University of Florida Thesis Or Dissertation Formatting

HIERARCHICAL PROCESSES STRUCTURING CROCODILE POPULATIONS IN CENTRAL AFRICA

MATTHEW HARTON SHIRLEY

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

2013

In memory of John Thorbjarnarson

ACKNOWLEDGMENTS

I would like to start by expressing my gratitude to my committee. Jim Austin was the ideal advisor for me because he allowed me to figure things out on my own, trusting that I would not get too lost along the way, while offering steadfast support when needed. John Thorbjarnarson was incredibly instrumental in the vision for and success of this work and his influence on my research and conservation philosophy cannot be understated. Perran Ross was providing opportunities to help me since before I even started my graduate career and that support has never flagged. Paul Reillo has been a constant example in my life for how to live conservation and I thank him for always taking an interest in me. I offer my sincerest gratitude to my other committee members,

Charlie Baer and Susan Cameron Devitt, for their enthusiasm, willingness to share resources, and accessibility. Kent Vliet, while not on my committee, was always available as a mentor, friend, colleague and collaborator.

I thank the governments of Senegal, Gambia, Cote-d’Ivoire, Ghana, Gabon, and the Democratic Republic of Congo (DRC) for granting me research and export permits.

In Gabon I would especially like to recognize ANPN, including conservateurs Brice

Meye, Joseph Okouyi, Sosthene Ndong Obiang, Solange Ngouessono, Jean

Tondanoye, and CENAREST, particularly Daniel Franck Idiata and Flore Koumba. The

Gabonese Division de la Faune et de la Chasse of MINEF was always gracious in providing my CITES export permits. Much logistic and in-kind support was provided by the Wildlife Conservation Society, WWF, Smithsonian, PPG, PROGRAM and the

Lukuru Foundation. Special thanks are owed to Ruth Starkey, Malcolm Starkey, Alden

Whittaker, Romain Calaque, Bas Huijbrechts, Bas Verhage, Lisa Korte, Marguerite

Butler, Nicolas Bout, Koumba Kombila, Zacharie Chifundera, Ashley Vosper, and John

and Therese Hart. Victor Mpira was the best boatmen I ever had in Africa. I would also like to give special thanks to Fondation Liambissi and Philippe and Sylvie du Plessis for their logistic and financial support, but most of all their friendship.

I thank Steve Rogers and the Carnegie Museum of Natural History, Max

Nickerson, Kenny Krysko and the Florida Museum of Natural History, Patrick Campbell and the British Museum, Addison Wynn and the Smithsonian Institute, Allison Murray and the University of Alberta, Jon Woodward, Jose Rosado and the Harvard Museum of

Comparative Zoology, David Kizirian, George Amato, Natalia Rossi and the American

Museum of Natural History, as well as the Naturkunde Museum – Frankfurt,

Seckenberg Museum of Natural History, Royal Belgian Institute for Natural Sciences, and the Royal Museum for Central Africa for providing skeletal specimens and pictures.

I thank Jennifer Nestler for providing pictures from the latter European institutions.

Loopy Lu Rabanal was kind enough to take photos at the British Museum. Ashley

Pearcy helped clarify some of her analytical methods and offered advice. Vicki

Villanova extracted tons of DNA greatly simplifying my workload. A very special mention goes to Mandie Carr who did a superb job illustrating the crocodile skulls and is, quite possibly, the best undergraduate research collaborator ever.

Labwork was supported by a Doctoral Dissertation Improvement Grant from the

National Science Foundation, the Riverbanks Zoo and Gardens Conservation Support

Fund, and the Austin Lab. Fieldwork was supported by the USFWS Wildlife Without

Borders program, Conservation, Food and Health Foundation, Columbus Zoo, Idea Wild

Foundation, St. Augustine Alligator Farm, IUCN/SSC Crocodile Specialist Group, AZA

Crocodilian Advisory Group, Minnesota Zoo, Fresno Chaffee Zoo, San Diego Zoological

Society, Mohamed bin Zayed Species Conservation Fund, Cleveland Metroparks Zoo,

Oklahoma City Zoo, Aspinall Foundation, Wildlife Conservation Society, WWF, Rotary

International, and Colin Stevenson. The Rare Species Conservatory Foundation was very gracious in serving as a zero-overhead funding fiduciary for these grants. I would like to thank the University of Florida and the Dept. of Wildlife Ecology & Conservation for providing fellowship, scholarship, assistantship, and travel grant support.

Mitch Eaton provided many tissue samples and sequences, and was always available to talk about my work, crocodiles, conservation and Africa. John Brueggen,

Kevin Torregrosa, and David Kledzik at the St. Augustine Alligator Farm helped train me to handle and sample crocodilians and I thank them for their continual support and friendship. Monica Lindberg, Caprice McRae, and Claire Williams were wonderfully tolerant in the management of my funds and my absurd travel schedule, and I thank them for always being happy to see me when I came back from the field. Many other friends and colleagues helped make this a success along the way. I’d like to thank the other members of the Austin Lab including Emily Saarinen, John Hargrove, Nate

Johnson, Joe Townsend, and Aria Johnson. I thank Erin McClure, Laura Schreeg, Alex

Moskalenko, Chris Brochu, John Groves, Peter Brazaitis, Michelle Davis, Lucy Keith,

Mike Cherkiss, Frank Mazzotti, Kate Ingenloff, Christine Lippai, Ginger Clark, Richard

Oslisly, David Sebag, Christine Lippai, Charlie Manolis, Tom Dacey, and Grahame

Webb. Emilie Fairet m’a enseigné que tout est possible et j’ai hâte de réaliser nos rêves ensemble. Finally, I thank my parents – Jim and Barbara Shirley, my sister – Liz

Shirley, and my aunt – Diane Keane, for always being supportive of my endeavors… even when that meant disappearing to Africa for months at a time to catch crocodiles!

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

ABSTRACT ...... 12

CHAPTER

1 CROSSING THE DIVIDE – POLYMORPHISM AND SPECIATION IN AFRICAN CROCODILES ...... 14

Hierarchical Processes in Ecology and Evolution ...... 14 Crocodiles of Central and West Africa ...... 17 Slender-snouted Crocodile (Mecistops cataphractus) ...... 18 African Dwarf Crocodile (Osteolaemus tetraspis) ...... 19 Integrated Genetic Approach for Studying Central African Crocodiles ...... 20

2 TOTAL EVIDENCE APPROACH REVEALS TWO CRYPTIC SPECIES IN THE SLENDER-SNOUTED CROCODILE ...... 23

Introduction ...... 23 Methods ...... 26 Taxon and Molecular Character Sampling ...... 26 Data Collection ...... 28 Data Analysis ...... 29 Results ...... 37 Distance and Population Aggregation Analysis ...... 37 Phylogenetic Analysis and Divergence Timing ...... 38 Morphological Analysis ...... 40 Discussion ...... 40

3 COMPARATIVE SPATIAL STRUCTURE OF MECISTOPS SP. NOV. AND OSTEOLAEMUS TETRASPIS REVEALED BY MITOCHONDRIAL AND NUCLEAR MARKERS ...... 66

Introduction ...... 66 Methods ...... 68 Taxon and Character Sampling ...... 68 Data Analysis ...... 70 Results ...... 79 Mitochondrial Sequences and Phylogeographic Analyses ...... 79 Microsatellite Genotypes and Spatial Clustering Analysis ...... 83

Discussion ...... 88

4 COMPARATIVE DEMOGRAPHIC HISTORY OF MECISTOPS SP. NOV. AND OSTEOLAEMUS TETRASPIS REVEALED BY MITOCHONDRIAL AND NUCLEAR MARKERS ...... 119

Introduction ...... 119 Methods ...... 123 Demographic Reconstruction From Mitochondrial Sequence Data ...... 123 Demographic Reconstruction Using Microsatellites ...... 126 Results ...... 128 Bayesian Skyride Reconstruction of Historic Demographics ...... 128 Microsatellite-Based Genealogical Reconstruction of Historic Demographics 128 Discussion ...... 130

5 CONCLUSIONS AND IMPLICATIONS FOR CONSERVATION ...... 154

APPENDIX

A REPRINT OF HISTORICAL DESCRIPTIONS OF MECISTOPS TAXA ...... 164

Cuvier 1824 – Crocodilus cataphractus ...... 164 Bennett 1835 – Crocodilus leptorhynchus ...... 165 Gray 1844 – Mecistops ...... 167 Baikie 1857 – Description of a Nigerian Skull ...... 168 Fuchs et al. 1974 – Crocodylus cataphractus congicus ...... 169

B DESCRIPTION OF MORPHOLOGICAL CHARACTERS FOUND TO VARY BETWEEN THE TWO MECISTOPS CLADES ...... 172

LIST OF REFERENCES ...... 174

BIOGRAPHICAL SKETCH ...... 202

LIST OF TABLES

Table page

2-1 Specimens used in phylogenetic analyses and their capture localities...... 49

2-2 Molecular markers used in this study...... 50

2-3 Slender-snouted crocodile specimens examined as part of the morphological analysis...... 51

2-4 Estimates of average sequence divergence within and between groups...... 54

2-5 Node divergence dates estimated by Beast and r8s...... 55

3-1 Mecistops specimens used in spatial analyses and their capture localities...... 96

3-2 Osteolaemus specimens used in spatial analyses and their capture localities...... 97

3-3 Microsatellite markers used to genotype both Mecistops and Osteolaemus individuals as part of this study...... 98

3-4 Summary statistics for select coalescent simulations run to determine the pattern and timing of lineage sorting in Mecistops...... 99

3-5 Summary statistics for microsatellite markers used to genotype both Mecistops and Osteolaemus individuals as part of this study...... 100

3-6 K-cluster models searched for Mecistops and Osteolaemus using three different clustering methods...... 101

4-1 Prior settings and posterior estimates of population parameters from msvar runs for Mecistops populations...... 139

4-2 Prior settings and posterior estimates of population parameters from msvar runs for Osteolaemus tetraspis populations...... 140

4-3 Posterior estimates of population parameters from the Bayesian skyride model for all populations of both Mecistops and Osteolaemus...... 141

LIST OF FIGURES

Figure page

2-1 Map of sampling localities...... 56

2-2 COI barcoding gap for recognition of Mecistops species...... 56

2-3 Cladistic network from CHA analysis...... 57

2-4 Phylogenetic relationships within the Crocodylia...... 58

2-5 Nuclear DNA phylogenetic relationships within the Crocodylia. Numbers above branches are branch lengths...... 60

2-6 Divergence date estimates within the Crocodylia...... 61

2-7 Comparative cranial morphology of ventral view of Mecistops from Central and West Africa...... 62

2-8 Comparative cranial morphology of dorsal view of Mecistops from Central and West Africa...... 63

2-9 Comparative cranial morphology of occipital view of Mecistops from Central and West Africa...... 64

2-10 Unrooted parsimony cladogram of Mecistops based on morphological character assignment...... 65

3-1 Map of Osteolaemus sampling localities...... 102

3-2 Diagrammatic illustrations of population models tested in coalescent simulations for Mecistops...... 103

3-3 Map of haplotype distribution for Mecistops...... 104

3-4 Map of haplotype distribution for O. tetraspis...... 105

3-5 Statistical parsimony networks created by TCS as part of the NCPA for Mecistops (top) and Osteolaemus (bottom)...... 106

3-6 Haplotype network created using dnapars and Haploviewer for Mecistops...... 107

3-7 Haplotype network created using dnapars and Haploviewer for Osteolaemus...... 108

3-8 Results from Bayesian clustering of all Mecistops microsatellite genotypes using STRUCTURE with K = 4...... 109

3-9 Results from Bayesian clustering of Central Mecistops microsatellite genotypes using STRUCTURE...... 110

3-10 Results from Bayesian clustering of O. tetraspis microsatellite genotypes using STRUCTURE...... 111

3-11 Eigenvalues of sPCA analysis for Mecistops multilocus genotypes...... 112

3-12 Interpolated maps of individual lagged scores for the first (top) and second (bottom) spatial principle components for Mecistops...... 113

3-13 DAPC results for the full Mecistops dataset...... 114

3-14 DAPC results for the Central Africa-only Mecistops dataset...... 115

3-15 Eigenvalues of sPCA analysis for Osteolaemus multilocus genotypes...... 116

3-16 Interpolated maps of individual lagged scores for the first (top), second (bottom left) and third (bottom right) spatial principle components for Osteolaemus...... 117

3-17 DAPC results for O. tetraspis...... 118

4-1 Bayesian skyride plots of effective population size change in Mecistops populations over time...... 142

4-2 Bayesian skyride plots of effective population size change in Osteolaemus populations over time...... 144

4-3 Magnitude and timing of population bottlenecks in Mecistops populations...... 148

4-4 Magnitude and timing of population bottlenecks in Osteolaemus tetraspis populations...... 149

4-5 Timing of population bottlenecks for Mecistops and Osteolaemus populations with time in linear scale...... 150

4-6 Magnitude and timing of population bottlenecks for alternative, test models in the coastal population of Mecistops...... 151

4-7 Magnitude and timing of population bottlenecks for alternative, test models in the Abanda and Akaka populations of Osteolaemus tetraspis...... 152

4-8 Bayesian skyride plot for the coastal population of Mecistops illustrating Ne estimates over time under three different fixed clock rates...... 153

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

HIERARCHICAL PROCESSES STRUCTURING CROCODILE POPULATIONS IN CENTRAL AFRICA

Matthew Harton Shirley

May 2013

Chair: James D. Austin Major: Wildlife Ecology and Conservation

Understanding how species respond to dynamic geologic, climatic and ecological processes is critical for continent-scale conservation planning. Evolutionary conservation offers a framework to consider evolutionary processes in conservation by prioritizing management action at and below the species level. Unfortunately, implementing evolutionary conservation is often hindered by insufficient knowledge of both species-specific and regional evolutionary processes, particularly where those species exist only in challenging geopolitical settings. Two such species are the slender-snouted crocodile (Mecistops cataphractus) – the least known crocodilian in the world – and the African dwarf crocodile (Osteolaemus tetraspis). To overcome this knowledge deficiency and plan for future species management, I investigated the interaction of regional abiotic processes and species ecology in shaping the phylogeographic histories of these sympatrically distributed African crocodiles using a comparative phylogeographic framework.

Using a total evidence phylogenetic approach I uncovered previously undescribed

Mecistops species isolated in the Upper Guinea and Congo biogeographic zones. The timing of lineage divergence for these taxa was congruent with Osteolaemus and

African Crocodylus lineages and identified the Cameroon Volcanic Line as a significant biogeographic barrier. Within the Congolian biogeographic zone, I uncovered similar patterns of spatial genetic structure between Mecistops and Osteolaemus, with evidence for historically widespread and well-connected populations in both species.

Varying degrees of genetic isolation between the Congo, Ogooué and coastal basins underscored the importance of interpreting regional biogeographic processes in light of ecological factors like associations with forest cover and species-specific habitat utilization. Finally, I demonstrated that both species underwent significant demographic declines related to major climatic events since the late Pleistocene from which they were unable to recover genetic diversity.

This multi-scale research considerably expanded our understanding of the processes structuring crocodilian diversity across western Africa. The observed congruence of speciation pattern, spatial genetic structure, and demographic histories between these sympatric crocodilians supports the importance of regional processes like geographic vicariance and climate change despite divergent ecological profiles. In addition, these results highlight the critical role that systematics plays in biodiversity conservation of West and Central African fauna, and draws attention to conservation concerns for two of the world’s least studied crocodilians.

CHAPTER 1 CROSSING THE DIVIDE – POLYMORPHISM AND SPECIATION IN AFRICAN CROCODILES

A recent paradigm shift in conservation biology recognizes the need to emphasize both ecological and evolutionary processes in conservation (Carroll and Fox 2008,

Crandall et al. 2000, Ferriere et al. 2004). While some proponents of evolutionary conservation have argued that this perspective dictates a shift away from species conservation (Myers and Knoll 2001, Noss 1996), I believe that the reality of biodiversity conservation implementation suggests that species are still the unit upon which conservation decisions are made, but these two viewpoints are not at odds. Species conservation efforts can adopt the evolutionary conservation paradigm by integrating multiple levels of temporal and spatial scale for a better understanding of both dynamic biological systems and the long-term conservation of species. My dissertation aims to lay a foundation for evolutionary conservation decision making in two, sympatric African crocodiles by investigating the phylogenetic and phylogeographic relationships of genetic diversity through space and time across their distributions. Specifically I seek to determine: 1) if the slender-snouted crocodile (Mecistops cataphractus) is, like other

African crocodiles, a cryptic species complex with reciprocally monophyletic lineages corresponding to major African biogeographic zones; 2) the processes structuring spatial genetic diversity in Central African Mecistops and Osteolaemus; and 3) if significant demographic events are detectable in populations of these species and, if so, at what time scales.

Hierarchical Processes in Ecology and Evolution

Ecological processes relevant to species persistence operate at varying temporal scales with different evolutionary and conservation consequences (Zellmer and

Knowles 2009). Similarly, evolutionary processes operate over varying spatial and temporal scales resulting in different impacts on population structure and connectivity

(Avise et al. 1987, Carroll et al. 2007). Field ecological data can provide insight on how environmental, landscape, and anthropogenic factors impact individual and population distribution, as well as behavior at the fine spatiotemporal scale necessary for conservation planning. Genetic methods can resolve the impact of history on future persistence potential by estimating otherwise challenging population parameters and uncovering recent and historic demographic events (Scribner et al. 2005). A major challenge for studies concerned with current genetic variation and spatial genetic structure is separating the effect of historic from contemporary processes (Epperson

2003).

Natural populations are dynamic entities that undergo changes in size and distribution in space and time. One ramification of such changes is the difficulty in delineating species or distinct populations for analysis (Baum and Donoghue 1995,

Crandall et al. 2000, Davis and Nixon 1992, Fraser and Bernatchez 2001, Moritz 1994,

Waples and Gaggiotti 2006). While this may appear to largely be an issue of temporal scale, the relevance of spatial scale cannot be understated (Avise et al. 1987). By incorporating both broader temporal and spatial perspectives our understanding of the processes governing population dynamics (e.g., vicariance and dispersal) can be enhanced (Hey et al. 2004). Assessing deep population history can provide a better understanding of broad-scale spatiotemporal processes like speciation that, when ignored, only exacerbate the issue of species and population delineation (e.g., Hebert et al. 2004).

At fine spatial and temporal scales, dispersal studies reveal general habitat associations, seasonal movements, and how these differ by demographic class (Cooke et al. 2004, Wittmer et al. 2005). While such data provide insight into the behavior and movement of individuals that are important for species conservation planning, they speak little to the broader ecological and evolutionary context of species persistence in the landscape (e.g., Clobert et al. 2001). For example, due to their narrow spatiotemporal focus, field ecological methods often provide no data on long-distance dispersal or on the success of dispersal events (i.e., gene flow) (Rousset 2001,

Rueness et al. 2003). In contrast, inferring dispersal events through genetic methods can be more informative about long-term population dynamics but genetic estimates of migration have an unknown correlation to individual dispersal probability (Peacock and

Ray 2001). An integrated approach combining field and genetic methods may be more useful in diagnosing metapopulation dynamics (Zellmer and Knowles 2009), as well as understanding inter-deme movement pathways and habitat selection in degraded habitats (Beier et al. 2008).

Demographic events (e.g., changes in population size) have important ramifications for ecological processes and evolutionary potential (Crandall et al. 2000,

Ferriere et al. 2004). While census counts of population size and surveys of distribution are critical components of conservation action, it is well known that census size does not represent population evolutionary or ecological adaptability (e.g., Wright 1938).

Effective population size (Ne) more accurately reflects historic population processes and represents future evolutionary potential (Nunney and Elam 1994). Similarly, knowing the distribution or habitat preference of a species is informative about its

potential space occupancy and population connectivity; however, distribution and habitat selection may belie recent, human-mediated shifts in species natural history characteristics away from historic optima. Phylogeographic assessments can reveal significant patterns of distribution change with historic habitat cycles (Bowie et al. 2006), though understanding genetic variation under more recent landscape dynamics may be more relevant than historic perspectives in the face of rapid, human-induced landscape alteration (Foll and Gaggiotti 2006, Zellmer and Knowles 2009).

Crocodiles of Central and West Africa

Crocodilians are both interesting and appropriate species for studying the influence of scale in spatiotemporal processes structuring populations. The crown group of the Crocodylia was present in the fossil record by the Late Cretaceous with most extant species, including all African taxa, present by some point in the Miocene.

Crocodilians are long-lived with overlapping generations, and exhibit high fecundity and expected high survival in the sub-adult and adult life stages. With rare exception they are widely distributed and are largely habitat generalists within the aquatic ecosystem spectra. Finally, they are inextricably linked to the anthropogenic landscape through conflicts arising from their ecological needs and importance as natural resources. This broadly defined conflict with humans has resulted in recent, large-scale population fragmentation and global population declines.

Recent research has shown that crocodilians underwent significant evolution driven by vicariant forces that geographically partitioned widespread populations (e.g.,

Eaton et al. 2009, Hekkala et al. 2011, Meredith et al. 2011); though crocodile population dynamics may have also been shaped by dispersal and secondary contact as climates shifted the distribution of habitats over time (Eaton 2008, Hekkala et al.

2010, Hrbek et al. 2008). Modern African crocodilian lineages, for example, are the result of geologic events, like the closure of the Tethys Sea or the formation of the

African Rift Valley and the Cameroon Volcanic Line, as well as climatologic events like the expansion-contraction cycles of the Afrotropical forests as a result of Pleistocene glaciations. My research focused on two of the three crocodile species currently recognized in Central and West Africa: Mecistops cataphractus and Osteolaemus tetraspis.

Slender-snouted Crocodile (Mecistops cataphractus)

The slender-snouted crocodile (Mecistops cataphractus) is distributed from the

Gambia River in West Africa to lakes Tanganyika and Mweru bordering the Democratic

Republic of Congo (Shirley 2010). It is a medium-bodied species capable of reaching

4.0 m total length. Very little is known of the ecology of Mecistops, though what we do know suggests it is a specialist of forested, freshwater wetland habitats. In West Africa it can also be found in wooded savannah and in Gabon it is irregularly found in semi- estuarine environments (summarized in Shirley 2010). No movement or habitat selection studies have been conducted on this species and we have no real understanding of its capacity for dispersal over long distances or through different habitat types, though anecdotal evidence suggests it is a highly aquatic species unlikely to conduct large overland dispersals.

Mecistops is a mound-nesting species that prefers to place its nests under closed canopy cover within 10 m of the water’s edge (Waitkuwait 1985, M. Shirley unpub. data), though nests have been found in more anthropogenically disturbed habitat

(Shirley et al. 2009). Nests contain an average of 16 eggs (Waitkuwait 1985, Shirley

2010) and eggs and hatchlings are very large relative to female size (Thorbjarnarson

1996, Waitkuwait 1985, 1989). No information is currently available on survival, mating frequency, success, or any other reproductive characteristic in this species; though

Mecistops is widely regarded as one of the most vocal crocodilians leading different researchers to hypothesize widely about its mating system (Shirley 2010).

The slender-snouted crocodile is currently listed as Data Deficient on the IUCN

Red List, though it was last assessed in 1996 and is in need of an update (IUCN 2011).

The major threats facing this species include large-scale loss of habitat, illegal hunting for the trade of leather and bushmeat, and conflict with artisanal fisheries. The latest information (summarized in Shirley 2010) suggests that these threats have resulted in highly fragmented, reduced populations and, as a result, this species may merit a higher threat designation.

African Dwarf Crocodile (Osteolaemus tetraspis)

The African dwarf crocodile is distributed throughout West and Central Africa from

Senegambia in the west to the Ituri Forest of the Congo Basin in the east (Eaton 2010).

It is a small-bodied crocodilian reaching a maximum total length of less than 2 m.

Relatively little is known of the ecology of Osteolaemus, though what we do know suggests that it is restricted to forested wetland habitats including those bordering coastal and other estuarine environments (summarized in Eaton 2010). Until recently, the evolutionary history of this species was under debate with two species in two separate genera recognized, though synonymized, at different periods. Eaton et al.

(2009) found significant molecular, morphological, and biogeographic support for the presence of three species within the genus Osteolaemus. The nominate O. tetraspis is restricted in distribution to the Ogooué Basin of western Central Africa.

A single study to date examined movements and population connectivity in this species (Eaton 2008), though it is as yet published and no other information is available.

Anecdotal evidence, however, suggests that Osteolaemus is capable of making extensive terrestrial movements (e.g., Waitkuwait 1989) and many researchers report seeing them throughout the forest away from water. Osteolaemus is a mound-nesting species that selects closed canopy nesting sites. They lay an average of 10 – 14 eggs

(Eaton 2010, Waitkuwait 1989) and there appears to be little in the way of parental care

(M. Shirley, pers. obs.). As for Mecistops, we know nothing about reproductive frequency, success, or survival in this species.

Integrated Genetic Approach for Studying Central African Crocodiles

The goal of this research is to address population structure of sympatric African crocodilians at multiple temporal and spatial scales by integrating molecular data from different markers and analytical methods. This will not only facilitate accurate assessment of the ecology and evolution of these species, but also better our understanding of regional ecological and evolutionary processes impacting wildlife population structure in the tropical forest wetland ecosystems of western Africa.

Following recent results concerning the systematic status and non-monotypy of both dwarf (Eaton et al. 2009) and Nile (Hekkala et al. 2011) crocodiles, it was necessary to first test the monotypy of Mecistops before further spatiotemporal analyses could be conducted. In Chapter 2 I tested the hypothesis that the range-wide distribution of Mecistops reflects cryptic vicariant processes leading to highly divergent, reciprocally monophyletic clades. I tested this hypothesis by examining the molecular phylogenetic relationships of slender-snouted crocodile populations throughout its

range. I provide further support in the form of divergence dating and cranial morphological analysis.

In Chapter’s 3 and 4 I used spatial and demographic analyses to test the hypothesis that ecological niche partitioning resulted in discordant phylogeographic histories between the sympatrically distributed Mecistops and Osteolaemus.

Comparative phylogeography provides an analytical framework under which to assess the importance of geographic and demographic processes on the population structure of regional biotas. Discordant genealogical structure, in this case between the slender- snouted and dwarf crocodiles, would suggest that differing natural history strategies, shifts in niche distribution, and disparate responses to anthropogenic threats may be more important in explaining crocodile population structure than regional biogeographic processes.

In Chapter 3 I compared the spatial structure of Central African Mecistops and

Osteolaemus tetraspis populations from both a historic and contemporary phylogeographic perspective. I used homologous mitochondrial DNA sequence data to test the null phylogeographic hypothesis of historic panmixia with incomplete lineage sorting against hypotheses of phylogeographic structure and patterns of population connectivity. I used microsatellite data to better understand population clustering as a proxy for gene flow across the landscape.

In Chapter 4 I compared the demographic history and timing of major demographic events between populations of the Central African Mecistops clade and Osteolaemus tetraspis. First, I used mitochondrial sequence data to test the hypothesis that major changes in historic crocodile population size followed significant climatologic or geologic

events against the null hypothesis of stable populations over time. To do this I reconstructed historic demographic profiles in these species using Bayesian skyline plot methods. Second, I tested the hypothesis that populations in both species underwent significant, contemporary population bottlenecks largely as a result of unsustainable hunting pressure and habitat loss in the 20th century.

CHAPTER 2 TOTAL EVIDENCE APPROACH REVEALS TWO CRYPTIC SPECIES IN THE SLENDER-SNOUTED CROCODILE

Introduction

Delimiting species can be difficult, particularly when lineages have only recently diverged or in cases where cryptic phenotypes mask more ancient divergences. While numerous species concepts have been proposed that emphasize different criteria for delimiting species, such as biological (e.g., reproductive compatibility or isolation) or phylogenetic (e.g., reciprocal monophyly) (Wheeler and Meier 2000 for a review), species may be best conceptualized as aggregates of populations that are evolving together as a metapopulation lineage independent from other such aggregates (de

Queiroz 2007). By any concept or conceptualization, however, accurate delimitation of species is critical because species are the fundamental unit for much of biology including biogeography, ecology and in prioritization and conservation planning (e.g.,

Agapow et al. 2004).

In many cases the delimitation of species seems relatively trivial owing to allopatric distributions or pre-zygotic reproductive barriers (e.g., different call types in birds and anurans). In reality, however, the process of species delimitation is increasingly obfuscated by the presence of cryptic variation and the limitations of many traditional species concepts (e.g., morphological or biological) to effectively recognize it (e.g.,

Tattersall 2007). Cryptic species, those that are seemingly morphologically indistinct, are increasingly being recognized as the products of “range-wide” biogeographic processes that geographically structure heterogeneous lineages (e.g., Quattro et al.

2005).

Significant lineage structuring may be more likely in widely distributed taxa subject to range-wide biogeographic or ecogeographic processes at continental or regional scales. The African continent has a long history of rift formation, volcanic uplift, desertification, and shifting habitat distributions due to continental drift, changing ocean currents, and climatic cycles (Axelrod and Raven 1978, Coetzee 1993, Fjeldsa and

Lovett 1997, Livingstone 1993, Plana 2004, Wieczorek et al. 2000). As a result, as many as thirty distinct biogeographic realms have been recognized in Africa (Grubb

1982, Udvardy 1975), most of which are distributed in the eastern and southern portions of the continent. A recent meta-analysis of African plant and vertebrate taxa identified two major zones in western Africa: the Congolian and Guinean (Linder et al. 2012). The division between these was alternately placed at the Sanagha River, the Cameroon

Volcanic Line (CVL), or the Dahomey Gap depending on the exemplar taxonomic group. Linder et al. (2012) also recognized the presence of a third biogeographic zone in this area for some taxa – the Cameroon-Gabon zone (i.e., Lower Guinean). Recent phylogenetic studies of two widely-distributed, African crocodile species found significant support for cryptic species that support all three of these biogeographic zones (Brochu 2007, Eaton et al. 2009, Hekkala et al. 2011).

The third crocodile species in Africa, Mecistops cataphractus, has never been the subject of any species-specific phylogenetic study despite having a taxonomic history suggestive of hidden diversity. The slender-snouted crocodile was recognized and described in 1824 as Crocodylus cataphractus from a specimen of unknown origin

(Cuvier 1824), though its type locality was later restricted to the Senegal River (Fuchs et al. 1974a). Bennett (1835) described C. leptorhynchus largely on the basis of slight

differences in head length to head width ratios from C. cataphractus. Its type locality was designated as “apud Fernando Po.” Gray (1844) replaced the nomen leptorhynchus with bennettii and, additionally, erected the genus Mecistops as he considered these two longirostrine crocodiles to not be ‘true crocodiles’ of the genus

Crocodylus. Mecistops was later synonymized with Crocodylus and bennettii synonymized with cataphractus (Boulenger 1889). However, all recent phylogenetic analyses, both molecular and morphological, demonstrate that cataphractus is not a

‘true crocodile’ and the resurrection of Mecistops has been suggested (e.g., Eaton et al.

2009, Hekkala et al. 2011, McAliley et al. 2006, Meredith et al. 2011, Oaks 2011). A single subspecies, C. cataphractus congicus, was proposed by Fuchs et al. (1974) on the basis of skin character differences. Unfortunately, details of the study that would facilitate testing of subspecies validity were not provided and it is generally not recognized.

Mecistops is widely distributed in western Africa ranging from Lakes Tanganyika and Mweru in the east and southeast to the Gambia River in the west, largely mirroring the distribution of Osteolaemus (Shirley 2010, Shirley et al. 2009). In this chapter I test two hypotheses regarding the range-wide systematic status of Mecistops. First, that

Mecistops is comprised of at least two cryptic species. Second, I test the hypothesis that these cryptic lineages are geographically structured corresponding to the

Congolian, Guinean, and Cameroon-Gabon biogeographic zones of Linder et al. (2012).

These hypotheses are tested against the null hypothesis of panmixia and phylogenetic homogeneity. To test these hypotheses I utilized a total evidence-based analytical approach incorporating both multi-marker molecular and morphological datasets.

Methods

Taxon and Molecular Character Sampling

Individual slender-snouted crocodiles were wild-caught from throughout their range including sites in The Gambia, Cote-d’Ivoire, Ghana, Gabon, Central African

Republic, Republic of Congo, and Democratic Republic of Congo (Table 2-1, Fig. 2-1).

Individuals were captured utilizing standard crocodilian capture techniques including by hand, pole-snaring, and darting (Cherkiss et al. 2004, Walsh 1987). A 0.5 ml blood sample was taken from the cervical vessel posterior the occiput and dried on Whatman filter paper. Samples were stored in paper coin envelopes at ambient environmental conditions. All captured animals were released at the site of capture. All samples were exported from the host country and imported into the USA with authorization from the respective CITES Management Authorities. In addition to wild caught specimens, I sampled virtually all known, unrelated slender-snouted crocodiles in captivity in the AZA and private collections throughout the USA. Sequences from these specimens were only compared to wild specimens and not used in any of the described analyses due to their unknown wild origins.

I examined sequence variation across a total 7,768 bp from 11 gene regions. Four regions (3,768 bp) were mitochondrial (mtDNA) and seven (4,000 bp) were nuclear

(nuDNA) as follows (Table 2-2): mtDNA - partial cytochrome b (cytb), partial 12S, partial cytochrome c oxidase subunit I (COI), and partial NADH dehydrogenase subunit 4

(ND4); nDNA - lactate dehydrogenase A (LDHA), recombination-activating gene 1

(rag1), and the flanking regions of five anonymous microsatellites (CpP205, CpDi13,

CpP4004, CpP3303, and CpP2902). Primers for mitochondrial markers were designed for this study from complete Mecistops mitogenome sequences published on GenBank

using the online version of Primer3 (Rozen and Skaletsky 2000), while primers for the nuclear genes were taken from previously published studies (Eaton et al. 2009) or, in the case of the microsatellites, redesigned from available clone sequences (Miles et al.

2008, 2009) to extend the flanking region.

Phylogenetic analyses included outgroup taxa from across the Crocodylia to both facilitate estimation of divergence timing and further understand the relationship of

Mecistops to the rest of the Crocodylidae. Individual Osteolaemus from all three putative species (Eaton et al. 2009) were sampled following the protocols described above and sequenced alongside Mecistops as part of this study. Outgroup sequences for other members of the Crocodylia were obtained from Genbank as follows:

Crocodylus suchus (mtDNA - JF502243 and JF502244, LDHA – JF315502, rag1 –

Burkina Faso and Gambia from doi:10.5061/dryad.s1m9h), C. rhombifer (mtDNA -

JF502247, LDHA – JF315538, rag1 - AY239172), C. niloticus (mtDNA – JF502245 and

JF502246, LDHA – JF315530 and JF315522, rag1 – MadagascarSE and TanzaniaB from doi:10.5061/dryad.s1m9h), C. acutus (mtDNA – JF502241, LDHA - JF315534, rag1 - doi:10.5061/dryad.s1m9h), C. intermedius (mtDNA – JF502242, LDHA -

JF315496, rag1 - AY239173), C. novaeguineae (mtDNA – JF502240, LDHA -

JF315494), C. porosus (mtDNA – DQ273698, LDHA - JF315542, rag1 - EU375509), C. mindorensis (mtDNA – NC_014670, LDHA - JF315536), C. moreletii (mtDNA –

NC_015235, LDHA - JF315514, rag1 - AY136680), C. palustris (mtDNA – HM4880007,

LDHA - JF315540), C. johnsoni (mtDNA – NC_015238, LDHA - JF315474), Tomistoma schlegelii (mtDNA – NC_011074, LDHA - JF315470, rag1 - AY239176), Gavialis gangeticus (mtDNA – AJ810454, LDHA - JF315485, rag1 - AF143725), Alligator

mississippiensis (mtDNA – NC_001922, LDHA - JF315508, rag1 - AF143724), Caiman crocodilus (mtDNA – NC_002744 and AJ404872, LDHA - JF315512 and JF315466, rag1 - AY239166 and AY136670), Paleosuchus palpebrosus (mtDNA – NC_009729,

JF315531, rag1 - AY239169), and P. trigonatus (mtDNA – AM493869, LDHA -

JF315510) (Eaton et al. 2009, Feng et al. 2010, Gatesy et al. 2003, Groth and

Barrowclough 1999, Hekkala et al. 2011, Janke and Arnason 1997, Janke et al. 2001,

2005, Li et al. 2007, Meganathan et al. 2011, Meredith et al. 2011, Oaks 2011, Ray and

Densmore unpub. data, Roos et al. 2007). Microsatellite sequences for C. porosus were the clone sequences developed by Miles et al. (2008). Microsatellite sequences for A. mississippiensis were obtained by BLAST searching preliminary assemblies of the alligator genome (available from http://www.crocgenomes.org/downloads.html; St

John et al. 2012). Microsatellite sequences for all other outgroup taxa were generated as part of this study from samples collected from captive specimens at the St. Augustine

Alligator Farm except those of C. suchus and C. niloticus which were wild caught specimens of known provenance.

