Caudata: Plethodontidae) Kenneth Paul Wray III

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2013 Patterns and Processes of Diversification in Salamanders of the Subfamily Spelerpinae (Caudata: Plethodontidae) Kenneth Paul Wray III

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] THE FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

PATTERNS AND PROCESSES OF DIVERSIFICATION IN SALAMANDERS OF THE

SUBFAMILY SPELERPINAE (CAUDATA: PLETHODONTIDAE)

KENNETH PAUL WRAY III

A Dissertation submitted to the Department of Biological Science in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Summer Semester, 2013 Kenneth Paul Wray III defended this dissertation on July 1, 2013. The members of the supervisory committee were:

Scott J. Steppan Professor Directing Dissertation

William C. Parker University Representative

Joseph Travis Committee Member

Austin R. Mast Committee Member

Peter Beerli Committee Member

D. Bruce Means Committee Member

The Graduate School has verified and approved the above-named committee members, and certifies that the dissertation has been approved in accordance with university requirements.

For Maria, Maya, Kylie, and Eirinn: Without you, none of this would have been possible. For Mom and Dad who always let me be me. And for Flavio Morrissiey, who reset me on this path one fateful summer day all those years ago.

iii

ACKNOWLEDGMENTS

I’d would like to acknowledge my fantastic committee for all of their mentoring and the knowledge they have passed on to me: Scott Steppan, Peter Beerli, Austin Mast, D. Bruce Means, William Parker, and Joseph Travis. I appreciate it more than you all will ever know. A very special thank you to Scott Steppan, the best advisor a grad student like myself could ever have asked for. Also much gratitude to my honorary committee member: John Schenk. I’d like to extend my gratitude to Joseph Apodaca, Dick Bartlett, Nathanael Herrera, Pierson Hill, Flavio Morrissiey, and Joseph Pfaller for providing assistance in the field and life long friendships out of it. Thank you all for making the daily grind worthwhile. I’d like to thank David Beamer, Roger Birkhead, Ron Bonett, Brandon Bowers, Robb Brumfield (Louisiana State University Collection of Genetic Resources), Daniel Dye, Michael Dye, Sean Graham, Troy Hibbitts, Toby Hibbitts (Texas Cooperative Wildlife Collection Division of Herpetology), John Himes, John Jensen, Ken Kozak, Barry Mansell, Ryan Means, Paul Moler, Tom Morris, Matt Nordgren, Rebecca Pfaller, Kim Sash, Jake Scott, Mike Steffen, Dirk Stevenson, Mizuki Takahashi, Wayne VanDevender, Larry Wilson, and John Willson for the loan of sequences, tissues and/or specimens. Matthew Niemiller, Brice Noonan and John Schenk provided valuable discussion on analyses and comments on early drafts of the manuscripts. I am grateful to Brice Noonan for opening his lab to me and for instruction on developing anonymous nuclear loci. Funding for this research was provided through a Robert B. Short Grant in Zoology, Texas Herpetological Society grant, East Texas Herpetological Society grant, and NSF funding to Scott J. Steppan (DEB-0841447).

TABLE OF CONTENTS

List of Tables ...... vi List of Figures...... viii Abstract...... x 1. HISTORICAL BIOGEOGRAPHY AND DIVERSIFICATION IN A MAJOR LINEAGE OF SALAMANDERS (PLETHODONTIDAE: SPELERPINAE) ...... 1

2. GENETIC DIVERGENCE IN THE DWARF SALAMANDER (Eurycea quadridigitata) COMPLEX: TESTING MECHANISMS OF SPECIATION ...... 42 3. SPECIES DELIMITATION AND MORPHOLOGICAL VARIATION IN THE DWARF SALAMANDER (Eurycea quadridigitata) COMPLEX ...... 74 REFERENCES ...... 129

BIOGRAPHICAL SKETCH ...... 142

LIST OF TABLES

1.1 Specimens and sequences used in the Spelerpinae phylogenetic analyses...... 37

1.2 Specimens and sequences used in the Spelerpinae phylogenetic analyses...... 40

1.3 Estimated ages of key spelerpine nodes from BI chronogram in Fig. 3 (95% HPD credible interval in millions of years)...... 41

2.1 Comparison of abiotic and biotic factors of Southeaster Coastal Plain wetlands known to be used as breeding sites for the Dwarf Salamander (Eurycea quadridigitata) ...... 68

2.2 Specimens and corresponding locality data, including the categorical scores for each hypothesis test...... 69

2.3 Amplification and sequencing primers for mitochondrial genes (tRNA-Met, ND2, tRNA- Trp, tRNA-Ala, tRNA-Asn, Olrep, tRNA-Cys, tRNA-Tyr, partial CO1) used in this study ...... 73

2.4 Results of Shimodaira-Hasegawa and Approximately Unbiased tests of the constrained topologies representing the five hypotheses compared to the unconstrained ML tree in Fig. 2 ...... 73

3.1 Samples used in molecular analysis of species delimitation in the Eurycea quadridigitata complex...... 115

3.2 Primers and PCR protocols used in amplification and sequencing ...... 117

3.3 Samples and corresponding measurements for 12 morphological traits in the Eurycea quadridigitata complex...... 118

3.4 Population sample means and standard deviations from 12 morphological traits of the Eurycea quadridigitata complex ...... 125

3.5 Principal component loadings for 11 morphological traits of the Eurycea quadridigitata complex...... 125

3.6 Results of one-way Type III SS ANOVAs and Kruskal-Wallis tests for 11 morphological traits of the Eurycea quadridigitata complex ...... 126

3.7 Results of pairwise t-tests and Wilcoxon Rank Sum tests for 11 morphological traits of the Eurycea quadridigitata complex ...... 127

3.8 Group means of discriminant functions and discriminant function loadings for the Eurycea quadridigitata complex ...... 128

vii

LIST OF FIGURES

1.1 Physiographic regions of eastern North America: AH (Appalachian Highlands), AP (Atlantic Plain), IH (Interior Highlands), IP (Interior Plains), LU (Laurentian Uplands) ...... 26

1.2 Maximum likelihood phylogram of the Spelerpinae (Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, and Urspelerpes) based on the mitochondrial CytB, ND2, and ND4 genes and nuclear RAG1...... 28

1.3 BI chronogram of the Spelerpinae (Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, and Urspelerpes) based on the mitochondrial CytB, ND2, and ND4 genes and nuclear RAG1...... 30

1.4 Lineage through time plots of the genus Eurycea and other Spelerpines (Gyrinophilus+Pseudotriton+Stereochilus clade) ...... 31

1.5 DEC ancestral range and split estimations of the Spelerpinae under the adjacent areas rates dispersal model using BI chronogram from Fig. 3...... 33

1.6 Sea level fluctuations of eastern North America from the Late Cretaceous through Pliocene...... 35

2.1 Geographic distribution of the E. quadridigitata complex (denoted by gray line) ...... 63

2.2 Maximum likelihood phylogram of the mitochondrial ND2, tRNAs, and partial CO1 genes ...... 65

2.3 Maximum likelihood phylogram from Fig. 2 showing just the E. quadridigitata complex and the five hypotheses (E = Ecoregion, Rs = River Subunit, Ra = River Accounting, Rc = River Cataloguing, and H = Habitat)...... 67

3.1 Map of Eurycea quadridigitata and E. chamberlaini samples used in A) species tree analysis for BPP species delimitation method and B) morphological analyses ...... 98

3.2 The species tree from the *BEAST analysis of four nuclear and one mitochondrial loci ...... 100

3.3 Anonymous locus 02 (AL02) gene tree from *BEAST analysis...... 102

3.4 Anonymous locus 21 (AL21) gene tree from *BEAST analysis...... 104

3.5 Anonymous locus 51 (AL51) gene tree from *BEAST analysis...... 106

3.6 Recombination activating gene 1 (RAG1) gene tree from *BEAST analysis...... 108

viii

3.7 Mitochondrial (CytB + ND2) gene tree from *BEAST analysis...... 110

3.8 Plot of PC1 versus PC2 for the 11 morphological traits (n = 199)...... 112

3.9 Plots of four linear discriminant functions (LD1-LD4) for the 11 morphological traits (n = 199)...... 114

ABSTRACT

Biologists have long been interested in the diversity of organisms on earth. With their joint presentation to the Linnean Society of London in 1858, Darwin and Wallace proposed natural selection as a clear mechanism to explain the diversity of life. In the 155 years since this seminal presentation, evolutionary biologists have explored the patterns and processes of diversification in a vast number of taxonomic groups, including proposing additional mechanisms and confirming the dominant presence of natural selection in the role of diversification. The last few decades have seen rigorous debates on the study of these patterns and processes (e.g. adaptive radiation, species concepts) and advances in theory, data acquisition, and analytical methods to address a number of questions associated with diversification. Yet, much of the attention has been on model systems, resulting in deficiencies in the knowledge of how widespread certain phenomena are (e.g. adaptive radiation) or how general certain modes of speciation may be. In the present study, I explore various patterns and processes responsible for the diversification of salamanders in the Spelerpinae (Caudata: Plethodontidae). I examine the role of adaptive radiation in the subfamily using a completely sampled phylogeny based on multiple loci and testing three widely held predictions of the process. I also test three hypotheses to explain the diversification patterns observed in a range-wide phylogeographic analysis of the

Eurycea quadridigitata species complex. Finally, I use a multilocus nuclear phylogeny and a

Bayesian species delimitation method to test whether genetic lineages that are linked to breeding habitats represent distinct species. Using the molecular species delimitation results, I look for congruence among 11 morphologic traits. I find that adaptive radiation is not a good model for the diversification of the Spelerpinae. Instead, early range expansion followed by several vicariant events and in situ diversification best explains the diversity in this group. In addition, I

show strong molecular evidence that habitat isolation has likely lead to ecological speciation in

at least three lineages of the E. quadridigitata complex, with moderate support from morphology.

CHAPTER ONE

HISTORICAL BIOGEOGRAPHY AND DIVERSIFICATION IN A MAJOR LINEAGE OF SALAMANDERS (PLETHODONTIDAE: SPELERPINAE)

1.1 Introduction

1.1.1 Background

Adaptive radiation results when an ancestral population diverges into multiple descendent species due to natural selection on ecologically important traits (Dobzhansky 1948; Simpson

1953; Schluter 2000; Gavrilets & Losos 2009). A key feature of adaptive radiations involves ecological opportunity, whereby the ancestral population is able to diversify into multiple lineages through one (or a combination) of three mechanisms: migration into a novel environment, evolution of key innovation(s), or extinction of competitors (Simpson, 1953; see

Givnish, 1997; Losos, 2010; and Yoder et al., 2010 for recent reviews of the role of ecological opportunity). When a population experiences one or more of these mechanisms, ecological release may occur, freeing the ancestral population from selection on some important ecological trait(s) which can lead to morphological and species diversification (Yoder et al. 2010).

Most definitions of adaptive radiation agree that ecological opportunity is the catalyst that sets in motion three key aspects of an adaptive radiation: 1) multiplication of species, 2) adaptive divergence, and 3) extraordinary diversity, though the latter is controversial (Givnish 1997;

Schluter 2000; Gavrilets & Losos 2009; Glor 2010). As organisms experience ecological release, they have the potential to move into new niches that were previously unavailable to them. This access to previously unavailable resources is expected to be followed by an early burst of speciation as populations rapidly diverge from one another and fill different parts of niche space.

This rapid speciation is expected to be closely tied to rapid divergence in phenotype, since 1 selection will favor adaptations that allow organisms to exploit these novel resources. As niches rapidly become filled, this burst of speciation and morphological diversification is expected to slow down compared to shortly after the ecologic opportunity (Schluter 2000; Losos & Mahler

2010; Glor 2010). The most contentious aspect of adaptive radiation is whether extraordinary diversification is a requirement and what constitutes “extraordinary” (for detailed reviews see

Givnish 1997; Losos & Mahler 2010). Glor (2010) effectively argued that this is an unnecessary requirement, giving suggestions as to how patterns due to adaptive radiation could be recognized and used to understand the underlying processes involved. I apply aspects from this general theory of adaptive radiation to examine whether patterns in a lineage of eastern North American salamanders are the likely result of an adaptive radiation.

1.1.2 Study System

Salamanders (Order: Caudata) are second only to frogs in diversity among modern amphibians and represent a major tetrapod radiation. The approximately 650 species (Frost

2013) are found throughout most temperate regions of the Northern Hemisphere, with a single radiation (subfamily Bolitoglossinae) occurring in parts of the Neotropics. The highest salamander diversity occurs in North America (particularly in the United States), where nine of the ten extant families occur. Consequently, salamanders are one of the better-studied tetrapod groups, having served as model systems for physiology, anatomy, community ecology, and numerous studies of evolutionary processes (see Duellman & Trueb (1986) for detailed review).

Despite the attention salamanders have received, some lineages are still poorly understood.

At 433 species (66%), the family Plethodontidae represents the largest family of salamanders (Frost 2013). Several major lineages within this family have received considerable

attention from evolutionary biologists studying adaptive and nonadaptive radiations (Wake 1987,

2006; Moritz et al. 1992; Wiens et al. 2006; Kozak, Weisrock, et al. 2006; Kozak & Wiens

2006, 2010; Vieites et al. 2007). Studies have focused on the high species richness of plethodontids in eastern North America and the processes responsible for this exceptional diversity. A number of studies have provided evidence that niche conservatism, the pattern wherein a group of organisms retain the ancestral niche and its associated ecological traits

(Wiens & Graham 2005), has played a major role in the diversification of this group (Kozak,

Weisrock, et al. 2006; Kozak & Wiens 2006, 2010). Niche conservatism has been particularly important in explaining the diversity seen in the species-rich genera Plethodon and

Desmognathus. These studies supported the hypothesis that the ancestral niche was similar to present-day conditions that prevail at mid-elevations throughout much of the Appalachian

Mountains (Kozak, Weisrock, et al. 2006; Kozak & Wiens 2006, 2010). In the past, when cooler and moister conditions prevailed, salamanders moved lower in elevation and into the surrounding valleys, allowing them to disperse to other areas. As the climate became warmer and drier, salamanders migrated to higher elevations following ancestral niche conditions, where they were cut off from other populations, eventually becoming reproductively isolated and experiencing allopatric speciation. Kozak & Wiens (2010) also suggested that niche conservatism might explain the hump-shaped pattern of elevation vs. species richness seen in these two genera. Despite this high species richness, Plethodon and Desmognathus show very little morphological disparity outside of body size divergence (Kozak et al. 2005; Kozak et al.

2006; Kozak et al. 2009), a pattern similar to that observed in the western Batrachoseps (Wake

2006).

A third, species-rich radiation of plethodontid salamanders exists in eastern North

America for which niche conservatism does not appear to explain the pattern of diversification

(Kozak & Wiens 2010). The subfamily Spelerpinae (cf. Chippindale et al. 2004) is comprised of

35 species in six genera: Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, and

Urspelerpes (Tilley et al. 2008). In examining the role niche conservatism has played in the diversification of eastern plethodontid salamanders, Kozak & Wiens (2010) found a monotonic pattern of species richness versus elevation in spelerpines, in which the highest observed diversity occurred at low elevations. They also found no correlation between the time an elevational band was occupied by spelerpines and its observed species richness (i.e., no support for the time-for-speciation effect of Stephens & Wiens (2003)) and in stark contrast to the direct relationship observed in Plethodon and Desmognathus.

The overall species richness within the Spelerpinae appears to be unbalanced. Of the six genera, three (Haideotriton, Stereochilus, and Urspelerpes) are monotypic, with Gyrinophilus

(four species) and Pseudotriton (two species) only slightly more diverse. This leaves the genus

Eurycea accounting for nearly 75% of spelerpine diversity. Further, this diversity is an

underestimate as several species complexes have been shown to consist of multiple, deep genetic

lineages, at least some of which represent undescribed species (Kozak et al., 2006a; Wray,

unpublished data). For example, based on the work of Kozak et al. (2006a), Adams et al. (2009)

estimated an additional 10 undescribed Eurycea species based on the work of Kozak et al.

(2006).

In recent years, many phylogenetic tools and methods have helped to identify and

understand the complexities of adaptive radiations (Losos & Mahler 2010; Yoder et al. 2010;

Glor 2010). Herein, I employ phylogeny, divergence time estimation, lineage-through-time plots,

diversification rate estimates, and ancestral geographic range estimation to test whether the

genus Eurycea may have experienced an adaptive radiation. Specifically, I examine whether

ecological opportunity, via migration into new environments, played a role in explaining the

unbalanced species richness observed in the spelerpine salamanders.

I also attempt to resolve the relationships of several enigmatic spelerpine taxa.

Gyrinophilus subterraneus has never been included in a molecular phylogenetic analysis and its

relationships to other species is unknown; specifically, it is unclear whether it is the sister taxon

to the other cave dwelling Gyrinophilus spp. or more closely related to the widespread surface

dwelling G. porphyriticus. The monotypic Haideotriton wallacei was included in an analysis of the Eurycea of the Edwards Plateau of central Texas (Chippindale et al. 2000), as it was considered to be closely allied with this genus. However, their analyses were inconclusive and it was unclear whether it was the sister taxon to Eurycea or nested within the genus. Finally, I examine the relationships of widespread species, E. quadridigitata, and its relationships within the genus. Previous studies (Chippindale et al. 2000; Kozak et al. 2009; Lamb & Beamer 2012)

have suggested that the species is paraphyletic with respect to the Edwards Plateau clade, but

these studies did not have complete spelerpine sampling.

1.2 Materials and Methods

1.2.1 Taxon Selection and Sampling

I sampled 34 species representing all currently described spelerpine salamanders, except

E. robusta, which has not been seen since it was first discovered in 1951 and for which no viable

tissue exists (Chippindale et al. 2000). In addition, the sampling included all but three currently

recognized subspecies (P. montanus montanus, P. m. flavissimus, and G. palleucus necturoides).

I also included a number of putative species, primarily from the E. quadridigitata complex

(Wray, unpublished data) and E. bislineata complex (Kozak et al. 2006). Whenever possible, I

included at least two samples of each taxon to help identify any misidentification of specimens,

hybridization, or mitochondrial introgression (Barraclough & Nee 2001). Following Vieites et al.

(2007), I chose five members of the genus Batrachoseps (representing each of the five species

groups), for outgroup analysis. In total, 82 individuals were included in the analyses (Table 1.1).

I generated 137 novel sequences for this study (Table 1.1), with the remaining sequences

(n = 156) downloaded from GenBank. Whenever possible, I attempted to generate or download

all sequences for an OTU from the same individual. However, in seven samples of the ingroup

and with all five outgroup samples, this was not possible. In the seven ingroup samples, the

species involved were either restricted to small, geographic ranges, did not overlap with a

morphologically similar, close relative (E. latitans, E. nana, E. naufragia, E. sosorum, E.

tonkawae, and E. troglodytes), or belonged to a monotypic genus (U. brucei). Additionally, all of

these species are sufficiently rare and are difficult to obtain tissue samples from (several are

federally or state protected), so I deemed it important to combine these samples and include them

in the analyses. Sequences generated in this study were obtained from tissue samples previously

collected in the field or borrowed from the Texas Cooperative Wildlife Collection Division of

Herpetology or from private collections. Tissue samples consisted of tail tips or liver samples and were preserved in 95% ethanol stored at -80°C.

1.2.2 DNA Extraction, Amplification, and Sequencing

Genomic DNA was extracted using the hot phenol-chloroform-isoamyl alcohol/chloroform-isoamyl alcohol method (Sambrook & Russell 2001). Extracts were

visualized on agarose gels and DNA concentration quantified using a Nanodrop ND-1000

Spectrophotometer (NanoDrop, Wilmington, DE, USA). Polymerase chain reaction (PCR) was

performed on the mitochondrial cytochrome b (CytB), NADH dehydrogenase subunit 2 (ND2),

and NADH dehydrogenase subunit 4 (ND4) genes, as well as the first section of the nuclear

recombination-activating gene 1 (RAG1). The following reagents and concentrations were used

in all PCR runs: 13.3 µl distilled water, 5 µl 5X Colorless GoTaq Reaction Buffer, 1.5 µl MgCl2

(25 mM), 1.5 µl dNTPs (2.5 mM), 0.2 µl GoTaq (5u/ul), 0.5 µl bovine serum albumin (10 mg/ml), 1.0 µl of each primer (10 ng/ul). Amplification was performed using the primers and thermal cycler programs listed in Table 1.2. A negative and positive was run for all PCR amplifications. Amplification products were cleaned enzymatically with Affymetrix-USB

ExoSAP-IT PCR Product Clean-up kits (USB Corporation, Cleveland, OH, USA).

Sequencing reactions were performed at the Florida State University Sequencing Facility using an Applied Biosystems 3130xl Genetic Analyzer with capillary electrophoresis (Applied

Biosystems Inc., Foster City, CA, USA) or at the DNA Analysis Facility at Yale University using an Applied Biosystems 3730xl Genetic Analyzer (Applied Biosystems Inc.). Sequencing was conducted using the amplifying primers and two internal sequencing primers were also used for ND2 (Table 1.2). Sample sequence lengths varied within genes: CytB (493–1112 bp), ND2

(499–1039 bp), ND4 (638–725 bp), RAG1 (481–1467 bp). All novel sequences generated for this study were deposited in GenBank (Table 1.1).

1.2.3 Phylogenetic Analyses and Divergence Time Estimates

A total of 293 sequences representing 39 plethodontid species (including all but one currently described member of the Spelerpinae) and a number of putative species and divergent

7 lineages were used in the alignment. Sequences were aligned and edited using Geneious v. 5.5.7

(Drummond et al. 2012). I used the Geneious alignment algorithm for initial alignment and then adjusted manually. Genes were translated to amino acids to ensure there were no premature stop codons and to verify the alignment. No premature stop codons were detected in any of the four protein-coding genes used, suggesting that the amplicons were not orthologous (Zhang & Hewitt

1996). The resulting alignment was unambiguous.

I used jModeltest v. 2.1.1 (Guindon & Gascuel 2003; Darriba et al. 2012) to fit 88 DNA- substitution models to the alignment, using the Akaike Information Criterion (AIC; Akaike,

1974) and determined the best substitution model of nucleotide evolution for each of the four genes. The AIC scores determined that GTR+ I+Γ was the best-fit model of evolution for each gene, which is consistent with other studies of plethodontid salamanders using these markers

(Kozak et al., 2005; Wiens et al., 2006; Kozak & Wiens, 2006; Vieites et al., 2007).

Phylogeny and divergence time estimates were then conducted on the four concatenated genes using maximum likelihood (ML) as implemented in PAUP* v. 4.0a126 (Swofford 2003) and using Bayesian inference (BI) as implemented in BEAST v. 1.7.4 (Drummond & Rambaut

2007). For the ML analysis, a heuristic search with 10 stepwise random-addition sequence replicates using the tree bisection-reconnection method was performed on the data set. To assess support for the ML tree, I also performed a nonparametric bootstrap analysis using 1000 pseudoreplicates with 100 stepwise random-addition sequence replicates. The ML tree from the initial search was used as a starting tree for the BEAST analysis, which was partitioned by gene and by codon.

For the BI analysis, I performed two independent runs of 2 x 108 generations, sampling every 1000 generations. I combined the resultant data sets using LogCombiner v. 1.7.4

(Drummond & Rambaut 2007). I checked for convergence using AWTY (Nylander et al. 2008) and Tracer (Rambaut & Drummond 2007). I used Tracer to check for stationarity and discarded the first 4 x 104 trees as burnin. The divergence time estimations utilized an uncorrelated lognormal relaxed molecular clock (Drummond et al. 2006), a birth-death speciation prior

(Gernhard 2008), and five fossil-based node calibrations. I used a lognormal prior distribution on all fossil calibrations (Ho 2007) with mean = 0 and standard deviation = 1.0. For the offset of each prior, I used the lower boundary of the time period the fossil was confirmed from, making this a conservative estimate of the node age. I constrained the outgroup (Batrachoseps) to be monophyletic and calibrated the basal node of Batrachoseps at 4.9 Ma using a fossil

Batrachoseps sp. from the late Hemphillian North American land mammal age (NALMA; Clark

1985; Holman 2006). This fossil is a second trunk vertebra and was identified to family level based on the presence of anterior and posterior spinal nerve foramina and to the genus level based on the presence of a posteriorly deflected diapophysis, a synapomorphy of the genus

Batrachoseps. A fossil sample of G. porphyriticus and one of P. ruber, both from the

Irvingtonian NALMA, were used to calibrate their respective nodes at 0.24 Ma (Holman 1977,

1995, 2006; Holman & Grady 1987). The Gyrinophilus fossil consisted of a presacral vertebra identified to species based on the very low neural spines with no raised central portion, widely separated transverse processes, and an unnotched distal end of the neural arch. The Pseudotriton fossil is also a presacral vertebra, similar in appearance to that of the Gyrinophilus, but is distinct in having prominent neural spines The final two fossil calibrations were of E. lucifuga and E. cirrigera and came from the same Rancholabrean NALMA deposit 0.011 Ma at Cheek Bend

Cave, Maury County, Tennessee (Miller 1992; Holman 2006). The E. lucifuga fossils were identified based on their relatively large size and straight, dorsal crest of the atlas. Eurycea

cirrigera fossils consisted of trunk vertebrae that are much smaller, have pseudo-opisthocoelous

centra, weakly developed neural ridges, and well developed basapophyses. The results of Kozak

et al. (2006) indicated that the E. cirrigera from this region of Tennessee are a distinct, divergent

lineage that may represent a distinct species. Given these results, combined with the widespread

present-day range of this species, I used this fossil to calibrate the node for the E. cirrigera A

(clade C of Kozak et al. 2006).

