Squamata: Gymnophthalmidae): the Diversity in Southern Ecuador Revisited

Home , Alopoglossus, Alopoglossus angulatus, Alopoglossus atriventris, Ameivula, Ameivula ocellifera, Anadia ocellata

Philosophy, practice and implications of phylogenetic inference as exemplified by Neotropical microteiid lizards (Reptilia: Gymnophthalmidae, Alopoglossidae)

Santiago J. Sánchez-Pacheco

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Ecology and Evolutionary Biology University of Toronto

© Copyright Santiago J. Sánchez-Pacheco, 2017 Philosophy, practice and implications of phylogenetic inference as exemplified by Neotropical microteiid lizards (Reptilia: Gymnophthalmidae, Alopoglossidae)

Santiago J. Sánchez-Pacheco

Doctor of Philosophy

Department of Ecology and Evolutionary Biology University of Toronto

2017

Abstract

Understanding evolutionary processes involved in the diversity of life first requires knowledge of phylogeny. This knowledge is necessary to explain the evolutionary origins of features and provides a predictive framework to guide research. Biodiversity demands explanation, and heritable variation provides the evidential basis for formulating and testing phylogenetic hypotheses. My goal herein is to advance understanding of phylogenetic inference by addressing theoretical, analytical and empirical problems, and the rationale behind my approaches to solving them. To this end, I use a combination of approaches both to make phylogenetic inferences and to advance the field. Consequently, I present a collection of broadly overlapping contributions. I begin with a review of median-joining (MJ), a method for phylogeographic analysis. I show that MJ networks are theoretically untenable for phylogenetic inference. I also discuss the concept of phylogenetic network. Given the confusion that afflicts the implementation of concepts in some methods, I explore the concept of cladogram and argue that cladograms do not necessarily require

ii synapomorphies, but synapomorphies are required to test and ultimately falsify cladograms. I conclude that both cladograms and synapomorphies are required to achieve phylogenetic explanation. I continue with an evaluation of the impact of phenotypic evidence on molecular datasets. I show that inclusion of phenotypic evidence can alter both topologies and support values in phylogenetic analyses. I use an optimal phylogenetic hypothesis to formulate and test biogeographic hypotheses involving Neotropical montane regions, and to analyze character evolution. I finish with the development of a theory of outgroup sampling, a fundamental step in phylogenetic analysis, grounded in the logic of scientific discovery. Its objective is to test hypotheses of ingroup topology and homology as severely as possible. This framework provides a logical basis for sampling and successively increases severity of hypothesis-testing, but it does not provide any grounds for limiting the sample. I then propose a heuristic procedure that provides an empirical basis to limit sampling. For most projects, I use Neotropical microteiid lizards. Finally, biodiversity also demands documentation, and my research often results in taxonomic novelties. Herein, I describe three new species and erect two new genera of microteiid lizards.

iii Dedication

Being a woman with a master’s degree in mathematics and teaching in a Colombian university in the late 70’s and in the 80’s was an extreme rarity due to strong cultural barriers. Stella Pacheco not only pursued an academic career under such circumstances, but also managed to raise a family at the same time.

I dedicate this work to Stella Pacheco, my mother, who passed away while I was completing my master’s degree in Brazil. She evidently shaped my life by giving me the opportunity to grow up amidst stacks of papers, books and exams, and often on campus. I will always be grateful for her early influence, unrelenting encouragement, and guidance by word, teaching and example, without which I surely would not have pursued a career in science. Her mathematics lessons made their way into this PhD thesis (Chapter 2).

iv Acknowledgments

My first acknowledgement goes to my supervisor Robert W. Murphy. I thank him for his encouragement, advice and friendship, and continued financial support. I am indebted to him for letting me follow my interests as they developed, and for constantly challenging me. He has greatly shaped the kind of scientist I strive to be.

For her unrelenting encouragement and patience, countless sacrifices, unconditional support, and never-ending love I am deeply grateful to Paola Pulido-Santacruz, my precious wife, without whom my graduate studies would not have been possible. My father, Luis F. Sánchez, and my siblings, Victoria E. Sánchez-Pacheco and Juan F.

Sánchez-Pacheco, have always provided encouragement in all my endeavors, often abroad.

I thank my Committee members for their numerous insights that greatly improved this dissertation: Deborah McLennan and David C. Evans. Douglas Currie, Nathan

Lovejoy, Sebastian Kvist and Michael Caldwell generously took part in my appraisal and final examinations. My approach to phylogenetics has been influenced heavily by

Robert W. Murphy and Taran Grant, and I acknowledge them for sharing their knowledge and causing me to question my assumptions.

I am grateful to my friends and colleagues—no one mentioned no one forgotten— who facilitated all aspects of my graduate studies. Collaborators on my PhD projects are included in the relevant chapters. The Royal Ontario Museum and the Department of Ecology and Evolutionary Biology at the University of Toronto (EEB) provided intellectual, stimulating environments and friendships during my five years as a PhD student.

v Throughout my PhD, I was financially supported by a COLCIENCIAS doctoral fellowship (Becas Francisco José de Caldas) from the Colombian Department of

Science, Technology and Innovation, an Ontario Graduate Scholarship at the

University of Toronto, and a number of grants provided by EEB. The research in this thesis was funded through a grant from the Natural Sciences and Engineering Research

Council (NSERC) to Robert W. Murphy.

vi Table of Contents

Dedication ...... iv

Acknowledgments ...... v

Table of Contents ...... vii

List of Tables ...... x

List of Figures ...... xi

Chapter 1 GENERAL INTRODUCTION ...... 1

1.1 Phylogenetic analysis and research objective ...... 1 1.2 Study group ...... 2 1.3 Research outline ...... 3 1.3.1 Chapter 2. On the use of median-joining networks in evolutionary biology. . 4 1.3.2 Chapter 3. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms...... 4 1.3.3 Chapter 4. Lizards of the genus Riama (Squamata: Gymnophthalmidae): The diversity in southern Ecuador revisited...... 5 1.3.4 Chapter 5. Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic evidence on molecular datasets, and the origin of the Sierra Nevada de Santa Marta endemic fauna...... 6 1.3.5 Chapter 6. Formal recognition of the species of Oreosaurus (Squamata: Gymnophthalmidae) from the Sierra Nevada de Santa Marta, Colombia...... 7 1.3.6 Chapter 7. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards...... 7 Chapter 2 ON THE USE OF MEDIAN-JOINING NETWORKS IN

EVOLUTIONARY BIOLOGY ...... 9

2 Abstract ...... 10 2.1 Introduction ...... 11 2.2 Theoretical overview ...... 12 2.3 MJNs in the literature ...... 23 2.4 Conclusions ...... 26 2.5 Acknowledgements ...... 26 2.6 Figures ...... 28 Chapter 3 CLADOGRAMS DO NOT NECESSARILY ENTAIL

SYNAPOMORPHIES, BUT SYNAPOMORPHIES FALSIFY CLADOGRAMS

...... 32

3 ...... 34

vii 3.1 Cladogram, cladograms without synapomorphies and the illegitimacy of “synapomorphy Î cladogram” ...... 34 3.2 The logically consistent relationship between cladogram and synapomorphy ...... 36 3.3 Acknowledgements ...... 38 Chapter 4 LIZARDS OF THE GENUS Riama (Squamata: Gymnophthalmidae):

THE DIVERSITY IN SOUTHERN ECUADOR REVISITED ...... 39

4 Abstract ...... 40 4.1 Introduction ...... 41 4.2 Material and methods ...... 41 4.3 Systematics ...... 42 4.4 Discussion ...... 61 4.5 Acknowledgments ...... 64 4.6 Figures ...... 65 Chapter 5 PHYLOGENY OF Riama (Squamata: Gymnophthalmidae), IMPACT

OF PHENOTYPIC EVIDENCE ON MOLECULAR DATASETS, AND THE

ORIGIN OF THE SIERRA NEVADA DE SANTA MARTA ENDEMIC FAUNA

...... 73

5 Abstract ...... 74 5.1 Introduction ...... 76 5.2 Materials and methods ...... 80 5.3 Results ...... 88 5.4 Discussion ...... 92 5.5 Acknowledgements ...... 109 5.6 Figures ...... 110 5.7 Tables ...... 123 Chapter 6 FORMAL RECOGNITION OF THE SPECIES OF Oreosaurus

(Squamata: Gymnophthalmidae) FROM THE SIERRA NEVADA DE SANTA

MARTA, COLOMBIA ...... 129

6 Abstract ...... 130 6.1 Introduction ...... 131 6.2 Materials and Methods ...... 132 6.3 Species description ...... 133 6.4 Acknowledgments ...... 140 6.5 Figures ...... 142 Chapter 7 OUTGROUP SAMPLING CRITERIA: SEVERITY OF TEST,

EXPANSION, STABILITY, AND ALOPOGLOSSID LIZARDS ...... 147

7 Abstract ...... 148 7.1 Introduction ...... 150

viii 7.2 Background and objective ...... 151 7.3 The scientific objectives of outgroup sampling ...... 156 7.4 Outgroup sampling and tests of topology ...... 157 7.5 Outgroup sampling and tests of homology ...... 159 7.6 Successive outgroup expansion ...... 161 7.7 Alopoglossid lizards and successive outgroup expansion ...... 164 7.8 Successive outgroup expansion and tree-alignment+maximum parsimony analysis (TA+MP) ...... 168 7.9 Successive outgroup expansion and similarity-alignment+maximum parsimony analysis (SA+MP) ...... 171 7.10 Successive outgroup expansion and similarity-alignment+maximum likelihood analysis (SA+ML) ...... 173 7.11 Discussion ...... 175 7.12 Figures ...... 181 7.14 Tables ...... 198 Chapter 8 CONCLUDING DISCUSSION ...... 205

8 ...... 205 8.1 Summary of thesis chapters ...... 205 8.1.1 Chapter 2. On the use of median-joining networks in evolutionary biology. 205 8.1.2 Chapter 3. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms...... 206 8.1.3 Chapter 4. Lizards of the genus Riama (Squamata: Gymnophthalmidae): The diversity in southern Ecuador revisited...... 206 8.1.4 Chapter 5. Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic evidence on molecular datasets, and the origin of the Sierra Nevada de Santa Marta endemic fauna...... 207 8.1.5 Chapter 6. Formal recognition of the species of Oreosaurus (Squamata: Gymnophthalmidae) from the Sierra Nevada de Santa Marta, Colombia...... 208 8.1.6 Chapter 7. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards...... 208 8.2 Avenues for future work ...... 210 Appendices ...... 214

9 Appendix S1 ...... 214 10 Appendix S2 ...... 218 11 Appendix S3 ...... 242 12 Appendix S4 ...... 250 13 Appendix S5 ...... 251 14 Appendix S6 ...... 260 15 Appendix S7 ...... 260 References ...... 262

ix List of Tables

Table 5.1. List of PCR and sequencing primers used in this study, and a summary of the PCR conditions...... 123

Table 5.2. GenBank accession numbers for loci and terminals sampled in this study. Asterisks indicate new sequences obtained for this study. Species are listed following the new taxonomy proposed herein. Numbers and letters following species names are identifiers of conspecific terminals. See Materials and Methods for institutional abbreviations...... 125

Table 7.1. List of PCR and sequencing primers used in this study, and a summary of the PCR conditions...... 198

Table 7.2. GenBank accession numbers for loci and terminals sampled in this study. Asterisks indicate new sequences obtained for this study. Species are listed following the new taxonomy proposed herein...... 199

Table 7.3. Summary of tree-alignment+maximum parsimony (TA+MP) and similarity- alignment+maximum parsimony (SA+MP) analyses of increasingly inclusive datasets to test

Ptychoglossus monophyly. Length is the equally weighted parsimony cost of the optimal tree(s).

Datasets include all terminals from the previous dataset plus the listed terminals...... 203

List of Figures

Figure 2.1. Absolute number of citations of Bandelt et al. (1999) between 1999 and (December)

2014 (ISI Web of Science)...... 28

Figure 2.2. A phylogenetic network. Note the absence of cycles due to the direction of the edges

(arrows). A reticulation event is represented by a node with in-degree 2 and out-degree 1 (v5). . 29

Figure 2.3. A Median Joining Network. Note the presence of cycles due to the absence of direction

(square). Each circle represents a unique haplotype where the diameter is proportional to the number of DNA sequences represented. Integers on each edge denote the position of nucleotides within the sequence that differ between haplotypes. Small solid circles indicate median vectors.

...... 30

Figure 2.4. The distance matrix D on S = {s1, s2, s3…, si} shown in (a) gives rise to three different minimum spanning trees (networks), shown in (b), (c) and (d), respectively. The corresponding minimum spanning (super)network N is shown in (e). Modified from Huson et al. (2010)...... 31

Figure 4.1. Riama aurea, new species. Dorsal, lateral, and ventral views of head of holotype

(QCAZ 07886, 57.4 mm SVL]...... 66

Figure 4.2. Riama aurea, new species. Dorsal and ventral views of holotype (QCAZ 07886, 57.4 mm SVL). Photos courtesy: Pedro H. Bernardo...... 67

Figure 4.3. Riama kiziriani, new species. Dorsal, lateral, and ventral views of head of holotype

(QCAZ 9667, 61.0 mm SVL]...... 68

Figure 4.4. Riama kiziriani, new species. Dorsal and ventral views of holotype (QCAZ 9667, 61.0 mm SVL) in life. Photos: Omar Torres-Carvajal...... 69

Figure 4.5. Riama vespertina. Dorsal and ventral views of holotype (AMNH 22130, 40 mm SVL).

Photos courtesy: David A. Kizirian...... 70

Figure 4.6. Riama vespertina. Dorsolateral and ventral views of the adult male in life QCAZ 10283

(56.0 mm SVL). Photos courtesy: Silvia Aldás-Alarcón...... 71

Figure 4.7. Distribution of Riama aurea, R. kiziriani and R. vespertina in southern Ecuador.

Arrows indicate type localities...... 72

Figure 5.1. Strict consensus of two most parsimonious trees (9680 steps) from the total evidence analysis. Values above branches are Goodman-Bremer support and below branches are REP support. Yellow = Riama sensu stricto (i.e. R. unicolor, type species of Riama, is included in this clade), including 14 nominal and three undescribed species; blue = a clade composed of nine nominal species currently referred to Riama; red = a clade comprising two nominal and two undescribed species referred to Riama plus Anadia mcdiarmidi. The non-monophyletic

Pantodactylus, which is embedded within Cercosaura, is highlighted in green. The new taxonomy proposed herein (Appendix S5) is represented...... 112

Figure 5.2. Strict consensus of five most parsimonious trees (9464 steps) from the molecular-only analysis. Values above branches are Goodman-Bremer support and below branches are REP support. Taxonomic changes proposed herein are adopted (Fig. 6. 1 and Appendix S5)...... 114

Figure 5.3. Map of northern South America and summary of the phylogeny and geographic distribution of Oreosaurus. SNSM = Sierra Nevada de Santa Marta, Colombia; TCM = tepuis from the Chimantá massif, Venezuela; CCC = Cordillera de la Costa Central, Venezuela; IT = island of

Trinidad; CCO = Cordillera de la Costa Oriental, Venezuela...... 115

Figure 5.4. Lateral view and asulcate face of the hemipenis of Riama orcesi (KU 142919).

Character 0, shape of hemipenial body; state 0, cylindrical. Char. 2(2): Character 2, flounce orientation on asulcate face of hemipenial body; state 2, horizontal (with no vertex). Char. 3(1):

Character 3, asulcate central nude area; state 1, narrow, restricted to a sagittal stripe. Char. 4(0):

Character 4, orientation of lateral body flounces; state 0, chevron shaped. Char. 5(1): Character 5,

xii

lateral body flounce ornamentation; state 1, present. Char. 6(1): Character 6, position of lateral body flounce ornamentation; state 1, distributed over entire flounce. Char. 7(0): Character 7, shape of lateral body flounce ornamentation; state 0, comb-like series of spicules. Char. 8(1): Character

8, isolated transversal flounces on proximal-central region of asulcate face; state 1, present. .. 116

Figure 5.5. Lateral view and asulcate face of the hemipenis of Riama balneator (DHMECN 4111).

Character 0, shape of hemipenial body; state 1, elongated. Char. 3(2): Character 3, asulcate central nude area; state 2, broad, occupying approximately 50% of the asulcate face. Char. 4(1): Character

4, orientation of lateral body flounces; state 1, extended diagonally from anterior (asulcate) to posterior (sulcate) face. Char. 6(0): Character 6, position of lateral body flounce ornamentation; state 0, distal, restricted to flounce extremities. Char. 7(1): Character 7, shape of lateral body flounce ornamentation; state 1, isolated hook-shaped spines...... 117

Figure 5.6. Sulcate and asulcate faces of the hemipenis of Riama crypta (KU 135104). Character

0, shape of hemipenial body; state 2, conical, with proximal region distinctly thinner than distal and lobes. Char. 1(0): Character 1, lobes; state 0, large, distinct from hemipenial body. Char. 9(1):

Character 9, distal filiform appendages on the hemipenial lobes, state 1, present...... 118

Figure 5.7. Asulcate face of the hemipenis of Riama simotera (ICN-R 9836). Character 0, shape of hemipenial body; state 3, globose. Char. 1(1): Character 1, lobes; state 1, narrow, uniform with the hemipenial body. Char. 2(0): Character 2, flounce orientation on asulcate face of hemipenial body; state 0, lateral (with a central vertex directed distally). Char. 3(0): Character 3, asulcate central nude area; state 0, absent, flounces extended across entire asulcate face. Char. 8(0):

Character 8, isolated transversal flounces on proximal-central region of asulcate face; state 0, absent. Char. 9(0): Character 9, distal filiform appendages on the hemipenial lobes; state 0, absent.

...... 119

xiii

Figure 5.8. Asulcate face of the hemipenis of Riama cashcaensis (KU 217206). Char. 2(1):

Character 2, flounce orientation on asulcate face of hemipenial body; state 1, medial (with a central vertex directed proximally)...... 120

Figure 5.9. Lateral view of the hemipenis of Riama striata (KU 217206). Char. 4(2): Character

4, orientation of lateral body flounces; state 2, horizontal...... 121

Figure 5.10. Lateral view of the hemipenis of Bachia flavescens (AMNH 140925). Char. 5(0):

Character 5, lateral body flounce ornamentation; state 0, absent...... 122

Figure 6.1. Oreosaurus serranus sp. n. (holotype, ROM 53608 [70.4 mm SVL]). Dorsal, lateral and ventral views of the head, and ventral view of the pelvic region...... 142

Figure 6.2. Oreosaurus serranus sp. n. (paratype, ROM 53609 [68.6 mm SVL]) in life. Photos:

Sergio Marques de Souza (top) and Jhon Jairo Ospina-Sarria (bottom)...... 143

Figure 6.3. Oreosaurus serranus sp. n. Sulcate (left), lateral (center) and asulcate (right) views of the right hemipenis of ROM 53610 (paratype)...... 144

Figure 6.4. Type locality (top) and habitat (bottom) of Oreosaurus serranus sp. n. in the Sierra

Nevada de Santa Marta, Colombia. Photos: Jhon Jairo Ospina-Sarria (top) and Sergio Marques de

Souza (bottom)...... 145

Figure 6.5. Summary of the phylogeny and geographic distribution of Oreosaurus (Sánchez-

Pacheco et al. 2017). SNSM = Sierra Nevada de Santa Marta, Colombia; TCM = tepuis from the

Chimantá massif, Venezuela; CCC = Cordillera de la Costa Central, Venezuela; IT = island of

Trinidad; CCO = Cordillera de la Costa Oriental, Venezuela. Oreosaurus luctuosus, from the CCC, and O. rhodogaster, from the CCO, were included in this genus due to the presumed close relationships of these species and O. achlyens and O. shrevei, respectively. Data taken from

Sánchez-Pacheco et al. (2017)...... 146

xiv

Figure 7.1. Outgroup sampling and tests of ingroup synapomorphies. Top: Using O1 (which could be a single terminal or an entire clade) to root the tree, character optimization cannot determine if characters 1 and 2 are ingroup or outgroup apomorphies. Treating characters 1 and 2 as ingroup synapomorphies requires the assumption that all character-states in O1 are plesiomorphic. Bottom:

Using outgroup taxon O2 to root the tree reveals that character 1 is apomorphic in O1 and character

2 is an ingroup synapomorphy, thus refuting the hypothesis that all character-states in O1 are plesiomorphic. Although there are now no ambiguous optimizations, there is also no evidence for the O1+ingroup clade and the root node should be shown as unresolved...... 181

Adding outgroup taxon O4 removes O3 from the ingroup. State 1 of character 1 is now explained as unique and unreversed in the ingroup and state 1 of characters 2 and 3 is independently derived in O3 and (B C D) and (C D), respectively...... 182

Figure 7.3. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony prior to outgroup expansion. Strict consensus of six most parsimonious trees from Analysis 1TA+MP (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus...... 183

Figure 7.4. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 25 most parsimonious trees from Analysis 2TA+MP, analysis of data in Analysis 1TA+MP plus eight additional outgroup terminals (details in text and

Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus. Position of P. bicolor differs from Analysis 1...... 184

Figure 7.5. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 47 most parsimonious trees from Analysis 3TA+MP, analysis of data in Analysis 2TA+MP plus four additional outgroup terminals (details in text and

Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic.

...... 185

Figure 7.6. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 85 most parsimonious trees from Analysis 4TA+MP, analysis of data in Analysis 3TA+MP plus four additional outgroup terminals (details in text and

Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic.

...... 186

Figure 7.7. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 90 most parsimonious trees from Analysis 5TA+MP, analysis of data in Analysis 4TA+MP plus four additional outgroup terminals (details in text and

Table 3). Goodman-Bremer support values given above branches. Ptychogloosus is polyphyletic.

...... 187

Figure 7.8. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony prior to outgroup expansion. Strict consensus of 12 most parsimonious trees from Analysis 1SA+MP (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus...... 188

Figure 7.9. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 60 most parsimonious trees from Analysis

2SA+MP, analysis of data in Analysis 1SA+MP plus eight additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is

xvi

paraphyletic with respect to Alopoglossus. Positions of P. bicolor and P. myersi differ from

Analysis 1SA+MP...... 189

Figure 7.10. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 24 most parsimonious trees from Analysis

3SA+MP, analysis of data in Analysis 2SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic. The second clade of Ptychoglossus (details in text) is the sister group of the remaining alopoglossids...... 190

Figure 7.11. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 48 most parsimonious trees from Analysis

4SA+MP, analysis of data in Analysis 3SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic. The first (instead the second) clade of Ptychoglossus (details in text) is the sister group of the remaining alopoglossids...... 191

Figure 7.12. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 44 most parsimonious trees from Analysis

5SA+MP, analysis of data in Analysis 4SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychogloosus is polyphyletic...... 192

Figure 7.13. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood prior to outgroup expansion. Optimal solution from Analysis 1SA+ML (details in text and Table 3).

Ptychoglossus is paraphyletic with respect to Alopoglossus...... 193

Figure 7.14. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 2SA+ML, analysis of data in Analysis

xvii

1SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is paraphyletic with respect to Alopoglossus...... 194

Figure 7.15. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 3SA+ML, analysis of data in Analysis

2SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic...... 195

Figure 7.16. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 4SA+ML, analysis of data in Analysis

3SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic...... 196

Figure 7.17. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 5SA+ML, analysis of data in Analysis

4SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic...... 197

xviii

Chapter 1 GENERAL INTRODUCTION

1.1 Phylogenetic analysis and research objective

Knowledge of phylogenetic relationships among extant and extinct organisms is paramount for understanding the evolutionary processes involved in the diversity of life. Detailed knowledge of phylogeny is necessary to explain the evolutionary origins of features and provides an essential predictive framework to guide research.

Phylogenetic tools allow us to identify natural groups of organisms and how they are related to each other (Donoghue et al. 1989; Grant et al., 2006).

Observed biodiversity (i.e., species and their taxonomy, genomes, physiology, behavior, anatomy, ecology, and biogeography) demands explanation. Heritable variation at the levels of DNA and phenotype provides the evidential basis for formulating and testing phylogenetic explanations (hypotheses). The fundamental task of phylogenetic analysis is to explain biological variation by inferring relationships among organisms and the unique transformation events that link them (Hennig, 1966).

Although evolutionary analysis requires additional assumptions and tests external to the phylogeny, inference of particular functions, mechanisms of adaptations, and constraints, and the many other processes that shaped the evolution of each group, can be informed by the results of phylogenetic analysis (Wheeler et al., 2006). The novel knowledge that emerges from phylogenetic analysis has implications beyond the immediate problems of systematics. By providing a causally relevant framework of reference, knowledge of phylogeny often leads to unanticipated insights and identifies novel problems for further investigation (Grant et al., 2006; e.g., biogeography and

character evolution). Further, observed biodiversity demands not only explanation, but also documentation, and improved knowledge of phylogeny facilitates taxonomic work by highlighting relevant comparisons. Progress in taxonomy forms the foundation for future research in diversity.

From afar, phylogenetic analysis may appear to be a simple exercise in point-and- click computing, but in reality it is a complex, theory-laden, analytically challenging undertaking. This is especially true of the present thesis, which aims to advance understanding of phylogenetic inference by addressing theoretical, analytical and empirical problems, and the rationale behind my approaches to solving them. Often, disagreements in inferred phylogenies stem as much from the use of different discovery operations and assumptions as from empirical conflict.

1.2 Study group

My empirical research focuses mostly on the phylogenetic diversification of non- avian reptiles and how it relates to the evolution of particular character systems, including morphological and molecular evolution. For most projects in this thesis

(Chapters 4–7), I use Neotropical gymnophthalmid and alopoglossid lizards (235 and

22 species, respectively), known collectively as microteiids, as the study group. They occur in a great diversity of habitats and exhibit multiple cases of semi-aquatic, arboreal and secretive and/or burrowing habits, different degrees of limb reduction and body elongation, loss of eyelids and external ear openings, and parthenogenesis. This variation offers an exceptional opportunity to address evolutionary questions, and has given rise to a growing interest in many areas of ecology and biology. Among the diverse studies are investigations of chromosomal variation (e.g., Pellegrino et al.,

1999), reproductive biology (e.g., Ramos-Pallares et al., 2010), comparative anatomy

(e.g., Roscito and Rodrigues, 2010), locomotor behavior (Höfling and Renous, 2004), ecological features (e.g., Vitt et al., 2007), dietary variation (Doan, 2008), origins of unisexual species (e.g., Kizirian and Cole, 1999), and microhabitat use and evolution of cranial design (Barros et al., 2011). Similarly, research in comparative and developmental morphology has revealed interesting structures (e.g., Tarazona and

Ramírez-Pinilla, 2008). Most of the species occur throughout South America, with relatively few species in Middle America. More specific information on the groups studied, including systematics and geographic distribution, and on the data used is provided in the relevant chapters.

1.3 Research outline

The integration of different lines of investigation forms an essential part of a progressive research program. In this thesis, I use a combination of approaches both to make phylogenetic inferences and to advance the field. As a consequence, herein I present a collection of broadly overlapping theoretical, analytical, and empirical contributions. In evolutionary biology, as in any other scientific discipline, theory

(including concepts and methods) determines which observations are scientifically relevant and how they may be analyzed. Only after theoretically defined concepts have been established, can valid operations be developed and/or implemented in empirical research (Grant, 2002). Phylogenetics is not an exception. It requires a general and logically consistent analytical framework (Wheeler et al., 2006). Accordingly, (i) theoretical foundations are necessary to allow consistent inferences to be made

(Chapters 2, 3, 5 and 7), (ii) rigorous analytical techniques allow otherwise complex

problems to become feasible (Chapters 5 and 7), (iii) heritable variation provides the evidential basis for formulating and testing evolutionary hypotheses that explain biodiversity (Chapters 5 and 7), and (iv) observed biodiversity demands documentation

(Chapters 4–7).

Below, I outline the research chapters of my thesis. All chapters were written as independent, publishable units and have been (or are being) published in peer-reviewed journals (citations are provided).

1.3.1 Chapter 2. On the use of median-joining networks in evolutionary biology.

I begin in this chapter with a detailed review of the efficacy of median-joining (MJ) as a method for phylogenetic inference in general, and phylogeographic analysis in particular. I show that MJ networks (distance-based, undirected branching diagrams with cycles) are theoretically untenable for evolutionary inference. I also discuss the concept of a phylogenetic network as opposed to a non-evolutionary network.

This chapter was published in Cladistics (Kong, S.*, Sánchez-Pacheco, S.J.*,

Murphy, R.W. 2016. On the use of median-joining networks in evolutionary biology.

Cladistics 32, 691–699. * = equal contribution, co-first authors).

1.3.2 Chapter 3. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms.

One of the messages that emerges from Chapter 2 is that after 50 years of post-

Hennigian debate, confusion still afflicts the implementation of phylogenetic concepts in some methods. Therefore, in this chapter I discuss the concept of cladogram (rooted,

directed branching diagrams), and clarify the only logically consistent relationship between cladogram and synapomorphy: cladograms do not necessarily require synapomorphies, but synapomorphies are required to test and ultimately falsify cladograms. I conclude that both cladograms and synapomorphies are required to achieve phylogenetic explanation.

This chapter was submitted to Cladistics and is currently under review (Sánchez-

Pacheco, S.J., Grant., T., Murphy, R.W. In review. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms. Cladistics).

1.3.3 Chapter 4. Lizards of the genus Riama (Squamata: Gymnophthalmidae): The diversity in southern Ecuador revisited.

Having established a theoretically and logically consistent framework in Chapters 2 and 3, I infer the phylogeny of Riama, the most speciose genus of gymnophthalmid lizards, to address questions in analytical techniques, biogeography, and character evolution (Chapter 5). However, it is important to test microevolutionary hypotheses

(e.g., species identities) before pursuing macroevolutionary studies. Therefore, in this chapter, I review the diversity of Riama in southern Ecuador, where the greatest void in the knowledge of the distribution of Riama occurs (Kizirian, 1996). I use morphological and molecular evidence to address taxonomic questions. My investigation results in the description of two new species and the redescription of another one. Further, I discuss morphological variation to propose explicit character- states, which I employ to delimite characters used in Chapter 5.

This chapter was published in South American Journal of Herpetology (Sánchez-

Pacheco, S.J., Aguirre-Peñafiel, V., Torres-Carvajal, O. 2012. Lizards of the genus

Riama (Squamata: Gymnophthalmidae): The diversity in southern Ecuador revisited.

South American Journal of Herpetology 7, 259–275).

1.3.4 Chapter 5. Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic evidence on molecular datasets, and the origin of the Sierra Nevada de Santa Marta endemic fauna.

Having made progress on alpha taxonomy and analysis of character variation, this chapter tests current knowledge of diversification in Riama as severely as possible by combining new and prior genotypic and phenotypic evidence in a total evidence (TE) analysis. Also, to evaluate the impact of phenotypic evidence on molecular datasets, I compare my TE results with those obtained from analyses using DNA sequence data only. Inclusion of a small amount of phenotypic evidence alters both the topology and support values of clades that do not differ, showing the relevance of non-molecular evidence in phylogenetic analyses. I then use the optimal phylogenetic hypothesis to formulate and test biogeographic hypotheses involving different Neotropical montane regions, to analyze character evolution, and to propose a genus-level monophyletic taxonomy that reflects inferred historical relationships.

This chapter was published in Cladistics (Sánchez-Pacheco, S.J., Torres-Carvajal,

O., Aguirre-Peñafiel, V., Nunes, P.M.S., Verrastro, L., Rivas, G.A., Rodrigues, M.T.,

Grant, T., Murphy, R.W. 2017. Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic evidence on molecular datasets, and the origin of the Sierra

Nevada de Santa Marta endemic fauna. Cladistics DOI: 10.1111/cla.12203).

1.3.5 Chapter 6. Formal recognition of the species of Oreosaurus (Squamata: Gymnophthalmidae) from the Sierra Nevada de Santa Marta, Colombia.

The species of Oreosaurus from the Sierra Nevada de Santa Marta plays a key role in understanding historical biogeographic patterns in the Neotropics (Chapter 5).

However, it remains undescribed. In this chapter, I name and describe this new species.

This chapter was submitted to Zookeys and is currently under review (Sánchez-

Pacheco, S.J., Nunes, P.M.S., Rodrigues, M.T., Murphy, R.W. In review. Formal recognition of the species of Oreosaurus (Squamata: Gymnophthalmidae) from the

Sierra Nevada de Santa Marta, Colombia. Zookeys).

1.3.6 Chapter 7. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards.

Outgroup sampling (OGS) is a fundamental step in phylogenetic analysis. Despite the pivotal role that OGS plays in testing hypotheses of ingroup topologies and homologies, a logically consistent analytical framework that guides sampling is lacking. In this chapter, I develop a theory of OGS grounded in the logic of scientific discovery. This framework provides a logical basis for sampling and successively increases severity of hypothesis-testing, but it does not provide any grounds for limiting the sample. I then propose a heuristic procedure that provides an empirical basis to limit sampling. I illustrate this procedure using a novel multi-locus DNA sequence dataset for alopoglossid lizards. Finally, I use the optimal phylogenetic hypotheses for alopoglossid lizards to propose a genus-level monophyletic taxonomy.

This chapter will be submitted to Systematic Biology (Sánchez-Pacheco, S.J., Grant,

T., Murphy, R.W. In prep. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards).

A by-product of this chapter was published (but not included herein) in Salamandra

(Sánchez-Pacheco, S.J., Rueda-Almonacid, J.V., Caicedo-Portilla, J.R., Souza, S.M.

2016. First record of Leposoma caparensis from Colombia, with confirmation for the presence of Ptychoglossus myersi and P. plicatus (Squamata: Gymnophthalmidae,

Alopoglossidae). Salamandra 52, 53–57). In this short contribution, I propose that unusual phenotypic variation in two syntopic, closely related species of alopoglossid lizards is consistent with the character-displacement hypothesis.

Nomenclatural Disclaimer. The taxonomic changes presented in this chapter, including new taxa, combinations, and synonymy, are disclaimed as nomenclatural acts and are not available, in accordance with Article 8.3 of the International Code of

Zoological Nomenclature.

Chapter 2 ON THE USE OF MEDIAN-JOINING NETWORKS IN EVOLUTIONARY BIOLOGY

A modified version of this chapter was published in Cladistics (Kong, S*, Sánchez-

Pacheco, S.J.* and R.W. Murphy. 2016. On the use of Median-Joining Networks in evolutionary biology. Cladistics. 32(6): 691-699).

*Equal contribution

2 Abstract

Median-Joining (MJ) was proposed as a method for phylogeographic analysis and it enjoys increasing popularity. Herein, I evaluate the efficacy of the approach as originally intended. I show that median-joining networks (MJNs) are theoretically untenable for evolutionary inference, and that confusion has afflicted their use for over

15 years. The approach has two obvious shortcomings: its reliance on distance-based phenetics (overall similarity instead of character transformations), and the lack of rooting (no direction or history). Given that evolution involves both change and time, and the absence of rooting removes time (ancestor–descendent relationships) from the equation, the approach cannot yield defensible evolutionary interpretations. I also examine the impact of MJ analyses on evolutionary biology via an analysis of citations and conclude that the spread of MJNs through the literature is difficult to explain, especially given the availability of character-based analyses.

The rising preeminence of Phylogenetic Systematics runs the risk of being self defeating, for it is

becoming more and more common for practitioners of other approaches to pay lip-service to

phylogenetic principles…This tendency seems to be most pronounced when the alternative

approaches are of a mathematical nature or are implemented by computer programs, and the

practice hinders continued development of truly phylogenetic methods.

[Farris et al., 1982, p. 317]

2.1 Introduction

Technological advances in computing and the flood of molecular data have catapulted research in evolutionary biology in general, and in phylogenetics, in particular, by testing hypotheses through user-friendly software. However, these advances have also prompted the proliferation of sophisticated-looking analyses without any consideration of the philosophy behind the methods (Grant et al., 2003). In this context, the median-joining (MJ) approach has been implemented for over 15 years.

The application of median-joining networks (MJNs) to evolutionary studies has dramatically grown, and there is no indication that this trend will wane in the foreseeable future (Fig. 2.1). As is necessary for many widely accepted approaches and concepts, such as the biological species concept, it is important to reevaluate the assumptions and limitations of MJ. Herein, I review its theoretical foundations, current applications, and the associated terminology. I discover that its implementation has been plagued by confusion since its conception. The approach overlooks basic principles of both evolutionary biology and phylogenetic analysis, and even the underlying prerequisites of MJ itself.

Bandelt et al. (1999) (hereinafter referred to as BEA99) introduced MJ as a method for inferring intraspecific phylogenies, stating that “[r]econstructing phylogenies from intraspecific data…is often a challenging task because of large sample sizes and small genetic distances between individuals” (emphasis added). They argued further that

“[t]he resulting multitude of plausible trees is best expressed by a network which displays alternative potential evolutionary paths in the form of cycles” (p. 37 [abstract], emphasis added). MJ constructs such networks. Although the non-evolutionary essence of MJ analysis was summarized exceptionally well in the two consecutive, introductory

sentences, MJNs continue to persist and flourish in the literature. Therefore, it is desirable to address the most conspicuous drawbacks of this method: its distance-based approach and the absence of rooting.

2.2 Theoretical overview

Phylogenetic networks: rooting matters

A phylogenetic tree is a directed (rooted) branching diagram that represents the hypothesized relationships among the organisms under study. Unlike undirected

(unrooted) branching diagrams (i.e. networks; but see below), historical statements must be made based on trees (Wheeler et al., 2006) because outgroup comparison roots the ingroup topology and polarizes character transformations (Farris, 1972, 1982), thereby converting a non-evolutionary network into an evolutionary hypothesis.

Although both trees and networks are cladograms (Farris, 1970)—diagrams with a branching pattern—that depict hypothesized relationships, the evolutionary history of the evidence (e.g. DNA sequences) and the organisms can only be inferred by ordering terminals, explaining characters and testing hypotheses on trees. However, because of the bifurcating pattern of phylogenetic trees (two, and only two, descendant branches arise from a single ancestral branch), reticulation events derived from non-vertical inheritance processes such as hybridization, recombination, and horizontal gene transfer cannot be visualized. Alternatively, networks are used frequently to represent such events.

The concept of a phylogenetic network has been used indiscriminately in the literature. In mathematics and computer science (specifically, in graph theory), a

“network” is an undirected cyclic graph (UCG), which is nothing more than an unrooted branching diagram with reticulation. This differs from a “tree”, which is a directed acyclic graph (DAG), or a rooted branching diagram without reticulation (Wheeler et al., 2006; Wheeler, 2012). In UCGs, a cycle is formed by a path over edges from a vertex back to itself where each intermediate edge between vertices is visited once. In contrast, cycles cannot be formed in DAGs because there is only one unique path between two different vertices, that is, the edges between vertices can be traversed in one direction only. For vertex v in graph G, the degree is the number of edges in G that contains v. In DAGs, in-degree and out-degree edges can be specified, and their sum is the degree. For instance, in Fig. 2.2 the degree of vertex v4 is 3, its in-degree and out- degree being 1 and 2, respectively. Note that trees are connected graphs because there are no vertices with degree 0 (all vertices are visited by a path over edges). They are composed of three types of vertices: the root (in-degree 0 and out-degree 2), the internal vertices (in-degree 1 and out-degree 2) and the leaves or terminals (in-degree 1 and out- degree 0) (Chung, 1986; Moret et al., 2004; Wheeler, 2012, 2014).

In a phylogenetic context, trees are basically a series of ancestor–descendant statements, as well as representations of sister-group relationships. Accordingly, the nodes (= vertices) signify both sister groups and ancestral conditions, and the branches

(= edges) that connect them contain the character transformations between ancestors and descendants (Nelson, cited in Eldredge and Cracraft, 1980; Wheeler, 2012).

Considering that “phylogenetics” refers to the evolutionary history of sets of organisms, and that direction through rooting is imperative to allow evolutionary inference, a

“phylogenetic network” is then defined as a DAG with at least one node with in-degree

2 and out-degree 1 (Moret et al., 2004; Wheeler, 2012; Fig. 2.2: v5), which denotes a reticulation event. In other words, a phylogenetic network is a tree with directed

reticulate branches (Wheeler, 2014). By comparison, a MJN is an UCG1 (Fig. 2.3;

“cyclic evolution” occurs) or an unrooted 2 branching diagram with reticulation (= network only), and is, therefore, not a phylogenetic network as introduced by BEA99 and as often understood. Considering these criteria, the statistical parsimony network

(Templeton et al., 1992), among other UCGs, is also a non-evolutionary network.

Conversely, hybridization (Maddison, 1997; Linder and Rieseberg, 2004), recombination (Griffiths and Marjoram, 1996; Song and Hein, 2005), and duplication- loss-transfer (DLT) (Delwiche and Palmer, 1996; Planet et al., 2003) networks can be considered as phylogenetic, evolutionary networks. Phyletic group-types for phylogenetic networks, in addition to the Hennigian mono-, para- and polyphyletic, are defined by Wheeler (2014).

The (distance-based) MJ method

Operational details of the MJN algorithm are available (BEA99; Huson et al., 2010).

Here, I focus on its phenetic nature. Unlike its preceding reduced median network

(Bandelt et al., 1995), MJN can handle large datasets, as well as multistate data, such as amino acid sequences, very rapidly. In an attempt to create an intermediate-sized network, MJ combines the minimum spanning network and quasi-median network

1Sometimes MJ analysis results in an undirected acyclic graph (i.e. an undirected, simple graph; e.g. Gangloff et al., 2013). 2Although the program NETWORK (fluxus-engineering.com; BEA99) offers the option to “root the [MJ] network” to determine “the ancestral node” by “comparing the network nodes with suitable outgroups” (p. 29, user guide, fluxus-engineering.com), which has been implemented in some studies (e.g. Sakaguchi et al., 2012), this procedure merely links the “outgroup” sequence (i.e. non-conspecific) to the most similar haplotype of the already produced “ingroup” network. Hence, it neither roots the ingroup topology nor polarizes character transformations. The direct addition of an outgroup sequence into the network construction process (i.e. not through the rooting option) is likely to yield unresolved and extremely confusing networks with complex 3-D cycles and multiple median vectors (unpublished analyses).

algorithms (the former produces too few nodes associated with a multiple alignment of sequences, and the latter produces too many; Huson et al., 2010). BEA99 (p. 37) developed the following reasoning:

The MJ method begins with the minimum spanning trees, all combined within a single (reticulate)

network. Aiming at parsimony, we subsequently add a few consensus sequences (i.e., median

vectors, or Steiner points) of three mutually close sequences at a time. These median vectors can

be biologically interpreted as possibly extant unsampled sequences or extinct ancestral sequences.

The median operation, also referred to as ‘‘Steinerization’’ in mathematics (in which the most

parsimonious realizations of MP trees are called Steiner trees; see Hwang, Richards, and Winter

1992), is basic to all fast MP heuristic algorithms, although it is typically applied in a very

restricted (‘‘greedy’’) manner in order to arrive at a single tree (Farris 1970). In contrast, the

unconstrained use of the median operation eventually generates the so-called full quasimedian

network…, which normally harbors all optimal trees, as well as numerous suboptimal

trees…With MJ, we take care that at each stage only those median vectors which have a good

chance of appearing as branching nodes in an MP tree are generated by considering only triplets

of sequences for which one sequence is linked to the other two in the network under processing.

An additional ranking of these candidate triplets according to a distance score (as proposed by

Tateno 1990) allows further refinement of the triplet selection. After each round of median

generation, the process restarts with the thus enlarged set of sequences.

Minimum spanning network. The MJ method starts with the generation of a minimum spanning network (BEA99). For a given set, S, of DNA sequences (s1, s2, s3…, si), the algorithm requires a multiple sequence alignment, A, with infrequent ambiguous states and with no recombination. The data are the basis of a distance matrix, D, on S. To calculate the distance, d, between two sequences in A, the algorithm employs the

Hamming distance, H, which is the number of differences between equal-length sequences (or to define D, then H(sx, sy) in A). Distance values between sequences are

then increasingly ordered (d1 < d2 < d3 <…< di). If character state differences between sequences (not “character changes” as claimed by BEA99, p. 39) are unequally weighted, then d(sx, sy) in A to define D is given by the sum of weights, ω, of all different states between sx and sy (Fig. 2.4a), or

i-th d(sx, sy) = Sω(sx, sy) ,

where i-th denotes any position at which sx and sy differ. This “weighted Hamming distance” of MJ (BEA99, p. 39) is better termed an “unequally weighted Hamming distance”. It would have been more precise because in the Hamming distance, differences are equally weighted. Ambiguous states are specified via a comparison

“with the definite states of the other minimally distant sequences”, and arbitrarily assigned by setting “the most common definite state of these sequences” (BEA99, p.

39).

As for the minimum spanning network itself, consider the graph G = (V, E), where

V represents the set of vertices (nodes), v, therefore V = S in this case, and E is the set of edges (branches), e, containing all possible edges between any two nodes in V. In a spanning tree of G, all vertices are connected, and, as with trees in general, cycles cannot be formed. Note that the usage of “tree” in this context (and in mathematics) refers to a connected—not rooted—graph. If the edges have weights (ω(e)) that reflect

D, that is, ω(e) = d(sx, sy) for every e in E, then ω(G) is given by the sum of the weights of all edges in G associated with A (G = (V, E, ω); i.e. G is a “weighted graph” [Wheeler

(2012), or a “distance graph” sensu Huson et al. (2010)]. In this case, the minimum spanning tree for G is the spanning tree (T) that connects all vertices of G and that minimizes the sum of ω(e) given by

ω(e) ω(T) = Se in T

(Huson et al., 2010; Fig. 2.4, e.g. b). In other words, it is the spanning tree whose weight is minimal (Wheeler, 2012).

Kruskal’s (1956) and Prim’s (1957) algorithms can construct minimum spanning trees. MJ analysis is based on the former. It can obtain different minimum spanning trees because it processes all edges that have the same weight consecutively in some arbitrary order. This procedure serves as an implicit “tie-breaking” rule. Thus, different input orders can produce different, but equally optimal solutions (Fig. 2.4b–d; Kruskal,

1956; Huson et al., 2010, p. 229). The minimum spanning network, N, for A is the subgraph of the weighted graph G whose E is given by the union of the E of all minimum spanning trees generated (Huson et al., 2010; Fig. 2.4e). Simply stated, N is the union of all minimum spanning trees by dropping the tie-breaking rule of Kruskal’s algorithm (BEA99). Cycles may be created initially because equally optimal solutions can be included and displayed simultaneously (Fig. 2.4e: triangle).

The tolerance parameter D (ε in BEA99) can be specified to restrict the distance values of accepted weighted edges into N. Naturally, parameter D in N is, by definition,

= 0. If, instead, D is increased (i.e. > 0), the distance criterion will be relaxed and, thus,

N will also contain all of those heavier edges of G whose weights do not surpass the heaviest weight in N by more than D. The resulting graph is known as the relaxed minimum spanning network (ND) (Huson et al., 2010). In practice, however, parameter

D is usually set = 0. The obtained minimum spanning network is the starting point for the last process to build the final MJN, the addition of median vectors.

The MJ algorithm and the resulting MJN. Based on the quasi-median operation (Huson et al., 2010), but guided by the minimum spanning network to avoid the full quasimedian resolution, the MJ algorithm produces “additional…sequence types” (BEA99) from existing ones. Specifically, it generates consensus sequences. The addition of extra vertices and associated weighted edges to the minimum spanning trees may further reduce the overall weight of the graph (Wheeler, 2012), a process known as

Steinerization. User-specified D affects the construction of the minimum spanning network, and it governs the generation of median vectors (i.e. the additional sequences).

Increasing D widens the search for potential new median vectors. However, because this parameter is usually set to zero, only minimal cost connections (i.e. vertices connected by minimal weighted edges) are considered. Basically, triplets of sequence types, where there are at least two feasible edges among them, are used to generate a median vector (or Steiner point; Fig. 2.3: small solid circles). These median vectors are added to the original pool of sequence-types, and the minimum spanning network is then recalculated with the newly enlarged set of sequences. Iterations occur until no further median vectors can be generated. Consequently, cycles are formed. Either the original cycles may be modified or new ones may be produced (Fig. 2.3: square). The final product (the MJN) shows all feasible links in minimal cost connections plus D at most (Fig. 2.3).

Discussion and implications. Although BEA99 devoted most of their attention to median generation, the construction of minimum spanning networks is central to the

MJ method. Therefore, the construction of MJNs is based entirely on a measure of similarity of DNA sequences. The reliance on a distance-based technique was described operationally and mentioned recurrently throughout their paper, but BEA99 neither

addressed this topic explicitly, nor discussed its implications. This suggests that they favored similarity-based, phenetic methods, which are unquestionably the most criticized approaches by (most) systematists. Alternatively, they may have failed to distinguish similarity from character transformation as the basis for delimiting groups.

This oversight has led some phylogeneticists to mistakenly use phenetic approaches to infer phylogenetic relationships (Grant and Kluge, 2004). Subtle, but significant, confusion supports this possibility (BEA99, p. 39, italics added):

The simplest way to obtain a distance measure between two sequences is to count the number of

character differences (the ‘‘Hamming distance’’). As a refinement, we may also weight character

changes, albeit only in a symmetrical fashion…

In one way or another, all phylogenetic methods (maximum parsimony [MP], maximum likelihood and Bayesian inference) aim to minimize character transformations, but assumptions about character evolution employed in MJ analysis rely on similarity alone. Likewise, due to the spread of this method, MJ practitioners have been led to the assumption that a cluster in the final network is a group of closely related subjects, but these relationships are based on overall similarity of sequences, which is the operational basis of phenetics. Indeed, even if MJ method employed outgroup rooting, it would behave like phenetic clustering. Unfortunately, this type of method is used frequently.

Median generation, as interpreted by BEA99, has its own problems. In an attempt to highlight the benefits of their method, they appealed to parsimony as a principle of MJ analysis (p. 37 [abstract]):

We present a method…for constructing networks…that combines features of Kruskal’s algorithm

for finding minimum spanning trees by favoring short connections, and Farris’s maximum-

parsimony (MP) heuristic algorithm, which sequentially adds new vertices called ‘‘median

vectors’’, except that our MJ method does not resolve ties.

BEA99 continued (p. 37):

Aiming at parsimony…[t]he median operation,…(in which the most parsimonious realizations of

MP trees are called Steiner trees),…is basic to all fast MP heuristic algorithms, although it is

typically applied in a very restricted (‘‘greedy’’) manner in order to arrive at a single tree (Farris

1970)…With MJ, we take care that at each stage only those median vectors which have a good

chance of appearing as branching nodes in an MP tree are generated by considering only triplets

of sequences for which one sequence is linked to the other two in the network under processing.

Likewise, they justified the addition of median vectors as follows (p. 38):

[T]he minimum spanning network is of little direct use for representing genetic data, since in

general a minimum spanning tree is far from being most parsimonious. It serves, however, as a

good point of departure in each recursive step of our MJ network construction for generating

additional inferred sequence types which reduce tree length.

The interpretation of MP by BEA99 differed from that of Farris (1970). Invoking principles of parsimony does not validate a phenetic technique as being a phylogenetic method, and the best Steiner trees are not the most parsimonious trees. Again, “tree” in the MJ context refers to a connected—not rooted—graph, or network. The generation of median vectors and the subsequent addition of extra vertices to the minimum spanning tree(s) may reduce the overall weight of the graph(s), which is then referred to as a Steiner tree(s). Consequently, the overall weight of the minimum spanning networks and MJNs may also be reduced. This Steinerization process in MJ analysis

serves to obtain MP, and the resulting best (unrooted) Steiner tree is what BEA99 assumed to be the most parsimonious tree. Although reducing the weight of graphs could constitute a form of Occam’s Razor, this form of MP does not empower MJ.

A Prim Network is an unrooted, connected graph in which the set of nodes (i.e. vertices) is identical to that of observed taxa (operational taxonomic units—OTUs), in this case DNA sequences. Thus, no median nodes (vectors) or hypothetical taxonomic units (HTUs) are constructed (Farris, 1970). This immediately leads to what is usually referred to as a Steiner-type problem in systematics: extant taxa cannot be ancestors of other extant taxa (Wheeler, 2012). Wagner or Steiner networks (also undirected, connected graphs), in turn, allow for the addition of extra vertices (i.e. HTUs or Steiner points) and associated edges. However, the Steiner problem expands on them because

OTUs and HTUs are placed indiscriminately on the network in some order that is determined by a given cost function and the absence of direction. In sharp contrast, a

(rooted) Steiner tree is a minimum cost tree with a set of terminal and internal vertices

(or Steiner points), and is, therefore, a Wagner tree (Farris, 1970). In Wagner trees,

OTUs are confined to terminal nodes (tips or leaves) and HTUs are placed at inner nodes. Thus, Wagner or Steiner trees overcome the Steiner problem (Platnick, 1977).

Under this scenario, all “trees” in MJ are actually Prim or Steiner networks. Thus, minimum spanning networks are best defined as “minimum spanning super-networks”

(Fig. 2.4e). Although BEA99’s interpretation of MP goes beyond the theoretical underpinnings of network- and tree-building differences and implications, they demark a relevant starting point for understanding the essential distinction between median construction of the MJ algorithm and that of Farris (1970): character optimization.

The Wagner Method (Farris, 1970; additive characters optimization) is a modified version of the original Wagner procedure (Wagner, 1961). It builds a single branching

diagram by adding OTUs sequentially (one at a time) in an arbitrary order. However, the optimal position on a growing tree is determined by choosing the immediately best option. Specifically applied for additive characters, the algorithm finds one optimization, but not all, as pointed out later by Farris (1983) and Goloboff (1993).

More parsimonious solutions are found by optimizing novel character combinations independently to inner nodes, creating HTUs. The resulting tree is commonly used as a starting point for several other heuristics (e.g. random addition sequences—RAS) to find more parsimonious solutions (Wheeler et al., 2006; Wheeler, 2012). Different approaches have also been proposed to find all possible optimizations of characters (e.g.

Goloboff, 1993), but their description is beyond the scope of this paper. The simple

Wagner algorithm (Farris, 1970), which BEA99 made reference to when claiming MP for MJ analysis, generates medians (HTUs) based on the characters states of the OTUs and other HTUs. Initially, the addition and placement of OTUs is determined by the advancement index, which establishes a rank order, and the interval distance formula, a relation on the character-states between HTUs and OTUs. New HTUs to connect

OTUs to the branching diagram are formed trough the median-state property, which specifies optimal HTUs one character-state at a time (Farris, 1970). The Wagner method uses patristic distances (character-state transformations, number of steps, tree- length) of a phylogenetic hypothesis (cladogram) to explain the observed character variation (Kluge and Grant, 2006; also see Farris, 1967), and not similarity or phenetic differences. Whereas the Wagner algorithm aims to minimize character transformations

(i.e. a character-based method), the MJ algorithm is governed completely by similarity.

Thus, Farris’ algorithm and the MJ method are similar only in their sequential addition of new vertices to a diagram under construction (i.e. a graphic procedure from opposite approaches).

BEA99 stressed that median vectors can be interpreted biologically as existing unsampled or extinct ancestral sequences (i.e. they can represent missing intermediates;

Fig. 2.3). However, a median vector in a MJ analysis is a majority-consensus generated sequence and a mathematically drawn point in the final MJN that connects a triplet of sequences. The resulting “evolutionary paths in the form of cycles” (BEA99, p. 37) merely illustrates the failure of the algorithm to choose between alternative, equally optimal connections due to the modification of Kruskal’s algorithm. Consequently, a cycle represents an analytical artifact rather than an evolutionary scenario (Salzburger et al., 2011).

Finally, BEA99 introduced a new meaning for homoplasy (p. 37): “[t]he MJ method…can be adjusted to the level of homoplasy by setting a [tolerance] parameter

ε [D]”, because (p. 39),

In practice, the quasimedian network generated by the given data may be somewhat large due to

homoplasy, such that only a portion should be heuristically constructed by carefully selecting

triplets of sequences for median generation.

