1 RELATEDNESS, INBREEDING, AND KINSHIP OF THE ENDANGERED FLORIDA MANATEE ( TRICHECHUS MANATUS LATIROSTRIS ) By MICHELLE CHRISTINE DAVIS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
2 Â© 2014 Michelle Christine Davis
3 To my parents, for always believing that I can do anything
4 ACKNOWLEDGMENTS I would like to thank my committee for believing in me and helping me achieve my goal of making a difference in an endangered species. Dr. Bob Bonde was the catalyst for my acceptance to the University of Florida. His encouragement and re commendation is the reason that I am here today. Dr. Margaret Hunter has provided me the foundation needed to persist and make it through any challenge. I am grateful for her guidance and expertise throughout this process. I am appreciative of all Bob a nd Maggie have done for me as my co advisors. Their constant patience, endless revisions, and the occasional need to reign me in has helped mold me into the scientist I am today. Dr. Mike Tringali was the first of my mentors and the main inspiration for me to continue my education. His encouragement and guidance did more than convince me to pursue a PhD; it enabled me to excel in everything I do. I will forever be grateful for his role in my life. Dr. Roger Reep took me in as one of his manatee student s and taught me the ways of graduate school, especially my first year. His encouraging words and suggestions were always the motivation I needed . I want to thank Dr. David Ostrov for his enthusiasm and willingness to join my committee and I look forward to incorporating the MHC work into Florida manatee research in the future. To my USGS family, Bob Bonde, Maggie Hunter, Cathy Beck, Gaia Meigs Friend, Susan Butler, Amy Teague, and Jim Reid, thank you so much for welcoming me into the Sirenia Project fam ily. I especially want to thank Cathy Beck for all of her help, guidance, and expertise. A special thank you to Dr. Dan Slone for helping me with some statistical analyses. To my fellow lab techs, John Butterfield and Theresa Floyd, thank you for helpin g me hone my teaching skills and making me feel like an expert in the lab.
5 To my FWC family, Mike Tringali, Alicia Alvarez, Seifu Seyoum, Marta Rodriguez Lopez, Jamie Galvin, Stacie Koslovsky, Margie Barlas, Jen Johnson, Cecilia Puchulutegui, and Kathy G uindon, you were my introduction to genetics. You not only showed me the way that science should be done, but you gave me the foundation and courage to continue on this path. Thank you all for your help, support, and friendship through the years. Marta, you introduced me to the manatee world and have encouraged and guided me every step of the way. To Seifu who taught me so much about genetics, thank you for your constant support and advice. Alicia, I am honored to continue to work with you and have you as a manatee genetics colleague. Stacie, thank you so much for saving me hours of sorting thorough data by answering all of my data request emails. Mammal Section, Genetics Lab , and Marine Mammal Pathobiology Lab for allowing me to use the manatee genotype and carcass recovery information. I am grateful to have met and worked with these fellow manatee researchers whom I now consider friends, Anmari Aleman, Coralie Nourisson, a nd Kayla DaCosta. here at UF. I also want to thank Dr. Charles Courtney, Dr. Ruth Francis Floyd, Dr. Iske Larkin, Heather Maness, Dr. Mike Walsh, Dr. Tom Waltzek, Dr. Ammon Pec k and everyone who is part of the Aquatic Animal Health Program at UF, for accepting and funding me through the program. I am thankful to Dr. Jim Austin for allowing me to t o John Hargrove and Matt Shirley for all your help, advice, and witty remarks. I am
6 also appreciative of the Graduate Editorial Office for helping me format my dissertation, allowing for much easier first and final submissions. A big thank you to my fel low (and former) UF graduate students and friends Leighann Skurupey, Lauren Harshaw, Noel Takeuchi, Jason Ferrante, and Lucy Keith Diagne for the needed lunch breaks, venting sessions, and overall encouragement. I could not have done it without you. Jacl yn Irwin and Vicki Villanova (my minion), thank you for showing me what it means to be a Florida Gator. To my friends, old and new, Kerry Jackson, Rachel Hatch, Ben Swartz, Marisa Kushner, Chris Sewell, Reggie Chaple, Sr. Kelly Francis, Darlene Saindon, D ave Christian, Dana Overbaugh, Erika Aenlle, Katie Martin, and Sara Martin, your constant support and friendship has meant the world to me. Thank you for always being there when I needed someone. No one is more blessed then I am to have a family as wond erful and supportive as the one I have. Though they may not understand what it is that I do, they are always there with a smile, hug, and a listening ear. Thank yo u Davis, Brandt Rizzi, Prosperi , and Lagasse families for your encouragement and love. Tha nk you especially to my Grandma for the prayers, love, and faith in all that I do. I also thank Darwin and Julius for their love and support during those long nights of analyzing and writing. To my parents, who have been on th is four year journey with me, t hank you for instilling in me the strength of character to never give up, the confidence to know that I cannot fail, and the reassurance to know that you will always be there for me. You gave me a solid foundation of intelligence, creativity, faith, lov e, sensitivity and wit that has served me well. You will never know how grateful I am to have such loving and
7 inspiring parents. You are the best parents and friends a girl could ever want. I love you all.
8 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 11 LIST OF FIGURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 Life History of the Manatee ................................ ................................ ..................... 15 The Florida Manatee ................................ ................................ ........................ 15 Family Unit and Kinship ................................ ................................ .................... 16 Manatee Identification Studies ................................ ................................ ......... 20 Manatee Genetic Studies ................................ ................................ ................. 21 Manatee Conservation Genetics ................................ ................................ ............. 21 Parentage and Pedigree Analysis ................................ ................................ ........... 49 Manatee Pedigree Studies ................................ ................................ ............... 51 Pedi gree Statistical Analysis ................................ ................................ ............ 54 Conservation Management Through the Use of Pedigree Studies ................... 57 2 ASSESSMENT OF LATENT COANCESTRY AND INBREEDING IN WILD FLORIDA MANATEES ................................ ................................ ............................ 68 Background ................................ ................................ ................................ ............. 68 Materials and Meth ods ................................ ................................ ............................ 73 Sample History, Collection, and DNA Extraction ................................ .............. 73 Microsatellite DNA Analysis ................................ ................................ .............. 74 Statistical Analysis ................................ ................................ ............................ 75 Results ................................ ................................ ................................ .................... 77 Florida ................................ ................................ ................................ .............. 77 By Coast ................................ ................................ ................................ ........... 79 Management Units ................................ ................................ ........................... 80 Differentiation Values ................................ ................................ ....................... 81 Discussion ................................ ................................ ................................ .............. 83 Informativeness and Power ................................ ................................ .............. 83 Relatedness ................................ ................................ ................................ ..... 85 Inbreeding ................................ ................................ ................................ ........ 86 Genetic Differences ................................ ................................ .......................... 87 Summary ................................ ................................ ................................ .......... 90
9 3 RELATEDNESS AND INBREEDING UTILIZING CAPTIVE FLORIDA MANATEE FAMILY UNITS ................................ ................................ ..................... 96 Background ................................ ................................ ................................ ............. 96 The Florida Manate e ................................ ................................ ........................ 96 Investigations of Relatedness ................................ ................................ ........... 98 Materials and Methods ................................ ................................ ............................ 99 Sample History, Collection, and DNA Extraction ................................ .............. 99 Study Locations ................................ ................................ .............................. 101 Microsatellite DNA Analysis ................................ ................................ ............ 101 Statistical Analysis ................................ ................................ .......................... 102 Parentage Analysis ................................ ................................ ........................ 103 Categorical Relationship Estimators ( R xy ) ................................ ................ 103 Relatedness Estimators ( r xy ) ................................ ................................ .... 104 Results ................................ ................................ ................................ .................. 105 Genetic Marker Assessment ................................ ................................ .......... 105 Re latedness and Relationship Estimates ................................ ....................... 106 Family Units ................................ ................................ ................................ .... 107 Expected Values of Relatedness ................................ ................................ .... 109 Case Studies ................................ ................................ ................................ .. 109 Discussion ................................ ................................ ................................ ............ 110 Estimators ................................ ................................ ................................ ...... 110 Romeo and Juliet ................................ ................................ ........................... 112 Ocean Reef and Foster ................................ ................................ .................. 112 Impacts of Latent Coancestry ................................ ................................ ......... 113 Wild Case Studies ................................ ................................ .......................... 115 Summary ................................ ................................ ................................ ........ 116 4 PARENTAGE ANALYSIS OF WILD MANATEES AT A LONG TERM STUDY SITE: CRYSTAL RIVER, FLORIDA ................................ ................................ ...... 123 Background ................................ ................................ ................................ ........... 123 Materials and Me thods ................................ ................................ .......................... 128 Sample History, Collection, and DNA Extraction ................................ ............ 128 Microsatellite DNA Analysis ................................ ................................ ............ 129 Statistical Analysis ................................ ................................ .......................... 130 Results ................................ ................................ ................................ .................. 133 Genetic Marker Assessment ................................ ................................ .......... 133 Maternity ................................ ................................ ................................ ......... 134 Paternity ................................ ................................ ................................ ......... 136 Discussion ................................ ................................ ................................ ............ 138 Maternity ................................ ................................ ................................ ......... 138 Paternity ................................ ................................ ................................ ......... 141 Summary ................................ ................................ ................................ ........ 145 5 CONCLUSIONS AND FUTURE DIRECTIONS ................................ .................... 153
10 Summary of Study ................................ ................................ ................................ 153 Future Dir ections ................................ ................................ ................................ .. 155 APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 2 ................................ .............. 162 West Coast ................................ ................................ ................................ ..... 162 Northwest ................................ ................................ ................................ ....... 163 Southwest ................................ ................................ ................................ ....... 163 East Coast ................................ ................................ ................................ ...... 167 Atlantic ................................ ................................ ................................ ............ 167 Upper St Johns ................................ ................................ ............................... 168 B SUPPORTING INFORMATION FOR CHAPTER 3 ................................ .............. 171 C SUPPORTING INFORMATION FOR CHAPTER 4 ................................ .............. 178 LIST OF REFERENCES ................................ ................................ ............................. 184 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 206
11 LIST OF TABLES Table page 1 1 Mitochondrial DNA sequence divergence comparison of three Trichechus species in six studies ................................ ................................ .......................... 65 2 1 Polymorphic microsatellite mar kers with manatee population diversity information for all of Florida ................................ ................................ ................ 92 2 2 Comparisons of the Florida manatee population and subpopulations based on differe nt microsatellite marker panels ................................ ............................ 93 2 3 Power of inference to distinguish between relationship dyads (PW R ) ................. 94 3 1 Life history information of the captive manatees used in this study. ................. 119 3 2 Expected relatedness and inbreeding values ................................ ................... 122 4 1 Presumed relationships with unexpected outcomes of calves 235cm in length ................................ ................................ ................................ ................ 146 4 2 The 21 maternal families with reconstructed and inferred paternity using thr ee parentage assignment methods ................................ .............................. 148 A 1 Statistical characterization of Florida manatee polymorphic microsatellite markers for the West Coast (WC) and management units No rthwest (NW) and Southwest (SW) ................................ ................................ ......................... 165 A 2 Statistical characterization of Florida manatee polymorphic microsatellite markers for the East Coast (EC) and management units Atlantic (ATL) and Upper S ................................ ................................ ........ 169 B 1 All pairwise relationships from captive pedigree dataset of Chapter 3 .............. 172 C 1 Cow Calf pairs from Crystal River, Florida used in Chapter 4 .......................... 178
12 LIST OF FIGURES Figure page 1 1 Manatees gather at Cryst al River, Florida in the winter ................................ ...... 64 2 1 Mean difference between gro ups ................................ ................................ ....... 95 3 1 Captive pedigree diagram ................................ ................................ ................ 118 3 2 Comparison of three relationship assignment estimators (Coance stry, ML Relate, and ML Relate Prior) to the relationships based on the North American Regional West Indian Manatee Studbook ................................ ........ 120 3 3 Expected relatedness values compared to Coancestry (TrioML and Wang) and ML Relate estimators. ................................ ................................ ............... 121 4 1 Probability of related ness by pairs. ................................ ................................ ... 147 4 2 Number of bull s siring between 1 and 9 calves ................................ ................ 152 A 1 Causes of death for the Florida manatee from 1974 2012. .............................. 162
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RELATEDNESS, INBREEDING, AND KINSHIP OF THE ENDANGERED FLORIDA MANATEE ( TRICHECHUS MANATUS LATIROSTRIS ) By Michelle Christine Davis August 2014 Chair: Robert Bonde Cochair: Margaret Hunter Major: Veterinary Medical Sciences The Florida manatee is an endangered marine mammal that lives in the coastal waters of the southeastern United States. Population sizes, food sources, habitat use, and individual life history have been assessed through photo identification, telemetry, and other behavioral studies over the past 40 years. Genetic studies have provided evidence of one mitochondrial haplotype, little genetic divergence between Florida subpopulations, and very low genetic diversity, suggesting a population vulnerable to natural and anthropogenic effects, including algal toxins, increased watercraft, habitat change, and loss of warm water sources. To better understand the health and fitness of the Florida manatee population, overall relatedness and individual inbreeding coefficients were quantified using 36 polymorphic microsatellite markers. Robust estimates of pow er to distinguish between higher order relationships and unrelated individuals was demonstrated, as was the ability to perform individual identification and detect parentage . Estimates revealed that Florida manatees are related ( r =0.115) and inbred ( F x =0. 108) at a level corresponding to first cousins. Pairwise comparisons dictate that 20 26% of pairs in the population are
14 one quarter related ; a substantially higher assessment than would be expected in a healthy, randomly mating population. To guide parent age and kinship studies of natural populations , captive manatee pedigrees, cow calf pairs, and life history data were assessed with various estimators. Categorical r elationship estimators more accurately distinguish ed between related and non related individuals than relatedness estimators. However, due to the significant amount of latent coancestry and high inbreeding levels found in the Florida manatee population, relationships between types of siblings and half sibl ings and non related individuals were difficult to assign. Finally, known relationships were assessed and paternity was inferred in 72 calves sampled over 19 years. Maternity was verified with an 80% success rate. Paternity was inferred in 83 males, with relatively few males (6.6%) assigned to a large number of offspring. This supports the hypothesis that the population is highly related with a limited gene pool. Population decline and low diversity can result in inbreeding depression and a cyclical red uction in fecundity. As new threats impact the population, genetic monitoring of relatedness and inbreeding levels are needed to ensure the survival of this endangered species.
15 CHAPTER 1 INTRODUCTION Life History of the Manatee Manatees are of the Order Sirenia, which are found world wide in tropical to subtropical environments. The order contains three extant species of manatees , one dugong ( Dugong dugon Hydrodamalis gigas ) . T he West Indian manatee, Trichechus manatus, is separated into two subspecies, T. m. latirostris and T. m. manatus . T he Florida manatee , T. m. latirostris , occupies both coasts of Florida . T he Antillean manatee, T. m. manatus , is distributed throughout the Caribbean and along the Central and South American coastline from Mexico to Brazil. T he Amazonian manatee, T. inunguis, is restricted to the Amazon River in Colombia, Ecuado r, Peru and Brazil. T he African manatee, T. senegalensis , has been documented in 21 countries of Africa, ranging along the west coast and inland from Senegal to Angola. All three species of manatees are considered vulnerable by the International Union fo r Conservation of Nature (IUCN) Red List of Threatened Species, with population status ranging from stable in the Florida manatee to decreasing in the Antillean and Amazonian manatees, to unknown for the West African manatee (IUCN 2012) , therefore making t he Trichechidae family w orthy of conservation efforts. The Florida Manatee The Florida manatee is an endangered aquatic mammal that exists in the coastal waters of the southeastern United States (Lefebvre & O'Shea 1995) . As a typical K selected breeder ( MacArthur & Wilson 1967) , manatees are very large reaching lengths of 2.7 3.0m (8.9 10 ft.) and weights of 400 550kg (882 1213 lbs.), and a can have long life spans, greater than 60 years in captivity (Reep & Bonde 2006) . They are generalist
16 herbivores an d typically consume 5 10% of their body weight per day from a large variety of aquatic plants and seagrasses. They have a very slow metabolism and are affected by prolonged exposure to cold, thereby seeking warm water temperatures of 20 23Â°C (68 74Â°F) in winter in order to maintain the most effective thermoneutral zone (Blair Irvine 1983; Worthy et al. 1999; Reep & Bonde 2006) . I n winter months , manatees occupy warm water locations such as C rystal R iver, Blue S pring, or so me of the many power plants located throughout Florida (Figure 1 1 ). Manatees are mostly solitary animals that aggregate in large groups at warm water sites in the winter. They typically do not form pods or clans as is common in other marine mammals, alt hough they are found in large breeding groups when mating . Manatees do not form monogamous mating pairs. Often, reproduction takes place in mating herds, involving one focal, receptive female and multiple, upwards of 40, males (Reynolds & Rommel 1999; Re ep & Bonde 2006) . These mating herds can last 3 4 weeks and are very fluid, with males joining and leaving the herd continuously (Rathbun et al. 1995; Larkin et al. 2005; Reep & Bonde 2006) . Manatees reach sexual maturity between 3 5 years of age, have a n approximate one year gestation period , and tend to calve every three years. Manatees usually have only one calf at a time, and although rare, twins do occur (1.5%). They also have an extended calf dependency period (1 3 years). Family Unit and Kinship offspring , to a unit of socialization, or even community or region (Parsons 1943; Pattison et al. 1975) . In science, family unit can refer t o a group of closely related individuals and specifically within the study of animals,
17 family units are commonly referred to as packs, pods, prides, colonies, clans, or herds , just to name a few. In some animal species, such as whales, elephants, and bees , family units are made up of matrilines, which are groups containing a head female , her female relatives , and all of their offspring (Richard et al. 1996; McComb et al. 2001; Danforth 2002; Yurk et al. 2002) . Male offspring will usually stay with the gro up until reaching sexually maturity and then disperse , which may be an avoidance mechanism for inbreeding, while adult males may join the group for a short time to mate (Dobson & Jones 1985; Nyakaana & Arctander 1999) . Other forms of family units are observed in wolves, birds, and humans a s the nuclear family, consisting of parents and their offspring, while solitary animals us ually consist of the mother and her offspring (Kleiman 1977; Greenwood 1980; Asa & Valdespino 1998) . In family units, the pare ntal figure teaches the offspring how to survive, where and how to find food and water, and how to avoid threats (Asa & Valdespino 1998; Kokko et al. 2002; ConstÃ¢ncia et al. 2004) . Genealogy is the study of families and their lineage, using history, gene tics, and other biological information to establish kinship and pedigrees (Kingman 1982; Newman et al. 2001) . Within pedigrees, the family unit consists of all related individuals depicted pictorially with connections as to how they are related (Frankham et al. 2002; Allendorf et al. 2012) . When establishing pedigrees where one or both parents are unknown, the family unit can be described as the offspring, known parent, and inferred parent based on genetic parentage analyses (de Castro et al. 2013; De Lor enzis et al. 2013) . Manatee family groups usually consist of a female and her calf, known as a cow calf pair. The calf dependency period is 1 4 years, with strong mother calf behavioral bonds (Rathbun et al. 1995; Reynolds & Rommel 1999; Reep & Bonde 200 6) . Learned
18 traditions are passed down from mother to calf and can be maintained in manatees through their strong site fidelity, where known individuals are detected in the same relative area year after year and sometimes for generations (Greenwood 1980; Deutsch et al. 2003; Reep & Bonde 2006) . Other traditional learning include s migration routes as well as the locati ons of fresh water sources, sea grass beds, and warm water in winter (Reynolds & Rommel 1999; Reep & Bonde 2006) . M al who pass it on to their calves. L arger abstract family units of matrilines have been observed through photo identification and can be constructed through genetic analysis (Beck & Reid 1995; Rathbun et al. 1995; Reid et al. 1995) . P aternity is unknown in the wild due to a lack of paternal care. W hen pedigree reconstruction takes place on wild manatees, the family unit will be defined as the known cow calf pair and the potential father(s) , detected by genetic analysis. Ki nship is a term that can be defined many ways. Biologically, this term refers to how related genetic ally analyzed individuals are to each other . It is represented by the coefficient of relatedness ( r) which ranges from zero in unrelated to one in monozygotic twins (Wright 1922) . Anthropologically, the word kinship refers to the social interactions and relationships between individuals and groups (Carsten 2000) . It can include both relationship by descent or by marriage , as well as whom one feels an affinity towards (i.e. a community, group , or friends) (LÃ©vi Strauss 1969) . It is a social term and used to maintain and identify groups of individuals. Kinship can also refer to network of social relationships (LÃ©vi Strauss 1969; Car sten 2000) . For example, adopted children may not be tend to develop a kinship or social affinity towards them . These feelings are usually stronger than towards
19 their biological parents whom they do not know. In a social sense, kinship has more to do with those who m you associate ( ha ve taken care of you ), and how you identify a strong social connection with them . This can be observed with animals that imprint (Klopfer et al. 1964) . For example, when geese fi rst hatch, they imprint on the first large moving thing they see (Hollis et al. 1991) . E ven though there is not a genetic relationship or sometimes a species relationship, the individuals will feel kinship towards what they imprinted upon early in the dev elopmental stages (Dyer & Gottlieb 1990) . Based on the current understanding of manatee population status, manatees do not portray many behaviors indicative of a strong relational kinship. They do form close mother calf bonds that last for 1 4 years, wh ere cow and calf are inseparable, but there is little evidence that most calves associate with their mothers after this time frame (Rathbun et al. 1995; Reynolds & Rommel 1999; Reep & Bonde 2006) . Female manatees rescued orphan calf and treat them as their own, but this is a rare behavior (Bonde 2012 pers. comm.) . Lactating females in the wild have been seen nursing multiple calves, which could be a case of twins, or could be an example of a societal kinship allow ing nursing of related individuals (Beck 2013 pers. comm.) . It has been reported t hat manatees have rubbing posts where rocks or stumps are used by manatees to rub their face, axilla, or genital areas, possibly marking it with a scent (Reep & Bonde 2006) . They have also been observed interacting with others through behaviors best described as kissing (touching muzzle to muzzle during greeting) and male groups have been detected displaying behaviors that are characteristic of play , fighting , or sexual are in nature (Reep & Bonde 2006) .
20 Generally speaking manatees may be showing signs of loose kinship behavior, but it is difficult to determine accurate motive and more study may be warranted in this area. Manatee Identification Studies Photo based mark res ighting studies have been conducted for the last 35 years by the U.S. Geological Survey Sirenia Project, the Florida Fish and Wildlife Conservation Commission, and Mote Marine Laboratory (Kendall et al. 2004; Langtimm et al. 2004) . Using scar patterns tha t manatees have acquired from boat propellers, entanglements , and/or infections or lesions, scientists are able to identify individual manatees and collect life history data on each. The goals of these studies a re to estimate movement patterns , reproducti ve rates , survival rates , and population growth rates of the Florida manatee (Beck & Reid 1995; Tringali et al. 2008b) . Difficulties arise when photo based identifications are not possible, such as with non scarred individuals or during poor photographic conditions. Assuming samples can be acquired, these problems are not encountered in a genetic mark recapture study, which could be used to assist photo ID by identifying more individuals. Genetic sampling of free swimming manatees has been in effect for t he last 25 years in Crystal River, Florida (Bonde 2009) and state wide for the last seven years (Carney et al. 2007) . Owing to the extended mother calf dependency period observed in manatees (Rathbun et al. 1995; Boyd et al. 1999) , it is relatively easy to learn about maternal lineages through both photo identification and genetic sampling. However, very little information is known about the paternity of these calves, because there is no way to identify the father.
21 Manatee Genet ic Studies Manatee populations throughout the world have been studied genetically for the last 40 years, with the majority of these studies focused on the West Indian manatee. Manatees in general have low genetic diversity, with some individual population s (e.g., Belize) having diversity levels that rival those of critically endangered species, such as the North Atlantic right whale (Frasier et al. 2007) . The Florida manatee has only one reported mitochondrial DNA (mtDNA) haplotype, indicative of either a population bottleneck or a founder event, and an average heterozygosity of 0.47, which is lower than that reported for demographically challenged populations (approximately 0.5) (Garcia Rodriguez et al. 1998; Garner et al. 2005; Vianna et al. 2006) . The Antillean manatee populations of Mexico and Belize both have even lower levels of heterozygosity, although both populations have at least three mtDNA haplotypes. The Amazonian manatee has very low levels of mtDNA genetic diversity. Multiple studies using samples from throughout the range of manatees have detected a number of haplotypes, but on the other hand nucleotide diversity was an order of magnitude less than T. manatus . This indicates a population with a large range yet low diversity, which may war rant stringent conservation management practices. Manatee Conservation Genetics Conservation genetics is the application of genetic methods to protect and restore biodiversity. Some methods implemented include: cytogenetics, the study of chromosomes; p hylogenetics, the study and classification of organisms based on their evolutionary relationships; phylogeography, the study of the genetic processes that determine geographic distribu tion based on genetic diversity; and population genetics, the study of g enetic diversity within and between populations (Rudin 1997; Beebee &
22 Rowe 2008) . In the last four decades, a multitude of conservation genetic studies have been conducted on manatees, with a n additional four graduate these s and dissertation chapters pert aining specifically to pedigree studies. Herein, will be a summary of the current knowledge of manatee conservation genetics, reviewing what is known about manatee pedigree analysis and what can be done to improve the method. Finally, the enrichment of c onservation management decisions using manatee pedigree studies will be discussed . In the 19 that described the karyotypes, or the characterization of manatee chromosomes (Loughman et al. 1970; White et al. 1976) . Chromosome studies can help to differentiate species based on diversity, by examining number, shape , or banding patterns within the chromosomes (Kellogg 2008) . Loughman et al . (1970) established that the male Amazonian manatee has a diploid number of 56 chromosomes, while White et al . (1976) deter mined that the Florida manatee has a diploid number of 48 chromosomes, pro viding evidence that these are two separate species. T. i nunguis cytogenetics was conducted using specimens from both Brazil and Colombia (Assis et al. 1988) . They confirmed that both male and female Amazonian manatees have a diploid number of 56 chromosomes and no differences were observed between study l ocations. They compared their findings to those of White et al . (1976) and found that there may be chromosome rearrangement between T. manatus and T. inunguis which could explain some of the divergence between the two species.
23 Chromosome banding (G and Q banding) was also performed on T. manatus latirostris (Gray et al. 2002) . Chromosome numbers agreed with previously published values (N=48) (Loughman et al. 1970; White et al. 1976) . The authors were able to construct karyotypes for both males and fem ales, produce the first analysis of banded chromosomes in detail, and propose an ideogram of these karyotypes for the Florida manatee. The first population level manatee conservation genetic study was conducted by (1988) . They est imated the genetic variation of the Florida manatee using 59 carcass specimen s from throughout the state of Florida based on winter aggregation sites. Variation was determined via electrophoresis of allozymes, which are distinct forms of proteins based on the size and charge of the enzyme. The 24 loci that were evaluated were visible in all individuals tested and 10 of these were polymorphic. Allele and genotype frequencies were calculated for each locus and proportions of polymorphic loci were reported for each region (mean P=0.3). The expected (mean H E =0.069) and observed (mean H o =0.05) heterozygosites were described and it was determined that there is less genetic diversity in the population than would be expected in a randomly mating population based on lower H o than H E values. F statistics, which estimate the division of genetic diversity within and between populations, were calculated for total ( mean F it =0.29), within ( mean F is =0.26), and among samples ( mean F st =0.058). High F it and F is (levels of inbreeding compared to total population and levels of inbreeding within a population , respectively) values suggest high levels of homozygosity in a population, while low F st , degree of population differentiation, values indicate high gene flow among samples or fragments of the
24 population (Frankham et al. 2002; Beebee & Rowe 2008) . explain that high levels of homozygosity could be due to either inbreeding, the Wahlund effect (individuals being pooled with genetically dissi milar indivi duals), or selection. Because these individuals were all carcasses, selection could be occurring against homozygosity in the population. Genetic diversity was measured for within and among samples and it was determined that little (~4%) of th e genetic variability is due to differences among locations. Genetic pairwise distances were also calculated across locations with an average genetic distance of 0.0148 reported. McClenaghan and low, high genetic diversity, and a homogenous Florida manatee population. T he mitochondrial cytochrome b (cyt b ) gene was sequenced for 75 codons (225 bases) of three Florida manatees (Bradley et al. 1993) . Mitochondrial DNA (mtDNA) is a small circular molecule of DNA that is inherited from the mother (Allendorf et al. 2012) . It is easy to isolate from a variety of small biological samples, such as blood, tissue, skin, hair, feces, and bone because it is highly concentrated in each cell and does not deg rade as easily as nuclear DNA. With high mutation rates and high variability within and among populations, mtDNA is a good choice for the study of recently divergent populations (Parker et al. 1998) . Mitochondrial DNA is also very useful in the study of phylogenetics, the comparing of DNA sequences of certain genes among and within species to determine evolutionary divergences (Rudin 1997) . The cyt b gene was used because there are both conserved and variable regions, allowing systematic questions about phylogenetic relationships at both the ancestral and more recent divergent levels to be answered (Farias et al. 2001) . Sequencing took place using the universal primers
25 of cyt b , MVZ3 and MVZ4. There was no intraspecific variation detected within the nuc leotide sequence of the three manatees. When compared to other spe cies, the manatee sequence had two unique codons and had two codons that it shared with the African elephant. There were 51 nucleotide changes between manatees and African elephants, while there were 53 changes between manatees and humans. This study was too small to make general conclusions about the population, but it appears to agree (1988) , that the manatees of Florida are genetically homogen ous. The first phylogeographic study on manatees focused on the West Indian manatee by determining how many populations there were , as well as how many taxa were represented (Garcia Rodriguez et al. 1998) . The authors did this through sequencing 410 base pairs (bp) of the mtDNA control region, which is found in the non coding region of the mtDNA and therefore mutates faster than other regions , such as cyt b which encodes for proteins (Tang et al. 2006) . Due to its ability to evolve faster, the mtDNA control region is thought to allow for a more thorough depiction of the diversity in the West Indian manatee. Eighty six specimens of T. manatus from eight locations (Florida, Mexico, Dominican Republic, Pue rto Rico, Colombia, Venezuela, Guyana and Brazil) as well as 16 specimens of T. inunguis , for comparison, were used in this study. Nucleotide and haplotype diversities were calculated , sequence divergence within and between populations , and genetic divers ity within and among samples (AMOVA) were estimated . The use of st , which is analogous to F st when using haploid data, was calculated to estimate gene flow. Migration rates, the number of haplotype site differences, and evolutionary relationships were a lso reported (Table
26 1 1). There were 16 haplotypes observed from the T. manatus samples with an overall high level of haplotype diversity ( h= 0.839) and nucleotide diversity ( =0.04). Although regionally there was substantial variation among haplotype div ersity, Florida, Mexico, and Brazil only had one haplotype each, while Colombia and Guyana had seven and five haplotypes, respectively. The migration rate ( m ) ranged from 0 3.5 between regions indicating that some regions exhibited more mixing while other regions had none. Pairwise sequence divergence levels were also high (ranging d =0.04 0.074). This finding corresponds to the geographic clustering of haplotypes reported in the Neighbor Joining tree of T. manatus and T. inunguis . Florida, Puerto Rico, Dominican Republic, and Colombia grouped together in what the authors called the Cluster I Atlantic C luster IV contained mostly T. inunguis from Brazil, three individuals from Guyana that morphologically appeared to be T. manatus but were genetically more closely rel ated to T. inunguis were also included. The inclusion of these individuals indicates the possibility of range expansion, haplotype retention, or a hybrid zone. These phylogeographic units exhibited strong separation by geography but do not correspond wit h the currently accepted species divisions of T. manatus and T. inunguis . Sequence divergence estimates between T. manatus and T. inunguis were in the same range ( d =0.04 0.08) as within T. manatus , leading the authors to suggest that the two species share d a common ancestor more recently than previously thought.
27 The authors suggest population structure has occurred due to the need of T. manatus for coastal/marine habitat, yet its ability for rare long distance travel allows colonization events to occur. G arcia Rodriguez et al . (1998) agree with McClenaghan & (1988) and Bradley et al . (1993) that the manatee population structure in Florida is one large unit as evidenced by a single haplotype (A). They describe two possibilities for this occurrence: (1) the population was colonized recently, within the last 12,000 years, by manatees from the West Indies, most likely Puerto Rico or the Dominican Republic, and (2) a population bottleneck occurred, significantly reducing the population and decreasing its genetic diversity. Manatees in Brazil also had no mtDNA diversity observed, yet unlike the F lorida manatee, they are known to have a small population and have been hunted for the last 300 years, which could contribute to their low diversity. Phylogeographically, the authors suggest there is a deep divergence (3.5 7.0 Myr) between the three linea ges (Clusters I III), due to isolation, expansion, and the physical geography of the range that these animals inhabit. The authors feel that a re evaluation of the taxonomic structure of T. manatus may be in order. A phylogeographic and population gene tic study was conducted on the Amazonian manatee (Cantanhede et al. 2005) . The goals of this study were to estimate levels of genetic variability within and among the Amazon River Basin, test correlations of geography using mtDNA control region haplotype diversity, and determine phylogeny of the Amazonian manatee. Amazonian manatee specimens (N= 68) were used to amplify and sequence a 361 bp segment of the mt DNA control region ( d loop ). Phylogenetic analyses included MODELTEST 3.06 to determine the best model of nucleotide evolution ( HKY85 with gamma distributed mutation probabilities ) and a full
28 heuristic max imum likelihood analysis in PAUP* to determine the most likely phylogeny. There were 63 variable positions observed in the complete alignment of ma natees , including 49 parsimony informative sites and 31 unique haplotypes. N one of the haplotypes from geographically close locations clustered together i n the max imum likelihood topology, although this tree is very similar in structure to the tree from G arcia Rodriguez et al . (1998) . The tree reports that T. manatus is paraphyletic to T. inunguis due to one of the T. manatus clusters ( C luster I) being more closely related to the T. inunguis clade but it is not statistically significant . L ow average genetic distance was estimated within T. inunguis (1.0%) with a higher average genetic distance between T. manatus and T. inunguis (~7.35%) . Population analyses of T. inunguis consisted of 74 ind ividuals with 33 haplotypes, grouped into six pop ulation s based on location. The number of haplotypes, haplotype diversity, nucleotide diversity (Nei) , and the number of polymorphic sites , were estimated for both T. inunguis and T. manatus ( Table 1 1) . Haplotype diversity was approximately equal for bo th species, while nucleotide diversity was much higher in T. manatus . There were a high number of polymorphic sites indicating a high level of genetic diversity . No significant difference was discovered between genetic variability and location. The authors suggest that the main force s of haplotype distribution are that gene flow and dispersal are limited . C oalescent theory , which determine s the most recent common ancestor by tracing the alleles of a ge ne within a population back to one ancestral copy , was used to estimate haplotype age at 4.0 Mya with the divergence occurring in the Plio Pleistocene. D test, which tests the neutrality of evolution of a sequence, and F s test , which also tests for neutrality given a derived level of
29 diversity, were both significantly negative (Table 1 1). This indicates that there is possible genetic disequilibrium and evidence of mutation, migration, and /or drift . These statistics normally test for effects of selection, but selection does not occur in the control region of mtDNA due to the fact that is in the non coding region and presumed neutral (Rand 1996) . The significantly negative test results could indicate a population bottleneck , a population expansion event occurred recently, or that the single population analyzed was actually more than one population (Wahlund effect ) . The f emale variance effective population size ( N ef ) was estimated at 4.55x10 5 which represents a 95% increase in population size due to manatees now being protected from h unting. The data also indicates that T. inunguis has a much larger N ef than T. manatu s. An a nalysi s of molecular variance (AMOVA), which tests pop ulation subdivision and structure , established that most of the variability is within populations (94.6%) , yet differences among populations were significant . Pairwise population F st significance test s found no significant differences between loc ations. A Mantel test was used to determine significance between genetic and geographic distances and found no si gnificant correlations. Although there may be a pattern of restricted gene flow , it is not enough to be an influential factor in population structure; a more likely factor is the large geographic range of T. inunguis . Similar to Garcia Rodriguez et al . (1 998) , Cantanhede et al . (2005) determined that there are four lineage s of Western Atlantic t richechids , three saltwater and one freshwater. They also agree that while T. inunguis is morphologically distinct from T. manatus , there is not more of a genetic differen ce between T. inunguis and T. manatus than there is among any of the three T. manatus clusters . G enetic disequilibrium was
30 reported when all four lineages were analyzed as one population which may indicate there is little to no gene flow between the lineages. The authors conclude the best explanation for this disequilibrium is population expansion that has occurred either due to decreased hunting pressure within the last 30 40 years , population expansion mitigat ing the effect of a population bott leneck (a drastic reduction in the population), or there was no population bottleneck (1988) , Bradley et al. (1993) , and Garcia Rodriguez et al. (1998) found in Florida, Cantanhede et al. (2005) have concluded that T. inungui s is a panmictic p opulation and genetically homogenous . In 2006, the largest comparative manatee genetics study was published. It included phylogenetics, phylogeography, cytogenetic, and population level hybridization analyses to gain a comprehensive und erstanding of the three extant Trichechus species ( T. manatus, T. inunguis, and T. senegalensis) (Vianna et al. 2006) . The mtDNA control region (550 bp) and cyt b (615 bp) region were amplified, sequenced and analyzed for 291 specimens ( T. manatus= 189 from 10 countries [the only new country was French Guiana, as compared to Garcia Rodriguez et al. (1998) ] , T. inunguis N = 93 from three countries , T. senegalensis N = 6 from four countries , D. dugon N = 3 ) . Sequences from Garcia Rodriguez et al. (1998) were us ed yet inconsistencies were found. When they were fixed, the haplotypes to which they were assigned changed. Number of haplotypes , haplotype diversity, nucleotide diversity (Nei), and the number of polymorphic sites were estimated for each species (Table 1 1). Haplotype diversity was slightly higher when compared to the same species in other studies (Garcia Rodriguez et al. 1998; Cantanhede et al. 2005) , while T. senegalensis was very high, with five haplotypes detected from six individuals. Nucleotide diversities were also fairly similar
31 (2005) estimate of T. manatus , which was an order of magnitude smaller than either of the other studies. This could have been a computation error, as other studies do not report nucleotide diversity as a percentage. Pairwise st were calculated and Mantel tests were performed. The number of migrant females was estimated ( N mf ), exact test of population differentiation was measured, and an AMOVA was performed. Most analyses used populations based on geographic location, but a few tests investigated what other population partitions could be discerned. Time to recent population bottleneck was estimated, as was the expansion coefficient to estimate population sizes throughout time. For T. manatus , population differentiation was high ( st =0.658), due to two potential gene flow barriers, one isolating Puerto Rico and the Dominican Republic and the other separating Guyana and Brazil. A possible explanation the authors give for this population differentiation is a linear stepping stone mode l. This model postulates that there will be a decrease in genetic diversity at the extremes of the range, in this case, Florida and Brazil, which have the least amount of genetic diversity (Garcia Rodriguez et al. 1998; Cantanhede et al. 2005) . As report ed by Cantanhede et al. (2005) there were no significant correlations between genetic distances and location (Mantel test), yet there was a significant correlation between genetic distances and coastline, supporting the notion that manatees travel along th e shallow waters of the coast. The phylogenetic tree created using the Tamura Nei model formed different geographic clusters than were reported by Garcia Rodriguez et al. (1998) and Cantanhede et al. (2005) , yet all three studies ha ve grouped T. manatus i nto three distinct clusters . Cluster I consists of Florida, Mexico , Puerto Rico, Dominican Republic, Belize,
32 Colombia, and Venezuela, Cluster II consists of Mexico , Belize, Colombia, and Venezuela, while Cluster III contains Brazil, French Guiana and Guya na. Tajima test of neutrality found only a significant D in Cluster III, indicating that a population bottleneck followed by a population expansion may have occurred approximately 183,000 years ago. For T. inunguis , haplotypes were very closely related, even though the individuals came from a very large geographic range, and because of this the effective female population size was able to be estimated between 1.3x10 5 5.1x10 5 individuals. (2005) estimate of 4. 55x10 5 T. inunguis females fall into t his range. These estimates appear to be very high. The mutation rate used (2%/million years) is very slow which could have over estimated the number of individuals. Population differentiation estimates were low ( st =0.18), denoting a single population. The Tajima test of neutrality was negative and significant, indicating a recent expansion event, while the time to recent bottleneck was calculated at approximately 130,000 years ago. The authors conclude that the g enetic diversity of the population is high and due to basal haplotypes being detected in the East, the population expanded from the mouth of the Amazon into the river system. Trichechus senegalensis sequences formed two clusters, one from Guinea Bissau a nd the other made up of individuals from Ghana, Chad, and Niger, this cluster had low genetic differentiation ( st not reported). These two clusters do not agree with the findings of Domning & Hayek (1986) that there is a morphological separation between river and coastal populations, but only a few samples were analyzed.