1.2.4 Ancestral Range Estimation

I used the likelihood-based method that employs the dispersal-extinction-cladogenesis

(DEC) model as implemented in LAGRANGE (Ree et al. 2005; Ree & Smith 2008) to estimate the ancestral ranges and descendant splitting events of the Spelerpinae. This was done with the chronogram generated from the BEAST analysis and five physiographic regions I chose (Fig.

1.1), encompassing the entire range of the extant Spelerpinae: Appalachian Highlands (AH),

Atlantic Plain (AP), Interior Highlands (IH), Interior Plains (IP), and Laurentian Uplands (LU).

These regions are based on landforms and not vegetation or climate, having persisted in their current areas and geographic relationships for over 100 million years (Fenneman 1916; Vigil et al. 2000). In fact, four of the five are part of the North American (Laurentian) Craton, with only the AP occurring outside of this tectonically stable region. Nonetheless, even the AP has existed, at least in part, during much of the last 60–70 million years. This long-term stability makes the use of physiographic regions informative for reconstructing the ancestral ranges of the

Spelerpinae.

LAGRANGE allows taxa to occur across multiple regions, so I scored all 82 OTUs for presence/absence in each of the five regions in the matrix. Given the small size and presumed

10 low vagility of plethodontid salamanders, I used an unequal rates model of dispersal between the five areas, allowing dispersal only between adjacent areas (Fig. 1.1). In order to test the robustness of these predictions, I also performed a more conservative, equal rates analysis in which dispersal was allowed between any two areas with equal probability and compared the ln likelihoods of each model. In LAGRANGE, statistical significance between two models is assessed using the method of Edwards (1992), wherein a score of two log-likelihood units or more is considered significantly different.

1.2.5 Lineage-through-time Plots, Gamma Statistic, and Diversification Rates

I used the APE package v. 3.0-8 (Paradis et al. 2004) as implemented in R statistical package v. 2.15.3 (R Core Team 2013) to create lineage-through-time (LTT) plots using the chronogram from the BEAST analysis (Paradis et al., 2004; R Development Core Team, 2013). I generated a LLT plot for the Eurycea clade and the Gyrinophilus+Pseudotriton+Stereochilus clade. I pruned the chronogram so that there was only one OTU per species before the LLT plots were created. The only exceptions to the pruning was if an OTU was thought to represent a putative species, or when a taxon was rendered paraphyletic, in which case I included samples of that taxon from each clade it belonged to.

I also used APE to calculate the gamma statistic, which indicates whether nodes are closer to the tips or root than would be expected under a pure birth model (Pybus & Harvey

2000). I then conducted a constant-rates test to test the significance of the gamma statistic against a pure birth model. In order to test for differences in rates of diversification I used the program

SYMMETREE v. 1.1 (Chan & Moore 2005), which examines whether a given tree has experienced significant diversification rates among its branches and which branches specifically

may have experienced these significant rate shifts. The method is robust to phylogenetic

uncertainty (such as in branch length estimates) and differs from some other methods of rate

diversification in that it is a topological approach, directly comparing the difference in species

diversity between select clades instead of temporally looking at the distribution of branching

events.

1.3 Results

1.3.1 Phylogenetic Analyses and Divergence Time Estimates

These results represent the most comprehensive Spelerpinae phylogeny to date, both in taxonomic and character (number of base pairs) sampling. The ML phylogram with bootstrap support (BS) and Bayesian posterior probabilities (PP) is reported in Figure 1.2. The ML and BI trees were almost completely congruent and most major nodes were strongly supported with very few exceptions (Fig. 1.2). The ML analysis had poor support for the relationships among the southern Edwards Plateau neotenes (M, Fig. 1.2), yielding a polytomy between the E. latitans, E. troglodytes, (E. rathbuni, E. waterlooensis), and ((((((E. neotenes A, E. neotenes B), E. sp.

Comal Springs), E. pterophila), E. tridentifera), E. sosorum), E. nana) clades, whereas the BI formed a well supported clade between the latter two, which in turn formed a well supported clade sister to a E. latitans+E. troglodytes clade. The other discrepancy between the two analyses was in the placement of the enigmatic H. wallacei. Both analyses placed all three samples in a well-supported ((H. wallacei A, H. wallacei C), H. wallacei B) clade. However, the ML analysis placed this clade in a strongly supported polytomy with the ((E. aquatica A, E. aquatica B), (E. junaluska A, E. junaluska B)) and ((E. cirrigera C, E. wilderae), E. wilderae/cirrigera B) clades.

In the BI tree, there was weak support for the H. wallacei clade being sister to the ((E. cirrigera

C, E. wilderae), E. wilderae/cirrigera B) clade, which, together, was sister with the E. aquatica+E. junaluska clade.

The analyses revealed two major lineages within the Spelerpinae (B, Fig. 1.2): one clade representing the genera Gyrinophilus, Pseudotriton, and Stereochilus (H, Fig. 1.2) and the other the genera Eurycea, Haideotriton, and Urspelerpes (C, Fig. 1.2). Pseudotriton was the sister taxon to Stereochilus (I, Fig. 1.2). There was a deep divergence between the two species of

Pseudotriton (L, Fig. 1.2). Additionally, there was also a deep divergence between the two subspecies of P. montanus used. Urspelerpes formed the deeply divergent sister taxon to

Eurycea and Haideotriton (C, Fig. 1.2).

Within Eurycea, five strongly supported lineages are recovered (D, Fig. 1.2). The first lineage is the clade containing the IH endemics (G, Fig. 1.2): (E. multiplicata, (E. tynerensis, E. spelaea)). These three species formed the sister clade to all other Eurycea. Within the remaining

Eurycea, two clades are represented. One of these (E, Fig. 1.2) contained the E. lucifuga complex (E. guttolineata, E. longicauda, and E. lucifuga; O, Fig. 1.2) and the E. bislineata complex (E. aquatica, E. bislineata, E. cirrigera, E. junaluska, and E. wilderae; N, Fig. 1.2). The troglobitic H. wallacei was contained within the E. bislineata complex clade. The sister group (F,

Fig. 1.2) to the clade containing the E. bislineata and E. lucifuga complexes consisted of the

Edwards Plateau neotenes (M, Fig. 1.2) and two samples of E. quadridigitata originating from

Mississippi and Texas (K, Fig. 1.2), sister to a group (J, Fig. 1.2) of several, deeply divergent lineages of E. quadridigitata with E. chamberlaini nested within these lineages.

The origin of the Spelerpinae (Fig. 1.3, Table 1.3) occurred approximately 72 Ma with a

95% highest posterior density confidence interval of 112–45 Ma (herein represented as 72 Ma

[112–45]), sometime in the Cretaceous through the early Eocene. The clades

Eurycea+Urspelerpes and Gyrinophilus+Pseudotriton+Stereochilus diverged at approximately

61 Ma [97–38.5] and 35 Ma [55–21.5], respectively. Within the latter, the first origin of

Pseudotriton (27 Ma [43–16.5]), and Stereochilus (33.5 Ma [53–20]) occurred much earlier than

Gyrinophilus (10 Ma [16–6.5]). The genus Eurycea first appeared around 42 Ma [65–26.5]

sometime between the Paleocene through Oligocene. In contrast to the genera in the

Gyrinophilus+Pseudotriton+Stereochilus clade, the major lineages within Eurycea split close in

time. The E. bislineata+E. lucifuga complex originated around 37 Ma [57–23.5] during the

Eocene or Oligocene. This was followed by the appearance of the E. quadridigitata and Edwards

Plateau neotenes (35 Ma [56–23]) and the IH lineage (35 Ma [56–21.5]), also during the Eocene

or Oligocene. The former diversified into eastern (29 Ma [45–18.5]) and western (27 Ma [43–

17]) lineages. Numerous additional diversifications have occurred within these major spelerpine

lineages, ranging from very shallow (e.g., some Edwards Plateau neotenes and the genus

Gyrinophilus) to late Oligocene–Miocene (e.g., E. quadridigitata and E. bislineata complexes).

1.3.2 Ancestral Range Estimation

The unequal rates of dispersal model had a significantly better ln likelihood (-136.82)

than the equal rates of dispersal model (-142.98), though the two dispersal models yielded

largely congruent results, with 69 out of a possible 76 nodes estimated with the same ancestral

range. Five of the seven nodes in disagreement differed in having the best estimate of the

ancestral range in one model swapped with the second best estimate in the other model and vice

versa. The only two nodes where the results were quite different were shallow divergences

within the E. bislineata complex (N, Fig. 1.5). Since it significantly fit the data better, herein, I

refer to the results from the unequal rates of dispersal model.

The best ancestral range estimation model supported a Spelerpinae ancestor occurring in

the AH and AP, with one immediate daughter lineage (H, Fig. 1.5) remaining in AH and AP, whereas the other (C, Fig. 1.5) experienced extinction in AP, but remained in AH. The

Gyrinophilus+Pseudotriton+Stereochilus clade (I, Fig. 1.5) is reconstructed as having occurred in the AH and AP, but subsequent vicariance left the Pseudotriton+Stereochilus lineage (I, Fig.

1.5) in the AP, while the lineage leading to Gyrinophilus (P, Fig. 1.5) remained in the AH. The

genera Pseudotriton and Stereochilus are both reconstructed as having diversified in the AP,

whereas Gyrinophilus originated and radiated in the AH.

The Eurycea+Urspelerpes lineage (C, Fig. 1.5) showed a much more dynamic history.

The ancestral range of Eurycea+Urspelerpes is reconstructed as AH. Whereas the monotypic

Urspelerpes remained in this region, the ancestral sister genus Eurycea is reconstructed as occurring in the AH, AP, IH, and IP. Within the genus Eurycea, there was a significant range expansion at the base of the clade (ca. 45-35 Ma), followed by several vicariant events, wherein further diversification took place in situ with little subsequent dispersal in the E. bislineata+E. lucifuga complex (E, Fig. 1.5). Vicariance then led to major lineages within the genus: E. bislineata+E. lucifuga complex (AH), E. quadridigitata complex (AP), and the Interior

Highlands Eurycea (IH).

1.3.3 Diversification Analyses

I did not detect a significant early burst of diversification in the lineage-through-time plots (Fig. 4), though the plot for Eurycea does not appear as clear-cut as the plot for the other spelerpines. The gamma statistic for the Eurycea clade was -1.130 and 0.515 for the

Gyrinophilus+Pseudotriton+Stereochilus clade, indicating that the nodes for the Eurycea clade

were closer to the root than expected based on a pure-birth model, but closer to the tips than

expected for the other spelerpines. Since I had essentially complete sampling, I used the one-

tailed test of significance of Pybus & Harvey (2000), which resulted in a failure to reject the null

hypothesis of a constant rate of diversification for both clades (γα=0.05 = -1.645, P-value = 0.129

for Eurycea and P-value = 0.303 for the other spelerpines). The SymmeTREE results (not

shown) indicated that there was no significant difference between any two clades in the

spelerpine phylogeny.

1.4 Discussion

1.4.1 Phylogenetic Relationships and Divergence Time Estimates

This study represents the most comprehensively sampled phylogeny of the Spelerpinae to

date (Fig. 1.2). These results mostly agree with previous, smaller-scale phylogenies of the

Spelerpinae. The reconstruction of the Gyrinophilus+Pseudotriton+Stereochilus clade had

Pseudotriton as the sister group to Stereochilus, matching the findings of Vieites et al. (2007)

based on three nuclear genes, but in contrast to Kozak et al. (2009) which had Gyrinophilus and

Stereochilus as sister taxa. Considering biogeography (Pseudotriton and Stereochilus are widespread in AP) and ecology (both often occurring in organic detritus of streams and floodplain swamps [Petranka, 1998]), combined with the strong support of these results and the independent results of Vieites et al. (2007), I suspect that Pseudotriton is the more likely sister

clade to Stereochilus, diverging sometime during the Eocene or Oligocene (Fig 1.3, Table 1.3).

Within the Pseudotriton ruber clade, P. r vioscai was the sister taxon to the other three subspecies and P. r ruber was paraphyletic with respect to P. r. nitidus. In the Pseudotriton montanus clade (the sister taxon to the P. ruber clade), deep branch lengths separated the

16 subspecies P. montanus diastictus and P. m. floridanus and were deeper than might be expected between populations of the same species. These subspecies are estimated to have split from one another approximately 15 Ma (Fig. 1.3, Table 1.3.). The two subspecies do differ in body size and color pattern and are the two most geographically distant subspecies, with two intervening subspecies separating them from each other. A range-wide phylogeographic analysis would shed further light on the relationships between these two forms.

A different pattern emerges within the genus Gyrinophilus, where three neotenic cave species render the much wider ranging G. porphyriticus paraphyletic. Niemiller et al. (2008), in a range-wide phylogeographic study, argued that G. palleucus and G. gulolineatus represented independent, recent divergence from the surface dwelling G. porphyriticus, with occasional gene flow. I found that there is less divergence between the three cave species of Gyrinophilus and the phylogenetically closest populations of G. porphyriticus, than there is between any two subspecies of G. porphyriticus (Fig. 1.3), also suggesting multiple invasions of caves and potentially cryptic species in G. porphyriticus.

These results also agree with previous molecular phylogenetic studies that placed the monotypic genus Urspelerpes as the sister taxon to Eurycea (Camp et al. 2009; Lamb & Beamer

2012) separating ca. 60 Ma [97–38.5]. The same relationships were found within the IH Eurycea clade as a previous mitochondrial phylogeny (Bonett & Chippindale 2004), including a deep divergence between the two E. multiplicata samples dating to sometime in the Miocene (Fig.

1.3). The enigmatic Haideotriton was strongly supported as being a member of the E. bislineata complex, although its exact placement is unclear as the support was low in both the ML and BI trees. Phylogenetically, it is either the sister group to the E. aquatica+E. junaluska clade, or sister to the southeastern coastal plain and southern Appalachian E. wilderae lineages. Since both

of these lineages are southern populations of the E. bislineata complex, it is clear that

Haideotriton originated from a southern lineage of the E. bislineata complex sometime during the early-to-mid Miocene (Figs. 1.2 and 1.3). Interestingly, geologic evidence supports its present day range (the Marianna Lowlands and Dougherty Plain of the central Florida panhandle and southwestern Georgia) as having been underwater during most of the Miocene (Fig. 1.6). It is unclear how to reconcile these two observations. Given the secretiveness of this species, it is possible that it evolved in underground caves and aquifers further north and migrated southward as sea levels begin to drop at the end of the Miocene. The true extent of its range is unclear, as one of the samples (H. wallacei C) was collected from a recently discovered population in

Washington Co., FL. If this taxon does represent the sister clade to the E. aquatica+E. junaluska clade, it is possible that during the Miocene it existed further north as a surface, spring dwelling form and only more recently (since Miocene ocean levels fell) invaded the Florida Aquifer of the

Florida panhandle and southwest Georgia. Evidence for this hypothesis may exist in the current geographic range and natural history of E. aquatica, a species that occurs near the Fall Line in the Ridge and Valley Provinces of Alabama, not far from the Southeastern Coastal Plain, where it inhabits permanent springs of the region (Petranka 1998). It’s possible that E. aquatica reflects the ancestral form and that H. wallacei occupied similar habitats across the northern edge of the southeastern coastal plain, expanding and contracting its range with fluctuating sea levels.

Testing this hypothesis would require further investigation; however, it is clear that Haideotriton is nested within the genus Eurycea, rendering the latter paraphyletic. I recommend that

Haideotriton be synonymized into Eurycea suggested by Dubois (2005).

Within the E. lucifuga complex, the subspecies E. l. melanopleura is the sister taxon to E. lucifuga, rendering the E. longicauda paraphyletic. This pattern could be explained by a number

of factors. One possibility is that the E. l. melanopleura specimen was misidentified, as it and E. lucifuga can be very similar in appearance in areas of sympatry (Petranka 1998). Unfortunately, I was unable to confirm the identity of the specimen. Alternatively, the specimen could be a hybrid or represent an example of mitochondrial introgression. Where the two forms are sympatric, particularly in the western portion of their range, they inhabit very similar habitats

(crevices in and around rock out crops and the twilight zone of caves) and are often found side- by-side (Petranka, 1998; Wray, personal observations), potentially allowing for occasional gene flow. I consider incomplete lineage sorting an unlikely scenario given the time since divergence of the crown clade (ca. 14 Ma [23–8]). Still, another possibility is that E. l. melanopleura represents an undescribed species that has been misplaced as a subspecies of E. longicauda. The two races have only a narrow contact zone along the eastern edge of the Ozark Plateau in

Missouri and along the Mississippi River in Illinois. Further investigation, such as a range wide phylogeographic study, is needed to fully resolve the relationships within this clade.

Perhaps the most surprising results from the Spelerpinae phylogeny are the degree of divergences observed within the E. quadridigitata complex. Eurycea quadridigitata sensu lato is rendered paraphyletic with respect to two other taxonomic groups: the Edwards Plateau neotenes

(M, Fig. 1.2) and E. chamberlaini of the Carolinas. Cryptic speciation in such a widespread species is not all that surprising. However, when the estimated divergence times are taken into account, the pattern is more unexpected. Three of the divergences date to a minimum of 14 Ma

(Fig. 1.3), deeper than any divergences observed within the E. bislineata complex.

1.4.2 Biogeography

Biogeographical reconstruction of the Spelerpinae suggested that the ancestor to all

spelerpines existed in the AH and AP approximately 72 Ma [112–45] during the Late Cretaceous

when sea levels were high and only the northern portion of the AP along the Mississippi River

was exposed (Miller et al., 2005; Fig. 1.6). The ancestor to Eurycea+Urspelerpes existed in the

AH sometime during the Late Cretaceous through early Eocene, a time that experienced extensive drops in sea level, exposing much of eastern North America (Miller et al., 2005; Fig.

1.6). By the time the ancestral Eurycea arose, ca. 42 Ma [65–26.5], it was widespread in the AH,

AP, IH, and IP. In contrast, the ancestor to Gyrinophilus+Pseudotriton+Stereochilus is reconstructed as occurring in the AH and AP, same as the ancestral spelerpine. The ancestor of

Pseudotriton+Stereochilus is reconstructed as occurring in the AP, where Stereochilus has remained to the present day. Though the ancestor of Pseudotriton is also reconstructed as occurring in the AP, there have been recent dispersals into the AH for both species, as well as into the IP for P. ruber. On the other hand, Gyrinophilus and its ancestors appear to have never left the AH, with the exception of G. palleucus, a neotenic cave species that is known to occur in a handful of caves of the Central Basin region of central Tennessee, just outside of the AH. The habitat of Gyrinophilus (collectively known as Spring Salamanders), consists of headwaters of small streams and cool mountain springs (Petranka 1998), habitats largely lacking in the relatively flat AP and IP.

1.4.3 Adaptive Radiation in Eurycea

Overall I find little support for an adaptive radiation in the genus Eurycea. One prediction of adaptive radiation is that ecological opportunity, via dispersal into a novel environment,

evolution of key innovations, and/or extinction of competitors, is thought to relax selective

pressures leading to ecological release, and ultimately an increase in species richness and

morphological diversity (Yoder et al. 2010). I hypothesized that the unbalanced species richness

observed in the Spelerpinae was the result of the ancestor to all Eurycea being more widely distributed across eastern North America compared to the ancestor of other spelerpines. If this were the case, ecological opportunity could have been the trigger for ecological release, resulting in an adaptive radiation of the genus. Ancestral range estimation provides a solid case that the genus Eurycea had ecological opportunity over the ancestors of other genera of spelerpines, via entry into three new physiographic regions that have largely been unoccupied by other spelerpines.

If the Eurycea clade has radiated due to ecological opportunity of the ancestor, there should be evidence of increased speciation early within the genus, coinciding with movement into these novel regions, followed by a later slow down as ecological niches became filled. The gamma statistic revealed no significant difference between the rate of diversification within

Eurycea and the pure-birth model, indicating a constant rate of diversification. This pattern was reflected in the lineage-through-time plots (Fig. 1.4). Adams et al. (2009) used a method-of- moments estimator, which takes into account extinction rate, to estimate the diversification rate of 15 major plethodontid salamander lineages using a wide variety of possible extinction rates and combinations since the true rates are unknown. Their results showed that the diversification rate of Eurycea was nearly three times that of the sister clade

Gyrinophilus+Pseudotriton+Stereochilus.

The third prediction is that an early burst of speciation would be followed closely by early morphological diversification. Although I did not test this prediction in the present study, in

their work on morphological evolution and rates of species diversification in plethodontids,

Adams et al. (2009) used seven morphological traits and Principal Component Analyses axes to show that among 15 major plethodontid clades, Eurycea had the highest rate variation of body

size, second highest rate variation of shape, and second highest level of morphological disparity.

However, there is a need to test whether this morphology is at least correlated with habitat

differences. Blankers et al. (2012) examined the relationship between microhabitat and

morphological variation in plethodontids (including most spelerpines), and found that, overall,

there was a significant relationship between the two, though they concluded this may be driven

primarily by the South American Bolitoglossinae. I agree with the criticism of Blankers et al.

(2012) that there may be other aspects of microhabitat that are important, but not captured in

their categorizations. Nonetheless, the overall pattern suggests that the high level of

morphological variation seen in Eurycea might be due to adaptation to a wide range of habitats.

The controversial requirement of “extraordinary” diversification was also examined using

a topological comparison of clade diversification within the spelerpines. The SymmeTREE

results indicated that there was no significant differences among any sister clade comparisons

throughout the tree, which does not meet the requirement, even though it may not be necessary.

1.4.4 Diversification within Eurycea

Although Eurycea may not have experienced an adaptive radiation via ecological opportunity, the biogeographic reconstruction indicates that the ancestral Eurycea was

widespread, occurring in four different physiographic regions, while the other spelerpines were

much more restricted (as they are today). Figure 1.5 demonstrates that the hypothetical ancestor

to the genus dispersed from the AH and into the AP, IH, and IP while also remaining in the AH.

From this point, four early lineages diversified in situ, with secondary colonization into other regions only recently and mostly in just one of these major lineages (E, Fig. 1.5). This pattern is consistent with early shifts in ecology, which lead to range expansion and subsequent vicariance and in situ diversification. This pattern predicts that ecologies within Eurycea would vary greatly

from typical plethodontids. Varied ecological differences would also be predicted given the high

degree of morphological disparity and variation in body size and shape.

Spelerpines fall into two broad categories: fully aquatic and semi-aquatic species. The

other major lineages of eastern plethodontids are either terrestrial (Plethodon) or aquatic/semi-

aquatic (Desmognathus), and only five of the 21 species of Desmognathus occur outside the AH.

Modern Eurycea occur in a wide variety of aquatic/semi-aquatic habitats. Many are classic

streamside salamanders, such as E. bislineata and E. lucifuga complex animals, but are known to

use different microhabitats within a stream system. For instance, in the Southeastern Coastal

Plain, E. cirrigera occurs more often in the first and second order streams, while E. guttolineata

occupies third order streams and floodplains of rivers (Means 2000). The E. quadridigitata

complex deviates from the classic Eurycea mold, breeding mostly in lentic environments, such

as ponds, swamps, and ephemeral wetlands, rather than the flowing waters of most other

Eurycea. They can also be found far from any body of water outside of the breeding season.

