niger. Parr (2000) reported the collection location of samples within each country , but no sample information was provided by Vianna et al . ( 2006 ). Analyses by Parr (2000) reported three clusters that were geographically differentiated: one from Guinea Bissau, one from Chad, and one that included samples from Cameroon, Gabon, and Ghana. Vianna et al. (2006) identified and published five control region haplotypes of 41 0 base pairs (bp) and three cytochrome b haplotypes of 615 bp from the four different locations. Haplotypes are a combination of alleles on adjacent loci that are inherited together , which are indicators of genetic diversity, and can inform researchers about relatedness between populations. The control region haplotypes identified in Vianna et al. (2006) formed two distinct clusters: cluster I which included Y01 and Y02 haplotypes (b oth from Guinea Bissau), and cluster II included the Y03 (Ghana), Y04 ( Chad) and Y05 ( Niger) haplotypes. These control region haplotypes were characterized by 15 polymorphic sites consisting of 14 transitions and 1 transversion (Vianna et al. 2006). No hap lotype was shared between locations . African manatees displayed the highest haplotype diversity ( h = 0.9333, n = 6) when compared with West Indian ( T richechus manatus , h = 0.8554, n = 224) and Amazonian ( T richechus inunguis ,
114 h = 0.8772, n = 92 ) manatees , s ugges ting a large population size, high mutation rate , and/or distinct populations for African manatees (Vianna et al. 2006). The Vianna et al. study concluded with the recommendation that a larger population genetics study of African manatees should be un dertaken with samples from throughout the species range. Detailed reports have been published on the phylogeographical distribution of manatees throughout their ranges, with the exception of Africa (Garcia Rodriguez et al . 1998, Cantanhede et al . 2005, Vi anna et al . 2006, Hunter et al. 2010, Hunter et al . 2012, Nourisson et al . 2011 ). Therefore, t he current study present s the first genetic analysis of African manatee populations from countries across the range. Based upon movement pattern data for Florida manatees and data from the first satellite tracking study for African manatees (Deutsch et al. 1998, Deutsch et al. 200 3 , Keith Diagne et al. in prep.), it is unlikely that African manatees have home ranges greater than several hundred kilometers, or that there is much genetic mixing between individuals from the northern end of the species range with those from the southern end. Therefore, it is unli kely this species is panmictic . The objective of this research was to u tiliz e genetics as a tool to charact erize African manatee populations and to define the geographic structure and diversity of those populations, both within specific waterways and countries, as well as in the larger range wide context. This information will also allow for comparison of manat ees sampled from different locations to examine population connectivity or isolation, both within and among countries. The non coding mtDNA control region was chosen for use in these analyses since it is a common ly used means of surveying intraspecific -
115 pop ulation genetic diversity . The control region is useful in determining genetic diversity between populations because it has a higher rate of mutation than coding regions such as cytochrome b (Frankham et al . 2010). Cytochrome b is adjacent to the control r egion on the mitochondrial genome and is also commonly used in population level studies because it is more conserved and can dete ct patterns o f both relatedness and divergence within populations and species (Frankham et al . 2010) . Methods Sample Collection Due to the difficulty in accessing African manatees across their large and remote range, all samples we re collected opportunistically. Samples were collected over eight years during annual fieldwork starting in Gabon in 2006, Senegal in 2009, and during s horter term work in Mali in 2010 . Samples for analysis include sub dermal or muscle tissue, or blood from live or dead manatees. In Gabon and Senegal manatee carcasses that died of both natural and anthropogenic causes , as well as live manatees were sample d . Additional samples from carcasses were pro vided by collaborators from Cameroon, Gabon, Guinea, Ivory Coast, Mali, Niger, and Senegal . Manatees captured alive (in rescue o r research capture situations) we re sampled according to standard protocols used fo r Florida manatees over the past 25 years: a skin sample approximately 2 cm x 2 cm in siz e wa s taken from the caudal margin of the tail using a cattle ear notcher (Bonde 2009) . Samples we re preserved as follows: all fresh tissue samples from live or dead m anatees were stored in SED buffer solution ( saturated NaCl; 250 mM EDTA, pH 7.5;20% DMSO) at room temperature. Dried samples from dead manatees , including dried skin and musc le tissue at least 2 cm x 2 cm , were stored dry, while blood wa s preserved using a lysis buffer ( 100mM Tris HCI, pH 8;
116 100mM EDTA, pH 8;10mM NaCI; 1.0% SDS (weight/volume) that can be stored long term at room temperature . S amples we re exported to the United States for analysis. A ll necessary permits to conduct this research and to expo rt / import African manatee samples from African countries into the United States were obtained (from the U.S. Fish and Wildlife Service, as well as the countries of Gabon and Senegal). Collaborators who collect ed and export ed samples for analysis obtain ed the appropriate research and export permits in their countries. Detailed records are kept for each sample, including ID number, other date and location (latitude, longitude, locality), any known information about th e manatee (age class, sex, total length , cause of death, photos taken), export and import permit numbers, and additi onal notes of interest. A portion of each genetic s sample collected for this study was archived for future analyses, to be processed as new techniques and informative studies evolve. A total of 65 African manatee tissue samples and one blood sample were collected for this study. Samples were collected from Cameroon (n=9), Gabon (n=21), Guinea (n=2), Ivory Coast (n=2), Mali (n=3), Niger (n=2) and Senegal (n=27). Published partial mtDNA control region 410 bp haplotype sequences for African manatees (n= 5) wer e downloaded from GenBank (www.ncb i.nlm.nih.gov) for comparison to new haplotypes identified by this study. Published partial mtDNA cytochr ome b 615 bp haplotypes for African manatees (n= 3 ) wer e also downloaded for analysis. Genetics Analysis Analysis of all samples was conducted at the U.S. Geological Survey, Southeast Ecological Science Center, Sirenia Project Conservation Genetics Laborat ory in Gainesville, Florida, USA. For phylogenetic analyses, mt DNA wa s extracted using
117 Qiagen DNeasy Blood and Tissue kits (Qiagen Inc., Valencia, CA) for fresh tissue and blood samples, and p henol c hloroform isolation for small or degraded samples (modifi ed from Hillis et al . 1990). Most samples in this study required extraction using the p henol c hloroform method. Primers CR 4 and CR 5 ( Southern et al . 19 8 8 ) we re used to amplify a 410 bp portion of the mtDNA control region displacement loop. Each PCR react ion included: 10 ng DNA, sterile PCR water, 1 x PCR buffer (10 mM Tris HCl, pH 8.3, 50 mM KCl, 0.001% gelatin; Sigma Aldrich, Inc., St. Louis MO), 0.8 mM dNTP, 3 .0 mM MgCl 2 , 0.25 mM BSA, 0.25 and 0.04 units of Sigma Jump Start Taq DNA po lymerase. All PCR amplifications we re carried out on a MJ thermal cycler (MJ Research, Waltham, MA, USA). The PCR cycling profile was as follows: , followed by 34 cycles of denature for then with a final 10 min utes at 72Â°C. For samples which did not amplify, likely due to degradation, a PCR gradient was run and 50 was selected as the optimum annealing temperature along w ith increasing MgCl 2 and BSA quantities in the PCR mix. For cytochrome b , primers previously used by Vianna et al . (2006) to amplify a 615 bp region of mt DNA for all three species of manatees ( MVZ3 and MVZ4, based on Kocher et al . 1989 ) we re used in this study, but resulted in short sequences of 329 338 bp in all samples tested. Therefore, we used a cyt ochrome b primer pair that had successfully amplified the complete cyt ochrome b gene in 40 species of mammals, including the dugong ( Dugon g dugon ) (Naidu et al . 2012). The forward MTCB F (5 CCHCCATAAATAGGNGAAGG 3 ) and reverse MTCB R (5 AGAAYTTCAGCTTTGGG 3 ) (Naidu et al . 2012) primer s were run in PCR reactions
118 that included the same mix as listed above for control region PCR. A gradient PCR ranging from 48 62 o C was run at the beginning to determine the optimal annealing temperature for the samples and the optimal amount of MgCl 2 for the mix. The following PCR cycling profile was then used , 60 then with a final 10 min utes at 72Â°C. PCR products were analyzed by gel electro phoresis using 5 Âµl of DNA and 2 Âµl of 1x load dye . Bands were compared to a 100 0 kb low mass ladder. PCR product was cleaned using QIA Quick kits (Valencia, CA) or ExoSap purification (Affymetrix, Santa Clara, CA) following guidelines. For ExoSap, 5 ÂµL of PCR product from each sample was mixed with 2 ÂµL of ExoSap and the mix was heated to 37Â°C for fifteen minutes, followed by heating to 80Â°C for fifteen minutes to inactivate the enzyme. Samples were sequenced following the Big Dye v.3kit protocol using an ABI sequencer. Representatives from each haploty pe and any ambiguous sequences we re sequenced in both directions to ensure the accuracy of nucleotide designations. Statistical Analyses Fifty samples were sequenced for control region and aligned to previou sly published control region sequences , trimmed to 410 bp and checked for errors by hand using G ENEIOUS v. 7.1 (Biomatters Ltd., Auckland, New Zealand). Any sample containing a novel haplotype or ambiguous base was sequenced in the opposite direction for c onfirmation. Control region sequences and unique haplotypes were also analyzed by regional location, defined as North (Senegal, Guinea Bissau and Guinea) or South (Mali, Ivory Coast, Ghana, Niger, Chad, Cameroon, and Gabon) based upon separation of these r egions in all evolutionary history models. Countries with greater
119 than seven control region sequences (Gabon, Senegal and Cameroon ) were also analyzed separately. For cytochrome b , 23 samples were successfully sequenced in both forward and reverse directi ons and concatenated in G ENEIOUS to create a consensus sequence for each sample. Consensus sequence lengths were trimmed to 1227 bp. In order to compare the new sequences with three previously published cytochrome b sequences ( Y1, Y3 and Y4, Vianna et al . 2006), all sequences were trimmed to 605 bp due to nucleotide discrepancies at bp position 10, and aligned in G ENEIOUS . For cytochrome b analyses, North was defined as samples from Senegal and Guinea, and South as samples from Gabon and Cameroon, based upo n these separations in all evolutionary history models. The 23 samples which were successfully sequenced for both control region and cytochrome b were concatenated and aligned in G ENEIOUS to analyze them using a multi locus approach. DNAsp software v 5.10 (Librado and Rozas 2009) was used to calculate summary statistics including of polymorphic sites, sequence divergence, population parameter of genetic diversity (Theta), and selective neutrality . For individual locus and concatenated sequences , 88 evolution models were tested using the software program jMODELTEST2 v 2.5 (Darriba et al . 2012). Due to small sample sizes, c orrected Akaike Information Criterion (AICc) used maximu m likelihood (ML) optimality criterion to determine the best model for the dataset in order to estimate rates of nucleotide substitutions and invariant sites. The resulting models were then used in MEGA6
120 software (Tamura et al . 2013) using a ML heuristic m ethod with 1000 bootstrap replicates to build phylogenetic trees. ARLEQUIN v 3.5 software (Excoffier and Lischer 2010) was used to calculate an analysis of molecular variance (AMOVA) . T ests of differentiation between populations using pairwise ST statist ics examine d haplotype frequencies and molecular divergence while exact tests examine d haplotype frequency distributions. Statistical significance for pairwise ST statistics, the exact test , and AMOVA were assessed using 10,000 permutations and a statisti cal significance of P <0.05. The Tamura genetic distance model was used to allow for unequal nucleotide frequencies, and so that overall nucleotide frequencies and transition and transversion ratios could be computed from the data ( Tamura 1992). The data w ere analyzed in a hierarchical manner to estimate variance components at the different spatial scales: the level of genetic differentiation was measured by distance among regions, among populations within regions, and within populations, respectively. Taj D neutrality test (Tajima 1989), which examines whether a population is at neutral drift mutation equilibrium by comparing nucleotide diversity with segregating sites, was tested using DNAsp. Fu's F S test of neutrality was also conducted because it i s a more sensitive indicator of population expansion and genetic hitchhiking than D S test (Fu 1997), which estimates mutation rate based upon the number of unique mutations was conducted using ARLEQUIN. Both tests were con ducted with statistical significance of P <0.05. Bayesian phylogenetic inference using Markov c hain Monte Carlo (MCMC) methods was used to reconstruct evolutionary relationships between all sequences
121 within new control region and cytochrome b haplotypes in this study, as well as concatenated control region / cytochrome b sequences. MRBAYES v 3.2.1 software was utilized for these analyses (Ronquist et al . 2012). For control region and cytochrome b haplotype sets run separately, GTR+I+G models were run for 1,000,000 generations with a relative burnin of 25% (250,000 generations). For concatenated sequences, a GTR+I+G model were run for 5,000,000 generations with a relative burnin of 25% (1,250,000 generations). Potential Scale Reduction Factor (PSRF) was use d as a convergence diagnostic; values approached 1.0 as runs converged. Consensus trees were produced using FIGTREE version 1.4.0 (Rambaut 2007). Results Mitochondrial DNA was isolated from 49 African manatee tissue samples and one blood sample. Intact DNA was not successfully isolated from an additional 17 samples after multiple attempts (three tries per sample); in all cases the manatee tissue was highly degraded and/or had been cooked prior to being collected for this study . These are the first African m anatee genetic samples collected and analyzed from four of seven countries in the study: Senegal, Guinea, Ivory Coast, and Mali. For the four countries where samples had previously been collected, the additional samples and the varied collection site local ities and waterways within these countries , provide d greater regional coverage of manatee populations to help further define population structure. Control R egion Fourteen new haplotypes from six countries were identified from control region sequences ( Tab le 2 1) . Including the five previously published haplotypes (Vianna et al . 2006), there are a total of 19 species haplotypes in 55 sequences . For all control region haplotypes ST = 0.80740). Twelve of 19
122 haplotypes (63%) were identified in more than one individual. Control region genetic diversity indices are shown in Table 2 2. Haplotype diversity ( HD = 0.9091, n = 55) is higher th an the averages previously reported for either of the two other manatee species, T. inunguis ( HD =.8772, n = 92) and T. manatus ( HD = .8554, n = 224) (Vianna et al ) fell between values reported in Vianna et al . (2006 ) for T. manatus T. inunguis T. manatus clusters reported in the same publication are considered separately (T.m. cluster I II T.m. cluster III then T. senegale nsis also has the highest nucleotide diversity of the three species. There were 25 polymorphic sites including 21 transitions and four transversions. Nucleotide frequencies were A = 31.13%, T = 25.73%, C = 14.24%, and G = 28.89%. test value ( D S test (F S = 1.412) were not significant ( P > 0.05) for the species , therefore the null hypotheses of selective neutrality could not be rejected. Analysis of the 19 haplotype sequences using the program jMODELTEST2 indicated the maxim um l ikelihood method based on the General Time Reversible model with invariant sites and gamma shape parameters (GTR+I+G) was the best evolutionary history substitution model for t his data (Nei and Kumar 2000). An initial tree for the heuristic search w as obtained by applying the neighbor j oining method to a matrix of pairwise distances estimated using the maximum composite l ikelihood (MCL) approach. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+ G , parameter = 0.0500)). The rate variation model allowed for some sites to be evolutionarily invariable (+ I, 21.4585% sites). Evolutionary analyses
123 were conducted in MEGA6 and the resulting tree with the highest log likelihood ( 886.054) and 1000 bootstrap replicates is shown in Figure 2 1. The control region consensus tree and all analyses identified two distinct groups separated geographically into North and South regions (Figure 2 2). North included published haplotypes Y01 and Y02, and samples from Seneg al, Guinea Bissau and Guinea. South included published haplotypes Y03, Y04 and Y05, as well as sequences from Ivory Coast, Mali, Niger, Chad, Cameroon and Gabon. Only one haplotype (TS CR03) was found in both the North and South regions: it was identified in one individual from southern Gabon and in one individual from southern Senegal. The North regional clade is further separated into two geographically separated populations . T he Senegal River population is separated from coastal Senegal, Guinea Bissau an d Guinea (Figure 2 3). The South regional clade is separated into distinct sub populations as well: a larger cluster of nine haplotypes from Gabon, Cameroon, Ghana , and Ivory Coast forms one subpopulation, and four haplotypes from the interior countries of Niger and Mali, as well as one haplotype from Cameroon form a separate group. However, few samples from countries located between the northern and southern regions have been collected and sequenced to date , so this separation should be treated with caution . The highest haplotype diversity from a specific country was recorded in Gabon ( HD = 0.813 , Figure 2 4), but sample sizes from most countries were very small. The AMOVA (Table 2 3) indicated that most of the genetic variation occurred between the North an d South regions (66%), with less variation among populations within regions (14.7%) and within countrie s (19.3%). All AMOVA results were significant ( P ST P values (<0.05)
124 between Senegal and Guinea, Ivory Coast, Cameroon and Gabon, and between Gabon and Guinea, Guinea Bissau, Mali, Niger and Cameroon (Table 2 4). The Bay e sian analysis of 19 co ntrol region haplotypes resulted in a consensus tree which closely resembled the ML tree with less fine scale structure (Figure 2 5). The harmonic means for the two runs were: chain 1: 889.09 and chain 2: 894.10 with a total harmonic mean of 893.42. Sum maries were based on a total of 4002 samples from 2 runs. Summary statistics for partitions with frequencies of >0.10 in at least one run were: average deviation of split frequencies = 0.018495 and maximum standard deviation of split frequencies = 0.110235 . Maximum PSRF parameter value was 1.012 . Of the 3002 trees, a 50% credible set included 1501 trees, and a 99% credible set included 2972 trees. The consensus tree indicated the same regional North / South separation, as well as sub populations with regions , of all haplotypes in the ML consensus tree. One notable difference from the ML analysis was that the Mali sample, TS CR06 , did not align with the inland haplotypes in the Bayesian analysis, but rather formed a separate group with an Ivory Coast sample, T S CR05. Of the 14 newly identified haplotypes, only four were found in more than one country. One of these (TS CR03) was identified both in coastal Senegal and coastal Gabon, at opposite ends of the African manatee range, over 4565 km apart (Figure 2 2). I n addition to the 14 new haplotypes, three of five previously published haplotypes from Vianna et al . (2006) were identified in new countries. The Y02 haplotype (GenBank accession AY963895), which was identified in Guinea Bissau by Vianna et al . (2006) , wa s documented in two individuals from the Delta Saloum region of central coastal Senegal. The Y03 haplotype (GenBank accession AY963896), which Vianna et al .
125 (2006) identified in Ghana, was documented in two new countries: coastal Ivory Coast (n=1) and Gabo n (n=8) . In Gabon, the haplotype was found in sequences from both the Ogooue River ( n=3) n=5 ). The Y04 haplotype (GenBank accession AY963897), which Vianna et al . (2006) identified from Lac TrÃ©ne in Chad, was documented by this study in a manatee from the Niger River in Niger. Cytochrome b Cytochrome b was sequenced in t wenty three samples from four countries (Senegal, Guinea, Cameroon, and Gabon) and resulted in nine new haplotypes for the species ( Table 2 1 ) . These were added to th ree previously published haplotypes ( Figure 2 6) . Although a gradient PCR was run with the new primers and samples were optimized, an additional 11 samples were unable to be successfully amplified during PCR, or sequences had to be removed from analyses du e to high levels of background noise. Genetic diversity indices for new cytochrome b haplotypes are shown in Table 2 5. Although the sample size is small, high haplotype diversity (HD = 0.8419) and low were identifie d, sim ilar to control region sequences. Although Vianna et al . (2006) did not publish the locations of origin for any of the T. senegalensis cytochrome b haplotypes, the cytochrome b locus is adjacent to the control region and a lack of recombination results in presumed linkage . Therefore, it is reasonable to assume that Y1 ( GenBank accession AY965882 ) originated from Guinea Bissau, Y3 ( GenBank accession AY965881 ) from Ghana, Y4 ( GenBank accession AY965880 ) from Chad, and the new haplotypes identified by the pres ent study would then represent four new countries of distribution. No haplotypes were shared between countries, but in Gabon one haplotype (TS
126 twelve new haplotypes (33%) oc curred in more than one individual, but this may be an artifact of low sample sizes . The AMOVA results (Table 2 6) indicated that most of the genetic variation occurred between the North and South regions (70.1%), but with more variation between population s within regions (18.7%) than within populations (11.2%). All AMOVA results were significant ( P < 0.05), but population results should be viewed ST differences calculated using the Tamura distance model were significant between all four countries ( P < 0.05), with the exception of Guinea and Cameroon, which may be due to small sample sizes from these countries (Table 2 7) . Analysis of the nine new and three previously published African manatee cytochrome b ha plotypes using the program jMODELTEST2 indicated the ML method based on the General Time Reversible model with invariant sites and gamma shape parameters (GTR+I+G) was the best model for this data (Nei and Kumar 2000) . MEGA6 was used to create ML GTR+I+G t rees for t he twelve sequences rooted in T. manatus cytochrome b haplotype M (Gen Bank accession AY965885 ). The tree with the highest log likelihood ( 889.716) is shown in Figure 2 7. A discrete Gamma distribution was used to model evolutionary rate differe nces among sites (5 categories (+ G , parameter = 0.0500)) and the rate variation model allowed for some sites to be evolutionarily invariable (+ I , 4.1725% sites). As a result of trimming the new haplotype sequences to 605 bp for analysis with the previously published haplotypes, the new sequences produced only two haplotypes and aligned geographically within the North and South regions . T he previously published haplotypes also f e ll within the two regions .