33 When comparing interspecific variability, 80 polymorphic sites were reported. The West African manatee had the highest haplotype diversity (5 haplotypes/6 samples), indicating substanti al divergence among samples and possibly regions. While the Amazonian manatee had the lowest nucleotide diversity ( =0.005), signifying a species with very low genetic variation. The phylogenetic tree of the control region, like the one presented in Cant anhede et al. (2005) , also reported that T. inunguis and C luster I of T. manatus are more closely related, and T. senegalensis is most basal. Yet, the cyt b phylogenetic trees, both neighbor joining and maximum parsimony, place T. senegalensis and T. mana tus more closely related while T. inunguis was most basal, with high bootstrap support. Dates of divergence between the dugong and the three species of manatees were calculated ( T. manatus =621,000 years ago, T. inunguis =371,000 years ago, and T. senegalen sis =308,900 years ago). There were four cases of potential hybrids, where morphologically the individuals were identified as one species but genetically w ere identified as another. To resolve this, karyotypes of three West Indian manatees from Brazil, in cluding a likely hybrid, were analyzed. Following the studies by Loughman et al. (1970) and White et al. (1976) , which illustrated the diploid number of chromosomes as 2n=56 and 2n=48 respectively for T. inunguis and T. manatus , Vianna et al. (2006) agree d with the T. manatus chromosome count but identified an animal from Brazil that morphologically was T. manatus yet genetically was T. inunguis and had a diploid number of 2n=50. This could be due to a hybrid female mating with a male T. manatus . The two most informative microsatellite loci (H11 and A09) (Garcia Rodriguez et al. 1998) were amplified, to determine if hybridization was a possibility using nuclear DNA. For H11, the Amazonian manatee
34 had seven alleles, while the West Indian manatee had only one, yet in three of the hybrids a mix of alleles were observed lending support for the theory of manatee hybridization in Brazil. Vianna et al. (2006) conclude that the most recently diverged species was the Amazonian manatee, then the West In dian and the African manatees. However, their phylogenetic trees using data from the two regions of mtDNA, control region and cyt b, disagree with each other as to which species is more basal, T. senegalensis or T. inunguis . They suggest that T. manatus originated in Colombia, because it has the most haplotype diversity and represents two of the haplotype clusters. However, the data suggests there are two points of origin, one east and one west of Puerto Rico and Dominican Republic. This would imply tha t there are two evolutionarily significant units within T. manatus. The authors also make a suggestion for management and conservation plans, to refrain from breeding animals in captivity from distant locations, in order to avoid hybridization. They conc lude that likely hybridization zones of T. manatus and T. inunguis are at the mouth of the Amazon River and at the mouth of the Orinoco River in Venezuela. A cytogenetic study was conducted to resolve the manatee phylogeny using chromosome painting and t o validate the super order Afrotheria (Kellogg et al. 2007) . Chromosome painting is used to identify the different chromosomes by attaching fluorescently labeled probes to unique sites on the chromosomes. This can allow for the detection of abnormalities within the chromosomes as well as comparative cytogenetic studies (Ried et al. 1998) . Afrotheria is a super order within the clade of Eutheria, or all mammals. It includes mammals of African origin with a wide range of
35 morphological features including, golden moles, elephant shrews, aardvarks, hyraxes, elephants and sirenians. The hyrax, elephant and sirenians form their own clade, Paenungulata, and within that elephants and sirenians their own clade, Tethytheria. Karyotype results of manatees in this study found that there is a diploid number of 48 chromosomes, agreeing with previous studies (White et al. 1976; Gray et al. 2002) . This study found that human chromosome paints hybridized to 20 manatee chromosome segments. Most provided signals at two c hromosomes, but some signaled on up to five different chromosomes. The human Y chromosome did not hybridize with any of the manatee chromosomes. The painted manatee karyotype was then compared to other taxa within Afrotheria and it was determined that al l of the species had eight chromosome associations in common, with five of those being found in all Eutherians. There were four associations in karyotype comparisons between sirenians and elephants. Two of the associations were found in all Afrotheria, w hich led the authors to support Afrotheria as a valid clade. The authors conclude that the data here confirm Tethytheria as a clade and support the clade Paenungulata within the super order Afrotheria. Another chromosome painting study was conducted to a ssess whether Paenungulata is actually monophyletic or if some of the taxa are more closely related (Pardini et al. 2007) . Chromosomes of the African elephant ( Loxodonta Africana ), the hyrax ( Procavia capensis ), and the Florida manatee ( Trichechus manatus latirostris ) were painted to identify similarities and differences. Within these, 29 33 segments were conserved and when compared with an out group (the aardvark, Orycteropus afer ) there were 36, 33, and 32 similar regions, respectively. It was inferred that
36 Paenungulata diverged from aardvarks 80 million year ago, while African elephants, hyraxes and Florida manatees diverged 62 million years ago. The authors conclude that Paenungulata are one clade, while unlike the findings reported in Kellogg et al. (2007) there was no evidence to support Tethytheria as its own clade within Paenungulata, due to a slow rate of evolution. The first set of manatee specific microsatellite markers (N=14) was published in 2000 (GarcÃa RodrÃguez et al.) . They reported eigh t polymorphic markers that yielded an average of 2.9 alleles/locus and an average heterozygosity of 0.41 in the Florida manatee. These markers were amplified across the other three extant species of sirenians with eleven polymorphic markers for the Antill ean and Amazonian manatees and nine for the dugong. In 2007, another ten manatee specific microsatellite markers were published (Pause et al. 2007) . They describe an average of 4.2 alleles/locus and an average heterozygosity of 0.501. Some of these mark ers were also polymorphic for the other sirenian species. An additional 18 polymorphic markers were reported, with 2.5 alleles/locus and an average heterozygosity of 0.35, as well as three sex determining markers (Tringali et al. 2008a; Tringali et al. 20 08b) . The average match probability, or probability of identity, which is the probability that two individuals will have the same genotype by chance, was computed and corrected using these markers and found to be 1.95x10 6 . In 2010, a cross species microsatellite panel of markers for the dugong (D . dugon) and the Florida manatee ( T. m. latirostris ) was published (Hunter et al. 2010 b ). In that study, they determined of the 32 dugong specific and the 25 manatee specific microsatellite markers tested, 30 dugong and 21 manatee markers were polymorphic. A set of microsatellite markers were amplified for both dugongs and
37 manatees to determine the most informative panel of markers for each species. For dugongs, the most informative panel consisted of 11 l oci, eight dugong primers and three manatee primers, giving a probability of identity of 1.24E 11 and an average heterozygosity level of 0.710. For manatees, the most informative panel consisted of 13 loci, four dugong and nine manatee primers, giving a p robability of identity of 2.70E 09 and an average heterozygosity level of 0.603. When comparing the manatee genetic diversity to a fish species with a large population, such as the Atlantic tarpon, ( Megalops atlanticu s ) which has an average of 7.7 allele s/locus over 15 loci and an average heterozygosity of 0.60 ( Seyoum et al. 2008 ), the manatees lack of genetic diversity could have major negative implications for the fitness of the population. Yet, a combination of all of these microsatellite markers inc luding a mix of species specific markers, as seen in Hunter et al. (2010b) , could allow for a very informative and highly polymorphic panel of markers that would enable studies of population structure, individual identification, and pedigree analysis. The sexing of sirenians both visually and genetically is important for estimating sex ratios, growth rates, fecundity, and survivorship (Lanyon et al. 2009) . Lanyon et al. ( 2009) compared the visual and genetic identification of two species of sirenians, dugongs (N=628) and Florida manatees (N=100), that are not only difficult to catch but do not display sexual dimorphism. Opportunistic capture of dugongs took place in Australia between 2001 and 2007 (N=460), with a majority (N=454) having their sex visua lly identified in the field. An additional 168 dugongs had skin biopsies taken and were not visually sexed. The Florida manatee samples were obtained from carcasses (N=64) recovered state wide and from free swimming manatees (N=36) between 1998
38 and 2005. All had their sex visually identified. The application of three sexing molecular markers, ZFX and two male specific (SRY and ZFY), resulted in 93% 96% accuracy of visual and genetic sex identification. The discrepancies identified were reportedly due t o inexperience in visual sex identification or other human error such as mislabeling samples or incorrectly recording the data. The authors suggest a more rigorous approach to visual sex identification including multiple experienced researchers identifyin g the sex, multiple identifications of the animal over time, photography of the genital region, and the use of secondary sex characteristics such as teat size, tusks, or calf associations. U sing both phylogeographic and population genetic methods , the g enetic variation and dispe rsal of the Antillean manatee in Belize was determined ( Hunter et al. 2010a) . uman impact s threaten this species. Phylogeny and phylogeographic studies included the amplification of a 410 bp segment of the mtDNA control region in 113 individuals , from three sites on the coast of Belize: Belize City Cayes, Southern Lagoon systems, a nd Placencia Lagoon . Three haplotypes were present country wide, with only one haplotype in the entire Southern Lagoon S ystem. The degree of differentiation, F st , and st were calculated for both Florida Belize and within Belize comparisons (Table 1 1) . D , genetic diversity, nucleotide diversity , and the number of polymorphic sites and nucleotide substitutions were calculate d. Within Belize, haplotypes and genetic differentiation estimates indicate that there is reduced mix ing between the populations in the Southern Lagoon System and the Belize City Cayes. A
39 In comparisons to the three previous phylogenetic and phylogeographic studies, st is approxima tely 18 times lower than T. manatus reported by Vianna et al. (2006) . When comparing haplotype numbers, the Belize population has only three, while T. manatus from Garcia Rodriguez et al. (1998) , Cantanhede et al. (2005) and Vianna et al. (2006) had 16, 40, and 20 haplotypes respectively. All three studies used T. manatus from a variety of countries and locations, while Hunter et al. (2010a) only used manatee samples obtained from Belize. Both genetic diversity and nucleotide diversity are lowe r for Antillean manatees from Belize as compared to the other studies as is the number of polymorphic sites. All of these statistical analyses indicate this population has very low diversity. Population studies included the amplification of 16 polymorph ic microsatellite markers in 118 individuals. Polymorphism was estimated via observed and expected heterozygosity and average number of alleles per locus. Genetic differentiation, F st , and R st w ere calculated for Florida Belize and within Belize comparis ons as was the coefficient of genetic relatedness (Table 1 2) . Analogous to F st , R st utilizes allele size differences in genes within populations, as opposed to using allele state differences (Hardy et al. 2003) . Calculating both F st and R st allows for q uestions of migration ( F st ) and mutation ( R st ) rates to be answered. The probability of identity was calculated as 4.6E 08 and the probability of sibship identity was estimated at 0.0004. Two genetic subdivisions were identified within Belize, the Southern Lagoon System and the Belize City Cayes, and two subdivisions were identified between Belize and Florida, indicating two separate populations with restricted gene flow. Limited genetic e xchange between
40 the Southern Lagoon System and the Belize City Cayes within Belize may indicate anthropogenic barriers that could be address ed through conservation management actions . Differences in haplotype proportions and a difference in F st values bet ween mtDNA and microsatellites indicate that females tend to stay in the same location while males move between locations. Overall , this study determined that the genetic diversity of both mtDNA and nuclear markers within manatees from Belize was lower th an some critically endangered species , such as the North Atlantic right whale. These findings establish that Belize has a small, isolated population that experienced a population bottleneck or was impacted heavily by anthropogenic threats. Ferreira et a l . (2011) conducted an in silico (computer simulated) study using published cyt b sequences from all extant sirenian species, as well as domestic animals commonly used as food, to describe nucleotide mutations (or single nucleotide polymorphisms, SNPs) whi ch will identify illegal manatee meat in markets. As the authors state, illegal hunting of sirenians is still being conducted even though all four species are listed as vulnerable on the IUCN red list and endangered/threatened in the Convention of Interna tional Trade in Endangered Species (CITES). SNPs are a useful molecular tool for identifying species specific changes and with the amount of sequence data in Gen Bank, this allows for an effective and inexpensive way to identify mutations. Sequence alignm ents were performed in BIOEDIT 6.0.7 and nucleotide and restriction enzyme sites were discovered using CLEAVER software. Five polymorphic sites were identified useful to discern sirenians from domestic animals, and two sites were able to specify within si renians. As was seen with the cyt b phylogenetic tree in Vianna et al. (2006) , sequences of T. manatus and T. senegalensis were unable to differentiate from
41 each other, implying they are more closely related to each other than to the other sirenian specie s. Nourisson et al. (2011) published on the genetic diversity and population structure of the Antillean manatees ( T. manatus manatus ) in Mexico. Samples from six different regions of Mexico: two from the Caribbean Sea (ChB/AB) and four from the Gulf of Mexico (GMx) were compared to Florida; all were genotyped using 13 polymorphic microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) . Subpopulations were identified, genetic distances were measured, effective population size, genetic di versity, genetic differentiation (allelic richness, heterozygosity, pairwise F st and R st , and inbreeding coefficient F is ), and population bottlenecks were estimated (Table 1 2). Three subpopulations were identified, one in Florida and two in Mexico separating the Gulf of Mexico from the Caribbean Sea regions. Although there were multiple specimen that had mixed ancestries, such as a genotype from the Gulf of Mexico with a haplotype from the Caribbean Sea, indicating that a female from the Caribbean Sea likely mated with a male of the Gulf of Mexico or a backcrossed male. Measures of genetic distance indicate two main clusters separating Mexico and Florida, yet within the Mexico C luster there appears to be two sub clusters representing the Gulf of Me xico at the north and Chetumal Bay at the south with a mixing zone of these in Ascencion Bay. Effective population sizes were measured for each region within Mexico ranging from 5 33 individuals. Heterozygosity and allelic richness levels were comparable to the other four microsatellite marker papers and Hunter et al. (2010b) . Pairwise F st and R st values also indicate that there are two main groupings of manatees in Mexico: the Gulf of Mexico group and the Caribbean Coast group. Both F st and R st
42 values were significant and relatively high between the Gulf of Mexico and all other groups, indicating low gene flow. The lowest F st value was between Chetumal Bay and Ascencion Bay which are both on the Caribbean coast. This value, although low, was significa nt; representing the fact that there are genetic differences between the two regions yet there is also high gene flow. These findings also support the theory that Ascencion Bay is a breeding ground for both the Gulf of Mexico and Chetumal Bay. The inbree ding coefficients throughout all regions of Mexico were low ( 0.059 0.035) indicating that there are high heterozygosity levels and low chances of inbreeding in the populations. The authors conclude, based on their findings and findings of unpublished mtDNA data, that there is higher genetic diversity on the Caribbean Coast than in the Gulf of Mexico, with three haplotypes identified; one in common with Florida, one with Belize, and one with the Gulf of Mexico population. They also concluded that the l ow diversity detected in the Gulf of Mexico was probably caused by both a founder effect (few individuals starting the population) and a population bottleneck. In comparisons with the Hunter et al. (2010a) Belize study, it was confirmed that Mexico shares not only a haplotype with Florida but also a higher level of nuclear gene flow. This is indicated by lower F st values between Florida and both the Gulf of Mexico and the Caribbean Coast than between Florida and Belize. The results of this study signify the possibility that the subspecies are breeding and/or migrating from Florida to Mexico. In a very comprehensive study using cytogenetic, phylogenetic, phylogeographic, and population genetic methods, the chromosomal, mitochondrial, and nuclear diversit y within Puerto Rico and between Puerto Rico and Florida were investigated ( Hunter et al.
43 2012) . Cytogenetic findings concluded that Antill e an manatees of Puerto Rican have 48 chromosomes which are consistent with chromosome numbers found in the Florida m anatee (Gray et al. 2002) . Phylogenetic and phylogeographic analyses consisted of the amplification of 410 bp of the mtDNA control region for 112 individuals from four sites in Puerto Rico and 28 individuals from four sites in Florida. Four haplotypes we re found in Puerto Rico, including a new haplotype found only in the northern region, while Florida had only one haplotype (A). The degree of differentiation, F st and st , were calculated for both Puerto Rico Florida and within Puerto Rico comparisons. N ucleotide diversity , genetic diversity, the number of polymorphic sites and nucleotide substitutions, and s D were calculated ( Table 1 1 ). When comparing these values to previous studies there is considerably less diversity in Puerto Rico. With only three polymorphic sites, 0.001 nucleotide diversity, low mtDNA genetic diversity values, and the high st values within Puerto Rico, it is clear that this population is in need of conservation efforts. There were also significant differences between the Puerto Rico and Florida populations in both F st and st estimates, indicating separate populations. Population genetic analyses were conducted on samples throughout Puerto Rico (N=110) and Florida (N=95) that were genotyped at 15 polymorphic loci (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) . Nuclear genetic diversity estimates (Table 1 2) were simi lar to previous reports for Florida, Belize, and Mexico, with a higher average number of alleles in Puerto Rico than Mexico and Belize. There were significant levels of genetic differentiation between Puerto Rico and Florida, indicating a lack of gene flo w between the two populations. Within Puerto Rico, there was a moderate amount of
44 genetic differentiation between the subpopulations, indicating limited gene flow between the north, east, and southwest regions. Structure and genetic landscape results cle arly indicate that Florida and Puerto Rico have a barrier to gene flow and that these two populations should be managed separately. Within Puerto Rico, structure results inferred two subpopulations (N E and S W) as was also seen in the mtDNA data. The au thors suggest that managing the two subpopulations within Puerto Rico separately would allow for more directed conservation efforts to be put in place according to each as was seen in both tracking studies and these genetic results. It was concluded from this and previous studies that there is a genetic barrier between Puerto Rico and everywhere west and south. Immigration to Puerto Rico is not likely, due to limited po pulation sizes of the surrounding islands and the lack of gene flow from Florida. The effective population size was estimated at 52 individuals which appears low, but is plausible given an estimated population size of 250. It was also concluded that due to regional haplotype diversity, yet evidence of gene flow between certain regions, that male regions. Overall, Puerto Rico has not only a small population with limited diversi ty in both mtDNA and nDNA, but it also has geographic barriers that may prevent ease of gene flow. It was concluded that Puerto Rico should be managed separately from Florida and requires a new recovery plan that calls for the incorporation of the distinc t threats posed on its two subpopulations. A phylogeography study of the critically endangered Antillean manatees of Brazil was conducted to characterize haplotype distribution throughout Brazil and identify
45 haplotypes of rescued and captive born manatees (Luna et al. 2012) . In the last quarter century, an increasing number of orphaned calves have been rescued and rehabilitated by the Stranding Network of Northeast Brazil and the Center of Research and Conservation of Aquatic Mammals. These strandings ar e attributed to habitat degradation leading to the female giving birth in open water and being separated from her calf. Blood and tissue samples were collected from rescued and captive born manatees, as well as recovered carcasses (N=73) since 1989. A 41 0 bp segment of the mitochondrial control region was amplified for all manatee samples, identifying three haplotypes with low genetic and nucleotide diversity (Table 1 1). The three haplotypes were not evenly distributed, representing genetic differences between the northern and southern locations. The most common haplotype (M01) was found in the south in seven of the eight sample sites. Of the two other haplotypes M03 was discovered only in the northern most site (MaranhÃ£o), while M04 was identified in two site s , MaranhÃ£o but also in PiauÃ ( a site directly east). Rehabilitated manatees are currently being released into one of two National Protected Areas, located in the southern region of Brazil. With limited genetic identification information availab le at this time , it is recommended that manatees with distinct haplotypes (M03 and M04) not formally identified in the southern regions should be released back to areas in the north (MaranhÃ£o and PiauÃ), until analyses utilizing nuclear DNA can be conducte d. The authors recommend this course of action to prevent genetic swamping (i.e. the loss of locally adapted alleles) and outbreeding depression (i.e. disruption of favorable genes) due to the mixing of individuals from genetically distinct populations ( F rankham et al. 2002; Allendorf and Luikart 2007 ). The
46 authors also suggest utilizing an area in southern Brazil, ( at the border between Alagoas and Pernambuco) that has an extinct population of manatees. This would not only re establish a population of m anatees, but reduce the risk of outbreeding depression and genetic swamping, as well as connect the northern and southern populations. The most recent manatee genetics study was a thorough evaluation of the genetic diversity and population structure of t he Florida manatee (Tucker et al. 2012) . Manatee samples from four management units (MU) of Florida (N=341) were genotyped using 18 polymorphic microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Hunter et al. 2010b) . Using different spatial scales: all of Florida, by coast, and by individual MU, typical summary statistics were calculated as were multiple genetic differentiation estimates, effective population size, and inbreeding coefficient (Table 1 2). Global genetic differentiati on was estimated using the fixation index G st as opposed to F st st D were used as controls for bias in the dataset and to ensure convergence of the estimates. Heterozygosity levels were consistent with what has been previously reported for manatees (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Nourisson et al. 2011; Hunter et al. 2012) , indicating low genetic diversity. It was noted that Florida manatee diversity levels were even lo wer than those reported for other species that were hunted, disturbed, or lived in fragmented populations. Moderate levels of inbreeding were observed in all spatial scales. Genetic differentiation levels were low yet significant, suggesting genetic stru cture (i.e. genetic differences) between coasts. Results also revealed possible genetic structure between coasts when all 18 loci were used in
47 analyses, however when only 11 loci were used only one population was revealed. Based on the 2010 Florida manat ee census of approximately 5,000 individuals (< 2,500 adults, Runge et al. 2007a; Runge et al. 2007b) the estimated effective population size reported (N e =1,260) is about 1/5 of the whole population and Â½ of the adult population. Although it is explained that these values may be biased and inaccurate, it is the best and most conservative estimate thus far. This information, in accordance with moderate inbreeding levels, supports the notion that although the Florida manatee population lacks genetic diversi ty, its population numbers are high and not overly inbred. It was concluded that while the population appears stable at the current time, that a rise in anthropogenic effects could severely impact this endangered population. In 2012, a review of the conse rvation genetic tools that have and could be used for research on the West Indian manatee was published (Bonde et al. 2012) . The coupling of genetic data with life history information was emphasized to improve management efforts. A multitude of genetic markers were reviewed, including some that have already been in use: allozymes, mitochondrial (mtDNA) cytochrome b and control region, microsatellites, and karyotype, banding patterns, and numbers of chromosomes. Novel marker studies for manatees include the use of mtDNA marker cytochrome oxidase 1 (CO1) gene and sequencing the entirety of the mitochondrial genome to provide new variation estimations, as well as to supplement the information provided by nuclear marker studies. The addition of 17 polymorph ic microsatellite markers to the panel of 36 currently in use may help to improve the power of pedigree and parentage assignments. Either developing more complex microsatellite markers (tri or tetra nucleotide repeats) or transitioning to single nucleoti de polymorphisms
48 (SNPs) were suggested for increasing genetic information for populations that exhibit low genetic diversity. With the sequencing of the whole Florida manatee genome, gene expression studies can explore topics such as diseases, immune resp onse, mate selection, and fitness. The authors recommend some guidelines for choosing the proper genetic tools for the research area of interest, including a solid sampling regimen, multiple marker types, neutral markers, and incorporating ecological lif e history data sets with genetics. The current status, research results, and implemented suggestions of the Florida manatee is described. With low genetic diversity, population size may be affected, which would ultimately lead to higher relatedness, inbreeding, and a decreased ability to adapt to a changing environment. By implementing these tools and suggestions in a continuous effort to genetically monitor the manatee and other imperiled populations, managers will be abl e to make more informed management decisions. The 19 conservation genetic studies summarized above all agreed on the same general conclusions. There are three geographic clusters of T. manatus with high population differentiation among them. The Florida manatee and the Amazonian manatee are very distinct based on chromosome numbers, phylogenetics, and phylogeographic studies, yet are both homogenous populations and population divisions by country are appropriate units for analyses. A re evaluation of the phylogenetic structure of the Trichechidae family was suggested due to a lack of consensus as to which species is the most recently diverged. At the population level, there was low genetic diversity observed across T. manatus while Florida has a single
49 i solated population and Belize and Mexico have two groups each. The low genetic diversity observed in these manatee conservation genetic studies leads to questions of inbreeding and relatedness that can be addressed through pedigree studies. Parentage an d Pedigree Analysis Parentage assessment in its most basic definition is the assigning of parents to offspring. This can be done through photo identification of potential parent offspring pairs, observations of copulation, and/or genotype data (Clapham & PalsbÃ¸ll 1997) . Ideally, each sampling event will collect data using a variety of techniques to increase the chances of correctly assessing parentage. The discovery and implementation of microsatellites into the field of population and conservation biolo gy have enabled parentage assessment to infer answers to issues on the topics of evolution, ecology, and conservation, such as mating systems and behavior, dis persal and recruitment patterns, captive breeding programs, reproductive success, impacts of inbr eeding, and determining effective popu lation size (Clapham & PalsbÃ¸ll 1997; Frankham et al. 2002; Jones & Ardren 2003; Allendorf et al. 2012) . Multilocus genotypes, using microsatellite markers, provide an easy way to determine parentage if the genotypes o f an offspring and one or ideally both potential parents are available. Directly comparing the genotypes of a parent to its offspring to verify that Mendelian inheritance took place, the exclusion method is the ideal method to be used for parentage assess ment. However, success of this technique depends on a lack of mutations (i.e. allele changes during the inheritance process), genotyping errors (i.e. the misreading or amplification failure of the genotype), and how many individuals are being considered ( Jones et al. 2010) .
50 When complete exclusion can not be attained , c ategorical allocation also uses a focal offspring an d a group of potential parents, however assignments are based on either maximum likelihood or Bayesian approaches (Smouse & Meagher 1994; Jones et al. 2010 ) . The genotypes of the potential parents are presumed to have different probabilities in their contributions towards (Jones et al. 2010) . Briefly, the maximum likelihood approach is used to determine the probabi lity of the data given the hypothesis and uses many data points to maximize the likelihood of the model (Kalinowski et al. 2006) . The Bayesian approach is used to determine the posterior probability of th e model given the data and uses prior or conditioni ng information (prior probabilities) (Neff et al. 2001; Hadfield et al. 2006; Tringali 2006 ) . P arental reconstruction relies on known sibling groups, both full and half siblings to reconst ruct possible parenta l genotypes (Jones 2001) . If the genotype of the shared parent is known, those alleles are removed from all the progenies genotype s , using a similar technique as seen in the exclusion method. The remaining alleles can be used to determine how many other parents are possible an d can reconstruct the potential genotypes of those parents. In order for this method to be the most effective, there shou ld be a large number of progeny and less than six potential parents (Jones et al. 2010) . If these criteria are not met, this method m ay not accurately reconstruct the potential genotypes due to the lack of all alleles being represented in the progeny array. Using more than one method is always recommended and ideally will produce congruent results. All of the discussed methods require that a strong sampling procedure is in place , with at least one relationship known or presumed or if that is not practical than the collection of a large number of individuals from the population (Jones
51 et al. 2010) . If at least one of the parents is know n, the power of the test is greatly enhanced. These methods also require many polymorphic markers to supply the genotypes. Jones et al. (2010) highly recommend microsatellite markers, which are highly polymorphic and repeatable. The occurrence of genoty ping error, mutations , and null alleles can be accounted and controlled for in the parentage analysis technique used. F ield sample and molecular genotype data are used to extrapolate family units, leading to the creation of pedigree diagrams , which in tur n give a more complete picture of the population in question. Manatee Pedigree Studies Pedigree analysis on manatees has been implemented by four graduate students since 2007. The first tested the sufficiency of 18 microsatellite loci (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) to identify captive pedigree relationships and pedigree reconstruction (Pause 2007) . In this study Pause (2007) used two methods, allele sharing and maximum likelihood to determine family units and relationships. Pairwise ge netic distances were then calculated to determine the proportion of shared alleles between individuals and a neighbor joining tree was constructed to visualize relationships. This method grouped the 14 captive individuals into their correct family units. The software program ML RELATE (Kalinowski et al. 2006) was used to calculate relatedness coefficients and identify relationships using a maximum likelihood algorithm. This method was able to correctly discriminate between related and unrelated pairs, bu t the correct relationship was not always identified when compared to known individuals. Pause further states that genotyping error, null alleles, and inbreeding, especially in Family Unit 1, could contribute to the incorrect relationship assignment. The author also concludes that the panel of 18 markers is not robust
52 enough to reconstruct pedigrees, yet is adequate to identify known family groups and suggests using more microsatellite markers for pedigree reconstruction in the future. The next study focused on the manatee population in Crystal River, Florida (Bonde 2009) . In this study the goal was to verify cow calf pedigrees designated by the Manatee Individual Photo identification System (MIPS) (Beck & Reid 1995) using a panel of 11 microsatellit e markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) . These markers were a subset of the most informative of the 18 markers Pause (2007) used in her pedigree study. The software programs CERVUS (Kalinowski et al. 2007) and ML RELATE (Kalinowski et al. 2006) were used for parentage assignment using a maximum likelihood framework to assign the most likely relationship between a pair of individuals. There were 45 mothers and 94 known calves used to genetically verify field observation identifications of cow calf pairs. Multiple sets of twins were analyzed to improve the long term manatee reproductive data available. Out of the three sets of potential twins, only one was confirmed as being related and the other two may be cases of adoption. There we re also four sets of multigenerational individuals: a grandmother, a mother, and multiple grandcalves; both analysis programs agreed with MIPS about these assigned relationships. In 2010, inbreeding and relatedness were investigated with attempts to infer paternity in 19 captive Antillean manatees in European zoos and aquaria (Jacobs 2010). Using 20 microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) , the same ones that Pause (2007) used plus two additional markers, Jacobs found there was a mean individual relatedness of r xy =0.401 among all individuals tested. The founder and outbred animals had negative r xy values, while the rest had a range of
53 positive values from r xy = 0.006 0.74, indicating a range of relationships. A low F IS value was observed for the population, indicating a high level of heterozygosity with limited inbreeding. Inbreeding coefficients could not be estimated for this population due to a lack of sampled and genotyped fathers. Paternity inferences were assessed for three offspring, using exclusion by eye and maximum likelihood estimates through CERVUS (Kalinowski et al. 2007) and PARENTE (Cercueil et al . 2002). All methods converged on the same result. Most recently, a pilot study to determine the accuracy of a p anel of 16 (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) versus 34 (an additional 18 by Tringali et al. 2008b ) microsatellite markers was conducted on a set of 17 captive manatees in Mexico with known pedigrees (Nourisson 2011). The author also wanted to determine the relationship between a potential twin pair, as well as associations of family groups in the wild. ML RELATE (Kalinowski et al. 2006) was used for pedigree reconstruction and to estimate relatedness values for both the captive and wild ma natees. Of the 34 markers only 30 worked for all specimens. The larger panel of markers yielded more accurate relationship assignments than did the panel of 16 markers, yet still reported inconsistent and inaccurate relationships. A program using poster ior probabilities with known population allele frequencies was used to identify the relationship between a potential twin pair. With 30 microsatellite markers, the probability of being twins was calculated at 99.2%, while the probability of being half sib lings was 95.7%. This compared to an individual from another population whose probability of relatedness was 1.7% with 15 markers, indicates a high probability of being twins. Pedigree reconstruction in the wild population did not yield more accurate res ults. Only 16
54 markers were used on this population and inaccurate and not possible relationships were assigned. Nourisson suggests that this was due to the low resolution of the markers. She also suggests using the additional primers that have been publ ished by Hunter et al. (2010b) for future pedigree analyses. The four pedigree studies described above have laid the foundation for more detailed research on T. manatus . Pause (2007) , Bonde (2009) and Nourisson (2011) determined that using fewer than 18 markers was not suitable for reconstructing accurate pedigrees of known lineages. Jacobs (2010) was able to assign fathers to offspring using 20 markers, but only when given a choice of two putative fathers. The reports by Broderick et al . (2007) and Hunter et al . (2010b) indicate that there are 17 polymorphic dugong markers and a new manatee marker. With 54 microsatellite markers available to choose from, a highly informative panel should be able to reconstruct accurate pedigrees of both captive and wild manatees. Pause ( 2007) , Bonde (2009) and Nourisson (2011) all used the program ML RELATE (Kalinowski et al. 2006) . Pedigree Statistical Analysis While a good first step to assigning relationships, ML RELATE (Kalinowski et al. 2006) tends to o verestimate the proportion of unrelated individuals which can lead to underestimating inbreeding levels, especially with few loci and low allelic diversity. Improvements may be made by converting relative likelihoods, which is the ML RELATE (Kalinowski et al. 2006) output, into posterior probabilities, by first subtracting the relative likelihood (LnL(R)) score from the highest likelihood score to produce actual likelihood scores(Ln(R)), and then convert the actual likelihood to odds by transforming to its exponent. Using the formula:
55 e LnU /( e LnU + e LnHS + e LnFS + e LnPO ) ( 1 1 ) The numerator is changed based on which relationship (U=Unrelated, FS=Full sibling, HS=Half sibling, PO=Parent Offspring) is being converted. Then all relationship odds are convert ed to posterior probabilities. Both relative likelihoods and posterior probabilities should be reported to indicate the robustness of the assignment (Tringali 2012 pers. comm.). Improvements may also be possible by using other available pedigree reconstruct ion programs. Jacobs (2010) used both CERVUS (Kalinowski et al. 2007) and PARENTE (Cercueil et al. 2002) which are parentage assignment programs. CERVUS (Kalinowski et al. 2007) is a categorical allocation program which computes a multilocus likelihood ratio (LOD score) for each parent offspring pair. This program can detect null alleles and also accounts for genotyping error and mutations by an input value for the error rate pe r locus. CERVUS (Kalinowski et al. 2007) can create inbreeding simulations using information regarding the level of relatedness between the parents, and the rate at which relatives in the population mate (Jones et al. 2010) . PARENTE (Cercueil et al. 2002 ) infers parentage from genotype data yet uses birth and death dates (if applicable) as priors to ensure that parents are reproductively viable during the birth of the offspring. Additionally, it allows for genotyping errors by allowing genetic incompatib ilities (Cercueil et al. 2002) . Nourisson (2011) used a manatee specific program, manateerelate.xls (Tringali 2010 pers. comm.), which computes pairwise probability of relatedness , similar to the hand calculation improvements to the ML RELATE (Kalinowski et al. 2006) output described above. Other programs that were not mentioned in the previous studies that may be beneficial to pedigree analysis studies include COANCESTRY (Wang 2011) , PARENTAGE (Emery et al. 2001) , and
56 COLONY (Jones & Wang 2010) . The sof tware program COANCESTRY (Wang 2011) , employs seven methods in two categories to estimate pairwise relatedness . The first category uses likelihood estimators that assign pairs of individuals to a specific relationship (i.e. unrelated, halfsibs, fullsibs, or parent offspring , etc.). These estimators take into consideration the possibility of inbreeding in the population. The second category uses moment estimators, which estimates via probability the relatedness of two individuals being identical by descen t. In this program, there are five moment estimators, all of which do not assume inbreeding. Once cow calf relationships are verified, parental reconstruction programs can be used to identify how many different males are siring offspring . PARENTAGE (Em ery et al. 2001) is a parental reconstruction technique that uses a Bayesian based approach. Less informative markers can be used to reconstruct parental genotypes, but may result in an escalated number of potential parents per progeny array. This progra m accounts for mutations and genotyping errors (Jones et al. 2010) . COLONY (Jones & Wang 2010) is used to determine a maximum likelihood ratio to assign parentage and sibling relationships. It does this by concurrent partitioning of the offspring into ma ternal and paternal sibling families and then assigns parents to the groups. If there is not a candidate parent assigned or available, a reconstruction of the potential parent genotypes will be created. This program accounts for null alleles, mutations, and genotyping errors (Jones et al. 2010; Jones & Wang 2010) . The output files from COLONY (Jones & Wang 2010) can be used to create pedigree diagrams. To date , manatee pedigree studies h ave not yielded robust results. Improvements may be made by utilizi ng and combining different software programs ,
57 more robust statistical analyses, and additional polymorphic markers. These changes should provide more thorough pedigree and parentage assignment studies which will have direct impacts on the conservation man agement of the t richechids. Conservat ion Management Through the Use of Pedigree S tudies Conservation management is the maintenance of wild populations, their habitats, and the interactions humans have with the population. This can be accomplished by answering questions pertaining to ecology, behavior, population dynamics, physiology, nutriti on, diseases, habitat loss and restoration, and genetics (TWS 2012) . Genetics plays a key role in conservation management by identifying populations that have diverged genetically and no longer have a high level of gene flow. This allows for the assignme nt of population management units and can be used to designate natural reserves (Beebee & Rowe 2008) . Genetic information can also be used to estimate movement, reproductive, survival, and growth rates at a population level, as well as identify gene diver sity within demographic units through phylogeny and phylogeographic studies (Beebee & Rowe 2008; Tringali et al. 2008b) . Genetic pedigree studies, as was discussed previously, describe the movement of genes through family lineages. These studies allow f or population life history assessments, answering questions about mating systems, and more specifically estimating effective population size, which is the number of individuals in an ideal population that have the same amount of genetic diversity loss as t hat of the adult census population (Lacy et al. 1995; Kruuk & Hill 2008; Tokarska et al. 2009; Lacy et al. 2012) . As was stated earlier, in manatees, reproduction takes place during mating herds, involving one female with multiple males, and an extended m other calf dependency period (Boyd et al. 1999), but very little information is known about the
58 paternity of calves since there is no visual way to identify the father. P aternity assessments are, however, very critical in determining fitness, dominance, r eproductive success, inbreeding depression, inbreeding avoidance, and heritability in a population (Sardell et al. 2010) . If successful males can be identified in the population, not only can reproductive success be identified but this information could c ontribute to the knowledge of how reproduction is divided among the population s , i.e. reproductive skew theory (Port & Kappeler 2010) . Knowledge of paternity could also identify genes unique to those males and may lead to information regarding whether fem ales have input regarding which of the males are selected for propagation . Inferring paternity will complete the pedigree and allow the relatedness of individuals to be compared more thoroughly. The relationship of any two manatees and the inbreeding coe fficient of a potential offspring could then be determined. The answers to questions about reproductive skew, dominance, and relatedness will enable better estimates of the fitness of the population. Pedigree studies also provide information for estimat ing individual inbreeding levels through the inbreeding coefficient ( F x ): F x n+1 (1+F A )] ( 1 2 ) where F x is the coefficient of inbreeding of individual x, n is the number of connecting links between two parents of x through common ancestors, and F A is the coefficient of inbreeding of common ancestor A. This is the probability that an offspring of two relate d individuals will inherit the same allele at a given locus from a common ancestor (Lacy et al. 1995; Frankham et al. 2002; Kruuk & Hill 2008; Tokarska et al. 2009; Allendorf et al. 2012) . To effectively calculate F x , a complete pedigree must be obtained including
59 offspring, parents, and all common ancestors. In most wild populations, manatees included, the pedigree is not known for all individuals making it difficult to calculate the inbreeding coefficient. Estimates of F can be identified by comparing the level of heterozygosity observed in the population to the expected level of heterozygosity if random mating occurred (Allendorf & Luikart 2007; Frankham et al . 2010). High individual inbreeding coefficients throughout the population can indicate a pop ulation with low genetic diversity, leading to reduced reproduction, declining fitness, and decreased survival, i.e. inbreeding depression, which could therefore signify a population in need of drastic conservation management efforts. These efforts could include the translocation of animals into different populations, starting a restocking, or captive breeding program (Lacy et al. 1995; Tokarska et al. 2009; Lacy et al. 2012) . While an excessive amount of inbreeding can cause detrimental effects to a pop ulation, t here are benefits to inbreeding if it occurs in limited amounts . A low level of inbreeding may purge the population of deleterious alleles, by selecting against these genes, removing them from the gene pool , and thereby improving the health of t he population (Lynch 1991; Coulson et al. 1999; Edmands 2007) . Individuals that are considered superior, especially in dogs and horses, are bred for those specific and desirable traits (Langlois 1996; Derry 2003; Valera et al. 2005; Calboli et al. 2008) . This type of artificial selection fixes those traits in the population (Dickerson 1973) . This can be advantageous and is often seen in livestock and agriculture to increase yield, growth , and production (Drolsom & Nielsen 1969; Wehner 1999; Derry 2003) . Yet, t oo much inbreeding can result in a loss of heterozygosity which lowers the amount of genetic diversity in a population (Coulson et al. 1999; Frankham et al. 2002) . Species need
60 genetic diversity to evolve and adapt to the environmental and demograp hic changes that occur. Increased homozygosity, which can lead to the fixation of alleles, increases the chances that deleterious recessive alleles are expressed in the gene pool, can cause inbreeding depression, which is the reduction of fitness in a pop ulation (Roelke et al. 1993; Balloux et al. 2004) . Inbreeding depression adversely affects reproductive characteristics including a reduction in fertility, offspring survival, longevity, and the quality and quantity of sperm (Roelke et al. 1993; Hedrick & Kalinowski 2000; Mc Parland et al. 2007) . Inbreeding can also affect body size and weight as well as other external characteristics such as kinked tails or cowlicks as has been reported in the Florida panther (Hedrick & Fredrickson 2010; Johnson et al. 2 010) . Inbreeding depression increases the risk of extirpation through an extinction vortex, which occurs when low levels of genetic diversity lead to less fit progeny. These individuals are reproductively inferior , which when combined with an interaction with environmental or demographic variation , contributes to a decrease in population size which leads to more inbreeding et al. 2006) . The pr obability of inbreeding increases in small isolated populations, such as those often observed in captivity (Leberg & Firmin 2008) . Captive breeding programs are essential in conserving the endangered and threatened species of the world (Snyder et al. 1996 ; McPhee 2004) . The se programs serve as a protected environment that allows species to persist that may not survive in the wild , due to isolated populations, habitat degradation, or other anthropogenic events (Foose & Ballou 1988; Griffiths & Pavajeau 200 8) . Animals that are bred in captivity can increase a natural population through supplementation, such as releasing captive -
61 bred individuals into the wild or by translocating animals to new areas to encourage genetic diversity (Wyner et al. 1999; Araki et al. 2007; Griffiths & Pavajeau 2008; Ballou et al. 2010) . While fundamentally a good idea, captive breeding can lead to high juvenile mortality rates, a loss of rare alleles, and a decrease o f genetic diversity, as well as reproductive dysfunction due to small population sizes (Snyder et al. 1996; Lynch & O'Hely 2001; Araki et al. 2007) . Genetic changes may also occur in captive populations, for example larger litter sizes or greater survivorship due to a more nutritious diet than what they would experie nce naturally (Lynch & O'Hely 2001; McPhee 2004; Araki et al. 2009) . These changes are not necessarily bad for the population in captivity but can be detrimental if used for supplementation of an existing wild population when reintroduction occurs (Snyder et al. 1996; McPhee 2004) . Captive breeding programs need to insure that genetic diversity is preserved in the population by breeding only individuals that are healthy and genetically distinct from one another while monitoring and tracking pedigrees of a ll captive bred individuals (Templeton & Read 1984; Foose & Ballou 1988; Kalinowski et al. 2000) . Evolutionary genetic questions , as well as questions pertaining to environmental change and the impact it has on the population, can also be addressed using p edigree data (Lacy et al. 1995; Pemberton 2008; Mucha & Windig 2009; Marsden et al. 2013) . The coefficient of relatedness and the inbreeding coefficient of individuals in a population enable inquiries into influences of selection on a population (Gandon 1 999; Hughes et al. 2008) . Examples include the effect of directed selection on natural genetic variation, the effect of environmental changes on inbreeding depression, and to what degree populations avoid inbreeding ( Kruuk & Hill 2008; Pemberton 2008; Lac y
62 2009; Marsden et al. 2013) . Manatees live in an environment that is expected to change significantly in the near future. With the impending retirement of many Florida power plants, the warm water winter environment in which manatees have become depende nt is going to be reduced (Reep & Bonde 2006) . Further study is required to understand the consequences of these adjustments. In order to effectively answer questions about selection and inbreeding using pedigree data, we need to know certain life histor y characteristics about a population. In most pedigree studies, assumptions are made that the founders of the population are both unrelated and outbred (Lacy 2012) . This complicates analyses, especially in small, isolated populations, such as that of th e Florida manatee, due to the fact that true relationships are not known (Koch et al. 2008; Ruiz Lopez et al. 2009) . Simulations can be run to adjust for relationship and inbreeding levels of founders to determine the correct starting point for pedigree a nalyses, as was done with endangered ungulates by Ruiz Lopez et al . (2009) . Other issues that should be considered when performing pedigree studies are mutations, genotyping errors, and null alleles, which can be addressed and accounted for through carefu l selection of genetic markers and appropriate analysis programs, as well as lacking adequate samples of possible parents (Koch et al. 2008) . These problems need to be addressed when using pedigrees to make informed conservation management decisions. In c onclusion, the aforementioned conservation genetic studies have established that genetic diversity is low in the Trichechidae family. Many phylogenetic, phylogeographic, and population genetic questions have been addressed with respect to manatees, but ne w and improved genetic methods and statistical analyses are
63 needed to verify phylogenetic and population genetic structure of all manatee species. Specific population level questions can be answered using pedigree studies, including inbreeding and related ness coefficients as well as correlations with environmental changes and mating systems. These and future studies will aid in conservation management of these imperiled species.