By homoplasy, BEA99 meant the multitude of nodes generated by the full quasimedian resolution, which can be constrained by implementing the MJ operation. The MJ analysis, like any other phenetic (not to mention undirected) method, does not recognize either homoplasy or homology, and it does not distinguish between them.

2.3 MJNs in the literature

To evaluate the impact of the MJ analysis on evolutionary biology, I conducted a

meta-analysis of studies that employed the MJ method to infer evolutionary relationships among organisms. Over the last 15 years, more than 3000 papers have cited BEA99 (ISI Web of Science; absolute number of citations [December, 2014]).

The number of citations increased exponentially from 118 (less than 5%) between 1999 and 2003 to over 2023 (more than 55%) between 2010 and 2014 (Fig. 2.1). I further examined an initial subset of 376 randomly chosen articles between 2006 (first year with more than 100 citations) and 2014 and without consideration of taxonomic group.

Theoretical papers that did not employ the method were filtered out. Recent applications of MJNs included population structure analysis (e.g. Escobar-Gutierrez et al., 2013), visualization of haplotype diversity (e.g. McCracken et al., 2013) and relationships (e.g. Castelin et al., 2012), complementary information of phylogenetic analyses (i.e. when phylogenetic trees could not provide “desired resolution”; e.g.

Klütsch et al., 2012), and phylogenetic inference. I further evaluated the latter application, although most discoveries also apply equally to the former three usages.

Among the 376 papers, 161 studies (42.8%) used MJNs primarily to infer

“phylogenetic” (e.g. Zhu et al., 2013), “evolutionary” (e.g. Turchetto et al., 2013) or

“genealogical” (e.g. Amaral et al., 2014) relationships, as stated in the Materials and

Methods section of the articles. Although MJNs were mostly used to infer phylogenies based on intraspecific haplotypes (e.g. Fan et al., 2013), the MJ method was also used to infer interspecific relationships (e.g. Zou et al., 2013). Additional inappropriate use of terms was common in the literature.

Authors often misinterpreted similar haplotypes grouped through MJN as “clades”

(e.g. Cunha et al., 2012), and the MJ product as “cladograms” (e.g. Cao et al., 2013).

Clades are monophyletic groups, that is, an ancestor and all its descendants, which is represented on a rooted cladogram by all terminals arising from a single node (Wheeler

et al., 2006). A grouping produced by MJ is simply a crowd of haplotypes that are similar to each other (Fig. 2.3). Hence, it is more appropriate to refer them as “clusters” or “haplogroups”. Similarly, MJNs should be termed “unrooted networks” or

“phenograms” only.

Confusion between MJNs and MP trees exist. For example, Malyarchuk et al. (2014) reported that optimal, shortest phylogenetic trees were reconstructed using the MP calculation in optional post-processing implemented in NETWORK (Polzin and

Daneshmand, 2003). Unfortunately, this is untrue. MP calculations of NETWORK do not produce MP trees, but rather they identify median vectors and edges produced in a full MJN that are not contained in the shortest graphs, and switches them off in the final display (Fig. 2.3). Sometimes authors included MJ analyses along with traditional phylogenetic tree-reconstruction methods. For example, Bataille et al. (2013, p. 4199) inferred phylogenetic relationships using “maximum parsimony, maximum-likelihood,

Bayesian inference, and Median-Joining Network methods”.

One of the prerequisites for MJ analysis is recombination-free input data (BEA99).

MJNs are not intended to detect recombination events and, indeed, such would be impossible. Notwithstanding, MJ was used to “infer recombination amongst haplotypes”

(Arnott et al., 2013, p. 4).

Comparative studies including MJNs are over-interpreted in many cases to justify the use of the method. Cassens et al. (2003; 2005) and Wooley et al. (2008) examined the relative performance of MJNs compared with several other approaches. These have served as reference studies. Cassens et al. (2003) stated that the MJ method yielded the

“best genealogy” because it required the least number of mutations to explain the data when compared with other unrooted networks. Further, Cassens et al. (2005) suggested the MJ approach worked well when haplotypes were relatively distant, yet it

occasionally failed to reconstruct the “correct topology”. In contrast, Wooley et al.

(2008) reported that MJNs performed well when the substitution rate was low, but performed significantly less accurately when the substitution rate was high. Some authors exaggerated the findings from Cassens et al. (2003) by stating that MJNs have been shown to yield “the best-resolved genealogies relative to other rooting and network procedures” (Lin et al., 2012, p. e36334; Zhou et al., 2011, p. 331). Pauperio et al. (2012, p. 6019) also justified use of MJNs by referring to Cassens et al. (2005) and Wooley et al. (2008) and stating “in the analysis of closely related sequences, networks are useful tools as they can provide more information than a strict consensus tree and still present a reliable estimate of the true genealogy” (emphasis added).

2.4 Conclusions

Other than fast computation and very attractive graphics, MJNs harbor no virtue for phylogenetic inference. MJNs are distance-based, unrooted branching diagrams with cycles that say nothing about the evolutionary history due to the absence of direction.

MJ was introduced in 1999 and, in contrast to most scientific ideas, its application has spread rapidly through copying the methods of others, and, unfortunately, with little further scrutiny. I hope that the theoretical arguments presented here can reverse this trend.

2.5 Acknowledgements

Funding for SJSP was provided by a COLCIENCIAS doctoral fellowship (Becas

Francisco José de Caldas), and an Ontario Graduate Scholarship (OGS) at the

University of Toronto. NSERC Discovery Grant 3148 supported the research.

2.6 Figures

600

500

400

300

200 Number of citations of Number

100

0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year

Figure 2.1. Absolute number of citations of Bandelt et al. (1999) between 1999 and

(December) 2014 (ISI Web of Science).

Figure 2.2. A phylogenetic network. Note the absence of cycles due to the direction of the edges (arrows). A reticulation event is represented by a node with in-degree 2 and out-degree 1 (v5).

Figure 2.3. A Median Joining Network. Note the presence of cycles due to the absence of direction (square). Each circle represents a unique haplotype where the diameter is proportional to the number of DNA sequences represented. Integers on each edge denote the position of nucleotides within the sequence that differ between haplotypes. Small solid circles indicate median vectors.

Modified from Huson et al. (2010).

Chapter 3 CLADOGRAMS DO NOT NECESSARILY ENTAIL SYNAPOMORPHIES, BUT SYNAPOMORPHIES FALSIFY CLADOGRAMS

A modified version of this chapter was submitted to Cladistics and is currently under review (Sánchez- Pacheco, S.J., Grant, T. and R.W. Murphy. In review. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms.

Cladistics).

Brower’s (2016) review of the origin, meaning and use of the term “cladogram” is a welcomed opportunity to discuss a concept that, similarly to that of a “phylogenetic network” (discussed by Kong et al., 2016; and citations therein), has been used indiscriminately in the literature. I agree, as would most cladists, with two of Brower’s general points. First, “cladogram” has been defined, in general, “as a graphical representation of an empirical hypothesis of relationships among taxa”. Whether this representation of a hypothesis of relationships is “based on evidence from synapomorphies alone” (p. 573; abstract) or not will be discussed below. And second, one would certainly expect that after 50 years of post-Hennigian debate, clear and unambiguous terminology would have been reached in the current literature on phylogenetic inference. Yet this is not the case. As correctly pointed out by Brower, many evolutionary biologists use the terms “phylogeny”, “tree” (either evolutionary and phylogenetic trees or tree graphs), “cladogram” and, I would add, non-phylogenetic

“networks” (see Kong et al., 2016) interchangeably, thereby confusing their meaning.

Similarly, other authors (e.g. Wägele, 2005; Parenti and Ebach, 2009) conflate

“cladogram” with “phenogram”, “phylogram” and “dendrogram” 3 without any consideration of the theory behind the concepts. Therefore, Brower called for retaining the “historical meaning” of the word “cladogram” (p. 575). But what is that historical meaning of “cladogram”? Unfortunately, in his timely attempt to clarify this term,

Brower may be perpetuating several inaccuracies and inconsistencies, which I address below.

Brower’s (2016) main premise is quite clear: a cladogram implies relative recency of common ancestry, as evident from the presence of, or based on evidence from,

3 This brief rebuttal is not meant to review and clarify these particular concepts, except for “cladogram”. Instead, readers are referred to the relevant literature. For instance, Wheeler (2012) offers an excellent compilation of clear definitions (e.g. graph, tree, network), and Brower (2016) himself provides some succinct meanings (e.g. phenogram).

synapomorphies alone (p. 573–574). In other words, synapomorphies are a requisite or an element of cladogram (hereinafter “synapomorphy Î cladogram”). Regrettably, this is demonstrably untrue. From the logic of cladistic parsimony analysis, here I discuss the legitimacy of Brower’s claim, further explore the historical concept of “cladogram”, and clarify the only logically consistent relationship between cladogram and synapomorphy: cladograms do not necessarily require synapomorphies, but synapomorphies are required to test and ultimately falsify cladograms. And both cladograms and synapomorphies are required to achieve phylogenetic explanation.

3 3.1 Cladogram, cladograms without synapomorphies and the illegitimacy of “synapomorphy Î cladogram”

The historical concept of cladogram

Brower’s (2016) review included the origins, meanings and current uses of

“cladogram”, but overlooked relevant contributions to the development of the concept itself. The term was first coined simultaneously by Camin and Sokal (1965) and Mayr

(1965). However, although Farris (1970) did not use the term cladogram as such— branching form (dendrogram) instead, the concepts of “cladistic” and “cladogram” were first given a cladistic treatment (i.e. a methodology based on character transformation events instead of similarity, the Wagner method) by Farris (contra

Williams and Ebach, 2008, p. 5). Nevertheless, the theoretical foundations of these concepts, as implemented in character-based methods, trace back to Wagner (1961), who first proposed the procedure upon which Farris (1970) built the Wagner method.

Likewise, Farris (1967) recognized cladistic relationships (cladogenic events, topology,

and hypotheses of monophyly) and patristic relationships (character-state transformation events and hypotheses of homology), thus defining conceptually the problem of phylogenetic explanation (Kluge and Grant, 2006).

Polytomies

After quoting and commenting on Hennig (1966, p. 194, 196) and a series of definitions of “cladogram”, Brower (2016) concluded (p. 574, italics added):

[A] cladogram is a special sort of dendrogram, depicting an empirically supported hypothesis of

branching order that implies relative recency of common ancestry, as evident from the presence

of shared, derived character states (synapomorphies), and which does not take into account degree

of similarity or difference, branch length or absolute time. So what’s the problem?

In Hennig’s sense, phylogenetic relationships are represented as nested sets of sister- groups only (Wheeler, 2012). This is included appropriately in Brower’s definition of cladogram (in terms of recency of common ancestry, as evidenced by the shared presence of derived character-states, i.e. synapomorphies). The problem is that, under a framework of refutation and corroboration of cladistic hypotheses, even in instances where it is not possible to claim the existence of any kind of hypothesized hierarchical branching order (i.e. nested sets), “there remains the completely unresolved proposition, the trichotomous cladogram [i.e. polytomy] in the potentially informative simplest case”

(Kluge, 1997, p. 86, emphasis added). Consider the same three-taxon example of A, B,

C provided by Brower (p. 573) and in detail by Kluge (1997, p. 87, 2003, p. 236). If a set of putative synapomorphies are distributed as 110, 101, and 011 (where state 0 is plesiomorphic and state 1 is apomorphic), respectively, then three equally optimal

rooted cladograms exist when conjoining the solutions with the congruent and incongruent synapomorphies: (A1,B1)C0 or (A1,C1)B0 or A0(B1,C1), where parentheses indicate relative recency of common ancestry. In each cladogram, the derived character-states of the incongruent synapomorphies are explained by independent transformation events. For example, for the cladogram (A,B)C, the taxonomic distribution of the incongruent synapomorphies is (A1,B0)C1 and (A0,B1)C1, respectively. There remains, however, the completely unresolved, rooted cladogram,

(ABC), as a competing (though less parsimonious) hypothesis. And what about a strict consensus cladogram? The strict consensus includes only clades that are unambiguously supported by the available evidence (Grant and Kluge, 2003, 2008).

However, in this example equally optimal solutions contradict all monophyletic groups, and a complete polytomy depicts the summarized knowledge of relationships (Grant et al., 2003). Consequently, complete polytomies may not be nested sets of sister-groups based on synapomorphies, but are still competing explanations of relationships. Within the logic of testability, therefore, cladograms are better defined conceptually as directed branching diagrams representing the hypothesized phylogenetic relationships among the terminal taxa under study (Farris, 1967). As evidenced in the case of polytomies, the statement “synapomorphy Î cladogram” does not hold in all instances of

“cladogram”. The capacity of a given cladogram, as a hypothesis, to explain synapomorphies is yet another relationship between cladogram and synapomorphy

(Kluge, 1999). However, it is also dependent on resolved cladograms.

3.2 The logically consistent relationship between cladogram and synapomorphy

Phylogenetic explanations (= phylogenetic hypotheses) are definitionally historical and, therefore, entail directed cladograms (Wheeler et al., 2006). Outgroup comparison roots the topology and polarizes character transformations (Farris, 1972, 1982). This action converts a non-evolutionary network, in the sense that it lacks ancestor– descendant relationships, as well as representations of sister-group relationships, into an evolutionary hypothesis, i.e. a phylogenetic tree. In phylogenetic trees, the nodes signify both sister-groups and ancestral conditions, and the branches that connect them contain the character transformations between ancestors and descendants (Nelson, cited in Eldredge and Cracraft, 1980; Wheeler, 2012). As such, the evolutionary history of the characters and the organisms can only be inferred by ordering terminals, explaining characters, and testing hypotheses on phylogenetic trees (Wheeler, 2012).

Within this phylogenetic, scientific context, testability refers to the logical relationship between cladograms (hypotheses), synapomorphies (evidence), and background knowledge [descent with modification sensu Darwin (1859)] (Kluge, 1997,

1999, 2003). Of particular interest here is the relationship between cladogram and synapomorphy. According to Kluge (1997, p. 82), falsifying cladograms through synapomorphies is “consistent with the logic of Popperian testability and its practice of refutation and corroboration (Popper, 1968; 1992)”. Therefore, it constitutes a form of hypothetico-deductive testing. Synapomorphies are evidence because only these empirical relations have the potential to refute (falsify) cladistic hypotheses

(cladograms) (Hennig, 1966; Kluge, 1997, 2003). Refutation, however, lies in incongruent synapomorphies (explained as homoplasy), because these shared character-states imply evidence for a different, competing cladogram (see the three- taxon example above and Kluge, 1997, 1999 for an expanded explanation). It is this

relationship between evidence and hypothesis that underlies a logically consistent relationship between synapomorphy and cladogram.

3.3 Acknowledgements

Funding for S.J.S-P. was provided by a COLCIENCIAS doctoral fellowship (Becas

Francisco José de Caldas), and an Ontario Graduate Scholarship (OGS) at the

University of Toronto. NSERC Discovery Grant 3148 supported the research.

Chapter 4 LIZARDS OF THE GENUS Riama (Squamata: Gymnophthalmidae): THE DIVERSITY IN SOUTHERN ECUADOR REVISITED

A modified version of this chapter was published in South American Journal of

Herpetology (Sánchez-Pacheco, S. J., V. Aguirre-Peñafiel, and O. Torres-Carvajal.

2012. Lizards of the genus Riama (Squamata: Gymnophthalmidae): the diversity in southern Ecuador revisited. South American Journal of Herpetology 7, 259–275).

4 Abstract

Following examination of recently collected and older specimens of Riama from southern Ecuador, I report morphological variation in R. vespertina and modify the species diagnosis and description accordingly; furthermore, I describe two new species, comment on additional diversity of the genus in this region and discuss some character- states, especially dorsal scale relief (specifically striated and keeled conditions). I provide an identification key to the species of Riama occurring in southern Ecuador.

4.1 Introduction

Southern Ecuador contains some of the least known gymnophthalmid species in the genus Riama. As a conclusion of his comprehensive monograph on Ecuadorian Riama

(then Proctoporus), Kizirian (1996: 148) stated “The greatest void in the knowledge of

Proctoporus distribution occurs in southern Ecuador.” One would expect that since then a myriad of Riama specimens from southern Ecuador, including the Andean provinces of Azuay, Cañar, El Oro, Loja, Zamora-Chinchipe, and the southern half of Morona-

Santiago, would have reached natural history collections; but this is not the case.

Besides the scarce material reported by Kizirian (1996), I have been able to locate only

29 additional specimens from this vast region. The secretive nature of these lizards and the lack of interest in collecting them would explain such a scenario. Despite the limited material I have gathered, a review of it is pivotal if we are to better understand the high diversity of Riama in the Ecuadorian Andes, especially to facilitate phylogenetic studies of the genus; it is important to test microevolutionary hypotheses (e.g., species identities) before pursuing macroevolutionary studies. To this end, I maximize the available evidence by analyzing traditional external morphological characters used in

Riama systematics along with molecular data. My investigation results in the description of two new species and the redescription of R. vespertina, a species described by Kizirian (1996) on the basis of a single male, followed by comments on additional Riama diversity in southern Ecuador, and a discussion of some character- states.

4.2 Material and methods

To facilitate species determination, Riama vespertina and the new species are diagnosed from each other as well as all other 27 currently recognized species of Riama from Ecuador, Colombia, Venezuela, Peru and Trinidad & Tobago. For this purpose, I examined specimens from all the species known to occur in Ecuadorian territory plus eight species from Colombia, Venezuela and Trinidad & Tobago (see Appendix S1).

Data for R. inanis, R. laudahnae, and R. rhodogaster were taken from the literature

(Doan and Schargel, 2003; Köhler and Lehr, 2004; Rivas et al., 2005). Measurements

(snout-vent length [SVL] and tail length) were taken to 0.1 mm with digital calipers.

Sex was determined by noting the presence of everted hemipenes in males and/or secondary sex characters. Characters and head-scale terminology follow Kizirian

(1996). Bilateral variation is reported as left/right.

Except for DHMECN (División de Herpetología, Museo Ecuatoriano de Ciencias

Naturales, Quito, Ecuador), EPNH (Escuela Politécnica Nacional, Colección

Herpetología, Quito, Ecuador), FHGO (Fundación Herpetológica Gustavo Orcés, Quito,

Ecuador), MHNCSJ (Museo de Historia Natural, Colegio San José, Medellin,

Colombia), MHNUC (Museo de Historia Natural, Universidad de Caldas, Manizales,

Colombia), PSO-CZ (Museo de Historia Natural de la Universidad de Nariño, Pasto,

Colombia), QCAZ (Museo de Zoología, Pontificia Universidad Católica del Ecuador,

Quito, Ecuador), and UV-C (Museo de Vertebrados, Universidad del Valle, Cali,

Colombia), institutional abbreviations are those of Sabaj Pérez (2012).

4.3 Systematics

Riama aurea, new species

Figures 5.1 and 5.2

Proctoporus oculatus (O’shaughnessy, 1879): Burt and Burt, 1931: 369 [part (for

AMNH 18310)]. Proctoporus striatus (Peters, 1862): Uzzell, 1958: 7 [part (for AMNH

18310)].

Proctoporus hypostictus Boulenger, 1902: Kizirian, 1996: 112 [part (for AMNH

18310)].

Riama sp.: Sánchez-Pacheco et al., 2011: 11 [part (for AMNH 18310)].

Holotype.—QCAZ 07886 (Figs. 1, 2), an adult male collected on december 5, 2006 by

Silvia Aldás-Alarcón at Guanazán, Provincia El Oro, Ecuador, 2789 m; 03°26’29”S;

79°29’39”W.

Paratypes.—QCAZ 09649–50, a male and a female, respectively, collected on august

22, 2009 by Silvia Aldás-Alarcón at El Panecillo, Provincia El Oro, Ecuador, 2775 m;

03°28’3”S; 79°28’59”W; EPNH 06196, a female collected on march 25, 1995 by A.

Almendariz and Pedro Chicai at Guishaguiña, Zaruma, Provincia El Oro, Ecuador.

Referred material.—AMNH 18310, a specimen (anterior portion only) collected on august 3, 1920 by H. E. Anthony at El Chiral, Provincia El Oro, Ecuador.

Etymology.—The specific epithet, to be treated as an indeclinable word, is an adjective derived from the Latin word aurum, meaning gold, and refers to the provenance of the species, El Oro Province.

Diagnosis.—Among the other five species of Riama currently known to occur in southern Ecuador, R. aurea differs from R. anatoloros in having two postparietals

(three in R. anatoloros), 19–20 longitudinal dorsal scale rows in males (22–27), 34–35 transverse dorsal scale rows (36–44) and four femoral pores per hind limb in males (7–

11). It differs from R. stigmatoral in having four femoral pores per hind limb in males

(9–11 in R. stigmatoral) and eight scales between medialmost femoral pores in males

(0–2). It can be distinguished from R. petrorum by the superciliary arrangement, 1+1,

1+2 or 2+2 (one, the anteriormost, in R. petrorum), by the second or second and fourth supraoculars contacting the ciliaries (second, third and fourth, or first, second and third in contact with ciliaries) and by the ventral coloration, cream with brown smears in scale centers gradually becoming more distinct ventrolaterally and on tail, forming nearly continuous longitudinal lines (venter unicolored olive or dark brown with cream along scale sutures); it further differs from R. petrorum in adult body size (maximum known SVL in R. aurea is about 57 mm for males and 52 mm for females, versus 72 mm and 76 mm for males and females, respectively, of R. petrorum). From R. vespertina, R. aurea differs in having four supraoculars (three in R. vespertina) and 1+1,

1+2 or 2+2 superciliaries (2+1). From R. kiziriani it differs in having four supraoculars

(three in R. kiziriani), 1+1, 1+2 or 2+2 superciliaries (2+1), four femoral pores per hind limb in males (seven) and venter cream with brown smears in scale centers gradually becoming more distinct ventrolaterally and on tail, forming nearly continuous longitudinal lines (venter dark brown to black with longitudinally arranged white stripes or spots on the scale sutures).

Riama aurea can be distinguished from the remaining Ecuadorian species, as well as Colombian, Venezuelan, Peruvian and Trinidadian congeners by the number of scales between medialmost femoral pores in males (eight in R. aurea versus six or fewer in the other species).

Description.—Riama aurea possesses the following characteristics: (1) maximum SVL in males 57.4 mm (n = 3), in females 52 mm (n = 2); (2) frontonasal distinctly shorter than frontal; (3) prefrontals absent; (4) nasoloreal suture absent [= loreal absent]; (5) supraoculars four, second in contact with ciliaries (one specimen with second and fourth in contact with ciliaries); (6) superciliary series incomplete, 1+1, 1+2 or 2+2; (7) supralabial-subocular fusion usually absent; (8) postoculars two; (9) postparietals two;

(10) supratympanic temporals three; (11) genials two pairs; (12) dorsal scales rectangular, juxtaposed, striated; (13) nuchal scales smooth; (14) longitudinal dorsal scale rows in males 19–20, in females 19–22; (15) transverse dorsal scale rows 34–35;

(16) ventral scales smooth, in 21 transverse rows; (17) lateral scale rows two or three;

(18) femoral pores per hind limb in males four, in females absent or four; (19) scales between medialmost femoral pores eight; (20) subdigital scales on toe I four or five;

(21) anterior cloacal plate scales paired; (22) dorsum brown with or without dark brown to black spots, pale dorsolateral stripe present; longitudinal row of ocelli on flanks of adult males present; venter cream, with brown smears in scale centers gradually becoming more distinct ventrolaterally and on tail, forming nearly continuous longitudinal lines.

Description of holotype.—Male (Figs. 5.1, 5.2), SVL 57.4 mm, tail (incomplete) length

46.7 mm; head scales smooth, glossy; rostral scale wider than long, higher than adjacent supralabials, in contact with frontonasal, nasals, and anteriormost supralabials posteriorly; frontonasal roughly quadrangular, widest posteriorly, much shorter than frontal, in contact with nasals laterally, and frontal posteriorly; prefrontals absent; frontal longer than wide, widest anteriorly, anterior suture slightly convex, lateral sutures slightly concave, posterior suture angular with point directed posteriorly, not in

contact with anteriormost superciliary anterolaterally on the left side, in contact with first and second supraoculars laterally, frontoparietals posteriorly; frontoparietals pentagonal, in contact with second and third supraoculars anterolaterally, parietals and interparietal posteriorly; interparietal hexagonal, longer than wide, in contact with parietals laterally, postparietals posteriorly; parietals in contact with third and fourth supraoculars anterolaterally, dorsalmost temporal scale laterally, and postparietals posteriorly, not in contact with dorsalmost postocular; postparietals two, in broad contact; supraoculars four, second and fourth in contact with ciliaries. Nasoloreal suture absent (= loreal absent), nasal roughly pentagonal; superciliaries one anteriorly, one posteriorly, separated from each other by second supraocular; anteriormost superciliary lies between nasal, frontal (right side only), first and second supraoculars, and anteriormost ciliaries, barely extending onto dorsal surface of head; palpebral disc divided into three large and several small scales, anteriormost scales unpigmented, posteriormost pigmented; frenocular roughly pentagonal, in contact with nasal anteriorly; circumorbital scales between posteriormost supraocular and frenocular five; postoculars two; temporals smooth, glossy, polygonal; supratympanic temporals three; supralabials six; infralabials five. Mental wider than long, in contact with anteriormost infralabials and postmental posteriorly; postmental roughly pentagonal, posterior suture angular, point directed posteriorly, in contact with first and second infralabials laterally; genials in two pairs, anteriormost pair roughly quadrangular, in contact with second and third infralabials; posterior pair polygonal, in contact with third and fourth infralabials on the right side, third, fourth and fifth infralabials on the left side; scale rows between genials and collar fold (along midventral line) eight; medialmost scales of posteriormost scale row distinctly enlarged, smooth; posteriormost gular row

enfolded posteriorly, concealing two small scale rows; lateral neck scales squarish or rounded, smooth.

Dorsal scales rectangular, longer than wide, juxtaposed, striated, in 34 transverse rows (one of them incomplete); longitudinal dorsal scale rows at fifth transverse ventral scale row 18, at 10th transverse ventral scale row 19, at 15th transverse ventral scale row

19; lateral scale rows at fifth transverse ventral scale row 4/5, at 10th transverse ventral scale row two, at 15th transverse ventral scale row three; lateral scales on body near insertion of forelimb small to granular; ventral scales smooth; complete transverse ventral scale rows 21; longitudinal ventral scale rows at midbody 10; anterior cloacal plate scales two; posterior cloacal plate scales five, medial scale in contact with anterior cloacal plate; scales on tail rectangular and juxtaposed; midventral subcaudals smooth, wider than adjacent scales, nearly square. Femoral pores per hind limb four; scales between medialmost femoral pores eight. In general, limb squamation as described for congeners (e.g., Kizirian, 1996).

Coloration of holotype (Fig. 5.2). — In preservative (70% ethanol), dorsal ground color brown with dark brown to black spots, and fine dark brown mottling visible microscopically; dorsolateral dark-bordered pale stripe extending posteriorly from temporal region onto body, disappearing posterior to hind limb. Well-defined ocelli laterally from neck to anterior portion of tail becoming less distinct posteriorly on tail

(tail incomplete); labial scales dark brown with cream sutures. Venter cream, with brown smears in scale centers gradually becoming more concentrated and distinct ventrolaterally and on tail, forming nearly continuous longitudinal lines.

Variation. — The paratypes consist of a male (SVL = 48.6 mm) and two females (SVLs

= 52 and 51.9 mm). Head shape, which tends to be more robust in sexually mature males, suggests that the referred specimen (anterior portion only) is presumably a male.

The paratypes and the referred specimen are similar to the holotype with the following noteworthy exceptions. Frontal divided horizontally in QCAZ 9650; second supraocular in contact with ciliaries in all paratypes and the referred specimen (second and fourth in the holotype); superciliaries 2+2 (EPNH 06196, QCAZ 09650 and AMNH

18310) or 1+2 (QCAZ 09649), 1+1 in the holotype; supralabial-subocular fusion present in QCAZ 09650. Females with pale dorsolateral stripe (not dark-bordered), without ocelli laterally, EPNH 06196 with some tiny cream spots on flanks.

Natural history. — One specimen (QCAZ 07886) was found on a branch 10 cm above the ground next to an unpaved road. To my knowledge, this is the first report of arboreality in Riama. Two specimens (QCAZ 09649 and 09650) were found under ground material near cultivated fields in an herbaceous páramo forest and dry montane scrub.

Distribution. — Riama aurea is known from four localities in northeastern El Oro

Province (Fig. 5.7). It occurs at elevations between 2775 and 2789 m.

Remarks.—Peters (1967), Kizirian (1996) and Sánchez-Pacheco et al. (2011) briefly discussed the identity of AMNH 18310 (anterior portion only) and agreed that it may represent an undescribed species, but that additional specimens were required to allow specific recognition. AMNH 18310 is conspecific with the new material from El Oro as evidenced by color pattern and cephalic squamation.

I found that sequences of 12S and 16S mtDNA are 5.5–6.8% and 3.3–4.4%, respectively, divergent (uncorrected pairwise distance) between the samples of Riama aurea and R. kiziriani, and 4.4% and 3.4–3.6% between the samples of R. aurea and R. vespertina, whereas the sequences are identical and nearly identical (0% and 0–0.2%) between the samples of R. aurea (unpubl. data derived from an ongoing phylogenetic analysis).

Riama kiziriani, new species

Figures 5.3 and 5.4

Holotype. — QCAZ 9667 (Figs. 5.3, 5.4), an adult male collected on august 20, 2009 by Elicio Tapia at San Antonio, Provincia de Azuay, Ecuador, 1900 m; 02°53’42”S;

79°24’19”W.

Paratype. — QCAZ 9607, a male collected on July 20, 2009 by Paola Mafla-Endara and Amaranta Carvajal-Campos at El Chorro de Girón, Girón, Provincia de Azuay,

Ecuador, 2546 m; 03°7’48”S; 79°9’57”W.

Etymology. — The specific epithet is a noun in the genitive case and a patronym for

David A. Kizirian, my friend and colleague, in recognition of his contribution to the knowledge of Riama diversity.

Diagnosis. — Riama kiziriani can be distinguished from its congeners, except R. vespertina, by the presence of two anteriorly and one posteriorly positioned superciliary scales. It can be distinguished from R. vespertina by the following characteristics

(condition for R. vespertina in parentheses): femoral pores per hind limb in males seven

(4–5); and venter dark brown to black with small white spots or narrow lines on longitudinal sutures (ventral coloration in males cream —yellowish in life—with faint brown spots on central portion of scales).

In addition, among the other four species of Riama currently known to occur in southern Ecuador, R. kiziriani differs from R. anatoloros in having two postparietals

(three in R. anatoloros), 20 longitudinal dorsal scale rows in males (22–27), and 32–34 transverse dorsal scale rows in males (36–42). It differs from R. stigmatoral in having three supraoculars (four in R. stigmatoral), seven femoral pores per hind limb in males

(9–11), and six scales between medialmost femoral pores in males (0–2). It can be distinguished from R. petrorum by the second supraocular contacting the ciliaries

(second, third and fourth, or first, second and third in contact with ciliaries in R. petrorum), by the number of femoral pores per hind limb in males, seven (4–5), and by the number of scales between medialmost femoral pores, six (eight); it further differs from R. petrorum in adult body size (maximum known SVL in R. kiziriani is about 61 mm for males, versus 72 mm in R. petrorum). From R. aurea it differs in having three supraoculars (four in R. aurea), seven femoral pores per hind limb in males (four) and venter dark brown to black with longitudinally arranged white stripes or spots on the scale sutures (venter cream with brown smears in scale centers gradually becoming more distinct ventrolaterally and on tail, forming nearly continuous longitudinal lines).

Description.—Riama kiziriani possesses the following characteristics: (1) maximum

SVL in males 61.0 mm (n = 2), females unknown; (2) frontonasal shorter than frontal;

(3) prefrontals absent; (4) nasoloreal suture absent [= loreal absent]; (5) supraoculars three, second in contact with ciliaries; (6) superciliary series incomplete, two anteriorly,

one posteriorly; (7) supralabial-subocular fusion absent; (8) postoculars two; (9) postparietals two; (10) supratympanic temporals three; (11) genials two pairs; (12) dorsal scales rectangular, juxtaposed, striated; (13) nuchal scales smooth; (14) longitudinal dorsal scale rows 20; (15) transverse dorsal scale rows 32–34; (16) ventral scales smooth, in 19–21 transverse rows; (17) lateral scale rows three; (18) femoral pores per hind limb seven; (19) scales between medialmost femoral pores six; (20) subdigital scales on toe I four; (21) anterior cloacal plate scales paired; (22) dorsum dark brown, distinct dorsolateral stripe present; venter dark brown with longitudinally arranged white stripes or spots on the scale sutures.

Description of holotype.—Male (Figs. 5.3, 5.4), SVL 61.0 mm, tail (regenerated) length 33.9 mm; head scales smooth, glossy; rostral scale wider than long, higher than adjacent supralabials, in contact with frontonasal, nasals, and anteriormost supralabials posteriorly; frontonasal longer than wide, widest posteriorly, shorter than frontal, in contact with nasals laterally, anteriormost superciliary and first supraocular posterolaterally, and frontal posteriorly; prefrontals absent; frontal longer than wide, widest anteriorly, anterior suture slightly convex, lateral sutures slightly concave, posterior suture angular with point directed posteriorly, not in contact with anteriormost superciliary anterolaterally, in contact with first and second supraoculars laterally, frontoparietals posteriorly; frontoparietals pentagonal, in contact with second and third supraoculars anterolaterally, parietals and interparietal posteriorly; interparietal roughly hexagonal, longer than wide, in contact with parietals laterally, postparietals posteriorly; parietals in contact with third supraocular anterolaterally, dorsalmost temporal scales and postocular laterally, and postparietals posteriorly; postparietals two, in broad contact; supraoculars three, second in contact with ciliaries. Nasoloreal suture

absent (= loreal absent), nasal roughly pentagonal; anterior superciliaries two, posterior superciliary one, separated by second supraocular; anteriormost superciliary lies between nasal, frontonasal, first supraocular, second superciliary and anteriormost ciliaries, barely extending onto dorsal surface of head; palpebral disc divided into three large and several small unpigmented scales; frenocular quadrangular, in contact with nasal anteriorly; circumorbital scales between posteriormost supraocular and frenocular five; postoculars two; temporals smooth, glossy, polygonal; supratympanic temporals three; supralabials six; infralabials five. Mental wider than long, in contact with anteriormost infralabials and postmental posteriorly; postmental roughly pentagonal, posterior suture angular, point directed posteriorly, in contact with first and second infralabials laterally; genials in two pairs, anteriormost pair roughly quadrangular, in contact with second infralabials; posterior pair polygonal, in contact with second and third infralabials; scale rows between genials and collar fold (along midventral line) eight; medialmost scales of posteriormost scale row distinctly enlarged, smooth; posteriormost gular row enfolded posteriorly, concealing small scales; lateral neck scales squarish or rounded, smooth.

Dorsal scales rectangular, longer than wide, juxtaposed, striated, in 34 transverse rows; longitudinal dorsal scale rows at fifth transverse ventral scale row 16, at 10th transverse ventral scale row 20, at 15th transverse ventral scale row 21; lateral scale rows at fifth transverse ventral scale row six, at 10th transverse ventral scale row three, at 15th transverse ventral scale row two; lateral scales on body near insertion of forelimb small to granular; ventral scales smooth; complete transverse ventral scale rows 21; longitudinal ventral scale rows at midbody 11; anterior cloacal plate scales two; posterior cloacal plate scales five, medial scale in contact with anterior cloacal plate; scales on tail rectangular and juxtaposed; midventral subcaudals smooth, wider than

adjacent scales, nearly square. femoral pores per hind limb seven; scales between medialmost femoral pores six. in general, limb squamation as described for congeners

(e.g., Kizirian, 1996).

Coloration of holotype (Fig. 5.4).—In life, dorsal ground color dark brown, dorsal surface of head with randomly arranged concentrations of light brown pigment; dorsolateral dark-bordered pale stripe present on neck, disappearing posterior to forelimb, appearing again on tail; ocelli laterally; ventral surface of head brown, center of postmental and genials with cream pigmentation; ventral aspect of neck, body and tail dark brown to black with small white spots or narrow lines on longitudinal sutures.

Variation in paratype. — The paratype (SVL 43.1 mm) is similar to the holotype with the following noteworthy exceptions: frontonasal not in contact with first supraocular posterolaterally; dorsum mottled; white longitudinal lines along ventral scale sutures more distinctive than in the holotype.

Natural history. — The holotype (QCAZ 9667) was found under a 30 × 40 cm rock next to a pasture in a dry highland forest. Paratype QCAZ 9607 was found on rocks next to a trail in a dry area covered with Marchantia (Marchantiophyta:

Marchantiaceae).

Distribution. — Riama kiziriani is known only from two localities in Azuay province, southern Ecuador, at elevations between 1900 and 2546 m (Fig. 5.7).

Remarks. — I found that sequences of 12S and 16S mtDNA are 5.5–6.8% and 3.3–

4.4%, respectively, divergent (uncorrected pairwise distance) between the samples of

Riama kiziriani and R. aurea, and 6–7.8% and 4–4.6% between the samples of R. kiziriani and R. vespertina, whereas the divergence is only 0–2.1% and 0–1.3% between the samples of R. kiziriani (unpubl. data derived from an ongoing phylogenetic analysis).

Riama vespertina (Kizirian, 1996)

Figures 5.5 and 5.6

Proctoporus vespertinus Kizirian, 1996: 142–145. Original description. Holotype, male

(AMNH 22130) from [Pampa] Chitoqúe, 6000 ft [Loja], Ecuador.

Riama vespertina (Kizirian, 1996): Doan and Castoe, 2005: 409 [first use of combination]; Reyes-Puig et al., 2008: 368 [for DHMECN 4113–14 from Reserva

Biológica Utuana, 48.3 km southeast to the type locality].

New Referred material. — QCAZ 10283, 10286, 10288, 10306–13, a series collected on February 25, 2010 by Silvia Aldás-Alarcón and Freddy Velásquez Alomoto at

Guachaurco, Loja, Ecuador, between 2824–2958 m; 04°2’33”S; 79°51’46”W.

Additional Examined Specimens. — AMNH 22130 (Holotype, Fig. 5.5) and

DHMECN 4113–14.

Etymology. — The specific epithet is a Latin adjective meaning “of the west” and alludes to the fact that Riama vespertina is the westernmost species of Riama (Kizirian,

1996).

Diagnosis. — Riama vespertina can be distinguished from its congeners, except R. kiziriani, by the presence of two anteriorly and one posteriorly positioned superciliary scales. It can be distinguished from R. kiziriani by the following characteristics

(condition for R. kiziriani in parentheses): femoral pores per hind limb in males 4–5

(seven); and venter cream—yellowish in life—with faint brown spots on central portion of scales (ventral coloration in males dark brown to black with small white spots or narrow lines on longitudinal sutures).

In addition, among the other four species of Riama currently known to occur in southern Ecuador, R. vespertina differs from R. anatoloros in having two postparietals

(three in R. anatoloros) and 4–5 femoral pores per hind limb in males (7–11). It differs from R. stigmatoral in having three supraoculars (four in R. stigmatoral), 4–5 femoral pores per hind limb in males (9–11) and 6–10 scales between medialmost femoral pores in males (0–2). It can be distinguished from R. petrorum by the second supraocular contacting the ciliaries (second, third and fourth, or first, second and third in contact with ciliaries in R. petrorum) and by the number of transverse dorsal scale rows in males, 34–35 (33) and in females, 34–36 (31–33); it further differs from R. petrorum in adult body size (maximum known SVL in R. vespertina is about 66 mm for males and 68 mm for females, versus 72 mm and 76 mm for males and females, respectively, of R. petrorum). From R. aurea, R. vespertina differs in having three supraoculars (four in R. aurea).

Description.—Riama vespertina possesses the following characteristics: (1) maximum

SVL in males 65.8 mm (n = 5), in females 68.6 mm (n = 9); (2) frontonasal usually shorter than frontal; (3) prefrontals absent; (4) nasoloreal suture absent [= loreal absent];

(5) supraoculars three (two specimens with four on the right side), second in contact with ciliaries; (6) superciliary series incomplete, two anteriorly, one posteriorly; (7) supralabial-subocular fusion absent; (8) postoculars usually two; (9) postparietals two;

(10) supratympanic temporals usually three; (11) genials two pairs; (12) dorsal scales rectangular, juxtaposed, striated; (13) anterior nuchal scales smooth, posterior nuchal scales slightly striated; (14) longitudinal dorsal scale rows in males 19–23, in females

19–26; (15) transverse dorsal scale rows in males 34–35, in females 34–36; (16) ventral scales smooth, in 19–22 transverse rows in males, 21–24 in females; (17) lateral scale rows 1–3; (18) femoral pores per hind limb in males 4–5, in females absent or four, usually absent; (19) scales between medialmost femoral pores 6–10; (20) subdigital scales on toe I 4–6; (21) anterior cloacal plate scales paired; (22) dorsum dark brown, pale dorsolateral stripe usually present in both sexes; ocelli laterally in males, absent in females; venter cream (yellowish in life) with faint brown spots on central portion of scales in males, distinctive brown markings ventrolaterally in females.

Description of holotype. — Kizirian (1996) provided a detailed description of the holotype of Riama vespertina.

Coloration in life (Fig. 5.6). — Reyes-Puig et al. (2008) provided photographs and color description of DHMECN 4113–14 in life.

Coloration in preservative (70% ethanol). — Dorsum in males dark brown (light brown in the holotype likely because of long preservation time), with fine darker brown mottling in the holotype and QCAZ 10283; indistinct pale dorsolateral stripe present on neck (also on the temporal region in the holotype and QCAZ 10313), conspicuous in QCAZ 10313 and DHMECN 4113, disappearing posterior to forelimb, appearing again above hind limbs and on anterior portion of tail in DHMECN 4113. Flanks with white spots surrounded by black blotches to form ocelli longitudinally arranged from neck to anterior portion of tail. Venter cream with faint concentrations of dark brown pigment on central portion of scales (distinctive brown markings in DHMECN 4113), becoming more concentrated on ventral surface of head. Dorsum in females dark brown; pale dorsolateral stripe present on neck, absent in QCAZ 10310. Lateral ocelli absent.

Venter cream with distinctive brown markings on central portion of scales, especially on ventral surface of head, ventrolaterally on the body, and subcaudally.

Variation. — Noteworthy variations among referred specimens include: frontonasal equal in length to frontal in QCAZ 10313; postoculars three in QCAZ 10286; supratympanic temporals two in QCAZ 10283; femoral pores absent in all but one female (DHMECN 4114, four femoral pores per hind limb). DHMECN 4114 and

QCAZ 10286 appear to have four supraoculars on the right side because the third supraocular is divided. Femoral pore number is the most evident sexually dimorphic character, with males having from 4–5 pores per hind limb and females usually lacking them.

Previous authors have mentioned that Riama vespertina has keeled or striated dorsal scales (Kizirian, 1996; Reyes-Puig et al., 2008). However, in all but one of the specimens examined by us, scales are striated, or slightly striated (DHMECN 4113).

The exception is QCAZ 10309, a juvenile female (SVL 31.7 mm) with smooth or slightly keeled dorsal scales (but see section on dorsal scale relief in Discussion below).

Natural history. — Ecological data of DHMECN 4113–14 can be found in Reyes-Puig et al. (2008). The specimens listed under “New Referred Material” above were found under rocks next to a secondary road in a high Andean forest, as well as a pine forest.

Two females (QCAZ 10310 and 10312) contained two eggs each. One specimen

(QCAZ 10313) was collected next to a communal nest with six eggs. All specimens were collected between 9am–2pm.

Distribution. — Riama vespertina has only been collected in southern Loja Province in extreme southwestern Ecuador, at elevations between 2600 m and 2958 m (Fig. 5.7).

Remarks. — It should be noted that Kizirian (1996: 142) was not very precise when reporting the type locality of Riama vespertina as “[Pampa] Chitoqúe, 6000 ft [Loja],

Ecuador”. However, in the same paper he clarifies that the type locality is “4.5 km south of Vicentino [79°55’54”W, 03°59’55”S]”, Cordillera de Celica, in extreme southwestern Ecuador (Kizirian, 1996: 144). The AMNH catalog further states that this locality lies “between San Bartolo & piñas (San Bartolo 8 mi NE of Alamor, Piñas 8 mi N of Alamor), Chitoque [Loja], Ecuador”.

Based on the difference in elevational ranges and femoral pore counts between the holotype of Riama vespertina and the specimens that Reyes-Puig et al. (2008) reported

(DHMECN 4113–14), the latter concluded that the new material may correspond to an undescribed species. However, I examined those specimens and conclude that they are indeed conspecific with R. vespertina. The variation Reyes-Puig et al. observed in

femoral pore counts is due to a methodological misunderstanding. Kizirian (1996) reported the number of femoral pores on the left leg (= per hind limb), which was defined in his characters section, whereas Reyes-Puig et al. reported the total number of femoral pores (= including both hind limbs). Thus, the 10 femoral pores reported by

Reyes-Puig et al. for the male DHMECN 4113 match Kizirian’s description of the holotype (five femoral pores on the left leg). DHMECN 4114 is the only female with femoral pores, which supports Reyes-Puig et al.’s claim; nevertheless, because in all other aspects both specimens fall well within the variation observed in R. vespertina, I consider this variation as individual. Similar intraspecific variation in presence (four) or absence of femoral pores in females has been documented for other species of Riama such as R. petrorum (Kizirian, 1996) and R. aurea (this study). I attribute the difference between the elevational ranges to an imprecise record of the holotype’s provenance.

I found that sequences of 12S and 16S mtDNA are 4.4% and 3.4–3.6%, respectively, divergent (uncorrected pairwise distance) between the samples of Riama vespertina and

R. aurea, and 6–7.8% and 4–4.6% between the samples of R. vespertina and R. kiziriani, whereas the sequences are identical (0%) between the samples of R. vespertina (unpubl. data derived from an ongoing phylogenetic analysis).

Comments on additional Riama diversity in southern Ecuador

Four specimens from Cordillera del Cóndor, Zamora-Chinchipe province, (QCAZ

9169, FHGO 2405 and 8617, and EPNH 12689) are tentatively assigned to Riama anatoloros. One of them, FHGO 2405, appears to correspond to the southern variation reported by Kizirian (1996): a complete nasoloreal suture, an incomplete superciliary series, and a high number of femoral pores per hind limb (11). The identity of the

remaining specimens is enigmatic as they have only the anteriormost superciliary scale and exhibit supralabial-subocular fusion, two features that have not previously been recorded in this species. Moreover, these specimens have a single pair of genials, a trait formerly reported in only one specimen of R. anatoloros (Kizirian, 1996) and extremely atypical in Riama. In the absence of a thorough morphological and molecular analysis involving the entire geographic range of R. anatoloros, which is beyond the scope of this paper, I refrain from recognizing these specimens as a different species.

The other two species of Riama occurring in southern Ecuador, R. petrorum and R. stigmatoral, remain rare in collections. This is especially true for R. petrorum, a species described on the basis of two type and one non-type specimens from Morona-Santiago province (Kizirian, 1996). In spite of my collection searching and the targeted field work carried out by VA and OTC, I was unable to find new material of this rare species.

Similarly, R. stigmatoral was only known from the 10 specimens used in its original description (Kizirian, 1996), but unlike R. petrorum I have reviewed an additional series of males (QCAZ 7374, 7884, 6657, 9946 and 11414) collected in Cañar and

Azuay provinces. These five new specimens do not significantly alter the diagnosis and description of the species. However, femoral pore per hind limb count of QCAZ 9946 is eight, one less than the range (9–11) reported by Kizirian (1996). Likewise, the number of transverse ventral scale rows in QCAZ 9946 (25) is one more (21–24), and the number of transverse dorsal scale rows in QCAZ 7374 (42) and QCAZ 6657 (43) is one and two more (36–41), respectively.

Riama anatoloros, R. petrorum and R. stigmatoral occur sympatrically in Cordillera

Zapote Naida, Morona-Santiago (Kizirian, 1996). No other case of sympatry between congeners in southern Ecuador was detected by us, but possible interspecific contact zones may be discovered as collections in unexplored areas increase.

4.4 Discussion

All six species of Riama in southern Ecuador have striated dorsal scales—also keeled in R. stigmatoral (see below for a detailed discussion on dorsal scale relief)— and all but one of them (R. stigmatoral) have a high number of scales (four or more) between the medialmost femoral pores in males. Furthermore, these taxa exhibit an incomplete series of superciliaries, though superciliary arrangement varies considerably among them (from one superciliary, the anteriormost, in R. petrorum and some specimens of R. anatoloros, to 2+2 in two specimens of R. aurea). Riama vespertina, R. aurea, R. kiziriani and R. petrorum share at least two additional character-states: frontonasal shorter than frontal and nasoloreal suture absent. Insofar as these states have been observed throughout Riama, their phylogenetic significance cannot yet be inferred. In contrast, the presence of two anteriorly and one posteriorly positioned superciliary scales is a rare condition in Riama that all known specimens of

R. vespertina and R. kiziriani exhibit. Kizirian (1996) and Sánchez-Pacheco (2010a) documented (exceptional) specimens of R. unicolor and R. columbiana, respectively, with the same feature (two of 132 R. unicolor and on one side in two of 16 R. columbiana).

Dorsal scale relief varies considerably among the species of Riama. Smooth scales are easily identified as well as the rugose condition—somewhat unclear when initially described and implemented by Kizirian (1996) but later clarified by Sánchez-Pacheco

(2010b). Both states are uncommon in the genus; the former occurs in four species (R. afrania, R. laevis, R. meleagris and R. simotera)4 whereas the latter in only two taxa (R. stellae and R. vieta). In contrast, as pointed out by Kizirian (1996: 91), differentiating

4 Dorsal scales on posterior part of body and on tail often with a low rounded keel in Riama meleagris and R. simotera.

between keeled and striated relief of dorsal scales can be confusing. According to

Kizirian, the striated condition “refers to scale relief characterized by two centrally positioned, parallel and longitudinal furrows” and the keeled condition “refers to a rounded, narrow, central keel in the middle of the scale. This rounded keel can be flanked by striations varying in depth such that the distinction between the keeled and striated conditions becomes unclear”. Further, in keeled species of Riama the keel is typically low, which confuses relief identification. Thus, in some species descriptions

Kizirian (1996) reported dorsal scale relief as “striated/keeled” (among them R. anatoloros, R. petrorum, R. stigmatoral and R. vespertina). Nonetheless, I believe that by treating these character-states independently, the differences become unambiguous.

Much of the confusion is attributable to perception; when striations occur, the central portion of the scale between them appears to be raised, and may be interpreted as a keel.

When a low, rounded keel is present, it is not flanked by striations. Hence, when striations are present, the scale relief condition must be described as striated, even if the central portion of the scale resembles a keel, and, when striations are absent but a keel is evident, the condition is to be described as keeled. Difficulties in recognizing both states can be minimized if the ethanol is permitted to evaporate from the scales and the angle of light is adjusted. All adult specimens of R. anatoloros, R. petrorum and R. vespertina that I have examined are actually striated. Adults of R. stigmatoral can be either striated or keeled. Dorsal scale ornamentation is well developed and conspicuous in adults, less developed in juveniles.

Artificial Key to the Southern Ecuadorian Species of Riama

The following key does not reflect phylogenetic relationships, and is largely based on adult male morphology.

1. Femoral pores per hind limb in males four or five...... 2

Femoral pores per hind limb in males seven or more...... 4

2. Transverse dorsal scale rows 31–33; superciliary series incomplete, one (anteriormost);

second, third and fourth, or first, second and third supraoculars contacting the ciliaries;

venter unicolored olive, or dark brown with cream along scale

sutures...... R. petrorum

Transverse dorsal scale rows 34–35; superciliary series incomplete, two or more; only

the second supraocular contacting ciliaries; venter cream with faint brown spots on

central portion of scales ...... 3

3. Superciliary series 2+1; supraoculars three………………………………R. vespertina

Superciliary series 1+1, 1+2 or 2+2; supraoculars four...... R. aurea

4. Frontonasal shorter than frontal; transverse dorsal scale rows 32–34; longitudinal

dorsal scale rows 20 ...... R. kiziriani

Frontonasal longer than or equal to frontal; transverse dorsal scale rows 36–44;

longitudinal dorsal scale rows 21–28 ...... 5

5. Scales between the medialmost femoral pores in males 0–2...... R. stigmatoral

Scales between the medialmost femoral pores in males four or more...... R. anatoloros

4.5 Acknowledgments

Funding for SJS-P was provided by an American Museum of Natural History

Collection Study Grant and a COLCIENCIAS doctoral fellowship. VA and OTC received funding from SENESCYT (PIC-08-0000470). Manuscript preparation was supported by a Natural Sciences and Engineering Research Council Discovery Grant

3148 to Robert W. Murphy. Robert W. Murphy, Ross D. MacCulloch, Tiffany M. Doan and David A. Kizirian offered many suggestions that greatly improved the paper. For specimen loans and access to collections I am grateful to Kevin de Queiroz and Traci

D. Hartsell (USNM), Jose rosado (MCZ), Ronald A. Nussbaum and Greg Schneider

(UMMZ), Alan Resetar (FMNH), Linda Trueb and Andrew Campbell (KU), S. O.

Kullander and E. Ahlander (NRM), Fernando Castro (UV-C), Jhon Jairo Calderón

(PSO-CZ), J. Salazar and H. F. Arias (MHNUC), A. Maldonado (IAvH), Marta L.

Calderón espinosa (ICN), Carol L. Spencer and Sarah Werning (MVZ), J. V. Vindum

(CAS-SUR), Ana Almendáriz (EPNH), Mario Yánez-Muñoz (DHMECN), Jorge

Valencia (FHGO) and Julio Mario Hoyos (MUJ). Pedro H. Bernardo and Will Shim provided the photographs of the Riama aurea holotype, and the head drawings of Riama aurea and R. kiziriani, respectively. Marco Rada’s expertise in image manipulation was helpful in making the figures.

4.6 Figures

Figure 4.1. Riama aurea, new species. Dorsal, lateral, and ventral views of head of holotype (QCAZ 07886, 57.4 mm SVL].

Figure 4.2. Riama aurea, new species. Dorsal and ventral views of holotype (QCAZ

07886, 57.4 mm SVL). Photos courtesy: Pedro H. Bernardo.

Figure 4.3. Riama kiziriani, new species. Dorsal, lateral, and ventral views of head of holotype (QCAZ 9667, 61.0 mm SVL].

Figure 4.4. Riama kiziriani, new species. Dorsal and ventral views of holotype

(QCAZ 9667, 61.0 mm SVL) in life. Photos: Omar Torres-Carvajal.

Figure 4.5. Riama vespertina. Dorsal and ventral views of holotype (AMNH 22130,

40 mm SVL). Photos courtesy: David A. Kizirian.

Figure 4.6. Riama vespertina. Dorsolateral and ventral views of the adult male in life

QCAZ 10283 (56.0 mm SVL). Photos courtesy: Silvia Aldás-Alarcón.

Figure 4.7. Distribution of Riama aurea, R. kiziriani and R. vespertina in southern

Ecuador. Arrows indicate type localities.

Chapter 5 PHYLOGENY OF Riama (Squamata: Gymnophthalmidae), IMPACT OF PHENOTYPIC EVIDENCE ON MOLECULAR DATASETS, AND THE ORIGIN OF THE SIERRA NEVADA DE SANTA MARTA ENDEMIC FAUNA

A modified version of this chapter was published in Cladistics (Sánchez-Pacheco, S.J.,

Torres-Carvajal, O., Aguirre-Peñafiel, V., Nunes, P.M.S., Verrastro, L., Rivas, G.A.,

Rodrigues, M.T., Grant, T. and R.W. Murphy. Phylogeny of Riama (Squamata:

Gymnophthalmidae), impact of the phenotypic evidence on molecular datasets, and the origin of the Sierra Nevada endemic fauna. Cladistics DOI: 10.1111/cla.12203).