127 A second cytochrome b un rooted ML GTR+I+G tree ( + I , 0.0010% sites , 5 categories (+ G , parameter = 200.0000)) was produced for the nine new haplotypes at 1227 bp lengths. The evolutionary history was inferred by using this model in MEGA6 and the tree with the hi ghest log likelihood ( 1767.987 ) is shown in Fig ure 2 8. Bayesian analyses of twelve cytochrome b haplotypes resulted in a consensus tree topology identical to the structure of the cytochrome b un rooted ML haplotype tree ( Figure 2 8). The harmonic means for the two runs were: chain 1: 880.51 and chai n 2: 879.82 with a total harmonic mean of 880.22. Summaries were based on a total of 3002 samples from two runs. Summary statistics for partitions with frequencies of >0.10 in at least one run were: average deviation of split frequencies = 0.008737 and m aximum standard deviation of split frequencies= 0.017901. Maximum PSRF parameter value was 1.006. Of the 3002 trees tested, a 50% credible set included 1500 trees and a 99% credible set included 2971 trees. Concatenated Control Region and Cytochrome b Seq uences Twenty three control region and cytochrome b sequences from four countries were concatenated to create fourteen unique 1637 bp sequences . Substitution models were tested with jMODELTEST2 and results indicated the GTR+I+G model was the best fit for t he data ( Log likelihood = 2515.537 ) . Using this model, phylogenetic trees were created in MEGA6. The tree with the highest log l ikelihood ( 2525.451 ) is shown in Figure 2 9 . Similar to both control region and cytochrome b phylogenetic trees, the trees for the concatenated sequences also strongly indicated geographic regional separation ST = 0.8891) . Haplotypes in the North region were from Senegal and Guinea, while haplotypes from the South were from Gabon and Cameroon. Genetic diversity indexes
128 number o f polymorphic sites (S = 33), sequence diversity (K = 10.846 ), mutation rate D (0.81530) . Nucleotide frequencies were A = 28.73%, T = 26.39%, C = 27 .34%, and G = 17.54%. AMOVA indicated that 69.66% of variation was betwee n regions, 19.26% between populations within regions and 11.09% between individual countries. All AMOVA values were significant ( P <0.05) and a strong phylogenetic structure was indicated overall ST = 0.8891 , P ST differences were s ignificant, with the exception of Cameroon with Guinea (Table 2 8) . Separate Bayesian analyses were conducted for the 23 concatenated sequences as well as the 14 unique haplotypes. The 14 haplotypes resulted in a higher likelihood consensus tree ( Figure 2 10) which was very similar to the concatenated sequence ML tree, with long branch lengths for the Senegal River and both Guinea sequences. The harmonic means for the two runs were: chain 1: 2586.94 and chain 2: 2584.59 with a total harmonic mean of 25 86.34. Summaries were based on a total of 7502 samples from 2 runs. Summary statistics for partitions with frequencies of >0.10 in at least one run were: average deviation of split frequencies = 0.036834 and maximum standard deviation of split frequencies= 0.213046. Maximum PSRF parameter value was 1.076. Of the 7502 trees, a 50% credible set included 5877 trees and a 99% credible set included 13228 trees. Discussion Control R egion Control region haplotypes exhibited high levels of diversity with 19 haplot ypes , 14 new, identified from 55 sequences. In contrast, the nucleotide diversit y values indicate d low levels of nucleotide divergence between haplotypes. Low mtDNA nucleotide diversity may also signal reductions in effective population sizes ( N e ). Many ha plotypes
129 within regions differed from each other only by 1 3 bp. The large number of haplotypes with minimal nucleotide differences suggest rapid expansion after experiencing a small effective population size. This may be the result of the colonization of the Africa n continent by a small population of manatees which then expanded along the coast, moved into river and lagoon systems, and diverged over time. Similar stepping stone population models have been reported for West Indian manatees (Vianna et al . 20 06, Nourisson et al . 2011). I n this study, only one of 19 haplotypes occurred regionally between the North and South, only four of 19 occurred in more than one country, and a majority of haplotypes identified in Senegal, Gabon and Cameroon we re locally end emic and did not occur outside specific waterways within those countries. This is a clear indication of rapid expansion along with possible local isolation within regions, as well as within populations. Since mtDNA ref lects matrilineal lin ag es, these data likely reflect small effective female population sizes . Additionally, given that the species occurs over such a large range and was generally considered abundant in many locations as recently as 100 years ago, it would not be surprising if large population sizes had existed in the recent past, and still may occur in certain countries such as Gabon today. D values were not significant (Table 2 2) . S test did have two negative significant values, for the South region (F S = 4.914 ) and for Gabon (F S = 3.250) , which signify recent population expansion (Holsinger 2012). These D being considered a more conservative test of nucleotide diversit S
130 incorporating mutation rates into the calculation . Caution should be used in interpreting D S test results in this study, as all sample sizes were low. ST differences and exact tests indicated significant diffe rentiation among populations within and between regions (Table 2 4). Gabon was significantly different from all other countries except Chad, Ivory Coast and Ghana, which is unsurprising given the high levels of haplotype diversity identified in that countr y. Cytochrome b The slower mutation rate of cytochrome b makes it useful in determining evolutionary history and genetic diversity at regional population scales. Comparison of the previously published haplotypes (Vianna et al. 2006) with the new cytochrome b haplotypes required trimming all sequences to 605 bp, which removed much of the diversity indicated by the longer sequences and resulted in only two haplotypes, one for the North region and one for the South (Figure 2 5). However, when the new 1227 bp l ength cytochrome b sequences were analyzed separately, nine new haplotypes resulted (Figure 2 6). The new haplotypes indicated strong North and South region al separation . Thus the 1227 bp sequences presented here greatly increased the haplotype diversity i nformation for the species, and almost doubled the sequence length reported for Trichechids. Corroborating the control region diversity results , cytochrome b haplotype diversity was high and nucleotide diversity was low over all 23 sequences analyzed. No h aplotypes occurred in more than one country. Phylogeography of the African M anatee The results of AMOVA for both control region and cytochrome b indicated that the majority of variation was found among regions, which is not surprising given the strong sepa ration observe d between the ST
131 differences and exact tests indicated significant differentiation among populations within and between regions (Table s 2 3 and 2 7). For cytochrome b , Senegal was significantly differe nt from all other countries for both tests, and Gabon was significantly different from all other countries except for Guinea in the exact test. Of specific i nterest were two manatees from Guinea, which both control region haplotype TS CR04 , but were found to be two different cytochrome b haplotypes (with three polymorphic sites) . Each control region, cytochrome b , and concatenated sequence analysis identified a strong North versus South geographic separation of haplotypes. One explanation for this may be d ue to regional matrilineal population differences due to low dispersal across the large range of the species . Additionally, some of this separation is likely due to having few samples from countries in the middle of the range (Ivory Coast and Ghana) and no ne from other countries (particu larly Sierra Leone and Liberia), but the separation of these two clades was strongly supported and may persist with additional data. Therefore, i t is possible that one or more geophysical barriers may inhibit manatee moveme nt between northwest Africa and the Gulf of Guinea. Barriers in that region include the convergence of major oceanic currents near Sierra Leone and Liberia (including the Canary Current, the North Equatorial Countercurrent, the Benguela Current, and the So uth Equatorial Current), well documented coastal upwelling of cold water (Ingham 1970, Sarnthein et al . 1982), and an extreme narrowing of the continental shelf in Sierra Leone (from approximately 145 km wide at the northern border of the country, to appro ximately 32 km wide at the southern border). Hunter et al . (2012) reported a similar separation of manatee populations in Puerto Rico due to geophysical barriers includi ng a narrow continental shelf. O nce samples from countries
132 between the two current regi ons are included in future analyses, i t will be of great interest to determine whether or not central areas indicate isolation by distance , or have a strong gene flow boundary , an d where the boundary is located . Samples are particularly needed from Sierra Leone and Liberia, but additional samples from Ivory Coast, Nigeria, Ghana, Togo and Benin will also be beneficial. The North regional group is further divided into two geographically separated populations: inland ( Senegal River ) and coastal ( Senegal, G uinea Bissau and Guinea ) . The inland Senegal River population has been isolated since the construction of the Diama Dam near the mouth of the river in 1984, however, due to the phylogenetic separation of mtDNA haplotypes from all other African manatee sequ ences , this population appears to have been naturally isolated before construction of the d am. The unique Senegal River haplotypes were not identified at any other locations and the evolutionary history inferred from the phylogenetic analyses suggest s a de ep separation of this population from all others . T here was only a single control region and cytochrome b haplotype from samples collected approximately 800 miles apart in the river . Geographically , the Senegal River is greatly isolated from any other fres hwater bodies: the Sahara Desert is to the north , and the Dakar peninsula, which extends out into the Atlantic Ocean to the edge of the continental shelf, likely acts as a barrier to north south movement. The nearest freshwater source to the Senegal River is a network of freshwater springs in coastal Delta Saloum, approximately 340 km to the sou th. Manatee dependence on fresh water could inhibit travel outside of isolated sources (Hunter et al ., 2012). Additionally, manatees inhabit the Senegal River up to 9 50 km inland from the mouth, but generally do not migrate greater than several
133 hundred kilometer s, possibly further allowing isolation of population s . Additional data may provide information about when this population diverged from other African manatee po pulations. Similarly, manatees in th e Niger River in Niger and Mali , living over 2000 km inland from the coast, may have been naturally isolated prior to the construction of the four major and many minor dams that now divide that river into smaller segmen ts as it flows to the sea. Identification of three published haplotypes in new countries greatly increases their geographic range. The Y02 haplotype , previously identifie d in Guinea Bissau, was documented in two new individuals from the Delta Saloum regio n of central coastal Senegal (Vianna et al . 2006). Although only the country of collection for the original six samples was reported by Vianna et al . (2006) , the distance from Delta Saloum to the Guinea Bissau border is only 178 km. This is well within the distance African manatees have been documented to travel . A 2009 satellite tracking study o f manatees in the Senegal River documented movements of 141 281 km by individual manatees within several weeks (Keith Diagne et al . in prep). The Y03 haplotype, whi ch Vianna et al . (2006) identified as collected in Ghana , was documented in coastal Ivory Coast and coastal and inland Gabon. In Gabon, the haplotype was found in sequences from both the Ogooue River ( n=3) n=5 ), and it accounted for eight of 19 (42%) samples from that country. Therefore, the Y03 haplotype has now been documente d at four sites over 2771 km apart. The Y04 haplotype, which Vianna et al . identified from Lac TrÃ©ne in Chad, was documented by this study in the Niger River i n Niger, a minimum of 1850 km away. These locations are separated by the Kainji Dam in Nigeria which was constructed in 1968, permanently separati ng populations. Therefore
134 the Y03 and Y04 haplotypes appear to be widespread across large portions of the rang e, which may indicate that these are more basal haplotypes. The single control region haplotype (TS CR03) identifi ed in both the North and South region s may also represent a widespread haplotype that has not yet been identified in other locations. Althoug h this sequence aligns within the South regional group, it also appears to be the least diverge nt , most basal haplotype in that group. The one TS CR03 manatee from Gabon was an orphan manatee that washed ashore from the Atlantic Ocean at the southernmost t own at the border with the Congo. This young manatee was believed to have been carried northward by prevailing ocean currents from a country south of Gabon, since there are no freshwater sources in Gabon south of y barnacles on the calf, he was likely at sea for a minimum of one to two weeks, and ma y have originated from the Republic of the Congo, D emocratic R epublic of the C ongo , or northern Angola. Fourteen of 19 control region haplotypes in this study were foun d only in a single water body (river or lagoon), which supports the hypothesis of rapid expansion after a population bottleneck, and may indicate local isolation in some cases. Detection of seven out of eight Gabon haplotypes in a single water body may ind icate strong local adaptation and low genetic mixing. Manatees traveling between major lagoons and rivers in Gabon would need to move between these sites using the ocean, which has very strong currents a nd a narrow continental shelf. African m anatees are r arely documented in the Atlantic off Gabon, and with abundant food resources in rivers and lagoons, general ly do not need to move between freshwater habitats. As discussed
135 above, similar geophysical barriers have been hypothesized to impede movements of We st Indian manatees (Hunter et al . 2012). T he South ern region analyses included the furthest inland record of the African manatee in Mali. The sample was collected a minimum of 2956 km up the Niger River from the Atlantic Ocean. The sequence also consistent ly exhibited long branch lengths in both ML and Bayesian analyses, making it highly derived and indicating that this haplotype has likely been isolated for a long period of time. Secondly, on e coastal Cameroon control region haplotype (TS CR08) grouped wit h inland (Mali, Niger, and Chad) haplotypes rather than the other coastal Cameroon and Gabon haplotypes . This may indicate movement along the coast to or from the Niger River, which has also been verified by sightings of manatees in this region (E. Eniang, pers. comm.). The area of coastal Cameroon where the two manatee samples of this haplotype were collected is only 465 km south of the mouth of the Niger River, well within the normal range of movement of manatees, and therefore mixing between these popula tions is p os sible. Additionally, the presumed linked corresponding cytochrome b and concatenated sequences from the same individuals strongly separated from the coastal samples, but in those analyses we did not have inland sequence data to include for comp arison. Implications for C onservation of T. senegalensis The results presented here are only the first step in our understanding of African manatee genetic diversity and population boundaries across their very broad range. However, this matrilineal derive d information does identify regional population structure, and local population structure within Senegal and Gabon. K nowledge of distinct populations will allow managers in those countries to focus conservation efforts. For example, trans boundary conserva tion and management efforts could be set up
136 between Senegal, Guinea Bissau , and Guinea sin ce their coastal populations are regionally distinct. The Senegal River manatee population has very low mtDNA diversity in addition to being permanently isolated from all other African manatee populations, therefore it would benefit from careful management to ensure its long term persistence. Nuclear DNA (paternally inherited) analyses will help to further assess fine scale structure and genetic diversity. It is hoped that the next step will be for other African manatee biologists to continue to opportunistically collect manatee samples so that genetic analyses can be expand ed to include other countries, and thereby increase the amount of information for the countries included in this study. There is also a critical need to calculate manatee population estimates across Africa to discover where populations are most at risk from anthropogenic threats and habitat loss . Effective population sizes can be estimat ed using mole cular tools , but these studies will require additional samples from more individuals, since accuracy increases as the number of samples analyzed increases. One limitation of using mtDNA is that it is only a measure of maternal gene flow since there is no paternal contribution. Microsatellites and other analyse s that include nuclear DNA are a logical next step in the evolution of genetic studie s of this species to address fine scale population structure, such as the lagoons with unique haplotypes and to bet ter address current genetic diversity . Environmental DNA (eDNA) that can detect the presence of animal DNA from samples obtained from water bodies, has great potential for gaining information about elusive species distribution, such as that of the African manatee .
1 37 Table 2 1. New African manatee control region and cytochrome b partial haplotypes identified by this study , and collection locations. Name Identification Location Number of S ample s Control Region TS CR01 Senegal River 12 TS CR02 Senegal c oast 3 TS CR03 Gabon Mayumba and Senegal coast 2 TS CR04 Guinea 2 TS CR05 Ivory Coast 1 TS CR06 Mali Niger River 1 TS CR07 Cameroon Douala Edea 1 5 TS CR08 Cameroon Douala Edea 2 2 TS CR09 Gabon Ogooue River 1 1 TS CR10 Gabon Ogooue River 2 2 TS CR11 Gabon Ogooue River 3 1 TS CR12 Gabon N'gowe Lagoon 2 TS CR13 Gabon N'dogo Lagoon 2 2 TS CR14 Gabon N'dogo Lagoon 1 2 Cytochrome b TS CY TB 01 Senegal River 7 TS CY TB 02 Guinea 1 1 TS CY TB 03 Guinea 2 1 TS CY TB 04 Cameroon Douala Edea 1 3 TS CY TB 05 Cameroon Douala Edea 2 1 TS CY TB 06 Gabon Ogooue River 1 1 TS CY TB 07 Gabon Ogooue River 2 1 TS CY TB 08 Gabon Ogooue River and N'dogo Lagoon 6 TS CY TB 09 Gabon N'dogo Lagoon 2
138 Table 2 2. African manatee control region haplotype diversity indices and tests of neutrality, including both new and previously published haplotypes. Y01 and Y02 are included in the North region, and Y03 Y05 are included in the South region. Source n h HD S K Tajima's D test Fu's Fs test All sequences 55 19 0.909 0.017 25 7.172 1.008 1.412 By Region : North (Senegal, Guinea Bissau, Guinea) 22 6 0.684 0.008 16 3.251 0.942 1.370 South (Gabon, Cameroon, Ghana, Ivory Coast, C had, Niger, Mali) 33 14 0.885 0.007 15 2.996 0.626 4.914 * Selected Countries : Senegal 18 4 0.543 0.007 14 2.804 1.171 2.693 Gabon 19 8 0.813 0.004 6 1.567 0.282 3.250 * Cameroon 7 2 0.476 0.006 5 2.381 0.826 3.754 n , number of ind ividual samples; h nucleotide diversity; S , polymorphic sites; K, sequence diversity. *Significant P values (<0.0 2 )
139 Table 2 3. Control region sequence a nalysis of molecular variance (AMOVA) for African manatee populatio ns, among and between North and South regions, and within countries . Source of variation d.f. Sum of Squares Variance Components Percentage of variation P value Among regions (North vs. South) 1 115.29 3.86 66.00 0.000 Â± 0.00* Among countries wi thin regions 8 32.44 0.86 14.74 0.000 Â± 0.00* Within countries 45 50.71 1.13 19.26 0.009 Â± 0.001* TOTAL 54 198.44 5.85 d.f., degrees of freedom *Significant P values (<0.05)
140 Table 2 4. African manatee pairwise ST estimates using Tamura distance estimation (below the diagonal) and P values for exact tests of pairwise haplotype frequency comparisons (above the diagonal) for ten African manatee populations by country. Country Guinea Bissau Guinea Senegal Chad Â§ I vory Coast Mali Â§ Ghana Â§ Niger Cameroon Gabon Guinea Bissau 0 0.336 Â± 0.002 0.101 Â± 0.002 1.000 Â± 0.000 1.000 Â± 0.000 1.000 Â± 0.000 1.000 Â± 0.000 1.000 Â± 0.000 0.055 Â± 0.001 0.065 Â± 0.002 Guinea 0.500 0 0.010 Â± 0.001* 0.334 Â± 0.001 0.335 Â± 0.003 0.332 Â± 0.002 0.333 Â± 0.003 0.335 Â± 0.002 0.055 Â± 0.002 0.034 Â± 0.002* Senegal 0.065 0.530* 0 0.103 Â± 0.003 0.021 Â± 0.001* 0.100 Â± 0.003 0.110 Â± 0.004 0.022 Â± 0.002* 0.000 Â± 0.000* 0.000 Â± 0.000* Chad Â§ 0.642 1.000 0.707 0 1.000 Â± 0.000 1.000 Â± 0.000 1.000 Â± 0.00 0 1.000 Â± 0.000 0.125 Â± 0.002 0.211 Â± 0.006 Ivory Coast 0.726 0.892 0.776* 0.459 0 1.000 Â± 0.000 1.000 Â± 0.000 1.000 Â± 0.000 0.057 Â± 0.003 0.474 Â± 0.014 Mali Â§ 0.771 1.000 0.827 1.000 0.604 0 1.000 Â± 0.000 1.000 Â± 0.000 0.125 Â± 0.003 0.189 Â± 0.005 Ghan a Â§ 0.642 1.000 0.746 1.000 1.000 1.000 0 1.000 Â± 0.000 0.124 Â± 0.002 1.000 Â± 0.000 Niger 0.701 0.892 0.718 1.000 0.504 0.540 0.336 0 0.054 Â± 0.002 0.067 Â± 0.004 Cameroon 0.781 0.853* 0.774* 0.425 0.242 0.724 0.289 0.422 0 0.000 + 0.000* Gabon 0.843* 0.885* 0.806* 0.664 0.235 0.839* 0.656 0.670* 0.322* 0 Â§ Comparisons for these populations used small sample sizes. *S ignificant P values (P < 0.05) .
141 Table 2 5. Cytochrome b 1227 bp haplotype diversity indices for the nine new hap lotypes identified b y this study . Source n h HD S K Tajima's D All sequences 23 9 0.8419 0.0033 13 3.976 0.4517 By Region : North (Senegal, Guinea) 9 3 0.4167 0.0009 4 1.0556 1.1494 South (Gabon, Cameroon) 14 6 0.7912 0.0012 7 1.4176 1.3136 n , number of individual samples; h nucleotide diversity; S , polymorphic sites; K, sequence diversity Table 2 6. Cytochrome b ( for nine new hap lotypes) sequence a nalysis of molecular va riance (AMOVA) African manatee populations, among and between North and South regions, and within countries . Source of Variation d.f. Sum of Squares Variance Components Percentage of Variation P value Among regions (North vs. South) 1 30.50 2.36 70. 10 0.000 Â± 0.000* Among countries within regions 2 6.30 0.63 18.70 0.000 Â± 0.000* Within countries 19 7.16 0.38 11.21 0.329 Â± 0.005* TOTAL 22 43.90 3.36 d.f., degrees of freedom *Significant P values (<0.05)
142 Table 2 7. Cytochrome b (n = 23 haplotypes) pairwise ST estimates using Tamura distance estimation (below the diagonal) and P values for exact tests of pairwise haplotype frequency comparisons (above the diagonal) for samples from four African manatee countr ies . Country Senegal Guin ea Cameroon Gabon Senegal 0 0.028 Â± 0.002* 0.004 Â± 0.001* 0.004 Â± 0.001* Guinea 0.790* 0 0.195 Â± 0.006 0.110 Â± 0.005 Cameroon 0.924* 0.545 0 0.005 Â± 0.001* Gabon 0.940* 0.783* 0.535* 0 * S ignificant P values (P < 0.05) . Table 2 8. Concatenated cont rol region and cytochrome b pairwise ST estimates using distance estimation for samples from four African manatee countr ies . Country Gabon Cameroon Senegal Guinea Gabon 0 Cameroon 0.403* 0 Senegal 0.922* 0.916* 0 Guinea 0.855* 0.766 0.939* 0 * S ignificant P values (P < 0.05) .
143 Figure 2 1. African manatee control region haplotype maximum l ikelihood 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 , p arameter = 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 .
144 Figure 2 2. Map of control region m i t ochondrial DNA haplotypes identifie d in African manatee sample s. Fourteen n ew (solid colors) and five p reviously published haplotypes (patterns ; Vianna et al . 2006) from 55 sequences are indicated by pie charts. Circle size corresponds to the total number of samples per country and slices are proportional to haplotypes found (see inset table). A sterisks (*) indicate p reviously published haplotypes identified at four new locations. Base map courtesy of Ellen McIlhinny.
145 F igure 2 3. Map of control region haplotypes identified in Senegal manatee sample s . Pie chart size corresponds to the number of samples in each indicated location and slices are proportional to haplotypes found. Three new haplotypes (solid colors) and one previously published haplotype (pattern; Vianna et al . 2006) are shown for the Senegal River (TS CR01, n=12 ), Delta Saloum ( TS CR02 (n=3) and Y02 (n=2)), and the Casamance River (TS CR 03 , n=1 ) .
146 Figure 2 4 . Map of control region haplotypes identified i n Gabon manatee sample s . Pie chart size corresponds to the number of samples in each indicated location and slices are proportional to haplotypes found. Seven new haplotypes (solid colors) and one previously published haplotype (pattern; Vianna et al . 2006 ) are shown for the Ogooue River (TS CR09 (n=1), TS CR10 (n=2), TS CR11 CR13 (n=2) , TS CR14 (n=2) and Y03 (n=5) ), and Mayumba (n=1).
147 Figure 2 5. Un rooted n eighbor joining tree of African manatee control region haplotypes indicating a strong separation of North region haplotypes (lower right) and South Region (upper left).
148 Figure 2 6. Map of cytochrome b mitochondrial DNA haplotypes identifie d in African manatee sample s . Nine new (solid colors) and three previously published haplotypes (patterns; Vianna et al . 2006) from 26 sequences are indicated by pie charts. Circle size corresponds to the total number of samples per country and slices are proportional to haplotypes found (s ee inset table). Gabon Lagoon (n=4). Base map courtesy of Ellen McIlhinny .
149 Figure 2 7. Cytochrome b maximum l ikelihood GTR+I+G tree (+I, 4.1725% sites; 5 categories +G, parameter = 0. 0500) for three previously published African manatee haplotypes and nine new haplotypes trimmed to 605 base pair for comparison (log likelihood = 889.716). The tree is rooted in Trichechus manatus haplotype STm03 (GenBank accession AY965885 ) and is drawn to scale, with branch lengths measured in the number of substitutions per site.
150 Figure 2 8 . Cytochrome b un rooted m aximum l ikelihood GTR+I+G tree (+ I , 0.0010 % sites ; 5 categories + G , parameter = 200.0000 ) for nine new Trichechus senegalensis 1227 bas e pair haplotypes (log likelihood = 1754.188 ) . Bootstrap values (1000 replicates) are shown above nodes . The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.
151 Figure 2 9 . U n rooted m aximum l ikelihood GTR+I +G tree (+ I , 1.8433 % sites ; 5 categories + G , parameter = 0.0500 ) for 14 Trichechus senegalensis concatenated control region and cytochrome b 1637 base pair sequence s (log likelihood = 2525.451) . Bootstrap values (1000 replicates) are shown above nodes . Th e tree is drawn to scale, with branch lengths measured in the number of substitutions per site.
152 Figure 2 10. Un rooted neighbor joining tree for African manatee concatenated control region and cytochrome b 1637 base pair sequences . North region sequenc es are shown at top, South region sequences at the bottom . Bootstrap values (1000 replicates) are shown at nodes.
153 CHAPTER 3 PHYLOGENY OF THE AFRICAN MANATEE IN RELATION TO THE OTHER EXTANT TRICHECHID SIRENIANS Study Rationale and Objectives The study o f manatee population genetics began in earnest only 25 years ago and few populations have been analyzed outside of the West Indian manatee ( Trichechus manatus; et al. 1993, Garcia Rodriguez et al . 1998, Cantanhede et al . 2005, Vianna et al . 2006, Pause et al. 2007, Tringali et al. 2008a, Hunter et al. 2010a, Hunter et al. 2010b, Hunter et al . 2012, Luna et al . 201 2 ) . The African manatee ( Trichechus senegalensis ) is the least studied of the three extant manatee species du e to the difficulty of studying this extremely elusive aquatic mammal that primarily inhabits muddy and tannic waterways in highly remote locations . Additionally, there are no known fossil remains for the species, and very few for the other two extant tric hechids. The first genetic study of the African manatee was a doctoral dissertation by Parr in 2000 and the subsequent publication of that work (Parr and Duffield 2002). The study used 398 control region and 125 cytochrome b base pairs and identified three distinct clusters for the African manatee: 1: Guinea Bissau, 2: Chad, and 3: Ghana, Cameroon , and Gabon (Parr 2000). A study by Vianna et al . ( 2006) is the most recent and widely referenced paper examining the phylogeny of the three extant trichechid spec ies: West Indian, Amazonian ( Trichechus inunguis ) , and African manatee s . At the time of the Vianna et al . (2006) study, only six African manatee samples from four countries were available, and these resulted in five published control region mitochondrial D NA (mtDNA) haplotypes and three cytochrome b mtDNA haplotypes. Conflicting results within and between loci were presented for the evolutionary order of the three manatee species which were not resolved in the
154 publication . A control region neighbor joining (NJ) tree created using a Tamura Nei model in the software MEGA v. 3.0 (Kumar et al . 2004) , identified the African species as paraphyletic and most basal to the three trichechid species. The West Indian and Amazonian manatees formed a monophyletic clade. H owever, in a median joining network (MJN) , the Af rican species was shown as more closely related to the Amazonian manatee, which was basal, but with a dashed line indicating long branch lengths. The mtDNA control region is a quickly mutating locus and not well suited to delineate distant speciation events as a single locus. In contrast, analysis of cytochrome b haplotypes in Vianna et al. (2006) and the resulting Maximum Parsimony (MP) and NJ consensus trees indicated that the Amazonian manatee was basal t o the African and West Indian species, with the African manatee being more derived and most closely related to the West Indian manatee. In these analyses, the cytochrome b trees indicated that all three of the trichechids form a monophyletic clade (Vianna et al . 2006). In summary, both the African and Amazonian manatee s were identified as most basal and the clade was identified as both paraphyletic and monophyletic. These results are possibly due to the small African manatee sample sizes, analysis , and para meter selection. Dates of divergence in Vianna et al . (2006) were calculated using a linearized approach in MEGA and an evolutionary rate of 0.0066 / site / million years for cytochrome b . The rate was calculated from the average divergence between Dugong dugon and Trichechus spp. assuming common ancestry between those lineages at a minimum of 20 million years ago (Mya), based on a previous morphological character analysis of 36 sirenian species (Domning 1994). For the African manatee, Vianna et al .