Data Collection

Total genomic DNA was extracted utilizing standard phenol chloroform extraction methods (Sambrock and Russell 2001), quantified and aliquoted into 30 ng/µl working concentrations. Extractions yielding less than 30 ng/µl were aliquoted and utilized at extraction concentration. Polymerase chain reactions (PCR) were performed utilizing the Qiagen Multiplex PCR kit (Qiagen) following manufacturer’s instructions scaled down to a volume of 15 µl containing 1.1 µl genomic DNA and primers at 0.2 µM.

Cycling conditions for all markers were as follows: 95°C 15 min, 35 cycles of 94°C 30 sec / 56.5°C 1:30 min / 72°C 1:45 min, 72°C 10 min final extension. ND4 had an

annealing temperature of 57°C. PCR products were exosap purified with 2 µl exosap and 5 µl PCR product. All gene regions were sequenced in both directions utilizing

BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Carlsbad, CA) following manufacturers specifications scaled back to 7 µl reactions containing 1.1 µl

PCR product template DNA. Cycle sequencing conditions for mtDNA markers were as follows: 96°C 3 min, 30 cycles of 96°C 30 sec / 55.7°C 30 sec / 60°C 1 min with no final extension. Cycle sequencing conditions for nuDNA markers were as follows: 96°C 3 min, 30 cycles of 96°C 30 sec / 57°C 30 sec / 60°C 1:40 min with no final extension.

Sequences were analyzed on an ABI 3130 Genetic Analyzer (Applied Biosystems).

Base calling was performed with Sequencher v4.1 (Genecodes Corp.). Forward and reverse sequences were assembled into contigs, base calls verified by eye, and marker datasets aligned in CLC v3.6.2. Individual alignments were checked by eye prior to concatenation. Both forward and reverse nuclear strands were included for each individual to account for heterozygous individuals. Microsatellite sequence alignment included the repeat section, though the repeat was removed from the alignment prior to all analyses.

Data Analysis

Distance and population aggregation analysis. Uncorrected p-distances were calculated in MEGA5 (Tamura et al. 2011) between individuals for each marker set individually and by genome (mtDNA vs. nDNA). Mecistops samples were aggregated following individual pairwise comparisons to reflect highest inter- versus lowest intra- grouping distances and mean intra- and inter-group distances were calculated as above. Intra- and inter-group distances for COI were plotted to diagnose a barcoding

gap (i.e., net genetic distance; Hebert et al. 2003a). Barcoding gaps of > 2% have been proposed as a minimum diagnostic for identifying unique species (Hebert et al. 2003b).

Population Aggregation Analysis (PAA; Davis and Nixon 1992) was used in two capacities. First, PAA was used as a discrete, character-based method for defining phylogenetic species under two, complimentary frameworks – PAA1 and PAA2 (Brower

1999). The full Mecistops alignment was searched for unique haplotypes (PAA1) and fixed nucleotide positions (PAA2) that were then geographically aggregated to test the hypothesis that populations from each of the three geographically defined regions will yield unique aggregates. Brower (1999) proposed a third test of character-based species delimitation – cladistic haplotype aggregation (CHA) – that may have better power to overcome possible homoplasies in the dataset through approximate phylogenetic reconstruction. I generated genealogical networks for the mtDNA dataset using the method of maximum parsimony implemented in dnapars of the Phylip-3.69 package (Felsenstein 1989, 2005) with haplotype networks reconstructed from the most parsimonious trees using Haploviewer (Ewing 2011).

Second, the Mecistops dataset was reduced to a minimum of one individual per purported population plus any additional individuals that showed unique haplotypes based on fixed character-based aggregates for the phylogenetic analyses. This was done to both reduce the computational requirements for phylogenetic analysis (i.e., by reducing the overall size of the dataset) given that non-unique sequences contribute nothing to the analysis. COI sequences from captive individuals were compared to the wild dataset as part of the PAA for clade assignment of specimens of unknown wild origin but were not included in subsequent phylogenetic analyses.

Phylogenetic analysis. Hypotheses of phylogenetic structure were generated under both maximum likelihood and Bayesian algorithms. Data from the Mecistops ingroup was analyzed with all outgroup taxa both by genome and concatenated for total evidence analysis (Kluge 1998, Maddison 1997). Individual marker datasets were tested for the maximum likelihood model of base substitution with jModelTest 0.1.1

(Posada 2008) for the Mecistops-only ingroup and for the full dataset including outgroups.

Maximum likelihood analysis was conducted in PhyML implemented on the

Montpellier Bioinformatics Platform (Guindon et al. 2010; http://www.atgc- montpellier.fr/phyml/). Each individual concatenated dataset (mtDNA, nDNA, combined) was analyzed under a GTR+G substitution model with the gamma parameter set at 0.272 as per the recommendation of jModelTest. Equilibrium allele frequencies were optimized from the dataset, the proportion of invariable sites was estimated by the program, and there were five substitution rate categories. Trees were searched from a starting tree created by BIONJ using the best of the SPR and NNI options with topologies and branch lengths optimized. Branch support was determined with the SH- like option of the Approximate Likelihood Ratio Test (aLRT) method (Anisimova and

Gascuel 2006). Trees were rooted by both the outgroup and mid-point rooting methods; both methods produced the same root point (Hess et al. 2007).

Bayesian inference of phylogenetic structure was conducted in MrBayes 3.2

(Ronquist et al. 2012). The concatenated dataset was partitioned by gene region with the substitution model (mtDNA – GTR+G, nDNA – HKY) implemented for each gene with all model parameters estimated by jModelTest. The analysis was run with and

without gaps (indels) coded as restriction site binary characters. Three simultaneous

Markov Chain Monte Carlo (MCMC) searches were run with five chains for 12,000,000 generations with trees sampled every 5000 generations. The adequacy of a 15% burn- in and posterior convergence of all parameters was assessed with Tracer v1.5

(Rambaut and Drummond 2009). A 50% majority rule tree was created after discarding the first 3000 burn-in trees. Trees were rooted by both the outgroup and mid-point rooting methods; both methods produced the same root point (Hess et al. 2007).

Estimation of divergence timing. To assess divergence timing within Mecistops clades, and compare this to sympatrically distributed Osteolaemus species, I estimated divergence timing under both Bayesian and maximum likelihood algorithms with fossil calibration dates.

Bayesian estimation of divergence timing was conducted in Beast v1.7.4

(Drummond et al. 2012) using a 36 taxa dataset including all outgroups, two members of each Osteolaemus clade, and five members of each Mecistops clade for the full

7,979 bp sequence dataset. I applied an uncorrelated lognormal relaxed clock

(Drummond et al. 2006) to the unpartitioned dataset and assumed a GTR+G substitution model with five categories of rate heterogeneity as selected by jModelTest.

I estimated a starting tree in RAxML (Stamatakis et al. 2008), also under a GTR+G substitution model, and scaled branch lengths to fall within the prior distributions for each calibration point.

I used a Yule Process prior for the tree model of speciation (Gernhard 2008) and a uniform prior (U (0,5)) for the uncorrelated lognormal relaxed clock mean rate with an initial value of 0.005 substitutions/site/my (Oaks 2011). Minimum fossil appearance

data from the rich fossil record of the crown Crocodylia was used to set node calibration dates as follows. I gave the tree root prior a normal distribution (N (78, 8)) placing upper and lower truncations of 68 and 115 mya, respectively, to allow the analysis to explore parameter space including estimated Campanian origin of the Crocodylia

(Brochu 2003, Salisbury et al. 2006) and the estimated 78 mya divergence between the

Alligatoridae and Crocodylidae (Brochu 2004a, 2004b). I used a gamma prior (Γ (2,

2.9)) with offset 60 mya for the divergence between Alligator and Caiman; the

Alligatorinae-Caimaninae split is considered one of the best fossil-based calibration constraints for all vertebrates with a minimum divergence estimated at 64 mya (Brochu

2004a, 2004b, Muller and Reisz 2005).

Previous divergence dating analyses in the Crocodylia have only included these two calibrations (e.g., Oaks 2011). However, limiting calibration points to near the root of trees makes estimation of more distant nodes difficult (e.g., Linder et al. 2005).

Likewise, providing calibrations for a limited portion of the major clades included in the analysis limits true exploration of among-lineage rate heterogeneity and may bias estimated ages for more divergent nodes from the calibration points (Ho and Phillips

2009). The robustness of the crocodilian fossil record and recent revisions of our understanding of the crown Crocodylia (e.g., Brochu 2003, Brochu et al. 2009) allow for the use of three additional fossil dates to calibrate internal nodes. Following recommendations and fossil/stratigraphic justifications in Brochu (2004a,b), I placed a gamma prior (Γ (3, 5.5)) with offset 30 mya for the divergence between Paleosuchus and

Caiman, a gamma prior (Γ (2, 2)) with offset 18 mya for the root of the Crocodylinae (i.e.,

Mecistops, Osteoleamus, and Crocodylus), and a gamma prior (Γ (2, 2.8)) with offset 10

mya for the root of the Crocodylus. The use of a gamma distribution offset to before the first fossil appearance of these calibration points (and the Alligator – Caiman split) ensures that the highest posterior density (HPD) includes the fossil/stratigraphic-based mean estimate while allowing for some uncertainty in stratigraphic or fossil interpretation

(i.e., soft upper and lower bounds; Ho and Phillips 2009, Yang and Rannala 2006). No hard upper bounds were set for any node calibrations, except that mentioned for the root of the tree, and no nodes were restricted to be monophyletic allowing the full extent of tree space to be explored.

All other priors and MCMC operators were left at their default settings. MCMC operators were allowed to automatically optimize over the run after the initial 500,000 generations. I ran 4 independent analyses for 5.0 x 107 generations sampling every

10,000 generations. The adequacy of a 15% burn-in and posterior convergence of all parameters was assessed with Tracer v1.5 (Rambaut and Drummond 2009). The results from independent runs were combined using LogCombiner v1.7.4 and a 50% majority rule tree annotated with posterior mean and 95% HPD node heights was created in TreeAnnotator v1.7.4 (Drummond et al. 2012).

Maximum likelihood estimation of divergence dates was conducted in r8s v1.71

(Sanderson 2003, 2006). There are four potential drawbacks to this method. First, node dates are estimated on an input tree containing branch lengths estimated by another program. This differs from the Bayesian method implemented in Beast which estimates the phylogeny as a random variable as part of the analysis. Second, constraining the age of the tree root invariably results in the date estimation running into the lower boundary. The resolution to this is to re-run the analysis multiple times with

the root age fixed for a series of likely ages. Third, the node age constraints for other calibration points are uninformative (i.e., uniformly distributed) hard upper and lower bounds. Finally, it is tedious and impractical to estimate confidence intervals on divergence date estimations in r8s. Despite these limitations, congruence between dates estimated using penalized likelihood in r8s and MCMC methods in Beast provides strong support for inferred divergence timing.

Divergence dates were searched using the penalized likelihood approach

(Sanderson 2002) with the truncated Newton method (TN). The TN method works best with bounded age constraints on node age calibrations. One of the limitations to this method is that short (i.e., near zero) branch lengths can confound estimated rates, and therefore the input tree was a shortened, 30 taxa tree that included all outgroups but only two representatives of each Mecistops and Osteolaemus clade for the full, concatenated sequence dataset. The smoothing parameter was estimated using the cross validation procedure described in Sanderson (2002) and set alternatively to 0.25 and 0.07. The smoothing parameter can range from 0 to infinity where as it approaches

0 there is near complete rate heterogeneity amongst lineages (i.e., deviation from a molecular clock). The gamma shape parameter for rate variation was set to 0.544 as estimated in Beast. Each analysis was restarted 10 times with a perturbation factor of

0.25. The root was alternatively fixed at 78 mya, 86 mya, and 92 mya. Final, bounded node calibrations were [62 mya, 76 mya], [20 mya, 60 mya], and [12 mya, 35 mya] for the Alligator – Caiman, Paleosuchus – Caiman, and Crocodylinae splits.

Morphological analysis. Cranial features of slender-snouted crocodile skulls were compared with respect to geographical origin and the results of molecular

analysis. Unfortunately, none of the original descriptions of M. cataphractus, M. leptorhynchus/bennettii, nor M. cataphractus congicus included discussion of cranial morphology and how it differs amongst the Mecistops taxa or between Mecistops and any other crocodilian (Appendix A). Therefore, initial examination was done through side-by-side comparison of two individuals from each diagnosed clade at the Florida

Museum of Natural History to preliminarily identify potential features of interest.

Discrete characters found to differ between the clades are described in Appendix B.

Following initial description of characters, I sampled skulls in institutions across the

USA and Europe, largely through a series of standard photos of specimens supplied by the institutions (Table 2-3). Each skull was coded for each character by two, independent observers in a blind procedure whereby the observers did not know the identity or geographic origin of the skull being coded. Characters that could not be coded due to incomplete skeletal material or obfuscated views were coded as unknown

(“?”). The character states were considered unordered and analyzed by Wagner parsimony with the program Pars in the phylip3.69 package (Felsenstein 2005).

Wagner parsimony is suitable for discrete, morphological data where the ancestral character states are unknown and coded characters are considered unordered (i.e., all state changes are considered equally likely). Wagner parsimony also assumes that characters and lineages evolve independently and that the rates of evolutionary change are sufficiently low such that one change along a branch is more likely than two (Kluge and Farris 1969). Prior to analysis, OTU x character matrices were compared between the two observers and consensus decisions made for inconsistently coded characters.

Additionally, completely ambiguous characters, specimens with less than 75% of

characters coded, and juveniles were removed from the dataset. I searched for the most parsimonious tree by jumbling the input order of the taxa 100 times and rearranging on up to 1000 equally most parsimonious trees.

Results

Distance and Population Aggregation Analysis

Up to 108 individual Mecistops were sequenced for each gene as part of this study

(Table 2-2). Distance analysis revealed the presence of two genetic clusters within

Mecistops corresponding geographically to Central and West Africa. Mean intra-group divergences within each of these groupings were estimated to be 0.0 – 0.0027 (± 0.001) while inter-group divergences were 0.0194 (± 0.0046) – 0.0789 (± 0.0083) depending on the gene region(s) analyzed (Table 2-4). The COI barcoding gap was 0.0491 (Fig.

2-2).

Both PAA1 and PAA2 offered unambiguous support for phylogenetic divisions between two geographic regions. For PAA1, No mitochondrial haplotypes were shared between regions. For PAA2, the mitochondrial alignment resulted in more fixed characters between, as well as more variable characters within, clusters per gene than the nuclear alignments. ND4 was the most variable region both between and within clusters having 77 fixed differences (7.2% of nucleotide positions) between, and 13 and

11 variable sites within, the Central and West clusters, respectively. Additionally, the

Central cluster had a fixed 2 bp insertion in the t-RNA region flanking ND4. Cytb was the second most variable having 47 fixed differences (5.0%) between, and 7 and 10 variable positions within, the Central and West clusters. COI contained 43 fixed differences (4.7%) between, and 7 and 10 variable positions within, the Central and

West clusters. 12S was the least variable between clusters, showing only 16 fixed differences (1.9%), and 3 variable positions within both the Central and West clusters.

The nuclear regions were considerably less variable with both LDHA and rag-1 not possessing any fixed positions between, or variable positions within, clusters.

Microsatellite flanking sequences, however, contained 2 – 5 fixed differences between clusters (0.45 – 0.88%) and 0 – 1 variable positions within clusters. Some individuals in the Central cluster showed a 1 bp insertion in marker CpP2902, and CpP4004 failed to sequence in the West cluster. This latter result was confirmed by failed genotyping reactions of this same marker in all West samples (Chapter 3).

CHA for both the mtDNA and nuDNA datasets additionally corroborated the hypothesis of geographic partitioning, though only into two clusters as shown in the distance and PAA results. The produced network showed all intra-regional samples clustered in two different, contiguous regions of the network separated from each other by a single branch significantly longer than any internal branch (Fig. 2-3). The spinal branch connecting the haplogroups was 116 steps compared to < 5 steps separating any haplotypes within each cluster.

Phylogenetic Analysis and Divergence Timing

Maximum likelihood and Bayesian analysis of the mtDNA and full, concatenated sequence datasets recovered all expected relationships within the Crocodylia including monophyletic Alligatoridae, Crocodylidae, Caiman, Paleosuchus, Crocodylus, and

Osteolaemus with most nodes receiving 100% aLRT and posterior probability support

(Fig. 2-4). Within the Crocodylus, the Nile crocodile was comprised of two distinct taxa paraphyletic in relation to the New World Crocodylus. Tomistoma and Gavialis were part of the Crocodylidae and formed a clade sister to the monophyletic Crocodylinae.

These results continue to support the presence of significant lineage diversity within

Osteolaemus, as well as a sister group relationship of Osteolaemus and Mecistops to the ‘true crocodiles’ of the genus Crocodylus. Maximum likelihood and Bayesian analysis of the concatenated nuDNA sequence dataset recovered all family groupings, as well as monophyletic genera, though it placed Mecistops as basal to a clade consisting of Osteolaemus sister to Crocodylus (Fig. 2-5). Analysis of individual nuDNA datasets largely recovered monophyletic genera as expected, though relationships amongst genera did not match expectations of total evidence-based Crocodylia systematics (results not shown).

In all analyses, Mecistops consisted of two, highly divergent, reciprocally monophyletic clades with one clade of individuals from West Africa and the other of

Central African individuals. The West African clade exhibited some regional structure with individuals from Senegambia forming a monophyletic clade distinct from individuals sampled from the Upper Guinea forest block of Ghana and Cote d’Ivoire. Similarly, the

Central African clade exhibited some regional structure with individuals from each of five zones – east of the Congo River, north of the Congo River, within the arch of the Congo

River, Sanagha drainage, Ogooué basin – loosely grouping together, though showing very little hierarchical structure. Branch lengths within the two major clades were incredibly short indicating little intra-clade divergence.

Estimated divergence times from Bayesian and maximum likelihood methods were highly congruent as evidenced by the MLE falling within the 95% HPD of the Bayesian means (Table 2-5). The root of the crown Crocodylia, placed at the split of the

Alligatoridae and the Crocodylidae, was estimated to be ±79 mya, while the most recent

common ancestor (MRCA) of the Crocodylidae was likely present by ±45 mya but may be as old as the late Paleocene (Fig. 2-6). The MRCA for the Crocodylinae was dated to the late Oligocene (±26 mya) with the MRCA of the Osteolaemus – Mecistops sister clade in the early Miocene and the Crocodylus radiation originating in the mid-Miocene.

The split between the two Mecistops lineages was estimated at ±7.4 mya, slightly after the estimated split between O. tetraspis and O. sp. nov. cf. tetraspis in West Africa (±8.6 mya), though the 95% HPD values for these two estimates overlap substantially.

Changing the rate smoothing parameter in r8s between 0.07 and 0.25 had no impact on estimated dates (results not shown), while extending the root age from 78 – 92 mya only slightly increased the estimated age of divergence for interior nodes (results only shown for 78 and 86 mya fixed root ages to be congruent with the 95% HPD of the

Bayesian estimates).

Morphological Analysis

Comparison of 103 skulls between the two genetically-defined West and Central

African lineages revealed the presence of at least 13 discrete, autapomorphic morphological characters distinguishing the lineages (Appendix B, Figs. 2-7 – 2-9). An additional 10 characters showed character state variation that was more frequent in one clade or the other, but both states were represented in each clade. Parsimony analysis of morphological characters recovered a single most parsimonious tree (Fig. 2-10). The unrooted cladogram supports two, reciprocally monophyletic clades that cluster following expectations based on the collection locality of each skull specimen.

Discussion

A growing body of evidence, particularly from the African taxa (e.g., Eaton et al.

2009, Hekkala et al. 2011), continues to support that crocodiles are not the

evolutionarily static entities long believed. The results presented here not only support previous studies that found significant lineage heterogeneity within the crown African

Crocodylus and Osteolaemus, but demonstrate that the genus Mecistops is also comprised of previously unrecognized species-level diversity. Total evidence-based analysis combining phenetic, character-based, and molecular phylogenetics support the presence of two, highly divergent lineages within the slender-snouted crocodile that last shared a most recent common ancestor in the late Miocene. Further, cranial morphology analyses revealed that the lineages can be unambiguously partitioned on the basis of discrete character variation. These results reject range-wide homogeneity in Mecistops, though only support the presence of two biogeographically defined lineages in the Upper Guinea and combined Congo – Ogooué basins. This finding clarifies historic, taxonomic uncertainties, provides a basis for further investigation into intra-lineage population process, and establishes a platform from which more effective conservation planning can be made for Mecistops.

This is the first study to show significant molecular and morphological divergence within the slender-snouted crocodile, and the strength of the evidence presented here does not warrant, or recognize, previously suggested subspecies status (e.g., Fuchs et al. 1974). My estimates of molecular divergence between the two Mecistops clades are congruent with the degree of molecular divergence between other recognized pairs of extant crocodilians (Table 2-4; also Eaton et al. 2009, Hekkala et al. 2011, McAliley et al. 2006, White and Densmore 2001). Further, fixed molecular and morphological character differences can be used to unambiguously diagnose the two Mecistops lineages (Brower 1999, Davis and Nixon 1992, Moritz 1994).

Neither the description of M. cataphractus nor M. leptorhynchus/bennettii included detailed discussion of the features distinguishing them as species, apart from perceived differences in the head length to head width ratio (Bennett 1835, Cuvier 1824). Despite this, I have demonstrated the presence of relatively many morphological characters distinguishing the two lineages within the genus. For example, multiple authors have reported finding only up to four cranial character differences separating Osteolaemus osborni from O. tetraspis, two species with significantly higher molecular divergence than seen in the two Mecistops clades (Brochu 2007, Eaton et al. 2009, Inger 1948).

Further work remains to determine the degree to which some of my identified characters represent continuums of variation; however, at least 13 of 23 identified characters represent discrete differences in cranial bone shape and configuration. Analysis of these features in skulls of unknown origin, for example, resulted in unambiguous morphological assignment.

Divergence times estimated for nodes within the crown Crocodylia generally agree with previously published results (e.g., Hekkala et al. 2011, Oaks 2011), and confirm baseline expectations for the MRCA between crocodilian sister taxa pairs and within crocodilian species. Any slight incongruence with dates estimated by similar methods are best explained by the inclusion of additional/different calibration points, mostly closer to the terminal nodes (Ho and Phillips 2009, Linder et al. 2005) and/or different molecular markers. I estimated the MRCA’s for virtually all intra-generic pairs of crocodilian sister taxa to the mid to late Miocene (±6 – 15 mya), while diversification within species (i.e., the genealogical MRCA) is dated to the late Pleistocene for taxa where multiple individuals were analyzed. This estimated divergence timing supports

the Miocene as the period of significant radiation not only for the Crocodylus (i.e., Oaks

2011), but for the other African genera as well, and establishes timelines of millions of years, compared to hundreds of thousands of years, as a reasonable expectation for crocodilian speciation across the Order (Avise and Mitchell 2007).

This study is the first to estimate divergence time between all pairs of established and putative crocodilians from the genera Mecistops and Osteolaemus. The MRCA between recovered Mecistops lineages was estimated to be 7.364 mya. This is congruent with the earliest fossil appearance of Mecistops in Africa stratigraphically dated to the Late Miocene (McAliley et al. 2006, Storrs 2003, Tchernov 1986).

However, these fossils are largely restricted to sites in Kenya and Egypt where the species is not contemporarily distributed; no fossil Mecistops have been found to date in

Central or West Africa.

While no samples were available for genetic analysis from Nigeria, these molecular divergence dates and morphological classification of skulls from Nigeria (this study; Baikie 1857) and Cameroon (this study) to the West and Central clades, respectively, coincide well with the formation of the Cameroon Volcanic Line (CVL) and/or the Benue Trough as the likely biogeographic barrier. Formation of the CVL began in the late Mesozoic but has continued into the Cenozoic with much of the middle volcanos and associated highlands (e.g., Mt. Cameroon, Bioko, Bambouto) not arising until the mid to late Miocene with final uplifts into the Pleistocene (Fitton and Dunlop

1985, Fitton 1987, Marzoli et al. 1999, Thouveny and Williamson 1988). The CVL is a significant geologic feature in the region (Cantagrel et al. 1978, Lee et al. 1994, Meyers

et al. 1998) acting as a zoogeographic barrier for other taxa (e.g., Fjeldså and Bowie

2008, Marks 2010), as well as the African Crocodylus complex (Hekkala et al. 2011).

The MRCA for the corresponding Osteolaemus clades (O. tetraspis – O. sp. nov. cf. tetraspis) was estimated to be 8.573mya, though the confidence intervals overlap considerably with that of the Mecistops divergence. While these two species pairs provide concordant phylogeographic support for Linder et al.'s (2012) Guinean and

Congolian biogeographic zones, patterns of intra-generic speciation between Mecistops and Osteolaemus are discordant within Central Africa. Mecistops showed no significant lineage diversification between the Ogooué and Congo Basins, while Central African

Osteolaemus shows deep divergence at around 12.45 mya supporting recognition of O. tetraspis and O. osborni as unique taxa. That speciation in Osteolaemus provides support for recognition of further division of Linder et al.'s (2012) Congolian biogeographic zone with recognition of the Gabon-Cameroon biogeographic zone may be related to the more direct dependence of Osteolaemus on forested regions and the centers of humid forest distribution in this region with aridity cycles.

The strength of the multiple lines of evidence presented here are several fold (de

Queiroz 2007). First, the multi-inferential approach to identification of lineage divergence within Mecistops, including the use of sequence data from both genomes, overcomes the risk of confounding organelle gene trees with species trees that are impacted not only by lineage isolation, but by other demographic processes like migration and changes in effective population size over time (e.g., Rosenberg and

Nordborg 2002). Second, support of these molecularly defined clades from fixed morphological characters should convince those skeptical of strictly molecular taxonomy

(DeSalle et al. 2005, Uilenberg et al. 2004). Finally, the ingroup samples utilized here cover the full extent of the geographic distribution of the slender-snouted crocodile and, additionally, included virtually all other extant members of the Crocodylia as outgroups.

This comprehensive sampling effort ensures that the confounding effects of incomplete taxon sampling are avoided (Zwickl and Hillis 2002). Additionally, the recovery of expected tree topologies congruent with other studies despite using different individuals as representatives of each species, and different genes regions, further supports the robustness of these results (Leaché 2009).

While it is true that some geographic areas are not represented by my sampling, these minor gaps will not likely impact the inferred phylogenetic structure (Rosenberg and Kumar 2001). Instead, additional genetic variation within these clades likely remains undiscovered that will be informative on the evolutionary origins and phylogeographic processes structuring populations within each of these taxa. For example, results comparing COI sequences from captive individuals to the known wild dataset demonstrated that all captive Mecistops originate exclusively from West Africa.

However, nearly half the individuals possessed a COI haplotype, differing at two positions, not found in the wild samples. While no documents exist confirming the wild origin of these captive animals, it has long been rumored that they originate from Liberia and, in fact, many of these individuals exhibit unique, almost melanistic phenotypes.

Incidentally, that all captive Mecistops in the USA originated in West Africa is also the reason why other studies focusing on the molecular systematics of Mecistops (i.e.,

McAliley et al. 2006), or utilizing multiple Mecistops individuals in their phylogenetic and biogeographic assessments of crocodilians (e.g., Oaks 2011), have not previously

uncovered the deep lineage divergence seen here. These previous studies exclusively sampled captive crocodilians for their research.

I highly recommend that additional sampling efforts are made throughout the

Benue and Niger drainages of Nigeria, as well as the forested drainages of Liberia,

Sierra Leone and Guinea, to facilitate better phylogeographic inference for M. cataphractus. For the new Central African Mecistops taxa, samples from the Lake

Mweru population in Zambia and the Malagarasi population in Tanzania (assuming they are not extinct), the Dja River in Cameroon, the upper Ivindo River in Gabon, and the

Uele drainage and central Congo River basins in the Democratic Republic of Congo would be most informative in explaining the general lack of intra-clade phylogenetic structure seen in this species.

By definition, cryptic species are those that are currently only identified as a single taxonomic unit where, in fact, two or more exist but they cannot be distinguished morphologically (Bickford et al. 2007). A recent review of the distribution of cryptic species across taxonomic groups found that cryptic species complexes are evenly distributed across taxa and suggested that “morphological stasis” may be a reasonable expectation even in the face of significant evolutionary change (Pfenninger and

Schwenk 2007). However, morphological variation thought to characterize a phenotypically diverse species can often be parsed into reciprocally monophyletic (even relatively “monotypic”) groups with the help of DNA evidence and in consideration of distribution-wide biogeography and species or population level ecology (e.g., Hebert et al. 2004). This notion is both congruent with the prevailing view of crocodiles as

morphologically static entities and supported by the divergent genetic diversity and cranial morphologies found in this study.

That these results are not the first to find highly divergent, cryptic lineages within members of the crown Crocodylia begs a shift in the crocodile conservation community away from a naïve view of species as conveniently binned entities subject to our limited perception of morphological divergence to recognition of species as benchmarks along the coalescent continuum (Cracraft 1989). This may not always result in recognition of unique crocodilian species from molecular evidence. For example, Caiman crocodilus and C. yacare show genetic signatures of either secondary hybridization or incomplete lineage sorting and, as a result, are not easily distinguished molecularly (Hrbek et al.

2008). In addition, they are even weakly differentiated on a limited morphological basis

(Busack and Pandya 2001). Similarly, Crocodylus intermedius and C. acutus, despite being readily distinguished morphologically, may not be genetically reciprocally monophyletic (Venegas-Anaya, unpub. data). Additionally, C. acutus has been shown to hybridize substantially with its sympatric congeners C. moreletii (Cedeño-Vásquez et al. 2008, Rodriguez et al. 2008) and C. rhombifer (Milián-García et al. 2011, Weaver et al. 2008) suggesting that species boundaries between these taxa may yet be porous

(Knowles and Carstens 2007).

These evolutionary scenarios illustrated across two highly divergent families of

New World crocodilians are in complete contrast to that seen in African taxa where all species complexes are completely reciprocally monophyletic and show no signs of hybridization in potential contact zones sampled to date (this study, Eaton et al. 2009,

Hekkala et al. 2010, 2011). In keeping with the age of taxonomic progress (Sangster

2009), the findings discussed here for African crocodilians, and other members of the

Crocodylia, should encourage the crocodilian systematics and, especially, conservation communities to reflect critically on how we are willing to define species within the

Crocodylia.

Table 2-1. Specimens used in phylogenetic analyses and their capture localities. No. is the maximum number of specimens sequenced from each site, though not all sequences were used for all analyses. Map Country Site Coordinates No. Label1 A The Gambia Gambia River NP N13.64207 W14.9608 5 B Cote-d’Ivoire Boubou River N5.30883 W5.21757 1 C Dipui Crocodile Farm2 N5.20454 W3.74519 3 D Goh River N5.23444 W6.64327 2 E Ghana Accra Zoo3 N5.70584 W0.21558 1 F Kumasi Zoo4 N6.70557 W1.61153 1 G Gabon Azingo River S0.49485 E9.99332 9 H Bongo River S2.52212 E10.09213 6 I Lac Divangui S1.94708 E9.98159 4 J Dji Dji River N0.05740 E12.70930 6 K Echira/N’gowe Rivers S2.20690 E9.70436 17 L Lac Evaro S0.90918 E10.10261 2 M Ogooué River5 S1.04988 E9.34485 1 N Banio Lagoon S3.72242 E11.12689 2 O Mbani River S2.51145 E10.57538 3 P Mpassa River S2.05323 E14.06599 9 Q Mpivie River S1.73650 E9.43047 6 R Nyanga River S3.00730 E10.35200 6 S Rabi River S1.90387 E9.51144 3 T Congo Lac Tele N1.37328 E17.15292 5 U CAR6 Sanagha River N2.482447 E16.2168 1 V DR Congo Kinshasa Reptile Park7 S3.54021 E16.09651 1 W Lomami River S2.68550 E25.07678 6 X Ituri River N1.49781 E28.67505 7 Y Lake Tanganyika8 S3.36735 E29.15647 2 1Corresponds to Figure 2-1. 2Specimens wild caught throughout the coastal lagoon system from Grand Bassam west to Abidjan (Lagoon Ebrie). 3Animal wild caught in Ghana, though of unknown specific locality, likely southern Ghana. 4Animal believed to have originated in the Subin River near Kumasi. 5Animal held at the Le Ranch Hotel in Port Gentil, of unknown specific origin but believed to have been wild caught in the Ogooué River between Port Gentil and Lambarene. 6Central African Republic. 7Captive held animal in Kinshasa of unknown origin, but believed to come from upstream the Congo River, perhaps area of Stanley Pool. 8Animals captive held at the Hydrobiological Research Station in Uvira of wild caught origin from the mouth of the Ruzizi River in Lake Tanganyika.

Table 2-2. Molecular markers used in this study. Gene N1 Size Primer Sequences Source Region (bp)2 5’  3’ 12S 107 838 GTTTAGTCCTAACCTTATTA This Study GTAAACATATTTCAGGTGTA Cyt-b 105 935 ACAATAATATTAATTCCATC This Study ACTTAGTCCTGTTAGAAAGA COI 106 922 TTTTATAGTAATACCAATCA This Study TATGTTTCAGAAAGAGTATG ND4 77 1073 ATAATATTAATTCCATCCAC This Study CTATGATTAAGTGTTCTCGT LDH-A 49 710 TGGCTGAAACTGTTATGAAGAACC Eaton et al. 2009 TGGATTCCCCAAAGTGTATCTG Rag-1 49 762 AGCCACAAGGAGATGGAAGGGAAA Eaton et al. 2009 TGGTCCACATCCATGCTTCTCACT CpP205 653 341 GTTTACCCATCATGTATTTTTAGA Miles et al. 2008 TTGGTTCCCTATTTCGATTCA This Study4 CpDi13 583 453 TTCCTGGCATATGAGGGTGT This Study4 CAGTCTCAGAGTATGCCTAGAA Miles et al. 2008 CpP4004 603 423 CACTGAATTGGGTGGAATAG Miles et al. 2008 AGCAGGGATCCCAGTTTTCT This Study4 CpP3303 573 629 GCTCATTGTCACCTCCCTGA This Study4 CCAGCTCTTTTGGCACTCTC CpP2902 722 682 TTCATTTCCTTGGTGTTAGTGC This Study4 GCAAATCAGGAGTTTGTCAGC 1Number of Mecistops sequences obtained for each gene fragment. 2Fragment size for M. cataphractus only, outgroup sequences may be of different length depending on the presence of indels. 3Sample size includes heterozygote alleles present in a single individual. 4Primer redesigned from C. porosus clone sequence developed by Miles et al. 2008.