64 Figure 1 1. Manatees gather at Crystal River, Florida in the winter. Picture by Joyce Kleen/USFWS
65 Table 1 1. Mitochondrial DNA sequence divergence comparison of three Trichechus species in six studies. Belize (BZ), Florida (FL), Puerto Rico (PR), Brazil (BR), number of individuals sampled ( n ), genetic distance (Fst and st) , number of haplotypes ( H D F ), genetic diversity ( h ), nucleotide diversity ( ), number of polymorphic sites (S),genetic and geographic distances (Mantel), and number of nucleotide s ubstitutions (NS) are indicated. * represents statistical significance. Source & Species n Fst H Tajima's D Fu's Fs h S Mantel NS Garcia Rodriguez et al. 1998 T. manatus 86 16 0.84 0.040 51 T. inunguis 16 8 0.88 0.005 Cantanhede et al. 2005 T. manatus 81 40 2.33* 2.05* 0.82 0.004 T. inunguis 74 33 2.09* 27.04* 0.89 0.006 42 0.0 9 Vianna et al. 2006 T. manatus 224 0.66 20 2.35 0.86 0.039 45 0.27 T. inunguis 92 0.18 31 2.12* 0.88 0.005 34 T. senegalensis 6 5 0.91 0.93 0.020 15 Hunter et al. 2010 BZ T. manatus 130 0.08 0.04 3 4.12 0.53 0.031 28 0.0 68 FL BZ T. manatus 0.63 0.30 Hunter et al. 2012 PR T. manatus 112 0.46* 4 0.08 0.49 0.001 3 0.0 07 North 22 0.48 0.54* 0.11 East 35 0.01 0.48 0.13 South 37 0.01 0.54 0.00 West 20 0.01 0.54 0.29 FL T. manatus 28 1 0.000 PR FL 0.66* 0.50 Luna et al. 2012 BR T. manatus 73 3 0.08 0.003 0.005
66 Table 1 2 . Population genetic statistical comparisons of Trichechus manatus in seven studies. Belize (BZ), Florida (FL), Gulf of Mexico (GMx), Chetumal Bay (ChB), Ascencion Bay (AB), Puerto Rico (PR), Gulf Coast (GC), East Coast (EC), sampled ( n ), heterozygosity observed (H o ), heterozygosity expected (H e ), average number of alleles per locus (N a ), genetic distance (F st and R st ), migration rate (G st st ), differentiation ( D ) , effective population size (N e ), coefficient of relatedne ss ( r xy ), and inbreeding coefficient ( F IS ) are indicated. * represents statistical significance. Source & Species n H o H e N a F st R st G st G' st D N e r xy F IS Garcia Rodriguez et al. 2000 0.41 2.9 Pause et al. 2007 0.50 4.2 Tringali et al. 2008 0.35 2.5 Hunter et al. 2010 0.63 Hunter et al. 2010 BZ 130 0.46 0.45 3.1 0.031 0.041 0.18 FL BZ 0.141 0.082 Nourisson et al. 2011 GMx 28 0.44 0.41 2.6 0.088 0.131* 0.087 0.114* 27.6 ChB 51 0.47 0.46 3.0 0.047 0.131* 0.016 0.114 32.7 AB 15 0.45 0.43 3.0 0.047 0.094* 0.014 0.087 4.9 FL 95 0.47 0.47 3.6 0.094 0.106* 0.014 0.089 424.3 Hunter et al. 2012 PR 110 0.45 0.45 3.9 0.101* 51.9 0.01 North 0.026 0.028* 0.04 East 0.024 0.026* 0.04 Southwest 0.024 0.028* 0.02 F L 95 0.01 PR FL 0.163* 0.119*
67 Table 1 2. Continued Source & Species n H o H e N a F st R st G st G' st D N e r xy F IS Tucker et al. 2012 All FL 341 0.4 6 0.4 8 4.8 0.012 0.016 0.015 1260 1404.3 0.045 GC 168 0.44 0.47 4.5 0.02* 429.4 1,106 0.046 EC 173 0.4 7 0.4 8 4.2 0.02* 197.2 310.1 0.027 SJR 51 0.44 0.4 7 4 .0 0.018 0.024 0.017 0.052 ATL 122 0.46 0.48 4 .0 0.018 0.024 0.017 0.051 NW 91 0.4 4 0.47 4.3 0.018 0.024 0.017 0.049 SW 77 0.4 6 0.4 7 4 .0 0.018 0.024 0.017 0.031
68 CHAPTER 2 ASSESSMENT OF LATENT COANCESTRY AND INBREEDING IN WILD FLORIDA MANATEES Background Conservation management is the maintenance of wild populations and their habitats as they change through human interactions. Genetic analysis plays a key role in conservation management by identifying populations that have low levels of extrinsic gene flo w resulting in genetic divergences. Assignment of population management units can be used to designate natural reserves of unique genetic diversity and habitats (Beebee & Rowe 2008) . The re are many genetic and environmental factors that determine the abi lity to evolve, including reproductive fitness, population size , relatedness, and inbreeding in . All are major concerns conservation biology, especially in small isolated populations, such as that of the Florida manatee . Relatedness ( r xy ) represents the amount of genetic material shared between two individuals (Weir & Cockerham 1984; Queller & Goodnight 1989; Ritland 1990, 1996; Lynch & Ritland 1999) . Populations with high relatedness values are most likely undergoing some level of inbreeding, reducing t heir fitness ( Amos et al. 2001 ; Chakraborty & Ji n 2014; Fitzpatrick et al. 2014 ) . Inbreeding ( F x ) represents the amount of genetic material that is inherited from a common ancestor, or identical by descent (IBD) (Wright 1922; Ballou 1983) . While relatedn ess values are not directly correlated with fitness, elevated F x values in a population can indicate that random, unrelated copulation is not consistently occurring, reducing the overall genetic health. Major genetic drift events, such as a founder effect or a population bottleneck, leave their mark in the DNA by altering the gene frequencies (Hedrick & Kalinowski 2000; Weber et al. 2004) . These changes can be used to quantify the levels of relatedness and inbreeding of that population over time
69 (Jorde et al. 1998; Balloux et al. 2004; Weir et al. 2006) . High coefficients of individual inbreeding could signal that a population with low genetic diversity is at increased risk from deleterious recessive alleles that are exposed in the gene pool , which could therefore signify a population in need of rigorous conservation management efforts (Crnokrak & Roff 1999; Keller & Waller 2002; Balloux et al. 2004; Charlesworth & Willis 2009) . This could lead to inbreeding depression, which includes reduced reproduction and fertility, declining fitness, and decreased survival and longevity (Roelke et al. 1993; Frankham et al. 2002; Balloux et al. 2004) . Inbreeding can also affect body size and weight , the quality and quantity of sperm , as well as other external characte ristics such as kinked tails or cowlicks as is observed in inbred Florida panther s (Roelke et al. 1993; Hedrick & Fredrickson 2010) . Inbreeding depression can increase the risk of extinction through the genetic process of an extinction vortex, which occur s when low levels of genetic diversity lead to less fit progeny , resulting in imminent extirpation . Reproductively inferior individuals in association with environmental or demographic variation can also contribute to a decrease in population size , which again results in further inbreeding (Gilpin & Soule 1986; Frankham et al. 2002; Fagan & Holmes 2006; . The Florida manatee ( Trichechus manatus latirostris) is an endangered aquatic mammal that lives in the coastal waters of the southe astern United States (Lefebvre & O'Shea 1995; Reep & Bonde 2006) . Their range is mainly peninsular Florida, but the ir extended range includes Rhode Island to the north, Bahamas to the east , and Texas to the west (Powell & Rathbun 1984; Deutsch et al. 2003 ; FWC 2007) . There are currently four management units (MUs) that divide Florida manatees into four subpopulations
70 (FWC 2007) based on winter site fidelity . T he east coast includes the Atlantic (ATL) and the Upper St. Johns (USJ) units , while the west coast includes the Northwest (NW) and the Southwest (SW) units . These MUs delineate groups of manatees that have similar winter distribution patterns and that also experience varying habitat types and anthropogenic concerns (FWS 2001) . W hile both coasts suffer from high levels of cold stress mortality, the s ubpopulation on the west coast experience s a higher number of red tide occurrences (FWRI 2014; NOAA 2014) . The two coasts have population s of approximately equal abundance; yet the AT L region has the highest abundance (47% of the total), followed by the SW (37%) and NW regions (12%) , while the USJ has the lowest (4%) (FWS 2001) . Manatees have been successfully genotyped with up to 36 microsatellite markers and are individually ident ifiable via their DNA throughout their range (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b; Hunter et al. 2010b; Tucker et al. 2012) . Low levels of mean heterozygosity (0.41; 0.501; 0.35; and 0.455) and allelic diversity (2.9, 4. 2, 2.5 and 4.8) were reported (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b; Tucker et al. 2012) . In comparison to the critically endangered North Atlantic r ight w hale ( Eubalaena glaciali s ) and the b lack r hino ( Diceros bicornis ), the genetic diversity of the manatee is considerably lower despite its larger population size (Waldick et al. 2002; Harley et al. 2005; NOAA 2013; WWF 2013) . While manatees do not have any natural predators, they are affected by both natural and anthrop ogenic effects. Natural causes such as perinatal, cold stress, and red tide make up over 60% of known determined causes of death, while human causes,
71 watercraft strikes, flood gates, and other human related incidents (entanglement, entrapment, loss of war m water habitat, etc.) make up the other 40% ( Figure A 1 ) (FWC 2007) . The numbers of watercraft related deaths have been steadily on the rise for the (FWS 2001) . Watercraft continue to be the largest identifiable anthropogenic contributor of manatee deaths (~25%) in the state of Florida (FWS 2001; FWC 2007) . Perinatal deaths, calves with body lengths of <150cm, result in 29% of deaths, the second highest single factor of all known manatee deaths (Figure A 1). Natural causes, which include red tide exposure and cold stress combined contribute 33% of known manatee deaths. While death categories are assigned to recovered carcasses, very little is known about the c ause of some of the major categories. It is understood that if manatees are exposed to red tide, a brevetoxin that induces neurologic and respiratory failure, that drowning is likely (Bossart et al. 1998) or if a manatee is severely struck by a watercraft the likelihood that it will succumb to complications from the injury is high. But could there be a genetic factor contributing to these deaths? Highly related and inbred individuals may have compromised immune systems preventing them from combating an i nfection that an otherwise healthy manatee could defeat. It is well documented that manatees overcome watercraft injuries by calcifying the fracture as a possible means to prevent additional breakage (Reep & Bonde 2006; Clifton et al. 2008) . It is also w ell known that the Florida manatee has low genetic diversity (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008a; Tringali et al. 2008b; Hunter et al. 2010b; Tucker et al. 2012) and with that there is a presumed elevated level of inbree ding, although individual inbreeding values have never been quantified. Once
72 individual inbreeding coefficients are known, associations between different groups can be investigated. In humans, it has been shown that mortality rates of offspring of inbred parents (first cousins or greater) are significantly higher than those of unrelated parents (Stoll et al. 1994; Stoltenberg et al. 1999; Jorde 2001) . O ffspring of consanguineous relationships had a significantly higher rate of illnesses and congenital mal formations than those of non related parents (Abdulrazzaq et al. 1997) . In marine mammals, delphinids such as the vaquita ( Phocoena sinus ) and phocids, have displayed physical characteristics that could be associated with inbreeding, such as polydactyly, deformed vertebrae, and cleft palate. Known inbred mammals (e.g. Florida panther) present with kinked tails, cowlicks, cryptorchidism (non descended, atrophied testicle), and poor semen quality (Roelke et al. 1993; Trupkiewicz et al. 1997; Rojas Bracho & Taylor 1999; Taylor & Rojas Bracho 1999; Berghan & Visser 2000; Ortega Ortiz et al. 2000; Cooper & Dawson 2009) . Some of these conditions including cleft flipper and stillborn births, have been observed in recovered manatee carcasses (Watson & Bonde 1986 ; FWRI 2014 ). In this study, we examine Florida manatee relationships to (1) assess the informativeness and power of multiple panels of polymorphic microsatellite loci, (2) quantify the level of relatedness, inbreeding, and genetic differentiation of the manatee population and subpopulations, and (3) investigate the relatedness and inbreeding levels of certain demographic classifications and biological conditions.
73 Material s and Methods Sample History, Collection, and DNA E xtraction A total of 2,654 specim ens was used in this study. Tissue of the tail from either free swimming (N=582) or recovered necropsied wild manatees (N=2,072) was collected and stored in either a DMSO solution or in 85% EtOH between 1975 and 2013. There were 1,142 individuals from th e east coast of Florida (ATL=1055, USJ=87) and 1,512 individuals from the west coast of Florida (NW=720 SW=792). Genomic CA, USA) (Val encia, CA, USA). F ollowing kit instructions , all specimens were isolated for a total DNA volume of 100 200 uL each. The quantity and quality of the DNA was measured and DNA concentrations were standardized. Of the 2,654 available specimens for use in th is study, only a limited number had complete genotypes, although all were used to calculate allele frequencies. Due to the nature of the analyses an excessive amount of missing data would be unacceptable, therefore a large percentage of the data was elimi nated from further study. The resulting genotypes (N=261) were analyzed in three different spatial groupings, all of Florida (N=200), east (N=60) and west (N=201) coasts of Florida, and each of the four management unit areas (NW [N=51], SW [N=145], ATL [N =61], and USJ [N=43]). For all subpopulations, a specimen was utilized if it had 23 or more scored loci, except USJ which accepted fewer, up to 17 missing loci, due to poor quality and smaller sample size. To ensure accurate comparisons all data were ana lyzed for summary statistics based on spatial groupings and then again at a subset of the loci that all groupings
74 shared. Data were also analyzed based on distinguishing features (Perinatal [N=313] and cleft flippers [N=11]) to assess correlations with re latedness and inbreeding values. Microsatellite DNA A nalysis A panel of 36 poly morphic microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b) divided into 12 multiplex reactions were used to amplify the 2,654 indi viduals. PCR amplifications were performed on either a PTC 100, PTC 200 thermal cycler (MJ Research, Waltham, MA, USA), or a n UnoII thermal cycler (Biometra, Goettingen, Germany). Isolated DNA was polymerase chain reaction (PCR) amplified in 13.4uL react ion volume containing: 0.7uL of 10ng DNA, 8uL DNA free water, 2uL 0.8mM dNTPs, 1.25uL 10x Sigma PCR Buffer,(10nM Tris HCl, pH 8.3, 50mM KCl, 0.001% gelatin) (TmaE08 and TmaE11 used 1.625uL), 1.25uL MgCl 2 ( Tma FWC14 and Tma FWC15 required 2uL MgCl 2 ) , 0.063 units Sigma Jump Start Taq Polymerase, and primer amounts varied from 0.037 0.220uM at a concentration of 100ng/uL (Table 2 1 ). The PCR reaction profile was: 95ÂºC for 5 min, 34x (95ÂºC for 30 s, 54 60ÂºC for 1 min, 72ÂºC for 1 min) final extension 72Âº C for 10 min with a 4ÂºC hold (Table 2 1 for annealing temperature and primer concentrations). Fragment assays used Genescan Rox500 size standard (Applied Biosystems, Inc., Foster City, CA, USA). Fragment analysis was performed on an Applied Biosystems 31 30XL automated genetic analyzer and genotyped using GENEMARKER software version 2.2.0 (Applied Biosystems, Inc., Foster City, CA, USA). Focusing on a subset of the 36 markers (N=18) (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007) , s amples that did not a mplify for all used markers were rerun in additional PCRs to complete the genotype if possible . Two negative controls, two positive controls , and eight individuals were run twice for each PCR to test for error rates and contamination.
75 Additionally, a su bset of microsatellite analyses were conducted at a second laboratory using MBS 0.2S thermocyclers (HYBAID, Potomac MD). PCR reactions contained 1 ÂµL template nDNA, 2.5 ÂµL 1.25mM dNTP mix, 0.125 ÂµL 0.1mg/mL BSA, 2.5 ÂµL Taq Polymerase 10x buffer containing 1.5mM MgCl 2 (Promega), varying amounts (10 pmol/ÂµL) of forward and reverse primers (Table 2 1) and 1.25 U Taq Polymerase (Promega). Fragments were visualized on Applied Biosystems 3130 and 31 30XL automated genetic analyzers and genotyped using GENEMAPPER software version 3.7 ( Applied Biosystems, Inc., Foster City, CA, USA ). A custom ROX labelled size standard was produced in house following DeWoody et al . (2004) . The most informative subset of the polymorphic markers (N=20) are currently being used, while the older data (prior to 2007) used a different subset of 28 markers (GarcÃa RodrÃguez et al. 2000; Tringali et al. 2008b) , and some individuals were run with all 36 markers. Negative and positive controls were included in every run and multiple individuals were run in duplicate to estimate error and detect any contamination. Scoring inconsistencies between labs were tested and accounted for via a multi specimen sample swap to process and analyze in each lab facility. Data were compared between the two labs and conflicting scores were corrected, resulting in consistent manatee genotype scoring throughout Florida. Statistical Analysis M icrosatellite markers were a ssessed for the presence of null alleles using M ICROCHECKER v.2.2.3 (Van Oosterhout et al. 2004) and deviations from H ardy Weinberg equilibrium (HWE), linkage disequilibrium (LD), and allele frequencies via G ENEPOP 4.2 (Raymond & Rousset 1995) . Bonferroni as applied for multiple comparisons. Mean observed (H o ) and expected (H e ) heterozygosity,
76 polymorphic information content (PIC), and null allele frequencies were estimated using C ERVUS v.3.0 (Kalinowski et al. 2007) . K ININFOR v1.0 (Wang 20 06) was used to assess the overall power of the markers to distinguish between relationships and infer relatedness for all six combinations of parent offspring (PO), full sibling (FS), half sibling (HS), and unrelated (U) hypotheses. Allele frequencies fo r all manatee population s in Florida and a significance level of 0.05 were used with equal priors of sharing zero, one , or two alleles identical by descent were assumed. Mean relatedness values ( r xy ) and individual inbreeding coefficients ( F x ) computed fo r all pairwise relationships via COANCESTRY are reported (Wang 2011) . When analyzing pedigree data, it is also important to choose a reliable and accurate estimator which depends on the markers, true relatedness estimates , and population structure of the data (Wang 2011) . In the case of the Florida manatee , with many polymorphic microsatellite loci, likelihood estimators may work better (indicating TrioML and DyadML ). With markers that are biallelic, as almost half of the markers in this data set are, three of the moment estimators ( LynchRd, Ritland , and QuellerGt ) report poor quality . In cases of known highly related individuals, the QuellerGt, LynchLi, and Wang estimators tend to yield more accurate results. Based on t his knowledge, the best estimators would be TrioML , Wang , LynchLi , and DyadML . The maximum likelihood estimator TrioML was used for the estimation of relatedness and inbreeding unless stated otherwise. From mean relatedness values, latent coancestry ( xy ) was calculated following Lynch and Ritland (1999) : r xy =2 xy ( 2 1 )
77 Simulated genotypes (N=200 each) of the three higher order relationships (PO, FS, and HS) and nonrelated pairs were conducted using COANCESTRY (Wang 2011) . Simulation parameters included 79,800 dyads, 200 bootstrapping samples, and 10 0 reference individuals, with varying numbers of loci , missing loci, typing error, and inbre eding were taken into account (Table 2 2) . Simulated and actual genotypes were analyz ed using ML RELATE to quantify overall relatedness of the population (Kalinowski et al. 2006) . GENALEX 6 (Peakall & Smouse 2006) and ARLEQUIN v.3.5 (Excoffier & Lischer 2010) were used to estimate genetic distance and F Statistics, and analysis of mole cular variance (AMOVA) was used to distinguish genetic differences between and among subpopulations in Florida. The likelihood of two individuals in a population sharing the same genotype, probability of identity P (ID) , and a more conservative estimate, probability of sibship P (ID)SIB , were also estimated using GENALEX 6 (Peakall & Smouse 2006) . To investigate alternative genetic differences between regions of Florida, pairwise comparisons of subpopulations and sex were analyzed to assess differences bet bootstrapping method (Wang 2011) . Inferences based on biological conditions (i.e. perinatal and cleft flipper) and relatedness and inbreeding levels were also estimated using the aforementioned tech niques ( Koslovsky & Bassett , Florida Fish and Wildlife Marine Mammal Pathobiology Laboratory 2014 pers. comm. ). Results Florida The 36 markers when an alyzed for all of Florida had a mean of 3.28 alleles per locus (range 2 to 9), a mean Ho of 0.36, and a mean He of 0.40 (Table 2 1). The mean
78 PIC (0.343) was low although seven loci had values greater than 0.44, indicating moderate informativeness (Hildeb rand et al. 1992). Estimates of power to assign categorical relationships ranged from PWR=0.42 to 1.00 (Table 2 3). Assignments between first order and nonrelated relationships proved to be robust, while second order and beyond were less robust. TMAA03, TMAE04, TMAE07, TMAE08, TMAE14, TMAE26, TMAFWC09, TMAFWC11, TMAFWC13, TMAFWC14, TMAH13, and TMASC05 had evidence of homozygote excess, indicating the potential presence of null alleles. Sixteen makers deviated from HWE expectations (p<0.05). Following B onferroni correction, seven deviations remained: TMAE02, TMAE04, TMAE08, TMAE11, TMAE14, TMAK01, and TMASC05. There were four that were corrected with null allele corrections. Eleven pairs were found to be in LD. Five markers were dropped from further a nalysis due to inconsistent scoring, LD, and HWE deviations : TMAA02, TMAE02, TMAE11, TMAFWC12, and TMAK01 ( Table 2 1). Using a subset of individuals representative of the geographic distribution of the Florida manatee population (N=200) and the 31 operat ional loci, the mean r xy was 0.115, while the mean individual F x was 0.108 (Table 2 2). Both values indicated that the Florida sample as a whole was related at the level of first cousins, with a latent coancestry of 0.058. Simulations for the Florida sam ple indicated that 21 % of the popu lation was related , while with observed data, 28% of the population was related (Table 2 2). When analyzed at only 23 loci, the Florida sample (N=262) had a mean of 3.13 alleles per locus and a PIC value of 0.340 . The me an r xy , F x , and simulated percent relatedness value s increased (0.141, 0.134, and 29% respectively). Observed percent relatedness decreased slightly (27% ) (Table 2 2).
79 By Coast West Coast . For the west coast dataset five markers were dropped from analyses due to HWE and LD deviations (summary statistics are given in Tables 2 2 and A 1). The power to assign categorical relationships ranged from PW R =0. 40 to 1.00 (Table 2 3 ). As expected, r obust assignments were observ ed for PO and U ; assignment powers for other relationships were less robust. There was a single private allele found at locus TmaSC05 in the west coast sample. Both simulated and observed results indicated that 24% of the population was related. Utilizi ng 201 individuals and 31 loci, t he overall mean r xy was 0.11 5, which is in the range of values expected for first and second cousins. The overall mean F x was 0.1 10, equivalent to the mating of first cousins/ half siblings (Table 2 2) . With only 23 loci (N=198), the west coast mean r xy value (0.13 1) , the mean F x value ( 0.127 ) , and the simulated percent relatedness within the population (28%) increased substantially, while the observed percent relatedness decreased slightly (27% ) (Table 2 2). East Coast . For the east coast dataset three markers were dropped from analysis due to deviations from HWE and LD (summary statistics can be found in Tables 2 2 and A 2). The power to assign categorical relationships ranged from PW R =0. 42 to 0.99 , with the most robust between PO and U relationships (Table 2 3 ). Using 60 individuals and 33 loci, the o verall r xy range was 0.008 0.15 3 , with a mean of 0.03 8 , which i s equivalent to second cousins and the o verall F x ranged from 0.018 to 0.114, with a mean of 0.06 3, which is in the range of values expected for first cousins . Simulations indicated that 26% of the east coast population was related, while with observed data 25% were related. Using only 23 loci, the east coast group had an
80 increase in mea n r xy (0.137), F x values (0.136), simulated percent (28%) and observed percent (29%) relatedness within the population. Management Units Northwest . Northwest summary statistics can be found in Tables 2 2 and A 1. Four markers were dropped from analyses due to HWE and LD deviations. Power ranged from PW R = 0.41 to 0.99 with the most power to distinguish between PO and U relationships (Table 2 3). Usin g a subset of samples (N=51) and the 32 optimal loci, the mean r xy was 0.118, while the mean F x was 0.120. Simulations indicate d that 21% of the population was related, while with observed data 34% were related. A single private allele was found at locus TMAE11. When analyzed at only 23 loci (N=50), the NW region had a mean of 2.74 alleles per locus and a PIC value of 0.317. The mean r xy increased to 0.1 29 , as did the mean F x value (0.133) and simulated percent relatedness (28%), while the observed perc ent relatedness decreased slightly (32% ) (Table 2 2). Southwest . The Southwest dataset dropped six markers from analysis due to deviations from HWE and LD (summary statistics can be found in Tables 2 2 and A 2). The power to assign categorical relations hips ranged from PW R =0. 41 to 0.97, with the most robust between PO and U relationships (Table 2 3 ). Simulation data indicate d that 21% of the population was related, while with observed data 33% were related. The SW region was related at the level of fir st to second cousins (mean r xy =0.109) and has an F x of 0.10. Private alleles were found at locus TMAFWC02. When analyzed with only 23 loci (n=150), the SW region increased in its mean r xy (0.150), mean F x (0.123), and simulated percent relatedness (28%), while observed percent relatedness decreased (29% ) (Table 2 2).
81 Atlantic . Utilizing 61 individuals and 33 loci, three markers were dropped from analyses due to HWE and LD deviations. Summary statistics can be found in Tables 2 2 and A 2. Power ranged from PW R = 0.42 to 0.9 8 (Table 2 3 ). As was seen with the other regions, assignments b etween PO and U were robust while other relationships were less robust. Simulations indicated that 20% of the ATL region was related, while with observed data 25% were related. The mean r xy was 0.113 , which is in the range of values expected for first and second cousins and the mean F x was 0.109 . Private alleles were found in multiple loci: TMAA02, TMAE01, TMAE26, TMAFWC0 8, TMAFWC09, TMAFWC17, and TMAM79. Using only 23 loci (N=61), the ATL groups mean r xy , F x , simulated and observed percent relatedness values increased (0.138, 0.134, 28%, and 30%, respectively). Upper St . Johns River . All 36 markers were used in analyses. Summary statistics can be found in Tables 2 2 and A 2. Power ranged from PW R = 0.52 to 1.0 (Table 2 3), with the most power being able to distinguish between PO and U relationships. Simulations indicate d that 19% of t he population was related, while with observed data 38% were related. Using a subset of samples (N=43) and all 36 loci, the mean r xy was 0.134, while the mean F x was 0.078. When analyzed at only 23 loci r xy increased to 0. 20 7 , as did the mean F x value (0.116), simulated (26%) and observed (43%) percent relatedness (Table 2 2). Differentiation Values Overall differentiation values were low (0 0.088) yet significant between coasts, MUs, and sexes. There was a significant diffe rence (p<0.05) found between the east and west coast of Florida as well as between the northwest and southwest MUs. Additionally, there were significant differences found in all pairwise comparisons
82 between the sexes between coasts, yet there was no signi ficant difference between the sexes among coasts. F statistics were estimated by MU and coast, respectively for total (F IT =0.080; 0.081), within (F I S =0.064; 0.069), and among samples (F S T =0.018; 0.014). AMOVA results indicated that most variation (98 99% ) was within populations, while only 1 2% of variation was due to differences between populations, both by coast and MUs. Probability of Identity (P (ID ) ) was 1.4E 12 to 8.2E 13 and quantified as between a one in a trillion and eight in 10 trillion chance of identifying two individuals with the same genotype, while probability of sibship (P ( ID )SIB ) was estimated as between a one in a million to eight in 10 million chance of assigning related individuals (Table 2 2). When using fewer loci (N=23) the P (ID ) a nd P (ID )SIB decreased substantially (ex. 1.1E 09 and 1.3E 04, respectively), while most of the fixation indices and F IS values increased considerably, indicating improved robustness with additional markers (Table 2 2). Significant relatedness value differences were observed within the west coast (NW versus SW) and between the NW and the ATL management units (Figure 2 1). There were also significant differences between males and females overall, as well as significant sex differences among the west c oast, east coast, and Atlantic groupings. When combining sex and location, significant differences were also observed between WC males and EC females, NW females and SW females, NW females and SW males, NW females and ATL males and NW males and ATL males, but no significance was detected between the sexes in the same location (Figure 2 1 A). Significant individual inbreeding coefficient differences were found between sexes within the SW, EC, and
83 ATL regions. The sex location combination indicated signifi cance between SW males and NW males as well as ATL females and SW males (Figure 2 1 B). In analyses based on biological conditions, a significant difference was observed between r xy values of manatees having cleft flippers (N=11; pairwise n=55) and the t otal population (N=198; pairwise n=19,568). No significant differences in F x values were detected for manatees having cleft flippers or in r xy or F x values for perinatal carcasses. Discussion Informativeness and Power The overall informativeness of multiple panels of polymorphic microsatellite loci was consistent with what has been previously reported for the Florida manatee (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b; Hunter et al. 2010b; Tucker et al. 2012) . Private all eles were identified primarily in the Atlantic region, which had the greatest genetic diversity of the four management units, yet one private allele was found in the west coast. The Florida manatee has a mean PIC of 0.34 3 , with only six to ten markers in the moderately informative range (all of them dinucleotide repeats) . According to Hildebrand et al. (1992) , a PIC of 0.7 is highly informative, while a PIC of 0.44 is moderately informative. The use of multiplexes increased efficiency when working with such a large number of markers. With the addition of 18 loci (Tringali et al. 2008b) , overall precision increased while the level of diversity at individual markers may not have. Many of the summary statistics were affected by the number of loci used, in cluding mean number of observed alleles, PIC, and fixation index. As the number of loci decreased so did the performance of the marker panel. This was also illustrated with the probability of identity. With 30 or more markers the P (ID) was highly robust and while using only 23
84 loci the P (ID) decreased yet was still substantial, indicating that there is enough power to identify individuals and to detect parentage in the Florida manatee. While the criteria used to identify the number of loci needed for this study indicated that the use of more markers yielded higher power and information, it was not substantial enough to warrant the cost associated with using the full set of markers. It was not until the mean relatedness and inbreeding estimates were as sessed that a substantial difference was observed. With fewer markers, estimates of relatedness and inbreeding typically increased by a classification (inferred relationship), in some cases from between second and first cousins to nearly half siblings. T his is important to consider, especially when quantifying the level of latent coancestry of a population. Latent coancestry needs to be considered when calculating inbreeding coefficients from pedigree data. With the use of 30+ loci, there was substantia l power to distinguish between higher order relationships and unrelated individuals, yet it wa s difficult to discriminate bet ween half sibling and unrelated relationships. Many loci are typically needed to distinguish half siblings from unrelated individu als (Mayor & Balding 2006) and , in this case, more loci are needed to accurately identify and assign specific relationships. Broderick et al. (2007) and Hunter et al. (2010b) report that there are 17 polymorphic dugong and another manatee marker available to use on the Florida manatee population . With access to these 18 loci , pedigrees for both captive and wild manatees have the capability to be reconstructed more accurately. T he ability to precisely identify related versus unrelated individuals is extre mely important when manag ing imperiled species in captivity, especially if considering the use of a captive breeding program as a way to reestablish or sustain a population (Lacy 2009; Ballou et al. 2010) . Further, the
85 ability to genetically identify mana tees that do not have identifiable markings can be used in mark recapture studies and to improve life history parameters for models to estimate more robust survival rates. Relatedness Florida manatee r elatedness levels were found to be related as approximately first cousins on average ( r =0.125). Inbreeding values also suggest that the population is related between the level of first cousins and half siblings. However, relationship estimators do not ty pically assign second order relationship s, therefore any relatedness value quantified as <0.25 was classified as an unrelated relationship . In populations of mammals and birds, it has been reported that 90 99% of pairs within pedigrees of each population had relatedness values of < 0.25 , suggestive of second order relationships (i.e. first and second cousins to unrelated) (CsillÃ©ry et al. 2006) . In manatees, 62 81% (simulated mean 78 % , observed mean 70% ) of pairs within spatial groupings had relatedness va lues of < 0.25 . The percent unrelated of the population and subpopulations can then be used as a biologically relevant prior when analyzing likelihood data. Bayesian procedures , which are designed to determine the probability of a hypothesis given the dat a, can combine the likelihood data and prior information to generate posterior probabilities (Tringali 2006) . Discrepancies between simulated and observed data are due to the actual population being more related than simulated populations. Simulated geno types were generated from allele frequencies for only the four basic relationships (PO, FS, HS, and U). Because second order relationships are not included in the analyses , the Florida manatee population may be more related than the simulations indicate . This is especially the case for the northwest and southwest
86 disparity was probably due to a high number of partial genotypes, indicating that when less data are availa ble there is an upward bias in estimates of relatedness. Therefore, the relatedness value for this region should be taken with caution and re evaluated when more complete genotypes become available. (2006) finding, manatees are much more related than would be expected in a healthy, randomly mating population. As Tucker et al. (2012) indicates, a population bottleneck, founder effect, or a combination of both, may be responsible for these moderate levels of inbreeding and low amo unts of genetic diversity. Presumptions of a recent founder effect have always been associated with the Florida manatee, due to the presence of a single mitochondrial haplotype (Vianna et al. 2006) and low genetic diversity. There is empirical genetic ev idence to suggest that there was a recent bottleneck i n the west coast subpopulation but not i n the east coast subpopulation (Tucker et al. 2012) . This could explain the private alleles detected i n the east coast subpopulation as well as the higher relate dness values for the west coast subpopulation. Inbreeding As would be expected in highly related, small populations, Florida manatee inbreedin g levels are higher than the reported averages for undisturbed mammals and birds ( F =0.053) (Grueber et al. 2008) . Relatedness and inbreeding levels often work in a cyclic nature, where the increase in one is then joined by the other, which is either currently elevated or will become elevated after a generation or two (Chapman et al. 2000; GutiÃ©rrez et al. 2005) . W hen relatedness and inbreeding levels across a population are inflated, consistent genetic evaluations can help to assess the health of the population.
87 The ability to infer individual inbreeding levels would also be a good criteria to consider during the rescue, rehabilitation and release protocols already in place for the Florida manatee (FWS 2001; Reep & Bonde 2006) . Between 1957 and 2013, 1 , 506 manatees were rescued, and of those, 62% survived (FWS 2012) . Many of these manatees (N=59) were rescued mu ltiple (2 8) times over the course of days to decades (FWS 2012) . Few of these rescued manatees were assessed for genetic diversity levels. As rehabilitating a manatee can be very expensive and as budgets diminish in the not so distant future, it may be imperative to consider the overall health and fitness of the population as a whole. This can be determined by examining the genetic fitness of each of the individuals. Limited funds may be more effectively spent on outbreed manatees that can improve the health of then endangered population. High levels of inbreeding can have drastic effects on a population, causing a fixation of alleles, a loss of diversity, and a more rapid and likely threat of extinction (Eldridge et al. 1999; Frankham et al. 2002; Weir et al. 2006; Cristescu et al. 2009) . At an individual level , high inbreeding can cause loss of reproductive ability, decreased immune function , and even mortality (O'Brien & Evermann 1988; Abdulrazzaq et al. 1997; Bener et al. 2007; Grueber et al. 2010) . High levels of inbreeding can also cause negative effects on the estimation of effective population size (N e ), which is the number of individuals in an ideal population that ha s the same loss of genetic diversity as that of the adult census population (L acy et al. 1995; FernÃ¡ndez et al. 2005; Beebee & Rowe 2008; Kruuk & Hill 2008; Tokarska et al. 2009; Lacy et al. 2012) . Genetic Differences The east and west coast subpopulations had significant levels of genetic differentiation, consistent with previous findings (Tucker et al. 2012) . Private alleles on
88 the east coast support the notion of a genetic divergence between the coasts. High F IT and F I S values suggest a population with high homozygosity and low F S T values imply high ge ne flow among populations, indicative of one randomly mating population. However , habitats within the respective coasts of Florida contain different threats, such as less available natural warm water refugia on the east coast and more red tide events on t he west coast. Each coast also provide unique conditions, especially in regard to the habitat, forage availability, and numbers of natural versus man made warm water complexes. Significant differences in the levels of relatedness between the NW unit, an d both ATL and SW units, indicates that the NW unit is highly divergent. The NW unit has a small population than both the ATL and SW units and is known to have multigenerational maternal family lineages; as detected by biologists throughout the year. As is known through radio tracking, photo ID, and behavior studies, K selected female manatees generally teach their calves about necessary habitat requirements to survive. This is reinforced through detection of locations of warm water in winter, food, and fresh water sources, as is evident by high site fidelity for both sexes (Beck & Reid 1995; Deutsch et al. 2003; Langtimm et al. 2004; Reep & Bonde 2006) . The significant relatedness difference between the sexes could be explained by an inbreeding avoidanc e mechanism through dispersal, which is a mating strategy employed by many mammal populations to avoid mating with close relatives. In many systems, adaptations of male biased dispersal is often due to high male mating competition and female choice of mat es (Lehmann & Perrin 2003) . Yet, due to the polygamous mating strategy employed by manatees, it is unlikely or unclear, if females have a choice in
89 mate selection (Reynolds & Rommel 1999; Reep & Bonde 2006) . In other large mammals such as elephants and d olphins, males tend to leave their natal groups and explore greater distances, likely to discover or detect potential mating partners (Nyakaana & Arctander 1999; Perrin & Mazalov 1999; Deutsch et al. 2003; MÃ¶ller & Beheregaray 2004; Lawson Handley & Perrin 2007) . Females tend to apply their energies towards competing for resources rather than competing for mates (Lehmann & Perrin 2003) . This may also be true for female manatees as genetic evidence points to philopatry and less dispersal than males. These findings are consistent with biological data obtained from telemetry studies, as has been documented that generally males travel farther than females, especially in the warmer months of the year when breeding is more likely (Deutsch et al. 2003) . Ratio nale behind the significant inbreeding differences detected in this study between females and males is less straightforward. There are many hypothetical causes, one being that inbred female offspring tend to survive longer, thereby developing to adulthood and reaching sexual maturity with less fecundity. Male offspring may be aborted before their genetic signatures could be implanted into the population. On the other hand, inbred males may out live their female counterparts but may not have been genetica lly sampled. It could possibly also be a recessive x linked abnormality, resulting in males dying while females survive. This concept of one sex having higher inbreeding values than the other has been observed in invertebrates, citing sexual selection as the main hypothesis (Bilde et al. 2009) . The m anatees with cleft flippers or ectrodactyly that were used in this study, (Watson & Bonde 1986) are significantly less related to each other than they are to the
90 total population. This is consistent with what is known from human studies, specific mutations on chromosome seven is the most common cause, although inheritance via autosomal dominance is also prevalent (de Mollerat et al. 2003; Duijf et al. 2003). Most manatee cleft samples were from the east coast of Florida, so there is possibly an epigenetic cause that has not yet been studied. While only a few manatees (N=11) with cleft flippers have been observed and genetically sampled, this significant statistical outcome may prompt additional research. For it has been noted in the wild, that a cow and her calf were both observed with cleft flippers ( Beck 2014 per s . comm.). With an increased sample size, more in depth studies can be implemented, in the future. Summary Given the panel of microsatellite markers, spatial datasets, and analytical methods employed, there was sufficient power and information to assess both individual identification as well as certain genetic relationships within the Florida manatee. Distinct ions could be identified between higher order relationships and unrelated individuals , h owever, assignments involving half sibling and unrelated relationships were generally not robust. This is the first time that relatedness and inbreeding metrics have be en used to compare groups of Florida manatees. Higher relatedness levels in females support observations of female philopatry, and of males dispersing farther distances. The significance of higher inbreeding levels in females is less clear, with many hyp othetical causes including female survivability or an x linked abnormality. Relatedness and inbreeding levels indicate that the Florida manatee population is related on average at the level of first cousins. This high level of latent coancestry and the fa ct that 20 26% of the population is at least a quarter related (HS or above)
91 suggests that the health and fitness of the population is at risk. While the current population appears to be growing, it is imperative that routine monitoring of relatedness and inbreeding levels continues to inform ongoing management efforts and promote the recovery of this endangered species.
92 Table 2 1. Polymorphic microsatellite markers with manatee population diversity information for all of Florida. Garcia Rodriguez et al. (2000) (GR), Pause et al. (2007) (P), Tringali et al. (2008b) (T), annealing temperature (Tm), number of alleles (k), dropped from analysis (**), null alleles (Â§), HWE (Â¶), heterozygosity observed (Ho), heterozygosity expected (HE), polymorphic inform ation content (PIC). See Appendix A for other subpopulations. Primer Tm Primer concentration (uM) Multiplex k H o H E PIC TmaA02 GR 59ÂºC 0.122 1 4 0.383** 0.397 0.320 TmaA03 GR 57ÂºC 0.127 2 2 0.173 Â§ 0.186 0.168 TmaE01 P 54ÂºC 0.134 9 6 0.588 0.596 0.543 TmaE02 GR 57ÂºC 0.067 2 3 0.475 Â¶ ** 0.478 0.364 TmaE04 P 57ÂºC 0.181 10 2 0.249 Â§Â¶ 0.287 0.246 TmaE07 P 57ÂºC 0.220 10 5 0.603 Â§ 0.629 0.557 TmaE08 GR 57ÂºC 0.069 2 3 0.472 Â§Â¶ 0.499 0.395 TmaE11 GR 57ÂºC 0.089 2 9 0.611 Â¶ ** 0.614 0.562 TmaE14 P 56ÂºC 0.103 11 6 0.602 Â§Â¶ 0.698 0.659 TmaE26 GR 59ÂºC 0.098 1 3 0.122 Â§ 0.131 0.123 TmaF14 GR 59ÂºC 0.071 1 2 0.369 0.362 0.296 Tma FWC01 T 58ÂºC 0.079 5 2 0.253 0.265 0.229 Tma FWC02 T 58ÂºC 0.090 3 6 0.546 0.560 0.471 Tma FWC03 T 58ÂºC 0.079 4 2 0.412 0.402 0.321 Tma FWC04 T 58ÂºC 0.098 7 2 0.151 0.154 0.142 Tma FWC05 T 58ÂºC 0.138 6 2 0.179 0.186 0.169 Tma FWC06 T 58ÂºC 0.090 3 3 0.280 0.276 0.257 Tma FWC07 T 58ÂºC 0.055 6 4 0.451 0.435 0.395 Tma FWC08 T 58ÂºC 0.130 7 3 0.254 0.271 0.235 Tma FWC09 T 58ÂºC 0.059 5 5 0.653 Â§ 0.678 0.615 Tma FWC10 T 58ÂºC 0.118 3 2 0.416 0.438 0.342 Tma FWC11 T 58ÂºC 0.039 4 3 0.363 Â§ 0.375 0.342 Tma FWC12 T 58ÂºC 0.157 3 2 0.367** 0.364 0.298 Tma FWC13 T 58ÂºC 0.126 4 3 0.416 Â§ 0.449 0.384 Tma FWC14 T 57ÂºC 0.186 8 2 0.337 Â§ 0.377 0.306 Tma FWC15 T 57ÂºC 0.037 8 2 0.149 0.155 0.143 Tma FWC16 T 58ÂºC 0.102 6 2 0.494 0.494 0.372 Tma FWC17 T 58ÂºC 0.055 7 3 0.134 0.140 0.131 Tma FWC18 T 58ÂºC 0.095 5 2 0.391 0.384 0.310 TmaH13 P 54ÂºC 0.099 9 3 0.311 Â§ 0.332 0.306 TmaJ02 P 60ÂºC 0.094 12 2 0.361 0.354 0.291 TmaK01 P 54ÂºC 0.107 9 6 0.599 Â¶ ** 0.464 0.418 TmaKb60 P 60ÂºC 0.091 12 2 0.509 0.491 0.370 TmaM79 GR 59ÂºC 0.086 1 3 0.489 0.487 0.369 TmaSC13 P 56ÂºC 0.118 11 3 0.394 0.383 0.310 TmaSC05 P 60ÂºC 0.118 12 4 0.533 Â§Â¶ 0.657 0.583 Average 3.28 0.359 0.401 0.343
93 Table 2 2. Comparisons of the Florida manatee population and subpopulations based on different microsatellite marker panels. All of Florida (FL), by coast [east coast (EC) and west coast (WC)], and by the four management units [Atlantic (ATL), Upper St . Johns ( USJ), Southwest (SW) and Northwest (NW)]. Number of samples (N), mean number of observed alleles (k), mean polymorphic information content (PIC), mean relatedness ( r xy ), mean individual inbreeding value ( F x ), percent of the population that is unrelated ba sed on simulations (Sim % U), percent of the population that is unrelated based on observed data (Obs % U), number of loci potentially containing null alleles (Null), probability of identity (P (ID) ), probability of sibship identity (P (ID)SIB ), individual to subpopulation inbreeding coefficient (F IS ), and fixation index (F). Two panels were used to investigate information and power of the markers at datasets with a large number of loci and a reduced set of loci. The reduced panel was chosen to easily comp are subpopulations for the same set of markers. Regions N # of Loci k PIC rxy F x Sim % U Obs % U Null P (ID) P (ID)SIB F IS F FL 200 31 3.28 0.343 0.115 0.108 79% 72% 12 1.4E 12 2.0E 06 0.026 0.027 EC 60 33 2.77 0.339 0.111 0.112 74% 75% 6 2.3E 12 2.3E 06 0.055 0.028 WC 201 31 2.71 0.322 0.115 0.110 76% 76% 5 1.8E 12 2.2E 06 0.031 0.026 ATL 61 33 3.08 0.345 0.113 0.109 80% 73% 7 2.7E 13 8.7E 07 0.028 USJ 43 36 2.53 0.334 0.134 0.078 81% 62% 0 1.1E 13 4.3E 07 0.023 NW 51 32 2.81 0.327 0.118 0.120 79% 66% 3 5.6E 12 3.3E 06 0.045 SW 145 30 2.89 0.339 0.109 0.100 79% 67% 5 8.2E 13 1.5E 06 0.011 FL 262 23 3.13 0.340 0.141 0.134 71% 73% 10 1.6E 09 5.8E 05 0.031 0.042 EC 106 23 2.7 0 0.335 0.137 0.136 72% 71% 4 3.1E 09 7.8E 05 0.005 0.033 WC 198 23 2.91 0.337 0.131 0.127 72% 73% 4 1.6E 09 5.7E 05 0.026 0.032 ATL 61 23 3 .00 0.336 0.138 0.134 72% 70% 4 3.0E 09 7.8E 05 0.039 USJ 45 23 2.57 0.339 0.207 0.116 74% 57% 0 6.6E 09 9.5E 05 0.064 NW 50 23 2.74 0.317 0.129 0.133 72% 68% 3 9.1E 09 1.3E 04 0.020 SW 150 23 2.83 0.340 0.150 0.123 72% 71% 3 1.1E 09 4.8E 05 0.044
94 Table 2 3. Power of inference to distinguish between relationship dyads (PW R ). The first number is the power to distinguish the first relationship from the second and second number is the reverse comparison. Parent offspring (PO), full sibling (FS), half sibling (HS), unrelated (U), Florida (FL), East Coast (EC), West Coast (WC), Atlantic (ATL), Upper St. Johns River (USJ), Northwest (N W), and Southwest (SW). Dyad FL EC WC ATL USJ NW SW PO U 0.99; 1.00 0.98; 0.99 0.98; 1.00 0.97; 0.98 0.99; 1.00 0.98 ; 0.99 0 .95; 0.97 FS U 0.95 ; 0.95 0.95 ; 0.94 0.93 ; 0.93 0.94 ; 0.93 0.97 ; 0.97 0.94 ; 0.94 0.91 ; 0.90 PO HS 0.80; 0.58 0.80 ; 0.61 0.79 ; 0.55 0.79 ; 0.61 0.87 ; 0.72 0.80 ; 0.58 0.76 ; 0.57 FS HS 0.55; 0.57 0.58 ; 0.57 0.55 ; 0.54 0.58 ; 0.56 0.64 ; 0.63 0.56 ; 0.55 0.55 ; 0.54 PO 0.42; 0.60 0.42; 0.61 0.40; 0.60 0.42 ; 0.60 0 .52; 0.71 0.41 ; 0.60 0.41 ; 0.58 HS U 0.48 ; 0.48 0.49 ; 0.50 0.46; 0.45 0.49 ; 0.50 0.54 ; 0.54 0.47 ; 0.48 0.45 ; 0.46
95 Figure 2 1 . Mean difference between groups. (A) Significant difference s between groups based on the TrioML relatedness estimator . (B) Significant difference s between groups based on in dividual inbreeding coefficients . Symbols used: Northwest (NW) Southwest (SW), East Coast (EC), West Coast (WC), Atlantic (ATL), Male (M), and Female (F). Out of the 26 mean relatedness pairwise comparisons based on location and sex by location, 11 were quantifi ed as significant. Females were more related than males and individuals from the NW region were more related than individuals from other locations. From the 26 mean inbreeding pairwise comparisons only six were reported as significant. Females were iden tified as being more inbred than males.