Though many species can be found at the mouth of caves and even in the twilight zone, three

species (representing two different, major lineages) have become cave specialists. Eurycea l.

melanopleura and E. lucifuga are both troglophiles, often encountered in caves, but can also be

found in surrounding rocky outcrops and other broken terrain (Petranka 1998). They are dorsal-

ventrally flattened, allowing them to squeeze into small crevices and between rocks. Eurycea

spelaea is a troglobyte with an odd life cycle. Larvae are found in surface streams and creeks that

usually exit from cave entrances. As they mature, they migrate upstream into the cave systems

where they transform into a cave adapted adult (Petranka 1998). Three of the four major lineages

(Fig. 1.5) have independently evolved permanently aquatic, neotenic species (species that obtain

sexual maturity while retaining the larval body form). The Edwards Plateau clade (M; Fig. 1.5) is

composed entirely of neotenic species, but neotenic forms exist in the IH clade (E. tynerensis)

and in the E. bislineata complex (E. wallacei). Several of these neotenic forms (e.g., H. wallacei and E. rathbuni) represent some of the most extreme examples of cave adapted animals known

(Wiens et al. 2003). Such varied ecologies in a single genus represent most of the extremes in the entire family, providing convincing evidence that a shift in ecologies early in the evolution of

Eurycea lead to colonization and allopatric diversification.

1.5 Conclusion

The species diversity observed in the genus Eurycea does not fit the classic model of adaptive radiation or the recently popularized pattern of niche conservatism. Instead, ancestral range reconstruction indicates that the ancestor to all Eurycea dispersed out of the Appalachian

Highlands, the cradle of eastern plethodontid diversity, and into surrounding physiographic regions during the Eocene, approximately 42 Ma, coinciding with drops in sea levels that exposed previously uninhabitable lowlands. This early range expansion was most likely the result of a shift in ecology, which facilitated dispersal and colonization. This pattern is seen in modern Eurycea, which represents one of the most morphologically diverse and disparate radiations of plethodontid salamanders, inhabiting an equally wide range of niches.

Figure 1.1 Physiographic regions of eastern North America: AH (Appalachian Highlands), AP (Atlantic Plain), IH (Interior Highlands), IP (Interior Plains), LU (Laurentian Uplands). Heavy black lines represent physiographic region boundaries. Arrows represent dispersal routes allowed in the ancestral estimation rate matrix. Numbers represent the number of species of Eurycea/other spelerpines/other plethodontids in the physiographic region. Gray shaded area represents the modern range of the Spelerpinae (= Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, Urspelerpes).

LU 1/0/3

IP 19/4/39 AH 9/7/67

IH 6/0/16

AP 6/3/32

500 km

Figure 1.2 Maximum likelihood phylogram of the Spelerpinae (Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, and Urspelerpes) based on the mitochondrial CytB, ND2, and ND4 genes and nuclear RAG1. Support values are reported at the nodes as ML bootstrap/BI posterior probabilities. Asterisks represent nodes supported by both BS values/PP that are ≥ 99/0.99. A = Batrachoseps+Spelerpinae, B = Spelerpinae, C = Eurycea+Urspelerpes, D = Eurycea, E = E. bislineata+E. lucifuga complexes, F = E. quadridigitata+Edwards Plateau neotenes, G = Interior Highlands Eurycea, H = Gyrinophilus+Pseudotriton+Stereochilus, I = Pseudotriton+Stereochilus, J = E. quadridigitata+E. chamberlaini, K = Western E. quadridigitata+Edwards Plateau neotenes, L = Pseudotriton, M = Edwards Plateau neotenes, N = E. bislineata complex, O = E. longicauda complex, P = Gyrinophilus.

* Eurycea quadridigitata A 84/1 Eurycea quadridigitata B 89/1 Eurycea quadridigitata M * Eurycea quadridigitata D * Eurycea quadridigitata E * * Eurycea quadridigitata C Eurycea quadridigitata K * Eurycea quadridigitata J * Eurycea quadridigitata L * Eurycea chamberlaini A * * Eurycea chamberlaini B Eurycea chamberlaini C * * Eurycea quadridigitata F J Eurycea quadridigitata I * Eurycea quadridigitata G Eurycea quadridigitata H 54/.63 Eurycea neotenes A 74/1 Eurycea neotenes B 53/.95 Eurycea sp. Comal Springs 86/.99 Eurycea pterophila 89/1 64/1 Eurycea tridentifera F * Eurycea sosorum --/1 Eurycea nana * Eurycea rathbuni * Eurycea waterlooensis --/1 Eurycea latitans 52/1 M Eurycea troglodytes 71/.81 Eurycea chisholmensis * * Eurycea tonkawae K Eurycea naufragia * Eurycea quadridigitata N Eurycea quadridigitata O 91/1 * Eurycea aquatica A * Eurycea aquatica B * Eurycea junaluska A Eurycea junaluska B * Haideotrion wallacei A 85/1 * Haideotriton wallacei C Haideotriton wallacei B 81/1 Eurycea cirrigera C * Eurycea wilderae N90/1 Eurycea wilderae/cirrigera B 98/1 Eurycea bislineata * D 94/1 Eurycea cirrigera A 88/1 Eurycea cirrigera B E Eurycea wilderae/cirrigera A 82/1 * Eurycea guttolineata A 83/1 Eurycea guttolineata B * * Eurycea l. longicauda A 94/1 O Eurycea l. longicauda B C * Eurycea lucifuga Eurycea l. melanopleura * Eurycea multiplicata A Eurycea multiplicata B G Eurycea spelaea 87/1 * Eurycea tynerensis * Urspelerpes brucei A Urspelerpes brucei B * Gyrinophilus gulolineatus A * * Gyrinophilus gulolineatus B B * Gyrinophilus palleucus A * Gyrinophilus palleucus B 80/1 Gyrinophilus p. porphyriticus B * Gyrinophilus subterraneus A 89/1 * Gyrinophilus subterraneus B Gyrinophilus p. porphyriticus A * Gyrinophilus p. dunni P * Gyrinophilus p. danielsi Gyrinophilus p. porphyriticus C * Pseudotriton r. nitidus A * * Pseudotriton r. ruber A H * Pseudotriton r. ruber B * Pseudotriton r. schencki 85/1 L Pseudotriton r. vioscai * Pseudotriton montanus diastictus I Pseudotriton m. floridanus 82/1 Stereochilus marginatus 65/.60 Batrachoseps diabolicus 66/.52 Batrachoseps nigriventris * Batrachoseps major * Batrachoseps attenuatus Batrachoseps wrightorum 0.08

Figure 1.3 BI chronogram of the Spelerpinae (Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, and Urspelerpes) based on the mitochondrial CytB, ND2, and ND4 genes and nuclear RAG1. The 95% highest posterior density credible intervals are reported as gold bars at nodes. Red circles denote fossil calibration points. Geologic time scale at bottom of figure is in millions of years. The Pliocene and Pleistocene are abbreviated as PO and PS, respectively. Node letters correspond to those in Fig. 1.2.

E. quadridigitata A E. quadridigitata B E. quadridigitata M E. quadridigitata D E. quadridigitata E E. quadridigitata C E. quadridigitata K E. quadridigitata J E. quadridigitata L E. chamberlaini A E. chamberlaini B E. chamberlaini C J E. quadridigitata F E. quadridigitata I E. quadridigitata G E. quadridigitata H E. neotenes B E. neotenes A E. sp. Comal Springs F E. pterophila E. tridentifera E. sosorum E. nana E. troglodytes E. latitans M E. rathbuni E. waterlooensis K E. tonkawae E. chisholmensis E. naufragia E. quadridigitata N E. quadridigitata O E. aquatica A E. aquatica B E. junaluska A E. junaluska B H. wallacei A H. wallacei C H. wallacei B E. cirrigera C N E. wilderae E. wilderae/cirrigera B D E. wilderae/cirrigera A E. cirrigera B E. bislineata E. cirrigera A E E. guttolineata A E. guttolineata B E. l. longicauda A C E. l. longicauda B O E. lucifuga E. l. melanopleura E. spelaea E. tynerensis G E. multiplicata A E. multiplicata B U. brucei A U. brucei B G. gulolineatus A B G. gulolineatus B G. palleucus A G. palleucus B G. p. porphyriticus B G. subterraneus A G. subterraneus B G. p. porphyriticus A G. p. dunni P G. p. danielsi G. p. porphyriticus C A H P. r. nitidus P. r. ruber A P. r. ruber B L P. r. schencki P. r. vioscai P. montanus diastictus P. m floridanus I S. marginatus B. attenuatus B. nigriventris B. diabolicus B. major B. wrightorum

EARLY CRETACEOUS LATE CRETACEOUS PALEO- EOCENE OLIGO- MIOCENE P P CENE CENE O S 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 30

Lineages Through Time Plot

Eurycea ќ

2WKHU6SHOHUSLQHV ќ Number of Lineages 1 2 5 10 20

-40 -30 -20 -10 0

Time (millions of years before present)

Figure 1.4 Lineage through time plots of the genus Eurycea and other Spelerpines (Gyrinophilus+Pseudotriton+Stereochilus clade).

Figure 1.5 DEC ancestral range and split estimations of the Spelerpinae under the adjacent areas rates dispersal model using BI chronogram from Fig. 3. Map inset from Fig. 1 with physiographic ranges color-coded. Circles at nodes are colored based on the most likely ancestral range estimation, not proportions of particular models (e.g., a circle at a node that is half yellow and half red reflects the ancestral range estimation as being AP and IH). Colored circles in front of tip names represent the modern range of that OTU. Colored branches reflect the most likely ancestral splits. Values at nodes represent the marginal likelihoods of ancestral state/ancestral splits. Asterisks represent marginal likelihoods of 1.0. Abbreviations on geological time scale at bottom stand for Late Cretaceous (Late K), Pliocene (PO ), and Pleistocene (PS). Node letters correspond to those in Fig. 1.2.

/ */* * * E. quadridigitata A / E. quadridigitata B * * E. quadridigitata M */* */* E. quadridigitata D LU / E. quadridigitata E */* * * E. quadridigitata C / E. quadridigitata K */* * * E. quadridigitata J E. quadridigitata L 0.96/0.24 0.99/0.99 0.64/0.64 E. chamberlaini A E. chamberlaini B / E. chamberlaini C * * 0.96/0.96 E. quadridigitata F J E. quadridigitata I IP */* E. quadridigitata G E. quadridigitata H AH */* E. neotenes A */* E. neotenes B */* E. sp. Comal Springs */* 0.68/0.68 */* E. pterophila F */* E. tridentifera IH */* E. sosorum / E. nana */* * * E. troglodytes E. latitans */* */* E. rathbuni AP M / E. waterlooensis */* */ * E. tonkawae K * * E. chisholmensis 500 km / E. naufragia * * E. quadridigitata N E. quadridigitata O 0.39/0.19 / / * * E. aquatica A * * / E. aquatica B / * * E. junaluska A * * / E. junaluska B / * * H. wallacei A 0.89/0.72 * * H. wallacei C H. wallacei B 0.62/0.57 E. cirrigera C 0.46/0.46 0.61/0.61 N E. wilderae 0.80/0.80 E. wilderae/cirrigera B 0.41/0.35 E. wilderae/cirrigera A D 0.35/ 0.72/0.72 E. cirrigera B 0.35 0.52/0.52 E. bislineata E E. cirrigera A 0.79/0.13 E. guttolineata A 0.27/0.27 E. guttolineata B 0.29/0.12 0.71/0.18 E. l. longicauda A 0.20/0.20 O E. l. longicauda B C 0.42/0.19 E. lucifuga / E. l. melanopleura / * * E. tynerensis * * E. spelaea G */* E. multiplicata A E. multiplicata B */* U. brucei A / U. brucei B * * G. gulolineatus A 0.23/0.18 0.86/0.86 G. gulolineatus B B 0.97/0.97 G. palleucus A */* */0.25 G. palleucus B / G. p. porphyriticus B / / * * G. subterraneus A * * * * G. subterraneus B */* G. p. porphyriticus A P G. p. dunni */* G. p. danielsi G. p. porphyriticus C 0.52/0.51 0.77/0.47 0.32/0.30 P. r. nitidus H 0.49/0.49 P. r. ruber A 0.54/0.29 P. r. ruber B 0.32/0.32 P. r. schencki P. r. vioscai L 0.39/0.39 P. montanus diastictus I P. m. floridanus 0.47/0.47 S. marginatus

LATE K PALEOCENE EOCENE OLIGOCENE MIOCENE POS P 70 60 50 40 30 20 10 0 33

Figure 1.6 Sea level fluctuations of eastern North America from the Late Cretaceous through Pliocene. Heavy black lines reflect same physiographic regions in Fig. 1. The gray shaded area represents the modern range of the Spelerpinae (= Eurycea, Gyrinophilus, Haideotriton, Pseudotriton, Stereochilus, Urspelerpes). Blue reflects shifting coastline.

LATE CRETACEOUS (ca. 75 Ma) CRETACEOUS-TERTIARY (ca. 65 Ma)

PALEOCENE (ca. 60 Ma) EARLY EOCENE (ca. 50 Ma)

LATE EOCENE (ca. 40 Ma) OLIGOCENE (ca. 25 Ma)

MIOCENE (ca. 15 Ma) PLIOCENE (ca. 3 Ma)

Table 1.1 List of samples used in this study.

Sample ID County, State Lat Long CytB ND2 ND4 RAG1 E. chamberlaini A KW0024 Anderson, SC 34.64 -82.7 E. chamberlaini B KW0025 Anderson, SC 34.64 -82.7 E. chamberlaini C KW0128 Pitt, NC 35.61 -77.36 E. multiplicata B KW0966 Pushmataha, OK 34.52 -94.98 E. neotenes B KW0987 Bexar, TX 29.66 -98.63 E. pterophila KW0986 Blanco, TX 30.08 -98.44 E. quadridigitata A KW0004 Leon, FL 30.64 -84.25 E. quadridigitata B KW0016 Bradford, FL 29.83 -82.16 E. quadridigitata C KW0026 Volusia, FL 29.07 -81.13 E. quadridigitata D KW0053 Washington, FL 30.53 -85.86 E. quadridigitata E KW0054 Washington, FL 30.53 -85.86 E. quadridigitata F KW0064 Washington, FL 30.45 -85.5 E. quadridigitata G KW0076 Santa Rosa, FL 30.84 -86.94 E. quadridigitata H KW0088 Mobile, AL 31.04 -88.19 E. quadridigitata I KW0090 Taylor, GA 32.64 -84.37 E. quadridigitata J KW0124 Berkeley, SC 33.21 -79.47 E. quadridigitata K KW0125 Marion, FL 29.34 -81.73 E. quadridigitata L KW0141 Barnwell, SC 33.26 -81.6 E. quadridigitata M KW0144 Liberty, FL 30.17 -84.68 E. quadridigitata N KW0951 Sabine, TX 31.36 -93.72 E. quadridigitata O KW0965 Jones, MS 31.44 -88.97 E. sp. Comal Springs KW0989 Comal, TX 29.71 -98.13 E. spelaea KW1200 Mayes, OK 36.19 -95.14 E. tynerensis KW0967 Crawford, AR 35.62 -94.45 H. wallacei A KW0959 Jackson, FL 30.78 -85.15 H. wallacei B KW0960 Jackson, FL 30.78 -85.15 H. wallacei C KW1086 Washington, FL 30.57 -85.84 G. gulolineatus A KW1161 Knox, TN

Table 1.1 Continued.

Sample ID County, State Lat Long CytB ND2 ND4 RAG1 G. gulolineatus B KW1164 Knox, TN G. palleucus A KW1158 Franklin, TN G. palleucus B KW1163 Franklin, TN G. p. porphyriticus B KW1160 DeKalb, TN G. p. porphyriticus C KW1162 Cocke, TN G. subterraneus A KW1159 Greenbrier, TN G. subterraneus B KW1165 Greenbrier, TN U. brucei B KW0634 Stephens, GA 34.66 -83.31 *This study E. aquatica A 61296 E. aquatica B 61299 E. bislineata A E. chisholmensis New E. cirrigera A KHK245 E. cirrigera B KHK419 E. cirrigera C KHK174 E. guttolineata A KHK204 E. guttolineata B KHK330 E. junaluska_A KHK208 E. junaluska_B KHK8.280 E. latitans JQ920628 JQ920812 E. l. longicauda_A KHK388 E. l. longicauda_B KHK238 E. lucifuga KHK698 E. l. melanopleura KHK771 E. multiplicata A CEA E. nana JQ920630 JQ920814 EF443117 JQ920772 E. naufragia JQ920627 JQ920811 JQ920772

Table 1.1 Continued.

Sample ID County, State Lat Long CytB ND2 ND4 RAG1 E. neotenes A CEA E. rathbuni 919 E. sosorum JQ920631 JQ920815 JQ920775 E. tonkawae New JQ920810 E. tridentifera HCC E. troglodytes JQ920629 JQ920813 JQ920773 E. waterlooensis AY014856 E. wildcirr A KHK71203 E. wildcirr B KHK8.234 E. wilderae KHK8.252 G. p. danielsi ABS010 G. p. dunni RRM3214 G. p. porphyriticus A KHK409 P. montanus diastictus 71757 P. m. floridanus 90248 P. r. nitidus ASB031 P. ruber ruber A KHK148 P. r. ruber B KHK563 P. r. schencki KHK263 P. r. vioscai KHK240 S. marginatus CEA U. brucei A JQ920618 JQ920802 JQ920763 *GenBank/Ken Kozak B. attenuatus NC006340 NC006340 NC006340 JF449382 B. diabolicus EU011257 EU117190 EU011259 EU020163 B. major AY691754 EU117194 AY691798 AY650126 B. nigriventris EU011253 EU117187 EU011255 EU020164 B. wrightorum NC006333 NC006333 NC006333 JF449380 *GenBank Outgroup 39

Table 1.2 Primers and PCR protocols used in amplification and sequencing.

Anneal Elong Primer Gene Primer Sequence (5'–3') Temp (Time) Time Cycles Reference MVZ15 CytB GAACTAATGGCCCACACWWTACGNAA 50 C (30 s) 90 s x 30 Moritz et al. 1992 MVZ16R CytB AAATAGGAARTATCAYTCTGGTTTRAT 50 C (30 s) 90 s x 30 Moritz et al. 1992 PGludg2 CytB GGTCTGAAAAACCAATGTTGTATTC 50 C (30 s) 90 s x 30 Wiens et al. 2006 L4437 ND2 AAGCTTTCGGGCCCATACC 50 C (35 s) 150 s x 25 Macey et al. 1997 H6159 ND2 GCTATGTCTGGGGCTCCAATTA 50 C (35 s) 150 s x 25 Weisrock et al. 2001 EqND2FI* ND2 GGAGGCCTAAATCAACCACA 50 C (35 s) 150 s x 25 This study EqND2RI* ND2 GTGATGTGGTGTACGCAAGG 50 C (35 s) 150 s x 25 This study ND4F ND4 CACCTATGACTACCAAAAGCTCATGTAGAAGC 55 C (30 s) 90 s x 30 Arévalo et al. 1994 Phist ND4 TTTYTAGGRTCACRGCCTA 55 C (30 s) 90 s x 30 Wiens et al. 2006 Ephist ND4 TCRTTTTTTAGGGTCACRGCCTAG 55 C (30 s) 90 s x 30 Wiens et al. 2006 EuryceaRag1F RAG1 GGTAYGATGTTGCATTGGTTGCCA 58 C (30 s) 60 s x 30 Timpe et al. 2009 Rag1midElongFb RAG1 TGCACTGTGAYATNGGGAATGCTG 58 C (30 s) 60 s x 30 Timpe et al. 2009 ElongRag1R RAG1 TTGACTGCCATCGCTTCCTCTCTT 58 C (30 s) 60 s x 30 Timpe et al. 2009 Rag1endElongRb RAG1 AACTTGGACTGCCTGGCGTTCATT 58 C (30 s) 60 s x 30 Timpe et al. 2009 * Internal primer

Table 1.3 Estimated ages of key spelerpine nodes from BI chronogram in Fig. 3 (95% HPD credible interval in millions of years). Node letters correspond to labeled nodes in Figs. 2 and 5.

Node Name Node Age (Ma) 95% HPD CI (Ma) A Batrachoseps+Spelerpinae 91 140–62 B Spelerpinae 72 112–45 C Eurycea+Urspelerpes 61 97–38.5 D Eurycea 42 65–26.5 E E. bislineata+E. lucifuga complexes 37 57–23.5 F E. quadridigitata+Edwards Plateau neotenes 35 56–23 G Interior Highlands Eurycea 35 56–21.5 H Gyrinophilus+Pseudotriton+Stereochilus 35 55–21.5 I Pseudotriton+Stereochilus 33.5 53–20 J E. quadridigitata+E. chamberlaini 29 45–18.5 K Western E. quadridigitata+Edwards Plateau neotenes 27 43–17 L Pseudotriton 27 43–16.5 M Edwards Plateau neotenes 25 39.5–16 N E. bislineata complex 24 37–15 O E. lucifuga complex 15 23–8.5 P Gyrinophilus 10 16–6.5

CHAPTER TWO

GENETIC DIVERGENCE IN THE DWARF SALAMANDER (Eurycea quadridigitata) COMPLEX: TESTING MECHANISMS OF SPECIATION

2.1 Introduction

2.1.1 Background

Identifying and understanding the patterns and mechanisms of population divergence is critical to the study of speciation (Coyne & Orr 2004). In the past, most attention focused primarily on patterns of diversification, particularly various geographic patterns, such as allopatric and sympatric speciation (Coyne & Orr 2004; Rundle & Nosil 2005). More recently, however, the field has concentrated on the mechanisms that drive the development of reproductive isolation of populations (Schluter 2000, 2001; Coyne & Orr 2004; Rundle & Nosil

2005), with mechanisms broadly falling into two categories: adaptive and nonadaptive. Adaptive divergence implicitly requires a correlation between population differentiation and ecology, which should decrease gene flow between populations, ultimately leading to reproductive isolation and speciation. In contrast, Gittenberger (1991) defined nonadaptive divergence as the result of diversification in the absence of significant niche differentiation, leading to the isolation of allopatric species due to competitive exclusion.

Traditionally, allopatric divergence has been thought of as most commonly being nonadaptive, though exceptions undoubtedly occur (Dieckmann et al. 2004). The classic example is of two populations becoming isolated from one another on either side of a physical barrier until genetic drift causes enough differentiation that when the barrier is overcome, the populations are no longer genetically compatible (Coyne & Orr 2004; Dieckmann et al. 2004).

More recently, some studies have argued that nonadaptive divergence via niche conservatism is 42 widespread and an important mechanism in the diversification of populations into species (Wiens

& Graham 2005; Rundell & Price 2009; Wiens et al. 2010). Under this concept, some organisms have a tendency to retain the ancestral fundamental niche and associated ecological traits (Wiens

& Graham 2005), which can result in nonadaptive divergence via allopatric speciation if populations are fragmented and restricted by this niche conservatism (e.g., thermal or physiological tolerances) over long periods of time (Kozak et al. 2006; Kozak & Wiens 2006,

2010).

The last two decades has seen renewed interest in the old concept of ecological speciation, whereby reproductive isolation results from divergent selection on traits during the process of adaptive divergence (Schluter 2000, 2001; Rundle & Nosil 2005). Somewhat surprisingly, the role of ecology in speciation has been largely neglected (Morell 1999). The concept of ecological speciation dates to at least Darwin and was mentioned during the modern synthesis by both Mayr (1942) and Dobzhansky (1946), but very few concrete examples or an explicitly outlined mechanism have been put forth until relatively recently (Schluter 2001; Coyne

& Orr 2004; Rundle & Nosil 2005). Several pre- and post-zygotic barriers have been implicated in reproductive isolation with ecological speciation (see Rundle & Nosil (2005) for a detailed review), but none have been as neglected as the pre-zygotic barrier of habitat isolation

(Schemske 2000; Coyne & Orr 2004). Habitat isolation takes place when different populations of a species possess genetic preferences for different habitats, decreasing the opportunity for gene flow between populations. As preference for different habitats increases, divergent selection on traits interacting in these different habitats may lead to reproductive isolation and, ultimately, ecological speciation. Despite the appeal of such a straightforward, plausible mechanism with clear expectations, unequivocal examples of habitat isolation are rare, with most examples

restricted to insects (Tavormina 1982; Katakura et al. 1989; Craig et al. 1993; Feder et al. 1994;

Funk 1998; Via 1999; Linn et al. 2003), though other examples exist in plants (Wang et al. 1997) and frogs (Lynch 1978; MacCallum et al. 1998).