Table 2-3. Slender-snouted crocodile specimens examined as part of the morphological analysis. Specimen Institution1 Collection Locality A priori Clade Number Assignment 39665 CM Cameroon Central 39664 CM Cameroon Central 166780 FLMNH Akaka, Gabon Central 166781 FLMNH Akaka, Gabon Central R54251 MCZ Kigoma, Tanzania Central 17962 RBINS Semliki River, DRC Central 17963 RBINS Semliki River, DRC Central 17964 RBINS Semliki River, DRC Central 17965 RBINS Semliki River, DRC Central 17966 RBINS Semliki River, DRC Central 17967 RBINS Semliki River, DRC Central 3302 RBINS Lake Mweru, DRC Central 4975 RBINS Lake Mweru, DRC Central 4976 RBINS Lake Mweru, DRC Central 4977 RBINS Lake Mweru, DRC Central 4978 RBINS Lake Mweru, DRC Central 4979 RBINS Lake Mweru, DRC Central 4980 RBINS Lake Mweru, DRC Central 4981 RBINS Lake Mweru, DRC Central 4982 RBINS Lake Mweru, DRC Central 4983 RBINS Lake Mweru, DRC Central 4985 RBINS Lake Mweru, DRC Central 4986 RBINS Lake Mweru, DRC Central 4987 RBINS Lake Mweru, DRC Central 4988 RBINS Lake Mweru, DRC Central 4989 RBINS Lake Mweru, DRC Central 4990 RBINS Lake Mweru, DRC Central 4991 RBINS Lake Mweru, DRC Central 4993 RBINS Lake Mweru, DRC Central 4994 RBINS Lake Mweru, DRC Central 4995 RBINS Lake Mweru, DRC Central 4996 RBINS Lake Mweru, DRC Central 4997 RBINS Lake Mweru, DRC Central 4998 RBINS Lake Mweru, DRC Central 4999 RBINS Lake Mweru, DRC Central 5000 RBINS Lake Mweru, DRC Central 5001 RBINS Lake Mweru, DRC Central 5225 RBINS Toa, Lake Tanganyika Central 5233 RBINS Albertville, DRC Central 6031 RBINS Lake Mweru, DRC Central

Table 2-3. Continued Specimen Institution1 Collection Locality A priori Clade Number Assignment 11385 RMCA Bukama, DRC Central 16119 RMCA Ubangui River, DRC Central 17078 RMCA Congo Central 182 RMCA Boma, DRC Central 219 RMCA Bas Congo Central 2955 RMCA Zambi, DRC Central 5403 RMCA Equateur Prov., DRC Central 5414 RMCA Nyonga Central 5416 RMCA Nyonga Central 5417 RMCA Nyonga Central 6882 RMCA Kununga Central 83-06-R-7 RMCA Kisangani, DRC Central 899 RMCA Lake Leopold, DRC Central 3287 RMCA Koteli, DRC Central 5408 RMCA Nyonga Central 5444 RMCA Congo Belge Central 5859 RMCA Koteli, DRC Central 10158 RMCA Ruki River, DRC Central R803 UA Congo Central R10074 AMNH Haut-Uele, DRC Central R10075 AMNH Belgian Congo Central R160902 AMNH Republic of Congo Central N/A Wild Ituri River, DRC Central 1900.2.27.1 BM Ogowe Central 1924.5.10.1 BM Lake Tanganyika Central 109154 FLMNH Captive2 West 145926 FLMNH Captive2 West 54119 Seckenberg Liberia West 83021 Seckenberg Liberia West 34091 Naturkunde West Africa West R22483 MCZ Liberia West 51734 Seckenberg Tappika, Liberia West 54119 Seckenberg Liberia West R107634 AMNH Liberia West CG55-27-09 Dakar Grand Lahou, Ivory Coast West N/A Wild Bia River, Ivory Coast West 86.12.30.2 BM Sierra Leone West 1904.9.9.2 BM Accra, Gold Coast West 32.6.30.8 BM Old Calabar West 1950.1.1.25 BM Sierra Leone West 1934.10.24.2 BM Nigeria West

Table 2-3. Continued Specimen Institution1 Collection Locality A priori Clade Number Assignment 1934.10.24.3 BM Nigeria West 64.4.6.1 BM West Africa Unknown 1947.3.6.35 BM C. leptorhynchus Type3 Unknown R29300 AMNH Unknown Unknown R57772 AMNH Unknown Unknown R75424 AMNH Cameroons Unknown N/A Dakar Unknown Unknown N/A Dakar Unknown Unknown 75-56-R-16 RMCA Lomie, Cameroon Unknown 5415 RMCA Unknown Unknown 43977 Seckenberg Bonambassi, Cameroon Unknown 47820 Seckenberg Unknown Unknown 36648 Naturkunde Unknown Unknown 36647 Naturkunde Loangi Unknown 37220 Naturkunde Unknown Unknown 28138 Seckenberg Unknown Unknown 1CM – Carnegie Museum of Natural History, Philadelphia, USA; FLMNH – Florida Museum of Natural History, Gainesville, USA; MCZ – Harvard Museum of Comparative Zoology, Cambridge, USA; RBINS – Royal Belgian Institute, Brussels, Belgium; RMCA – Royal Museum of Central Africa, Tervuren, Belgium; UA – University of Alberta, Alberta, Canada; Seckenberg – Seckenberg Museum of Natural History, Frankfurt, Germany; Naturkunde – Naturkunde Museum, Berlin, Germay; AMNH – American Museum of Natural History; Dakar – Specimen in the Dakar University Museum; Wild – Skull specimen never collected but seen at the listed locality. 2Specimen from a captive collection in Florida, USA of unknown origin but presumed West due to genetic typing of all other captive specimens in the USA. 3Specimen was a complete, stuffed juvenile crocodile. Morphological character assignment was done through xrays and characteristics visible through the skin.

Table 2-4. Estimates of average sequence divergence within and between groups. Shown are uncorrected p-distances with standard error estimated from 500 bootstrap replicates. Other intra-genera pairwise comparisons are shown to illustrate baseline expectations for divergence between closely related crocodilian species. Comparison1 12S ND4 COI Cyt-b CpP CpP CpP CpP Cp LDH-A RAG-1 205 2902 3303 4004 Di13 Within Central 0.001 0.0011 0.0005 0.0004 0.0000 0.0006 0.0000 0.0002 0.0013 0.0000 0.0000 ±0.001 ±0.001 ±0.0002 ±0.0002 ±0.000 ±0.001 ±0.000 ±0.0002 ±0.001 ±0.000 ±0.000 Within West 0.001 0.0027 0.0009 0.0026 0.0000 0.0006 0.0006 N/A 0.0000 0.0000 0.0000 ±0.001 ±0.001 ±0.001 ±0.001 ±0.000 ±0.001 ±0.001 ±0.000 ±0.000 ±0.000 Central - West 0.021 0.0787 0.0498 0.0550 0.0084 0.0059 0.0056 N/A 0.0047 0.0000 0.0000 ±0.004 ±0.008 ±0.007 ±0.008 ±0.005 ±0.003 ±0.003 ±0.003 ±0.000 ±0.000 O. osborni – 0.0666 N/A 0.1622 0.1060 0.0000 0.0929 0.0016 0.0047 0.0157 0.0068 0.0052 O. frontatus ±0.009 ±0.010 ±0.000 ±0.012 ±0.002 ±0.003 ±0.005 ±0.003 ±0.003 O. osborni – 0.0564 N/A 0.0859 0.1032 0.0064 0.0955 0.0056 0.0053 0.0164 0.0041 0.0052 O. tetraspis ±0.008 ±0.009 ±0.011 ±0.004 ±0.012 ±0.003 ±0.004 ±0.005 ±0.002 ±0.003 O. tetraspis – 0.0522 N/A 0.1002 0.0865 0.0042 0.0016 0.0072 0.0062 0.0034 0.0027 0.0026 O. frontatus ±0.008 ±0.010 ±0.004 ±0.002 ±0.003 ±0.004 ±0.002 ±0.002 ±0.002 C. niloticus – 0.0513 0.0640 0.0499 0.0808 0.0049 0.0071 0.0000 N/A 0.0091 0.0014 0.0022 C. suchus ±0.007 ±0.007 ±0.007 ±0.009 ±0.007 ±0.003 ±0.000 ±0.006 ±0.001 ±0.002 Intra – New 0.0237 0.0543 0.0400 0.0469 0.00293 0.00293 0.00163 N/A 0.00873 0.0022 0.0008 World Crocodylus ±0.004 ±0.005 ±0.005 ±0.005 ±0.003 ±0.002 ±0.002 ±0.005 ±0.001 ±0.001 Intra – Asian 0.0582 0.0907 0.0813 0.0941 0.00874 0.01044 0.00494 N/A 0.01424 0.0075 0.00654 Crocodylus ±0.006 ±0.006 ±0.006 ±0.007 ±0.005 ±0.004 ±0.003 ±0.008 ±0.002 ±0.004 P. palpebrosus – 0.0722 0.1184 0.0770 0.0898 N/A 0.0029 N/A N/A N/A 0.0043 N/A P. trigonatus ±0.012 ±0.010 ±0.008 ±0.010 ±0.002 ±0.003 Alligator – 0.2239 0.2964 0.1356 0.1957 N/A N/A N/A N/A N/A 0.0885 0.0158 Caiman ±0.014 ±0.014 ±0.010 ±0.014 ±0.010 ±0.005 1Intra – New World and Intra – Asian sequence divergences were estimated using the mean within group function, as for the within Mecistops group calculations. 2Values taken from Eaton et al. 2009, K2P model of sequence evolution. Distances may be disproportionately large because the sequenced region is not 100% congruent to that used in this study. 3The full assemblage of New World crocodilians was not sequenced for this marker; this distance is that between C. moreletii and C. niloticus which are in sister clades. 4The full assemblage of Asian crocodilians was not sequenced for this marker; this distance is that between C. porosus and C. suchus which are both the basal members of sister clades.

Table 2-5. Node divergence dates estimated by Beast and r8s. Dates are in millions of years before present. Mean and 95% HPD are Bayesian estimates derived in Beast and the MLE in r8s. # tMRCA Mean 95% HPD MLE1 MLE2 1 Alligatoridae – Crocodylidae 79.002 69.696 – 88.066 78 86 2 Alligatorinae – Caimaninae 63.312 60.112 – 67.606 62 62.124 3 Caiman – Paleosuchus 38.999 33.176 – 45.316 38.868 38.570 4 Paleosuchus 14.785 10.386 – 19.310 15.356 15.225 5 Crocodylidae 45.199 36.598 – 53.870 46.136 50.901 6 Tomistoma – Gavialis 21.75 15.325 – 28.573 24.078 26.630 7 Crocodylinae 26.592 22.044 – 31.346 24.562 27.081 8 Crocodylus 16.415 13.459 – 19.473 13.992 15.436 9 Crocodylus suchus – C. niloticus/New World 10.656 8.430 – 13.033 9.772 10.827 10 C. niloticus – New World 8.753 6.803 – 10.607 8.137 9.025 11 New World 7.717 5.941 – 9.578 7.225 8.0123 12 C. rhombifer – C. moreletii 6.588 4.904 – 8.406 6.373 7.0651 13 C. acutus – C. intermedius 1.606 0.948 – 2.289 1.455 1.613 14 C. porosus – Asian 14.758 11.973 – 17.665 12.382 13.645 15 C. palustris – Asian 13.534 10.905 – 16.326 11.352 12.499 16 C. johnsoni – Asian 11.915 9.308 – 14.572 9.958 10.948 17 C. mindorensis – C. novaguineae 6.128 4.384 – 7.970 4.776 5.236 18 Osteolaemus – Mecistops 21.670 17.398 – 26.006 20.018 22.060 19 O. osborni – O. tetraspis/O. sp. nov. West Africa 12.448 9.508 – 15.632 11.228 12.365 20 O. tetraspis – O. sp. nov. West Africa 8.573 6.038 – 11.222 8.028 8.862 21 Diversification within O. osborni 0.394 0.120 – 0.710 0.242 0.266 22 Diversification within O. tetraspis 0.119 0.013 – 0.261 0.071 0.079 23 Diversification within O. sp. nov. West Africa 0.158 0.016 – 0.348 0.088 0.097 24 Mecistops cataphractus – M. sp. nov. 7.364 5.194 – 9.658 6.647 7.339 25 Diversification within M. cataphractus 0.426 0.220 – 0.651 0.208 0.231 26 Diversification within M. sp. nov. 0.596 0.323 – 0.884 0.168 0.185 1MLE from r8s analysis fixing the root age to 78 mya. 2MLE from r8s analysis fixing the root age to 86 mya.

Figure 2-1. Map of sampling localities. The base map is colored to reflect topographic and elevation features of the landscape across the sampling distribution. Sample points are color-coded by corresponding clade (Blue – M. cataphractus (West), Green – M. sp. nov. (Central)) and labels correspond to localities detailed in Table 2-1.

Figure 2-2. COI barcoding gap for recognition of Mecistops species.

Figure 2-3. Cladistic network from CHA analysis. Two distinct, parsimonious haplogroups are evident, one representing all Central African samples and the other all West African samples, separated by 108 mutational steps. Wedge- colors represent unique sampling localities.

Figure 2-4. Phylogenetic relationships within the Crocodylia. Phylogenetic trees showing reconstructed relationships within the Order Crocodylia including individuals representing all putative biogeographic zones in the African crocodiles using a concatenated mtDNA + nDNA (top) and mtDNA only datasets (bottom). Numbers above branches are branch lengths. * at nodes represent >99% PP and >95% aLRT support for associated clades.

Figure 2-4 Continued.

Figure 2-5. Nuclear DNA phylogenetic relationships within the Crocodylia. Numbers above branches are branch lengths. * at nodes represent >98% PP and >90% aLRT support for associated clades.

Figure 2-6. Divergence date estimates within the Crocodylia. Mean dates are shown as node labels with blue bars representing the 95% HPD values. Branch labels are posterior probability support values for each node following the label.

Figure 2-7. Comparative cranial morphology of ventral view of Mecistops from Central and West Africa. Labeled characters correspond to character descriptions in Appendix B.

Figure 2-8. Comparative cranial morphology of dorsal view of Mecistops from Central and West Africa. Labeled characters correspond to character descriptions in Appendix B.

Figure 2-9. Comparative cranial morphology of occipital view of Mecistops from Central and West Africa. Labeled characters correspond to character descriptions in Appendix B.

Figure 2-10. Unrooted parsimony cladogram of Mecistops based on morphological character assignment. Specimens in the green clade were collected in West Africa, while specimens in the black clade were collected in Central Africa.

CHAPTER 3 COMPARATIVE SPATIAL STRUCTURE OF MECISTOPS SP. NOV. AND OSTEOLAEMUS TETRASPIS REVEALED BY MITOCHONDRIAL AND NUCLEAR MARKERS

Introduction

The contemporary distribution and structure of intraspecies genetic diversity is shaped by the joint effects of dynamic historic processes (e.g., range shifts accompanying climate change and habitat cycling) and intrinsic species characteristics

(e.g., vagility, fecundity, lifespan). Understanding the joint effects of history and species-specific traits on intra-specific lineages forms the basis for interpreting patterns of population connectivity and historical processes (e.g., population and/or geographic expansion/contraction) at various scales (Avise 2009). In addition to better understanding of the processes driving population genetic structure, examining intraspecific lineages can facilitate species and population conservation through the delineation of operational taxonomic units (Leaché 2009), Evolutionary Significant Units

(ESU’s) and Conservation Units (CU’s) (Crandall et al. 2000, Moritz 1994, 2002).

Phylogeographic analyses of wide-ranging species provide a unique opportunity to assess the impact and timing of multiple geographic features, or other landscape processes like climate change, on spatial genetic structure (e.g., Burbrink et al. 2008).

Comparative phylogeography offers a framework through which genealogical and spatial concordance detected in separate species reflect the concurrent effects of geographic features and climate change in structuring regional biotas (Avise 2000).

Discordant multilocus genealogical patterns, or geographical partitioning of lineages, suggest that differing natural history strategies (e.g., habitat selection, dispersal behavior), ecological niche partitioning, or shifts in niche distribution due to paleoclimate

change may be more important than strictly geologic features on interspecies phylogeographic structure (Carnaval et al. 2009, Carstens and Richards 2007).

Crocodiles of the genus Mecistops and Osteolaemus make appropriate, and unique, study systems for examining comparative phylogeography of western Africa.

First, they are largely sympatric throughout the region. Despite this, these two species have already been shown to have slightly divergent evolutionary histories where

Mecistops is comprised of two lineages distributed in West and Central Africa (Chapter

2), while Osteolaemus is comprised of three lineages distributed in the Congo, Lower

Guinea, and Upper Guinea bioregions (Chapter 2, Eaton et al. 2009, Oaks 2011).

Second, vicariant landscape features, including the development of the Cameroon

Volcanic Line (or Benue Trough and Nigerian plateaus) kept West and Central African

Mecistops and Osteolaemus lineages in allopatry since ± 8 mya (Chapter 2). Third, both species groups have similar ecological traits that include dependence on forest canopy cover for nesting, and both lay small clutches and demonstrate the least parental care in the Crocodylia (Eaton 2010, Shirley 2010, Thorbjarnarson 1996).

Fourth, both species are believed to have been historically widespread and abundant, while contemporary populations in much of their ranges are locally fragmented due to hunting and habitat loss (Eaton 2010, Shirley 2010). Finally, within Central African

Mecistops and Osteolaemus lineages there is very little genetic diversity or spatially- explicit hierarchical structure (Chapter 2, Eaton et al. 2009) suggesting high levels of population connectivity.

In addition to these five characteristics, Mecistops and Osteolaemus also have trait differences that may shape their phylogeographic patterns. For example, while

both species have similar nesting habitat needs non-reproductive Mecistops seem less dependent on closed canopy forest habitats than do Osteolaemus. Disparate dependence on forest habitats may result in discordant population histories or population connectivity across heterogeneous landscapes as a result of changes in forest cover over time (Eaton 2010, Shirley 2010). In contrast, Mecistops seems to actively avoid large coastal lagoon networks and saline conditions while Osteolaemus is often distributed continuously throughout these habitats (M. Shirley, unpub. data).

In this chapter I use a multi-marker, multi-analytic approach to compare the phylogeographic and spatial patterns of genetic diversity in Central African Mecistops and O. tetraspis. I used Nested Clade Phylogeographic Analysis of mitochondrial sequences to identify hypotheses of lineage history for both species that are then empirically tested under expectations of lineage structure derived from coalescent and neutral theory. I then use multilocus microsatellite data to identify the presence and positioning of explicit crocodile clusters on the landscape.

Methods

Taxon and Character Sampling

Analyses in this chapter deal with both Mecistops and Osteolaemus individuals, utilizing samples from throughout Central Africa for Mecistops (Table 3-1, Fig. 2-1) while only O. tetraspis specimens from Gabon were utilized for Osteolaemus (Table 3-2, Fig.

3-1). While the geographic scale of sampling between these two species is different, it allows two objectives to be accomplished. First, all analyses for both genera are conducted at the intra-species level; inclusion of broader samples for Osteolaemus would introduce interspecific variation within this genera confounding results. Second, it allowed intraspecies analyses to be conducted at the full geographic extent of the

sampling for each species. Comparisons between the two genera were additionally accomplished through repeated analyses following the protocols below utilizing only samples from both species that fell within the geographic extent of co-sampling (i.e., the

Ogooué Basin). All crocodiles were captured in the wild following the methods detailed in Chapter 2; no captive crocodiles were used in these analyses.

Mitochondrial sequences. Generation of Mecistops sequences (12S, cytb, COI,

ND4) was described in Chapter 2. Osteolaemus individuals were sequenced for partial

COI and cytb-control regions using the primer sequences developed by Eaton et al.

(2009) though following the PCR and sequencing protocols described in Chapter 2.

Additional sequences for individuals not collected as part of this study were provided by

M.J. Eaton. Sequences were concatenated for each individual and utilized as a single locus due to the non-independence of mtDNA loci. Diversity statistics π (Nei 1987) and

θS (Watterson 1975) were calculated at the species level using Arlequin v3.5 (Excoffier and Lischer 2010).

Microsatellites. Two hundred and sixteen Mecistops and 225 Osteolaemus individuals were genotyped for the same 20 microsatellite markers as follows. I screened microsatellite loci developed from C. porosus (Miles et al. 2008) that were previously shown to cross-amplify in Mecistops and Osteolaemus (Miles et al. 2009).

Ultimately, only 15 and 12 consistently amplified and proved informative for use with

Mecistops and Osteolaemus, respectively (Table 3-3). Microsatellites primers were

M13-tagged for annealing of the fluorescent dyes FAM and HEX (Schuelke 2000) and

PCR-amplified via the protocol described in Chapter 2. PCR products were run on an

ABI 3130 Genetic Analyzer (Applied Biosystems) using Rox 400HD size standard, or

Rox 500 for markers whose allele sizes exceeded 400 bp. Allele scoring was done in

Genemarker® software (Softgenetics, State College, PA).

Evidence of null alleles and scoring error due to stutter or large allele dropout was examined using Micro-checker version 2.2 partitioning the data by population (i.e., sampling locality or aggregates thereof; van Oosterhout et al. 2004). Scoring error was additionally verified by randomly re-amplifying 8 – 29% of individuals for all markers to confirm allele morphology and scoring (Hoffman and Amos 2005). Genepop version 4

(Rousset 2008) was used to test for linkage disequilibrium among pairs of loci using

Fisher’s global test and heterozygote deficiency and excess using exact tests.

Significance was adjusted using posthoc sequential Bonferroni tests (Rice 1989).

Summaries of allelic diversity and expected (He) and observed (Ho) heterozygosity for each locus grouped by species were conducted in Genalex v6.5 (Peakall and Smouse

2012).

Data Analysis

Phylogeographic analysis. Phylogeographic analyses were conducted under two complementary paradigms: Nested Clade Phylogeographic Analysis (NCPA;

Templeton et al. 1995) and statistical phylogeography (Knowles and Maddison 2002).

NCPA was a widely-adopted means of detecting phylogeographic structure after its introduction by Templeton et al. (1995). However, the method recently came under significant scrutiny due to a perceived lack of rigor and falsifiability, and for a perceived lack of hypothesis testing capacity (e.g., Beaumont and Panchal 2008, Knowles and

Maddison 2002, Petit 2008). Despite these criticisms, NCPA can be of utility in objectively generating phylogeographic hypotheses for further testing. For example, inferred patterns of isolation-by-distance or incomplete lineage sorting can then be

tested for using model-based approaches (Garrick et al. 2008). Statistical phylogeography has been proposed as a robust analytical framework for testing hypotheses of phylogeographic structure generated by NCPA and other means

(Knowles and Maddison 2002).

NCPA was implemented in ANeCA v1.2 (Clement et al. 2000, Panchal 2007,

Posada et al. 2000) using the full mtDNA dataset for both species. Creation of the haplotype network followed the protocol outlined in Templeton et al. (1992) using TCS

(Clement et al. 2000) with the network connection limit set to a minimum 95% connection parsimony probability. Ambiguities in the haplotype network were resolved following the parsimony criteria suggested by Crandall and Templeton (1993): 1) rare haplotypes are more likely to be found at tip positions (i.e., haplotypes with only one connection to another haplotype) and common haplotypes are more likely to be positioned at interior (i.e., more than 1 connection) nodes of a network; and 2) a haplotype represented by a single individual is more likely to be connected to haplotypes from the same population than to haplotypes from different populations

(Mardulyn 2001). The resolved haplotype network was used for the remainder of the

NCPA steps. Clade nesting (Templeton and Sing 1993, Templeton et al. 1987), statistical tests of geographical and clade association (Posada et al. 2000, Posada et al.

2006, Templeton et al. 1995), and the inference key (06 Jan 2011 version) were run automatically in ANeCA (Panchal 2007), and results were confirmed manually. In a series of rigorous simulation studies, Salzburger et al. (2011) found that TCS performed relatively poorly at recovering true genealogical networks. In order to confirm the TCS network and ambiguity resolutions, I generated genealogical networks using the method

of maximum parsimony implemented in dnapars of the Phylip-3.69 package

(Felsenstein 1989, 2005) with haplotype networks reconstructed from the most parsimonious trees using Haploviewer (Ewing 2011).

In addition to NCPA analyses, for Osteolaemus I ran two tests on sequence data to detect evidence of past demographic and spatial change. Pairwise mismatch distributions were used to compare the observed pairwise nucleotide site differences with distributional expectations for both demographically and spatially expanding populations. The distribution is usually multi-modal if the population is in demographic or spatial equilibrium but unimodal under expansion (Excoffier and Lischer 2010,

Harpending 1994, Schneider and Excoffier 1999). Goodness of fit to the simulated mismatch distributions were assessed through both the sum of squared deviations

(SSD) and Harpending’s raggedness index (Harpending et al. 1998). These results were compared with Fu’s Fs (Fu 1997) and Tajima’s D (Tajima 1989a, 1989b) which both assess the probability of a population being at neutral mutation-drift equilibrium.

Under assumptions of selective neutrality, deviations from zero are taken as indication of demographic change. The above tests were implemented in Arlequin v3.5 (Excoffier and Lischer 2010) on samples grouped by haplogroup and species.

I used coalescent simulations in Mesquite v2.75 (Maddison and Maddison 2011) to determine if a widespread, fragmented ancestor population model better fits the dat,a as suggested by both the phylogenetic analysis (Chapter 1) and NCPA, than a series of other more structured models for Mecistops (Fig. 3-2). The fit of my maximum likelihood generated mtDNA gene tree into population tree models was determined using the number of deep coalescences (nDC). The nDC test statistic measures the fit

of a gene tree into a population tree model by counting the number of extra-population lineages in each proposed population as a result of stochasticity in the coalescent process. Coalescent theory predicts that the process of genetic drift will eventually restrict genealogical lineages to a single population (i.e., nDC = 0) in a process known as lineage sorting (Maddison 1997, Knowles and Maddison 2002). Deviations from nDC = 0 are assumed to be due to incomplete lineage sorting making it an ideal test statistic for my system (Maddison 1997, Knowles and Maddison 2002).

Population models hypothesize relationships between proposed aggregations of individuals (i.e., “populations”) that correspond to, in this case, geographical groupings.

The null hypothesis is the Fragmented Ancestor model in which all contemporary population divergences were effectively concurrent in the past and the result of geographical sorting (i.e., isolation-based fragmentation) of a widely distributed common ancestor. Alternative, more complex population models considered hierarchical and/or nested divergence, as well as isolation by distance, relationships amongst populations.

Populations were defined in two alternate manners – by sampling locality (n = 12), which largely reflect river basins, and by haplogroup partitioned by broad-scale geographic basin (n = 6) (Fig. 3-2). In addition to geographic structure, population models varied clade depths (i.e., time in generations) to test different hypotheses of time since lineage diversification/fragmentation. Assuming a generation time of 15 years, I tested tree and clade depths ranging from 10 to 39,000 generations which corresponds to roughly 85 – 585,000 YBP. This time range tests lineage sorting as a result of recent, anthropogenically-driven population fragmentation, fragmentation corresponding to Pleistocene and Holocene climate events (Chapter 4 contains

additional discussion), and the age of first diversification within the Central African

Mecistops lineage estimated in divergence timing results detailed in Chapter 1.

Branch widths (i.e., Ne) for each sample grouping were inferred from θ values calculated in Migrate-n version 3.3 (Beerli 2009) with the following parameters: one long and eight short, heated chains (utilizing the automated heating procedure implemented in migrate-n) for 15,000,000 generations sampled every 50th generation following a burn-in of 100,000 generations. Transition/transversion ratios (R) used in the sequence model were estimated in MEGA5 (Tamura et al. 2011). Migrate-n additionally estimated an averaged migration rate (M) amongst population pairs using a migration matrix that approximately followed a step-wise migration pattern where migration was only allowed between populations where it was deemed biologically and geographically likely.

Bayesian estimates of θ and M were achieved from four simultaneous, independent runs to ensure convergence upon similar values (Beerli and Felsenstein 2001, Beerli

2006). I converted the total θ to Ne using the equation for mtDNA, θ = Neµ, where µ =

2.28 x 10-6/site/generation calculated in Beast v1.7 assuming a generation time of 15 years for Mecistops (Chapter 4). I summed the Ne for all areas to calculate Total Ne and then scaled the branch widths of my hypothesized population models using the proportion of Total Ne comprised by each population. Internal branches on the various structured models were scaled such that the sum total branch widths at any given point in time equaled the Total Ne (i.e., 1) between coalescent events (Pyron and Burbrink

2009, Carstens et al. 2005). The average probability of migration was estimated by averaging the proportion of Ne for each population comprised by migrants (both immigrants and emigrants) for all populations.

Using the Bayesian estimate of Total Ne and the average probability of migration from Migrate-n, I simulated 500 gene trees under a neutral coalescent process both with and without migration to generate an expected normal distribution for nDC under each population model. Both the ML gene tree and simulated gene trees were assumed unrooted for fit into each population model. The most likely population model was chosen in consideration of two criteria. First, the population tree model was only considered if the nDC calculated from the observed genealogy (maximum likelihood gene tree) for a given population model fell within the 95% confidence interval of the simulated nDC normal distribution (Knowles and Maddison 2002). Second, amongst models that fit nDC expectations, the most likely model should be the one that minimizes the number of deep coalescences in the observed genealogy because as nDC approaches zero lineage sorting approaches completeness and, therefore, this model best reflects the actual lineage sorting process (Maddison 1997).

Spatial clustering of microsatellite genotypes. Two different methods were used to determine the number and geographic distribution of genetic clusters from microsatellite data. I implemented the Bayesian clustering algorithm employed by

STRUCTURE v2.3.3 (Pritchard et al. 2000) to search for genetic clusters without a priori geographic information bias. This type of Bayesian clustering attempts to identify natural groupings of individual multilocus genotypes by arranging samples into K clusters in a way that minimizes deviations from Hardy-Weinberg Equilibrium and linkage disequilibrium. I ran the admixture model, setting the program to infer the alpha admixture parameter from a uniform U(10.0, 0.025) prior distribution and a starting value of 1.0 (François and Durand 2010). I implemented the correlated allele frequency

model with prior FST mean 0.01 and s.d. 0.05 (Falush et al. 2003). I did not use sampling location priors. For each analysis I conducted 20 independent replicate runs for each a priori assumed number of clusters (K) where K-values varied from 1 – 15 depending on the subset of samples analyzed. Each run consisted of a 2.0x105 burn-in steps followed by 1.0x106 post burn-in replicates.

Results were analyzed with STRUCTURE HARVESTER (Earl and Von Holdt

2011) which implements the Evanno method of detecting the number of clusters

(Evanno et al. 2005). The ΔK method finds the breakpoint in the slope of the distribution of deviation information criterion scores to infer K, however, it may be unreliable for K = 1 clusters or where multi-modality in log likelihood scores makes selection of K from ΔK difficult. Therefore, bar plots of individual Q-values from the chosen K were visually compared to bar plots from other K-values and the final most likely number of K clusters was chosen combining the ΔK method and consideration of crocodile species ecology and the landscape. Cluster matching from each independent run replicate for relevant K-values was conducted in CLUMPP v1.1.2 (Jakobsson and

Rosenberg 2007) and bar plots constructed in DISTRUCT v1.1 (Rosenberg 2003).

STRUCTURE analysis was run on the full sample set for both species, and then on the

Mecistops dataset excluding West African samples.

Analysis using STRUCTURE makes basic assumptions about population models including Hardy-Weinberg and linkage equilibrium to identify clusters (Jombart et al.

2008, Pritchard et al. 2000, Song et al. 2009). Additionally, STRUCTURE may not resolve the true number of clusters accurately outside of basic island or slightly hierarchically structured models (Jombart et al. 2010). Unfortunately, these may not be

reasonable assumptions for threatened species persisting in small, fragmented populations or when the species was formerly fairly widespread with well-connected populations as in a stepping stone model or clines (Jombart et al. 2008). Because of this, I implemented two secondary analyses that strip away equilibrium assumptions and instead focus on spatial patterns of allele distribution as multivariate datasets

(Jombart et al. 2008, Lee et al. 2009). Additionally, multivariate methods have been proposed as independent inference methods for verifying population structure seen in

Bayesian clustering methods (François and Durand 2010).

Principal components analysis (PCA) was first applied to genetic data over three decades ago (Menozzi et al. 1978) and has developed as a robust method for studying structure via summarizing genetic variability between individuals and populations (e.g.,

Hanotte et al. 2002, Patterson et al. 2006). PCA is a multivariate ordination method that seeks to explain variability in a dataset by reducing that variability to as few explanatory vectors (i.e., principle components) as possible. Interpolation of principal component scores onto georeferenced surfaces can then be used to infer spatially-explicit patterns of genetic variation via synthesis mapping (Hanotte et al. 2002, Manel et al. 2003).

However, this technique may not be wholly appropriate in the search for spatially- explicit genetic variability as it seeks to maximize the variance among genotypes regardless of spatial distribution. In response to this, Jombart et al. (2008) proposed the spatial principle components analysis (sPCA) which modifies PCA such that principal component scores must be spatially auto-correlated. I implemented sPCA in the adegenet package (Jombart 2008) of R (R Core Team 2012). The k-nearest neighbor algorithm was implemented to form the connection network between individuals using

half the total sample size as k. While utilizing a large k-value can result in highly globalized individual classifications that fail to distinguish localized patterns, this may be preferable to highly localized individual classifications that infer spurious small-scale patterns. The number of eigenvalues retained was determined by visually plotting the relative influence of global and local structure and the number of eigenvalues to be interpreted was determined from a decomposition plot of the spatial and variance components of eigenvalues (Jombart 2012a). Lagged spatial principle component scores were mapped to visually assess patterns. sPCA analysis was run on the full O. tetraspis dataset but only on the Central African Mecistops dataset due to small sample sizes of western individuals.

While sPCA may adequately describe spatial patterns in genetic diversity, it may not adequately distinguish discrete clusters as it focuses on describing global levels of genetic variability (i.e., both the within and between group components). Instead, emphasis should be placed on maximizing between-group variation while minimizing within-group variation, as is achieved through the multivariate discriminant analysis

(Lachenbruch and Goldstein 1979). I implemented the Discriminant Analysis of

Principle Components (DAPC) to identify clusters maximizing between group variation

(Jombart et al. 2010). DAPC utilizes a K-means clustering procedure when group priors are unknown and utilizes Bayesian Information Criterion (BIC) to assess the best supported K-cluster model (Jombart et al. 2010). DAPC was implemented in the adegenet package (Jombart 2008) of R (R Core Team 2012) and up to 15 clusters were searched. At the PCA stage all principle components were used to ensure the maximum variability in the dataset was retained, but at the dimension reduction step I

used the α-score optimization procedure to identify the number of principal components that best describes the variation in the data without over-parameterizing the model

(Jombart 2012b). One risk of DAPC, like all spatial clustering algorithms, is the potential for inferring spurious discrete groups where genetic diversity is continuously distributed. Therefore, I visually analyzed scatterplots and cluster assignments against sampling localities for all K-values that were both within 3 BIC points (Kass and Raftery

1995) and smaller than the K selected by the graphical method outlined by Jombart

(2012) and Jombart et al. (2010). DAPC analysis was run on the full Mecistops dataset, the Central Africa-only dataset, and the full O. tetraspis dataset.