96 CHAPTER 3 RELATEDNESS AND INBREEDING UTILIZING CAPTIVE FLORIDA MANATEE FAMILY UNITS Background Incorporating genetic pedigree information with life history data is a crucial step in developing a robust conservation m anagement plan for imperiled species (Lacy et al. 1995; Frankham et al. 2002; Pemberton 2008) . The joining of life history and genetic information is best seen in captive settings . Here complete family trees are documented and can be genetically confirme d through pedigree analysis. Multigenerational pedigree analyses often involve behavioral observations of copulation and genetic surveys of neutral gene variation (Frankham et al. 2002; Allendorf et al. 2012) , however b ehavioral observation s are challengi ng in wild manatee populations . Constant visualization is difficult, as is the ability to repeatedly sample the same group of individuals, document their reproductive events, and collect data over multiple generations (MacArthur 1967) . Therefore, related ness and relationships between wild manatees must be inferred through the computation of genetic metrics (Grueber et al. 2011; Allendorf et al. 2012) . The Florida Manatee The Florida manatee is an endangered aquatic mammal that lives in the marine and fresh waters of the southeast United States. The census population was estimated at 5,000 individuals with a majority of the population distributed along the east coast of Fl orida (FWRI 2010; FWS 2013) . Manatees reach lengths of 2.7 3.0m (8.9 10 ft.) and weights of 400 550kg (882 1213 lbs.), and a can have long life spans, greater than 60 years in captivity (Reep & Bonde 2006) . Manatees are mostly solitary mammals that aggre gate in large groups at warm water sites in the winter. They reach sexual maturity
97 between 3 5 years of age, have an approximate one year gestation period , and tend to have calves every three years (Rathbun et al. 1995; Reid et al. 1995; Reynolds & Rommel 1999) . Manatees usually have only one calf at a time, and although rare, twins do occur (1.5%). Manatee family units usually consist of a female and her calf, known as a cow calf pair. The calf dependency period is 1 4 years, with strong mother calf bo nds (Reid et al. 1995) . Learned traditions are passed down from mother to calf and can be seen in manatees through their strong site fidelity, where known animals are seen in the same relative area year after year and sometimes for generations (Beck & Rei d 1995) . Other traditions include, migration routes , as well as the locations of fresh water sources, seagrass beds, and warm water sites. Being that mothers pass this knowledge onto their calves and those calves pass it on to their calves, larger abstra ct family units of matrilines have been witnessed through photo identification and can be constructed through genetic analysis (Beck & Reid 1995) . Captive Florida manatees are managed using a studbook (FWS 2012) . All manatees that are brought into captivi ty are assigned a unique number, which is used to document its movement among facilities , life history events, and any subsequent release. W ith the enactment of the Marine Mammal Protection Act (1972) , the Endangered Species Act (1973) , and the Florida Ma natee Sanctuary Act (1978) , manatees may not be captured for the sole purpose of public display. Captive breeding was never an official management strategy, and is now forbidden under current captive conditions. Historic pregnancies and births were likel y a by product of insufficient planning or a desire for the publicity generated by births of calves. The fourteen captive
98 manatees used in this study are documented within the studbook and their ancestry has been confirmed by direct observation of captive husbandry teams and managers. Investigations of Relatedness For years, researchers have been interested in determining p edigree r elationships by using multilocus genetic data (reviewed in Jones et al . 2010). Various metrics for quantifying relatedness can be computed based on pairwise comparison s of genotypes between individuals. Prior studies conducted on all subspecies of Triche chus manatus were not able to reconstruct pedigrees of known manatee lineages using up to 18 microsatellite loci (Pause 2007; Bonde 2009; Nourisson 2011) . One study was able to assign paternity using 20 loci, but only when given a choice between two putat ive fathers (Jacobs 2010) . Limited in the number of loci, these studies may also have been hindered by the analytical approaches used. Fortunately, additional loci are available for the manatee and a suite of analytical methods and estimators can be expl ored. These additions could allow for more robust investigations of pedigrees and assignments of categorical relationships than previously achieved. In addition to examining specific pairwise relationships among members of the Florida manatee population, knowledge of the overall level of inbreeding/diversity is important to understanding its vulnerability to natural and anthropogenic effects, such as new diseases, carrying capacity limits, and habitat and/or climate change. A healthy and stable populatio n is more likely to withstand challenges, whereas a highly related and inbred population would be more vulnerable to these challenges , if their fitness was compromised. Currently , there are no published estimate s for the level of relatedness of the Florid a manatees. To date, values in manatees are only available in the Belize manatee. O verall levels of relatedness in th is Antillean manatee ( Trichechus manatus
99 manatus ) population were high (0.184) and approximately 20% of its members were full siblings (H unter et al. 2010a) . The level of population relatedness for the Belize Antillean manatee is among the highest recorded for marine mammals. By comparison, relatedness in the Sperm Whale ( Physeter macrocephalus ) (Ortega Ortiz et al. 2012) was estimated to be 0.05, while in Polar Bears ( Ursus maritimus ) it was 0.04 (Zeyl et al. 2009) and in male Australian bottlenose dolphins ( Tursiops aduncus ) it was 0.09, with females slightly higher at 0.16 (MÃ¶ller et al. 2001; MÃ¶ller et al. 2006) . Usin g a captive population with a known pedigree, baseline genetic data from manatees east coast population, the objectives of this study were to : (1) apply the existing microsatellite l oci and latent coancestry to a captive population to investigate pedigree reconstruction , (2) identify the relationship and relatedness estimators that are best suited to the population , (3) determine baseline diversity and inbreeding statistics from a cap tive population, and (4) apply this information to wild calves with their mothers. Materials and Methods Sample H istory, C ollection, and DNA E xtraction Fourteen captive manatees from three family groups were used in this study (Figure 3 1). Manatees were housed at facilities around the state of Florida, including Homosassa Springs Wildlife State Park (HSWSP), Lowry Park Zoo (LPZ), Mote Marine Laboratory (MML), Miami Seaquarium (MSQ), and Sea World Orlando (SWF) (Table 3 1). Family unit one, originally mai ntained at MSQ, consist ed of the parental pair Romeo and Juliet, and their five known offspring, Lorelei, Foster, Hurricane, Buffet, and Aurora (genetic sample unavailable ; Table 3 1 ) . Family unit one also included the inbred offspring, Hugh and Stoneman, both fathered by the patriarch, Romeo with his
100 daughters Lorelei and A urora. Family unit two consisted of Gene, Rita, and Dundee all formerly housed at SWF. Family unit three consisted of Ocean Reef, Patch and Pumpkin all formerly managed at MSQ in 2006 . Specimens for the two wild case studies consisted of (1) 18 deceased calves as well as two potential mothers that were encountered during health assessments of live animals in Brevard County, FL in winter 2010. During the assessments, tissue samples w ere collected and stored in a DMSO solution for genetic analyses. Two lactating adult females were radio tagged, released, and monitored for a period of 90 days . No calves were observed with or around these females before or after assessments (Deutsch & Barlas 2011) . Tissues from the 18 deceased calves were obtained from Florida Fish and Wildlife Conservation Commission (FWC) staff and stored in 9 5% EtOH; carcasses were recovered from Brevard County and the adjoining Indian River County, FL, also in wint er 2010. The second (2) case study included genotype data from 34 pregnant adult female carcasses and the unborn fetuses (N=68) (Fish and Wildlife Conservation Commission, 2014). There were 20 pairs from the east coast and 14 pairs from the west coast. A limited number of loci (N=16) were run with this set of data. The majority (N=63) had a genotype with 14 or more loci while four were missing three loci and one was missing four loci. Genomic DNA was isolated from the 34 pedigree and case study specimen s using the DNeasy Blood and Tissue Extraction k it (Qiagen, Valencia, CA, USA), quantities and purities were measured and all concentrations were standardized to 10 ng/Âµl.
101 S tudy L ocations All captive locations, HSWSP, LPZ, MML, MSQ and SWF have hous ed manatees since the 1970s. Four of the facilities, MSQ, SWF, LPZ, and HSWSP , are part of the Manatee Rescue, Rehabilitation, and Release Program, caring for sick, injured, and o rphaned manatees with the goal of releasing them back into the wild. M ote Marine Laboratory is an educational/research facility housing two captive born manatees. Case study, wild samples were collected from t he two power plants in Brevard County, Florid a Power and Light (FPL) and Orlando Utilities Commission, are known manatee warm water site s (Deutsch & Barlas 2011) . This region has been well studied for the last 35 years with over 100 manatees radio tagged to provide movement information on behavior and habitat use. Microsatellite DNA A nalysis A panel of 33 polymorphic microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b) divided into 12 multiplex reactions were used to amplify the 68 individuals. Published primers , TmaA03, TmaE02, and TMAFWC02 , were not used due to inconsistent scores . All PCR amplifications were performed on either a PTC 100, PTC 200 thermal cycler (MJ Research, Waltham, MA, USA), or a UnoII thermal cycler (Biometra, Goettingen, Germany). Isolated DNA was polymerase chain reaction (PCR) amplified in 13.4Âµl reaction volume containing: 0.7Âµl of 10ng DNA, 8Âµl DNA free water, 2Âµl 0.8mM dNTPs, 1.25Âµl 10x Sigma PCR Buffer,(10nM Tris HCl, pH 8.3, 50mM KCl, 0.001% gelatin) (TmaE08 and TmaE11 used 1.625Âµl), 1.25Âµl MgCl 2 ( Tma FWC14 and Tma FWC15 required 2uL MgCl 2 ) , 0.063 units Sigma Jump Start Taq Polymerase, and primer amounts varied from 0.037 0.220uM at a concentration of
102 100ng/Âµl (Table 2 1 ). The PCR reaction profile was: 95ÂºC for 5 min, 34x (9 5ÂºC for 30 s, 54 60ÂºC for 1 min, 72ÂºC for 1 min) final extension 72ÂºC for 10 min with a 4ÂºC hold ( Table 2 1 for annealing temperature and primer concentrations). Fragment assays used Genescan Rox500 size standard (Applied Biosystems, Inc., Foster City, CA , USA) and was performed on an Applied Biosystems 3130XL automated genetic analyzer and genotyped using GENEMARKER software version 2.2.0 (Applied Biosystems, Inc., Foster City, CA, USA). Two negative controls, two positive controls, and eight individuals were run twice for each PCR to test for error rates and contamination . Statistical Analysis All 33 microsatellite markers were assessed for the presences of null alleles using M ICROCHECKER v.2.2.3 (Van Oosterhout et al. 2004) and deviations from Hardy We inberg equilibrium (HWE) and linkage disequil i brium (LD) via GENPOP 4.2 (Raymond & Rousset 1995) . was applied for multiple comparisons. Mean observed (H o ) and expected (H e ) heterozygosity, polymorphic information content (PIC), an d null allele frequencies were estimated using CERVUS v.3.0 (Kalinowski et al. 2007) . KININFOR v1.0 (Wang 2006) was used to assess marker informativeness, the overall power of the markers to distinguish between relationships and infer relatedness for all six combinations of parent offspring (PO), full sibling (FS), half sibling (HS), and unrelated (U) hypotheses. Parameters include diploid diploid dyad, 1.0 for all Beta priors, 0.05 Significance level, 0.01 precision, and a length and simulation length of 1,000,000. Allele frequencies for the manatee population of the east coast of Florida and a significance level of 0.05 were used and equal priors of sharing zero, one , or two alleles identical by descent were assume d. Mean relatedness values and inbreeding coefficients for the east coast of Florida
103 manatees were determined via COANCESTRY (Wang 2011) . Simulation parameters included 85 , 905 dyads of 1,200 individuals, 200 bootstrapping samples, and 100 reference individuals, with 31 loci. Parentage Analysis The relatedness coefficient ( r xy ) represents the proportion of genes identical by descent (IBD) between two in dividuals. For any two related offspring x and y , expected values of r xy were calculated following the path analysis approach described in Lynch and Walsh (1998) using the equations : 1/4(1+ F P + 2 P ) for full siblings ( 3 1 ) And 1/8(1+ F P ) + Â¾( P ) for half siblings ( 3 2 ) Herein, P denotes the coefficient of coancestry between pairs of parents. F P is the inbreeding coefficient within a given parent P (Wright 1922) and was computed using the equation, where n is the number of generations from the parents of parent P to a common ancestor A, and F A is the coefficient of inbreeding of A: F P n+1 (1+ F A )] ( 3 3 ) Because P is associated with the coefficient of relatedness; i.e., r xy = 2 P (Lynch & Ritland 1999) , the kinship Table 3 2 ) were used to estimate expected r xy values ( r xy = 8 Categorical Relationship Estimators ( R xy ) CERVUS. Assigning parentage was performed using CERVUS 3.0 (available from http://www.fieldgenetics.com , Kalinowski et al . 2007 ). CERVUS is a categorical allocation program which computes a multilocus likelihood ratio (LOD score and Delta) for each parent offspring pair. This program can detect null alleles and also accounts
104 for genotyping error and mutations via error rate per locus. Simulation parameters included 100,000 offspring, 300 candidate mothers with a 75% sampling rate, and 500 candidate fathers with a 60% sampling rate. To accommodate for mistyping, null alleles, and other genotyping error, a 97.8% loci typed and 0. 0 196% error rate were used in calculations . These means were derived from duplicate samples run and analyzed for Florida manatee samples . Confidence levels were strict (95%) and relaxed (80%) delta scores. In the captive study, all males (N=9) and femal es (N=5) were considered fathers and mothers respectively, while only those individuals (N=9) that were known to be offspring were considered offspring. ML RELATE. C ategorical relationships , R xy , were estimated among pairs using the methods described in Kalinowski et al . (2006) . Specific relationships were classified (U , HS, FS, and PO ) . The null allele option was implemented to account for presumptive null allelism in TmaA02, TmaE08, TmaE04, TmaE07, TmaE14, and TmaSC5. To improve assignments, relativ e likelihoods were converted into posterior probabilities . Initially, incorporating neutral priors, where all relationships were equally likely to be assigned and then, using a biologically relevant prior that was derived from the overall mean relatedness of the population. The specific value was calculated using east coast allele frequencies, which simulated 200 PO, FS, HS, and U pairwise genotypes (N=1600 total). These simulated genotypes were then analyzed in ML Relate for percent unrelated. Relatedn ess Estimators ( r xy ) ML RELATE. Relatedness coefficients r xy were calculated using the software program ML RELATE (Kalinowski et al. 2006) , which implements a maximum likelihood
105 method of individuals having zero, one, or two genes identical by decent. Th e same parameters as described above were used for this estimate as well. COANCESTRY. Parentage and sibling relationships were evaluated using the software program COANCESTRY 22.214.171.124 (Wang 2011) , which offers seven different relatedness estimators: two likelihood ( TrioML and Dyad ML ) and five moment estimators ( Wang , LynchLi , LynchRd , Ritland , and QuellerGt ) (Lynch 1988; Queller & Goodnight 1989; Li et al. 1993; Ritland 1996; Lynch & Ritland 1999 ; Wang 2002; Milligan 2003; Wang 2007) . Likelihood estimators assign pairs of individuals a likelihood of specific a relationship (i.e. 0=U, 0.25=HS, 0.5=FS, or 0.5=PO) and take into account the possibility of inbreeding in the population. Moment estimat ors take into account the nine descent modes, which track the inheritance of a gene between two individuals. Herein, only the likelihood estimator TrioML and the moment estimator Wang were evaluated. Analytical parameters included 100 reference dyads and 200 bootstrap replicates; missing loci, typing error, and inbreeding were taken into account. Inbreeding coefficient ( F x were computed for all pairwise relationships. Simulations were conduc ted to determine plausible ranges of r xy for each relationship and cut off values were determined to assign relationships to values (<0.23 U, 0.231 0.433 HS, >0.4331 FS/PO) . Results Genetic Marker A ssessment The 33 markers used in this study had a mean of 3.06 alleles per locus (range 2 to 6). TMAA02, TMAE04, TMAE07, TMAE08, TMAE14, and TMASC05 had evidence of homozygote excess , indicating the potential presence of null alleles. Twelve markers deviated from HWE expectations (p<0.05). Following Bonferroni correction, seven
106 deviations remained: TMAA02, TMAE07, TMAE08, TMAE14, TMAK01, and TMASC05. Of these, all except TMAK01 were corrected for with null allele corrections in all parentage programs used. Two pairs of markers were found to be in linkage dise quilibrium : TMAE08 TMAFWC09 and TMAFWC10 TMAFWC12 . Due to LD and HWE deviations, t wo markers were dropped from analyses : TMAK01 and TMAFWC10 . The mean PIC value was 0.345 (range of 0.044 0.621) (Table s 2 2 and A 2 ). The power to assign categorical relationships ranged from PW R =0.49 to 0.98 (Table 2 3 ). Robust assignments could be expected between PO and U whereas assignment powers for other relationships were shown to be less robust. Overall relatedness values from COANCESTRY ranged from 0.0080 0 .1527, with a mean of 0.0377, which is equivalent to second cousins. Overall inbreeding coefficients ranged 0.0177 0.1144, with a mean of 0.0626 , equivalent to first cousins . Relatedness and Relationship Estimates There were 91 pair wise comparisons, whi ch included 15 PO, 7 FS, 10 HS, and 59 U presumed relationships. Estimated values of R were converted from log likelihoods to posterior probabilities (PP) using a uniform prior. ML RELATE assigned 16 PO, 14 FS, 28 HS, and 33 U relat ionships (Figure 3 2 ) . Relatedness value ranges can be seen in Figure 3 3 . Expected inbreeding and relatedness can be seen in Table 3 2 and Table B 1. Inbreeding estimates from TrioML ranged from 0 0.495 (Table 3 1 ). Of 59 assumed u nrelated relationships , 54.4% were correct ly assigned . Half sibling relationships were assigned to 42% of the presumed unrelated individuals. The posterior probability of the relationship between Foster and Ocean Reef was highest for PO (PP=0.61). When using the exclusion method Foster and Ocea n Reef were assigned a 79% probability of being PO and a 21% probability of being HS. Simulation
107 results indicated that 74% of the east coast population was unrelated , which could then be used as biologically relevant prior information in posterior probab ility calculations . Using 0.74 as a prior U, the accuracy for assigning unrelated relationships increased from 54.4% to 94.74%. Correct assignments were inferred for s even of the eight cow calf relationships. Strict confidence (95%) was assigned to 62. 5% (5 of 8) of relationships, 25% of relationships were assigned with relaxed confidence (80%), and one was not assigned but was correctly indicated as the most likely candidate parent ( Juliet with Hurricane). Bull calf relationships were not as accuratel y assigned as the cow calf pairs. One pair was assigned with strict confidence (Gene and Dundee), one pair with relaxed confidence (Romeo Hugh), and five relationships were not assigned parentage. Hurricane was not paired with his presumed father, Romeo. He was paired with Foster and Buffett, with Buffett being the more likely candidate. Ocean Reef and Foster were assigned PO with relaxed confidence. Of the six offspring with both parents known (Buffett, Dundee, Foster, Hugh, Hurricane, and Lorelei) CER VUS correctly assigned both parents to three relationships . Romeo Juliet Buffett and Gene Rita Dundee were assigned with strict confidence. Hugh Juliet Lorelei and Hugh Juliet Hurricane were incorrectly assigned as parent offspring trio s . Romeo or Hugh were the two most likely candidate fathers for Hurricane and Lorelei, with Hugh incorrectly assigned as the most likely father. Family Units Family Unit 1 . There were 28 pair wise comparisons within this family unit. In total, 64% (18/28) of the relationships were correctly assigned. The majority of incorrect assignments (8/10) involved the two inbred individuals (Stoneman and Hugh) , when
108 removed, 87% (13/15) of relationships were correctly assigned . Using 0.74 as a prior, the percentage of corr ectly assigned relationships decreased to 61%. Romeo and Juliet, the parents of this family unit and presumed to be unrelated, were initially assigned a joint posterior probability of 0.93 of being related (as opposed to unrelated). Employing the prior, this joint posterior probability decreased to 0.60. Both relatedness and relationship estimates indicated at least a HS relationship ( r xy =0.21 0.34). All presumed full sibling relationships had relatedness values approximating 0.5, indicative of full si blings ( r xy =0.26 0.75) with 95% confidence intervals (CIs) ranging from ( 0.03 1.21). The ten presumed half sibling relationships ranged from 0.16 to 0.6 and all had very large 95% CIs. These included seven that were both half sibling and aunt/uncle and one both half sibling and first cousin, which would increase relatedness values above r xy =0.25 of the typical half sibling. The eleven presumed PO pairs all had r xy values ranging from 0.41 to 0.94. Family Unit 2 : All assigned relationships were correc t. Two (Rita Gene and Gene Dundee) had high probabilities (PP>0.74), while Rita and Dundee were assigned a PO with low er probability (PP=0.59). Rita and Dundee did not match at one locus, due to Dundee missing four alleles, which caused the low er relatio nship probability. Using the prior U did not change these relationships. All estimators indicated that Gene and Rita were unrelated ( r xy = 0.1 6 0), yet with large 95% CIs ( 0.7 5 0.3 9 ). Relatedness values between Dundee and his parents were as expected w ith r xy =0.4 1 0.5 8 but with wide 95% CIs (0 0.34) . Family Unit 3 : All assigned relationships were correct when compared to captive records. Both HS (PP=0.04) and FS (PP=0.97) were identified for the Patch and
109 Pumpkin pair , because the father was unknown . Relatedness values between Patch and Pumpkin were approximate ly 0.5 ( r xy =0.5 1 0.5 5 ), indicative of a full sibling relationship. Using a prior U did not change these relationships. Relatedness values between Ocean Reef and offspring Patch and Pumpkin were slightly higher than 0.5 ( r xy =0.54 0.7 6 ). All 95% CIs were large. Ocean Reef and Foster, a presumably non related pair, had high relatedness values ( TrioML r xy =0.38 and Wang r xy =0.61 ), indicating a potential parent offspring pair or another higher order relationship . Expected Values of Relatedness Expected relatedness values ( r xy ) for all pairs of relationships can be seen in Table B 1. Expected values were seen for relationships within family units 2 and 3 and between family units 1 and 2. Observed r e latedness values were higher than expected for family unit 1. Values for FS and most PO were 0.64, however, Romeo with his inbred sons was 0.7 7 , and Lorelei with Hugh was 0.8 3 , suggesting prior inbreeding. R fspring and the two inbred sons were 0.4 9 , the HS/first cousin relationship of Hugh and Stoneman was 0.3 8 , and the grandmother relationship of Juliet to Hugh and Stoneman was 0. 35, also indicative of inbreeding . Relatedness values for the plausible relati onships between family unit 1 and family unit 3 can be seen in Table 3 2 and Table B 1 . Case Studies Brevard County maternity test . There was a consensus with all estimators that cow 2 and calf 12 and cow 1 and calf 16 are potential parent offspring pairs . ML RELATE reported a posterior probability of 0.63 PO for cow 2 calf 12 and a 0.50 PO for cow 1 calf 12 ( TrioML= 0.66 and Wang =0.63) and cow 1 calf 16 ( TrioML= 0.4 1 and Wang =0.3 6 ) had relatedness
110 values consistent with PO, but also considered two other pairs ( TrioML= 0.46 and Wang =0.4 2; TrioML= 0.4 1 and Wang =0.3 7 ) to be PO pairs as well. Within CERVUS, maternity for calf 12 was assigned to cow 2 with strict confidence and a large delta score (4.82E+ 00) and maternity for calf 16 was assigned to cow 1 with relaxed confidence (delta score: 3.97E+00). Maternity was not assigned for the other two PO pairs (delta scores: 2.45E+00 and 1.45E+00, respectively). Cow fetus maternity tests. All estimators as signed PO to 97% of cow fetus pairs. O ne pair was assigned as HS ( r xy =0.34, joint related PP=0.89). All others had r xy (9/34) with 95% confidence, 26.5% (9/34) with 80% confid ence, and 26.5% (9/34) not assigned but most likely. The remaining seven pairs were not assigned. Discussion The overall goal of this study was to establish that parentage analysis is possible for the Florida manatee. With robu st informativeness and pow er of 33 polymorphic microsatellite markers, and the ability to distinguish between related and unrelated pairs, this was accomplished. While it wa s difficult to discriminate between half sibl ing and unrelated relationships, the use of the biologically re levant prior (0.74 U) helped considerably. The identification of the latent coancestry and usage of a prior within the east coast population, increased accuracy of relatedness estimators. Estimators Two of the objectives, (2) to identify the estimator(s) best suited to the population, and (3) to determine baseline diversity and inbreeding statistics were accomplished using both relatedness and relationship estimators. Each estimator brings a unique set of characteristics that enhanced the overall outcome. Using multiple
111 estimators allowed for comparisons of values and relationships to establish greater confidence in the results. T he reliability and accuracy of relatedness (r xy ) estimators depends upon the set of markers used and the population structure of the data (Wang 2011) . Taking the most accurate of each type of estimator, maximum likelihood estimator s ML Relate and TrioML , and the moment estimator Wang , were used. Although superior, these estimators tended to overestimate the number of related pa irs, especially half siblings while non related individuals were underestimated. With many biallelic microsatellite markers and a population that is innately related to the level of first cousins ( r xy =0. 111 ), the Florida manatee require s a n estimator tha t will assign a definitive relationship. Large 95% CIs, sliding relatedness values, and incorrect values for identity coefficients resulted in many discrepancies in relationship assignments. These concerns lead us to look to relationship estimators for m ore accurate results It has been stated that relationship (R xy ) estimators tend to overestimate the proportion of unrelated individuals, which can lead to underestimating inbreeding levels, especially with few loci and low allelic diversity (Henkel et al. 2012) . Converting relative likelihoods into posterior probabilities can improve results . Maximum likelihood estimators do not account for inbreeding or population structure, which small populations in need of conservation management tend to have in abund ance (Henkel et al. 2012) . Issues may occur with small populations that have low allelic diversity and highly related individuals, as is the case with the wild Florida manatee population and also
112 captive populations . In this study, related associations were overestimated as compared to presumed relationships based on studbook information (Figure 3 2 ). Adding a relatedness prior helped to correctly estimate the proportion of unrelated individuals but then caused an overestimation of higher order relationships, while unde restimating the proportion of half siblings (Figure 3 2). Overestimating relationships leads to a type I error, as well as the overestimation of inbreeding levels. While inflated , over estimating inbreeding levels could lead to m ore stringent management efforts , as opposed to under estimating inbreeding levels which may allow a false sense of security in the wellbeing of an endangered population . Accurate results are necessary and new statistical methods are being developed for t he Florida manatee . If we consider that family unit 1 and 3 are related, then related and non related relationships were estimated accurately. Romeo and Juliet All estimators inferred that Romeo and Juliet are more related than unrelated, with a half sibling relationship as the most probable. When priors were added, the probability decreased, however, there was still substantial evidence of an underlying relatedness. Empirically, there is evidence to support this finding. Manatees are ta ught by their mothers the location of warm and fresh water sites, as well as abundant feeding grounds (Reep & Bonde 2006) . This leads to high site fidelity between related offspring. Both Romeo and Juliet were captured as adults for display purposes in l ate 1957 to early 1958 , respectively, in Miami Dade County and may be independently related . Ocean Reef and Foster All relatedness and relationship estimators indicated that Foster could be Ocean The exclusion method indicated a high prob ability of relatedness (100%)
113 versus unrelated. While estimators did tend to overestimate relationships, when taking the prior and latent coancestry into account, which usually decreased relatedness if it was incorrectly overestimated, values only decreas ed slightly and r xy values still maintained a PO relationship. Other individuals with an unknown parent (ex. Patch and Stoneman) did not mat ch with any potential parents. Foster was released from captivity in 1998 in Monroe County at the age of five and died in 2012 of internal trauma injuries . According to Rathbun et al . (1995) , the average minimum age of males seen in a mating herd is 3.75 years and males three years of age are spermatogenic (Hernandez et al. 1995) . Ocean Reef , a small adult (269 cm) (Deutsch et al. 2003) was rescued from Monroe County in 2006 due to watercraft injuries . She gave birth two months later. I f we consider age of female sexual maturity to be 3 5 years and a 12 month gestation period (Rathbun et al. 1995) , then Ocean Reef was born no later than 20 0 2. Considering both genetic and analytical evidence, it is plausible that Foster was the father of Ocean Reef. R elatedness between family unit 1 and family unit 3 could explain 42% of the misidentified relationships within this dataset. It is also possible that the manatee numbers in Monroe County are low and there could be siblings (full or half) of Romeo and/or Juliet still in the area. If these potential siblings had offspring in the area, they would be half first cousins ( r xy =0.063) to family unit 1. This plausible hypothesis would also connect family 1 with family 3. While we desire concrete answers, when it comes to wildlife we can only make conjectures, so we run our estimates and infer answers using the data we have. Impacts of Latent Coancestry It has been reported that 90 99% of individuals within populations of wild mammals are unrelated (CsillÃ©ry et al. 2006)
114 wa s estimated to be 74% unrelated corroborating the high relatedn ess ( r xy =0.111) and inbreeding ( F =0.112) values in the population. This latent coancestry ( =0.056) within the population, approximately first cousins , contributed to the higher than expected relatedness values for all pairs of individuals ( Table B 1 ). The high r elatedness values for family u nit 1, supports a half sibling relationship between R omeo and Juliet . Not only did documented inbreeding occur with Romeo mating with both of his daughters (Lorelei and Aurora) to conceive Hugh and Stoneman , respect ively , but as half sibling parents their relatedness coefficient with their offspring will be 0.6 4 rather than 0.5, indicative of an extremely inbred family unit. The expected relatedness value for Lorelei Hugh, which would be both parent offspring and ha lf sibling, would increase to 0.83 which correlates with what was observed ( r xy =0.74 0.94). The elevated relatedness values between all individuals within Family unit 1 caused discrepancies in correct relationship assignments. Because Hugh was inbred, he was assigned as the likely father for Hurricane and Lorelei. Due to the nature of Mendelian inheritance, it is likely that Lorelei passed on genes to Hugh that she inherited from Romeo, that Romeo also passed on to Hugh and Hurricane. This would make Hu gh a better candidate for genetic paternity assignments than Romeo. All estimators tended to overestimate the relatedness values for unrelated pairs . This could be due to the latent relatedness found within all manatees. Based on simulations and observed data, Florida as a whole is between 21 29% related, and the east coast specifically is between 25 29%. While t here is no physical proof of a common ancestor linking family unit 2 to the other two families, r xy values indicate that relationships are likel y. Due to the high site fidelity demonstrated by manatees, the fact
115 that all parents were taken from the same general area, and the high latent coancestry of manatees, an ancestor common to all three families is not an unreasonable conclusion. Knowledge o f captive manatee inbreeding levels can guide expectations for wild populations. Having known pedigrees and cases of inbreeding allowed calculations of expected relatedness and inbreeding values for a variety of plausible scenarios. These values, while s eemingly extreme for a wild population, give the baseline data needed a s we continue to genotype and analyze the greater manatee populat ion. Considering that the manatee population is related on the level of first cousins on average, levels of inbreeding are going to start out greater than a fully outbred population. While the current population of Florida manatees has stayed consistently around 5,000 individuals for the last four years, if population sizes decline to levels observed 20 years ago, there i s a considerable chance that inbreeding levels will increase (FWRI 2010) . If strong trends are detected, more severe management actions could be implemented to improve the outcomes for the population , as was the case with the Mexican wolf and the Florida panther (Hedrick & Fredrickson 2010) . Wild Case Stud ies Both types of estimators were able to accurately assign relationships to the known cow fetus pairs. The few (N=2) that were not correctly assigned could be due to partial genotypes for those individ uals. Relatedness values of one (as was observed in one pair) are indicative of a genotypic match. This could have been due to the sampling protocol during necropsy, possibly contamination or mislabeling. Having examples of undisputed parentage, such as a female with a fetus, can not only test the reliability of sampling protocols, molecular tools, but also of the estimators.
116 Relatedness and relationship estimators agreed on assignments and could not exclude two calves from being each assigned as poten tial offspring to two lactating females. Calf 12 (152 cm) was found deceased on December 22 and Calf 16 (203 cm) was found on December 30 both due to cold stress in the Indian River. It is possible that Cow 1 was weaning her second year calf, (Calf 16) in an attempt to start the reproductive cycle again (Reep & Bonde 2006) . The 2010 winter season was a very cold year, with 282 mortalities due to cold stress (FWRI 2014). This particular telemetry study was being conducted in Brevard County due to the repowering of the Florida Power & Light Company power plant, which is a frequently used warm water site for manatees on the east coast. T hese tools and methods provided plausible answer s to life history questions within a natura l population . C ommonly , dependent calves remain with a lactating adult female for approximately 2.5 years before weaning . The fact that we did not see dependent calves with either of these lactating females, which is contrary to our expectations caused q uestions to arise: maybe the calf died prior to capture, or maybe manatees utilize nursery areas, where calves remain at warm water sites while mothers venture to food sources, as has been seen in sperm whales ( Physeter macrocephalus ), grey seals ( Halichoe rus grypus ), and in the mongoose ( Suricata suricatta ) (Whitehead 1996; Perry et al. 1998; Clutton Brock et al. 2000) . These and other questions can be investigated using parentage assignment methods. Summary Given the set of markers and analytical metho ds used here, there was sufficient information and power to assign parent offspring relationships within the Florida manatee. Distinctions could be made between higher order relationships and unrelated
117 individuals when joint posterior probabilities were c onsidered; however, assignments involving half sibling and unrelated relationships were generally not robust. Compared to the relatedness estimators, the estimators for categorical relationship assignments performed better. These assignments were able to more accurately distinguish between related and non related individuals. In future studies, additional markers and analytical improvements could improve power and accuracy. Overall, low levels of genetic diversity and significant levels of latent coances try were observed for the Florida manatee population. Consequently, the estimators that accounted for inbreeding and latent coancestry provided more accurate results than those assuming outbred relationships. This knowledge will benefit future relatednes s and inbreeding studies involving the endangered Florida manatee population.
118 Figure 3 1. Captive pedigree diagram redrawn from Pause 2007 . Squares represent males; circles represent females. Romeo is represented three times in Family unit 1. Gene tic samples were not available for Aurora or an u nknown wild male (Unk).
119 Table 3 1. Life history information of the captive manatees used in this study. Family unit is associated with Figure 1. U nk wild indicates the parent was unknown and assumed to be wild. 2012 Status locations: Miami Seaquarium (MSQ), Released (Rel), Homosassa Springs Wildlife State Park (HSWSP), Mote Marine Laboratory (MML). Age is given in years unless otherwise specified. + indicates exact age was not known, minimum age given. Inbreeding Estimates and 95% Confidence Intervals assigned from the TrioML estimator within COANCESTRY. Studbook # Name Sex Family Unit Captured Cause Born Bull Cow Sample 2012 Status Age (years) F F 95% CI 3 Romeo M 1 10/10/1957 Pre act U nk wild U nk wild Blood MSQ 56+ 0.00 7 0.0 2 0.12 4 Juliet F 1 1/1/1958 Pre act U nk wild U nk wild Blood MSQ 55+ 0.006 0.03 0.1 2 20 Hurricane M 1 11/23/1983 Romeo 3 Juliet 4 Blood Rel 2008, Dead 2010 27 0.055 0. 10 0.24 5 Lorelei F 1 5/3/1975 Romeo 3 Juliet 4 Blood HSWSP 38 0.00 1 0.0 3 0.1 3 21 Hugh M 1 6/28/1984 Romeo 3 Lorelei 5 Blood MML 28 0.301 0.28 0.45 26 Buffett M 1 5/16/1987 Romeo 3 Juliet 4 Blood MML 25 0.495 0.4 4 0.6 4 29 Aurora F 1 3/13/1989 Romeo 3 Juliet 4 None Rel 1998, MIA 23 239 Foster M 1 3/19/1993 Romeo 3 Juliet 4 Blood Rel 1998, Dead 2012 20 0.064 0.09 0.22 273 DB Stoneman M 1 10/6/1994 Romeo 3 Aurora 29 Blood Rel 2006, MIA 19 0.14 3 0.1 7 0.37 60 Gene M 2 2/16/1977 Watercraft U nk wild U nk wild Blood Rel 2007, Dead 2010 33+ 0.108 0.10 0.2 6 130 Rita F 2 4/12/1982 Entanglement U nk wild U nk wild Tissue Rel 2009, Dead 2009 27+ 0.13 4 0.1 5 0.29 23 Dundee M 2 7/11/1986 Gene 60 Rita 130 Blood Rel 2006, Dead 2008 22 0.000 0.01 0. 10 Ocean Reef F 3 2/22/2006 Watercraft U nk wild U nk wild Blood Rel 2007, MIA 7+ 0.274 0.23 0.41 Patch M 3 4/20/2006 U nk wild Ocean Reef Blood Dead 2006 15 days 0 .000 0.01 0.10 Pumpkin F 3 4/20/2006 U nk wild Ocean Reef Blood Rel 2007, MIA 7 0.135 0.14 0.33
120 Figure 3 2. Comparison of three relationship assignment estimators (Coancestry, ML Relate, and ML Relate Prior) to the relationships based on the North American Regional West Indian Manatee Studbook (Presumed) and relationships if Foster and Ocean Reef are related (Presumed OR F). Usin g the 14 individuals with 91 pairwise comparisons, 64.8% of the pairs are assumed to be unrelated. Coancestry and ML Relate under estimated unrelated relationships and over estimated related relationships. ML Relate Prior, which uses the prior informatio n that 74% of manatees are unrelated, under estimated HS relationships. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Presumed Presumed ORF Coancestry ML-Relate ML-Relate Prior Frequency of assignment Program Relationship Assignments Unrelated Related Other Half Sibling Parent-Offspring /Full Sibling
121 Figure 3 3. Expected r elatedness values compared to Coancestry (TrioML and Wang) and ML Relate estimator s . 0 5 10 15 20 25 0 0.2 0.4 0.6 0.8 1 0 -0.2 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0.1 0.3 0.5 0.7 0.9 Frequency of Observations Expected TrioML Wang ML Relate Relatedness values (rxy) Relatedness Assignments PO FS HS U
122 Table 3 2. Expected relatedness and inbreeding values. T he five c ategories represent common relationship estimators: r xy , and F values. zero pairs of genes identical by descent at a locus, respectively. The upper seven are common outbred relationships. The lower 13 relationships are manatee specific and were cal c ulated using the Wright (1922) coefficient of inbreeding equation and the Lynch and Walsh (1998) equations for relatedness. Relationship r xy F Self 1 1 0 0 0.5 000 Parent Offspring 0.5 0000 0 1 0 Full Siblings 0.5 0000 0.25 0.5 000 0.25 00 0.25 00 Half Siblings 0.25 000 0 0.5 000 0.5 000 0.125 0 First Cousins 0.125 00 0 0.25 00 0.75 00 0.0625 Second Cousins 0.03125 0 0.0625 0.9375 0.0156 Unrelated 0 0 0 1 Parent Offspring/Half Siblings 0.8 29 Parent Offspring/Grandparent 0.7 66 Parent Offspring 0.5 65 0.639 Full Siblings 0.5 65 0.639 Half Sibling/nephew 0.49 0 Half Sibling/1st cousin 0.3 81 Avuncular 0.378 Grandmother/half aunt 0. 347 Half uncle/1st cousin 0.2 62 Great avuncular 0.1 57 Great grandparent 0.14 1 Half great avuncular/ 2nd cousins 0.1 31 Unrelated 0. 129
123 CHAPTER 4 PARENTAGE ANALYSIS OF WILD MANATEES AT A LONG TERM STUDY SITE: CRYSTAL RIVER, FLORIDA Background In the last four decades, many conservation genetic studies have been conducted on manatees ( Garcia Rodriguez 2000; Pause et al. 2007; Tringali et al. 2008b; Hunter et al. 2010; Nourisson et al. 2011; Hunter et al. 2012; Tucker et al. 2012) , although none have been able to accurately reconstruct pedigrees with known manatee lineages (Pause 2007; Bonde 2009; Jacobs 2010; Nourisson 2011) . There have been no studies directly investigating paternity of Fl orida manatees in the wild. Pedigree analyses, specifically paternity assessments, can provide information on individual fitness (i.e. who will survive and who will reproduce) and levels of inbreeding (Jones & Wang 2010a) . The discovery and implementatio n of advanced modern molecular tools, namely microsatellites , into the field of population and conservation biology have enabled parentage assessment to infer answers to issues on the topics of evolution, ecology, and conservation . These include mating sy stems and behavior, dis persal and recruitment patterns, captive breeding programs, reproductive success, impacts of inbreeding, and determining effective popu lation size (Clapham & PalsbÃ¸ll 1997; Frankham et al. 2002; Jones & Ardren 2003; Allendorf et al. 2012) . Parentage assessment in its most basic definition is the assigning of parents to offspring. This can be done through photo identification of potential parent offspring pairs, observations of copulation, and/or genotype data (Clapham & PalsbÃ¸ll 1997 ) . Ideally, each sampling event will collect data using a variety of techniques to increase the chances of correctly assessing parentage.