155 (2006) obtained a divergence date of 308,900 years before present from a trichechid ancestor, which was more recent than dates obtained for T . manatus ( 621,000 years before present) and T . inunguis ( 371,000 years before present ) . These dates of divergence for the three extant trichechids appear inexplicably recent , given the high levels of genetic differentiation. A previous publication by Domning (2002) estimated the divergence of the Trichechidae from the Dugongidae at approximately 33 Mya. Further, Blair et al . (2013) calculated a much faster mutation rate of 0.248 / site / million years for dugongs in Australia using mtDNA control region sequences. The recent identification of 23 new African manatee mtDNA control region and cytochrome b haplotypes in 50 sample s (Chapter 2) provided an opportunity to better assess African manatee diversity in relation to previously published haplotypes for extant trichechid species. Maximum likelihood (ML) and Bayesian analyses were utilized in order to clarify the evolutionary placement of African manatees in relation to the other trichechids. Additionally, Bayesian Markov chain Monte Carlo (MCMC) analysis methodologies were conducted to calculate more robust estimates of divergence times of extant trichechids from ancestral spe cies, and in particular to elucidate the time of divergence of the African manatee from the other trichechids. Methods All currently published mtDNA control region 410 bp haplotype sequences for West Indian (n= 23), Amazonian (n= 34), and African (n= 5) ma natees and the dugong (n=56) (GenBank ;www.ncbi.nlm.nih.gov) were compared to the 14 new, unpublished African manatee control region haplotypes identified in this study. Sequences were aligned using GENEIOUS software v7.1 (Biomatters Ltd., Auckland, New Z ealand). Using the software jMODELTEST2 v 2.1.5, the data was modeled using maximum
156 parsimony (MP), distance, and maximum likelihood (ML) optimality criterion using 88 models and 11 substitution schemes to determine the best criterion for the dataset. M axi mum likelihood was also utilized in order to estimate rates of nucleotide substitutions and invariant sites (Darriba et al . 2012). M aximum l ikelihood heuristic searches were then performed utilizing MEGA6 software (Tamura et al . 2013) to identify the most likely phylogenetic trees. A bootstrap analysis of 1000 replicates with a heuristic search of two replicates of random addition sequence was performed. Published mtDNA cytochrome b 615 bp haplotypes for West Indian (n=4), Amazonian (n= 4), and African (n = 3) manatees (GenBank), and one published partial mtDNA cytochrome b 1140 bp haplotype for the dugong were aligned in GENEIOUS with nine unpublished African manatee sequences (Chapter 2), and trimmed to 605 bp. Published trichechid sequences (Vianna et al . 2006) were originally 615 bp, but we found discrepancies between all the published samples with all the new haplotypes at position 10 (published were G while all new were T in both T. senegalensis and T. manatus ) . Therefore, the first 10 bps we re removed in all sequences prior to analysis. Analyses for control region and cytochrome b haplotypes were performed separately. MRBAYES v 3.2.1 software (Ronquist et al . 2012) was utilized for Bayesian phylogenetic inference using Markov chain Monte Carlo (MCMC) m ethods to reconstruct evolutionary relationships among all sequences within new African control region and cytochrome b haplotypes. For control region, a Hasagawa Kishino Yano (HKY) model (Hasagawa et al . 1985) ran for 2,000,000 generations with a relative burnin of 25% (500,000 generations). For cytochrome b , a GTR+I+G model ran for 1,000,000 generations with a relative burnin of 25% (250,000 generations). Potential
157 scale reduction f actor (PSRF) was used as a convergence diagnostic; values approached 1.0 a s runs converged. Consensus trees were produced using FIGTREE software v 1.4.0 (Rambaut 2007). Dating analyses were performed using the BEAST 1.7.5 package (Bouckaert et al . 2014). BEAST uses a Bayesian MCMC phylogenetics framework for estimating divergen ce dates through molecular clock analyses, and constructs phylogenies within a coalescent framework using the best fit substitution models (Bouckaert et al . 2014). The BEAST file was generated using BEAUti v 1.7.5. For the cytochrome b data set, samples we re divided into two taxon sets: all trichechids ( T. inunguis , T. manatus and T. senegalensis ), and all trichechids plus the dugong. The sequence data were input into jMODELTEST2 and Akaike Information Criterion (AIC) was used to determine the optimal model of evolution. The HKY model was determined to be the most appropriate substitution model and was used for cytochrome b sequence analysis . The age of the the range o f divergence dates referenced by Domning (2002). A normal distribution around this calibration point was set with the 95% confidence interval between 35.9 and 29.5 Mya. A strict clock was used, based on the documented clocklike nucleotide evolution in mamm alian cytochrome b (Brown 1983, Avise 2004). A random starting tree with a Yule prior was used . Ten million generations of MCMC were analyzed and parameters were logged every 1000 generations. Convergence of the MCMC runs and sufficient effective sample si zes (ESS) were assessed using TRACER . Ages were estimated for all internal nodes . The BEAST output trees were then summarized using
158 TREE ANNOTATER 1.7.5 with a 10% burnin (1 00, 000 trees). The summary tree was visualized in FIGTREE 1.4. The identified tri chechid diversification date was then used to calibrate BEAST analyses for the more rapidly evolving mtDNA control region data, to obtain additional support for the more recent divergences within Trichechus . BEAST was run again using si milar parameters for trichechid control region haplotypes. HKY+I+G was used as the model of evolution as determined by AIC jMODELTEST2. A normal distribution around the calibration point was utilized , with 1.3 standard deviations on either side of the point. As with cytochrom e b analyses, a strict clock was used with a random starting tree and a Yule prior , and node ages were estimated. BEAST ran for 10 million generations and logged par ameters every 1000 generations. Tracer was used to assess adequate ESS, TREE ANNOTATER summ arize d trees with a 10% burnin (1 00, 000 trees), and FIGTREE visualize d the output. Results Control region phylogenetic analyses were conducted on West Indian , Amazonian, and African manatees, and the dugong sequences. An initial analysis was run using all 55 dugong haplotypes, but as expected , there was no difference in the phylogenetic structure of the trichechid portions of the tree as a result of dugong phylogeny, so a single dugong haplotype (h1, GenBank accession EU835761) was used as an outgroup to r oot all subsequent ML and Bayesian control region trees. Results of jMODELTEST2 indicated that the HKY+G model had the highest likelihood to estimate rates of nucleotide substitutions. The resulting control region phylogenetic tree had a likelihood score o f 1711.89 and an estimated value of Gamma shape parameter of 0.2061. A 50% majority rule bootstrap consensus tree is shown in Figure 3 1. African
159 manatees were the most basal species after the dugong outgroup, and were distinctly separated from Amazonian and West Indian manatees, which formed a monophyletic clade. West Indian manatee clades corresponded to previously defined clusters I, II and III (Vianna et al . 2006). The topology for the Amazonian manatee species remained the same in all control region trees, with strong supporting bootstrap values at main nodes separating them from the other species. The Bayesian consensus analysis for the same 77 control region haplotypes produced a similar consensus tree (Figure 3 2) to the ML tree in terms of the di vergence of the three trichechid species, but within the African manatee it indicate d further divergence of the North and South regions. Separation of North haplotypes from the South haplotypes (posterior probability (pp) = 0.834) was not significant ( P > 0.05). However clades T. manatus (pp = 0.990) and T. inunguis (pp = 0.978) were supported with significant values ( P < 0.05 ) . Harmonic means for the two runs were: chain 1: 2009.14 and chain 2: 2003.86 with a total harmonic mean of 2008.45. Summaries we re based on a total of 6002 samples from two runs. Summary statistics for partitions with frequencies of >0.10 in at least one run were: average deviation of split frequencies = 0.009006 and maximum standard deviation of split frequencies= 0.043355. Of the 6002 trees, a 50% credible set included 3001 trees and a 99% credible set included 5942 trees. Maximum PRSF for parameter values was 1.009. Evolutionary history for 21 cytochrome b h aplotypes from West Indian, Amazonian, and African manatees and the dugo ng were also inferred by using the ML method, but for this gene, jMODELTEST2 indicated the General Time Reversible model with Gamma distribution (+ G , parameter = 0.2616) and invariable sites (GTR+I+G, Ne i
160 and Kumar 2000). Initial trees for the heuristic se arch were obtained by applying the NJ method to a matrix of pairwise distances estimated using the maximum composite l ikelihood (MCL) approach. The tree with the highest log likelihood ( 1287.0519) was visualized in MEGA6 (Tamura et al. 2013) and is shown in Figure 3 3. In this analysis the Amazonian manatee is the most basal species after the dugong outgroup and the African manatee is the most derived. In this topology, the African manatee is most closely related to the West Indian manatee. Bayesian cytoc hrome b analysis for all tric hechids and the dugong outgroup resulted in harmonic means for the two runs: chain 1: 1045.69 and chain 2: 1041.58, with a total harmonic mean of 1045.01. Summaries were based on a total of 3002 samples from 2 runs. Summary statistics for partitions with frequencies of >0.10 in at least one run were: average deviation of split frequencies = 0.002779 and maximum standard deviation of split frequencies= 0.007537. A NJ consensus tree is shown in Figure 3 4. Of the 3002 trees, a 50% credible set included 1501 trees and a 99% credible set included 2972 trees. Maximum PRSF for parameter values was 1.008. The tree was rooted in D. dugong cytochrome b (GenBank Accession U07567). Identical topology to the cytochrome b ML tree (Figure 3 3) and displayed clear separation of clades in all three species which align geographically, including the African manatee regional North and South population delineations. Divergence dating analyses utilizing 19 cytochrome b haplotypes estimated the div ersification of the extant trichechids (Figure 3 5). The time to most recent common ancestor (TMRCA) was run for the three trichechid species and results estimated that T. inunguis diverged from T. manatus and T. senegalensis at 4.91 Mya (95% distribution
161 range = 2.79 7.49 Mya, pp = 1). Trichechus senegalensis diverged from T. manatus at 3.49 Mya (95% distribution range = 1.94 5.37 Mya, pp = 1). Additionally, the analysis indicated a separation of the T. senegalensis North and South regional clades at 1 .51 Mya (95% distribution range = 0.69 2.56 Mya, pp = 0.93). Further divergence was found in T. inunguis (at 2.48 Mya and again at 0.73 Mya) and T. manatus (at 2.74 Mya and again at 0.59 Mya). The cytochrome b evolutionary clock rate for dugongs and tric hechids calculated by BEAST was 0.0194 / site / million years, which is much faster than the rate of 0.0066 / site / million years calculated by Vianna et al . (2006). The obtained trichechid diversification date (4.91 Mya), was then utilized to calibrate BEAST analyses for 76 mtDNA control region haplotypes (Figure 3 6). This tree indicated the separation of T. inunguis from the other two trichechid species, as seen in the cytochrome b BEAST analysis. Subsequent TMRCA results for T. inunguis indicated rapi d radiation of haplotypes beginning at 1.33 Mya. At 4.55 Mya (pp = 1) the T. manatus cluster I diverged from the rest of T. manatus ( II and III ) and T. senegalensis , which is a very different separation than that which was indicated by cytochrome b . The T. manatus cluster I subsequently diverged further at 1.08 Mya (pp = 1). Then at 3.98 Mya (pp = 0.83) T. senegalensis diverged from T. manatus II and III (similar to the cytochrome b divergence at 3.49 Mya), and subsequently splits into North and South clade s at 2.03 Mya (pp = 1). In the North, the Senegal River population diverged from coastal populations at 1.04 Mya (pp = 0.99), while in the South the inland haplotypes (Niger, Chad, and Mali, as well as one Ivory Coast and one Cameroon haplotype) diverged f rom coastal haplotypes at 1.09 Mya (pp = 0.33). In all extant
162 trichechid species a rapid radiation of control region haplotypes began approximately 1.33 0.91 Mya. Discussion This study investigated the evolution of the extant trichechids using additiona l African manatee sequences and three different statistical methodologies. Maximum l ikelihood phylogenetic analyses provided a preliminary examination of data and was used to compare to previously published results. New methodologies examining trichechids included Bayesian analyses, which provided posterior probabilities, and coalescent Bayesian BEAST analyses, which utilized the most rigorous testing available to estimate evolutionary rates. This is the first time Bayesian analyses have been conducted fo r trichechid mtDNA, and this widely used methodology increases confidence and reinforces morphological evidence supporting the evolutionary placement of these species. All analyses for both control region and cytochrome b loci resulted in consensus trees t hat exhibited short branch lengths, which indicate rapid evolution ( radiation) of the three extant t richechids over approximately five million years. The ML phylogenies for both loci for T. manatus and T. inunguis are the same as previously published in Ca ntanhede et al . (2005) and Vianna et al . (2006), but the inclusion of additional African manatee haplotypes with the same results revalidate s the topology of the Trichechidae family tree . In the case of the African manatee, control region Bayesian analysis posterior probabilities and ML bootstrap evolutionary histories place it as the most basal species to the trichechid group. To the contrary, cytochrome b Bayesian analyses suggest that the Amazonian manatee is basal and the African and West Indian species likely descended from a common ancestor after the Amazonian manatee diverged. Posterior
163 probabilities and ML bootstraps, as well as both BEAST analyses for both loci strongly support the African manatee as being most closely related to the Antillean subs pecies of the West Indian manatee found in the northern and Atlantic coast regions of South America ( T. manatus clusters II and III ). This result agrees with previous morphology studies which have also reported that the closest relative of the African mana tee is the West Indian manatee ( Domning 1982, Domning and Hayek 1986). Although conflicting, the results of the cytochrome b locus are more likely to be accurate. The control region, a non coding gene, is one of the most variable portions of t he mammalian mtDNA genome and is commonly variable at the intraspecific level, whereas cytochrome b , a coding gene, typically has only moderate levels of intraspecific variation. These attributes make the control region more useful for genetic variability studies amon g populations, while the more conserved cytochrome b gene is more useful for examining deeper phylogenies among species. The markedly high control region diversity may affect the evolutionary order. Additionally, control region ML tree bootstrap values and Bayesian posterior probabilities were not as strong as those for cytochrome b . The control region divergence dating analysis for African manatees indicated separations within North and South regional populations at 1.04 and 1.09 Mya respectively. The Nor th separation of the Senegal River from the coastal populations had a very strong pp (0.99), while the South separation of the inland from the coastal population was not significant (0.33). The inland population did remain separate and distinct in all phyl ogenetic analyses, and may currently be starting to diverge.
164 The cytochrome b evolutionary clock rate for dugongs and trichechids calculated by BEAST was 0.0194 / site / million years, which is much faster than the rate of 0.0066 / site / million years c alculated by Vianna et al . (2006). This increased rate may have resulted from the longer 1227 bp cytochrome b sequences used in this study than the 615 bp sequences which were available to Vianna et al . (2006). However, the main difference in evolutionary clock rate is likely due to use of the more rigorous BEAST analysis in this study, which provided a more accurate estimate. The divergence date trees for cytochrome b and control region haplotypes are in agreement for divergence estimates of individual sp ecies, with the exception of T. manatus cluster I diverging at 4.55 Mya from the rest of the species haplot ypes on the control region tree. This did not result in the cytochrome b analysis. This may be due to the greatly reduced number of available haploty pes for the cytochrome b analysis (only four vs. 23 for control region) that also did not include all clusters (there are currently no cytochrome b haplotypes for T. manatus cluster III ). In the control region dating analysis, the divergence of T. manatus cluster I at 4.55 Mya might be explained by trichechid species that were already living in the Caribbean and Florida during this time period (Domning 2005). During this period manatees would have been geographically separated from T. manatus in the norther n portion of South America. For T. senegalensis , both loci indicated speciation between 3.5 4 Mya. Domning (2005) reported that African manatees probably diverged from T. manatus through dispersal from South America across the Atlantic Ocean during the P liocene (5 1.8 Mya) or Pleistocene (1.8 0.01 Mya). The TMRCA values can vary considerably between simulations and therefore should be interpreted cautiously, but we believe that
165 the additional new African manatee haplotype sequences, the analyses of bo th cytochrome b and control region for all extant trichechids , and the rigorous divergence dating methodology used in this study have increased the accuracy of the divergence time estimates. Divergence date estimates between 3.5 4 .0 Mya for African manat ees imply rapid speciation from the ancestral species during the early Pliocene Epoch. It is of interest to understand the environmental conditions that occurred in the tropical Atlantic Ocean during that time, which may have produced favorable conditions for ancestral African manatees to cross the Atlantic. Approximately five to four million years ago in the early Pliocene, land masses were in approximately the same positions as today, and the closure of the Central American Seaway by the formation of the Isthmus of Panama (by the mid Pliocene) caused large scale oceanic changes around the globe (Domning 2001b, Federov et al . 2006 ). Domning (2001b) theorized that the closure of the Central American Seaway led to the die off of larger seagrass species that C aribbean dugongids depended upon, leading to their disappearance in the region, and concurrently filled by trichechids moving into the Caribbean from South America. The reorganization of ocean systems caused a permanent El NiÃ±o effec t that created a much warmer Pacific Ocean, and the closure of the Isthmus of Panama increased movement of warm Atlantic water towards the poles, so that all the o C warmer than today (Federov et al . 2006, Filippelli and Flores 2009 ). On the West African coast, ocean temperatures were as much as 10 o C higher during the Pliocene and cold surface waters in this reg ion did not exist until 3 Mya, when the climate cooled at the end of the Pliocene (Federov et al .
166 2006, Federov et al . 2013). Sea level was 25 m higher than today due to reduced ice sheets at the poles, and precipitation levels were much higher, particularly across the tropical Atlantic Ocean from the Caribbean to West Africa (5 9 mm/day in the Pliocene vs. 2 7 mm/d ay at present; Federov et al . 2006). Due to the warmer oceans, there were significant decreases in coastal upwelling, wind driven systems, and major ocean circulatio n (Sar n thein et al . 1982, Filippelli and Flores 2009). Overall the Pliocene was characterized by climatic stability, high sea level, calm oceanic and prevailing zonal wind circulation, and continental wetness ( Sarnthein et al . 1982, Federov et al . 2013). P revailing currents in the Atlantic today, which primarily circulate east to west, were largely dampened, but the North Equatorial Countercurrent (NECC) , which moves from west to east in the central Atlantic, strengthens during warmer time periods (Billups et al . 1999, HÃ¶ll et al. 2000, Bischof et al. 2004) and therefore may have been stronger during the Pliocene . Vianna et al. (2006) reported that analysis of T. manatus mtDNA control region samples showed the probable center of origin of the species to be i n Colombia, based upon high haplotype diversity. From that region it is believed T. manatus expanded to the east and west along the coastal Caribbean and South America (Vianna et al. 2006). The results of the current study indicate that the African manatee is most closely related to West Indian manatee mtDNA hapl otypes that originated from Colo mbia and the Atlantic coast of South America and crossed the Atlantic. The NECC originates in the Atlantic Ocean near French Guyana and Brazil. This current lies bet ween 3 10 o N latitude within the Intertropical Convergence Zone (ITCZ),
167 which is strongly influenced by seasonal west to east trade winds ( HÃ¶ll et al. 2000, Bischof et al. 2004). The NECC sweeps eastward across the Atlantic at its narrowest point, and is s trongest in the warm summer months when it shifts northward, reaching the African continent near Gambia and Guinea Bissau. In the winter the NECC decreases and shifts further south, reaching the African continent near Liberia (Richardson and Reverdin 1987, Bischof et al. 2004). Other recent work has determined that there is a second north core of the NECC that extends as far as 15 o N latitude which reaches Africa near Dakar, Senegal (Urbano et al. 2006). The NECC has also been found to weaken or disappear i n the winter months (Bischof et al. 2004). This current is believed to have existed in the Pliocene (Billups et al . 1999), and could have been a possible method of dispersal for the African manatee. It is plausible that as manatees moved coastally along So uth America during this warm period, they could have been caught up in this or a similar current. After 3 Mya, when the climate became cooler with the onset of a glacial interval, the NECC weakened or dissipated (Billups et al . 1999). Collection and analy ses of additional African manatee mtDNA could identify the likely locations where the species first colonized the continent. Additionally, in order to confirm or refute the regional separation of African manatee haplotypes between the North and South regio collected and analyzed. Of particular importance will be samples from countries that between the regions (Gu inea, Ivory Coast, and Ghana).
168 Few analyses for West Indian and Amazonian manatees have utilized cytochrome b , likely due to low levels of diversity identified using previous techniques that resulted in 615 bp sequences. Additional African manatee samples with nine new haplotypes and longer sequences , have resulted in new evolutionary inferences and divergence times for the three trichechid species. These techniques have made it possible for further investigations of cytochrome b in the other trichechids wh ich may be useful in identifying additional mtDNA diversity in those species.
169 Figure 3 1. HKY+G (+ G , parameter = 0.2061) maximum l ikelihood tree for control region 410 bp haplotypes for Trichechus senegalensis , T. manatus clusters I,II,III , and T. inu nguis . The tree is rooted in D. dugong h1 haplotype. B ranch lengths are measured in the number of substitutions per site.
170 Figure 3 2. Bayesian consensus tree of 76 t richechid control region haplotypes created with MRBAYES software v 3.2.1 and rooted in the D. dugong h1 haplotype. Posterior probabilities are shown at nodes. Scale bar represents branch length equal to the amount of genetic change indicated by the number.
171 Figure 3 3 . Cytochrome b 605 bp haplotype maximum l ikelihood GTR+I+G phyloge netic tree for Trichechus senegalensis , T. manatus , and T. inunguis rooted in the dugong. Gamma distribution (+ G , 5 categories, parameter = 0.9974) and invariable sites (+ I , 0.0000% sites). The tree is drawn to scale, with branch lengths measured in the nu mber of substitutions per site, and bootstrap values are shown above nodes.
172 Figure 3 4. Bayesian analysis of cytochrome b haplotypes using MRBAYES and visualized in a NJ tree rooted in the dugong. The T. inunguis (Ti branches) are the most basal after the dugong outgroup and separated from the T. manatus (Tm branches) and the T. senegalensis (TS branches, lower right clusters). Posterior probabilities are shown at nodes.
173 Figure 3 5. BEAST cytochrome b output tree of s irenian dates of divergence (i ncluding D. dugong ). Divergence date estimates, based on the separation of dugongs from trichechids at 32.67 Mya (Domning 2001a) , are shown at nodes with purple bars indicating 95% distribution of dates. Time scale (millions of years before present) is sho wn below the tree.
174 Figure 3 6. Un rooted BEAST control region output tree of t richechid divergence. Species and clades within species for Trichechus senegalensis (North, South) and T. manatus ( I, II, III ) are indicated to the right of the tree. Dive rgence date estimates ( at nodes ) are derived from the estimated 32.7 Mya split of dugongs and trichechids (Domning 2001 a ).P osterior probabilities are in parentheses, and t ime scale (millions of years before present) is shown below the tree.
175 CHAPTER 4 CARBON AND NITROGEN STABLE ISOTOPE ANALYSIS TO DETERMINE THE DIET OF THE AFRICAN MANATEE Study Rationale and Objectives The African manatee is a highly elusive species which inhabits numerous remote waterways in 21 countries. In Africa, manatees live in lagoons within equatorial rainforests, in rivers at the edge of the Sahara Desert, around coastal islands in the Atlantic Ocean, and in many other habitats in between. Plant species and African manatee feeding strategies vary widely between habitats (Powe ll 1996, Keith Diagne unpublished data). Florida manatees have been reported to consume over 60 species of plants (Ames et al . 1996), and 70 species of plants have been documented for African manatees to date (Appendix A; Villiers and Bessac 1948, Husar 19 78, Reeves et al . 1988, Powell 1996, Akoi 2004, Ogogo et al . 2013, Keith Diagne this study). There is also strong anecdotal evidence in many African countries that African manatees eat fish from nets, and that they eat freshwater and estuarine mollusks in addition to plants (Powell 1996, Keith Diagne unpublished data). There are also several reports of mollusks seen in the stomachs of dead manatees (Reeves et al . 1988, Powell 1996). Stable carbon and nitrogen isotope analysis can help verify or refute this, and if accurate it would be evidence of a significant dietary difference from the other extant manatee species. African manatees have only rarely been observed feeding, and opportunities to collect stomach samples from fresh carcasses are almost non exist ent, therefore stable isotope analysis offers the best option for documenting the diet of this species. Additionally, becau 13 15 N values or data between the habitats where African manatees live (freshwater, marine and estuarine), stable isotope analyses show great potential for inferring both habitat use and movement patterns for
176 the species. These data can identify diversities in diets between different populations and habitat types. Stable isotopes are an exceptionally useful tool in animal studies, particularly for wildlife that is difficult to observe and/or may travel over large distances (Hobson an d Wassenaar 2008). Stable carbon isotope ratios in animal tissues ( 13 C/ 12 C) are used to identify the carbon in food sources (Peterson and Fry 1987, Hobson 1999). For example, carbon stable isotope ratios can indicate differences between C 3 and C 4 plants, a nd differentiate between terrestrial, freshwater, and marine ecosystems (Fry 2006). Differences in the stable isotope ratios of nitrogen ( 15 N/ 14 N) in animal tissues indicate the trophic level of an animal (Peterson and Fry 1987, Hobson 1999). Stable isotop as parts per mil ( 0 / 00 ) relative difference to the standard for the element of interest (Hobson and Wassenaar 2008). The delta value is determined by the standard delta equ ation: 0 / 00 ) = ( R sample / R standard 1) x 1000 (4 1) where R sample and R standard are the absolute isotope ratios of the sample and standard respectively, for example 13 C/ 12 C or 15 N/ 14 N. Bone has two components that can be used in stable isotope an alyses. Bioapatite is the mineralized (inorganic) component of mammalian bone, and collagen is composed of proteins (organic portion) that are the main constituent of animal connective tissues (Clementz et al . 2007, Nelson and Cox 2008). Bioapatite and col lagen incorporate carbon from different components of diet: carbon from the protein part of the diet is more strongly labeled in collagen, whereas carbon in blood (produced
177 through respiration and bulk dietary carbon, including carbohydrates, lipids, and p roteins) is more strongly conserved in bioapatite (Clementz et al . 2009a). Collagen reflects the growth aspect of diet, and bioapatite reflects the energy aspect (Lee Thorp et al . 1989). Bone collagen has been shown to have a slow turnover rate of ~7 years for terrestrial mammals, and is also considered to reflect a running average lifetime diet because although bone is remodeled, some of the original bone remains throughout the life of the organism (Hobson and Clark 1992, Jim 2000, Clementz et al . 2007). T urnover rates for bone bioapatite from terrestrial mammals have been estimated at ~4 years, but turnover rates specifically for manatee bone (apatite or collagen) have yet to be determined (Clementz et al . 2007). An enrichment range of 2 6 0 / 00 13 C for bo ne collagen compared to diet has been documented for animals overall, and ungulate 13 C enrichment rate of 5 0 / 00 relative to diet (Peterson and Fry 1987, Lee Thorp et al . 1989). Bioapatite is generally en 13 C by 12 14 0 / 00 relative to diet for herbivore ungulates (Passey et al. 2005). Using bone samples from D. dugong , T. manatus , and the extinct H. gigas , Clementz et al 13 C apatite collagen values were statistically distin ct between the different sirenian species studied. 15 N per trophic level for bone collagen (Schoeninger and DeNiro 1984, Bocherens and Drucker 2003). Schoeninger and DeNiro (1984) studied 15 N in bone collagen for 66 species of 15 N enrichment in the mean values between herbivorous terrestrial animals (5.3 marine animals with a fish di et (16.5 ) and marine
178 animals with an invertebrate diet (13.3 ). 15 N trophic shifts are much more complex to calculate, since nutritional stress (starvation or fasting), water conservation, and digestion can influence values (Sponheimer et al . 20 03). Manatees are hindgut fermenters, and most studies of mammalian hindgut fermenters place them at a 15 N values due to the location of the microflora in their digestive tracts (Sponheimer et al . 2003). How ever, manatees have been shown to be atypical of what is expected of a hindgut fermenter, due to both specialized features of their digestive tracts, as well as their low fiber diets of aquatic plants which slows the passage of food through their digestive tracts (Lanyon and Marsh 1995, Clementz et al . 2007, Larkin et al . 2007). Manatees exhibit highly efficient digestion, particularly of cellulose (64 97%) (Clementz et al 15 N values can shift with environmental factors including salinity, annual rainfall and soil acidity, and these values have been shown to increase specifically in bone collagen, making this tissue inappropriate to use to estimate to tal diet (Bocherens et al. 2005). Both the inorganic and organic components of bone are remodeled during growth et al . 2007). Bone remodeling is accomplished through the removal of mineralized bone by osteoclasts, followed by the formation of bone matrix through the osteoblasts that subsequently become mineralized ( Hadjidakis and Androulakis 2006) . Hadjidakis and Androulakis (2006) describe t he bone remodeling cycle as consisting of three consecutive phases: resorp tion, during which osteoclasts digest old bone; reversal, when mononuclear cells appear on the bone surface; and formation, when osteoblasts lay down new bone until the resorbed bone is completely replaced.