Results

Mitochondrial Sequences and Phylogeographic Analyses

Sequences from 96 Mecistops covering four, concatenated mitochondrial gene regions for a total 3,768 bp were utilized in these analyses. Only 14 unique haplotypes were identified despite the specimens covering a geographic distribution over ± 2,200 km. The geographic distribution of haplotypes loosely followed expectations with geographically isolated haplogroups discovered in the Gabon coast south of the

Ogooué River, Lomami River, Lac Tele, and Lake Tanganyika. Geographically mixed haplogroups covered the main Ogooué drainages east to the Ituri Forest and the northern tributaries of the Congo River (Fig. 3-3). At the species level, mean nucleotide diversity (π) was 0.000845 (±0.00048) and θS was 6.0354 (±1.798). Genetic diversity using only the two markers congruent with Osteolaemus (i.e., cytb and COI) was lower.

Mean nucleotide diversity (π) was 0.000635 (±0.00047) and θS was 2.92 (±1.016).

Genetic diversity using only the two markers (i.e., cytb and COI) and sampling

distribution (i.e., Ogooué Basin) congruent with Osteolaemus was even lower still.

Mean nucleotide diversity (π) was 0.000371 (±0.00032) and θS was 1.03 (±0.514).

Sequences from 166 Osteolaemus covering two, concatenated mitochondrial gene regions (1,494 bp) were utilized in these analyses. Only 21 unique haplotypes were discovered despite the distribution of specimens covering a linear geographic distribution of ± 700 km. The spatial distribution of haplotypes did not follow geographic expectations and exhibited a couple of unusually common, widely distributed haplotypes with a few secondary haplotypes, few of which were geographically restricted (Fig. 3-4).

At the species level, mean nucleotide diversity (π) was 0.003964 (s.d. 0.00211) and θS was 4.0492 (s.d. 1.213).

Nested clade phylogeographic analysis. The genealogical networks inferred in

TCS as part of the NCPA analysis for both species were highly reticulate (i.e., unresolved) networks of haplotypes each connected to each other, with few exceptions, via 1 – 2 steps (Fig. 3-5). However, the manually resolved networks were congruent with those generated by the method of Salzburger et al. (2011) indicating it was appropriate for continued inference (Fig. 3-6 and 3-7).

For Mecistops, two, high-frequency, widely distributed internal haplotype nodes were linked to each other by a single mutational step and formed six sub-networks; though only three showed significant clade-geography association values (Fig. 3-6).

Within these three sub-networks, the null hypothesis of no geographic relationship amongst haplotypes from sampled populations could not be rejected, which is likely the result of panmixia in sexually reproducing species. Only three second-order clades resulted from nesting: one grouping all coastal Gabon samples south of the Ogooué

River, one for the Lake Tanganyika samples, and a third grouping all other samples from the Ogooué, Sanagha, Congo, and Lomami drainages. The latter clade demonstrated significant geographic-clade relationship supporting restricted gene flow with isolation by distance as a structuring mechanism while the former two clades could not reject the null hypothesis of no geographic-clade structure. At the level of the total cladogram, Mecistops exhibited either restricted gene flow/dispersal over intermediate areas not occupied by the species or past gene flow followed by extinction of intermediate populations.

For Osteolaemus, a single star-like pattern (i.e., a centrally-located common ancestral haplotype to which recently derived haplotypes are each connected) was evident in which eight sub-networks were linked to each other by a single mutational step (Fig. 3-7); though only six showed significant clade-geography association values.

Within these six sub-networks, the null hypothesis of no geographic relationship amongst sampled populations, which is thought to be predominantly caused by panmixia in sexually reproducing species, could not be rejected. Four second order clades resulted from nesting, though none of the clades contained strict geographic partitioning: one corresponding to most of the samples excluding individuals at one coastal site (Petite Loango), and the other three comprised of crocodiles off the Fernan

Vaz lagoon and random other individuals from Lope and Petite Loango. The former showed support for contiguous range expansion while the latter three could not reject the null hypothesis. At the level of the total cladogram, the outcome was inconclusive though there was weak support for past range expansion.

At the level of the species both Tajima’s D (-0.37510, p < 0.0001) and Fu’s Fs (-

25.9855, p < 0.0001) were significantly negative indicating population expansion, however, based on raggedness index and sum of squared deviations (SSD) mismatch distribution was not significant for either the demographic expansion (r = 0.083, p =

0.777; SSD = 0.0213, p = 0.07) or spatial expansion models (r = 0.083, p = 0.87; SSD =

0.0054, p = 0.803). When the sample was broken into haplogroup, Petite Loango

(Tajima’s D = -0.9235, p = 0.025, Fu’s Fs = -3.4E38, p < 0.000001), all other coastal samples (Tajima’s D = -0.26270, p < 0.00001, Fu’s Fs = -26.226, p < 0.00001), Abanda

(Fu’s Fs = -26.05, p < 0.000001), and the Ogooué (Tajima’s D = -1.9667, p < 0.00001,

Fu’s Fs = -3.4E38, p < 0.000001) all showed significant signs of population expansion.

However, like results for the combined dataset, mismatch distributions for sudden demographic (mean r = 0.164, p = 0.78, mean SSD = 0.1368, p = 0.224) and spatial expansion (mean r = 0.164, p = 0.6955, mean SSD = 0.0091, p = 0.521) were not significant.

Statistical phylogeography and coalescent simulations. For Mecistops, significantly fewer deep coalescent events were needed to fit the observed gene tree into the Fragmented Ancestor population model irrespective of Ne and clade/tree depth than models that assumed some sort of geographic structure. In addition, the

Fragmented Ancestor assuming six populations (Gabon coast, Ogooué, western

Congo, Ituri, Lomami, Tanganyika) had fewer nDC’s than that assuming 12 populations

(5 independent coastal sites, 3 independent Ogooué sites, western Congo, Lomami,

Ituri, Tanganyika). Mean expected nDC decreased with increasing tree depth (i.e., time) but increased with the inclusion of migration for any given tree depth, as expected

from coalescent theory. Mean expected nDC also increased with increased population structure and no models with migration ever fit the expected normal distribution for nDC and, therefore, migration models were rejected in favor of non-migration models (Table

3-4).

While the Fragmented Ancestor is the best model under both selection criteria, several other population models showed observed nDC values that fell within the 95% probability limits of the expected normal distribution from coalescent simulations. This may indicate some level of support for parameters that were congruent amongst these models and with the Fragmented Ancestor model, notably tree depth and geographic structure secondary to initial lineage sorting. All models with support under acceptance criteria one (i.e., fit between the observed and simulated test statistic), as well as the best Fragmented Ancestor model, had tree depths corresponding to 150 – 300 generations. Assuming a 15 year generation time, this suggests that the lineage sorting process began in earnest 2,200 – 4,500 YBP. Additionally, all secondary models provided some support for east versus west geographic structure, though there was no objective support for any particular hierarchical population relationship in these models.

Microsatellite Genotypes and Spatial Clustering Analysis

One hundred and ninety six individual Mecistops were successfully genotyped for

15/15 markers, while the remaining 20 individuals were genotyped for 14/15 (n = 15),

13/15 (n = 4), and 12/15 (n = 1) microsatellites (Tables 3-3, 3-5). There was evidence for possible null alleles in locus CpP1610 for half of the populations and CpDi21 for two of the populations. Re-amplification and rescoring a total 1022 (15.9% of 6428 possible) alleles resulted in 14 alleles incorrectly scored during the first scoring –

0.002% error. Allele scoring errors were restricted to four of the 15 markers (including

CpP1610) and were likely related to null alleles given that all rescores resulted in former homozygotes becoming heterozygotes. Marker CpP1610 was removed from all subsequent analyses.

One hundred and twenty five individual O. tetraspis were successfully genotyped for 12/12 markers, while the remaining 100 individuals were genotyped for 11/12 (n =

57), 10/12 (n = 22), and 9/12 (n = 12). All individuals genotyped for less than 9/12 markers were discarded from further analysis (Tables 3-3, 3-5). Re-amplification and rescoring a total 247 (5% of 5028 possible) alleles resulted in 24 alleles incorrectly scored during the first scoring – 0.4% error. Allele scoring errors were restricted to six of the 12 markers and were likely related to null alleles given that most rescores resulted in former homozygotes becoming heterozygotes.

Bayesian clustering analysis. Bayesian clustering of multi-locus microsatellite data in STRUCTURE produced a single best ΔK (K = 4) for the global Mecistops dataset (Fig. 3-8). The clusters correspond to expectations from both phylogenetic and mtDNA-based phylogeographic analyses identifying a West African cluster, a Gabonese coastal cluster, an Ogooué basin cluster and a Congo basin cluster. Individuals from the Minkané River area of the lower Ogooué showed ancestry from both the Ogooué

River and coastal clusters, which may indicate a cline as this area is halfway between the Ogooué Delta, which feeds into the northern most lagoons of the coast region, and the tributaries of the upper and middle Ogooué. Analysis of the West Africa-only dataset diagnosed K = 2 clusters with individuals sampled in the Upper Guinea forest zone and Senegambia basins clustering separately (results not shown). While this corresponds to results from the phylogenetic analyses discussed in Chapter 2 (i.e.,

possible ESU in the Senegambia area), the results should be viewed cautiously until further samples are available. Analysis of the Central clade resulted in ΔK of K = 3 clusters (Table 3-6, Fig. 3-9) corresponding to the same three Central African clusters identified in the results of the full analysis (Fig. 3-8). However, examination of the K = 4 and K= 5 show clustering patterns that suggest samples from the upper, middle and lower Ogooué may represent a cline with the Congo Basin where group assignment is inconsistent. Analysis of the Ogooué-only subset of samples resulted in recognition of the same clusters as the full Central dataset (results not shown).

Analysis of O. tetraspis multilocus genotypes also recovered a single best ΔK (K =

4) (Table 3-6, Fig. 3-10). The four recovered clusters show loose geographic affiliation with clusters of individuals from the middle Ogooué, Abanda, and two clusters forming a

Q-value cline from north to south along the coast suggesting a pattern of slight isolation by distance though with gene flow through the intermediate locales. For all K-values both the middle Ogooué and Abanda populations were continuously identified as largely

“pure” with admixed individuals likely the result of unsampled intermediary populations for the Ogooué or actual migrants in the case of Abanda. Increased values of K (results not shown) simply increased admixture proportions within samples without identifying new, or spatially-explicit, groups.

Multivariate clustering analysis. In the sPCA analysis of Mecistops multilocus genotypes a single global component appeared to be most important in interpreting the spatial pattern of genetic variation, though a second global component may illustrate a secondary pattern (Fig. 3-11). Interpolation of the lagged component 1 spatial principal component scores showed three zones of differentiated allele frequencies

corresponding to the Gabonese coast, the Ogooué drainage, and the Congo Basin, which actually appeared to have a cline in allele frequencies from west to east (Fig. 3-

12). The lagged component 2 spatial principal component scores were less clear in their interpretation though suggest a slight north to south pattern of variation within the three zones and indicate that the Plateau Bateke samples may be distinct from the rest of the Ogooué drainage (Fig. 3-12). Analysis of the Ogooué-only subset of samples resulted in recognition of the same clusters as the full Central dataset (results not shown).

For both the global and Central Africa-only Mecistops dataset, DAPC revealed that the first eight principal components were sufficient to discriminate genetic clusters regardless of the chosen cluster number. For the global dataset, 8 was the BIC- selected k-value though using BIC-scores alone it may not be possible to distinguish 6,

7, or 8 as the most likely k-value. At all k-values, West African Mecistops comprised a unique cluster with no proportion of admixture and were, additionally, the most divergent

(Fig. 3-13). Additionally, at k=6 and k=8, West Africa was comprised of two unique clusters representing the Upper Guinea forest zone and the Senegambia basin. Central

African Mecistops were broken into discrete clusters in the Gabonese coast, Ogooué and the Congo Basin. At k=6, the Ogooué showed lower and upper drainage clusters with an admixed middle drainage, while these three regions were generally distinguished at k=7 and k=8. For the Central Africa-only Mecistops dataset, 6 was the

BIC-selected k-value though distinguishing k=4 through k=7 may not be possible on the basis of BIC-scores alone. For all k-values, recovered clusters generally represented those recovered by analysis of the full dataset (Fig. 3-14). In DAPC analyses of both

the full and Central-only Mecistops datasets, increasing k-value usually increased cluster identification and admixture in the Gabon coast cluster, though additional clusters within this region did not show any biological (sex, size class) or geographic

(sampling location, directionality) significance (Figs. 3-13 and 3-14). Analysis of the

Ogooué-only subset of samples resulted in recognition of the same clusters as the full

Central dataset (results not shown).

In the sPCA analysis of Osteolaemus multilocus genotypes as many as three global components were most important in interpreting the spatial pattern of genetic variation (Fig. 3-15). Interpolation of the lagged component 1 spatial principal component scores showed three zones of differentiated allele frequencies corresponding to the Gabonese coast, which showed some signs of a north to south cline, the Abanda area including the caves, and the Ogooué drainage, which some signs of an east to west cline (Fig. 3-16). The lagged component 2 spatial principal component scores highlighted the isolation of the Abanda area including the caves and showed signs of a generalized north/west to south/east cline. The lagged component 3 spatial principle components indicated possible segregation of coastal allele frequencies from north to south (Fig. 3-16).

DAPC analysis of Osteolaemus revealed that the first seven principal components were important in discriminating genetic clusters regardless of the chosen cluster number. Four was the BIC-selected k-value though using BIC-scores alone it may not be possible to distinguish 3, 4, or 5 as the most likely k-value. At all k-values, the

Ogooué drainages, Abanda region, and coastal areas formed generally recognizable clusters, though the latter showed considerable admixture while virtually no individuals

in the former two were admixed or assigned to other clusters (Fig. 3-17). At k=3, all individuals from the Nkomi Lagoon area (Abanda, Abanda Caves, Mpivie River) were recognized as a cluster unique from other coastal samples, while at k=4 individuals from the Abanda Caves were discriminated from other Nkomi samples. At k=5, only the coastal samples south of the Nkomi showed increased cluster distinction and admixture, though additional clusters within this region did not show any biological (sex, size class) or real geographic (sampling location, directionality) significance (Figs. 3-17) and, therefore, this is not likely to be the true number of spatially relevant clusters.

Discussion

My results show similar patterns of spatial genetic structure between sympatrically-distributed crocodiles, providing clear support for the strong influence of regional processes structuring crocodile populations in western Central Africa.

However, the strength of perceived inter-regional partitioning differs between these species which provides considerable support for the importance of the interaction between regional process and species-specific ecology. For example, there is significant support for species level distinction in Osteolaemus between the Ogooué and

Congo Basins (this study, Eaton et al., 2009). Mecistops, while not divergent at the species-level, showed relative mitochondrial lineage sorting and clear microsatellite- based partitioning between these two basins. That both of these crocodiles showed some degree of basin isolation supports vicariance as a regional process; however, the incongruence of the extent of basin partitioning is likely best explained by species- specific factors like the degree of dependence on closed canopy forest systems and aquatic versus terrestrial movements.

Within each of the Ogooué and Congo Basins both crocodiles showed semi- congruent patterns of spatial partitioning, though likely due to disparate species-specific process. In the Congo Basin, both species showed clear mitochondrial support for lineage partitioning through geographically-explicit haplotypes. Mecistops lineages seem to be partitioned by drainage though there were shared haplotypes across the region north of the Congo River (i.e., Ituri to the Sanagha). The limited movement data available to date suggests that slender-snouted crocodiles maintain small home ranges and have low probabilities of long distance dispersal (M. Shirley, unpub. data) suggesting that Mecistops was historically continuously distributed across this landscape. In contrast, Osteolaemus showed a more generalized east – west partition

(Chapter 2, M. Shirley unpub. data, Eaton et al. 2009) that follows patterns associated with the location of Pleistocene forest refugia (e.g., Beresford and Cracraft 1999, Marks

2010).

In the Ogooué Basin both Mecistops and Osteolaemus exhibited patterns of population partitioning between the main Ogooué drainages and the coastal drainages to the south. However, the degree of mitochondrial lineage sorting and microsatellite cluster distinction exhibited by Mecistops is more likely due to hydrologic basin partitioning while shared mtDNA haplotypes and more admixed microsatellite structure in Osteolaemus suggests a distance-based cline or expansion from multiple forest refugia (Leal 2001, Maley 1987). Identification of discrete lineages and genetic clusters from the main Ogooué drainage and the southern coast of Gabon follows expectations from major geologic features on the landscape. The Lower Guinea bioregion is characterized by a single hydrologic basin, the Ogooué, which can be separated into

significant interior and coastal regions. Interior portions of the basin, extending from the eastern boundary with the Congo Craton (i.e., the Bateke and Belinga Plateau’s) to within ±100 km, are separated from the coastal region by a north to south ranging series of low mountains including the Mont Crystal, Massif du Chaillu, Mont Doudou, and the

Mayombe ranges. All drainages east of this topography ultimately run into the Ogooué.

The coastal region, however, is comprised of north and south portions where drainages coming off the western relief run directly into the ocean, as in the north, or into a series of low-lying coastal lagoons, as in the south, and do not actually communicate with the

Ogooué or its tributaries. The two coastal regions are separated from each other by the

Ogooué Delta and, geologically, the Nkomi fracture.

Genetic homogeneity seen throughout the coast for Mecistops does not meet expectations based on the contemporary landscape as all rivers in this region run east to west with seemingly no connectivity, especially across the Mont Doudou and

Mayombe highlands. However, cycles in sea level and rainfall since the Last Glacial

Maximum (LGM; ± 18,000 YBP) resulted in drastic changes to the southern lagoon landscape. The forests of Central Africa regressed and fragmented considerably ending around 18,000 YBP and lasting until the start of the Holocene some 10,000 YBP

(Kim et al. 2010, Lézine and Vergnaud-Grazzini 1993), throughout which time lake levels and river discharges were noticeably decreased (Gasse 2000). Most recently,

±2,000 YBP, the northern N’gowe and N’dougou lagoons may have formed a single mega-lagoon from the Rabi to the Nyanga River (Vande weghe 2007). Additionally, the entire southern coast is spotted with much smaller beach front lagoons that are predominantly the result of alluvial sediment deposit and shifting river mouth locations.

It is possible that these smaller lagoons facilitated connectivity throughout this region as well.

While both crocodiles showed clear support for incomplete maternal lineage sorting, the extent of lineage sorting exhibited by Mecistops matches expectations for a species with historically widespread and well-connected populations followed by a combination of late Pleistocene/Holocene and relatively recent fragmentation. Though the LGM was followed by a period of increased lake level and river discharge associated with monsoon conditions and the spread of lowland forests (Adegbie et al.

2003, Kim et al. 2010, Lézine and Vergnaud-Grazzini 1993), there is strong evidence for a second period of aridity in Africa around 4, 000 - 5,000 YBP resulting in significantly reduced precipitation and forest fragmentation through to 2,100 YBP (Giresse et al.

2008, Kim et al. 2010, Maley 2002, Vincens et al. 1999, Wirrman et al. 2001).

Coalescent simulations identified the start of the lineage sorting process for the coast versus Ogooué and Congo drainage at ±225 Mecistops generations ago, which is congruent with this mid-Holocene aridification event. Additionally, this mitochondrial pattern is in strong agreement with the pattern exhibited by microsatellites though perhaps reflective of the longer time involved in mitochondrial lineage sorting compared to microsatellite frequency-based cluster differentiation (e.g., Howes et al. 2006).

In contrast, Osteolaemus mitochondrial sequence variation suggests that this species underwent a range expansion at some time in the “recent” past. However, the patterns exhibited by mtDNA and microsatellite loci were somewhat incongruent in that most crocodiles and sampling localities exhibited a single, common mitochondrial haplotype with few geographically restricted rarer haplotypes. In contrast, the

microsatellites distinguished clusters between the Ogooué and coastal regions with a secondary cline of connectivity from north to south along the coast. Osteolaemus appears to be highly forest dependent for most of its lifecycle and this pattern is likely indicative of expansion from one or more forest refugia in Gabon with relatively high contemporary population connectivity (Eaton 2008, Monsen and Blouin 2003, Phillips et al. 2004). The most recent climate-driven forest fragmentation events (2,100 – 4,000

YBP) most affected forest edges and exposed wetland habitats, which would not likely limit connectivity in Osteolaemus given its capacity for overland dispersal.

Like the phylogenetic analysis presented in Chapter 2, this is the first comprehensive, intraspecies analysis for Mecistops and only the second for O. tetraspis

(Eaton 2008, Eaton et al. 2009). Using microsatellite data, Eaton (2008) described statistically significant, though low, genetic differentiation between O. tetraspis populations in three Gabonese national parks also included in this study (Ivindo,

Loango, Mayumba). Additionally, he identified up to 10 genetic clusters, though there was very little support for geographic structure in more than 3 clusters corresponding to the Ogooué drainage (Ivindo) samples from the coast with, at best, weak support for north and south coastal groups. The increased geographic sample for Osteolaemus presented here generally supports these results, though the inclusion of samples from intermediate areas along the coast suggests that the weak differentiation found by

Eaton between extreme coastal samples are actually just end points along a cline.

Unfortunately, no other phylogeographic or landscape genetic studies have been carried out on aquatic vertebrates with an extensive sampling distribution in Gabon and so no information exists to compare patterns seen in these two crocodiles, especially

Mecistops. Elsewhere in Africa, Crocodylus niloticus (Hekkala et al. 2010) and C. suchus (Cunningham et al. unpub. data) show similar patterns to Mecistops of genetic structuring by river basin. Within Central Africa, comparison with other forest dependent species may be relevant, especially for Osteolaemus due to the high degree of forest dependence and overland movements. Analysis of western lowland gorilla populations, including sampling throughout Gabon, suggests a pattern of expansion from a single refuge south of the Ogooué River (Anthony et al. 2007, Clifford et al. 2004), perhaps centered on the Massif du Chaillu or Monts Doudou regions (Maley 1987, 1996). Either or both are possibilities for Osteolaemus.

Aside from the uncovered spatial patterns, the other remarkable results of this study are the discrepancy in mtDNA haplotype and nucleotide diversity between

Mecistops and Osteolaemus. Mecistops diversity was similar to the low variation reported in A. mississippiensis (Glenn et al. 2002) and insular populations of Crocodylus porosus (Russello et al. 2007) but was considerably lower than in other crocodilians including Caiman sp. (Godshalk 2006, Hrbek et al. 2008, Vasconcelos et al. 2006,

Venegas-Anaya et al. 2008) and Crocodylus moreletti (Ray et al. 2004). Osteolaemus nucleotide diversity was an order of magnitude higher than these species and comparable, but triple, to that seen in Melanosuchus niger (Vasconcelos et al. 2008).

The Melanosuchus results are, perhaps, an interesting comparison to the results from Mecistops and Osteolaemus as the sample sizes and extent of geographic sampling distribution were comparable, additionally Melanosuchus fulfills a similar niche in the Amazon as Mecistops in the Congo. Vasconcelos et al. (2008) recovered a similar pattern of few widespread haplotypes; however they found significantly more

unique haplotypes using less sequence data than in Mecistops. Additionally, they reported significant negative values for Tajima’s D and Fu’s F, indicating a strong signature of demographic expansion like the results presented here for Osteolaemus.

This genetic pattern may be congruent with what we know about census population increases in Melanosuchus populations (Thorbjarnarson 2010), but is generally incongruent to what we expect is happening with both Mecistops and Osteolaemus where census populations are declining. While Mecistops sp. nov. and O. tetraspis have largely divergent ecologies, they both show patterns of relative panmixia across their ranges compared to the other crocodilians discussed above which may partially explain the relatively high levels of genetic diversity compared to the more fragmented/structured species (e.g., López-Castro and Rocha-Olivares 2005, Santos et al. 2007, Taylor et al. 2010).

No studies have been able to empirically quantify rates of evolution (i.e., substitution or mutation) for crocodiles; however, that all crocodilians studied to date exhibit low diversity compared to other vertebrates is interesting and worth comment.

Mitochondrial DNA, for example, is hypervariable across lineages and shows hypermutability compared to most nuclear DNA (Lynch et al. 2006). However, the reasons for the extensive inter-lineage heterogeneity are not well understood. A strong negative correlation has been posited between mutation rate and generation time

(Bromham et al. 1996, Ohta 1993) and longevity/lifespan (Barja and Herrero 2000,

Kujoth et al. 2007, Nabholz et al. 2008), while a more positive linear relationship has been proposed for metabolic rate (e.g., Gillooly et al. 2005). While most of these studies used mammal lineages as model systems, there is some indication that some

combination of these factors is relevant to crocodilian evolution (Bromham 2002,

Gillooly et al. 2005, Nabholz et al. 2008). Given their large body sizes, relatively low metabolic rates, and long expected lifespans it may be that crocodiles simply have low mutation rates and, therefore, do not easily recover genetic diversity after major demographic events (additional discussion Chapter 4).

This study is the first to explore comparative phylogeographic and spatial genetic structure in Central African aquatic taxa. The results are significant in their support of the influence that regional vicariant (e.g., Basin entrapment) and climate change (e.g., via the contraction/expansion of forests and wetlands) processes have had on genetic structure for both highly aquatic (Mecistops) and semi-terrestrial (Osteolaemus) forest- dependent species. Distinction of the Congo and Ogooué Basins as unique bio/phylogeographic regions is not new; however, this is the first vertebrate-based evidence for biogeographic distinction between the main Ogooué and coastal zones of the Lower Guinea bioregion. The degree of phylogeographic and spatial genetic structure seen between the Ogooué and coastal drainages warrants further investigation in other aquatic taxa, but should be considered distinct zones for crocodile conservation planning.

Table 3-1. Mecistops specimens used in spatial analyses and their capture localities. No. is the number of specimens used from each site. Map Country Site Coordinates No. No. Label1 mtDNA msats A The Gambia Gambia River NP N13.64207 W14.9608 5 5 B Cote-d’Ivoire Boubou River N5.30883 W5.21757 1 1 C Dipui Crocodile Farm2 N5.20454 W3.74519 3 3 D Goh River N5.23444 W6.64327 2 2 E Ghana Accra Zoo3 N5.70584 W0.21558 1 1 G Gabon Azingo River S0.49485 E9.99332 9 24 H Bongo River S2.52212 E10.09213 6 17 I Lac Divangui S1.94708 E9.98159 4 4 J Dji Dji River N0.05740 E12.70930 6 23 K Echira/N’gowe Rivers S2.20690 E9.70436 17 68 L Lac Evaro S0.90918 E10.10261 2 2 M Ogooué River4 S1.04988 E9.34485 1 1 N Banio Lagoon S3.72242 E11.12689 2 2 O Mbani River S2.51145 E10.57538 3 3 P Mpassa River S2.05323 E14.06599 9 17 Q Mpivie River S1.73650 E9.43047 6 11 R Nyanga River S3.00730 E10.35200 6 6 S Rabi River S1.90387 E9.51144 3 5 T Congo Lac Tele N1.37328 E17.15292 5 5 U CAR5 Sanagha River N2.482447 E16.2168 1 0 V DR Congo Kinshasa Reptile Park6 S3.54021 E16.09651 1 1 W Lomami River S2.68550 E25.07678 6 6 X Ituri River N1.49781 E28.67505 7 7 Y Lake Tanganyika7 S3.36735 E29.15647 2 2 1Corresponds to Figure 2-1. 2Specimens wild caught throughout the coastal lagoon system from Grand Bassam west to Abidjan (Lagoon Ebrie). 3Animal wild caught in Ghana, though of unknown specific locality, likely southern Ghana. 4Animal held at the Le Ranch Hotel in Port Gentil, of unknown specific origin but believed to have been wild caught in the Ogooué River between Port Gentil and Lambarene. 5Central African Republic. 6Captive held animal in Kinshasa of unknown origin, but believed to come from upstream the Congo River, perhaps area of Stanley Pool. 7Animals captive held at the Hydrobiological Research Station in Uvira of wild caught origin from the mouth of the Ruzizi River in Lake Tanganyika.

Table 3-2. Osteolaemus specimens used in spatial analyses and their capture localities. No. is the number of specimens used from each site for each different marker type. All samples analyzed came from sampling sites in Gabon. Map Site Coordinates No. No. Label1 mtDNA msats A Abanda2 S1.43948 E9.69149 14 14 A Abanda Caves S1.42808 E9.69024 18 18 B Lac Azingo S0.48916 E9.95614 1 1 C Bongo River S2.52212 E10.09213 7 7 D Dji Dji River N0.05740 E12.70930 5 11 E N’gowe Lagoon S1.87233 E9.32890 28 7 F Lac Divangui S1.94708 E9.98159 2 2 G Lac Ogoumué S1.15380 E9.99486 1 1 H Echira/N’gowe Rivers S2.20690 E9.70436 13 30 I Mbani River S2.51145 E10.57538 7 7 J Mpivie River S1.73650 E9.43047 4 4 K Nyanga River S3.00730 E10.35200 4 4 L Rabi River S1.90387 E9.51144 7 35 M Mboumie River S0.39917 E10.57840 1 1 N Lope River S0.16829 E11.62324 17 17 O Sette Cama S2.51631 E9.78500 8 8 P Tassi Lagoon S2.03065 E9.4101 X 4 P Louri Lagoon S2.00218 E9.37908 8 27 Q Petite Loango S2.29716 E9.6033 18 0 R Bame Lagoon3 S3.78277 E11.02256 S Banio Lagoon3 S3.72242 E11.12689 S N’djoungou River3 S3.61218 E10.93948 T Dounvou River3 S3.43319 E10.6939 23 183 U Louzibi River3 S3.72863 E11.12165 V Mbokou Wara River3 S3.68279 E11.07821 X Tchimbiya Lagoon3 S3.87099 E11.03717 1Corresponds to the map in Figure 3-1. 2Individuals captured in forest stream network outside of the cave system. 3Samples collected in Mayumba National Park and considered a single aggregation.

Table 3-3. Microsatellite markers used to genotype both Mecistops and Osteolaemus individuals as part of this study. N is the number of each species that were successfully genotyped for each marker out of 216 for Mecistops (Mcat) and 219 for O. tetraspis (Otet). Re-Amp is the number of individuals re-genotyped for error checking. Gene Region Motif N - Mcat Re-Amp N - Otet Re-Amp Primer Sequences (5’  3’)1 CpDi06 Di 216 (100%) 30 (14%) - - CCAGTCGGGCGTCATCATGTTGGGCACTTTGAAC GTTTAAGAAAAATGGTGGAAAAC CpF509 Di 215 (99%) 31 (14%) - - CCAGTCGGGCGTCATCAACACAAAGGAGCATACAC GTTTAGCCAATTCCCATATCT CpDi21 Di 216 (100%) 56 (26%) - - CCAGTCGGGCGTCATCAAAACAGTTGGCTCTGTG GTTTATACTTCCTGTGGCATCAT CpDi29 Di 214 (99%) 48 (22%) - - CGAAACAGCCAAATGTGAG CAGTCGGGCGTCATCAGGTAGCTCCAAGTAGTTTATT CpP305 Di 214 (99%) 40 (18%) - - CGTTTGTAGCTGGAACCTGATAGTG CAGTCGGGCGTCATCAGGTTAACACGTGGTAACTACA CpP314 Tetra 216 (100%) 37 (17%) - - CGTTTGAAATGCCACTAATACACACA CAGTCGGGCGTCATCACCAATTCTTCAGGTCCTTAT CpP205 Tetra 214 (99%) 42 (19%) 205 (94%) 17 (8%) CGTTTACCCATCATGTATTTTTAGA CAGTCGGGCGTCATCAGAATCACCACCCAAAG CpP4004 Tetra 207 (96%)2 29 (13%) 208 (95%) 7 (3%) CCAGTCGGGCGTCATCACTGAATTGGGTGGAATAG GTTTATCCACATTTTTCCATGAC CpP4208 Tetra 216 (100%) 18 (8%) 219 (100%) 29 (13%) CCAGTCGGGCGTCATCACTCTTATTGCACCAGGTATA GTTTAGAGAGATGCGTGATGAT CpP1401 Tetra 216 (100%) 33 (15%) 208 (95%) 16 (7%) CGTTTAAAAGTAACCCAAAATACACA CAGTCGGGCGTCATCACTTCGGCATCTGATTC CpP1610 Tetra 216 (100%) 20 (9%) 217 (99%) 34 (16%) CCAGTCGGGCGTCATCATAGAGGGATTTTGACTGT GTTTGATTATTTTGTCTGGGTTCTT CpP2206 Tetra 216 (100%) 22 (10%) 199 (91%) 37 (17%) CGTTTAGGCCAGTTCTTATCTACAT CAGTCGGGCGTCATCAAAGTTCTCCCCACTAAA CpP3004 Tetra 216 (100%) 19 (9%) 211 (96%) 10 (5%) CCAGTCGGGCGTCATCAGGAGTGAATCTATGCCAGC GTTTAAAATGTTTTCATATGGTCG CpP3313 Tetra 213 (98%) 64 (29%) 176 (80%) 47 (21%) CCAGTCGGGCGTCATCACTTCTGTTACTTAGGGACTG GTTTAAAAACCCAGGCAAATA CpP2514 Tetra 212 (98%) 22 (10%) 199 (91%) 6 (3%) CGTTTCTTTGCACACCCTATCTATC CAGTCGGGCGTCATCAGAGTTGAAGGGAAGTTTC CpP1404 Tetra - - 208 (95%) 21 (10%) CGTTTCCCAATAACTCCATAACATAG CAGTCGGGCGTCATCAGGGCTTCAGCACACTA CpP121 Tetra - - 200 (91%) 20 (9%) CCAGTCGGGCGTCATCAATATTTGTTTCTGGGATCA GTTTAGGAAATGAGCCCTAATAGT CpP914 Tetra - - 205 (94%) 3 (2%) CCAGTCGGGCGTCATCAACATGGCAACTTCAGAG GTTTCGAATAAATGCAGCATAA 1All forward primers were tagged with the following M13 primer for dye annealing: CACGACGTTGTAAAACGA 2Reduced percentage of individuals genotyped here reflects failed reactions for all West African animals, 100% of Central African individuals were genotyped.