124 Multilocus genotypes, using microsatellite markers, provide statistically sound method to determine parentage , espe cially when the genotypes of both parents and the offspring are available (Kalinowski et al. 2007; Jones & Wang 2010b) . The exclusion method includes directly comparing the genotypes of a parent to its offspring to ensure that Mendelian inheritance took p lace (Vandeputte et al. 2006) . While, the method is typically used for parentage assessment, success depends upon stability of the genome during meiosis and somatic cell development , genotyping errors (i.e. the misreading or amplification failure of the g enotype), and the number of assessed individuals (Jones et al. 2010) . When complete exclusion can not be attained , c ategorical allocation also uses a focal offspring an d a group of potential parents, however assignments are based on either maximum likeliho od, Bayesian approaches, or parental reconstruction (Smouse & Meagher 1994; Jones et al. 2010 ) . The genotypes of the potential parents are presumed to have different probabilities in their contributions towards genotype (Jones et al. 2010) . Briefly, the maximum likelihood approach is used to determine the probability of the data given the hypothesis and uses many data points to maximize the likelihood of the model (Kalinowski et al. 2006) . The Bayesian approach is used to determine the po sterior probability of th e model given the data and uses prior or conditioning information (prior probabilities) (Neff et al. 2001; Hadfield et al. 2006; Tringali 2006 ) . The third method, p arental reconstruction, relies on known sibling groups, both full a nd half siblings, to reconst ruct possible parenta l genotypes (Jones 2001) . If the genotype of the shared parent is known, those allel
125 genotype s , using a similar technique as seen in the exclusion method. The remaining all eles can be used to determine how many parents of the sibling group there are and can reconstruct the potential genotypes of those parents. In order for this method to be the most effective, there needs to be a large number of progeny and less than six po tential parents (Jones et al. 2010) . If these criteria are not met, this method may not accurately reconstruct the potential genotypes due to the lack of all alleles being represented in the progeny array. Using more than one method is always recommended and ideally will derive concurrent results. All of the discussed methods require that a strong sampling procedure is in place . This includes at least one known or presumed relationship or if that is not practical, then the collection of a large number of individuals from the population (Jones et al. 2010) . If at least one of the parents is known, the power of the test is greatly enhanced. These methods also require many polymorphic markers to supply the genotypes. Jones et al. (2010) highly recommend m icrosatellite markers, which are highly polymorphic and repeatable. The occurrence of genotyping error, mutations , and null alleles can be accounted and controlled for in the parentage analysis technique used. F ield sample and molecular genotype data are used to extrapolate family units, leading to the creation of pedigree diagrams , which in turn give a more complete picture of the population in question. The Florida manatee ( Trichechus manatus latirostris) is an endangered aquatic mammal that lives in the coastal waters of the southeastern United States (Lefebvre & O'Shea 1995) . Most manatees, especially in Florida, have been marred and mutilated by watercraft strikes (O'Shea et al. 1985; Wright et al. 1995; Reep & Bonde 2006) . I f
126 the manatee survives, each strike leaves a unique scar pattern allowing for p hoto based mark resighting studies . The Manatee Individual Photo identification System (MIPS) is an electronic database containing photo and life history information from Flori (Beck & Reid 1995) . This research has been conducted for the last 40 years to estimate movement, reproductive, survival, and population growth rates of the Florida manatee (Kendall et al. 2004; Langtimm et al. 2004 ; Tringali et al. 2008b) . With an extended calf dependency period, the cow teaches her calf migration routes, locations of fresh water sources, seagrass beds and warm water in winter. Through the use of photo identification and detailed field observatio n , potential maternal lineages have been collected over the years, with known females (cows) being seen with likely calves, and those calves (if female) being seen with their presumed calves (Bonde 2009) . Difficulties arise when photo based identification s are not possible, such as in non photographic conditions or with non scarred individuals . Fortunately, the Crystal and Homosassa r ivers have conditions that support long term photo identification and ~ 80% of manatees in this study location have scars wh ich make them identifiable (Rathbun et al. 1995) . However, cow calf pairs can only be assumed, based on nursing and other behaviors , even with substantial cow re sighting rates , given that calves typically have not accumulated scars and consequently canno t be identified through photo ID methods . Genetic parentage studies are therefore a powerful tool to assist and confirm observation based studies. Prior studies of paternity in manatees have been limited to captive samples and a choice between two puta tive fathers. Because there is no visual way to identify the
127 fathe r, v ery little information is know n about the paternity of calves in the wild . P aternity assessments are critical in determining, dominance, reproductive success, inbreeding , and heritabil ity in a population (Sardell et al. 2010) . Knowledge of paternity could identify genes unique to those males and may lead to information regarding whether or not females have input as to which of the males are chosen for breeding . Inferring paternity wil l complete the pedigree and allow the relatedness of individuals to be more thoroughly compared. The answers to questions about reproductive skew, dominance, and relatedness will enable better estimates of the fitness of the population. Reproduction wit hin the Florida manatee takes p lace in mating herds, involving one female with multiple males (O'Shea & Hartley 1995; Reid et al. 1995; Boyd et al . 1999 ; Reynolds & Rommel 1999; Reep & Bonde 2006) . The Florida manatee mating system has been classified as a scramble competition polygamous sys tem, where the males mate with the receptive female and then leave to find another (Reynolds & Rommel 1999; Reep & Bonde 2006) . This is evident by the lack of paternal care. Manatees could also be classified as a prom iscuous species, because females may also mate with many males during a mating herd. A similar mating system is observed in the North Atlantic right whale where it has been suggested that females encourage competition between males by ensuring that copula tion is difficult and that only the most fit males breed (Frasier et al. 2007) . There are many benefits for a polyandrous or promiscuous mating system, especially in an endangered marine mammal. A multitude of hypotheses surround the idea that females sh ould mate with multiple males to give their offspring the best chance of survival as well as make it worth the investments and costs of pregnancy (Zeh & Zeh
128 1996; Stockley 2003) multiple father s reduces sibling competition and/or reduces the risk of the offspring coping with an uncertain environment through increased diversity. that multi ple matings are to offset matings with males with low sperm quality or fitness. with closely related individuals, mating promiscuously may reduce the impact of inbreedi have more superior sperm or ejaculates, due to either sperm competition or the (Zeh & Zeh 1996) . This study employs 33 mic rosatellite markers, three methods of parentage analysis, and 40 years of photo identification data to evaluate relationships of wild manatee maternal family units at a long term study site, Crystal River, Florida. The two main objectives of this study ar e to (1) a ssess and verify cow calf associations determined vi a photo ID using genetic methods, and (2) i nvestigate the possibility of inferring paternity from maternal families using a pool of males and reconstructing paternal genotypes . Materials and Met hods Sample H istory, C ollection, and DNA E xtraction A total of 267 west coast free swimming or re covered necropsied wild manatee specimen s was used in this study. Tissue was collected from the tail and stored i n DMSO or 9 5% EtOH (Bonde 2009) . F rom 1990 2 011 , 75 photo identified adult females, were sampled along with their offspring ( N=161, 85 female, 74 male, 2 unknown) in Crystal River, Florida (N=236). An additional 34 adult male carcasses from
129 the west coast of Florida (2003, 2004, and 2010 ) were included as potential fathers. Genomic DNA was isolated following instructions (Valencia, CA, USA). The quantity and quality of the DNA were measured and DNA concentrations were standardized. Microsatellite DNA A nalysis A panel of 33 polymorphic microsatellite markers (GarcÃa RodrÃguez et al. 2000; Pause et al. 2007; Tringali et al. 2008b) divided into 12 multiplex reactions was used to amplify the 267 individuals. Published primers TmaA03, TmaE02, and TM AFWC02 were not used due to inconsistent allele scores . All PCR amplifications were performed on either a PTC 100, PTC 200 thermal cycler (MJ Research, Waltham, MA, USA), or a UnoII thermal cycler (Biometra, Goettingen, Germany). Isolated DNA was polymer ase chain reaction (PCR) amplified in 13.4uL reaction volume containing: 0.7uL of 10ng DNA, 8uL DNA free water, 2uL 0.8mM dNTPs, 1.25uL 10x Sigma PCR Buffer,(10nM Tris HCl, pH 8.3, 50mM KCl, 0.001% gelatin) (TmaE08 and TmaE11 used 1.625uL), 1.25uL MgCl 2 ( Tm a FWC14 and Tma FWC15 required 2uL MgCl 2 ), 0.063 units Sigma Jump Start Taq Polymerase, and primer amounts varied from 0.037 0.220uM at a concentration of 100ng/uL ( Table 2 1 ). The PCR reaction profile was: 95ÂºC for 5 min, 34x (95ÂºC for 30 s, 54 60ÂºC for 1 min, 72ÂºC for 1 min) final extension 72ÂºC for 10 min with a 4ÂºC hold (Table 2 1 for annealing temperature and primer concentrations). Fragment assays used Genescan Rox500 size standard (Applied Biosystems, Inc., Foster City, CA, USA). Fragment analysis was performed on an Applied Biosystems 3130XL automated genetic analyzer and genotyped using GENEMARKER software version 2.2.0 (Applied Biosystems, Inc., Foster City, CA, USA). Samples that did not amplify for all 33 markers were rerun in additional PCRs to complete the genotype. Two
130 negative controls, two positive controls , and eight individuals were run twice for each PCR to test for error rates and contamination. Statistical Analysis All 33 microsatellite markers were assessed for the presences of n ull alleles using M ICROCHECKER v.2.2.3 (Van Oosterhout et al. 2004) and deviations from Hardy Weinberg equilibrium (HWE) and linkage disequilibrium (LD) via G ENEPOP 4.2 (Raymond & Rousset 1995) . was applied for multiple comparisons. Mean observed (H o ) and expected (H e ) heterozygosity, polymorphic information content (PIC), and null allele frequencies were estimated using C ERVUS v.3.0 (Kalinowski et al. 2007) . K ININFOR v1.0 (Wang 2006) was used to a ssess marker informativeness, which is the overall power of the markers to distinguish between relationships and infer relatedness for all six combinations of parent offspring (PO), full sibling (FS), half sibling (HS), and unrelated (U) hypotheses. To te st hypotheses, a llele frequencies for the manatee population of the we st coast of Florida and a significance level of 0.05 were used and equal priors of sharing zero, one , or two alleles identical by descent were assumed. The likelihood of two individuals in a population sharing the same genotype, the probability of identity P (ID) , and a more conservative estimate, the probability of sibship P (ID)SIB , were estimated using GENECAP (Wilberg & Dreher 2004) . Mean relatedness values, inbreeding coefficients an d simulations for the west coast of Florida were determined via C OANCESTRY (Wang 2011) . Simulation parameters included 79 , 800 dyads of 200 pairs of PO, FS, HS, and U relationships , 200 bootstrapping samples, and 100 reference individuals, with 28 loci.
131 C ategorical relationship estimator. Assigning parentage was performed using CERVUS 3.0 (Kalinowski et al. 2007) . This program computes a multilocus likelihood ratio (LOD score and Delta) for each parent offspring pair and can detect null alleles , while al so accounting for genotyping error and mutations via error rate per locus. Simulation parameters included 100,000 offspring, 300 candidate mothers with a 98% sampling rate, and 500 candidate fathers with a 50% sampling rate. To accommodate for mistyping, null alleles, and other genotyping error, a 97.8% loci typed and 0.196% error rate were used in calculations. Confidence levels were strict (95%) and relaxed (80%) delta scores. All adult males (N=34) and adult females (N=73) were considered fathers and mothers respectively, while only those individuals (N=144) that were known to be offspring were considered offspring. In a subsequent analysis, all females, including both known mothers and calves (N=148) , we re considered candidate mothers, while all mal es were considered candidate fathers (N=101). Parent pairs were assessed in two separate analyses, one i dentifying the known mothers and the other allowing CERVUS to assign maternity to all calves. Another categorical relationship estimator, ML RELATE, was used to estimate relationships ( R ) among pairs using the methods described in Kalinowski et al. (2006) . Specific relationships were classified as U, HS, FS, and PO. Null allele s were account ed for in TmaA02, TmaE04, TmaE07, Tma H13 , and TmaSC5. To im prove assignments, relative likelihoods were converted into posterior probabilities using a prior that incorporated overall mean relatedness, as simulated for the west coast manatee population (See Chapter 2) . Joint related probabilities were calculated b y taking the sum of all related relationship probabilities (PO, FS, and HS).
132 Relatedness. Overall, inbreeding and relatedness estimates of both the West Coast Florida manatee population and the Crystal River study population were evaluated using the softw are program COANCESTRY 126.96.36.199 (Wang 2011) , which offers seven different relatedness estimators: two likelihood and five moment estimators. The likelihood estimator, TrioML , assign s pairs of individuals a relatedness coefficient (i.e. 0=U, 0.25=HS, 0.5=FS , or 0.5=PO). Analytical parameters included 100 reference dyads and 200 bootstrap replicates; missing loci, typing error, and inbreeding were taken into account. Individual i nbreeding ( F x ) and relatedness ( r xy ) coefficients were computed for all pairwis e relationships and mean values are reported. The common inbreeding values for the mating between FS is 0.25, the mating between HS is 0.125, the mating between first cousins is 0.0625 and the mating between second cousins is 0.0156. Reconstruction. Pare ntage was assigned or inferred and sibship was estimated using a maximum likelihood reconstruction algorithm through the software program COLONY 188.8.131.52 (Jones & Wang 2010) . Offspring we re clustered into full or half sibling groups a nd candidate females a nd males we re assigned as parents to these clusters. If there wa s not a candidate parent assigned or available, a reconstruction of the potential parent genotypes were created. S econd order relationships we re assumed to be absent, yet null alleles, mutat ions , genotyping errors , and inbreeding were taken in to account . Two replicates using different random seed numbers to initiate simulations and analytical parameters include d medium precision and short run lengths with the FPLS analysis method. Maternal families (N=21) with three or more genotyped offspring (N=74) were analyzed. Paternity was estimated from 101 potential males with
133 a 0.30 probability of having sampled the fa ther. Males were excluded from being potential fathers if they were not sexually mature at time of conception, estimated as a year before first year calves and two years prior to second year calves. Calves 200cm or under were classifi ed as a first year calf, and over 200cm as a second year calf. Three years were added to first year calves and two years to the second year calves for the min imum age of sexual maturity. To determine accurate year of birth, for 1 st and 2 nd year calves if first year calves were sampled between January and June they were considered born in the previous year and July Dec ember born in that year. The same month ranges applied to second year calves with the addition of two years . Results Genetic Marker A sses sment The 33 microsatellite markers used in this study had a mean of 2.88 alleles per locus (range 2 to 9), mean Ho=0.392, and He=0.396. H omozygote excess indicating the potential presence of null alleles were found in five markers : TMAA02, TMAE04, TMAE07 , TMAH13, and TMASC05. Following Bonferroni correction, two HWE deviations remained: TMAE11 and TMAK01 (p<0.05). Li nkage disequilibrium (LD) was found in t welve pairs of markers, eight were corrected for with null allele corrections in all parentage prog rams used. Due to LD and HWE deviations, f ive markers were dropped from further analyses : TMAK01, TMAE0 1, TMAE08, TMAE11, and TMAFWC12 . The mean PIC value was 0.338 (range of 0.054 0.6) (Table 2 1 ). The power to assign categorical relationships ranged from PW R =0.36 to 0.97 (Table 2 3 ). Robust assignments could be expected between PO and U whereas assignment powers for
134 other relationships were shown to be less robust. The overall mean relatedness value was 0.125, which is equivalent to first cousins. The overall mean inbreeding coefficient was 0.123 equivalent to the mating of first cousins/half siblings. Simulation results indicated that 77% of the population was unrelated. Of the total 236 specimen s from Crystal River, 145 were also used in another study using less microsatellite markers (Bonde 2009). There were 19 specimens that were not able to be used due to DNA degradation or PCR amplification issues. The Crystal River sample s had a mean relatedness value of 0.168 equal to between FS and HS an d a mean inbreeding estimate of 0.116 equivalent to between first cousin and HS mating . There were nine pairs that had identical genotypes (Table 4 1) . Of these, seven were verified by photo ID data. Multiple pairs (N=3) were known to have two genetic samples before genetic analysis, and four pairs were discovered to have been sampled twice : one as a first year calf and again as a second year cal f. Two pairs could not be verified using photo ID data. Maternity For the maternity stu dy 311 presumed related individuals based on MIPS with 149 P O pairs, 136 sibling pairs, 19 a vuncular and eight matches were used in this study set. A categorical relationship estimator, ML RELATE , assigned 81% as related. Out of the 206 correct assignmen ts: 1 4 9 were PO , 21 FS, and 55 HS. There were 42 pairs correctly assigned as being related but based on presumed photo ID relationships had been assigned an incorrect genetic relationship: eight presumed PO assigned as HS, one presumed HS assigned as PO, three presumed HS assigned as FS, and 30 presumed siblings assigned as PO. Of the presumed related pairs 57 were incorrectly
135 assigned as U via genetic assignments : 36 presumed siblings, nine presumed avuncular, and 12 presumed PO. Of the presumably 23 , 1 17 U relationships, 33% were assigned as related, with 75% as HS. Many pairs had joint related probabilities greater than 0.60 (N=6,642) , although when using 0.77 as a prior U that value decreased to 1 , 417 pairs (Figure 4 1 ). Another categorical relations hip estimator, CERVUS, assigned multilocus likelihood relationship estimates to 148 known cow calf pairs. T he top two most likely females were chosen as potential mothers for each calf, and then one was assigned maternity with either strict (SC) or relaxed confidence (RC) or did not assign but indicated as most likely candidate mother (NAL). Maternity was correctly assigned to 57% of the photo ID pairs, 12 SC, 31 RC, and 41 NAL. Another 15% of pairs had the correct female as one of the top two choices but the other female was assigned maternity: 1 SC, 2 RC, and 19 NAL. Forty two pairs did not have the correct female as one of the top two potential mothers. Accuracy was verified by the comparison of th e two categorical relationship estimator res ults. Overall, t hirteen assigned cow calf pairs were not possible based on the age of sexual maturity of the mother. Of these, 62% had assignments of PO with d FS with joint related probabilities of 1.00, and 1 was assigned U. CERVUS assigne d s ix pairs of replicates , either the same individual sampled twice or twins , as PO pairs with each individual being both the parent and the offspring . As compared to ML R ELATE which assigned all as FS relationships. B oth categorical estimators assigned t wo pairs of presumed siblings , based on photo identification of the mother, as PO. O ne of these PO assignments is plausible based on
136 year of birth, while the other is no t . Of the remaining pairs, b oth categorical estimators assigned 43 pairs as PO which agreed with photo ID data . The remaining 1/3 of cow calf assignments did not match with their presumed mothers. ML RELATE agreeing with 81% of these PO pairings, assigne d three as FS and one as HS relationships. Paternity Using a pool of 16 sexually mature males from the west coast of Florida begat 2,482 possible father calf genetic pairs. Of these , 84 (3 % ) were assigned PO/FS relationships. The elimination of 16 pairs based on the feasibility of the year the male died and the age of the calf, left 68 possible paternity assignments. Between 1 and 9 calves were indicated as being likely sired by 15 males. Out of 51 unique calves, 36 were assi gned to only one potential father, 13 calves assigned to two potential fathers, and two claves could have been sired by three potential fathers. All candidate fathers assigned to more than four offspring, were found to have likely sired more than one offspring in a single year. When using male calves as possible fathers of all offspring, there were 10,440 possible father calf genetic pairs. Only 8% were assigned PO/FS relationships. Of these, 194 were possible relationships based on sexual maturity of the males and the year the offspring was born. When examining only those males assigned with confidence, 12 males were assigned to 14 bull calf p airs; two males were assigned to two different calves. Strict confidence was assigned to three pairs. All three were male calves and one mated with its presumed mother. Relaxed confidence was assigned to 11 pairs. A few male calf (N=7) assignments were not possible based on offspring birth year as well as five pairs or trios that are questionable based on age of sexual maturity or year of death. Confidence was assigned to eight males within an offspring mother father trio: two with
137 SC and six with RC. Of those, two of the assigned mothers were not the presumed mother and two males were assigned to multiple calves. CERVUS assigned 11 bull calf pairs with which ML RELATE agreed with 90%, one was assigned as a HS relationship . Three of these bull calf pairs had the potential fathers mating with their mother, yet only two had F values that resembled inbreeding (see Â§ in Table 4 2). An other son mother breeding pair had F values indicative of inbreeding , yet it was only assigned by one estimator (see Â§ in Table 4 2). The nine offspring mother father trios that CERVUS assigned had two of the females not match their presumed calves, yet all were assigned PO/FS relationships by ML RELATE. One female was assigned a HS relationship to her presumed calf, while all other females were assigned PO/FS relationships. Two males were assigned HS relationships by ML RELATE, yet had joint related probabilities of 0.79 and 0.82 but these values dropped significantly (0.27, 0.32) when a prior which incorporated overall m ean relatedness was used. There was a high degree of convergence between the two COLONY runs, inconsistent findings were excluded. There were 16 male genotypes reconstructed as possible fathers for 72 offspring between 1992 and 2011. Potential fathers we re assigned to between one and nine calves with ten of these males potentially siring more than one calf in a given year ( Table 4 2 and Figure 4 2 ). S even maternal families had evidence of full siblings indicating the same potential male sired two offspri ng with the same female in different years (Table 4 1 and 4 2 ). One pair (CR251c2 and c3 ) was found to be the same individual sampled in separate years. COLONY assigned all as FS dyads, but only four as pairwise FS dyads, a stricter test. Five pairs wer e also
138 assigned FS/PO relationships in ML RELATE with joint related probabilities of 0.98 1.00. While t wo HS relationships with joint related probabilities of 0.67 and 0.85 were assigned . Three candidate males were assigned as fathers to offspring, yet o nly one had concurrent relationship assignments in CERVUS and ML RELATE. Within the 21 maternal family groups used in the analy ses, four sets of siblings had been seen with their mother within 21 23 months of each other. Photo ID data confirmed sightings in each winter and genetic data confirmed parent offspring relationships (Table 4 1). Discussion Maternity The first main objective to a ssess and verify cow calf associations determined vi a photo ID using genetic methods was met. Within the west coast of Florida, there was sufficient power and information in the markers used, yet there was a substantial amount of latent coancestry, or unexpected and underlying average level of relatedness. The latent coancestry was found to be at the level of first cou sins . With this level of innate relatedness, difficulty in assigning relationships would be expected. Considering the method of reproduction, a scramble competition polygamous system, where males can mate multiple times every year, there would likely be many half siblings in the population. When these half siblings become sexually mature, even if no direct inbreeding occurs, there would still be a substantial number of half first cousins in the population. Being that categorical relationship estimators do not assign second order relationships, this elevated level of relatedness combined with the latent pair as half siblings. Using genotype information, we can now identify relationships that would not have be en considered using photo ID methods alone . This data can assist in the development of new studies
139 to investigate deeper pedigrees and second order family units. Questions concerning the interact ions of these extended relationships and how those relation ships effect genetic dispersal can be addressed . The collaborative effort of using both genetic and photo ID data allows for the verification of dually sampled individuals and the possible identification of twins . The nine genetic matches corresponded to photo ID recaptures and the MIPS database verified or was corrected due to these findings (Langtimm et al. 1998) . Of the two genetic pairs that could not be verified using photo ID, there is a possibility that one could be the same individual due to size and independent dates matching what is known to be true for manatees (Beck 2014 pers. comm.). The two samples were both females sampled nine years apart , while the second sample was from a n older manatee larger than the first. The second pair (CR041c1 an d CR547c1) gives evidence to the fact that manatees are highly related and have low genetic diversity, especially in Crystal River ( Table 4 1). This could not be a case of recapture or twins due to the dates and sizes of the manatees at the time of sampli ng. There were 13 years between samples and the smaller calf was sampled second. It is evident that although there is enough power and information in the molecular markers used, the high latent coancestry found within all regions of Florida has led to t he lack of consistent assignments between presumed related and unrelated individuals (Chapter 2) (Pause 2007; Bonde 2009; Nourisson 2011). While, the genetic methods used here were able to give above average results of maternity assignment, the ability to include photo identification and behavioral observations allows for a more thorough and accurate estimate of relationships within the Florida manatee population. For
140 example, female calves were often assigned as mothers when not excluded, likely due to t he high relatedness of the Crystal River population. Implausible relationships, such as these, were eliminated based on date of birth and ultimately improved the accuracy of the findings. P hoto ID and genetic data can also be combined to identify sibling relationships using matrilineal groups . M aternal families that had three or more offspring (N= 21 ) were used to identify the possibility of male dominance in the system, the number of potential fathers per family , as well as to identify twins, matching individuals , and sibling relationships. Of the potential twins identified in a previous study, one was confirmed as twins both genetically and through photo ID, although it is unclear if they are full siblings or half siblings (B onde 2009). The categorical method indicated the likelihood of full siblings ( PP= 0.48) as slightly higher than that of half siblings ( PP= 0.43), yet the maximum likelihood reconstruction method indicated a half sibling relationship ( Probability= 1.00) . Bay esian probability testing identified a full sibling relationship to be twice as likely as half siblings, while two paternity assignment programs identified different fathers for each twin. Due to the inconsistent findings, more work is recommended . Altho ugh if the twins are half siblings, this would be a case of heteropaternity superfecundation in manatees. Heteropaternity superfecundation occurs when fraternal twins have different fathers and has been well documented in humans, primates, and a multitude of other mammal species (Terasaki et al. 1979; Verma et al. 1992; Bercovitch et al. 2002; Bonin et al. 2012; Liu et al. 2013) . This finding would be consistent with the promiscuous or polygamous mating strategy of manatees. Sperm can stay viable for 3 5 days in the female body after ejaculation, if the female is in the preovulatory period (Pallone &
141 Bergus 2009) . If many males successfully ejaculate during copulation in mating herds, then there would be a greater chance of twins having different fathers . Twinning in manatees is rare, approximately 1 2% of all births ( Rathbun et al. 1995; Reynolds & Rommel 1999; Reep & Bonde 2006) . There have been two recorded insistences of a female carcass carrying twins and ten cases of newborn twin calf mortalities (FWC P athobiology L aboratory pers. c omm. ). The use of these in situ samples would provide an excellent experiment, as twins would be confirmed with a known mother. Paternity The second main objective to i nvestigate the possibility of inferring paternity from a pool of males , as well as reconstructing paternal genotypes from maternal families , was also met. This is the first reported paternity study of the Florida manatee. While limited in its scope, due to a lack of sampled males, there was evidence of paternity with 17 males potentially siring 48 offspring and the reconstruction of 16 male genotypes from 72 calves. There were also three cases of potential inbreeding between mother and son. The inbreeding value of the calf was indicative of parent off spring mating ( F ~ 0 .25+). The male son was at least eight years old at the time of potential mating based on MIPS data. With what is known about manatee site fidelity and the lack of knowledge on mating herd structure and the interrelationships among its members, this is a plausible scenario (Reep & Bonde 2006). This gives more credence to the fact that the Florida manatee has low genetic diversity. The finding that 1/3 of 21 families with three or more offspring had full siblings, lends itself to two trains of thought. E ither there is a shortage of unique genotypes which sired these offspring due to the lack of diversity within the population, or there is a signature of male dominance in the population. To understand more fully the breeding
142 success of males, as has been noted in numerous marine mammal publications, a n exhaustive sampling of the males in the population should be undertaken (Amos et al. 2001; Gemmell et al. 2001; Cerchio et al. 2005; Frasier et al. 2007) . While a complete sampling can be difficult in a mar ine mammal with a relatively large population size, as opposed to the North Atlantic right whale (Frasier et al. 2007) , it is possible to make a more focused effort on adult males especially along the west coast of Florida. Although observation of copulat ion in the wild is rare, it provides information that the female is truly in estrus and the duration of the estrus cycle (Reep & Bonde 2006) . Therefore, focused sampling of mating herds and the resultant calf could supply answers to some dominance and est rus questions. Sampling the genetics, fecal, and behavior through photo ID of as many males in a mating herd as possible, the focal female, and her calf the following year would provide a semi closed population in which to determine paternity. It would a llow investigations into the genetic and hormonal differences of individuals in a mating herd. Fecal collections would enable immunoassay studies to investigate hormone fluctuations of both males and females during a mating herd (Larkin 2000) . If blood a nd/or tissue samples can be obtained from mating herds, immunological assessments could be explored. Furthermore, research investigating mating herds can identify those males that participate at a higher frequency and any correlations with the successful bull. Examining male reproductive success can help to determine whether the observed number of successful breeders is greater or less than expected from random mating. These studies can estimate correlations between successful males and their age and loc ations. They can also inform estimates of effective population size, identifying how many males are contributing to the population
143 and from where. These answers could aid in determining if the Florida manatee should be considered one population or multip le subpopulations. If only a few males are contributing to the population it may be necessary to establish stringent management practices, as was implemented with the Florida panther ( Johnson et al. 2010) . Manatees live in an environment that is expected to change significantly in the near future. With the impending retirement of many Florida power plants, the numerous warm water environments in which manatees have become accustomed to are going to diminish. In the last four years, 33% of all manatee dea ths have been due to environmental factors (cold stress and red tide) (FWRI 2014) . A s the environment continues to change, the Florida manatee will have to adapt. These adaptations and evolutionary genetic questions can be quantified and addressed using pedigree data (Pemberton 2008) . Quantifying the relatedness and inbreeding coefficients of individuals in the population will allow inquiries into the variation of traits and the ways in which the population responds to selection (Pemberton 2008). Knowl edge of these values, especially within the study area of interest, can also assist in drawing conclusions surrounding mating system hypotheses, as was briefly discussed earlier. It is now known that inbreeding levels are high and the population is highly related (Chapter 2). This would lead one to believe that promiscuous mating may be due to the inbreeding avoidance hypothesis. In situations where inbreeding may not be avoided, the more males the female is able to mate with can reduce the impact of inb reeding. If the female mates with many different males during her time in estrous, the chances that she mates with only related individuals is reduced. Also, given such a high inbreeding level ( F =0.123), there are likely detrimental genetic consequences in
144 effect, such as lower fecundity due to abnormal sperm (Asa et al. 2007; Hedrick & Fredrickson 2010; Wilson et al. 2011) . Fertility issues can be associated with low diversity through inbreeding depression, as well as anthropogenic and environmental str essors (Asa et al. 2007; Wilson et al. 2011) . The genetic benefit and genetic incompatibility avoidance hypotheses could also be important factors to consider. Females could be mating with many males to give their offspring the best chance of a healthy l ife. If males with reduced quality sperm are competing with males with normal quality sperm and sperm competition is the main mating strategy, it would be reasonable to believe that the healthier, less inbred male would be more successful (Reynolds et al. 2004) . A healthier population better equipped to withstand these challenges would then ensue. In the future, correlations between directional selection and genetic variation can be investigated (Wilson et al. 2006) . This could be assessed through more in depth paternity studies. There may be characteristics or genes that successful males have that other males do not. There may be genes that are favored to establish immunity and/or resilience to environmental challenges or potential diseases. Other f uture studies could include estimating r elatedness and inbreeding coefficients and comparing them to longevity and fecundity in the population (Pemberton 2008) . Genetic samples could be collected from known individuals, of both sexes, to establish the rel atedness and inbreeding values of the population of interest (ex. Crystal River). Then, utilizing the vast amount of life history information from the MIPS database, factors such as number of offspring, number of re sightings, and longevity can be compare d to these genetic parameters for potential correlations. Moreover, t he effect of environmental
145 changes on inbreeding depression can be inferred through substantial shifts of the inbreeding coefficient of the population between years with large environmen tal differences (Keller & Waller 2002) . Because there are long term datasets pertaining to the Florida manatee, both life history and genetic associations between population or subpopulation, inbreeding values and environmental or anthropogenic changes can be investigated. There have been some years with large m ortality events due to extreme cold winters (2009 2010, N=480 deaths, FWRI 2014) or red tide events that may have caused the population numbers to fluctuate (2014 unusual mortality event on the east coast of Florida). The genetic consequences of these flu ctuations could be investigated. Summary Overall it is concluded that m anatees of the west coast, and more specifically in Crystal River, Florida, have a substantial amount of latent coancestry and are highly related making it difficult to assign accurate relationships . Yet, with the combined efforts of genetic and photo ID data, unexpected relationships were inferred, such as the three cases of a male offspring potentially mating with its mother, and the nine cases of matching genotypes. Focused sampling efforts could help to shed light on these relationships in the future. Additionally, relatively few males were assigned paternity to a large number of offspring. This further supports the hypothesis that the population is highly related with low diversi ty and a limited gene pool.
146 Table 4 1. Presumed relationships with unexpected outcomes of calves 235cm in length. Full Sibling (FS), Half Sibling (HS), Parent Offspring (PO), Unrelated (U). Calf year: a 1 st (<200cm) or 2 nd (>200cm) year calf . Match cow: was the calf genetically assigned to the presumed mother. The reconstruct analysis method assigned sibling relationships to dyads (Sib Dyad) and pairwise dyads (Pair Sib Dyad), which is a stricter test. ML output of categorical rela tionship with and without a prior (R and Prior R) and relative likelihood (LnL(R)). Joint related posterior probability of relationship s with and without a prior (Joint related and Joint related prior) , Presumed relationship before genetic analyses (Presum ption). Calf 1 Calf year Calf Sex Year Match Cow? Calf 2 Calf year Calf Sex Year Match Cow? Sib Dyad Pai r Sib Dyad R LnL(R) Joint related Prior R Joint related prior Presumption Inferences CR032c4a 1st M Dec. 1999 yes CR032c1 2nd F Feb. 1992 yes FS HS 49.64 0.85 U 0.36 Siblings Full Siblings CR104c5 1st F Nov. 2000 yes CR104c2 2nd M Jan. 1997 no FS HS 58.04 0.67 U 0.17 Siblings Full Siblings CR123c3 2nd F Feb. 1999 yes CR123c1 2nd M Feb. 1992 yes FS PO 49.95 0.99 PO 0.91 Siblings Full Siblings CR130c3 2nd M Nov. 2010 yes CR130c1 2nd M Jan. 1998 yes FS FS PO 51.16 1.00 PO 0.99 Siblings Full Siblings CR205c3 2nd F Jan. 2001 yes CR205c1b 1st F Dec. 1996 yes FS FS FS 45.49 0.98 FS 0.86 Siblings Full Siblings CR251c3 2nd F Oct. 2005 yes CR251c2 2nd F Nov. 2006 yes FS FS FS 42.76 1.00 FS 1.00 Siblings Full Siblings CR363c2 1st F Feb. 2000 yes CR363c1 1st F Jan. 1997 yes FS FS PO 56.47 1.00 PO 0.99 Siblings Full Siblings CR032c2 2nd F Dec. 1993 yes CR032c1 2nd F Feb. 1992 yes HS HS PO 46.53 1.00 PO 0.97 Sibs, maybe twins Siblings CR123c4 2nd M Nov. 2000 yes CR123c3 2nd F Feb. 1999 yes HS HS PO 47.27 0.93 U 0.55 Sibs, maybe twins Siblings CR363c3 2nd M Nov. 2001 yes CR363c2 1st F Feb. 2000 yes HS HS HS 52.1 0.86 U 0.39 Sibs, maybe twins Siblings CR541c2 1st F Dec. 2007 yes CR541c1 2nd M Jan. 2006 yes HS 56.83 0.63 U 0.15 Sibs, maybe twins Siblings CR205c1b 1st F Dec. 1996 yes CR205c1a 1st M Dec. 1996 yes HS HS FS 45.52 0.91 U 0.51 Sibs, maybe twins Twins CR251c3 2nd F Oct. 2005 yes CR251c2 1st F Nov. 2006 yes FS FS FS 42.76 1.00 FS 1.00 Siblings Same individual CR111c4 1st F Dec. 2010 yes CR111c3 1st U Dec. 2009 yes FS 32.62 1.00 FS 1.00 Siblings Same individual CR263c2 2nd F Dec. 2002 yes CR263c1 1st F Nov. 2001 yes FS 34.88 1.00 FS 1.00 Siblings Same individual CR354c2 2nd M Jan. 2002 yes CR354c1 1st M Jan. 2001 yes FS 35.35 1.00 FS 1.00 Siblings Same individual CR431c3 2nd F Jan. 2007 yes CR431c2 2nd F Jan. 2006 yes FS 40.91 1.00 FS 1.00 Siblings Same individual CR046c1 1st F Feb. 1995 yes CR506 Adult F Nov. 2006 yes FS 30.07 1.00 FS 1.00 Same Same individual CR321c1 2nd F Jan. 1995 yes CR458 Adult F Jan. 2007 yes FS 40.49 1.00 FS 1.00 Same Same individual CR041c1 Adult F Jan. 1993 yes CR547c1 1st U Nov. 2006 yes FS 34.81 1.00 FS 1.00 Unrelated Matching genotype CR125c2 2nd F Nov. 1997 yes CR523 Adult F Nov. 2006 yes FS 33 1.00 FS 1.00 Unrelated Matching genotype
147 Figure 4 1 . Probability of relatedness by pairs. Blue line indicates joint related probabilities, orange line indicates joint related probabilities using th e prior that 77% of the Crystal River population is unrelated. 0 2000 4000 6000 8000 10000 12000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Pairs Probability Probability of Relatedness All Data Joint Related Joint Related Prior
148 Table 4 2. The 21 maternal families with reconstructed and inferred paternity using three parentage assignment methods. Inbreeding coefficient ( F ), COLONY assigned numbers to the 16 paternal genotypes reconstructed (Reconstructed Father ID) Female (Cow) Male (Bull). Calf year: a 1 st (<200cm) or 2 nd (>200cm) year calf. Maternity confirmed: was the calf genetically assigned to the presumed mother. Offspring Sex Year of Birth Calf F Calf year Maternity Confirmed? Cow F Reconstructed Father ID Inferred paternity Bull F CR027 0.015 CR027c1 F 1997 0.304 2nd Yes^ *2 CR027c2 F 1999 0.0 10 1st Yes^ *10 CR027c3 M 2002 0.014 2nd Yes^ *14 CR027c4 F 2011 0.00 6 1st Yes^ *12 CR032 0.09 2 CR032c1 F 1992 0.025 2nd Yes^+ *3 CR032c2 F 1993 0.058 2nd Yes^ *11 CR032c4a M 1999 0.018 1st Yes^ *3 CR032c6 F 2002 0.01 1 1st Yes^*+ *10 MSW03141+# 0.069 CR046 0.001 CR046c1 F 1995 0.229 1st Yes^ *4 CR046c2 F 1997 0 .000 2nd Yes^ *1 CR046c3 F 2006 0.311 1st Yes^ *5 CR104 0.107 CR104c2 M 1997 0.09 1 2nd No *5 CR104c4 M 2000 0.338 2nd No *12 CR104c5 F 2000 0.005 1st Yes^ *5 CR104c6 M 2002 0.156 1st No# *8 MNW0334+# 0.02 1 CR111 0.301 CR111c1 M 1992 0.019 2nd Yes^+ *6 CR111c2 M 1997 0.01 2 2nd Yes^ *13 MNW0338*^ 0.074 CR111c3 F 2009 0.001 1st Yes^ *14
149 Table 4 2 . Continued Offspring Sex Year of Birth Calf F Calf year at sampling Maternity Confirmed? Cow F Reconstructed Father ID from COLONY Inferred paternity Bull F CR123 0.39 9 CR123c1 M 1992 0.404 2nd Yes^ *7 CR123c2 F 1996 0.01 5 2nd Yes^+ *2 MNW0334+^ 0.02 1 CR123c3 F 1999 0.0 10 1st Yes^ *7 CR123c1^ 0.404 CR123c4 M 2000 0.04 7 1st Yes^ *5 CR123c5 F 2011 0.214Â§ 1st Yes^+ *12 CR123c1^ 0.404 CR125 0.616 CR125c1 M 1995 0.15 1 2nd Yes^ *8 CR125c2 F 1997 0.267 2nd Yes^ *1 CR125c3 M 2000 0.01 6 1st No *15 CR130 0.29 3 CR130c1 M 1998 0.112 2nd Yes^*+ *3 CR130c2b F 2003 0 .000 2nd Yes^*+ *2 CR130c1^ 0.112 CR130c3 M 2010 0.008 2nd Yes^*+ *3 CR130c1+^ 0.112 CR164 0.012 CR164c1 F 1997 0.046 1st No *6 CR164c2 M 2001 0.181 1st Yes^ *7 CR164c3 F 2010 0.00 5 1st Yes^ *4 CR205 0.117 CR205c1a M 1996 0.206 1st Yes^ *3 CR205c1b F 1996 0.039 1st Yes^ *4 CR205c2 M 2005 0.01 3 2nd Yes^+ *11 CR205c1a^ 0.206 CR205c3 F 2001 0.121 1st Yes^ *4 CR235 0.40 2 CR235c1 M 1998 0.41 2 1st Yes^ *3 CR235c2 F 2003 0.00 5 2nd Yes^ *4 CR235c1^ 0.41 2 CR235c3 M 2010 0.00 1 1st Yes^ *2
150 Table 4 2. Continued Offspring Sex Year of Birth Calf F Calf year at sampling Maternity Confirmed? Cow F Reconstructed Father ID from COLONY Inferred paternity Bull F CR266 0.000 CR266c1 F 1995 0.00 6 1st No# *9 CR266c2 M 1998 0.00 1 1st No *11 CR266c4 F 2003 0.08 3 1st Yes^ *2 MNW0411^ 0.06 7 CR271 0.009 CR271c1 M 1992 0.169 1st Yes^ *8 CR271c2a F 2000 0.329 1st No *11 CR271c2b M 2000 0.14 5 No# *9 CR271c3 M 2002 0.009 1st Yes^ *14 CR321 0.205 CR321c1 F 1995 0.003 2nd Yes^ *7 MNW0334+^ 0.02 1 CR321c2 F 1997 0.101 1st Yes^ *5 CR321c3 M 2001 0.096 1st Yes^ *15 CR321c4 M 2005 0.001 1st Yes^ *11 CR385c1*^ 0.21 0 CR321c5 M 2007 0.080 2nd No# *8 CR341 0.07 9 CR341c1 F 1995 0.00 5 1st Yes^ *2 MNW0334+# 0.02 1 CR341c2 M 1998 0.011 1st Yes^ *8 MSW03145+^ 0.0 20 CR341c3 F 2010 0.125 1st Yes^ *4 CR341c2^ 0.011 CR358 0.000 CR358c1 F 1997 0.21 0 1st Yes^ *2 CR358c2 F 2001 0.2 80 1st Yes^+ *8 CR358c3 F 2006 0.017 1st Yes^* *7 CR171c1*^ 0.097 CR360 0.369 CR360c1 M 1999 0.1 60 1st Yes^ *2 CR360c2 M 2007 0.00 1 1st Yes^ *6 CR360c3 F 2001 0.157 2nd Yes^ *1
151 Table 4 2. Continued Offspring Sex Year of Birth Calf F Calf year at sampling Maternity Confirmed? Cow F Reconstructed Father ID from COLONY Inferred paternity Bull F CR363 0.15 9 CR363c1 F 1997 0.202 1st Yes^ *6 CR363c2 F 2000 0.10 5 1st Yes^+ *6 CR363c3 M 2001 0.000 2nd Yes^ *16 CR436 0.63 2 CR436c1 M 2001 0.009 2nd No *8 CR436c2 M 2001 0.008 1st Yes^*+ *2 CR436c3 F 2005 0.001 2nd Yes^+ *7 CR436c2^ 0.008 CR554 0.00 1 CR554c1 M 1999 0.00 5 2nd No *5 CR554c2 F 2002 0.173 2nd No *9 CR554c3 M 2008 0.261Â§ 2nd No *8 CR554c1+^ 0.00 5 CR251 0 .000 CR251c1 M 1998 0.187 1st Yes^+ *3 CR251c3 F 2005 0.1 20 2nd Yes^+ *10 CR171 0.604 CR171c1 M 1999 0.097 1st Yes^*+ CR171c2 F 2007 0.473Â§ 1st Yes^* CR171c1*^ 0.097 * 95% Confidence in categorical assignment + 80% Confidence in categorical assignment ^ P arent O ffspring categorical assignment # H alf S ibling categorical assignment & F ull S ibling categorical assignment Â§ values indicative of inbreeding
152 Figure 4 2 . Number of bulls siring between 1 and 9 calves. Each segment going up represents a single male (0 6). Each wedge represents the number of calves sired from left to right (1 9). There were 31 males, 16 reconstructed genotypes (simulations) and 15 from the wild population (outlined in yellow). A majority of potential bulls sired between one and three offspring, with the occasional male siring between seven to nine offspring over the course of 19 years (1992 2011). F igure created by Katherine Jackson.