Herein, I examine the phylogeographic pattern of a widespread species of salamander, the

Dwarf Salamander (Eurycea quadridigitata) and its relative, Chamberlain’s Dwarf Salamander

(E. chamberlaini). I find several deep genetic divergences in the phylogeographic analysis.

Whether these molecular divergences among populations in this salamander complex are best explained via a nonadaptive or an adaptive mechanism is unclear. In order to choose between adaptive and nonadaptive scenarios, I tested five hypotheses: 1) ecoregion, 2) river subregion, 3) river accounting unit, 4) river cataloguing unit, and 5) habitat isolation. The first tests for congruence with ecoregion boundaries. The next three are classic allopatric models hypothesizing vicariance via intervening uplands between river drainages. The last one is a habitat isolation model, based on breeding habitats.

2.1.2 Study System

Plethodontidae is one of nine extant families of salamander, making up 66% of salamander species diversity worldwide (Frost 2013). Nearly all of this diversity occurs in North

America, with 44% of plethodontids occurring in the U.S. alone (Tilley et al. 2008). The

Appalachian Mountains of the eastern United States harbor an abundance of these plethodontids, with several studies suggesting that niche conservatism has played a significant role in generating the high species diversity seen in the genera Plethodon and Desmognathus (Kozak et al. 2006; Kozak & Wiens 2006, 2010). Kozak and Wiens (2010) demonstrated that the highest species richness observed in Plethodon and Desmognathus occurs at intermediate elevations,

hypothesizing that these elevations harbor conditions matching that of the ancestral niche. As

past climatic conditions fluctuated, salamander populations have migrated up and down in

elevation following ancestral niche conditions. Present day intermediate elevations are suspected

to be most similar to this ancestral niche, isolating salamander populations via inhospitable

conditions (e.g., higher average temperatures) in the surrounding lowlands and effectively

blocking migration between populations. Such a scenario sets up long-term isolation and,

ultimately, species divergence. A third group of salamanders inhabiting the eastern U.S., the

Spelerpinae, do not follow this same elevational pattern of species richness. The highest species

richness in spelerpines occurs at low elevations and steadily decreases with increasing elevation

(Kozak & Wiens 2010). It is unclear why this group of salamanders deviates from the pattern

demonstrated by Plethodon and Desmognathus.

Spelerpinae is comprised of five genera: Eurycea, Gyrinophilus, Pseudotriton,

Stereochilus, and Urspelerpes. Gyrinophilus and Urspelerpes are largely confined to the

Appalachian Mountains and Piedmont (the plateau region between the Appalachians and low lying coastal plains), whereas the other three genera range far into the adjacent lowlands of the

Atlantic and Gulf coastal plains (Conant & Collins 1998). Even in these lowlands, however, most species inhabit cooler microhabitats in and around flowing water and seepage, similar to the hypothesized ancestral plethodontid niche of Kozak and Wiens (2010).

Eurycea quadridigitata is a small (5.0–9.0 cm snout-vent length) species of salamander restricted to the coastal plains of the southeastern United States, from the eastern third of Texas eastward to the southern tip of Florida and northward through the Carolinas (Conant & Collins

1998; Petranka 1998; Fig. 2.1). Though as many as four different subspecies of E. quadridigitata were once recognized (see Mittleman 1947), none are currently recognized (Tilley et al. 2008).

Chamberlain’s Dwarf Salamander (Eurycea chamberlaini) occupies the central portions of the

Carolinas (Harrison & Guttman 2003) and was formally considered a color variant of E. quadridigitata. My work with E. quadridigitata and E. chamberlaini, as well as that of Lamb &

Beamer (2012), suggests that much more diversity than is currently recognized exists and,

herein, I refer to this group as the E. quadridigitata complex.

Within the range of the E. quadridigitata complex, there are no fewer than 12 ecoregions

(Slaats 1999). Ecoregions are recognized based on various biotic and abiotic factors, such as historical biogeography, vegetation type, soil composition, geomorphology, and climate, that are more homogenous within the given area than across a larger scale (Omernik 1987; Omernik &

Griffith 1991; Olson et al. 2001). Some recent studies have found good congruence between

specific taxonomic groups and ecoregion boundaries (Johnson 2000; Heino et al. 2002), while

others have had mixed results (Wright et al. 1998; Marchant et al. 2000).

A number of phylogeographic studies have demonstrated the strong role the numerous river systems of the southeastern U.S. have played in restricting gene flow between populations on opposite sides (Avise 1996; Walker et al. 1998; Burbrink et al. 2000; Pauly et al. 2007;

Zellmer et al. 2012). However, for salamanders, Means (1974) proposed that southeastern river

drainages should provide corridors for gene flow, if suitable habitat is present throughout,

whereas the intervening uplands would greatly restrict gene flow between populations. In this

model, intervening uplands serve as a vicariant barrier, with the expectation that animals within a

given drainage will be more closely related to each other than to animals from adjacent river

drainages.

The southeastern U.S. is rich in various still-water (= lentic) and flowing-water (= lotic)

habitats, with some (e.g., steephead ravines, Everglades) found nowhere else in the world

(Means 1977, 2000). Although adult E. quadridigitata complex animals are terrestrial and may be found far from water, during the breeding season they must return to wetlands to mate and deposit eggs. The aquatic larvae then hatch and spend 2–6 months feeding and growing in these wetlands before metamorphosing and becoming terrestrial (Petranka 1998). Unlike the vast majority of species in the genus Eurycea that breed and inhabit lotic systems (e.g., flowing seepage, streams, creeks, and surface/subterranean rivers), members of the E. quadridigitata complex are more terrestrial and breed in a variety of lentic habitats and, in some cases, are the only plethodontid salamander present in southeastern coastal plain (SECP) wetlands (Petranka

1998; Means 2000).

Abiotic and biotic factors vary greatly in wetlands known to harbor breeding populations of E. quadridigitata (Table 2.1). Cypress domes, strands, and swamps are dominant wetlands of the SECP and commonly support large populations of these salamanders (Means 2000). These sluggish or still water wetlands experience short, dynamic hydroperiods driven by seasonal rains

(Myers & Ewel 1990; Whitney et al. 2011). These conditions are essential for the life cycle of cypress (Taxodium sp.), which dominate these wetlands, providing dense canopies and thick layers of organic nutrients in the basins. Though often in close proximity to these cypress wetlands, hillside seepage bogs have much more open canopies, dominated by small herbaceous undergrowth and shrubs (Whitney et al. 2011). Many plants living in these wetlands are adapted to the extremely acidic soils (e.g., pitcher plants, Sarracenia spp.). Though flow rate varies throughout the year, seepages usually retain some level of moisture at least near the seepage headwaters (Myers & Ewel 1990; Whitney et al. 2011). A third type of wetland is not traditionally thought of as a breeding wetland for E. quadridigitata (Petranka 1998), though they occasionally utilize them. Steepheads and ravines both form from erosion of soil, though in very

different ways. Steepheads are formed from the lateral undercutting of exposed banks of sandy

soil by percolation of subsurface water, whereas ravines form gullied depressions through

surface erosion of top soil (Means 2000; Whitney et al. 2011). Both systems are characterized by

having mixed forests with multiple canopy layers and constant water flow, which provides cool,

moist conditions year round. This results in some of the highest species diversity of salamanders

in the SECP (Means 1977, 2000). Finally, E. quadridigitata complex animals in some locations

can be found utilizing whitewater, streamside habitats for breeding. These third order streams

have permanent water flow and low levels of organic detritus, lacking the tea-colored water of

black water streams. In the SECP, such habitats are most common along the Fall Line, the area

where bedrock from the coastal plain is first exposed. Given the utilization of such diverse

breeding habitats, the potential for habitat isolation in E. quadridigitata exists.

2.2 Materials and Methods

2.2.1 Taxon Selection and Sampling

I collected samples (61) from throughout the range of the E. quadridigitata complex and supplemented the tissue sampling with loans from private individuals (42), the Texas

Cooperative Wildlife Collection Division of Herpetology (three) and from the Louisiana State

University Collection of Genetic Resources (five), for a total of 111 samples of E. quadridigitata from 64 populations (Fig. 2.1, Table 2.2), including representatives of the recently described E. chamberlaini (five individuals, three populations). I was unable to secure any samples representing the Eurycea clade from the Edwards Plateau of central Texas; a group of neotenic

species that have been demonstrated to form a strongly supported clade. Recent work (Lamb &

Beamer 2012; Wray unpublished) suggests that this Edward Plateau clade may render E.

quadridigitata paraphyletic. Sample sizes ranged from one to eight individuals per locality.

Sequence from an additional sample of E. quadridigitata was downloaded from GenBank. Ten other species from the genus Eurycea were included in the phylogenetic analysis in order to

thoroughly access the phylogenetic position of the E. quadridigitata complex (Table 2.2). These

ten species were represented by 34 samples that were either generated for this study (nine) or

downloaded from GenBank (25), including 23 samples from a phylogeographic study of the E.

bislineata complex (Kozak et al. 2006). Outgroups consisted of one member from each of the

remaining Spelerpinae genera: Urspelerpes brucei (this study) and Pseudotriton montanus,

Gyrinophilus porphyriticus, and Stereochilus marginatus (GenBank). Tissue samples consisted

of either liver or tail tips preserved in 95% ethanol and/or stored at -80°C. Voucher specimens

were set in 10% formalin and stored in 70% ethanol.

2.2.2 DNA Extraction, Amplification, and Sequencing

Genomic DNA was extracted from liver or tail tips using the hot phenol-chloroform-

isoamyl alcohol/chloroform-isoamyl alcohol method (Sambrook & Russell 2001). Extracts were

visualized on agarose gels and DNA concentration quantified using a Nanodrop ND-1000

Spectrophotometer or an Invitrogen Qubit Flurometer. Polymerase chain reaction (PCR) was

performed on a section of the mitochondrial genome consisting of the tRNA-Met, the entire

NADH dehydrogenase subunit 2 (ND2) gene, tRNA-Trp, tRNA-Ala, tRNA-Asn, the origin for

light-strand replication (OLrep), tRNA-Cys, tRNA-Tyr, and 226 base pairs of the cytochrome

oxidase subunit 1 (CO1) gene. I used the following reagents and concentrations: 13.3 µl distilled

water, 5 µl 5X Colorless GoTaq Reaction Buffer, 1.5 µl MgCl2 (25 mM), 1.5 µl dNTPs (2.5

mM), 0.2 µl GoTaq (5u/ul), 0.5 µl bovine serum albumin (10 mg/ml), 1.0 µl of each primer (10

49 ng/ul). Amplification was performed using primers listed in Table 2.3 (Macey et al. 1997;

Weisrock et al. 2001) with the following thermal cycler program: denaturation at 94°C for 25 sec, annealing at 50°C for 35 sec, and extension at 70°C for 150 sec, for a total of 25 cycles. I used a negative and positive control for all PCR amplifications. Amplification products were cleaned enzymatically with Affymetrix-USB ExoSAP-IT PCR Product Clean-up kits.

Sequencing reactions were performed at the Florida State University Sequencing Facility using an Applied Biosystems 3130xl Genetic Analyzer with capillary electrophoresis or at the DNA

Analysis Facility at Yale University using an Applied Biosystems 3730xl Genetic Analyzer.

Sequencing was conducted using the amplifying primers and two internal sequencing primers

(Table 2.3) that were designed specifically for this study using the program Geneious v. 5.5.7

(Drummond et al. 2012). Sample sequence lengths ranged from 1078 bps to 1730 bps. All sequences were deposited in GenBank (Table 2.2).

2.2.3 Phylogenetic Analyses

All sequences were aligned and edited using Geneious v. 5.5.7 (Drummond et al. 2012). I used the Geneious alignment algorithm for initial alignment and then adjusted manually. Regions encoding ND2 and CO1 were translated to amino acids to ensure there were no premature stop codons and to verify the alignment. Sequences were then checked for redundant haplotypes using

Collapse v. 1.2 (Posada 2004).

Phylogenetic reconstruction was carried out using maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference (BI). The MP and unconstrained ML analyses were run using PAUP* 4.0b10 (Swofford 2003). A heuristic search with 1000 stepwise random addition sequence replicates and using the tree bisection-reconnection branch swapping method was

conducted for the MP analysis. Substitutions were weighted equally and gaps were treated as

missing data. Using these same parameters, 860 bootstrap pseudoreplicates were conducted with

100 random addition replicates to access support. I used the Akaike Information Criterion

(Akaike 1974) as implemented in Modeltest v. 3.7 (Posada & Crandall 1998) to determine that

GTR + I + Γ was the most appropriate model of evolution for the ML and BI analyses, consistent

with other studies of plethodontid salamanders using these markers (Kozak et al. 2005; Wiens et

al. 2006; Kozak et al. 2006; Vieites et al. 2007). Parameters (base frequencies, rate matrix,

proportion of invariable sites, and shape) were estimated from the data. A heuristic search using

1000 stepwise random addition sequence replicates was performed. In addition, a partitioned ML

analysis was performed using RAxML v 7.2.8 (Stamatakis 2006). The data were divided into

eight partitions, 5’ to 3’, in the following manner: tRNA-Met, ND2 by codon position, remaining

tRNAs and OLrep, and CO1 by codon position. RAxML only allows GTR enforced with the

addition of Γ-distributed rate heterogeneity (GTRGAMMA), so this was the model applied to

each partition and using 100 stepwise random addition sequence replicates. In addition, support

was measured using 1000 bootstrap pseudoreplicates in RAxML v 7.2.8.

A Bayesian analysis was conducted using MrBayes v 3.1.2 (Huelsenbeck & Ronquist

2001; Ronquist & Huelsenbeck 2003). The same partitions used in the ML run were used for the

Bayesian analysis and an unlinked GTR + I + Γ model was applied to all partitions. Two runs, utilizing 6 chains each (five heated and one cold), were run for 5 x 107 generations, sampling every 1000 generations. In addition to the average standard deviation of split frequencies, the programs AWTY (Nylander et al. 2008) and Tracer (Rambaut & Drummond 2007) were used to check for convergence of the runs, while Tracer was used to check for stationarity and determine burnin (= 5 x 106).

2.2.4 Hypothesis Testing

I assigned each population to an ecoregion using The Nature Conservancy’s designations

(Slaats 1999). Because these ecoregions cover such expansive areas, I was able to assign all samples to an ecoregion, even when locality data were restricted to state and county (Table 2.2).

This ecoregion hypothesis makes the specific prediction that populations within a particular ecoregion will be more closely related to one another than will populations across ecoregions.

To test the riverine hypotheses, I assigned populations using boundaries of southeastern watersheds designated by the United States Geological Survey (Seaber et al. 1987). Given the small size and presumed low vagility of this species complex, it is unclear exactly how large a river system and its surrounding uplands would need to be to reasonably expect restrictions in gene flow of these salamanders. In order to deal with this uncertainty, I tested the riverine hypothesis as three different, but nested hypotheses corresponding to successively smaller watershed units (subregions > accounting units > cataloguing units). However, due to restrictions in locality data of some samples, I was not able to classify, and therefore constrain, every sample to a clade. In such cases, these samples were allowed to float, along with all other non-Eurycea quadridigitata complex animals, in the tree. These hypotheses make the specific predictions that populations within a particular watershed will be more closely related to one another than will populations across different watersheds.

The final hypothesis test assigned samples to breeding habitats. Nearly all samples were collected during the winter and spring breeding season and had detailed field notes, as well as geographic coordinates, associated with each locality. From these data, I was able to classify 84 specimens used in the phylogenetic analysis into one of four breeding habitats: cypress wetlands

(CW), hillside seepage bogs (HS), steepheads and ravines (SR), and whitewater stream sides

(SS). The remaining individuals were allowed to float within the constrained tree. This hypothesis predicts that adaptation to local habitats will lead to population divergence and therefore, samples collected from the same types of habitats will be more closely related to each other than to animals occurring in different breeding habitats.

The Shimodaira-Hasegawa test (SH test; Shimodaira & Hasegawa 1999) and the approximately unbiased test (AU test; Shimodaira 2002), as implemented in PAUP* 4.0b10

(Swofford 2003), were used for the hypothesis testing. These methods test for significant differences between the ln likelihood of an unconstrained, best phylogenetic tree (the null phylogenetic hypothesis) and that of constrained trees (the alternative phylogenetic hypotheses).

In particular, the SH test corrects for multiple comparisons and is considered to be a conservative approach to hypothesis testing (Buckley 2002). I adjusted the significance level for multiple hypothesis tests using a Bonferonni correction.

2.3 Results

2.3.1 Phylogenetic Analyses

A total of 150 sequences representing 16 species of plethodontid salamanders from the subfamily Spelerpinae were used in the alignment. There were no premature stop codons detected in the ND2 or CO1 genes, strongly suggesting that the amplicons were indeed mitochondrial in nature and not the result of a duplication event inserted into the nuclear genome

(Zhang & Hewitt 1996). The resulting alignment was unambiguous except for a series of eight indels with the following locations: 9 bps between ND2 and tRNA-Trp, 16 bps between tRNA-

Trp and tRNA-Ala, 6 bps between tRNA-Ala and tRNA-Asn, 4 bps within tRNA-Asn, 2 bps within OLrep, 2 bps within OLrep, 1 bp between tRNA-Cys and tRNA-Tyr, and 6 bps within

tRNA-Tyr. These ambiguous indels were excluded from all analyses. The Collapse run found 17

redundant haplotypes. In each case, the redundant haplotype was from the same population as its

duplicate, so the shorter of the two sequences was removed from the alignment. This resulted in

a final alignment of 133 sequences used in all phylogenetic analyses.

All four phylogenetic analyses (MP, unpartitioned ML, partitioned ML, and BI)

reconstructed nearly identical trees (the partitioned and unpartitioned ML trees were identical),

with differences detailed herein. The ML tree from the PAUP* analysis with bootstrap support

values from the MP and RAxML analyses and posterior probabilities from the BI

(MP/RAxML/BI) are reported (Fig. 2.2).

The monotypic genus Urspelerpes was the sister taxon to a strongly supported, monophyletic, Eurycea. All three analyses strongly supported the monophyly of the Ozark

Eurycea complex, the E. lucifuga complex, and the E. bislineata complex. Within the E. bislineata complex, the 23 GenBank samples had the exact same relationship as reported in

Kozak et al. (2006). There was poor support for the monophyly of the E. quadridigitata complex in all three analyses, with the proposed relationships to other Eurycea varying. The MP tree consisted of a polytomy made up of the Ozark complex, the E. lucifuga complex, the E. bislineata complex, and three well supported clades of the E. quadridigitata complex: a clade composed of samples from Alabama eastward, including a sample from Louisiana (clades A + B

+ E. chamberlaini, Fig. 2.2), a clade composed of two sites in Santa Rosa County, FL and

Mobile County, AL (clade C, Fig. 2.2) and a clade composed of samples from south central

Mississippi westward (clade D, Fig. 2.2). The ML and BI trees were similar to the MP tree in these relationships, with a few exceptions. In these trees, clades A, B, C, and E. chamberlaini formed a strongly supported monophyletic group (Fig. 2.2). However, both trees showed weak

54 support for a monophyletic E. quadridigitata complex, with clade E as the sister group to other

Eurycea complexes, rendering E. quadridigitata paraphyletic.

Nearly all other deep clades in the E. quadridigitata complex were strongly supported, the exception being the placement of the strongly supported E. chamberlaini clade. In the MP and BI trees, the E. chamberlaini clade was sister to clade A, though this relationship was weakly supported (MP bootstrap = 56, BI posterior probability = 0.33). In both trees, clade B was sister to E. chamberlaini + clade A. In all analyses, clade D was strongly supported as being monophyletic, regardless of its position in the phylogeny. Within the remainder of the E. quadridigitata complex, several divergent clades are strongly supported (Fig. 2.2). These strongly supported divergences were the focus of the hypothesis testing.

2.3.2 Speciation Models

The results of the SH and AU tests are reported in Table 2.4. The constrained trees for the ecoregion and all three riverine hypotheses were found to be significantly worse than the unconstrained best tree in both the AU test and the more conservative SH test (p < 0.0001, all trees more than 1700 ln likelihood units worse). However, the constrained habitat isolation tree was not found to be significantly different from the best tree (1.09 ln likelihood units worse).

Indeed, when the population assignments for each hypothesis were mapped onto the best tree, a striking pattern for the habitat isolation hypothesis can be seen (Fig. 2.3). Each basal clade (A, B,

E. chamberlaini, and C) is restricted to a single habitat. In contrast, there is no one-to-one correspondence between these lineages and ecoregions or river drainage systems.

2.4 Discussion

2.4.1 Phylogenetic Analyses

The results agree with those of Camp et al. (2009), strongly supporting Urspelerpes as the sister group to Eurycea. There is also strong support for a monophyletic Eurycea, however the basal relationships within the genus are not well supported. One unresolved relationship is whether the E. quadridigitata complex is monophyletic or not. The data are unable to resolve this point, though Lamb & Beamer (2012), using four mitochondrial and two nuclear markers, found the E. quadridigitata complex to be paraphyletic with respect to the Edwards Plateau

Eurycea of central Texas. Within the E. quadridigitata complex, there was strong support for many deeply divergent lineages. This is somewhat unexpected given the absence of recognized subspecies. Three of these divergences are deeper than any divergence seen within any other species or species complex within the genus (Fig. 2.2). This is most evident when compared with the E. bislineata complex, which is composed of six described species and several putative ones

(Kenneth Kozak, pers. comm.). This strongly suggests that E. quadridigitata sensu lato is composed of at least three different species.

2.4.2 Mechanisms of Divergence

These results reveal an example of ecological divergence between sister species involving habitat isolation that, to my knowledge, is the first example in temperate plethodontid salamanders. These results contrast with the recent work in the species rich region of eastern

North America, which found nonadaptive radiations via allopatry and niche conservatism to be common mechanisms explaining the diversity seen in some of the larger genera (Kozak, Blaine, et al. 2006; Kozak, Weisrock, et al. 2006; Kozak & Wiens 2006). Though multiple species of

Desmognathus often inhabit the same stream systems and surrounding forest, this variation

involves microhabitat preferences with as many as three species occupying the same

microhabitat (Kozak et al. 2005; Bruce 2011). Indeed, among plethodontid salamanders,

ecological divergence between sister species due to habitat isolation has only been demonstrated

within the neotropical radiations of the subfamily Bolitoglossinae (Wake 1987), whereby

salamander assemblages partition first by major habitat types (e.g., fossorial, arboreal, bromeliad

specialists) and then by microhabitat preferences.

These data rejected the ecoregion hypothesis and all of the riverine hypotheses. Although

ecoregions often reflect common biogeographic boundaries across many different taxa, they

don’t necessarily reflect those of all taxonomic groups within the region (McDonald et al. 2005).

Another ecoregion scheme may better reflect the observed divergences, however alternative ecoregions maps of the southeastern U.S. show little variation, and the World Wildlife Fund’s designations is slightly less inclusive than the one used in this hypothesis test (Olson et al. 2001).

Rivers have long been thought of as classic examples of vicariant barriers to gene flow

(Wallace 1852) and a number of studies have demonstrated the phylogeographic role played by rivers in the southeastern U.S. (Avise 1996; Walker et al. 1998; Burbrink et al. 2000; Pauly et al.

2007; Zellmer et al. 2012). Most spelerpine salamanders, including most Eurycea, live in and around lotic environments. Means (1974) and Means & Karlin (1989) proposed that the vast river systems in the SECP act as dispersal corridors, whereas the intervening uplands restrict gene flow. Yet, despite breaking river systems into three different spatial classes, combined with life history traits (small body size, presumed low vagility, biphasic life cycle) that might indicate otherwise, I found no evidence for this in the E. quadridigitata complex.

2.4.3 Habitat Isolation, Life History Evolution, and Ecological Speciation

I hypothesize that the evolution of the E. quadridigitata complex into different lentic habitats was driven by departures from the ancestral Eurycea life history. Most Eurycea rarely venture far from the margins of lotic environments, where they carry out their entire life cycle

(Petranka 1998). Lotic environments persist in even the driest of conditions, and larval stages are known to last as long as two years (Petranka 1998). In contrast, animals of the E. quadridigitata complex are thought to rarely occur in lotic environments, such as steephead ravines and first order streams, but are abundant in a variety of lentic environments, such as pond margins and floodplain swamps. Temperature, water level, and other environmental conditions vary widely in these habitats. These preferred environments are often seasonally ephemeral, leaving the salamanders without a permanent water source for much of the year, and consequently placing a wide range of selective pressures on these salamanders that is not encountered in other Eurycea.