5 Abstract

Riama is the most speciose genus of the Neotropical lizard family

Gymnophthalmidae. Its more than 30 montane species occur throughout the northern

Andes, the Cordillera de la Costa (CC) in Venezuela, and Trinidad. I present the most comprehensive phylogenetic analysis of Riama to date based on a total evidence (TE) approach and direct optimization of molecular and morphological evidence. Analyses use DNA sequences from four loci and 35 phenotypic characters. The dataset consists of 55 ingroup terminals representing 25 of the 30 currently recognized species of Riama plus five undescribed taxa, including an endemic species from the Sierra Nevada de

Santa Marta (SNSM) in Colombia, and 66 outgroup terminals of 47 species. Analysis results in a well-supported hypothesis in which Riama is polyphyletic, with its species falling into three clades. The Tepuian Anadia mcdiarmidi nests within one clade of

Riama, and the recently resurrected Pantodactylus nests within Cercosaura.

Accordingly, I propose a monophyletic taxonomy that reflects historical relationships.

Analysis of character evolution indicates that the presence/absence of prefrontals—a cornerstone of the early genus-level taxonomy of cercosaurines—is optimally explained as having been plesiomorphically present in the most recent common ancestor of Cercosaurinae and lost in that of the immediately less inclusive clade.

Multiple independent reversals to present and subsequent returns to absent occur within this clade. To evaluate the impact of phenotypic evidence on my results, I compare my

TE results with results obtained from analyses using only molecular data. Although phenotypic evidence comprises only 1.2% of the TE matrix, its inclusion alters both the topology and support values of the clades that do not differ. Finally, current phylogenetic evidence reveals a SNSM–CC–Trinidad–tepuis biogeographic link. I

hypothesize that an ancient connection facilitated the exchange of species between the

SNSM and the CC.

5.1 Introduction

Gymnophthalmidae Fitzinger, 1826 is a species-rich family of small to medium

Neotropical lizards. With 236 named species divided into 46 genera (Uetz and Hošek,

2017) and distributed throughout South America (with relatively few representatives in

Middle America), Gymnophthalmidae is one of the most important components of the lizard fauna in the Neotropics. Although multiple molecular-based phylogenetic analyses in the 21st Century have led to great improvement in gymnophthalmid systematics (e.g. Pellegrino et al., 2001; Castoe et al., 2004; Pyron et al., 2013; Colli, et al. 2015; Kok, 2015; Goicoechea et al., 2016), much of the current genus-level taxonomy still relies on early phenetic clusterings. This is particularly true for the

Cercosaurinae Gray, 18385, which holds for over 50% of gymnophthalmid diversity.

Despite considerable progress in unravelling the phylogenetic relationships of cercosaurines (e.g. Torres-Carvajal et al., 2016), compelling evidence for the monophyly of some genera is lacking. Many species were never included in phylogenetic analyses, and, thus, their placement in one genus or another is based mainly on overall similarity. The questionable generic placement of some species (e.g. based on geographic proximity) and the lack of rigorous tests of monophyly of genera exacerbate the problem.

This study focuses on Riama Gray, 1858 (Cercosaurinae). With 30 recognized species, it is by far the largest genus of gymnophthalmids and additional species are awaiting description (S.J.S-P. pers. obs.; this study). Although the molecular studies by

Castoe et al. (2004; see also Doan and Castoe, 2005), Aguirre-Peñafiel et al. (2014) and

Torres-Carvajal et al. (2016) advanced understanding of the phylogeny of Riama, the

5 I follow Torres-Carvajal et al.’s (2016) delimitation of Cercosaurinae, which corresponds to Cercosaurini in the sense of Goicoechea et al. (2016).

current delimitation of this genus is suspect. As recognized by Doan and Castoe (2005),

Castoe et al. (2004) sampled five of the 24 species of Riama only. Torres-Carvajal et al. (2016) added seven species and analyses nested one of them within Proctoporus

Tschudi, 1845 sensu stricto. Thus, neither the monophyly of Riama nor the relationships within the genus have been tested rigorously. Further, the trees from

Castoe et al. (2004) and subsequent authors differ from the morphological assessment of Doan (2003a), which was limited to pholidosis. Finally, all nominal and undescribed species of Riama have narrow, montane ranges with strikingly disjunct distributions.

Most species occur along the tropical Andes from 1100 to 3340 m above sea level (a.s.l.) and expansive geographic barriers (e.g. depressions) separate them from species occurring from 650 to 2165 m a.s.l. on the Cordillera de la Costa in Venezuela, the

Sierra Nevada de Santa Marta (SNSM) in Colombia, and northern Trinidad. Thus, I test the monophyly of Riama and explore the phylogenetic relationships among its species.

My phylogenetic analysis of cercosaurine lizards uses nuclear and mitochondrial DNA sequences as well as morphological data. Further, the high degree of endemism of

Riama species offers an exceptional opportunity to test historical biogeographic hypotheses involving different Neotropical montane regions. Accordingly, analyses test hypotheses on the origin of the montane SNSM endemic vertebrate fauna.

Phylogenetic analyses also explore the evolution of prefrontal scales in

Cercosaurinae—a cornerstone of the early genus-level taxonomy of this group (see below)—as well as the relationships of Cercosaura Wagler, 1830, Pantodactylus

Dumeril and Bibron, 1839 and Proctoporus.

Impact of phenotypic evidence on molecular datasets

Technological advances in DNA sequencing and molecular phylogenetics facilitate analyses based on genotypic evidence. In contrast, analyses of phenotypic evidence suffer stagnation due to (1) the ease of obtaining large molecular datasets and (2) the time-consuming coding of morphological characters requires expertise with a group of organisms and it yields much less characters than molecular procedures for analysis (de

Sá et al., 2014). Further, given the relative sizes of the phenotypic and genotypic matrices (tens or hundreds of characters vs. thousands of sites, respectively), the notoriously disproportional quantity of molecular data may overwhelm phenotypic data in total evidence analyses (Kluge, 1983). To test this prediction, I explore the impact of a modest morphological dataset on an analysis dominated by a larger DNA sequence matrix, a phenomenon that has received recent attention (e.g. de Sá et al., 2014;

Mirande, 2016).

Systematics background

External head morphology has played an important role in the systematics of

Cercosaurinae. For example, some cercosaurine genera were traditionally grouped by the presence or absence of prefrontal scales. However, compelling phylogenetic evidence for that grouping is lacking, as molecular based phylogenetic analyses (e.g.

Castoe et al., 2004; Goicoechea et al., 2012) have shown that these genera do not correspond to monophyletic groups.

The former Proctoporus sensu lato (i.e. Petracola+Proctoporus+Riama) and several of its traditionally assumed relatives (e.g. Euspondylus and Opipeuter xestus) exemplify this scenario. The absence of prefrontals served to diagnose Proctoporus s.l.

(e.g. Peters and Donosos-Barros, 1970), which represents the model case of presumably

monophyletic groups in the Cercosaurinae. In contrast, Kizirian and Coloma (1991) and Kizirian (1995, 1996) questioned the monophyly of Proctoporus s.l. Doan (2003a) resolved Proctoporus s.l. as a monophyletic group based on external morphology, but

DNA analysis of Gymnophthalmidae by Castoe et al. (2004) supported Kizirian’s view.

Their molecular analyses resolved a polyphyletic Proctoporus s.l. and depicted scattered distributions of the absence and presence of prefrontals in the Cercosaurinae.

Subsequently, Doan and Castoe (2005) split Proctoporus s.l. into three genera. First,

Proctoporus sensu stricto held species from the Andes of southern Peru and Bolivia. It contained the type species, Proctoporus pachyurus, and its relatives (i.e. the

Proctoporus pachyurus group sensu Uzzell, 1970). Second, Petracola Doan and Castoe,

2005 hosted species from the Andes of central and northern Peru. It included the

Proctoporus ventrimaculatus group (sensu Uzzell, 1970) plus one species. Finally,

Riama was resurrected for the remaining 24 species from the Andes of Peru, Ecuador,

Colombia and Venezuela, the Cordillera de la Costa in Venezuela, and Trinidad.

Subsequently, many new species were referred to Proctoporus s.s. (Doan et al., 2005;

Goicoechea et al., 2013; Mamani et al., 2015), Petracola (Kizirian et al., 2008;

Echevarría and Venegas, 2015) and Riama (Rivas et al., 2005; Arredondo and Sánchez-

Pacheco, 2010; Sánchez-Pacheco, 2010a; Sánchez-Pacheco et al., 2011; Sánchez-

Pacheco et al., 2012; Aguirre-Peñafiel et al., 2014). In their phylogenetic study of

Proctoporus s.s., Goicoechea et al. (2012) transferred into Proctoporus two species of

Euspondylus Tschudi, 1845, a genus whose monophyly had also been questioned

(Köhler and Lehr, 2004), and the monotypic genus Opipeuter Uzzell, 1969, which have prefrontals. Similarly, in their recent phylogenetic analysis of Cercosaurinae, Torres-

Carvajal et al. (2016) transferred three additional species of Euspondylus, as well as R. laudahnae, into Proctoporus.

Kizirian (1996) and Doan and Castoe (2005) provided a detailed systematic documentation of the diversity of Riama. Subsequently, most taxonomic activity on

Riama has consisted of species descriptions (but see Torres-Carvajal et al., 2016).

5.2 Materials and methods

Taxon sampling

Taxa were selected based on availability of tissues (or DNA sequences in GenBank) and specimens for morphological study. The former criterion was considered for exclusion and in the absence of the latter, character-states were retrieved from the literature. When multiple terminals for a given taxon were available, I chose those for which the greatest number of gene sequences were available, first, and the greatest length of gene sequences, second. Based on published and preliminary analyses, I selected divergent terminals in order to maximize intraspecific variation. When possible, I used two terminals per species.

The ingroup included 55 terminals representing 25 of the 30 (>80%) currently recognized species of Riama plus five undescribed taxa from Colombia and Venezuela.

My sampling added 19 species to the 11 species (R. anatoloros, R. balneator, R. cashcaensis, R. colomaromani, R. labionis, R. meleagris, R. orcesi, R. simotera, R. stigmatoral, R. unicolor and R. yumborum) included by Torres-Carvajal et al. (2016) as follows: R. achlyens, R. afrania, R. aurea, R. columbiana, R. “Cordillera Central”,

R. “Cordillera Occidental”, R. crypta, R. hyposticta, R. kiziriani, R. laevis, R. “Nariño”,

R. oculata, R. raneyi, R. shrevei, R. “Sierra Nevada”, R. striata, R. “Venezuela”, R. vespertina and R. vieta. Tissues for DNA extraction were not available for R. inanis, R. luctuosa, R. petrorum, R. rhodogaster and R. stellae.

To test the monophyly of Riama, and due to the uncertain phylogenetic affinities of most species of Riama, 38 species representing 11 of the remaining 13 nominal cercosaurine genera were added to the outgroup (Anadia Gray, 1845, Cercosaura,

Echinosaura Boulenger, 1890, Macropholidus Noble, 1921, Neusticurus Duméril and

Bibron, 1839, Pantodactylus, Petracola, Pholidobolus Peters, 1862, Placosoma

Tschudi, 1847, Potamites Doan and Castoe, 2005, and Proctoporus). My analyses did not include the recently erected Gelanesaurus Torres-Carvajal et al., 2016, and

Euspondylus, for which no tissues were available. The outgroup also included representatives of the following taxa: Alopoglossidae Goicoechea et al., 2016

(Ptychoglossus brevifrontalis), which forms the sister taxon of

Teiidae+Gymnophthalmidae (Goicoechea et al., 2016); Teiidae Gray, 1827 (Ameivula ocellifera and Kentropyx calcarata), the sister group of Gymnophthalmidae

(Goicoechea et al., 2016). The outgroup also contained the following gymnophthalmid subfamilies: Bachiinae Colli et al., 2015 (Bachia flavescens), or Bachiini in the sense of Goicoechea et al. (2016); Ecpleopodinae Fitzinger, 1843 (Ecpleopus gaudichaudii), or Ecpleopodini in the sense of Goicoechea et al. (2016); Gymnophthalminae Fitzinger,

1826 (Gymnophthalmus vanzoi); Rachisaurinae Pellegrino et al., 2001 (Rachisaurus brachylepis); and Riolaminae Kok, 2015 (Riolama leucosticta). Overall, the outgroup included 47 species. Ptychoglossus brevifrontalis was designated as the root for all analyses. Recent progress in understanding the diversification of cercosaurines led to correcting the identities of two terminals. KU 212687, Proctoporus cf. ventrimaculatus of Castoe et al. (2004) was included in the type series of Petracola waka (Kizirian et al., 2008). Similarly, ROM 22892, referred to as “Neusticurus sp. Guyana” by Fu

(2000), was included as a paratype in the original description of Echinosaura sulcarostrum (Donnelly et al., 2006). As suggested by Castoe et al. (2004: appendix 2),

I swapped the 12S and 16S sequences of Potamites juruazensis (GenBank accession numbers AF420758, AF420704, respectively) with those of Ptychoglossus brevifrontalis (AF420757, AF420697).

Genotypic evidence

Most published molecular analyses of gymnophthalmid relationships used DNA sequences from four loci: nuclear DNA (nDNA) oocyte maturation factor gene (C-mos), mitochondrial DNA (mtDNA) NADH dehydrogenase subunit IV (ND4), and mitochondrial rRNA subunits 12S and 16S. Therefore, I sampled the same fragments obtaining up to 374, 860, 400, and 515 base pairs (bp), respectively. Primers and their sources are provided in Table 1. I analyzed a total of 2149 bp of sequences. Novel sequences were deposited in GenBank. My own data were augmented with sequences in GenBank from Fu (2000), Pellegrino et al. (2001), Castoe et al. (2004), Goicoechea et al. (2012), Kok et al. (2012), Torres-Carvajal and Mafla-Endara (2013), Aguirre-

Peñafiel et al. (2014), Kok (2015), and Torres-Carvajal et al. (2016). Voucher specimens and GenBank accession numbers are listed in Table 2.

DNA isolation, sequencing, and editing. Total genomic DNA was extracted from frozen and ethanol-preserved liver or muscle tissues using either the DNeasy kit

(Qiagen, Valencia, CA, USA), following the manufacturer’s guidelines, or a guanidinium isothiocyanate protocol. Amplification of fragments of C-mos, ND4, 12S and 16S was performed with 25 µL final reactions. Negative controls were run on all amplifications to check for contamination. Primers and PCR conditions are detailed in

Table 1. Double-stranded PCR-amplified segments were cleaned and then sequenced in both directions using standard protocols and conventional Sanger sequencers. Novel sequences constituted a consensus of both DNA strands. Sequences were visualized,

assembled and edited using Geneious v.6.1.8 (http://www.geneious.com, Kearse et al.,

2012).

Morphological evidence

Phenotypic character sampling targeted variation among cercosaurines. Thirty-five characters were scored, including 25 derived from external morphology (scutellation) and 10 from hemipenial anatomy (Appendix S2). Other sources of variation, such as osteology and myology, have been examined in too few cercosaurine taxa, and in too few Riama species in particular, to have allowed for their inclusion (for example, see

Montero et al. 2002). Twenty-seven characters were binary and the remaining eight multi-state characters were treated as non-additive (Appendix S2). Characters 10, 13,

14, 20, 21, 28, 30, 31 and all hemipenial characters (0–9) have not been included previously in phylogenetic analyses of gymnophthalmid lizards. In her morphology- based phylogenetic analyses of Proctoporus s.l. and Cercosaura s.l., Doan (2003a,b) used a set of 62 external morphological characters. My characters 16 and 17 corresponded to her characters 1 and 2, and my characters 11, 18, 15, 19, 29, 32, 33, 34 and (22–27) were modifications of her characters 3, 10, 13, (24–25), 36, 45, (40, 42),

41 and 46, respectively; parenthetic notation denotes multiple characters being treated as a single one. Character 12 was a modification of character 10 of Rodrigues et al.

(2005). The remaining 50 mostly meristic, but also morphometric, characters of Doan were not included herein because Doan coded intraspecific polymorphism using the

Generalized Frequency Coding method (Smith and Gutberlet, 2001), following argumentation by Wiens (1995, 1998), among others, which was shown to yield untenable results (Murphy and Doyle, 1998). Taxon phylogeny is to be inferred from hypothesized character-states transformations (Hennig, 1966). Only transformations

from one character-state to another (a à a’) constitute evidence for relationships. In general, changes in the distribution of states among organisms, or in frequency of states in populations (a à aa’ à a’), do not entail additional character-state transformations.

Thus, methods that convert polymorphism into frequencies conflate population-level similarity with character transformation events (Grant and Kluge, 2003, 2004). Further, because frequencies are not heritable (Murphy, 1993; Wiens, 2000) I consider them irrelevant in phylogenetic inference6. In the case of meristic, continuous variation (i.e. counts), Doan kindly provided me with her raw data, which complemented my own observations. However, although I agree that continuous variation carries phylogenetic information (Goloboff et al., 2006), a defensible method of incorporating meristically continuous variables into phylogenetic inference awaits development (but see Goloboff et al. 2006).

Character-states for most species were coded directly from observations taken using a stereoscope and complemented with published data. For unavailable material, data were taken exclusively from the literature. Polymorphic species had multiple states that were scored for a character. When it was not possible to collect all phenotypic data for a particular species, unknown character-states were treated as missing (“?”).

Inapplicable characters were scored as “–”. Phenotypic characters and their states are described in detail in Appendix S2 (Analysis and description of phenotypic characters).

The morphological matrix (Nexus format) is deposited in Morphobank (O’Leary and

Kaufman, 2011, 2012: permalink: http://morphobank.org/permalink/?P2601). The list of specimens examined is given in Appendix S3.

6This position does not reflect that of all coauthors (see Torres-Carvajal, 2007, for example), but did not interfere with my final phenotypic character sampling.

Hemipenial morphology. Hemipenes were prepared following the procedures described by Manzani and Abe (1988) as modified by Pesantes (1994) and Zaher (1999). The retractor muscle was severed manually and the everted organ was filled with stained petroleum jelly. Following Uzzell (1973) and Nunes et al. (2012), calcareous hemipenial structures were stained in an alcoholic solution of alizarin red. Terminology follows Dowling and Savage (1960), Savage (1997), and Nunes et al. (2012).

Hemipenes examined are listed in Appendix S4.

Institutional acronyms

Institutional abbreviations for specimen repositories generally follows Sabaj Pérez

(2014). To this, I added the following collections: CORBIDI (Centro de Ornitología y

Biodiversidad, Lima, Peru), DHMECN (División de Herpetología, Museo Ecuatoriano de Ciencias Naturales, Quito, Ecuador), EPNH (Escuela Politécnica Nacional,

Colección Herpetología, Quito, Ecuador), FHGO (Fundación Herpetológica Gustavo

Orcés, Quito, Ecuador), MHNCSJ (Museo de Historia Natural, Colegio San José,

Medellin, Colombia), MHNUC (Museo de Historia Natural, Universidad de Caldas,

Manizales, Colombia), PSO-CZ (Museo de Historia Natural de la Universidad de

Nariño, Pasto, Colombia), and UV-C (Museo de Vertebrados, Universidad del Valle,

Cali, Colombia).

Phylogenetic analyses

I performed a total evidence analysis of the molecular and phenotypic data under the maximum parsimony optimality criterion. The rationale for this approach was advanced by Farris (1983) and discussed, among others, by Kluge (1989, 2004), Goloboff (2003),

Goloboff and Pol (2005) and Kluge and Grant (2006). The cladogram that minimizes the transformations required to explain the observed character variation maximizes evidential congruence and provides the greatest explanatory power (Farris, 1983; Kluge and Grant, 2006). While maintaining that the total evidence analysis of all available evidence identifies the optimal explanation (Kluge, 1989, 2004), I also analyzed the molecular data separately using the same parameters to evaluate the effect of a modest morphological matrix on an analysis dominated by a larger DNA sequence dataset (cf. de Sá et al., 2014). Following Padial et al. (2014), I employed POY 5.1.1 (Varón et al.,

2010) for tree-alignment (i.e. direct optimization or dynamic homology; e.g. Sankoff,

1975; Wheeler, 1996; Varón and Wheeler, 2012, 2013). The method tested hypotheses of nucleotide homology dynamically by optimizing unaligned DNA sequences directly onto alternative topologies (Kluge and Grant, 2006; Wheeler et al., 2006; Grant and

Kluge, 2009) while simultaneously optimizing prealigned transformation series as standard static matrices.

I searched for optimal trees using the Museu de Zoologia da Universidade de São

Paulo’s high-performance computing cluster (Ace), as described in detail by Padial et al. (2014) and de Sá et al. (2014). I calculated tree-costs using the standard direct optimization algorithm for unaligned data (Wheeler et al., 2006) with all transformations weighted equally. Following Grant et al. (2006: 56–57), I treated each sequenced individual as a separate terminal and duplicated the phenotypic data coded for the species as a whole for each conspecific terminal. I used the same search parameters to analyze both the total evidence and molecular-only datasets. Each analysis involved four 4-hr searches on 704 CPUs (giving a total of 11 264 CPU-hours).

I used the command “search,” which implemented a driven search that included random addition sequence Wagner builds, Subtree Pruning and Regrafting (SPR) and Tree

Bisection and Reconnection (TBR) branch swapping (RAS+swapping; Goloboff,

1996), Parsimony Ratcheting (Nixon, 1999), and Tree Fusing (Goloboff, 1999). The shortest trees of each independent run were stored and used to perform a final round of

Tree Fusing on the pooled trees. I then submitted the resulting trees to a final round of swapping using the iterative pass algorithm (Wheeler, 2003a). To verify the length reported during the tree-alignment analyses and search for additional optimal trees, I calculated the implied alignment (i.e. the matrix version of the tree-alignment; Wheeler,

2003b) and performed an additional 1280 random addition sequence Wagner builds plus TBR searches, saving five minimum-length trees per build.

I estimated clade support (Grant and Kluge, 2008a) using the Goodman-Bremer measure (GB; Goodman et al., 1982; Bremer, 1988; Grant and Kluge, 2008b), by determining the length difference between the optimal trees and all trees visited during a TBR swap of one of the optimal tree using the corresponding implied alignment.

Although it is possible for shorter suboptimal trees to be found by calculating the optimal tree-alignment for each visited topology, the time requirements would be prohibitively costly unless each search was made extremely superficial. Further, Padial et al. (2014) found that using the implied alignment to estimate support overestimates

GB values considerably less than when GB is calculated using a MAFFT (Katoh et al.,

2005) similarity-alignment.

To evaluate the impact of the small morphological dataset on my results, I assessed differences between the total evidence and molecular-only analyses by examining clade-by-clade incongruences and by comparing the standardized support values obtained in the two analyses. For clades shared by both results I calculated the ratio of explanatory power (REP) value (Grant and Kluge, 2007), which scales the observed support for a given clade relative to its maximum possible support (Grant and Kluge,

2010). I obtained the lengths of least parsimonious trees by conducting 1280 random addition sequence Wagner builds plus TBR searches with all characters assigned a weight of -1 and taking the absolute value of the resulting lengths. To make REP values more manageable, I multiplied them by 10,000 and report them to two significant figures.

Biogeography and character evolution

The novel knowledge that emerges from phylogenetic analysis has implications beyond the problems of systematics. By providing a causally relevant framework of reference, knowledge of phylogeny often leads to unanticipated insights and identifies novel problems for further investigation (Grant et al., 2006). In the section

“Biogeographical commentary”, I analyzed the implications of my phylogenetic results for patterns of distribution among major biogeographic units in the Neotropics. Thus, rather than performing a detailed biogeographic analysis, I explored the connection between the biogeographic units, as implied from phylogenetic evidence, within a hypothesis-testing framework. Similarly, in the section “Character evolution”, I analyzed the implications of the total evidence phylogeny for the evolution of prefrontal scales in the Cercosaurinae.

5.3 Results

General total evidence results

Analysis of the total evidence dataset completed 35 533 replicates of random addition sequence Wagner builds plus TBR branch swapping, 189 895 rounds of Tree

Fusing, and 17 405 iterations of Ratcheting. The analyses identified two optimal trees

of 9702 steps. A final round of swapping under the iterative pass algorithm and additional searches using the implied alignment resulted in two equally parsimonious trees of 9680 steps (not shown). Only one ingroup node collapsed in the strict consensus tree (Fig. 6.1), which involved terminals of R. crypta and resulted in the polytomy of a clade also composed of R. hyposticta and R. oculata (tree-alignment matrix and consensus tree deposited in MorphoBank; permalink: http://morphobank.org/permalink/?P2601). Analyses resulted in a least parsimonious tree of 22 325 steps, which I used for calculating REP values.

Riama polyphyly and general ingroup relationships

Analyses recovered a monophyletic Cercosaurinae (sensu Torres-Carvajal et al.,

2016; Cercosaurini in the sense of Goicoechea et al., 2016) (Fig. 6.1). As defined currently, Riama was polyphyletic. Representatives fell into three clades, with no pair of clades as sister taxa (Fig. 6.1: yellow, blue and red). In sequence, (i) an Andean clade

(yellow), Riama sensu stricto, included the type species R. unicolor along with R. balneator, R. orcesi, R. striata, R. “Cordillera Occidental”, R. “Cordillera Central”, R. columbiana, R. anatoloros, R. raneyi, R. “Nariño”, R. colomaromani, R. simotera, R. cashcaensis, R. stigmatoral, R. meleagris, R. yumborum, and R. labionis. (ii) A second

Andean clade (blue) was composed of R. vieta, R. laevis, R. afrania, R. hyposticta, R. crypta, R. oculata, R. kiziriani, R. aurea, and R. vespertina. Finally, (iii) a third clade

(red) was comprised of R. “Sierra Nevada” (from the Sierra Nevada de Santa Marta,

Colombia), the Tepuian Anadia mcdiarmidi, R. achlyens (from the Cordillera de la

Costa, Venezuela, CC), the Trinidadian R. shrevei, and R. “Venezuela” (also from CC).

The density of taxon sampling allowed the coherent delimitation of these three clades,

whose monophyly was well supported: GB = 42, 29 and 11, respectively. This permitted development of a phylogenetic taxonomy (Appendix S5).

Following the new taxonomy, Riama s.s. contained two major clades (GB = 24 and

15) and was the sister group of (Andinosaura (Oreosaurus (Anadia

(Macropholidus+Pholidobolus))) (Echinosaura+Petracola (Cercosaura)

(Potamites+Proctoporus))). Andinosaura also had two major clades (GB = 29 and 2)

(Fig. 6.1).

Relevant outgroup relationships

Outgroup sampling included representatives of Proctoporus sensu stricto and

Petracola, both of which, plus Riama sensu lato, formed the former Proctoporus s.l.

My analysis recovered both Proctoporus s.s. and Petracola as monophyletic (GB = 15 and 10, respectively), but they were not closely related to each other, nor to any of the three clades of Riama s.l. Further, the Tepuian Anadia mcdiarmidi nested within clade three of Riama s.l. (i.e. Oreosaurus), and the recently resurrected Pantodactylus

(Goicoechea et al., 2016) nested within Cercosaura (Fig. 6.1).

Molecular-only results and comparison with the total evidence analysis

Analysis of the molecular-only dataset resulted in five most parsimonious trees of

9464 steps (not shown). Three outgroup nodes collapse in the strict consensus tree (Fig.

6.2), involving terminals of Pholidobolus prefrontalis, Cercosaura schreibersii and

Proctoporus pachyurus (tree-alignment matrix and consensus tree deposited in

MorphoBank; permalink: http://morphobank.org/permalink/?P2601). Analyses

resulted in a least parsimonious tree of 21 720 steps, which I used for calculating REP values.

Although the ingroup clades Riama, Andinosaura, and Oreosaurus (following the new taxonomy) are monophyletic in both analyses, and the relationships within them are identical, topologies from the total evidence and molecular-only analyses present important differences. My molecular-only analysis placed Andinosaura as the sister group of Riama and Oreosaurus as sister to a clade composed of Potamites, Petracola,

Cercosaura and Proctoporus (Fig. 6.2). In contrast, my total evidence analysis recovered Riama as the sister group of the remaining cercosaurines except for

Placosoma+Neusticurus, followed by Andinosaura and Oreosaurus as sister to a clade composed of Anadia, Macropholidus and Pholidobolus (Fig. 6.1). Among outgroup taxa, incongruence between both analyses includes the placement of Echinosaura sulcarostrum and Potamites and relationships within Proctoporus (Figs 6.1 and 6.2).

In addition to topological differences, comparison of REP values for the ingroup clades between the two analyses shows that support for the three clades Riama,

Andinosaura, and Oreosaurus increased with the inclusion of morphological evidence.

REP support for Riama is 3 in the molecular-only analysis and 3.3 in the total evidence analysis. Similarly, REP supports for Andinosaura and Oreosaurus are 1.7 and 0.3 in the molecular-only analysis and increased to 2.2 and 0.8, respectively, in the total evidence analysis (Figs 6.1 and 6.2). Among outgroup taxa, REP values increased for four clades (Neusticurus, Pholidobolus, Potamites and Cercosaura), remained the same for one clade (Macropholidus), and decreased for four clades (Placosoma, Anadia,

Petracola and Proctoporus; Figs 6.1 and 6.2).

5.4 Discussion

Ingroup relationships

When Doan and Castoe (2005: 408) resurrected Riama, they cautioned that “[w]e much prefer to take the chance of creating a paraphyletic Riama [than to create a paraphyletic Proctoporus], because there is much greater likelihood that the northern

Proctoporus s.l. species belong there”. However, my analysis, which is the first to combine both the molecular and phenotypic evidence of cercosaurine lizards, resolves

Riama sensu Doan and Castoe (2005) as a polyphyletic genus (Fig. 6.1). Torres-

Carvajal et al. (2016) recently reported non-monophyly for Riama sensu Doan and

Castoe (2005) because “R”. laudahnae nested deeply within (and was transferred to)

Proctoporus. Previous molecular-only phylogenetic analyses dealing with

Cercosaurinae diversification only included representatives (usually five species) of

Riama sensu stricto. My analyses resolve Riama s.s. as the sister group of the remaining cercosaurines except for Placosoma+Neusticurus. This finding agrees with the results of Castoe et al. (2004)7 (following the current taxonomy), Goicoechea et al. (2012,

2016), and Colli et al. (2015), but disagrees with those of Pyron et al. (2013), Kok (2015) and Torres-Carvajal et al. (2015). The latter three studies resolved Riama s.s. as the sister group of all other cercosaurines. Using denser taxon sampling of cercosaurines,

Torres-Carvajal et al. (2016) recently found Riama s.s. to be the sister of the remainder

7 Castoe et al.’s (2004) Bayesian analysis of the concatenated nuclear and mitochondrial data recovered Riama s.s. as the sister group of all cercosaurines except Placosoma+Neusticurus sensu stricto (Doan and Castoe, 2005), while their strict consensus of two most parsimonious trees embedded Riama s.s. within Cercosaurinae (sensu Torres-Carvajal et al., 2016) in an unresolved position.

cercosaurines except for Placosoma+Neusticurus and Echinosaura (excluding E. sulcarostrum).

Based on morphological similarity, Uzzell (1958) recognized the Proctoporus luctuosus group, as currently consisting of Riama achlyens, R. laevis, R. luctuosa (not included herein), R. oculata and R. shrevei. My results differ because R. laevis and R. oculata nest within the second clade of Riama (i.e. Andinosaura), but R. achlyens and

R. shrevei nest within the third clade (i.e. Oreosaurus). The non-monophyly of the luctuosus group is congruent with Doan (2003a) in her morphological analysis of

Proctoporus sensu lato. Uzzell (1958: 12) also suggested that R. achlyens and R. shrevei were sister-species, and Doan recovered them as sister taxa. When describing

R. rhodogaster (not included herein), Rivas et al. (2005) mapped seven morphological traits onto Doan’s phylogeny. They hypothesized the clade (R. achlyens (R. rhodogaster+R. shrevei)). My analyses resolve R. achlyens, R. shrevei and R.

“Venezuela” as being closely related, and they nest together along with R. “Sierra

Nevada” and Anadia mcdiarmidi. This placement of A. mcdiarmidi is surprising for two reasons. First, Kok and Rivas (2011) and Montero et al. (2002), among others, associated Anadia with other cercosaurine genera, such as Euspondylus, based on morphology. Second, Kok et al. (2012) and Kok (2015) included for the first time a species of Anadia (A. mcdiarmidi) in analyses of DNA sequence data (16S and ND1; and 16S, ND4 and C-mos, respectively). Kok et al. (2012: Suppl. info., fig. 2) recovered

A. mcdiarmidi as part of a cercosaurine clade that contained two species of Potamites,

Cercosaura ocellata, and Echinosaura sulcarostrum. Because the position of the latter species was unresolved, a sister species relationship for A. mcdiarmidi was also unresolved. Later, Kok (2015: fig. 8) recovered a cercosaurine clade of similar content; it included Proctoporus and Petracola, which were not sampled in Kok et al.’s (2012)

study. His analyses resolved Anadia (mcdiarmidi) as the sister taxon of Echinosaura

(sulcarostrum). This relationship was also found by Goicoechea et al. (2016: fig. 12), although they also recovered A. mcdiarmidi as the sister taxon of Potamites (their figs

4 and 8). Because these studies lacked rigorous taxon sampling of cercosaurines, the phylogenetic relationships of A. mcdiarmidi were inconclusive. The denser taxon sampling of Torres-Carvajal et al. (2016), which included for the first time additional members of Anadia (the Andean A. petersi and A. rhombifera), resulted in a non- monophyletic Anadia. Anadia mcdiarmidi formed the sister taxon of a clade composed of A. petersi+A. rhombifera, their unnamed clade 1, and Macropholidus+Pholidobolus.

My results corroborate the non-monophyly of Anadia. Anadia mcdiarmidi does not form a monophyletic group with A. rhombifera (Fig. 6.1). Thus, although the placement of A. mcdiarmidi within my third clade of Riama (i.e. Oreosaurus) is surprising, no data challenge this result.

The dearth of taxon sampling of Riama s.s. in most previous molecular-based analyses precludes meaningful comparisons with my results. Aguirre-Peñafiel et al.

(2014) and Torres-Carvajal et al. (2016) added four and six species (seven including

“R”. laudahanae, now in Proctoporus), respectively, to the five taxa of Castoe et al.

(2004). The study by Aguirre-Peñafiel et al. lacked rigorous outgroup sampling, but their main objective was to infer the phylogenetic position of R. yumborum rather than to resolve the phylogeny of Riama. Nevertheless, they resolved R. unicolor as the sister taxon of (R. cashcaensis (R. meleagris+R. stigmatoral (R. yumborum+R. labionis))).

My results (Fig. 6.1), and those of Torres-Carvajal et al. (2016), corroborate this clade.

Previously, Kizirian and Coloma (1991) suggested that R. cashcaensis was probably most closely related to R. unicolor, and Doan (2003a) found R. stigmatoral, R. labionis,

R. meleagris, and R. unicolor to be closely related. Torres-Carvajal et al.’s (2016)

analysis placed R. orcesi and R. balneator as sister species. My results corroborate this hypothesis.

Doan’s (2003a) densely sampled phylogenetic analysis of Proctoporus s.l. yielded a poorly resolved strict consensus tree of two reconstructions that used different weighting schemes. However, her tree included several hypotheses of sister-species relationships supported by evidence. Further, several authors have proposed additional sister-species relationships on the basis of morphological similarity and/or geographic proximity. Doan (2003a) found Riama colomaromani and R. simotera to be sister- species. However, she used specimens of an undescribed species to score morphological characters for R. simotera (Sánchez-Pacheco et al. 2010; herein referred to as R. “Nariño”). Therefore, Doan actually found R. colomaromani and R. “Nariño” to be sister-species. My analysis recovers R. “Nariño” as the sister taxon of R. colomaromani+R. simotera (Fig. 6.1). Although Doan found R. striata and R. vieta to be sister-species, my results assign R. striata to Riama s.s., and R. vieta within the second clade of Riama (i.e. Andinosaura). My results place R. vieta as the sister taxon of R. laevis+R. afrania. Arredondo and Sánchez-Pacheco (2010) previously hypothesized that R. laevis and R. afrania were sister-species based on morphological similarity and relative geographic proximity. Similarly, Sánchez-Pacheco et al. (2011) interpreted the shared occurrence of single, distal filiform appendages on the hemipenial lobes of R. crypta and R. hyposticta as a putative synapomorphy uniting these two species. My analysis nests R. crypta within a clade also composed of R. hyposticta and R. oculata, but forming a polytomy (Fig. 6.1). The placement of R. oculata in this clade provides a prediction that filiform appendages may also occur on the hemipenial lobes of R. oculata. Finally, Sánchez-Pacheco et al. (2012), based on morphological similarity, suggested that R. vespertina, R. aurea, and R. kiziriani (and

R. petrorum, not included in this study) were closely related, which my results also support (Fig. 6.1).

Outgroup relationships

My study is designed to test, as severely as possible—in terms of taxon and character sampling—monophyly of Riama sensu lato, and to explore the relationships among its species. Because my outgroup sampling is largely drawn from Cercosaurinae, my topology (Fig. 6.1) deserves some discussion. I comment on Cercosaura (including

Pantodactylus) and Proctoporus.

Cercosaura and Pantodactylus. Pellegrino et al.’s (2001) molecular phylogenetic analyses of Gymnophthalmidae, based on mtDNA 16S, 12S and ND4, and nuDNA C- mos and 18S, included representatives of Cercosaura (one species), Pantodactylus (two species), and Prionodactylus O’Shaughnessy, 1881 (three species). Pellegrino et al. sampled the type species of Cercosaura (C. ocellata Wagler, 1830) and Pantodactylus

(P. d’orbignyi Dumeril and Bibron, 1939 = P. schreibersii). The species of

Pantodactylus and Prionodactylus formed a clade with C. ocellata. Based on 61 morphological characters for all 11 species of Cercosaura, Pantodactylus, and

Prionodactylus, Doan (2003b) found Prionodactylus to be paraphyletic with respect to

C. ocellata and species of Pantodactylus. Consequently, she relegated Pantodactylus and Prionodactylus junior synonyms of Cercosaura.

Castoe et al. (2004) included five species of Cercosaura in their molecular phylogenetic analysis of Gymnophthalmidae. Cercosaura was monophyletic in their preferred Bayesian inference tree (Castoe et al., 2004: fig. 6), but it was polyphyletic in the strict consensus of their two most parsimonious solutions (their fig. 1) because C.

quadrilineata was sister to a clade composed of Potamites, Pholidobolus, Petracola,

Cercosaura s.s, and Proctoporus (following the current taxonomy). Pyron et al. (2013) included six species of Cercosaura. They recovered a monophyletic Cercosaura.

Torres-Carvajal et al. (2015; see also Torres-Carvajal et al., 2016) performed the largest molecular phylogenetic analysis of Cercosaura to date. They sampled 11 species, including the type species of Prionodactylus (P. manicatus O’Shaughnessy, 1881= C. manicata). Their analysis found Cercosaura, as defined by Doan (2003b), to be non- monophyletic because C. dicra and C. vertebralis nested deeply within Pholidobolus.

They transferred both of these species to Pholidobolus and in doing so redefined

Cercosaura.

Goicoechea et al. (2016) included six species of Cercosaura. They resolved a monophyletic Cercosaura (Goicoechea et al., 2016: SA+PA and SA+ML analyses) and a non-monophyletic Cercosaura (their TA+PA analysis), because C. quadrilineata was sister to a clade composed of Anadia mcdiarmidi, Potamites, Cercosaura s.s., and

Proctoporus. Ultimately, Goicoechea et al. (2016) resurrected Pantodactylus from the synonymy of Cercosaura for C. quadrilineata and C. schreibersii, arguing that (p. 34):

It is possible that the name Prionodactylus O’Shaughnessy (1881) is available but the type species

of that genus, Cercosaura manicata O’Shaughnessy, 1881, was not included in our analysis. We

suspect that “Cercosaura” quadrilineata will ultimately be placed in its own genus, but at present

we cannot exclude the possibility that Prionodactylus is the appropriate assignment for this

species and Cercosaura manicata. Until more data are available, we tentatively resurrect

Pantodactylus to allocate C. quadrilineata and C. schreibersi [sic].

However, Goicoechea et al. (2016) found the type species of Pantodactylus (C. schreibersii) nested consistently within a monophyletic group containing the type

species of Cercosaura (C. ocellata) in all of their analyses. Multiple phylogenetic studies, including mine (Fig. 6.1), have also imbedded C. schreibersii within

Cercosaura s.s. (e.g. Pellegrino et al., 2001; Castoe et al., 2004; Torres-Carvajal et al.,

2015, 2016). Furthermore, in none of their trees did Goicoechea et al. (2016) recover a sister species relationship between C. schreibersii and C. quadrilineata (which they proposed as members of Pantodactylus). My study corroborates the non-monophyly of

Pantodactylus sensu Goicoechea et al. (2016) (Fig. 6.1). Therefore, the placement of C. schreibersii within Cercosaura and the non-monophyly of Pantodactylus render the resurrection of Pantodactylus by Goicoechea et al. (2016) an arbitrary change because they overlooked existing evidence regarding the phylogenetic relationships of C. schreibersii. In addition, the denser taxon sampling of Torres-Carvajal et al. (2016), in terms of Cercosaura (including C. manicata) and Cercosaurinae species diversity, resulted in a monophyletic Cercosaura, as defined by Torres-Carvajal et al. (2015).

My total evidence parsimony analysis includes a denser sampling of cercosaurines than the molecular analyses of Goicoechea et al. (2016), as well as the same terminals of Cercosaura. My results (Fig. 6.1) resemble those of Pellegrino et al. (2001), Castoe et al. (2004: Bayesian analysis), Pyron et al. (2013), and Torres-Carvajal et al. (2015,

2016—excluding Pholidobolus dicrus and P. vertebralis) regarding the monophyly of

Cercosaura. This contrasts with the TA+PA analysis of Goicoechea et al. (2016) and the parsimony analysis of Castoe et al. (2004). Similar to these analyses, my near- optimal trees (not shown) resolve a non-monophyletic Cercosaura due to the placement of C. quadrilineata. Therefore, I return Pantodactylus to the synonymy of Cercosaura

(Appendix S5).

Proctoporus. My analysis resolves monophyly for Proctoporus (Fig. 6.1), as delimited by Goicoechea et al. (2012). This finding agrees with the results of Torres-Carvajal et al. (2015) and Goicoechea et al. (2016: TA+PA and SA+PA analyses). Using denser sampling of cercosaurines, analysis by Torres-Carvajal et al. (2016) nested

Euspondylus rahmi, E. spinalis, E. oreades and Riama laudahnae within Proctoporus s.s. Therefore, Torres-Carvajal et al. (2016) transferred them to this genus. Further,

Proctoporus sensu Goicoechea et al. (2012) was polyphyletic. The polyphyly of

Proctoporus remains to be tested, and, insofar as Torres-Carvajal et al. (2016) did not propose taxonomic changes, I recognize the content of Proctoporus as defined by

Goicoechea et al. (2012; 2013), and as complemented by Torres-Carvajal et al. (2016).

The relationships within Proctoporus are uncertain owing to conflicting topologies

(Goicoechea et al., 2012, 2016; Torres-Carvajal et al., 2016; my work). These include two critical considerations. First, my analysis (Fig. 6.1) corroborates the hypothesis of

Torres-Carvajal et al. (2016) that P. pachyurus, as delimited by Goicoechea et al. (2012;

2013), is a composite of at least two species from Junin and Cusco, Peru. Torres-

Carvajal et al. (2016) included one sample of P. pachyurus from Cusco used by

Goicoechea et al. (2012); it clustered with P. rahmi. Although the phylogeny of

Goicoechea et al. (2012) was consistent with the recognition of a single species, levels of genetic divergences, the findings of Torres-Carvajal et al. (2016), and my results suggest the occurrence of two species. Second, different hypotheses of relationships within Proctoporus (Goicoechea et al., 2016; Torres-Carvajal et al., 2016; this work) challenge the biogeographic scenario of Goicoechea et al. (2012).

Impact of phenotypic evidence on molecular datasets

Epistemologically, the increased explanatory power that results from including additional evidence validates total evidence analysis (Grant and Kluge, 2003; Kluge,

2004). However, to evaluate the effect of phenotypic evidence on an analysis dominated by a larger DNA sequence dataset, I repeated the analyses using only the molecular evidence. Assuming that truly optimal trees were obtained in both heuristic searches

(i.e. total evidence and molecular-only analyses), differences between the results of the two analyses must be due to the inclusion of the morphological evidence.

Topologies from the two analyses present important differences (Results and Figs

6.1 and 6.2). Topological incongruence between the two analyses is surprising given that the total evidence analysis is dominated by molecular evidence. Morphological evidence comprises only 1.2% of the total evidence matrix (2719 aligned nucleotides,

35 morphological characters). This finding is similar to those of de Sá et al. (2014), who found that the inclusion of phenotypic data resulted in 15 topological differences.

In addition to topological incongruence, comparison of REP values for clades that do not differ between the two analyses demonstrates that support varies with the inclusion of morphological evidence. More importantly, support for the ingroup clades increases. de Sá et al. (2014) also found differences in REP values between both analyses.

In conclusion, relative to the size of the entire dataset phenotypic evidence had a disproportionately large impact on my total evidence results, showing that the inclusion of a comparatively small amount of morphology can alter both the topology and support values in analyses dominated by DNA sequence data.

Character evolution

My results identify the optimal phylogenetic explanation of the species diversity of

Riama sensu lato and have implications for the evolutionary origins, losses, and

100

reversals of a number of morphological characters, particularly those related to scutellation and hemipenial anatomy. Below I focus on the evolution of prefrontal scales in the Cercosaurinae. However, the future inclusion of key taxa (e.g.

Euspondylus and Gelanesaurus) might overturn my hypotheses and favor alternative evolutionary explanations.

Prefrontal scales

The occurrence of prefrontals (Character 11; Appendix S2 has played a pivotal role in cercosaurine systematics, having been used both to diagnose genera and species in

Cercosaurinae and to infer phylogenetic relationships. For example, the absence of prefrontal scales was used to diagnose the former, polyphyletic Proctoporus s.l. (e.g.

Peters and Donosos-Barros, 1970) and as a synapomorphy of this group (Doan, 2003a).

The ancestral state for Cercosaurinae optimizes unambiguously as plesiomorphically present in my analysis. Prefrontals were then lost in the most recent common ancestor of the immediately less inclusive clade (i.e. Cercosaurinae excluding

Placosoma+Neusticurus), with no fewer than five independent reversals to present

(Oreosaurus mcdiarmidi, Anadia, Pholidobolus prefrontalis and P. affinis, and the most recent common ancestor of the clade composed of Echinosaura sulcarostrum,

Petracola, Cercosaura, Potamites and Proctoporus). Further, two subsequent returns to absent occur in Petracola and Proctoporus (excluding Proctoporus xestus), with one gain in Proctoporus chasqui+Proctoporus sp.

Biogeographic commentary

My results also have implications for patterns of distribution among major biogeographic units, and especially the connection between the Sierra Nevada de Santa

101

Marta in Colombia (SNSM), the Cordillera de la Costa in Venezuela (CC), the island of Trinidad, and the tepuis (Venezuelan Guyana and Guyana Shield) (Fig. 6.3). My inclusion of the undescribed, SNSM endemic Oreosaurus “Sierra Nevada”, two endemic species from the CC (O. achlyens and O. “Venezuela”), the Trinidadian endemic O. shrevei (Aripo Northern Range), and the tepui endemic O. mcdiarmidi allows me to test previous biogeographic hypotheses. Although each of these highland complexes has a unique geological history, cumulative phylogenetic evidence suggests an ancient connection between them. Species form monophyletic groups despite the considerable geographic distances that separate them. For example, my analysis recovers O. “Sierra Nevada” as the sister of the remaining species of Oreosaurus, followed by O. mcdiarmidi. Oreosaurus achlyens is the sister of O. shrevei+O.

“Venezuela” (Figs 6.1 and 6.3). The distribution of Oreosaurus (SNSM, CC,

Trinidadian highlands and tepuis) constitutes a biogeographic pattern (as a whole) not repeated in other vertebrates. This distribution strongly implies an ancient biogeographic connection between the SNSM and the CC as part of the explanation for the origin of the montane SNSM endemic vertebrate fauna.

Cordillera de la Costa (CC)–island of Trinidad

The Venezuelan CC extends eastwards along the Caribbean coast from the Andean

Cordillera de Mérida and it has two main sections. First, the Barquisimeto Depression separates the CC Central (locality of Oreosaurus achlyens and O. luctuosus) from the

Cordillera de Mérida. Second, the CC Oriental extends farther east along the coast toward the island of Trinidad, which lies 12 km off the northeastern coast of Venezuela

(Doan and Schargel, 2003; Rivas et al. 2005) (Fig. 5.3). In turn, the CC Oriental consists

102

of two separate mountain chains situated in extreme northeastern Venezuela: the

Península de Paria, the type locality of O. rhodogaster, and the massif of Turimiquire, locality of O. “Venezuela” (Rivas et al., 2005). The CC originally extended from the

Mérida Andes onto the island of Trinidad (Liddle, 1946). Thus, the northern range of

Trinidad, the type locality of O. shrevei, was stratigraphically contiguous with the coastal range of Venezuela during the Pliocene and Pleistocene. This connection facilitated the exchange of species through land connections. A Miocene downwarping event severed the land connection (Liddle, 1946; Doan, 2003a; Rivas and Freitas, 2015).

Accordingly, the CC Oriental shares many species with Trinidad (Rivas and Freitas,

2015). Steyermark (1979, 1982) and Kaiser et al. (2015) suggested that some species endemic to the massif of Turimiquire and the Península de Paria (CC Oriental) may be closely related to those of the CC Central and the island of Trinidad. Brown and

Lomolino (1998) suggested that the ancestor of O. shrevei may have arrived in Trinidad using an ephemeral Pleistocene landbridge.

The relationship between Trinidadian and CC taxa is, therefore, not surprising. For example, Manzanilla et al.’s (2009) analysis of Mannophryne (Anura: Aromobatidae) placed the Trinidadian endemic M. trinitatis and the CC endemic M. venezuelensis as sister species. As for Oreosaurus, Doan (2003a) reported a sister relationship for O. shrevei and O. achlyens. Rivas et al. (2005) hypothesized O. achlyens as sister of O. shrevei+O. rhodogaster (not included herein). My analyses resolve O. achlyens as sister of O. shrevei+O. “Venezuela”.

Cordillera de la Costa (CC)+island of Trinidad–Tepuis

103

The Venezuelan Guayana contains remnants of the ancient Guyana Shield, which supports an endemic biota, especially in the tepuis (Doan and Schargel, 2003; and citations therein). Tepuis are geographically isolated mountains derived from the

Precambrian sandstone of the Roraima Group (southern Venezuela, western Guyana, and northern Brazil), and surrounded by savannas and tropical forests (Bonaccorso and

Guayasamin, 2013; and citations therein). The origin of the tepuis’ biodiversity remains uncertain (Rull, 2009; Kok and Rivas, 2011; Bonaccorso and Guayasamin, 2013). The closest relatives of some taxa, including endemic fauna, (may) occur in the surrounding lowlands, the Península de Paria and the massif of Turimiquire (CC Oriental), and the

Andes (Chapman, 1931; Steyermark, 1974, 1979, 1982; Gorzula, 1987; Pérez-Emán,

2005; Schargel et al., 2005; Mauck and Burns 2009; Bonaccorso et al., 2011; Salerno et al., 2012; Bonaccorso and Guayasamin, 2013; Kok et al., 2013; Rivas and Freitas,

2015; Rivas et al., 2005). My results support the hypothesis of a biogeographic association between the CC and the tepuis. Oreosaurus mcdiarmidi, which is endemic to the summit of Abakapá-tepui, and O. “Venezuela”, which occurs endemically on the

Turimiquire massif (CC Oriental), are placed as sister species (Figs 6.1 and 6.3).

Similarly, Pérez-Emán (2005) found Myioborus pariae (Aves: Parulidae), which is endemic to the Península de Paria (CC Oriental), to be the sister taxon of a clade composed of three species of Myioborus that are endemic to the tepuis. Three hypotheses exist for the origin of the tepuis’ biota that have close affinities to other montane regions. First, tepuis taxa come from widely distributed species in the

Neotropical montane regions and vicariance drove their speciation. Second, multiple dispersal events from the Andes and other montane regions account for these taxa.

Third, an ‘ancient tepui’ biota has contributed elements to younger montane regions, including the Andes (Bonaccorso and Guayasamin, 2013; and citations therein).

104

Sierra Nevada de Santa Marta (SNSM): origin of its montane, endemic vertebrate fauna

Lower elevation dry forests, xeric shrublands, and wide alluvial plains isolate the

SNSM (ca. 5800 m maximum elevation) from other Andean wet forests. It does not have a physical association with the other Andean ranges, and seems to be older than the Andes (Tschanz et al., 1974; Simpson, 1975; Lynch and Ruíz-Carranza, 1985;

Guayasamin et al., 2009). Consequently, the SNSM is a center of endemism for multiple plant and animal groups, especially in the highland cloud forests and páramos

(Ruthven, 1922; Cadena et al., 2016 and citations therein). Several hypotheses have been proposed to explain the origin and associations of the endemic vertebrates of the

SNSM. Ruthven (1922) and Carriker (in Ruthven, 1922) attributed the fauna to the lowland forest of the Magdalena basin. Carriker8, Walker and Test (1955), Rivero

(1961), and Lynch (in Duellman, 1979) tied it to the Cordillera de la Costa (CC). Lynch

(1976) suggested affinities with the West Indies (Antilles). Lynch and Ruíz-Carranza

(1985) pointed to the (northern) Andes and/or other Santa Martan species. The inclusion of montane, endemic terrestrial vertebrates from the SNSM in modern phylogenetic studies using DNA sequences (including the present analysis) allow testing of these historical biogeographic scenarios.

In their phylogenetic analysis of the (current) genus Arremon (Aves: Emberizidae),

Cadena et al. (2007; see also Cadena and Cuervo, 2010) included the SNSM endemic

A. basilicus (300–1200 m). Their combined analysis of mitochondrial and nuclear

8Carriker (in Ruthven, 1922) stated that “we have in the mountain mass an absolutely isolated area containing its own distinctive habitats and faunal characteristics, into which enter but two outside influences, that of the Magdalena basin on the southwest and that of the central plateau of Venezuela through the Goajira Peninsula on the east” [sic]. I interpret the latter as the CC Central.

105

genes recovered A. basilicus and A. phygas, which is endemic to the CC Oriental

(Serranía de Turimiquire and Peninsula de Paria), as sister species. The phylogeographic analysis of Henicorhina leucophrys (Aves: Troglodytidae) by Caro et al. (2013; see also Cadena et al., 2016) found H. anachoreta (1800–3600 m) and H. l. bangsi (600–2100 m), two congeneric (and until recently conspecific) taxa that are endemic to the SNSM, to be more closely related to different populations of H. leucophrys on the northern Andes than to each other. Notwithstanding, the high capacity of birds to disperse might have played an important role in initially colonizing the SNSM from geographically proximate areas (e.g. Serranía de Perijá). This scenario leaves uncertain the phylogenetic relationships of SNSM endemic non-avian terrestrial vertebrates. Guayasamin et al. (2009) included the SNSM endemic Ikakogi tayrona

(980–1790 m) in their molecular-based phylogenetic analyses of glassfrogs (Amphibia:

Centrolenidae). Their combined analysis of mitochondrial and nuclear genes found I. tayrona to be the sister taxon of all other Centrolenidae (150 species distributed throughout Tropical Central America, Tropical Andes, CC, Tobago, Guiana Shield,

Amazon Basin, and Atlantic Forest of Brazil). Thus, this study does not shed light on specific biogeographic patterns. In contrast, Castroviejo-Fisher et al. (2015) included the SNSM endemic Cryptobatrachus boulengeri (360–1790 m) in their molecular phylogenetic analysis of egg-brooding frogs (Anura: Hemiphractidae). This species nested within a monophyletic Cryptobatrachus, a genus also composed of five montane species from the northern Andes of Colombia and Venezuela (e.g. Serranía de Perijá).