153 CHAPTER 5 CONCLUSIONS AND FUTURE DIRECTIONS Summary of Study Building upon four decades of behavioral studies and a decade and a half of genetic inquiry, this research took what has been known about the Florida manatee and applied it to a specific genetic niche, inbreeding and relatedness. Previous investigations o f parentage analysis had been unable to yield definitive results (Pause 2007; Bonde 2009; Jacobs 2010; Nourisson 2011) , citing too few markers, not enough power, low allelic resolution, and presumed excessive levels of relatedness and inbreeding. While lo w allelic diversity in and of itself cannot be overcome without extreme efforts, utilizing a greater number of polymorphic microsatellite markers can compensate in some ways. This study was able to take these suggestions and expound on them to enhance the knowledge base of relatedness and inbreeding in the Florida manatee. Using all 36 polymorphic microsatellite markers it was demonstrated that there was enough power to consistently distinguish between first order relationships and nonrelated but specific relationships, such as discriminating between types of siblings (full or half), was inconsistent. A novel investigation using relatedness and inbreeding metrics to distinguish between groups of Florida manatees illustrated that females are both more relat ed and inbred than males. Higher relatedness suggest female philopatry, while higher inbreeding is less clear with a multitude of possibilities with room for future examination. As this and previous research suggest, high levels of both relatedness and i nbreeding were revealed for the Florida manatee. Biologically relevant genetic
154 priors were devised via both simulated and observed pair wise comparisons, revealing that 20 26% of pairs share at least a quarter of their genes (Chapter 2). This latent coa ncestry was surprising, yet when applied to calculations of pairwise relatedness the elevated relatedness values were comparable to inferred relatedness values for known pedigrees. This lends credence to the idea that Florida manatees are related at level s substantially greater than would be expected in a healthy, randomly mating population. Categorical relationship estimators were more accurate in assignments of known relationships than relatedness estimators. However, due to the simplistic pedigree ass umptions, relationships other than parent offspring, full siblings, half siblings, and unrelated cannot be assigned. Future studies will have to take this into account when investigating deeper relationships. It is now known that the Florida manatee has latent coancestry that cannot be accounted for using current categorical relationship programs. In the future, new programs could be developed that take these second order relationships into account or relatedness estimators can be used and compared to ex pected relatedness with coancestry considered (Chapter 3). Using a full panel of microsatellite markers, the most accurate estimators, and an understanding of the coancestry of the Florida manatee, known relationships were assessed and the ability to infer paternity was explored. Maternity was verified for a majority of the relationships known through the Manatee Individual Photo identification System (MIPS). Collaborative assistance between the genetic data and the MIPS database allowed authentication fo r both sets of data, regarding potential twins, cow calf pairs, and matching genotypes. When paternity was assessed very few males were assigned as potential fathers and genotype reconstruction specified 16 male genotypes
155 to have sired 72 calves from 21 m aternal family groups. While few males were assigned to potential calves, there was not enough evidence to establish if there are dominant males in the breeding system. Unexpected relationships were assigned and plausible, including at least one mother s on mating pair. It is important, especially with the Florida manatee, to acknowledge that unexpected relationships are possible (Chapter 4). Overall, it can be concluded that the Florida manatee is highly related and inbred. The molecular tools, analytic programs, and samples available are robust enough to distinguish between related and unrelated individuals but more samples, additional microsatellite markers, and more specific relationship programs may need to be used to address more definitive relation ship questions. While currently thriving, the Florida manatee population should be continuously monitored and evaluated to secure their future. Future Directions Much of the previous genetic research on manatees has focused on the population level quantif ying genetic differences between groups, whether based on location, sex, or season. A few of the more recent studies have brought it to the individual level, which will likely be a strong focus in future manatee genetic studies. Studies characterizing in dividual manatees can focus on specific ailments, malformations, or unique genetic signatures. These can then be compared to the larger population or subpopulation to assess significance. Conclusions can then be applied to the population as a whole which will help inform more effective management decisions . A few examples of both population and individual level studies that I would like to see in the future include identifying immune system genes and investigating correlations
156 with mate choice and envir onmental hazards, such as red tide and extreme cold events. Also, I feel the expansion of this study , including a more in depth assessment of relatedness and inbreeding specifically for paternity studies and female reproduction , would be beneficial . Anot her example would be to explore associations with individual inbreeding levels further, examining multiple causes of death including infections, red tide, cold stress, perinatal, and diseases. The major histocompatibility c omplex (MHC) is a multigene fam ily that codes for proteins on the cell surface that are important in immune response (Beebee & Rowe 2008; Vassilakos et al. 2009) . MHC genes are highly polymorphic which is presumed to allow the species to better cope with a large variety of pathogens (B eebee & Rowe 2008; Eizaguirre et al. 2009; Vassilakos et al. 2009) . MHC studies can answer questions about inbreeding, mate choice, population structure, and the overall health of the population. Inbreeding leads to lower genetic diversity, which can be observed in low levels of MHC diversity. Yet, if there is MHC diversity, even though there may be low levels of genetic diversity, there may be enough polymorphism in the MHC to cope with a variety of pathogens, producing a strong immune system. A good e xample of this is the San Nicolas Island fox (Aguilar et al. 2004) , whose genetic diversity was essentially monomorphic at neutral markers, yet had an adequate amount of diversity (observed heterozygosity values ranging from 0.33 0.62) at MHC loci. The au thors concluded that severe balancing selection may have occurred to maintain genetic diversity in MHC loci and suggest using loci under selection or genes that influence fitness to assess fitness of small populations (Aguilar et al. 2004) .
157 S elect ing mates based on MHC diversity is a good mechanism for a species with low genetic diversity to give their offspring the strongest pathogen resistant genes possible. A heightened immune system is very beneficial, especially to a species that is exposed to e nvironmental pathogens, for example, red tide. Since it is known that manatees have low genetic diversity , determining the amount of MHC diversity in the Florida manatee could be beneficial (Tucker et al. 2012) . A previous study looked at two MHC loci, T rimaDRB and TrimaDQB, and found three and four haplotypes respectively in ten individuals ( Trichechus manatus manatus ) from European zoos (Jacobs 2010) . It c ould be of interest to managers to determine if there is a correlation between cause of death and MHC diversity, especially when examining cold stress and red tide events. Based on previous research, we know that there is population structure in the Florida manatees between the east and west coasts of Florida and it is also known that there are differ ences in the amount of red tide that e ffect each coast (Tucker et al. 2012; FWRI 2014) . T herefore , correlation s could be investigated between MHC diversity and death due to environmental hazards. It would also be of interest to know if MHC divers ity has an effect on mate choice. What is the MHC diversity of the manatees within a mating herd? W e suspect that the Florida manatee went through a severe population bottleneck and there is evidence of elevated inbre e d ing . Oliver and Piertney (2012) examined MHC diversity of water voles as they went through a population bottleneck. MHC allele frequencies fluctuated during the four year bottleneck but returned to parity once the population stabilized. The authors conclude that selection is an important mechan ism in maintaining MHC heterozygosity in a small population lacking genetic diversity.
158 Another topic that may produce interesting and informative results is a more focused analysis of paternity in the Florida manatee. This study set the ground work for th ose paternity studies. We know it is possible with the markers available, now we need additional male samples. A concerted effort to collect genetic samples from adult males and ideally those individuals observed in mating herds, would help address quest ions regarding dominance, reproductive success, inbreeding avoidance, and heritability in a population (Sardell et al. 2010) . Focusing on paternity will allow reproductive skew theory (i.e. how reproduction is divided among a population) to be explored wi th respect to manatees (Port & Kappeler 2010) . This theory is based on the idea that in a highly skewed population, there is a (or a couple) dominant male(s) who has control over the breeding, and conversely in a low skewed population there is more equal breeding (Johnstone 2000) . In theory, tracking either a tagged estrous adult female or a known scarred female during the warm months would led to mating herd sightings. The radioimmunoa ssay from fecal samples collected while tracking to gauge when she may be in estrous (Larkin 2000) . When mating herds are observed genetic sampling via biopsy can be conducted for as many males as possible, and the calf the following year. If tracking is not a possibility, focusing genetic sampling on adult males throughout Florida and especially those that participate in mating herds would be beneficial. If dominan ce among males in the population can be inferred, reproductive and survival estimates , and effective population size can be more accurately calculated . It could also establish if there is something unique in dominant male manatee genes that distinguish them from the rest.
159 A study assessing whether correlations exist between inbreeding levels and reproductive success would be interesting. It is presumed that if a female is reproductively viable she is reproducing. Yet, studies have determined that high levels of inbreeding lead to lower fecundity (Ober et al. 1999; Saito et al. 2000; Radwan 2003) . This type of study could be both a behavioral and genetic study incorporating the 40 years of photo identification (MIPS) data and genetic samples to determine reproductive rates and inbreeding levels for known female manatees. A more comprehensive examination of apparent causes of death including infections, red tide exposure, and cold stress could shed light on diversity levels and genes associated with combating infection and other ailments. Why do some manatees survive exposure to red tide (or cold stress), while others do not? Could there be a genetic reason? With boat strikes, cold stress, and red tide as the leading causes of death of the Florida manatee, these questions should not go unanswered. Linking these causes with inbreeding and/or relatedness could lead to a greater understanding of the fitness of the population. This would give managers and researchers alike a chance to propose ways to increase the health of the manatee population. Extreme measures could be taken, as was demonst rated with the Florida panther and other species, to introduce a small number of individuals to the population thereby increasing the genetic diversity and rescuing a population on the brink of extinction (Fenster & Galloway 2000; Edmands 2007; Johnson et al. 2010) . There is always the possibility of developing new tools such as restriction site associated DNA (RAD) markers used to quickly sequence, genotype, and identify single nucleotide polymorphisms (SNPs) to be used for phylogeographic, population, and
160 quantitati ve genetic studies (Ginzinger 2002; Davey & Blaxter 2010) . The identification of SNPs could help increase the power of current manatee specific polymorphic loci. As the affordability of high throughput sequencing increases, the potential to develop more powerful markers also increases. With the available genome and transcriptome data of the Florida manatee, studies of gene expression can investigate more deeply the interactions between the species and the environment through the identification of immune genes, vulnerability to diseases, and other health parameters (Bonde et al. 2012) . Another future goal is to have a fully genotyped set of international manatee genetic samples and defined panel of molecular markers for different types of studies that wi ll help facilitate collaborations between researchers . The scoring of microsatellite data can be subjective, so to ensure datasets that are useful to all researchers , consistent scoring needs to be implemented. If scientists start with the same standardi zed data and marker panel , then there could be world wide comparative manatee studies. This may be beneficial for regions where manatee genetic studies are just beginning and little information is known. Thereby, providing a starting point and a large co mparative dataset (e.g. the Florida manatee) with which to work. With the possible re classification of the Florida manatee, these findings, especially links between immune response and environment, will be critical for management decisions at the local, S tate, and Federal levels. The manatee, as Florid a flagship endangered species , brings substantial monetary contributions to the State, through its appeal to tourist s and locals alike. The conservation of this species will not only contribute to Florida , but more importantly continue to
161 provide encounters and interactions with a gentle and curious wild animal, as long as we continue to do our part .
162 APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 2 Figure A 1. Causes of death for the Florida manatee from 1974 2012. West Coast Utilizing 201 individuals and 31 loci, the west coast had a mean of 2. 92 alleles per locus (range 2 to 9), a mean H o of 0.392, and a H e of 0.396. The mean PIC value was 0.3 22 (range of 0.054 0.6 0 ) (Table A 1 ). Of the 36 markers, f ive were found to have homozygote excess indicating the potential presence of null alleles: TMAA02, TMAE04, TMAE07, TMAH13, and TMASC05. Tests of HWE found s ix markers deviated from expectations (p<0.05) , f ollowing Bonferroni corre ction, two deviations remained: TMAE11 and TMAK01. Twelve pairs of markers were in linkage Human, Other 2% Gate/Lock 2% Verified, not recovered 3% Cold stress 10% Natural 14% Undetermined, other 6% Undetermined, decomposed 20% Perinatal 20% Watercraft 23% CAUSES OF DEATH 1974 2012
163 disequilibrium, eight were accounted for through error estimates, while five markers: TMAK01, TMAE01, TMAE08, TMAE11, and TMAFWC1 2, were dropped from analyses. Nor thwest The northwest region had a mean of 2.81 alleles per locus (range 2 to 9), a mean Ho of 0.38, and a mean He of 0.39. The mean PIC was 0.327 and power ranged from PW R = 0.41 to 0.99 (Table 2 3), with the most power to distinguish between PO and U rela tionships. The potential presence of null alleles was identified in: TMAE04, TMAE14, and TMASC05 which had evidence of homozygote excess. Deviations from HWE expectations (p<0.05), even following Bonferroni corrections, remained in five markers: TMAE11, TMAE14, TMAK01, TMAM79, and TMASC05. There were two that were corrected for with null allele corrections. Five pairs were in linkage disequilibrium. Four markers were dropped from further analysis: TMAJ02, TMAM79, TMAE11, and TMAK01 (Table A 1). Sou thwest The SW region had 2.89 alleles per locus, a mean H o of 0.391, and a mean H e of 0.398. Marker informativeness was assessed as a PIC of 0.339 and a power range of PW R = 0.41 0.97. In tests of HWE, there were seven loci that deviated from expec tations (p<0.05), although following Bonferroni corrections only five remained: TMAE08, TMAE11, TMAE14, TMAK01, and TMASC05. Excess levels of homozygosity were discovered in TMAE04, TMAE08, TMAE14, TMAFWC14, and TMASC05, indicating the potential for null alleles. These were accounted for in all analysis programs used, and corrected three of the five loci not found in HWE. Following Bonferroni corrections six pairs of loci were in linkage disequilibrium. S ix loci were
164 dropped from further analysis: TMA A0 3 , TMAE1 1, TMA FWC04 , TMAFWC12, TMAFWC18, and TMAK01 (Table A 1).
165 Table A 1. Statistical characterization of Florida manatee polymorphic microsatellite markers for the West Coast (WC) and management units Northwest (NW) and Southwest (SW). Garcia Rodrig uez et al . (2000) (GR), Pause et al. (2007) (P), Tringali et al. (2008b) (T), annealing temperature (Tm), number of alleles (k), dropped from analysis (**), null alleles ( Â§ ), HWE ( Â¶ ), heterozygosity observed (H o ), heterozygosity expected (H E ), polymorphic information content (PIC). Primer WC k WC N WC H o WC H E WC PIC NW k NW N NW H o NW H E NW PIC SW k SW N SW H o SW H E SW PIC Private alleles TmaA02 GR 3 1144 0.399 Â§ 0.396 0.319 2 454 0.368 0.362 0.296 3 677 0.419 0.415 0.331 TmaA03 GR 2 550 0.133 0.133 0.124 2 55 0.109 0.136 0.126 2 495 0.135** 0.133 0.124 TmaE0 1 P 5 983 0.608** 0.610 0.557 5 684 0.632 0.630 0.573 5 286 0.549 0.542 0.500 TmaE0 2 GR 2 547 0.510 0.491 0.370 2 54 0.463 0.484 0.364 2 493 0.515 0.492 0.371 TmaE04 P 2 976 0.243 Â§Â¶ 0.275 0.237 2 682 0.207 Â§ 0.224 0.199 2 279 0.323 Â§ 0.377 0.305 TmaE07 P 5 1016 0.622 Â§ 0.628 0.555 5 695 0.614 0.610 0.539 3 306 0.644 0.656 0.581 NW TmaE0 8 GR 3 1355 0.472** 0.487 0.377 3 687 0.495 0.478 0.370 3 653 0.449 Â§Â¶ 0.495 0.383 TmaE11 GR 9 1377 0.597 Â¶ ** 0.595 0.538 9 691 0.616 Â¶ ** 0.608 0.544 7 671 0.575 Â¶ ** 0.571 0.522 TmaE1 4 P 3 967 0.624 0.646 0.572 5 649 0.641 Â§Â¶ 0.691 0.637 6 306 0.592 Â§Â¶ 0.680 0.629 TmaE2 6 GR 2 1074 0.055 0.055 0.054 2 411 0.027 0.026 0.026 2 651 0.074 0.074 0.071 TmaF14 GR 2 1109 0.374 0.366 0.299 2 443 0.386 0.390 0.314 2 653 0.368 0.349 0.288 Tma F WC01 T 2 596 0.225 0.230 0.204 2 56 0.357 0.296 0.250 2 540 0.211 0.223 0.198 Tma F WC02 T 6 577 0.522 0.528 0.454 4 52 0.385 0.490 0.428 6 525 0.535 0.532 0.456 SW Tma FWC03 T 2 593 0.393 0.374 0.304 2 56 0.500 0.440 0.341 2 537 0.382 0.366 0.299 Tma F WC04 T 2 596 0.131 0.134 0.125 2 56 0.232 0.234 0.205 2 540 0.120** 0.123 0.115 Tma F WC05 T 2 573 0.136 0.139 0.129 2 54 0.148 0.138 0.128 2 519 0.135 0.139 0.129 Tma FWC06 T 3 581 0.281 0.279 0.257 3 53 0.226 0.250 0.225 3 528 0.286 0.282 0.260 Tma F WC07 T 4 585 0.432 0.418 0.375 3 55 0.327 0.373 0.331 4 530 0.443 0.423 0.380 SW Tma F WC08 T 2 596 0.270 0.287 0.246 2 56 0.196 0.207 0.184 2 540 0.278 0.295 0.251 Tma FWC09 T 4 595 0.642 0.670 0.600 4 56 0.607 0.618 0.545 4 539 0.646 0.672 0.602 Tma F WC10 T 2 568 0.410 0.425 0.334 2 50 0.280 0.347 0.284 2 518 0.423 0.431 0.338 Tma F WC11 T 3 595 0.361 0.380 0.347 3 56 0.286 0.311 0.284 3 539 0.369 0.387 0.352 Tma F WC12 T 2 554 0.404** 0.394 0.316 2 49 0.306 0.403 0.320 2 505 0.414 0.393 0.316 Tma F WC13 T 3 594 0.508 0.534 0.458 3 56 0.446 0.430 0.361 3 538 0.515 0.543 0.466 Tma F WC14 T 2 552 0.288 0.316 0.266 2 50 0.360 0.389 0.311 2 502 0.281 Â§ 0.308 0.261 Tma F WC15 T 2 585 0.171 0.176 0.160 2 56 0.143 0.134 0.124 2 529 0.174 0.180 0.164 Tma FWC16 T 2 590 0.507 0.500 0.375 2 55 0.527 0.502 0.374 2 535 0.505 0.499 0.374 Tma F WC17 T 2 596 0.206 0.206 0.185 2 56 0.179 0.164 0.149 2 540 0.209 0.211 0.188
166 Table A 1. Continued Primer WC k WC N WC H o WC H E WC PIC NW k NW N NW H o NW H E NW PIC SW k SW N SW H o SW H E SW PIC Private alleles Tma FWC18 T 2 596 0.391 0.404 0.322 2 56 0.375 0.434 0.337 2 540 0.393** 0.402 0.321 TmaH13 P 3 1016 0.307 Â§ 0.325 0.300 3 694 0.298 0.312 0.289 3 307 0.329 0.356 0.327 TmaJ0 2 P 2 1030 0.362 0.353 0.291 2 704 0.344** 0.350 0.289 2 311 0.395 0.351 0.289 TmaK01 P 5 989 0.587 Â¶ ** 0.458 0.413 3 687 0.590 Â¶ ** 0.458 0.413 5 289 0.581 Â¶ ** 0.458 0.413 TmaKb 60 P 2 1025 0.507 0.495 0.372 2 700 0.524 0.500 0.375 2 310 0.477 0.462 0.355 TmaM7 9 GR 2 1080 0.497 0.484 0.367 2 443 0.558 Â¶ ** 0.482 0.365 2 625 0.454 0.485 0.367 TmaSC13 P 2 1027 0.380 0.371 0.302 2 700 0.376 0.372 0.302 2 312 0.385 0.366 0.299 TmaSC 05 P 4 985 0.540 Â§Â¶ 0.657 0.583 4 677 0.563 Â§Â¶ 0.653 0.579 4 296 0.490 Â§Â¶ 0.657 0.582 WC Avera ge 2.92 0.36 0 0.39 0 0.34 0 2.8 9 0.380 0.387 0.327 2.89 0.391 0.398 0.339
167 East Coast Utilizing 60 individuals and 33 loci, t he east coast had a mean of 3.08 alleles per locus (range 2 to 7 ) , a mean H o of 0.3 7 , and a H e of 0. 41 . The informativeness of the markers was 0.345 ( mean PIC ) and t he power ranged from PW R = 0.42 to 0.9 9 (Tables A 2 and 2 3 ). As was seen with the other regions, assignments b etween PO and U were robust while other relationships were less robust. TMAA02, TMAE04, TMAE07, TMAE08, TMAE14, and TMASC05 had evidence of homozygote excess , indicating the potential presence of null alleles. Twelve markers deviated from HWE expectations (p<0.05) , while f ollowing Bonferroni correction, seven deviations remained: TMAA02, TM AE02 , TMAE07, TMAE08, TMAE14, TMAK01, and TMASC05. Of these, all except TMAK01 were corrected for with null allele corrections. Two pairs of markers were in linkage disequilibrium : TMAE08 TMAFWC09 and TMAFWC10 TMAFWC12. TMAE08 was corrected for with nul l allele corrections, while three markers, TMAE02, TMAK01 and TMAFWC10, were dropped from analyses (Table A 2) . Atlantic The ATL region had a mean of 3.08 alleles per locus (range 2 to 7 ) , a mean H o of 0.3 9 , and a H e of 0. 41 . The informativeness of the markers was 0.345 ( mean PIC ) and t he power ranged from PW R = 0.42 to 0.9 8 (Tables A 2 and 2 3 ). As was seen with the other regions, assignments b etween PO and U were robust while other relationships were less robust. TMAA02, TMAA03 , TMAE04, TMAE07, TMAE08, TMAE14, and TMASC05 had evidence of homozygote excess , indicating the potential presence of null alleles. Twelve markers deviated from HWE expectations (p<0.05) , after Bonferroni correction s , six deviations remained: TMA E 02, TMAE07, TMAE08, TMAE14, TMAK0 1, and TMASC05. Of these, all except TMAE02 and TMAK01 were corrected for with null
168 allele corrections. Two pairs of markers were in linkage disequilibrium : TMAE08 TMAFWC09 and TMAFWC10 TMAFWC12. TMAE08 was corrected for with null allele corrections, wh ile three markers, TMAE02, TMAK01 and TMAFWC12 , were dropped from analyses (Table A 2) . Upper St Johns The USJ region had a mean of 2.7 alleles per locus (range 2 to 5), a mean Ho of 0.40, and a mean He of 0.40. The mean PIC was 0.341 and power ranged f rom PW R = 0.52 to 1.0 (Table 2 3), with the most power to distinguish between PO and U relationships. No null alleles were identified. There were three markers that deviated from HWE expectations (p<0.05), although were corrected following Bonferroni corre ctions (Table A 2).
169 Table A 2. Statistical characterization of Florida manatee polymorphic microsatellite markers for the East Coast (EC) and Rodriguez et al. (2000) (GR) , Pause et al. (2007) (P), Tringali et al. (2008b) (T), annealing temperature (Tm), number of alleles (k), dropped from analysis (**), null alleles (Â§), HWE (Â¶), heterozygosity observed (H o ), heterozygosity expected (H E ), polymorphic information content (P IC). Primer EC k EC N EC H o EC H E EC PIC ATL k ATL N ATL H o ATL H E ATL PIC USJ k USJ N USJ H o USJ H E USJ PIC Private alleles TmaA02 GR 4 391 0.352 Â§Â¶ 0.404 0.326 4 533 0.351 Â§Â¶ 0.401 0.324 2 22 0.364 0.444 0.340 ATL TmaA03 GR 2 377 0.220* 0.244 0.214 2 596 0.210 Â§ 0.231 0.204 2 58 0.172 0.187 0.168 TmaE01 P 6 43 0.500 0.526 0.471 6 144 0.493 0.524 0.470 4 35 0.457 0.540 0.474 ATL TmaE02 GR 3 361 0.438* 0.453 0.350 3 576 0.446 Â¶ ** 0.462 0.356 2 61 0.410 0.450 0.347 TmaE04 P 2 64 0.283 Â§ 0.352 0.290 2 183 0.284 Â§ 0.349 0.287 2 38 0.263 0.305 0.255 TmaE07 P 4 62 0.503 Â§Â¶ 0.634 0.562 4 181 0.508 Â§Â¶ 0.639 0.567 3 35 0.486 0.610 0.517 TmaE08 GR 3 378 0.460 Â§Â¶ 0.523 0.428 3 581 0.463 Â§Â¶ 0.523 0.428 3 59 0.542 0.489 0.416 TmaE11 GR 7 392 0.645 0.656 0.608 7 583 0.647 0.660 0.612 5 50 0.640 0.558 0.520 TmaE14 P 5 62 0.535 Â§Â¶ 0.597 0.552 5 231 0.528 Â§Â¶ 0.584 0.539 5 45 0.467 0.591 0.534 TmaE26 GR 3 371 0.272 0.278 0.246 3 452 0.268 0.277 0.245 2 17 0.353 0.299 0.248 ATL TmaF14 GR 2 378 0.358 0.352 0.290 2 476 0.357 0.347 0.286 2 19 0.316 0.444 0.339 Tma FWC01 T 2 422 0.270 0.292 0.249 2 503 0.276 0.295 0.252 2 55 0.327 0.323 0.269 Tma FWC02 T 4 394 0.581* 0.571 0.477 4 473 0.564 0.568 0.472 4 57 0.632 0.569 0.468 Tma FWC03 T 2 416 0.440 0.435 0.340 2 427 0.440 0.436 0.341 2 16 0.438 0.417 0.323 Tma FWC04 T 2 421 0.181 0.183 0.166 2 429 0.186 0.188 0.170 1 16 0.000 0.000 0.000 Tma FWC05 T 2 396 0.240 0.248 0.217 2 384 0.240 0.249 0.218 2 15 0.267 0.239 0.204 Tma FWC06 T 3 402 0.279 0.270 0.253 3 450 0.276 0.268 0.251 3 51 0.333 0.309 0.285 Tma FWC07 T 4 409 0.479 0.462 0.422 4 396 0.472 0.456 0.417 3 16 0.625 0.583 0.503 ATL Tma FWC08 T 3 422 0.237 0.252 0.222 3 471 0.236 0.253 0.222 2 18 0.222 0.203 0.178 ATL Tma FWC09 T 5 420 0.663 0.680 0.621 5 545 0.659 0.683 0.625 4 52 0.712 0.619 0.538 ATL Tma FWC10 T 2 387 0.447** 0.456 0.352 2 431 0.413 0.453 0.350 2 48 0.500 0.449 0.346 Tma FWC11 T 3 420 0.359 0.364 0.333 3 525 0.364 0.366 0.335 3 61 0.361 0.388 0.345 Tma FWC12 T 2 374 0.331 0.336 0.279 2 457 0.330** 0.336 0.279 2 52 0.288 0.276 0.236 Tma FWC13 T 3 413 0.315 0.325 0.279 3 500 0.320 0.330 0.283 2 59 0.322 0.337 0.278 Tma FWC14 T 2 355 0.413 0.445 0.346 2 346 0.413 0.445 0.346 2 14 0.429 0.476 0.354 Tma FWC15 T 2 403 0.118 0.124 0.116 2 398 0.116 0.118 0.111 2 16 0.125 0.226 0.195 Tma FWC16 T 2 403 0.470 0.480 0.364 2 389 0.478 0.481 0.365 2 16 0.375 0.444 0.337
170 Table A 2. Continued Primer EC k EC N EC H o EC H E EC PIC ATL k ATL N ATL H o ATL H E ATL PIC USJ k USJ N USJ H o USJ H E USJ PIC Private alleles Tma FWC17 T 3 421 0.041 0.044 0.044 3 442 0.041 0.044 0.044 2 18 0.056 0.056 0.053 ATL Tma FWC18 T 2 422 0.394 0.365 0.298 2 560 0.391 0.365 0.298 2 53 0.396 0.364 0.295 TmaH13 P 3 62 0.333 0.363 0.333 3 155 0.342 0.363 0.333 3 33 0.303 0.360 0.324 TmaJ02 P 2 63 0.348 0.376 0.305 2 126 0.381 0.380 0.307 2 9 0.111 0.425 0.321 TmaK01 P 6 59 0.627 Â¶ ** 0.499 0.453 6 169 0.669 Â¶ ** 0.501 0.445 3 38 0.632 0.476 0.410 TmaKb60 P 2 62 0.519 0.448 0.347 2 126 0.484 0.438 0.341 2 8 0.750 0.500 0.359 TmaM79 GR 3 342 0.480 0.493 0.376 3 453 0.477 0.493 0.375 2 17 0.412 0.515 0.375 ATL TmaSC13 P 3 64 0.448 0.656 0.579 3 189 0.450 0.430 0.339 2 45 0.489 0.433 0.337 TmaSC05 P 3 51 0.432 Â§Â¶ 0.432 0.340 3 110 0.473 0.658 Â§Â¶ 0.580 3 9 0.333 0.660 0.544 Average 3.08 0.365 0.406 0.346 3.08 0.391 0.405 0.345 2.7 0 0.399 0.404 0.341
171 APPENDIX B SUPPORTING INFORMATION FOR CHAPTER 3
172 Table B 1. All pairwise relationships from captive pedigree dataset of Chapter 3. Columns 3 8 are expected values. Column 3 is presumed relationships based on s tudbook information, columns 4 8 are expected relatedness values for outbred (r xy ) and Ocean Reef and Foster as PO , and as HS, respectively. Column 9 is presumed relationships based on studbook information and considering Foster and Ocean Reef as PO. Columns 10 13 are COANCESTRY relatedness estimators TrioML and Wang with their respective 95% CIs. Column 14 is the relatedness v alue from ML RELATE. Columns 15 17 are relationship estimates from ML Relate: relative likelihood (LnL(R)) , posterior probability of relationship (PP best) , and posterior probability with relationship including prior (PP best with prior) . Ind1 Ind2 R r xy OR+F PO OR+F HS R OR F TrioML r xy (95% CI) Wang r xy (95% CI) ML Relate r xy LnL(R) PP best PP best w/ prior Romeo Juliet U 0.250 0.347 0.347 0.347 0.347 HS 0.212 (0 0.71) 0.337 ( 0.12 0.65) 0.218 64.45 HS; 0.55 U; 0.40 Juliet Buffet PO 0.500 0.565 0.639 0.639 0.639 PO 0.747 (0.44 1.1) 0.456 (0.26 0.66) 0.500 60.43 PO; 0.88 PO; 0.88 Romeo Buffet PO 0.500 0.565 0.639 0.639 0.639 PO 0.688 (0.44 1.19) 0.591 (0.37 0.78) 0.534 55.96 PO; 0.75 PO; 0.75 Juliet Foster PO 0.500 0.565 0.639 0.639 0.639 PO 0.583 (0.35 0.86) 0.516 (0.27 0.69) 0.546 62.18 PO; 0.68 PO; 0.68 Romeo Foster PO 0.500 0.565 0.639 0.639 0.639 PO 0.563 (0.37 1.10) 0.564 (0.26 0.71) 0.546 59.85 PO; 0.65 PO; 0.65 Lorelei Hugh PO 0.813 0.829 0.829 0.829 0.829 PO 0.941 (0.83 1.34) 0.744 (0.61 0.94) 0.748 55.44 FS; 0.67 FS; 0.67 Romeo Hugh PO 0.750 0.766 0.766 0.766 0.766 PO 0.655 (0.53 1.02) 0.511 (0.36 0.69) 0.500 56.36 PO; 0.87 PO; 0.87 Juliet Hurricane PO 0.500 0.565 0.639 0.639 0.639 PO 0.556 (0.36 1) 0.688 (0.48 0.85) 0.548 52.46 FS; 0.87 FS; 0.87 Romeo Hurricane PO 0.500 0.565 0.639 0.639 0.639 PO 0.460 (0.28 1) 0.563 (0.28 0.81) 0.574 54.23 FS; 0.5 FS; 0.49 Juliet Lorelei PO 0.500 0.565 0.639 0.639 0.639 PO 0.414 (0.24 0.75) 0.479 (0.22 0.67) 0.541 63.3 PO; 0.49 PO; 0.48 Romeo Lorelei PO 0.500 0.565 0.639 0.639 0.639 PO 0.615 (0.50 0.93) 0.701 (0.47 0.86) 0.700 54.53 FS; 0.8 FS; 0.8 0 Romeo Stoneman PO 0.750 0.766 0.766 0.766 0.766 PO 0.544 (0.28 1.21) 0.558 (0.30 0.80) 0.518 54.22 PO; 0.51 PO; 0.46
173 Table B 1. Continued Ind1 Ind2 R rxy OR+F PO OR+F HS R OR F TrioML rxy (95% CI) Wang rxy (95% CI) ML Relate rxy LnL(R) PP best PP best w/ prior Buffet Foster FS 0.500 0.565 0.639 0.639 0.639 FS 0.445 (0.35 1) 0.348 (0.01 0.72) 0.500 63.19 FS; 0.86 FS; 0.85 Buffet Hurricane FS 0.500 0.565 0.639 0.639 0.639 FS 0.752 (0.43 1.21) 0.538 (0.24 0.74) 0.500 55.51 PO; 0.79 PO; 0.79 Buffet Lorelei FS 0.500 0.565 0.639 0.639 0.639 FS 0.467 (0.16 0.93) 0.261 ( 0.56) 0.330 65.04 HS; 0.87 HS; 0.71 Foster Hurricane FS 0.500 0.565 0.639 0.639 0.639 FS 0.604 (0.40 1.02) 0.578 (0.41 0.75) 0.553 55.28 PO; 0.7 PO; 0.7 0 Foster Lorelei FS 0.500 0.565 0.639 0.639 0.639 FS 0.483 (0.34 0.94) 0.421 (0.06 0.69) 0.437 64.57 FS; 0.66 FS; 0.61 Hurricane Lorelei FS 0.500 0.565 0.639 0.639 0.639 FS 0.515 (0.33 0.91) 0.547 (0.36 0.77) 0.578 55.74 PO; 0.48 PO; 0.48 Hugh Stoneman HS 0.375 0.381 0.381 0.381 0.381 HS/ 1st cousin 0.627 (0.40 1.08) 0.456 (0.04 0.75) 0.437 54.71 FS; 0.8 FS; 0.71 Buffet Hugh HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.045 (0 0.59) 0.207 ( 0.98) 0.024 66.6 U; 0.77 U; 0.96 Buffet Stoneman HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.194 (0 0.73) 0.054 ( 0.48 0.58) 0.137 59.06 HS; 0.49 U; 0.87 Foster Hugh HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.453 (0.25 0.88) 0.21 ( 0.27 0.56) 0.363 64.31 FS; 0.56 FS; 0.43 Foster Stoneman HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.342 (0.18 0.94) 0.319 (0.01 0.73) 0.361 57.69 FS; 0.68 FS; 0.56 Hurricane Hugh HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.584 (0.29 0.98) 0.384 (0.12 0.60) 0.500 57.96 PO; 0.73 PO; 0.65 Hurricane Stoneman HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.567 (0.13 0.92) 0.452 (0.21 0.64) 0.500 52.21 PO; 0.61 PO; 0.53 Lorelei Stoneman HS 0.481 0.493 0.490 0.490 0.490 HS/ Nephew 0.509 (0.25 1.08) 0.425 (0.17 0.74) 0.406 57.57 FS; 0.55 FS; 0.43 Juliet Hugh HS 0.250 0.347 0.347 0.347 0.347 Grandmother / Half Aunt 0.172 (0.02 0.62) 0.153 ( 0.24 0.51) 0.153 67.05 HS; 0.54 U; 0.86 Juliet Stoneman HS 0.250 0.347 0.347 0.347 0.347 Grandmother / Half Aunt 0.267 (0.08 0.79) 0.281 ( 0.14 0.66) 0.281 59.77 FS/ HS; 0.45 U; 0.49
174 Table B 1. Continued Ind1 Ind2 R rxy OR+F PO OR+F HS R OR F TrioML rxy (95% CI) Wang rxy (95% CI) ML Relate rxy LnL(R) PP best PP best w/ prior Foster Ocean Reef U 0.000 0.129 0.129 0.595 0.378 PO 0.607 (0.14 1.05) 0.379 (0.19 0.60) 0.474 64.46 PO; 0.61 PO; 0.55 Foster Patch U 0.000 0.129 0.129 0.378 0.157 Grandfather 0.105 (0 0.51) 0.025 ( 0.29 0.26) 0.092 61.62 U; 0.51 U; 0.90 Foster Pumpkin U 0.000 0.129 0.129 0.378 0.157 Grandfather 0.271 (0.02 0.75) 0.019 ( 0.33 0.33) 0.246 66.56 HS; 0.52 U; 0.71 Romeo Ocean Reef U 0.000 0.129 0.129 0.282 0.141 Grandfather 0.170 (0 0.80) 0.205 ( 0.22 0.51) 0.214 62.7 HS; 0.71 U; 0.74 Juliet Ocean Reef U 0.000 0.129 0.129 0.282 0.141 Grandmother 0.387 (0.18 1.0) 0.253 ( 0.19 0.63) 0.323 65.62 FS; 0.52 FS; 0.43 Hugh Ocean Reef U 0.000 0.129 0.129 0.262 0.131 Half uncle/ 1st cousin 0 .000 (0 0.47) 0.266 ( 0.79 0.24) 0.000 65.91 U; 0.9 U; 0.99 Stoneman Ocean Reef U 0.000 0.129 0.129 0.262 0.131 Half uncle/ 1st cousin 0.194 (0 0.63) 0.039 ( 0.57 0.39) 0.000 57.27 U; 0.72 U; 0.96 Lorelei Ocean Reef U 0.000 0.129 0.129 0.378 0.157 Aunt 0.080 (0 0.66) 0.102 ( 0.27 0.48) 0.114 67.46 HS; 0.54 U; 0.87 Buffet Ocean Reef U 0.000 0.129 0.129 0.378 0.157 Uncle 0.402 (0.25 0.967) 0.148 ( 0.77) 0.297 66.65 HS; 0.71 HS; 0.56 Hurricane Ocean Reef U 0.000 0.129 0.129 0.378 0.157 Uncle 0.332 (0 0.70) 0.14 ( 0.24 0.49) 0.170 60.47 HS; 0.65 U; 0.77 Romeo Patch U 0.000 0.129 0.129 0.141 0.071 Great grandfather 0.045 (0 0.33) 0.009 ( 0.48 0.32) 0.089 58.36 U; 0.65 U; 0.94 Romeo Pumpkin U 0.000 0.129 0.129 0.141 0.071 Great grandfather 0.016 (0 0.39) 0.047 ( 0.40 0.40) 0.066 62.88 U; 0.58 U; 0.92 Juliet Patch U 0.000 0.129 0.129 0.141 0.071 Great grandmother 0.053 (0 0.44) 0.099 ( 0.43 0.32) 0.090 62.01 HS; 0.5 U; 0.89 Juliet Pumpkin U 0.000 0.129 0.129 0.141 0.071 Great grandmother 0.157 (0 0.54) 0.033 ( 0.50 0.40) 0.171 66.75 HS; 0.7 U; 0.73 Lorelei Patch U 0.000 0.129 0.129 0.157 0.078 Great aunt 0.071 (0 0.46) 0.029 ( 0.24 0.42) 0.119 62.25 HS; 0.49 U; 0.87 Lorelei Pumpkin U 0.000 0.129 0.129 0.157 0.078 Great aunt 0.012 (0 0.27) 0.086 ( 0.58 0.29) 0.060 66.8 U; 0.79 U; 0.97 Buffet Patch U 0.000 0.129 0.129 0.157 0.078 Great uncle 0.105 (0 0.46) 0.255 ( 0.91) 0.111 64.22 U; 0.5 U; 0.89 Buffet Pumpkin U 0.000 0.129 0.129 0.157 0.078 Great uncle 0.213 (0.05 0.59) 0.085 ( 0.53 0.36) 0.188 67.49 HS; 0.75 U; 0.69 Hurricane Patch U 0.000 0.129 0.129 0.157 0.078 Great uncle 0.059 (0 0.50) 0.008 ( 0.42 0.42) 0.109 60.16 HS; 0.5 U; 0.87
175 Table B 1. Continued Ind1 Ind2 R rxy OR+F PO OR+F HS R OR F TrioML rxy (95% CI) Wang rxy (95% CI) ML Relate rxy LnL(R) PP best PP best w/ prior Hurricane Pumpkin U 0.000 0.129 0.129 0.157 0.078 Great uncle 0.030 (0 0.50) 0.176 ( 0.60 0.34) 0.081 61.69 U; 0.78 U; 0.97 Hugh Patch U 0.000 0.129 0.129 0.131 0.065 Half greatuncle/ 2nd cousin 0.031 (0 0.61) 0.19 ( 0.73 0.20) 0.050 61.15 U; 0.64 U; 0.94 Hugh Pumpkin U 0.000 0.129 0.129 0.131 0.065 Half greatuncle/ 2nd cousin 0 .000 (0 0.33) 0.348 ( 0.87 0.10) 0.000 65.07 U; 0.96 U; 0.99 Stoneman Patch U 0.000 0.129 0.129 0.131 0.065 Half greatuncle/ 2nd cousin 0.000 (0 0.31) 0.125 ( 0.52 0.20) 0.000 56.37 U; 0.88 U; 0.98 Stoneman Pumpkin U 0.000 0.129 0.129 0.131 0.065 Half greatuncle/ 2nd cousin 0.084 (0 0.69) 0.093 ( 0.37 0.24) 0.000 56.51 U; 0.78 U; 0.97 Gene Rita U 0.000 0.129 0.129 0.129 0.129 U 0 .000 (0 0.39) 0.158 ( 0.72 0.38) 0.002 65.96 U; 0.92 U; 0.99 Gene Dundee PO 0.500 0.565 0.565 0.565 0.565 PO 0.570 (0.35 0.89) 0.501 (0.24 0.73) 0.500 55 PO; 0.73 PO; 0.72 Rita Dundee PO 0.500 0.565 0.565 0.565 0.565 PO 0.575 (0.42 0.89) 0.409 (0.09 0.70) 0.557 56.5 PO; 0.59 PO; 0.59 Ocean Reef Patch PO 0.500 0.565 0.565 0.565 0.565 PO 0.756 (0.54 1.31) 0.619 (0.46 0.85) 0.588 53.33 FS; 0.5 FS; 0.5 Ocean Reef Pumpkin PO 0.500 0.565 0.565 0.565 0.565 PO 0.616 (0.22 1.0) 0.523 (0.35 0.76) 0.620 58.62 PO; 0.5 PO; 0.5 Patch Pumpkin FS 0.500 0.565 0.565 0.565 0.565 FS 0.507 (0.23 1.05) 0.548 (0.05 0.84) 0.530 53.57 FS; 0.97 FS; 0.97 Buffet Dundee U 0.000 0.129 0.129 0.129 0.129 U 0.117 (0 0.6) 0.235 ( 0.78) 0.119 65.66 HS; 0.5 U; 0.89 Buffet Gene U 0.000 0.129 0.129 0.129 0.129 U 0.316 (0 0.75) 0.092 ( 0.41 0.48) 0.242 63.83 HS; 0.71 U; 0.65 Buffet Rita U 0.000 0.129 0.129 0.129 0.129 U 0.177 (0.04 0.75) 0.199 ( 0.97 0.47) 0.177 68.03 U; 0.84 U; 0.98 Dundee Foster U 0.000 0.129 0.129 0.129 0.129 U 0.180 (0 0.58) 0.018 ( 0.30 0.36) 0.163 62.71 HS; 0.57 U; 0.84 Dundee Hugh U 0.000 0.129 0.129 0.129 0.129 U 0.038 (0 0.48) 0.028 ( 0.53 0.41) 0.024 62.56 U; 0.78 U; 0.97