Such pressures should be accompanied by changes in life history of E. quadridigitata complex animals. Indeed, there are differences with the larvae of this complex when compared to other Eurycea, having much shorter developmental periods that have been documented to happen in as little as two months (Brimley 1923). In addition, since they are lungless and utilize their skin for respiration, these salamanders must seek out some other source of moisture when these environments dry up. They may do this by going underground or in/under surface objects that retain moisture (e.g., leaf litter, decaying logs). Eurycea quadridigitata are often found far from any apparent body of water (Carr 1940; personal observations). Salamanders have been found within dry pine logs in east Texas, more than 500 m from the nearest ephemeral or

permanent water source (personal observations). These salamanders may react to dwindling

water sources by migrating in search of other sources.

This necessity to adapt to the constantly fluctuating conditions of ephemeral aquatic

environments may be why there was a strong association with habitat. As the ancestral animals

began to depart from inhabiting the more stable lotic habitats and started moving into various

ephemeral habitats, they were exposed to strong selective forces that led to reduced reproduction

between lineages and genetic divergence. As animals migrated out of drying wetlands to seek out

necessary moisture, they would have dispersed across the landscape, encountering an increasing

range of habitat types. The ability to disperse far from permanent water sources combined with

shorter times to metamorphosis, would allow salamanders to breed in almost any temporary body

of water (e.g., I have observed larvae in Longleaf Pine [Pinus palustris] extraction stump holes), greatly expanding ecological opportunity.

Ecological speciation occurs when barriers to gene flow evolve due to ecologically divergent selection (Rundle & Nosil 2005). I propose that the observed mitochondrial lineages in the E. quadridigitata complex are the result of ecological speciation. As increasingly different lentic environments were colonized (e.g., sphagnum bogs vs. cypress swamps), strong selective forces resulted in temporal changes in larval development and increasing terrestriality in adults, resulting in prezygotic reproductive isolation via this habitat isolation.

2.4.4 Marker choice

The criticism of single locus phylogeographic studies is well documented and in many cases legitimate (Knowles & Madison 2002; for a recent review, see Brito & Edwards 2009). It is now well understood that individual loci often differ in their genealogical histories, sometimes

59 substantially, which can impede interpretation of single locus phylogenies (Avise 2000).

Furthermore, the use of mitochondrial markers has its own set of criticisms, such as introgression and, particularly, its maternal inheritance (Avise 2000). Despite these criticisms, mitochondrial markers have advantages over nuclear markers. Since the mitochondria is effectively haploid, it coalesces much faster since its effective population size is a quarter of the nuclear genome

(Kingman 1982a; b; Avise 2000). Indeed, when dealing with deeper internal nodes, the mitochondrial genome is much more likely to reflect the species tree than even a moderate amount (< 16) of nuclear loci (Moore 1995) since the latter is more likely to suffer from incomplete lineage sorting (though see Hoelzer 1997).

I contend that the choice of mitochondrial loci does not invalidate these conclusions. I used a marker less likely to suffer from incomplete lineage sorting, to investigate the phylogenetics of a closely related species complex. The use of the resultant phylogeny to test hypotheses of the observed genetic divergences reveals a surprising, and strongly supported, pattern and outcome. Though I do not claim that the relationships recovered within this tree should be taken as a correct interpretation of the true species tree, I argue that the likelihood of getting such a pattern that fits perfectly with the ecological isolation hypothesis is extremely low.

Perhaps the pattern reflects maternal preferences in breeding habitat, ignoring paternal contributions (though I do not know of any information to support such a scenario). Even so, this scenario would still be a surprising case of maternally mediated ecological isolation. Finally, independent work of Lamb & Beamer (2012) using four mitochondrial and two nuclear loci yielded similar genetic divergences to this work, strengthening the argument that the single locus analysis is not being misled by introgression or driven strictly by maternal inheritance.

2.5 Conclusion

The E. quadridigitata complex harbors several deeply divergent, mitochondrial clades, at

least two (possibly three) of which are more divergent from the rest of the complex than the

recently described E. chamberlaini. The most divergent clade (clade D, Figs. 2 and 3) may be

more closely related to other Eurycea spp. than it is to other E. quadridigitata complex animals.

Despite being an important factor in the explanation of plethodontid species richness in the

Appalachian Mountains, niche conservatism does not appear to play a role in this low elevation plethodontid species complex. Ecoregions and river systems are poor explanations for the observed divergences, despite the presumed low vagility and other life history traits of these small salamanders. Instead, habitat isolation via adult breeding habitats best explains the observed pattern of genetic divergences. I hypothesize that ecological opportunity in numerous lentic habitats of the Southeastern Coastal Plain generated strong selection on various life history traits (e.g., terrestrial adult stage and shorter larval period) in this salamander complex, which further drove diversification into lentic environments. This is the first example of adaptive divergence due to habitat isolation in temperate plethodontid salamanders. Further work, using multiple nuclear loci and morphology, is necessary to assess which of these clades may represent cryptic species.

Figure 2.1 Geographic distribution of the E. quadridigitata complex (denoted by gray line). Circles represent the 64 populations (total of 111 individuals) sampled for this study. Colors of the circles correspond to the clades in Fig. 2.

Figure 2.2 Maximum likelihood phylogram of the mitochondrial ND2, tRNAs, and partial CO1 genes. Support values are reported on nodes as MP bootstrap/RAxML bootstrap/Bayesian PP. Nodes denoted with solid circles or asterisks represent clades in which the support values were ≥ 95/95/.95 or ≥ 99/99/.99, respectively. Some support values for recent nodes were not shown for clarity. The five, color labeled clades in the E. quadridigitata complex correspond to the colored populations in Fig 1.

KW0045 ClinchGA KW0964 IrwinGA KW0049 WareGA KW0041 WayneGA A KW0067 LongGA KW0035 ApplingGA KW0066 LongGA KW0070 LongGA KW0120 NassauFL B KW0010 BradfordFL KW0016 BradfordFL KW0118 NassauFL 81/84/.97 KW0119 NassauFL KW0017 BradfordFL KW0117 MarionFL E. chamberlaini KW0012 BradfordFL KW0094 SumterSC KW0134 BakerFL * KW0001 LeonFL KW0004 LeonFL C KW0005 LeonFL KW0008 LeonFL KW0006 LeonFL KW0078 ThomasGA * KW0079 ThomasGA D KW0009 JeffersonFL KW0019 JeffersonFL KW0048 LowndesGA KW0051 ClinchGA 77/91/.99 KW0132 ThomasGA KW0147 AikenSC KW0020 LeonFL KW0003 LeonFL KW0114 LeonFL KW0099 AlachuaFL KW0144 LibertyFL KW0129 TaylorFL KW0131 TaylorFL KW0113 LeonFL 83/98/1 KW0115 LeonFL KW0007 LeonFL KW0097 BakerFL * KW0098 BakerFL KW0111 LibertyFL KW0112 LibertyFL KW0083 LibertyFL 96/81/.96 61/82/.99 KW0054 WashingtonFL KW0080 OkaloosaFL KW0053 WashingtonFL t KW0037 BayFL * KW0081 OkaloosaFL KW0177 StTammanyLA * KW0023 SarasotaFL * * KW0145 GladesFL * KW0026 VolusiaFL 68/93/.89 KW0125 MarionFL KW0137 AikenSC 99/92/1 KW0141 BarnwellSC 91/ KW0143 BarnwellSC 100/1 KW0044 EffinghamGA * KW0135 JasperSC -/54/- 93/96/1 KW0095 DarlingtonSC KW0096 DarlingtonSC KW0124 BerkeleySC 97/95/.82 KW0126 CharlestonSC KW0039 MaconAL * KW0071 MaconAL 100/89/.99 KW0056 MaconAL KW0090 TaylorGA -/100/- KW0091 CrawfordGA * 98/92/t KW0089 MeriwetherGA .99 KW0040 WashingtonFL * KW0064 WashingtonFL KW0046 CalhounFL Eurycea -/62/.80 KW0024 AndersonSC -/94/1 * KW0127 AndersonSC quadridigitata KW0025 AndersonSC * KW0635 AndersonSC Complex KW0128 PittNC 63/63/.95 KW0085 MobileAL KW0088 MobileAL * t KW0087 MobileAL KW0075 SantaRosaFL t KW0076 SantaRosaFL 89/100/1 E. quadridigitata (DQ018387) * KW0965 JonesMS KW0166 LivingstonLA KW0936 BossierLA * KW0158 GrantLA KW0159 DeSotoLA * -/99/.99 KW0030 SanAugustineTX * KW0032 SanAugustineTX * KW0951 SabineTX * KW0955 JasperTX KW0956 JasperTX * E. aquatica (DQ018651) E. aquatica (DQ018654) t * E. junaluska (DQ018655) 71/92/1 E. junaluska (DQ018656) 94/99/1 E. cirrigera (DQ018486) * E. wilderae (DQ018649) 83/90/1 E.cirrigera (DQ018505) 79/80/1 E. wilderae (DQ018571) E. wilderae (DQ018578) 75/93/1 E. cirrigera (DQ018552) E. wilderae (DQ018613) t E. cirrigera (DQ018511) 75/93/1 H. wallacei KW0959 JacksonFL H. wallacei KW0961 JacksonFL * H. wallacei KW0963 JacksonFL * H. wallacei KW0962 JacksonFL H. wallacei KW0960 JacksonFL E. bislineata (AY728217) t E. bislineata (DQ018391) * * E. bislineata (DQ018405) 79/89/1 E. bislineata (DQ018409) E. cirrigera (DQ018468) * t E. cirrigera (DQ018474) E. cirrigera (DQ018456) t * E. cirrigera (DQ018448) 90/87/1 E. wilderae (DQ018624) * * E. cirrigera (DQ018429) E. cirrigera (DQ018418) * E. guttolineata KW0033 LibertyFL * E. guttolineata KW0034 LibertyFL * E. longicauda (AY916023) 74/92/.99 E. longicauda (DQ018386) E. multiplicata KW0966 PushmatahaOK E. tynerensis KW0967 CrawfordAR U. brucei KW0634 StephensGA S. marginatus (AY916022) * P. montanus (AY916021) G. porphyriticus (NC006341) 0.07

Figure 2.3 Maximum likelihood phylogram from Fig. 2 showing just the E. quadridigitata complex and the five hypotheses (E = Ecoregion, Rs = River Subunit, Ra = River Accounting, Rc = River Cataloguing, and H = Habitat). Note the striking relationship between the Habitat hypothesis and the tree topology. The classification abbreviations are as follows: AH (Alapaha), AJ (St. Andrew-St. Joseph Bays), AL (Alabama), AM (Apalachee Bay-St. Marks), AP (Apalachicola), AT (Altamaha), AS (Altamaha-St. Marys), AU (Aucilla), AW (Aucilla- Waccasassa), BL (Black), BW (Blackwater), CE (Choctawhatchee-Escambia), CH (Choctawhatchee), CN (Canoochee), CP (Chipola), CT (Coosa-Tallapoosa), CW (Cypress Wetlands), EC (Econfina-Steinhatchee), EG (East Gulf Coastal Plain), ES (Edisto-Santee), FC (Florida Panhandle Coastal), FP (Florida Peninsula), HS (Hillside Sphagnum Seep), KI (Kissimmee), LA (Lower Angelina), LC (Lower Choctawhatchee), LL (Lower Leaf), LN (Lower Neches), LO (Lower Ochlockonee), LP (Lower Pee Dee), LS (Lower Savannah), LT (Lower Tallapoosa), MA (Mid-Atlantic Coastal Plain), MB (Mobile Bay-Tombigbee), MK (Myakka), MN (Middle Neuse), MO (Mobile-Tensaw), MS (Middle Savannah), MT (Mobile-Tombigbee), NA (Nassau), NC (Neches), NE (Neuse), NP (Neuse-Pamlico), OC (Ochlockonee), OG (Ogeechee), OK (Oklawaha), OS (Ogeechee-Savannah), PA (Pascagoula), PB (Pensacola Bay), PD (Pee Dee), PE (Peace), PM (Piedmont), PT (Peace-Tampa Bay), SA (Santee), SB (Sabine), SC (South Atlantic Coastal Plain), SE (Seneca), SF (Santa Fe), SJ (St. Johns), SM (St. Marys), SO (Southern Florida), SR (Steepheads and Ravine), SS (Stream Side), SU (Suwannee), SV (Savannah), TB (Toledo Bend Reservoir), UE (Upper East Gulf Coastal Plain), UF (Upper Flint), UJ (Upper St. Johns), UO (Upper Ochlockonee), US (Upper Suwannee), UW (Upper West Gulf Coastal Plain), WG (West Gulf Coastal Plain), WO (Western Okeechobee Inflow).

E Rs Ra Rc H KW0045 ClinchGA SC SU SU US CW KW0964 IrwinGA SC SU SU AH CW KW0049 WareGA SC SU SU US CW KW0041 WayneGA SC AS AT AT CW A KW0067 LongGA SC OS OG CN CW KW0035 ApplingGA SC AS AT AT CW KW0066 LongGA SC OS OG CN CW KW0070 LongGA SC OS OG CN CW KW0120 NassauFL SC SJ SJ NA CW B KW0010 BradfordFL FP SU SU SF CW KW0016 BradfordFL FP SU SU SF CW KW0118 NassauFL SC SJ SJ NA CW E. chamberlaini KW0119 NassauFL SC SJ SJ NA CW KW0017 BradfordFL FP SU SU SF CW KW0117 MarionFL FP SJ SJ OK CW KW0012 BradfordFL FP SU SU SF CW KW0094 SumterSC MAPD LP BL CW C KW0134 BakerFL SC AS SM SM CW KW0001 LeonFL EG OC OC LO CW KW0004 LeonFL EG OC OC LO CW KW0005 LeonFL EG OC OC LO CW D KW0008 LeonFL EG OC OC LO CW KW0006 LeonFL EG OC OC LO CW KW0078 ThomasGA EG OC OC LO CW KW0079 ThomasGA EG OC OC LO CW KW0009 JeffersonFL EG OC OC AM CW KW0019 JeffersonFL EG SU AW AU CW KW0048 LowndesGA SA SU SU AH CW KW0051 ClinchGA SA SU SU US CW KW0132 ThomasGA EG OC OC AM CW KW0147 AikenSC SA OS SV MS CW KW0020 LeonFL EG OC OC LO CW KW0003 LeonFL EG OC OC LO CW KW0114 LeonFL EG OC OC LO CW KW0099 AlachuaFL FP SJ SJ OK CW KW0144 LibertyFL EG OC OC LO CW KW0129 TaylorFL EG SU AW EC CW KW0131 TaylorFL EG SU AW EC CW KW0113 LeonFL EG OC OC LO CW KW0115 LeonFL EG OC OC LO CW KW0007 LeonFL EG OC OC LO CW KW0097 BakerFL SA AS SM SM CW KW0098 BakerFL SA AS SM SM CW KW0111 LibertyFL EG AP AP AP CW KW0112 LibertyFL EG AP AP AP CW KW0083 LibertyFL EG AP AP AP CW KW0054 WashingtonFL EG CE CH LC CW KW0080 OkaloosaFL EG CE FC PB CW KW0053 WashingtonFL EG CE CH LC CW KW0037 BayFL EG CE FC AJ CW KW0081 OkaloosaFL EG CE FC PB CW KW0177 StTammanyLA EG ?? ?? ?? ?? KW0023 SarasotaFL FP PT PE MY CW KW0145 GladesFL FP SO KI WO CW KW0026 VolusiaFL FP SJ SJ US CW KW0125 MarionFL FP SJ SJ US CW KW0137 AikenSC SA OS SV MS CW KW0141 BarnwellSC SA OS SV MS CW KW0143 BarnwellSC SA OS SV MS CW KW0044 EffinghamGA SA OS SV LS CW KW0135 JasperSC SA OS SV LS CW KW0095 DarlingtonSC MA PD LP LP CW KW0096 DarlingtonSC MA PD LP LP CW KW0124 BerkeleySC MA ES SA SA CW KW0126 CharlestonSC MA ES SA SA CW KW0039 MaconAL UE AL CT LT SR KW0071 MaconAL UE AL CT LT SR KW0056 MaconAL UE AL CT LT SR KW0090 TaylorGA SA AP AP UF SR KW0091 CrawfordGA SA AP AP UF SR KW0089 MeriwetherGA PM AP AP UF SR KW0040 WashingtonFL EG CE CH LC SR KW0064 WashingtonFL EG CE FC SS SR KW0046 CalhounFL EG AP AP CP SR KW0024 AndersonSC PM OS SV SE SS KW0127 AndersonSC PM OS SV SE SS KW0025 AndersonSC PM OS SV SE SS KW0635 AndersonSC PM OS SV SE SS KW0128 PittNC MA NP NE MN ?? KW0085 MobileAL EG MT MB MO HS KW0088 MobileAL EG MT MB MO HS KW0087 MobileAL EG MT MB MO HS KW0075 SantaRosaFL EG CE FC PB HS KW0076 SantaRosaFL EG CE FC PB HS E. quadridigitata (DQ018387) ?? ?? ?? ?? ?? KW0965 JonesMS EG PA PA LL ?? KW0166 LivingstonLA EG ?? ?? ?? ?? KW0936 BossierLA UW ?? ?? ?? ?? KW0158 GrantLA WG ?? ?? ?? ?? KW0159 DeSotoLA UW ?? ?? ?? ?? KW0030 SanAugustineTX WG SB NC LA ?? KW0032 SanAugustineTX WG SB NC LA ?? KW0951 SabineTX WG SB SB TB ?? KW0955 JasperTX WG SB NC LN ?? KW0956 JasperTX WG SB NC LN ??

0.07

Table 2.1 Comparison of abiotic and biotic factors of Southeaster Coastal Plain wetlands known to be used as breeding sites for the Dwarf Salamander (Eurycea quadridigitata) .

Habitat Dominant Water Water Hydroperiod pH Soil Detritus Fire Vegetation Movement Source Type Regime Cypress Cypress seasonally shallow ground short to acidic to organic/ high/low moderate Wetlands (Taxodium sp.) flooded water/rain moderate ca. neutral mineral Hillside herbaceous semi-permanent shallow ground long acidic organic high moderate Seepage water to high Steepheads/ mixed forest permanent shallow ground short to ca. neutral organic high low to Ravines water/rain long to alkaline moderate Streamside mixed hardwood permanent rain long ca. neutral mineral low low to alkaline Modified from Myers & Ewel 1990; Whitney et al. 2011

Figure 2.2 List of specimens and corresponding locality data, including the categorical scores for each hypothesis test.

Species ID County, State Lat. Long. GenBank # Ecoregion River Subregion River Accounting River Cataloguing Habitat

E. chamberlaini KW0024 Anderson, SC 34.64 -82.7 XX-XXXX Piedmont Ogeechee-Savannah Savannah Seneca Stream Side E. chamberlaini KW0025 Anderson, SC 34.64 -82.7 XX-XXXX Piedmont Ogeechee-Savannah Savannah Seneca Stream Side E. chamberlaini KW0127 Anderson, SC 34.6 -82.76 XX-XXXX Piedmont Ogeechee-Savannah Savannah Seneca Stream Side E. chamberlaini KW0128 Pitt, NC 35.61 -77.36 XX-XXXX Mid-Atlantic Coastal Plain Neuse-Pamlico Neuse Middle Neuse Stream Side E. quadridigitata KW0635 Anderson, SC 34.64 -82.7 XX-XXXX Piedmont Ogeechee-Savannah Savannah Seneca Cypress Wetland E. quadridigitata KW0001 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0003 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0004 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0005 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0006 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0007 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0008 Leon, FL 30.64 -84.25 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0009 Jefferson, FL 30.6 -83.89 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Apalachee Bay-St. Marks Cypress Wetland E. quadridigitata KW0010 Bradford, FL 29.83 -82.16 XX-XXXX Florida Peninsula Suwannee Suwannee Santa Fe Cypress Wetland E. quadridigitata KW0012 Bradford, FL 29.83 -82.16 XX-XXXX Florida Peninsula Suwannee Suwannee Santa Fe Cypress Wetland E. quadridigitata KW0016 Bradford, FL 29.83 -82.16 XX-XXXX Florida Peninsula Suwannee Suwannee Santa Fe Cypress Wetland E. quadridigitata KW0017 Bradford, FL 29.83 -82.16 XX-XXXX Florida Peninsula Suwannee Suwannee Santa Fe Cypress Wetland E. quadridigitata KW0019 Jefferson, FL 30.58 -83.61 XX-XXXX East Gulf Coastal Plain Suwannee Aucilla-Waccasassa Aucilla Cypress Wetland E. quadridigitata KW0020 Leon, FL 30.58 -84.36 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0023 Sarasota, FL 27.21 -82.24 XX-XXXX Florida Peninsula Peace-Tampa Bay Peace Myakka Cypress Wetland E. quadridigitata KW0026 Volusia, FL 29.07 -81.13 XX-XXXX Florida Peninsula St. Johns St. Johns Upper St. Johns Cypress Wetland E. quadridigitata KW0030 San Augustine, TX 31.33 -94.22 XX-XXXX West Gulf Coastal Plain Sabine Neches Lower Angelina N/A E. quadridigitata KW0032 San Augustine, TX 31.33 -94.22 XX-XXXX West Gulf Coastal Plain Sabine Neches Lower Angelina N/A E. quadridigitata KW0035 Appling, GA 31.92 -82.27 XX-XXXX South Atlantic Coastal Plain Altamaha-St. Marys Altamaha Altamaha Cypress Wetland Choctawhatchee- Florida Panhandle St. Andrew-St. Joseph E. quadridigitata KW0037 Bay, FL 30.43 -85.54 XX-XXXX East Gulf Coastal Plain Escambia Coastal Bays Cypress Wetland Upper East Gulf Coastal E. quadridigitata KW0039 Macon, AL 32.51 -85.6 XX-XXXX Plain Alabama Coosa-Tallapoosa Lower Tallapoosa Steephead/Ravine Choctawhatchee- E. quadridigitata KW0040 Washington, FL 30.72 -85.53 XX-XXXX East Gulf Coastal Plain Escambia Choctawhatchee Lower Choctawhatchee Steephead/Ravine E. quadridigitata KW0041 Wayne, GA 31.49 -81.84 XX-XXXX South Atlantic Coastal Plain Altamaha-St. Marys Altamaha Altamaha Cypress Wetland E. quadridigitata KW0044 Effingham, GA 32.39 -81.3 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Lower Savannah Cypress Wetland E. quadridigitata KW0045 Clinch, GA 30.65 -82.53 XX-XXXX South Atlantic Coastal Plain Suwannee Suwannee Upper Suwannee Cypress Wetland E. quadridigitata KW0046 Calhoun, FL 30.43 -85.17 XX-XXXX East Gulf Coastal Plain Apalachicola Apalachicola Chipola Steephead/Ravine E. quadridigitata KW0048 Lowndes, GA 30.94 -83.17 XX-XXXX South Atlantic Coastal Plain Suwannee Suwannee Alapaha Cypress Wetland E. quadridigitata KW0049 Ware, GA 31.1 -82.56 XX-XXXX South Atlantic Coastal Plain Suwannee Suwannee Upper Suwannee Cypress Wetland E. quadridigitata KW0051 Clinch, GA 31.03 -82.87 XX-XXXX South Atlantic Coastal Plain Suwannee Suwannee Upper Suwannee Cypress Wetland Choctawhatchee- E. quadridigitata KW0053 Washington, FL 30.53 -85.86 XX-XXXX East Gulf Coastal Plain Escambia Choctawhatchee Lower Choctawhatchee Cypress Wetland

Figure 2.2 Continued.