My study is the first to include a SNSM endemic non-avian reptile in a phylogenetic analysis using DNA sequences. Again, Oreosaurus “Sierra Nevada” (2156 m) is the sister of its congeners, which occur in the CC, on the island of Trinidad, and the tepuis

(Figs 6.1 and 6.3). Thus, O. “Sierra Nevada” does not have a close relationship with

106

Andean radiations of cercosaurines (e.g. Riama). This leads to two scenarios for relationships of montane SNSM endemic vertebrates. First, Caro et al. (2013) suggested allopatric speciation for two SNSM endemic birds (Henicorhina anachoreta and H. l. bangsi) after independent colonization events from geographically proximate Andean ranges (e.g. Serranía de Perijá). The phylogenetic relationships of one SNSM endemic frog (Cryptobatrachus boulengeri; Castroviejo-Fisher et al., 2015) support the hypothesis of a biogeographic association between the SNSM and the northern Andes

(Lynch and Ruíz-Carranza, 1985). Geologically, the SNSM and the Serranía de San

Lucas (northern Cordillera Central of Colombia) are both “composed of Paleozoic and

Precambrian continental rocks intruded by Mesozoic and Cenozoic plutons”, with the

SNSM being displaced by the Santa Marta-Bucaramanga fault (Taboada et al., 2000:

797). Second, for the case of one bird (Arremon basilicus) and one lizard (O. “Sierra

Nevada”), I hypothesize that an ancient biogeographic connection facilitated the exchange of species between the SNSM and the CC, but geological events subsequently severed the connection. Isolation drove genetic differentiation and speciation. The phylogenetic evidence provided by Cadena et al. (2007) and my analyses cannot reject this hypothesis. Previously, Carriker (in Ruthven, 1922), Walker and Test (1955),

Rivero (1961) and Lynch (in Duellman, 1979) suggested affinities between the biota of the SNSM and CC. The Serranía de Macuira may be a remnant of this connection. This mountain range (maximum height of ca. 864 m) stands in the middle of the La Guajira dessert (Guajira Peninsula, Colombia). It is isolated from the SNSM and the Andean

Cordillera Oriental of Colombia, and located approximately halfway (northward) between the SNSM and the westernmost portion of the CC. Its vegetation is composed mainly of wet dwarf and cloud forests. Geological data from the Sevilla Complex of the SNSM include the occurrence of rocks such as Paleozoic orthogneisses and schists,

107

which were intruded by Permian-Late Triassic syntectonic granitoids. These rocks are similar to those observed in the Macuira Formation (Cardona Molina et al., 2006).

Future phylogenetic analysis can test my hypothesis by using Macuira endemic species, such as the aromobatid frog Allobates wayuu.

Relevant to the proposed biogeographic scenario, Oreosaurus rhodogaster and O. luctuosus from the CC were not included herein. Both species were hypothesized to be closely related to O. achlyens, also from CC, and O. shrevei from Trinidad, which were included. Further, Riama inanis from the Venezuelan Mérida Andes was not included in my analyses but was referred herein to Andean Riama (Appendix S5). Although ultimately these actions may be found to be false, only the embedding of R. inanis within Oreosaurus might overturn my hypothesis and favor alternative biogeographic explanations.

Lynch (1978) and Duellman (1979) said the SNSM fauna was Andean, but later

Lynch and Ruíz-Carranza (1985) pointed out these conclusions were based on shared, widely distributed species. Based on morphology, Lynch and Ruíz-Carranza (1985) suggested that SNSM frogs of the (current) genus Pristimantis (including eight endemic species) were not closely related to taxa in the Antilles (Lynch, 1976) and the CC, but to either species found in the northern Andes of Colombia or to other SNSM species.

These hypotheses remain to be tested. Similarly, the influence of the surrounding lowlands (e.g. Magdalena basin; Ruthven, 1922) on the origin of the montane SNSM endemic vertebrates awaits clarification. Finally, future phylogenetic analyses should try to include multiple SNSM endemic congeneric species (e.g. frogs of the genus

Pristimantis and lizards of the genus Anolis).

108

5.5 Acknowledgements

Funding for S.J.S-P. was provided by a COLCIENCIAS doctoral fellowship (Becas

Francisco José de Caldas), an Ontario Graduate Scholarship (OGS) at the University of

Toronto, a Coordenadoria de Aperfeiçoamento de Pessoal de Ensino Superior (CAPES) master’s fellowship, and an American Museum of Natural History Collection Study

Grant. NSERC Discovery Grant 3148 supported the research. T.G. was supported by

Conselho Nacional de Desenvolvimento Científico e Tecnológico Proc. 307001/2011-

3 and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) Proc.

2012/10000-5. O.T-C. and V.A-P. received funding from Secretaría de Educación

Superior, Ciencia, Tecnología e Innovación (SENESCYT). M.T.R. was supported by

Conselho Nacional de Desenvolvimento Cientifico e Tecnológico and FAPESP Proc.

2003/10335-8 and 2011/50146-6. P.M.S.N. was supported by Fundação de Amparo à

Ciência e Tecnologia do Estado de Pernambuco (FACEPE) and FAPESP Proc.

2012/00492-8. I thank M. O’Leary, J.M. Padial and an anonymous reviewer for constructive criticism of the manuscript. S.J.S-P. thanks D. Frost and D.A. Kizirian for workspace and other facilities provided during his stay at the AMNH. For access to collections and specimen and tissue loans I am grateful to D.A. Kizirian (AMNH), K. de Queiroz and T.D. Hartsell (USNM), J. Rosado (MCZ), R.A. Nussbaum and G.

Schneider (UMMZ), A. Resetar (FMNH), L. Trueb and A. Campbell (KU), S.

Kullander and E. Åhlander (NRM), F. Castro and W. Bolívar (UV-C), J.J. Calderón

(PSO-CZ), V. Páez (MHUA), A. Zamudio and A. Ortiz (MHNCSJ), J. Salazar and H.F.

Arias (MHNUC), J.A. Maldonado (IAvH), M.L. Calderón Espinosa (ICN), C.L.

Spencer and S. Werning (MVZ), J.V. Vindum (CAS-SUR), A. Almendáriz (EPNH),

M. Yánez-Muñoz (DHMECN), J. Valencia (FHGO), P.J. Venegas and L.Y. Echevarría

109

(CORBIDI), M. Borges Martins (UFRGS), J.M. Hoyos (MUJ), M.G. Rutherford

(UWI), M. Rada, A. Mejia Tobón, and J.J. Mueses Cisneros. S.J.S-P. thanks P. Pulido-

Santacruz, S.B. Arroyo, J.J. Ospina-Sarria, M. Anganoy-Criollo, S. Marques Sousa, L.

Saboyá-Acosta, and M. Targino Rocha for arranging logistics and collaborating on fieldwork in Colombia.

5.6 Figures

110

111

112

113

(Fig. 6. 1 and Appendix S5).

114

O. “Sierra Nevada” SNSM O. mcdiarmidi TCM O. achlyens CCC O. shrevei IT O. “Venezuela” CCO

Caribbean Sea

Trinidad CCC CCO and Tobago Panama

Venezuela Guyana

French Colombia Suriname Guiana Pacific Ocean Brazil

Ecuador

Figure 5.3. Map of northern South America and summary of the phylogeny and geographic distribution of Oreosaurus. SNSM = Sierra Nevada de Santa Marta,

Colombia; TCM = tepuis from the Chimantá massif, Venezuela; CCC = Cordillera de la Costa Central, Venezuela; IT = island of Trinidad; CCO = Cordillera de la Costa

Oriental, Venezuela.

115

Figure 5.4. Lateral view and asulcate face of the hemipenis of Riama orcesi (KU

142919). Character 0, shape of hemipenial body; state 0, cylindrical. Char. 2(2):

Character 2, flounce orientation on asulcate face of hemipenial body; state 2, horizontal

(with no vertex). Char. 3(1): Character 3, asulcate central nude area; state 1, narrow, restricted to a sagittal stripe. Char. 4(0): Character 4, orientation of lateral body flounces; state 0, chevron shaped. Char. 5(1): Character 5, lateral body flounce ornamentation; state 1, present. Char. 6(1): Character 6, position of lateral body flounce ornamentation; state 1, distributed over entire flounce. Char. 7(0): Character

7, shape of lateral body flounce ornamentation; state 0, comb-like series of spicules.

Char. 8(1): Character 8, isolated transversal flounces on proximal-central region of asulcate face; state 1, present.

116

Figure 5.5. Lateral view and asulcate face of the hemipenis of Riama balneator

(DHMECN 4111). Character 0, shape of hemipenial body; state 1, elongated. Char.

3(2): Character 3, asulcate central nude area; state 2, broad, occupying approximately

50% of the asulcate face. Char. 4(1): Character 4, orientation of lateral body flounces; state 1, extended diagonally from anterior (asulcate) to posterior (sulcate) face. Char.

6(0): Character 6, position of lateral body flounce ornamentation; state 0, distal, restricted to flounce extremities. Char. 7(1): Character 7, shape of lateral body flounce ornamentation; state 1, isolated hook-shaped spines.

117

Figure 5.6. Sulcate and asulcate faces of the hemipenis of Riama crypta (KU 135104).

Character 0, shape of hemipenial body; state 2, conical, with proximal region distinctly thinner than distal and lobes. Char. 1(0): Character 1, lobes; state 0, large, distinct from hemipenial body. Char. 9(1): Character 9, distal filiform appendages on the hemipenial lobes, state 1, present.

118

Figure 5.7. Asulcate face of the hemipenis of Riama simotera (ICN-R 9836). Character

0, shape of hemipenial body; state 3, globose. Char. 1(1): Character 1, lobes; state 1, narrow, uniform with the hemipenial body. Char. 2(0): Character 2, flounce orientation on asulcate face of hemipenial body; state 0, lateral (with a central vertex directed distally). Char. 3(0): Character 3, asulcate central nude area; state 0, absent, flounces extended across entire asulcate face. Char. 8(0): Character 8, isolated transversal flounces on proximal-central region of asulcate face; state 0, absent. Char. 9(0):

Character 9, distal filiform appendages on the hemipenial lobes; state 0, absent.

119

Figure 5.8. Asulcate face of the hemipenis of Riama cashcaensis (KU 217206). Char.

2(1): Character 2, flounce orientation on asulcate face of hemipenial body; state 1, medial (with a central vertex directed proximally).

120

Figure 5.9. Lateral view of the hemipenis of Riama striata (KU 217206). Char. 4(2):

Character 4, orientation of lateral body flounces; state 2, horizontal.

121

Figure 5.10. Lateral view of the hemipenis of Bachia flavescens (AMNH 140925).

Char. 5(0): Character 5, lateral body flounce ornamentation; state 0, absent.

122

5.7 Tables

Table 5.1. List of PCR and sequencing primers used in this study, and a summary of the PCR conditions.

Gene Primer Sequence (5’–3’) Source PCR protocol region Name

12S 12S1L CAAACTGGGATTAGATACCCCACTAT Kocher et 1 cycle: 3 min al. (1989) 94 °C 12S2H AGGGTGACGGGCGGTGTGT 33 cycles: 30 s 92 °C, 30 s 57 °C, 1:50 min 72 °C

1 cycle: 10 min 72 °C

16S 16SF.0 CTGTTTACCAAAAACATMRCCTYTAGC Pellegrino 1 cycle: 3 min et al. 96 °C (2001) 40 cycles: 30 s 16SR.0 TAGATAGAAACCGACCTGGATT Whiting et 95 °C, 1 min 51 al. (2003) °C, 1 min

16SL CGCCTGTTTAACAAAAACAT Harris et 72 °C al. (1998) 16SH CCGGTCTGAACTCAGATCACGT 1 cycle: 10 min 72 °C

ND4 ND4L CACCTATGACTACCAAAAGCTCATGTAGAAGC Arévalo et 1 cycle: 3 min al. (1994) 94 °C Leu CATTACTTTTACTTGGATTTGCACCA 33 cycles: 30 s 92 °C, 30 s 57 °C, 1:50 min 72 °C

1 cycle: 10 min 72 °C

123

1 cycle: 3 min 96 °C

40 cycles: 30 s 95 °C, 1 min 52 °C, 1 min

72 °C

1 cycle: 10 min 72 °C

C-mos G73 GCGGTAAAGCAGGTGAAGAAA Saint et al. 1 cycle: 3 min (1998) 96 °C G74 TGAGCATCCAAAGTCTCCAATC 40 cycles: 30 s 95 °C, 1 min 52 °C, 1 min

72 °C

1 cycle: 10 min 72 °C

1 cycle: 3 min 96 °C

35 cycles: 25 s 95 °C, 1 min 52 °C, 2 min

72 °C

1 cycle: 10 min 72 °C

124

Table 5.2. GenBank accession numbers for loci and terminals sampled in this study.

Asterisks indicate new sequences obtained for this study. Species are listed following the new taxonomy proposed herein. Numbers and letters following species names are identifiers of conspecific terminals. See Materials and Methods for institutional abbreviations.

Terminal ID Voucher C-mos 12S 16S ND4 Ameivula ocellifera MRT 946089 AF420862 AF420706 AF420759 AF420914 Anadia rhombifera1 QCAZ 11061 KU902052 KU902133 KU902214 KU902289 Anadia rhombifera2 QCAZ 11862 KU902053 KU902135 KU902216 KU902291 Andinosaura afrania1 RH – KY670680* KY681098* – Andinosaura afrania2 RM – KY670681* KY681099* – Andinosaura aurea1 QCAZ 9649 KY670647* KY670682* KY681100* KY710831* Andinosaura aurea2 QCAZ 9650 KY670648* KY670683* KY681101* KY710832* Andinosaura crypta1 QCAZ 10455 KY670649* KY670684* KY681102* KY710833* Andinosaura crypta2 QCAZ 6154 KY670650* KY670685* KY681103* KY710834* Andinosaura hyposticta1 PSO-CZ 085 – KY670686* KY681104* – Andinosaura hyposticta2 DHMECN 1360 KY670651* KY670687* KY681105* – Andinosaura kiziriani1 QCAZ 9607 KY670652* KY670688* KY681106* KY710835* Andinosaura kiziriani2 QCAZ 9667 KY670653* KY670689* KY681107* KY710836* Andinosaura laevis WB 1330 KY670654* KY670690* KY681108* KY799165* Andinosaura oculata1 QCAZ 10410 KY670655* KY670691* KY681109* KY710837* Andinosaura oculata2 QCAZ 5474 KY670656* KY670692* KY681110* KY710838* Andinosaura vespertina1 QCAZ 10286 KY670657* KY670693* KY681111* KY710839* Andinosaura vespertina2 QCAZ 10306 KY670658* KY670694* KY681112* KY710840* Andinosaura vieta1 QCAZ 10456 KY670659* KY670695* KY681113* KY710841* Andinosaura vieta2 QCAZ 5287 KY670660* KY670696* KY681114* KY710842* LSUMZ Bachia flavescens H12977 AF420859 AF420705 AF420753 AF420869 LSUMZ Cercosaura argula H12591 AF420838 AF420698 AF420751 AF420896 Cercosaura eigenmanni MRT 976979 AF420828 AF420690 AF420728 AF420895 Cercosaura ocellata MRT 977406 AF420834 AF420677 AF420731 AF420883 LSUMZ Cercosaura oshaughnessyi H13584 AF420852 AF420696 AF420750 AF420893 Cercosaura quadrilineata LG 936 AF420830 AF420672 AF420717 – Cercosaura schreibersii1 LG 1168 AF420856 AF420650 AF420729 AF420882 Cercosaura schreibersii2 LG 927 AF420817 AF420686 AF420749 AF420911 Echinosaura sulcarostrum ROM 22892 – AF206584 AF206584 – Ecpleopus gaudichaudii LG 1356 AF420855 AF420660 AF420738 AF420901 Gymnophthalmus vanzoi MRT 946639 AF420827 AF420687 AF420743 AF420867

125

Kentropyx calcarata MRT 978224 AF420864 AF420707 AF420760 AF420913 Macropholidus annectens1 QCAZ 11120 – KC894341 KC894355 KC894369 Macropholidus annectens2 QCAZ 11121 – KC894342 KC894356 KC894370 CORBIDI Macropholidus huancabambae1 10492 – KC894343 KC894357 – CORBIDI Macropholidus huancabambae2 10493 – KC894344 KC894358 KC894372 Macropholidus ruthveni CORBIDI 4281 – KC894354 KC894368 KC894382 Neusticurus bicarinatus MRT 968462 AF420816 AF420671 AF420708 – Neusticurus rudis MRT 926008 – AF420689 AF420709 AF420905 Oreosaurus achlyens ACHL 11010 – KY670697* KY681115* KY799160* Oreosaurus mcdiarmidi IRSNB 2674 KP283385 – JQ742263 KP283392 Oreosaurus “Sierra Nevada”1 JJS 543 KY670661* KY670698* KY681116* KY799163* Oreosaurus “Sierra Nevada”2 JJS 548 KY670662* KY670699* KY681117* KY799164* Oreosaurus shrevei UWIZM 2011.7 KY670663* KY670700* KY681118* – Oreosaurus “Venezuela” EBRG 5962 – KY670701* KY681119* KY799161* Petracola ventrimaculata KU 219838 AY507910 AY507863 AY507883 AY507894 Petracola waka KU 212687 AY507903 AY507864 AY507876 – Pholidobolus affinis1 QCAZ 9641 – KC894348 KC894362 KC894376 Pholidobolus affinis2 QCAZ 9900 – KC894349 KC894363 KC894377 Pholidobolus macbrydei1 KU 218406 AY507896 AY507848 AY507867 AY507886 Pholidobolus macbrydei2 QCAZ 9914 – KC894352 KC894366 KC894380 Pholidobolus montium1 KU 196355 AF420820 AF420701 AF420756 AF420884 Pholidobolus momtium2 QCAZ 4051 – KC894346 KC894360 KC894374 Pholidobolus prefrontalis1 QCAZ 9908 – KC894350 KC894364 KC894378 Pholidobolus prefrontalis2 QCAZ 9951 – KC894351 KC894365 KC894379 Placosoma cordylinum LG 1006 AF420823 AF420673 AF420734 AF420879 Placosoma glabellum LG 940 AF420833 AF420674 AF420742 AF420907 Potamites ecpleopus MRT 0472 AF420829 AF420656 AF420748 AF420890 LSUMZ Potamites juruazensis H13823 AF420857 AF420697 AF420757 AF420878 Potamites strangulatus KU 212677 – AY507847 AY507866 AY507885 Proctoporus bolivianus1 MNCN 8989 JX436040 JX435940 JX435994 JX436071 Proctoporus bolivianus2 MNCN 43679 JX436043 JX435943 JX435997 JX436069 Proctoporus cf bolivianus Ca1a UTA R-52945 – AY968825 AY968832 AY968813 Proctoporus cf bolivianus Ca1b MHNC 5322 JX436045 JX435945 JX435988 – AMNH R- Proctoporus cf bolivianus Ca2 150695 – AY968821 AY968828 AY968812 Proctoporus carabaya1(a) MHNC 5428 JX436016 JX435912 JX435979 JX436083 Proctoporus carabaya2(b) MHNC 5429 JX436019 JX435915 JX435982 JX436086 Proctoporus chasqui1 MHNC 6771 JX436003 JX435887 JX435946 JX436051 Proctoporus chasqui2 MNCN 44407 JX436004 JX435888 JX435947 JX436052 Proctoporus guentheri1(2) UTA R-51515 AY507900 AY507849 AY507872 AY225185 Proctoporus guentheri2(4) UTA R-51517 AY507901 AY507854 AY507873 AY225169 Proctoporus iridescens1(a) MNCN 44224 JX436021 JX435920 JX435987 JX436078

126

Proctoporus iridescens2(b) MHNC 6005 JX436049 JX435927 JX435966 JX436079 Proctoporus kiziriani1(a) MNCN 44216 JX436022 JX435900 JX435972 JX436096 Proctoporus kiziriani2(b) MHNC 5367 JX436011 JX435907 JX435978 JX436091 Proctoporus lacertus1(a) UTA R-51487 AY507897 AY507850 AY507868 AY225180 Proctoporus lacertus2(b) UTA R-51506 AY507898 AY507851 AY507869 AY225175 Proctoporus pachyurus1 UTA R-52949 – AY968824 AY968834 AY968816 Proctoporus pachyurus2 MHNC 4599 JX436024 JX435891 JX435952 JX436055 Proctoporus sp Ca MNCN 23305 JX436006 JX435890 JX435949 JX436054 Proctoporus sucullucu1(2) UTA R-51478 AY507905 AY507857 AY507878 AY225171 Proctoporus sucullucu2(4) UTA R-51496 AY507906 AY507858 AY507879 AY225177 Proctoporus unsaacae1(3) UTA R-51488 AY507908 AY507859 AY507881 AY225186 Proctoporus unsaacae2(5) UTA R-51477 AY507909 AY507860 AY507882 AY225170 Proctoporus xestus1 MNCN 6160 – JX435898 JX436002 JX436101 Proctoporus xestus2 MNCN 2425 JX436007 JX435899 JX436001 JX436100 Ptychoglossus brevifrontalis MHNSM AY507911 AY507865 AY507884 AY507895 Rhachisaurus brachylepis MRT 887336 AF420853 AF420665 AF420737 AF420877 Riama anatoloros1 QCAZ 9169 KY670664* KY670702* KY681120* KY710843* Riama anatoloros2(3) QCAZ 9201 KY670665* KY670703* KY681121* KY710844* Riama balneator1 QCAZ 11101 KY670666* KY670704* KY681122* KY710845* Riama balneator2 QCAZ 11099 KU902115 KU902196 KU902271 KU902352 Riama cashcaensis1 KU 217205 – – AY507870 AY507887 Riama cashcaensis2 QCAZ 10686 KJ948210 KJ948180 KJ948122 KJ948162 Riama colomaromani1 KU 217209 AY507899 AY507853 AY507871 AY507888 Riama colomaromani2(3) QCAZ 8753 KY670667* KY670705* KY681123* KY710846* Riama columbiana1 ICN 11298 KY670668* KY670706* KY681124* KY710847* Riama columbiana2(8) ICN 11294 KY670669* KY670707* KY681125* KY710848* Riama “Cordillera Central”1 A1 KY670670* KY670708* KY681126* – Riama “Cordillera Central”2 A2 KY670671* KY670709* KY681127* – Riama “Cordillera Occidental” JJM 2251 KY670672* KY670710* KY681128* KY799162* Riama labionis1 QCAZ 10411 KJ948218 KJ948171 KJ948120 KJ948147 Riama labionis2 QCAZ 10412 KJ948207 KJ948172 KJ948121 KJ948148 Riama meleagris1 QCAZ 9840 KJ948214 KJ948182 KJ948129 KJ948164 Riama meleagris2 QCAZ 9846 KJ948212 KJ948183 KJ948133 KJ948168 Riama “Nariño”1 SSP 058 KY670673* KY670711* KY681129* KY710849* Riama “Nariño”2(5) SSP 076 KY670674* KY670712* KY681130* KY710850* Riama orcesi1 KU 2212772 – AY507855 AY507874 AY507889 Riama orcesi2(3) QCAZ 9035 KY670675* KY670713* KY681131* KY710851* Riama raneyi1 QCAZ 10090 – KY670714* KY681132* KY710852* Riama raneyi2 QCAZ 9034 – KY670715* KY681133* KY710853* Riama simotera1(2) QCAZ 879 AY507904 AY507861 AY507877 AY507892 Riama simotera2(4) QCAZ 4120 KY670676* KY670716* KY681134* KY710854* Riama stigmatoral1 QCAZ 7374 KJ948217 KJ948187 KJ948128 KJ948161 Riama stigmatoral2 QCAZ 11412 KJ948209 KJ948189 KJ948124 KJ948158 Riama striata1(2) MAR 333 KY670677* KY670717* KY681135* –

127

Riama striata2(3) MAR 933 KY670678* KY670718* KY681136* KY710855* Riama unicolor1 KU 217211 AY507907 AY507862 AY507880 AY507893 Riama unicolor2 QCAZ 9662 KY670679* KY670719* KY681137* KY710856* Riama yumborum1 QCAZ 10822 KJ948213 KJ948186 KJ948125 KJ948169 Riama yumborum2 QCAZ 10827 KJ948216 KJ948195 KJ948142 KJ948170 Riolama leucosticta1 VUB 3767 KP283389 – JQ742254 KP283396 Riolama leucosticta2 VUB 3263 – – JQ742256 –

128

Chapter 6 FORMAL RECOGNITION OF THE SPECIES OF Oreosaurus (Squamata: Gymnophthalmidae) FROM THE SIERRA NEVADA DE SANTA MARTA, COLOMBIA

A modified version of this chapter was published in Zookeys (Sánchez- Pacheco, S.J.,

P.M.S. Nunes, S.M. Souza, M.T. Rodrigues, and R.W. Murphy. 2017. Formal recognition of the species Oreosaurus (Squamata: Gymnophthalmidae) from the Sierra

Nevada de Santa Marta, Colombia. Zookeys).

129

6 Abstract

Oreosaurus is one of the two genera extracted from the former Riama sensu lato, which was recently recognized as polyphyletic. Oreosaurus is a small clade (five named and two undescribed species) of montane gymnophthalmid lizards and exhibits an exceptional distributional pattern. Its nominal and undescribed species are discontinuously distributed on the Cordillera de la Costa of Venezuela, the tepuis from the Chimantá massif in Venezuela, the highlands of the island of Trinidad, and the

Sierra Nevada de Santa Marta in Colombia (SNSM). Herein, I describe the species of

Oreosaurus that is endemic to the SNSM. Historically, this species associates with two names that are currently nomina nuda: Proctoporus serranus and P. specularis. Formal nomenclatural recognition of Oreosaurus serranus sp. n. renders specularis a permanently unavailable name for this taxon. Oreosaurus serranus sp. n. is the sister of all remaining congeners, and differs primarily from them in having only one pair of genial scales, as well as a unique pattern of scutellation. I provide an identification key to the species of Oreosaurus.

130

6.1 Introduction

Oreosaurus Peters, 1862 (Reptilia: Gymnophthalmidae) contains five named species of montane lizards that have discontinuous distributions on the Cordillera de la

Costa and tepuis from the Chimantá massif in Venezuela, and the Aripo northern range in the Caribbean island of Trinidad (Sánchez-Pacheco et al. 2017). An additional species that is the sister of all remaining congeners and is endemic to the Sierra Nevada de Santa Marta in Colombia (SNSM) remains undescribed. Sánchez-Pacheco et al.

(2017) referred to it as “Sierra Nevada”.

Over 30 years ago, Ayala and Castro reviewed the Colombian lizard fauna in their unpublished but widely distributed book “Lizards of Colombia”. Their work included brief descriptions of several species and they referred to informal specific epithets associated with authors to indicate that formal descriptions were not yet published, but were forthcoming. Among these species, Ayala and Castro included ‘Proctoporus’ serranus, a gymnophthalmid lizard from the Serranía de San Lorenzo, SNSM, and they provided a reference for the description (Harris, dated to 1984). However, Harris’ formal description of this taxon was never published. Although Ayala and Castro included a brief description (based on an undetermined number of specimens), the name

“serranus” is a nomen nudum because it does not have a reference, and therefore fails to conform to ICZN (1999) Art. 11. Similarly, Ayala (1986) published a list of

Colombian lizards, which included undescribed species referred to names within quotes

(“”) and associated with authors to indicate imminent formal descriptions. Most of these names were the same ones provided by Ayala and Castro (unpublished data), the exception being “Proctoporus” “specularis”, also from San Lorenzo, SNSM.

Nevertheless, both the locality and the given reference (Harris, but this time dated to

131

1986—also never published) were strongly suggestive that “serranus” and “specularis” referred to the same species. However, in accordance with ICZN (1999) Art. 13, the absence of a description for “specularis” (Ayala 1986) renders this name a nomen nudum.

While carrying out field work in the SNSM, I had the opportunity to collect a series of specimens that conform to the unpublished description of “serranus”. Two terminals labeled “Sierra Nevada” 1 and 2 were included in a recently published phylogenetic analysis of Riama Gray, 1858 sensu lato (Sánchez-Pacheco et al. 2017), which recovered this species as part of the resurrected Oreosaurus. Although “serranus” and

“specularis” are currently nomina nuda, and by definition unavailable names (i.e., they fail to conform to ICZN Arts. 11 and 13), both of them have reached the modern literature (Rueda-Almonacid et al. 2012 and de Albuquerque et al. 2012, respectively).

A nomen nudum can be made available (or validated) if it is published again in a way that meets the criteria of availability (ICZN, 1999). Anadia altaserrania Harris and

Ayala, 1987, another endemic gymnophthalmid lizard from the SNSM, is a pertinent example. It was included in Ayala and Castro’s unpublished book (with reference to

Harris, Ayala and Castro, 1984) and listed by Ayala (1986; this time with reference to

Harris and Ayala, 1986), but finally published formally by Harris and Ayala (1987).

The situation with Oreosaurus is not unlike that of Anolis in which Poe et al. (2009) provided examples of nomen nudum species of Anolis lizards listed by Ayala (1986).

Below I provide a name and a description for the Oreosaurus from the SNSM.

6.2 Materials and Methods

132

For comparative purposes, specimens of Oreosaurus achlyens (Uzzell, 1958), O. luctuosus Peters, 1862, O. shrevei (Parker, 1935) and the undescribed O. “Venezuela” were examined (Appendix S6). Data for O. mcdiarmidi (Kok & Rivas, 2011) and O. rhodogaster (Rivas et al., 2005) were taken from the literature (Kok and Rivas 2011 and Rivas et al. 2005, respectively). Measurements (snout-vent length [SVL] and tail length) were taken to 0.1 mm with a digital caliper. Sex was determined by noting the presence of hemipenes in males and/or secondary sex characters, such as the number of femoral pores. To facilitate comparisons with other species of Oreosaurus, scutellation and head-scale terminology follows Kizirian (1996). Bilateral variation is reported as left/right. Hemipenes were prepared following the procedures described by

Manzani and Abe (1988) as modified by Pesantes (1994) and Zaher (1999). The retractor muscle was severed manually and an everted organ was filled with stained petroleum jelly. Following Uzzell (1973) and Nunes et al. (2012), calcareous hemipenial structures were stained in an alcoholic solution of alizarin red. Terminology follows Dowling and Savage (1960), Savage (1997) and Nunes et al. (2012).

The following collection abbreviations are used herein: AMNH (American Museum of Natural History, New York), EBRG (Museo de la Estación Biológica de Rancho

Grande, Maracay, Venezuela), MCZ (Museum of Comparative Zoology, Harvard

University, Cambridge, USA), ROM (Royal Ontario Museum, Toronto, Canada), and

USNM (National Museum of Natural History, Washington D.C., USA).

6.3 Species description

Oreosaurus serranus sp. n.

Figures 6.1 – 6.3

133

Holotype.—ROM 53608 (field number JJS 548; Fig. 7.1), an adult female collected by

S.J.S-P., P.M.S.N., Sergio Marques de Souza, Liliana Saboyá-Acosta, Jhon Jairo

Ospina-Sarria, Sandy B. Arroyo, and Mariane Targino Rocha in Colombia, Sierra

Nevada de Santa Marta, Departamento de Magdalena, headwaters of the Río

Guachacos, Corregimiento de Minca, finca Vista Hermosa, approximately 2156 m,

June 2013. This locality is situated at approximately 11°05′N, 74°01′W.

Paratypes.—ROM 53609 (adult female, Fig. 2), ROM 53610 (subadult male), ROM

53611 (subadult female), ROM 53612–13 (juvenile females), and ROM 53614

(juvenile male), all with same data as holotype.

Diagnosis.—Oreosaurus serranus sp. n. can be distinguished from all its congeners by the number of genial pairs (1 in O. serranus sp. n. versus 2 in the other species). It also differs from all other species of Oreosaurus, except O. mcdiarmidi, by the number of supraoculars (3 in O. serranus sp. n. and O. mcdiarmidi versus 4 in the other species), and dorsal scale relief (smooth in O. serranus sp. n. and O. mcdiarmidi versus keeled or slightly keeled in the other species). Oreosaurus serranus sp. n. also differs from O. mcdiarmidi by the absence of prefrontal scales (present in O. mcdiarmidi).

Description.—Oreosaurus serranus sp. n. possesses the following characteristics: (1) maximum known SVL in males 60 mm (n = 2), in females 70.4 mm (n = 5); (2) frontonasal equal to or longer than frontal; (3) prefrontal scales absent; (4) nasoloreal suture complete [= loreal present]; (5) supraoculars three, all in contact with ciliaries;

(6) superciliary series incomplete, formed just by the anteriormost superciliary scale;

134

(7) supralabial-subocular fusion absent; (8) postoculars two; (9) postparietals two; (10) supratympanic temporals two; (11) genials one pair; (12) dorsal scales rectangular, juxtaposed, smooth; (13) nuchal scales smooth; (14) longitudinal dorsal scale rows 10–

11; (15) transverse dorsal scale rows 33–36; (16) ventral scales smooth, in 21–22 transverse scale rows; (17) lateral scale rows (oval, non-granular scales) 4–6; (18) femoral pores per hind limb in males 7–9, in females 2–3 (located proximally); (19) scales between medialmost femoral pores two; (20) subdigital scales on toe I four; (21) anterior cloacal plate scales four or six; (22) posterior cloacal plate scales seven; (23) dorsum dark brown to black with fine brown mottling; distinct dorsolateral stripes absent; lateral ocelli (i.e., white spots surrounded by dark blotches) absent (white or cream spots instead); venter black with conspicuous whitish spots mostly on scale sutures; (24) hemipenial body globose, slightly bilobed, ornamented by 14–15 chevron- shaped flounces on each side.

Description of holotype.—Adult female (Fig. 6.1), SVL = 70.4 mm, tail length = 72.4 mm; head scales smooth, glossy; rostral scale wider than long, higher than adjacent supralabials, in contact with frontonasal, nasals, and anteriormost supralabials posteriorly; frontonasal roughly quadrangular, longer than wide, widest posteriorly, equal in length to frontal, in contact with nasals and loreals laterally, and frontal posteriorly; prefrontals absent; frontal longer than wide, anterior suture convex, lateral sutures concave, posterior suture angular with point directed posteriorly, in contact with anteriormost supraoculars and superciliaries posterolaterally, and frontoparietals posteriorly; frontoparietals pentagonal, in contact anterolaterally with all supraoculars on the left side and second and third supraoculars on the right side, and posteriorly with parietals and interparietal; interparietal hexagonal, longer than wide, lateral sutures

135

concave, in contact with parietals laterally, postparietals posteriorly; parietals in contact with third supraoculars anterolaterally, dorsalmost temporal and postocular scales laterally, and postparietals posteriorly; postparietals pentagonal, two, in broad contact; supraoculars three, all in contact with ciliaries. Nasoloreal suture complete (= loreal scale present), nasal quadrangular; loreal quadrangular, not in contact with second supralabial; superciliary series incomplete, formed just by the anteriormost superciliary scale, which barely extends onto dorsal surface of head, and lies between loreal, frontal, first supraocular, and anteriormost ciliaries; palpebral disc of lower eyelid divided into three large, unpigmented scales; frenocular quadrangular, in contact with loreal and nasal anteriorly; circumorbital scales between posteriormost supraocular and frenocular five; postoculars two; temporals smooth, glossy, polygonal; supratympanic temporals two; supralabials seven; infralabials four. Mental wider than long, in contact with anteriormost infralabials and postmental posteriorly; postmental roughly pentagonal, posterior suture angular with point directed posteriorly, in contact with first and second infralabials laterally; genials in one pair, roughly quadrangular, in contact with second and third infralabials; scale rows between genials and collar fold (along midventral line) eight, medialmost scales of posteriormost scale row distinctly enlarged, smooth; posteriormost gular row enfolded posteriorly, concealing one small scale row; lateral neck rounded, smooth.

Dorsal scales rectangular, longer than wide, juxtaposed, smooth, in 35 transverse rows; longitudinal dorsal scale rows at fifth transverse ventral scale row nine, at 10th transverse ventral scale row 10, at 15th transverse ventral scale row 11; lateral scale rows at fifth transverse ventral scale row 6/5, at 10th transverse ventral scale row four, at 15th transverse ventral scale row four; lateral scales on body near insertion of forelimb small to granular; ventral scales quadrangular, smooth; complete transverse ventral

136

scale rows 22; longitudinal ventral scale rows at midbody 10; anterior cloacal plate scales six; posterior cloacal plate scales seven, medialmost scale with a horizontal suture; scales on tail rectangular and juxtaposed; midventral subcaudals smooth, wider than adjacent scales, nearly square. Femoral pores per hind limb two, located proximally; scales between medialmost femoral pores two.

Coloration of holotype.—In life, dorsal ground color dark brown to black with fine brown mottling; dorsal surfaces of head, body and tail with an iridescent bluish shine.

White or cream spots laterally from neck to posterior portion of body, becoming less distinct posteriorly. Ventral surfaces of head and body predominantly black, with conspicuous whitish spots mostly on scale sutures; subcaudally black without spots. In preservative (70% ethanol), dorsal ground color brown with fine light brown mottling; dorsal surfaces of head, body and tail without the iridescent bluish shine. Ventral surfaces of head and body brown with cream spots on scale sutures.

Hemipenial morphology.—Right organ of subadult male ROM 53610 (Fig. 6.3) is partially everted and filled. Basal and lobular regions are partially damaged.

Hemipenial body is roughly globose, ending in two small and partially everted, barely visible lobes. Partial eversion and some damages precluded the detection of folds, or any other ornamentation, on the lobes.

The sulcus spermaticus, central in position, originates at the base of the organ and proceeds in a straight line towards the lobes. It is bordered by two narrow and parallel nude areas, and divided before reaching lobes’ base by a fleshy fold. Branches of the sulcus spermaticus are not visible. Two columns of at least 14 chevron-shaped flounces ornament the sides of the organ and the borders of the sulcate and asulcate faces of the

137

hemipenial body. Although these flounces do not present calcified comb-like spicules, it is possible that such absence is due to the age of the specimen. These calcified structures are present in adults of most species of Cercosaurinae that have their hemipenial morphology described, including species of Oreosaurus (e.g., Kok and

Rivas 2011, Nunes 2011, Rivas et al. 2005). A broad nude area occupies at least 50% of the asulcate face. Some damages at the basis of the organ precluded the detection of the isolated horizontal flounces on the proximal-central region of the asulcate face that are often present in species of Cercosaurinae (e.g., Kok and Rivas 2011, Nunes 2011,

Rivas et al. 2012, Sánchez-Pacheco et al. 2011).

Variation.—Paratypes consist of an adult female (ROM 53609, SVL = 68.6 mm, Fig.

2), a subadult female (ROM 53611, SVL = 56 mm), two juvenile females (ROM

53612–13, SVLs = 50.5 and 41.4 mm, respectively), a subadult male (ROM 53610,

SVL = 60 mm), and a juvenile male (ROM 53614, SVL = 40.4 mm). The paratypes are similar to the holotype with the following noteworthy exceptions. Frontonasal longer than frontal in ROM 53609–12 and 53614; loreal scale in contact with second supralabial in ROM 53612–13; ventralmost postocular fused with posteriormost subocular on the right side in ROM 53613; medialmost scale of the posterior cloacal plate not divided horizontally in ROM 53610–11 and 53614; palpebral disc of the lower eyelid divided into two large, pigmented scales in ROM 53609; femoral pores per hind limb in female ROM 53612 three. Femoral pore number is the most evident sexually dimorphic character, with males having from 7–9 pores per hind limb (ROM 53610 8/9,

ROM 53614 9/7), whereas females have from 2–3.

138

Distribution and natural history.—Oreosaurus serranus sp. n. is known exclusively from the type locality (Fig. 6.4: top) and adjacent San Lorenzo (Ayala and Castro unpublished data, Ayala 1986), two cloud forest localities on the northwestern slopes of the Sierra Nevada de Santa Marta (SNSM) at elevations of about 1800–2156 m. This forest-dwelling lizard (Fig. 6.4: bottom) is often found under fallen, rotten trunks or logs. Holotype and paratypes were collected manually during the day. The new species was found at the type locality in sympatry with Anadia pulchella, another gymnophthalmid endemic to the SNSM.

Etymology.—The specific epithet serranus, which is an adjective derived from the

Spanish adjective serrano (meaning from the sierra), refers to the location of the species’ type locality in the Sierra Nevada de Santa Marta, and preserves the original etymological intent of Harris, as stated by Ayala and Castro (unpublished data).

Comments.—Formal nomenclatural recognition of Oreosaurus serranus sp. n. renders specularis (Ayala 1986) a permanently unavailable name for this taxon. Specimens reported by Ayala and Castro (unpublished data) were not included herein because they are presumably lost (S.J.S-P., personal observation).

Oreosaurus is one of the two genera extracted from the former Riama sensu lato, which was recently found to be non-monophyletic (Sánchez-Pacheco et al. 2017). The other clade, Andinosaura Sánchez-Pacheco et al., 2017, includes 11 Andean species and Riama sensu stricto is also an exclusively Andean radiation of 16 named species.

Proctoporus Tschudi, 1845 sensu stricto and Petracola Doan & Castoe, 2005 are other genera that include species of the former, polyphyletic Proctoporus sensu lato.

139

Sánchez-Pacheco et al. (2017) discussed the disjunct geographic distributions of species of Oreosaurus, as well as their phylogenetic relationships. Figure 6.5 summarizes these findings. All species of Oreosaurus share the absence of a narrow band of differentiated granular lateral scales (present in species of Andinosaura,

Petracola, Proctoporus, and Riama).

Key to the species of Oreosaurus

1 One pair of genial scales……………….………...…Oreosaurus serranus sp. n.

– Two pairs of genial scales………………..……………………………………2

2 Prefrontal scales present.………………..…………...... O. mcdiarmidi

– Prefrontal scales absent……………………...……………………………..…3

3 Loreal scale absent……………………………..………………………...……4

– Loreal scale present………………………………..………………….……….5

4 Anterior cloacal plate row composed of a small scale..………..…….O. shrevei

– Anterior cloacal plate row composed of two large scales….…...O. “Venezuela”

5 Dorsal body scales hexagonal……………………………………..…………..6

– Dorsal body scales rectangular………………………..…….…..…O. luctuosus

6 42–44 transverse dorsal scale rows..…..……………….…….…O. rhodogaster

– 37–40 transverse dorsal scale rows………..…...……………..……O. achlyens

6.4 Acknowledgments

Funding for S.J.S-P. was provided by a COLCIENCIAS doctoral fellowship (Becas

Francisco José de Caldas), an Ontario Graduate Scholarship (OGS) at the University of

140

Toronto, and an AMNH Collection Study Grant. NSERC Discovery Grant 3148 supported the research. P.M.S.N. and M.T.R. are grateful to Fundação de Amparo à

Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de

Desenvolvimento Científico e Tecnológico (CNPq) and P.M.S.N. is grateful to

Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco (FACEPE) for financial support. S.J.S-P. thanks D. Frost and D.A. Kizirian for work space and other facilities provided during his stay at the AMNH. For specimen loans and access to collections I am grateful to D.A. Kizirian (AMNH), J. Rosado (MCZ), and K. de

Queiroz and T.D. Hartsell (USNM). I thank P. Pulido-Santacruz, S.B. Arroyo, J.J.

Ospina-Sarria, S. Marques de Souza and M. Targino Rocha for arranging logistics and collaborating on fieldwork at Sierra Nevada de Santa Marta. Jessica Hsiung provided the excellent drawings of the holotype.

141

6.5 Figures

Figure 6.1. Oreosaurus serranus sp. n. (holotype, ROM 53608 [70.4 mm SVL]).

Dorsal, lateral and ventral views of the head, and ventral view of the pelvic region.

142

Figure 6.2. Oreosaurus serranus sp. n. (paratype, ROM 53609 [68.6 mm SVL]) in life. Photos: Sergio Marques de Souza (top) and Jhon Jairo Ospina-Sarria (bottom).

143

Figure 6.3. Oreosaurus serranus sp. n. Sulcate (left), lateral (center) and asulcate

(right) views of the right hemipenis of ROM 53610 (paratype).

144

Figure 6.4. Type locality (top) and habitat (bottom) of Oreosaurus serranus sp. n. in the Sierra Nevada de Santa Marta, Colombia. Photos: Jhon Jairo Ospina-Sarria (top) and Sergio Marques de Souza (bottom).

145

O. serranus sp. nov. SNSM

O. mcdiarmidi TCM

O. achlyens CCC

O. shrevei IT

O. bisbali CCO

Figure 6.5. Summary of the phylogeny and geographic distribution of Oreosaurus

(Sánchez-Pacheco et al. 2017). SNSM = Sierra Nevada de Santa Marta, Colombia;

TCM = tepuis from the Chimantá massif, Venezuela; CCC = Cordillera de la Costa

Central, Venezuela; IT = island of Trinidad; CCO = Cordillera de la Costa Oriental,

Venezuela. Oreosaurus luctuosus, from the CCC, and O. rhodogaster, from the CCO, were included in this genus due to the presumed close relationships of these species and O. achlyens and O. shrevei, respectively. Data taken from Sánchez-Pacheco et al.

(2017).

146

Chapter 7 OUTGROUP SAMPLING CRITERIA: SEVERITY OF TEST, EXPANSION, STABILITY, AND ALOPOGLOSSID LIZARDS

A modified version of this chapter will be submitted to Systematic Biology (Sánchez-

Pacheco, S.J.*, Grant, T.*, Murphy, R.W. In prep. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards).

* Equal contribution

147

7 Abstract

Outgroup sampling is a fundamental step in the design of any phylogenetic investigation, independent of optimality criterion, taxonomic group, or source of evidence. Recent studies demonstrate the efficient analysis of thousands of terminals, all of which could be included in any empirical investigation, yet outgroup samples typically include only a small number of terminals. Most papers discuss outgroup sampling criteria in terms of employing “correct” or “appropriate” outgroup terminals to increase “accuracy” or “reliability” by preventing “errors” such as long branch attraction and “incorrect” ingroup rooting. Alternatively, I develop a theory of outgroup sampling, whereby the objective of outgroup sampling is to test nested hypotheses of ingroup topology and character-state transformations as severely as possible by incorporating outgroup terminals in unconstrained, simultaneous analysis of all included terminals, using background knowledge to select the outgroup terminals that have the greatest chance of refuting those hypotheses. This framework provides a logical basis for selecting outgroup taxa, but it does not provide any grounds for limiting the outgroup sample. Therefore, I propose the ancillary procedure of successively expanding the outgroup sample until hypotheses of ingroup topology and homology become stable (insensitive) to increased sampling, with each expansion guided by the scientific objectives of outgroup sampling. This is a heuristic procedure that does not exclude sampling of more outgroup terminals or guarantees that ingroup hypotheses will remain insensitive to further outgroup expansion, and it has no bearing on the objective support of a given hypothesis. Nevertheless, it provides an empirical basis to limit sampling in a given research cycle. I illustrate this procedure by examining the effect of outgroup expansion on a novel multi-locus DNA sequence

148

dataset for alopoglossid lizards (Reptilia: Alopoglossidae). Based on my results, I propose a monophyletic taxonomy for alopoglossids that reflects historical relationships.

149

7.1 Introduction

A fundamental step in the design of all phylogenetic studies, independent of optimality criterion, taxonomic group, or source of evidence, is the selection of ingroup and outgroup taxa. By convention, the ingroup is a putative clade composed of the terminals whose relationships are the primary focus of study. The outgroup is the complement of the ingroup and terminals are sampled from it for comparison to the ingroup. Outgroup comparison is usually conceptualized as a method of rooting the topology and directing or polarizing character transformations (Farris, 1972, 1982), thereby converting a network of abstract connections into a theory of concrete evolutionary events (Lundberg, 1972).

In principle, outgroup comparison can be based on a single terminal or an entire outgroup, but outgroup sampling usually falls somewhere between these two extremes.

A number of practical considerations constrain taxon sampling generally, such as specimen availability and resources for fieldwork, laboratory analysis, and computation.

Systematists focus appropriately on maximizing the density of the ingroup sample.

Consequently, taxon sampling tends to be extremely imbalanced, with ingroup terminals comprising 75–99% of a given dataset. However, the collective efforts of the global scientific community to make DNA sequences and phenotypic evidence available through open access online databases, such as the GenBank

(www.ncbi.nlm.nih.gov), Morphobank (www.morphobank.org), and treeBASE

(www.treebase.org), enable the inclusion of phylogenetic evidence from vast numbers of outgroup terminals using nothing more than a laptop computer and an internet connection. Further, Goloboff et al. (2009) demonstrated that a truly staggering number

150

of terminals—over 73,000!—can be thoroughly analyzed in a reasonable amount of time using modest computing facilities. In large datasets with extensive character conflict, even distantly related terminals can impact ingroup hypotheses.

Thus, must every phylogenetic study include all outgroup terminals no matter how small the ingroup or narrow the phylogenetic question? Although this approach would approximate the ideal of total evidence, such extensive outgroup sampling would be overkill; most of the human and computational effort would be directed at parts of the tree that are so distant from the ingroup that they are irrelevant to the research question

(e.g. spider relationships in a study of geckoes). However, if outgroup sampling need not be exhaustive, then what scientific criteria should guide the selection of outgroup taxa and on what basis should sampling be limited? Below I aim to answer these questions.

7.2 Background and objective

At best, most empirical studies include statements to the effect that the outgroup taxa were sampled “based on” the results of one or more previous phylogenetic studies.

Studies rarely provide any details or justification about those results that informed outgroup selection, or why outgroup comparison was limited to those terminals.

Beyond emphasizing the importance of sampling the presumed sister-group and at least one more outgroup terminal (e.g. Wiley and Lieberman, 2011), little other guidance exists.

Farris (1972, 1982) identified the logical relationship between outgroup comparison and parsimony (an outgroup terminal occurring inside the ingroup refutes ingroup monophyly) yet did not discuss outgroup sampling criteria. Debate over methods of

151

testing or establishing the direction of character-state transformations, including outgroup comparison, was prevalent throughout the 1980s and 1990s (e.g. de Queiroz,

1985; Kluge, 1985; Kluge and Strauss, 1985; Kraus, 1988; Bryant, 1997), but criteria for outgroup sampling (beyond the perceived need to sample the ingroup’s sister taxon to correctly infer ingroup ancestral states) were not specifically addressed.

Nixon and Carpenter (1993) thoroughly reviewed the literature through 1992 and corrected numerous errors and misconceptions subsequent to Farris’s seminal papers.

In particular, they emphasized that the selected outgroup terminals need not form a clade, be restricted to, or even include the ingroup’s presumed sister group, or be primitive with respect to the ingroup, and that the truly globally most parsimonious tree is obtained by unconstrained, simultaneous analysis of all terminals (i.e. without distinguishing between ingroup and outgroup or constraining the outgroup topology prior to parsimony analysis).

In the section titled “Outgroup selection,” Nixon and Carpenter (1993, p. 421) suggested the following guidelines for sampling: (1) select one or more outgroup terminals on the basis of more inclusive synapomorphies shared with the ingroup; (2) related to this, select the ingroup’s sister-group; (3) sample more densely than sparsely, and, if more inclusive synapomorphies are not known, select outgroup terminals based on (4) previous classifications, or (5) similarity. The authors noted explicitly that these recommendations did not ensure “correct” inferences, and they did not propose theoretical justifications for them or for limiting the outgroup sample. These were put forth as pragmatic guidelines at a time when the available data were a fraction of what they are today.

As molecular data became more prevalent, discussion of outgroups shifted to focus on analyses using DNA sequences. Wheeler (1990) observed that rooting with a taxon

152

that is phylogenetically distant from the ingroup may be equivalent to rooting with random data. He suggested criteria to identify random outgroup effects, including the expectation that random outgroup terminals will incorrectly root the tree by joining the longest ingroup branch. Smith (1994) reviewed the errors that may occur when rooting analyses of DNA sequences and suggested strategies to improve the chances of correctly rooting the ingroup. He offered several analytical strategies, including fixing the ingroup topology prior to including outgroup terminals and eliminating ambiguously aligned gene regions. Smith also argued that sampling the sister taxon as densely as possible would increase the accuracy of the ingroup root, which would make the tree more balanced and decrease the ingroup root branch-length, and less so by adding more distantly related taxa. Sanderson and Shaffer (2002) also reviewed the potential sources of error in rooting DNA sequence analyses and repeated earlier recommendations to increase accuracy and reliability, including sampling the ingroup’s sister group and sampling taxa in such a way that the root branch is subdivided.

Many studies have analyzed the effects of taxon sampling, but few have addressed outgroup sampling per se. Most studies based on empirical data either held outgroup taxa constant to focus on the effects of ingroup sampling (e.g. Poe, 1998; Johnson, 2001;

Miller and Hormiga, 2004) or did not isolate outgroup sampling effects from ingroup sampling effects (e.g. Yoder and Irwin, 1999; Rydin and Källersjö, 2002; Hovenkamp,

2006). However, some authors have examined outgroup sampling effects specifically.

Dalevi et al. (2001) studied the effect of outgroup sampling on the monophyly of bacterial divisions by running analyses with different three-terminal outgroup samples.

In one dataset, the ingroup was monophyletic with respect to all outgroup samples, whereas the second dataset had variable ingroup monophyly. Braun and Kimball (2002) examined outgroup sampling effects on inferred relationships of birds and nodal

153

support by repeating analyses with subsamples of the outgroup sample (two turtles and two crocodilians). They found that outgroup perturbations had little effect on ingroup topology or support values but did alter the placement of the ingroup root. Ward et al.

(2010) included six outgroup species in their study of ant relationships and examined the effect that removal of one of them, a relict species thought to be sister to the ingroup, had on ingroup relationships. They also found no effect on ingroup topology but high sensitivity to placement of the ingroup root. Several authors have performed similar subsampling experiments (e.g. Puslednik and Serb, 2008; de la Torre-Bárcena et al.,

2009; Graham and Iles, 2009; Spaulding et al., 2009). They reported dramatic effects on ingroup topology and character optimizations, often extending deep within the tree.

Takezaki and Nishihara (2016) studied the effects of using either ray-finned fishes or cartilaginous fishes as the only outgroup for analyses of lobe-finned vertebrates

(tetrapods, coelocanths, and lungfishes).

Simulated data have been used extensively to study the effects of taxon sampling

(e.g. Hillis, 1998; Zwickl and Hillis, 2002; Heath et al., 2008) and have overwhelmingly found that increasing taxon sampling greatly increases accuracy (contra Rosenberg and

Kumar, 2001). However, simulated data have rarely been used to study outgroup effects specifically. In a study of mammal systematics, Sullivan and Swofford (1997) simulated randomized DNA sequences to demonstrate that the selected outgroup terminal behaved the same as a random outgroup terminal. They related this to the problem of long branch attraction and proposed that a more “reliable” root could be obtained by including additional ingroup terminals to subdivide the long branch between the ingroup and the outgroup. Qiu et al. (2001) used simulated sequences to evaluate the “appropriateness” of using gymnosperms to root an analysis of angiosperms, and Takezaki and Nishihara (2016) employed a similar approach to

154

examine the effect of branch-lengths in determining if ray-finned fishes or cartilaginous fishes are the most appropriate outgroup in analyses of lobe-finned fishes. Graham et al. (2002) found that simulated outgroup terminals composed of randomized DNA sequences tended to join an ingroup of a commelinoid monocot dataset on the longest terminal ingroup branches (cf. Wheeler, 1990), but rarely joined internal branches, regardless of length. Holland et al. (2003) found that a single, distantly related outgroup terminal frequently misplaced the ingroup root and disrupted the ingroup topology.