Table 3-4. Summary statistics for select coalescent simulations run to determine the pattern and timing of lineage sorting in Mecistops. Lower Upper No. of Depth Observed Model Migration Total N 95% 95% Pops e (Generations) nDC Bound Bound Fragmented Ancestor 12 N 10000 10 69 61 74 Fragmented Ancestor 12 N 10000 20 69 53 67 Fragmented Ancestor 12 N 10000 50 69 40 53 Fragmented Ancestor 12 N 10000 1000 69 1 8 Fragmented Ancestor 6 N 10000 20 30 58 73 Fragmented Ancestor 6 N 10000 50 30 47 64 Fragmented Ancestor 6 N 10000 100 30 35 52 Fragmented Ancestor 6 N 10000 200 30 24 38 Fragmented Ancestor 6 N 10000 500 30 11 23 Fragmented Ancestor 6 N 3500 50 30 29 43 Fragmented Ancestor 6 N 3500 100 30 18 29 Fragmented Ancestor 6 N 3500 200 30 9 20 Fragmented Ancestor 6 Y 10000 20 30 61 75 Fragmented Ancestor 6 Y 10000 100 30 48 65 Fragmented Ancestor 6 Y 10000 500 30 46 63 Fragmented Ancestor 6 Y 10000 5000 30 46 63 Two Basin 12 N 10000 20 122 123 153 Two Basin 6 N 10000 200 72 58 86 Two Basin 6 N 3500 50 58 59 86 Two Basin 6 N 3500 90 58 56 84 Two Basin 6 Y 10000 100 72 11 22 Two Basin 6 Y 10000 200 72 0 0 Three Basin 12 N 10000 20 98 102 129 Three Basin 6 N 10000 200 60 42 62 Three Basin 6 N 3500 90 60 46 68

Table 3-5. Summary statistics for microsatellite markers used to genotype both Mecistops and Osteolaemus individuals as part of this study. Size range is for the full product including flanking sequence. NA is the total number of alleles. Ne is the effective number of alleles for each marker. He and Ho are the expected and observed heterozygosity. All statistics for Mecistops are based on the full sample including both West and Central clades while for Osteolaemus they were only calculated for O. tetraspis individuals. Mecistops Osteolaemus Gene Region Size Range NA Ne He Ho Size Range NA Ne He Ho CpDi06 251 – 276 10 1.688 0.408 0.338 - - - - - CpF509 340 – 390 17 4.150 0.759 0.565 - - - - - CpDi21 183 – 223 16 2.006 0.501 0.335 - - - - - CpDi29 251 – 276 9 2.161 0.537 0.451 - - - - - CpP305 199 – 263 17 2.453 0.592 0.446 - - - - - CpP314 265 – 275 3 1.119 0.106 0.000 - - - - - CpP205 341 – 385 10 1.161 0.139 0.037 353 – 389 9 3.146 0.682 0.635 CpP4004 391 – 409 3 1.846 0.458 0.348 408 – 452 10 3.078 0.675 0.610 CpP4208 208 – 286 17 9.724 0.897 0.759 210 – 274 18 6.690 0.851 0.769 CpP1401 189 – 209 6 1.581 0.368 0.248 189 – 217 7 1.138 0.121 0.073 CpP1610 310 – 342 8 1.976 0.494 0.111 294 – 358 13 3.691 0.729 0.688 CpP2206 262 – 270 3 1.114 0.102 0.028 242 – 274 5 1.423 0.297 0.259 CpP3004 130 – 185 13 4.335 0.769 0.647 127 – 131 2 1.116 0.104 0.091 CpP3313 382 – 414 9 4.329 0.769 0.535 390 – 414 6 1.269 0.212 0.194 CpP2514 192 – 216 6 2.758 0.637 0.481 188 – 224 8 2.710 0.631 0.638 CpP1404 - - - - - 263 – 309 5 1.278 0.271 0.165 CpP121 - - - - - 175 – 224 6 3.494 0.714 0.670 CpP914 - - - - - 274 – 316 12 2.244 0.554 0.429

100

Table 3-6. K-cluster models searched for Mecistops and Osteolaemus using three different clustering methods. Model selection-related values are provided for each clustering procedure. *Indicates the selected model under each selection regime. **Indicates models indistinguishable from “best” model based on BIC alone. K-value Pritchard2 Evanno3 BIC Mecistops1 1 N/A N/A 322.82 2 4.4E-288 - 303.07 3 4.6E-192 11.398* 297.01 4 3.5E-184 0.6555 293.88** 5 4.3E-66 6.0886 292.19** 6 1.5E-55 2.4435 290.61* 7 0.99999* - 289.62 8 N/A N/A 289.00 9 N/A N/A 289.81 10 N/A N/A 291.09 11 N/A N/A 292.60 12 N/A N/A 294.36 13 N/A N/A 295.77 14 N/A N/A 298.08 15 N/A N/A 300.72 Osteolaemus 1 N/A N/A 305.29 2 0.0 - 298.26 3 1.7E-60 2.9680 293.00** 4 0.99999* 11.065* 291.79* 5 1.2E-06 1.9398 291.70** 6 1.4E-96 0.9521 292.18 7 4.9E-130 1.6910 293.35 8 1.1E-209 3.1873 294.18 9 1.1E-189 2.4052 295.90 10 2.4E-241 0.2770 298.29 11 3.3E-281 0.7457 300.02 12 N/A N/A 301.89 13 N/A N/A 304.53 14 N/A N/A 306.37 15 N/A N/A 307.94 1Model selection values for the Central clade-only Mecistops dataset. 2Ad hoc method proposed by Pritchard et al. (2000) – Pr(K|X) 3ΔK proposed by Evanno et al. (2005)

101

Figure 3-1. Map of Osteolaemus sampling localities. The base map is colored to reflect topographic and elevation features of the landscape across the sampling distribution. The major river seen in the middle is the entire Ogooué, while the Congo can be seen in the SE corner and the Dja-Sanagha in the north to east. Sample labels correspond to localities detailed in Table 3-2.

102

Figure 3-2. Diagrammatic illustrations of population models tested in coalescent simulations for Mecistops. These are simplified dummy models illustrating the types of hierarchical structure tested against the Fragmented Ancestor null model (upper left). Arrows on the right and left correspond to temporal change points in effective population size (i.e., tree and clade height). Letters on nodes correspond to hypothetical populations or haplogroups (C = coast, O1 = Ogooué 1, O2 = Ogooué 2, CBW = Congo Basin West, CBE = Congo Basin East, LT = Lake Tanganyika). Models considered Ogooué Basin versus Congo Basin (top middle), 3 Basin (coast, Ogooué, Congo – top right), 2 Basin with hierarchical structure in the basin (bottom left), coast versus Ogooué plus Congo basins (bottom middle), and complete isolation by distance from west to east (bottom right) and east to west. Simulated datasets were fit into all variations of these models (including others not shown) varying tree and clade heights to correspond to known climatic and landscape events in Central Africa.

103

Figure 3-3. Map of haplotype distribution for Mecistops. Each circle represents the haplotype composition of sampling areas color-coded to follow haplotypes in Fig. 3-5. The geographic extent of the map is from the Atlantic coast of Central Africa to Lake Tanganyika (±2,200 km).

104

Figure 3-4. Map of haplotype distribution for O. tetraspis. Each circle represents the haplotype composition of sampling areas color-coded to follow haplotypes in Fig. 3-6. Green areas are Gabonese National Parks.

105

Figure 3-5. Statistical parsimony networks created by TCS as part of the NCPA for Mecistops (top) and Osteolaemus (bottom). Each circle/square represents a unique haplotype, the size of which reflects the frequency of that haplotype in the sample. Each connection line represents a single mutational difference between haplotypes and small x’s (top network) represent extinct or unsampled haplotypes. This figure is shown only to illustrate the highly reticulate nature making NCPA inference impossible prior to application of the described ambiguity resolution technique (Crandall and Templeton 1993). Final networks are depicted in Figs. 3-6 and 3-7.

106

Figure 3-6. Haplotype network created using dnapars and Haploviewer for Mecistops. This represents the final network analyzed in NCPA. Haplotype circle size, and interior label, reflects the frequency of that haplotype in the sample and each haplotype is color-coded to reflect the proportion found in each of 15 sampling localities. Letters are assigned to each haplotype which correspond to color-coded pie charts in Fig. 3-3. Each connection line represents a single mutational difference between haplotypes and small circles represent extinct or unsampled haplotypes.

107

Figure 3-7. Haplotype network created using dnapars and Haploviewer for Osteolaemus. This represents the final network analyzed in NCPA for Osteolaemus. Haplotype circle size, and interior label, reflects the frequency of that haplotype in the sample and each haplotype is color-coded to reflect the proportion found in each of 15 sampling localities. Letters are assigned to each haplotype which correspond to color-coded pie charts in Fig. 3-4. Each connection line represents a single mutational difference between haplotypes and small circles represent extinct or unsampled haplotypes.

108

Figure 3-8. Results from Bayesian clustering of all Mecistops microsatellite genotypes using STRUCTURE with K = 4. Columns correspond to the multi-locus genotypes of each crocodile partitioned into colors representing the probability of ancestry assigned to each cluster (K). Clusters correspond to the Gabonese coast, the Ogooué and its drainages, the Congo Basin, and West Africa.

109

Figure 3-9. Results from Bayesian clustering of Central Mecistops microsatellite genotypes using STRUCTURE. Columns correspond to the multi-locus genotypes of each crocodile partitioned into colors representing the probability of ancestry assigned to each cluster (K) for three different K values. The ΔK selected clustering, K = 3 (top), model recovered the same three Central African clusters as the K = 4 model for all Mecistops data. The K = 4 model has clusters corresponding to the Gabonese coast, the lower Ogooué, the middle/upper Ogooué, and the Congo Basin, while the K = 5 completely distinguished the lower from the middle/Upper Ogooué.

110

Figure 3-10. Results from Bayesian clustering of O. tetraspis microsatellite genotypes using STRUCTURE. Columns correspond to the multi-locus genotypes of each crocodile partitioned into colors representing the probability of ancestry assigned to each cluster (K) for three different K values. Samples are order geographically from the middle Ogooué downstream and south along the coast to Mayumba. The ΔK selected clustering, K = 4 (bottom), model loosely recovered clusters from the middle Ogooué (green), Abanda Caves (blue), and a large cline moving south along the coast.

111

Figure 3-11. Eigenvalues of sPCA analysis for Mecistops multilocus genotypes. The barplot (left) illustrates that the dataset contains at least one significant global structure and not likely any important local structure. The decomposed plot (right) illustrates that the first eigenvalue is the most important in explaining the spatial variance pattern.

112

Figure 3-12. Interpolated maps of individual lagged scores for the first (top) and second (bottom) spatial principle components for Mecistops. Black dots represent actual samples plotted in real space.

113

Figure 3-13. DAPC results for the full Mecistops dataset. The left column is scatterplots of individual assignment in discriminated clusters connected by a minimum spanning network. Inset shows the cumulative % of variance for eigenvalues (used in black). The right column is barplots of individual assignment probability to each cluster, similar to the Q-value barplot of STRUCTURE. Labels above the barplots signify geographic zones: Gabonese coast (Coast), Lower (L), Middle (M) and Upper (U) Ogooué, Congo Basin (C), and West Africa (WA). Cluster color is the same in both visual representations for each k-value (6 – 8, top to bottom).

114

Figure 3-14. DAPC results for the Central Africa-only Mecistops dataset. The left column is scatterplots of individual assignment in discriminated clusters connected by a minimum spanning network. Inset shows the cumulative % of variance for eigenvalues (used in black). The right column is barplots of individual assignment probability to each cluster, similar to the Q-value barplot of STRUCTURE. Labels above the barplots signify geographic zones: Gabonese coast (Coast), Lower (Low), Middle (Mid) and Upper (Up) Ogooué, and Congo Basin (CB). Cluster color is the same in both visual representations for each k-value (4 – 6, top to bottom).

115

Figure 3-15. Eigenvalues of sPCA analysis for Osteolaemus multilocus genotypes. The barplot (left) illustrates that the dataset contains at least two significant global structures and not likely any important local structure. The decomposed plot (right) illustrates that the first two eigenvalues are the most important in explaining the spatial variance pattern, while the third may contain secondary variance.

116

Figure 3-16. Interpolated maps of individual lagged scores for the first (top), second (bottom left) and third (bottom right) spatial principle components for Osteolaemus. Black dots represent actual samples plotted in real space.

117

Figure 3-17. DAPC results for O. tetraspis. The left column is scatterplots of individual assignment in discriminated clusters connected by a minimum spanning network. Inset shows the cumulative % of variance for eigenvalues (used in black). The right column is barplots of individual assignment probability to each cluster, similar to the Q-value barplot of STRUCTURE. Labels above the barplots signify geographic zones: Ogooué (O), Abanda and Abanda Caves area (A), coast area from North to South. Cluster color is the same in both visual representations for each k-value (4 – 6, top to bottom).

118

CHAPTER 4 COMPARATIVE DEMOGRAPHIC HISTORY OF MECISTOPS SP. NOV. AND OSTEOLAEMUS TETRASPIS REVEALED BY MITOCHONDRIAL AND NUCLEAR MARKERS

Introduction

The diversity of alleles within and among populations can provide useful insight into the demographic and evolutionary history of species. While apportionment of diversity between populations is determined by the time since divergence from a common ancestor and the level of inter-population gene flow (e.g., Nm; Slatkin 1993,

Wright 1931), diversity within populations is driven by their effective population size (Ne), gene flow from other populations, and the mutation rate (µ) of the markers under observation (Gillespie 2004, Wright 1931). Over the past 70 years numerous models have been derived that calculate parameters associated with these processes (e.g.,

Excoffier et al. 1992, Nielsen and Wakeley 2001, Wright 1931). Amongst these parameters, population demographic history is, arguably, the most influential characteristic shaping the abundance and spatial distribution of genetic diversity at both the population and species level.

Rapid population declines (i.e., bottlenecks) are thought to have profound effects on species evolutionary potential and persistence via the ability to cope with changing environments due to a loss of genetic diversity and, ultimately, a decrease in heterozygosity (Frankham et al. 1999, Lande and Shannon 1996). Populations that have undergone drastic reductions in size (i.e., bottlenecks) are increasingly subject to genetic drift and a loss of genetic diversity. However, processes that drive a loss of genetic diversity at the population level may not decrease genetic diversity at the species level by fixing different alleles in different populations. Similarly, post-bottleneck

119

expansions may result in large geographic swaths of genetic homogeneity at the species level if the expansion occurs from single refugia. Alternatively, populations expanding from multiple refugia may increase heterozygosity as individuals come back into contact (Excoffier et al. 2009).

Detection of historic demographic events shaping population history using molecular tools has had a long history. Early efforts to detect changes in population size relied on metrics of sampled sequence diversity like average heterozygosity, number of alleles, and differences in segregating sites versus nucleotide differences

(Chakraborty and Nei 1977, Nei et al. 1975, Tajima 1989a, Watterson 1986). However, introduction of the coalescent (Kingman 1982) and coalescent-based genealogical modeling (e.g., Donnelly and Tavare 1995, Hudson 1991) have greatly improved demographic reconstruction by providing explicit, model-based expectations for the relationship between gene genealogies and changes in effective population sizes (e.g.,

Kitchen et al. 2008, Rambaut et al. 2008). Coalescent-based methods also allow for the estimation of the timing of demographic events (e.g., Drummond et al. 2005, Storz and

Beaumont 2002) permitting explicit inference on the impact of climatological or anthropogenic factors on populations to better understand regional biogeography (e.g.,

Goossens et al. 2006, Heller et al. 2008, Shapiro et al. 2004), as well as the potential future influence of climate change, habitat loss, and other human-mediated global changes on biodiversity.

Coalescent-based approaches to demographic reconstruction have been used relatively few times with contemporary wildlife species. The demographic histories of

African savannah buffalo (Heller et al. 2008), African elephants (Okello et al. 2008),

120

baboons (Storz et al. 2002), and Cynopterus fruit bats (Storz and Beaumont 2002) were demonstrated to be most significantly impacted by mid-Holocene climatic events despite known severe contemporary bottlenecks due to disease, overharvesting, or the perceived capacity for demographic and range expansion based on changes in ecological and climatic suitability. In contrast, Goossens et al. (2006) found that demographic collapse in orangutans is heavily correlated with human-induced deforestation. The difference in association between explanatory events is likely due to some combination of the strength of the genetic signal coupled with the severity of the demographic event, the timing relative to generation time, and the mutation rate of the markers in question (Beaumont 1999, Drummond et al. 2005).

The demographic history of crocodilians has not yet been studied in any detail using any genetic method, including those based on the coalescent. In the previous chapter I explored the spatial relationship of the, sparse, genetic diversity seen in

Central African populations of Mecistops and Osteolaemus. Two demographic scenarios which could account for the lack of genetic diversity and its insignificant spatial structuring are: 1) both crocodile species maintained significantly large historic populations followed by recent collapse, possibly as a result of human pressures like the well-documented hunting-related declines of the 19th and 20th centuries, or 2) historic population bottlenecks following major climatic events with not enough time since the demographic event to allow for the accumulation, or fixation, of genetic diversity (i.e., due to low mutation rate).

The first is a widely discussed hypothesis amongst the community of crocodilian researchers supported largely by the presence of extreme density crocodile populations

121

in swaths of undisturbed habitat with little hunting pressure (e.g., Florida, Pantanal, etc…). Additionally, historic accounts suggest that crocodilians were formerly abundant wherever anecdotes are available (e.g., Audubon 1827, Bartram 1791). Both Mecistops and Osteolaemus in Central Africa are known to be under significant contemporary hunting pressure and are susceptible to degradation of wetland habitats, both of which have caused significant declines of certain populations, even localized extinctions, since about 1960 (e.g., Behra 1987a, 1987b, 1987c, Eaton 2010, Shirley 2010). However, crocodilians are long lived, multiparous and have overlapping generations, so it remains questionable whether “recent” demographic events would even be detectable genetically.

The second hypothesis draws logical support from documentation of changes in climatic suitability and resultant habitat distribution in regions where crocodiles occur

(e.g., the tropics). In Central Africa, Pleistocene glacial cycles and Holocene climatic perturbations have resulted in fluctuations of temperature, humidity and precipitation

(Mayewski et al. 2004), all of which could have significant implications for crocodile abundance and distribution. For example, all crocodilians have distributional limits primarily controlled by temperature (Markwick 1998) and, tropical crocodilians especially, may not respond favorably to temperature decreases (e.g., Brandt and

Mazzotti 1990). Humidity and precipitation changes in the tropics regulate the availability of critical crocodile habitat – wetlands. And, while crocodiles are not typically considered forest obligate species, both Mecistops and Osteolaemus rely on tree cover for at least nesting and may not, therefore, respond favorably to decreases in forest cover.

122

Both of these hypotheses are, as yet, untested. In this chapter, I use coalescent- based demographic reconstructions to determine if demographic changes in crocodile populations can be detected and if the timing of these changes provides support for one of these alternative hypotheses. Both mitochondrial sequences and microsatellite genotypes were employed to detect temporally referenced declines or expansion.

Methods

Collection of data used in the following analyses was described in Chapters 2 and

3. Mitochondrial and nuclear sequence data was used as described below. All microsatellite markers used for spatial analyses in Chapter 3 were used for microsatellite-based demographic reconstruction discussed below.

Demographic Reconstruction From Mitochondrial Sequence Data

Coalescent events represent stochastic processes impacted most significantly by effective population size and changes therein (Hudson 1991, Kingman 1982).

Therefore, distinguishing the relative importance of contemporary population connectivity from historic affiliation on the lineage sorting process may, in fact, hinge on an accurate understanding of historic population profiles. Many parametric methods today assume simplistic demographic models (e.g., exponential or logistic growth) to determine point estimates of θ, or other indices of Ne, as well as generalized increasing or decreasing trend statements (Cornuet and Luikart 1996, Garza and Williamson 2001,

Harpending et al. 1998, Weiss and von Haeseler 1998). However, point estimates of Ne at time t0 do not allow further inference of the relationship between, for example, demographic and climatic or geologic events that ultimately impact coalescent segregation of lineages (Drummond et al. 2002).

123

Skyline plots are a suite of non-parametric methods designed to reconstruct changes in population size that have occurred moving backward through time based on genealogical reconstruction and, therefore, are not constrained by a priori assumptions of a possible demographic model (Pybus et al. 2000). The original skyline plot introduced by Pybus et al. (2000), now referred to as the classic skyline plot, is limited by a critical assumption that the true genealogy is known without error – an assumption that does not take into account phylogenetic uncertainty, especially amongst unsorted populations and genealogies with short inter-coalescent intervals (e.g., Drummond et al.

2005).

In response to this limitation, Bayesian skyline plot models were introduced

(Drummond et al. 2005, Minin et al. 2008, Heled and Drummond 2008). These methods are advantageous in that they recognize the uncertainty in the coalescent process and account for it through simultaneous estimation of the genealogy, demographic history, and substitution model parameters. Bayesian Skyline Plot (BSP) and Bayesian Skyride (BSR) algorithms assume some degree of autocorrelation in population sizes between coalescent steps; however, BSR population size estimates from one coalescent event to the next are penalized based on the length of the coalescent interval where longer intervals are penalized less – the functional equivalent of gradual population size change over time (Ho and Shapiro 2011). While this usually results in smoother demographic curves and may confound detection of sharp changes

(i.e., time-restricted bottlenecks or rapid expansion events), the method may be preferable because there is no a priori assumption of the number of coalescent interval groupings which can be problematic to determine for relatively uninformative (i.e.,

124

homogeneous) datasets (Minin et al. 2008). Extended Bayesian Skyline Plot (EBSP) methods are equivalent to the BSP; however, they further allow the user to overcome coalescent uncertainty associated with using a single marker by incorporating multiple, unlinked loci in the analysis, as well as handling group estimation similar to the BSR

(Heled and Drummond 2008, Ho and Shapiro 2011).

I implemented BSR, and trialed EBSP, models in Beast v1.7.4 (Drummond et al.

2012). Analysis was conducted individually by major haplogroup as resolved from the spatial analyses detailed in Chapter 3 (Mecistops – coast and Ogooué drainage;

Osteolaemus – Abanda, Abanda Caves, coastal sets with and without cave crocodiles,

N’gowe Lagoon, Rabi River, Lope River, and Ogooué drainage), as well as on the full

Gabonese sample set for both species. BSR analyses were conducted on the concatenated mtDNA dataset, while the EBSP model included all unlinked nuclear markers. Unfortunately, given the almost complete lack of intra-population (and inter- population, for that matter) nuclear haplotype diversity seen in Chapter 1, the EBSP method turned out to be uninformative and analytically intractable (J. Heled, pers. comm.), and was abandoned after initial explorations.

mtDNA analyses were implemented under the HKY substitution model for both species, as selected by the Bayesian Information Criteria model selection procedure in jModelTest (Posada 2008), with empirically estimated base frequencies. I used a strict clock model, as there is no a priori expectation of rate heterogeneity amongst genealogical lineages within populations of either species, with uniform rate priors

(U(0,0.9)) and a starting rate of 5e-4/site/thousand years. A uniform prior of (U(0, 300)) thousand YBP was set for the root height of each tree model based on expectations for

125

within lineage divergence, estimated by Beast and r8s (Chapter 2), as well as the estimated date of lineage fragmentation from the coalescent simulations detailed in

Chapter 3. This root height prior distribution, therefore, covers the last two major glacial-interglacial cycles. The kappa parameter for the HKY model was given as a lognormal(1, 1.25) prior with a starting value of 2. All other priors and operators were left at their default settings. Four independent, identical MCMC chains were run for 2.5 x 108 generations with trees sampled every 10,000 iterations. The adequacy of a 10% burn-in and posterior convergence of all parameters was assessed with Tracer v1.5

(Rambaut and Drummond 2009). The results from independent BSR chains were combined using LogCombiner v1.7.4 and the BSR plot reconstructed in Tracer v1.5.

Demographic Reconstruction Using Microsatellites

I implemented a hierarchical Bayesian model that relies on the genealogical information content of microsatellite allele frequency distributions (Beaumont 1999,

Storz and Beaumont 2002). The model uses multi-locus genotypes to simultaneously estimate four natural parameters of interest: N0 (current effective population size), N1

(ancestral population size at the time of the demographic event), T (the generation- scaled time of the demographic event), and µ. The advantage to this hierarchical model structure is that it allows the posterior distribution of the mutation rate and demographic parameters to vary among loci making the inference robust to aberrant markers (Storz and Beaumont 2002). It assumes a single stepwise mutation model for microsatellites, which appears appropriate for this dataset, though slight deviations to a more generalized stepwise mutation model may be tolerated (Girod et al. 2011). The analysis is implemented in msvar1.3 (Storz and Beaumont 2002).

126

I set priors that varied widely between loci for all four parameters capitalizing on ranges of Ne estimated by both Migrate-n (Chapter 3) and skyline plots (above), and estimated timing of the demographic event to incorporate both recent declines due to intensive hunting in the 20th century, as well as population expansion or decline following patterns of forest refugia due to glacial cycling in Africa (Tables 4-1, 4-2). I ran exploratory chains using a variety of initial prior values, as well as hyperpriors, for the coastal populations of Gabon for both species to assess the robustness of the results to prior parameterization. Consistency in the posterior distributions during exploratory runs indicated appropriate selection of priors to achieve reliable parameter estimates.

All species and population divisions outlined above were analyzed with five independent chains using identical hyperpriors but variable starting values for each marker from run to run. Additional exploratory runs investigated population expansion, instead of decline, as well as the effect of generation time (7, 15, and 25 years) on the results.

Each chain resulted in 20,000 draws from the posterior over a total chain length of 1.2 x

109 generations. Stationary convergence of independent chains was assessed using the Heidelberger and Welch stationarity test (Heidelberger and Welch 1983) where chains were assumed to reach stationarity if they passed either the stationarity or half- width tests (Smith 2007). Convergence between independent chains was assessed using the Brooks, Gelman, and Rubin convergence diagnostic statistic (Brooks and

Gelman 1998, Gelman and Rubin 1992) where chain convergence was assumed for all diagnostic statistic values below 1.2 (Smith 2007). Convergence diagnostics were run using the boa package (Smith 2007) in R v2.15.1 (R Core Team 2012). The last half of each chain was used to make a single, combined chain of 50,000 data points which was

127

assumed to summarize the posterior distributions of the demographic parameters.

Parameters of interest for each population were plotted in R v2.15.1 (R Core Team

2012).

Results

Bayesian Skyride Reconstruction of Historic Demographics

Posterior convergence of all BSR model parameters for all populations of both species was confirmed by consistent parameter estimation between independent runs, as well as high (>10,000) effective sample sizes signifying low correlation between chain sampling events. The matrilineal effective population size amongst Mecistops populations and amongst Osteolaemus populations were similar, both contemporarily and backwards in time (Figs. 4-1, 4-2). However, the median estimated effective population sizes were bigger in Mecistops than in Osteolaemus populations.

Unfortunately, the confidence intervals surrounding the estimated effective population sizes were large questioning the validity of these comparisons. Regardless, the observed trend for all populations of both species was slightly decreasing with a sharp increase sometime in the last few hundred years (Figs. 4-1, 4-2). The mean estimated root height of the genealogies ranged from 1,800 – 3,700 and 218 – 438 YBP for

Mecistops and Osteolaemus, respectively (Table 4-3). The mean estimated substitution rate was 1 – 2 orders of magnitude slower for Mecistops than for Osteolaemus (Table 4-

3).

Microsatellite-Based Genealogical Reconstruction of Historic Demographics

Analysis of both Mecistops (Table 4-1, Fig. 4-3) and Osteolaemus (Table 4-2, Fig.

4-4) indicated severe decline, > 97%, for all populations. Based on mean current (N0) and ancestral (N1) population sizes, the severity of the bottleneck was relatively equal

128

amongst populations of both species (Tables 4-1, 4-2). Mean estimated ancestral population sizes for all populations of both species were both large (11,000 – 19,000) and similar between species (Tables 4-1, 4-2). Despite this similarity of ancestral population size, in all cases Osteolaemus showed higher contemporary effective population sizes than Mecistops.

The upper and lower bounds of the estimated time since population decline overlapped considerably between all populations of both species (Figs. 4-3, 4-4); though estimated means ranged from 1,949 to 11,967 YBP for Mecistops (Table 4-1) and 2,113 – 13,458 YBP for Osteolaemus (Table 4-2). While the estimated time since population decline was relatively congruent between species, mean estimates from comparable sampling sites (i.e., Akaka and middle Ogooué) were slightly incongruent.

In all cases, the majority of the posterior density supports population bottleneck dates at or after the Last Glacial Maximum, though posterior support is relatively equal for dates throughout the late Pleistocene to mid-Holocene (Fig. 4-5). In all cases, the lower 95%

HPD bounds for the timing of population decline suggests that both crocodile species underwent significant population declines prior to the start of contemporary commercial trade for meat or leather (Fig. 4-5).

The exploratory runs using different model scenarios (e.g., population expansion versus contraction) and prior settings (e.g., generation time) for selected populations of both species converged on similar posteriors to the full runs described above for all demographic parameters (Figs. 4-6 and 4-7). In particular, population expansion models all converged on population decline and the estimated timing since population decline was similar. Though, for all populations of both species, changing the

129

generation time resulted in slightly lower or higher mean estimates of the time since the population decline, but the time difference was not proportional to the difference in generation time (Figs. 4-6 and 4-7). Convergence on similar contemporary and ancestral effective population sizes, as well as time since population decline, for all populations under these alternative models and prior scenarios supports posterior robustness to prior settings and the general validity of the model results.

Discussion

In this chapter I reconstructed the demographic histories for two crocodile species using coalescent-based genealogical models for two different marker types. The results reveal three striking patterns that provide support for my second demographic hypothesis - historic population bottlenecks following major climatic events. First, demographic signals from the two different marker types indicated the presence of a significant population decline for both crocodile species after the Last Glacial Maximum

(LGM) of the late Pleistocene followed by stable effective populations throughout the late Holocene. Second, the timing of this major demographic decline was markedly congruent between the two crocodile species. Third, the estimated mutation rates for both markers were notably slower than that estimated for other species. This study represents the first historic demographic reconstruction for any crocodilian species and, as such, there is little comparative data available from across the Order much less across the study region. However, as the study itself compared demographic patterns amongst two sympatrically distributed crocodilian species the results are informative about regional phylogeographic processes and their impacts on crocodilians, which may be applicable to crocodilians globally.

130

Microsatellite-based genealogical reconstruction revealed a severe population bottleneck with mean estimates dating to the late Pleistocene/mid-Holocene (± 2,000 –

18,500 YBP). Mitochondrial genealogy reconstructions were hampered by the lack of intra-population mitochondrial diversity, but suggested that no further demographic change occurred in the late Holocene. From these results, it is clear that there is no posterior support for hunting-driven declines in effective population size. This may seem counterintuitive as, for example, we are empirically aware of dramatic census population declines due to hunting within the last century. However, it is likely the case that climate-driven declines since the LGM significantly reduced genetic diversity within these populations and, due to low mutation rates, their effective population sizes simply had not increased prior to contemporary census population declines. The timing of spatial segregation of lineages estimated for Mecistops in the coalescent simulations from Chapter 3 is amazingly congruous with this period suggesting that these population bottlenecks may have been the driving force behind geographic lineage sorting.

Reconstructions of climatic conditions in the Congo Basin since the late

Pleistocene have revealed a relatively constant temperature of 20° - 21°C at the time of the LGM slowly rising to about 25°C only 5,000 YBP (Bonnefille et al. 1992, Weijers et al. 2007, Wu et al. 2007). Even though the Congo Basin in general has the least understood climatic history in Africa, and site specific records present slightly different pictures in terms of magnitude and timing, several trends are evident. The forests of

Central Africa regressed and fragmented considerably ending around 18,000 YBP and lasting until the start of the Holocene some 10,000 YBP (Kim et al. 2010, Lézine and

131

Vergnaud-Grazzini 1993), throughout which time lake levels and river discharges were noticeably decreased (Gasse 2000). This period is followed by sea level rise, spread of mangrove and coastal wetlands, increased lake level and river discharge associated with monsoon conditions and the spread of lowland forests (Adegbie et al. 2003, Kim et al. 2010, Lézine and Vergnaud-Grazzini 1993). Finally, a second period of aridity began in Africa around 4,000 – 5,000 YBP resulting in significantly reduced lake levels and fragmented lowland forests (Giresse et al. 2008, Kim et al. 2010, Maley 2002, Vincens et al. 1999, Wirrman et al. 2001). Unfortunately, my results had high HPD support for this entire period and, therefore, isolating one climatic event as the sole cause may not be possible. Additionally, the different mean time estimates since population decline for the populations examined in this study may reflect site specific variability in habitat change (Elenga et al. 1994, Jahns 1996, Wirrman et al. 2001). Climate change may drive demographic change in crocodilians through two predominant mechanisms: 1) exposure to unsuitable thermal conditions and 2) shifting the availability and distribution of suitable habitat.

Temperature has many effects on crocodilian life cycles from embryonic development to the survival and growth of individuals at all life stages via optimal metabolic function. Incubation in mound nests, usually through the decomposition of vegetation or association with insect activity, allows egg chamber temperatures to rise as much as 6°C, though more normally ±4°C, above the mean daily ambient temperature (Magnusson et al. 1985, Waitkuwait 1985, 1989). Additionally, regardless of nest type or incubation efficiency, ambient temperature directly impacts egg chamber temperatures where 3°C change in the former can cause ≥ 1°C change in the egg

132

chamber, supporting the notion that ambient temperature provides a necessary platform from which thermal optima are reached (Magnusson 1979). Crocodilian embryonic development is optimal at temperatures ranging from 28 - 33°C, while temperatures below 27°C seem to result in the complete failure of embryonic development (reviewed by Magnusson et al. 1985). Incubation temperature also has been shown to effect post- hatching survival and growth rates (e.g., Lang 1985). It is highly likely, therefore, that prolonged climatic cooling events can significantly impact crocodile populations through decreased reproductive success (i.e., fitness) – a notion supported by the results presented here.

Temperature and thermoregulation play critical roles in crocodilian growth and survival. Both parameters are influenced by food intake, metabolic conversion, time dedicated to thermoregulation, and the ability to fight infection. In all cases, the thermal optimum for crocodilians seems to be 25 - 35°C with temperatures outside this range causing significant behavioral and physiological disruptions, even death (Lang 1987).

At below 24°C most crocodilians are sluggish, hesitant to eat, have difficulty metabolizing meals, and cannot fight infection (Lang 1987), while prolonged exposure to temperatures below 20°C can be fatal (Brandt and Mazzotti 1990, Coulson and

Hernandez 1983). Additionally, cooler temperatures may have a significant impact on survival through alteration of social dynamics (Lang 1977, 1987), anti-predatory defense mechanisms (Lang 1987, Markwick 1998, Warner and Huggins 1978), or growth rates

(Blake and Loveridge 1975, Hutton 1989, Perez and Escobedo-Galván 2009). Finally, cooler temperatures will result in decreased humidity that not only impacts distribution

133

and abundance of habitat (see below), but may also impact egg incubation through desiccation (e.g., Joanen and McNease 1989, Markwick 1998)

The crocodiles examined as part of this study can be considered habitat specialists, breeding exclusively in forested or wooded wetland habitats (Eaton 2010,

Shirley 2010, Waitkuwait 1989). To date no nests for either species have been found outside of canopy cover, even when such habitat is amply available in occupied sites

(e.g., Akaka, Gabon). This is in contrast to other mound nesting species that regularly build nests in floating grass mats, reed beds, and other open habitats. Climatic changes that impacted forest distribution and wetland abundance likely affected the distribution and abundance of Mecistops and Osteolaemus, and other forest dependent crocodilian species, simply as a result of habitat availability and, possibly, through increased competition from competitively superior sympatric crocodilians. The sympatric Crocodylus niloticus and C. suchus rely on open nesting and basking habitats and decreases in forest cover may have increased habitat availability in these two species which are thought to out-compete Mecistops and Osteolaemus for space and food resources (M. Shirley, pers. obs.).

Estimated pre-bottleneck effective population sizes were both comparable between species and large. While there was some incongruence between the two different marker datasets, generally speaking effective population sizes for both crocodiles were small (i.e., < 50 - 1000) since the bottleneck. Effective population size is directly related to population, or species, level genetic diversity and, as such, increases in Ne are not only affected by census size but by the combined forces generating genetic diversity within populations (Charlesworth 2009). The ratio of

134

census to effective population size has been found to range from 4 – 10 or more (e.g.,

Frankham 1995), which may still indicate large census crocodilian populations throughout the study area.