Species ID County, State Lat. Long. GenBank # Ecoregion River Subregion River Accounting River Cataloguing Habitat Choctawhatchee- E. quadridigitata KW0054 Washington, FL 30.53 -85.86 XX-XXXX East Gulf Coastal Plain Escambia Choctawhatchee Lower Choctawhatchee Cypress Wetland Upper East Gulf Coastal E. quadridigitata KW0056 Macon, AL 32.43 -85.64 XX-XXXX Plain Alabama Coosa-Tallapoosa Lower Tallapoosa Steephead/Ravine Choctawhatchee- Florida Panhandle St. Andrew-St. Joseph E. quadridigitata KW0064 Washington, FL 30.45 -85.5 XX-XXXX East Gulf Coastal Plain Escambia Coastal Bays Steephead/Ravine E. quadridigitata KW0066 Long, GA 31.9 -81.77 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Ogeechee Canoochee Cypress Wetland E. quadridigitata KW0067 Long, GA 31.9 -81.77 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Ogeechee Canoochee Cypress Wetland E. quadridigitata KW0070 Long, GA 31.9 -81.77 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Ogeechee Canoochee Cypress Wetland Upper East Gulf Coastal E. quadridigitata KW0071 Macon, AL 32.51 -85.6 XX-XXXX Plain Alabama Coosa-Tallapoosa Lower Tallapoosa Steephead/Ravine Choctawhatchee- Florida Panhandle E. quadridigitata KW0075 Santa Rosa, FL 30.84 -86.94 XX-XXXX East Gulf Coastal Plain Escambia Coastal Blackwater Hillside Seepage Choctawhatchee- Florida Panhandle E. quadridigitata KW0076 Santa Rosa, FL 30.84 -86.94 XX-XXXX East Gulf Coastal Plain Escambia Coastal Blackwater Hillside Seepage E. quadridigitata KW0078 Thomas, GA 30.83 -84 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Upper Ochlockonee Cypress Wetland E. quadridigitata KW0079 Thomas, GA 30.83 -84 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Upper Ochlockonee Cypress Wetland Choctawhatchee- Florida Panhandle E. quadridigitata KW0080 Okaloosa, FL 30.42 -86.77 XX-XXXX East Gulf Coastal Plain Escambia Coastal Pensacola Bay Cypress Wetland Choctawhatchee- Florida Panhandle E. quadridigitata KW0081 Okaloosa, FL 30.42 -86.77 XX-XXXX East Gulf Coastal Plain Escambia Coastal Pensacola Bay Cypress Wetland E. quadridigitata KW0083 Liberty, FL 29.91 -84.96 XX-XXXX East Gulf Coastal Plain Apalachicola Apalachicola Apalachicola Cypress Wetland E. quadridigitata KW0085 Mobile, AL 31.04 -88.19 XX-XXXX East Gulf Coastal Plain Mobile-Tombigbee Mobile Bay-Tombigbee Mobile-Tensaw Hillside Seepage E. quadridigitata KW0087 Mobile, AL 31.04 -88.19 XX-XXXX East Gulf Coastal Plain Mobile-Tombigbee Mobile Bay-Tombigbee Mobile-Tensaw Hillside Seepage E. quadridigitata KW0088 Mobile, AL 31.04 -88.19 XX-XXXX East Gulf Coastal Plain Mobile-Tombigbee Mobile Bay-Tombigbee Mobile-Tensaw Hillside Seepage E. quadridigitata KW0089 Meriwether, GA 33.11 -84.52 XX-XXXX Piedmont Apalachicola Apalachicola Upper Flint Steephead/Ravine E. quadridigitata KW0090 Taylor, GA 32.64 -84.37 XX-XXXX South Atlantic Coastal Plain Apalachicola Apalachicola Upper Flint Steephead/Ravine E. quadridigitata KW0091 Crawford, GA 32.69 -84 XX-XXXX South Atlantic Coastal Plain Apalachicola Apalachicola Upper Flint Steephead/Ravine E. quadridigitata KW0094 Sumter, SC 33.94 -79.97 XX-XXXX Mid-Atlantic Coastal Plain Pee Dee Lower Pee Dee Black Cypress Wetland E. quadridigitata KW0095 Darlington, SC 34.38 -79.73 XX-XXXX Mid-Atlantic Coastal Plain Pee Dee Lower Pee Dee Lower Pee Dee Cypress Wetland E. quadridigitata KW0096 Darlington, SC 34.38 -79.73 XX-XXXX Mid-Atlantic Coastal Plain Pee Dee Lower Pee Dee Lower Pee Dee Cypress Wetland E. quadridigitata KW0097 Baker, FL 30.17 -82.41 XX-XXXX South Atlantic Coastal Plain Altamaha-St. Marys St. Marys St. Marys Cypress Wetland E. quadridigitata KW0098 Baker, FL 30.17 -82.41 XX-XXXX South Atlantic Coastal Plain Altamaha-St. Marys St. Marys St. Marys Cypress Wetland E. quadridigitata KW0099 Alachua, FL 29.51 -82.22 XX-XXXX Florida Peninsula St. Johns St. Johns Oklawaha Cypress Wetland E. quadridigitata KW0111 Liberty, FL 30.11 -85.02 XX-XXXX East Gulf Coastal Plain Apalachicola Apalachicola Apalachicola Cypress Wetland E. quadridigitata KW0112 Liberty, FL 30.11 -85.02 XX-XXXX East Gulf Coastal Plain Apalachicola Apalachicola Apalachicola Cypress Wetland E. quadridigitata KW0113 Leon, FL 30.35 -84.67 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0114 Leon, FL 30.35 -84.67 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0115 Leon, FL 30.35 -84.67 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland E. quadridigitata KW0117 Marion, FL 29.43 -81.72 XX-XXXX Florida Peninsula St. Johns St. Johns Oklawaha Cypress Wetland E. quadridigitata KW0118 Nassau, FL 30.63 -81.86 XX-XXXX South Atlantic Coastal Plain St. Johns St. Johns Nassau Cypress Wetland E. quadridigitata KW0119 Nassau, FL 30.63 -81.86 XX-XXXX South Atlantic Coastal Plain St. Johns St. Johns Nassau Cypress Wetland

Figure 2.2 Continued.

E. quadridigitata KW0120 Nassau, FL 30.63 -81.86 XX-XXXX South Atlantic Coastal Plain St. Johns St. Johns Nassau Cypress Wetland E. quadridigitata KW0124 Berkeley, SC 33.21 -79.47 XX-XXXX Mid-Atlantic Coastal Plain Edisto-Santee Santee Santee Cypress Wetland E. quadridigitata KW0125 Marion, FL 29.34 -81.73 XX-XXXX Florida Peninsula St. Johns St. Johns Upper St. Johns Cypress Wetland E. quadridigitata KW0126 Charleston, SC 33.14 -79.54 XX-XXXX Mid-Atlantic Coastal Plain Edisto-Santee Santee Santee Cypress Wetland E. quadridigitata KW0129 Taylor, FL 29.72 -83.46 XX-XXXX East Gulf Coastal Plain Suwannee Aucilla-Waccasassa Econfina-Steinhatchee Cypress Wetland E. quadridigitata KW0131 Taylor, FL 29.72 -83.46 XX-XXXX East Gulf Coastal Plain Suwannee Aucilla-Waccasassa Econfina-Steinhatchee Cypress Wetland E. quadridigitata KW0132 Thomas, GA 30.76 -84.01 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Apalachee Bay-St. Marks Cypress Wetland E. quadridigitata KW0134 Baker, FL 30.37 -82.39 XX-XXXX South Atlantic Coastal Plain Altamaha-St. Marys St. Marys St. Marys Cypress Wetland E. quadridigitata KW0135 Jasper, SC 32.48 -81.17 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Lower Savannah Cypress Wetland E. quadridigitata KW0137 Aiken, SC 33.25 -81.74 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Middle Savannah Cypress Wetland E. quadridigitata KW0141 Barnwell, SC 33.26 -81.6 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Middle Savannah Cypress Wetland E. quadridigitata KW0143 Barnwell, SC 33.26 -81.64 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Middle Savannah Cypress Wetland E. quadridigitata KW0144 Liberty, FL 30.17 -84.68 XX-XXXX East Gulf Coastal Plain Ochlockonee Ochlockonee Lower Ochlockonee Cypress Wetland Western Okeechobee E. quadridigitata KW0145 Glades, FL 26.9 -81.32 XX-XXXX Florida Peninsula Southern Florida Kissimmee Inflow Cypress Wetland E. quadridigitata KW0147 Aiken, SC 33.39 -81.9 XX-XXXX South Atlantic Coastal Plain Ogeechee-Savannah Savannah Middle Savannah Cypress Wetland E. quadridigitata KW0158 Grant, LA XX-XXXX West Gulf Coastal Plain N/A N/A N/A N/A Upper West Gulf Coastal E. quadridigitata KW0159 DeSoto, LA XX-XXXX Plain N/A N/A N/A N/A E. quadridigitata KW0166 Livingston, FL XX-XXXX East Gulf Coastal Plain N/A N/A N/A N/A E. quadridigitata KW0177 St. Tammany, LA XX-XXXX East Gulf Coastal Plain N/A N/A N/A N/A Upper West Gulf Coastal E. quadridigitata KW0936 Bossier, LA XX-XXXX Plain N/A N/A N/A N/A E. quadridigitata KW0951 Sabine, TX 31.36 -93.72 XX-XXXX West Gulf Coastal Plain Sabine Sabine Toledo Bend Reservoir N/A E. quadridigitata KW0955 Jasper, TX 31.03 -94.28 XX-XXXX West Gulf Coastal Plain Sabine Neches Lower Neches N/A E. quadridigitata KW0956 Jasper, TX 31.03 -94.28 XX-XXXX West Gulf Coastal Plain Sabine Neches Lower Neches N/A E. quadridigitata KW0964 Irwin, GA 31.52 -83.36 XX-XXXX South Atlantic Coastal Plain Suwannee Suwannee Alapaha N/A E. quadridigitata KW0965 Jones, MS 31.44 -88.97 XX-XXXX East Gulf Coastal Plain Pascagoula Pascagoula Lower Leaf N/A * E. quadridigitata complex samples generated in this study

E. guttolineata KW0033 Liberty, FL 30.52 -84.97 XX-XXXX

E. guttolineata KW0034 Liberty, FL 30.52 -84.97 XX-XXXX E. multiplicata KW0966 Pushmataha, OK 34.52 -94.98 XX-XXXX E. tynerensis KW0967 Crawford, AR 35.62 -94.45 XX-XXXX E. wallacei KW0959 Jackson, FL 30.78 -85.15 XX-XXXX E. wallacei KW0960 Jackson, FL 30.78 -85.15 XX-XXXX E. wallacei KW0961 Jackson, FL 30.78 -85.15 XX-XXXX E. wallacei KW0962 Jackson, FL 30.78 -85.15 XX-XXXX E. wallacei KW0963 Jackson, FL 30.78 -85.15 XX-XXXX U. brucei KW0634 Stephens, GA 34.66 -83.31 XX-XXXX

Figure 2.2 Continued.

*Outgroup samples generated in this study

Species ID County, State Lat. Long. GenBank # Ecoregion River Subregion River Accounting River Cataloguing Habitat E. aquatica DQ018651 E. aquatica DQ018654 E. bislineata AY728217 E. bislineata DQ018391 E. bislineata DQ018405 E. bislineata DQ018409 E. cirrigera DQ018418 E. cirrigera DQ018429 E. cirrigera DQ018448 E. cirrigera DQ018456 E. cirrigera DQ018468 E. cirrigera DQ018474 E. cirrigera DQ018486 E. cirrigera DQ018505 E. cirrigera DQ018511 E. cirrigera DQ018552 E. junaluska DQ018655 E. junaluska DQ018656 E. longicauda AY916023 E. longicauda DQ018386 E. quadridigitata DQ018387 E. wilderae DQ018489 E. wilderae DQ018571 E. wilderae DQ018578 E. wilderae DQ018613 E. wilderae DQ018624 G. porphyriticus NC006341 P. montanus AY916021 S. marginatus AY916022 *Samples borrowed from GenBank

Table 2.3 Amplification and sequencing primers for mitochondrial genes (tRNA-Met, ND2, tRNA-Trp, tRNA-Ala, tRNA-Asn, Olrep, tRNA-Cys, tRNA-Tyr, partial CO1) used in this study.

Primer Primer Sequence Reference L4437 5’-AAGCTTTCGGGCCCATACC-3’ Macey et al. 1997 H6159 5’-GCTATGTCTGGGGCTCCAATTA-3’ Weisrock et al. 2001 EqND2FI 5’-GGAGGCCTAAATCAACCACA-3’ This study EqND2RI 5’-GTGATGTGGTGTACGCAAGG-3’ This study

Table 2.4 Results of Shimodaira-Hasegawa and Approximately Unbiased tests of the constrained topologies representing the five hypotheses compared to the unconstrained ML tree in Fig. 2.

Hypothesis ln Difference in SH Test Weighted SH AU Test likelihood ln likelihood (p-value) Test (p-value) (p-value) Ecoregion –22909.57 1713.50 < 0.0001 < 0.0001 ~0 River Subregion –23157.67 1961.60 < 0.0001 < 0.0001 ~0 River Accounting –23621.33 2425.26 < 0.0001 < 0.0001 ~0 River Cataloguing –23112.44 1916.37 < 0.0001 < 0.0001 ~0 Habitat –21197.16 1.09 0.9282 0.9064 0.486

CHAPTER THREE

SPECIES DELIMITATION AND MORPHOLOGICAL VARIATION IN THE DWARF SALAMANDER (Eurycea quadridigitata) COMPLEX

3.1 Introduction

3.1.1 Background

Species are the fundamental units of biology. As such, the patterns and processes responsible for speciation, as well as the recognition of new species, is central to most major disciplines of biology, such as comparative anatomy and physiology, ecology, evolution, and conservation biology (Coyne & Orr 2004). The last four decades have seen rigorous debate on the definition of species, with various arguments centered around abstract or theoretical

(conceptual) definitions vs. more pragmatic (operational) definitions (see Coyne & Orr 2004 and

Sites & Marshall 2004 for detailed reviews). Various lineage concepts, such as the Evolutionary

Species Concept (Simpson 1951), have found prominence in the study of speciation. Such concepts define species as distinct lineages of ancestral-descendant populations that maintain their distinctiveness from other lineages and evolve with their own separate evolutionary trajectories or fate (Simpson 1951; Wiley 1978). The General Lineage Concept (GLC) of de

Queiroz (1998) is a broad lineage concept, which proposes that various lines of evidence (e.g., reciprocal monophyly, reproductive isolation, morphological divergence, etc.) represent temporal benchmarks on the continuum of speciation in a lineage. Therefore, later stages of the process will be supported by increasingly different types of evidence, making the case for speciation and delimitation stronger.

The use of molecular data has played a major role in the study of speciation over the last two decades (Avise 2000; Coyne & Orr 2004). The initial role of species delimitation was mostly restricted to single locus studies that delineated species based on monophyly, despite a number of early studies demonstrating the inherent problems of inferring the species tree from a single gene tree (Pamilo & Nei 1988; Takahata 1989; Avise & Wollenberg 1997). The earliest methods showed marked improvements by providing tests of alternative hypotheses, incorporating morphology (Templeton 2001; Wiens & Penkrot 2002) or multiple loci (Baum & Shaw 1995); however, these methods still suffered from complications of gene trees vs. species trees and issues with Type I errors (Knowles & Madison 2002; Knowles 2008; Wakeley 2008). Given the stochastic nature of inheritance, gene genealogies often differ from one another and from the true pattern of phylogenetic relatedness of species (i.e. the species tree). One extreme example of this is that of mitochondrial vs. nuclear inheritance. In most organisms, the mitochondrial genome is inherited maternally, resulting in a smaller effective population size and faster coalescent time than nuclear markers (Kingman 1982a; b; Avise 2000). Furthermore, retention of ancestral polymorphisms in slower evolving loci can lead to incomplete lineage sorting (Avise et al. 1987;

Pamilo & Nei 1988; Takahata 1989; Maddison 1997). Mitochondrial introgression, hybridization, and gene duplication can also lead to gene trees that are very different from that of the species tree (Maddison 1997). Recent species delimitation methods have attempted to alleviate gene trees-species tree issues due to stochasticity and incomplete lineage sorting through modeling of the coalescent (Knowles & Carstens 2007; Noble et al. 2010; Yang &

Rannala 2010; Leaché & Fujita 2010; Rannala & Yang 2013). Such methods are invaluable in detecting speciation events among closely related taxa that are likely to suffer from lineage sorting and morphological crypsis. One method in particular, the Bayesian Phylogenetics and

Phylogeography (BPP) method of Rannala & Yang (2003) and Yang & Rannala (2010), uses a

Bayesian approach to integrate over uncertainty that arises between gene trees due to coalescent

stochasticity. This method specifically assumes reproductive isolation among a priori chosen populations or putative species, which is the major requirement of the Biological Species

Concept (BSC) of Mayr (1942), however it has been shown to be robust to low levels of gene flow between populations or species (Zhang et al. 2011). The complete and permanent genetic isolation of a population or species represents a point in the continuum of the speciation process at which a lineage actually has its own evolutionary trajectory or fate. Therefore, I view the BSC as the temporally earliest of the various lineage concepts that serves as both a reasonable conceptual and operationally testable definition of species.

Although morphological crypsis is a possibility of recent reproductive isolation, the utility of molecular data to reveal cryptic species is great. However the issue of delineating and describing species with molecular data alone is highly contentious (see Fujita & Leache 2010,

Leaché & Fujita 2010, and Bauer et al. 2011 for a recent example). Nevertheless, the presence of

morphological crypsis does not imply that there is no morphological variation between recently

diverged species, just that it has gone undetected. As with molecular methods, the number of

sophisticated exploratory, descriptive, and hypothesis testing statistics for identifying

morphological divergence has increased over the last decade (Elewa 2004; Zelditch et al. 2012),

allowing researchers to compare morphological change with molecular diversity. The

combination of coalescent-based species delimitation methods using multilocus molecular data

and morphometric data can only strengthen the case for speciation and matches the expectations

of lineage concepts of speciation, such as the BSC of Mayr (Mayr 1942) and the GLC of (de

Queiroz 1998). Such multifaceted approaches should be the standard in research of species

delimitation (Yang & Rannala 2010).

3.1.2 Study System

The Dwarf Salamander (Eurycea quadridigitata) is a species of plethodontid salamander that occurs throughout the southeastern U.S., from the eastern third of Texas eastward through southern Arkansas, Louisiana, the southern halves of Mississippi, Alabama, and Georgia,

Florida, and the coastal lowlands of the Carolinas (Petranka 1998; Fig. 3.1). This small salamander is distinguished from all other members of the genus by having four digits on both the front and hind limbs. It is much more terrestrial than other members of the genus, often found far from permanent water, and is commonly the only plethodontid salamander where it occurs

(Means 2000). Like other members of the genus, it has a biphasic lifecycle, but unlike other

Eurycea which breed in lotic (= flowing water) habitats, E. quadridigitata breed in a variety of lentic (= non-flowing or sluggish water) habitats. Here they lay eggs in water that hatch and remain through metamorphosis (Petranka 1998). Perhaps due to breeding in such highly variable and ephemeral systems, E. quadridigitata has a greatly reduced larval period when compared to other Eurycea, sometimes transforming in as little as two months (Petranka 1998; Means 2000).

Long considered a color morph of E. quadridigitata from the Carolinas, E. chamberlaini was elevated by Harrison & Guttman (2003). Currently, no subspecies are recognized within either species (Tilley et al. 2008).

Chapter 2 revealed deep genetic lineages within a range-wide phylogeographic study of

E. chamberlaini and E. quadridigitata using ~1700 bp mitochondrial fragment. I tested five competing hypotheses of population divergence demonstrating that ecoregions and various

77 vicariant riverine models did not explain the phylogeographic pattern. Instead, breeding habitats best explained the observed phylogeographic pattern. This striking pattern of ecological isolation due to breeding habitats was unexpected, as many previous studies have suggested that plethodontid salamanders experience niche conservatism, unable to expand beyond the narrow niche conditions of their ancestors (Wake 2006; Kozak et al. 2006; Kozak & Wiens 2006, 2010).

I hypothesize that a shift from the ancestral condition of inhabiting lotic environments into a more terrestrial existence and utilization of various lentic breeding habitats lead to habitat isolation, reducing gene flow among these diverging habitat populations resulting in reproductive isolation. This hypothesis predicts an overall agreement in genome-wide genetic divergences and morphological trait changes associated with these observed mitochondrial lineages. If they are reproductively isolated, these habitat populations should cluster in a multilocus molecular reconstruction and may have morphological differences associated with breeding habitat differences. Although these population assignments are based on a single mitochondrial gene tree, the striking correlation between clades and breeding habitats sets up a test of a realistic biological scenario: habitat isolation facilitating speciation via reproductive isolation. Herein, I test the predictions of this hypothesis using a multilocus nuclear gene tree-species tree approach with the Bayesian coalescent species delimitation method of Rannala & Yang (2003; 2010), though I do not directly test reproductive isolation. I also use a variety of exploratory and frequentist hypothesis tests to see how well a morphological data set consisting of 12 size and shape variables agrees with the molecular data.

3.2 Materials and methods

3.2.1 Molecular Sampling

I collected 23 tissue samples from throughout the range of the E. quadridigitata complex and supplemented my tissue sampling with 27 samples from other collections, for a total of 50 samples of E. quadridigitata and E. chamberlaini (Fig. 3.1, Table 3.1). I based my sampling strategy on the results from Chapter 2, sampling at least two individuals from each divergent mitochondrial lineage in order to test if these lineages represent distinct species. Tissue samples consisted of either liver or tail tips preserved in 95% ethanol and/or stored at -80°C. Voucher specimens were set in 10% formalin and stored in 70% ethanol.

3.2.2 Molecular Methods

Genomic DNA was extracted using the hot phenol-chloroform-isoamyl alcohol/chloroform-isoamyl alcohol method (Sambrook & Russell 2001). Extracts were visualized on agarose gels and DNA concentration quantified using a Nanodrop ND-1000

Spectrophotometer (NanoDrop, Wilmington, DE, USA) or an Invitrogen Qubit Fluorometer

(Life Technologies Corporation, Carlsbad, CA, USA). Polymerase chain reaction (PCR) was performed using the following reagents and concentrations: 13.3 µl distilled water, 5 µl 5X

Colorless GoTaq Reaction Buffer, 1.5 µl MgCl2 (25 mM), 1.5 µl dNTPs (2.5 mM), 0.2 µl GoTaq

(5u/ul), 0.5 µl bovine serum albumin (10 mg/ml), 1.0 µl of each primer (10 ng/ul). I amplified a

713 bp fragment of the mitochondrial cytochrome b (CytB), a 988 bp fragment of the

mitochondrial NADH dehydrogenase subunit 2 (ND2) and portion of the adjacent tRNA-Trp,

and an 1146 bp fragment of the nuclear recombination activating gene 1 (RAG1). Additionally, I

developed three anonymous nuclear loci (ANL) using the method of Noonan & Yoder (2009).

This method digests genomic DNA using restriction sites and then scans for preferred target-size

fragments using gel visualization. Target-sized fragments are then transformed into E. coli and

cultured, at which point colonies are picked at random for amplification and visualized on gels to

determine size of the amplicons. Fragments are then selected for sequencing and primer

development. Sequences were then blasted against known sequences in GenBank. The 405 bp

anonymous locus 2 (EqAL02), the 405 bp anonymous locus 21 (EqAL21), and the 231 bp

anonymous locus 51 (EqAL51), combined with the other three loci, represent a total of 3888 bp

of sequence.

All amplifications were performed using the primers and thermal cycler programs listed

in Table 3.2. I used a negative and positive control for all PCR amplifications. Amplification

products were purified enzymatically with Affymetrix-USB ExoSAP-IT PCR Product Clean-up

kits (USB Corporation, Cleveland, OH, USA). Sequencing reactions were performed at the

University of Mississippi using an Applied Biosystems 3130xl Genetic Analyzer with capillary electrophoresis (Applied Biosystems Inc., Foster City, CA, USA) or at the DNA Analysis

Facility at Yale University using an Applied Biosystems 3730xl Genetic Analyzer (Applied

Biosystems Inc.). Sequencing was conducted using the amplifying primers and internal sequencing primers listed in Table 3.2. Anonymous loci and some internal sequencing primers were designed using the program Geneious v. 5.5.7 (Drummond et al. 2012). All 50 individuals

were sequenced for all six genes with the exception of one sample for EqAL02 (KW0125) and

six samples for CytB (KW0010, KW0070, KW0075, KW0087, KW0125, and KW0126). All

sequences were deposited in GenBank (Table 3.1).

Nuclear data was phased using PHASE v. 2.1.1 (Stephens et al. 2001; Stephens & Scheet

2005) as implemented in DnaSP v. 5.10.1 (Librado & Rozas 2009), and the most probable alleles

used in the analyses. All sequences were aligned and edited using Geneious v. 5.5.7. I used the

Geneious alignment algorithm for initial alignment and then adjusted manually. I translated all

six sequence-reading frames into amino acids to check for stop codons and to verify the

alignment. In each of the three anonymous loci, a reading frame existed that contained no stop

codons. Additionally, there were relatively few variable sites and most occurred at positions that were multiples of three. I interpreted this as evidence of these loci being exons and, accordingly, applied a model of nucleotide evolution to each ANL. All alignments were unambiguous.