Shavit et al. (2007) detected these effects as well, but they also found that a two-taxon outgroup sample was more accurate than a one-taxon outgroup sample and that the greatest accuracy was attained when no outgroup sample was included. Such studies are useful to understand methods, but they provide little guidance for empirical research

(Grant, 2002).

Some outgroup selection criteria are, in reality, outgroup exclusion criteria. To infer accurate relationships and correctly root the ingroup, candidate outgroup terminals are selected and analyzed such that problematic terminals may be identified and eliminated.

The most renowned of the outgroup exclusion methods is RASA’s ill-fated “optimal outgroup analysis” (Lyons-Weiler et al., 1998), which enjoyed brief and high profile popularity before being demolished in a flurry of papers (Faivovich, 2002; Farris, 2002;

Simmons et al., 2002). A decade later, Rota-Stabelli and Telford (2008) proposed choosing outgroup taxa that (1) exhibit low substitution rates, (2) have base composition similar to the ingroup, (3) avoid random outgroup effects, and (4) are phylogenetically closely related to the ingroup. These criteria may conflict (e.g. phylogenetically closest terminals may have high substitution rates or different base composition), and no instructions were given to resolve this problem. More importantly, only the last criterion actually serves to select candidate outgroup terminals; the others

155

are all criteria to exclude terminals. Any procedure that eliminates evidence is scientifically problematic, and these exclusion methods actually do not provide criteria for the selection of candidate outgroup terminals in the first place and, therefore, they beg the question. For a strict application of these methods, the 73 thousand terminals of Goloboff et al. (2009) would have to be evaluated to identify which of them should be excluded.

Most studies have discussed the criteria for outgroup sampling in terms of employing “correct”, “appropriate” or “suitable” outgroup terminals to increase

“accuracy” or “reliability” and avoid “incorrect tree topologies” by preventing “errors” such as long branch attraction and “incorrect” ingroup rooting. This is consistent with the view that phylogenetic systematics aims to accurately or reliably reconstruct evolutionary history. However, for those who view accuracy and reliability as lying outside the purview of science (e.g. Watkins, 1984), a research program that incessantly searches for such “magic bullets and golden rules” (Cummings and Meyer, 2005) is destined for failure. As an alternative, I develop a theory of outgroup sampling grounded in the logic of scientific discovery. First, I identify criteria for outgroup sampling that increase the severity of the tests of ingroup hypotheses. Next, I propose a heuristic procedure to defensibly limit outgroup sampling in a given study.

Subsequently, I illustrate this procedure by applying it to a novel DNA sequence dataset for alopoglossid lizards (Reptilia: Squamata: Alopoglossidae).

7.3 The scientific objectives of outgroup sampling

Pylogenetic systematics seeks to formulate and test explanations of history as severely as possible. Such explanations compound hypotheses of topology (monophyly,

156

cladistic events, cladistic relationship or distance) and homology (character-state transformation events, transformation series, patristic relationship or distance) (Kluge and Grant 2006; Grant and Kluge 2009). Consequently, outgroup sampling seeks to test nested hypotheses of ingroup topology and homology by incorporating outgroup terminals in unconstrained, simultaneous analysis of all included terminals.

Evidence causally related to a hypothesized explanation, such that it is logically capable of refuting that hypothesis, effects testing; the greater the probability that the evidence will refute the hypothesis, the severer the test (Popper, 1989). Although severity of a test is expressed in terms of probabilities, these need not be calculated numerically (e.g. Popper, 1979, p. 18). As Popper (1959, p. 402) emphasized “one cannot completely formalize the idea of a sincere and genuine attempt [at refutation]”

(see also Popper, 1983, p. 254). In phylogenetic systematics, evidence consists of the character-states of terminal taxa (Hennig, 1966; Grant and Kluge, 2004, 2009). In addition to the character-states of the terminals whose relationships are the primary focus of study (i.e. the ingroup terminals), the character-states of other terminals (i.e. outgroup terminals) are logically capable of refuting ingroup hypotheses, and, therefore, their inclusion increases the severity of a test.

7.4 Outgroup sampling and tests of topology

The inclusion of outgroup taxa in phylogenetic analyses potentially refutes, and therefore tests, ingroup cladistic hypotheses in at least three ways. First, one or more outgroup terminals may optimally place within the ingroup, which refutes the hypothesis of ingroup monophyly as well as topological relationships among ingroup taxa. Second, outgroup comparison may refute the hypothesized direction of

157

cladogenetic events and, therefore, tests hypotheses of monophyly throughout the ingroup. Third, even without violating ingroup monophyly, outgroup taxa can alter the optimal ingroup topology.

It is logically impossible for the terminal (or clade) designated to root the tree to be placed within the ingroup. Consequently, at least two outgroup terminals are necessary to test ingroup monophyly including one that serves as the root and the other that is free to potentially fall inside the ingroup. Increasing the size of the outgroup sample increases the severity of the test of ingroup topology because this increases the number of terminals that are free to be placed optimally inside the ingroup. However, not all outgroup taxa are equally crucial as a test of ingroup topology, and the severity of the test can be increased further by targeting terminals most likely to violate ingroup monophyly.

The primary sampling criterion to increase the severity of this test is cladistic proximity. Taxa previously found to be closely related to the ingroup have the greatest potential to violate ingroup monophyly because less of the prior evidence would have to be overturned (i.e. there is less evidence separating those terminals from the ingroup than there is separating more distantly related terminals from the ingroup). The closer the putative relationship and the shorter the patristic distance, the stronger the test, with the most plesiomorphic terminals of the presumed sister-group therefore providing the strongest test. In addition to the results of phylogenetic analyses to inform outgroup sampling, taxonomy, shared character-states, biogeography, and other empirical observations that suggest a close relationship with the ingroup is sufficient reason to target an outgroup sample when testing for ingroup monophyly.

158

7.5 Outgroup sampling and tests of homology

The inclusion of outgroup terminals also tests hypothesized ingroup transformations, including those that diagnose the ingroup, in at least three ways. First, analysis of character variation of outgroup terminals tests the delimitation and homology of character-states prior to simultaneous analysis. Outgroup taxa can possess both plesiomorphic and apomorphic character-states that provide evidence to delimit transformation series. Equally important, they can also reveal that what was once delimited as a single transformation series or character-state actually involves multiple characters or states. The importance of these extra states is most obvious in studies of phenotypic characters, but outgroup DNA sequences can also have a profound effect on sequence alignment. Second, rooting using an outgroup terminal or clade can refute the hypothesized direction of character-state transformations and, therefore, tests hypotheses of homology throughout the ingroup. Third, outgroup comparison tests the identity of ingroup homologies, primarily those that are hypothesized to delimit the ingroup.

At least two outgroup terminals are necessary to effect tests of homology, assuming the free, non-rooting outgroup terminal is not placed optimally within the ingroup. This is for the unambiguous optimization of synapomorphies to the ingroup root node. In character optimization, the initial down pass is sufficient to determine the minimum tree cost, but a second up pass is required for final state assignment (e.g. Swofford and

Maddison, 1992). If a single outgroup terminal is employed, or if the entire outgroup sample is treated as the ingroup’s sister clade by rooting between the ingroup and outgroup (Nixon and Carpenter, 1993), then either all potential ingroup synapomorphies optimize ambiguously, with it being equally optimal for character

159

variation to be explained as apomorphies of the ingroup or the root-lineage (Fig. 1), or it must be assumed that outgroup states are plesiomorphic with respect to the ingroup.

In the latter case, this assumption can be tested (and is often refuted) by adding more outgroup terminals and rooting by using one of them.

Increasing the number of outgroup terminals sampled also increases the probability that ingroup character-states will be detected in the outgroup and/or alter the optimal ingroup state assignments. This action increases the severity of the test of ingroup homologies. The specific targeting of taxa known or suspected to possess ingroup synapomorphies, based on either phylogenetic proximity or observed character variation, serves to increase further the severity of the test.

Donoghue et al. (1989) examined the causes of topological change brought about by adding taxa generally, and these discoveries apply to outgroup taxa specifically. The character-states ascertained through outgroup examination have the potential to alter hypothesized transformation series, which strengthens individual tests of homology

(Grant and Kluge, 2004). Outgroup taxa may also possess unique combinations of character-states that introduce character conflict, altering both the optimal topology and, consequently, character-state optimizations (hypotheses of homology).

Outgroup terminals that are phylogenetically nearest to the ingroup likely will have the greatest effect on hypotheses of ingroup character-state transformations. In Figure

2, the addition of outgroup taxon O3 violates ingroup monophyly. State 0 of character

1 in that taxon is optimally explained as newly evolved (Grant and Kluge, 2009).

However, the addition of outgroup sample O4 alters the optimal position of O3 and upholds ingroup monophyly; the presence of character 6 in the ingroup of four taxa is a unique and unreversed synapomorphy, and state 1 of characters 2 and 3 is independently derived in O3 and (B C D) and (C D), respectively. Donoghue et al.

160

(1989) underscored how such “plesiomorphic sister taxa” can effect changes to topologies and transformation series. Thus, dense sampling of the most plesiomorphic terminals of clades that are closely related to the ingroup (including the sister-group) provides the severest test of ingroup homologies.

The adding of distantly related terminals may also alter homologies near or within the ingroup by changing distant optimization ambiguities and reversing polarities that cause changes to topology and homology that ripple across the tree (for an empirical example, see Spaulding et al., 2009). Prior to phylogenetic analysis, the outcome of the complex interactions among large numbers of conflicting characters is unknown.

Nevertheless, as emphasized by Donoghue et al. (1989), the effect of taxon addition depends on the strength of the existing character evidence. Areas of a tree that are weakly supported (sensu Grant and Kluge, 2008 b) are most susceptible to change with increased taxon sampling. Because the least amount of evidence supports these regions of a tree, and alterations in one portion of a tree may lead to changes in distant portions of that tree, the severity of the test can increase by targeting terminals from areas of the outgroup topology that were weakly supported in previous analyses.

7.6 Successive outgroup expansion

We have a logical basis for selecting outgroup taxa, but the resulting sampling criteria do not provide guidelines for limiting the outgroup sample. Indeed, the inclusion of all terminals logically maximizes the explanatory power and testability, and thus no criterion derived solely from the logic of scientific discovery can provide this constraint. Some other basis is required to limit sampling.

161

The ancillary procedure I propose is to successively expand the outgroup sample, with each expansion guided by the criteria outlined above, until ingroup hypotheses remain constant. Once hypotheses of ingroup relationships and homologies become stable (insensitive) to increased outgroup sampling, outgroup expansion may be defensibly stopped. This assumes that it would be useless to expand an outgroup sample to include spiders in a study of geckoes because it would have no impact on the ingroup; this procedure discovers the point at which stability occurs. This heuristic procedure does not prevent the sampling of more outgroup terminals or guarantee that ingroup hypotheses will remain insensitive to further outgroup expansion. However, it provides a rational “stopping rule” for outgroup sampling. Thus, the approach is a step forward from the static, non-explicit and, therefore, often unrepeatable outgroup selection procedures employed currently.

Stability can be reached whether the ingroup is monophyletic or not, but the criteria chosen to assess stability may differ in each case. For monophyletic ingroups, assessment of sensitivity to increased outgroup sampling must address both ingroup topology and ingroup homologies. Small ingroups will require visual inspection of the trees of each sequential analysis only to assess variation in ingroup topology. However, very large datasets will require a measure of topological distance or incongruence that compares trees, such as the topological incongruence length difference index of

Wheeler (1999). Given a stable ingroup topology, evaluation of the sensitivity of hypotheses of ingroup homology requires examination of the synapomorphies that optimize to the ingroup root node only. The difference in branch lengths of the ingroup root node provides a quick numerical assessment, but when the branch length difference is zero, direct comparison of the optimized synapomorphies is required for confirmation.

162

For non-monophyletic ingroups, the criteria employed to assess stability are likely to depend on the degree of para- or polyphyly and the specific research questions. For example, if only a few outgroup terminals are inserted into an otherwise monophyletic ingroup, then the same criteria applied to monophyletic ingroups are relevant. However, if the ingroup occurs in a much larger clade, or if it is polyphyletic, then the stability of the general pattern of relationships (e.g. the sister-group relationships with outgroup terminals if the ingroup is paraphyletic or the composition of each independent clade if it is polyphyletic) may be more relevant than the stability of the specific sister-group relationships or character optimizations within each ingroup clade.

The criteria outlined above provide an objective basis for selecting highest priority outgroup terminals and a defensible stopping rule for outgroup sampling. However, they do not necessarily allow choosing between increasing the sampling density of particular outgroup clades and expanding outgroup sampling to more distant outgroup clades. Prior knowledge of the relationships within outgroup clades is crucial to determine if the highest priority terminals have been included, such as those that are nearest to the ingroup or possess particular character states. Further, an approach to evaluate the expected effect of increasing the sampling density of particular outgroup clades is to sequentially remove or subsample those clades and examine the effect on the ingroup (cf. Spaulding et al., 2009). If the removal or thinning of a particular clade leads to alterations of the ingroup, then the ingroup may be sensitive to increased sampling of that clade. However, outgroup deletion is not an alternative to outgroup expansion because sensitivity to terminal removal may not relate to sensitivity to terminal addition.

163

7.7 Alopoglossid lizards and successive outgroup expansion

Neotropical lizards of the Alopoglossidae inhabit the leaf litter of moist forests from sea level to 2190 m above sea level. They occur throughout northwestern South

America, and with two representatives in Panama and Costa Rica. The family, which contains the genera Alopoglossus (seven species) and Ptychoglossus (15 species), are part of the superfamily Gymnophthalmoidea (405 species), which also contains the

Teiidae and Gymnophthalmidae (Harris, 1994; Torres-Carvajal and Lobos, 2014;

Goicoechea et al., 2016). Therefore, the phylogenetic relationships of Alopoglossidae are crucial to understanding the evolution of Gymnophthalmoidea, which holds for over

6% of the global diversity of lizards.

For over a decade, phylogenetic analyses of DNA sequence data recovered

Alopoglossinae as being part of the Gymnophthalmidae (e.g. Pellegrino et al., 2001;

Castoe et al., 2004; Pyron et al., 2013). Goicoechea et al. (2016) obtained mixed results for the Alopoglossinae. Their similarity-alignment (SA)+maximum likelihood (ML) analysis was consistant with previous studies, but tree-alignment (TA)+maximum parsimony (MP) and SA+MP analyses resolved Alopoglossinae as the sister-group of

Gymnophthalmidae+Teiidae, or it formed a polytomy with Teiidae and the remainder of Gymnophthalmidae, respectively. Consequently, they elevated Alopoglossinae to family rank. Despite conflict regarding the phylogenetic position of alopoglossids within the Gymnophthalmoidea, there is broad consensus that they form a monophyletic group (e.g. Castoe et al., 2004; Pyron et al., 2013; Torres-Carvajal and

Lobos, 2014; Goicoechea et al., 2016). However, the phylogeny of alopoglossids remains poorly understood.

164

Previous studies used sequences from one to four loci for alopoglossids and included one species of Ptychoglossus only. Most evaluations considered little diversity of

Alopoglossus. Torres-Carvajal and Lobos (2014) included six of the seven species of

Alopoglossus as did Goicoechea et al. (2016). Thus, compelling evidence for the monophyly of Alopoglossus and Ptychoglossus is lacking, and especially for

Ptychoglossus. Previous analyses of Ptychoglossus included only one species of 15 species that were assigned to it based on phenotypic similarity. Thus, neither the monophyly of Ptychoglossus and Alopoglossus nor the relationships within

Alopoglossidae have been tested rigorously.

I use a de novo, multi-locus DNA sequence dataset for alopoglossid lizards and perform a series of increasingly inclusive analyses that sequentially add 4–8 outgroup taxa following the criteria outlined above. The datasets have variable amounts of missing data, but this is not an impediment to analysis (Wiens, 2006). I assign

Ptychoglossus as the ingroup.

Ingroup sampling

The ingroup included 28 terminals representing nine of the 15 (= 60%) recognized species of Ptychoglossus plus one undescribed species from Colombia. I used 1–11 terminals per species. My sampling added nine species to the only one (P. brevifrontalis) included in previous studies as follows: P. bicolor, P. cf. bicolor, P. festae, P. gorgonae,

P. myersi, P. plicatus, P. romaleos, P. stenolepis and P. vallensis. Tissues for DNA extraction were not available for P. bilineatus, P. danieli, P. eurylepis, P. grandisquamatus, P. kugleri and P. nicefori.

165

Evidence

I sampled DNA sequences for four mitochondrial and four nuclear loci obtaining up to 5253 base pairs per terminal. Non-coding mitochondrial genes included rRNA subunits 12S and 16S and protein-coding mitochondrial genes included NADH dehydrogenase subunits II (ND2) and IV (ND4). Nuclear protein-coding genes included brain-derived neurotrophic factor (BDNF), oocyte maturation factor (C-mos), prolactin receptor (PRLR) and recombination activating protein 1 (RAG1). Primers and their sources were provided in Table 1. Novel sequences were deposited in GenBank.

Sequences in GenBank from Pellegrino et al. (2001), Reeder et al (2002), Castoe et al.

(2004), Mott and Vieites (2009), Wiens et al. (2010; 2012), Mulcahy et al. (2012),

Torres-Carvajal and Lobos (2014), Kok et al. (2012), Kok (2015), Torres-Carvajal et al. (2015), Arteaga et al. (2016) and Murphy et al. (2016) augmented the de novo data.

Voucher specimens and GenBank accession numbers were listed in Table 2.

DNA isolation, sequencing and editing. Total genomic DNA was extracted from frozen and ethanol-preserved liver or muscle tissues using a NaCl protocol.

Amplification of fragments of 12S, 16S, ND2, ND4, BDNF, C-mos, PRLR and RAG1 was performed with 25-µL final reactions. Negative controls were run on all amplifications to check for contamination. PCR conditions were detailed in Table 1.

Double-stranded PCR-amplified segments were cleaned and then sequenced in both directions using standard protocols and a conventional Sanger sequencer. Novel sequences constituted a consensus of both DNA strands. Sequences were visualized, assembled and edited using Geneious v.6.1.8 (http://www.geneious.com; Kearse et al.,

2012).

166

Phylogenetic analyses

The primary objective of this study is to develop a theory of outgroup sampling grounded in the logic of scientific discovery, independent of alignment method or optimality criterion. Nucleotide homology (i.e. alignment methods), optimality criteria, and models and model selection generally followed Padial et al. (2014). For simplicity,

I limited my analyses to maximum parsimony (MP) and maximum likelihood (ML), although my conclusions applied equally to Bayesian inference (BI). To assess the effects of adding outgroups successively, I employed different assumptions about both nucleotide homology and evolutionary processes. Analyses used three different approaches. First, tree-alignment (i.e. direct optimization or dynamic homology;

Sankoff, 1975; Wheeler, 1996, 2001; Wheeler et al., 2006; Grant and Kluge, 2009) used

MP where the optimality criterion was the minimization of hypothesized changes required to explain the observed variation in DNA sequences. The method tested hypotheses of nucleotide homology dynamically by optimizing unaligned DNA sequences directly onto alternative topologies (Kluge and Grant, 2006; Wheeler et al.,

2006; Grant and Kluge, 2009) while simultaneously optimizing prealigned transformation series as standard static matrices. Second, ML estimation used the optimality criterion of the maximization of accuracy by incorporating several assumptions about the process of evolution. The method extended from a priori, similarity-based alignment and assumed a probabilistic model of molecular evolution

(Felsenstein, 2004). Third, to discern between the effects that different assumptions, alignment methods and tree selection criteria had on phylogenetic inferences, a MP

167

analysis used the same similarity-based alignment for ML. All analyses used the same datasets.

7.8 Successive outgroup expansion and tree- alignment+maximum parsimony analysis (TA+MP)

Homologous fragments were identified and formatted following Grant et al. (2006, p. 56). POY 5.1.1 (Varón et al., 2010) was employed for tree-alignments (Sankoff, 1975;

Wheeler, 1996; Varón and Wheeler, 2012, 2013). Optimal tree-searching used the

Museu de Zoologia da Universidade de São Paulo’s high-performance computing cluster (Ace) (Padial et al., 2014; de Sa et al., 2014). Tree-costs were calculated using the standard direct optimization algorithm for unaligned data (Wheeler et al., 2006) with all transformations weighted equally. Each analysis involved one 3-hr search on

768 CPUs (= 2304 CPU-h). The command ‘search’ was used to implement a driven search that included random addition sequence Wagner builds, Subtree Pruning and

Regrafting (SPR) and Tree Bisection and Reconnection (TBR) branch swapping (RAS

+ swapping; Goloboff, 1996), Parsimony Ratcheting (Nixon, 1999) and Tree Fusing

(Goloboff, 1999). The shortest trees of each independent run were stored and used to perform a final round of Tree Fusing on the pooled trees. The resulting trees were submitted to a final round of swapping using the iterative pass algorithm (Wheeler,

2003a). To verify the length reported by POY and search for additional optimal trees, the implied alignment (i.e. the matrix version of the tree-alignment; Wheeler, 2003b) was calculated and used to perform additional searches using TNT, Willi Hennig

Society Edition (Goloboff et al., 2008). For all analyses, the Goodman-Bremer measure

(GB; Goodman et al., 1982; Bremer, 1988; Grant and Kluge, 2008b) was used to assess

168

clade support (Grant and Kluge, 2008a). The details of the datasets and results were presented in Table 3 and summarized below.

Analysis 1TA+MP.—This analysis included a sparse sampling of Alopoglossus (18 terminals representing four species), the putative sister-group of Ptychoglossus. The tree was rooted with A. angulatus. All six optimal trees (Figure 3; strict consensus) depicted a paraphyletic Ptychoglossus with respect to Alopoglossus.

Analysis 2TA+MP.—An increased sampling density of Alopoglossus included all available representatives (26 terminals representing six species) and while rooted the tree as before. These eight additional outgroup terminals continued to resolve a paraphyletic Ptychoglossus with respect to Alopoglossus in 24 optimal trees (Figure 4; strict consensus). However, relationships within Ptychoglossus differed from those of

Analyses 1 in the position of P. bicolor.

Analysis 3TA+MP.—Outgroup sampling was extended beyond alopoglossids to include gymnophthalmid taxa that were hypothesized to be closely related to

Alopoglossus and Ptychoglossus. When Uzzell (1973) erected Riolama to accommodate Prionodactylus leucostictus, he hypothesized a close relationship of it with Ecpleopus, Alopoglossus and Ptychoglossus based mainly on the shared presence of plicae on the anterior, dorsal surface of the tongue, which occurs rarely in gymnophthalmoids. Myers and Donnelly (2001) and Myers et al. (2009) echoed and elaborated this view. Variation in this morphology has been documented. Riolama uzzelli, R. luridiventris and some unnamed species share a fully plicate tongue (Molina and Señaris, 2003; Esqueda et al., 2004; Myers and Donnelly, 2001) with alopoglossids, but R. leucosticta and R. inopinata, as monotypic Ecpleopus, Kaieteurosaurus and

Pantepuisaurus have a midsection of scale-like papillae that interrupts the oblique plicae on the anterior and posterior portions of the tongue (Uzzell, 1973; Kok, 2015).

169

A similar condition was reported in monotypic Adercosaurus (Myers and Donnelly,

2001). Myers et al. (2009) provisionally assigned this genus to Alopoglossinae (i.e.

Alopoglossidae) based on its hemipenial, lingual, and physiognomic resemblances to

Ptychoglossus, but shortly thereafter Pyron et al. (2013) allocated it to Ecpleopodinae

(Gymnophthalmidae) without comment. The remaining gymnophthalmids have scale- like papillae over most of the tongue and posterior plicae, as nearly all gymnophthalmoids do. Thus, analyses targeted four of the five species of

Gymnophthalmidae with plicae on the anterior, dorsal surface of the tongue for which molecular data were available (Riolama inopinata, R. leucosticta, Ecpleopus gaudichaudii and Kaieteurosaurus hindsi). The tree was rooted with R. inopinata.

Ptychoglossus was polyphyletic in all 47 optimal trees (Figure 5; strict consensus), with terminals divided among three clades. Ptychoglossus plicatus, P. myersi and P. romaleos formed a clade that was sister to the remaining species of Alopoglossidae.

Most species of Ptychoglossus, including P. cf. bicolor, P. bicolor, P. gorgonae, P. festae and P. brevifrontalis, formed a second clade that was sister of the remainder alopoglossids. Ptychoglossus stenolepis and P. vallensis formed a third clade that was sister to all terminals of Alopoglossus; thus, Alopoglossus was paraphyletic with respect to these two species of Ptychoglossus.

Analysis 4TA+MP.—Given the polyphyly of Ptychoglossus in Analysis 3, the outgroup was expanded to include the following: (i) the other gymnophthalmid with plicae on the anterior, dorsal surface of the tongue for which molecular data were available

(Pantepuisaurus rodriguesi); (ii) a gymnopthalmid that was associated with

Ptychoglossus due to shared head and body scutellation (Arthrosaura reticulata; Harris,

1994); (iii) a terminal from well within Cercosaurinae (Pholidobolus macbrydei), the most speciose subfamily of the Gymnophthalmidae; and (iv) a representative of Teiidae

170

(Ameivula ocellifera), the remaining family of the Gymnophthalmoidea. Ameivula ocellifera was used to root the tree. Analyses continued to resolve polyphyly for

Ptychoglossus in the 85 optimal trees (Figure 6; strict consensus); the ingroup had the same three clades as in Analysis 3, but with polytomies in the internal relationships of the second clade (P. brevifrontalis).

Analysis 5TA+MP.—Given the consistent polyphyly of Ptychoglossus and the polytomy within P. brevifrontalis, four additional terminals were sampled, including from near the roots of Teiidae (Teius teyou and Tupinambis cuzcoensis [2]) and from

Amphisbaenia (Geocalamus acutus), the sister group of Gymnophthalmoidea (Pyron et al. (2013). The amphisbaenian was used to root the tree. As found in Analyses 3 and

4, Ptychoglossus remained polyphyletic in the 90 optimal trees (Figure 7; strict consensus). The species of Ptychoglossus divided as in Analysis 4, with the same polytomies within P. brevifrontalis.

A priori, it was logically impossible to predict that adding additional outgroup taxa would not modify the current topology of Ptychoglossus terminals. However, the consistent polyphyly and relationships in the last three successive outgroup expansions suggested that further expansion was not warranted and provided an empirical basis to limit outgroup sampling.

7.9 Successive outgroup expansion and similarity- alignment+maximum parsimony analysis (SA+MP)

Similarity-alignments for MP and ML analyses of static matrices used MAFFT v.7.017 and the G-INS-i strategy for small-scale alignments (Katoh and Standley, 2013).

MAFFT was used to minimize the weighted pairwise distance summed over all

171

sequence pairs, and to sum a consistency score (COFFEE-like score) between the multiple alignment and pairwise alignments in the objective function (weighted sum- of-pair score, WSP). The strategy performed global alignment with a Fast Fourier

Transform approximation progressively on a phenetic (modified UPGMA) guide tree followed by iterative edge refinement that evaluated the consistency between the multiple alignment and pairwise alignments. The iterative refinement was repeated either until no improvement in the WSP score was observed, or 1000 cycles were completed (maxiterate = 1000). Following Padial et al. (2014), analyses used the default transition:transversion cost ratio of 1:2, while changing the gap opening penalty from three times substitutions to one time substitutions to avoid penalizing insertions and deletions more than done in the tree-alignment analyses.

For MP analyses, all transformations were weighted equally and gaps were treated as a fifth state. MP analyses to perform the same analyses described above for the tree- alignment matrices used TNT (Goloboff et al., 2008). Driven searches continued until stable strict consensuses were reached. For all analyses, the Goodman-Bremer measure was used to assess clade support. The details of the datasets and results are presented in Table 3 and summarized below.

Analysis 1SA+MP.—Ptychoglossus was paraphyletic with respect to Alopoglossus in the 12 optimal trees (Figure 8; strict consensus), as it was in Analysis 1TA+MP. Unlike

Analysis 1TA+MP, the position of P. brevifrontalis within the second clade of

Ptychoglossus (as described above) was unresolved in the strict consensus.

Analysis 2SA+MP.—Ptychoglossus was paraphyletic with respect to Alopoglossus in the 60 optimal trees (Figure 9; strict consensus), as it was in Analysis 2TA+MP. Like

TA+MP, the position of P. bicolor varied between Analyses 1 and 2. Unlike TA+MP,

172

the position of P. myersi and relationships within P. brevifrontalis differed between

Analyses 1 and 2.

Analysis 3SA+MP.—Ptychoglossus was polyphyletic in the 24 optimal trees (Figure

10; strict consensus), with terminals divided among the same three clades as in Analysis

3TA+MP. However, relationships of the first and second clades (as described above) differed drastically from TA+MP (Figures 5 and 10). Specifically, in Analysis 3SA+MP the second (instead of the first) clade of Ptychoglossus formed the sister-group of the remaining alopoglossids.

Analysis 4SA+MP.—Ptychoglossus was polyphyletic in the 48 optimal trees (Figure

11; strict consensus). The terminals divided among the same three clades as in Analysis

3. However, the first (instead the second) clade of Ptychoglossus formed the sister- group of the remaining alopoglossids (as in Analyses 3–5TA+MP).

Analysis 5SA+MP.—As before, Ptychoglossus was polyphyletic in the 44 optimal trees (Figure 12; strict consensus) and it had the same relationships for the three clades as in Analysis 4SA+MP. Thus, topology from Analysis 5SA+MP was nearly identical to that of Analysis 5TA+MP.

As in TA+MP, consistent polyphyly and relationships of Ptychoglossus terminals were obtained after four successive outgroup expansions. Therefore, this stability provided an empirical basis to limit the outgroup sampling and suggested that further expansion was not warranted.

7.10 Successive outgroup expansion and similarity- alignment+maximum likelihood analysis (SA+ML)

173

PartitionFinder v2 (Lanfear et al., 2016) was used to select the optimal partition schemes and substitution models for the datasets under the Akaike Information

Criterion (Akaike, 1974). ML analyses used GARLI 2.0 (Zwickl, 2006), which allowed more thorough searching of the tree-space than RAxML (Stamatakis, 2006; Morrison,

2007). Analyses were run on Ace, as described above. Tree searching used 352 replicates of an enhanced strategy consisting of the modification of a set of default parameters that, according to Zwickl (2006), should improve tree searching, albeit at the expense of computational time. Analyses used random addition sequence starting trees (streefname = random; default = stepwise) with 1000 attachments per terminal

(attachmentspertaxon = 1000; default = 50), run termination threshold of 100000 generations without topology improvement (genthreshfortopoterm = 100000; default =

20000), and maximum SPR distance of 30 branches away from original location

(limsprrange = 30; default = 6).

Analysis 1SA+ML.—Ptychoglossus was paraphyletic with respect to Alopoglossus

(Figure 13).

Analysis 2SA+ML.—Ptychoglossus continued to be paraphyletic with respect to

Alopoglossus (Figure 14). However, the position of P. stenolepis differed drastically from Analysis 1SA+ML.

Analysis 3SA+ML.—The inclusion of non-alopoglossids resolved a polyphyletic

Ptychoglossus (Figure 15); terminals divided among the same three clades as in

Analyses 3TA+MP and 3SA+MP. Ptychoglossus stenolepis returned to the same position as in Analysis 1SA+ML (Figure 13; P. stenolepis and P. vallensis were sister species). The first clade of Ptychoglossus (as described above) was the sister-group of the remaining alopoglossids.

174

Analysis 4SA+ML.—Ptychoglossus was polyphyletic, but terminals were divided among four clades instead of three (Figure 16). Ptychoglossus vallensis formed the sister-taxon of (P. stenolepis+Alopoglossus). Unlike Analysis 3SA+ML, the second clade of Ptychoglossus (as described above) was the sister-group of the remaining alopoglossids.

Analysis 5SA+ML.—A polyphyletic Ptychoglossus (Figure 17) had the same relationships as in Analysis 4SA+ML.

As in TA+MP and SA+MP, the trees displayed consistent polyphyly and relationships of the terminals of Ptychoglossus after four successive outgroup expansions. This stability provides an empirical basis for not pursuing further expansions.

7.11 Discussion

The analyses of Ptychoglossus exemplify the outgroup expansion procedure and, by extension, provide a test of alopoglossid relationships. Analyses recover a polyphyletic

Ptychoglossus regardless of optimality criteria and alignment methods. The consistent polyphyly of Ptychoglossus in the last three sequential outgroup expansions reveals that the paraphyly of Ptychoglossus with respect to Alopoglossus in the first two analyses owes to inadequate outgroup sampling. Likewise, some hypotheses of sister-groups and sister-species relationships in the last two sequential expansions of the outgroup reveal that competing hypotheses in the first three analyses owe to inadequate outgroup sampling, even when non-alopoglossids are first included (Analyses 3). The density of taxon sampling in Analysis 5 allows the coherent delimitation of three (TA+MP and

SA+MP) or four (SA+ML) clades of Ptychoglossus. This permits development of a

175

phylogenetic taxonomy (Appendix S7). The dearth of taxon sampling of Ptychoglossus in all previous studies precludes meaningful comparisons with my results.

This exercise illustrates the importance of targeting outgroup terminals specifically to test ingroup relationships instead of arbitrarily expanding the size of the outgroup.

The first expansion beyond alopoglossids (Analyses 3) is sufficient to refute paraphyly for Ptychoglossus and establish the general pattern of relationships (i.e. polyphyly) that subsequent outgroup expansions uphold.

The tree of life is composed of nested clades and their respective homologies. The distinction between ingroup and outgroup is artificial and subjective. No operational or theoretical imperative requires that phylogenetic hypotheses must be tested through outgroup comparison. Hypotheses of ingroup topology (but not monophyly) and homology may be tested without reference to the character-states of other taxa, and polarity of state changes may be tested through methods such as mid-point rooting,

Lundberg rooting, ontogenetic sequence and comparison with ingroup fossils (Bryant,

2001; Hess and Russo, 2007). Notwithstanding, the inclusion of additional taxa generally increases the severity of testing, and the inclusion of outgroup taxa in unconstrained phylogenetic analyses can refute ingroup hypotheses, such as ingroup monophyly and fossil plesiomorphies, that cannot be tested by the other methods and must be assumed. Consequently, the artificial subdivision of taxa into ingroup and outgroup representatives is a useful heuristic approach that greatly strengthens the testing of phylogenetic hypotheses.

None of the proposed outgroup sampling criteria justifies the exclusion of outgroup terminals or guarantees the reliability, accuracy, robustness or stability of either the ingroup topology or root. Ultimately, the more taxa included, the more severe the test.

Further outgroup expansion might allow refutation of ingroup hypotheses, even if prior

176

rounds of testing show ingroup hypotheses to be insensitive to outgroup sampling. This consideration points the way to increased testing in future research cycles (cf. Kluge,

1997). Rather, these sampling criteria heuristically guide outgroup sampling by preferentially targeting high priority terminals. Ingroup stability via outgroup expansion is a rational operational criterion that can constrain the outgroup sample in a given study.

This heuristic approach to outgroup sampling is analogous to heuristic tree- searching strategies. Criteria encoded in search algorithms (e.g. swapping, ratcheting, tree fusing, drifting, sectorial searching; Goloboff, 1999) guide tree-searching

(sampling) and ancillary criteria such as minimum number of independent hits or stable consensus serve as indicators of search-sufficiency. Likewise, there is no guarantee that increasing the sample of trees by running a longer search will not alter the results, and, assuming appropriate search algorithms are used, more exhaustive searches are always stronger than less exhaustive searches.

Many analyses other than outgroup expansion can evaluate the sensitivity of ingroup hypotheses to outgroup sampling. For example, Bielawski et al. (2001) jackknifed outgroup terminals and evaluated topological differences using Robinson and Foulds’s

(1981) index, but they interpreted the results as indicating the robustness of ingroup hypotheses and not as a basis for guiding or limiting outgroup sampling. Similarly, numerous authors have varied outgroup sample composition to determine the effect on the ingroup topology (e.g. Dalevi, 2001; Puslednik and Serb, 2008; de la Torre-Bárcena et al., 2009; Graham and Iles, 2009). However, these evaluations only provide a subjective notion of the reliability of ingroup inferences and do not relate to the objectives herein. Further, ingroup hypotheses may be less sensitive to permutation of a smaller outgroup sample than a larger outgroup sample, yet the latter yields a stronger

177

test and, therefore, is always better scientifically. Similarly, clade support (sensu Grant and Kluge, 2008b) might also be used to assess ingroup sensitivity to outgroup expansion. However, this requires time-consuming searching, and perceived sensitivity

(or insensitivity) may be due to insufficient searching instead of outgroup expansion per se.

Provided that ingroup monophyly is not violated, the choice of an outgroup terminal to root the tree has no bearing on ingroup cladistic or patristic hypotheses. Because the root terminal cannot fall within the ingroup, it is customary to root a tree with the terminal assumed to be most distantly related to the ingroup. Although this assumption is neither necessary nor defensible in terms of the severity of testing an ingroup, it also has no negative impact on tests of ingroup hypotheses and has the advantage of allowing outgroup hypotheses to be tested as well. That is, rooting on the terminal assumed to be farthest from the ingroup increases the number of tested hypotheses and, therefore, testability. Future analyses can test this assumption by further expanding outgroup sampling. This practice is consistent with the cyclic nature of empirical research (Kluge, 1997) and does not entail an infinite regress (contra Colless, 1969,

1985; see below).

Barriel and Tassy (1998) suggested choosing the strict consensus of the directed topologies that result from rooting on each outgroup terminal as the optimal tree. They suggested this to avoid rooting on a single outgroup terminal or clade, which they viewed as entailing an extra assumption and ad hoc hypothesis. Although designating the root entails an extra assumption, it is only ad hoc if it lacks independent empirical evidence. In practice, identification of the most distantly related outgroup terminal is based on the empirical evidence analyzed in previous phylogenetic studies. Moreover,

Barriel and Tassy’s (1998) procedure does not actually succeed in escaping the

178

assumption they eschew. Instead, by rooting on each outgroup terminal, the procedure compounds it by assuming that all outgroup terminals lie outside the ingroup. The only way to avoid the assumption that some terminal or clade is external to the ingroup is to not make that assumption, i.e. by depicting the tree as an undirected network.

The proposed sampling criteria rely heavily on prior knowledge, and especially on the optimal hypotheses from previous phylogenetic analyses. Early critics of outgroup comparison were quick to point out what they perceived as the circular reasoning inherent in the use of prior knowledge of higher relationships to direct phylogenetic hypotheses. The criticism is valid insofar as outgroup terminals were chosen to identify the correct outgroup sample (e.g. the true sister-group) and obtain accurate or reliable results. However, as conceptualized here, outgroup comparison is performed to test ingroup hypotheses as severely as possible, for which the results of prior analyses are crucial. As Popper (1989, p. 240, italics in original) clarified,

“A serious empirical test always consists in the attempt to find a refutation, a

counterexample. In the search for a counterexample, we have to use our

background knowledge;…we always look in the most probable kinds of

places for the most probable kinds of counterexamples—most probable in the

sense that we should expect to find them in light of our background

knowledge.”

Thus, targeting the outgroup terminals that will most probably refute ingroup hypotheses increases the severity of a test. This requires prior knowledge. This use of prior knowledge is exactly contrary to the use of prior knowledge in Bayesian inference, which uses prior knowledge to constrain hypothesis-preference to conform to expectations. I use prior knowledge to increase the probability that hypotheses will be refuted.

179

I exemplify the effect of outgroup expansion by using different methods of alignment and two approaches to tree construction (ML and MP). The selection of outgroup terminals is a fundamental step in the design of any phylogenetic study, independent of alignment method or optimality criterion, and there is no reason to believe that other approaches are affected differently. For example, Puslednik and Serb

(2008) analyzed manually corrected Clustal W (Thompson et al., 1994) alignments using MP and Bayesian analysis and reported the same outgroup sampling effects for both methods, as did de la Torre et al. (2009) in their analyses of MUSCLE (Edgar,

2004; spawned from ASAP, Sarkar et al., 2008) alignments using MP and ML, and

Ward et al. (2010) in their Bayesian and ML analyses of manually corrected Clustal X v1.81 (Thomson et al., 1997) alignments. In this paper, I outline the scientific criteria that should guide the selection of outgroup taxa and propose a basis on which outgroup sampling can be defensibly limited, regardless of phylogenetic method, optimality criterion, taxonomic group or source of evidence.

Acknowledgments

Funding for S.J.S-P. was provided by a COLCIENCIAS doctoral fellowship (Becas

Francisco José de Caldas) and an Ontario Graduate Scholarship (OGS) at the University of Toronto. NSERC Discovery Grant 3148 supported the research. For access to collections and specimen and tissue loans I am grateful to M.T. Rodrigues, W. Bolívar

(UV-C), M. Rada and G. Cháves. S.J.S-P. thanks P. Pulido-Santacruz, P.M.S. Nunes,

S.B. Arroyo, J.J. Ospina-Sarria, M. Anganoy-Criollo, S. Marques Sousa, L. Saboyá-

Acosta and M. Targino Rocha for arranging logistics and collaborating on fieldwork in

Colombia.

180

7.12 Figures

Figure 7.1. Outgroup sampling and tests of ingroup synapomorphies. Top: Using O1

(which could be a single terminal or an entire clade) to root the tree, character optimization cannot determine if characters 1 and 2 are ingroup or outgroup apomorphies. Treating characters 1 and 2 as ingroup synapomorphies requires the assumption that all character-states in O1 are plesiomorphic. Bottom: Using outgroup taxon O2 to root the tree reveals that character 1 is apomorphic in O1 and character 2 is an ingroup synapomorphy, thus refuting the hypothesis that all character-states in O1 are plesiomorphic. Although there are now no ambiguous optimizations, there is also no evidence for the O1+ingroup clade and the root node should be shown as unresolved.

181

Figure 7.2. Alterations to ingroup hypotheses caused by increases in outgroup sampling. Top: The ingroup A–D is monophyletic. Middle: Adding outgroup taxon O3 violates ingroup monophyly and state 0 of character 1 in that taxon is optimally explained as the origin of a new state. Bottom: Adding outgroup taxon O4 removes O3 from the ingroup. State 1 of character 1 is now explained as unique and unreversed in the ingroup and state 1 of characters 2 and 3 is independently derived in O3 and (B C

D) and (C D), respectively.

182

Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_buckleki_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2

2.0

Figure 7.3. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony prior to outgroup expansion. Strict consensus of six most parsimonious trees from

Analysis 1TA+MP (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus.

183

Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_atriventris_Peru2 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_bicolor_Colombia Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_buckleki_Ecuador Alopoglossus_copii_Ecuador

3.0 Figure 7.4. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 25 most parsimonious trees from

Analysis 2TA+MP, analysis of data in Analysis 1TA+MP plus eight additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus. Position of P. bicolor differs from Analysis 1.

184

Riolama_inopinata Riolama_leucosticta Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi Alopoglossus_buckleki_Ecuador Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_angulatus_Peru1 Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2

2.0

Figure 7.5. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 47 most parsimonious trees from

Analysis 3TA+MP, analysis of data in Analysis 2TA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic.

185

Ameivula_ocellifera Pholidobolus_macbrydei Arthrosaura_reticulata Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi Pantepuisaurus_rodriguesi Riolama_inopinata Riolama_leucosticta Alopoglossus_buckleki_Ecuador Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_angulatus_Peru1 Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2

2.0

Figure 7.6. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 85 most parsimonious trees from

Analysis 4TA+MP, analysis of data in Analysis 3TA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic.

186

Geocalamus_acutus Pholidobolus_macbrydei Arthrosaura_reticulata Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi Pantepuisaurus_rodriguesi Riolama_inopinata Riolama_leucosticta Alopoglossus_buckleki_Ecuador Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_angulatus_Peru1 Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ameivula_ocellifera Teius_teyou Tupinambis_cuzcoensis1 Tupinambis_cuzcoensis2

2.0

Figure 7.7. Phylogeny of Ptychoglossus under tree-alignment+maximum parsimony following outgroup expansion. Strict consensus of 90 most parsimonious trees from

Analysis 5TA+MP, analysis of data in Analysis 4TA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychogloosus is polyphyletic.

187

Alopoglossus_atriventris_Peru2 Alopoglossus_copii_Ecuador Alopoglossus_buckleki_Ecuador Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1

2.0

188

Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_atriventris_Peru2 Alopoglossus_copii_Ecuador Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_bicolor_Colombia Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_sp_Ecuador2 Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Alopoglossus_buckleki_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia

2.0

Figure 7.9. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 60 most parsimonious trees from Analysis 2SA+MP, analysis of data in Analysis 1SA+MP plus eight additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is paraphyletic with respect to Alopoglossus. Positions of P. bicolor and P. myersi differ from Analysis 1SA+MP.

189

Riolama_inopinata Riolama_leucosticta Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_angulatus_Peru3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleki_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi

2.0

Figure 7.10. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 24 most parsimonious trees from Analysis 3SA+MP, analysis of data in Analysis 2SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic. The second clade of Ptychoglossus

(details in text) is the sister group of the remaining alopoglossids.

190

Ameivula_ocellifera Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_buckleki_Ecuador Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_angulatus_Peru3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Pholidobolus_macbrydei Arthrosaura_reticulata Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi Pantepuisaurus_rodriguesi Riolama_inopinata Riolama_leucosticta

2.0

Figure 7.11. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 48 most parsimonious trees from Analysis 4SA+MP, analysis of data in Analysis 3SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychoglossus is polyphyletic. The first (instead the second) clade of

Ptychoglossus (details in text) is the sister group of the remaining alopoglossids.

191

Geocalamus_acutus Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_angulatus_Peru3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_buckleki_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_bicolor_Colombia Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_sp_Ecuador2 Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Arthrosaura_reticulata Pholidobolus_macbrydei Ecpleopus_gaudichaudii Kaieteurosaurus_hindsi Pantepuisaurus_rodriguesi Riolama_inopinata Riolama_leucosticta Ameivula_ocellifera Teius_teyou Tupinambis_cuzcoensis1 Tupinambis_cuzcoensis2

2.0

Figure 7.12. Phylogeny of Ptychoglossus under similarity alignment+maximum parsimony following outgroup expansion. Strict consensus of 44 most parsimonious trees from Analysis 5SA+MP, analysis of data in Analysis 4SA+MP plus four additional outgroup terminals (details in text and Table 3). Goodman-Bremer support values given above branches. Ptychogloosus is polyphyletic.

192

Alopoglossus_angulatus_Brazil Alopoglossus_copii_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleki_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_sp_Brazil2 Ptychoglossus_sp_Brazil1 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Ptychoglossus_romaleos_Colombia2 Ptychoglossus_romaleos_Colombia1 Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil2 Alopoglossus_atriventris_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil1 Alopoglossus_sp_Brazil3

0.05

Figure 7.13. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood prior to outgroup expansion. Optimal solution from Analysis 1SA+ML (details in text and Table 3). Ptychoglossus is paraphyletic with respect to Alopoglossus.

193

Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_atriventris_Peru2 Alopoglossus_buckleyi_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleki_Ecuador Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Peru1 Alopoglossus_atriventris_Ecuador Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_bicolor_Colombia Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil1 Ptychoglossus_sp_Brazil2 Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Alopoglossus_copii_Ecuador Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_angulatus_Peru3

0.05

Figure 7.14. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 2SA+ML, analysis of data in Analysis 1SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is paraphyletic with respect to Alopoglossus.

194

Riolama_leucosticta Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_plicatus_Colombia Ptychoglossus_myersi_Colombia Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_sp_Ecuador1 Ptychoglossus_festae_Colombia2 Ptychoglossus_festae_Colombia1 Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil2 Ptychoglossus_sp_Brazil1 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_bicolor_Colombia Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_angulatus_Peru3 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_sp_Colombia Alopoglossus_angulatus_Ecuador Alopoglossus_angulatus_Peru1 Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil2 Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_spn_Brazil Alopoglossus_atriventris_Brazil1 Alopoglossus_buckleki_Ecuador Alopoglossus_buckleyi_Peru2 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_copii_Ecuador Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Kaieteurosaurus_hindsi Ecpleopus_gaudichaudii Riolama_inopinata

0.08

Figure 7.15. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 3SA+ML, analysis of data in Analysis 2SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic.

195

Ameivula_ocellifera Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_bicolor_Colombia Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_festae_Colombia1 Ptychoglossus_festae_Colombia2 Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil2 Ptychoglossus_sp_Brazil1 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Peru Alopoglossus_festae_Ecuador Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Peru2 Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_angulatus_Peru3 Alopoglossus_angulatus_Peru1 Alopoglossus_angulatus_Ecuador Alopoglossus_sp_Colombia Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_buckleyi_Peru2 Alopoglossus_buckleyi_Peru1 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleki_Ecuador Alopoglossus_angulatus_Brazil Alopoglossus_sp_Brazil1 Alopoglossus_sp_Brazil3 Alopoglossus_atriventris_Brazil2 Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil2 Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_spn_Brazil Ptychoglossus_vallensis_Colombia4 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Ptychoglossus_romaleos_Colombia2 Ptychoglossus_romaleos_Colombia1 Riolama_inopinata Riolama_leucosticta Pholidobolus_macbrydei Ecpleopus_gaudichaudii Pantepuisaurus_rodriguesi Kaieteurosaurus_hindsi Arthrosaura_reticulata

0.07

Figure 7.16. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 4SA+ML, analysis of data in Analysis 3SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic.

196

Geocalamus_acutus Ptychoglossus_romaleos_Colombia1 Ptychoglossus_romaleos_Colombia2 Ptychoglossus_myersi_Colombia Ptychoglossus_plicatus_Colombia Alopoglossus_carinicaudatus_angulatus_Brazil4 Alopoglossus_angulatus_Peru3 Alopoglossus_angulatus_Peru1 Alopoglossus_carinicaudatus_angulatus_Brazil1 Alopoglossus_sp_Colombia Alopoglossus_angulatus_Ecuador Alopoglossus_carinicaudatus_angulatus_Brazil3 Alopoglossus_carinicaudatus_angulatus_Brazil2 Alopoglossus_atriventris_Peru2 Alopoglossus_copii_Ecuador Alopoglossus_atriventris_Brazil1 Alopoglossus_sp_Brazil1 Alopoglossus_atriventris_Brazil2 Alopoglossus_sp_Brazil3 Alopoglossus_angulatus_Brazil Alopoglossus_atriventris_Ecuador Alopoglossus_atriventris_Peru1 Alopoglossus_spn_Brazil Alopoglossus_sp_Brazil2 Alopoglossus_buckleyi_Peru1 Alopoglossus_angulatus_Peru2 Alopoglossus_buckleyi_Peru2 Alopoglossus_buckleki_Ecuador Alopoglossus_viridiceps_Ecuador Alopoglossus_festae_Ecuador Alopoglossus_festae_Peru Ptychoglossus_stenolepis_Colombia1 Ptychoglossus_stenolepis_Colombia2 Ptychoglossus_vallensis_Colombia3 Ptychoglossus_vallensis_Colombia4 Ptychoglossus_vallensis_Colombia1 Ptychoglossus_vallensis_Colombia2 Ptychoglossus_cf_bicolor_Colombia2 Ptychoglossus_cf_bicolor_Colombia1 Ptychoglossus_bicolor_Colombia Ptychoglossus_sp_Ecuador1 Ptychoglossus_gorgonae_Ecuador Ptychoglossus_brevifrontalis_Venezuela Ptychoglossus_brevifrontalis_MHNSM Ptychoglossus_sp_Brazil2 Ptychoglossus_sp_Brazil1 Ptychoglossus_brevifrontalis_Ecuador1 Ptychoglossus_brevifrontalis_Ecuador5 Ptychoglossus_brevifrontalis_Ecuador2 Ptychoglossus_sp_Ecuador2 Ptychoglossus_brevifrontalis_Ecuador3 Ptychoglossus_brevifrontalis_Ecuador6 Ptychoglossus_brevifrontalis_Ecuador4 Ptychoglossus_festae_Colombia2 Ptychoglossus_festae_Colombia1 Tupinambis_cuzcoensis1 Tupinambis_cuzcoensis2 Ameivula_ocellifera Teius_teyou Riolama_leucosticta Riolama_inopinata Pholidobolus_macbrydei Arthrosaura_reticulata Kaieteurosaurus_hindsi Pantepuisaurus_rodriguesi Ecpleopus_gaudichaudii

0.2

Figure 7.17. Phylogeny of Ptychoglossus under similarity alignment+maximum likelihood following outgroup expansion. Optimal solution from Analysis 5SA+ML, analysis of data in Analysis 4SA+ML plus eight additional outgroup terminals (details in text and Table 3). Ptychoglossus is polyphyletic.

197

7.14 Tables

Table 7.1. List of PCR and sequencing primers used in this study, and a summary of the PCR conditions.

Gene Primer Sequence (5’–3’) Source PCR protocol region Name 12S 12S1L CAAACTGGGATTAGATACCCCACTAT Kocher et al. 1 cycle: 3 min 94 °C (1989) 33 cycles: 30 s 92 °C, 30 s 57 °C, 1:50 min 72 12S2H AGGGTGACGGGCGGTGTGT °C 1 cycle: 10 min 72 °C 16S 16SF.0 CTGTTTACCAAAAACATMRCCTYTAGC Pellegrino et 1 cycle: 3 min 96 °C al. (2001) 40 cycles: 30 s 95 °C, 1 16SR.0 TAGATAGAAACCGACCTGGATT Whiting et min 51 °C, 1 min al. (2003) 72 °C 16SL CGCCTGTTTAACAAAAACAT Harris et al. 1 cycle: 10 min 72 °C 16SH CCGGTCTGAACTCAGATCACGT (1998)

ND4 ND4L CACCTATGACTACCAAAAGCTCATGTAGAAGC Arévalo et 1 cycle: 3 min 94 °C Leu CATTACTTTTACTTGGATTTGCACCA al. (1994) 33 cycles: 30 s 92 °C, 30 s 57 °C, 1:50 min 72 °C 1 cycle: 10 min 72 °C 1 cycle: 3 min 96 °C 40 cycles: 30 s 95 °C, 1 min 52 °C, 1 min 72 °C 1 cycle: 10 min 72 °C

C-mos G73 GCGGTAAAGCAGGTGAAGAAA Saint et al. 1 cycle: 3 min 96 °C G74 TGAGCATCCAAAGTCTCCAATC (1998) 40 cycles: 30 s 95 °C, 1 min 52 °C, 1 min 72 °C 1 cycle: 10 min 72 °C 1 cycle: 3 min 96 °C 35 cycles: 25 s 95 °C, 1 min 52 °C, 2 min 72 °C 1 cycle: 10 min 72 °C BDNF BDNFF GACCATCCTTTTCCTKACTATGGTTATTTCATACTT Same as C-mos BDNFR CTATCTTCCCCTTTTAATGGTCAGTGTACAAAC PRLR PRLRF1 GACARYGARGACCAGCAACTRATGCC Same as C-mos PRLRR3 GACYTTGTGRACTTCYACRTAATCCAT RAG1 Mart.FL1 AGCTGCAGYCARTAYCAYAARATGTA Same as C-mos Amp.R1 AACTCAGCTGCATTKCCAATRTCA ND2 Metf6(F) AAGCTTTCGGGCCCATACC Leach and Same as ND2 McGuire (2006) AsnR2-R TTGGGTGTTTAGCTGTTAA

198

Table 7.2. GenBank accession numbers for loci and terminals sampled in this study.

Asterisks indicate new sequences obtained for this study. Species are listed following the new taxonomy proposed herein.