Several studies have shown a positive linear relationship between mitochondrial genetic diversity and census population size for mammals (e.g., Mulligan et al. 2006), which may be related to the possible negative relationship between per generation mutation rate and effective population size (Piganeau and Eyre-Walker 2009). In this study, model-based estimates of crocodilian mitochondrial mutation rate were low (<

1%/site/million years ) compared to estimates for mammals (e.g., Pesole et al. 1999), for example, but were congruent for previous estimates from other ectotherms (Avise et al. 1992, Caccone et al. 2004, Formia et al. 2006). This suggests support for the relationship posited by Piganeau and Eyre-Walker (2009) and the maternal effective population sizes estimated from mitochondrial demographic reconstructions may simply reflect an inability to recover post-bottleneck genetic diversity, and therefore effective population size, in crocodilians despite large census sizes (though see Nabholz et al.

2009).

While the results presented here are dramatic, they should be viewed cautiously.

The demographic reconstruction methods used as part of this study are sensitive to both prior parameterization and several key model assumptions. The prior inputs for the genealogical root height in the BSR analyses may have introduced too much uncertainty in the coalescent time parameter to precisely estimate the skyride, as suggested by the wide confidence limits. That these priors were taken from a deep phylogenetic analysis may, at least partially, be the problem and could be overcome by

135

more informative prior estimation of intra-lineage coalescent times and/or clock rates

(Ho and Larson 2006). However, independent runs were also made fixing the clock rate to the 5%, 50% and 95% posterior estimates from the unrestricted analyses, which only decreased the confidence intervals on Ne slightly, suggesting the data were not greatly informative (see below), but showed similar patterns of population stability/increase, for different time scales (example Fig. 4-8). In contrast, it is clear that results from the microsatellite model were robust as changes in priors on generation time and starting conditions did not qualitatively change them. However, most parameters estimated by this model also had fairly wide confidence limits suggesting that additional loci may be needed for better precision.

Perhaps the most critical assumption for the results of both models is that an appropriate rate of substitution is established. Fortunately, the models implemented here are advantageous because they can estimate substitution rates as parameters of the models when reliable, external rate calibrations are available (Beaumont 1999,

Drummond et al. 2005, Storz and Beaumont 2002). In my case, genealogical root calibrations were available from calibrated divergence time estimation detailed in

Chapter 1 for both species. Additionally, the skyride analysis was implemented without calibration on the root height where the results (not shown) showed similar demographic profiles though on an unknown timescale. While the resolved demographic patterns seem to be robust to prior parameterization, results from both models would certainly be improved with more informative prior estimations of mutation rate if only through increased precision in Ne estimates.

136

Both methods are sensitive to substructure in the dataset and assume panmixia

(Drummond et al. 2005, Goossens et al. 2006, Ho and Shapiro 2011). The effect of population substructure on demographic reconstruction methods has yet to be explicitly tested, however, as genetic structure is known to effect estimates of mutation rate (e.g.,

Miller et al. 2009) and demographic reconstructions rely explicitly on mutation rate, it is likely that demographic profiles estimated from structured datasets are, in fact, representative of that structure and not the demographic history (Goossens et al. 2006,

Pannell 2003). Results detailed in Chapter 3 suggest that for both species there is no expectation of intra-dataset substructure and, therefore, the patterns observed here are actually representative of demographic history. However, the realities of crocodilian mating systems make it possible that the individuals sequenced as part of this study are closely related and the use of relatives has been known to confound coalescent-based analyses in a way similar to dataset structure (Strimmer and Pybus 2001).

Finally, both methods are subject to dataset information content, i.e., relative genetic diversity (Girod et al. 2011, Ho and Shapiro 2011). This is probably the largest concern for the datasets presented here, especially as evidenced in the wide confidence limits of the BSR plots (Drummond et al. 2005). Sequence diversity within mitochondrial datasets was low (Chapter 3) and it may be that multiple, or distant, changes in mitochondrial effective population size are simply not detectable.

Interestingly, it is likely that the low mitochondrial genetic diversity is reflective of the population bottlenecks diagnosed from the microsatellites. Unfortunately, there are no studies that quantify the level of diversity needed to estimate demographic histories over varying timescales. To combat this, future efforts might consider looking for more

137

variable sequence markers (e.g., control region, though the increased probability of homoplasy makes the use of this marker questionable) and/or sequence a much larger sample size of individuals to ensure more rare alleles are captured. Unfortunately, neither of these options is likely viable at the intra-population level for crocodilians either for the reasons discussed in Chapter 3 (i.e., low haplotype/sequence diversity for all crocodilians) or because the information returns may not be worth the effort and cost.

Additionally, several realities of crocodilian biology, including extensively overlapping generations and known recent declines (and recoveries in some cases) of census population sizes, may make demographic reconstruction results questionable (e.g., see

Luikart et al. 2010 for a review of Ne estimation model/method assumptions).

138

Table 4-1. Prior settings and posterior estimates of population parameters from msvar runs for Mecistops populations. The same range of priors was used for each chain, though starting values varied widely both between chains and between loci within chains. Values were converted from log to linear scale. Lower and upper bounds are 95% HPD values. Population Parameter Lower Bound Mean Upper Bound N0 Prior 10 1480 15000 N1 Prior 10 6323 96960 T Prior 10 6.36E08 9.0E09 Akaka N0 Posterior 3 41 555 N1 Posterior 1324 13001 134586 T Posterior 320 5559 75858 Azingo N0 Posterior 1 15 296 N1 Posterior 1599 16218 156314 T Posterior 83 1949 37239 Bongo N0 Posterior 1 14 432 N1 Posterior 1967 19230 186638 T Posterior 54 2239 58614 Dji Dji N0 Posterior 2 50 914 N1 Posterior 1091 13152 162555 T Posterior 190 5754 127350 Mpassa N0 Posterior 1 24 478 N1 Posterior 1387 25527 452898 T Posterior 425 11967 266072 Coast N0 Posterior 4 70 1208 N1 Posterior 1241 12560 132129 T Posterior 264 6011 103992 Ogooué N0 Posterior 2 49 929 N1 Posterior 1276 13092 146893 T Posterior 160 4064 93972

139

Table 4-2. Prior settings and posterior estimates of population parameters from msvar runs for Osteolaemus tetraspis populations. The same range of priors was used for each chain, though starting values varied widely both between chains and between loci within chains. Values were converted from log to linear scale. Lower and upper bounds are 95% HPD values. Population Parameter Lower Bound Mean Upper Bound N0 Prior 10 1697 15000 N1 Prior 10 7272 96960 T Prior 10 5.59E08 9.0E09 Akaka N0 Posterior 10 145 2198 N1 Posterior 1343 11803 104954 T Posterior 158 3273 56885 Abanda N0 Posterior 3 91 2307 N1 Posterior 1517 12677 112461 T Posterior 56 2113 58749 Abanda Caves N0 Posterior 2 57 1282 N1 Posterior 1172 11695 125314 T Posterior 102 3273 72611 Iguela N0 Posterior 20 236 2831 N1 Posterior 1435 15959 178238 T Posterior 528 9397 170608 Loango N0 Posterior 50 458 3945 N1 Posterior 1905 15996 142233 T Posterior 972 10691 108393 Lope N0 Posterior 15 309 4677 N1 Posterior 1256 12162 116145 T Posterior 288 9226 234423 Mayumba N0 Posterior 27 274 2754 N1 Posterior 1409 15241 161808 T Posterior 931 13458 194536 Rabi N0 Posterior 16 253 3192 N1 Posterior 1175 13490 143549 T Posterior 358 9572 231740

140

Table 4-3. Posterior estimates of population parameters from the Bayesian skyride model for all populations of both Mecistops and Osteolaemus. Root height is years before present. Clock rate is substitutions/site/thousand years. Lower and upper bounds are 95% HPD values. Root Height Clock Rate Species Population Lower Median Mean Upper Lower Median Mean Upper Mecistops Coast 500 1,037 3,709 10,840 1.23e-7 6.32e-5 7.53e-5 1.87e-4 Ogooué 250 519 2,209 5,652 6.28e-8 1.50e-4 1.90e-4 5.11e-4 All Gabon 250 510 1,878 5,353 1.72e-7 1.34e-4 1.52e-4 3.54e-4 Osteolaemus Abanda 50 107 433 1,191 2.01e-6 4.14e-3 5.30e-3 1.43e-2 Caves 50 105 438 1,155 1.76e-6 4.01e-3 4.99e-3 1.29e-2 Abanda + 50 104 407 1,145 4.45e-6 3.29e-3 3.84e-3 9.40e-3 Caves Coast I 50 99 368 942 6.29e-6 4.22e-3 4.62e-3 9.56e-3 Coast II 50 101 431 1,077 3.47e-6 4.92e-3 5.17e-3 1.09e-2 Lope 25 53 218 541 7.69e-7 3.16e-3 4.77e-3 1.46e-2 Ogooué 25 52 256 572 7.75e-6 8.93e-3 1.09e-2 2.81e-2 All Gabon

141

Figure 4-1. Bayesian skyride plots of effective population size change in Mecistops populations over time. The solid line represents the change in the median estimated Ne over time, while the dashed lines are the upper and lower confidence limits of Ne. Population sizes were un-scaled by generation (i.e., divided by the relative generation time, in this case 0.015 thousand years) prior to being put in log scale.

142

Figure 4-1. Continued.

143

Figure 4-2. Bayesian skyride plots of effective population size change in Osteolaemus populations over time. The solid line represents the change in the median estimated Ne over time, while the dashed lines are the upper and lower confidence limits of Ne. Population sizes were un-scaled by generation (i.e., divided by the relative generation time, in this case 0.008 thousand years) prior to being put in log scale.

144

Figure 4-2. Continued.

145

Figure 4-2. Continued.

146

Figure 4-2. Continued.

147

Figure 4-3. Magnitude and timing of population bottlenecks in Mecistops populations. N0 is the estimated contemporary effective population sizes while N1 is the effective population size prior to the demographic event. T is the estimated time since the noted demographic event for each population.

148

Figure 4-4. Magnitude and timing of population bottlenecks in Osteolaemus tetraspis populations. N0 is the estimated contemporary effective population sizes while N1 is the effective population size prior to the demographic event. T is the estimated time since the noted demographic event for each population.

149

Figure 4-5. Timing of population bottlenecks for Mecistops and Osteolaemus populations with time in linear scale. The figure illustrates the posterior support for estimated time since population decline. Vertical lines represent (from left to right) 50 YBP, 500 YBP, 10000 YBP, and 18000 YBP.

150

Figure 4-6. Magnitude and timing of population bottlenecks for alternative, test models in the coastal population of Mecistops. N0 is the estimated contemporary effective population sizes while N1 is the effective population size prior to the demographic event. T is the estimated time since the noted demographic event for each population. Vertical lines on the T plot represent the mean estimated time since population decline from the final models.

151

Figure 4-7. Magnitude and timing of population bottlenecks for alternative, test models in the Abanda and Akaka populations of Osteolaemus tetraspis. N0 is the estimated contemporary effective population sizes while N1 is the effective population size prior to the demographic event. T is the estimated time since the noted demographic event for each population. Vertical lines on the T plot represent the mean estimated time since population decline from the final models.

152

Figure 4-8. Bayesian skyride plot for the coastal population of Mecistops illustrating Ne estimates over time under three different fixed clock rates. The 50% (top), 5% (bottom right) and 95% (bottom left) posterior rate estimates from runs using priors on root height to concurrently estimate the clock rate. The Y-axis is generation-scaled Ne and the X-axis is time in thousands of years before present. It is clear that removing uncertainty in the clock rate only slightly affected confidence limits on the estimated Ne’s while more clearly illustrating the trend. Vertical dotted lines are the lower (light) and upper (dark) 95% posterior estimates on tree root height. Estimates from before the lower 95% HPD limit should be viewed cautiously.

153

CHAPTER 5 CONCLUSIONS AND IMPLICATIONS FOR CONSERVATION

The central objective of this research was to investigate the phylogenetic, phylogeographic and spatial relationships of genetic diversity in two African crocodiles.

Specifically I set out to determine: 1) if the African slender-snouted crocodile (Mecistops cataphractus) was, like other African crocodiles, a cryptic species complex with reciprocally monophyletic lineages corresponding to major African biogeographic zones;

2) the spatial patterns of and processes structuring genetic diversity in Central African

Mecistops and Osteolaemus; and 3) if significant demographic events were detectable in these species and at what time scales. Exploring these issues in a comparative framework utilizing two, sympatrically distributed crocodiles with divergent ecologies not only significantly increased our understanding of the evolutionary history of Mecistops and Osteolaemus lineages in western Africa, but provided considerable insight into our understanding of regional Central African phylogeography which, until this study, was restricted to studies of terrestrial biota like elephants, gorillas, and forest flora.

I described two divergent, allopatric lineages within Mecistops whose distributions correspond to the Upper Guinea and Congolian biogeographic zones. The geography and timing of Mecistops lineage vicariance was congruent with the pattern seen in both

Osteolaemus and African Crocodylus where each genera showed a split ±8 mya at the

Cameroon Volcanic Line (this study, Eaton et al. 2009, Hekkala et al. 2011).

Unfortunately, none of these studies, including the present dissertation, incorporated samples collected from known localities in Nigeria and, to date, only a single genetic sample from Cameroon for Osteolaemus has been published. However, Mecistops skull specimens collected from unknown localities in Cameroon and Nigeria were

154

morphologically diagnosed to the Central African clade. This is congruent with unpublished data for Osteolaemus which found O. tetraspis to extend across the CVL into western Cameroon (N. Smolensky, pers. comm.). These results, however, may not refute the proposal of the CVL as the predominant vicariant feature today. Final uplift of these highlands was temporally heterogeneous allowing populations of Central African lowland species to bleed through, as is seen, for example, with the isolated Cross River gorilla (Gorilla g. diehli) and drill (Mandrillus leucophaeus) – members of otherwise entirely Congolian genera found between the Cross and Sanagha Rivers. Bleed- through and relict populations in western Cameroon/eastern Nigeria would be isolated from the Upper Guinea biome by the north-south series of Nigerian plateaus (i.e.,

Mambilla, Jos, Adamawa, and Obudu). Sampling in the isolated Cross and Niger River basins would improve our understanding of species distribution boundaries and the biogeography of the Upper versus Lower Guinea regions.

The spatial genetic structure and inferred divergent processes of basin isolation and population expansion in Central African Mecistops and Osteolaemus suggest that both regional processes and species-specific ecology are important in shaping the evolutionary histories of Central African fauna. Mecistops exhibited spatial partitioning of both mtDNA and nuclear markers corresponding to coastal Gabon, Ogooué drainage, western Congolian region of Lac Tele, and the eastern Congolian Lomami drainage and

Lake Tanganyika. However, widely distributed haplotypes ranging from the Ogooué drainage through the Sanagha to the Ituri region of NE Congo belie a history of large, well-connected populations throughout the distribution of Mecistops that may have only recently (< 3,000 YBP) been fragmented. All analyzed Mecistops populations revealed

155

a significant demographic decline corresponding to the period at or after the Last Glacial

Maximum. Unlike Mecistops, Osteolaemus showed significant species-level divergence between the Ogooué and Congo Basins (this study, Eaton et al. 2009). Additionally,

Osteolaemus exhibited very limited geographic structuring between the Gabonese coast and the Ogooué drainage; but unique mitochondrial haplotypes limited to the coast, as well as a north to south clinal pattern in microsatellite allele frequencies, suggests a possible pattern of population expansion from one or more core refugia followed by isolation by distance. Similar to Mecistops, however, all analyzed Osteolaemus populations exhibited significant demographic decline corresponding to the period at or after the Last Glacial Maximum with stable effective populations contemporarily.

Together, the patterns of genetic variation exhibited by Mecistops and

Osteolaemus can be interpreted with respect to what is known about their biology. For example, despite geologic isolation of the Congo and Ogooué basins, Mecistops was able to maintain enough population connectivity for common mitochondrial and nuclear alleles to be found over a distance of nearly 2,200 km, while Osteolaemus populations from each basin are reciprocally monophyletic. This might suggest that, while both species are dependent on forested waterways for reproduction, Osteolaemus is more dependent on extensive closed canopy, flooded or well-drained, forests and would retract into forest refuges during glacial maxima and other periods of low temperature/humidity as seen in climate cycles starting in the late Miocene through the mid-Holocene. However, proliferation in lowland forests should facilitate relative homogeneity across heterogeneous, largely “terrestrial” landscapes. In contrast, the ready genetic distinction of Mecistops in isolated drainage areas supports dependence

156

on larger rivers and wetlands which, when not well connected, would lead to divergence despite large populations. Both of these ideas are supported by crocodilian habitat selection where Mecistops is predominantly found in the main river courses and larger wetlands while Osteolaemus prefers smaller, more heavily vegetated watercourses and flooded forests. Additionally, Mecistops is rarely found away from water while

Osteolaemus has been known to make extensive terrestrial forays.

Radio telemetry data from both Mecistops (M. Shirley, unpub. data) and

Osteolaemus (M. Eaton, unpub. data) suggest that these crocodiles maintain much smaller home ranges and probabilities of movement compared to other crocodiles.

Small home ranges and species-specific habitat preferences suggest that populations of both species would have to have been large and/or relatively evenly distributed to maintain such large swaths of genetic homogeneity. The demographic analyses in

Chapter 4 showed that both species had effective population sizes numbering in the thousands to tens of thousands prior to population collapse in the late Pleistocene or

Holocene suggesting that this was the case. Identification of demographic declines dating to the period at or after the Last Glacial Maximum highlights for the first time the importance of regional environmental trends like climate change in shaping crocodilian genetic diversity and structure despite known contemporary declines due to anthropogenic threats.

The most likely mechanisms by which this reduction in genetic diversity occurred

(i.e., significant reduction of forest habitat and population size through decreased temperatures) is being replicated by human-mediated land cover change and population decreases with hunting in western Africa. Projecting into the future, these continued

157

pressures may, therefore, never permit the recovery of genetic diversity and will likely increase inbreeding and isolation. The long-term effects of low genetic diversity, inbreeding depression, and the resultant decreased adaptability to changing conditions, are only barely understood beyond a theoretical basis (reviewed by Amos and Balmford

2001, Frankham 2005). The results of my dissertation, and other studies of crocodilians, suggest that the Crocodylia are an interesting study system for this considering that they are genetically depauperate compared to most vertebrates but within the Order show levels of genetic variation that differ by an order magnitude.

Despite their low genetic diversity, crocodilians are viewed as incredibly robust taxa with high resistance to disease, infection, and parasite loading, no known modern species or population extinctions due to non-anthropogenic factors, and unparalleled ability to recover from highly diminished, fragmented populations (Forrester and Sawyer 1974,

Merchant et al. 2006). This is perhaps best exemplified by the recovery of the American alligator which now numbers in the millions following dramatic hunting-related declines since the 1800’s (Elsey and Woodward 2010), has very low genetic diversity (Davis et al. 2002, Dessauer and Elsey 1993, Glenn et al. 2002, Menzies et al. 1979), and shows no signs of reduced fitness, developmental capacity, or other long-term effects of inbreeding depression.

The conservation implications of my results are significant. Samples utilized in this research were collected as part of crocodile population surveys covering ±3000 km of crocodile habitat at nearly 200 sites split evenly between West and Central Africa

(Shirley et al. 2009, M. Shirley unpub. data). While nearly 1,800 individual Mecistops were encountered over the course of these surveys, only 40 confirmed sightings were in

158

West Africa – 14 in the Upper Guinea region of Ghana and Cote-d’Ivoire and 26 in a single national park in The Gambia. Further, only one of the 40 animals encountered in

West Africa was in the adult size class (>2 m total length). In contrast, surveys in

Gabon and DRC regularly encountered a heterogeneous array of size classes, as well as evidence of reproduction (M. Shirley, unpub. data). The slender-snouted crocodile was last reviewed for the IUCN Red List in 1996 at which time its status was listed as

Data Deficient. It is now likely that we have enough information to revise its status and recognizing distinct Mecistops taxa in West and Central Africa underscores the urgency.

Abundant populations of the Central African Mecistops sp. nov. are more or less restricted to Gabon with evidence for fragmented and declining populations in other range states (Shirley 2010, unpub. data). Throughout Central Africa, however, it is continually threatened by bushmeat hunting, artisanal fisheries, and habitat loss. For these reasons Central African Mecistops likely qualifies for Vulnerable status. In contrast, there are no known abundant M. cataphractus populations in West Africa and the only ones that may be close to sustainable are, at best, tenuously protected in a few national parks (e.g., River Gambia National Park, Sapo National Park, Tai National

Park, Comoe National Park, Bia Resource Reserve, and the Ankasa Resource

Reserve). The major threat to this species is predominantly habitat loss, though hunting has certainly contributed to the declines. For these reasons, Mecistops cataphractus likely qualifies for Endangered or Critically Endangered status (Shirley et al. 2009).

To date, only a single paper has attempted to identify units for crocodile conservation below the species level. Thorbjarnarson et al. (2006) identified 69

Crocodile Conservation Units (CCU) in nine delineated bioregions focusing on habitat

159

priorities explaining that “planning for threatened species conservation ideally requires maintenance of viable populations across the full range of ecosystems in which they exist.” Justification for this perspective was largely based on maximizing the conservation of ecological variability – which in taxa like crocodilians may or may not reflect the actual processes contributing to intraspecies diversification. In other words, are American crocodiles in Caribbean Central America actually ecologically or evolutionarily different from those in Caribbean South America as posited by

Thorbjarnarson et al. (2006)? Is it likely that coastal populations in both regions are even isolated and progressing on their own ecological or evolutionary trajectories? The answers to these questions, at least for Crocodylus acutus, are beyond the scope of this dissertation. However, an alternative justification for delineating priority regions within which to identify CCU’s may be found in the evolutionary conservation paradigm which advocates the conservation of evolutionary process through evolutionary significant lineages or populations (Crandall et al. 2000, Ferriere et al. 2004, Moritz 1994). Under this context, the results of my research can help us to better define priority regions for crocodile conservation in Central and West Africa.

In Central Africa it is clear from both Mecistops and Osteolaemus that crocodiles in the Congo Basin and Lower Guinea bioregions are evolving along independent trajectories and face different levels of threat and, therefore, should be viewed as different bioregions for conservation decision-making. In the Congo Basin, both species exhibited considerable genetic distinction between drainages running north versus south into the Congo River and, perhaps, these two regions should be considered separately. The most divergent haplotypes found in Mecistops were from samples

160

taken in Lake Tanganyika suggesting it has been isolated from populations further west and that it, possibly even split into east and west banks, as well as Lake Mweru, should be designated as distinct bioregions. In the Lower Guinea biome, populations along the coast and from the main Ogooué drainages showed high degrees of divergence and recognition. While the coast can likely be considered a single unit for conservation, the conservation regions along the Ogooué may need to be partitioned at the tributary level pending validation through additional sampling, especially of the northern tributaries

(i.e., Ivindo River drainage). At this stage it is unclear into which region the tributaries of the Sanagha fit. Delineating and protecting CCU’s within these 7 or so regions should be a straightforward proposition. Protected areas (mostly national parks) are already well placed within each region and can ensure the sustainability of viable populations regardless of changes in legislation that may affect populations outside of protected areas. Fortunately, all crocodile species are listed as integrally protected species in the six Central African countries providing a legislative framework for their conservation.

Unfortunately, unprotected crocodile populations are already highly reduced in five of these six countries. Because of this, and ongoing threats to both protected and unprotected populations, future conservation efforts can succeed in protecting CCU’s across Central Africa by simply enforcing existing species and protected area policies to mitigate threats from bushmeat, habitat loss, and unsustainable fishing practices.

Delineating and protecting CCU’s is more complicated in West Africa. Few samples were available west of Cameroon; however, the available data from both

Mecistops and Osteolaemus suggests that the Upper Guinea and Senegambian forest blocks should be considered distinct bioregions for crocodile conservation planning.

161

Additionally, unique haplotypes found in nearly all Mecistops samples suggest that crocodile populations in the Upper Guinea region are highly structured by drainage, which predominantly run north to south and show little lateral connectivity. Because of this, further sampling will be required before evolutionary-based CCU’s can be identified in the Upper Guinea region. Like for Central Africa, crocodiles are integrally protected in all West African countries, but the similarities end there. Protected areas, especially those covering forest blocks, are few and far between and, due to already extensive forest degradation, are becoming more ideal habitat for Crocodylus suchus. Both protected and unprotected populations of these crocodiles are under severe threat of habitat loss, hunting for food, and an extensive trade for traditional medicine and religious uses. Conserving these species in West Africa will require action beyond simply enforcing existing legislation and may need to incorporate activities like captive breeding and reintroduction. That being said, without a considerable change in the environmental ethic in West Africa the future for all wildlife is bleak.

By utilizing an integrative molecular framework combining phylogenetics, spatial genetics, and coalescent-based modeling of population structure and demographic history, I have highlighted critical processes in western Africa that maintain and structure intraspecies diversity. The futures of both Mecistops and Osteolaemus in

West Africa, especially in light of taxonomic distinction, are increasingly tenuous.

However, Central African Mecistops and Osteolaemus, especially within the Lower

Guinea Ogooué Basin region, may have more secure futures not only due to larger population sizes but the convenient placement of national parks already coinciding with identified evolutionary significant units. Throughout this dissertation I have tried to

162

underscore the importance of evolutionary thinking in future crocodile conservation endeavors and, because of this, believe I have established a solid foundation from which existing conservation legislation in Africa can be applied to the long-term sustainability of crocodile populations.

163

APPENDIX A REPRINT OF HISTORICAL DESCRIPTIONS OF MECISTOPS TAXA

The following are transcriptions and/or translations to English of documents describing crocodile taxa now recognized as Mecistops cataphractus. The purpose of this is two-fold: 1) provide ready access to difficult to find historic, taxonomic literature on this species and, 2) provide English translations for ease of accessibility to otherwise

French and German documents. In all cases, the texts were transcribed and/or translated without alteration. The original texts are written in block quote format while my comments precede and follow in the formatting style found throughout the rest of the dissertation.

Cuvier 1824 – Crocodilus cataphractus

Cuvier (1824) originally described Crocodilus cataphractus as part of his influential series: « Recherches sur les ossemens fossiles de quadrupedes ou l’on retablit du plusieurs speces d’animaux que les revolutions du globe paroissent avoir detruites ».

He gave the crocodile the common name “crocodile a nuque cuirassée” which roughly means “armor-necked crocodile” presumably in reference to the extra rows of nuchal scutes compared to other crocodiles. This description is based on a single specimen observed in the Museum of Surgeons – London (now the Royal College of Surgeons

Museum – London). The following is translated from French and comprises the entire text provided by Cuvier:

I observed this species in 1818 at the Museum of Surgeons in London, where we preserve a dried specimen. Its snout is much longer and narrower than the crocodile of Saint Domingue [now Hispaniola]; the length of its head was 2.5 times the width. Its snout has neither the convex protrusion characteristic of the crocodile of St. Domingue nor any other notable mark. It has 16 teeth on each side of the upper jaw and 15 per lower. The cranial fossa can be seen through the skin as in other crocodiles. Aside from the snout, it is best characterized by the nuchal

164

shield; after two, isolated, ovular plates, and a row of four small, equally ovular and isolated plates, come five scale rows which are contiguous between them and with the dorsal scales and are formed of two large square scales each. The first two pairs are largest; the next three gradually reducing in size; and together they form an armor plate on the nuchal stronger than any caiman or gavial. The dorsal scales are keeled and situated in transverse rows of six each, except the first two which only have four. This species is obviously distinct from all the others that I described in my first edition; unfortunately we don’t know anything about its origin.

The most unfortunate part of this description, as becomes a repeated theme with descriptions of cataphractus and its congeners/conspecifics, is a real lack of description of more than the obvious characters which immediately diagnose it from other crocodilians (e.g., head length to head width ratio and nuchal scalation) prohibiting a solid understanding of the character variation within this species. Additionally, since there was no description of any cranial features this type description does not provide a basis for assigning specific epithets on the basis of morphology. Unfortunately, the specimen from which Cuvier based this description can no longer be found at the Royal

College of Surgeons Museum and is presumed to have been destroyed during the extensive bombing of London during WWI and WWII, or to have been transferred to, and subsequently lost from, the British Museum – where it also cannot be found. That

Cuvier did not know the origin of the specimen/species at the time of his description further hampers our ability to assign a specific epithet to the molecularly diagnosed clades despite the fact that its type locality was later restricted to the Senegal River

(Fuchs et al. 1974a).

Bennett 1835 – Crocodilus leptorhynchus

Bennett (1835) offered the description of Crocodilus leptorhynchus at a meeting of the Zoological Society of London and his oral description of the species was transcribed and published by Yarrell (1835). Here I provide an abridged version of the account,

165

which largely revolves around description of the internal organs and viscera and not relevant for comparison to M. cataphractus.

[W]hich he regarded, while it was living in the Society’s Gardens, as referable, on account of the length of its head and the extent of the shielding at the back of its neck, to the Crocodilus cataphractus, Cuv. A more close examination of it, however, subsequent to its death, had shown him that its head was still more prolonged than that part is described to be in Croc. cataphractus, its length being to its breadth as 3 to 1, instead of as 2.5 to 1 : it is also deficient of the second post-occipital series of four small plates noticed as occurring in Croc. cataphractus. On these accounts principally he stated that he considered it as representing a previously undescribed species, which he characterized as

CROCODILUS LEPTORHYNCHUS. Croc. rostro elongate, capitis latitudine longitudinis partem tertiam æquante; scutis post-occipitalibus ovalibus parvis duobus, nuchalibus per paris quator cataphractus, cum dorsi seriebus continuis.

Long. Tot. 27 une.; cranii, 4·5/8; cranii, ad maxillarum commissuram, lat. 1·7/8. Hab. apud Fernando Po.

There are several points of note here. First, the description of C. leptorhynchus is based almost entirely on the ratio of head length to head width. Second, Bennett additionally remarks on the absence of certain post-occipital scutes, a character that is notoriously variable in all living crocodilians. Finally, the examined specimen was a captive at the Zoological Society of London’s Gardens and is of unknown wild origin.

The locality notation “apud” is commonly used in the botanical taxonomic literature to signify “near”, “next to,” or “beside,” in this case Fernando Po. In reality, Fernando Po was an incredibly poor choice of geographic descriptor, even with the modifier, as this island has never had a population of Mecistops and, further, is an extension of the proposed feature on the landscape keeping the western and central clades in vicariance

– the Cameroon Volcanic Line.

166

Gray 1844 – Mecistops

Gray (1844) ascribes the genus Mecistops to both Crocodylus cataphractus and

C. leptorhynchus, as well as changes the specific epithet of leptorhynchus to bennettii.

This latter change of specific epithet is invalid because the International Code of

Zoological Nomenclature only allows for a change in specific epithet with a change in genera if the change in genera will result in a secondary, or later, specific homonym. As there was no conflict with a pre-existing species Mecistops leptorhynchus, M. bennettii is an invalid name and as such I designate it is a junior synonym of M. leptorhynchus.

Additionally, the code requires correct adoption of Latin grammar when designating names. In this case, the correct spelling of M. bennettii is actually M. bennetti as it is named after a male person. This alone does not invalidate the name, but would require formal correction if the name were valid. Gray offers the following descriptions of the genus Mecistops and two species:

MECISTOPS. Gavialis, Muller. – Jaws oblong, slender, flat, without ridges. Teeth unequal, lower canines fitting into a notch in the side of the upper jaw. Feet fringed; toes webbed to tip. The cervical plates in 3 or 4 cross series, united to the dorsal shield. Males without any swelling in front of the nostrils.

BENNETT’S FALSE GARIAL. Mecistops Bennettii. Crocodilus leptorhynchus, Bennett, Proceedings of the Zoological Society 1835, 129. C. cataphractus var. Bennett, P.Z.S. 1834, 110.

Muzzle elongate, very slender, rather tapering, with a small tubercle on each side over the second canines, which are rather in front of the middle; nuchal shields 2, far apart, small, with some scattered ones; cervical shields 4, in two bands; dorsal shields equal, square, bluntly keeled, the two first small.

THE FALSE GARIAL. Mecistops cataphractus. Crocodilus cataphractus, Cuv. Oss. Foss. V. t. 5, f. 1,2. Gray, S.59. Dum. Et Bib, E. G. iii. 126.

Nuchal shields 6, in two rows, two front apart, four hinder in a line; cervical shields 10, in pairs, two front large, three hinder pairs smaller. Inhab. W.

167

Africa, in the River Galbar, near Sierra Leone. Probably only a var. of the preceding.

A third species, M. journei, was additionally described. However, this description clearly applies to Tomistoma schlegelii and is not presented here. As with Bennett

(1835), Gray relies on very few morphological characters to distinguish these two species, though he seems to recognize the inherent variability in the characters used and suggests that the two species are likely the same. However, it is unclear why he suggests M. cataphractus is likely a variant of M. bennettii and not the other way around considering that M. cataphractus has temporal priority.

Baikie 1857 – Description of a Nigerian Skull

Baikie (1857) offers the following (abbreviated) description of a Mecistops skull collected in the town Ojogo along the “Binue River” in Nigeria. The underlined texts are character descriptions that fit well with my descriptions of the same skull characters in

M. cataphractus (West clade).

The skull seems from its appearance to be that of an adult animal. Its extreme length is 22.25 inches, the greatest breadth being 9.25 inches, or nearly in proportion of 2.5 to 1. From this it may be inferred to be most probably M. cataphractus, that being the proportion of the length to the breadth in that species, while in M. bennettii (if distinct) it is said to be as 3 to 1… The skull is considerably depressed, much produced anteriorly, and the extremity of the snout somewhat enlarged. Upper surface smooth. Forehead nearly flat, pitted, sides not raised, converging anteriorly. Cranial fossae nearly circular, resembling those of the Gavial. Orbits rather more convergent than in the Crocodiles, and the nasal aperture more circular. Nasal bones more prolonged than in Gavialis, yet not reaching, as in Crocodili, the nasal opening, but distant from it an inch and a half. Anterior spine of the middle-frontal very long, slender, tapering, and pointed. Lacrymal bones lengthened and narrow… Anterior palatine foramen small. Palatine bones tapering and pointed anteriorly.

Baikie also describes anecdotally how rare this species seems to be, this being the first specimen of such he has ever seen and relating stories of colleagues who

168

were, presumably, as well travelled in West Africa and never seen an individual. While this has no taxonomic bearing, it supports the notion that this species may have always been rare in West Africa, especially in relation to Nile crocodiles, and may be revealing about its relative status today.

Fuchs et al. 1974 – Crocodylus cataphractus congicus

Fuchs et al. (1974) reviewed the status of C. cataphractus and included descriptions of its skin characteristics. As part of the research for this description, they feel as though there was evidence for a unique subspecies, C. c. congicus, on the basis of differing scalation characteristics, particularly those of the ventral and flanks. This reprint is translated from the original German with the assistance of R. Sommerlad and

A. Meurer. I provide the Fuchs. et al. (1974) descriptions of both C. c. cataphractus and

C. c. congicus to aid the diagnostic process.

Crocodylus cataphractus cataphractus, Cuvier.

TYPE: Cuvier described the species from a skeleton whose geographic origin was not known and cannot be discerned by the study of recognizable osteological characteristics as such differences between the subspecies diagnosed here are not yet determined. It is also not known to us if this type still exists in the natural history museum of Paris. Linnaeus mentions for his mongrel species (Gavialis gangeticus + Crocodylus cataphractus) partim Senegal, and also Latreille gives the Senegal River as the locality for his Crocodilus Niger, which is possibly the same as Crocodylus cataphractus. Hence it is the obvious choice to designate the Senegal River as terra typica for the nominal race of this species.