Models of nucleotide evolution were then chosen using jModeltest v. 2.1.1 (Guindon &

Gascuel 2003; Darriba et al. 2012), with the Akaike Information Criterion (AIC; Akaike 1974)

used to determine the most appropriate model for each locus. The AIC results from the

jModelTest runs indicated the following nucleotide substitution models as the best fit of the

alignments: TPM1uf+I+Γ (AL02), TPM3uf+I (AL21), TrNef+I (AL51), TPM3+I+Γ (RAG1),

GTR+I+Γ (CytB), and TIM1+I+Γ (ND2). Due to restrictions in available nucleotide substitution

models in *BEAST, I implemented the next most complex model for each locus, resulting in the

following models: TrNef+I with equal rates (AL51), GTR+I with equal rates (AL21), and

GTR+I+Γ (AL02, RAG1, CytB, and ND2).

3.2.3 Population Assignments

Many recent species delimitation methods have used programs that implement model

based cluster methods (e.g., STRUCTURAMA; Huelsenbeck et al. [2011]) to a priori assign

individuals to populations or species (Noble et al. 2010; Leaché & Fujita 2010; Pinzón &

LaJeunesse 2011). Though these methods are improvements over more traditional methods (e.g.,

relying on monophyletic groups of single gene trees), they can oversimplify biological reality

(e.g., assume no admixture among populations; François & Durand 2010). My approach to

population assignment was to assign each nuclear allele and the mitochondrial haplotypes to one

of five populations based on the mitochondrial phylogeographic analysis (Chapter 2), in which

four well-supported lineages of E. quadridigitata represented clades that utilized different, distinctive breeding habitats described below.

Cypress wetlands (CW) are dominant across most of the southeastern U.S. and range from sluggish backwaters of floodplain swamps to the completely still-bodied ephemeral wetlands. These wetlands are constantly fluctuating in water levels due to floods and seasonal and long term droughts. They often have significant emergent vegetation around their perimeters and thick accumulated detritus layers on their bottoms. Hillside seepages (HS) are characterized by acidic soils and are often dominated by carnivorous plants (e.g., pitcher plants, Sarracenia

sp.) and other acidic-adapted species, forming bogs with minimal canopies due to low soil

nutrient content. Flow rates are constant but slow, due to the abundant ground vegetation, and

though water levels fluctuate, source seeps usually stay damp in all but the driest periods. In

contrast, steepheads and ravines (SR) are heavily canopied with permanent (in the case of

steepheads) water flow. Steephead ravines are found nowhere else in the world outside of a very

narrow area in the Florida panhandle and adjacent parts of Alabama and Georgia, and differ

significantly from traditional ravines in their formation, but often house very similar floral and

faunal communities (Means 1977, 2000). These habitats are kept relatively cool by the various

over story levels and, consequently, have some of the highest salamander species richness and

densities found in the southeastern U.S. Streamside (SS) communities are permanent water

sources characterized by high flow rates. Larvae of most salamanders that live in these situations

82 are stream-adapted with distinctive morphologies. In this study, streamside populations correspond to the samples of E. chamberlaini.

A fifth lineage exists and is herein referred to as the Western population (WP). It is not clear whether this fifth lineage corresponds to a distinctive breeding habitat, as data were lacking for these specimens; however, this lineage was strongly supported and evidence suggests that it is more closely related to other species of Eurycea than to other E. quadridigitata (Lamb &

Beamer 2012) to which it is currently classified.

3.2.4 Species Tree Estimation

To estimate the species tree from the five loci and habitat-derived population designations, I used the multilocus Bayesian model implemented in the program *BEAST v.

1.7.4 (Heled & Drummond 2010), which integrates over uncertainty in the gene trees, the coalescent, and models of nucleotide evolution. I used the phased nuclear alleles in combination with the mitochondrial haplotypes. I unlinked the nucleotide substitution and clock models for all six loci and the tree model for the four nuclear loci, leaving the mitochondrial tree model linked.

I partitioned the data by codon position across all loci and used uniform priors at each position.

An uncorrelated lognormal relaxed clock model was used for each data partition. All tree priors were set to Yule process and utilized a random starting tree. I performed two independent analyses, each run for 2.5 x 108 generations and sampled every 2.5 x 104 generations, to ensure convergence. I combined the resultant data sets using LogCombiner v. 1.7.4 (Drummond &

Rambaut 2007). I checked for convergence using AWTY (Nylander et al. 2008) and Tracer

(Rambaut & Drummond 2007). I used Tracer to check for stationarity and discarded the first

2000 trees as burnin.

3.2.5 Bayesian Species Delimitation

I used a modified version of the Bayesian species delimitation method of Leaché & Fujita

(2010), assigning individuals to breeding habitats or to WP, a divergent lineage that is more

closely related to the Edwards Plateau neotenes of central Texas, rather than using a model-based

cluster analysis of allele frequencies (e.g., STRUCTURAMA). The program Bayesian

Phylogenetics and Phylogeography (BPP, Rannala & Yang 2003; Yang & Rannala 2010) was

used to conduct the species delimitation analysis. This method takes into account both the

species phylogeny and issues of lineage sorting due to ancestral polymorphism. I chose BPP

since it has been shown to outperform other Bayesian delimitation methods and because it does

not assume that the gene trees are known without error (Camargo et al. 2012). I used the tree from the *BEAST analysis (based on habitat populations) as the guide tree with the default species model prior (uniform rooted trees). BPP utilizes a gamma distribution for its priors on the parameters, with G(α, β) having m = α/β and s2 = α/β2 (Rannala & Yang 2003; Yang & Rannala

2010). It is recommended to try a variety of parameter settings on the root priors and the ancestral population sizes, with congruence among results a strong indicator of confidence in the species delimitation (Yang & Rannala 2010; Leaché & Fujita 2010; Rannala & Yang 2013). In order to assess the robustness of the guide tree and population assignments, I employed gamma priors with three combinations of parameter values for the population sizes (θs) and root age of the species tree (τ0): (1) θ = G(1, 10), τ0 = G(1, 10), reflecting a large effective population size and deep root divergence; (2) θ = G(2, 2000), τ0 = G(2, 2000), representing a small effective population size and shallow root divergence; and (3) θ = G(1, 10), τ0 = G(2, 2000), representing

a large effective populations size and shallow root divergence. The latter combination of

parameter values is considered a conservative combination, in that it should favor fewer species

delimitations (Yang & Rannala 2010; Leaché & Fujita 2010). Other divergence time parameters were assigned a Dirichlet prior (Yang & Rannala 2010: Equation 2). In order to insure proper mixing of the reverse jump Markov chain Monte Carlo (rjMCMC), I ran one analysis with each species delimitation model as the starting tree (n = 5) and ran at least one analysis utilizing each of the rjMCMC algorithms (Yang & Rannala 2010; Equations 3 and 4 and Equations 6 and 7, respectively). The starting trees are equivalent to the guide tree with various nodes collapsed

(e.g., “0000” is the starting tree representing all lineages as a single species, collapsed at the root, and “1111” represents the fully resolved guide tree with five species). Each analysis was run for

50,000 generations with a burnin of 5000, which has been shown to be long enough to ensure convergence in many large datasets (Yang & Rannala 2010).

3.2.6 Morphological Data

I collected data for 12 morphological characters on 199 preserved salamanders (Table

3.3). This total included specimens used in the molecular analyses, specimens from the exact same locality as animals used in the molecular analyses, or from animals that were collected at a breeding wetland that was identified (or could be identified), so that individuals could be confidently placed into one of the five populations. Sex can be difficult to ascertain in most salamanders due to a lack of obvious sexual dimorphism (Petranka 1998). I did not dissect specimens to determine sex since specimens are small and I wanted to avoid the risk of permanent damage or destruction. Whenever possible, animals were sexed based on the presence of cirri (secondary sexual character on the snout of males) or when ova could be detected

85 through the thin abdominal wall of a female. Measurements consisted of (1) snout-vent length

(SVL; distance from tip of snout to posterior edge of cloaca), (2) tail length (TL; distance from posterior edge of cloaca to tip of tail), (3) tail height (TH; dorsal-ventral distance immediately behind cloaca), (4) tail width (TW; lateral distance immediately behind cloaca), (5) hind limb length (HLL; outstretched distance between the limb insertion and longest toe), (6) fore limb length (FLL; outstretched distance between the limb insertion and longest toe), (7) head length

(HL; midline distance between tip of snout and gular fold), (8) head width (HW; widest point on head), (9) head depth (HD; dorsal-ventral distance in front of gular fold), (10) canthus distance

(CD; distance between anterior corner of eye and midline tip of snout), (11) interocular distance

(IO; distance between anterior corners of eyes), and (12) ocular distance (OD; distance between anterior and posterior corners of eye). Limb measurements were taken on the animals left side. It was noted when tails had been broken or were regenerated. All measurements were taken to the nearest 0.1 mm using 150 mm digital calipers. Samples sizes ranged from seven (SS population) to 136 (CW population).

3.2.7 Morphological Analyses

All analyses were conducted using the R statistical package v. 2.15.3 (R Development

Core Team 2013). I performed a Principal Components Analysis (PCA) on the correlation matrix of the morphological traits in order to obtain standardized principal components (PC) to explore the variance due to size and shape among the populations. When variable loadings of PC1 are of similar magnitude and are all the same sign, this is often interpreted as principally indicating general size (Jolicoeur 1963). Subsequent PC axes are then interpreted as estimates of the contribution of other morphological traits to body shape.

I calculated the mean and standard deviation of each morphological trait for each population. I checked for normality of the morphological traits with histograms, quantile- quantile probability plots, and the Shapiro-Wilk test and homogeneity of variance with Levene’s test. Traits that were normally distributed and homoscedastic were first analyzed with a one-way, type III analysis of variance (ANOVA) of each morphological variable with population as the factor, to first test whether there was a significant difference among populations. If a significant difference was found, that variable was then used in pairwise t-tests among all populations to see which populations were different for the trait of interest. Since the pairwise t-test conducts multiple tests simultaneously, I used the conservative Bonferroni correction to adjust the p-value.

Traits that were not normally distributed were analyzed using nonparametric versions of the aforementioned parametric tests. Kruskal-Wallis tests were run on these variables to test for significant differences among populations. If there was a significant difference among populations, I performed a Mann-Whitney U test with a Bonferroni correction on the variable to test for pairwise differences among populations. Finally, I carried out a Discriminant Function

Analysis (DFA) on all significantly different variables to test their utility in classifying individuals to populations.

3.3 Results

3.3.1 Species Tree

The gene trees from the *BEAST analysis were congruent in most major aspects (Figs.

3.3-3.7). In all five trees, the WP and HS populations formed well-supported clades with posterior probabilities (PP) = 0.96–1.0. In two gene trees (RAG1 and mtDNA; Figs. 3.6-3.7), the

WP was the sister clade to all other populations, but in the AL02 tree (Fig 3.3), the HS

population was the sister clade to all other populations. In the AL21 (PP = 0.88; Fig. 3.4) and

AL51 (PP = 0.86; Fig. 3.5), the WP and HS populations formed a clade that was sister to all

other populations. The CW, SR, and SS populations formed a clade with PP = 1.0 in all five gene

trees. In the RAG1 and mtDNA trees, the SR (PP = 0.98 and 1.0, respectively) and SS (PP = 0.64

and 1.0, respectively) populations were monophyletic, but were not for the anonymous loci. The

remaining relationships among the CW, SR, and SS populations were in conflict between most

trees and showed moderate (RAG1) to poor (anonymous loci) support, with the mtDNA tree

being the exception. This is expected, given the smaller effective populations size and faster

coalescence of the haploid mitochondria.

The species tree from the *BEAST run is shown in Figure 3.2. The WP (for which habitat

data were lacking) was sister to the other four populations. The remaining populations formed a

clade (PP = 0.85) with the HS population sister to the remaining three. The divergence between

the HS population and these populations was relatively deep in the tree. The next clade was

supported with a PP of 1.0 and was much shallower than the previous node. It was composed of

a clade containing the SR population and E. chamberlaini (PP 0.45), which was sister to the CW

population.

3.3.2 Species Delimitation

The results from the BPP species delimitation analysis are shown in Figure 3.2. There

was strong support for the five species model (“1111”) across all combinations of starting trees

and parameters, with the exception of a single run with large θ (G[1, 10]) and large τ0 (G[1, 10]) that used the “1100” model as the starting tree. In that analysis, the “1100” model was selected with a speciation probability of 1.0 for both nodes, supporting a three species delimitation.

3.3.3 Morphological Analyses

The means and variances for each variable are reported in Table 3.4. All traits were

homoscedastic and I failed to reject normality for TL, HLL, FLL, and HL. The remaining traits

were not normally distributed (SVL, TW, TH, HW, HD, CL, IO, and OD) and were analyzed

with the nonparametric tests. Due to the fact that TL was highly variable and tails are easily

broken and regenerated to various degrees, I excluded TL from further analyses.

Principal component loadings from the PCA on the correlation matrix of morphological

traits are reported in Table 3.5. The first PC explained 60% of the total variation and loadings

were negative and roughly of the same magnitude, so I consider this axis to represent size. I

cautiously interpret the remaining axes to represent variation due to shape, although I

acknowledge that size may still explain some of the variation in these axes (Mosimann & James

1979). PC2 explained 12% of the variation, with high positive loadings on FLL and HL, as well

as large negative loadings on TH, TW, and HD. Because the component variances starting with

PC3 were below 1.0 and it only explained 6% of the variance, I did not interpret beyond the first

two PCs, though all PC axes are reported in Table 3.4.

The results of the ANOVAs and Kruskal-Wallis test are reported in Table 3.6. All

-11 variables were significantly different among populations: CL (H[4] = 55.72, p = 2.30 x 10 ),

-3 -8 FLL (F[4,194] = 4.69, p = 1.24 x 10 ), HD (H[4] = 39.36, p = 5.88 x 10 ), HL(F[4,194] = 9.98, p =

-7 -5 -12 2.3 x 10 ), HLL (F[4,194] = 6.94, p = 3.06 x 10 ), HW (H[4] = 60.16, p = 2.69 x 10 ), IO (H[4] =

-7 -9 -10 36.95, p = 1.85 x 10 ), OD (H[4] = 46.59, p = 1.86 x 10 ), SVL (H[4] = 49.28, p = 5.09 x 10 ),

-9 -10 TH (H[4] = 46.47, p = 1.97 x 10 ), and TW (H[4] = 49.48, p = 4.63 x 10 ).

The results of the pairwise t-tests and Mann-Whitney U tests are reported in Table 3.7

(for brevity, only p-values are reported in the table). The WP significantly differed from all populations in every character, with the exception of FLL, HLL, IO, and SVL when compared to the SS population. The CW population differed significantly from the SR population in SVL,

TH, and TW and from the HS population in CL, HW, SVL, TH, and TW. The only character that was different between CW and SS was TW. Among the HS, SR, and SS populations, no traits significantly varied.

The DFA scores yielded mixed results in regards to properly placing specimens into the appropriate a priori groups: CW (98%), WP (70%), HS (28%), SR (6%), and SS (0%). Plots of

LD1 vs. LD2, LD3, and LD4 (Fig. 3.9), respectively, indicate that the WP animals are distinct from the other populations. However, additional plots of LD2 vs. LD3, LD2 vs. LD4, and LD3 vs. LD4 indicate that the populations are not differentiated. The traits that discriminated best among populations were HL, IO, OD, and SVL (Table 3.8). Ocular diameter had the strongest discriminatory power in the first, second, and fourth discriminant functions. Interocular distance was the second strongest predictor in the first discriminant function and the strongest predictor in the third discriminant function. The group means on the discriminant variables indicate that the

WP is best discriminated from the other populations by HL, IO, OD, and SVL. The HS population is best discriminated from the CW, SR, and SS populations by IO and further discriminated from the SR population by HL and OD. The HS population also differs from the

SS population in OD and can be further distinguished from CW and SS populations by SVL.

Snout-vent length and HL best discriminate between the CW population and the SR population.

Interocular distance best discriminates between the CW and SS populations. Finally, the SR and

SS populations are best discriminated by IO and OD.

3.4 Discussion

I present strong molecular and ecological evidence that E. quadridigitata sensu lato represents a cryptic species complex composed of at least four species. General concordance of multiple independent loci has been considered strong evidence for deep genetic divergences between populations, indicative of gene flow restriction, rather than shallow, more recent gene flow (Avise & Ball 1990). The five gene trees in this analysis were largely congruent and generally supported the monophyly of each of the five populations. These gene trees were generated using a variety of sequence data, including the maternally inherited, haploid mitochondria and three ANL developed specifically for use in E. quadridigitata. The strength of using ANL is that they represent a random sampling of the nuclear genome and are not expected to be linked (though they could be). These results suggest that the mitochondrial lineages strongly associated with distinctive breeding habitats reflect deep genetic divergences.

Furthermore, these divergences are most likely due to long term reductions in gene flow. Such a scenario is indicative of reproductive isolation.

The habitat isolation model of ecological speciation predicts the evolution of pre-zygotic reproductive isolation via preferences for different habitats (Coyne & Orr 2004). Divergence, and ultimately speciation, is likely to occur more rapidly if the preferred habitats involved are associated with mating and reproduction. The biphasic life cycle of the E. quadridigitata complex allows for non-mating, terrestrial adults to interact with individuals from other populations, however, during the mating season, salamanders are forced to return to wetlands for courtship, egg deposition, and larval development, effectively isolating them from individuals with preferences for different breeding sites.

This striking pattern of clade divergence based on breeding habitat was further supported

with the BPP species delimitation analysis. I found strong support for the recognition of five

species conducting multiple BPP analyses using a wide variety of combinations of settings for

the means (0.1–0.001) and variances (0.01–5x10-7) of the ancestral populations size (θ) and the

age of the root (τ0), as well as varying the rjMCMC algorithm and starting trees (Fig. 3.2). All speciation probabilities were 1.0 for all analyses, with the exception of one run. This particular analysis favored a three species model (speciation probabilities all = 1.0), collapsing the node containing the CW, SR, and SS populations into a single species. In contrast, Leaché & Fujita

(2010) found that this same combination of priors led to increased speciation probabilities favoring more species. These two parameter settings represent prior probabilities of small ancestral population sizes and shallow divergences, both conditions that allow for rapid coalescence and shorter times to complete lineage sorting. If, however, the lineages in the tree actually had large ancestral populations and deep divergences, coalescence would require much longer times, with lineage sorting persisting for long periods. An unrealistic model (in this case a small ancestral population size and shallow divergence) imposed on data that actually came from a large ancestral population size and diverged further back in time, could interpret lineage sorting as gene flow and would collapse these populations into a single lineage. The node in question

(uniting CW, SR, SS) is estimated to have arisen approximately 22 Ma (35-14 Ma 95% HPD credible interval) based on a fossil calibrated phylogeny (Chapter 1). Additionally, E. quadridigitata animals are abundant salamanders throughout most of their range, including within the areas where these three populations occur (Petranka 1998; Means 2000). This suggests that the assumptions of shallow divergences and small ancestral population sizes present an

unrealistic model for this group. I interpret this odd result as a case of fitting unrealistic

parameters to the data.

Selection may favor traits associated with preference for divergent habitats if they

increase fitness in the preferred habitat or decrease fitness of hybrids via reinforcement (Coyne

& Orr 2004; Rundell & Price 2009). Four of the five populations utilize very different breeding habitats, each of which is strongly supported by the molecular data. The results of the PCA suggests that most of the variation is related to size, as there is nearly identical component variance and in the same direction within the first axis among all of the traits involved (Table

3.5, Fig. 3.8). Most of this size variation appears to be driven by the WP (Tables 3.4 and 3.5).

PC2 is correlated with limb length and tail shape. Most permanently aquatic salamanders are characterized by having elongated bodies with short limbs and tall, thick tails that are used to propel them through aquatic environments (e.g., Amphiumidae, Cryptobranchidae, Proteidae,

Sirenidae, neotenic Gyrinophilus). The SS population shows a correlation between reduction in limb length and increase in tail width and height as body size increases (Fig. 3.8). This population also breeds in a lotic system (streams) with permanent, flowing waters. All other populations show the opposite (or no) correlation between limb size and tail proportions. This suggests that selection could be favoring this morphology in the SS populations due to its breeding habitat. However, I only had seven samples from this population making, so any interpretation is tenuous at best.

The statistical hypothesis tests of pairwise morphological comparisons among populations only support the differentiation of the CW and WP populations (Table 3.6). The WP has been separated from other E. quadridigitata complex animals the longest (Fig. 3.2) and other evidence has suggested that E. quadridigitata sensu lato is not a monophyletic group, with the

WP being sister to a radiation of Eurycea inhabiting the Edwards Plateau of central Texas

(Chapter 1; Fig. 1.2, clade K). The HS, SR, and SS populations did not differ significantly among each other in any trait. These morphological tests support the recognition of three populations; WP, CW, and HS/SR/SS.

The DFA performed relatively poorly in classifying samples. Individuals in the CW and

WP populations were placed with the most success (98% and 70%, respectively), with HL and

SVL being the best traits for placing specimens (Table 3.8). I suspect that this may be a combination of two things: sample size and geographic scale. More specimens were examined from these two populations than from the other three. Additionally, these two clades were sampled from throughout wide geographic areas, possibly capturing more of the variation for those particular populations than for the remaining populations. Additional sampling from the

HS, SR, and SS populations may capture more variation than exists in the present study, which may result in better placements of individuals. A third explanation is that these populations may be morphologically constrained due to a common, but unknown, selective force shared among these habitat clades. Additionally, two of the populations (RS and SS) are closely related and have the most recent, relative divergence time. Thee populations may not have been isolated long enough for significant morphological differences to have a formed.

A final possibility is that the HS, SR, and SS populations do not represent distinct species. The morphological data is unable to distinguish the populations from one another, suggesting that they belong to the same species. One of these (SS) is represented in this study by specimens that belong to the recently described E. chamberlaini. This species was separated from E. quadridigitata sensu lato using allozymes, tooth counts and position, osteological characters, color pattern, coastal groove count, and SVL and relative limb length (Harrison &

Guttman 2003). The data I collected overlaps with only two of these characters (SVL and

relative limb length). The PCA results disagree with the conclusion of Harrison & Guttman

(2003) that E. chamberlaini has relatively longer limbs, but their sample size was much larger (n

= 47). The molecular data also agree with the distinctiveness of E. chamberlaini from E. quadridigitata, though perhaps not from the SR populations (PP = 0.45, Fig. 3.2). Without examining more specimens, it is difficult to dismiss the results of Harrison & Guttman (2003).

Furthermore, though the HS populations could not be told apart morphologically from the

SS (= E. chamberlaini) and SR populations, molecularly they were very divergent and sister to a clade that contained E. chamberlaini, CW, and the SR population (PP = 0.88, Fig. 3.2). Since the

CW populations were significantly different compared to all other populations in at least one trait and could be placed with 98% accuracy in the DFA, it seems unlikely that a molecularly divergent clade, such as the HS populations, that differs significantly from the CW population in five traits, is the same species as the CW populations. Since the CW population is more closely related to the SR and SS populations, it seems more plausible that the HS population represents a distinctive species and that examining more individuals from a greater geographic area or looking at other features of morphology (osteology, color, etc.) may reveal differences that separate them from the other populations.

This may not be the case for the SR and SS populations, which could not be distinguished from one another morphologically and had low support in the species tree, though the species delineation method and differences in breeding habitat suggest they are different. Since the two populations are sister taxa and have the most shallow divergence of all the populations examined, the low support may be due to incomplete lineage sorting. Although they have different breeding habitats, there are similarities between the two types (e.g., flowing, permanent

water). It is possible that there is overlap between these two lineages not captured by the

sampling in this study that would explain the lineage sorting, or that might allow for recent gene

flow that would obfuscate the history of these two lineages.

3.5 Conclusion

The data strongly supported a case of ecological speciation due to habitat isolation in the

E. quadridigitata complex. Populations, corresponding to different breeding habitats, form well-

supported molecular clades based on six molecular loci. The BPP coalescent-based species

delimitation method of Rannala & Yang (2003) and Yang & Rannala (2010) strongly supports a

five species delimitation model under a wide variety of initial parameter settings reflecting very

different evolutionary scenarios. The WP is the most distinctive population, most likely because

it represents a cryptic species that is more closely related the Edwards Plateau neotenes of central

Texas than to the E. quadridigitata complex. The CW and HS should be recognized as distinct species based on molecular and morphological data. The SS populations correspond to the currently described E. chamberlaini, and the molecular data indicates they are different from the

SR populations. However, they could not be told apart based on morphology, so further work needs to be conducted before a recommendation can be made on the SR populations. These putative species display varying degrees of morphological differentiation, particularly in size, limb proportions, and tail proportions. To the best of my knowledge, this represents one of the few well-documented examples of breeding habitat isolation leading to ecological speciation in vertebrates.