Voucher Terminal ID C-mos 12S 16S ND4 Alopoglossus angulatus Brazil 329 Alopoglossus angulatus Ecuador LSUMZH12692 Alopoglossus angulatus Peru1 CORBIDI 2010-4770 Alopoglossus angulatus Peru2 CORBIDI 2010-9944 Alopoglossus angulatus Peru3 CORBIDI 2010-8321 Alopoglossus atriventris Brazil1 LSUMZH 13856 Alopoglossus atriventris Brazil2 INPA-H20301 Alopoglossus atriventris Ecuador QCAZ 5622 Alopoglossus atriventris Peru1 CORBIDI 2010-5969 Alopoglossus atriventris Peru2 CORBIDI 2010-8794 Alopoglossus buckleyi Ecuador MZUTI 4007 Alopoglossus buckleyi Peru1 CORBIDI 2012-10060 Alopoglossus buckleyi Peru2 CORBIDI 2010-9437 Alopoglossus carinicaudatus LG1026 angulatus Brazil 1 Alopoglossus carinicaudatus Uniban 1802 angulatus Brazil2 Alopoglossus carinicaudatus MSB 016 angulatus Brazil3 Alopoglossus carinicaudatus MTR 6280 angulatus Brazil4 Alopoglossus copii Ecuador QCAZ 8314 Alopoglossus festae Ecuador MZUTI 4134 Alopoglossus festae Peru CORBIDI 2010-3709 Alopoglossus sp Brazil1 18643 Alopoglossus sp Brazil2 MTR 10175 Alopoglossus sp Brazil3 18983 Alopoglossus spn Brazil MTR 10.083 Alopoglossus sp Colombia MAR 891 Alopoglossus viridiceps Ecuador QCAZ 10670 Ameivula ocellifera MTR 946089 Arthrosaura reticulata MTR 976977 Ecpleopus gaudichaudii LG 1356 MVZ 232837 FMNH Geocalamus acutus 262014 Kaieteurosaurus hindsi MVZ 232837 VUB 3253 IRSNB Pantepuisaurus rodriguesi 2650

199

KU 218406 KU Pholidobolus macbrydei 218402 Ptychoglossus bicolor Colombia GAC 034 Ptychoglossus brevifrontalis QCAZ 8191 Ecuador1 Ptychoglossus brevifrontalis QCAZ 10577 Ecuador2 Ptychoglossus brevifrontalis QCAZ 10637 Ecuador3 Ptychoglossus brevifrontalis QCAZ 10860 Ecuador4 Ptychoglossus brevifrontalis QCAZ 13683 Ecuador5 Ptychoglossus brevifrontalis QCAZ 13965 Ecuador6 Ptychoglossus brevifrontalis MHNSM MHNSM Ptychoglossus brevifrontalis 01/6 Venezuela Ptychoglossus cf bicolor JJS 438 Colombia1 Ptychoglossus cf bicolor JJS 439 Colombia2 Ptychoglossus festae Colombia1 JDL 30005 Ptychoglossus festae Colombia2 JDL 30048 Ptychoglossus gorgonae Ecuador QCAZ 14866 Ptychoglossus myersi Colombia MAR 1589 Ptychoglossus plicatus Colombia MAR1520 Ptychoglossus romaleos JJS 484 Colombia1 Ptychoglossus romaleos JJS 486 Colombia2 Ptychoglossus sp Brazil1 18843 Ptychoglossus sp Brazil2 19157 Ptychoglossus sp Ecuador1 QCAZ 4648 Ptychoglossus sp Ecuador2 QCAZ 5065 Ptychoglossus stenolepis WB 1916 Colombia1 Ptychoglossus stenolepis WB1923 Colombia2 Ptychoglossus vallensis SSP 071 Colombia1 Ptychoglossus vallensis JJS 576 Colombia2 Ptychoglossus vallensis CAH01 Colombia3 Ptychoglossus vallensis CAH03 Colombia4 Riolama inopinata IRSNB 2680 Riolama leucosticta VUB 3767 Teius teyou REE 150 Tupinambis cuzcoensis1 KU 205023 Tupinambis cuzcoensis2 CHUNB 00485

200

Voucher Terminal ID RAG1 BDNF PRLR ND2 Alopoglossus angulatus Brazil 329 Alopoglossus angulatus Ecuador LSUMZH12692 Alopoglossus angulatus Peru1 CORBIDI 2010-4770 Alopoglossus angulatus Peru2 CORBIDI 2010-9944 Alopoglossus angulatus Peru3 CORBIDI 2010-8321 Alopoglossus atriventris Brazil1 LSUMZH 13856 Alopoglossus atriventris Brazil2 INPA-H20301 Alopoglossus atriventris Ecuador QCAZ 5622 Alopoglossus atriventris Peru1 CORBIDI 2010-5969 Alopoglossus atriventris Peru2 CORBIDI 2010-8794 Alopoglossus buckleyi Ecuador MZUTI 4007 Alopoglossus buckleyi Peru1 CORBIDI 2012-10060 Alopoglossus buckleyi Peru2 CORBIDI 2010-9437 Alopoglossus carinicaudatus LG1026 angulatus Brazil 1 Alopoglossus carinicaudatus Uniban 1802 angulatus Brazil2 Alopoglossus carinicaudatus MSB 016 angulatus Brazil3 Alopoglossus carinicaudatus MTR 6280 angulatus Brazil4 Alopoglossus copii Ecuador QCAZ 8314 Alopoglossus festae Ecuador MZUTI 4134 Alopoglossus festae Peru CORBIDI 2010-3709 Alopoglossus sp Brazil1 18643 Alopoglossus sp Brazil2 MTR 10175 Alopoglossus sp Brazil3 18983 Alopoglossus spn Brazil MTR 10.083 Alopoglossus sp Colombia MAR 891 Alopoglossus viridiceps Ecuador QCAZ 10670 Ameivula ocellifera MTR 946089 Arthrosaura reticulata MTR 976977 Ecpleopus gaudichaudii LG 1356 MVZ 232837 FMNH Geocalamus acutus 262014 Kaieteurosaurus hindsi MVZ 232837 VUB 3253 IRSNB Pantepuisaurus rodriguesi 2650 KU 218406 KU Pholidobolus macbrydei 218402 Ptychoglossus bicolor Colombia GAC 034 Ptychoglossus brevifrontalis QCAZ 8191 Ecuador1

201

Ptychoglossus brevifrontalis QCAZ 10577 Ecuador2 Ptychoglossus brevifrontalis QCAZ 10637 Ecuador3 Ptychoglossus brevifrontalis QCAZ 10860 Ecuador4 Ptychoglossus brevifrontalis QCAZ 13683 Ecuador5 Ptychoglossus brevifrontalis QCAZ 13965 Ecuador6 Ptychoglossus brevifrontalis MHNSM MHNSM Ptychoglossus brevifrontalis 01/6 Venezuela Ptychoglossus cf bicolor JJS 438 Colombia1 Ptychoglossus cf bicolor JJS 439 Colombia2 Ptychoglossus festae Colombia1 JDL 30005 Ptychoglossus festae Colombia2 JDL 30048 Ptychoglossus gorgonae Ecuador QCAZ 14866 Ptychoglossus myersi Colombia MAR 1589 Ptychoglossus plicatus Colombia MAR1520 Ptychoglossus romaleos JJS 484 Colombia1 Ptychoglossus romaleos JJS 486 Colombia2 Ptychoglossus sp Brazil1 18843 Ptychoglossus sp Brazil2 19157 Ptychoglossus sp Ecuador1 QCAZ 4648 Ptychoglossus sp Ecuador2 QCAZ 5065 Ptychoglossus stenolepis WB 1916 Colombia1 Ptychoglossus stenolepis WB1923 Colombia2 Ptychoglossus vallensis SSP 071 Colombia1 Ptychoglossus vallensis JJS 576 Colombia2 Ptychoglossus vallensis CAH01 Colombia3 Ptychoglossus vallensis CAH03 Colombia4 Riolama inopinata IRSNB 2680 Riolama leucosticta VUB 3767 Teius teyou REE 150 Tupinambis cuzcoensis1 KU 205023 Tupinambis cuzcoensis2 CHUNB 00485

202

Table 7.3. Summary of tree-alignment+parsimony (TA+MP) and similarity- alignment+parsimony (SA+MP) analyses of increasingly inclusive datasets to test monophyly of Ptychoglossus. Length is the equally weighted parsimony cost of the optimal tree(s). Datasets include all terminals from the previous dataset plus the listed terminals.

Analysis Ingroup Ingroup Outgroup Outgroup Length Ingroup sample terminals sample terminals mono-, para- size size or monophyly

1 10 spp. Ptychoglossus 4 spp. 18: TA+MP: TA+MP and bicolor (1), P. cf. Alopoglossidae SA+MP: SA+MP: bicolor (2), P. (Alopoglossus 5660 paraphyletic brevifrontalis buckleyi, A. copii, with respect to (11), P. festae (2), A. festae, A. Alopoglossus P. gorgonae (2), viridiceps [1]) P. myersi (1), P. plicatus (1), P. romaleos (2), P. stenolepis (2), P. vallensis (4) 2 – – 6 ssp. 26: TA+MP: TA+MP and Alopoglossidae SA+MP: SA+MP: (Alopoglossus 6232 paraphyletic angulatus, A. with respect to atriventris) Alopoglossus

3 – – 10 spp. 30: TA+MP: TA+MP and Gymnophthalmi SA+MP: SA+MP: dae (Riolama 6996 polyphyletic inopinata [1], R. leucosticta [1], Ecpleopus gaudichaudii [1], Kaieteurosaurus hindsi [1]) 4 – – 14 spp. 34: TA+MP: TA+MP and Gymnophthalmi SA+MP: SA+MP: dae 8348 polyphyletic (Pantepuisaurus rodriguesi [1], Arthrosaura reticulata [1], Pholidobolus macbrydei [1]) Teiidae (Ameivula ocellifera [1])

203

5 – – 17 ssp. 38: TA+MP: TA+MP and Amphisbaenidae 9476 SA+MP: (Geocalamus SA+MP: polyphyletic acutus [1]) 9830 Teiidae (Teius teyou [1], Tupninambis cuzcoensis [2])

204

Chapter 8 CONCLUDING DISCUSSION

8 In this thesis, I combine theoretical, analytical and empirical approaches to advance the field of phylogenetic inference and to address evolutionary questions. First, I discuss the efficacy of median-joining as a method for phylogenetic inference. Second,

I explore the concepts of cladogram and synapomorphy, and their role in achieving phylogenetic explanation. Third, I evaluate the impact of phenotypic evidence on molecular datasets, formulate and test biogeographic hypotheses involving different

Neotropical montane regions, and analyze character evolution. Finally, I develop a theory of outgroup sampling grounded in the logic of scientific discovery and propose a heuristic procedure to limit sampling. For most of these projects I use Neotropical microteiid lizards as model organisms. Some taxonomic novelties arose during the development of my research, which I document herein. Below I summarize the results and main insights from the previous chapters and then highlight a relevant analytical problem to be addressed in the current context of phylogenetic inference.

8.1 Summary of thesis chapters

8.1.1 Chapter 2. On the use of median-joining networks in evolutionary biology.

Median-joining (MJ) was proposed as a method for phylogeographic analysis and is enjoying increasing popularity. In this chapter, I evaluate the efficacy of MJ as a method for phylogenetic inference in general, and phylogeographic analysis in particular. I

205

show that MJ networks are distance-based, unrooted (undirected) branching diagrams with cycles and, therefore, are theoretically untenable for evolutionary inference. That confusion has afflicted research for over 15 years. Specifically, I argue that while the approach offers fast computation and very attractive graphics, it has two obvious shortcomings: its reliance on distance-based phenetics (overall similarity instead of character transformations) and the lack of rooting (no direction or history). Given that evolution involves both change and time, and the absence of rooting removes time

(ancestor–descendant relationships) from the equation, the approach cannot yield defensible evolutionary interpretations. I also discuss the concept of phylogenetic network as opposed to non-evolutionary network.

8.1.2 Chapter 3. Cladograms do not necessarily entail synapomorphies, but synapomorphies falsify cladograms.

Given the confusion that afflicts the implementation of phylogenetic concepts in some methods, in this chapter I discuss the concept of a cladogram (rooted, directed branching diagrams), and argue that cladograms do not necessarily require synapomorphies, but synapomorphies are required to test and ultimately falsify cladograms. I conclude that both cladograms and synapomorphies are required to achieve phylogenetic explanation.

8.1.3 Chapter 4. Lizards of the genus Riama (Squamata: Gymnophthalmidae): The diversity in southern Ecuador revisited.

Having established a theoretically and logically consistent framework in Chapters 2 and 3, I infer the phylogeny of Riama, the most speciose genus of gymnophthalmid

206

lizards, to address questions in analytical techniques, biogeography and character evolution (Chapter 5). However, because it is important to test microevolutionary hypotheses (e.g., species identities) before pursuing macroevolutionary studies, in this chapter I review the diversity of Riama in southern Ecuador, the greatest void in the knowledge of the distribution of Riama. I use morphological and molecular evidence to address taxonomic questions. My investigation results in the description of two new species and the redescription of another one. Further, I discuss variation in dorsal scale relief to propose explicit character-states, which I employ to delimite characters used in Chapter 5.

8.1.4 Chapter 5. Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic evidence on molecular datasets, and the origin of the Sierra Nevada de Santa Marta endemic fauna.

Having made progress on alpha taxonomy and analysis of character variation, in this chapter I test current knowledge of the diversification of Riama as severely as possible by combining new and prior genotypic and phenotypic evidence in a total evidence (TE) analysis. Also, to evaluate the impact of phenotypic evidence on molecular datasets, I compare my TE results with those obtained from analyses of DNA sequence data only.

Although phenotypic evidence comprises only 1.2% of the TE matrix, yet its inclusion alters both the topology of the tree and support values of the clades that do not differ.

This shows the relevance of non-molecular evidence in phylogenetic analyses. I then use the optimal phylogenetic hypothesis to formulate and test biogeographic hypotheses involving different Neotropical montane regions. In particular, I hypothesize that an ancient connection facilitated the exchange of species between the

Sierra Nevada de Santa Marta in Colombia and the Cordillera de la Costa in Venezuela

207

(two montane systems isolated from the Andes). I also use the optimal phylogenetic hypothesis to analyze character evolution. I focus on the evolution of prefrontal scales, a cornerstone of the early genus-level taxonomy of cercosaurine lizards. Finally, I propose a genus-level monophyletic taxonomy that reflects inferred historical relationships. I redefine Riama, resurrect Oreosaurus, erect a new genus (Andinosaura), and return Pantodactylus to the synonymy of Cercosaura.

8.1.5 Chapter 6. Formal recognition of the species of Oreosaurus (Squamata: Gymnophthalmidae) from the Sierra Nevada de Santa Marta, Colombia.

Despite its relevance in understanding historical biogeographic patterns in the

Neotropics, the species of Oreosaurus from the Sierra Nevada de Santa Marta in

Colombia remains undescribed (Chapter 5). In this chapter, I name and describe this new species.

8.1.6 Chapter 7. Outgroup sampling criteria: severity of test, expansion, stability, and alopoglossid lizards.

Outgroup sampling (OGS) is a fundamental step in phylogenetic analysis. It plays a pivotal role in testing hypotheses of ingroup topologies and homologies. However,

OGS has been discussed historically in terms of employing “correct” or “appropriate” outgroup terminals to increase “accuracy” or “reliability” by preventing “errors” such as long branch attraction and “incorrect” ingroup rooting. Due to the absence of a logically consistent analytical framework that guides sampling, in this chapter I develop a theory of OGS grounded in the logic of scientific discovery. Its objective is to test

208

nested hypotheses of ingroup topology and character-state transformations as severely as possible by incorporating outgroup terminals in unconstrained, simultaneous analysis of all included terminals, using background knowledge to select the outgroup terminals that have the greatest chance of refuting those hypotheses. This framework provides a logical basis for sampling and successively increases severity of hypothesis- testing, but it does not provide any grounds for limiting the sample. Therefore, I propose the ancillary procedure of expanding the outgroup sample until hypotheses of ingroup topology and homology become stable (insensitive) to increased sampling, with each expansion guided by the scientific objectives of OGS. This heuristic procedure does not prevent more outgroup terminals from being sampled or guarantees that ingroup hypotheses will remain insensitive to further outgroup expansion, and it has no bearing on the objective support of a given hypothesis. Nevertheless, it provides an empirical basis to limit sampling. I illustrate this procedure using a novel multi-locus DNA sequence dataset for alopoglossid lizards. Finally, I use the optimal phylogenetic hypothesis for alopoglossid lizards to propose a genus-level monophyletic taxonomy.

Specifically, I redefine Alopoglossus and Ptychoglossus, and erect a new genus

(Plicaglossus).

In a by-product of this chapter that was not included herein (Sánchez-Pacheco, S.J.,

Rueda-Almonacid, J.V., Caicedo-Portilla, J.R., Souza, S.M. 2016. First record of

Leposoma caparensis from Colombia, with confirmation for the presence of

Ptychoglossus myersi and P. plicatus (Squamata: Gymnophthalmidae,

Alopoglossidae). Salamandra 52, 53–57), I proposed that unusual phenotypic variation in two syntopic, closely related species of alopoglossid lizards is consistent with the character-displacement hypothesis.

209

8.2 Avenues for future work

Relevant chapters consider some directions for future work to further examine theoretical, analytical and empirical questions addressed throughout this thesis. For example, in Chapter 5 I argue that a defensible method of incorporating meristically continuous variables into phylogenetic inference awaits development. Further, all hypotheses formulated herein—biologic or otherwise—are subject to testing. Thus, rather than outlining potential extensions of this thesis, below I highlight a relevant analytical problem that requires immediate attention in the field of phylogenetic inference.

Epistemologically, increased explanatory power results from including additional evidence (Grant and Kluge, 2003; Kluge, 2004). Therefore, it is essential to include additional evidence for quantitative phylogenetic analysis. This is secondary to the need to review the analytical framework in which cumulative data are being processed.

Continuous development of Next-Generation Sequencing (NGS) techniques results in unprecedented amounts of DNA data for analysis. However, increased ability to produce large quantities of sequence data has not been paralleled by a comparable revision of the analytical techniques, including computational tools (e.g., searching algorithms), to handle these large datasets.

The datasets analyzed in this thesis (Chapters 5 and 7) exemplify the need to accelerate the development of more efficient computational tools under the maximum parsimony optimality criterion (MP). Dynamic homology (i.e., tree-alignment; DH) seeks to combine the two historically disconnected processes of multiple alignment and tree searching into a single-step procedure, that is, DH simultaneously evaluates nucleic acid sequence homologies and trees. The algorithm Direct Optimization (Wheeler,

210

1996) operationalizes the concept of DH. Although a multiple sequence alignment is not performed, it is implied by the cladogram (Wheeler et al., 2006). Wheeler et al.

(2006) hypothesized that “the separation of alignment and cladogram searching (static homology) is less efficient when searching the optimal solution.” Therefore, DH

“generates more efficient explanations of sequence variation (shorter, optimal trees)”.

Different studies have corroborated this hypothesis (Padial et al., 2014; Goicoechea et al., 2016; Chapter 7 of this thesis). However, implementation of DH is computationally challenging. It requires high-performance computing clusters, even to process some thousands of base pairs (bp) per terminal “only” (some NGS techniques can produce millions of bp for analysis!). Consequently, it would not have been possible to thoroughly analyze in a reasonable amount of time the two datasets for Chapter 5 (i.e., total evidence and molecular-only datasets) under this approach using desktop computing. Further, high-performance computing using available algorithms does not guarantee carrying out extra searches that increase rigor. The time requirements for additional, rigorous estimations of clade support in Chapter 5 would have been prohibitively costly, unless each search had been made extremely superficially (details in Chapter 5). The problem is compounded when performing a series of increasingly inclusive analyses to address analytical questions (Chapter 7), as parallel computing is necessary not only for multiple instances of tree searching, but also of clade support estimation. But parallelization is not exclusive to DH. If multiple analyses under static homology are required, desktop computing is inefficient. However, parallelization of available programs in cluster computing is not always possible (e.g., TNT analyses for

Chapter 7 were carried out inefficiently one after another, despite the fact that algorithms implemented in TNT are extremely efficient).

211

Although I exemplify the problem of analyzing increasingly large datasets under MP, there is no reason to believe that other approaches are affected differently. In fact, performing the sampling experiments for Chapter 7 under the Maximum Likelihood optimality criterion (ML) was also computationally challenging. Analyses required high-performance, parallel computing due to the searching for the best-fitting partitions and models per (sub)dataset and the number of replicate analyses for tree searching.

It could be argued that NGS datasets for phylogenetic analysis (i.e.,

“phylogenomics”) are currently being analyzed using available computational tools.

However, these analyses either incorporate only a limited number of terminals

(including relevant outgroup sampling, Chapter 7) or sacrifice analytical rigor for the sake of desired phylogenetic resolution, or both. Streicher et al. (2016) summarized the analytical context in which phylogenomics is taking place. They analyzed an NGS dataset (up to 2.4 million bp) for pleurodont iguanian lizards (a group comprising more than 1020 species divided among 12 families). Their taxon sampling included 35 species only. Eight species of acrodont iguanian lizards with available data were used as outgroups (i.e., outgroup sampling was not guided by scientific objectives; Chapter

7). When analyzing different (sub)datasets they stated that (p. 132–133):

On each concatenated data set, we performed a maximum-likelihood analysis

(single tree search)…applying a single…model across the alignment. In theory,

we could have applied partitions to different subsets of the data. However,

searching for the best-fitting partitions would have been computationally

problematic given the large number of loci involved…Moreover, any existing

heterogeneity in rates should be accounted for (at least in part) by including the

gamma distribution of rates among sites…We used rapid botstrapping feature

212

with an automatic cutoff. Using bootstrap values obtained from rapid bootstrapping…is controversial because they can potentially be inflated or misleading…Given the large sizes of many of our data sets (i.e., computational limitations) and a desire to employ the same methodology when generating nodal support for all concatenated analyses, we had little choice but to use this methodology.

213

Appendices

9 Appendix S1

Material Examined

Riama achlyens: VENEZUELA: Aragua: Rancho Grande (AMNH 137260,

137267-69, 137271–76, 137278–82, 137297). Riama afrania: COLOMBIA:

Antioquia: municipio de Urrao, 13 km northeast on Urrao-Caicedo road, Valle Real,

2350 m (MHNCSJ 1048 holotype, MHNCSJ 801–03, 1044, 1051–52, IAvH-R 3957,

3959–60 paratypes), vereda El Chuscal, quebrada Las Juntas, 2430–2490 m (ICN 9513 paratype). Riama anatoloros: ECUADOR: Napo: La Bonita (USNM 229706–45);

Napo-Pastaza [= Napo]: Abitagua (AMNH 38821–22) Zamora-Chinchipe: cuenca del

Rio Jamboe, Romerillos, Parque Nacional Podocarpus, 1700 m (FHGO 2405). Riama cf. anatoloros: ECUADOR: Zamora-Chinchipe: Las Orquídeas, tepuy, 4 km desde el rio Namgaritza (Gabarra) hacía el tepuy (QCAZ 9169); Yanzatza, parroquia Los

Encuentros, localidad Colibrí (FHGO 8617); Yanzatza, Los Encuentros (EPNH 12689).

Riama aurea: ECUADOR: El Oro: Guanazán, 2789 m (QCAZ 07886 holotype); El

Panecillo, 2775 m (QCAZ 09649–50 paratypes); Guishaguiña, Zaruma (EPNH 06196 paratype); El Chiral (AMNH 18310). Riama cashcaensis: ECUADOR: Bolivar:

Guaranda, 2640 m (KU 135019–21 paratypes); 4.0 km E Guanujo, 2870 m (QCAZ 877 paratype). Riama colomaromani: ECUADOR: Pichincha: 19.8 km W Chillogallo, ca

Quito-Chiriboga rd (KU 221737 paratype); Carchi: 26.9–27.3 km from Maldonado on road to Tulcan (KU 217209); 58 km E Tulcán, 2900 m (QCAZ 4250, 4252). Riama columbiana: COLOMBIA: probably Antioquia: municipio de Sonsón (NRM 1631

214

lectotype, NRM 1633, 1634, 6168 paralectotypes); Caldas: municipio de Villa María, vereda Montaño, 2450 m (MHNUC 0088), predio La Mesa, Bosques de la CHEC, 2640 m (ICN 11295–98), 2600 m (ICN 11299–01); municipio de Neira, vereda La Cristalina, finca La Cristalina, 2300 m (ICN 11302); Quindio: between the haciendas El Brillante and San Julian, vereda San Julian, municipio de Calarcá, 2100 m (ICN 6479); Risaralda:

Santuário de Fauna y Flora Otún Quimbaya (IAvH-R 4941); parque municipal Campo

Alegre, municipio de Santa Rosa de Cabal (IAvH-R 5194). Riama crypta: ECUADOR:

Cotopaxi: Pilaló, 2700 m (KU 121153, 121154 paratypes); 2500 m (KU 135100–02 paratypes, 135103 holotype, 135104–15 paratypes); 2400 m (KU 179455–65 paratypes); 2320 m (KU 196386–89 paratypes); 3 km W Pilaló on Quevedo-Latacunga road (USNM 229638–39 paratypes). Riama hyposticta: COLOMBIA: Nariño: municipio de Barbacoas, corregimiento Altaquer, vereda El Barro, Reserva Natural Rio

Ñambí (PSO-CZ 085). Riama kiziriani: ECUADOR: Azuay: San Antonio, 1900 m

(QCAZ 9667, holotype); El Chorro de Girón, Girón, 2546 m (QCAZ 9607, paratype).

Riama laevis: COLOMBIA: Valle del Cauca: municipio Cumbre, 2000 m (IAv 4916), vereda Chicoral (UV-C 11266); 15 km al oeste del Cairo, base Cerro del Ingles, ca.

2000 m (UV-C 10103). Riama luctuosa: VENEZUELA: Aragua: Rancho Grande

(AMNH 137270, 137277, MCZ 100410, USNM 196336), Parque Nacional Henry

Pittier, Rancho Grande (USNM 259170). Riama meleagris: ECUADOR:

[Tungurahua]: Baños (FMNH 28037–42, 28049 [six specimens]). in error: El Oro:

Machala (USNM 196264–65). Riama oculata: ECUADOR: Pichincha: Nanegal

(USNM 229640), 3 km E of Nanegal Chico (USNM 229642). Cotopaxi: San Francisco de las Pampas (UMMZ 188630). Riama orcesi: ECUADOR: Napo: 12 km W (via road)

Baeza (AMNH 124044 paratype); 31 km N Jondachi, 2190 m (QCAZ 2829, 2835); vertiente del volcán Sumaco, 2200 m (QCAZ 931–40). Riama petrorum: ECUADOR:

215

Morona-Santiago: trail between Sevilla de Oro and Mendez on E slopes of the mountains between Cerro Negro and Pailas (Tambos) (USNM 196266 paratype), San

Vicente, slightly S of W of Limon and 35 km E Gualeceo by road (USN 196268).

Riama raneyi: ECUADOR: Napo: 3.3 km ESE Cuyuja, 170 m (KU 142903 paratype);

Sucumbios: near Santa Barbara (MCZ 175160–62); Napo [= Sucumbios]: inmediate environs of Santa Barbara (USNM 229750); 2 km E of Santa Barbara (USNM 229749);

3 km SW Santa Barbara at bridge (covered) over river (USNM 229748). Sucumbios:

32 km E Julio Andrade on road to Santa Barbara, 2610 m (QCAZ 2827). Carchi: Santa

Bárbara, Santa Bárbara-Guanderal, 2980 m (QCAZ 1379). Riama shrevei: TRINIDAD

& TOBAGO: Horne Tucuche (MCZ 62506–07); El Teluche [in error, probably

Tucuche] (MCZ 100466–68); mt. Tucuche (MCZ 160065–66). Riama simotera:

ECUADOR: Carchi: 14.6 km NW El Carmelo, 3130 m (KU 179478); km 13 carretera a El Carmelo, 3300 m (ICN 9823–34); km 16 Tulcán-Tufino, 3130–3160 m (ICN

9835–36); 15.3 km W Tulcán on road to Tufino, 3080 m (QCAZ 915, 918); km 13 desvío carretera Panamericana, El Ángel (ICN 9837). COLOMBIA: Nariño: municipio de Pupiales (IAvH [formerly IND-R] 1553); municipio de Túquerres, km 10 carretera

Túquerres-Guachucal, hacienda Alsacia, 3140 m (ICN 9817); municipio de Cumbal, km 4 Cumbal-volcán Cumbal, 3260 m (ICN 9818–22). Riama stellae: COLOMBIA:

Nariño: municipio de Barbacoas, corregimiento de Ricaurte, reserva La Planada (PSO-

CZ 102 holotype; PSO-CZ 103 and 109, ICN 12068 [formerly PSO-CZ 108] paratypes).

Riama stigmatoral: ECUADOR: Azuay: Sevilla de Oro (USNM 229644 paratype);

San Vicente, camino a las antennas (QCAZ 11414); Morona-Santiago: Pailas, a Tambo on trail between Sevilla de Oro and Mendez, on E or NE facing slope (USNM 229648 paratype); between Tambos called Cerro Negro and Pailas on trail Sevilla de Oro and

Mendez (USNM 229643 paratype); between Pailas and Mirador, on trail between

216

Sevilla de Oro and Mendez (USNM 229645 paratype), Pailas, a Tambo on trail between

Sevilla de Oro and Mendez, on E or NE facing slope (USNM 229647 paratype); San

Juan Bosco, a posada on trail between Limon (General Plaza) and Gualeceo, slightly S of W of Limon (USNM 229649); El Cruzado, a posada on trail between Limon

(General Plaza) and Gualeceo, slightly S of W of Limon (ca. 0.5 hour up trail from San

Juan Bosco) (USNM 229650). No other data (AMNH 32778). San Jose (AMNH 38820);

Cañar: Mazar, reserva Mazar (QCAZ 7374, 7884, 6657); Biblian, iglesia de Biblian

(QCAZ 9946). Riama striata: COLOMBIA: Boyacá: municipio de Villa de Leyva, sector rural vereda El Roble (IAvH 4895); Pesca (IAvH [formerly IND-R] 0665–66); municipio de Turmequé, vereda Joyagua (MUJ 816); Toquilla, Vadohondo, km 71 carretera Sogamoso-Pajarito (ICN 2800). Cundinamarca: Bogotá (CAS-SUR 8280,

MCZ 14166–67, 16979–80, 16982–83, 17129, 110415–16, USNM 75969, 153974–82,

194744, ICN 2181); Bogotá, salón de clases de la Pontificia Universidad Javeriana

(MUJ 229); Bogotá, instalaciones del Laboratorio de Fauna “Venado de Oro,” vivero

Inderena (IAvH [formerly IND-R] 1100–01, 1499, 1602, 3006); Bogotá, laboratorio de

Fauna Unifem Inderena (IAvH [formerly IND-R] 3130–31, 3934, 4163, 4262); Bogotá, ladera del cerro Guadalupe (IAvH [formerly IND-R] 3503, ICN 2436, 2535, 2537, 2541,

2543–44, 2546); mt. Guadalupe (FMNH 177075–81, 177243–47); Salto del

Tequendama (IAvH [formerly IND-R] 3985); municipio de San Francisco, vereda

Sabaneta (ICN 5991), cerro Cueva Grande, 2590 m (ICN 5737), finca La Quebrada, quebrada El Vino, 2540–2560 m (ICN 9759–65); páramo Cruz Verde, 3100 m (ICN

675–76); municipio de Fómeque (ICN 2232); between Alban and Sasaima, 50 NW

Bogotá, D.C. (MVZ 191880); 6 km S Alban on road to Bogotá D.C. (MVZ 191878); represa del Hato, south of Usme, ca 2800 m (FMNH 165800–03, ICN 2371, 2373,

2375); municipio de Suesca, vereda El Hatillo, microcuenca Santa Helena, 2950 m

217

(MUJ 644–48), 3 km al sur de la laguna Suesca, 2860 m (ICN 7276); no other data

(UMMZ 56448, 56760 [11 specimens], 89417 [seven specimens], 123315, 203767–70,

ICN 2362); PNN Chingaza, 3300 m (IAvH [formerly IND-R] 4241–42), sitio

Monteredondo, 3030 m (MUJ 906–09), sector de Chuza (IAvH [formerly IND-R]

3891–94), embalse cerca del casino (MUJ 228). Santander: SFF Guanenta, alto Rio

Fonce, 2650 m (MUJ 910); municipio de Virolin, Cañaverales km 72 carretera a

Charalá, Rio Cañaverales, 1830 m (ICN 9783). not located: tanques de Vitelma (IAvH

[formerly IND-R] 0649). Riama unicolor: ECUADOR: Carchi: Montufar Atal-

Vinculo, 2700 m (UMMZ 105895–97). Imbabura: Lago de Cuicocha (MCZ 154515–

16, 154628). Pichincha: Quito (MCZ 22154, 164616, 164662, 164665–68, 164670);

Pasochoa volcano forest, 40 km SE Quito, 2800–2880 m (175052–53); Machachi

(QCAZ 758). not located: Chillo (MCZ 21070). not located (QCAZ 6122). Riama vespertina: ECUADOR: Loja: [Pampa] Chitoque, between San Bartolo and Piñas

(AMNH 22130 holotype); Reserva Biológica Utuana, 48.3 km southeast to the type locality, 2600 m (DHMECN 4113–14); Guachaurco, 2824–2958 m (QCAZ 10283,

10286, 10288, 10306–13). Riama vieta: ECUADOR: Guayas: km 85 on Durán-Tambo road (USNM 142601).

10 Appendix S2

Analysis and description of phenotypic characters

The following analysis of character variation and transformation series individuation follows the current taxonomy of cercosaurines.

218

Hemipenial anatomy (characters 0–9)

The hemipenial anatomy of cercosaurines has been documented traditionally in the context of alpha taxonomy (e.g. Uzzell, 1970; Kizirian, 1996; Sánchez-Pacheco et al.,

2011) and interpreted phylogenetically by several authors (e.g. Myers et al., 2009;

Nunes et al., 2014). Although hemipenial morphology of cercosaurines is unquestionably a useful source of evidence of phylogenetic relationships, variation has not been coded as transformations series and incorporated into phylogenetic analyses.

Consequently, I delimited a number of novel hemipenial characters and character-states based mainly on the study of gymnophthalmid hemipenial morphology by Nunes

(2011). Some characters required attention because they were highly prone to artifacts of preparation and/or preservation of the organs (e.g. characters 0 and 9). Thus, I coded the following characters on the basis of distinct character-states as observed only in fully everted and sufficiently filled hemipenes.

Character 0. Shape of hemipenial body (SHB; Figs 4–7): cylindrical = 0; elongated

= 1; conical, with proximal region distinctly thinner than distal and lobes = 2; globose

= 3. Non-additive.

SHB was coded as cylindrical when its lateral borders approximately paralleled to each other in sulcate and asulcate views, and without evidence of curvatures or expansions (Fig. 4; e.g. Riama orcesi; state 0). It was coded as elongated when its lateral borders paralleled each other and its length (height) was more than twice the largest width (Fig. 5; e.g. R. balneator; state 1). Likewise, I considered SHB to be conical when its proximal region was distinctly thinner than its distal region and lobes (Fig. 6; e.g. R.

219

crypta; state 2). Globose hemipenes had lateral borders that were curved uniformly, thus providing the body with a roughly rounded condition (Fig. 7; e.g. R. simotera; state

3). The absence of intermediacy among states suggested non-additivity, and, therefore,

I scored this transformation series as being non-additive.

Character 1. Lobes (Figs 6 and 7): large, distinct from hemipenial body = 0; narrow, indistinct from hemipenial body = 1.

Hemipenial lobes were prominent and distinct from the rest of the organ by projecting considerably from the distal limits of the hemipenial body (Fig. 6; e.g. Riama crypta; state 0). Alternatively, hemipenial lobes were small and indistinct from the rest of the organ by projecting only slightly from the hemipenial body, often making it difficult to delimit lobes from body (Fig. 7; e.g. R. simotera; state 1).

Character 2. Flounce orientation on asulcate face (Figs 4, 7–8): lateral (central vertex directed distally, in contact on the center of asulcate face) = 0; medial (central vertex directed proximally, in contact on the center of asulcate face) = 1; horizontal

(with no vertex) = 2. Non-additive.

All ingroup taxa had flounces on both lateral regions of the hemipenial body, which expand from opposite lateral regions and invade the asulcate face, either reaching the center of it, or not. Species in which the flounces reach the center of the asulcate area were assigned to one of three patterns: flounces continuous forming chevron-shaped lines with the vertex directed either distally (“Λ”; Fig. 7; e.g. Riama anatoloros, R. colomaromani, R. columbiana, R. laevis, R. “Nariño”, R. simotera and R. stigmatoral; state 0) or proximally (“V”; Fig. 8; e.g. R. cashcaensis and R. yumborum; state 1); and

220

flounces forming discontinuous (but nearly complete) horizontal lines, with no vertex

(“– –”; Fig. 4; e.g. R. orcesi, R. raneyi and R. striata; state 2). Species in which the flounces did not reach the center of the asulcate face were coded as inapplicable. No intermediate state(s) in flounce orientation on asulcate face of the hemipenial body existed. Thus, I evaluated this transformation series to being non-additive.

Character 3. Asulcate central nude area (Figs 4–5, 7): absent, flounces extended across entire asulcate face = 0; narrow, restricted to a sagittal stripe = 1; broad, occupying approximately 50% of the asulcate face = 2. Non-additive.

Flounces expanding from opposite lateral regions of the hemipenial body and occupying the asulcate face formed continuous or discontinuous lines on this face. A vertical “nude” area of the asulcate face was formed where the flounces were discontinuous, (nearly) horizontal and equal-length lines (e.g. Fig. 4). Depending on the proximity between flounces from opposite lateral regions, nude areas varied in width. Nude areas were coded narrow when they were restricted to a sagittal stripe (Fig.

4; e.g. Riama orcesi, R. raneyi, R. stigmatoral and R. striata; state 1) and broad if they occupied at least 50% of the asulcate face (Fig. 5; e.g. R. balneator; state 2). When the flounces formed continuous lines that extended across the entire asulcate face, I considered the nude area to be absent (Fig. 7; e.g. R. anatoloros, R. cashcaensis, R. colomaromani, R. columbiana, R. laevis, R. maxima, R. meleagris, R. simotera, R. unicolor and R. yumborum; state 0). The increasing (or decreasing) degree of proximity between flounces from opposite lateral regions was indicative of additivity. However,

I did not treat this character as additive because no ontogenetic evidence suggested that transformations between states 0 and 2 passed through state 1.

221

Characters 2 and 3 were transformationally independent and, as such, were coded separately. No evidence indicated a relationship between flounce (dis)continuity and orientation on the center of the asulcate face. Therefore, I did not consider horizontal flounce orientation (character 2, state 2; “– –”) and a narrow, vertical nude area

(character 3, state 1; “≡ ≡”), as well as continuous, chevron-shaped flounces (character

2, e.g. state 1; “V”) and the absence of vertical nude areas (character 3, state 0; “︾”), as identical character-states.

Character 4. Orientation of lateral body flounces (Figs 4–5, 9): chevron shaped = 0; extended diagonally from anterior (asulcate) to posterior (sulcate) face = 1; horizontal

= 2. Non-additive.

All ingroup taxa had flounces on both lateral regions of the hemipenial body. In most species, these lateral flounces formed chevron-shaped lines with the vertex directed proximally (“V”; Fig. 4; e.g. Riama orcesi; state 0). In other species, the lateral flounces either extended diagonally from the asulcate (anterior) to the sulcate (posterior) face

(“/”; Fig. 5; e.g. R. balneator; state 1) or were directed horizontally (“—”; Fig. 9; e.g.

R. striata; state 2). The absence of intermediacy among states suggested non-additivity.

Character 5. Lateral body flounce ornamentation (Figs 4 and 10): absent = 0; present = 1.

Most gymnophthalmids have calcified structures ornamenting the lateral body flounces. Estes et al. (1988) suggested this character-state was a putative synapomorphy of Gymnophthalmidae. In contrast, based on the topology presented by Pellegrino et al.

222

(2001), Nunes et al. (2014) interpreted the presence of such structures as either a synapomorphy of a less inclusive group within the family (Gymnophthalmidae except

Alopoglossinae—now Alopoglossidae according to Goicoechea et al., 2016) with subsequent independent reversals, or as independently evolved at least twice within

Gymnophthalmidae. Alternatively, based on the topology presented by Castoe et al.

(2004), Nunes et al. (2014) interpreted this condition as the result of three independent origins within the family.

My assessment of the hemipenes revealed the presence of calcified ornamentation on the lateral body flounces (Fig. 4; state 1) in all species of Riama and most cercosaurines. These structures are absent (Fig. 10; state 0) in two cercosaurines

(Neusticurus bicarinatus and N. rudis), Bachia flavescens, Ecpleopus gaudichaudii,

Rachisaurus brachylepis, and the non-gymnophthalmid taxa (Kentropyx calcarata,

Ameivula ocellifera, and Ptychoglossus brevifrontalis).

Character 6. Position of lateral body flounce ornamentation (Figs 4 and 5): distal, restricted to flounce extremities = 0; distributed over entire flounce = 1.

Among the sampled species with ornamented lateral flounces, the calcified structures may be distributed over the entire flounce without interruption (Fig. 4; e.g.

Riama orcesi; state 1), or restricted to the extremities of the flounce, with the rest of the flounce being “nude” (Fig. 5; Echinosaura sulcarostrum, R. anatoloros, R. balneator, and Riolama leucosticta; state 0).

223

Character 7. Shape of lateral body flounce ornamentation (Figs 4 and 5): comb- like series of spicules = 0; isolated hook-shaped spines = 1; calcified lamina = 2. Non- additive.

The calcified ornamentation on the lateral body flounces varied in shape and disposition. Most cercosaurines and a few gymnophthalmines possess small calcified spicules organized in series along each flounce (Fig. 4; e.g. Riama orcesi; state 0), which have been generally referred to as “comb-like” series of spicules (e.g. Myers and

Donnelly, 2001; Rodrigues et al., 2005, 2007; Nunes et al., 2014). Sometimes calcified spines on less conspicuous lateral flounces were enlarged and elongated, isolated and hook-shaped (Fig. 5; Echinosaura sulcarostrum, R. balneator and Riolama leucosticta; state 1). Some species of Gymnophthalmus Merrem, 1820 had expanded calcified lamina, as represented by G. vanzoi (state 2). I analysed this transformation series as non-additive because of the absence of intermediate states.

Character 8. Isolated horizontal flounces on proximal-central region of asulcate face (Figs 4 and 7): absent = 0; present = 1.

Many cercosaurines, including approximately half of the species of Riama included herein, possess isolated, nearly horizontal flounces on the proximal-central region of the asulcate face (Fig. 4; e.g. R. orcesi; state 1). The other species did not have it (Fig.

7; e.g. R. simotera; state 0). Ornamentation occurs on these flounces but interspecific variation in shape and disposition was much more complicated than that observed in lateral body flounces. Because I was unable to delimit transformation series objectively,

I did not code ornamentation of proximal flounces of the asulcate face.

224

Character 9. Distal filiform appendages on the hemipenial lobes (Figs 6 and 7): absent = 0; present = 1.

Sánchez-Pacheco et al. (2011) documented the occurrence of single, distal filiform appendages on the hemipenial lobes of Riama crypta and R. hyposticta (Fig. 6), which they interpreted as a putative synapomorphy uniting these two species. They cautioned that the detection of such structures required the hemipenial lobes to be fully everted, which may not have been the case for the described hemipenes of other putatively related species (e.g. R. afrania; Arredondo and Sánchez-Pacheco, 2010). These appendages were distinctly thinner than the rest of the lobe (Fig. 6). Among other gymnophthalmids, similar appendages have been documented for Cercosaura manicata and an unnamed species of Iphisa Gray, 1851 (Nunes, 2011; Nunes et al.,

2012). However, these differed in length, especially those of Iphisa sp. Köhler and Lehr

(2004: 510, Fig. 7) described and illustrated the hemipenis of Proctoporus laudahnae, and reported the “apex [of the organ] with two large protrusions separated by the distal end of the sulcus spermaticus”. These protuberances seemed to be more integrated with other lobular folds (i.e. less distinct from the rest of the lobe) than those of R. crypta and R. hyposticta, and the hemipenes of four more cercosaurines (R. “Cordillera

Occidental”, Proctoporus bolivianus, P. guentheri, and P. pachyurus) examined herein had similar structures. Apical soft-tissue papillae also occurred atop each lobe in the cercosaurines Anadia ocellata and A. blakei (Myers et al., 2009). However, these structures were paired and distinctly robust. Cercosaura manicata, Iphisa sp., P. laudahnae, A. ocellata, and A. blakei were not included in the current study.

225

Variation in length and degree of integration with the rest of the lobe suggested that the single distal filiform appendages on the hemipenial lobes of Riama crypta and R. hyposticta may not be homologous with those of R. “Cordillera Occidental”,

Proctoporus bolivianus, P. guentheri and P. pachyurus. It was also possible that two transformation series have been conflated under “present”, i.e. one involving occurrence of the appendages, the other variation in their length. For the purpose of the present phylogenetic analysis, I recognized only two character-states: single, distal filiform appendages on the hemipenial lobes absent (Fig. 7; e.g. R. simotera; state 0) and single, distal filiform appendages on the hemipenial lobes present (Fig. 6; e.g. R. crypta; state 1).

External morphology (scutellation, characters 10–34)

Traditionally, variation in scutellation has been used descriptively and comparatively in taxonomic studies of cercosaurines. Because of this, I reviewed its usage in the context of cercosaurine phylogenetic systematics. I defined a number of novel characters and states (characters 10, 13, 14, 20, 21, 28, 30, 31), modified a series of characters that have been used in previous phylogenetic analyses (11, 12, 15, 18, 19,

22–27, 29, 32–34), and perpetuated the use of two characters (16 and 17).

Character 10. Minute tubercles on dorsal head scales: absent = 0; present = 1.

In state 0, all dorsal head scales were smooth and glossy because they lacked minute tubercles or granules (e.g. Petracola spp., Proctoporus spp., Riama spp.). State 1 consisted of dense rounded, minute tubercles or granules on the dorsal head scales and

226

was confined to some species in the outgroup (Neusticurus spp., Potamites spp.,

Ameivula ocellifera, and Kentropyx calcarata).

Character 11. Prefrontal scales occurrence: absent = 0; present = 1.

Because occurrence of prefrontals has been used to diagnose genera and species, and to hypothesize phylogenetic relationships, it has played a pivotal role in cercosaurine systematics. The absence of prefrontals has been used to diagnose the former

Proctoporus s.l. (e.g. Peters and Donosos-Barros, 1970), and interpreted as a synapomorphy of this group (Doan, 2003a). Although prefrontals vary in size and degree of medial contact, which was the basis of Doan’s (2003b; character 3) coding of this character, prefrontal scales were absent in all ingroup species and some relevant outgroup taxa (e.g. Petracola spp., most of Proctoporus spp.) (but see below).

Therefore, I scored only the absence (state 0) and presence (state 1) of prefrontals, despite the occurrence of conflated transformation series under “present”.

Intraspecific variation in the occurrence of prefrontal scales was found to rarely occur in cercosaurines. An undescribed species referred to herein as Riama “Cordillera

Occidental”, as well as Macropholidus huancabambae, Pholidobolus macbrydei and

Proctoporus sucullucu exhibited polymorphism, though the presence of prefrontal scales is rare in R. “Cordillera Occidental” and P. sucullucu. Therefore, I coded these species as polymorphic (absent/present).

Characters 12–14. Nuchal scale relief

227

Dorsal scale relief has been used descriptively and comparatively in alpha taxonomic studies of cercosaurines, and of Riama in particular (e.g. Kizirian, 1996). Studies have rarely distinguished posterior and nuchal scales probably because of the usually uniform relief of dorsal scales, such as when both nuchal and posterior scales were keeled. However, the texture of posterior dorsal scales and nuchal scales sometimes differed. For example, Sánchez-Pacheco (2010a) noted that the nuchals of R. striata and two undescribed species were rugose but the posterior scales were striated. Species of Riama had smooth nuchals but keeled or striated posterior dorsals (Sánchez-Pacheco et al., 2011, 2012; Aguirre-Peñafiel et al., 2014). Thus, the textures of nuchal and posterior dorsal scales have independent transformations, which further examination of additional ingroup and outgroup taxa reinforced. Therefore, I coded nuchal (characters

12–14) and posterior dorsal scale relief (characters 22–28) separately.

The texture of nuchal scales has been noted sporadically in diagnoses and descriptions. Uzzell (1958) first reported that the nuchal scales of R. achlyens and R. luctuosa (not included herein) were keeled. Sánchez-Pacheco (2010a) reported the nuchals of R. “Cordillera Occidental” (R. “sp. 2” in that study), R. “Cordillera Central”

(R. “sp. 1”), R. striata, R. vieta and R. stellae as being rugose. Sánchez-Pacheco (2010b),

Sánchez-Pacheco et al. (2011, 2012) and Aguirre-Peñafiel et al. (2014) described the nuchal scales of R. aurea, R. columbiana, R. crypta, R. hyposticta, R. kiziriani and R. yumborum as smooth. In general, character states related to scale relief, including the smooth condition, have been combined into a single character under the assumption that they represented a unique transformation series (e.g. Doan, 2003a: character 46;

Rodrigues et al., 2005: characters 10 and 11). However, this oversimplification of coding involved independent transformations. For example, the absence of keels or rugosities (or striations in the case of posterior dorsal scales), does not necessarily imply

228

smoothness. In contrast, the absence of all these different conditions determines smoothness. In addition, minute tubercles and keels may occur together on nuchal scales (e.g. Neusticurus spp., Placosoma cordylinum), or may not (e.g. Potamites spp., only tubercles; R. achlyens and R. shrevei, only keels). Therefore, I coded the different conditions as independent characters. Kizirian (1996), Sánchez-Pacheco (2010a) and

Sánchez-Pacheco et al. (2012) discussed variation in dorsal scale relief among species of Riama. They proposed explicit character-states, which I employed in delimitating characters related to nuchal scale relief, specifically characters 12 and 13.

Character 12. Keels on nuchal scales: absent = 0; present = 1.

Following Sánchez-Pacheco et al. (2012), state 1 involved a (generally weak) keel in the middle of the nuchal scale, not flanked by striations (e.g. Riama achlyens).

Species lacking such keels were scored as state 0. No intraspecific variation in this character was detected.

Character 13. Rugosities on nuchal scales: absent = 0; present = 1.

Kizirian (1996) and Sánchez-Pacheco (2010a) referred the rugose condition to scales with multiple, minute, nearly longitudinally positioned ridges (state 1; e.g. Riama striata). Species without rugosities were scored as state 0. Intraspecific variation occurred in Pholidobolus macbrydei, which was coded as absent/present.

Character 14. Minute tubercles on nuchal scales: absent = 0; present = 1.

229

Species with rounded, minute tubercles or granules distributed densely over the nuchal scales were scored as state 1. Scale texture appeared granular under high magnification. This character-state only occurred in the outgroup species Neusticurus spp., Potamites spp., Placosoma spp., and Kentropyx calcarata. Species lacking these tubercles were scored as 0. No intraspecific variation was detected.

Character 15. Palpebral disc of lower eyelid: undivided = 0; divided = 1.

The palpebral disc of the lower eyelid has long been used in gymnophthalmid systematics (e.g. Boulenger, 1885). Uzzell (1958, 1970) used this structure to diagnose species groups in the former Proctoporus s.l. Most gymnophthalmids possess fully movable eyelids. The palpebral disc of their lower eyelid is either undivided (e.g. most

Proctoporus s.s. spp., Macropholidus spp.; state 0) or divided into several scales (e.g.

Riama spp., Petracola spp.; state 1). Potamites juruazensis exhibited intraspecific variation and I coded this species as polymorphic (single/divided). Doan (2003a: character 13) noted interspecific variation in the number of scales in the palpebral disc of the lower eyelid. I scored only the condition of the palpebral disc—undivided or divided—because, when divided, variation was continuous (i.e. meristic), which must be accounted for when individuating characters (see Materials and Methods:

Phenotypic evidence). Some gymnophthalmids lack eyelids (tribe Gymnophthalmini

Fitzinger, 1826, except Tretioscincus Cope, 1862), including Gymnophthalmus vanzoi, the only eyelidless gymnophthalmid included herein. This species was coded as missing (“–”).

Character 16. Frontonasal scale: undivided = 0; divided = 1.

230

All species of Riama possess an undivided frontonasal scale (state 0), and it is both undivided and divided (state 1) in my outgroup species. Differences occur in frontonasal division. Vertical divisions into two scales occurs in some taxa (e.g.

Cercosaura argula and C. oshaughnessyi), and into three scales in others (e.g.

Echinosaura sulcarostrum). Only several outgroup taxa exhibit variation. Therefore, I treated the division of frontonasal scale as a single character-state. This character corresponded to character 1 of Doan (2003b).

Character 17. Relative frontonasal length: shorter than frontal = 0; same length as frontal = 1; longer than frontal = 2. Non-additive.

Given the consistency of frontal length, this reference point was used to compare interspecific variation in the size of frontonasals. Following Kizirian (1996) and character 2 of Doan (2003a,b), coding of the frontonasal length was based on it being shorter (state 0), equal to (state 1), or longer than the frontal (state 2). The considerable intraspecific variation that occurred within Riama was accounted for by polymorphic coding (e.g. 0/1/2). Although state 1 was intermediate in the degree of frontonasal

“expansion”, no ontogenetic evidence suggested that transformations between states 0 and 2 passed through state 1. Further, the frontonasal was either shorter or longer than the frontal in some species (e.g. R. laevis). Therefore, I evaluated this character as being non-additive.

Character 18. Nasoloreal suture: absent = 0; incomplete = 1; complete (i.e. loreal scale present) = 2. Non-additive.

231

Occurrence of the loreal scale usually has been reported as absent or present.

However, Kizirian (1996: 89) pointed out that in Riama “[t]he nasoloreal suture can be absent (= loreal absent), complete (= loreal present), or incomplete. In some cases, where the suture is incomplete, the scale is referred to as a nasoloreal scale [= loreal absent]”. Doan (2003a: 373) modified the terminology by coding this character as

“nasal condition: undivided 0 (i.e. nasal and loreal fused), incompletely divided 1, divided into separate nasal and loreal scales 2”. Although Doan’s intention was the same as Kizirian’s, her character delimitation described the observed variation somewhat less precisely. For this reason, my coding followed Kizirian.

In state 0, specimens lacked a nasoloreal suture. When present, the nasoloreal suture extended from the supralabial or frenocular scales part-way to the frontonasal scale

(state 1), or all the way to the frontonasal (state 2), forming the loreal scale. For illustrations of these three states, see Kizirian (1996). State 1 was intermediate in the degree of suture development, yet no ontogenetic evidence suggested it was the transformational state between states 0 and 2. In some species the nasoloreal suture is either absent or complete (e.g. Riama meleagris). Therefore, the character was evaluated as non-additive.

Unlike most cercosaurine and gymnophthalmid genera, the degree of development of the nasoloreal suture varied considerably among conspecifics of Riama. For example, whereas intraspecific variation has been reported for only three of the outgroup taxa included herein, it has been detected in 16 of the sampled ingroup species. Thus, I coded these terminals as polymorphic (0/1, 0/2, 1/2, 0/1/2). Interspecific variation was common in Riama, but less so in the outgroup genera. Incomplete nasoloreal sutures

232

occurred in some Riama spp., several specimens of Proctoporus bolivianus (Uzzell,

1970; Goicoechea et al., 2013) and the holotype of Riolama leucosticta (Uzzell, 1973).

Character 19. Scale organs on labials: absent = 0; present = 1.