There were continuous and excited discussions on the status of Crocodilus Niger. Gray (1862) designated this taxon as a synonym to Crocodilus frontatus Murray (Osteolaemus tetraspis). Strauch (1868) insisted on its synonymy with Crocodylus cataphractus, and Vaillant (1897) believed to be able to prove that the debatable species is in reality identical with Paleosuchus trigonatus. Should it turn out in the future that Crocodilus Niger really is the same as Crocodylus cataphractus, this would be no danger to the later name cataphractus as after the revised version of Article 79 of the "International Code of Zoological

169

Nomenclature" it has been used by far more than 5 authors in an immense number works since 1825.

DIAGNOSIS: A large-scaled subspecies with 20 to 24 transverse rows along the belly shields, posterior the nuchal collar to anterior the cloaca. Scales on the flanks (between the back shields and the belly shields) are relatively large and strongly keeled; only 4 flank scales in the middle cross row of the body. Collar clearly distinct between cross rows the belly shields before it and behind it; its shields strongly ossified; the middle shields twice as long as the middle shields of the cross row situated before it. Pores on the belly shields faint or not recognizable.

DISTRIBUTION: Western Africa, from Senegal eastwards and southwards up to northern Angola, excluding the distributional area of the eastern subspecies. The precise boundary to the other subspecies is not yet known.

Crocodylus cataphractus congicus, nov. ssp.

TYPE: A belly skin of 114 cm length. Catalogue No. 68129. Collection: Seckenberg Museum, Frankfurt am Main.

TERRA TYPICA: Middle Congo

DIAGNOSIS: A small-scaled subspecies with 25 to 26 transverse rows of scales along the belly shields (from posterior the nuchal collar to the anterior edge of the cloaca). Scales on the flanks are relatively small, only mildly ossified and only faintly keeled; in the middle of the body 6 scales in a transverse row. Collar less distinctive than with the nominal race; its shields only mildly ossified; the middle shields at most 1.5 times (mostly less) as long as the middle shields of the transverse row situated before it. Pores on the belly shields always visible, also in tanned and processed skins.

DISTRIBUTION: Middle Africa (Congo area, Zaire). A precise range delimitation compared to the distribution of the nominal race is not possible yet.

Future investigations will presumably unearth further differentiating characteristics between for these two subspecies. A possibly important characteristic may be the sides of the tail, which are dark colored in juvenile slender-snouted crocodiles (Mertens, in 1943). Whether this is with all representatives of the kind, whether this applies only for one or both subspecies, or whether this characteristic disappears with one race later and with the other race is persistent, should be object of further research.

170

It is unclear from these descriptions and, in fact, the entire study how many skin specimens were examined and if they were of known provenance. Unfortunately, another of Fuchs works (Fuchs et al. 1974b), which ascribed subspecies to Nile crocodiles based on skin characteristics, has never been well received primarily because the skins used were of largely unknown locality and therefore the extent to which character aggregations represent true geographic/population clusters is unknown.

It is likely the same is happening here, especially if this subspecies is based entirely on a single skin, for most of these scale meristic characters are known to vary widely within species. Based on a preliminary examination of my wild caught specimens from throughout the range of Mecistops, these scale count partitions do not hold up as specimens caught in The Gambia show scalation patterns described for C. c. congicus.

171

APPENDIX B DESCRIPTION OF MORPHOLOGICAL CHARACTERS FOUND TO VARY BETWEEN THE TWO MECISTOPS CLADES

Character A: Palatal Fenestra. More convex lateral and medial margins giving a generally more rounded appearance in the West clade compared to straighter lateral and medial margins and more narrow/squared appearance in the Central clade. This character description is being revised to assess longest length to fixed position width ration.

Character B: Palatine. In ventral view at the suture with the pterygoid, the posterior edge of the palatine is not acute, but transversely linear or broadly curved in the West clade compared to acute anterior projection of the pterygoid as a point along the midline in the Central Clade.

Character C: Premaxilla 1. In the dorsal view, the posterior-most point of the premaxilla is level with or anterior the second maxillary tooth in the West clade compared to even with or posterior the third maxillary tooth in the Central clade.

Character D: Premaxilla 2. In the dorsal view, the nasal aperture ranges from round to somewhat gumdrop-shaped in the West clade compared to heart-shaped to somewhat gumdrop-shaped in the Central clade. This character shows overlapping variation in the individuals that are more gumdrop-shaped, but round versus heart-shaped is fixed to either clade.

Character E: Premaxilla 3. In the dorsal view, the foramen of the first mandibular is small and most often not eroded through even in large individuals in the West clade compared to larger, often eroded, even in smaller individuals, in the Central clade.

Character F: Vomer. The vomer intrudes very little, if at all, into the nasal aperture in the West clade compared to actual intrusion in the Central clade. This character will be very difficult to code in some specimens depending on the preparation and damage to the vomer.

Character G: Nasal. In the dorsal view, the anterior most point of the nasal bone extends to even with or anterior the first maxillary tooth in the West clade compared to posterior the first maxillary tooth in the Central clade.

Character H: Nasal. Generally speaking the nasal bone is also much wider throughout its length in the West clade compared to the Central clade in which it continually tapers to a point. This character will be re-described to consider ratio of widest width to width at second maxillary tooth.

Character I: Prefrontal. In the dorsal view, the medial margin is in contact with the posterior point of the nasal bone for less than 40% the length of the suture in the West clade compared to more than 40% the length of the suture in the Central clade.

172

Character J: Lacrimal 1. The longest length to widest width ratio significantly smaller in the West clade compared to significantly longer in the Central clade.

Character K: Lacrimal 2. In the dorsal view, the posterio-lateral suture intersects the lateral margin of the orbit at a point either < 1/5 (West) or > 1/4 (Central) the length to the orbital bar.

Character L: Frontal 1. In the dorsal view, the anterior projection of the frontal tapers anteriorly evenly after contact with the orbit in the West clade compared to relatively parallel-sided until contact with the posterior-most nasal after which it tapers to a point in the Central clade.

Character M: Frontal 2. In the dorsal view, the frontal-prefrontal suture angles anterio- medially either transversely (West clade) or obliquely (Central clade).

Character N: Parietal. In dorsal view, the parietal either has a medial notch to accommodate the supraoccipital, which is a spine-like projection (West clade), or it does not because the supraoccipital is rounded and does not project into the parietal (Central clade).

Character O: Supraoccipital Tuberocities. In dorsal/occipital view, the supraoccipital tuberocities are highly flared/enlarged in the West clade compared to highly reduced/virtually non-existent in the Central clade. This character is incredibly hard to score in photographs and is very much dependent on how the skull was prepared.

Character P: Foramen Aërum. In dorsal/occipital view, this foramen is either posterior to (West clade) or parallel to or anterior (Central clade) the posterior-most projection of the paraoccipital process.

Character Q: Basioccipital – Exoccipital. In occipital view, the lateral margins are broadly flared dorsally in the West clade compared to more parallel-sided in the Central clade.

Character R: Pterygoids. In the occipital view they are less vaulted dorso-ventrally in the West clade compared to the Central clade.

173

LIST OF REFERENCES

Adegbie, A. T., R. R. Schneider, U. Röhl, and G. Wefer. 2003. Glacial millennial-scale fluctuations in central African precipitation recorded in terrigenous sediment supply and freshwater signals offshore Cameroon. Palaeogeography, Palaeoclimatology, Palaeoecology 197:323–333.

Agapow, P.-M., O. R. P. Bininda-Emonds, K. A. Crandall, J. L. Gittleman, G. M. Mace, J. C. Marshall, and A. Purvis. 2004. The impact of species concept on biodiversity studies. The Quarterly Review of Biology 79:161–179.

Amos, W., and A. Balmford. 2001. When does conservation genetics matter? Heredity 87:257–265.

Anisimova, M., and O. Gascuel. 2006. Approximate Likelihood-Ratio Test for branches: a fast, accurate, and powerful alternative. Systematic Biology 55:539 – 552.

Anthony, N. M., M. Johnson-Bawe, K. Jeffery, S. L. Clifford, K. A. Abernethy, C. E. Tutin, S. A. Lahm, L. J. T. White, J. F. Utley, E. J. Wickings, and M. W. Bruford. 2007. The role of Pleistocene refugia and rivers in shaping gorilla genetic diversity in central Africa. Proceedings of the National Academy of Sciences of the United States of America 104:20432–6.

Audubon, J. J. 1827. Observations on the natural history of the alligator. Edinburgh New Philosophical Journal 2:270–280.

Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. A. Reeb, and N. C. Saunders. 1987. Intraspecific phylogeography: the mitochondrial-DNA bridge between population genetics and systematics. Annual Review of Ecology and Systematics 18:489 – 522.

Avise, J. C., B. Bowen, T. Lamb, A. Meylan, and E. Bermingham. 1992. Mitochondrial DNA evolution at a turtle’s pace: evidence for low genetic variability and reduced microevolutionary rate in the Testudines. Molecular Biology and Evolution 9:457– 473.

Avise, J. C., and D. Mitchell. 2007. Time to standardize taxonomies. Systematic Biology 56:130–3.

Avise, J. C. 2000. Phylogeography: The History and Formation of Species. Harvard University Press, Cambridge.

Avise, J. C. 2009. Phylogeography: retrospect and prospect. Journal of Biogeography 36:3–15.

174

Axelrod, D. I., and P. H. Raven. 1978. Late cretaceous and tertiary vegetation history of Africa in J. A. Werger, editor Biogeography and Ecology of Southern Africa Junk, The Hague, Netherlands.

Baikie, B. 1857. On the skull of a species of Mecistops inhabiting the River Bínuë or Tsádda, in Central Africa. Proceedings of the Zoological Society of London 25:57– 59.

Barja, G., and A. Herrero. 2000. Oxidative damage to mitochondrial DNA is inversely related to maximum life span in the heart and brain of mammals. FASEB Journal 14:312–318.

Bartram, W. 1791. Travels through North and South Carolina, Georgia, East and West Florida. James andJohnson, Philadelphia.

Baum, D. A., and M. J. Donoghue. 1995. Choosing among alternative “phylogenetic” species concepts. Systematic Botany 20:560 – 573.

Beaumont, M. A., and M. Panchal. 2008. On the validity of nested clade phylogeographical analysis. Molecular Ecology 17:2563–5.

Beaumont, M. A. 1999. Detecting population expansion and decline using microsatellites. Genetics 153:2013–2029.

Beerli, P., and J. Felsenstein. 2001. Maximum likelihood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proceedings of the National Academy of Sciences of the United States of America 98:4563–8.

Beerli, P. 2006. Comparison of bayesian and maximum-likelihood inference of population genetic parameters. Bioinformatics 22:341–5.

Beerli, P. 2009. How to use migrate or why are markov chain monte carlo programs difficult to use? in G. Bertorelle, M W Bruford, H. C. Hauffe, A. Rizzoli, and C. Vernesi, editors Population Genetics for Animal Conservation Cambridge University Press, Cambridge, UK.

Behra, O. 1987a. Etude de repartition des populations de crocodiles du Congo, du Gabon et de la R.C.A. 3eme partie: R.C.A. Secretariat de la Faune et de la Flore. Museum National d’Histoire Naturelle de Paris. Paris, France.

Behra, O. 1987b. Etude de repartition des populations de crocodiles du Congo, du Gabon et de la R.C.A. 2eme partie: Gabon. Secretariat de la Faune et de la Flore. Museum National d’Histoire Naturelle de Paris. Paris, France.

175

Behra, O. 1987c. Etude de repartition des populations de crocodiles du Congo, du Gabon et de la R.C.A. 1ere partie: Congo. Secretariat de la Faune et de la Flore. Museum National d’Histoire Naturelle de Paris. Paris, France.

Beier, P., D. R. Majka, and W. D. Spencer. 2008. Forks in the road: choices in procedures for designing wildland linkages. Conservation Biology 22:836–51.

Bennett, E. T. 1835. Crocodilus leptorhynchus. Proceedings of the Zoological Society of London 128–132.

Beresford, P., and J. Cracraft. 1999. Speciation in African forest robins (Stiphronis): species limits, phylogenetic relationships, and molecular biology. American Museum Novitates 3720:1 – 22.

Bickford, D., D. J. Lohman, N. S. Sodhi, P. K. L. Ng, R. Meier, K. Winker, K. K. Ingram, and I. Das. 2007. Cryptic species as a window on diversity and conservation. Trends in Ecology & Evolution 22:148 – 155.

Blake, D. K., and J. P. Loveridge. 1975. The role of commercial crocodile farming in crocodile conservation. Biological Conservation 8:261–272.

Bonnefille, R., F. Chalié, J. Guiot, and A. Vincens. 1992. Quantitative estimates of full glacial temperatures in equatorial Africa from palynological data. Climate Dynamics 6:251–257.

Boulenger, G. 1889. Catalogue of the Chelonians, Rhynchocephalians, and Crocodiles in the British Museum (Natural History), New. The Trustees, London.

Bowie, R. C. K., J. Fjeldså, S. J. Hackett, J. M. Bates, and T. M. Crowe. 2006. Coalescent models reveal the relative roles of ancestral polymorphism, vicariance, and dispersal in shaping phylogeographical structure of an African montane forest robin. Molecular Phylogenetics and Evolution 38:171 – 188.

Brandt, L. A., and F. J. Mazzotti. 1990. The behavior of juvenile Alligator mississippiensis and Caiman crocodilus exposed to low temperature. Copeia 1990:867–871.

Brochu, C. A., J. R. Wagner, S. Jouve, C. D. Sumrall, and L. D. Densmore. 2009. A correction corrected: consensus over the meaning of Crocodylia and why it matters. Systematic Biology 58:537–543.

Brochu, C. A. 2003. Phylogenetic approaches toward Crocodylian history. Annual Review Earth and Planetary Science 31:399–427.

Brochu, C. A. 2004a. Calibration age and quartet divergence date estimation. Evolution 58:1375–1382.

176

Brochu, C. A. 2004b. Patterns of calibration age sensitivity with quartet dating methods. Journal of Paleontology 78:7–30.

Brochu, C. A. 2007. Morphology, relationships, and biogeographical significance of an extinct horned crocodile (Crocodylia, Crocodylidae) from the Quaternary of Madagascar. Zoological Journal of the Linnaean Society 150:835–863.

Bromham, L., A. Rambaut, and P. H. Harvey. 1996. Determinants of rate variation in mammalian DNA sequence evolution. Journal of Molecular Evolution 43:610 – 621.

Bromham, L. 2002. Molecular clocks in reptiles: life history influences rate of molecular evolution. Molecular Biology and Evolution 19:302–309.

Brooks, S., and A. Gelman. 1998. General methods for monitoring convergence of iterative simulations. Journal of Computational and Graphical Statistics 7:434 – 455.

Brower, A. V. Z. 1999. Delimitation of phylogenetic species with DNA sequences: a critique of Davis and Nixon’s Population Aggregation Analysis. Systematic Biology 48:199 – 213.

Burbrink, F. T., F. Fontanella, R. Alexander Pyron, T. J. Guiher, and C. Jimenez. 2008. Phylogeography across a continent: the evolutionary and demographic history of the North American racer (Serpentes: Colubridae: Coluber constrictor). Molecular Phylogenetics and Evolution 47:274–88.

Busack, S. D., and S. Pandya. 2001. Geographic variation in Caiman crocodilus and Caiman yacare (Crocodylia: Alligatoridae): Systematic and legal implications. Herpetologica 57:294–312.

Caccone, A., G. Gentile, C. E. Burns, E. Sezzi, W. Bergman, M. Ruelle, K. Saltonstall, and J. R. Powell. 2004. Extreme difference in rate of mitochondrial and nuclear DNA evolution in a large ectotherm, Galápagos tortoises. Molecular Phylogenetics and Evolution 31:794–8.

Cantagrel, J., C. Jamond, and M. Lassere. 1978. Le magmatisme alcalin de la ligne du Cameroun au Tertiaire inferieur: donnees geochronologiques K⁄ Ar. Comptes Rendus Sommaire de la Societe Geologique de France 6:300 – 303.

Carnaval, A. C., M. J. Hickerson, C. F. B. Haddad, M. T. Rodrigues, and C. C. Moritz. 2009. Stability predicts genetic diversity in the Brazilian Atlantic Forest hotspot. Science 232:785 – 789.

Carroll, S. P., and C. W. Fox. 2008. Conservation Biology:Evolution in Action. Oxford University Press, Oxford, New York ; :392.

177

Carroll, S. P., A. P. Hendry, D. N. Reznick, and C. W. Fox. 2007. Evolution on ecological time-scales. Functional Ecology 21:387–393.

Carstens, B. C., J. D. Degenhardt, A. L. Stevenson, and J. Sullivan. 2005. Accounting for coalescent stochasticity in testing phylogeographical hypotheses: modelling Pleistocene population structure in the Idaho giant salamander Dicamptodon aterrimus. Molecular Ecology 14:255–65.

Carstens, B. C., and C. L. Richards. 2007. Integrating coalescent and ecological niche modeling in comparative phylogeography. Evolution 61:1439–54.

Cedeño-Vásquez, J. R., D. Rodriguez, S. Calmé, J. P. Ross, L. D. Densmore, and J. B. Thorbjarnarson. 2008. Hybridization between Crocodylus acutus and Crocodylus moreletii in the Yucatan Peninsula: I. Evidence from Mitochondrial DNA and Morphology. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:661 – 673.

Chakraborty, R., and M. Nei. 1977. Bottleneck effects on average heterozygosity and genetic distance with the stepwise mutation model. Evolution 31:347 – 356.

Charlesworth, B. 2009. Fundamental concepts in genetics: effective population size and patterns of molecular evolution and variation. Nature Reviews - Genetics 10:195– 205.

Cherkiss, M. S., H. E. Fling, F. J. Mazzotti, and K. G. Rice. 2004. Counting and capturing crocodilans. Wildlife Ecology and Conservation Department, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL ; :1–14.

Clement, M., D. Posada, and K. A. Crandall. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9:1657–9.

Clifford, S. L., N. M. Anthony, M. Bawe-Johnson, K. A. Abernethy, C. E. G. Tutin, L. J. T. White, M. Bermejo, M. L. Goldsmith, K. McFarland, K. J. Jeffery, Michael W Bruford, and E. J. Wickings. 2004. Mitochondrial DNA phylogeography of western lowland gorillas (Gorilla gorilla gorilla). Molecular Ecology 13:1551–65, 1567.

Coetzee, J. A. 1993. African Flora since the Terminal Jurassic in P. Goldblatt, editor Biological Relationships between Africa and South America Yale University Press, New Haven.

Cooke, S. J., S. G. Hinch, M. Wikelski, R. D. Andrews, L. J. Kuchel, T. G. Wolcott, and P. J. Butler. 2004. Biotelemetry: a mechanistic approach to ecology. Trends in Ecology & Evolution 19:334–43.

178

Cornuet, J. M., and G. Luikart. 1996. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144:2001–14.

Coulson, R. A., and T. Hernandez. 1983. Alligator Metabolism: Studies on Chemical Reactions in vivo. Pergamon Press, Oxford, New York.

Cracraft, J. 1989. Speciation and its ontology in D. Otte and J. Endler, editors Speciation and Its Consequences Sinauer, Sunderland, MA.

Crandall, K. A., O. R. P. Bininda-Emonds, G. M. Mace, and R. K. Wayne. 2000. Considering evolutionary processes in conservation biology. Trends in Ecology & Evolution 15:290–295.

Crandall, K. A., and A. R. Templeton. 1993. Empirical tests of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics 134:959 – 969.

Cuvier, G. 1824. Recherches sur les ossemens fossiles de quadrupedes ou l’on retablit du plusieurs speces d’animaux que les revolutions du globe paroissent avoir detruites, 3rd ed. Dufour & d’Ocagne, Paris ; :58 – 59.

Davis, J. I., and K. C. Nixon. 1992. Populations, genetic variation, and the delimitation of phylogenetic species. Systematic Biology 41:421–435.

Davis, L. M., T. C. Glenn, D. C. Strickland, L. J. Guillette, R. M. Elsey, W. E. Rhodes, H. C. Dessauer, and R. H. Sawyer. 2002. Microsatellite DNA analyses support an east-west phylogeographic split of American alligator populations. Journal of Experimental Zoology 294:352–372.

De Queiroz, K. 2007. Species concepts and species delimitation. Systematic Biology 56:879–86.

DeSalle, R., M. G. Egan, and M. Siddal. 2005. The unholy trinity: taxonomy, species delimitation and DNA barcoding. Philosophical Transactions of the Royal Society B: Biological Sciences 360:1905 – 1916.

Dessauer, H. C., and R. M. Elsey. 1993. Molecular genetics of the American alligator. FASEB Journal 7:A814.

Donnelly, P., and S. Tavare. 1995. Coalescents and genealogical structure under neutrality. Annual Review of Genetics 29:410 – 421.

Drummond, A. J., S. Y. W. Ho, M. J. Phillips, and A. Rambaut. 2006. Relaxed phylogenetics and dating with confidence. PLoS Biology 4:e88.

179

Drummond, A. J., A. Rambaut, B. Shapiro, and O. G. Pybus. 2005. Bayesian coalescent inference of past population dynamics from molecular sequences. Molecular Biology and Evolution 22:1185–92.

Drummond, A. J., M. A. Suchard, D. Xie, and A. Rambaut. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29:1969 – 1973.

Earl, D. A., and B. M. Von Holdt. 2011. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4:359–361.

Eaton, M. J., A. Martin, J. B. Thorbjarnarson, and G. D. Amato. 2009. Species-level diversification of African dwarf crocodiles (Genus Osteolaemus): a geographic and phylogenetic perspective. Molecular Phylogenetics and Evolution 50:496–506.

Eaton, M. J. 2008. Systematics, population structure and demography of the African dwarf crocodile (Osteolaemus spp.): a perspective from multiple scales. Dissertation, University of Colorado.

Eaton, M. J. 2010. Dwarf Crocodile Osteolaemus tetraspis in S C Manolis and C. Stevenson, editors Crocodiles. Status Survey and Conservation Action Plan, Third Edit. Crocodile Specialist Group: Darwin.

Elenga, H., D. Schwartz, and A. Vincens. 1994. Pollen evidence of late Quaternary vegetation and inferred climate changes in Congo. Palaeogeography, Palaeoclimatology, Palaeoecology 109:345–356.

Elsey, R. M., and A. R. Woodward. 2010. American Alligator Alligator mississippiensis in S C Manolis and C. Stevenson, editors Crocodiles. Status Survey and Conservation Action Plan, Third Edit. Crocodile Specialist Group, Darwin.

Epperson, B. K. 2003. Geographical Genetics. Princeton University Press, Princeton.

Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14:2611–20.

Ewing, G. B. 2011. Haploviewer. .

Excoffier, L., M. Foll, and R. J. Petit. 2009. Genetic consequences of range expansions. Annual Review of Ecology, Evolution, and Systematics 40:481–501.

180

Excoffier, L., and L. Lischer. 2010. Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10:564 – 567.

Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479 – 491.

Falush, D., M. Stephens, and J. K. Pritchard. 2003. Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies. Genetics 164:1567 – 158.

Felsenstein, J. 1989. PHYLIP - Phylogeny Inference Package (Version 3.2). Cladistics 5:164 – 166.

Felsenstein, J. 2005. PHYLIP (Phylogeny Inference Package) version 3.6. .

Feng, G., X Wu, P Yan, and X. Li. 2010. Two complete mitochondrial genomes of Crocodylus and implications for crocodilians phylogeny. Amphibia-Reptilia 31:299 – 309.

Fitton, J. G., and H. M. Dunlop. 1985. The Cameroon Line, West Africa, and its bearing on the origin of oceanic and continental alkali basalt. Earth and Planetary Science Letters 72:23–38.

Fitton, J. G. 1987. The Cameroon Line, West Africa: a comparison between oceanic and continental alkaline volcanism in J. G. Fitton and B. C. G. Upton, editors Alkaline Igneous Rocks Geological Society of London Special Publication.

Fjeldsa, J., and J. C. Lovett. 1997. Geographical patterns for relict and young species in African forest biota: the significance of montane areas as evolutionary centres. Biodiversity and Conservation 6:325 – 346.

Fjeldså, J., and R. C. K. Bowie. 2008. New perspectives on the origin and diversification of Africa’s forest avifauna. African Journal of Ecology 46:235–247.

Formia, A., B. J. Godley, J.-F. Dontaine, and Michael W Bruford. 2006. Mitochondrial DNA diversity and phylogeography of endangered green turtle (Chelonia mydas) populations in Africa. Conservation Genetics 7:353–369.

Forrester, D. J., and R. T. Sawyer. 1974. Placobdella multilineata (Hirudinea) from the American alligator in Florida. Journal of Parasitology 60:673.

181

Frankham, R., K. Lees, M. E. Montgomery, P. R. England, E. H. Lowe, and D. A. Briscoe. 1999. Do population size bottlenecks reduce evolutionary potential? Animal Conservation 2:255–260.

Frankham, R. 1995. Effective population size/adult population size ratios in wildlife: a review. Genetical Research 66:95–107.

Frankham, R. 2005. Genetics and extinction. Biological Conservation 126:131–140.

François, O., and E. Durand. 2010. Spatially explicit Bayesian clustering models in population genetics. Molecular Ecology Resources 10:773–84.

Fraser, D. J., and L. Bernatchez. 2001. Adaptive evolutionary conservation: towards a unified concept for defining conservation units. Molecular Ecology 10:2741 – 2752.

Fu, Y. X. 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:915–925.

Fuchs, K. H., R. Mertens, and H. Wermuth. 1974a. Zum status von Crocodylus cataphractus und Osteolaemus tetraspis. Stuttgarter Beitraege zuer Naturkunde, Serie A. Biologie 266:1–8.

Fuchs, K. H., R. Mertens, and H. Wermuth. 1974b. Die unterarten des Nilkrokodils, Crocodylus niloticus. Salamandra 10:107–114.

Garrick, R. C., R. J. Dyer, L. B. Beheregaray, and P. Sunnucks. 2008. Babies and bathwater: a comment on the premature obituary for nested clade phylogeographical analysis. Molecular Ecology 17:1401–3; discussion 1404.

Garza, J. C., and E. G. Williamson. 2001. Detection of reduction in population size using data from microsatellite loci. Molecular Ecology 10:305–318.

Gasse, F. 2000. Hydrological changes in the African tropics since the Last Glacial Maximum. Quaternary Science Reviews 19:189–211.

Gatesy, J., G. D. Amato, M. Norell, R. Desalle, and C. Hayashi. 2003. Combined support for wholesale taxic atavism in gavialine crocodylians. Systematic Biology 52:403–422.

Gelman, A., and D. B. Rubin. 1992. Inference from iterative simulation using multiple sequences. Statistical Science 7:457 – 511.

Gernhard, T. 2008. The conditioned reconstructed process. Journal of Theoretical Biology 253:769–78.

182

Gillespie, J. H. 2004. Population Genetics: A Concise Guide, 2nd ed. Johns Hopkins University Press, Baltimore, Maryland.

Gillooly, J. F., A. P. Allen, G. B. West, and J. H. Brown. 2005. The rate of DNA evolution: effects of body size and temperature on the molecular clock. Proceedings of the National Academy of Sciences of the United States of America 102:140–5.

Giresse, P., J. Maley, and A. Ngomanda. 2008. Late-Holocene equatorial environments inferred from deposition processes, carbon isotopes of organic matter, and pollen in three shallow lakes of Gabon, west-central Africa. Journal of Paleolimnology 41:369–392.

Girod, C., R. Vitalis, R. Leblois, and H. Fréville. 2011. Inferring population decline and expansion from microsatellite data: a simulation-based evaluation of the Msvar method. Genetics 188:165–79.

Glenn, T. C., J. L. Staton, A. T. Vu, L. M. Davis, J. R. A. Bremer, W. E. Rhodes, I. L. Brisbin, and R. H. Sawyer. 2002. Low mitochondrial DNA variation among American alligators and a novel non-coding region in crocodilians. Journal of Experimental Zoology 294:312–324.

Godshalk, R. E. 2006. Phylogeography and conservation genetics of the Yacare caiman (Caiman yacare) of South America. Wildlife Ecology and Conservation 1–228.

Goossens, B., L. Chikhi, M. Ancrenaz, I. Lackman-Ancrenaz, P. Andau, and Michael W Bruford. 2006. Genetic signature of anthropogenic population collapse in orangutans. PLoS Biology 4:e25.

Gray, J. 1844. Catalogue of the tortoises, crocodiles, and amphibaenians in the collection of the British Museum. London.

Groth, J. G., and G. F. Barrowclough. 1999. Basal divergences in birds and the phylogenetic utility of the nuclear RAG-1 gene. Molecular Phylogenetics and Evolution 12:115 – 123.

Grubb, P. 1982. Refuges and Dispersal in the Speciation of African Forest Mammals in G. T. Prance, editor Biological Diversification in the Tropics: Proceedings of the 5th International Symposium of the Association for Tropical Biology Columbia University Press, New York.

Guindon, S., J. F. Dufayard, V. Lefort, M. Anisimova, W. Hordijk, and O. Gascuel. 2010. New algorithms and methods to estimate maximum likelihood phylogenies: Assessing the performance of PhyML 3.0. Systematic Biology 59:307 – 321.

183

Hanotte, O., D. G. Bradley, J. W. Ochieng, Y. Verjee, E. W. Hill, and J. E. O. Rege. 2002. African pastoralism: genetic imprints of origins and migrations. Science 296:336–9.

Harpending, H. C., M. A. Batzer, M. Gurven, L. B. Jorde, A. R. Rogers, and S. T. Sherry. 1998. Genetic traces of ancient demography. Proceedings of the National Academy of Sciences 95:1961–1967.

Harpending, H. C. 1994. Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Human Biology 66:591–600.

Hebert, P. D. N., A. Cywinska, S. L. Ball, and J. R. de Waard. 2003. Biological identifications through DNA barcodes. Proceedings of the Royal Society B: Biological Sciences 270:313 – 321.

Hebert, P. D. N., E. H. Penton, J. M. Burns, D. H. Janzen, and W. Hallwachs. 2004. Ten species in one: DNA barcoding reveals cryptic species in the neotropical skipper butterfly Astraptes fulgerator. Proceedings of the National Academy of Sciences of the United States of America 101:14812–7.

Hebert, P. D. N., S. Ratnasingham, and J. R. DeWaard. 2003. Barcoding animal life: cytochrome c oxidase subunit 1 divergences among closely related species. Proceedings of the Royal Society B: Biological Sciences 270 Suppl:S96–9.

Heidelberger, P., and P. Welch. 1983. Simulation run length control in the presence of an initial transient. Operations Research 31:1109 – 1144.

Hekkala, E., G. D. Amato, R. DeSalle, and M. J. Blum. 2010. Molecular assessment of population differentiation and individual assignment potential of Nile crocodile (Crocodylus niloticus) populations. Conservation Genetics 11:1435–1443.

Hekkala, E., M. H. Shirley, G. D. Amato, J. D. Austin, S. Charter, J. B. Thorbjarnarson, K. A. Vliet, M. L. Houck, R. Desalle, and M. J. Blum. 2011. An ancient icon reveals new mysteries: mummy DNA resurrects a cryptic species within the Nile crocodile. Molecular Ecology 20:4195–4215.

Heled, J., and A. J. Drummond. 2008. Bayesian inference of population size history from multiple loci. BMC Evolutionary Biology 8:289.

Heller, R., E. D. Lorenzen, J. B. A. Okello, C. Masembe, and H. R. Siegismund. 2008. Mid-Holocene decline in African buffalos inferred from Bayesian coalescent-based analyses of microsatellites and mitochondrial DNA. Molecular Ecology 17:4845–58.

Hess, P. N., C. A. de Moraes, and M. Russo. 2007. An empirical test of the midpoint rooting method. Journal of Heredity 669–674.

184

Hey, J., Y.-J. Won, A. Sivasundar, R. Nielsen, and J. A. Markert. 2004. Using nuclear haplotypes with microsatellites to study gene flow between recently separated Cichlid species. Molecular Ecology 13:909–919.

Ho, S. Y. W., and G. Larson. 2006. Molecular clocks: when times are a-changin’. Trends in Genetics 22:79–83.

Ho, S. Y. W., and M. J. Phillips. 2009. Accounting for calibration uncertainty in phylogenetic estimation of evolutionary divergence times. Systematic Biology 58:367–80.

Ho, S. Y. W., and B. Shapiro. 2011. Skyline-plot methods for estimating demographic history from nucleotide sequences. Molecular Ecology Resources 11:423–34.

Hoffman, J. I., and W. Amos. 2005. Microsatellite genotyping errors: detection approaches, common sources and consequences for paternal exclusion. Molecular Ecology 14:599–612.

Howes, B. J., B. Lindsay, and S. C. Lougheed. 2006. Range-wide phylogeography of a temperate lizard, the five-lined skink (Eumeces fasciatus). Molecular Phylogenetics and Evolution 40:183 – 194.

Hrbek, T., W. R. Vasconcelos, G. Rebelo, and I. P. Farias. 2008. Phylogenetic relationships of South American alligatorids and the caiman of Madeira River. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:588–599.

Hudson, R. R. 1991. Gene genealogies and the coalescent process in K. J. Futuyama and J. Antonovics, editors Oxford Surveys in Evolutionary Biology Oxford University Press, Oxford.

Hutton, J. M. 1989. Movements, home range, dispersal and the separation of size classes in Nile crocodiles. American Zoologist 29:1033–1049.

Inger, R. F. 1948. The systematic status of the crocodile Osteoblepharon osborni. Copeia 1948:15–19.

Jahns, S. 1996. Vegetation history and climate changes in West Equatorial Africa during the Late Pleistocene and Holocene, based on a marine pollen diagram from the Congo fan. Vegetation History and Archaeobotany 5:207–213.

Jakobsson, M., and N. A. Rosenberg. 2007. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23:1801–6.

185

Janke, A., and U. Arnason. 1997. The complete mitochondrial genome of Alligator mississippiensis and the separation between recent archosauria (birds and crocodiles). Molecular Biology and Evolution 14:7.

Janke, A., D. Erpenbeck, M. Nilsson, and U. Arnason. 2001. The mitochondrial genomes of the iguana (Iguana iguana) and the caiman (Caiman crocodylus): implications for amniote phylogeny. Proceedings of the Royal Society B: Biological Sciences 268:623–631.

Janke, A., A. Gullberg, S. Hughes, R. K. Aggarwal, and U. Arnason. 2005. Mitogenomic analyses place the gharial (Gavialis gangeticus) on the crocodile tree and provide Pre-K/T divergence times for most crocodilians. Journal of Molecular Evolution 61:620–626.

Joanen, T., and L. McNease. 1989. Ecology and physiology of nesting and early development of the American alligator. American Zoologist 29:987–998.

Jombart, T., S. Devillard, and F. Balloux. 2010. Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics 11:94.

Jombart, T., S. Devillard, A.-B. Dufour, and D. Pontier. 2008. Revealing cryptic spatial patterns in genetic variability by a new multivariate method. Heredity 101:92–103.

Jombart, T. 2008. adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24:1403–5.

Jombart, T. 2012a. A tutorial for the spatial Analysis of Principal Components using adegenet 1.3-4. .

Jombart, T. 2012b. A tutorial for Discriminant Analysis of Principal Components (DAPC) using adegenet 1.3-4. .

Kass, R. E., and A. E. Raftery. 1995. Bayes Factors. Journal of the American Statistical Association 90:773 – 795.

Kim, S.-Y., J. Scourse, F. Marret, and D.-I. Lim. 2010. A 26,000-year integrated record of marine and terrestrial environmental change off Gabon, west equatorial Africa. Palaeogeography, Palaeoclimatology, Palaeoecology 297:428–438.

Kingman, J. 1982. The coalescent. Stochastic Processes and Their Applications 13:235 – 248.