176 Table B 1. Continued Ind1 Ind2 R rxy OR+F PO OR+F HS R OR F TrioML rxy (95% CI) Wang rxy (95% CI) ML Relate rxy LnL(R) PP best PP best w/ prior Dundee Hurricane U 0.000 0.129 0.129 0.129 0.129 U 0.083 (0 0.54) 0.115 ( 0.47 0.51) 0.098 61.72 U; 0.59 U; 0.92 Dundee Juliet U 0.000 0.129 0.129 0.129 0.129 U 0.173 (0.05 0.62) 0.232 ( 0.16 0.61) 0.203 62.37 HS; 0.51 U; 0.63 Dundee Lorelei U 0.000 0.129 0.129 0.129 0.129 U 0.058 (0 0.47) 0.062 ( 0.37 0.42) 0.075 63.74 U; 0.73 U; 0.96 Dundee Ocean Reef U 0.000 0.129 0.129 0.129 0.129 U 0.141 (0 0.51) 0.001 ( 0.46 0.38) 0.139 63.51 HS; 0.55 U; 0.87 Dundee Patch U 0.000 0.129 0.129 0.129 0.129 U 0.228 (0.06 0.62) 0.123 ( 0.32 0.57) 0.247 61.44 HS; 0.62 U; 0.62 Dundee Pumpkin U 0.000 0.129 0.129 0.129 0.129 U 0 .000 (0 0.32) 0.129 ( 0.51 0.21) 0.000 64.15 U; 0.89 U; 0.99 Dundee Romeo U 0.000 0.129 0.129 0.129 0.129 U 0.029 (0 0.36) 0.104 ( 0.25 0.46) 0.041 59.82 U; 0.6 U; 0.92 Dundee Stoneman U 0.000 0.129 0.129 0.129 0.129 U 0.019 (0 0.48) 0.111 ( 0.24 0.42) 0.016 55.15 U; 0.58 U; 0.92 Foster Gene U 0.000 0.129 0.129 0.129 0.129 U 0.154 (0.04 0.60) 0.025 ( 0.36 0.45) 0.138 65.47 U; 0.69 U; 0.95 Foster Rita U 0.000 0.129 0.129 0.129 0.129 U 0.296 (0.05 0.69) 0.113 ( 0.20 0.39) 0.207 68.69 HS; 0.72 U; 0.74 Gene Hugh U 0.000 0.129 0.129 0.129 0.129 U 0.059 (0 0.59) 0.217 ( 0.77 0.35) 0.000 63.14 U; 0.93 U; 0.99 Gene Hurricane U 0.000 0.129 0.129 0.129 0.129 U 0.385 (0 0.69) 0.124 ( 0.30 0.57) 0.223 58.94 HS; 0.61 U; 0.75 Gene Juliet U 0.000 0.129 0.129 0.129 0.129 U 0.280 (0 0.55) 0.084 ( 0.47 0.44) 0.169 65.46 HS; 0.57 U; 0.84 Gene Lorelei U 0.000 0.129 0.129 0.129 0.129 U 0.152 (0.01 0.49) 0.061 ( 0.55 0.55) 0.125 64.87 U; 0.67 U; 0.94 Gene Ocean Reef U 0.000 0.129 0.129 0.129 0.129 U 0.348 (0.06 0.81) 0.095 ( 0.22 0.41) 0.107 64.5 U; 0.54 U; 0.91 Gene Patch U 0.000 0.129 0.129 0.129 0.129 U 0.052 (0 0.55) 0.022 ( 0.45 0.47) 0.046 59.19 U; 0.68 U; 0.95 Gene Pumpkin U 0.000 0.129 0.129 0.129 0.129 U 0.028 (0 0.44) 0.013 ( 0.39 0.31) 0.000 63.66 U; 0.84 U; 0.98 Gene Romeo U 0.000 0.129 0.129 0.129 0.129 U 0.185 (0 0.52) 0.093 ( 0.41 0.56) 0.116 62.33 U; 0.51 U; 0.89
177 Table B 1. Continued Ind1 Ind2 R rxy OR+F PO OR+F HS R OR F TrioML rxy (95% CI) Wang rxy (95% CI) ML Relate rxy LnL(R) PP best PP best w/ prior Gene Stoneman U 0.000 0.129 0.129 0.129 0.129 U 0 .000 (0 0.21) 0.276 ( 0.98 0.26) 0.000 57.12 U; 0.95 U; 0.99 Hugh Rita U 0.000 0.129 0.129 0.129 0.129 U 0.240 (0.02 0.82) 0.099 ( 0.42 0.52) 0.118 67.37 U; 0.52 U; 0.90 Hurricane Rita U 0.000 0.129 0.129 0.129 0.129 U 0.071 (0 0.53) 0 .000 ( 0.45 0.40) 0.000 64 U; 0.86 U; 0.98 Juliet Rita U 0.000 0.129 0.129 0.129 0.129 U 0.118 (0 0.42) 0.029 ( 0.41 0.35) 0.061 70.08 U; 0.69 U; 0.95 Lorelei Rita U 0.000 0.129 0.129 0.129 0.129 U 0.169 (0.01 0.70) 0.267 ( 0.14 0.59) 0.179 68.93 HS; 0.49 U; 0.85 Ocean Reef Rita U 0.000 0.129 0.129 0.129 0.129 U 0.289 (0.06 0.71) 0.127 ( 0.33 0.42) 0.240 65.48 HS; 0.84 U; 0.55 Patch Rita U 0.000 0.129 0.129 0.129 0.129 U 0.181 (0 0.54) 0.004 ( 0.46 0.34) 0.214 60.76 HS; 0.75 U; 0.68 Pumpkin Rita U 0.000 0.129 0.129 0.129 0.129 U 0.089 (0 0.54) 0.046 ( 0.59 0.37) 0.096 66.51 U; 0.74 U; 0.96 Rita Romeo U 0.000 0.129 0.129 0.129 0.129 U 0.157 (0 0.68) 0.265 ( 0.11 0.54) 0.129 66.19 HS; 0.56 U; 0.84 Rita Stoneman U 0.000 0.129 0.129 0.129 0.129 U 0.500 (0.08 1.07) 0.322 ( 0.04 0.62) 0.300 59.82 HS; 0.67 U; 0.58
178 APPENDIX C SUPPORTING INFORMATION FOR CHAPTER 4 Table C 1. Cow Calf pairs from Crystal River, Florida used in Chapter 4. Symbols used: also included in the Bonde (2009) study (^), dropped from analysis due to incomplete genotype (*). Sample Name Date Size Sex Comments CCR 11 18 10 Nov 11 274 F CCR 11 18 lactating CCR 11 18c1 10 Nov 11 232 F CCR 11 17calf of CCR 11 18 CCR 11 21 10 Nov 11 339 F CCR 11 21 lactating CCR 11 21c1 10 Nov 11 260 M CCR 11 20 calf of CCR 11 21? CR026^ 14 Nov 06 300 F CR026c1^ 28 Dec 05 230 M CR027^ 22 Nov 06 300 F CR027c1^ 02 Dec 97 230 F CR027c2^ 16 Nov 99 175 F CR027c3^ 26 Nov 02 220 M CR027c4 7 Jan 11 180 F 11 007 02 CR032^ 28 Nov 07 350 F MAGGIE CR032c1^ 12 Feb 92 220 F CR032c2^ 1 Dec 93 230 F CR032c3*^ 10 Dec 97 180 F CR032c4a^ 28 Dec 99 170 M Duplicate sample CR032c4b*^ 29 Dec 99 170 M Duplicate sample CR032c6 7 Jan 02 200 F GS319, TM884, 02 007 07 CR041 12 Jan 09 320 F CCR 09 04 CR041c1 20 Jan 93 235 F Calf of CR041 CR045^ 15 Jan 07 340 F CR045c1 19 Dec 00 160 F GS269, 00 354 02 CR046^ 22 Nov 06 300 F TWINS WITH CR049 CR046c1^ 01 Feb 95 180 F same as CR506 CR046c2^ 10 Dec 97 230 F CR046c3 30 Mar 06 170 F GS510, 06 089 01 CR049^ 8 Oct 06 300 F TWINS WITH CR046 CR049c1^ 28 Nov 00 210 M CR054^ 6 Dec 06 340 F CR054c1^ 12 Feb 92 230 F CR054c2*^ 6 Dec 96 220 M CR054c3^ 16 Dec 02 190 F CR060*^ 16 Nov 06 320 F CR060c1^ 24 Feb 99 160 M CR060c2^ 30 Oct 01 210 M CR060c3^ 16 Nov 06 160 F CR061^ 16 Nov 06 290 F CR061c1^ 11 Jan 01 220 F
179 Table C 1. Continued Sample Name Date Size Sex Comments CR070^ 29 Jan 07 320 F ATLANTA, CALF IS CR130 CR071^ 18 Apr 07 LA F PIETY, CALF IS CR104 CR071c1^ 8 Mar 96 230 F CR071c2*^ 08 Jan 03 180 M CR075 16 Feb 08 330 F 08 047 01, CR146? CR075c1 3 Jan 90 ? F 90 003 01, CR149? CR075c2 14 Jan 00 180 F 00 014 05, CR148? CR093^ 14 Nov 07 320 F CR093c1^ 19 Nov 02 230 M CR093c2 19 Jan 05 160 M GS401, 05 019 03 CR104^ 15 Jan 07 320 F NARNIA, CALF OF CR071 CR104c1*^ 20 Jan 93 240 M CR104c2^ 17 Jan 97 230 M CR104c3*^ 24 Apr 98 160 M CR104c4^ 04 Feb 00 220 M CR104c5^ 20 Nov 00 180 F CR104c6^ 17 Jan 02 180 M CR111 18 Nov 10 320 F 10 322 01, GS758 CR111c1 10 Feb 92 220 M 92 041 01, GS006 CR111c2 6 Dec 97 220 M 97 340 01, GS122 CR111c3 22 Dec 09 170 ? 09 356 02 CR111c4 29 Dec 10 170 F 10 363 01 CR123^ 29 Jan 07 320 F CR123c1^ 12 Feb 92 240 M CR123c2^ 25 Nov 96 235 F CR123c3^ 24 Feb 99 200 F CR123c4^ 20 Nov 00 200 M CR123c5 9 Feb 11 140 F 11 040 04 CR125^ 16 Nov 06 280 F CR125c1^ 31 Jan 95 210 M CR125c2^ 04 Nov 97 230 F CR125c3^ 01 Dec 00 200 M CR130^ 8 Oct 06 300 F Calf of Atlanta CR130c1^ 12 Jan 98 220 M CR130c2a*^ 08 Jan 03 230 M NOT TWIN CR130c2b^ 17 Jan 03 210 F NOT TWIN Adopted? CR130c3 8 Nov 10 LC M 10 312 02 CR133^ 4 Nov 06 300 F CR133c1^ 22 Dec 97 200 F CR133c2^ 01 Dec 00 160 M CR139 7 Jan 11 320 F CR139c1 12 Feb 92 220 F Calf of CR139, GS008
180 Table C 1. Continued Sample Name Date Size Sex Comments CR157c1*^ 07 Nov 01 160 F CR164^ 4 Nov 06 290 F CR164c1^ 29 Dec 97 170 F CR164c2^ 13 Nov 01 190 M CR164c3 3 Feb 10 160 F 10 034 02 CR171^ 28 Nov 07 320 F CR171c1^ 24 Feb 99 180 M CR171c2^ 28 Nov 07 170 F CR205^ 7 Feb 07 290 F CR205c1a^ 06 Dec 96 140 M TWIN Paternal CR205c1b^ 06 Dec 96 140 F TWIN Paternal CR205c2^ 22 Jan 05 220 M CR205c3^ 05 Jan 01 200 F CR235^ 14 Nov 07 290 F CR235c1^ 19Feb98 190 M CR235c2^ 13 Nov 03 210 F CR235c3 31 Oct 10 160 M 10 304 02 CR236 31 Oct 10 300 F GS756 CR236c1 7 Nov 01 180 F CALF OF CR236, GS306, TM867 CR251^ 4 Nov 06 300 F CR251c1^ 9 Feb 98 190 M CR251c2^ 4 Nov 06 220 F SAME AS CR251c3 CR251c3 27 Oct 05 SC F 06 326 04, GS421, SAME AS CR251c2 CR263^ 31 Jan 07 280 F CR263c1^ 06 Nov 01 210 F SAME AS CR263c2 CR263c2 3 Dec 02 230 F TM693, 02 337 03, SAME AS CR263c1 CR263c3 29 Jan 07 230 F GS579, 07 029 06, CR266^ 17 Jan 08 300 F CR266c1^ 1 Feb 95 200 F CR266c2^ 14 Jan 98 160 M CR266c3*^ 22 Jan 03 180 F CR266c4 17 Jan 03 170 F GS374, 03 017 02 CR271^ 22 Nov 06 300 F CR271c1^ 10 Feb 92 140 M CR271c2a^ 21 Nov 00 180 F NOT TWIN, ADOPTED CR271c2b^ 01 Dec 00 200 M NOT TWIN, ONE ALLELE OFF MOM CR271c3^ 30 Dec 02 170 M CR271c4^ 22 Nov 06 180 M CR272^ 5 Nov 06 300 F CR272c1^ 09 Feb 98 240 M CR272c2^ 22 Dec 04 190 M
181 Table C 1. Continued Sample Name Date Size Sex Comments CR277^ 4 Nov 06 340 F CR277c1^ 23 Jan 96 180 M CR277c2^ 11 Nov 03 190 M CR278^ 22 Nov 06 290 F CR278c1^ 4 Feb 93 180 F CR278c2^ 22 Nov 06 200 M CR299 4 Jan 08 320 F GS651 CR299c1 4 Dec 97 210 M Calf of CR299, GS121 CR321^ 14 Nov 06 300 F CALF IS CR458 CR321c1^ 27 Jan 95 230 F SAME AS CR458 CR321c2^ 08 Nov 97 200 F CR321c3^ 29 Oct 01 200 M CR321c4^ 22 Jan 05 190 M CR321c5^ 05 Nov 07 210 M CR333^ 08 Oct 06 290 F CR333c1^ 08 Oct 06 190 M CR339^ 22 Nov 06 280 F CR339c1^ 26 Nov 02 220 M CR341^ 29 Jan 07 300 F CR341c1^ 21 Feb 95 195 F CR341c2^ 18 Dec 98 200 M CR341c3 3 Feb 10 180 F 10 034 01 CR354^ 5 Nov 06 300 F CR354c1^ 08 Jan 01 200 M SAME AS CR354c2 CR354c2^ 16 Jan 02 220 M SAME AS CR354c1 CR354c3^ 5 Nov 06 180 F CR358^ 4 Nov 06 290 F CR358c1^ 10 Dec 97 160 F CR358c2^ 10 Jan 01 190 F CR358c3^ 4 Nov 06 160 F CR360^ 5 Mar 07 280 F CR360c1^ 4 Jan 99 200 M CR360c2^ 5 Mar 07 180 M CR360c3 7 Nov 01 230 F 01 311 01 CR363^ 6 Mar 07 320 F CR363c1^ 16 Jan 97 180 F CR363c2^ 8 Feb 00 180 F CR363c3^ 6 Nov 01 210 M CR366 22 Nov 06 300 F GS547 CR366c1 6 Feb 98 170 M Calf of CR366, GS140 CR366c2 7 Feb 07 220 F Calf of CR366, Nursing, GS618
182 Table C 1. Continued Sample Name Date Size Sex Comments CR385^ 4 Nov 06 300 F CR385c1^ 07 Jan 99 200 M CR385c2^ 25 Oct 06 170 F CR390 11 Nov 07 300 F GS633 CR390c1 8 Jan 01 210 M Calf of CR390, GS278 CR401^ 5 Nov 06 300 F CR401c1^ 16 Nov 99 190 F CR413^ 11 Nov 07 320 F CR413c1^ 2 Dec 02 220 F CR431 14 Nov 06 300 F GS538 CR431c1 2 Dec 02 170 M TM685 CR431c2 18 Jan 06 200 F 07 029 18(CR431C3), Nursing, GS467 CR431c3 29 Jan 07 220 F Equals 06 018 02,(CR431C2), GS589 CR436 5 Nov 06 300 F GS529 CR436c1 29 Oct 01 210 M Calf of CR436, GS290 CR436c2 7 Nov 01 160 M Calf of CR436, GS305, TM866 CR436c3 13 Dec 05 LC F Calf of CR436, GS432 CR443^ 9 Nov 07 340 F CR443c1^ 9 Nov 07 160 M CR458^ 31 Jan 07 280 F CALF OF CR321 SAME AS CR321c1 CR458c1^ 22 Jan 05 180 F CR474^ 22 Nov 06 300 F CR474c1^ 07 Dec 00 180 M CR485*^ 16 Jan 01 230 F SAME ANIMAL, NO CALF KNOWN CR485*^ 17 Jan 08 280 F SAME ANIMAL, NO CALF KNOWN CR506^ 14 Nov 06 280 F SAME AS CR046c1 MOM IS CR046 CR506c1^ 01 Dec 00 200 M CR506c2^ 08 Mar 06 220 F CR509^ 22 Nov 06 270 F CR509c1^ 22 Nov 06 170 M NOT CALF? CR509c2 29 Nov 10 LC F 10 333 01 CR523 14 Nov 06 320 F CR523c1 5 Feb 06 180 F Equals 07 318 02, Nursing CR541 10 Dec 07 294 F CCR 07 16 CR541c1 18 Jan 06 200 M Calf of CR541, Nursing CR541c2 10 Dec 07 158 F calf of CR541 CCR 07 17, CR547 5 Nov 06 340 F Same as GS754 CR547c1 5 Nov 06 180 ? Calf of CR547 CR547c2 21 Oct 10 154 M CALF OF CR547 CCR 10 07, CR549 31 Oct 06 265 F CR549c1 14 Nov 02 170 F Calf of CR549 CR551 31 Jan 07 270 M CR551c1* 16 Dec 02 170 F Calf of CR551
183 Table C 1. Continued Sample Name Date Size Sex Comments CR554 14 Jan 07 300 F CR554c1 2 Mar 99 210 M CALF OF CR554 CR554c2 5 Feb 02 210 F CALF OF CR554 CR554c3 29 Jan 08 210 M Calf of CR554 CR555 3 Jan 07 310 F Pregnant CR555c1 5 Dec 07 SC M Calf of CR555 CR561 14 Jan 11 LA F same as GS757 CR561c1 8 Nov 10 LC M Calf of CR561 CR566 22 Nov 06 300 F CR566c1 23 Dec 09 210 F CALF OF CR566 CR567 9 Nov 07 300 F CR567c1 22 Jan 05 190 F CALF OF CR567 CR567c2 9 Nov 07 160 M CALF OF CR567 CR569 2 Feb 07 300 F CR569c1 2 Feb 07 210 F CALF OF CR569 CR577 3 Dec 09 270 F CR577c1 6 Nov 01 170 F CALF OF CR577 CR577c2 3 Dec 09 140 F CALF OF CR577 CR580 8 Dec 08 290 F CR580c1* 24 Jan 03 180 F CALF OF CR580 CR580c2* 8 Dec 08 190 F CALF OF CR580 CR LUMP 14 Jan 00 180 F Calf of Unknown, Lump CR LUMPc1 7 Nov 01 210 F Calf of Lump DUCR02129 13 Jun 12 SA F Muse with calf DUCR02129c1 17 Apr 12 SC F Calf of Muse DUCR02129 TB339 29 Jan 07 290 ? TB339c1 21 Dec 09 170 M CALF OF TB339
184 LIST OF REFERENCES Abdulrazzaq Y, Bener A, Al Gazali LI, Al Khayat A, Micallef R, Gaber T (1997) A study of possible deleterious effects of consanguinity. Clinical G enetics , 51 , 167 173. Aguilar A, Roemer G, Debenham S, Binns M, Garcelon D, Wayne RK (2004) High MHC diversity maintained by balancing selection in an otherwise genetically monomorphic mammal. Proceedings of the National Academy of Sciences of the United States of America , 101 , 3490 3494. Allendorf FW, Luikart GH, Aitken SN (2012) Conservation and the G enetics of P opulations . John Wiley & Sons. Hoboken, New Jersey. Amos W, Wilmer JW, Fullard K, Burg T, Croxall J, Bloch D, Coulson T (2001) The influence of parental relatedness on reproductive success. Proceedings of the Royal Society of London Series B: Biological Sciences , 268 , 2021 2027. Araki H, Cooper B, Blouin MS (2007) Genetic effects of captive breeding cause a rapid, cumulative fitness decline in the wild. Science , 318 , 10 0 103. Araki H, Cooper B, Blouin MS (2009) Carry over effect of captive breeding reduces reproductive fitness of wild born descendants in the wild. Biology L etters , 5 , 621 624. Asa C, Miller P, Agnew M, Rebolledo J, Lindsey S, Callahan M, Bauman K (2007) R elationship of inbreeding with sperm quality and reproductive success in Mexican gray wolves. Animal Conservation , 10 , 326 331. Asa CS, Valdespino C (1998) Canid reproductive biology: an integration of proximate mechanisms and ultimate causes. American Zoo logist , 38 , 251 259. Assis M, Best RC, Barros R, Yassuda YY (1988) Cytogenetic study of Trichechus inunguis (Amazonian manatee). Revista Brasileira de GenÃ©tica , 11 , 41 50. Ballou J (1983) Calculating inbreeding coefficients from pedigrees. Genetics and Con servation: A Reference for Managing Wild Animal and Plant Populations Menlo Park, California: Benjamin/Cummings, 509 520. Ballou JD, Lees C, Faust LJ, Long S, Lynch C, Bingaman Lackey L, Foose TJ (2010) Demographic and genetic management of captive populations. Wild Mammals in Captivity: Principles and Techniques for Zoo M anagement , pp. 219 246 . Balloux F, Amos W, Coulson T ( 2004) Does heterozygosity estimate inbreeding in real populations? Molecular Ecology , 13 , 3021 3031. Beck CA, Reid JP (1995) An automated photo identification catalog for studies of the life history of the Florida manatee. In: O'Shea TJ, Ackerman BB, Perci val F (eds)
185 In: Population Biology of the Florida M anatee . National Biological Service Informat ion and Technology Report 1. Washington, DC pp. 56 62. Beebee TJC, Rowe G (2008) An Introduction to Molecular E cology . Oxford University Press. Oxford, United Ki ngdom. Bener A, Hussain R, Teebi AS (2007) Consanguineous marriages and their effects on common adult diseases: studies from an endogamous population. Medical Principles and Practice , 16 , 262 267. Bercovitch FB, Widdig A, Berard JD, NÃ¼rnberg P, Kessler MJ, Schmidtke J, Trefilov A, Krawczak M (2002) Multiple sirehood in free ranging twin rhesus macaques (Macaca mulatta). American Journal of P rimatology , 57 , 31 34. Berghan J, Visser I (2000) Vertebral column malformations in New Zealand delphinids with a revi ew of cases worldwide. Aquatic Mammals , 26 , 17 25. Bilde T, Maklakov AA, Meisner K, la Guardia L, Friberg U (2009) Sex differences in the genetic architecture of lifespan in a seed beetle: extreme inbreeding extends male lifespan. BMC Evolutionary Biology , 9 , 33. Blair Irvine A (1983) Manatee metabolism and its influence on distribution in Florida. Biological Conservation , 25 , 315 334. Bonde RK (2009) Population genetics and conservation of the florida manatee: past, present, and f uture . PhD 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. Boness DJ, Craig MP, Honigman L, Austin S (1998) Fostering behavior and the effect of female density in Hawaiian monk seals, Monachus schauinslandi . Journal of Mammalogy , 79 , 1060 1069. Bonin CA, Goebel ME, O'Corry Crowe GM, Burton RS (2012) Twins or not? Genetic analysis of putative twins in Antarctic fur seals, Arctocephalus gazella , on the South Shetland Islands. Journal of Experimental Marine Biology and Ecology , 412 , 13 19. Bossart GD, Baden DG, Ewing RY, Roberts B, Wright SD (1998) Brevetoxicosis in manatees ( Trichech us manatus latirostris ) from the 1996 epizootic: gross, histologic, and immunohistochemical features. Toxicologic Pathology , 26 , 276 282. Boyd IL, Lockyer C, Marsh H (1999) Reproduction in Marine M ammals. In: Reynolds III J, Twiss Jr J (eds) Biology of Marine Mammals . Sm ithsonian Institution Press. Washington, DC. pp. 218 286.
186 Bradley JL, Wright SD, McGuire PM (1993) The Florida manatee: c ytochrome b DNA sequence. Marine Mammal Science , 9 , 197 202. Broderick D, Ovenden J, Slade R, Lanyon JM (2007) Charac terization of 26 new microsatellite loci in the dugong ( Dugong dugon ). Molecular Ecology Notes , 7 , 1275 1277. Calboli FC, Sampson J, Fretwell N, Balding DJ (2008) Population structure and inbreeding from pedigree analysis of purebred dogs. Genetics , 179 , 5 93 601. Cantanhede A, Da Silva V, Ferreira M, Farias IP, Hrbek T, Lazzarini S, Alves Gomes J (2005) Phylogeography and population genetics of the endangered Amazonian manatee, Trichechus inunguis Natterer, 1883 (Mammalia, Sirenia). Molecular Ecology , 14 , 4 01 413. Carney SL, Bolen EE, Barton SL, Scolardi KM, Englund CC, Tringali MD, Reynolds JE (2007) A minimally invasive method of field sampling for genetic analyses of the Florida manatee ( Trichechus manatus latirostris ). Marine Mammal Science , 23 , 967 975. Carsten J (2000) Cultures of R elatedness: New A pproaches to the S tudy of K inship . Cambridge University Press. Cambridge, United Kingdom. Cerchio S, Jacobsen JK, Cholewiak DM, Falcone EA, Merriwether DA (2005) Paternity in humpback whales, Megaptera novaeangliae : assessing polygyny and skew in male reproductive success. Animal B ehaviour , 70 , 267 277. Cercueil A, Bellemain E, Manel S (2002) PARENTE: computer program for parentage analysis. Journal of Heredity , 93 , 458 459. Chakraborty R, Jin L (2014) D etermination of r elatedness between i ndividuals u sing DNA f ingerprinting. Human Biology , 65 , 6: 2. Chapman T, Crespi B, Kranz B, Schwarz M (2000) High relatedness and inbreeding at the origin of eusociality in gall inducing thrips. Proceedings of the Nation al Academy of Sciences , 97 , 1648 1650. Charlesworth D, Willis JH (2009) The genetics of inbreeding depression. Nature Reviews Genetics , 10 , 783 796. Clapham PJ, PalsbÃ¸ll PJ (1997) Molecular analysis of paternity shows promiscuous mating in female humpback whales ( Megaptera novaeangliae , Borowski). Proceedings of the Royal Society of London Series B: Biological Sciences , 264 , 95 98. Clifton K, Yan J, Mecholsky J, Reep R (2008) Material properties o f manatee rib bone. Journal of Zoology , 274 , 150 159.
187 Clutton Brock T, Brotherton P, O'Riain M, Griffin A, Gaynor D, Sharpe L, Kansky R, Manser M, McIlrath G (2000) Individual contributions to babysitting in a cooperative mongoose, Suricata suricatta . Proceedings of the Royal Society of London Series B: Biolo gical Sciences , 267 , 301 305. ConstÃ¢ncia M, Kelsey G, Reik W (2004) Resourceful imprinting. Nature , 432 , 53 57. Cooper LN, Dawson SD (2009) The trouble with flippers: a report on the prevalence of digital anomalies in Cetacea. Zoological Journal of the Lin nean Society , 155 , 722 735. Coulson T, Albon S, Slate J, Pemberton J (1999) Microsatellite loci reveal sex dependent responses to inbreeding and outbreeding in red deer calves. Evolution , 53 , 1951 1960. Cristescu R, Cahill V, Sherwin WB, Handasyde K, Carly on K, Whisson D, Herbert CA, Carlsson BLJ, Wilton AN, Cooper DW (2009) Inbreeding and testicular abnormalities in a bottlenecked population of koalas ( Phascolarctos cinereus ). Wildlife R esearch , 36 , 299 308. Crnokrak P, Roff DA (1999) Inbreeding depression in the wild. Heredity , 83 , 260 270. CsillÃ©ry K, Johnson T, Beraldi D, Clutton Brock T, Coltman D, Hansson B, Spong G, Pemberton JM (2006) Performance of marker based relatedness estimators in natural populat ions of outbred vertebrates. Genetics , 173 , 2091 2101. Danforth BN (2002) Evolution of sociality in a primitively eusocial lineage of bees. Proceedings of the National Academy of Sciences , 99 , 286 290. Davey JW, Blaxter ML (2010) RADSeq: next generation po pulation genetics. Briefings in Functional Genomics , 9 , 416 423. de Castro ES, Moraes AS, Del L, Nassif S, Mino CI (2013) Conspecific brood parasitism in the w hite faced ibis Plegadis chihi (Aves: Pelecaniformes) revealed by microsatellites' based k inship r econstruction. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology , 319 , 277 284. De Lorenzis G, Imazio S, Biagini B, Failla O, Scienza A (2013) Pedigree reconstruction of the Italian grapevine Aglianico ( Vitis vinifera L.) from Campania. Molecular B iotechnology , 54 , 634 642. de Mollerat XJ, Gurrieri F, Morgan CT, Sangiorgi E, Everman DB, Gaspari P, Amiel J, Bamshad MJ, Lyle R, Blouin J L (2003) A genomic rearrangement resulting in a tandem duplication is associated with split hand split foot malformation 3 (SHFM3) at 10q24. Human Molecular G enetics , 12 , 1959 1971. Derry ME (2003) Bred for P erfection: Shorthorn C attle, Collies, and Arabian H orses S ince 1800 . JHU Press. Baltimore, Maryland.
188 Deutsch CJ, Barlas ME (2011) Ma natee response to the conversion of the F PL Cape Canaveral power plant: m ovements, warm water habitat use, and thermal regime of satellite tagged manatees during winter 2010 2011 . FWC/FWRI. St. Petersburg, Florida. Deutsch CJ, Reid JP, Bonde RK, Easton DE, Kochman HI, O'Shea TJ (2003) Seasonal movements, migratory behavior, and site fidelity of West Indian manatees along the Atlantic coast of the United States. The Wildlife Society , Wildlife Monographs , 151 , 1 77. DeWoody JA, Schupp J, Kenefic L, Busch J, M urfitt L, Keim P (2004) It ROX! BioTechniques , 37 , 348 352. Dickerson GE (1973) Inbreeding and heterosis in animals. Journal of Animal Science , 1973 , 54 77. Dobson FS, Jones WT (1985) Multiple causes of dispersal. American naturalist , 126 , 855 858. Domning DP, Hayek LAC (1986) Interspecific and intraspecific morphological variation in manatees (Sirenia: Trichechus ). Marine Mammal Science , 2 , 87 144. Drolsom P, Nielsen E (1969) Use of self fertility in the improvement of Bromus inermis and Phleum prat ense . Crop Science , 9 , 710 713. Duijf PH, van Bokhoven H, Brunner HG (2003) Pathogenesis of split hand/split foot malformation. Human Molecular G enetics , 12 , R51 R60. Dyer AB, Gottlieb G (1990) Auditory basis of maternal attachment in ducklings ( Anas platy rhynchos ) under simulated naturalistic imprinting conditions. Journal of Comparative Psychology , 104 , 190 194 . Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of inbreeding and outbreeding for conservation and management . Molecular Ecology , 16 , 463 475. Eizaguirre C, Yeates SE, Lenz TL, Kalbe M, Milinski M (2009) MHC based mate choice combines good genes and maintenance of MHC polymorphism. Molecular Ecology , 18 , 3316 3329. Eldridge MD, King JM, Loupis AK, Spencer P, Taylor AC, Pope LC , Hall GP (1999) Unprecedented low levels of genetic variation and inbreeding depression in an island population of the b lack footed r ock w allaby. Conservation Biology , 13 , 531 541.
189 Emery A, Wilson I, Craig S, Boyle P, Noble L (2001) Assignment of paternity groups without access to parental genotypes: multiple mating and developmental plasticity in squid. Molecular Ecology , 10 , 1265 1278. Excoffier L, Lischer HE (2010) Arlequin suite ver 3.5: a new series of programs to perform population geneti cs analyses under Linux and Windows. Molecular Ecology Resources , 10 , 564 567. Fagan WF, Holmes E (2006) Quantifying the extinction vortex. Ecology Letters , 9 , 51 60. Farias IP, OrtÃ G, Sampaio I, Schneider H, Meyer A (2001) The cytochrome b gene as a phylogenetic marker: the limits of resolution for analyzing relationships among cichlid fishes. Journal of Molecular Evolution , 53 , 89 103. Fenster CB, Galloway LF (2000) Inbreeding and outbreeding depression in natural populations of Chamaecrista fasciculata (Fabaceae). Conservation Biology , 14 , 1406 1412. FernÃ¡ndez J, Villanueva B, Pong Wong R, Toro MÃ (2005) Efficiency of the use of pedigree and molecular marker information in conservation programs. Genetics , 170 , 1313 1321. Ferreira PB, Torres RA, Garcia JE (2011) Single nucleotide polymorphisms from cyto chrome b gene as a useful protocol in forensic genetics against the illegal hunting of manatees: Trichechus manatus , Trichechus inunguis , Trichechus senegalensis , and Dugong dugon (Eutheria: Sirenia). Zoologia (Curitiba), 28 , 133 138. Fitzpatrick LJ, Gaspa rini C, Fitzpatrick JL, Evans JP (2014) Male female relatedness and patterns of male reproductive investment in guppies . Biology letters , 10 , 5: 20140166. Foose TJ, Ballou JD (1988) Population management: theory and practice. International Zoo Yearbook , 27 , 26 41. Frankham R, Briscoe DA, Ballou JD (2002) Introduction to C onservation G enetics . Cambridge University Press. Cambridge, United Kingdom. Frasier T, Hamilton P, Brown M, Conger L, Knowlton A, Marx M, Slay C, Kraus S, White B (2007) Patterns of male reproductive success in a highly promiscuous whale species: the endangered North Atlantic right whale. Molecular Ecology , 16 , 5277 5293. FWC (197 8) The Florida Manatee Sanctuary Act . Tallahassee, Florida. FWC (2007) Florida Manatee Management Plan . Tallahassee, Florida.
190 FWRI (2010) Manatee Synoptic Surveys . http://myfwc.com/research/manatee/projects/population monitoring/synoptic surveys/ . Accessed January 15 2014 . FWRI (2014) Manatee Mortality Statistics . http://www.myfwc.com/research/manatee/rescue mortality response/mortality statistics/ . Accessed May 28 2014. FWS (1973) Endangered Species Act of 1973 . Library of Congress, Washington DC. FWS (20 01) Florida Manatee Recovery P lan ( Trichechus manatus latirostris ), Third revision. US Fish and Wildlife Service Atlanta, Georgia, 144. FWS (2012) Captive Manatee Database . https:// www.fws.gov//manatee . Accessed 8/30/2013 . FWS (2013) Florida Manatee Recovery Facts . Fish and Wildlife Service. http://www.fws.gov/northflorida/Manatee/manatee gen facts.htm . Accessed January 15 2014 . Gandon S (19 99) Kin competition, the cost of inbreeding and the evolution of dispersal. Journal of Theoretical Biology , 200 , 345 364. Garcia Rodriguez A (2000) Genetic structure in the West Indian manatee (Trichechus manatus) . Ph D Dissertation, Universit y of Florida. Garcia Rodriguez A, Bowen B, Domning D, Mignucci Giannoni A, Marmontel M, Montoya Ospina R, Morales Vela B, Rudin M, Bonde R, McGuire P (1998) Phylogeography of the West Indian manatee ( Trichechus manatus ): how many populations and how many taxa? Molecular Ecology , 7 , 1137 1149. GarcÃa RodrÃguez AI, Moraga Amador D, Farmerie W, Mcguire P, King TL (2000) Isolation and characterization of microsatellite DNA markers in the Florida manatee ( Trichechus manatus latirostris ) and their application in selected Siren ian species. Molecular Ecology , 9 , 2161 2163. Garner A, Rachlow JL, Hicks JF (2005) Patterns of genetic diversity and its loss in mammalian populations. Conservation Biology , 19 , 1215 1221. Gemmell NJ, Burg TM, Boyd IL, Amos W (2001) Low reproductive success in territorial male Antarctic fur seals ( Arctocephalus gazella ) suggests the existence of alternative mating strategies. Molecular Ecology , 10 , 451 460. Gilpin ME, Soule ME (1986) Minimum viable populations: processes of species extinction. In: ME Soule, (ed) Conservation Biology . The science of scarcity and diversity . Sinauer Associates, Sutherland, Massachusetts. pp 19 34
191 Ginzinger DG (2002) Gene quantification using real time quantitative PCR: an emerging technology hits the mainstream. Experimen tal H ematology , 30 , 503 512. Gray BA, Zori RT, Mcguire PM, Bonde RK (2002) A first generation cytogenetic ideogram for the Florida manatee ( Trichechus manatus latirostris ) based on multiple chromosome banding techniques. Hereditas , 137 , 215 223. Greenwood PJ (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal B ehaviour , 28 , 1140 1162. Griffiths RA, Pavajeau L (2008) Captive breeding, reintroduction, and the conservation of amphibians. Conservation Biology , 22 , 852 861. Grueber CE, Laws RJ, Nakagawa S, Jamieson IG (2010) Inbreeding Depression Accumulation across Life History Stages of the Endangered Takahe. Conservation Biology , 24 , 1617 1625. Grueber CE, Wallis GP, Jamieson IG (2008) Heterozygosity fitness correlations and their relevance to studies on inbreeding depression in threatened species. Molecular Ecology , 17 , 3978 3984. Grueber CE, Waters JM, Jamieson IG (2011) The imprecision of hete rozygosity fitness correlations hinders the detection of inbreeding and inbreeding depression in a threatened species. Molecular Ecology , 20 , 67 79. GutiÃ©rrez JP, Marmi J, Goyache F, Jordana J (2005) Pedigree information reveals moderate to high levels of inbreeding and a weak population structure in the endangered Catalonian donkey breed. Journal of Animal Breeding and G enetics , 122 , 378 386. Hadfield J, Richardson D, Burke T (2006) Towards unbiased parentage assignment: combining genetic, behavioural and spatial data in a Bayesian framework. Molecular Ecology , 15 , 3715 3730. Hardy OJ, Charbonnel N, FrÃ©ville H, Heuertz M (2003) Microsatellite allele sizes: a simple test to assess their significance on genetic differentiation. Genetics , 163 , 1467 1482. Harle y EH, Baumgarten I, Cunningham J, O'Ryan C (2005) Genetic variation and population structure in remnant populations of black rhinoceros, Diceros bicornis , in Africa. Molecular Ecology , 14 , 2981 2990. Hedrick PW, Fredrickson R (2010) Genetic rescue guidelines with examples from Mexican wolves and Florida panthers. Conservation Genetics , 11 , 615 626. Hedrick PW, Kalinowski ST (2000) Inbreeding depression in conservation biology. Annual Review of Ecology and Systematics , 139 162.
192 Henkel JR, Jones KL, Hereford SG, Savoie ML, Leibo S, Howard JJ (2012) Integrating microsatellite and pedigree analyses to facilitate the captive management of the endangered Mississippi sandhill crane ( Grus canadensis pulla ). Zoo B iology , 31 , 322 335. Hernandez P, Reynolds III JE, Marsh H, Marmontel M (1995) Age and seasonality in spermatogenesis of Florida manatees. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: Population Biology of the Florida M anatee . National Biological Ser vice Informat ion and Technology Report 1. Washington, DC pp. 84 95. Hildebrand C, Torney D, Wagner R (1992) Informativeness of polymorphic DNA markers. Los Alamos Science , 20 , 100 102. Hollis KL, ten Cate C, Bateson P (1991) Stimulus representation: a subp rocess of imprinting and conditioning. Journal of Comparative Psychology , 105 , 307 317 . Hughes WO, Oldroyd BP, Beekman M, Ratnieks FL (2008) Ancestral monogamy shows kin selection is key to the evolution of eusociality. Science , 320 , 1213 1216. Hunter ME, Auil GÃ³mez NE, Tucker KP, Bonde RK, Powell J, McGuire PM (2010a) Low genetic variation and evidence of limited dispersal in the regionally important Belize manatee. Animal Conservation , 13 , 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 primers. Molecular Ecolog y 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. Conservation Genetics , 13 , 1623 1635. IUCN (2012) 2012 Red List of I UCN Threatened Species. http://www.iucnredlist.org . May 28 2012. Jacobs S (2010) A molecular toolbox for the establishment of a genetically based studbook for captive bred m anatees ( Trichechus manatus manatus ). M S Thesis, Technische Universitat Munchen . Johnson WE, Onorato DP, Roelke ME, Land ED, Cunningham M, Belden RC, McBride R, Jansen D, Lotz M, Shindle D (2010) Genetic restoration of the Florida panther. Science , 329 , 1641 1645. Johnstone RA (2000) Models of reproductive skew: a review and synthesis (invited article). Ethology , 106 , 5 26.
193 Jones AG (2001) GERUD 1. 0: a computer program for the reconstruction of parental genotypes from progeny arrays using multilocus DNA data. Molecu lar Ecology Notes , 1 , 215 218. Jones AG, Ardren WR (2003) Methods of parentage analysis in natural populations. Molecular Ecology , 12 , 2511 2523. Jones AG, Small CM, Paczolt KA, Ratterman NL (2010) A practical guide to methods of parentage analysis. Molecu lar Ecology Resources , 10 , 6 30. Jones O, Wang J (2010a) Molecular marker based pedigrees for animal conservation biologists. Animal Conservation , 13 , 26 34. Jones OR, Wang J (2010 b ) COLONY: a program for parentage and sibship inference from multilocus gen otype data. Molecular Ecology Resources , 10 , 551 555. Jorde LB (2001) Consanguinity and prereproductive mortality in the Utah Mormon population. Human H eredity , 52 , 61 65. Jorde LB, Bamshad M, Rogers AR (1998) Using mitochondrial and nuclear DNA markers to reconstruct human evolution. Bioessays , 20 , 126 136. Kalinowski ST, Hedrick PW, Miller PS (2000) Inbreeding depression in the Speke's gazelle captive breeding program. Conservation Biology , 14 , 1375 1384. Kalinowski ST, Taper ML, Marshall TC (2007) Revising how the computer program CERVUS accommodates genotyping error increases success in paternity assignment. Molecular Ecology , 16 , 1099 1106. Kalinowski ST, Wagner AP, Taper ML (2006) ml relate: a computer program for maximum likelihood estimation of relatedness and relationship. Molecular Ecology Notes , 6 , 576 579. Keller LF, Waller DM (2002) Inbreeding effects in wild populations. Trends in Ecology & Evolution , 17 , 230 241. Kellogg ME (2 008) Sirenian conservation genetics and Florida manatee (Trichechus manatus latirostris) cytogenetics . PhD Dissertation, University of Florida . 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 E volutionary B iology , 7 , 6. Kendall WL, Langtimm CA, Beck CA, Runge MC (2004) Capture recapture analysis for estimating manatee reproductive rates. Marine Mammal Science , 20 , 424 437. Kingman JF (1982) On the genealogy of large populations. Journal of Applied Probability , 19 , 27 43.
194 Kleiman DG (1977) Monogamy in mammals. Quarterly Review of Biology , 52 , 39 69. Klopfer PH, Adams DK, Klopfer MS (1964) Maternal "Imprinting" in Goats. Proceedings of the National Academy of Sciences of the United States of America , 52 , 911. Koch M, Hadfield JD, Sefc KM, Sturmbauer C (2008) Pedigree reconstruction in wild cichlid fish populations. Molecular Ecology , 17 , 4500 4511. Kokko H, Johnstone RA, Wright J (2002) The evolution of parental and alloparental effort in cooperatively breeding groups: when should helpers pay to stay? Behavioral Ecology , 13 , 291 300. Kruuk L, Hill W (2008) Introduction. Evolutionary dynamics of wild populations: the use of long term pedigree data. Proceedings of the Royal Society B: Biological Sciences , 275 , 593 596. Lacy RC (2009) Stopping evolution: genetic management of captive populations. Conservation Genetics in the Age of G enomics , 1st edn Columbia University Press. New York, New York. Lacy RC (2012) Extending pedigree analysis for uncertain parentage and diverse breeding systems. Journal of Heredity , 103 , 197 205. Lacy RC, Ballou JD, Pollak JP (2012) PMx: software package for demographic and genetic analysis and management of pedigreed populations. Methods in Ecology and Evolution , 3 , 433 437. Lacy RC, Ballou JD, Princee F, Starfield A, Thompson EA (1995) Pedigree analysis for population management. Popula tion Management for Survival and R ecovery , Columbia university Press, New York, New York. pp 57 75. Langlois B (1996) A consideration of the genetic aspects of some current practices in Thoroughbred horse breeding. Annales de Zootechnie , 45 , 41 51. Langtimm C, O'shea T, Pradel R, Beck C (1998) Estimates of annual survival probabiliti es for adult Florida manatees ( Trichechus manatus latirostris ). Ecology , 79 , 981 997. Langtimm CA, Beck CA, Edwards HH, Fick Child KJ, Ackerman BB, Barton SL, Hartley WC (2004) Survival e stimates f or Florida m anatees f rom t he p hoto i dentification o f i ndivi duals. Marine Mammal Science , 20 , 438 463. Lanyon JM, Sneath HL, Ovenden JR, Broderick D, Bonde RK (2009) Sexing s irenians: v alidation of v isual and m olecular s ex d etermination in both w ild d ugongs ( Dugong dugon ) and Florida m anatees ( Trichechus manatus latirostris ). Aquatic Mammals , 35 , 187 192.
195 Larkin IL, Gross TS, Reep RL (2005) Use of faecal testosterone concentrations to monitor male Florida manatee ( Trichechus manatus latirostris ) reproductive status. Aquatic Mammals , 31 , 52 61. Larkin ILV (2000) Re productive endocrinology of the Florida manatee (Trichechus manatus latirostris): estrous cycles, seasonal patterns and behavior . PhD Dissertation, University of Florida . Lawson Handley L, Perrin N (2007) Advances in our understanding of mammalian sex bias ed dispersal. Molecular Ecology , 16 , 1559 1578. Leberg PL, Firmin BD (2008) Role of inbreeding depression and purging in captive breeding and restoration programmes. Molecular Ecology , 17 , 334 343. Lefebvre L, O'Shea T (1995) Florida manatees. Our living resources -coastal and marine ecosystems. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: Population Biology of the Florida M anatee . National Biological Service Informat ion and Technology Report 1. Washington, DC pp. 267 269. Lehmann L, Per rin N (2003) Inbreeding avoidance through kin recognition: choosy females boost male dispersal. The American Naturalist , 162 , 638 652. LÃ©vi Strauss C (1969) The Elementary S tructures of K inship. Beacon Press. Boston, Massachusetts. Li C, Weeks D, Chakravar ti A (1993) Similarity of DNA fingerprints due to chance and relatedness. Human H eredity , 43 , 45 52. Liu D, Ma Y, Li H, Xu Y, Zhang Y, Dahmer T, Bai S, Wang J (2013) Simultaneous polyandry and heteropaternity in tiger ( Panthera tigris altaica ): Implications for conservation of genetic diversity in captive populations of felids. Chinese Science Bulletin , 58 , 2230 2236. Loughman WD, Frye FL, Herald ES (1970) The chromosomes of a male manatee. International Zoo Yearbook , 10 , 151 152. Luna FO, Bonde RK, Attademo FL, 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: Marine and Freshwater Ecosystems , 22 , 6 65 672. Lynch M (1988) Estimation of relatedness by DNA fingerprinting. Molecular Biology and Evolution , 5 , 584 599. Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding depression. Evolution , 45 , 622 629.