Sensory scale organs occurred on the labial scales in all species of Riama. Among the sampled outgroup taxa, scale organs on labials were both present (e.g. Proctoporus spp., Petracola spp.; state 1) and absent (e.g. Neusticurus spp., Placosoma spp.,

Rhachisaurus brachylepis; state 0). No intraspecific variation was detected. Doan

(2003a,b) treated the variation in number of scale organs on the first supralabial and the first infralabial as characters 24 and 25, respectively. I did not follow Doan’s coding for several reasons. First, if present, minute scale organs occur in most, if not all labials, and I do not consider exclusion of the remaining labials to be precise. Likewise, delimitation of the number of scale organs between the first infralabial and the first supralabial is arbitrary, and I do not consider a distinction between both scales to constitute different transformations series. Second, variation is continuous (i.e. meristic), which must be accounted for when individuating transformations series (see

Materials and Methods: Phenotypic evidence). Third, as coded by Doan, the character excludes absence of scale organs on labials, which was precise in that study given the taxon sampling (only species with scale organs), but it does not consider my observed variation.

Character 20. Association of supralabial and subocular: separated = 0; fused = 1.

233

Kizirian (1996) diagnosed Riama labionis based on the “supralabial–subocular fusion”, and summarized observed variation (both intra- and interspecific) in

Ecuadorian species as absent or present. Subsequent alpha taxonomic studies of some cercosaurine genera (e.g. Petracola, Riama) have reported the supralabial–subocular association. Although Kizirian described the holotype of R. labionis as having the

“subocular fused to fourth supralabial”, he did not explicitly mention which subocular was fused, making it difficult to recognize homologous fusions in other species.

However, figure 12 of Kizirian (1996) showed the third subocular and fourth supralabial fused. Kizirian noted that some specimens of R. cashcaensis, R. raneyi and

R. simotera exhibited the supralabial–subocular fusion. Close examination of specimens of these species (Appendix 3) revealed that, when fusion occurs, variation encompasses only the supralabial being fused with the subocular (fourth supralabial in

R. simotera and R. caschaensis, fifth supralabial in R. raneyi). In all cases, the third subocular fused with the fourth or fifth supralabial.

Among the included outgroup taxa, this unusual fusion has been detected in Anadia mcdiarmidi (Kok and Rivas, 2011), and Riolama leucosticta (Molina and Señaris, 2003:

16, fig. 3). In both cases, the fifth supralabial fused with the third subocular, as in specimens of Riama raneyi. Terminology has varied in the literature. Molina and

Señaris (2003: 15, translated freely from the Spanish) termed the fusion as “the supralabial series interrupted by a subocular scale”. Kok and Rivas (2011: 5) stated

“one [subocular] scale slightly protruding downward between 4th and 5th supralabial

[sic]”.

Variation involving the supralabials being fused with the third subocular suggested non-homology. However, I treated the variants as homologous and considered all fusions as a single character-state. When present, the second and third supralabials

234

fused in an elongated “second” supralabial in Riama labionis (Kizirian, 1996: 115, fig.

12), R. cashcaensis and R. simotera. Hence, the supralabial scale that fused with the third subocular was the same as in the other species. Herein, I delimited this character as an association of the third subocular and the fourth or fifth supralabials as follows: separated (state 0), and fused (state 1). Intraspecific variation has been detected in seven species of Riama (Kizirian, 1996; Sánchez-Pacheco et al., 2012; this study); these were scored as polymorphic (separated/fused).

Character 21. Association of anterior supraocular and anterior superciliary: separated = 0; fused = 1.

In most cercosaurines, a suture separates the anterior supraocular and anterior superciliary medially (state 0). When an enlarged scale occupied the area shared by both scales (e.g. Petracola spp., some Proctoporus spp.), I, like some other authors

(Uzzell, 1973; Kizirian, 1996; Köhler and Lehr, 2004; Kizirian et al., 2008), interpreted this as evidence of fusion (state 1). This state has also been referred to as the first superciliary expanding onto the dorsal surface of the head (e.g. Uzzell, 1970;

Goicoechea et al., 2013). Intraspecific variation (0/1) has been observed in Proctoporus lacertus (Goicoechea et al., 2013), Riolama leucosticta (Uzzell, 1973), and Riama

“Cordillera Central” (this study), the only species of Riama in which supraocular– superciliary fusion was detected.

Characters 22–28. Dorsal scale relief.

235

In addition to the keels, rugosities and minute tubercles described in characters of nuchal scale relief (12–14), there is interspecific variation in posterior dorsal scale relief involving striations and longitudinal rows of tubercles, as well as the morphology of keels and striations. Therefore, I individuated additional transformations series encompassing the observed variation. Kizirian (1996), Sánchez-Pacheco (2010a) and

Sánchez-Pacheco et al. (2012) discussed variation in dorsal scale relief among species of Riama. They proposed explicit character-states that I employed in the delimitation of the following characters (specifically characters 23–27).

Character 22. Minute tubercles on posterior dorsal scales: absent = 0; present = 1.

Species with rounded, minute tubercles or granules distributed densely over the posterior dorsal scales, making scale-texture appear granular under high magnification, were scored as state 1. This character-state only occurred in the outgroup (Neusticurus spp., Potamites spp., Placosoma spp., and Kentropyx calcarata). Species that lacked minute tubercles on posterior dorsals were scored as 0. No intraspecific variation in this character was detected.

Character 23. Keels on posterior dorsal scales: absent = 0; present = 1.

Following Sánchez-Pacheco et al. (2012), presence of a keel in the middle of the dorsal scale, not flanked by striations, was scored as state 1. Species that lacked these keels were scored as state 0. Intraspecific variation was detected in some species of

Pholidobolus, Proctoporus and Riama. I coded these occurrences as polymorphisms

(absent/present).

236

Character 24. Morphology of keels on posterior dorsal scales: weak (low, rounded keel) = 0; strong (prominent keel) = 1.

When present, keels on posterior dorsals were weak (i.e. low and rounded; state 0) or strong (i.e. prominent). When present, keels in species of Riama were usually weak but several species had strong keels (e.g. R. achlyens, R. shrevei and R. “Venezuela”).

Intraspecific variation was not detected. Species without keels on posterior dorsal scales were coded as inapplicable.

Character 25. Striations on posterior dorsal scales: absent = 0; present = 1.

Following Kizirian (1996) and Sánchez-Pacheco et al. (2012), the striated condition referred to two centrally positioned, longitudinal and parallel furrows (state 1). Species without these striations were scored as state 0. Intraspecific variation occurred in some species of Pholidobolus, Proctoporus and Riama. These species were coded as polymorphic (absent/present).

Character 26. Morphology of striations on posterior dorsal scales: weak (shallow furrows) = 0; strong (deep furrows) = 1.

When present, striations on posterior dorsals were either weak (shallow furrows; state 0) or strong (deep furrows; state 1). When present, the striations were usually weak in species of Riama. However, several species exhibited strong striations (e.g. R. striata

237

and undescribed R. “Cordillera Occidental” and R. “Cordillera Central”). Intraspecific variation was not detected. Species without the striations were coded as inapplicable.

Character 27. Rugosities on posterior dorsal scales: absent = 0; present = 1.

Kizirian (1996) and Sánchez-Pacheco (2010a) considered the rugose condition to indicate scales with multiple, minute, nearly longitudinally positioned ridges (state 1).

This condition occurred in the posterior dorsals of Riama vieta and R. stellae only (one of the five species of Riama not included in this study). The remaining ingroup and outgroup species were scored as state 0 (absent). Intraspecific variation was not observed.

Character 28. Longitudinal rows of tubercles on dorsum: absent = 0; present = 1.

Longitudinal rows of tubercles on the dorsum (state 1) were confined to the outgroup species Neusticurus spp. and Potamites spp. The remaining outgroup and ingroup species were scored as state 0 (absent). Intraspecific variation was not detected.

Characters 29–30. Anterior row of cloacal plate.

Kizirian (1996) reviewed the occurrence of, and variation in the anterior cloacal plate row of Ecuadorian Riama. The different conditions have been reported consistently for both ingroup taxa (e.g. Sánchez-Pacheco, 2010b; Aguirre-Peñafiel et al., 2014) and outgroup species (e.g. Goicoechea et al., 2013; Proctoporus spp.). Doan

(2003a,b: character 36) coded variation in cloacal plate rows as “number of cloacal

238

plate rows (1 or 2)”, where “1” referred to the absence of the anterior row. I did not follow this coding because it lacked explicit delimitation of character-states, and two transformations series were combined under her character state “2” (i.e. anterior row present; see below).

Character 29. Anterior cloacal plate row: absent = 0; present = 1.

The anterior cloacal plate row was either absent (e.g. Bachia flavescens and

Ecpleopus gaudichaudii; state 0) or present (e.g. Riama columbiana; state 1). Rivas et al. (2005: 463, fig. 2; absence (top), presence (bottom)) and Aguirre-Peñafiel et al.

(2014: 252, fig. 3; absence (left), presence (right)) illustrated examples. Some species of Riama exhibited intraspecific variation, which was coded as a polymorphism (0/1).

Character 30. Condition of anterior cloacal plate row: one scale = 0; paired scales

= 1.

When present, the anterior cloacal plate row was composed of either a small scale

(e.g. Riama meleagris and R. shrevei; state 0) or two large scales (e.g. R. crypta; state

1). Species without the anterior plate row were coded as inapplicable. Kizirian (1996:

102, fig. 5; a single small scale (bottom), two large scales (top)) illustrated examples.

Intraspecific variation in Ptychoglossus brevifrontalis and several species of Riama was coded as a polymorphism (0/1).

Character 31. Rugosities on ventral scales: absent = 0; present = 1.

239

Rugose ventral scales are extremely rare in Gymnophthalmidae. Herein, only Riama vieta and undescribed R. “Cordillera Occidental” possessed rugosities (as defined in characters 13 and 27; state 1) on their ventrals. Not included herein, R. stellae had rugose ventrals. All remaining ingroup and outgroup taxa were scored as state 0.

Intraspecific variation was not detected.

Character 32. Dorsal scale shape: approximately quadrangular = 0; rectangular = 1; hexagonal or sub-hexagonal = 2; irregular = 3. Non-additive.

The shape of the dorsal scales has been used in gymnophthalmid systematics for decades (e.g. Boulenger, 1885). Considerable variation occurred within Cercosaurinae.

My species possessed roughly quadrangular (e.g. Placosoma spp.; state 0), rectangular

(e.g. Proctoporus spp.; state 1), hexagonal or sub-hexagonal (e.g. Pholidobolus spp.; state 2), or irregular (e.g. Potamites spp.; state 3) dorsals. Most species of Riama exhibited state 1, whereas a few taxa from Venezuela and Trinidad had state 2.

Intraspecific variation was detected only in Kentropyx calcarata, which was coded accordingly (2/3). This character corresponded to character 45 of Doan (2003a,b), except for the inclusion of irregularly shaped scales and the exclusion of the states

“rhomboid” and “pyramidal”, presumably due to differences in taxon sampling and observed variation. The absence of intermediate states suggested non-additive analyses.

Characters 33–34. Occurrence of femoral pores.

The occurrence and number of femoral pores are two of the most useful characters for discriminating species of Riama. These two characters are sexually dimorphic, with

240

males usually possessing more femoral pores than females. In some species, females lacked them (Kizirian, 1996). Therefore, I coded males and females as separate semaphoronts. Distinction between preanal and femoral pores has been commonly used in cercosaurine systematics. As such, Doan (2003a,b: characters 40–42) coded preanal and femoral pores separately. However, explicit discrimination of preanal and femoral pores has been generally lacking. Kizirian (1996: 92) stated “[p]reanal pores are femoral pores that occur medially, inside a line congruent with the outside edge of the tail. The presence of femoral pores that are preanal in position is actually difficult to determine”. Consequently, the distinction between preanal and femoral pores has been arbitrary. As pointed out by Grant et al. (2006: 66) “[a]lthough such arbitrariness is relatively harmless in descriptive taxonomic studies, the cumulative effect of arbitrary delimitations can be disastrous in phylogenetic analyses.” Therefore, I considered the complete pore series to be femoral pores. In addition, femoral pores varied in number, and Doan (2003a,b) coded this variation accordingly. However, frequencies do not entail additional character-state transformations and mistakenly equate population- level similarity with transformation events (see Materials and Methods: Phenotypic evidence). Therefore, I scored only the absence and presence of femoral pores, despite the fact that I have conflated additional transformation series under “present”.

Character 33. Femoral pores in males: absent = 0; present = 1.

Males usually had femoral pores (e.g. Riama spp.; state 1). Nevertheless, several male cercosaurines lacked femoral pores (e.g. Macropholidus huancabambae,

Pholidobolus montium and P. prefrontalis; state 0). Intraspecific variation in M. annectens, P. affinis and P. macbrydei was coded as a polymorphism (absent/present).

241

Character 34. Femoral pores in females: absent = 0; present = 1.

Females lacked femoral pores (e.g. Riama cashcaensis and R. orcesi; state 0) or had them (e.g. R. achlyens and R. oculata; state 1). Intraspecific variation was more common than in males (e.g. eight species of Riama). Therefore, I coded these cases as polymorphisms (absent/present).

11 Appendix S3

Specimens examined

The following list of specimens examined includes material used to individuate the transformation series and score the character-states, as well as for species identification and generic allocation. Ingroup species are listed following the new taxonomy proposed below. See Materials and Methods for institutional abbreviations.

Ingroup taxa

Andinosaura

A. afrania: COLOMBIA: Antioquia: municipio de Urrao, 13 km NE on Urrao-

Caicedo road, Valle Real, 2350 m (MHNCSJ 1048 (holotype), MHNCSJ 801–03, 1044,

1051–52, IAvH-R 3957, 3959–60 (paratypes)), vereda El Chuscal, quebrada Las Juntas,

2430–2490 m (ICN 9513 (paratype)). A. aurea: ECUADOR: El Oro: Guanazán, 2789

242

m (QCAZ 07886 (holotype)); El Panecillo, 2775 m (QCAZ 09649–50 (paratypes));

Guishaguiña, Zaruma (EPNH 06196 (paratype)); El Chiral (AMNH 18310). A. crypta:

ECUADOR: Cotopaxi: Pilaló, 2700 m (KU 121153–54 (paratypes)); 2500 m (KU

135100–02 (paratypes), 135103 (holotype), 135104–15 (paratypes)); 2400 m (KU

179455–65 (paratypes)); 2320 m (KU 196386–89 (paratypes)); 3 km W Pilaló on

Quevedo-Latacunga road (USNM 229638–39 (paratypes)). A. hyposticta: COLOMBIA:

Nariño: municipio de Barbacoas, corregimiento Altaquer, vereda El Barro, Reserva

Natural Río Ñambí (PSO-CZ 085). A. kiziriani: ECUADOR: Azuay: San Antonio,

1900 m (QCAZ 9667 (holotype)); El Chorro de Girón, Girón, 2546 m (QCAZ 9607

(paratype)). A. laevis: COLOMBIA: Valle del Cauca: municipio Cumbre, 2000 m

(IAvH 4916), vereda Chicoral (UV-C 11266); 15 km al oeste del Cairo, base cerro del

Ingles, ca. 2000 m (UV-C 10103). Risaralda: río San Rafael, Municipio de Tatamá,

2250m (IAvH 5197-98); quebrada Risaralda, vereda Planos de San Rafael, Parque

Municipal San Rafael, municipio de Santuario (IAvH 5199, 5200); quebrada San

Rafael, municipio de Santuario, 2820m (IAvH [formerly IND-R] 3967). A. oculata:

ECUADOR: Pichincha: Nanegal (USNM 229640), 3 km E of Nanegal Chico (USNM

229642). Cotopaxi: San Francisco de las Pampas (UMMZ 188630). A. petrorum:

ECUADOR: Morona-Santiago: trail between Sevilla de Oro and Mendez on E slopes of the mountains between Cerro Negro and Pailas (tambos) (USNM 196266 (paratype));

San Vicente, slightly S of W of Limon and 35 km E Gualeceo by road (USNM 196268).

A. stellae: COLOMBIA: Nariño: municipio de Barbacoas, corregimiento de Ricaurte, reserva La Planada (PSO-CZ 102 (holotype), PSO-CZ 103 and 109, ICN 12068

[formerly PSO-CZ 108] (paratypes)). A. vespertina: ECUADOR: Loja: [Pampa]

Chitoque, between San Bartolo and Piñas (AMNH 22130 (holotype)); reserva biológica

Utuana, 48.3 km southeast to the type locality, 2600 m (DHMECN 4113–14);

243

Guachaurco, 2824–2958 m (QCAZ 10283, 10286, 10288, 10306–13). A. vieta:

ECUADOR: Guayas: km 85 on Durán-Tambo road (USNM 142601).

Oreosaurus

O. achlyens: VENEZUELA: Aragua: Rancho Grande (AMNH 137260, 137267–69,

137271–76, 137278–82, 137297). O. “Venezuela”: VENEZUELA: Anzoátegui: Cerro

El Guamal, Macizo del Turimiquire, municipio Freites, 2150 m (EBRG 5962). O. luctuosus: VENEZUELA: Aragua: Rancho Grande (AMNH 137270, 137277, MCZ

100410, USNM 196336), Parque Nacional Henri Pittier, Rancho Grande (USNM

259170). O. “Sierra Nevada”: COLOMBIA: Magdalena: Santa Marta, corregimiento

Minca, finca Vista Hermosa, cabecera río Guachacos, 2156 m (seven specimens awaiting assignment of permanent collection number). O. rhodogaster: VENEZUELA:

Sucre: between Las Melenas and Cerro Humo, Península de Paria, ca. 650 m (MHNLS

16645 (holotype), 15730–31 (paratypes)). O. shrevei: TRINIDAD & TOBAGO: Horne

Tucuche (MCZ 62506–07); El Teluche [in error, probably Tucuche] (MCZ 100466–

68); Mt. Tucuche (MCZ 160065–66).

Riama

R. anatoloros: ECUADOR: Napo: La Bonita (USNM 229706–45); Napo-Pastaza [=

Napo]: Abitagua (AMNH 38821–22). Zamora-Chinchipe: cuenca del río Jamboe,

Romerillos, Parque Nacional Podocarpus, 1700 m (FHGO 2405). R. balneator:

ECUADOR: Tungurahua: San Antonio mountains, eastern slope of the Tungurahua volcano, 10.5 km north of the type locality (DHMECN 04111–12). R. cashcaensis:

244

ECUADOR: Bolivar: Guaranda, 2640 m (KU 135019–21 (paratypes)); 4.0 km E

Guanujo, 2870 m (QCAZ 877 (paratype)). R. colomaromani: ECUADOR: Pichincha:

19.8 km W Chillogallo, ca. Quito-Chiriboga rd (KU 221737 (paratype)). Carchi: 26.9–

27.3 km from Maldonado on road to Tulcán (KU 217209); 58 km E tulcán, 2900 m

(QCAZ 4250, 4252). R. columbiana: COLOMBIA: probably Antioquia: Municipio de

Sonsón (NRM 1631 (lectotype), NRM 1633, 1634, 6168 (paralectotypes)). Caldas: municipio de Villa María, vereda Montaño, 2450 m (MHNUC 0088), predio La Mesa, bosques de la CHEC, 2640 m (ICN 11295–98), 2600 m (ICN 11299–301); municipio de Neira, vereda La Cristalina, finca La Cristalina, 2300 m (ICN 11302). Quindio: between the haciendas El Brillante and San Julian, vereda San Julian, municipio de

Calarcá, 2100 m (ICN 6479). Risaralda: Santuário de Fauna y Flora Otún Quimbaya

(IAvH-R 4941); parque municipal Campo Alegre, municipio de Santa Rosa de Cabal

(IAvH-R 5194). R. “Cordillera Central”: COLOMBIA: Antioquia: 7 km N (by road)

Santa Rosa de Osos on road to Yarumal (MVZ 190563); municipio de Yarumal, 3.5 km N por La carretera a Llanos de Cuivá, 2700 m (ICN 6471–78); municipio de

Yarumal, 5 km N por La carretera a Llanos de Cuivá, 2650 m (ICN 6480–89); municipio de Yarumal, Llanos de Cuivá, 1–2 km hacia San José de las Montañas (ICN

11262); San Antonio de Prado, Medellin, 2000 m (MLS 1223); municipio de Santa

Rosa de Osos, vereda Vallecitos, El Ventiadero (MHUA 10573); municipio de Caldas,

Alto de San Miguel, vereda La Clara (MHUA 10009); Medellin (AMNH 32706–07,

89833); Santa Rosa (AMNH 19960–68, 32708–13). Tolima: 15 km W Cajamarca, 2350 m (KU 169940–42); Quindio mts [Quindio–Tolima border?] (MCZ 15942–47, 15949–

52); No other data (UMMZ 56443 [five specimens]). Not located (AMNH 32764–65).

R. “Cordillera Occidental”: COLOMBIA: Caquetá: municipio de Florencia, vereda

Gabinete, 0.7–1.1 km abajo del Alto Gabinete, 2300–2310 m (ICN 9780–81). Cauca:

245

Puracé, 2700 m (ICN 2171–73); km 41 carretera Belalcazar–Tacuello, 2600–2800 m

(ICN 2992), km 34.5 carretera Belalcazar–Tacuello, 2400 m (ICN 2993); entre Paletara y quebrada Bujios, 2960 m (ICN 4463–69 and 4540); 27 km por carretera E de

Coconuco, quebrada Ullucos, 2930 m (ICN 4471–74); municipio de Inza, vereda Río

Sucio, km 66–67 carretera Popayan–Inza (ICN 6029–32). Huila: 39 km NW arriba de

San José de Isnos, 2880 m (ICN 4470); No other data (UMMZ 121034). Valle del

Cauca: Cali, Parque Nacional Natural Los Farallones, campamento Corea, 2600 m

(ICN 2930–31), 2700 m (UV-C 7821). R. inanis: VENEZUELA: Portuguesa:

Carreterra Chabasquén-Córdova, Sierra de Portuguesa, 1450 m (MCNG 827 (holotype),

825–26 and 828 (paratypes)). Barinas: Los Alcaravanes, Calderas, 1100 m (MBLUZ

952). R. labionis: ECUADOR: Cotopaxi: Naranjito, Reserva de Bosque Integral

Otonga, 1985 m (QCAZ 10411–12). R. meleagris: ECUADOR: [Tungurahua]: Baños

(FMNH 28037–42, 28049 [six specimens]). In error: El Oro: Machala (USNM 196264–

65). R. “Nariño”: COLOMBIA: Nariño: municipio de San Juan de Pasto, corregimiento de Genoy (PSO-CZ 28, 29, 242–46); corregimiento de Obonuco, predios Corpoica

(UV-C 14982–83); municipio de Buesaco, km 17 carretera Pasto-Buesaco, Páramo de

Daza, 2850m (ICN 9816); Municipio de Tangua, Quebrada Chavez (UMMZ 171659–

61, formerly UV-C 2778–79 and 2781 respectively); 3 km al norte de Tangua, quebrada

Chavez (UMMZ 171662); 8 km NE Pasto, 3020m (KU 169943, 169945). R. orcesi:

ECUADOR: Napo: 12 km W (via road) Baeza (AMNH 124044 (paratype)); 31 km N

Jondachi, 2190 m (QCAZ 2829, 2835); vertiente del volcán Sumaco, 2200 m (QCAZ

931–40). R. raneyi: ECUADOR: Napo: 3.3 km ESE Cuyuja, 2350 m (KU 142903

(paratype)). Sucumbios: near Santa Barbara (MCZ 175160–62); Napo [= sucumbios]: inmediate environs of Santa Barbara (USNM 229750); 2 km E of Santa Barbara

(USNM 229749); 3 km SW Santa Barbara at bridge (covered) over river (USNM

246

229748). Sucumbios: 32 km E Julio Andrade on road to Santa Barbara, 2610 m (QCAZ

2827). Carchi: Santa Bárbara, Santa Bárbara-Guanderal, 2980 m (QCAZ 1379). R. simotera: ECUADOR: Carchi: 14.6 km NW El Carmelo, 3130 m (KU 179478); km 13 carretera a El Carmelo, 3300 m (ICN 9823–34); km 16 Tulcán-Tufino, 3130–3160 m

(ICN 9835–36); 15.3 km W Tulcán on road to Tufino, 3080 m (QCAZ 915, 918); km

13 desvío carretera Panamericana, el Ángel (ICN 9837). COLOMBIA: Nariño: municipio de Pupiales (IAvH [formerly IND-R] 1553); municipio de Túquerres, km 10 carretera Túquerres-Guachucal, hacienda Alsacia, 3140 m (ICN 9817); municipio de

Cumbal, km 4 Cumbal-volcán Cumbal, 3260 m (ICN 9818–22). R. stigmatoral:

ECUADOR: Azuay: Sevilla de Oro (USNM 229644 (paratype)); San Vicente, camino a las antenas (QCAZ 11414). Morona-Santiago: Pailas, a tambo on trail between Sevilla de Oro and Mendez, on E or NE facing slope (USNM 229647–48 (paratypes)); between tambos called Cerro Negro and Pailas on trail Sevilla de Oro and Mendez (USNM

229643 (paratype)); between Pailas and Mirador, on trail between Sevilla de Oro and

Mendez (USNM 229645 (paratype)); San Juan Bosco, a posada on trail between Limon

(General Plaza) and Gualeceo, slightly S of W of Limon (USNM 229649); El Cruzado, a posada on trail between Limon (General Plaza) and Gualeceo, slightly S of W of

Limon (ca. 0.5 hour up trail from San Juan Bosco) (USNM 229650). No other data

(AMNH 32778); San Jose (AMNH 38820). Cañar: Mazar, reserva Mazar (QCAZ 7374,

7884, 6657); Biblian, iglesia de Biblian (QCAZ 9946). R. striata: COLOMBIA:

Boyacá: municipio de Villa de Leyva, sector rural vereda El Roble (IAvH 4895); Pesca

(IAvH [formerly IND-R] 0665–66); municipio de Turmequé, vereda Joyagua (MUJ

816); Toquilla, Vadohondo, km 71 carretera Sogamoso-Pajarito (ICN 2800).

Cundinamarca: Bogotá (CAS-SUR 8280, MCZ 14166–67, 16979–80, 16982–83,

17129, 110415–16, USNM 75969, 153974–82, 194744, ICN 2181); Bogotá, salón de

247

clases de la Pontificia Universidad Javeriana (MUJ 229); Bogotá, instalaciones del laboratorio de fauna “Venado de Oro”, vivero Inderena (IAvH [formerly IND-R] 1100–

01, 1499, 1602, 3006); Bogotá, laboratorio de fauna Unifem Inderena (IAvH [formerly

IND-R] 3130–31, 3934, 4163, 4262); Bogotá, ladera del cerro Guadalupe (IAvH

[formerly IND-R] 3503, ICN 2436, 2535, 2537, 2541, 2543–44, 2546); mt. Guadalupe

(FMNH 177075–81, 177243–47); Salto del Tequendama (IAvH [formerly IND-R]

3985); municipio de San Francisco, vereda Sabaneta (ICN 5991), cerro Cueva Grande,

2590 m (ICN 5737), finca La Quebrada, quebrada El Vino, 2540–2560 m (ICN 9759–

65); páramo Cruz Verde, 3100 m (ICN 675–76); municipio de Fómeque (ICN 2232); between Alban and Sasaima, 50 NW Bogotá, D.C. (MVZ 191880); 6 km S Alban on road to Bogotá D.C. (MVZ 191878); represa del Hato, south of Usme, ca. 2800 m

(FMNH 165800–03, ICN 2371, 2373, 2375); municipio de Suesca, vereda el Hatillo, microcuenca Santa Helena, 2950 m (MUJ 644–48), 3 km al sur de la laguna Suesca,

2860 m (ICN 7276); no other data (UMMZ 56448, 56760 [11 specimens], 89417 [seven specimens], 123315, 203767–70, ICN 2362); PNN Chingaza, 3300 m (IAvH [formerly

IND-R] 4241–42), sitio Monteredondo, 3030 m (MUJ 906–09), sector de Chuza (IAvH

[formerly IND-R] 3891–94), embalse cerca del casino (MUJ 228). Santander: SFF

Guanentá, alto río Fonce, 2650 m (MUJ 910); municipio de Virolín, Cañaverales km

72 carretera a Charalá, río Cañaverales, 1830 m (ICN 9783). Not located: tanques de

Vitelma (IAvH [formerly IND-R] 0649). R. unicolor: ECUADOR: Carchi: Montufar

Atal-Vinculo, 2700 m (UMMZ 105895–97). Imbabura: lago de Cuicocha (MCZ

154515–16, 154628). Pichincha: Quito (MCZ 22154, 164616, 164662, 164665–68,

164670); Pasochoa Volcano Forest, 40 km SE Quito, 2800–2880 m (MCZ 175052–53);

Machachi (QCAZ 758). Not located: Chillo (MCZ 21070). Not located (QCAZ 6122).

248

R. yumborum: ECUADOR: Pichincha: Nanegal, Santa Lucía Cloud Forest Reserve,

1580–1591 m (QCAZ 10827 (holotype), 10822, 11077, 11079–81 (paratypes).

Outgroup taxa

Bachia flavescens: GUYANA: Essequibo: Tukeit beach (ROM 20515). Cercosaura argula: PERU: Cuzco: La Convención, Bajo Puyantimari, 1176 m (CORBIDI 9795).

Loreto: Datem, Sector 4, 210 m (CORBIDI 8754). Madre de Dios: Tambopata,

Baltimore, 204 m (CORBIDI 5458). Cercosaura ocellata: GUYANA: District 8:

Paramakatoi vicinity, 750 m (ROM 28352). PERU: Cuzco: La Convención, CC NN

Poyentimari, 725 m (CORBIDI 8330). San Martin: Picota, Area de Conservación

Municipal Chambira, 679 m (CORBIDI 6374). Cercosaura schreibersii: BRAZIL: Rio

Grande do Sul: Santana do Livramento, Cerros Verdes (UFRGS 5114). Mato Grosso:

Itiquira, fazenda Espigão (UFRGS 6205). Echinosaura sulcarostrum: GUYANA:

District 8: mount Wokomung, vicinity of camp 1 to camp 2, 698 m (ROM 43805).

Northwest: Baramita, vicinity of camp, 100 m (ROM 22893 (holotype), 22892, 22894

(paratypes)). Macropholidus huancabambae: PERU: Piura: Huancabamba, Las Pozas, camino El Tambo–La Mina, 2732 m (CORBIDI 10492–93, 10501–02). Macropholidus ruthveni: PERU: Lambayeque: Lambayeque, El Totoral, quebrada Palacios, distrito de

Salas, 837 m (CORBIDI 4281). Neusticurus rudis: GUYANA: District 7: Mount

Ayanganna, northeast plateu, 1490 m (ROM 39497–500). District 8: mount Wokomung, vicinity of camp 1, 698 m (ROM 42642); 1234 m (ROM 42643–44). Essequibo: Tukeit, banks of Tukei creek, 500 m from mouth, 100 m(ROM 42228, 20517). Petracola waka:

PERU: Cajamarca: Cajamarca, cataratas de Llacanora, 2705 m (CORBIDI 8639).

Petracola ventrimaculata: PERU: Cajamarca: Cajamarca, Tantahuatay, 3593 m

249

(CORBIDI 3628, 3633); Michiquillay-Quinuayoc, 3817 m (CORBIDI 9247).

Potamites ecpleopus: PERU: Cuzco: La Convención, Pongo de Mainique, 670 m

(CORBIDI 5519); CC NN Poyentimari, 725 m (CORBIDI 8331); Shokoriari, 602 m

(CORBIDI 9753). Loreto: Datem, sector 4, 210 m (CORBIDI 8690); Loreto, Andoas,

187 m (CORBIDI 4746). San Martin: Picota, Chambirillo, puesto de control 16, Parque

Nacional Cordillera Azul (CORBIDI 8834). Potamites juruazensis: PERU: Cuzco: La

Convención, Pagoreni Norte, Malvinas, 428 m (CORBIDI 10004). Potamites strangulatus: PERU: San Martin: Picota, puesto de control 16 Chambirillo, Cordillera

Azul, 1122 m (CORBIDI 9209); Tarapoto, carretera Tarapoto-Yurimaguas, cerro

Escalera, 771 m (CORBIDI 6368). Proctoporus bolivianus: PERU: Ayacucho: Yucai,

3126 m (CORBIDI 10757–58). Proctoporus chasqui: PERU: Ayacucho: Chiquintirka,

2635 m (CORBIDI 8413, 8420). Proctoporus pachyurus: PERU: Junin: Tarma, distrito de Palca, anexo Huandunga, 2429 m (CORBIDI 11808–09). Ptychoglossus brevifrontalis: BRAZIL: Acre: Serra do Divisor (MTR 28305, 28462).

12 Appendix S4

Hemipenes examined

Ingroup species are listed following the new taxonomy proposed below. See

Materials and Methods for institutional abbreviations. See Appendix 2 for locality data, if not included below.

Ingroup taxa

250

Andinosaura crypta: KU 135104 (paratype). Riama orcesi: ECUADOR: Napo: 3.3 km ESE Cuyujua, 2350 m (KU 142919 (paratype)). Riama balneator: DHMECN 4111.

Riama simotera: ICN 9836. Riama cashcaensis: ECUADOR: Chimborazo: 35.0 km

SW Cajabamba on road to Pallatanga, 2860 m (KU 217206). Riama striata: KU 217206.

Outgroup taxa

Bachia flavescens: AMNH 140925.

13 Appendix S5

A monophyletic taxonomy

The following classification reflects my total evidence phylogeny (Fig. 1). Given that Riama is polyphyletic, I redefine it, describe a new genus, and resurrect Oreosaurus.

A paucity of samples precluded the inclusion of five species in my analysis: R. inanis,

R. luctuosa, R. petrorum, R. rhodogaster and R. stellae. Regardless, my assessment includes the examination of all species currently allocated to Riama (Appendix 2).

Therefore, following the arguments of Grant et al. (2006: 149), I tentatively refer them to a genus. In addition, the recently resurrected Pantodactylus (Goicoechea et al., 2016) nests within Cercosaura. Therefore, I return it to the synonymy of Cercosaura.

Subfamily: Cercosaurinae Gray, 1838 (tribe Cercosaurini of Goicoechea et al., 2016) sensu n.

251

Content: Anadia Gray, 1845; Andinosaura new genus; Cercosaura Wagler, 1830;

Echinosaura Boulenger, 1890; Euspondylus Tschudi, 1845; Gelanesaurus Torres-

Carvajal et al., 2016; Macropholidus Noble, 1921; Neusticurus Duméril and Bibron,

1839; Oreosaurus Peters, 1862; Petracola Doan and Castoe, 2005; Pholidobolus Peters,

1862; Placosoma Tschudi, 1847; Potamites Doan and Castoe, 2005; Proctoporus

Tschudi, 1845; Riama Gray, 1858.

Comments: The analysis of Torres-Carvajal et al. (2016) recovered both

Echinosaura and Proctoporus as non-monophyletic genera. Their analysis also identified three unnamed clades composed of undescribed species from Peru and southern Ecuador. To maintain supraspecific rank equivalency within Cercosaurinae, these three clades await formal nomenclatural recognition as genera (Torres-Carvajal et al., 2016).

Genus: Riama Gray, 1858.

Type species: Riama unicolor Gray, 1858, by original designation.

Content (16 species): Riama anatoloros (Kizirian, 1996); R. balneator (Kizirian,

1996); R. cashcaensis (Kizirian and Coloma, 1991); R. cephalolineata (García-Pérez and Yustiz, 1995) new combination (see below); R. colomaromani (Kizirian, 1996);

R. columbiana (Andersson, 1914); R. inanis (Doan and Schargel, 2003); R. labionis

(Kizirian, 1996); R. meleagris (Boulenger, 1885); R. orcesi (Kizirian, 1995); R. raneyi

(Kizirian, 1996); R. simotera (O’shaughnessy, 1879); R. stigmatoral (Kizirian, 1996);

R. striata (Peters, 1862); R. unicolor Gray, 1858; R. yumborum Aguirre-Peñafiel,

Torres-Carvajal, Nunes, Peck, and Maddock, 2014.

252

Characterization and diagnosis: All unambiguously optimized synapomorphies for this clade are from DNA sequences. Phenotypic synapomorphies are not known. Other characteristics include: (1) head scales smooth; (2) frontoparietal and parietal scales paired; (3) interparietal, frontal and frontonasal scales single; (4) prefrontal scales usually absent (occasionally present in specimens of R. “Cordillera Occidental”; present in R. cephalolineata, but see comments below); (5) lower eyelid divided into several scales; (6) loreal scale absent or present; (7) scale organs on labials present; (8) anteriormost supraocular and anteriormost superciliary scales unfused (fused in some specimens of R. “Cordillera Central”); (9) dorsal surface of the tongue covered in scale- like papillae; (10) nuchal scales smooth in most species (rugose in R. striata, and undescribed R. “Cordillera Central” and R. “Cordillera Occidental”); (11) dorsal body scales rectangular; keeled (low, rounded keel) or striated (shallow or deep furrows), smooth in some specimens of R. simotera and R. meleagris; (12) ventral body scales smooth (rugose in R. “Cordillera Occidental”); (13) limbs pentadactyl; digits clawed;

(14) femoral pores in males present, in females absent or present; (15) hemipenial lobes usually narrow, indistinct from hemipenial body (large, distinct from hemipenial body in R. balneator).

Riama differs from Andinosaura by (usually) having hemipenial lobes narrow, indistinct from hemipenial body. It differs from Oreosaurus by having a narrow band of differentiated granular lateral scales.

Distribution: northern Andes of South America (Colombia, Ecuador, and Venezuela) above 1500 m a.s.l. Riama is an exclusively Andean radiation.

Comments: Currently, Riama contains 16 described species. The descriptions of four additional species from Colombia (including R. “Cordillera Occidental”, R. “Nariño”

253

and R. “Cordillera Central” from the present analysis) are currently in manuscript form

(S.J. Sánchez-Pacheco and D.A. Kizirian, in preparation).

The generic placement of Proctoporus cephalolineatus from the Venezuelan Mérida

Andes has been uncertain. In the original description, García-Pérez and Yustiz (1995) suggested it was closely related to P. achlyens and P. shrevei (both of which were in

Riama, but now in Oreosaurus, see below). However, Doan and Schargel (2003: 73) argued against its assignment to Proctoporus sensu lato due to the apparent presence of prefrontal scales in the holotype (the only known specimen of this species), while asserting “it most likely belongs to Euspondylus or Pholidobolus”. Consequently, relevant phylogenetic studies and generic classifications (e.g. Doan and Castoe, 2005;

Goicoechea et al., 2012) overlooked it. Without comment, Goicoechea et al. (2016) included this species in Proctoporus sensu stricto. Doan and Schargel (2003) described

Proctoporus inanis (prefrontals absent) from the Venezuelan Mérida Andes, and Doan and Castoe (2005) transferred it to Riama. However, Rivas et al. (2012) noted its resemblance to Proctoporus cephalolineatus, and suggested synonymy. The geographically proximate type localities and similar levels of individual variation in presence or absence of prefrontals in other cercosaurines (e.g. Macropholidus huancabambae, Reeder, 1996; Proctoporus spinalis, Köhler and Lehr, 2004;

Proctoporus sucullucu Doan and Castoe, 2003; Andinosaura stellae new combination,

Sánchez-Pacheco, 2010a; and Riama “Cordillera Occidental”, herein) are consistent with the hypothesis of conspecificity. However, I retain R. inanis as a valid species pending a review. The Andean distribution and resemblance to other species predicts that R. cephalolineata new combination and R. inanis belong to Riama; this assignment awaits confirmation.

254

Genus: Andinosaura new genus9.

Type species: Riama crypta Sánchez-Pacheco, Kizirian and Nunes, 2011.

Content (11 species): Andinosaura afrania (Arredondo and Sánchez-Pacheco, 2010) new combination; A. aurea (Sánchez-Pacheco, Aguirre-Peñafiel and Torres-Carvajal,

2012) new combination; A. crypta (Sánchez-Pacheco, Kizirian and Nunes, 2011) new combination; A. hyposticta (Boulenger, 1902) new combination; A. kiziriani

(Sánchez-Pacheco, Aguirre-Peñafiel and Torres-Carvajal, 2012) new combination; A. laevis (Boulenger, 1908) new combination; A. oculata (O’shaughnessy, 1879) new combination; A. petrorum (Kizirian, 1996) new combination; A. stellae (Sánchez-

Pacheco, 2010a) new combination; A. vespertina (Kizirian, 1996) new combination;

A. vieta (Kizirian, 1996) new combination.

A. stellae); (2) frontoparietal and parietal scales paired; (3) interparietal, frontal and frontonasal scales single; (4) prefrontal scales usually absent (occasionally present in specimens of A. stellae); (5) lower eyelid divided into several scales; (6) loreal scale absent or present; (7) scale organs on labials present; (8) anteriormost supraocular and anteriormost superciliary scales unfused; (9) dorsal surface of the tongue covered in scale-like papillae; (10) nuchal scales smooth in most species (rugose in A. stellae and

A. vieta); (11) dorsal body scales rectangular; smooth, keeled (low, rounded keel),

9This nomenclatural act has been registered in Zoobank: urn:lsid:zoobank.org:act:F228513F- 0DDB-4F3F-A771-68A0B5CD902E.

255

striated (shallow furrows) or rugose; (12) ventral body scales smooth (rugose in A. stellae and A. vieta); (13) limbs pentadactyl; digits clawed; (14) femoral pores in males present, in females absent or present; and (15) hemipenial lobes large, distinct from hemipenial body.

Andinosaura differs from Riama by having hemipenial lobes large, distinct from hemipenial body. It differs from Oreosaurus by having a narrow band of differentiated granular lateral scales.

Distribution: northern Andes of South America (Colombia and Ecuador) above 1100 m a.s.l. Andinosaura is an exclusively Andean radiation.

Etymology: Andinosaura (gender feminine), formed from the Spanish Andino (from the Andes) and the Latin Sauria (lizard), in reference to its distribution. The suffix is used commonly for gymnophthalmid genera.

Comment: Andinosaura contains nearly half of the species previously referred to the large, polyphyletic genus Riama sensu lato. I include A. stellae in this genus on the basis of its distinctive rugose scales covering almost the entire body, a condition also found in A. vieta. Although A. petrorum occurs sympatrically with Riama anatoloros and R. stigmatoral in southern Ecuador, I include this species in Andinosaura instead of Riama because Doan (2003a) found it to be the sister species of A. vespertina, and

Sánchez-Pacheco et al. (2012) suggested that it might be closely related to A. vespertina,

A. aurea and A. kiziriani.

Genus: Oreosaurus Peters, 1862

Type species: Ecpleopus (Oreosaurus) luctuosus Peters, 1862, by subsequent designation by Burt and Burt (1931).

256

Content (five species): Oreosaurus achlyens (Uzzell, 1958) new combination; O. luctuosus (Peters, 1862) new combination; O. mcdiarmidi (Kok and Rivas, 2011) new combination; O. rhodogaster (Rivas, Schargel, and Meik, 2005) new combination; O. shrevei (Parker, 1935) new combination.

(6) loreal scale usually present (absent in O. “Venezuela” and O. shrevei); (7) scale organs on labials present; (8) anteriormost supraocular and anteriormost superciliary scales unfused; (9) dorsal surface of the tongue covered in scale-like papillae; (10) nuchal scales keeled or smooth; (11) dorsal body scales hexagonal (or sub-hexagonal) or rectangular; smooth or keeled (prominent keel); (12) ventral body scales smooth;

(13) limbs pentadactyl; digits clawed; (14) femoral pores in both sexes present; (15) hemipenial lobes large, distinct from hemipenial body (narrow, indistinct from hemipenial body in O. “Sierra Nevada”).

Oreosaurus differs from Riama and Andinosaura by lacking a narrow band of differentiated granular lateral scales (oval, nongranular scales instead).

Distribution: Sierra Nevada de Santa Marta in northern Colombia, coastal mountain ranges (Cordillera de la Costa, CC), massif of Turimiquire and montane areas of the

Península de Paria (Cordillera Oriental—part of the CC) in northern Venezuela, Aripo

Northern Range in the Caribbean island of Trinidad, and tepuis from the Chimantá

Massif in southeastern Venezuela. Elevation: 650–2242 m a.s.l.

257

Comments: Peters (1862) erected Oreosaurus as a subgenus of Ecpleopus for his new species E. (O.) striatus and E. (O.) luctuosus (in sequence). Boulenger (1885) elevated Oreosaurus to genus rank. Andersson (1914) synonymized it with

Proctoporus, and Doan and Castoe (2005) implicitly placed it in the synonymy of

Riama. Peters (1862) did not designate a type species. Although O. striatus was first described, Burt and Burt (1931), without comment (but still a valid action), designated

O. luctuosus as the type species (Uzzell, 1958: 2; contra Peters and Donoso-Barros,

1970 and Kizirian, 1996: 86). A dearth of available material precluded my inclusion of the type species. However, following the arguments of Grant et al. (2006: 299) and

Frost et al. (2008: 387), I apply the name Oreosaurus to this clade based on the presumed close relationship of O. luctuosus to O. achlyens and O. shrevei (Uzzell, 1958;

Doan and Schargel, 2003). Oreosaurus achlyens and O. luctuosus occur in sympatry at

Rancho Grande on the coastal range of Venezuela. In the unlikely event that O. luctuosus is not part of this clade, there are no other available generic names (but see below).

I refer O. rhodogaster to this genus based on the assertion of Rivas et al. (2005) that

O. shrevei and O. rhodogaster are likely sister species. As such, Oreosaurus has five nominal species only, although O. “Sierra Nevada” and O. “Venezuela” are being described (Sánchez-Pacheco et al. and Rivas et al., unpublished).

Anadia is a large, broadly distributed and morphologically heterogeneous genus

(Oftedal, 1974). Despite the inclusion of the tepuian A. mcdiarmidi (herein referred to

Oreosaurus) and two Andean species from Ecuador, A. petersi and A. rhombifera

(Torres-Carvajal et al., 2016; this study), the relationships of most species of Anadia remain enigmatic (Myers et al., 2009; Kok and Rivas, 2011). My phylogenetic analysis

258

corroborates the non-monophyly of Anadia (Torres-Carvajal et al., 2016), and unexpectedly nests A. mcdiarmidi within Oreosaurus (Fig. 1).

It could be construed as more prudent to refrain from resurrecting Oreosaurus until additional species of Anadia and Euspondylus (including generic type species; see below), especially from northern South America (see Torres-Carvajal et al., 2016 and

Kok and Rivas, 2011), are included in phylogenetic analyses. However, this solution renders both Anadia and Riama polyphyletic because it overlooks existing evidence of the relationships of these genera (Torres-Carvajal et al., 2016; this study). This solution also ignores empirical knowledge on the allocations of the type species of Oreosaurus

(O. luctuosus; see above) and Anadia (A. ocellata). Oftedal (1974), based on morphological similarity, considered A. ocellata to be closely related to A. vittata, A. rhombifera and A. petersi (i.e. the A. ocellata group). Instead, I prefer to propose a taxonomic change as both a better representation of current knowledge of cercosaurine phylogeny and an invitation for refutation (for similar arguments see Grant et al. 2006 and Frost et al. 2008). Thus, I resurrect the available name Oreosaurus from the synonymy of Riama to accommodate the species forming this clade plus putatively closely related taxa. The recognition of Oreosaurus secures a monophyletic Riama and remedies the polyphyly of Anadia (Torres-Carvajal et al., 2016; this study). That said, the future discovery that Anadia marmorata (type species of Argalia Gray, 1846) is imbedded within this clade would render Oreosaurus a junior subjective synonymy of

Argalia. Because A. marmorata, O. achlyens and O. luctuosus occur in sympatry, and given the number of species currently referred to Anadia and their morphological diversity, this genus will likely experience additional partitioning.

259

14 Appendix S6

Comparative material examined

Oreosaurus achlyens: VENEZUELA: Aragua: Rancho Grande (AMNH 137260,

137267–69, 137271–76, 137278–82, 137297). O. luctuosus: VENEZUELA: Aragua:

Rancho Grande (AMNH 137270, 137277, MCZ 100410, USNM 196336), Parque

Nacional Henry Pittier, Rancho Grande (USNM 259170). O. shrevei: TRINIDAD &

TOBAGO: Horne Tucuche (MCZ 62506–07); El Teluche [in error, probably Tucuche]

(MCZ 100466–68); Mt. Tucuche (MCZ 160065–66). O. “Venezuela”: VENEZUELA:

Anzoátegui: Cerro El Guamal, Macizo del Turimiquire, municipio Freites, 2150 m

(EBRG 5962 (holotype)).

15 Appendix S7

A monophyletic taxonomy

The following classification reflects my results (Figures 7, 12 and 17; i.e. last expansions of the outgroup under three different approaches). Given that Ptychoglossus is polyphyletic, I redefine it, describe a new genus, and transfer two species into

Alopoglossus. A paucity of samples precluded the inclusion of six species in my analyses: P. bilineatus, P. danieli, P. eurylepis, P. grandisquamatus, P. kugleri and P. nicefori. Regardless, following the arguments of Grant et al. (2006: 149), I tentatively refer them to a genus.

260

Alopoglossus Boulenger, 1885. Content: A. angulatus (Linnaeus, 1758); A. atriventris Duellman, 1973; A. buckleyi (O’Shaughnessy, 1881); A. copii Boulenger,

1885; A. festae Peracca, 1904; A. lehmanni Ayala and Harris, 1984; A. stenolepis

(Boulenger, 1908) comb.n.; A. vallensis (Harris, 1994) comb.n.; A. viridiceps Torres-

Carvajal, 2014.

Ptychoglossus Boulenger, 1890. Content: P. bicolor (Werner, 1916); P. bilineatus

(Boulenger, 1890); P. brevifrontalis Boulenger, 1912; P. danieli Harris, 1994; P. eurylepis Harris and Rueda, 1985) P. festae (Peracca, 1896); P. gorgonae Harris, 1994;

P. grandisquamatus Rueda, 1985; P. kugleri Roux, 1927; P. nicefori (Loveridge, 1929).

Plicaglossus gen.n.

Type species: Alopoglossus plicatus Taylor, 1949.

Content: P. myersi (Harris, 1994) comb.n.; P. plicatus (Taylor, 1949) comb.n.; P. romaleos (Harris, 1994) comb.n.

261

References

Aguirre-Peñafiel, V., Torres-Carvajal, O., Nunes, P.M.S., Peck, M.R., Maddock, S.T.,

2014. A new species of Riama Gray, 1858 (Squamata: Gymnophthalmidae) from

the tropical Andes. Zootaxa 3866, 246–260.

Amaral, A.R., Lovewell, G., Coelho, M.M., Amato, G., Rosenbaum, H.C., 2014.

Hybrid speciation in a marine mammal: the Clymene dolphin (Stenella clymene).

PLoS One 9, e83645.

Andersson, L.G., 1914. A new Telmatobius and new teiidoid lizards from South

America. Arkiv Zool. 9, 1–12.

Arévalo, E.S., Davis, S.K., Sites, J.W. Jr., 1994. Mitochondrial DNA sequence

divergence and phylogenetic relationships among eight chromosome races of the

Sceloporus grammicus complex (Phrynosomatidae) in central Mexico. Syst. Biol.

43, 387–418.

Arnott, A., Mueller, I., Ramsland, P.A., Siba, P.M., Reeder, J.C., Barry, A.E., 2013.

Global population structure of the genes encoding the malaria vaccine candidate,

Plasmodium vivax apical membrane Antigen 1 (PvAMA1). PLoS Neglect. Trop. D.

7, e2506.

Arredondo, J.C., Sánchez-Pacheco, S.J., 2010. New endemic species of Riama

(Squamata: Gymnophthalmidae) from northern Colombia. J. Herpetol. 44, 610–617.

262

Ayala, S. C. 1986. Saurios de Colombia: lista actualizada, y distribución de ejemplares

colombianos en los museos. Caldasia 15:555–575.

Ayala, S. C., and F. Castro. Unpublished. Lagartos de Colombia/Lizards of Colombia.

Bandelt, H.J., Forster, P., Sykes B.C., Richards M.B., 1995. Mitochondrial portraits of human populations. Genetics 141, 743–753.

Bandelt, H.J., Forster, P., Röhl, A., 1999. Median-joining networks for inferring

intraspecific phylogenies. Mol. Biol. Evol. 16, 37–48.

Barriel, V., Tassy, P., 1998. Rooting with multiple outgroups: consensus versus

parsimony. Cladistics 14, 193–200.

Barros, F. C., Herrel, A., Kohlsdorf, T., 2011. Head shape evolution in

Gymnophthalmidae: does habitat use constrain the evolution of cranial design in

fossorial lizards? Journal of Evolutionary Biology 24, 2423–2433.

Bataille, A., Fong, J.J., Cha, M., Wogan, G.O., Baek, H.J., Lee, H., Min, M.S.,

Waldman, B., 2013. Genetic evidence for a high diversity and wide distribution of

endemic strains of the pathogenic chytrid fungus Batrachochytrium dendrobatidis

in wild Asian amphibians. Mol. Ecol. 22, 4196–4209.

263

Bonaccorso, E., Guayasamin, J.M., Peterson, A.T., Navarro-Siguenza, A.G., 2011.

Molecular phylogeny and systematics of Neotropical toucanets in the genus

Aulacorhynchus (Aves, Ramphastidae). Zool. Scripta 40, 336–349.

Bonaccorso, E., Guayasamin, J.M., 2013. On the origin of Pantepui montane biotas: a

perspective based on the phylogeny of Aulacorhynchus toucanets. PLoS ONE,

e67321.

Boulenger, G.A., 1885. Catalogue of the Lizards in the British Museum (Natural

History) 2, 1– 497.

Braun, E.L., Kimball, R.T., 2002. Examining basal avian divergences with

mitochondrial sequences: model complexity, taxon sampling, and sequence length.

Syst. Biol. 51, 614–625.

Bremer, K. 1988. The limits of amino acid sequence data in angiosperm phylogenetic

reconstruction. Evolution 42, 795–803.

Brower, A.V.Z., 2016. What is a cladogram and what is not? Cladistics 32, 573–576.

Brown, J.H., Lomolino, M.V., 1998. Biogeography. Sinauer Associates, Sunderland.

Bryant, H.N., 1997. Hypothetical ancestors and rooting in cladistic analysis. Cladistics

13, 337–348.

264

Bryant, H.N., 2001. Character polarity and the rooting of cladograms. In: Wagner, G.P.,

Ed., The Character Concept in Evolutionary Biology. Academic Press, New York,

pp. 321–340.

Burt, C.E., Burt, M.D., 1931. South American lizards in the collection of the American

Museum of Natural History. Bull. Am. Mus. Nat. Hist. 61, 227–395.

Cadena, C.D., Klicka, J., Ricklefs, R.E., 2007. Evolutionary differentiation in the

Neotropical montane region: molecular phylogenetics and phylogeography of

Buarremon brush-finches (Aves, Emberizidae). Mol. Phylogen. Evol. 44, 993–1016.

Cadena, C.D., Cuervo, A.M., 2010. Molecules, ecology, morphology, and songs in

concert: how many species is Arremon torquatus (Aves: Emberizidae)? Biol. J. Linn.

Soc. 99, 152–176.

Cadena, C.D., Caro, L.M., Caycedo, P.C., Cuervo, A.M., Bowie, R.C.K., Slabbekoorn,

H., 2016. Henicorhina anachoreta (Troglodytidae), another endemic bird species

for the Sierra Nevada de Santa Marta, Colombia. Ornitología Colombiana 15, 82–

89.

Camin, J.H., Sokal, R.R., 1965. A method for deducing branching sequences in

phylogeny. Evolution 19, 311–326.

Cao, L., Liang, X.-F., Du, Y., Zheng, H., Yang, M., Huang, W., 2013. Genetic

population structure in Siniperca scherzeri (Perciformes: Siniperca) in China

265

inferred from mitochondrial DNA sequences and microsatellite loci. Biochem. Syst.

Ecol. 51, 160–170.

Cardona Molina, A., Cordani, U.G., MacDonald, W.D., 2006. Tectonic correlations of

pre-Mesozoic crust from the northern termination of the Colombian Andes,

Caribbean region. J. South Am. Earth Sci. 21, 337–354.