Kitchen, A., M. M. Miyamoto, and C. J. Mulligan. 2008. A three-stage colonization model for the peopling of the Americas. PLoS ONE 3:e1596.

186

Kluge, A. G., and J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18:1 – 32.

Kluge, A. G. 1998. Total evidence or taxonomic congruence: cladistics or consensus classification. Cladistics 14:151 – 158.

Knowles, L. L., and B. C. Carstens. 2007. Delimiting species without monophyletic gene trees. Systematic Biology 56:887–95.

Knowles, L. L., and W. P. Maddison. 2002. Statistical phylogeography. Molecular Ecology 11:2623–2635.

Kujoth, G. C., P. Bradshaw, S. Haroon, and T. Prolla. 2007. The role of mitochondrial DNA mutations in mammalian aging. PLoS Genetics 3:161 – 173.

Lachenbruch, P. A., and M. Goldstein. 1979. Discriminant Analysis. Biometrics 35:69 – 85.

Lande, R., and S. Shannon. 1996. The role of genetic variation in adaptation and population persistence in a changing environment. Evolution 50:434 – 437.

Lang, J. W. 1977. Studies of the thermal behaviour and body temperature of crocodilians. Dissertation, Dept. of Biology, University of Minnesota.

Lang, J. W. 1985. Incubation temperature affects thermal selection of hatchling crocodiles. Amercian Zoologist 25:18A.

Lang, J. W. 1987. Crocodilian thermal selection. in G J W Webb, S C Manolis, and P. J. Whitehead, editors Wildlife Management: Crocodiles and Alligators Surrey Beatty and Sons Pty Ltd., Chipping Norton, New South Wales, Australia. 522 pp.

Leaché, A. D. 2009. Species tree discordance traces to phylogeographic clade boundaries in North American fence lizards (Sceloporus). Systematic Biology 58:547–59.

Leal, M. A. 2001. Microrefugia, small scale Ice Age forest remnants. Systematics and Geography of Plants 71:1073 – 1077.

Lee, C., A. Abdool, and C.-H. Huang. 2009. PCA-based population structure inference with generic clustering algorithms. BMC Bioinformatics 10 Suppl 1:S73.

Lee, D.-C., A. N. Halliday, J. G. Fitton, and G. Poli. 1994. Isotopic variations with distance and time in the volcanic islands of the Cameroon Line: evidence for a mantel plume origin. Earth and Planetary Science Letters 123:119 – 138.

187

Lézine, A.-M., and C. Vergnaud-Grazzini. 1993. Evidence of forest extension in West Africa since 22,000 BP: A pollen record from the eastern tropical Atlantic. Quaternary Science Reviews 12:203–210.

Li, Y., Xiaobing Wu, X. Ji, Peng Yan, and G. D. Amato. 2007. The complete mitochondrial genome of salt-water crocodile (Crocodylus porosus) and phylogeny of crocodilians. Journal of Genetics and Genomics 34:119–128.

Linder, H. P., C. R. Hardy, and F. Rutschmann. 2005. Taxon sampling effects in molecular clock dating: An example from the African Restionaceae. Molecular Phylogenetics and Evolution 35:569 – 582.

Linder, H. P., H. M. de Klerk, J. Born, N. D. Burgess, J. Fjeldså, and C. Rahbek. 2012. The partitioning of Africa: statistically defined biogeographical regions in sub- Saharan Africa. Journal of Biogeography 39:1189 – 1205.

Livingstone, D. A. 1993. Evolution of African Climate in P Goldblatt, editor Biological Relationships Between Africa and South America Yale University Press, New Haven.

López-Castro, M. C., and A. Rocha-Olivares. 2005. The panmixia paradigm of eastern Pacific olive ridley turtles revised: consequences for their conservation and evolutionary biology. Molecular Ecology 14:3325–34.

Luikart, G., N. Ryman, D. A. Tallmon, M. K. Schwartz, and F. W. Allendorf. 2010. Estimation of census and effective population sizes: the increasing usefulness of DNA-based approaches. Conservation Genetics 11:355–373.

Lynch, M., B. Koskella, and S. Schaack. 2006. Mutation pressure and the evolution of organelle genomic architecture. Science 311:1727–30.

Maddison, W. P., and D. R. Maddison. 2011. Mesquite: a modular system for evolutionary analysis. .

Maddison, W. P. 1997. Gene trees in species trees. Systematic Biology 46:523 – 536.

Magnusson, W. E., A. P. Lima, and R. A. Sampaio. 1985. Sources of heat for nests of Paleosuchus trigonatus and a review of crocodilian nest temperatures. Journal of Herpetology 19:199–207.

Magnusson, W. E. 1979. Maintenance of temperature of crocodile nests (Reptilia, Crocodilidae). Journal of Herpetology 13:439–443.

188

Maley, J. 1987. Fragmentation de la foret dense humide africaine et extension des biotopes montagnards au quaternaire recent: nouvelles donnees polliniques et chronologiques. Implications paleoclimatiques et biogeographiques. Palaeoecology of Africa and the Surrounding Islands 18:307 – 334.

Maley, J. 1996. The African rain forest - main characteristics of changes in vegetation and climate from the Upper Cretaceous to the Quaternary. Proceedings of the Royal Society of Edinburgh 104B:31 – 73.

Maley, J. 2002. A catastrophic destruction of African forests about 2,500 years ago still exerts a major influence on present vegetation formations. IDS Bulletin 33:13–30.

Manel, S., M. K. Schwartz, G. Luikart, and P. Taberlet. 2003. Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology & Evolution 18:189–197.

Mardulyn, P. 2001. Phylogeography of the Vosges mountains populations of Gonioctena pallida (Coleoptera: Chrysomelidae): a nested clade analysis of mitochondrial DNA haplotypes. Molecular Ecology 10:1751–1763.

Marks, B. D. 2010. Are lowland rainforests really evolutionary museums? Phylogeography of the green hylia (Hylia prasina) in the Afrotropics. Molecular Phylogenetics and Evolution 55:178–84.

Markwick, P. J. 1998. Fossil crocodilians as indicators of Late Cretaceous and Cenozoic climates: implications for using palaeontological data in reconstructing palaeoclimate. Palaeogeography, Palaeoclimatology, Palaeoecology 137:205–271.

Marzoli, A., P. R. Renne, E. M. Piccirillo, C. Francesca, G. Bellieni, A. J. Melfi, J. B. Nyobe, and J. N’ni. 1999. Silicic magmas from the continental Cameroon Volcanic Line (Oku, Bambouto and Ngaoundere): 40Ar-39Ar dates, petrology, Sr-Nd-O isotopes and their petrogenetic significance. Contributions to Mineralogy and Petrology 135:133–150.

Mayewski, P. A., E. E. Rohling, J. C. Stager, W. Karlén, K. A. Maasch, L. David Meeker, E. A. Meyerson, F. Gasse, S. van Kreveld, K. Holmgren, J. Lee-Thorp, G. Rosqvist, F. Rack, M. Staubwasser, R. R. Schneider, and E. J. Steig. 2004. Holocene climate variability. Quaternary Research 62:243–255.

McAliley, L. R., R. E. Willis, D. A. Ray, P Scott White, C. A. Brochu, and L. D. Densmore. 2006. Are crocodiles really monophyletic? Evidence for subdivisions from sequence and morphological data. Molecular Phylogenetics and Evolution 39:16 – 32.

189

Meganathan, P. R., B. Dubey, M. A. Batzer, D. A. Ray, and I. Haque. 2011. Complete mitochondrial genome sequences of three Crocodylus species and their comparison within the Order Crocodylia. Gene 478:35 – 41.

Menozzi, P., A. Piazza, and L. Cavalli-Sforza. 1978. Synthetic maps of human gene frequencies in Europeans. Science 201:786 – 792.

Menzies, R. A., J. Kushlan, and H. C. Dessauer. 1979. Low degree of genetic variability in the American alligator (Alligator mississippiensis). Isozyme Bulletin 12:55.

Merchant, M. E., K. Mills, N. Leger, E. Jerkins, K. A. Vliet, and N. McDaniel. 2006. Comparison of innate immune activity of all 23 known living crocodylian species. Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology 143:133–137.

Meredith, R. W., E. Hekkala, G. D. Amato, and J. Gatesy. 2011. A phylogenetic hypothesis for Crocodylus (Crocodylia) based on mitochondrial DNA: Evidence for a trans-Atlantic voyage from Africa to the New World. Molecular Phylogenetics and Evolution 60:183–191.

Meyers, J. B., B. R. Rosenthal, G. A. Harrison, and Z. D. Ding. 1998. Deep imaging seismic and gravity results from offshore Cameroon Volcanic Line, and speculation of African hotlines. Tectonophysics 284:31 – 63.

Miles, L. G., S. R. Isberg, C. Moran, C. Hagen, and T. C. Glenn. 2008. 253 Novel polymorphic microsatellites for the saltwater crocodile (Crocodylus porosus). Conservation Genetics 10:963–980.

Miles, L. G., S. L. Lance, S. R. Isberg, C. Moran, and T. C. Glenn. 2009. Cross-species amplification of microsatellites in crocodilians: assessment and applications for the future. Conservation Genetics 10:935–954.

Milián-García, Y., M. Venegas-Anaya, R. Frias-Soler, A. J. Crawford, R. Ramos- Targarona, R. Rodríguez-Soberón, M. Alonso-Tabet, J. B. Thorbjarnarson, O. I. Sanjur, G. Espinosa-López, and E. Bermingham. 2011. Evolutionary history of Cuban crocodiles Crocodylus rhombifer and Crocodylus acutus inferred from multilocus markers. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 315:358–75.

Miller, H. C., J. A. Moore, F. W. Allendorf, and C. H. Daugherty. 2009. The evolutionary rate of tuatara revisited. Trends in Genetics 25:13–5; author reply 16–8.

Minin, V. N., E. W. Bloomquist, and M. A. Suchard. 2008. Smooth skyride through a rough skyline: Bayesian coalescent-based inference of population dynamics. Molecular Bology and Evolution 25:1459–71.

190

Monsen, K. J., and M. S. Blouin. 2003. Genetic structure in a montane ranid frog: restricted gene flow and nuclear-mitochondrial discordance. Molecular Ecology 12:3275–3286.

Moritz, C. C. 1994. Defining “evolutionarily significant units” for conservation. Trends in Ecology & Evolution 9:373 – 375.

Moritz, C. C. 2002. Strategies to protect biological diversity and the evolutionary processes that sustain it. Systematic Biology 5:238 – 254.

Muller, J., and R. R. Reisz. 2005. Four well-constrained calibration points from the vertebrate fossil record for molecular clock estimates. BioEssays 27:1069 – 1075.

Mulligan, C. J., A. Kitchen, and M. M. Miyamoto. 2006. Comment on “Population size does not influence mitochondrial genetic diversity in animals”. Science 314:1390.

Myers, N., and A. H. Knoll. 2001. The biotic crisis and the future of evolution. Proceedings of the National Academy of Sciences of the USA 98:5389 – 5392.

Nabholz, B., S. Glémin, and N. Galtier. 2008. Strong variations of mitochondrial mutation rate across mammals--the longevity hypothesis. Molecular Biology and Evolution 25:120–30.

Nabholz, B., S. Glémin, and N. Galtier. 2009. The erratic mitochondrial clock: variations of mutation rate, not population size, affect mtDNA diversity across birds and mammals. BMC Evolutionary Biology 9:54.

Nei, M., T. Maruyama, and R. Chakraborty. 1975. The bottleneck effect and genetic variability in populations. Evolution 29:1 – 10.

Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York, NY.

Nielsen, R., and J. Wakeley. 2001. Distinguishing migration from isolation: a Markov Chain Monte Carlo approach. Genetics 158:885 – 896.

Noss, R. F. 1996. Ecosystems as conservation targets. Trends in Ecology & Evolution 11:351.

Nunney, L., and D. R. Elam. 1994. Estimating the effective population size of conserved populations. Conservation Biology 8:175 – 184.

Oaks, J. R. 2011. A time-calibrated species tree of Crocodylia reveals a recent radiation of the true crocodiles. Evolution 65:3285–3297.

191

Ohta, T. 1993. An examination of the generation-time effect on molecular evolution. Proceedings of the National Academy of Sciences 90:10676–10680.

Okello, J. B. A., G. Wittemyer, H. B. Rasmussen, P. Arctander, S. Nyakaana, I. Douglas-Hamilton, and H. R. Siegismund. 2008. Effective population size dynamics reveal impacts of historic climatic events and recent anthropogenic pressure in African elephants. Molecular Ecology 17:3788–3799.

Panchal, M. 2007. The automation of Nested Clade Phylogeographic Analysis. Bioinformatics 23:509 – 510.

Pannell, J. R. 2003. Coalescence in a metapopulation with recurrent local extinction and recolonization. Evolution 57:949–961.

Patterson, N., A. L. Price, and D. Reich. 2006. Population structure and eigenanalysis. PLoS Genetics 2:e190.

Peacock, M. M., and C. Ray. 2001. Dispersal in pikas (Ochotona princeps): combining genetic and demographic approaches to reveal spatial and temporal patterns in J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, editors Dispersal Oxford University Press, Oxford.

Peakall, R., and P. E. Smouse. 2012. GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research--an update. Bioinformatics 28:2537–9.

Perez, O., and A. H. Escobedo-Galván. 2009. Potential effects of El Nino-South Oscillation (ENSO) on growth of the American crocodile, Crocodylus acutus (Crocodylia: Crocodilydae) in captivity. Journal of Thermal Biology 34:14–16.

Pesole, G., C. Gissi, A. De Chirico, and C. Saccone. 1999. Nucleotide substitution rate of mammalian mitochondrial genomes. Journal of Molecular Evolution 48:427–434.

Petit, R. J. 2008. On the falsifiability of the nested clade phylogeographic analysis method. Molecular Ecology 17:1404–1404.

Pfenninger, M., and K. Schwenk. 2007. Cryptic animal species are homogeneously distributed among taxa and biogeographical regions. BMC Evolutionary Biology 7:121.

Phillips, B. L., S. J. E. Baird, and C. C. Moritz. 2004. When vicars meet: A narrow contact zone between morphologically cryptic phylogeographic lineages of the rainforest skink, Carla rubrigularis. Evolution 58:1536–1548.

Piganeau, G., and A. Eyre-Walker. 2009. Evidence for variation in the effective population size of animal mitochondrial DNA. PloS One 4:e4396.

192

Plana, V. 2004. Mechanisms and tempo of evolution in the African Guinea-Congolian rainforest. Philosophical Transactions of the Royal Society of London Series B: Biology 359:1585 – 1594.

Posada, D., K. A. Crandall, and A. R. Templeton. 2000. GeoDis: a program for the cladistic nested analysis of the geographical distribution of genetic haplotypes. Molecular Ecology 9:487–8.

Posada, D., K. A. Crandall, and A. R. Templeton. 2006. Nested clade analysis statistics. Molecular Ecology Notes 6:590–593.

Posada, D. 2008. jModelTest: phylogenetic model averaging. Molecular Biology and Evolution 25:1253–6.

Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure using multilocus genotype data. Genetics 155:945–59.

Pybus, O. G., A. Rambaut, and P. H. Harvey. 2000. An integrated framework for the inference of viral population history from reconstructed genealogies. Genetics 155:1429–1437.

Pyron, A. R., and F. T. Burbrink. 2009. Lineage diversification in a widespread species: roles for niche divergence and conservatism in the common kingsnake, Lampropeltis getula. Molecular Ecology 18:3443–57.

Quattro, J. M., D. S. Stoner, W. B. Driggers, C. A. Anderson, K. A. Priede, E. C. Hoppmann, N. H. Campbell, K. M. Duncan, and J. M. Grady. 2005. Genetic evidence of cryptic speciation within hammerhead sharks (Genus Sphyrna). Marine Biology 148:1143–1155.

R Core Team. 2012. R: A Language and Environment for Statistical Computing. .

Rambaut, A., and A. J. Drummond. 2009. Tracer v1.5: MCMC Trace Analysis Tool. .

Rambaut, A., O. G. Pybus, M. I. Nelson, C. Viboud, J. K. Tautenberger, and E. C. Holmes. 2008. The genomic and epidemiological dynamics of human influenza A virus. Nature 453:615 – 619.

Ray, D. A., J. A. Dever, S. G. Platt, T. R. Rainwater, A. G. Finger, S. T. McMurry, M. A. Batzer, B. Barr, P. J. Stafford, J. McKnight, and L. D. Densmore. 2004. Low levels of nucleotide diversity in Crocodylus moreletii and evidence of hybridization with C. acutus. Conservation Genetics 5:449–462.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223 – 225.

193

Rodriguez, D., J. R. Cedeño-Vásquez, M. R. J. Forstner, and L. D. Densmore. 2008. Hybridization between Crocodylus acutus and Crocodylus moreletii in the Yucatan Peninsula: II. Evidence from microsatellites. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:674–686.

Ronquist, F., M. Teslenko, P. van der Mark, D. L. Ayres, A. Darling, S. Höhna, B. Larget, L. Liu, M. A. Suchard, and J. P. Huelsenbeck. 2012. MrBayes 3.2: Efficient Bayesian phylogenetic inference and model choice across a large model space. Systematic Biology 61:539 – 542.

Roos, J., R. K. Aggarwal, and A. Janke. 2007. Extended mitogenomic phylogenetic analyses yield new insight into crocodylian evolution and their survival of the Cretaceous-Tertiary boundary. Molecular Phylogenetics and Evolution 45:663–673.

Rosenberg, M. S., and S. Kumar. 2001. Incomplete taxon sampling is not a problem for phylogenetic inference. Proceedings of the National Academy of Sciences 98:10751 – 10756.

Rosenberg, N. A., and M. Nordborg. 2002. Genealogical trees, coalescent theory and the analysis of genetic polymorphisms. Nature Reviews. Genetics 3:380–90.

Rosenberg, N. A. 2003. distruct: a program for the graphical display of population structure. Molecular Ecology Notes 4:137–138.

Rousset, F. 2001. Genetic approaches to the estimation of dispersal rates in J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, editors Dispersal Oxford University Press, Oxford.

Rousset, F. 2008. genepop’007: a complete re-implementation of the genepop software for Windows and Linux. Molecular Ecology Resources 8:103–6.

Rozen, S., and H. J. Skaletsky. 2000. Primer3 on the WWW for general users and for biologist programmers in S. Krawetz and S. Misener, editors Bioinformatics Methods and Protocols: Methods in Molecular Biology Humana Press, Totowa, NJ.

Rueness, E. K., P. E. Jorde, L. Hellborg, N. C. Stenseth, H. Ellegren, and K. S. Jakobsen. 2003. Cryptic population structure in a large, mobile mammalian predator: the Scandinavian lynx. Molecular Ecology 12:2623–2633.

Russello, M. A., P. Brazaitis, J. Gratten, G. J. Watkins-Colwell, and A. Caccone. 2007. Molecular assessment of the genetic integrity, distinctiveness and phylogeographic context of the saltwater crocodile (Crocodylus porosus) on Palau. Conservation Genetics 8:777–787.

194

Salisbury, S. W., R. E. Molnar, E. Frey, and P. M. A. Willis. 2006. The origin of modern crocodyliforms: new evidence from the Cretaceous of Australia. Proceedings of the Royal Society B: Biological Sciences 273:2439–2448.

Salzburger, W., G. B. Ewing, and A. Von Haeseler. 2011. The performance of phylogenetic algorithms in estimating haplotype genealogies with migration. Molecular Ecology 20:1952–63.

Sambrock, J., and D. W. Russell. 2001. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York.

Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: A penalized likelihood approach. Molecular Biology and Evolution 19:101 – 109.

Sanderson, M. J. 2003. r8s: Inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics 19:301 – 302.

Sanderson, M. J. 2006. r8s. .

Sangster, G. 2009. Increasing numbers of bird species result from taxonomic progress, not taxonomic inflation. Proceedings of the Royal Society B: Biological Sciences 276:3185 – 3191.

Santos, M. C. F., M. L. Ruffino, and I. P. Farias. 2007. High levels of genetic variability and panmixia of the tambaqui Colossoma macropomum (Cuvier, 1816) in the main channel of the Amazon River. Journal of Fish Biology 71:33–44.

Schneider, S., and L. Excoffier. 1999. Estimation of past demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: Application to human mitochondrial DNA. Genetics 152:1079–1089.

Schuelke, M. 2000. An economic method for the fluorescent labeling of PCR fragments. Nature Biotechnology 18:233 – 234.

Scribner, K. T., J. A. Blanchong, D. J. Bruggeman, B. K. Epperson, C.-Y. Lee, Y.-W. Pan, R. I. Shorey, H. H. Prince, S. R. Winterstein, and D. R. Luukkonen. 2005. Geographical Genetics: Conceptual foundations and empirical applications of spatial genetic data in wildlife management. Journal of Wildlife Management 69:1434–1453.

195

Shapiro, B., A. J. Drummond, A. Rambaut, M. C. Wilson, P. E. Matheus, A. V Sher, O. G. Pybus, M. T. P. Gilbert, I. Barnes, J. Binladen, E. Willerslev, A. J. Hansen, G. F. Baryshnikov, J. A. Burns, S. Davydov, J. C. Driver, D. G. Froese, C. R. Harington, G. Keddie, P. Kosintsev, M. L. Kunz, L. D. Martin, R. O. Stephenson, J. Storer, R. Tedford, S. Zimov, and A. Cooper. 2004. Rise and fall of the Beringian steppe bison. Science 306:1561–5.

Shirley, M. H., W. Oduro, and H. Yaokokore-Beibro. 2009. Conservation status of crocodiles in Ghana and Côte-d’lvoire, West Africa. Oryx 43:136–145.

Shirley, M. H. 2010. Slender-snouted Crocodile Crocodylus cataphractus in S.C. Manolis and C. Stevenson, editors Crocodiles. Status Survey and Conservation Action Plan, Third Edit. Crocodile Specialist Group, Darwin.

Slatkin, M. 1993. Isolation by distance in equilibrium and nonequilibrium populations. Evolution 47:264 – 279.

Smith, B. J. 2007. boa: An R package for MCMC output convergence assessment and posterior inference. Journal of Statistical Software 21:1 –37.

Song, B.-H., A. J. Windsor, K. J. Schmid, S. Ramos-Onsins, M. E. Schranz, A. J. Heidel, and T. Mitchell-Olds. 2009. Multilocus patterns of nucleotide diversity, population structure and linkage disequilibrium in Boechera stricta, a wild relative of Arabidopsis. Genetics 181:1021–33.

St. John, J. A., E. L. Braun, S. R. Isberg, L. G. Miles, A. Y. Chong, J. Gongora, P. Dalzell, C. Moran, B. Bed’hom, A. Abzhanov, S. C. Burgess, A. M. Cooksey, T. A. Castoe, N. G. Crawford, L. D. Densmore, J. C. Drew, S. V Edwards, B. C. Faircloth, M. K. Fujita, M. J. Greenwold, F. G. Hoffmann, J. M. Howard, T. Iguchi, D. E. Janes, S. Y. Khan, S. Kohno, A. J. de Koning, S. L. Lance, F. M. McCarthy, J. E. McCormack, M. E. Merchant, D. G. Peterson, D. D. Pollock, N. Pourmand, B. J. Raney, K. A. Roessler, J. R. Sanford, R. H. Sawyer, C. J. Schmidt, E. W. Triplett, T. D. Tuberville, M. Venegas-Anaya, J. T. Howard, E. D. Jarvis, L. J. Guillette, T. C. Glenn, R. E. Green, and D. A. Ray. 2012. Sequencing three crocodilian genomes to illuminate the evolution of archosaurs and amniotes. Genome Biology 13:415.

Stamatakis, A., P. Hoover, and J. Rougemont. 2008. A rapid bootstrap algorithm for the RAxML Web servers. Systematic Biology 57:758–71.

Storrs, G. W. 2003. Late Miocene–Early Pliocene crocodilian fauna of Lothagam, southwest Turkana Basin, Kenya. in M. G. Leakey and J. M. Harris, editors Lothagam: The Dawn of Humanity in Eastern Africa Columbia Univ. Press, New York.

196

Storz, J. F., M. A. Beaumont, and S. C. Alberts. 2002. Genetic evidence for long-term population decline in a savannah-dwelling primate: Inferences from a hierarchical Bayesian model. Molecular Biology and Evolution 19:1981–1990.

Storz, J. F., and M. A. Beaumont. 2002. Testing for genetic evidence of population expansion and contraction: An empirical analysis of microsatellite DNA variation using a hierarchical Bayesian model. Evolution 56:154–166.

Strimmer, K., and O. G. Pybus. 2001. Exploring the demographic history of DNA sequences using the Generalized Skyline Plot. Molecular Biology and Evolution 18:2298–2305.

Tajima, F. 1989a. The effect of change in population size on DNA polymorphism. Genetics 123:597–601.

Tajima, F. 1989b. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585–595.

Tamura, K., D. Peterson, N. Peterson, G. Stecher, M. Nei, and S. Kumar. 2011. MEGA5: Molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Molecular Bology and Evolution 28:2731–9.

Tattersall, I. 2007. Madagascar’s lemurs: Cryptic diversity or taxonomic inflation. Evolutionary Anthropology 16:12 – 23.

Taylor, S. A., C. B. Zavalaga, G. Luna-Jorquera, A. Simeone, D. J. Anderson, and V. L. Friesen. 2010. Panmixia and high genetic diversity in a Humboldt Current endemic, the Peruvian Booby (Sula variegata). Journal of Ornithology 152:623–630.

Tchernov, E. 1986. Évolution des Crocodiles en Afrique du Nord et de l’Est. Éditions du Centre National de la Recherche Scientifique, Paris.

Templeton, A. R., E. Boerwinkle, and C. F. Sing. 1987. A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. I. Basic theory and an analysis of alcohol dehydrogenase activity in Drosophila. Genetics 117:343–351.

Templeton, A. R., K. A. Crandall, and C. F. Sing. 1992. Cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. DNA Sequence 132:619 – 633.

Templeton, A. R., E. Routman, and C. A. Phillips. 1995. Separating population structure from population history: A cladistic analysis of the geographical distribution of mitochondrial DNA haplotypes in the tiger salamander, Ambystoma tigrinum. Genetics 140:767 – 782.

197

Templeton, A. R., and C. F. Sing. 1993. A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. IV. Nested analyses with cladogram uncertainty and recombination. Genetics 134:659–69.

Thorbjarnarson, J. B., F. J. Mazzotti, E. Sanderson, F. Buitrago, M. Lazcano, K. Minkowski, M. Muñiz, P. Ponce, L. Sigler, R. Soberon, A. M. Trelancia, and A. Velasco. 2006. Regional habitat conservation priorities for the American crocodile. Biological Conservation 128:25–36.

Thorbjarnarson, J. B. 1996. Reproductive characteristics of the order Crocodylia. Herpetologica 52:8–24.

Thorbjarnarson, J. B. 2010. Black Caiman Melanosuchus niger in S C Manolis and C. Stevenson, editors Crocodiles. Status Survey and Conservation Action Plan., Third Edit. Crocodile Specialist Group, Darwin.

Thouveny, N., and D. Williamson. 1988. Palaeomagnetic study of the Holocene and Upper Pleistocene sediments from Lake Barombi Mbo, Cameroun: first results. Physics of the Earth and Planetary Interiors 52:193–206.

Udvardy, M. D. F. 1975. A Classification of the Biogeographical Provinces of the World. Morges, Switzerland.

Uilenberg, G., F. Thiaucourt, and F. Jongejan. 2004. On molecular taxonomy: what is in a name? Experimental and Applied Aracology 32:301 – 312.

Van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. micro- checker: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4:535–538.

Vande weghe, J. P. 2007. Gabons National Parks. Loango, Mayumba, and the Lower Ogooue. Wildlife Conservation Society, Libreville, Gabon.

Vasconcelos, W. R., T. Hrbek, R da Silveira, B de Thoisy, L. Ruffeil, and I. P. Farias. 2008. Phylogeographic and conservation genetic analysis of the Black caiman (Melanosuchus niger). Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:600–613.

Vasconcelos, W. R., T. Hrbek, Ronis Da Silveira, Benoit de Thoisy, B. Marioni, and I. P. Farias. 2006. Population genetic analysis of Caiman crocodilus (Linnaeus, 1758) from South America. General Molecular Biology 29:220–230.

Venegas-Anaya, M., A. J. Crawford, A. H. E. Galván, O. I. Sanjur, L. D. Densmore, and E. Bermingham. 2008. Mitochondrial DNA phylogeography of Caiman crocodilus in Mesoamerican and South America. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:614–627.

198

Venegas-Anaya, M. Cryptic species as a frame on use and conservation of Crocodylus acutus in Crocodiles: Proceedings of the 19th Working Meeting of the Crocodile Specialist Group, IUCN - The World Conservation Union, Gland, Switzerland and Cambridge, UK.

Vincens, A., Dominique Schwartz, H. Elenga, I. Reynaud-Farrera, A. Alexandre, J. Bertaux, A. Mariotti, L. Martin, J.-D. Meunier, F. Nguetsop, M. Servant, S. Servant- Vildary, and D. Wirrman. 1999. Forest response to climate changes in Atlantic Equatorial Africa during the last 4000 years BP and inheritance on the modern landscapes. Journal of Biogeography 26:879–885.

Waitkuwait, W. E. 1985. Investigations of the breeding biology of the West African slender-snouted crocodile Crocodylus cataphractus. Amphibia-Reptilia 6:387–399.

Waitkuwait, W. E. 1989. Present knowledge of the West African slender-snouted crocodile, Crocodylus cataphractus Cuvier 1824 and the West African dwarf crocodile Osteolaemus tetraspis, Cope 1861. in Crocodile Specialist Group of the Species Survival Commission, editor Crocodiles: Ecology, Management and Conservation International Union for Conservation of Nature, Gland, Switzerland.

Walsh, B. 1987. Crocodile capture methods used in the Northern Territory of Australia. in G J W Webb, S C Manolis, and P. J. Whitehead, editors Wildlife Management: Crocodiles and Alligators Surrey Beatty and Sons Pty Ltd., Chipping Norton, New South Wales, Australia. 522 pp.

Waples, R. S., and O. Gaggiotti. 2006. What is a population? An empirical evaluation of some genetic methods for identifying the number of gene pools and their degree of connectivity. Molecular Ecology 15:1419–39.

Warner, B. F., and S. E. Huggins. 1978. An electroencephalographic study of sleep in young caimans in a colony. Comparative Biochemistry and Physiology Part A: Physiology 59:139–144.

Watterson, G. A. 1975. On the number of segregating sites in genetical models without recombination. Theoretical Population Biology 7:256 – 276.

Watterson, G. A. 1986. Homozygosity test after a change in population size. Genetics 112:899 – 907.

Weaver, J. P., D. Rodriguez, M. Venegas-Anaya, J. R. Cedeño-Vásquez, M. R. J. Forstner, and L. D. Densmore. 2008. Genetic characterization of captive Cuban crocodiles (Crocodylus rhombifer) and evidence of hybridization with the American crocodile (Crocodylus acutus). Journal of Experimental Zoology Part A: Ecological Genetics and Physiology 309A:649–660.

199

Webb, Grahame J W, R. Buckworth, and S Charlie Manolis. 1983. Crocodylus johnstoni in the McKinlay River, N.T. VI. Nesting biology. Australian Wildlife Research 10:607–637.

Weijers, J. W. H., E. Schefuss, S. Schouten, and J. S. Sinninghe Damsté. 2007. Coupled thermal and hydrological evolution of tropical Africa over the last deglaciation. Science 315:1701–4.

Weiss, G., and A. von Haeseler. 1998. Inference of population history using a likelihood approach. Genetics 149:1539–1546.

White, P S, and L. D. Densmore. 2001. DNA sequence alignment and data analysis methods: their effect on the recovery of crocodylian relationships. in G. C. Grigg, F. Seebacher, and C. E. Franklin, editors Crocodilian Biology and Evolution. Surrey Beatty & Sons Pty. Limited, Chipping Norton, N.S.W., Australia.

Wieczorek, A. M., R. C. Drewes, and A. Channing. 2000. Biogeography and evolutionary history of Hyperolius species: application of molecular phylogeny. Journal of Biogeography 27:1231 – 1243.

Wirrman, D., J. Bertaux, and A. Kossoni. 2001. Late holocene paleoclimatic changes in western Central Africa inferred from mineral abundance in dated sediments from Lake Ossa (southwest Cameroon). Quaternary Research 56:275 – 287.

Wittmer, H. U., B. N. McLellan, D. R. Seip, J. A. Young, T. A. Kinley, G. S. Watts, and D. Hamilton. 2005. Population dynamics of the endangered mountain ecotype of woodland caribou (Rangifer tarandus caribou) in British Columbia, Canada. Canadian Journal of Zoology 83:12.

Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97 – 159.

Wright, S. 1938. Size of population and breeding structure in relation to evolution. Science 87:430 – 431.

Wu, H., J. Guiot, S. Brewer, and Z. Guo. 2007. Climatic changes in Eurasia and Africa at the last glacial maximum and mid-Holocene: reconstruction from pollen data using inverse vegetation modelling. Climate Dynamics 29:211–229.

Yang, Z., and B. Rannala. 2006. Bayesian estimation of species divergence times under a molecular clock using multiple fossil calibrations with soft bounds. Molecular Biology and Evolution 23:212–26.

Yarrell, W. 1835. Character of a new species of crocodile (Crocodylus leptorhynchus). Proceedings of the Zoological Society of London 3:128–133.

200

Zellmer, A. J., and L. L. Knowles. 2009. Disentangling the effects of historic vs. contemporary landscape structure on population genetic divergence. Molecular Ecology 18:3593–602.

Zwickl, D. J., and D. M. Hillis. 2002. Increased taxon sampling greatly reduces phylogenetic error. Systematic Biology 51:588 – 598.

2000. Species Concepts and Phylogenetic Theory: A Debate (Q. D. Wheeler and R. Meier, editors). Columbia University Press, New York.

2001. Dispersal (J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, editors). Oxford University Press, Oxford.

2004. Evolutionary Conservation Biology (R. Ferriere, U. Dieckmann, and D. Couvet, editors). Cambridge University Press, Cambridge.

201

BIOGRAPHICAL SKETCH

Matthew (Matt) H. Shirley received his PhD in wildlife ecology and conservation from the University of Florida in the spring of 2013. He previously earned a Master of

Science, also in wildlife ecology and conservation from the University of Florida, in 2007 and Bachelor of Science in ecology and evolutionary biology from Yale University in

2003. Matt was born in Tulsa, Oklahoma in 1981 and is the son of Jim and Barbara

Shirley from whom he learned to respect and love nature. His formative teen and high school years were spent in Sarasota, Florida and it was during this time that he caught the travel bug visiting Nicaragua, Mexico and Spain. During his bachelor’s program he studied abroad in Australia and conducted senior thesis research on gigantothermy in sea turtles. He first visited Central Africa when he implemented bird surveys as part of a post-grad research project designed with friends at Yale. Before starting graduate school he was a bird guide in Brazil, continued bird surveys in Equatorial Guinea, and lived in Dominica for a year scuba diving and looking for endemic parrots. While at the

University of Florida he has had the good fortune of visiting 13 additional African countries, mostly for research but not without lots of fun seeing many wildlife species others only know through books and documentaries. While his research has focused on many aspects of crocodile evolution, he has enjoyed expanding the conservation focus of his work by prioritizing research on the status, distribution and ecology of the least known crocodilians in the world, as well as working to develop capacity and enthusiasm for conservation amongst African students and technicians. In 2012 he was recognized for these efforts by the IUCN/SSC Crocodile Specialist Group with the

Castillo Award for significant contributions to global crocodile conservation.

202