179 For manatees, the periosteal (ear) bone has been used to determine the age of individuals using a technique developed by Marmontel et al . (1996) in which the periotic dome of the tympano periotic complex is sliced, and annual growth layer groups (GLG) are read to estimate age. Figure 4 1 illustrates the distinct layers of the manatee ear bone, and the location of the annual GLGs that were used to determine age (Marmontel et al . 1996). In a study of Florida manatees using 1213 periotic bones, resorption was shown to occur in the growth layer groups after 15 years of age, and increased with age et al . 1996). Although the bone layers beneath the GLGs have not yet been studied in manatees, since all bone is remodeled hese lower layers are also remodeled. Marmontel et al . (1996) also analyzed ear bones from Antillean manatees (n=18) and found results in the GLGs similar to Florida manatees after 20 years of age. The Marmontel study (1996) classified levels of ear bone G LG resorption into categories (none, light, moderate, or heavy) based upon the number and extent of osteons (also known as Haversian systems) and secondary osteons that occurred within the bone due to remodeling as manatees age. These features disrupt or o bliterate annual adhesion lines used to age manatees, but some of the original bone persists, allowing researchers to read annual lines (Marmontel et al . 1996). Thus, as the bone is remodeled, new tissue forms within it, in the form of osteons and secondar y osteons, but some of the original bone persists, and stable isotope sampling will include both old Bone is also easily preserved and can tell us about the diet and life histories of a nimals from the past, since very old and fossilized bones can also be used for stable
180 isotope analyses (Peterson and Fry 1987). These data can identify diets, as well as diversities in diets between different populations and habitat types. In Africa, manat ees live in many diverse habitats and are believed to feed on mollusks and fish as well as plants. However, this evidence of a significant dietary difference from the other manatee species has not yet been quantified in the scientific literature. Feeding s trategies may also vary widely among habitats, which can influence the baseline carbon source, and corresponding primary and secondary consumer carbon isotope values (Peterson and Fry 1987). West Indian manatees ( Trichechus manatus ) have also been reported eating fish and mollusks, but this has been though t to be a rare occurrence (Powell 1978, Hartman 1979, Corbis and Worthy 2003). If African manatees are feeding on mollusks 15 N bone collagen values that are enriched by more 15 N values of strictly herbivorous manatees (Schoeninger and DeNiro 1984, Bocherens and Drucker 2003). Animals with lower protein diets are more likely to exhibit greater overall diet tissue discrimination for 15 N than those species with higher protein die ts (Hobson and Wassenaar 2008). Nutritionally, high nitrogen protein sources are beneficial, but if African manatees regularly feed on other animals as part of their diet, then they are also dependent upon the habitats and environmental parameters that can support these other animal species, which may be more sensitive to environmental changes than plants and manatees are. Stable isotope research is a non invasive way to collect valuable manatee foraging data that can greatly benefit managers in knowing wha t types of habitat to protect. In addition to the tissues themselves, the time of collection of samples is
181 lifetime diet and the ecosystem(s) in which it lives, but mus eum specimens can give Wassenaar 2008). Some ear bone samples collected for this study were taken from a museum collection in Senegal and were primarily collected in the of African manatee diet from samples collected in the past seven years with samples collected 70 or more years ago in the same locations gives a historical perspective which could have implications for conservation and management of t he species today. For manatees living on the coast of Senegal, there have been changes including decreased mangrove habitat, increased coastal erosion, and sea level rise, which are attributed to both an increasing human population in the region and global climate change (Dia Ibrahima 2012). In the Senegal River, drastic changes to manatee habitat have occurred in the past 70 years, including the construction of two major dams (the Diama Dam, Senegal completed in 1984 and the Manatali Dam, Mali completed in 1988) which now control all water flow in the river, numerous smaller dams (most notably the Taouey and Keur Momar Sarr dams which control all water flow into and out of Lac de Guiers) and increased agriculture. These changes may have altered the plant, m ollusk and fish species that occur in this region, and increased or reduced their abundance and availability to manatees. We would expect to detect these differences in stable isotope values. In the eastern Senegal River, annual flooding extents have decre ased since the completion of the Manatali Dam (Sandholt et al . 2003), which may reduce the amount of time manatees have to forage on seasonal floodplains each year and increase the amount of time they may need to rely on dry season forage. Additionally, as agricultural industries have increased in the past 70 years along these
182 waterways, they may now generate more pollution (fertilizers, pesticides, animal waste), and as a result we would expect to see an increase 15 N stable isotope signal due to increased nitrogen in the environment (Carpenter et al . 1998, Fry 2006). Increased nitrogen (and other nutrients such as phosphorous) in aquatic systems can cause problems such as toxic algal blooms, loss of oxygen, fish kills, loss of aquatic plant beds, and loss of biodiversity (Carpenter et al . 1998). Very little stable isotope research has been conducted for sirenians, and none for the African manatee previous to the current study (Ames et al . 1996, Walker and Macko 1 999, Clementz 2002, Reich and Worthy 2006, Clementz et al . 2007, Alves Stanley and Worthy 2009, Alves Stanley et al . 2010). African manatee ear bones have recently been sampled for age determination and stable carbon and nitrogen isotope analysis for the f irst time, providing an opportunity to also utilize them to determine lifetime average species diet using this tissue. This is the first study to use ear bones of any sirenian species for stable isotope analyses. We focused on manatee samples from Gabon an d Senegal to examine average lifetime diets. The objectives of this research were to examine whether isotopic differences were present in African manatee lifetime average diet based upon three diverse ecosystems at opposite ends of the species range: Gabon freshwater Senegal River. We also investigated whether or not the sampled African manatees consumed mollusks and fish in addition to plants. Additionally, carbon and nitrogen stab le isotope values from manatee ear bones collected >70 years ago in Senegal were compared to present day samples to investigate possible differences in diet due to changes in the habitat and environment over time.
183 Methods Sample C ollection Reference food s ource samples collected in Senegal included 20 plant species, three fish species and four mollusk species from four villages spread out over approximately 172 km (107 miles) on the Senegal River in eastern Senegal, and from Tocc Tocc Wildlife Reserve at La c de Guiers in northeastern Senegal. Lac de Guiers is adjacent to, and fed by, the Senegal River. One additional mollusk species and one seagrass species were collected from the Atlantic coast of Senegal at Delta Saloum. From Gabon, seven aquatic and shore Lagoon, seven shoreline plant species were collected from the Ogooue lakes (Ezanga and Oguemou Ã© ) in central Gabon, and one seagrass species was collected from Corisco Bay on the Atlantic coast in northern Gabon . One hermit crab species was also collected from Banio Lagoon in southern Gabon. Three plant species ( Crinum natans , Jussiaea repens , and Nymphea lotus ) were collected in both Gabon and Senegal. For each reference species at each study site, one to 18 ind ividual samples (whole animals/plants or plant parts including roots, leaves, stems, seeds, and or flowers) were collected, for a total of 239 reference samples from 43 species. Twenty four manatee ear bones were opportunistically collected from Senegal ( n= 16) and Gabon (n= 8). Individuals ranged in age from less than one year to 39 years old (Keith Diagne and Brill, unpublished data). In Gabon, manatee ear bones were collected from carcasses between 2005 (n= 1), a nd Lake Onangue (n= 1) in central Gabon. For the sixteen ear bones collected in Senegal, six came from carcasses recovered in two locations between 2009 2013: five from the Senegal River in eastern Senegal, and one from Delta Saloum on the
184 Atlantic coast i n central Senegal. Ten additional ear bones were sampled from specimens at the (IFAN) Museum in Dakar, (n=1) and central coastal Senegal (n= 6). The remaining three museum specimens were not labeled with their collection location within Senegal, so their dates of collection are known, but not specifically where they came from. Samples of all plants and two fish species from Senegal were ini tially dried in the field (using sunlight, oven drying and/or packaging with silica) prior to shipping to the USA. Plants were pressed using a standard biological plant press. Freshwater mussels and one fish species ( Alestes macrolepidotus ) from the Senega l River were initially preserved in gin (40% alcohol) in the field, and were transferred to 70% ethyl alcohol once they reached the laboratory, approximately two months later. Barrow et al . (2008) tested methods of preservation for sea turtle samples and d 13 C and 15 N values for tissues preserved in 70% ethanol were not significantly different from those of tissues dried at 60 o C. Therefore we expect the alcohol preservative in the gin 13 15 N values of the samples. However, we were unable to test for an effect of preserving mussels and fish in gin. Plant samples were imported under USDA permits to Import Prohibited Plant Material for Research Purposes issued to the PI. All African manatee samples were dr ied prior to shipping and were imported to the USA under U.S. Fish and Wildlife Service and CITES import and export permits issued to the PI and/or Dr. Robert Bonde. All mollusk and fish samples were declared to U.S. Fish and Wildlife Service agents upon t heir arrival in the USA, but
185 none were protected species and all were dead and preserved, therefore no additional permits were necessary. Laboratory Analysis Processing of all food source reference samples was conducted at the U.S. Geological Survey, Sout heast Ecological Science Center, Benthic Ecology Laboratory in Gainesville, Florida, USA. Once samples reached the laboratory, they were dried for 24 hours at 60 o C in a drying oven. They were then chopped, crushed into powder, placed in glass vials, and dr ied again for 24 hours at 60 o C. The powder was measured and sealed into single boated tin capsules. Plant samples were measured to between 1.0 1.5 mg per capsule, and animal samples were measured to between 0.400 0.425 mg per capsule. Samples were then placed in numbered trays and sent for combustion. Hermit crab samples were acidified with platinum chloride prior to combustion (Demopoulos et al . 2010). Every tenth sample was a standard (apple leaves for plants and bovine liver for animals) and every fi fth sample was duplicated and analyzed adjacent to each other in order to verify both the sample results and the accuracy of the mass spectrometer. Analyses were performed using a Costech elemental analyzer (Valencia, USA) interfaced with a GV Instruments (Manchester, UK) Isoprime isotope ratio mass spectrometer (Demopoulos et al . 2010). All reference plant, mollusk, and fish samples were analyzed in the laboratory of Dr. Ray Lee at Washington State University. Bone sampling was conducted by Dr. John Krig baum (University of Florida, Department of Anthropology) who runs their Bone Chemistry Lab and uses bone stable isotope analysis to study human and animal ecology, and who developed the technique for this study in collaboration with the PI. Since this was the first stable isotope study of manatee ear bones, a pilot study examined five manatee ear bones (three from Senegal
186 and two from Gabon, ranging in age from 1 34 years) in order to investigate and verify the accuracy of the sampling techniques. All perio tic bones used in this study had previously been sliced in half for age determination (techniques and further results for age analysis will be presented elsewhere). For stable isotope analysis, bone powder was extracted from three different regions of each periotic bone. Bone powder was micro comprising the primary vascular bone), a second central layer from the middle of the r bone and compact regions of the ear bones are shown in Figure 4 2. Bone Sample M echanical Preparation Periotic manatee ear bones were sectioned in half (sagittal) and one half section was retained with the collection and the other was used for age determination and made available for isotopic analysis. Two rounds of sampling and anal ysis were conducted: Round I (n=5) in Fall 2013; Round II (n=19) in Spring 2014. To facilitate sampling for ering blade and custom built chucks to secure the sample. The lateral section was reserved while the C), which in turn, were cored for bone collagen and bone apatite iso tope analysis. For Round I, layers A C were sampled using a dental drill (Brassler USA, Savannah, GA) for all samples. One sample which was known to be from a manatee calf <2 years old (MSNU1218) was sampled only for layers B C. Drilled holes delimited
187 the perimeter cored for growth layers A and B. The outer perimeter layer C was abraded with a Dremel tool and a tapered carbide bit, and then the dental drill was used to delimit the sample removed. A total of 14 bone powder samples were collected for Round I (n= 4 A layers, n= 5 B layers, and n =5 C layers). For Round II, it was determined that layer A would not be sampled because of similarity observed in Round I results between A and B layers, and due to cost. For Round II, a different approach to sampling layers was developed. Instead of defining the perimeter of the sample, we instead used a Sherline 5100 vertical milling machine ( Vista, California) outfitted with a ended cutter bit. This improved the spatial resolution of each sample collecte d. Layer B was actually cored 3 4 times and sample powder collected. The outer s ample powder (~ 200 mg) was produced. The revised technique allowed for greater accuracy in sampling layers. The remaining 19 ear bones (14 from Senegal and 5 from Gabon, ranging in age of the manatee from less than one year to 39 years) were subsequently sampled using this method. A total of 38 bone powder samples were collected for Round II (n= 19 B layers, and n = 19 C layers). Once sample bone powder was obtained, each sample was sieved, the smallest sample fraction (<0.25 mm) was reserved for bone apat ite isotopic analysis, and the larger sample fraction (0.25 0.5 mm) was reserved for bone collagen analysis. It should be noted that layers A and B were dense bone and were easily powdered during sampling, while layer C (the outermost layer) was more cryst alized and keratin like, which did not affect collagen yields.
188 Bone Sample Chemical Preparation Bone collagen preparation procedures followed a modified Longin protocol (Longin 1971) which involved demineralization of bone using hydrochloric acid (HCl) and gelatinizing the solution at 90Â° C. For Round I, 200 mg of the bone collagen fraction was demineralized in a 15 ml centrifuge tube with 12 ml of 0.5 M HCl acid which was refreshed 2 3 times, and demineralization was complete after 48 72 hours. The same pr ocedure was used for Round II samples, however, 1 M HCl was used and demineralization was complete after 24 36 hours, with only one HCl acid refresh. Each sample was then rinsed to neutral pH (3 4 times) using deionized 2x distilled H 2 O (DI H 2 O). Once sa mples were at normal pH, 12 ml of 0.125 M NaOH was added to each sample for 16 hours. A sodium hydroxide (NaOH) soak has been suggested to remove humic acids, often followed by various filtration techniques to remove acid insoluble components in solution ( Ambrose 1990). Afterwards each sample was rinsed again to neutral pH using DI H 2 O, 12 ml of 10 3 M HCl was added, and the entire solution was transferred to pre weighed 20 ml glass scintillation vial and placed in a 95Â° C oven for 4 5 hours. To acidify ea ch sample and completely dissolve the collagen, 30 40 Âµl of 1 M HCl was added and the samples were returned to the 95Â° C oven for another 4 5 hours. Each sample was then poured back into its respective 0 rpm for 20 minutes. The vial was placed in a 65Â° C oven until the volume was reduced to approximately 2 ml. Samples were then frozen for 24 hours and lyophilized fo r 48 72 hours, and then loaded for isotope ratio mass spectrometry (IRMS).
189 In order to extract bone apatite, a modified Krueger protocol was used (Lee Thorp et al . 1989), which involved an oxidation step to remove exogenous organics (collagen and non col lagenous proteins) followed by pretreatment with dilute acetic acid (CH 3 COOH). For both Round I and Round II, 50 mg of the bone apatite fraction was placed in a 15 ml centrifuge tube and 12 ml 50:50 bleach (2.5% sodium hypochlorite, NaOHCl) was added and a llowed to react for 16 hours. Samples were then rinsed 3 4 times to neutral pH with DI H 2 O. After each sample was neutralized, 12 ml of 0.2 M acetic acid was added and allowed to react for another 16 hours. Following this, each sample was then rinsed to ne utral pH 3 4 times, frozen for 24 hours, lyophilized for 48 72 hours, and then loaded for IRMS. Stable isotopes were measured using a continuous flow isotope ratio mass spectrometer (CF IRMS) which converts samples to pure CO 2 , H 2 O or N 2 gases (Hobson and Wassenaar 2008). Powdered samples were combusted to convert them to their ultrapure gaseous components and their isotopic ratios were measured in relation to a reference gas sample of the same type (Fry et al. 1992). Bone apatite samples were loaded into stainless steel boats and placed into a Kiel device coupled with a Finnigan MAT 252 IRMS. All isotope results were reported in standard delta notation relative to Vienna Pee Dee Belemnite (VPDB). The standard used was NBS 19. For bone collagen samples, a T hermo Scientific Delta V IRMS with a ConFlo IV interface linked to a Carlo Erba NA1500 Costech ECS elemental analyzer (to determine atomic C:N was used. All carbon isotopic results were expressed in standard delta notation relative to VPDB, and all nitroge n isotopic results were expressed in standard delta notation relative to atmospheric N 2 . Manatee ear bone samples were analyzed by Dr.
190 Jason Curtis at the Light Stable Isotope Mass Spectrometry Laboratory, Department of Geological Sciences at the Universit y of Florida. Statistical Analyses 13 15 N values across manatee bone samples, as well as 13 15 N values of reference food sources, were tested using parametric and non parametric analyses. Descriptive statistics (mean, standard deviation, ran ge) were calculated in Microsoft Excel 2010. One factor analysis of variance (ANOVA) , and tests for normality, homogeneity of variances, and equality of means were conducted using IBM SPSS software v 21 (SPSS Inc., New York). The Shapiro Wilk test was use d to determine normality of datasets, and a Levene statistic was used to homogeneity of variances. Robust tests for equality of means included Forsythe tests. The statistical significance of differences in mean values between sampled groups were assessed using (ANOVA), followed by the nonparametric Mann Whitney U test to test for equal distributions between groups. Preliminary analysis to estimate sources that potentially contributed to manatee diet was conducted using scatterplots in Microsoft Excel. Once food sources were plotted, a trophic correction box assuming manatee diet tissue enric hment values of 13 15 N was placed in the scatterplot for each country (Figures 4 3 and 4 4). These enrichment factors fall within the published range for mammals (Schoeninger and DeNiro 1984). Potential food sources were incorporated into the mixing model. Bayesian mixing model analysis software Stable Isotope Analysis in R (SIAR) v 4 (Parnell et al . 2010) was used to estimate the proportional contribution of potential food sources. The SIAR model can incorporate multiple dietary sour ces and the associated variance in isotopic values and uncertainties in trophic enrichment
191 factors (TEF; Parnell et al. 2010). Output from Bayesian mixing models are more robust than previous models because they allow for multiple dietary sources and gener ate potential dietary compositions as true probability distributions (Parnell et al. 2010). This analysis estimated the proportion of reference food sources that contributed to the entire diet of sampled manatees. Alves Stanley and Worthy (2009) studied Florida manatees and determined that 13 15 13 15 N mean enrichment of 13 C mean 15 our study, TEF values for bone collagen were assumed to be +4.0 Â± 2 13 C based upon the published range of 2 15 (Schoeninger and DeNiro 1984, Peterson and Fry 1987). Although the accuracy of the 15 N enrichment value has been recently been called into et al . 2012), it was used in this study due to the similarity of the value published by Clementz (2002) and the absence of any previ ous data for African manatees in Gabon and Senegal. Manatee samples were analyzed by SIAR for location (Senegal River, Senegal coastal, Senegal unknown location, and Gabon), periotic bone layer (GLG, interior vascular bone), age of the manatee (less than 15 years old, greater than 15 years, age unknown), and for Senegal recent versus historic samples (recent samples collected for
192 this study between 2009 2013, and museum samples collected >70 years ago in the same locations). Trophic level for manatees at e ach location was estimated by adding 15 15 N values for 15 N baseline plant values in Gabon and the Senegal River, multiple trophic level s were estimated for each of these locations. In this way, manatees were determined to be either primary (trophic level=2), secondary (trophic level=3), or tertiary consumers (trophic level=4). Results Food Source Isotope Analysis 13 C values for po tential plant food sources from Gabon ranged from 30.3 Â± 0.0 to 10.9 Â± 0.6 31.7 to 15 N ranged from 0.5 Â± 1.0 to 9.1 Â± 2.3 1). In Gabon, individuals of only one potential prey was collected, 13 C value of 26.7 Â± 1.9 (range = 29.8 to 24.1 ) and a mean 15 N value of 0.3 Â± 1.6 (range = 2.4 to 3.0 2) . 13 C values for potential plant food sources from Senegal ranged from 28.5 Â± 0.7 to 1 2.6 Â± 0.4 (range = 29.3 to 11.38 ) and mean 15 N ranged from 0.1 Â± 1.9 to 10.4 Â± 3.0 range = 2.5 to 12.5 1). For potential prey species 13 C values for potential sources ranged from 26.7 Â± 1.9 to 17 .5 Â± 0.5 26.3 to 16.9 ) 15 N ranged from 9.2 Â± 0.9 to 12.4 Â± 0.3 2). Three aquatic plant species which occurred in both Gabon and Senegal ( Crinum natans , Jussiaea repens , and Nymphea lotus ) were tested for 13 15 N differences between countries (Figure 4 5). Crinum natans was significantly enriched in
193 both 13 C (ANOVA, df= 10, F= 22.0, P <0.05) and in 15 N in Senegal (ANOVA, df= 10, F= 19.1, P >0.05) compared to Gabon. Jussiaea repens was not si gnificantly different in 13 C between Gabon and Senegal, but 15 N was significantly higher in Gabon (ANOVA, df = 9, F= 20.8, P <0.05). Nymphea lotus in Gabon was significantly depleted in 13 C compared to Senegal (ANOVA, df= 8, F= 41.2, P <0.05), but there was n o significant difference in 15 N between Gabon and Senegal. Manatee Bone Isotope Analysis In order to examine overall dietary differences between Gabon and Senegal, all 13 C collagen 13 C apatite (Figure 4 6). The collagen and apatite portions of bone assimilate carbon from different components o f the diet (Clementz et al 13 C collagen values (reflecting 13 C apatite (bulk dietary carbon, including carbohydrates, lipids and proteins), which designates apatite 13 C bioapatite collagen = 13 C bioapatite 13 C collagen , Clementz et al . 2009a). This reflects the value of different nutrients (proteins, carbohydrates and lipids) in different diets, which also changes with trophic level (Lee Thorp et al . 1989, Clementz et al . 2009a). Since significan t 13 15 N baseline values of the three plant species sampled from both Gabon and Senegal, data will need to be normalized in order to compare manatee stable isotope data between Gabon and Senegal. Stable carbon isotope 13 C collagen ) from Gabon manatees 13 C collagen values for any location in Senegal, and 13 C apatite (Table 4 13 C collagen values were significantly higher in Senegal than Gabon (ANOVA, df= 41, F= 88.36, P 13 C apatite was also significantly higher in
194 Senegal than Gabon (ANOVA, df= 47, F= 112.41, P 13 C apatite collagen values varied by country but were not s 13 C apatite collagen mean was 6.5 Â± 1.1 (range 4.9 13 C apatite collagen mean was 6.8 Â± 1.8 (range 4.7 9.5). 15 N median values for Gabon and Senegal were 9.74 and 11. 57 respectively; and the distributions in the two groups differed significantly (ANOVA, df= 47, F= 17.33, P < 0.05, Mann Whitney U = P < 0.05 ; Table 4 4 and Figure 4 15 N within bone layers were also significant ( ANOVA, GLG: df= 23, F= 13.35, P <0.05; secondary vascular bone: df= 23, F= 5.13, P <0.05) 13 15 N between bone layers (GLG and secondary vascular tissue, n= 8) were not significant ( ANOVA, P >0.05) indicating that over all diet did not change throughout life for the sampled 13 15 N between differing age categories (<15 years old, n= 4, and >15 years old, n= 12) were also not significant ( ANOVA, P >0.05), indicating that diet values we re similar between age groups. A comparison of bone layer groups by location for Senegal manatees (Figure 4 8) did not indicate any significant trends. Senegal bone samples were not significantly 13 15 N between bone layers (GLG a nd secondary vascular 13 15 N within bone layers between locations (Senegal 13 C nor 15 N were significantly different by location for historical sa mples (ANOVA, Senegal coast, n= 5; Senegal River, n= 2; and Senegal unknown location, n=2; Table 4 5). For recent samples collected from 2009 2013 (Senegal coast, n= 5; Senegal River, n= 11),
195 13 C was not significantly different between locations. However, 15 N for recent samples P <0.05) than those in the Senegal River. Gabon Mixing Model Results Gabon plants that were included in the mixing model analysis included both C3 p lants ( Crinum natans, Hybiscus tiliaceus, Jussiaea repens, Najas pectinata, Nymphea lotus , and Polygonum salicifolium ) and C4 grasses (two Echinochloa species and Vossia cuspidata ) (Figure 4 9). The C3 plants include both aquatic and emergent vegetation. H ermit crab was also included in the model. These potential diet sources were included based upon scatterplot results; they either fell within the box representing the range of stable isotope values expected for manatee food sources, or they were located ju st outside the border of the box (Figure 4 3). Based upon scatterplot results, the following Gabon plant species were determined as unlikely to have been consumed by manatees and were removed prior to SIAR analyses: Ceratophyllum demersum, Dissotis roundif olia , and Ruppia sp. Mean diet proportions were as follows: for manatees <15 years of age, plants made up 90% of diet (individual plant species ranged from 0 to 23%) and hermit crab 10% (range 0 to 20%), regardless of ear bone sampling layer (Table 4 6). For manatees >15 years of age, all plant species combined in the model composed 93 94% of mean diet (individual plant species ranged from 0 to 25%) and hermit crab 6 7% (range 0 to 16%), based upon ear bone sampling layer (GLG and secondary vascular tissue , secondary vascular tissue, n= 8) were not significant (ANOVA, P >0.05) indicating that overall diet did not change throughout life for the sampled manatees. Differences for
196 bo years old, n= 12) were also not significant (ANOVA, P >0.05), indicating that diet values were similar between age groups. N mean value of all plant 4), regardless of age, were above primary consumers but below secondary consumers. This indic ates manatees fed on mixtures of higher and lower trophic level food resources. Additionally, the single for two bone layers) was a secondary consumer. Senegal Mixing Model Res ults Senegal manatee samples collected seventy years ago (historic), regardless of collection location, indicated a more diverse diet than recent manatee samples collected in the past five years (Figure 4 10). Values for historic collagen samples ranged fr om 20.2 to 13 15 N, whereas recent collagen samples ranged from 15.9 to 13 15 13 15 N means were not significantly different between recent and historical samples, F tests for two 15 N (df= 3, F= 34.5, P <0.05) variance for the Senegal River was significantly higher than recent sample values. F tests for two 13 C between rec ent and historical samples. Only two food source items were able to be collected and analyzed for coastal Senegal manatees: the seagrass Halodule wrightii and the estuarine clam Senilia senilis . In SIAR analyses of manatees by bone layer group (Figure 4 11 ), mean
197 proportions for Senilia senilis were estimated at 48% of total diet for GLG samples (SD= 0.2, range 0.29 to 0.70), and 54% diet for the secondary vascular bone layer (SD= 0.2, range 0.37 to 0.72). The remaining proportions (51%, SD= 3.3, range 0.30 to 0.71; and 46%, SD= 4.9, range 0.28 to 0.63respectively) were estimated to be Halodule wrightii (Figure 4 15 N values based upon age 15 N mean va lue for Halodule wrightii For Senegal River manatees, plants represented the highest proportional contribution to their diet for both recent and historic manatee samples. Sources which contributed to manatee diet for that lo cation are shown in Figure 4 13. Based upon ear bone sampling layer, plants composed 46 57% of diet (Table 4 8). Fish and mollusks ranged from 24 27% and 19 24% respectively. Based upon the trophic correction plot (Figure 4 4), Senegal River plant sp ecies that were determined as unlikely to have been isotopically important diet sources and were removed prior to SIAR analyses included: Acacia sp., Azolla africana , Crinum natans , Jussiaea repens , Ludwigia leptocarpa , Nymphea lotus , Nymphea maculata , and Nymphoides indica Potamogeton coloratus , Salvinia nymphellula , Scirpus cubensis , Thypha australis , and five species of Echinochloa . The isotope data were also not consistent with manatees consuming snails as a significant food source. The trophic level f 15 N values (mean = 15 N mean values for all plant species
198 tertiary consumer, indicating a mi xture of higher and lower trophic level food resources, and a definite contribution of animal food sources. Discussion The current results identify diet diversities between two different manatee populations and habitat types. Lifetime dietary differences were expected between the Gabon and Senegal populations because of the very different ecosystems in which manatees live within those two countries, over 2600 kilometers apart. Gabon is a Central African rainforest country, with thousands of aquatic and sh oreline plant species, high rainfall (3050 mm annually), natural lagoon and lake systems, and low human density and development (human population is approximately 1.6 million). Senegal is a desert country, with less than 50 plant species available to manat ees, low rainfall (340 1550 mm annually from north to south), with water flow in the Senegal River completely controlled by humans/dam structures, and a high human density with rapidly increasing development (human population is approximately 13.7 million) . Therefore, not only are there habitat differences between these two countries, but distinct anthropogenic differences which may have differing influences on the ecology of African manatees. Similarities in diet values between manatees at study locations within each country were also expected based upon limited movement patterns for the species (Deutsch et al. 1998, Deutsch et al. 2003, Keith Diagne et al. in prep.), and genetic isolation of populations in the Senegal River and possibly within specific wat erways in Gabon (Chapter 2). Two species represented potential food resources for manatees for the first time by this study: Hybiscus tiliaceus and Najas guadelupensis . Although African manatees
199 throughout their large range have been observed and documente d eating all the plant species which were removed from analyses, results of the present bone samples did not indicate that they were likely part of the diet of the individual manatees sampled in this study. Also, it is difficult to obtain samples of all po ssible plant food resources for such a generalist forager, and there are likely many other plants consumed by African manatees that were not sampled by this study. 13 C collagen apatite values for sirenians to be r 13 C collagen apatite values corresponded to higher fiber content of diet. The study reported that for herbivores consuming low fiber content diets (such as aquatic plants), the lack of 1 3 C depleted methane production causes bone apatite fractionation to be similar to carnivores (Clementz 2002). For African manatees in this study, both Gabon and Senegal manatees had similar proportions of aquatic and terrestrial plants in overall mean diet s. Therefore, the variation between Gabon manatee diet (with depleted values for both 13 C collagen 13 C apatite compared to Senegal manatee diet) may indicate the difference between a primarily plant based diet (Gabon) and an omnivorous diet (Senegal) ( Figure 4 6). Both Gabon and Senegal manatee bone were depleted in 13 C collagen 13 C apatite . This is likely due to the carbohydrate and lipid components of apatite. 13 C values of manatee bone between Gabon and Senegal placed Gabon samples primarily in the range of C 3 plants, and most Senegal samples in the C 4 range (Figure 4 15 N, however, both Gabon and Senegal manatee values 15 N for Gabon was 9.4
200 4 4, Figure 4 15 N values indicate that some manatees in both countries were secondary or tertiary consumers. Therefore, throughout their lives these individuals were likely consuming invertebrates and fish in addition to plants. In contrast, th e results of the only previous bone stable isotope study to consider 15 N for wild West Indian manatees indicated that the species is a primary consumer (Clementz 2002). However, due to previous reports of West Indian manatees feeding on fish and invertebra tes (Powell 1978, Courbis and Worthy 2003), it would be of interest to investigate the potential contribution on animal sources in future studies of other sirenians. The African manatee is not the first durophagous manatee (Clementz et al. 2009b). Miosire n kocki , a Miocene Trichechid, is believed to have had a shellfish diet, based upon its denser palate and 13 13 C enamel values of in stable isotope analyses (Clementz et al . 2009b, Marsh et al. 2012). Gabon Manatee Diet Gabon manatees were documented eating both aquatic and shoreline plants (Reeves et al . 1988, Powell 1996, Ogogo et al . 2013, Keith et al . 2006, Keith and Collins 2007 , Keith Diagne unpublished d ata). Hermit crabs, or other animal 13 15 N values, contributed to diet in all age classes and ear bone layers sampled. Ruppia sp. is only available at the mouths of lagoons where there is high tidal influence from the Atlant ic Ocean. The fact that it did not appear to be part of the Gabon manatee diets may mean that the individuals sampled in this study were feeding further inside lagoons.