Caro, L.M., Caycedo-Rosales, P.C., Bowie, R.C.K., Slabbekoorn, H., Cadena, C.D.,

2013. Ecological speciation along an elevational gradient in a tropical passerine bird?

J. Evol. Biol. 26, 357–374.

Cassens, I., Van Waerebeek, K., Best, P.B., Crespo, E.A., Reyes, J., Milinkovitch, M.C.,

2003. The phylogeography of dusky dolphins (Lagenorhynchus obscrus): a critical

examination of network methods and rooting procedures. Mol. Ecol. 12, 1781–1792.

Cassens, I., Mardulyn, P., Milinkovich, M.C., 2005. Evaluating intraspecific “network”

construction methods using simulated sequence data: do existing algorithms

outperform the Global Maximum Parsimony approach? Syst. Biol. 54, 363–372.

Castelin, M., Lorion, J., Brisset, J., Cruaud, C., Maestrati, P., Utge, J., Samadi, S., 2012.

Speciation patterns in gastropods with long-lived larvae from deep-sea seamounts.

Mol. Ecol. 21, 4828–4853.

266

Castoe, T.A., Doan, T.M., Parkinson, C.L., 2004. Data partitions and complex models

in Bayesian analysis: the phylogeny of gymnophthalmid lizards. Syst. Biol. 53, 448–

469.

Castroviejo-Fisher, S., Padial, J.M., De la Riva, I., Pombal Jr., J.P., Da Silva, H.R.,

Rojas-Runjaic, F.J.M., Medina-Méndez, E., Frost, D.R., 2015. Phylogenetic

systematics of egg-brooding frogs (Anura: Hemiphractidae) and the evolution of

direct development. Zootaxa 4004, 1–75.

Chapman, F.M., 1931. The upper zonal bird-life of Mts. Roraima and Duida. Bull. Am.

Mus. Nat. Hist. 63, 1–135.

Chung, F.R.K., 1986. Diameters of communication networks. Proc. Symp. Appl. Math.

34, 1–18.

Colless, D.H., 1969. The interpretation of Hennig's "Phylogenetic Systematics"—a

reply to Dr. Schlee. Syst Zool. 18, 134–144.

Colless, D.H., 1985. On the status of outgroups in phylogenetics. Syst. Zool. 34, 364–

366.

Colli, G.R., Hoogmoed, M.S., Canatella, D.C., Cassimiro, J., Oliveira Gomes, J.,

Ghellere, J.M., Nunes, P.M.S., Pellegrino, K.C.M., Salerno, P., Marques de Souza,

S., Rodrigues, M.T., 2015. Description and phylogenetic relationships of a new

genus and two new species of lizards from Brazilian Amazonia, with nomenclatural

267

comments on the taxonomy of Gymnophthalmidae (Reptilia: Squamata). Zootaxa

4000, 401–427.

Cummings, M.P., Meyer, A., 2005. Magic bullets and golden rules: Data sampling in

molecular phylogenetics. Zoology 108, 329–336.

Cunha, R.L., Coscia, I., Madeira, C., Mariani, S., Stefanni, S., Castilho, R., 2012.

Ancient divergence in the trans-oceanic deep-sea shark Centroscymnus crepidater,

PLoS One 7, e49196.

Darwin, C., 1859. On the Origin of Species by means of Natural Selection, or the

Preservation of Favoured Races in the Struggle for Life. John Murray, London [1964.

Facsimile reprint of the first edition. Harvard University Press, Cambridge,

Massachusetts].

Davis, J.I., 1993. Character removal as a means for assessing stability of clades.

Cladistics 9, 201–210.

de Albuquerque, N. R., P. Passos, and S. W. Gotte. 2012. Leptophis santamartensis

(Serpentes, Colubridae), a junior synonym of Leptophis ahaetulla occidentalis.

Journal of Herpetology 46:248–252.

de la Torre-Bárcena, J.E., Kolokotronis, S.-O., Lee, E.K., Stevenson, D.W., Brenner,

E.D., Katari, M.S., Coruzzi, G.M., DeSalle, R., 2009. The impact of outgroup choice

268

and missing data on major seed plant phylogenetics using genome-wide EST data.

PLoS ONE 4, e5764.

de Queiroz, K., 1985. The ontogenetic method for determining character polarity and

its relevance to phylogenetic systematics. Syst. Zool. 34, 280–299.

de Sá, R.O., Grant, T., Camargo, A., Heyer, W.R., Ponssa, M.L., Stanley, E., 2014.

Systematics of the Neotropical genus Leptodactylus Fitzinger, 1826 (Anura:

Leptodactylidae): Phylogeny, the relevance of non-molecular evidence, and species

accounts. South Am. J. Herpetol. 9, S1–S128.

del Pino, E.M., 1980. Morphology of the pouch and incubatory integument in marsupial

frogs (Hylidae). Copeia 1980, 10–17.

Delwiche, C.F., Palmer, J.D., 1996. Rampant horizontal transfer and duplication of

rubisco genes in eubacteria and plastids. Mol. Biol. Evol. 13, 873–882.

Doan, T.M., 2003a. A south-to-north biogeographic hypothesis for Andean speciation:

evidence from the lizard genus Proctoporus (Reptilia, Gymnophthalmidae). J.

Biogeogr. 30, 361–374.

Doan, T.M., 2003b. A new phylogenetic classification for the gymnophthalmid genera

Cercosaura, Pantodactylus and Prionodactylus (Reptilia: Squamata). Zool. J. Linn.

Soc. 137, 101–115.

269

Doan, T.M., Castoe, T.A., 2003. Using morphological and molecular evidence to infer

species boundaries within Proctoporus bolivianus Werner (Squamata:

Gymnophthalmidae). Herpetologica 59, 432–449.

Doan, T. M., and W. E. Schargel. 2003. Bridging the gap in Proctoporus distribution:

a new species (Squamata: Gymnophthalmidae) from the Andes of Venezuela.

Herpetologica 59:69–75.

Doan, T.M., Castoe, T.A., 2005. Phylogenetic taxonomy of the Cercosaurini (Squamata:

Gymnophthalmidae), with new genera for species of Neusticurus and Proctoporus.

Zool. J. Linn. Soc. 143, 405–416.

Doan, T.M., Castoe, T.A., Arizábal Arriaga, W., 2005. Phylogenetic relationships of

the genus Proctoporus sensu stricto (Squamata: Gymnophthalmidae), with a new

species from Puno, southeastern Peru. Herpetologica 61, 325–336.

Doan, T. M., 2008. Dietary Variation within the Andean Lizard Clade Proctoporus

(Squamata: Gymnophthalmidae). Journal of Herpetology 42, 16–21.

Donnelly, M.A., MacCulloch, R.D., Ugarte, C.A., Kizirian, D., 2006. A new riparian

gymnophthalmid (Squamata) from Guyana. Copeia 3, 396–403.

Donoghue, M.J., Doyle, J.A., Gauthier, J., Kluge, A.G., 1989. The importance of fossils

in phylogeny reconstruction. Annu. Rev. Ecol. Syst. 20, 431–460.

270

Dowling, H.G., Savage, J.M., 1960. A guide to the snake hemipenis: a survey of basic

structure and systematic characteristics. Zoologica 45, 17–28.

Duellman, W.E., 1979. The herpetofauna of the Andes: patterns of distribution, origin,

differentiation, and present communities. In: Duellman (Ed.), The South American

Herpetofauna: its origin, evolution, and dispersal. Monogr. Mus. Nat. Hist., Univ.

Kans. 7, pp. 371–459.

Echevarría, L.Y., Venegas, P.J., 2015. A new elusive species of Petracola (Squamata:

Gymnophthalmidae) from the Utcubamba basin in the Andes of northern Peru.

Amphib. Reptile Conserv. 9, 26–33.

Ecuador. Check List 4:366–372.

Edgar, R.C., 2004. MUSCLE: multiple sequence alignment with high accuracy and

high throughput. Nucl. Acids Res. 32, 1792–1797.

Eldredge, N., Cracraft, J., 1980. Phylogenetic patterns and the evolutionary process.

Columbia University Press, New York.

Escobar-Gutiérrez, A., Soudeyns, H., Larouche, A., Carpio-Pedroza, J.C., Martinez-

Guarneros, A., Vazquez-Chacon, C.A., Fonseca-Coronado, S., Yamasaki, L.H.,

Ruiz-Tovar, K., Cruz-Rivera, M., 2013. Vertical transmission of hepatitis C virus: a

tale of multiple outcomes. Infect. Genet. Evol. 20, 465–470.

271

Estes, R., Queiroz, K., Gauthier, J. 1988. Phylogenetic relationships within the

Squamata. In: Estes, R., Pregill, G. (Eds.), Phylogenetic relationships of the lizard

families - Essays commemorating Charles L. Camp. Stanford: Stanford University

Press. p 119–281.

Faivovich, J., 2002. On RASA. Cladistics 18, 324–333.

Fan D.M., Yue, J.P., Nie, Z.L., Li, Z.M., Comes, H.P., Sun, H., 2013. Phylogeography

of Sophora davidii (Leguminosae) across the ‘Tanaka-Kaiyong line’, an important

phytogeographic boundary in Southwest China. Mol. Ecol. 22, 4270–4288.

Farris, J.S., 1967. The meaning of relationship and taxonomic procedure. Syst. Zool.

16, 44–51.

Farris, J.S., 1970. Methods for computing Wagner trees. Syst. Zool. 19, 83–92.

Farris, J.S., 1972. Estimating phylogenetic tress from distance matrices. Am. Nat. 106,

645–668.

Farris, J.S., 1982. Outgroups and parsimony. Syst. Zool. 31, 328–334.

Farris, J.S., Kluge, A.G., Mickevich, M.F., 1982. Phylogenetic analysis, the monothetic

group method, and myobatrachid frogs. Syst. Zool. 31, 317–327.

272

Farris, J.S., 1983. The logical basis of phylogenetic analysis. In: Platnick, N.I., Funk,

V.A. (Eds.), Advances in Cladistics: Proceedings of the Third Meeting of the Willi

Hennig Sciety II. Columbia University Press, New York, Vol. 2, pp. 7–36.

Farris, J.S., 2002. RASA attributes highly significant structure to randomized data.

Cladistics 18, 334–353.

Frost, D.R., Grant, T., Faivovich, J., Bain, R., Haas, A., Haddad, C.F.B., de Sá, R.O.,

Donnellan, S.C., Raxworthy, C.J., Wilkinson, M., Channing, A., Campbell, J.A.,

Blotto, B.L., Moler, P., Drewes, R.C., Nussbaum, R.A., Lynch, J.D., Green, D.,

Wheeler, W.C., 2006. The amphibian tree of life. Bull. Amer. Mus. Nat. Hist. 297,

1–370.

Fu, J., 2000. Toward the phylogeny of the family lacertidae—Why 4708 base pairs of

mtDNA sequences cannot draw the picture. Biol. J. Linn. Soc. 71, 203–217.

Gangloff, B., Zino, F., Shirihai, H., González-Solís, J., Couloux, A., Pasquet, E.,

Bretagnolle, V., 2013. The evolution of North-East Atlantic gadfly pretels using

statistical phylogeography. Mol. Ecol. 22, 495–507.

García-Pérez, J.L., Yustiz, E.E., 1995. Una nueva especie de Proctoporus (Sauria:

Gymnophthalmidae) de los Andes de Venezuela. Rev. Ecol. Latinoam. 4, 1–5.

Goicoechea, N., Frost, D.R., De la Riva, I., Pellegrino, K.C.M., Sites Jr, J., Rodrigues,

M.T., Padial, J.M., 2016. Molecular systematics of teioid lizards

273

(Teioidea/Gymnophthalmoidea: Squamata) based on the analysis of 48 loci under

tree-alignment and similarity-alignment. Cladistics 32, 624–671.

Goicoechea, N., Padial, J.M., Chaparro, J.C., Castroviejo-Fisher, S., De la Riva, I.,

2012. Molecular phylogenetics, species diversity, and biogeography of the Andean

lizards of the genus Proctoporus (Squamata: Gymnophthalmidae). Mol. Phylogen.

Evol. 65, 953–964.

Goicoechea, N., Padial, J.M., Chaparro, J.C., Castroviejo-Fisher, S., De la Riva, I.,

2013. A taxonomic revision of Proctoporus bolivianus Werner (Squamata:

Gymnophthalmidae) with the description of three new species and resurrection of

Proctoporus lacertus Stejneger. Am. Mus. Novit. 3786, 1–32.

Goloboff, P.A. 1993. Character optimization and calculation of tree lengths. Cladistics

9, 433–436.

Goloboff, P.A., 1996. Methods for faster parsimony analysis. Cladistics 12, 199–220.

Goloboff, P.A., 1999. Analyzing large data sets in reasonable times: solutions for

composite optima. Cladistics 15, 415–428.

Goloboff, P.A., 2003. Parsimony, likelihood, and simplicity. Cladistics 19, 91–103.

274

Goloboff, P.A., Pol, D. 2005. Parsimony and Bayesian phylogenetics. In: Albert, V.A.

(Ed.), Parsimony, Phylogeny, and Genomics. Oxford University Press, Oxford, pp.

148–159.

Goloboff, P.A., Mattoni, C.I., Quinteros, A.S., 2006. Continuous characters analyzed

as such. Cladistics 22, 589–601.

Goloboff, P.A., Catalaño, S.A., Mirande, J.M., Szumik, C.A., Ariasa, J.S., Källersjö,

M., Farris, J.S., 2009. Phylogenetic analysis of 73 060 taxa corroborates major

eukaryotic groups. Cladistics 25, 1–20.

Goodman, M., Olson, C.B., Beeber, J.E., Czelusniak, J., 1982. New perspectives in the

molecular biological analysis of mammalian phylogeny. Acta Zool. Fenn. 169, 19–

35.

Gorzula, S., 1987. Una revisión de los origenes de la fauna de vertebrados del Pantepui.

Pantepui 3, 4–10.

Graham, S.W., Iles, W.J.D., 2009. Different gymnosperm outgroups have (mostly)

congruent signal regarding the root of flowering plant phylogeny. Am. J. Bot. 96,

216–227.

Graham, S.W., Olmstead, R.G., Barrett, S.C.H., 2002. Rooting phylogenetic trees with

distant outgroups: a case study from the commelinoid monocots. Mol. Biol. Evol.

19, 1769–1781.

275

Grant, T., 2002. Testing methods: the evaluation of discovery operations in

evolutionary biology. Cladistics 18, 94–111.

Grant, T., Faivovich, J., Pol, D., 2003. The perils of ‘point-and-click’ systematics.

Cladistics 19, 275–285.

Grant, T., Frost, D.R., Caldwell, J.P., Gagliardo, R., Haddad, C.F.B., Kok, P.J.R.,

Means, D.B., Noonan, B.P., Schargel, W.E., Wheeler, W.C., 2006. Phylogenetic

systematics of dart-poison frogs and their relatives (Amphibia: Athesphatanura:

Dendrobatidae). Bull. Am. Mus. Nat. Hist. 299, 1–262.

Grant, T., Kluge, A.G., 2003. Data exploration in phylogenetic inference: scientific,

heuristic, or neither. Cladistics 19, 379–418.

Grant, T., Kluge, A.G., 2004. Transformation series as an ideographic character

concept. Cladistics 20, 23–31.

Grant, T., Kluge, A.G., 2007. Ratio of explanatory power (REP): a new measure of

group support. Mol. Phylogen. Evol. 44, 483–487.

Grant, T., Kluge, A.G., 2008. Clade support measures and their adequacy. Cladistics

24, 1051–1064.

276

Grant, T., Kluge, A.G., 2008a. Credit where credit is due: The Goodman-Bremer

support metric. Mol. Phylogenet. Evol. 49, 405–406.

Grant, T., Kluge, A.G., 2008b. Credit where credit is due: The Goodman-Bremer

support metric. Mol. Phylogen. Evol. 49, 405–406.

Grant, T., Kluge, A.G., 2008c. Support measures and their adequacy. Cladistics 24,

1051–1064.

Grant, T., Kluge, A.G., 2009. Parsimony, explanatory power, and dynamic homology

testing. Syst. Biodivers. 7, 357–363.

Grant, T., Kluge, A.G., 2010. REP provides meaningful measurement of support across

datasets. Mol. Phylogen. Evol. 55, 340–342.

Griffiths, R.C., Marjoram, P., 1996. Ancestral inference from samples of DNA

sequences with recombination. J. Comput. Biol. 3, 479–502.

Guayasamin, J.M., Castroviejo-Fisher, S., Ayarzagüena, J., Trueb, L., Vilá, C., 2008.

Phylogenetic relationships of glassfrogs (Centrolenidae) based on mitochondrial and

nuclear genes. Mol. Phylogenet. Evol. 48, 574–595.

Haas, A., 2003. Phylogeny of frogs as inferred from primarily larval characters

(Amphibia: Anura). Cladistics 19, 23–89.

277

Harris, D. M., and S. C. Ayala. 1987. A new Anadia (Sauria: Teiidae) from Colombia

and restoration of Anadia pamplonensis Dunn to species status. Herpetologica

43:182–190.

Harris, D.J., Arnold, E.N., Thomas, R.H., 1998. Relationships of lacertid lizards

(Reptilia: Lacertidae) estimated from mitochondrial DNA sequences and

morphology. Proc. Roy. Soc. Lond. B Biol. Sci. 265, 1939–1948.

Harris, D.N., 1994. Review of the teiid lizard genus Ptychoglossus. Herpetol. Monogr.

8, 226–275.

Heath, T.A., Hedtke, S.M., Hillis, D.M., 2008. Taxon sampling and the accuracy of

phylogenetic analyses. J. Syst. Evol. 46, 239–257.

Heinicke, M.P., Duellman, W.E., Trueb, L., Means, B.D., MacCulloch, R.D., Hedges,

S.B., 2009. A new frog family (Anura: Terrarana) from South America and an

expanded direct-developing clade revealed by molecular phylogeny. Zootaxa 2211,

1–35.

Hennig, W., 1966. Phylogenetic Systematics. University of Illinois Press, Urbana, IL.

Hess, P.N., Russo, C.A.M., 2007. An empirical test of the midpoint rooting method.

Biol. J. of Linnean Soc. 92, 669–674.

278

Hillis, D.M., 1998. Taxonomic sampling, phylogenetic accuracy, and investigator bias.

Syst. Biol. 47, 3–8.

Höfling, E. and S. Renous., 2004. High frequency of pauses during intermittent

locomotion of small South American gymnophthalmid lizards (Squamata:

Gymnophthalmidae). Phyllomedusa 3, 83–94.

Hovenkamp, P., 2006. Can taxon-sampling effects be minimized by using branch

supports? Cladistics 22, 264–275.

Huson, D.H., Rupp, R., Scornavacca, C., 2010. Phylogenetic networks: concepts,

algorithms and applications. Cambridge University Press, Cambridge.

Hwang, F.K., Richards, D.S., Winter, P., 1992. The Steiner tree problem. North-

Holland, Amsterdam.

ICZN (International Commission on Zoological Nomenclature). 1999. Code of

Zoological Nomenclature. Fourth edition. The International Trust for Zoological

Nomenclature, London.

Johnson, K.P., 2001. Taxon sampling and the phylogenetic position of Passeriformes:

evidence from 916 avian cytochrome b sequences. Syst. Biol. 50, 128–136.

279

Kaiser, H., Barrio-Amorós, C.L., Rivas, G.A., Steinlein, C., Schmid, 2015. Five new

species of Pristimantis (Anura: Strabomantidae) from the coastal cloud forest of the

Península de Paria, Venezuela. J. Threat. Taxa 7, 7047–7088.

Katoh, K., Kuma, K., Toh, H., Miyata, T., 2005. MAFFT, a novle method for rapid

multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30,

3059–3066.

Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., Buxton,

S., Cooper, A., Markowitz, S., Duran, C., Thierer, T., Ashton, B., Mentjies, P.,

Drummond, A., 2012. Geneious Basic: an integrated and extendable desktop

software platform for the organization and analysis of sequence data.

Bioinformatics 28, 1647–1649.

Kizirian, D.A., Coloma, L.A., 1991. A new species of Proctoporus (Squamata:

Gymnophhalmidae) from Ecuador. Herpetologica 47, 420–429.

Kizirian, D.A., 1995. A new species of Proctoporus (Squamata: Gymnophthalmidae)

from the Andean Cordillera Oriental of northeastern Ecuador. J. Herpetol. 29, 66–

72.

Kizirian, D.A., 1996. A review of Ecuadorian Proctoporus (Squamata:

Gymnophthalmidae) with descriptions of nine new species. Herpetol. Monogr. 10,

85–155.

280

Kizirian, D.A., Cole, C.J., 1999. Origin of the unisexual lizard Gymnophthalmus

underwoodi (Gymnophthalmidae) inferred from mitochondrial DNA nucleotide

sequences. Molecular Phylogenetics and Evolution 11, 394–400.

Kizirian, D., Bayefsky-Anand, S., Eriksson, A., Le, M., Donnelly, M.A., 2008. A new

Petracola and re-description of P. ventrimaculatus (Squamata: Gymnophthalmidae).

Zootaxa 1700, 53–62.

Kluge, A.G., 1983. Cladistics and the classification of the great apes. In: New

interpretations of Ape and Human Ancestry (R.I. Cicochon & R.S. Corruccini, eds),

pp. 151–177. Plenum Press, New York.

Kluge, A.G., 1985. Ontogeny and phylogenetic systematics. Cladistics 1, 13–27.

Kluge, A.G., Strauss, R.E., 1985. Ontogeny and systematics. Annu. Rev. Ecol. Syst. 16,

247–268.

Kluge, A.G., 1989. A concern for evidence and a phylogenetic hypothesis of

relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38, 7–25.

Kluge, A.G., 1997. Sophisticated falsification and research cycles: Consequences for

differential character weighting in phylogenetic analysis. Zool. Scr. 26, 349–360.

Kluge, A.G., 1997. Testability and the refutation and corroboration of cladistic

hypotheses. Cladistics 13, 81–96.

281

Kluge, A.G., 1999. The science of phylogenetic systematics: Explanation, prediction,

and test. Cladistics 15, 429–436.

Kluge, A.G., 2003. On the deduction of species relationships: A précis. Cladistics 19,

233–239.

Kluge, A.G., 2004. On total evidence: for the record. Cladistics 20, 205–207.

Kluge, A.G., Grant, T., 2006. From conviction to anti-superfluity: old and new

justifications for parsimony in phylogenetic inference. Cladistics 22, 276–288.

Klütsch, C.F.C., Manseau, M., Wilson, P.J., 2012. Phylogeographic analysis of mtDNA

data indicates postglacial expansion from multiple glacial refugia in Woodland

caribou (Rangifer tarandus caribou). PLoS One 7, e52661.

Kocher, T.D., Thomas, W.K., Meyer, A., Edwards, S.V., Pääbo, S., Villablanca, F.X.,

Wilson, A.C., 1989. Dynamics of mitochondrial DNA evolution in animals:

amplification and sequencing with conserved primers. Proc. Natl. Acad. Sci. USA

86, 6196–6200.

Köhler, G., Lehr, E., 2004. Comments on Euspondylus and Proctoporus (Squamata:

Gymnophthalmidae) from Peru, with the description of three new species and a key

to the Peruvian species. Herpetologica 60, 501–518.

282

Kok, P. J. R., and G. A. Rivas. 2011. A new species of Anadia (Reptilia, Squamata)

from the Venezuelan ‘Lost World’, northern South America. European Journal of

Taxonomy 3:1–18.

Kok, P.J.R., MacCulloch, R.D., Means, D.B., Roelants, K., Bocxlaer, I.V., Bossuyt, F.,

2012. Low genetic diversity in tepui summit vertebrates. Curr. Biol. 22, 589–590.

Kok, P.J.R., Means, D.B., Rivas, G.A., 2013. First record of the genus Anadia (Reptilia:

Squamata: Gymnophthalmidae) from Guyana based on an enigmatic specimen from

mount Kopinang, Wokomung massif. Bol. Mus. Para. Emílio Goeldi Cienc. Nat. 8,

27–39.

Kok, P.J.R., 2015. A new species of the Pantepui endemic genus Riolama (Squamata:

Gymnophthalmidae) from the summit of Murisipán-tepui, with the erection of a new

gymnophthalmid subfamily. Zool. J. Linn. Soc. 174, 500–518.

Kong, S., Sánchez-Pacheco, S.J., Murphy, R.W., 2016. On the use of median-joining

networks in evolutionary biology. Cladistics 32, 691–699.

Kraus, F., 1988. An empirical evaluation of the use of the ontogeny polarization

criterion in phylogenetic inference. Syst. Zool. 37, 106–141.

Kruskal, J.B., 1956. On the shortest spanning subtree of a graph and the traveling

salesman problem. Proc. Amer. Math. Soc. 7, 48–50.

283

Liddle, R.A., 1946. The geology of Venezuela and Trinidad. Paleontological Research

Institution, Ithaca.

Lin, L.-H., Qu, Y.-F., Li, H., Zhou, K.,-Y, Ji, X., 2012. Genetic structure and

demographic history should inform conservation: Chinese cobras currently treated

as homogenous show population divergence. PLoS One 7, e36334.

Linder, C.R., Rieseberg, L.H., 2004. Reconstructing patterns of reticulate evolution in

plants. Am. J. Bot. 91, 1700–1708.

Lundberg, J.G., 1972. Wagner networks and ancestors. Syst. Zool. 21, 398–413.

Lynch, J.D., 1976. The species groups of the South American frogs of the genus

Eleutherodactylus (Leptodactylidae). Occas. Pap. Mus. Nat. Hist. Univ. Kansas 61,

1–24.

Lynch, J.D., 1978. A new eleutherodactyline frog from the Andes of northern Colombia

(Leptodactylidae). Copeia 1978, 17–21.

Lynch, J.D., Ruíz-Carranza, P.M., 1985. A synopsis of the frogs of the genus

Eleutherodactylus from the Sierra Nevada de Santa Marta, Colombia. Occas. Pap.

Mus. Zool. Univ. Mich. 711, 1–59.

Lyons-Weiler, J., Hoelzer, G.A., Tausch, R.J., 1998. Optimal outgroup analysis. Biol.

J. Linnean Soc. 64, 493–511.

284

Maddison, W.P., 1997. Gene trees in species trees. Syst. Biol. 46, 523–536.

Malyarchuk, B., Derenko, M., Mikhailova, E., Denisova, G., 2014. Phylogenetic

relationships among Neoechinorhynchus species (Acanthocephala:

Neoechinorhynchidae) from North-East Asia based on molecular data. Parasitol. Int.

63, 100–107.

Mamani, L., Goicoechea, N., Chaparro, J.C., 2015. A new species of Andean lizard

Proctoporus (Squamata: Gymnophthalmidae) from montane forest of the Historic

Sanctuary of Machu Picchu, Peru. Amphib. Reptile Conserv. 9, 1–11.

Manzani, P.R., Abe, A.S., 1988. Sobre dois novos métodos de preparo do hemipenis de

serpentes. Memórias do Instituto Butantan 50, 15–20.

Manzanilla, J., La Marca, E., García-París, M., 2009. Phylogenetic patterns of

diversification in a clade of Neotropical frogs (Anura: Aromobatidae:

Mannophryne). Biol. J. Linn. Soc. 97, 185–199.

Mauck, W.M., Burns, K.J., 2009. Phylogeny, biogeography, and recurrent evolution of

divergent bill types in the nectar-stealing flowerpiercers (Thraupini: Diglossa and

Diglossopis). Biol. J. Linn. Soc. 98, 14–28.

Mayr, E., 1965. Numerical phenetics and taxonomic theory. Syst. Zool. 14, 73–97.

285

McCracken, K.G., Wilson, R.E., Peters, J.L., Winker, K., Martin, A.R., 2013. Late

Pleistocene colonization of South Georgia by yellow-billed pintails pre-dates the

Last Glacial Maximum. J. Biogeogr. 40, 2348–2360.

Miller, J.A., Hormiga, G., 2004. Clade stability and the addition of data: A case study

from erigonine spiders (Araneae: Linyphiidae, Erigoninae). Cladistics 20, 385–442.

Mirande, J.M., 2016. Combined phylogeny of ray-finned fishes (Actinopterygii) and

the use of morphological characters in large-scale analyses. Cladistics. doi:

10.1111/cla.12171

Molina, C., Señaris, J.C., 2003. Una nueva especie del género Riolama (Reptilia:

Gymnophthalmidae) de las tierras altas del estado Amazonas, Venezuela. Mem.

Fund. La Salle Cienc. Nat. 155, 5–19.

Montero, R., Moro, S.A., Abdala, V., 2002. Cranial anatomy of Euspondylus

acutirostris (Squamata: Gymnophthalmidae) and its placement in a modern

phylogenetic hypothesis. Russ. J. Herpetol. 9, 215–228.

Moret, B.M.E., Nakhleh, L., Warnow, T., Linder, C.R., Tholse, A., Padolina, A., Sun,

J., Timme, R., 2004. Phylogenetic networks: modeling, reconstructibility, and

accuracy. IEEE/ACM Trans. Comput. Biol. Bioinform. 1, 13–23.

286

Murphy, R.W., 1993. The phylogenetic analysis of allozyme data: invalidity of coding

alleles by presence/absence and recommended procedures. Biochem. Syst. Ecol. 21,

25–38.

Murphy, R.W., Doyle, K.D., 1998. Phylophenetics: frequencies and polymorphic

characters in genealogical estimation. Syst. Biol. 47, 737–761.

Myers, C.W., Donnelly, M.A., 2001. Herpetofauna of the Yutajé–Corocoro massif,

Venezuela: second report from the Robert G. Goelet American Museum–Terramar

expedition to the Northwestern tepuis. Bull. Am. Mus. Nat. Hist. 261, 1–85.

Myers, C.W., Rivas Fuenmayor, G., Jadin, R.C., 2009. New species of lizards from

Auyantepui and La Escalera in the Venezuelan Guayana, with notes on ‘‘microteiid’’

hemipenes (Squamata: Gymnophthalmidae). Am. Mus. Novit. 3660, 1–31.

Nixon, K.C., Carpenter, J.M., 1993. On outgroups. Cladistics 9, 413–426.

Nixon, K.C. 1999. The parsimony ratchet, a new method for rapid parsimony analysis.

Cladistics 15, 407–414.

Nunes, P. M. S. 2011. Morfologia hemipeniana dos lagartos microteídeos e suas

implicações nas relações filogenéticas da família Gymnophthalmidae (Teioidea:

Squamata). PhD Thesis, Departamento de Zoologia, Universidade de São Paulo, São

Paulo.

287

Nunes, P. M. S. A. Fouquet, F. F. Curcio, P. J. R. Kok, and M. T. Rodrigues. 2012.

Cryptic species in Iphisa elegans Gray, 1851 (Squamata: Gymnophthalmidae)

revealed by hemipenial morphology and molecular data. Zoological Journal of the

Linnean Society 166:361–376.

Nunes, P.M.S., Curcio, F.F., Roscito, J.G., Rodrigues, M., 2014. Are hemipenial spines

related to limb reduction? A spiny discussion focused on gymnophthalmid lizards

(Squamata: Gymnophthalmidae). Anat. Rec. 297, 482–495.

O’Leary, M.A., Kaufman, S.G., 2011. MorphoBank: phylophenomics in the ‘cloud’.

Cladistics 27, 1–9.

Oftedal, O.T., 1974. A revision of the genus Anadia (Sauria, Teiidae). Arqu. Zool. 25,

203–265.

Padial, J.M., Grant, T., Frost, D.R., 2014. Molecular systematics of terraranas (Anura:

Brachycephaloidea) with an assessment of the effects of alignment and optimality

criteria. Zootaxa 3825, 1–132.

Parenti, L.R., Ebach, M.C., 2009. Comparative Biogeography. University of California

Press, Berkeley, CA.

Paupério, J., Herman, J.S., Melo-Ferreira, J., Jaarola, M., Alves, P.C., Searle, J.B., 2012.

Cryptic speciation in the field vole: a multilocus approach confirms three highly

divergent lineages in Eurasia. Mol. Ecol. 21, 6015–6032.

288

Pellegrino, K.C.M., Rodrigues, M.T., Yonenaga-Yassuda, Y., 1999. Chromosomal

polymorphisms due to supernumerary chromosomes and pericentric inversions

in the eyelid-less microteiid lizard Nothobachia ablephara (Squamata,

Gymnophthalmidae). Chromosome Research 7, 247–254.

Pellegrino, K.C.M., Rodrigues, M.T., Yonenaga-Yassuda, Y., Sites Jr, J.W., 2001. A

molecular perspective on the evolution of microteiid lizards (Squamata,

Gymnophthalmidae), and a new classification for the family. Biol. J. Linn. Soc. 74,

315–338.

Pérez-Emán, J., 2005. Molecular phylogenetics and biogeography of the Neotropical

redstars (Myioborus: Aves, Parulinae). Mol. Phylogen. Evol. 37, 511–528

Pesantes, O. S. 1994. A method for preparing the hemipenis of preserved snakes.

Journal of Herpetology 28:93–95.

Peters, J. A. 1967. The lizards of Ecuador, a check list and key. Proceedings of the

United States National Museum 119:1–49.

Peters, J.A., Donosos-Barros, R., 1970. Catalogue of the Neotropical Squamata, Part II,

lizards and amphisbaenians. Bulletin of the US National Museum 297, 1–293.

289

Peters, W., 1862. Über Cercosaura und die mit dieser Gattung verwandten Eidechsen

aus Südamerica. Aus Den Abhandlungen der Königl. Akademie der Wissenschaften

zu Berlin 1862, 165–225.

Planet, P.J., Kachlany, S.C., Fine, D.H., DeSalle, R., Figurski, D.H., 2003. The

widespread colonization island of Actinobacillus actinomycetemcomitans. Nat.

Genet. 34, 193–198.

Platnick, N.I., 1977. Cladograms, phylogenetic trees, and hypothesis testing. Syst. Zool.

26, 438–442.

Poe, S. 1998. Sensitivity of phylogeny estimation to taxonomic sampling. Syst. Biol.

47, 18–31.

Poe, S., J. Velasco, K. Miyata, and E. E. Williams. 2009. Descriptions of two nomen

nudum species of Anolis lizard from northwestern South America. Breviora 516: 1–

16.

Polzin, T., Daneshmand, S.V., 2003. On Steiner trees and minimum spanning trees in

hypergraphs. Oper. Res. Lett. 31, 12–20.

Popper, K.R., 1959. The Logic of Scientific Discovery. Routledge, London [1992

reprint of 1968 edition].

Popper, K., 1968. The Logic of Scientific Discovery. Harper and Row, New York.

290

Popper, K.R., 1979. Objective Knowledge: An Evolutionary Approach. Oxford

University Press, New York.

Popper, K.R., 1983. Realism and the Aim of Science. Routledge, London.

Popper, K., 1992. Realism and the Aim of Science. Routledge, London.

Popper, K.R., 1989. Conjectures and Refutations: The Growth of Scientific Knowledge.

Routledge, London [1996 reprint].

Prim, R.C., 1957. Shortest connection networks and some generalizations. Bell

Pyron, R.A., Burbrink, F.T., Wiens, J.J., 2013. A phylogeny and revised classification

of Squamata, including 4161 species of lizards and snakes. BMC Evol. Biol., 13: 93.

Ramos-Pallares, E., Serrano-Cardozo, V.H., Ramirez-Pinilla, M.P., 2010.

Reproduction of Ptychoglossus bicolor (Squamata: Gymnophthalmidae) in an

Andean Coffee Shade Plantation in Colombia. South American Journal of

Herpetology 5, 143–150.

Reeder, T.W., 1996. A new species of Pholidobolus (Squamata: Gymnophthalmidae)

from the Huancabamba depression of northern Peru. Herpetologica 52, 282–289.

291

Reyes-Puig, J. P., M. Altamirano B., and M. H. Yánez-Muñóz. 2008. Reptilia,

Squamata, Gymnophthalmidae, Riama balneator and Riama vespertina: distribution

extension,

Rivas G. A., P. M. S. Nunes, J. R. Dixon, W. E. Schargel, J. R. Caicedo, T. R. Barros,

E. G. Camargo, and C. L. Barrio-Amorós. 2012. Taxonomy, hemipenial morphology

and natural history of two poorly known species of Anadia (Gymnophthalmidae)

from northern South America. Journal of Herpetology 46:33–40.

Rivas, G., W. E. Schargel, and J. M. Meik. 2005. A new species of Riama (Squamata:

Gymnophthalmidae), endemic to the Península de Paria, Venezuela. Herpetologica

61:461–468.

Rivas, G.A., de Freitas S., M., 2015. Discovery of the critically endangered Golden

Tree Frog, Phytotriades auratus (Boulenger, 1917), in eastern Venezuela, with

comments on its distribution, conservation, and biogeography. Herpetol. Rev. 46,

153–157.

Rivas, G.A., Molina, C.R., Ugueto, G.N., Barros, T.R., Barrio-Amorós, C.L., Kok,

P.J.R., 2012. Reptiles of Venezuela: an updated and commented checklist. Zootaxa

3211, 1–64.

Rivero, J.A., 1961. Salientia of Venezuela. Bull. Mus. Comp. Zool. 126, 1–207.

292

Robinson, D., Foulds, L., 1981. Comparison of phylogenetic trees. Math. Biosci. 53,

131–147.

Rodrigues M.T., Pellegrino, K.C.M., Dixo M., Verdade, V.K., Pavan, D., Argolo,

A.J.S., Sites-Jr, J.W., 2007. A new genus of microteiid lizard from the Atlantic

forests of state of Bahia, Brazil, with a new generic name for Colobosaura mentalis,

and a discussion of relationships among the Heterodactylini (Squamata,

Gymnophthalmidae). Am. Mus. Novit. 3565:1–27.

Rodrigues, M.T., Freire, E.M.X., Pellegrino, K.C.M., Sites Jr, J.W., 2005. Phylogenetic

relationships of a new genus and species of microteiid lizard from the Atlantic forest

of north-eastern Brazil (Squamata, Gymnophthalmidae). Zool. J. Linn. Soc. 144,

543–557.

Roscito, J.G., Rodrigues, M.T., 2010. Comparative cranial osteology of fossorial

lizards from the tribe Gymnophthalmini (Squamata, Gymnophthalmidae). Journal

of Morphology 271, 1352–1365.

Rosenberg, M.S., Kumar, S., 2001. Incomplete taxon sampling is not a problem for

phylogenetic inference. P. Natl. Acad. Sci. USA 98, 10751–10756.

Rota-Stabelli, O., Telford, M.J., 2008. A multi criterion approach for the selection of

optimal outgroups in phylogeny: Recovering some support for Mandibulata over

Myriochelata using mitogenomics. Mol. Phylogenet. Evol. 48, 103–111.

293

Rueda-Almonacid, J. V., A. Amézquita, G. Andrade-G., and O. Cortés. 2012. Fauna

silvestre amenazada del Departamento de la Guajira, Colombia. Corporación

Autónoma Regional de la Guajira - Corporación Colombia en Hechos. Serie Libros

Rojos Regionales, 194 Pp.

Rull, V., 2009. Pantepui. In: Gillespie, R.G., Clague, D.A. (Eds.), Encyclopedia of

Islands, University of California Press, pp. 717–720.

Ruthven, A.G., 1922. The amphibians and reptiles of the Sierra Nevada de Santa Marta,

Colombia. Misc. Publ. Mus. Zool. Univ. Michigan 8, 1–69.

Rydin, C., Källersjö, M., 2002. Taxon sampling and seed plant phylogeny. Cladistics

18, 485–513.

Sabaj Pérez, M. H. (ed). 2012. Standard symbolic codes for institutional resource

collections in herpetology and ichthyology: an Online reference. Version 3.0 (23

February 2012). Electronically accessible at www.asih.org, American Society of

Ichthyologists and Herpetologists, Washington D.C.

Sabaj Pérez, M.H. 2014. Standard symbolic codes for institutional resource collections

in herpetology and ichthyology: an Online Reference. Version 5.0 (22 September

2014). Electronically sccessible at http://www.asih.org/, American Society of

Ichthyologists and Herpetologists, Washington, DC.

294

Saint, K.M., Austin, C.C., Donnellan, S.C., Hutchinson, M.N., 1998. C-mos, a nuclear

marker useful for squamate phylogenetic analysis. Mol. Phylogenet. Evol. 10, 259–

263.

Sakaguchi, S., Qiu, Y.X., Liu, Y.H., Qi, X.S., Kim, S.H., Han, J., Takeuchi, Y., Worth,

J.R., Yamasaki, M., Sakurai, S., Isagi, Y., 2012. Climate oscillation during the

Quaternary associated with landscape heterogeneity promoted allopatric lineage

divergence of a temperate tree Kalopanax septemlobus (Araliaceae) in East Asia.

Mol. Ecol. 21, 3823–3838.

Salerno, P.E., Ron, S.R., Señaris, J.C., Rojas-Runjaic, F.J.M., Noonan, B.P., Cannatella,

D.C., 2012. Ancient tepui summits harbor young rather than old lineages of endemic

frogs. Evolution 66, 3000–3013.

Salzburger, W., Ewing, G.B., von Haeseler, A., 2011. The performance of phylogenetic

algorithms in estimating haplotype genealogies with migration. Mol. Ecol. 20,

1952–1963.

Sánchez-Pacheco, S.J., Rueda-Almonacid, J.V., Rada, M., 2010. Notes on the

occurrence of Riama simotera (Squamata, Gymnophthalmidae) in Colombia.

Herpetol. Bull. 113, 11–13.

Sánchez-Pacheco, S. J. 2010a. Lectotype designation and redescription of the

gymnophthalmid lizard Riama columbiana (Andersson, 1914) with notes on the type

locality. Papéis avulsos de Zoologia 50:31–41.

295

Sánchez-Pacheco, S. J. 2010b. A new “microteiid” lizard (Squamata:

Gymnophthalmidae: Riama) from southwestern Colombia. Herpetologica 66:349–

356.

Sánchez-Pacheco, S. J., D. A. Kizirian, and P. M. S. Nunes. 2011. A new species of

Riama from Ecuador previously referred to as Riama hyposticta (Boulenger, 1902)

(Squamata: Gymnophthalmidae). American Museum Novitates 3719:1–15.

Sánchez-Pacheco, S.J., Aguirre-Peñafiel, V., Torres-Carvajal, O., 2012. Lizards of the

genus Riama (Squamata: Gymnophthalmidae): the diversity in southern Ecuador

revisited. South Am. J. Herpetol. 7, 259–275.

Sánchez-Pacheco, S.J., Rueda-Almonacid, J.V., Caicedo-Portilla, J.R., Souza, S.M.,

2016. First record of Leposoma caparensis from Colombia, with confirmation for

the presence of Ptychoglossus myersi and P. plicatus (Squamata:

Gymnophthalmidae). Salamandra 52, 53–57.

Sánchez-Pacheco, S. J., O. Torres-Carvajal, V. Aguirre-Peñafiel, P. M. S. Nunes, L.

Verrastro, G. A. Rivas, M. T. Rodrigues, T. Grant, and R. W. Murphy. 2017.

Phylogeny of Riama (Squamata: Gymnophthalmidae), impact of phenotypic

evidence on molecular datasets, and the origin of the Sierra Nevada de Santa Marta

endemic fauna. Cladistics DOI:10.1111/cla.12203.

296

Sankoff, D., 1975. Minimal mutation trees of sequences. SIAM J. Appl. Math. 28, 35–

42.

Sarkar, I.N., Egan, M.G., Coruzzi, G., Lee, E.K., DeSalle, R., 2008. Automated

simultaneous analysis phylogenetics (ASAP): an enabling tool for phlyogenomics.

BMC Bioinformatics 9: 103.

Savage, J. M. 1997. On terminology for the description of the hemipenes of squamate

reptiles. Herpetological Journal 7:23–25.

Schargel, W.E., Rivas F., G., Myers, C.W. 2005. An enigmatic new snake from cloud

forest of the Península de Paría, Venezuela (Colubridae: genus Taeniophallus?). Am.

Mus. Novit. 3484, 1– 22.

Siddall, M.E. 1995. Another monophyly index: Revisiting the jackknife. Cladistics 11,

33–56.

Simmons, M.P., Randle, C.P., Freudenstein, J.V., Wenzel, J.W., 2002. Limitations of

Relative Apparent Synapomorphy Analysis (RASA) for measuring phylogenetic

signal. Mol. Biol. Evol. 19, 14–23.

Simpson, B., 1975. Pleistocene changes in the flora of the tropical Andes. Paleobiology

1, 273– 294.

297

Smith, E.N., Gutberlet Jr., R.L., 2001. Generalized frequency coding: a method of

preparing polymorphic multistate characters for phylogenetic analysis. Syst. Biol.

50, 156–169.

Song, Y.S., Hein, J., 2005. Constructing minimal ancestral recombination graphs. J.

Comput. Biol. 12, 147–169.

Spaulding, M., O'Leary, M.A., Gatesy, J., 2009. Relationships of Cetacea (Artiodactyla)

among mammals: Increased taxon sampling alters interpretations of key fossils and

character Evolution. PLoS ONE 4, e7062.

Steyermark, J.A., 1974. Relación florística entre la Cordillera de la Costa y la zona de

Guayana y Amazonas. Acta Botánica Venezuelica 9: 245–252.

Steyermark, J.A., 1979. Plant refuge and dispersal centres in Venezuela: their relict and

endemic element. In: Larsen, K., Holm-Nielsen, L.B. (Eds.), Tropical Botany.

Academic Press, London, pp. 185–221.

Steyermark, J.A., 1982. Relationship of some Venezuelan forest refuges with lowland

tropical floras. In: Prance, G.T. (Ed.), Biological diversification in the tropics.

Columbia University Press, New York, pp. 182–220.

Streicher, J.W., Schulte II, J.A., Wiens, J.J., 2016. How should genes and taxa be

sampled for phylogenomic analyses with missing data? An empirical study in

iguanian lizards. Systematic Biology 65, 128–145.

298

Sullivan, J., Swofford, D., 1997. Are guinea pigs rodents? The importance of adequate

models in molecular phylogenetics. J. Mammal. Evol. 4, 77–86.

Systems Techn. 36, 1389–1401.

Taboada, A., Rivera, L.A., Fuenzalida, A., Cisternas, A., Philip, H., Bijwaard, H.,

Olaya, J., Rivera, C., 2000. Geodynamics of the northern Andes: Subductions and

intracontinental deformation (Colombia). Tectonics 19, 787–813.

Takezaki, N., Nishihara, H., 2016. Resolving the phylogenetic position of coelacanth:

the closest relative is not always the most appropriate outgroup. Genome Biol. Evol.

8, 1208–1221.

Tarazona, O.A., Ramírez-Pinilla, M.P., 2008. The unusual orbitosphenoid of the

snakelike lizard Bachia bicolor. Journal of Anatomy 213, 120–130.

Tateno, Y., 1990. A method for molecular phylogeny construction by direct use of

nucleotide sequence data. J. Mol. Evol. 30, 85–93.

Templeton, A.R., Crandall, K.A., Sing, C.F., 1992. A cladistic analysis of phenotypic

associations with haplotypes inferred from restriction endonuclease mapping and

DNA sequence data. III. Cladogram estimation. Genetics 132, 619–633.

Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving the

sensitivity of progressinve multiple sequence alignment through sequence weighting,

299

position specific gap penalties and weight matrix choice. Nucl. Acids Res. 22, 4673–

4680.

Torres-Carvajal, O., 2007. Phylogeny and biogeography of a large radiation of Andean

lizards (Iguania, Stenocercus). Zool. Scr. 36, 311–326.

Torres-Carvajal, O., Lobos, S.E., 2014. A new species of Alopoglossus lizard

(Squamata, Gymnophthalmidae) from the tropical Andes, with a molecular

phylogeny of the genus. Zookeys 410, 105–120.

Torres-Carvajal, O., Lobos, S.E., Venegas, P.J., 2015. Phylogeny of Neotropical

Cercosaura (Squamata: Gymnophthalmidae) lizards. Mol. Phylogen. Evol. 93, 281–

288.

Torres-Carvajal, O., Lobos, S.E., Venegas, P.J., Chávez, G., Aguirre-Peñafiel, V.,

Zurita, D., Echevarría, L.Y., 2016. Phylogeny and biogeography of the most diverse

clade of South American gymnophthalmid lizards (Squamata, Gymnophthalmidae,

Cercosaurinae). Mol. Phylogen. Evol. 99, 63–75.

Torres-Carvajal, O., Mafla-Endara, P., 2013. Evolutionary history of Andean

Pholidobolus and Macropholidus (Squamata: Gymnophthalmidae) lizards. Mol.

Phylogen. Evol. 68, 212–217.

300

Tschanz, C.M., Marvin, R.F., Cruzb, J., Mehnert, H.H., Cebula, G.T., 1974. Geologic

evolution of Sierra-Nevada-de-Santa-Marta, northeastern Colombia. Geol. Soc. Am.

Bull. 85, 273–284.

Turchetto, C., Fagundes, N.J., Segatto, A.L., Kuhlemeier, C., Solís Neffa, V.G.,

Speranza, P.R., Bonatto, S.L., Freitas, L.B., 2014. Diversification in the South

American pampas: the genetic and morphological variation of the widespread

Petunia axillaris complex (Solanaceae). Mol. Ecol. 23, 374–389.

Uetz, P., Hošek, J., 2017. The Reptile Database (accessed January 30, 2017).

Available at http://www.reptile-database.org.

Uzzell, T. 1973. A revision of lizards of the genus Prionodactylus, with a new genus

for P. leucostictus and notes on the genus Euspondylus (Sauria, Teiidae). Postilla

159:1–67.

Uzzell, T., 1970. Teiid lizards of the genus Proctoporus from Bolivia and Peru. Postilla

142, 1– 39.

Uzzell, T.M., 1958. Teiid lizards related to Proctoporus luctuosus, with the description

of a new species from Venezuela. Occas. Pap. Mus. Zool. Univ. Mich. 597, 1–15.

Varón, A., Vinh, L.S., Wheeler, W.C., 2010. POY version 4, phylogenetic analysis

using dynamic homologies. Cladistics 26, 72–85.

301

Varón, A., Wheeler, W.C., 2012. The tree alignment problem. BMC Bioinformatics 13,

293.

Varón, A., Wheeler, W.C., 2013. Local search for the generalized tree alignment. BMC

Bioinformatics 14, 66.

Vitt, L.J., Avila-Pires, T.C.S., Esposito, M.C., Sartorius, S.S., Zani, P.A., 2007.

Ecology of Alopoglossus angulatus and A. atriventris (Squamata:

Gymnophthalmidae) in western Amazonia. Phyllomedusa 6, 11–21.

Wägele, J.-W., 2005. Foundations of Phylogenetic Systematics. Verlag Dr. Friedrich

Pfeil, München.

Wagner, W.H., 1961. Problems in the classification of ferns. In: Recent Advances in

Botany. University of Toronto Press, Toronto, pp. 841–844.

Walker, C.F., Test, F.H., 1955. New Venezuelan frogs of the genus Eleutherodactylus.

Occas. Pap. Mus. Zool. Univ. Michigan 561, 1–10.

Watkins, J.W.N, 1984. Science and scepticism. Princeton: University Press, and

London: Hutchinson.

Wheeler, W., Aagesen, L., Arango, C.P., Faivovich, J., Grant, T., D’Haese, C., Janies,

D., Smith, W.L., Varón, A., Giribet, G., 2006. Dynamic Homology and Phylogenetic

Systematics: A Unified Approach Using POY. American Museum of Natural

302

History, New York.

Wheeler, W.C., 1990. Combinatorial weights in phylogenetic analysis: a statistical

parsimony procedure. Cladistics 6, 269–275.

Wheeler, W.C., 1996. Optimization alignment: the end of multiple sequence alignment

in phylogenetics? Cladistics 12, 1–9.

Wheeler, W.C., 2003a. Iterative pass optimization of sequence data. Cladistics 19, 254–

260.

Wheeler, W.C., 2003b. Implied alignment: a synapomorphy-based multiple-sequence

alignment method and its use in cladogram search. Cladistics 19, 261–268.

Wheeler, W.C., 2012. Systematics: A Course of Lectures. Wiley-Blackwell, Oxford.

Wheeler, W.C., 2014. Phyletic groups on networks. Cladistics 30, 447–451.

Wheeler, W.C., Gladstein, D., De Laet, J., 1996–2003. POY: Phylogeny

Reconstruction via Optimization of DNA Data, version 3.0.

ftp://ftp.amnh.org/pub/molecular/poy.

Whiting, A.S., Bauer, A.M., Sites, J.W., Jr, 2003. Phylogenetic relationships and limb

loss in sub-Saharan African scincine lizards (Squamata: Scincidae). Mol.

Phylogenet. Evol. 29, 582–598.

303

Wiens, J.J. 2000. Coding morphological variation within species and higher taxa for

phylogenetic analysis. In: Wiens, J.J. (Ed.), Phylogenetic Analysis of Morphological

Data. Smithsonian Institution Press, Washington, pp. 115–145.

Wiens, J.J., 1995. Polymorphic characters in phylogenetic systematics. Syst. Biol. 44,

482–500.

Wiens, J.J., 1998. Testing phylogenetic methods with tree congruence: phylogenetic

analysis of polymorphic morphological characters in phrynosomatid lizards. Syst.

Biol. 47, 427–444.

Wiens, J.J., 2006. Missing data and the design of phylogenetic analyses. J. Biomed.

Inform. 39, 34–42.

Wiens, J.J., Fetzner, J.W., Parkinson, C.L., Reeder, T.W., 2005. Hylid frog phylogeny

and sampling strategies for speciose clades. Syst. Biol. 54, 719–748.

Wiens, J.J., Kuczynski, C.A., Duellman, W.E., Reeder, T.W., 2007. Loss and re-

evolution of complex life cycles in marsupial frogs: does ancestral trait

reconstruction mislead? Evolution 61, 1886–1899.

Wiley, E.O., Lieberman, B.S., 2011. Phylogenetics: theory and practice of phylogenetic

systematics, 2nd ed. Hoboken, NJ: Wiley-Blackwell.

304

Williams, D.M., Ebach, M.C., 2008. Foundations of Systematics and Biogeography.

Springer Science+Business Media, New York, NY.

Wooley, S.M., Posada, D., Crandall, K.A., 2008. A comparison of phylogenetic

network methods using computer simulation. PLoS One 3, e1913.

Yoder, A.D., Irwin, J.A., 1999. Phylogeny of the Lemuridae: Effects of character and

taxon sampling on resolution of species relationships within Eulemur. Cladistics 15,

351–361.

Zaher, H. 1999. Hemipenial morphology of the South American xenodontine snakes,

with a proposal for a monophyletic Xenodontinae and a reappraisal of colubroid

hemipenes. Bulletin of the American Museum of Natural History 240:1–168.

Zhou, C., Li, M., Yuan, K., Hu, N., Peng, W., 2011. Phylogeography of Ascaris

lumbricoides and A. suum from China. Parasitol. Res. 109, 329–338.

Zhu, L., Hu, Y., Qi, D., Wu, H., Zhan, X., Zhang, Z., Bruford, M.W., Wang, J., Yang,

X., Gu, X., Zhang, L., Zhang, B., Zhang, S., Wei, F., 2013. Genetic consequences

of historical anthropogenic and ecological events on giant pandas. Ecology 94,

2346–2357.

Zou, J., Sun, Y., Li, L., Wang, G., Yue, W., Lu, Z., Wang, Q., Liu, J., 2013. Population

genetic evidence for speciation pattern and gene flow between Picea wilsonii, P.

morrisonicola and P. neoveitchii. Ann. Bot. 112, 1829–1844.

305

Zwickl, D.J., Hillis, D.M., 2002. Increased taxon sampling greatly reduces

phylogenetic error. Syst. Biol. 51, 588–598.

306