196 Lynch M, O'Hely M (2001) Captive breeding and the genetic fitness of natural populations. Conservation Genetics , 2 , 363 378. Lynch M, Ritland K (1999) Estimation of pairwise relatedness with molecular markers. Genetics , 152 , 1753 1766. Lynch M, Walsh B (19 98) Genetics and Analysis of Quantitative Traits . Sinauer Associates, Sunderland. Massachusetts. MacArthur RH, Wilson EO (1967) The Theory of Island Biogeography . Princeton University Press. Princeton, New Jersey Maniscalco JM, Harris KR, Atkinson S, Parke r P (2007) Alloparenting in Steller sea lions ( Eumetopias jubatus ): correlations with misdirected care and other observations. Journal of Ethology , 25 , 125 131. Marsden C, Verberkmoes H, Thomas R, Wayne R, Mable B (2013) Pedigrees, MHC and microsatellites: an integrated approach for genetic management of captive African wild dogs (Lycaon pictus). Conservation Genetics , 14 , 171 183. Mayor LR, Balding DJ (2006) Discrimination of half siblings when maternal genotypes are known. Forensic Science I nternational , 159 , 141 147. McClenaghan Jr LR, O'Shea TJ (1988) Genetic variability in the Florida manatee ( Trichechus manatus ). Journal of Mammalogy , 481 488. McComb K, Moss C, Durant SM, Baker L, Sayialel S (2001) Matriarchs as repositories of social knowledge in Afri can elephants. Science , 292 , 491 494. Mc Parland S, Kearney J, Rath M, Berry D (2007) Inbreeding trends and pedigree analysis of Irish dairy and beef cattle populations. Journal of Animal Science , 85 , 322 331. McPhee ME (2004) Generations in captivity increases behavioral variance: considerations for captive breeding and reintroduction programs. Biological Conservation , 115 , 71 77. Milligan BG (2003) Maximum likelihood estimation of relatedness. Genetics , 163 , 1 153 1167. MÃ¶ller LM, Beheregaray LB (2004) Genetic evidence for sex biased dispersal in resident bottlenose dolphins ( Tursiops aduncus ). Molecular Ecology , 13 , 1607 1612. MÃ¶ller LM, Beheregaray LB, Allen SJ, Harcourt RG (2006) Association patterns and kins hip in female Indo Pacific bottlenose dolphins ( Tursiops aduncus ) of southeastern Australia. Behavioral Ecology and Sociobiology , 61 , 109 117.
197 MÃ¶ller LM, Beheregaray LB, Harcourt RG, KrÃ¼tzen M (2001) Alliance membership and kinship in wild male bottlenose dolphins ( Tursiops aduncus ) of southeastern Australia. Proceedings of the Royal Society of London Series B: Biological Sciences , 268 , 1941 1947. Mucha S, Windig J (2009) Effects of incomplete pedigree on genetic management of the Dutch Landrace goat. Journ al of A nimal Breeding and G enetics , 126 , 250 256. Neff BD, Repka J, Gross MR (2001) A Bayesian framework for parentage analysis: the value of genetic and other biological data. Theoretical Population Biology , 59 , 315 331. Newman DL, Abney M, McPeek MS, Ober C, Cox NJ (2001) The importance of genealogy in determining genetic associations with complex traits. American Journal of H uman G enetics , 69 , 1146. NOAA (1972) Marine Mammal Protection Act . Silver Spring. Maryland. NOAA (2013). North Atlantic Right Whales . http://www.nmfs.noaa.gov/pr/species/mammals/cetaceans/rightwhale_northatlant ic.htm . Accessed July 29 2013 . NOAA (2014) Water Temperature Table of All Coastal Regions . http://www.nodc.noaa.gov/dsdt/cwtg/all_meanT.html . Accessed May 30 2014. Nourisson C (2011) Genetic structure of the m anatees in Mexico . PhD Dissertation, El Colegio de la Frontera Sur . Nourisson C, Morales Vela B, Padilla SaldÃvar J, Tucker KP, Clark A, Olivera GÃ³mez LD, Bonde R, McGuire P (2011) Evidence of two genetic clusters of manatees with low genetic diversity in Mexico and implications for their conservation. Genetica , 139 , 833 842. Nyakaana S, Arctander P (1999) Population genetic structure of the African elephant in Uganda based on variation at m itochondrial and nuclear loci: evidence for male biased gene flow. Molecular Ecology , 8 , 1105 1115. O'Brien SJ, Evermann JF (1988) Interactive influence of infectious disease and genetic diversity in natural populations. Trends in Ecology & Evolution , 3 , 254 259. Ober C, Hyslop T, Hauck WW (1999) Inbreeding effects on fertility in humans: evidence for reproductive compensation. The American Journal of Human Genetics , 64 , 225 231. Oliver MK, Piertney SB (2012) Selection maintains MHC diversity through a nat ural population bottleneck. Molecular Biology and Evolution , 29 , 1713 1720.
198 Ortega Ortiz JG, Engelhaupt D, Winsor M, Mate BR, Rus Hoelzel A (2012) Kinship of long term associates in the highly social sperm whale. Molecular Ecology , 21 , 732 744. Ortega Orti z JG, Villa RamÃrez B, Gersenowies JR (2000) Polydactyly and other features of the manus of the vaquita, Phocoena sinus . Marine Mammal Science , 16 , 277 286. levels of inbreedi ng depression strongly affect extinction risk in wild populations. Biological Conservation , 133 , 42 51. O'Shea TJ, Beck CA, Bonde RK, Kochman HI, Odell DK (1985) An analysis of manatee mortality patterns in Florida, 1976 81. The Journal of Wildlife M anagem ent , 49 , 1 11. O'Shea TJ, Hartley WC (1995) Reproduction and early age survival of the manatees at Blue Spring, Upper St.Johns River, Florida. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: Population Biology of the Florida M anatee . National Biological Service Informat ion and Technology Report 1. Washington, DC pp. 157 176. Pallone SR, Bergus GR (2009) Fertility Awareness Based Methods: Another Option for Family Planning. The Journal of the American Board of Family Medicine , 22 , 147 157. Parker PG, Snow AA, Schug MD, Booton GC, Fuerst PA (1998) What molecules can tell us about populations: choosing andusing a molecular marker. Ecology , 79 , 361 382. Pardini A, O'Brien P, Fu B, Bonde R, Elder F, Ferguson Smith M, Yang F, Robinson T (2007) Chromosome paintin g among Proboscidea, Hyracoidea and Sirenia: Support for paenungulata (Afrotheria, Mammalia) but not tethytheria. Proceedings of the Royal Society B: Biological Sciences , 274 , 1333 1340. Parsons T (1943) The kinship system of the contemporary United States . American A nthropologist , 45 , 22 38. Pattison EM, DeFrancisco D, Wood P, Frazier H, Crowder J (1975) A psychosocial kinship model for family therapy. American Journal of Psychiatry , 132 , 1246 1251. Pause KC (2007) Conservation genetics of the Florida mana tee, Trichechus manatus latirostris . PhD Dissertation, University of Florida . Pause KC, Nourisson C, Clark A, Kellogg ME, Bonde RK, McGuire PM (2007) Polymorphic microsatellite DNA markers for the Florida manatee ( Trichechus manatus latirostris ). Molecular Ecology Notes , 7 , 1073 1076.
199 Peakall R, Smouse PE (2006) GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes , 6 , 288 295. Pemberton J (2008) Wild pedigrees: the way forward. Proceedings of t he Royal Society B: Biological Sciences , 275 , 613 621. Perrin N, Mazalov V (1999) Dispersal and inbreeding avoidance. The American Naturalist , 154 , 282 292. Perry EA, Boness DJ, Fleischer RC (1998) DNA fingerprinting evidence of nonfilial nursing in grey s eals. Molecular Ecology , 7 , 81 85. Port M, Kappeler PM (2010) The utility of reproductive skew models in the study of male primates, a critical evaluation. Evolutionary Anthropology: Issues, News, and Reviews , 19 , 46 56. Powell J, Rathbun G (1984) Distribu tion and abundance of manatees along the northern coast of the Gulf of Mexico. Northeast Gulf Science , 7 , 1 28. Queller DC, Goodnight KF (1989) Estimating relatedness using genetic markers. Evolution , 258 275. Radwan J (2003) Inbreeding depression in fecun dity and inbred line extinction in the bulb mite, Rhizoglyphus robini . Heredity , 90 , 371 376. Rand DM (1996) Neutrality tests of molecular markers and the connection between DNA polymorphism, demography, and conservation biology. Conservation Biology , 10, 665 671. Rathbun GB, Reid JP, Bonde RK, Powell JA (1995) Reproduction in free ranging Florida manatees. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: Population Biology of the Florida M anatee . National Biological Service Informat ion and Technology Repor t 1. Washington, DC pp. 135 156. Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity , 86 , 248 249. Reep R, Bonde RK (2006) The Florida Manatee: Biology and Conservation . Unive rsity Press of Florida. Gainesville, Florida. xvi + 189 pp. tagged and recognizable manatees on the Atlantic coast of Florida. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: Populati on Biology of the Florida M anatee . National Biological Service Informat ion and Technology Report 1. Washington, DC pp. 171 191.
200 Reynolds J E, Rommel SA (1999) Biology of Marine Mammals . Smithsonian Institution Press Washington, DC. 578 pp. Reynolds J, Rommel SA, Pitchford ME (2004) The Likelihood of Sperm Competition in Manatees -Explaining an Apparent Paradox. Marine Mammal Science , 20 , 464 476. Richard K, Dillon M, Whitehead H, Wright JM (1996) Patterns of kinship in groups of free living sperm whales (Physeter macrocephalus) revealed by multiple molecular genetic analyses. Proceedings of the National Academy of Sciences , 93 , 8792 8795. Ried T, SchrÃ¶ck E, Ning Y, Wienberg J (1998) Chromosome painting: a useful art. Human Molecular Genetics , 7 , 1619 1626. Ritland K (1990) Inferences about inbreeding depression based on changes of the inbreeding coefficient. Evolution , 1230 1241. Ritland K (1996) Estimators for pairwise relatedness and individual inbreeding coefficients. Genetical R esearch , 6 7 , 175 186. Roelke ME, Martenson JS, O'Brien SJ (1993) The consequences of demographic reduction and genetic depletion in the endangered Florida panther. Current Biology , 3 , 340 350. Rojas Bracho L, Taylor BL (1999) Rish Factors Affecting The Vaquita ( Phoc oena sinus ). Marine Mammal Science , 15 , 974 989. Roulin A (2002) Why do lactating females nurse alien offspring? A review of hypotheses and empirical evidence. Animal B ehaviour , 63 , 201 208. Rudin N (1997) Dictionary of Modern Biology . Barron's Educational Series, Hauppauge. New York. Runge MC, Sanders Reed CA, Fonnesbeck CJ (2007a) A core stochastic projection model for Florida manatees ( Trichechus manatus latirostris ). U.S. Geological Survey Open File Report 2007 1082. pp. 41 Runge MC, Sanders Reed CA, Langtimm CA, Fonnesbeck CJ (2007b) A quantitative threats analysis for the Florida manatee ( Trichechus manatus latirostris ). U.S. Geological Survey Open File Report 2007 1086. pp. 34.
201 Ruiz Lopez MJ, Roldan ER, Espeso G, Gomendio M (2009) P edigrees and microsatellites among endangered ungulates: what do they tell us? Molecular Ecology, 18, 1352 1364. Saito Y, Sahara K, Mori K (2000) Inbreeding depression by recessive deleterious genes affecting female fecundity of a haplo diploid mite. Journ al of Evolutionary Biology , 13 , 668 678. Sardell RJ, Keller LF, Arcese P, Bucher T, Reid JM (2010) Comprehensive paternity assignment: genotype, spatial location and social status in song sparrows, Melospiza melodia . Molecular Ecology , 19 , 4352 4364. Seyou m S, Tringali MD, Higham M (2008) Development of 15 polymorphic microsatellite markers in the Atlantic tarpon (Megalops atlanticus) for capture recapture studies. Molecular Ecology Resources , 8 , 126 128. Snyder NF, Derrickson SR, Beissinger SR, Wiley JW, S mith TB, Toone WD, Miller B (1996) Limitations of captive breeding in endangered species recovery. Conservation Biology , 10 , 338 348. Stockley P (2003) Female multiple mating behaviour, early reproductive failure and litter size variation in mammals. Proceedings of the Royal Society of London Series B: Biological Sciences , 270 , 271 278. Stoll C, Alembik Y, Dott B, Feingold J (1994) Parental consanguinity as a cause of increased incidence of birth defects in a study of 131, 760 consecutive birthsxs. Ame rican Journal of Medical G enetics , 49 , 114 117. Stoltenberg C, Magnus P, Skrondal A, Lie RT (1999) Consanguinity and recurrence risk of stillbirth and infant death. American Journal of Public Health , 89 , 517 523. Tang Q, Liu H, Mayden R, Xiong B (2006) Com parison of evolutionary rates in the mitochondrial DNA cytochrome b gene and control region and their implications for phylogeny of the Cobitoidea (Teleostei: Cypriniformes). Molecular Phylogenetics and Evolution , 39 , 347 357. Taylor BL, Rojas Bracho L (19 99) Examining the risk of inbreeding depression in a naturally rare cetacean, the vaquita ( Phocoena sinus ). Marine Mammal Science , 15 , 1004 1028. Templeton AR, Read B (1984) Factors eliminating inbreeding depression in a captive herd of Speke's gazelle (Ga zella spekei). Zoo B iology , 3 , 177 199. Terasaki PI, Gjertson D, Bernoco D, Perdue S, Mickey MR, Bond J (1979) Twins with two different fathers identified by HLA. Obstetrical & Gynecological Survey , 34 , 435 436.
202 Tokarska M, Marshall T, Kowalczyk R, WÃ³jcik J, Pertoldi C, Kristensen T, Loeschcke V, Gregersen VR, Bendixen C (2009) Effectiveness of microsatellite and SNP markers for parentage and identity analysis in species with low genetic diversity: the case of European bison. Heredity , 103 , 326 332. Tringali MD (2006) A Bayesian approach for the genetic tracking of cultured and released individuals. Fisheries Research , 77 , 159 172. Tringali MD, Davis MC, RodrÃguez LÃ³pez MA, Bolen EE, Sullivan JG, Haubold EM (2008a) Simultaneous use of the X and Y chr omosome genes SMCX, SMCY, and DBY for sex determination in the Florida manatee ( Trichechus manatus latirostris ). Marine Mammal Science , 24 , 218 224. Tringali MD, Seyoum S, Careny SL, Davis MC, Rodriguez Lopez MA, Reynolds III JE, Haubold E (2008b) Eighteen new polymorphic microsatellite markers for the endangered Florida manatee, Trichechus manatus latirostris . Molecular Ecology Resources , 8 , 328 331. Trupkiewicz J, Gulland F, Lowenstine L (1997) Congenital defects in northern elephant seals stranded along the central California coast. Journal of Wildlife D iseases , 33 , 220 225. Tucker KP, Hunter ME, Bonde RK, Austin JD, Clark AM, Beck CA, McGuire PM, Ol i MK (2012) Low genetic diversity and minimal population substructure in the endangered Florida manatee: implications for conservation. Journal of Mammalogy, 93 , 1504 1511. TWS (2012) The Wildlife Society . Careers in Wildlife Conservation . http://joomla.wi ldlife.org/index.php?option=com_content&task=view&id=73 . Accessed June 21 2012 2012 . Valera M, Molina A, GutiÃ©rrez JP, GÃ³mez J, Goyache F (2005) Pedigree analysis in the Andalusian horse: population structure, genetic variability and influence of the Carth usian strain . Livestock Production Science , 95 , 57 66. Van Oosterhout C, Hutchinson WF, Wills DP, Shipley P (2004) MICRO CHECKER: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes , 4 , 535 538. Vandeputte M, Mauger S, Dupont Nivet M (2006) An evaluation of allowing for mismatches as a way to manage genotyping errors in parentage assignment by exclusion. Molecular Ecology Notes , 6 , 265 267. Vassilakos D, Natoli A, Dahlheim M, Hoelzel AR (2009) Bal ancing and directional selection at exon 2 of the MHC DQB1 locus among populations of odontocete cetaceans. Molecular Biology and Evolution , 26 , 681 689.
203 Verma R, Luke S, Dhawan P (1992) Twins with different fathers. The Lancet , 339 , 63 64. Vianna JA, Bond e RK, Caballero S, Giraldo JP, Lima RP, Clark A, Marmontel M, Morales Vela B, De Souza MJ, Parr L (2006) Phylogeography, phylogeny and hybridization in trichechid sirenians: implications for manatee conservation. Molecular Ecology , 15 , 433 447. Waldick R, Kraus S, Brown M, White B (2002) Evaluating the effects of historic bottleneck events: an assessment of microsatellite variability in the endangered, North Atlantic right whale. Molecular Ecology , 11 , 2241 2249. Wang J (2002) An estimator for pairwise rela tedness using molecular markers. Genetics , 160 , 1203 1215. Wang J (2006) Informativeness of genetic markers for pairwise relationship and relatedness inference. Theoretical Population Biology , 70 , 300 321. Wang J (2007) Triadic IBD coefficients and applica tions to estimating pairwise relatedness. Genetical R esearch , 89 , 135. Wang J (2011) coancestry: a program for simulating, estimating and analysing relatedness and inbreeding coefficients. Molecular Ecology Resources , 11 , 141 145. Wang J, Whitlock MC (2003) Estimating effective population size and migration rates from genetic samples over space and time. Genetics , 163 , 429 446. Watson A, Bonde R (1986) Congenital malformations of the flipper in three West Indian manatees, Trichechus manatus , and a proposed mechanism for development of ectrodactyly and cleft hand in mammals. Clinical Orthopaedics and Related R esearch , 202 , 294 301. Weber DS, Stewart BS, Lehman N (2004) Genetic Consequences of a Severe Population Bottleneck in the Guad alupe Fur Seal ( Arctocephalus townsendi ). Journal of Heredity , 95 , 144 153. Wehner T (1999) Heterosis in vegetable crops. In: Coors, J & Pandey S (eds.) The Genetics and Exploitation of H eterosis in Crops. American Society of Agronomy , Madison, WI, pp. 387 397. Weir BS, Anderson AD, Hepler AB (2006) Genetic relatedness analysis: modern data and new challenges. Nature Reviews Genetics , 7 , 771 780. Weir BS, Cockerham CC (1984) Estimating F statistics for the analysis of population structure. Evolution , 1358 1 370.
204 White J, Harkness D, Isaacks R, Duffield D (1976) Some studies on blood of the Florida manatee, Trichechus manatus latirostris . Comparative Biochemistry and Physiology Part A: Physiology , 55 , 413 417. Whitehead H (1996) Babysitting, dive synchrony, and indications of alloparental care in sperm whales. Behavioral Ecology and Sociobiology , 38 , 237 244. Wilberg MJ, Dreher BP (2004) Genecap: a program for analysis of multilocus genotype data for non invasiv e sampling and capture recapture population estimation. Molecular Ecology Notes , 4 , 783 785. Wilson AJ, Pemberton JM, Pilkington J, Coltman DW, Mifsud D, Clutton Brock TH, Kruuk LB (2006) Environmental coupling of selection and heritability limits evolutio n. PLoS B iology , 4, e216. Wilson R, Reynolds III J, Wetzel D, Schweirzke Wade L, Bonde R, Breuel K, Roudebush W (2011) Secretion of anti Mullerian hormone in the West Indian manatee (Trichechus manatus) with implications for assessing conservation status. Endangered Species Research , 10 , 3354. Worthy G, Miculka T, Wright S (1999) Manatee response to cold: how cold is too cold. In: Florida Manatees and Warm water: Proceedings of the Warm Water Workshop, Jupiter, Florida , pp. 1 6. Wright S (1922) Coefficients of inbreeding and relationship. American. Naturalist , 56 , 330 338. Wright SD, Ackerman BB, Bonde RK, Beck CA, Banowetz DJ (1995) Analysis of watercraft related mortality of manatees in Florida, 1979 1991. In: O'Shea TJ, Ackerman BB, Percival F (eds) In: P opulation Biology of the Florida M anatee . National Biological Service Informat ion and Technology Report 1. Washington, DC pp. 259 268. WWF (2013) Black Rhino. World Wildlife Foundation. http://www.worldwildlife.org/species/black rhino . Accessed July 19 2013 . Wyner YM, Amato G, DeSalle R (1999) Captive breeding, reintroduction, and the conservation genetics of black and white ruffed lemurs, Varecia variegata variegata. Molecular Ecology , 8 , S107 S115. Yurk H, Barrett Lennard L, Ford J, Matkin C (2002) Cultural transmission within maternal lineages: vocal clans in resident killer whales in southern Alaska. Animal B ehaviour , 63 , 1103 1119. Zeh JA, Zeh DW (1996) The evolution of polyandry I: i ntragenomic conflict and genetic incompatibility. Proceedings of the Royal Society of London Series B: Biological Sciences , 263 , 1711 1717.
205 Zeyl E, Aars J, Ehrich D, Bachmann L, Wiig Ã˜ (2009) The mating system of polar bears: a genetic approach. Canadian J ournal of Zoology , 87 , 1195 1209.
206 BIOGRAPHICAL SKETCH Michelle C. Davis was born in Fairfield, California. As the only daughter of two Air Force officers, she spent the first quarter of her life moving every two to three years living in New York, Germany, and Massachusetts. She graduated from Littleton Jr. Sr. High School in 2001 and left New England to follow her dreams of saving the endangered Florida manatee. Michelle earned her B.S. (cum laude) in biology with minors in chemistry and psychology in 2005 from Eckerd College in Saint Petersburg, Florida. After graduating, Michelle worked for the Florida Fish and Wildlife Conservation years honing her population genetic skills on multiple species including tarpon, red drum , and the Florida manatee. She has attended multiple genetic workshops including Applied Conservation Genetics presented by the US Fish and Wildlife Service in Shepherdstown, West Virginia and Recent Advances in Conservation Genetics hosted by the Nationa l Cancer Institute and Smithsonian Tropical Research Institute. Michelle has trained at least a dozen volunteers and interns, is skilled at troubleshooting and maintaining the Applied Biosystems automated DNA analyzers and continues to be very proficient at scoring pesky microsatellite marker peaks. Her vested interest and six years of experience working with the Florida manatee, effectively trained her for her research to follow at the University of Florida under the guidance of Drs. Bob Bonde and Marga ret Hunter. During her time at UF, she has been an active member of the Veterinary Graduate Student Association (VGSA) maintaining the roll of Vice President for two years, as well as a member of the e was also fundraiser and
207 student representative of the Florida Chapter of the Society for Marine Mammalogy. She volunteered for the US Geological Survey, which became her second home, as she processed and analyzed many samples there. She has participate d in outreach programs teaching children about manatee research and is an active member of the young adult group at her church. Michelle is looking forward to starting the next phase of her life and continuing her career as a geneticist.
BeyondBonferroni:LessconservativeanalysesforconservationgeneticsShawnR.NarumColumbiaRiverInter-TribalFishCommission,3059-FNationalFishHatcheryRoad,Hagerman,ID,83332, USA(Correspondingauthor:Phone:+1-208-837-4072;Fax:+1-208-837-6047;E-mail:email@example.com)Received4March2005;accepted4September2005Keywords: Bonferroni,conservationgenetics,falsediscoveryrate,multiplecomparisontests Abstract Studiesinconservationgeneticsoftenattempttodeterminegeneticdierentiationbetweentwoormore temporallyorgeographicallydistinctsamplecollections.Pairwise p -valuesfromFisherÂsexacttestsor contingencyChi-squaretestsarecommonlyreportedwithaBonferronicorrectionformultipletests.While theBonferronicorrectioncontrolstheexperiment-wise a ,thiscorrectionisveryconservativeandresultsin greatlydiminishedpowertodetectdierentiationamongpairsofsamplecollections.Analternativeisto controlthefalsediscoveryrate(FDR)thatprovidesincreasedpower,butthismethodonlymaintains experiment-wise a whennoneofthepairwisecomparisonsaresigniÂ“cant.RecentmodiÂ“cationstotheFDR methodprovideamoderateapproachtodeterminingsigniÂ“cancelevel.Simulationsrevealthatcritical valuesofmultiplecomparisontestswithboththeBonferronimethodandamodiÂ“edFDRmethod approachaminimumasymptoteverynearzeroasthenumberoftestsgetslarge,buttheBonferroni methodapproacheszeromuchmorerapidlythanthemodiÂ“edFDRmethod.IcomparedpairwisesigniÂ“cancefromthreepublishedstudiesusingthreecriticalvaluescorrespondingtoBonferroni,FDR,and modiÂ“edFDRmethods.ResultssuggestthatthemodiÂ“edFDRmethodmayprovidethemostbiologically importantcriticalvalueforevaluatingsigniÂ“canceofpopulationdierentiationinconservationgenetics.Ultimately,morethoroughreportingofstatisticalsigniÂ“canceisneededtoallowinterpretationof biologicalsigniÂ“canceofgeneticdierentiationamongpopulations. Introduction GeneticdierentiationamongconspeciÂ“cpopulationsisoftentheresultofreproductiveisolation, geneticdrift,orlocaladaptationandisofprinciple concerninconservationgenetics.Inorderto maintainthegeneticdiversityofspeciesatriskto declineorextinction,geneticallydistinctpopulationsthatrepresentanevolutionarylegacyofthe speciesmustberecognizedandmanagedasdistinct populations(Waples 1995).However,interpretationofpopulationdierentiationmayvary dependingonthecriticalvaluedeterminedforstatisticaltests.Thisstudyconcentratesonevaluating biologicallymeaningfulcorrectionsformultiple comparisontestsforuseinconservationgenetics studies. Avarietyofmolecularmarkers(i.e.,restriction fragmentlengthpolymorphisms,microsatellite loci,singlenuclearpolymorphisms)canbeusedto assaysamplecollectionsandtestforhomogeneity ofallelefrequenciesamongpopulations.SigniÂ“canceamongpopulationsisfrequentlyreported aspairwise p -valuesfromFisherÂsexacttestsor contingencyChi-squaretests.TypeIerror,the probabilityoffalselyrejectingthenullhypothesis, isreducedbyselectingan a-priori signiÂ“cancevalue ( a )todeterminestatisticalsigniÂ“canceofindividual hypothesistests.Whenmultiplehypothesistests areperformed,theexperiment-wise(EW)TypeIConservationGenetics(2006)7:783Â…787 Springer2006 DOI10.1007/s10592-005-9056-y
errorisquicklyincreasedattherateof1 ) (1 ) a )k; where k isthenumberofhypothesistestsperformed.Forexample,if10hypothesistests areperformedat a =0.05,theEWTypeIerroris 1 ) (0.95)10=0.401.Frequently,Bonferronicorrections(Rice1989)havebeenusedtocontrol experiment-wise a Ã° aEWÃž atapredeterminedlevel (i.e.0.05)usingtheformula: aEW= a / k .TheBonferronicorrectedcriticalvaluefortheexamplewith k =10and a =0.05,wouldbe0.05/10=0.005. Bonferronicorrectionsareeectiveatcontrolling aEW,butthecorrectionisveryconservativeand power(theproportionofthefalsenullhypotheses thatarecorrectlyrejected)isgreatlyreduced (Hochberg1988;RymanandJorde2001;Garcia 2004).WhilemanystudiesfocusonreducingTypeI errorwithBonferronicorrections,reducingTypeII error(orincreasingpower)isoftenneglected inmultiplecomparisontests.SequentialBonferroni corrections(ÂÂstep-upÂÂHolm1979;Rice1989;ÂÂstepdownÂÂHochberg1988)providesomeimprovements topower,butassumeindependenceamongtests. Additionalmultiplecomparisonprocedureshave beenreviewed(e.g.,Ludbrook1991;Ludbrook 1998),butthemostpowerfulapplyonlytolimited numbersoftests(<Â“vetests;Dunn-Sidak),normally distributeddata(Tukey-Kramer),orindependent tests(Ryan-Peritz-Welsch).Thishighlightstheneed foramultiplecomparisonprocedureinconservation geneticsthatcanaccommodatelargenumbersof potentiallydependenttestswhilebalancingrisks ofTypeIandIIerrors.Failuretorejectthenull hypothesiswhenitisfalsecanleadtoconservation managementstrategiesthataredetrimentalrather thaneectiveatpreservinggeneticdiversityof species.Thus,measurestoreduceTypeIIerrorare necessarytoadequatelydeterminesigniÂ“cantgenetic structureamongpopulations. InthisstudyIevaluatedalternativestothe Bonferronimethodfordeterminingcriticalvaluesof multiplecomparisontestsinstudiesofconservation genetics.Thealternativemethodsfocusaround theconceptoffalsediscoveryrate(FDR)initially presentedbyBenjaminiandHochberg( 1995). Methods Alternativecorrectionsformultiplecomparisontests BenjaminiandHochberg( 1995)havepresentedan alternativemultipletestingprocedurecalledFalse DiscoveryRate(FDR).FDRcallsforcontrollingthe expectedproportionoffalselyrejectedhypotheses ratherthancontrollingallfalselyrejectedhypotheses asin aEW.FDRprocedureshavebeensuccessfully usedinsituationswithverylargenumbersofpairwise testssuchasmicroarraygeneexpression(Reiner etal.2003)andneuroimaging(Genoveseetal. 2002).TheFDRprocedurepresentedbyBenjamini andHochberg(1995)ishere-to-forreferredtoasthe BÂ…Hmethodandisperformedasfollows: (1) Order p -values p1 p2 pkwhere k =numberofpairwisetests (2) Startingwiththelargest p -value,Â“ndtheÂ“rst individual p -value( pi)thatsatisÂ“es: piÂ£ i / k * a where i = ithobservation (3) The pithatsatisÂ“estheconditionabove becomesthecriticalvaluefortheexperiment. Theprocedureisillustratedinthefollowing example: k =15withordered p( i )sof0.0001,0.0010, 0.0062,0.0101,0.0214,0.0227,0.0273,0.0292, 0.0311,0.0323,0.0441,0.0490,0.0573,0.1262, 0.5794.UsingtheFDRprocedurewith a =0.05, each p( i )iscomparedsequentiallywith i /15*0.05, startingwith p(15).TheÂ“rst p -valuetosatisfy pi i / k * a is p(10)since p(10)=0.032 Â£ 10/15*0.05= 0.0333.Thuspairwisetestsintheexperimentwith p valueslessthanorequalto0.0333rejectthenull hypothesis.ThegaininpowerwiththeBÂ…Hmethod FDRovertheBonferroniprocedure(criticalvalues of0.0333and0.003respectivelyinthisexample)are substantial.Whileproportionoffalsediscoveriesis controlled,theEWTypeIerrorisnotcontrolled withtheBÂ…HmethodFDRinthisexamplesince somehypothesistestswerefalse(i.e.,ÂÂweakcontrolÂÂof aEW). SeveralmodiÂ“cationsoftheBÂ…HmethodFDR havebeenpresented(e.g.,Storey 2002;Bickel 2004),butmanyhavenarrowapplicationorare diculttoadopttotestsofhomogeneityofallele frequencydata.IfocusonamodiÂ“edFDRprocedurebyBenjaminiandYekutieli(2001)referred toastheBÂ…Ymethodfromthispointforward. TheBÂ…Ymethodisacceptablewithdependent testsandiscalculatedwithapredetermined a that isdividedbyaquantityrelatedtothenumberof hypothesistests.IntheBÂ…Ymethod,thecritical valueisdeterminedby: a = Rk i Â¼ 1Ã° 1 = i Ãž 784
Withanexampleof15hypothesistests( k )and a =0.05, a isdividedbythesumof 1 = 1 Ã¾ 1 = 2 Ã¾ 1 = 3 Ã¾ 1 = 4 Ã¾ 1 = 5 Ã¾ 1 = 6 Ã¾ 1 = 7 Ã¾ 1 = 8 Ã¾ 1 = 9 Ã¾ 1 = 10 Ã¾ 1 = 11 Ã¾ 1 = 12 Ã¾ 1 = 13 Ã¾ 1 = 14 Ã¾ 1 = 15 Â¼ 0 : 05 = 3 : 319 Â¼ 0 : 015.ThiscriticalvalueisintermediaterelativetothosecalculatedfromBonferroni (0.0033)andBÂ…HmethodFDR(0.0333). Resultsanddiscussion PowerandTypeIerrorcomparison Therelativepowerofthethreemultiplecomparisontestshasbeendemonstratedtoincrease fromBonferroni,toBÂ…YmethodFDR,toBÂ…H methodFDR.Simulationbasedestimates(BenjaminiandHochberg 1995)betweentheBonferroniandBÂ…HmethodFDRrevealthatthepower oftheBÂ…Hmethodisuniformlylargerthanthe Bonferronimethod,andimprovementinpower withFDRisdramaticwithincreasingnumbersof hypothesistests(as k increasesfrom4to64in simulations).TheBÂ…YmethodFDRwasdevelopedtoaccommodatedependenceamong hypothesistestsandismoreconservativethanthe BÂ…HmethodFDR(seeBenjaminiandYekutieli 2001forproofoftheorem).Thiscorrespondsto largerTypeIerrorwiththeBÂ…HmethodFDR relativetoBonferonniasdemonstratedinsimulationsbyWelleretal.(1998),andBÂ…Ymethod providingamoreconservativeTypeIerrorrate thanBÂ…Hmethod(BenjaminiandYekutieli 2001). CriticalvalueswerecalculatedandarecontrastedinFigure 1withBonferroniandBÂ…Y methods(upto100multiplecomparisons).(BÂ…H methodFDRwasnotincludedascriticalvalues ofthismethodarespeciÂ“ctoeachexperiment.) Bothofthemethodsapproachzerowith increasingnumberoftests,buttheratethatthe Bonferroniprocedureapproacheszeroismuch morerapid.Infact,thenumberoftests( k )toreachcriticalvaluesof0.01,0.001,0.0001forthe Bonferronicorrectionare5,50,and500andfor theBÂ…Ymethodare74,2302,and24574, respectively. Applicationofthethreeprocedurestopublished data Inordertoevaluatetheutilityofthesevarious methodstostudiesinconservationgenetics,the criticalvalueofeachofthethreeprocedures Figure1. Criticalvaluesoftwoproceduresforcorrectionsofmultiplecomparisontests:BenjaminiandYekutieli(2001;BÂ…Ymethod FDR)andBonferronicorrectionforincreasingnumbersofhypothesistests( k ).785
(Bonferroni,BÂ…HmethodFDR,andBÂ…Ymethod FDR)wasdeterminedforthreeexamplesfrom recentpeerreviewedliterature.Criteriafor choosingthethreestudieswere:conservation geneticsresearch,allpairwise p -valuesreported, taxonomydierbetweenstudies,anddateof publicationinthelast10years.Completelistsof pairwise p -valueswerenotcommonlyreportedin aliteraturereviewofgreaterthan300conservationgeneticsmanuscripts,sothiscriterionbecamethemoststringent.Thethreeselected studiesincludeconservationofredkangaroos (Cleggetal.1998;mtDNAcontrolregion sequence),whitecappedalbatrosses(Abbottand Double2003;sixmicrosatelliteloci),andsteelheadtrout(Narumetal.2004;sixmicrosatellite loci). Criticalvaluesforthethreeproceduresand signiÂ“canceforthethreestudiesarepresented inTable 1.Ingeneral,theTypeIerrorrateof theBÂ…Hmethodistoohightobeappliedwidely inconservationgenetics.Ontheothersideofthe spectrum,theBonferronicorrectionprovided veryconservativecriticalvaluessuchthatmany fewertestsweresigniÂ“cantineachstudy.The expectedintermediarysigniÂ“canceoftheBÂ…Y methodrelativetotheothertwomethodswas conÂ“rmedintheempiricalstudies.Further,when thesigniÂ“canttestsareevaluatedwithrespectto eachoverallexperiment,theBÂ…Ymethodappearstodeliverthemostbiologicallymeaningful results.Forexample,theBonferronicorrection inNarumetal.(2004)leadstoinconsistentsigniÂ“canceofpairwisetestsinproximatepopulations.Ontheotherhand,whenthecriticalvalue oftheBÂ…Ymethodisappliedtotheresults, signiÂ“canceismoreconsistentandsensiblefor thegivenexperiment.Further,theBÂ…Ymethod FDRcanbeusedinsituationswheretestsmay bedependentandtheunderlyingdistributionis notnormalorunknown.Asinvestigators attempttoachieveabalanceofTypeIandII errors,theBÂ…Ymethodprovidesintermediate criticalvaluesasdemonstratednotonlyinsimulations,butalsoasappliedtoempiricalgenetic dataanalyses. Theevidencepresentedheresuggeststhatthe appropriatemethodofcorrectionformultiple comparisontestsshouldbechosenonacase-tocasebasisasdeterminedbypriorityofTypeI andTypeIIerrors,anddependencyoftests. InvestigatorsshouldconsiderpresentingsigniÂ“canceofempiricaldataatcriticalvaluesas determinedbyboththeBonferroniandBÂ…Y methodsbyincludingtworespectivereference markswithpairwisevalues(e.g., FSTÂ¼ 0 : 018Ã¾, where*and+indicaterespectivesigniÂ“canceby BonferroniandBÂ…Ymethods).TheBÂ…Ymethod FDRmaybeusefulforothertypesofdata pertinenttotheÂ“eldofconservationgenetics suchasevaluationofHardyÂ…Weinbergequilibriumatmultiplelociandmultivariateanalysisof morphometrictraits.However,moreorless conservativeapproachesmaybewarranted dependingonthehypothesesevaluated(e.g., multilocustestsofHardyÂ…Weinbergequilibrium maynecessitatehighlyconservativetestsdueto thelargenumberofassociatedassumptions). Aquickreferencechartofcriticalvaluesforthe BÂ…YmethodandBonferronicorrectionfor k of 1Â…100at a =0.05aregiveninAppendixA. ReportingstatisticalsigniÂ“canceinathorough mannerwillallowreaderstomorebroadly interpretresultsfromthestudyandpotentially assistinconservationeorts. Acknowledgements Thismanuscriptwasimprovedfollowingreviews fromVincentBuonaccorsi,RishiSharma,andtwo anonymousreviewers. Table 1. Criticalvalue(c.v.)andnumberofsigniÂ“canttests/totaltestsforthreemethodstocorrectformultiplecomparisonsfrom threeempiricalstudies MethodCleggetal.(1998)AbbottandDouble(2003)Narumetal.(2004) Bonferronic.v.=0.00216/24c.v.=0.00312/15c.v.=0.00310/15 BÂ…YFDRc.v.=0.01318/24c.v.=0.01512/15c.v.=0.01512/15 BÂ…HFDRc.v.=0.03820/24c.v. 0 : 04313/15c.v.=0.04715/15786
ReferencesAbbottCL,DoubleMC(2003)Geneticstructure,conservation geneticsandevidenceofspeciationbyrangeexpansionin shyandwhitecappedalbatrosses. Mol.Ecol. , 12 ,2953Â…2962. BenjaminiY,HochbergY(1995)Controllingthefalsediscoveryrate:Apracticalandpowerfulapproachtomultiple testing. J.R.Stat.Soc.B , 57 ,289Â…300. BenjaminiY,YekutieliD(2001)Thecontroloffalsediscovery rateunderdependency. Ann.Stat. , 29 ,1165Â…1188. BickelD(2004)Degreesofdierentialgeneexpression: detectingbiologicallysigniÂ“cantexpressiondierencesand estimatingtheirmagnitudes. Bioinformatics , 20 ,682Â…688. CleggSM,HaleP,MoritzC(1998)Molecularpopulation geneticsoftheredkangaroo( Macropusrufus ):mtDNA variation. Mol.Ecol. , 7 ,679Â…686. GarciaLV(2004)EscapingtheBonferroniironclawinecologicalstudies. OIKOS , 105 ,657Â…663. GenoveseCR,LazarNA,NicholsT(2002)Thresholdingof statisticalmapsinfunctionalneuroimageanalysisusingthe falsediscoveryrate. Neuroimage , 15 ,870Â…878. HochbergY(1988)AsharperBonferroniprocedureformultipletestsofsigniÂ“cance. Biometrika , 75 ,800Â…803. HolmS(1979)Asimplesequentialrejectivemultipletestprocedure. Scand.J.Stat. , 6 ,65Â…70. LudbrookJ(1991)Onmakingmultiplecomparisonsinclinical andexperimentalpharmacologyandphysiology. Clin.Exp. Pharmacol.Physiol. , 18 ,379Â…392. LudbrookJ(1998)Multiplecomparisonproceduresupdated. Clin.Exp.Pharmacol.Physiol. , 25 ,1032Â…1037. NarumSR,ContorC,TalbotA,PowellM(2004)Genetic divergenceofsympatricresidentandanadromousformsof Oncorhynchusmykiss intheWallaWallaRiverand ColumbiaRiverBasin,U.S.A. J.FishBio. , 65 ,471Â…488. ReinerA,YekutieliD,BenjaminiY(2003)Identifyingdierentiallyexpressedgenesusingfalsediscoveryratecontrolling procedures. Bioinformatics , 19 ,368Â…375. RiceWR(1989)Analyzingtablesofstatisticaltest. Evolution , 43 ,223Â…225. RymanN,JordePE(2001)Statisticalpowerwhentestingfor geneticdierentiation. Mol.Ecol. , 10 ,2361Â…2373. StoreyJD(2002)Adirectapproachtofalsediscoveryrates. J. R.Stat.Soc.B , 64,479Â…498. WaplesRS(1995)EvolutionarysigniÂ“cantunitsandtheconservationofbiologicaldiversityundertheEndangeredSpeciesAct. Am.Fish.Soc.Symp. , 17 ,8Â…27. WellerJI,SongJZ,HeyenDW,LewinHA,RonM(1998)A newapproachtotheproblemofmultiplecomparisonsin thegeneticdissectionofcomplextraits. Genetics , 150 , 1699Â…1706. AppendixA. Referencetablewithcriticalvaluesformultiple tests( k )rangingfrom1to100forboththeBonferroniandBÂ…Y method(BenjaminiandYekutieli2001)k BonferroniBÂ…Ymethod k BonferroniBÂ…Ymethod 10.050000.05000510.000980.01106 20.025000.03333520.000960.01101 30.016670.02727530.000940.01096 40.012500.02400540.000930.01092 50.010000.02190550.000910.01087 60.008330.02041560.000890.01082 70.007140.01928570.000880.01077 80.006250.01840580.000860.01073 90.005560.01767590.000850.01068 100.005000.01707600.000830.01064 110.004550.01656610.000820.01059 120.004170.01611620.000810.01054 130.003850.01572630.000790.01050 140.003570.01538640.000780.01046 150.003330.01507650.000770.01041 160.003130.01479660.000760.01037 170.002940.01454670.000750.01032 180.002780.01431680.000740.01028 190.002630.01409690.000720.01024 200.002500.01390700.000710.01020 210.002380.01372710.000700.01015 220.002270.01355720.000690.01011 230.002170.01339730.000680.01007 240.002080.01324740.000680.01003 250.002000.01310750.000670.00999 260.001920.01297760.000660.00995 270.001850.01285770.000650.00991 280.001790.01273780.000640.00987 290.001720.01262790.000630.00983 300.001670.01252800.000630.00979 310.001610.01242810.000620.00975 320.001560.01232820.000610.00972 330.001520.01223830.000600.00968 340.001470.01214840.000600.00964 350.001430.01206850.000590.00960 360.001390.01198860.000580.00957 370.001350.01190870.000570.00953 380.001320.01183880.000570.00949 390.001280.01175890.000560.00946 400.001250.01169900.000560.00942 410.001220.01162910.000550.00938 420.001190.01156920.000540.00935 430.001160.01149930.000540.00931 440.001140.01143940.000530.00928 450.001110.01138950.000530.00924 460.001090.01132960.000520.00921 470.001060.01127970.000520.00918 480.001040.01121980.000510.00914 490.001020.01116990.000510.00911 500.001000.011111000.000500.00908Appendix 787