201 Diet was not statistically significant between age groups for Gabon manatees based upon SIAR data. Not only does this indicate that manatee diet was similar between age groups, but for manatee samples from individuals <15 years old, the bone may be a reflection of maternal diet, since based on a study of Florida and Amazonian manatee ear bon es, the bone may not have experienced tissue turnover until over 15 years of age (Marmontel et al . 1996). One Gabon manatee (MGAO1302) was an outlier in all analyses (Figures 4 3, 4 4, and 4 6). This sample was collected from a freshwater lake system in i nterior Gabon that had few aquatic plants but abundant shoreline vegetation (primarily C 4 grasses) that overhung waterways, and which manatees can access. Therefore, it is unsurprising that this manatee sample fell in the C 4 13 C. All other Gabon manatee samples were collected from coastal lagoons which are primarily freshwater, but experience tidal fluctuations and saltwater intrusion during the annual dry season. The lagoons have abundant aquatic plants as well as shoreline vegetation (mangroves, grasses, tropical forest) primarily composed of C 3 plants, therefore it is unsurprising that 13 C values as C 3 plants. Future Gabon dietary studies should consider including mollusks and fish s ince these prey items are reported for manatees throughout Africa, including Central Africa, and were shown to constitute a high proportion of manatee diet in Senegal. Senegal Manatee Diet Estimated food resources for coastal Senegal manatees were likely artificially biased by incorporating only two food sources into the SIAR model. However, the sources used in the model yield the first indication of diet for African manatees in marine habitat. Additional sources would likely change dietary proportions of sources in this
202 study. Future collection efforts will include mangrove species, oysters, and additional seagrass samples. For Senegal River manatee samples, all four plant values that contributed to diet fell in the C 4 13 C = 18.47 to 11.38). Thi s was unexpected for aquatic plants in a freshwater system, but the values for the two species identified ( Crinum natans and Najas guadelupensis 13 C = 30 to 15 N = The higher trophic level, 15 N values observed in Senegal manatees compared to Gabon manatees was expected, based upon previous reports that African manatees are frequently observed eating mollusks and fish, particularly in the dry season (Reeves et al . 1988, Powell 1996, Keith Diagne unpublished data). There are also two published reports of carnivory for West Indian manatees: Powell (1978) reported on manatees eating fish from nets in Jamaica, and Courbis and Worthy (2003) reported observing Florida manatees feedin g on barnacles, bivalves, gastropods, crabs, tunicates, and possibly other invertebrates on dock pilings. In general, though, manatees are considered herbivores and have been thought to only choose other food resources such as fish and mollusks when plants are not available (VÃ©lez Juarbe 2014). However, this study indicates that African manatees in the Senegal River, and also possibly coastal Senegal, have high percentages of fish and mollusks in their average lifetime diets. This indicates that these food sources are regularly consumed and are an important component of the diet of these populations. These results are also notably different from all previous stable isotope studies for wild
203 West Indian manatees (Clementz 2002, MacFadden et al . 2004, Reich and Worthy 2006, Clementz et al . 2007, Alves Stanley et al . 2010). Although historic vs. recent diets were not found to be significantly different, the 13 15 N in average lifetime diet of the historic samples indicates manatees consu med a wider range of food resources 70 years ago than they do today. For Senegal coastal manatees, food resource changes over the past 70 years may be linked to destruction of seagrass and mangrove habitats in the region over that time period (Cunha and Ar aÃºjo 2009, Conchedda et al . 2011, Sakho et al . 2011). Large scale reforestation of mangrove habitats in Senegal in the last 20 years have also been documented and may return availability of that food source to historic levels in some areas (Conchedda et al . 2011, Sakho et al . 2011, Sall and Durin 2012). Seagrass beds, however, continue to be subjected to destruction due to large scale trawling for fishing (Cunha and AraÃºjo 2009), and there appears to be little if any current effort for protection and restor ation of this resource. For the Senegal River manatee population, the reduction in values from historic to recent diet is likely due to a major anthropogenic change: the Diama dam was built near the mouth of the Senegal River in 1984. This permanently isol ated the river manatee population from the Atlantic Ocean and removed the availability of coastal food resources such as seagrass, mangroves and marine mollusks from their diet. Knowledge of African manatee lifetime diet provides a baseline understanding of the foraging and habitat needs for the species in two very diverse ecosystems. This study provides a starting point for future studies of diet utilizing other tissues, such as skin and hair, which could identify seasonal and migratory differences in die tary and
204 habitat resources. The continued collection of present day samples and their comparison to historical data for Senegal will aid understanding of how manatees have adapted to changes in their environment from the past. Managers need to understand t he differing habitat requirements of the African manatee across an extremely large range in Africa. Therefore the comparison of diet in a Central African rainforest country versus a West African Sahelian system, and freshwater versus marine habitats will e stablish a baseline which will be can inform managers of the best habitats to protect within larger eco regions.
205 Table 4 1. 15 13 C Â±SD of plant species analyzed in this study. Species are listed by collection location within each c ountry, then ranked by 13 C values. Plant Species n Collection Location 13 15 Ruppia sp. 3 Gabon, Corisco Bay 10.9 Â± 0.6 0.5 Â± 1.0 Echinochloa sp. 1 12 Gabon, Lac OguemouÃ© 11.3 Â± 0.4 3.6 Â± 1.7 Echinochloa sp. 2 17 Gabon, La c OguemouÃ© 11.8 Â± 0.6 9.1 Â± 2.3 Polygonum salicifolium 10 Gabon, Lac OguemouÃ© 25.1 Â± 7.9 5.5 Â± 2.0 Jussiaea repens 7 Gabon, Lac Ezanga 27.8 Â± 1.0 6.2 Â± 1.6 Dissotis roundifolia 18 Gabon, Lac OguemouÃ© 29.1 Â± 4.4 1.8 Â± 2.1 Hybiscus tiliaceus 8 Gabon, Lac OguemouÃ© 29.5 Â± 0.4 6.5 Â± 1.6 Vossia cuspidata 7 Gabon, Ogooue River 12.3 Â± 1.1 4.6 Â± 1.2 Crinum natans 6 Gabon, N'dogo Lagoon 22.9 Â± 4.4 6.9 Â± 3.4 Nymphea lotus 3 Gabon, N'dogo Lagoon 27.6 Â± 0.7 1.7 Â± 1.8 Najas pectinata 3 Gabon, N'dogo Lagoo n 28.1 Â± 0.3 5.2 Â± 0.8 Ceratophyllum demersum 1 Gabon, N'dogo Lagoon 30.3 2.9 Halodule wrightii 2 Senegal, Atlantic Ocean 12.6 Â± 0.4 6.2 Â± 0.6 Echinochloa sp. 4 1 Senegal, Senegal River 11.4 6.4 Echinochloa sp. 3 3 Senegal, Senegal River 12.5 Â± 0. 5 7.2 Â± 0.3 Crinum natans 6 Senegal, Senegal River 14.8 Â± 0.5 3.9 Â± 2.0 Najas guadelupensis 6 Senegal, Senegal River 14.8 Â± 2.7 0.9 Â± 4.3 Acacia sp. 2 Senegal, Senegal River 27.9 Â± 0.2 10.4 Â± 3.0 Eclipta prostratra 1 Senegal, Senegal River 28.1 0. 0 Scirpus cubensis 6 Senegal, Lac de Guiers 17.3 Â± 7.4 0.9 Â± 0.6 Potamogeton coloratus 6 Senegal, Lac de Guiers 26.3 Â± 0.5 1.7 Â± 1.0 Nymphea lotus 6 Senegal, Lac de Guiers 23.1 Â± 1.1 0.9 Â± 1.4 Nymphea maculata 1 Senegal, Lac de Guiers 23.6 0.5 Ny mphoides indica 4 Senegal, Lac de Guiers 24.7 Â± 1.3 0.1 Â± 1.9 Salvinia nymphellula 2 Senegal, Lac de Guiers 26.9 Â± 0.3 2.2 Â± 1.2 Thypha australis 4 Senegal, Lac de Guiers 27.8 Â± 0.4 3.2 Â± 1.1 Azolla africana 1 Senegal, Lac de Guiers 27.9 1.2 Jussi aea repens 3 Senegal, Lac de Guiers 28.5 Â± 0.7 1.9 Â± 0.5 Ludwigia leptocarpa 1 Senegal, Lac de Guiers 28.4 0.5 n, number of samples ; SD, standard deviation
206 Table 4 2. 15 13 C Â± 1 standard deviation (SD) of animal prey species ana lyzed in the present study. Species are listed by collection 13 C values. Animal Species n Collection Location 13 15 Hermit crab sp. 11 Gabon, Banio Lagoon 26.7 Â± 1.9 0.3 Â± 1.6 Senilia senili s (clam) 5 Senegal, Atlantic Ocean 17.5 Â± 0.5 12.4 Â± 0.3 Synodontis sp. (catfish) 6 Senegal, Senegal River 20.1 Â± 2.1 11.1 Â± 0.9 Alestes macrolepidotus (fish) 5 Senegal, Senegal River 21.3 Â± 0.5 11.2 Â± 1.1 Clarias anguillaris (catfish) 5 Senegal, Sen egal River 22.7 Â± 1.3 12.0 Â± 0.4 Mutela rostrata (mussel) 2 Senegal, Senegal River 25.0 Â± 0.7 9.0 Â± 1.4 Coelatura aegyptiaca (mussel) 7 Senegal, Senegal River 25.1 Â± 0.4 9.2 Â± 0.9 Aspatharia dahomeyensis (mussel) 9 Senegal, Senegal River 25.7 Â± 0.5 9.9 Â± 0.9 Snail sp. 1 Senegal, Senegal River 26.2 9.3 n , number of samples ; SD, standard deviation
207 Table 4 3. 13 C collagen and apatite values for study samples by location, and references from previous sirenian studies. 13 13 Species Location n Tissue Mean SD Range Mean SD Range Reference T. senegalensis Gabon 8 Bone: this study GLG 21.3 3.5 23.1 to 12.8 14.1 2.9 15.9 to 7.1 Internal 20.4 3.9 23.4 to 11.6 14.6 3.4 17.0 to 6.5 Senegal River 7 Bone: GLG 13.7 1.9 16.3 to 11.3 6.4 1.3 7.9 to 4.6 Internal 12.9 1.4 14.6 to 11.2 6.2 1.0 7.8 to 4.9 Senegal, coastal 7 Bone: GLG 13.2 3.6 20.2 to 8.5 6.2 3.1 11.5 to 1.6 Internal 12.9 3.6 17.8 to 7.6 7.0 2.8 10.1 to 2.0 Senegal, unknown 2 Bone: GLG 13.0 4.0 15.8 to 10.1 4.9 2.8 6.8 to 2.9 Internal 12.6 3.7 15.2 to 10.0 4.4 1.9 5.7 to 3.1 T. manatus Florida 17 Bone 14.5 5.0 8.0 4.7 Clementz 2002 D. dugong Australia 8 Bone 6.5 0.9 1.3 0.8 H. gigas Museum collection (extinct) 11 Bone 7.3 0.8 H. gigas Museum collection (extinct) 8 Bone 15.4 0.8 T. manatus Florida (wild) 6 Bone 19.3 1.9 12.5 2.0 Clementz et al . 2007 H. gigas Museum collection (extinct) 10 Bone 15.2 1.0 7.3 0.8 layer.
208 Table 4 4. 15 N collagen values for study samples by location, and references from previous sirenian studies. 15 Species Location n Tissue Mean SD Range Reference T. senegalensi s Gabon 8 Bone collagen this study GLG 9.4 1.7 7.1 to 11.7 Internal 9.9 1.8 6.9 to 12.4 Senegal River 7 Bone collagen GLG 11.2 0.8 9.9 to 12.5 Internal 10.7 1.3 8.0 to 11.6 Senegal, coastal 7 Bone collagen GLG 12.1 0.9 11.1 to 13.4 Internal 12.4 0.7 11.6 to 13.4 Senegal, unknown location 2 Bone collagen GLG 10.7 2.7 8.8 to 12.7 Internal 10.6 3.0 8.5 to 12.7 T. manatus Florida 17 Bone colla gen 8.1 2.0 Clementz 2002 D. dugong Australia 8 Bone collagen 3.5 1.0 H. gigas Museum collection (extinct) 8 Bone collagen 11.1 0.6 secondary vascular bo
209 Table 4 5. 13 C 15 N values for Senegal recent and historic samples by bone layer. Recent samples were collected from 2009 2013, 13 15 Location n Recent / Historic B one Layer Mean SD Range Mean SD Range Senegal River 5 Recent GLG 13.1 1.8 Â±4.6 11.1 0.2 Â±0.5 5 Internal 12.7 1.4 Â±3.4 11.1 0.5 Â±1.0 2 Historic GLG 15.2 1.5 Â±2.1 11.2 1.9 Â±2.7 2 Internal 13.2 1.9 Â±2.7 9.8 2.5 Â±3.6 Senegal, c oastal 2 Recent GLG 12.8 0.0 Â±0.0 11.6 0.8 Â±1.1 2 Internal 12.6 1.6 Â±2.3 12.2 0.9 Â±1.3 5 Historic GLG 13.4 4.4 Â±11.7 12.3 0.9 Â±2.0 5 Internal 13.0 4.3 Â±10.1 12.4 0.7 1.7 Senegal, unknown location 2 Historic GLG 13.0 4.0 Â±5.7 10.7 2.7 Â±3.9 2 Internal 12.6 3.7 Â±5.2 10.6 3.0 Â±4.2
210 Table 4 6. Gabon manatee mean diet proportions by ear bone layer (GLG, internal) and age category (<15 years old, >15 years old or age unknown) based on SIAR analysis. GLG, <15 years GLG, >15 years GLG, Unknown Age Internal, <15 years Internal, >15 years Internal, Unknown Age Diet Source Mean Range Mean Range Mean Ra nge Mean Range Mean Range Mean Range Crinum natans 0.10 0.0 to 0.20 0.12 0.0 to 0.23 0.10 0.0 to 0.20 0.10 0.0 to 0.20 0.12 0.0 to 0.24 0.10 0.0 to 0.20 Echinochloa sp. 1 0.08 0.0 to 0.18 0.07 0.0 to 0.17 0.10 0.0 to 0.20 0.09 0.0 to 0.18 0.06 0.0 to 0.1 5 0.10 0.0 to 0.20 Echinochloa sp. 2 0.08 0.0 to 0.19 0.12 0.0 to 0.22 0.09 0.0 to 0.20 0.09 0.0 to 0.19 0.10 0.0 to 0.20 0.10 0.0 to 0.20 Hybiscus tiliaceus 0.12 0.0 to 0.23 0.12 0.0 to 0.23 0.10 0.0 to 0.20 0.11 0.0 to 0.21 0.13 0.0 to 0.25 0.10 0.0 to 0.20 Jussiaea repens 0.11 0.0 to 0.22 0.12 0.0 to 0.23 0.10 0.0 to 0.20 0.11 0.0 to 0.21 0.13 0.0 to 0.24 0.10 0.0 to 0.20 Najas pectinata 0.11 0.0 to 0.22 0.10 0.0 to 0.21 0.10 0.0 to 0.20 0.11 0.0 to 0.21 0.11 0.0 to 0.22 0.10 0.0 to 0.20 Nymphea lot us 0.10 0.0 to 0.21 0.08 0.0 to 0.18 0.10 0.0 to 0.20 0.10 0.0 to 0.20 0.08 0.0 to 0.19 0.10 0.0 to 0.20 Polygonum salicifolium 0.12 0.0 to 0.23 0.12 0.0 to 0.23 0.10 0.0 to 0.20 0.11 0.0 to 0.21 0.13 0.0 to 0.24 0.10 0.0 to 0.20 Vossia cuspidata 0.08 0. 0 to 0.18 0.09 0.0 to 0.19 0.10 0.0 to 0.20 0.09 0.0 to 0.19 0.07 0.0 to 0.17 0.10 0.0 to 0.20 Plant total proportion 0.90 0.94 0.90 0.90 0.93 0.90 Hermit crab proportion 0.10 0.0 to 0.20 0.06 0.0 to 0.16 0.10 0.0 to 0.20 0.10 0.0 to 0.20 0.07 0.0 to 0.16 0.10 0.0 to 0.20
211 Table 4 7 . 13 C 15 N values for study samples by location, manatee bone sampling layer , and age class. 13 15 Location n Bone Layer Age Class Mean SD Range Mean SD Range Gabon 2 GLG <15 years 22.7 0.5 22.3 to 23.0 8.8 1.2 7.9 to 9.6 2 Internal <15 years 20.8 1.9 22.2 to 19.5 8.8 2.7 6.9 to 10.6 5 GLG >15 years 22.4 0.8 23.1 to 21.0 9.2 1.7 7.1 to 11.7 5 Internal >15 years 22.0 1.5 23.4 to 19.8 10.1 1.7 8.0 to 12.4 1 GLG Unk Age 12.8 11.7 1 Internal Unk Age 11.6 11.0 Senegal River 2 GLG <15 years 15.0 1.8 16.3 to 13.7 10.5 0.9 9.9 to 11.2 2 Internal <15 years 14.6 0.0 14.6 to 14.6 9.4 2.0 8.0 to 10.8 5 GLG >15 years 13.2 1.9 15.9 to 11.3 11.4 0.7 11.0 to 12.5 5 Interna l >15 years 12.2 0.9 13.6 to 11.2 11.2 0.5 10.6 to 11.6 Senegal, coastal 0 <15 years 3 GLG >15 years 12.5 1.6 13.9 to 10.7 12.0 1.1 11.1 to 13.2 3 Internal >15 years 14.7 0.8 15.3 to 13.7 12.2 0.6 11.6 to 12.7 1 GLG Unk Age 12.7 1 2.1 1 Internal Unk Age 11.5 12.9 Senegal, unknown 2 GLG <15 years 18.0 3.1 16.3 to 13.7 10.3 2.1 9.9 to 11.2 2 Internal <15 years 16.5 1.8 17.8 to 15.2 10.3 2.6 8.5 to 12.1 3 GLG >15 years 11.6 4.0 13.7 to 8.5 12.5 1.0 11.3 to 13.4 3 Internal >15 years 9.1 1.3 10.0 to 7.6 12.6 0.9 11.7 to 13.4
212 Table 4 8. Senegal River manatee mean diet proportions b y ear bone layer (GLG, internal) and collection time (recent= 2009 2013, SIAR analysis. Diet Source GLG, Recent GLG, Historic Internal, Recent Internal, Historic A. dahomeyensis (mussel) 0.06 0.09 0.08 0.09 C. aegyptiaca (mus sel) 0.06 0.09 0.09 0.09 M. rostrata (mussel) 0.06 0.09 0.09 0.09 Mollusk total proportion 0.19 0.26 0.25 0.27 A. macrolepidotus (silverfish) 0.08 0.09 0.10 0.09 C. anguillaris (catfish) 0.07 0.08 0.09 0.09 Synodontis sp. (catfish) 0.09 0. 09 0.10 0.09 Fish total proportion 0.24 0.26 0.29 0.27 C. natans 0.14 0.12 0.11 0.11 Echinochloa sp. 3 0.15 0.12 0.12 0.12 Echinochloa sp. 4 0.16 0.12 0.12 0.12 N. guadelupensis 0.12 0.12 0.11 0.11 Plant total proportion 0.57 0.48 0 .46 0.46
213 Figure 4 1. Diagrams and caption courtesy of Marmontel et al . (1996). Photomicrograph (A) and line drawings (B) of whole section (original magnification 10Ã—) and detail (original magnification 100Ã—) from the central portion of the periotic d ome of an age class 8 Florida manatee. PV, primary vascular bone; SV, secondary vascular bone; CL, compact lamellar bone; WR, white rim. Numerals 1 8 mark the growth layer groups.
214 Figure 4 2 . Sampling regions of African manatee ear bones, including th e type of bone found in each sample. The first five bones were sampled at A, B and C layers, and the remainder (n= 21) were sampled at B and C layers only .
215 Figure 4 3. Gabon manatee samples by bone layer and food resource species collected for this stu dy. Plant species are represented by squares and diamonds, hermit crab samples are represented by circles, and manatee samples are represented by triangles. The black box represents the range of stable isotope values expected for manatee food sources. Food sources within the box were used in the SIAR mixing model. -4 -2 0 2 4 6 8 10 12 14 -35 -30 -25 -20 -15 -10 -5 15 N 13 C Gabon Manatees and Food Sources by Species C. demersum C. natans D. roundifolia Echinochloa sp.1 Echinochloa sp.2 H. tiliaceus J. repens N. pectinata N. lotus P. salicifolium Ruppia sp. V. cuspidata Hermit crab Manatee internal layer Manatee GLG
216 Figure 4 4. Senegal River manatee samples by bone layer and food resource species collected for this study. Plant species are represented by squares and diamonds, mollusks and fish are repre sented by circles, and manatee samples are represented by triangles. The black box represents the range of stable isotope values expected for manatee food sources. Food sources within the box were used in the SIAR mixing model. -4 -2 0 2 4 6 8 10 12 14 -35 -30 -25 -20 -15 -10 -5 15 N 13 C Senegal River Manatees and Food Sources by Species Acacia sp. A. africana A. dahomeyensis A. macrolepidotus C. aegyptiaca C. anguillaris C. natans E. prostratra Echinochloa sp. 3 Echinochloa sp. 4 J. repens L. leptocarpa M. rostrata N. guadelupensis N. indica N. lotus N. maculata P. coloratus P. subaldidum S. nymphellula S. cubensis Snail sp. Synodontis sp. T. australis Manatee internal layers Manatee GLGs
217 Figure 4 5. Comparison 13 15 N values for three aquatic plant species colle cted in both Senegal and Gabon. -4 -2 0 2 4 6 8 10 -30 -25 -20 -15 -10 15 N 13 C Baseline Values of Plant Species from Both Gabon and Senegal Senegal C. natans Senegal J. repens Senegal N. lotus Gabon C. natans Gabon J. repens Gabon N. lotus
218 Figure 4 13 C collagen 13 C apatite (reflecting total diet) for Gabon and Senegal manatees by periotic bone sampling layer. layer. Figure 4 13 C and 15 N values for all Gabon and Senegal manatees by ; internal, -25 -20 -15 -10 -5 0 -25 -20 -15 -10 -5 0 Senegal GLG Senegal Internal Gabon GLG Gabon Internal 0 5 10 15 -25 -20 -15 -10 -5 0 15 N 13 C Senegal GLG Senegal Internal Gabon GLG Gabon Internal
219 Figure 4 8. Senegal manatees shown by location and ear bone sampling layer. GLG, Figure 4 9. SIAR output of Gabon manat ees by ear bone sampling layer (GLG, internal) and by age class (<15 years old, > 15 years old, or age unknown) , shown in relation to food sources . 6 7 8 9 10 11 12 13 14 -25 -20 -15 -10 -5 0 15 N 13 C Senegal Manatee Samples by Bone Layer Senegal River GLG Senegal River internal Senegal Coast GLG Senegal Coast Internal Senegal Unk Loc GLG Senegal Unk Loc Internal
220 Figure 4 13 15 N for Senegal manatees 2013) sample collection by location. Both GLG and internal bone layers are included, therefore each manatee is represented by two points. 6 8 10 12 14 -25 -20 -15 -10 -5 0 15 N 13 C Senegal Historic (1940's) and Recent (2009 2013) by Location Historic Coastal Historic Senegal River Historic Unknown Location Recent Coastal Recent Senegal River
221 Figure 4 11. SIAR output of Senegal coast manatee samples by ear bone sampling layer shown in relation to food sources (colored cross hairs). Standard error for Senilia senilis 13 15 N = 0.11. Standard error for Halodule wr ightii 13 15 A B Figure 4 12. Boxplots of Senegal coast manatee diet sources by ear bone sampling layer: A) growth layer group, and B) secondary vascular bone.
22 2 Figure 4 13. SIAR output of S enegal River manatee samples by ear bone sampling
223 CHAPTER 5 CONCLUSIONS Next Steps for African Manatee Genetics and Stable Isot ope Research We are at the very beginning of our understanding of the population genetics and feeding ecology of the African manatee. The present study has identified manatee regional population structure across Africa, as well as local population structure within Gabon and Senegal, which will h elp inform management efforts for the species. My hope is that this preliminary work will inspire others a cross Africa to collect genetic samples and to collaborate with this ongoing work so that we can conti nue to add to our knowledge of manatee diversity in other African countries. Additionally, through stable isotope analyses this study has, for the first time, documented food resources that African manatees consume, and has identified diet diversity betwee n Central African rainforest and West African Sahelian ecosystems. Mollusks and fish play an important role in the diets of at least some African manatee populations, and as with the genetics, hopefully this study will inspire other African researchers to conduct additional dietary studies for the species. There is much more to discover regarding the feeding ecology of African manatees and these discoveries will also clarify critical food resource and habitat needs. Aside from including more genetic sampl es from additional locations across Africa , other mtDNA genes should be investigated, and the entire genome should be sequenced for the species. Determining the complete DNA sequence for the African manatee would provide further information on the origin a nd evolution of the species, frequency of mutations, and would allow more extensive comparisons with closely related species such as the West Indian manatee. The chromosome number is also still
224 unknown. Nuclear DNA and microsatellite analyses have yet t o b e conducted for the African manatee, and this is work I plan to begin soon in collaboration with genetics colleagues. These methodologies can estimate diversity and delineate fine scale population structure, important parameters in conservation and natural resource decisions. There are many more recently developed genetic methodologies that could and should be applied to African manatees , including environmental DNA (eDNA) sampling, r estriction site associated DNA (RAD) sequencing for use in identifying sin gle nucleotide polymorphisms (SNPs) and genome wide variation , e xon primed intron crossing ( EPIC ) primers, and more. A portion of each genetic sample collected for this study has be en archived at the USGS Sirenia Project in Gainesville, FL, to be available for future an aly ses. Hopefully this will lead to continued collaboration with African researchers and will allow for more detailed an alyses over the long term. There are also many other techni ques for addressing feeding ecology and habitat use of the African manatee using stable isotopes. Given that manatees replace their teeth throughout life, analysis of tooth enamel has the potential to show diet over time by taking samples from older and yo ung er teeth. Manatee molars are replaced approximately every seven years (Domning 1982, Domning and Hayek 1984) and enamel is laid down in incremental layers that can be assessed seasonally or annually (Clementz 2002). T herefore , teeth could be used to exa mine shorter time frames than ear bones provide. Afri can manatees also possess nails on their flippers , which may be a good tissue to analyze from carcasses , as it may provide information for time frames that span several months and could indicate seasonal differences in diet. Manatees throughout Africa are reported to migrate between wet and dry season habitats (Powell
225 1996, Keith et al . 2007), and dietary shifts are recorded in tissues with faster turnover rates. Skin samples are likely to have turnover r ates similar to Florida manatees , and isotope signatures have been shown to vary between regions for manatees sampled in Florida (Reich and Worthy 2006). However, controlled studies with African manatees to verify turnover rates will be difficult to achiev e in the near future , due to the small number of captive animals worldwide (almost all of which are in Asia) . Manatee hair and vibrissae have not previously been analyzed for stable isotopes, but could provide an intermediate chronological record between the longer record of bone, and the shorter term record of skin. During this study, hair was collected from two manatees, including serial collection over three years from an orphan manatee calf raised in captivity in Gabon. The next step will be to analyze these hair samples to determine the length of time recorded and to try to identify seasonal diet shifts. Whenever possible, hair and skin samples will now be regularly collected from lived captured manatees in Africa by this project , and those tissues as well as bone and skin will be collected from carcasses. Other Research and Conservation Needs The African manatee is believed to be in decline throughout much of its range (Keith Diagne and Powell 2014) , but without accurate baseline information it is impo ssible to know how to conserve these manatees, and conservation is also unlikely without capacity building within the African countries where the species exists . A lack of long term, committed fundi ng and trained researchers, as well as the difficulty of a ccessing the extremely remote regions where African manatees live, murky water past studies , and challenge future studies .
226 The most critical research need for the conserva tion of African manatees in the near future is the documentation of illegal hunting and the bushmeat trade. African manatee meat is openly sold in markets and consumed in villages across Africa, yet there a re almost no record s of the numbers of manatees h unted and sold, and therefore there is no documentation available to encourage and pressure African wildlife law enforcement agencies to enforce the laws that are supposed to protect the species. Genetic samples also need to be collected from manatee popul ations which are known to be heavily hunted, since genetic analyses are able to estimate population sizes as well as detect population decreases (and increases). These methodologies can inform researchers, managers, and law enforcement of population conser vation priorities within countries and at regional scales. Additionally, other key threats to manatees need attention. Documentation and mitigation of rates of manatee bycatch are needed due to the huge number of artisanal fisheries and fishermen t hroughout coastal Africa which incidentally catch manatees. Many manatees drown in nets, but even those unintentionally caught alive are not released, usually because manatee meat can sell for a higher profit than fish. The st marine mammal stranding network last year documented 13 stranded manatees in ten months during 2013 (Takoukam et al . 2013). Additional stranding networks need to be implemented in other countries , both to document manatee bycatch and to conduct educatio nal and incentive campaigns in order to reduce it. Manatee population isolation due to the construction of dams is a major issue, particularly in the Niger and Senegal Rivers and Lake Volta. As more and more large
227 hydroelectric and smaller agricultural da ms are built across Africa, the problem is intensifying. Studies of these populations are needed to assess levels of population genetic diversity and population sizes. Although sample size was small (n= 12), results of the current study for the Senegal Riv er indicated that mtDNA genetic diversity of that population is low, therefore strong conservation measures are needed to preserve as many individuals as possible. The next step for this project is to increase sample sizes for the Senegal River population so that effective population size can be determined. We will also work directly with Senegal government agencies including the Senegal River Basin Authority, and ministries of Water and Forestry and the Environment to train their field staff in manatee res earch and conservation techniques. We will also work towards the same goals with project collaborators on the Niger River and Lake Volta. Many other species besides manatees are impacted by dams (economically valuable fish species, turtles, etc.) so there are good opportunities to join forces with other biologists, but solutions to will require long term studies and a healthy amount of creativity. Emerging threats for manatees in Africa include increased habitat destruction and increasing levels of motorize d watercraft. Increasing human population levels along due to development and pollution. As human growth continues, b aseline studies should be implemented to monitor ch anges in manatee habitat and populations. Baseline health and life history parameter studies on the African manatee are also needed . We currently have v ery little basic information about this species , and growth rates, age of
228 first reproduction, calving interval , longevity, and many other attributes are still unknown . Many new African manatee conservation initiatives have begun since 2008. From 2007 2009 the Earthwatch Institute (UK) funded annual training workshops for African biologists interested to begin man atee research and conservation activities. The workshops were held at Lake Volta, Ghana and were led by P. Ofori Dansen, C. Self Sullivan and myself . Over three years 33 participants from 17 African countries were trained, and equally importantly, a spirit of collaboration was born. When the Earthwatch funding ended, I continued the training works hops in other countries, and have built an African manatee researcher network to increase and sustain communication between researchers in different countries, inc rease data collection, conservation activities , and educational awareness programs. As of 2014 this project has trained over 72 African biologists from 18 countries, and year round manatee research now occurs in nine countries as a direct result . The long term strategy is to create a sustainable and cohesive network of African researchers who will determine population sizes and status of African manatees in a majority of the 21 range countries, as well as develop and implement management plans for conservat ion of the species. As the network is set up and more collaborators are trained, detailed and more focused research efforts are beginning, such as the protection of specific habitats where manatee use is found to be high, and targeting anti poaching enforc ement to known hunting areas. Collaborators also share their knowledge and train others, further increasing the number of biologists working with the species.
229 For a long term effort such as this, operating on a continental scale, there will always be chal lenges. Ivory Coast had one of the most important manatee conservation programs from 1989 2009 (Akoi 1992, Akoi 2004) but unfortunately it ended with the funding was withdrawn after his death. This was a hard lesson that too often manatee programs in Africa are dependent up on the work of specific individuals. More people need to be trained and efforts need to be integrated into broader conservation programs so that man atee conservation initiatives do not end if one person is no longer able to continue the work. M ore university level students need to be involved, and more laboratories set up for sample analysis. O f course there will always be the need to secure funding t o continue all of these efforts. Conservation of this elusive species over its enormous and mostly remote range will take the dedicated effort of as many researchers working on the ground as possible. This species is only likely to be conserved through a n etwork of grassroots, localized efforts by African researchers dedicated to long term conservation and education efforts in their countries, and through increased funding of research, management, and conservation activities.
230 APPENDIX AFRICAN MANATEE FOO D RESOURCES Table A 1. Documented food resources of the African manatee. Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference PLANTS Acroceras zizanioides Angola, Cameroon, DRC, Re p ublic of the C ongo, Gabon, Ghana, Guinea, Guinea Bissau, Ivory Coast, Nigeria, Senegal Ogogo et al . 2013 Aeschynomene crassicaulis Angola, Benin, DRC, Mali, Nigeria, Senegal Kombo and Toko 1991 Alternanthera sessilis Sessile Joyweed Ghana, Ivory Coast, Senegal Ako i 2004, Keith Diagne Avicennia germinans Black Mangrove Angola, Benin, Cameroon, DRC, Republic of the Congo, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Nigeria, Senegal, Sierra Leone Powell 1996 Azolla africana Benin, Cameroon, Chad, Gabon, Ghana, Ivory Coast, Republic of the Congo, Senegal Powell 1996, Keith Diagne and Powell 2014 Brachiaria mutica, Urochloa mutica Para grass Ghana Ofori Dansen pers. comm. Brachiaria ramosa, Urochloa ramosa Browntop Mill et Ivory Coast Akoi 2004 Calamus deerratus Ivory Coast Akoi 2004 Ceratophyllum demersum Coontail, Hornwort Gabon, Ghana, Republic of the Congo, Mali, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diagne and Powell 2014 Ceratopteris cornuta B road leafed Water Sprite Ghana, Republic of the Congo Keith Diagne
231 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Chrysophyllum sp. Ivory Coast Akoi 2004 Colocasia antiquorum Elephant Ear, Taro Gabon, Republic of the Congo, Nigeria Ogogo et al . 2013, Keith Diagne Crinum natans African Onion Gab on, Republic of the Congo, Senegal Powell 1996, Keith Diagne Cryptosperma senegalensis Gabon, Republic of the Congo Powell 1996, Keith Diagne Cymodocea nodosa Slender Seagrass Angola, Gambia, Mauritania, Senegal, Sierra Leone Villiers and Bessac 1948, Husar 1978, Powell 1996 Cyperus papyrus Papyrus Angola, Benin, Cameroon, DRC, Gabon, Guinea, Ivory Coast, Republic of the Congo, Senegal, Togo Keith Diagne Dalbergia ecastaphyllum Coin vine Cameroon, Gabon, Ghana, Ivory Coast, Liberia, Nigeria, Seneg al Akoi 2004 Dissotis roundifolia Pink Lady Angola, Cameroon, Congo, Gabon Akoi 2004 Drepanocarpus lunatus Ivory Coast, Nigeria Powell 1996, Akoi 2004 Eichhornia crassipes Water Hyacinth Angola, Cameroon, Gabon, Ghana, Republic of the Congo, Senegal Husar 1978, Powell 1996, Keith Diagne and Powell 2014 Echinochloa crus pavonis Gulf Cockspur Grass Nigeria Ogogo et al. 2013, Keith Diagne and Powell 2014 Echinochloa pyramidalis Antelope Grass Cameroon, Gabon, Gambia, Ghana, Ivory Coast, Mali, Niger , Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Akoi 2004, Keith Diagne and Powell 2014 Echinochloa stagnina Cameroon, Nigeria Ogogo et al . 2013, Keith Diagne and Powell 2014
232 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Ficus asperifolia Sandpaper tree Angola, Benin, Cameroon, Chad, Republi c of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Akoi 2004 Ficus exasperata Brahma's Banyan, forest sandpaper fig, rough banyan, sand paper fig Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Ogogo et al . 2013 Flagellaria guineensis Kanoti grass Benin, Gabon, Ivory Coast, Nigeria Akoi 2004 Halodule wrightii Shoal grass Angola, Benin, Cameroon, Equatorial Guinea, Gabon, Ghana, Guinea Bissau, Mauritania, Nigeria, Senegal, Sierra Leone, Togo Green and Short 2003 , Keith Diagne Hybiscus tiliaceus Sea Hibiscus, Beach Hibiscus, Coastal (or Coast) Hibiscus Gabon, Republic of the Congo Keith Diagne Ipomoea aquatica Water Morning Glory, Water Spinach Gabon, Ghana, Nigeria Ogogo et al . 2013, Keith Diagne Ipomoea rep tans Water Spinach Ivory Coast Akoi 2004
233 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Jussiaea repens (also known as Ludwigia stolonifera ) Floating Primrose Willow Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast , Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996 Leersia hexandra Ghana, Senegal Powell 1996 Lemna aequinoctialis, Spirodela polyrhiza Tropical Duckweed Gabon, Ghana, Republic of the Congo Powell 1996, Keith Diagne a nd Powell 2014 Lemna paucicostata Duckweed Ghana Ogogo et al . 2013, Keith Diagne and Powell 2014 Leptochloa caerulescens Ghana, Nigeria Ogogo et al . 2013 Ludwigia leptocarpa Anglestem Primrose willow Angola, Benin, Cameroon, Chad, Republic of the Con go, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diagne and Powell 2014 Macaranga sp . Ivory Coast Akoi 2004 Manihot esculenta Manioc / Cassava Gabon, Republic of the Congo, Senegal Powell 1996, Keith Diagne Manihot palmata Nigeria Ogogo et al . 2013 Merremia hederacea Gabon, Ghana, Ivory Coast, Senegal Akoi 2004 Mimosa pigra Sensitive Tree Gabon, Ghana Powell 1996, Keith Diagne Myriophyllum sp. Watermilfoil Senegal Keith Diagne
234 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Najas pectinata Gabon, Cameroon, Chad, Ghana, Guinea, Nigeria, Senegal, Sierra Leone Powell 1996, Keith Diagne Najas sp. (guadelupensis?) Senegal Keith Diagne Neptunia oleracea Water mimosa Ghana Keith Diagne Nymphea lotus Egyptian White Water Lily Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diagne Nymphea maculata Red African Tiger Lotus Lily Gabon, Republic of the Congo, Senegal Powell 1996, Keith Diagne Nymphoides indica Water Snowflake Gabon, Republic of the Congo, Senegal Powell 1996, Keith Diagne Oryz a sp. Rice Ghana, Senegal, Sierra Leone Reeves et al . 1988, Keith Diagne Panicum subaldidum Bunchgrass, Buffalograss Angola, Cameroon, Ghana, Mali, Nigeria, Senegal Ogogo et al . 2013 Paspalidium geminatum (also known as Echinochloa geminata ) Egyptian P aspalidium (tall grass) Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Akoi 2004, Keith Diagne Paspalum vaginatum Seashore Paspal um (short grass) Ivory Coast Powell 1996, Akoi 2004
235 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Pennisetum purpureum Elephant Grass, Napier Grass or Ugandan Grass Sierra Leone Reeves et al . 1988 Phragmites karka Common Reed Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diagne and Powell 2014 Pistia stratiotes Water Lettuc e Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diagne and Powell 2014 Poly gonum salicifolium / Polygonum persicaria (now reclassified as Persicaria maculosa ) Redshank Gabon, Ivory Coast Husar 1978, Powell 1996, Akoi 2004, Keith Diagne and Powell 2014 Polygonum senegalense Knotweed, Knotgrass Ghana, Senegal Villiers and Bessac 1948, Powell 1996, Keith Diagne and Powell 2014 Potamogeton coloratus Fen Pondweed Senegal Keith Diagne Rhizphora harrisonii Red Mangrove Angola, Benin, Cameroon, Gabon, Gambia, Ghana, Guinea Bissau, Guinea, Liberia, Nigeria, Senegal, Sierra Leone, T ogo Husar 1978, Keith Diagne and Powell 2014
236 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Rhizophora mangle Red Mangrove Angola, Benin, Cameroon, DRC, Republic of the Congo, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Nigeria, Senegal, Sierra Leone Husar 1 978 Rhizophora racemosa Red Mangrove Angola, Benin, Cameroon, DRC, Republic of the Congo, Equatorial Guinea, Gabon, Ghana, Ivory Coast, Liberia, Nigeria, Sierra Leone Husar 1978, Powell 1996, Akoi 2004, Keith Diagne Rhynchospora corymbosa Golden Beak S edge Angola, Benin, Chad, Congo, DRC, Gabon, Ghana, Nigeria, Senegal, Togo Powell 1996, Keith Diagne Ruppia sp. Beaked Tasselweed Angola, Gabon, Ghana Powell 1996, Green and Short 2003, Keith Diagne Salvinia molesta Water Fern, Giant salvinia DRC, Ghana , Nigeria, Senegal, Togo Keith Diagne Salvinia nymphellula Benin, Gabon, Ghana, Senegal Powell 1996, Keith Diagne Schoenoplectus senegalensis Club rush Nigeria Ogogo et al . 2013 Scirpus cubensis Cuban Bulrush Cameroon, Ghana, Senegal Keith Diagne Thypha australis Cattail Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Powell 1996, Keith Diag ne
237 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Utricularia sp Bladderworts Angola, Cameroon, Chad, Congo, DRC, Gabon, Niger, Nigeria Powell 1996 Vossia cuspidata Hippo Grass Angola, Benin, Cameroon, Chad, Republic of the Congo, DRC, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau , Ivory Coast, Liberia, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Togo Husar 1978, Powell 1996, Keith Diagne SHELLFISH Aspatharia dahomeyensis Freshwater Mussel Benin, CÃ´te d'Ivoire, Gambia, Ghana, Guinea, Guinea Bissau, Mali, Mau ritania, Niger, Nigeria, Senegal, Togo Keith Diagne Coelatura aegyptiaca Freshwater Mussel Cameroon, Chad, DRC, Gambia, Guinea, Ivory Coast, Niger, Nigeria, Senegal Keith Diagne Egeria congica Freshwater and marine clam Angola, DRC only Keith Diagne Mutela rostrata Freshwater Mussel Benin, Cameroon, Chad, Republic of the Congo, DRC, CÃ´te d'Ivoire, Gambia, Ghana, Guinea, Guinea Bissau, Mali, Mauritania, Niger, Nigeria, Senegal, Togo Keith Diagne Senilia senilis Heavy African Ark Gambia, Guinea B issau, Mauritania, Senegal, Sierra Leone Keith Diagne
238 Scientific Name Common Name Known Countries of Occurrence within T. senegalensis range Manatee Reference Tympanotonus fuscatus Mud flat Periwinkle Angola, Benin, Cameroon, Congo, CÃ´te d'Ivoire, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Liberia, Nigeria, Senegal, Sierra Leone, Togo Perez Rivera, pers. comm. FISH Alestes macrolepidotus Silver fish Cameroon, Ghana, Liberia, Niger, Senegal, Sierra Leone, Togo Keith Diagne Clarias anguillaris Mudfish Mali, Senegal Keith Diagne Synodontis sp. Catfish Ghana, Ivory Coast, Mal i, Senegal Keith Diagne
239 LIST OF REFERENCES ABE (Agenc (1999) RÃ©pertoire des indicateurs environnementaux de d Ã©veloppement durable et de compendium s tatistique du BÃ©nin . Cotonou, BÃ©nin . Adjakpa JB (2002) Ã‰cologie du l amanti Trichechus senegalensis (Link, 1995) dans la partie BÃ©ninoise d u b assin du fleuve Niger et ses affluents. CEROE / Wetlands International . Affomasse TM (1999) Inventaire et etude de la stratÃ©gie de protection des populations de l amantins ( Tric hechus senegalensis ) des les zones humides du s ud BÃ©nin. Programme d . Akoi K (1992) Education et sensibilisation des populations pour la conservation du lamantin ouest A fricain ( Trichechus senegalensis Wild life Conservation Society, New York . Akoi K (1994) Une enquÃªte prÃ©liminaire sur les lamantins dans les eaux de la r Ã©serve de la Conkouati au sud du Congo. CanopÃ©e , 4 , 10 Akoi K (2000) Projet de conservation du lamantin ouest A fricaine en Cote d'Ivoire (No te de prÃ©sentation Ã l'atelier sur les petits cÃ©tacÃ©s) . Akoi K (2004) Fishers and the West African manatee in the Fresco Lagoon complex, om mon property, conflict and c onservation. Dissertation, University of Kent at Canterbury, United King dom . Allen GM (1942) Extinct and Vanishing M ammals of the Western H emisphere . American Committee for International Wildlife Protection, Special Publication No. 11. The Intelligencer Printing Co., Lancaster, PA . Alves Stanley CD (2007) Stable isotope turn over rates and diet tissue discrimination in the skin of West Indian manatees: implications for evaluating their feeding ecology and habitat use. Master s Thesis , University of Central Florida. Orlando, FL . Alves Stanley CD, Worthy GAJ (2009) Carbon and n itrogen stable isotope turnover rates and diet tissue discrimination in Florida manatees ( Trichechus manatus latirostris ). J ournal of Experimental Biol ogy , 212 , 2349 2355 . Alves Stanley CD, Worthy GAJ , Bonde RK (2010) Feeding preferences of West Indian ma natees in Florida, Belize, and Puerto Rico as indicat ed by stable isotope analysis. Mar ine Ecol ogy Prog ress Series , 402 , 255 267 Ambrose SH (1990) Preparation and characterization of bone and tooth collagen for isotopic analysis. Journal of Archaeological Science , 17 , 431 451.
240 A mes AL, Van Vleet ES, Sackett WM (1996) The use of stable carbon isotope analysis for determining the dietary habits of the Florida m anatee, Trichechus manatus latirostris . Mar ine Mam mal Sci ence , 12 (4), 55 563 Arnason U, Adegoke JA , Bodin K, Born EW , Esa YB , Gullberg A , Nilsson M , Short RV , Xu X , Janke A (2002) Mammalian mitogenomic relationships and the root of the eutherian tree. Proceedings of the National Academy of Sciences of the United States of America , 99 (12), 8151 8156 . A sano S, Sakamoto S (1997) Toba a quarium acquires West African manatees. Sirenews , 27 , 13 14 . Avis e JC (2004) Molecular Markers, Natural History, and E volution. Sinauer Associates, Sunderland, MA . Awobamise A (2008) Nigeria. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 75 78. UNEP, Nairobi, Kenya and Wetlands International A frica, Dakar, Senegal . Ba E, Diop MD, Sarr K, Niane AM, Jaulin M (2008) Senegal. In: Conservation strategy for the West African manatee ( eds Dodman T, Diop MD, Sarr K ) , pp. 26 30, UNEP, Nairobi, Kenya and Wetlands Inter national Africa, Dakar, Senegal . Bal G, BrÃ©heret, N (2007) Etude des lamantins dans la zone allant du Lac Youbi au f lueve Kouilou en Repub lique du Congo. Renatura Report. B alkie B (1857) On the skull of a Manatus from Western Africa. Proceedings of the Zoolog ical Society of London , 25 , 29 33 . Barnes LG , Domning DP , Ray CE (1985) Status of studies on fossil marine mammals. Mar ine Mamm al Sci ence , 1 (1), 15 53 . Barnett AA, Pr angley ML (1997) Mammalogy in the Republic of Guinea: an overview of research from 1946 to 1996, a preliminary check list and a summary of research recommendations for the future. Mammal Review , 27 , 115 164 . Barrow LM , Bjorndal KA , Reich KJ (2008) Effects of preservation method on stable c arb on and nitrogen isotope values. Physiological and Biochemical Zoology , 81 (5), 688 693 . Benoit J , Adnet S , El Mabrouk E , Khayati H, Ben Haj Ali M, Marivaux L, Merzeraud G , Merigeaud S , Vianey Liaud M , Tabuce R (2013) C ranial remain from Tunisia provides new clues for the origin and e volution of Sirenia (Mammalia, Afrotheria) in Africa. PLoS ONE , 8 (1), e54307. doi:10.1371/journal.pone.0054307 .
241 Berthe S (2011) Rapport du suivi lamantin sur le Bani, un affluent du Niger a u Mali. Niger River Basin Authority, San, Mali . Best RC (1983) Apparent dry season fasting in Amazonian manatees (Mammalia: Sirenia). Biotropica , 15 , 61 64 . Bischof B , Mariano AJ, Ryan EH (2004) The No rth Equatorial Counter Current. Ocean Surface Curren ts. http://oceancurrents.rsmas.miami.edu/atlantic/north equatorial cc.html . Accessed March 20, 2012 . Blair D, McMahon A, McDonald B, Tikel D, Waycott M, Marsh H (2013) Pleistocene sea level fluctuations and the phylogeography of the dugong in Australian w aters. Mar ine Mamm al Sci ence, doi : 10.1111/mms.12022 . Bocherens H, Drucker D (2003 ) Trophic level isotopic enrichment of carbon and nitrogen in bone collagen: case studies from recent and a ncient terrestrial e cosystems. International Journal of Osteoarcha eol ogy , 13 , 46 53 . Bocherens H, Mashkour M, Drucker DG , Moussa I, Billiou D (2005) Stable isotope evidence for palaeodiets i n southern Turkmenistan during H istorical period and Iron Age. J ournal of Archaeol ogical Sci ence , 33 , 253 264 . Boekschoten GJ, Bes t MB (1988) Fossil and r ecent shallow water corals from the Atlantic islands off western Africa. Zoologische Med edelingen (Leiden) , 62 (8), 99 112 . Bolobo EP (2001) Document de Perspectiva RÃ©pu blica de Guinea Ecuatorial. El estudio de perspectivas del se ctor f orestal en Ãfrica (FOSA)/ Forestry Sector Outlook Studies, FA O, Rome . Bonde RK (2009) Population genetics and c onser vation of the Florida m anatee: past, present and f uture. Dissertation , University of Florida . Bonde RK, McGuire PM, Hunter ME (2012) 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. Bossart GD (2001) Manatees. In : CRC handbook of marine mammal medicine , 2nd e d., (eds Dierauf LA, Gulland FMD), CRC Press , Washington DC . Boubacar B ( 2010 ) Ecobiologie et determinants de la distribution spatio temporelle du lamantin ( Trichechus senegalensis ) dans le Parc W Niger. S tudy plan, Univ ersitÃ© Abdou Moumouni de Niamey, N iger . Boubacar B (2012) Death of a female manatee in Niger. Niamey, Niger .
242 Bouckaert R, Heled J, KÃ¼hnert D, Vaughan TG, Wu C H, Xie D, Suchard MA, Rambaut A, Drummond AJ (2014) BEAST2: A software platform for Bayesian evolutionary a nalysis. PLOS Computat ional Biol ogy , 10 (4), e1003537. doi:10.1371/journal.pcbi.1003537 . Bradley JL , Wright SD , McGuire PM (1993) The Florida manatee: cytochrome b DN A sequence. Mar ine Mamm al Sci ence , 9 , 197 202 . Broderick D , Ovenden J , Slade R , Lanyon JM (2007) Characterizati on of 26 new microsatellite loc i in the dugong ( Dugong dugon ). Molec ular Ecology Notes , 7 , 1275 1277 . Brown WM (1983) Evolution of animal mitochondrial DNA. In: Evolution of genes and proteins (eds Nei M, Koehn RK), pp. 62 88, Sin auer Associates, Sunderla nd, MA. Buchardt B , Bunch V , Helin P (2007) Fingernails and diet: s table isotope signatures of a marine hunting community from mod ern Uummannaq, North Greenland. Chemical Geology , 244 , 316 329 . Camara O (2011) Suivi des sites et campagnes de sensibilisa tion des communautÃ©s des pÃªcheurs et des scolaires pour la protection du lamantin ouest africain ( Trichechus senegalensis ) le long du littoral GuinÃ©en. Centre National des Sciences Halieutiques de Boussoura (CNSHB), Guinea . Cantanhede AM , Da Silva VMF , Fa rias IP , Hrbek T , Lazzarini SM , Alves Gom es J ( 2005 ) P hylogeography and population genetics of the endangered Amazonian manatee, Trichechus inunguis Natt erer, 1883 (Mammalia, Sirenia). Molecular Ecology , 14 , 401 413 . Carpenter SR, Caraco NF, Correll DL, H owarth RW, Sharpley AN, Smith VH (1998) Nonpoint Pollution of Surface Waters with Phosphorus and Nitrogen. Ecological Applications, 8 (3), 559 568. Cerling TE, Harris JM (1999) Carbon isotope fractionation between diet and bioapatite in ungulate mammals an d implications for ecological and paleoecological studies. Oecologia , 120 , 247 363 . Cerling TE, Wittemyer G , Ehleringer JR , Remiene CH, and Douglas Hamilton I (2009) History of Animals using Isotope Records (HAIR): A 6 year dietary history of o ne family o f African elephants. Proceedings of the National Academy of Sciences , 106 (20), 8093 8100 . Chabi YaourÃ© F (2012) Le dÃ©fi de la sauvegarde des espÃ¨ces menacÃ©es au BÃ©nin , u .
243 Ciofolo I, Sadou I (1996) L e Lamantin du Niger ( Trichechus senegalensis ). Technical report 7 CACP/MIR/O82. Ministry of Finance and P lanning. European Union. GenÃ¨va , Switzerland. Clementz MT (2002) Sea cows, seagrasses, and stable isotopes: biogeochemical evaluation of the ecology a nd evolution of the Sirenia and Desmostylia. Dissertation, University of California at Santa Cruz , CA. Clementz MT , Goswami A , Gingerich PD , Koch PL (2 006) Isotopic records from early whales and sea cows: contrasting patterns of ecological transition. Jou rnal of Vertebrate Paleontology , 26 (2), 355 370 . Clementz M , Koch P (2001) Differentiating aquatic mammal habitat and foraging ecology with stable isotopes in toot h enamel. Oecologia , 129 , 461 472 . Clementz MT , Koch PL , Beck CA (2007) Diet induced differ ences in carbon isotope fractionation between siren ians and terrestrial ungulates. Marine Biology , 151 , 1 773 1784 . Clementz MT , Fox Dobbs K , Wheatley PV , Koch PL , Doak DF (2009 a ) Revisiting old bones: coupled carbon isotope analysis of bioapatite and coll agen as an ecolog ical and palaeoecological tool. Geological Journal , 44 (5), 605 620 . Clementz MT , Sorbi S , Domning DP (2009 b ) Evidence of Cenozoic environmental and ecological change from stable isotope analysis of Sirenian remains from the Tethys Mediter ranean region. Geology , 37 , 307 310 . Codron D , Codron J , Sponheimer M , Bernasconi SM , Clauss M (2011) When animals are not quite what they eat: diet digestibility influences 13 C incorporation rates and apparent discriminatio n in a mixed feeding herbivor e. Canadian J ournal of Zoology 89 ( 6), 453 465 . Collins T, Keith L, Rosenbaum H (2011) Marine mammal monitoring project final report, inshore and Congo River marine m ammals . Wildlife Conservation Society , New York. Conchedda G, Lambin EF, Mayaux P (2011) Between land and s ea: Livelihoods and environmental c hanges in mangrove e cosystems of Senegal. Annals of the Association of American Geographers , 101 ( 6 ) , 1259 1284 . Courbis SS, Worthy GAJ (2003) Opportunistic carnivory by Florida manatees ( Trichechus ma natus latirostris ) . Aquatic Mammals , 29 (1), 104 107 . Crawford K , McDonald RA , Bearhop S (2008) Applications of stable isotope techni ques to the ecology of mammals. Mammal Review , 38 (1), 87 107 .
244 Cunha AH, AraÃº jo A (2009) New distribution limits of seagras s beds in West Africa . Journal of Biogeography , 36 , 1613 1622. Dalerum F, AngerbjÃ¶rn A (2005) Resolving temporal variation in vertebrate diets using natur ally occurring stable isotopes. Oecologia , 144 , 647 658 . Demopoulos AWJ, Gualtieri D, Kovacs K (2010 ) Food web structure of seep sediment macrobenthos from the Gulf of Mexico. Deep Sea Research II , 57 , 1972 1981 . DeNiro M J, Epstein S (1978) Influence of diet on distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta , 42 , 495 506 . De N iro MJ, Epstein S (1981) Influence of diet on the distribution of nitrogen isotopes in animals. Geochimica et Cosmochimica Acta , 45 , 341 351 Deutsch CJ, Bonde RK, Reid JP ( 1998) Radio trackin g manatees from land and s pace: tag design, implementation, and lessons l earned from long term study. Marine T echnology Society Journal , 32 (1), 18 29 . Deutsch CJ , Reid JP, Bonde RK, Easton DE, Kochman HI, O'Shea TJ ( 2003 ) Seasonal movements, migratory behavior, and site fidelity of the West Indian manatees along the Atlantic coast of the United States . Wildlife Monographs , 151 , 1 77 Dia Ibrahima M ( 2012 ) Vulnerability Assessment of Central Coast Senegal (Saloum) and The Gambia Marine Coast and Estuary to Climate Change Induced Effects. Coastal Resources Center and WW F WAMPO, University of Rhode Island . Diarra ST (2011) West Africa: Niger River under pressure from d ams. IPS News website ( http://www.ipsnews.net/2011/10/west africa niger river under pressure from dams/ ) , Accessed March 21, 2014 . Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2: more models, new heuristics and parallel computing. Nature Methods , 9 (8), 772. doi:10.1038/nmeth.2109 . Diop MD (2006) Conservation du Lamantin ouest A fricain Trichechus senegalensis le long du littoral Ouest Afr icai n: EnquÃªte de base dans la rÃ©serve de b iosphÃ¨re du delta du Saloum (SÃ©nÃ©gal) September 2005. Survey team of IUCN and Parc Nationale Delta du Saloum. Wetlands International, Dakar, Senegal . Dodman T (2008 a ) Angola . In: Conservation strategy for the West Af rican manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 109 112. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Dodman T (2008 b ) Democratic Republic of the Congo . In: Conservation strategy for the West African manatee (eds Dodman T, Di op MD, Sarr K ) , pp. 107 108. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal .
245 Dodman T (2008 c ) Equatorial Guinea . In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 99 100. UNEP, Nairobi, Ken ya and Wetlands International Africa, Dakar, Senegal . Dodman T, Akoi K, Ajonina G, Kaya P (2006) Report on an exchange p r ogramme at coastal wetlands of t he Congo, April 2005. Wetlands International, Dakar, Senegal . Dodman T, Diop NMD, Khady S (eds) (2008 ) Conservation strategy for the West African Manatee . UNEP, Nairobi, Kenya and Wetlands Interna tional Africa, Dakar, Senegal Domning DP (1994) A Phylogenetic a nalysis of the Sirenia. In: Contributions in Marine Mammal Paleontology Honoring Proceedings of San Diego Society of Natural History . (eds Whitmore F C Jr, Berta A, DemÃ©rÃ© T A ), 29 , 177 189 . Domning DP (1981) Sea cows and seagrasses. Paleobiology , 7 , 417 420 . Domning DP (1982) Evolution of manatees, a speculative h istory. J ournal of Paleontology , 56 ( 3), 599 619 . Domning DP (2000) The readaptation of Eocene sirenians to life in water. Historical Biology , 14 , 115 119 . Domning DP (2001a) The earliest known fully quadrupedal sirenian. Nature , 413 (6856), 625 62 7 . Domning DP (2001b) Sirenians, seagrasses , and Cenozoic ecological change in the Caribbean. Palaeogeography Palaeoclimatology Palaeoecology , 166 (1 2), 27 50 . Domning DP (2002) Sirenia n evolution. In: The Encyclopedia of Marine Mammals (eds Perrin WF , Wursig B , Thewissen JGM), pp. 1083 1086 . Acade mic Press, San Diego. Domning DP (2005) Fossil Sirenia of the West Atlantic and Caribbean region. VII. Pleistocene Trichechus ma natus Linnaeus, 1758. J ournal of Vertebrate Paleontology , 25 (3), 685 701 . Domning DP, Gingerich PD (1994) Protosiren smithae , new species (Mammalia, Sirenia), from the late Middle Eocene of Wadi Hitan, Egypt. Contributions from the Museum of Paleontology , University of Michigan , 29 (3), 69 87 . Domning DP, Hayek LC (1986) Interspecific and i ntraspecific morphological variation in the m anatees (Sirenia: Trichechus ). Mar ine Mamm al Sci ence , 2 (2), 87 144 . Domning DP, Morgan GS, Ray CE ( 1982) North American Eocene sea cows (Mammalia: Sirenia). Smithsonian Co ntributions to Paleobiology , 52 , 1 69.
246 Domning DP, Thomason J , Corbett DG (2 0 07) Steller's sea c ow in the Aleutian Islands. Marine Mammal Science , 23 (4), 976 983 . Dossou BodjrÃ©nou JS (2003) Projet education, conservation et r echerche sur le BÃ©nin. Unpublished report NT ONG NÂ° 001 PL NT, CBDD/UUIC Pays Bas, Nature Tropicale Benin . Dossou BodjrÃ©nou J, Chabi YaourÃ© F, Zannou J ( 2006 ) la gestion rationnelle et communautaire des ressources biologiques et des Ã©cosystÃ¨mes des sites et des couloirs de migratio humides du Sud BÃ©nin. Nature Tropicale Benin . Dumas D, Mietton M, Hamerlynck O , Pesneaud F , Kane A , Coly A , Duvail S, Baba MLO (2010) Large dams and uncertainties: the c ase of the Senegal River (West Afri ca). Society and Natural Resources: An International Journal , 23 ( 11 ) , 1108 1122 . Endamne LN (2007 ) Analyse de la problÃ©matique de conservation dans le secteur riviÃ¨re Komo et affluents avec un accent particulier sur le lamantin. WWF, Libreville, Gabon . E va AN (1980) Pre Miocene seagrass communities in the Carib bean. Palaeontology , 23 , 231 236 . Excoffier L, Lisc her HEL (2010) Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol ecular Ecol ogy R esources , 10 (3), 564 56 7 . Farias IP , Orti G , Sampaio I , Schneider H, Meyer A (2001) The c ytochrome b gene as a phylogenetic marker: the limits of resolution for a nalyzing relationships among Cichlid f ishes. Journal of Molecular Evolution , 53 , 89 103 . F ederov AV, Dekens PS, McCarthy M, Ravelo AC, deMenocal PB, Barriero M, Pacanowski RC, Philander SG (2006) The Plio cene Paradox (mechanisms for a p ermanent El NiÃ±o). Science , 312 ( 5779 ) , 1485 1489 . Federov AV, Brierley CM, Lawrence KT, Liu Z, Dekens PS , Ra velo AC (2013) P atterns and mechanisms of early Pliocene warmth . Nature , 496 , 43 52 . doi:10.1038/nature12003 . Ferreira EO, Loseto LL, Ferguson SH (2011) Assessment of claw growth layer groups from ringed seals ( Pusa hispida ) as biomonitors of inter and i ntra 15 N, and 13 C variation. Canadian Journal of Zoology Revue Canadienne De Zoologie , 89 (9), 774 784 . Filippelli GM, Flores J A (2009) From the warm Pliocene to the cold Pleistocene: A tale of two oceans. Geology , 37 , 959 960 .
247 Frankham R , Ballou JD , Briscoe DA (2010) Introduction to Conservation G enetics , Second Edition. Cambr idge University Press, New York . ISBN: 978 0 521 70271 3. Fry B (2006) Stable Isotope Ecology . Springer , New York . Fry B, Brand W , Mersch FJ, Tholke K, Garritt R (19 92) Automated analysis system for coupled delta C 13 and delta N 15 measurements. Anal ytical Chemistry , 64 (3), 288 291 . Fu Y X (1997) Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics , 1 47 , 915 925 . Garcia Rodriguez AI (2000) Genetic studies of the West Indian manatee ( Trichechus manatus ). Dissertation , University of Florida . Garcia Rodriguez AI, Bowen BW , Domning DP , Mignucci Giannoni AA, Marmontel M , Montoya Ospina RA , Morales Vela B , Rudin M , Bonde RK, McGuire PM (1998) Phylogeography of the West Indian manatee ( Trichechus manatus ): how many populations and how many taxa? Molecular Ecology , 7 , 1137 1149 . Green EP, Short FT (2003) World Atlas of Seagrasses. UNEP World Conservation Mo nitoring Centre. University of California Press, Berkeley, CA . Grigione MM (1996) Observations on the status and distribution of the West African manatee in Cameroon . African Journal of Ecology , 34 (2), 189 195 . Guedegbe B, Kidjo CF, Guedou R, Affomasse M (2000) Inventaire et Ã©tude de stratÃ©gie de conservation des populations de mammifÃ¨res menacÃ©es: le Lamantin ( Trichechus senegalensis Hippopotamus amphibius ), le Sitatunga ( Tragelaphus spekeii ). Rappor t de recherches, Cotonou, Benin . H adj idakis DJ, A ndroulakis II (2006) Bone r emodeling. Annals of New York Academy of Science , 1092 , 385 396 . Hartman DS (1979) Ecology and behavior of the manatee (Trichechus manatus) in Florida. American Society of Mammalogy , Special Publication No. 5, Pittsb urgh, PA . Hasegawa M, Kishino H, Yano T ( 1985 ) Dating the human ape split by a molecular clock of mitochondrial DNA. J ournal of Mol ecular Evol ution , 22 :160 174 . Hatt RT (1934) The American Museum Congo expedition manatee and other recent manatees. Bullet in of the American Museum of Natural History , 66 , 533 566 .
248 Hautier L , Sarr RL , Tabuce R , Lihoreau F , Adnet S , Domning DP, Samb M , Hameh PM (2012) First Prorastomid sirenian fossil from Senegal (Western Africa) and the old world origin o f seacows. J ournal of Vertebra te Paleontology , 32 (5), 1218 1222 . Hellborg L, Ellegren H (2003) Y chromosome conserved anchored tagged sequences (YCATS) for the analysis of mamma lian male specific fragments. Mol ecular Ecol ogy , 12 , 283 291 . Hill JE, Grubb P, Jones TS, Davies AG, Edberg E, Starin ED (1998) Mammals of Ghana, Sierra Leone and The Gambia . Trendrine Press, St Ives, Cornwall . Hillis DM , Larson A , Davis SK , Zimmer EA (1990) Nucleic Acids III: Sequencing. In : Molecular S ystematics , 2 nd ed ition (eds Hillis DM, Moritz C), Sinauer Associates, Sunderla nd, Massachusetts . Hillis DM , Mable BK , Larson A , Davis SK , Zimmer EA (1996) Nucleic ac ids IV: sequencing and cloning. In: Molecular systematics 2 nd edition (eds Hillis DM, Moritz C) , Sinauer Associates, Bos ton, Massachuset ts . Hobson KA (1999) Tracing origins and migration of wildlife u sing stable isotopes: a review. Oecologia , 120 , 314 326 . Hobson KA, Clark RG (1992) Assessing avian diets using stable isotopes 1. Turnover of 13 C in tissues. Condor , 94 , 181 188 . Hobson KA , Schell DM, Renouf D, Noseworthy E (1996) Stable carbon and nitrogen isotopic fractionation between diet and tissues of captive seals: implications for dietary reconstructions. Canadian Journal of Fisheries and Aquatic Science , 53 , 528 533. Hobson KA, Wa ssenaar LI (e ds ) (2008) Tracking Animal Migration with Stable I sotopes . Terrestrial Ecology Series London, Academic Press . ISBN: 978 0 12 373867 7. Hobson KA, Welch HE (1992) Determination of trophic relationships within a high Arctic marine food web usin Marine Ecology Progress Series , 84 , 9 18. HÃ¶ll C, Zonneveld KAF , Willems H (2000) Organic walled dinoflagellate cyst assemblages in the tropical Atlantic Ocean and oceanographical changes over the last 140 ka. Palaeogeography, Pa laeoclimato logy, Palaeoecology , 160 , 69 90 . Holsinger KE (2012) Tajima's D , Fu's F S , Fay and Wu's H , and Zeng et al .'s E . http://darwin.eeb.uconn.edu/eeb348/lecturenotes/mol evol tajima.pdf . Accessed 22 May 2014 . Hopkins JB , Ferguson JM (2012) Estimating the diets of animals using stable isotopes and a compr ehensive Bayesian mixing model. P loS O ne , 7 (1), e28478 .
249 Hunter ME , Auil Gomez NE , Tucker KP , Bonde RK , Powell J, McGui re PM (2010a) Low genetic variation and evidence of limited dispersal in the regio nally important Belize manatee. Animal Conservation , 13 (6), 592 602 . Hunter MK , Broderick D , Ovenden JR , Tucker KP , Bonde RK , McGuire PM , Lanyon JM (2010b) Characterization of highly informative cross species microsatellite panels for the Australian dugong ( Dugong dugon ) and Florida manatee ( Trichechus manatus latirostris ) including five novel primer s. Mol ecular Ecol ogy Resources , 10 , 368 377 Hunter ME, Mignucci Giannoni AA , Tucker KP, King TL, Bonde RK, Gray BA, McGuire PM (2012) Puerto Rico and Florida manatees represent genetically distinct groups. Conserv ation Genet ics , doi : 10.1007/s10592 012 0414 2 . Husar S (1978) Trichechus senegalensis . Mammalian Species , 89 , 1 3 . Ichola RO, Tchibozo S (2008) Benin. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 71 74. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Idriss MH (2008) Chad. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 91 94. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal Ingham MC (1970) Coastal upwelling in the northwestern Gulf of Guinea. Bull etin of Mar ine Sci ence , 20 (1), 1 34 . Issa AM (2008) Niger. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 83 86. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . IUCN and Pa rc National du Delta du Saloum (2005) Conservati on du Lamantin ouest A fricain , Trichechus senegalensis , le long du littoral ouest africain : EnquÃªte de base dans la RÃ©serve de BiosphÃ¨re du delta du Saloum (SÃ©nÃ©gal). Wetlands International , Dakar , Senegal. Jallow AW (2008) The Gambia. In: Conservation s trategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 31 34. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Jim S (2000) 13 C values as a new source of palaeodietary information: qualitative and quantitative models of its use in conjunction 13 C values. D iss ertation, University of Bristol . Kataoka T, Asano S, Wakai Y (2000) Update on sirenians at Toba Aquarium. Sirenews , 33 , 12 13 .
250 Kaya P (2005) Exploration des lamantins Africai ns dans le fl euve LoÃ©mÃ© . Kaya JAP (2008) Republic of the Congo. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 101 106. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Keith LW, Collins T, Mbadinga N (2006) Report on West African m an atee ( Trichechus senegalensis ) survey a ctivities in Gabon 2006 . Keith LW, Collins T (2007) Report on West African m an atee ( Trichechus senegalensis ) survey a ctivities in Gabon 2007 . Keith Diagne L (2011) West African m anatee ( Trichechus senegalensis ) research and c onservation in Gabon. Mission report of scientific studies in the nationa l p arks of Gabon. Keith Diagne L (2012) The forgotten sirenian. Seagrass Watch , 46 , 14 17 . K eith Diagne LW, Powell JA (2014) Trichechus senegalensis . IUCN Red List of Threatened Species Version 2014.1. < www.iucnredlist.org > . Kellogg ME, Burkett S , Dennis TR , Stone G , Gray BA , McGuire PM , Zori RT, Stanyon R (2007) Chromosome painting in the manatee supports Afrotheria and Paenungulata. BMC Evolutionary Biology , 7 ( 6) , doi:10.1186/1471 2148 7 6. Kienta M (1982) Preliminary Investigations on the manatee ( Tri chechus senegalensis ) at Lac Debo, Mali, West Africa . Kienta M, K one B, Timbo S (2008) Mali. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 79 82. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Kimura M (1980) A simple method for estimating evolutionary rates of base substitutions through comparative s tudies of nucleotide sequences. J ournal of Mol ecular Evol ution , 16 , 111 120 Kocher TD , White TJ (1989) Evolutionary analysis via PCR. I n : PCR T echnology (ed Erlich HA), Stockton Press, New York, NY . Kombo Y, Toko. J M (1991) Notes sur les lamantins du Lac Debo. Rappo rt UICN Programme Pour Le Sahel . Kumar S, Tamura K, Nei M (2004) MEGA3: integrated software for molecular evolutionary gene tics analysis and sequence alignment. Briefings in Bioinformatics , 5 , 150 163 .
251 Kuntner M, May Collado LJ, Agnarsson I (2011) Phylogeny and conservation priorities of afrotherian mammals (Afrotheria , Mammalia). Zoologica Scripta , 40 (1), 1 15 . Lanyon JM, M arsh H (1995) Digesta passage times in the d ugong. Australian Journal of Zoology , 43 (2) , 119 127 . Larkin ILV, Fowler VF, Reep RL (2007) Digesta passage r ates in the Florida m anatee ( Trichechus manatus latirostris ). Zoo Biology , 26 , 503 515. Lee Thorp J A, Scaly JC, van der Merwe NJ (1989) Stable carbon isotope ratio differences between bone collagen and bone apatite, and their relationship to diet. J ournal of Archaeological Sci ence , 16 , 585 599 . L ibrado P, Rozas J (2009) DnaSP v5: a software for compreh ensive analysis of DNA polymorphism data. Bioinformatics , 25 , 1451 1452 . Longin R (1971) New method of collagen extraction for radiocarbon dating. Nature , 230 , 241 242. Louembet S (2008) Analyse des prÃ©lÃ¨veme nts des mammifÃ¨res aquatiques: lamantin et h ip popotame 2005 . Loughman WD, Frye FL, Herald ES (1970) The chromosomes of a male manatee, Trichechus inunguis . International Zoo Yearbook , 10 , 151 152 . Lowes RHG (1970) Destruction i n Sierra Le one. Oryx , 10 , 309 310 . Luna FO , Bonde RK , Attademo FLN , Saunders JW , Meigs Friend G , Passavante JZO, Hunter ME (2012) Phylogeographic implications for release of critically endangered manatee calves rescued in Northeast Brazil. Aquatic Conservation: Mari ne and Freshwater Ecosystems , 22 , 665 672. M acFadden BJ , Higgins P , Clementz MT, Jones DS (2004) Diets, habitat preferences, and niche differentiation of Cenozoic sirenians from Florida: evidence from stable isotopes. Paleobiology , 30 (2), 297 324. Machad o A (1998) Borrador de anteproyecto de ley de areas protegidas de Guinea Ecuatorial. Documento Tecnico No. 14. Proyecto C.U.R.E.F., Bata, Guinea Ecuatorial. Manni F, GuÃ©rard E, H eyer E (2004) Geographic pattern of (genetic, morphologic, linguistic) variat ion: how barriers can be detect Human Biology , 76 , 173 190 .
252 Marmontel M, TJ , Kochman HI , Humphrey SR (1996) Age d etermination in manatees using growth l ayer group counts in bone. Mar ine Mamm al Sci ence , 12 (1), 54 88. Marsh H TJ, Reynolds JE (2012) Ecology and C onservation of the Sirenia . Cambridge University Press. Mbina C (2001) Evaluation and status of m anatees ( Trichechus senegalensis ) of the Ogooue b asin in Gabon . McClenaghan LR, O'Shea TJ(1988) Genetic v ariability in the Florida m anatee ( Trichechus manatus ). J ournal of Mammalogy , 69 (3), 481 488 . McCue MD, Pollock ED (2008) Stabl e isotopes may provide evidence for starvation in reptiles. Rapid Commun ications in Mass Spectrom etry , 22 , 2307 2314 . McCutchan JH Jr, Lewis WM , Kendall C , McGrath CC (2003) Variation in trophic shift for stable isotope ratios of carbon, nitrogen , and sulfur. Oikos , 102 , 378 390 . McDonald B (2005) Population genetic s of dugongs around Australia: i mplicati ons of gene flow and migra tion. Dissertation, James Cook University, Townsville, Australia . McKenna MC (1975) Toward a phylogenetic classificati on of the Mammalia. In: Phylogeny of the Primates, a M ultidisciplin ary A pproach (eds Luckett WP, Szalay FS ) , Plenum, New York . Moore WS, Smith SM, Prychitko T (1999) Nuclear gene introns versus mitochondrial genes as molecular clocks. In: Proceedings of the 22 nd International Ornithological Congress, Durban (eds Adams NJ, Slotow RH), pp. 745 753 . Johannesburg: BirdLife South Africa . Morai s M (2006) AvailaÃ§Ã£o da condiÃ§Ã£o e d i stribuiÃ§Ã£o do m anatim Africana ( Trichechus senegalensis ) ao Longo do Rio Cuanza. Luanda, Angola . Muller Karger FE , McClain CR, Richardson PL (1988). The dispersal of the A water. Nature , 333 , 56 59 . Mullis KB , Faloona F (1987) Specific synthesis of DNA in vitro via a polymerase catalyzed chain reaction. Me thods in Enzymology , 155 , 335 350 . Murata Y, Nikaido M , Sasaki T , Cao Y , Fukumoto Y , Hasegawa M, Okada N (2003) Afrotherian phylogeny as inferred from complet e mitochondrial genomes. Mol ecular Phylogenetics and Evolution , 28 , 253 260 . Mvele C, Arrowood H (2013) Rapport de vente illegale de la viande de brousse, MarchÃ© Isaac, LambarÃ©nÃ©. Organisation Ecotouristique de Lac OguemouÃ© , Gabon .
253 Naidu A, Fitak RR, Mu nguia Vega A, C ulver M (2012) Novel primers for complete mitochondrial cytochrome b gene sequencing in mammals. Mol ecular Ecol ogy Resources , 12 , 191 196 . Navaza R , Burnham O (1998) Senegal manatees close to extinction . Sirenews , 29 , 7 8 . Nei M (1987) Mol ecular Evolutionary G enetics . Colu mbia University Press, New York . Nelson DL, Cox MM (2008) Lehninger Principles of B iochemistry . WH Freeman and Company , New York . Newsome SD , Clementz MT , Koch PL (2010) Using stable isotope biogeochemistry to study mar ine mammal ecology. Mar ine Mamm al Sci ence , 26 (3), 509 572 . Nicole M, Wadja ME, Schmidt M (1994) A preliminary i nventory of coastal wetlands of CÃ´ te voire. IUCN, Gland Swi t zerland . Nishiwaki M (1984) Current status of the African ma natee. Acta Zoologic a Fennica , 172 , 135 136 . Nishiwaki M, Yamaguchi M , Shokita S , Uchida S , Kataoka T (1982) Recent s urvey on the distribution of the African m anatee. Scientific Report of the Whale Research Institute (Japan) 34 , 137 147 . Noupa P (2008) Cameroon. In: Conserv ation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 87 90. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal Nourisson C, Morales Vela B, Padilla Saldivar J, Tucker KP, Clark AM , Olivera Gomez LD , Bonde R , and McGuire P (2011) Evidence of two genetic clusters of manatees with low genetic diversity in Mexico and implications for their cons ervation. Genetica , 139 , 833 842 . Novacek MJ (1992) Mammali an phylogeny shaking the tree. Nature , 356 (6365), 121 125 . N ovacek MJ (2001) Mammalian phylogeny: genes and supertrees. Current Biology , 11 (14), R 573 R575 . Oboto EA ( 2 002) Manatee Status in River Benue and River Niger (Nigeria). Wetlands International, Dakar, Senegal . GC (2012) The diet b ody o f fset in human nitrogen i sotopic values: A controlled dietary s tudy. American Journal of Physical Anthropology , 149 , 426 434 .
254 Ofori Danson PK (1995) Survey of manatees in Volta Lake. Institute for A quatic Biology, Achimota, Ghana . Ofori Danson PK, Ag bogah K (1995) Oceans and coastal areas (OCA) Programme Activity Centre (PAC) of United Natio ns Environment Programme (UNEP) . Ogogo AU, Eniang EA, Nkamenyin OO (2013) Ecology and c o nservation of the West African m anatee ( Trichechus se negalensis ) in Eniong Creek, South N igeria. International J ournal of Research in Applied, Natural and Social Sciences , 1 (1), 19 24 . Ortiz RM , Worthy GAJ, MacKenzie DS (1998) Osmoregulation in wild and captive West Indian manatees ( Trichechus manatus ). Phy siological Zoology , 71 (4), 449 457 . Hartley WC (1995) Reproduction and early age survival of manatees at Blue Spring, Upper St. Johns River, Florida. In : Population Biology of the Florida M anatee (eds TJ , Ackerman BB , Percival HF) , pp. 1 57 170. U.S. Dept . of the Interior . Ozawa T , Hayashi S, Mikhelson VM (1997) Phylogenetic position of mammoth and emonstrated by mitochondrial DNA sequences. Journal of Molecular Evolution , 44 , 406 413 . Pardini AT Brien PCM , Fu B , Bonde RK , Elder FFB , Ferguson Smith MA , Yang F , Robinson TJ (2007) Chromosome painting among Proboscidea, Hyracoidea and Sirenia: support for Paenungulata (Afrotheria, Mammalia) but not Tethytheria. Proceedings of the Royal Society , 274 , 1 333 1340 . P arnell AC, I nger R , B earhop S, J ackson AL (2010) Source partitioning using stable isotopes: coping with too much variation. PloS One , 5 , e9672. Parr L (2000) The position of Sirenia within the subungulates and comparison of intra and inter sp ecific mitochondrial DNA variation in the extant species of manatees. D issertation, Portland State University. Parr LA, Duffield D A (200 2 ) Interspecific comparison of m itochondrial DNA variation among extant s pecies of Sirenians . In: Molecular and Cell B iology of Marine Mammals (ed Pfeiffer CJ ), pp. 152 158. Krieger Publishing Company, Malabar, FL. Passey BH, Robinson TF, Ayliffe LK, Cerling TE, Sponheimer M, Dearing MD, Roeder BL, Ehleringer JR (2005) Carbon isotope fractionation between diet, breath CO 2, and b ioapatite in different mammals. J ournal of Archaeological Sci ence , 32 (10), 1459 1470 . Pause KC , Nourisson C , Clark A , Kellogg ME , Bonde RK , McGuire PM (2007) Polymorphic microsatellite DNA markers for the Florida manatee ( Trichechus manatus latiro stris ). Mol ecular Ecol ogy Notes , 7 (6), 1073 1076 .
255 Perrin WF (2001) Conservat ion status of the West African m anatee. Sirenews , 36 , 5 13. Peterson BJ, Fry B (1987) Stable isotopes in ecosystem studies. Ann ual Rev ue of Ecol ogical Syst ems , 18 , 293 320 . Phil lips DL , Gregg JW (2001) Uncertainty in source partitioning using stable isotopes. Oecologia , 127 , 171 179 . Phillips DL , Koch PL (2002) I ncorporating concentration dependence i n stable isotope mixing models. Oecologia , 130 (1), 114 125 . Poche R (1973) Nig Oryx , 12 , 216. Posada D, Crandall KA (1998) Modeltest: Testing the model of DNA substitution. Bioinformatics , 14 , 817 818 . Powell JA (1990) Manatees in the Bijagos Archipelago: recommendations for their conser vation. IUCN Wetlands Programme . Powell JA (1996) The distribution a nd biology of the West African m anatee ( Trich echus senegalensis , Link 1795). United Nations Environmental Program, Regional Seas Programme, Ocean and Coastal Areas, Nairobi, Kenya . Rambaut A (2007) FIGTREE v 1.4.0 . http://tree.bio.ed.ac.uk/software/ FIGTREE / Accessed 28 May 2014 . Reep RL, Bonde RK (2006) The Florida Manatee, Biology and Conservation . Gainesvill e, University of Florida Press . Reeves RR, Tuboku Metzger D, Kapindi RA (1988) Distribution and expl oitation of manatees in Sierra Leone. Oryx , 22 (2), 75 84 . Reich KJ (2001) Regional variability in feeding habits of the Florida m anatee ( Trichechus manatus latirostris ): a preliminary assessment u sing 15 N and 13 C. Master s Thesis , Wildlife and Fisheries S ciences, Texas A&M University . Reich KJ, Worthy GAJ (2006) An isotopic assessment of the feeding h abits of free ranging manatees. Marine Ecology Progress Series , 322 , 303 309 . Reid JP (1995) S irenews , 24 , 9 11 . RÃ©publica de Guinea Ecuatorial. (2005) Informe Nacional para la i mplementacion de la convencion de las Naciones Unidas de lucha contra la d esertificacion. National report to UNCC D .
256 Richardson PL, Reverdin G (1987) Seasonal c ycle of v el ocity in the Atlantic Nort h Equatorial Countercurrent as measured by surface drifters, current meters, and ship d rifts. J ournal of Geophysical Research , 92 (C4), 3691 3708 . Robinson PT (1971) Wildli fe trends in L iberia and Sierra Leone. Oryx , 11 , 117 122 . Ron T (1998) Monitoring and Conservation of African Manatees ( Trichechus senegalensis ) in Angola. Project proposal, Luanda, Angola . Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, HÃ¶hna S, Huelsenbeck JP (2012) MrBayes 3.2: efficient Bayesia n phylogenetic inference and model choice across a large model space. Systematic Biology , 61 ( 3) , 539 42. Roth HH, Waitkuwait E ( 1986 ) RÃ©partition et statut des grandes espÃ¨ces de m ammifÃ¨res en CÃ´te d'Ivoire, III Lamantins. Mammalia , 50 , 227 242 . Sakho I , Mesnage V , Deloffre J , Lafite R , Niang I , Faye G (2011 ) The influence of natural and anthropogenic factors on ma ngrove dynamics over 60 years: t he Somone Estuary, Senegal. Estuarine, Coastal and Shelf Science , 94 (1) , 93 101. Salkind JH (1998) Ã‰tude prÃ©liminaire sur les lamantins du Tchad. RÃ©vue S cientifique du Tchad , 17 , 41 50 . Sall I, Durin G (2012) Oceanium Dakar: The daily struggle for the integrated community Field Actions Science Report , Speci al Issue 7 , URL : http://factsreports.revues.org/2564 Sandholt I , Nyborg L , Fog B , LÃ´ M , Bocoum O, Rasmussen K (2003) Remote sensing techniques for flood m onitoring in the S enegal River v alley. Danish Jou rnal of Geography , 103 (1), 71 81 . Sarnthein M, Thie de J, Pflaumann U, Erlenkeuser H, FÃ¼tterer D, Koopman B, Lange H, Seibold E (1982) Atmospheric and oceanic circulation p atterns o ff Northwest Africa during the past 25 million y ears. In: Geology of the Northwest African Continental Margin (eds von Rad U, H in z K, Sarnthein M, Seibold E ) , pp. 545 604 . Springer Verlag, New York. Savage RJG , Domning DP , Thewissen JGM (1994) Fossil Sirenia of the w est Atlantic and Caribbean region V. The most primitive known sirenian, Prorastomus sirenoides , Owen 1855. J ournal of Verteb r ate Paleontol ogy , 14 , 427 449 . Schoeninger MJ, DeNiro MJ(1984) Nitrogen and carbon isotopic composition of bone collagen from marine and terrestrial animals. Geochimica et Cosmochimica Acta , 48 , 625 639 .
257 Schuhmann HJ (1995) Der Manati, Trichech us senegalensis , im Rio Geba, Guinea Bissau. Natur und Museum , 125 , 402 409 . Segniagbeto HG, Akpamou KG, TrÃ©vÃ© TK, Alassani D, Kotchikpa O (2008) Togo. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 63 71. UNEP , Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Short FT, Carruthers TJR, Waycott M , Kendrick GA, Fourqurean JW, Callabine A, Kenworthy WJ, Dennison WC (2010) Ruppia maritima . In: IUCN 2013. IUCN Red List of Threatened Species. Version 2013.2. < www.iucnredlist.org >. Downloaded on 07 May 2014 . Siaffa DD, Jalloh A (2006) C onservation of the West Africa m anatee Trichechus senegalensis along the West African coastal z one. Baseline survey on the W est Africa Manatee along the Sierra Leone Coastal Areas, February 2006. Wetlands International, Dakar , Senegal. Siaffa DD, Jalloh A (2008) Sierra Leone. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 47 50. UNE P, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Silva MA, AraÃºjo A (2001) Distribution and current status of the West African manatee ( Trichechus senegalensis ) in Guinea Bissau. Mar ine Mamm al Sci ence , 17 (2), 418 424 . Simpson GG (1945 ) The principle of classification and a classification of m ammals. Bulletin of the American Museum of Natural History , 85 , 1 350 . Southern SO , Southern PJ , Dizon AE (1988) Molecular characterization of a clone d dolphin mitochondrial genome. J ournal of Mol ecular Evol ution , 28 , 32 42 . S ponheimer M, Robinson T, Ayliffe L, Roeder B, Hammer J, Passey B, West A, Cerling T, D earing D, E hleringer J (2003) Nitrogen isotopes in m ammalian h erbivores: hair 15 N v alues from a c ontroll ed feeding s tudy. International J ournal of Osteoarchaeology , 13 , 80 87 . Sprent JFA (1983) Ascaridoid nematodes of sirenians a new sp ecies in the Senegal manatee. J ournal of Helminthology , 57 , 69 76 . Springer MS , Amrine HM , Burk A , Stanhope MJ(1999) Additional support for Afrotheria and Paenungulata, the performance of mitochondrial versus nuclear genes, and the impact of data partitions with heterogeneous base composition. Systematic Biol ogy , 48 (1), 65 75 . Sun JX, Mullikin JC, Patterson N, R eich DE (2011) Microsatellites are molecular clocks that support a ccurate i nferences about h istory. Molecular Biology and Evolution , 26 (5), 1017 1027 .
258 Swofford DL (2002) PAUP* Phylogenetic Analysis Using P arsimony (*and other methods). v.4 . Sinauer Associates, Sunderland , MA. Sykes S (1974) How to save the m ermaids. Oryx , 12 , 465 470 . Sykes S (2010) ousins . M irage Newberry, United Kingdom . Tabuce R , Marivaux L , Adaci M , Bensalah M , Hartenberger JL , Mahboubi M , Mebrouk F , Tafforeau P , Ja eger JJ (2007) Early Tertiary mammals from North Africa reinforce the molecular Afrotheria clade. Proceedings of the Royal Society , 274 , 1159 1166 . Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetic s , 123 ( 3), 585 595 . Takoukam AK, (2012) Activity center, h abitat use and c o nservation of the West African m anatee ( Trichechus senegalensis Link, 1795) in the Doua la Edea and Lake Ossa wildlife r eserves. Master s Thesis, University of Dschang, Cameroon . T akoukam AK, Ayissi I , Ngafack PR , Njanjouo GN , Nana A, Keith Diagne L , Collins T , Van Waerebeek K (2013) Assessing the distribution, bycatch, and strandings of marine mammals in the Cameroon coastal areas. 20 th Biennial Conference on the Biology of Marine Mammals, Dune din, New Zealand, December 2013 . 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 , 678 687 . Ta mura K, Stecher G, Pet erson D, Filipski A , Kumar S ( 2013 ) MEGA6: Molecular Evolutionary Genetics Analysis version 6.0. Molecular Biology and Evolution , 30 , 2725 2729 . Tchibozo S (2002) Etude pour la prospection des dernieres populations de lamantins ( Trichechus senegalensis ) d nin Minister of the Environment . ThÃ©ophile R, Ibrahima C, K poghomou CN, Diallo A, Dabo A, Bangoura CAK (2008) Guinea. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , p p. 41 46. UNEP, Nairobi, Kenya and Wetlands International Africa, Dakar, Senegal . Timbo S (2010) Necropsie lamantin a Ko ulikoro. Niger River Basin Authority, Bamako, Mali .
259 Trimble MJ, Van Aarde RJ (2010) Species i nequalit y in scientific s tudy. Conserv ati on Biol ogy , 24 , 886 890 . Tringali M , Davis MC , Rodriguez Lopez MA , Bolen EE , Sullivan JG, Haubold EM (2008a) Simultaneous use of the X and Y chromosome genes SMCX, SMCY, and DBY for sex determination in the Florida manatee ( Trichechus manatus latirostris ). Mar ine Mamm al Sci ence , 24 (1), 218 224 . Tringali MD , Seyoum S , Carney SL , Davis MC , Rodriguez Lopez MA , Reynolds JE, Haubold E (2008b) Eighteen new polymorphic microsatellite markers for the endangered Florida manatee, Trichechus manatus latirostris . Mo l ecular Ecol ogy Resources , 8 , 328 331 . Urbano DF , Jochum M , da Silveira ICA (2006) Rediscovering the second core of the Atlantic NECC. Ocean Modeling , 12 , 1 15 . van d er Winden J, Siaka A (2005) The West African m anatee in Sierra Leone. Sirenews , 43 , 15 . VÃ©lez Juarbe J (2014) Ghost of seagrasses past: Using sirenians as a proxy for historical distribution of seagrasses . Palaeogeography, Palaeo climatology, Palaeoecology , 400 , 41 49 . V Ã© lez Juarbe J , Domning DP, Pyenson ND (2012) Iterative evolution of symp atri c seacow (Dugongidae, Sirenia) a ssemblages during t he past ~ 26 million years. PloS One , 7 (2 ) , e 31294 . Vianna JA, Bonde RK, Caballero S , Giraldo JP , Lima RP , Clark A , Marmontel M , Morales Vela B , De Souza MJ , Parr L , Rodriguez Lopez MA , Mignucci Gian noni AA , Powell JA, Santos FR (2006) Phylogeography, phylogeny and h ybridiz ation in trichechid s irenians: i mplica tions for manatee c onservation. Mol ecular Ecol ogy , 15 (2), 433 447 . Villiers A, Bessac H (1948) Le l amantin du S enegal. La Nature , 3158 , 188 18 9. Walker JL, Macko SA (1999) Dietary studies of marine mammals using stable carbon and nit rogen isotopic ratios of teeth. Mar ine Mamm al Sci ence , 15 (2), 314 334 . Whi te JR , Harkness DR , Isaacks RE , Duffield DA ( 1976) Some studies on blood of the Florida manatee, Trichechus manatus latirostris . Comparative Biochemis try and Physiology , 55A , 413 417 . Wiles DL, Makor J (2008) Liberia. In: Conservation strategy for the West African manatee (eds Dodman T, Diop MD, Sarr K ) , pp. 51 52. UNEP, Nairobi, Kenya and W etlands Inter national Africa, Dakar, Senegal . Xavier M (2011) Report of the Lake CaÃºmba m anatee. Luanda, Angola .
260 Xu J, Yang Q, et al (2011) Preservation effects on stable isotope ratios and consequences for the reconstruction of energetic pathways. Aquat ic Ecology , 45 (4), 483 492 .
261 BIOGRAPHICAL SKETCH Lucy Ward Keith Diagne was born in Gle n Ridge, New Jersey in 1965 . She grew up in New Vernon, New Jersey and attended Harding Township School followed by Kent Place School in Summit, New Jersey , where she gr aduated in 1983. Lucy earned her Bachelor of Science degree in b iology from St. Lawrence University in 1987. During her senior year at St. Lawrence, Lucy was introduced to radio telemetry by her advisor, Biology professor Dr. John Green, and she conducted a VHF tracking study of the Eastern Coyote on Barnhar t Island in the St. Lawrence Seaway. In the following years, Lucy held jobs as a n Aquarist and Veterinary Services Technici an at the New England Aquarium i n Boston, Massachusetts , as a penguin and seabir d researcher for a National Science Foundation grantee at King George Island and Palmer Station, Antarctica, and four years as a field research team leader for the National Marine Fisheries Service Hawaiian Monk Seal Project in the Northwestern Hawaiian Is lands. From 1997 1999 Lucy complet ed her Master of S cience degree in m arine b iology from the Boston University Marine Program in Woods Hole, Massachusetts . Her thesis research used VHF radio telemetry to examine the haul out behavior of we aned pup and juve nile Hawaiian monk s eals at Midway Atoll. In 1998 Lucy was hired by the Florida Fish and Wildlife Conservation Commission to run their Florida manatee field station that cover s ten counties in southwest Florida. During her six years in that position , Lucy and her staff recovered approximately 42% of the statewide total in manatee carcasses annually , rescued over 10 0 sic k and injured manatees, and conducted a manatee winter habitat use study at Warm Mineral Springs, in North P o rt, Florida . In 2003 Lucy recei ved the Manatee Conservation Award from the U.S. Fish and Wildlife Service.
262 In 2004 Lucy took a research position at the University of New England in Biddeford, Maine to conduct satellite telemetry on Harbor seals , but i n 2005 she returned to Florida and began working with the non profit research organization Wildlife Trust to research Florida manatees using satellite telemetry and aerial surveys. In this position she also made trips to Belize and Costa Rica to assist with Antillean manatee captures and t racking. I n September 2006 , Lucy made her first trip to Gabon, Central Africa and began manatee surveys throughout the country , which continued for much of the next seven years. During this time Lucy began writing grants to raise funds to continue her work in Africa. From 2007 2009 Lucy also conducted manatee surveys of the lower Congo River in Angola. In 2008 and 2009 Lucy and Dr. Patrick Ofori Dansen of the University of Ghana co taught manatee research training workshops for African biologists from 17 co untries at Lake Volta, Ghana. In 2009 Lucy traveled to Senegal to participate in the rescue of manatees trapped behind agricultural dams and to attach satellite telemetry tags to three manatees to track their movements post release. This was the first sate llit e tracking study of the African manatee . The same year Lucy began her doctoral studies at the University of Florida. From 2010 through 2013 Lucy conducted African manatee research training workshops in Mali, the Gambia, and Gabon as well as continu ing fieldwork in Gabon, Senegal, Republic of the Congo, Guinea Bissau and the Gambia. S he has now trained over 72 African biologists in manatee research techniques and has supplied basic field equipment to researchers in 11 countries. During these workshops Lucy realized the only effective way to study and conserve the African manatee across its enormous 21 country range wa s to build a collaborative network for African manatee researchers in order to inc rease research
263 efforts for the sp ecies, accurately document threats, share information , increase educati onal outreach, and communicate results. She began this network in 2008 and now has participating researchers affiliated with government agencies, universities and NGOs in 1 9 African cou ntries . Year round African manatee research now occurs in nine countries as a direct result of In 2010 Lucy married Senegalese turtle and manatee researcher Tomas Diagne. They are based in Gainesville, Florida and Popenguine , S enegal. Lucy continues her work in Central and West Africa , and is developing long term study sites in the Senegal River, Lac de Guiers , and th e Casamance River in Senegal. She plans to continue to study and work towards conservation of the African manatee for the rest of her career.