Figure 3.1 Map of Eurycea quadridigitata and E. chamberlaini samples used in A) species tree analysis for BPP species delimitation method and B) morphological analyses. Circle sizes represent relative sample size among all populations (larger circles have more individuals). Red circles = cypress wetland populations (CW), orange circles = hillside seepage populations (HS), green circles = steephead/ravine populations (SR), yellow circles = stream-side/E. chamberlaini populations (SS), and blue circles = western populations (WP). Gray line represents the geographic range of the complex.

Figure 3.2 The species tree from the *BEAST analysis of four nuclear and one mitochondrial loci. Posterior probabilities from *BEAST analysis are located under respective nodes. Boxes above nodes represent species probabilities from runs with different parameter settings in the Bayesian Phylogenetics and Phylogeography species delimitation analysis: top = large ancestral population size and deep root age [θ = G(1, 10), τ0 = G(1, 10)], middle = small ancestral population size and shallow root age [θ = G(2, 2000), τ0 = G(2, 2000)], and bottom = large ancestral population size and shallow root age [θ = G(1, 10), τ0 = G(2, 2000)].

Stream side populations 1.0 (E. chamberlaini) 1.0 1.0

0.45

1.0 Steephead/Ravine 1.0 populations 1.0

1.0

1.0 1.0 Cypress Wetland 1.0 populations 0.85

1.0 1.0 1.0 Hillside Seepage populations 1.0

Western populations 0.0030

100

Figure 3.3 Anonymous locus 02 (AL02) gene tree from *BEAST analysis. Only PP > 0.80 are shown.

101

0.97

1.0

1.0 0.89 1.0 1.0 0.96

1.0

0.99 1.0 1.0 1.0 1.0 1.0

0.96 1.0 0.98 0.96 1.0 0.0070

102

Figure 3.4 Anonymous locus 21 (AL21) gene tree from *BEAST analysis. Only PP > 0.80 are shown.

103

0.83

0.82

1.0

0.91

0.95

1.0 0.94

0.94

0.88

1.0 1.0 1.0 0.88

0.99 0.02 1.0 0.84

104

Figure 3.5 Anonymous locus 51 (AL51) gene tree from *BEAST analysis. Only PP > 0.80 are shown.

105

0.90

0.92 0.90 1.0

0.86

0.99 1.0 0.88

0.86

1.0

0.86 1.0 0.01 1.0 1.0

106

Figure 3.6 Recombination activating gene 1 (RAG1) gene tree from *BEAST analysis. Only PP > 0.80 are shown.

107

1.0

0.97

1.0

0.98 1.0

0.97 1.0 1.0 0.87

1.0

0.98

0.88 0.99 0.85 1.0

1.0 0.98 1.0 1.0 1.0 1.0

1.0 1.0

0.93 1.0 1.0 0.96 0.0040

108

Figure 3.7 Mitochondrial (CytB + ND2) gene tree from *BEAST analysis. Only PP > 0.80 are shown.

109

1.0 0.93 1.0 1.0 1.0 1.0 1.0 1.0 1.0

1.0 1.0 1.0 1.0 1.0

1.0

0.94 1.0 1.0 0.96

0.97 1.0 0.98 1.0 1.0 1.0 0.98 0.99 0.80 1.0 1.0 0.96 0.99

1.0

1.0 1.0 1.0 1.0

1.0 1.0

1.0 1.0 1.0

1.0 1.0 1.0 0.99 0.85 1.0

1.0

1.0 0.99 1.0 0.83 1.0 1.0 1.0

1.0 1.0 1.0 1.0 0.99

1.0 1.0 1.0 1.0 1.0 1.0 0.0040

110

Figure 3.8 Plot of PC1 versus PC2 for the 11 morphological traits (n = 199). Red circles = cypress wetland populations (CW), orange circles = hillside seepage populations (HS), green circles = steephead/ravine populations (SR), yellow circles = stream-side/E. chamberlaini populations (SS), and blue circles = western populations (WP).

111

PC1

CW SR HS WP SS

¦ PC2

112

Figure 3.9 Plots of four linear discriminant functions (LD1-LD4) for the 11 morphological traits (n = 199). A = cypress wetland populations (CW), B = steephead/ravine populations (SR), C = hillside seepage populations (HS), D = western populations (WP), E = stream-side/E. chamberlaini populations (SS).

113

ï ï ï ï 012

D D D DD D D D D D D D D D D D D D D D DD D D D D D D D DDD D DD D DD D D E D E D E D A A A AAA A A A AA A A A A A A AA ADA AA A A AD AAA A AAA D AA A BAAABBAA A A A AA ABAAAA BA B A A A AA BAA ABBAAA LD1 D A A AAA BAAA A AAAAAADB A A A DBAA A A A ACACADBADAACABAAAAABEEA A A A A AEDAADAACAAAAAC ABACAABBAE A ACABAACACABDADAEABAAAA E AA A A A AA A A A A 024 C AAAAAAAAAAAAAAAAAAA AAA A AAAAACAAAAAAAAAAA AA AA AA AAAAAAAAAACAAA A AA C BACCA AAAB ABAA AA A AACA BAAAAABACBAA A CAA AA BABBCAAAA A A A BC CE ABAABAE ADAAAAA C ADCAAAAA AA BABEB E A A C BEBAAEACAADAB A AA C C CAAAA AAA A C A ACA AAA A CA AAA A CAA C A C C C CAA A AAA C ACAAACA A A AA A CAAAA C CA CA EEAC AA A A EA ACCE C AE AAECA AAA A A A A A AA A AA A A

C A A A C A CAA ï B B B A B BB A BBA B BB B B B B AA A A

A A A A AA A A D A A D A AD A AA AAAA A A A A AAAAA D AA AA D A AA AA A DA A AA A A A AAA E A A EA A AA AAE AAAA AAAAAAA D A A AAAAAADAA A AAAAAA ADAAA AA A A ABAAA A D BAAA DA A A A A A AAAABDAA A AA AADAAABE AA A AA DAEA A AAAABA E A AAAABAEADAA A E A A AC AA AAABAA A D AAAAAAABA CADA A BA AAAAAAAC A D A AA AAA AAAA A DD AD A A ADA AAA A A A AA D AAA ADAAA A EAA EAABABAABBA A D A DAA BAABAAAE AEBB AA BEAABAAAAEDABB E ABAAACAA AAA AA A AEAAA AA C B A A AAAABACE AA AAA A 012 C A A D A C DA CD A A B AABAAAD AABA D LD2 A ADABAAA BAB D A A AAA A DBDB AB A A B A AA ADAB A A AADA AAA B B A A AABAD AB AA B BC CCEAAAC DA C A C ADCCA B AB E BAB CDECA CAC A CC CACA D D C D ACDCC A D CD AC CA C ACAAA A D A CA A D CA A ACDCA A A A B CB A A A D CADA ABB ACDB A BA BCC D C C B D C B DC C D C D D C

ï ï C C C

A A A D D D A A A A D A D D A A A E AE E AEA A A A E A E C BA B AC B B C A B A BEA B B BEAA A BE AB AB A B ACAA A ABA BCAAAA A B CAA ABA AA B ABAAAAA D B D BA AA A A A BBADA A A BB EA A BA B D D B B B AE DBAA D A BB A BB E A D ACC AAABCAAA AAA D C BA C AAAACAAADAA A CAAACAA AABA AADC AA A ABAACAA AAA E D DD DC AAAAABAD AADAAE DA AAC AABDEADAAAA

A A A 012 CA AAAAAAAAAA D CAAAADAAAA AA A AA A AAD A AAC AAA A EACAAAAAACABABA A ACCAAAEAAB BAAA A A B ACAE AAA AC ABA A B A AAAAAAADAA ADA D A AADAB AAAAAAAA A A A A AA ADAAADAABAA A AAA ABADAA D D AD ADA A A BAADA A DAA ADABD A A AAA CAADA A D D C D ADAAADAAA /' A ADAA ADACDA ACCAACDA AAEAA C CA C A A ADEA AAA A C AA AEADCAC A A C CAAA CCA CA A C C AA A A A AA A A A AA A A A AA A A AA D D A A D A A A C A C A A A C A AA A AA A A AA A C C A A C A A A ï ï

A A A A A A A A A A AA A AA A A A 2 A A A A A A A A A C A A C A A A C A AEA D D A AA DE D A A DD EA A A A A A A A AA ACAA AAA BA C A AB AA A A A A C ABAA AA A A BAA AB A ABABAA AA A AA AA AAB B CC CAAAAAA AA D ACCA A ADAACA ACD AAC AAC A A CAE ADACAAB D D CD CB AE AAD DA DAC A AC DAE BA D B A AAAADAEAAA E D A A DBAAAD AEA E A EADABAAADEAA A A A A ACCADBAAB A D CCCDB AAAABAADAA C ACAAD A D BA A 01 A A A AAEA AEBABAAAA D D DA EADEAB A BAADAA BAADAAEA BAA ADA ED B AAACAABAACA D B C DCB ACAAAA AAA A DAAACA AABBC C LD4 A BEBAAAAAAA D B A AAADABEA AAA A A AA AAAAAD AB EB B CA AABAAA D A DB CA ABAAAA C DAA ABA AA B A A AAABAD D D A AAABD AA A DA AAAD A A BA BB C C CA A D CDB CCBA A CD AACCBB AA A A A A A A A AA AA A AAAA A A AA A A A AA A A C AA C A AA AA C A AA A A A A A A A D D A A A A A DA A A

ï ï A A A A A AA A A A A A

ï 024 ï ï 012

1 14

Table 3.1 List of samples used in molecular analysis of species delimitation in the Eurycea quadridigitata complex.

Sample ID County, State Lat Long CytB ND2 AL02 AL21 AL51 RAG1 Habitat E. quadridigitata KW0003 Leon, FL 30.64 -84.25 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0004 Leon, FL 30.64 -84.25 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0005 Leon, FL 30.64 -84.25 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0006 Leon, FL 30.64 -84.25 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0007 Leon, FL 30.64 -84.25 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0009 Jefferson, FL 30.6 -83.89 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0010 Bradford, FL 29.83 -82.16 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0016 Bradford, FL 29.83 -82.16 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0017 Bradford, FL 29.83 -82.16 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0019 Jefferson, FL 30.58 -83.61 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0020 Leon, FL 30.58 -84.36 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0023 Sarasota, FL 27.21 -82.24 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. chamberlaini KW0024 Anderson, SC 34.64 -82.7 XXXX XXXX XXXX XXXX XXXX XXXX Stream Side E. quadridigitata KW0026 Volusia, FL 29.07 -81.13 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0030 San Augustine, TX 31.33 -94.22 XXXX XXXX XXXX XXXX XXXX XXXX Western Population E. quadridigitata KW0035 Appling, GA 31.92 -82.27 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0041 Wayne, GA 31.49 -81.84 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0051 Clinch, GA 31.03 -82.87 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0053 Washington, FL 30.53 -85.86 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0054 Washington, FL 30.53 -85.86 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0064 Washington, FL 30.45 -85.5 XXXX XXXX XXXX XXXX XXXX XXXX Steephead/Ravine E. quadridigitata KW0066 Long, GA 31.9 -81.77 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0070 Long, GA 31.9 -81.77 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0075 Santa Rosa, FL 30.84 -86.94 XXXX XXXX XXXX XXXX XXXX XXXX Hillside Seepage E. quadridigitata KW0076 Santa Rosa, FL 30.84 -86.94 XXXX XXXX XXXX XXXX XXXX XXXX Hillside Seepage E. quadridigitata KW0078 Thomas, GA 30.83 -84.01 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0079 Thomas, GA 30.83 -84.01 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0083 Liberty, FL 29.91 -84.96 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0087 Mobile, AL 31.04 -88.19 XXXX XXXX XXXX XXXX XXXX XXXX Hillside Seepage E. quadridigitata KW0088 Mobile, AL 31.04 -88.19 XXXX XXXX XXXX XXXX XXXX XXXX Hillside Seepage

115

Table 3.1 Continued.

Sample ID County, State Lat Long CytB ND2 AL02 AL21 AL51 RAG1 Habitat E. quadridigitata KW0089 Meriwether, GA 33.11 -84.52 XXXX XXXX XXXX XXXX XXXX XXXX Steephead/Ravine E. quadridigitata KW0090 Taylor, GA 32.64 -84.37 XXXX XXXX XXXX XXXX XXXX XXXX Steephead/Ravine E. quadridigitata KW0091 Crawford, GA 32.69 -84.01 XXXX XXXX XXXX XXXX XXXX XXXX Steephead/Ravine E. quadridigitata KW0112 Liberty, FL 30.11 -85.02 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0113 Leon, FL 30.35 -84.67 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0114 Leon, FL 30.35 -84.67 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0115 Leon, FL 30.35 -84.67 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0124 Berkeley, SC 33.21 -79.47 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0125 Marion, FL 29.34 -81.73 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0126 Charleston, SC 33.14 -79.54 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. chamberlaini KW0128 Pitt, NC 35.61 -77.36 XXXX XXXX XXXX XXXX XXXX XXXX Stream Side E. quadridigitata KW0132 Thomas, GA 30.76 -84.01 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0134 Baker, FL 30.37 -82.39 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0135 Jasper, SC 32.48 -81.17 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0137 Aiken, SC 33.25 -81.74 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0141 Barnwell, SC 33.26 -81.6 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0143 Barnwell, SC 33.26 -81.64 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0144 Liberty, FL 30.17 -84.68 XXXX XXXX XXXX XXXX XXXX XXXX Cypress Wetlands E. quadridigitata KW0951 Sabine, TX 31.36 -93.72 XXXX XXXX XXXX XXXX XXXX XXXX Western Population E. quadridigitata KW0965 Jones, MS 31.44 -88.97 XXXX XXXX XXXX XXXX XXXX XXXX Western Population

116

Table 3.2 Primers and PCR protocols used in amplification and sequencing.

Anneal Temp Elong Primer Gene Primer Sequence (5'–3') (Time) Time Cycles Reference MVZ15 CytB GAACTAATGGCCCACACWWTACGNAA 50 C (30 s) 90 s x 30 Moritz et al. 1992 MVZ16R CytB AAATAGGAARTATCAYTCTGGTTTRAT 50 C (30 s) 90 s x 30 Moritz et al. 1992 PGludg2 CytB GGTCTGAAAAACCAATGTTGTATTC 50 C (30 s) 90 s x 30 Wiens et al. 2006 L4437 ND2 AAGCTTTCGGGCCCATACC 50 C (35 s) 150 s x 25 Macey et al. 1997 H6159 ND2 GCTATGTCTGGGGCTCCAATTA 50 C (35 s) 150 s x 25 Weisrock et al. 2001 EqND2FI* ND2 GGAGGCCTAAATCAACCACA 50 C (35 s) 150 s x 25 This study EqND2RI* ND2 GTGATGTGGTGTACGCAAGG 50 C (35 s) 150 s x 25 This study EuryceaRag1F RAG1 GGTAYGATGTTGCATTGGTTGCCA 58 C (30 s) 60 s x 30 Timpe et al. 2009 Rag1midElongFb RAG1 TGCACTGTGAYATNGGGAATGCTG 58 C (30 s) 60 s x 30 Timpe et al. 2009 ElongRag1R RAG1 TTGACTGCCATCGCTTCCTCTCTT 58 C (30 s) 60 s x 30 Timpe et al. 2009 Rag1endElongRb RAG1 AACTTGGACTGCCTGGCGTTCATT 58 C (30 s) 60 s x 30 Timpe et al. 2009 AL02F EqAL02 ATGGGTCATCATCGTTATCGATATC 52 C (25 s) 45 s x 30 This study AL02R EqAL02 TGGTACACATTGATCCTAGATCTAG 52 C (25 s) 45 s x 30 This study AL21F EqAL21 TTGATCTATCGATATGCTCTAG 52 C (25 s) 45 s x 30 This study AL21R EqAL21 TATGCTCTCGCACAYATGATC 52 C (25 s) 45 s x 30 This study AL51F EqAL51 TGTCGACACATCAATGGGTGCAC 52 C (25 s) 45 s x 30 This study AL51R EqAL51 CACTAGACTAGCATAAATGCAGT 52 C (25 s) 45 s x 30 This study * Internal primers

117

Table 3.3 List of samples and corresponding measurements for 12 morphological traits in the Eurycea quadridigitata complex.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD KW0001 CW 30.8 55.3 1.9 2.5 7.2 6.7 7.1 3.9 2.2 1.9 1.6 1.7 KW0003 CW 27.5 29.4 1.9 2 7.3 6.1 5.5 3.1 1.8 1.6 1.5 1.3 KW0004 CW 27.3 35 1.9 2.2 7.1 6 5.1 3.1 1.8 1.5 1.5 1.4 KW0005 CW 25.3 34.9 1.9 2.2 7.2 5.8 5.6 3 1.6 1.6 1.5 1.3 KW0006 CW 26 42.6 2 2.3 6.8 6 5.6 3.1 1.7 1.8 1.5 1.4 KW0007 CW 29.7 32.7 2 2.4 7.4 6 5.6 3.6 1.7 2.1 1.7 1.5 KW0008 CW 28.8 21.5 2.1 2.4 6.9 5.9 5.9 3.3 1.8 2 1.6 1.5 KW0009 CW 26 18.3 1.8 2.5 7.1 6 4 3.1 2.3 1.4 1.4 1.2 KW0010 CW 27 15.2 2.1 2.4 7.3 6.3 6.6 3 2.1 1.8 1.6 1.4 KW0011 CW 28.4 48.7 2.2 2.2 6.3 5.6 5.3 3.3 1.7 2 1.5 1.5 KW0012 CW 29 39.1 2.6 2.6 7.2 5.9 5.5 3.2 2 1.7 1.5 1.4 KW0013 CW 27.3 39.8 2.3 2.7 7.3 6.2 5.4 3.7 2 1.7 1.5 1.5 KW0014 CW 28 44.1 2.3 2.5 7.5 5.9 5.6 3.6 2.2 1.8 1.5 1.5 KW0015 WP 28.3 42.8 2.4 2.6 7.1 6.1 5.6 3.3 2 1.6 1.5 1.5 KW0016 WP 26.1 41.5 1.9 2.5 7.4 6.1 5.7 3 1.8 1.7 1.6 1.4 KW0017 WP 26.2 30.4 2 2.3 6.3 5.9 5.4 3.1 1.9 1.7 1.5 1.4 KW0019 WP 30.3 38.9 1.7 2.4 7.8 5.4 5.6 3.5 1.8 1.8 1.5 1.5 KW0020 WP 26.7 24.4 2.6 2.8 7.7 5.4 5.7 3.4 2.8 1.4 1.4 1.4 KW0023 SR 24.6 10.3 1.9 2.4 7.1 5.6 5.6 3.1 1.8 1.7 1.5 1.3 KW0024 SR 23.5 28.8 1.3 2.1 6.6 5.9 5 2.9 1.7 1.7 1.4 1.2 KW0025 CW 23.5 13.3 1.3 2.1 6.5 5.9 5 2.9 1.9 1.7 1.3 1.3 KW0026 CW 24.5 32.5 1.2 2 6.2 5.5 5.4 3 1.6 1.6 1.5 1.4 KW0028 CW 34.6 50.3 2.4 3.7 8.9 6.7 6 3.6 2.1 1.9 1.9 1.5 KW0029 CW 26 29.1 1.8 2.1 7.5 5.5 5.6 3.3 2.1 2 1.5 1.5 KW0030 CW 27.9 33.2 2 2.2 7.6 5.8 5.6 3.4 2.1 2 1.4 1.4 KW0031 CW 27 37.4 1.8 2.5 7.6 5.5 5.7 3.3 2 2 1.6 1.6 KW0032 CW 26.8 34.3 1.9 2.2 7.3 5.3 5.7 3.3 2 2 1.5 1.5 KW0035 SR 30 24.7 1.6 2.4 8.5 6.7 5.3 3.3 2.1 1.9 1.7 1.4 KW0038 SR 27.5 40.8 2.2 2.4 7.5 5.7 5.2 3.3 2.2 1.5 1.5 1.5

118

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD KW0039 SR 29.5 15.1 2.1 2.6 7.2 6.6 6 3.4 2.3 1.9 1.7 1.5 KW0040 SR 22.1 33.7 1.1 2.2 6 5.4 5 2.9 1.8 1.3 1.2 1.2 KW0041 SR 29.3 22.2 2.7 3.3 7.7 6 5.8 3.8 2.4 2.1 1.7 1.6 KW0044 CW 29.5 41 2 2.5 6.7 6.2 5.3 3.6 2.3 1.6 1.5 1.5 KW0045 CW 23.4 34.4 1.8 2.5 6.1 4.7 5.2 2.9 2 1.6 1.6 1.4 KW0046 CW 26.6 28.9 2 2.4 7.1 5.8 5.5 3.2 2.2 1.6 1.7 1.5 KW0047 CW 27.6 30.6 1.8 2.2 7.5 6.3 5.3 3.5 2 1.7 1.4 1.4 KW0048 CW 23.1 20.2 1.8 2.3 7 5.2 5.3 3.1 2 1.4 1.3 1.3 KW0049 SR 27.5 16.1 2.3 2.6 7 5.2 5.3 3.1 2 1.8 1.3 1.5 KW0050 SR 29.6 30.7 2.3 2.4 7.2 6.3 6.3 3.3 2.1 1.7 1.6 1.5 KW0051 SR 27.4 23 1.9 2.4 7.3 5.7 5.7 3.2 1.7 1.6 1.4 1.5 KW0052 HS 29.1 23.4 2 2.6 7 6.2 6.4 3.4 2 1.6 1.5 1.5 KW0053 HS 25.6 36.4 1.6 2.7 7 5.7 5.2 3.1 1.9 1.5 1.5 1.4 KW0054 HS 26 39.4 1.1 2 7.1 6.2 5.2 3 1.6 1.5 1.5 1.2 KW0055 HS 23.4 28.7 1.6 1.9 6.9 5.5 4.7 2.7 1.6 1.5 1.5 1.3 KW0056 CW 24.8 21.6 1.4 2 7 5.5 4.8 3 1.9 1.6 1.6 1.4 KW0057 CW 25.8 n/a 2 2 7 5.8 5.2 3.2 2.2 1.5 1.5 1.4 KW0058 CW 22.3 31.9 1.8 2 6 5 5.2 3.1 1.9 1.4 1.5 1.4 KW0059 CW 21.2 15.5 1.7 2.2 7.1 5.4 5.1 2.9 1.8 1.5 1.5 1.4 KW0061 CW 18 10.5 1 1.7 5.1 4 4.4 2.6 1.5 1.2 1.2 1.2 KW0062 CW 19.1 21.8 1.1 1.8 5.6 4.6 4.4 2.7 1.6 1.2 1.2 1.1 KW0063 CW 18 18.9 1 1.7 5.2 4.4 4.2 2.4 1.4 1.1 1.2 1.1 KW0064 CW 18.8 20.2 1.1 1.7 5.3 4.3 4.3 2.7 1.5 1.2 1.2 1.2 KW0066 CW 29.4 42.3 1.5 2.8 9 6.4 5.2 3.2 1.8 1.5 1.5 1.4 KW0067 CW 28 42.1 1.5 2.2 8.5 7.1 5.4 3.2 1.8 1.5 1.5 1.4 KW0068 CW 28.9 42.2 1.3 2.6 8.9 7.4 5.8 3.4 1.8 1.7 1.5 1.5 KW0069 CW 30.4 n/a 1.9 2.7 9.7 7.8 6 3.5 2.2 1.7 1.7 1.5 KW0070 CW 27.9 35.9 1.5 2.6 8.5 6.7 5.8 3.1 1.9 1.7 1.7 1.5 KW0071 SS 24.8 40.1 1.9 2.6 7.2 6.2 5.3 3.4 2.1 1.8 1.6 1.3

119

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD KW0072 CW 27.2 42.7 1.7 2.4 6.9 6.6 5.2 3 1.7 1.6 1.6 1.4 KW0073 CW 27.5 44.6 1.9 2.3 6.8 6.7 5.5 3.4 2 1.8 1.8 1.4 KW0074 CW 25.4 33.6 2.2 2.4 8.1 6.8 5.4 3.3 2.1 1.7 1.7 1.4 KW0084 CW 24.6 35.9 1.7 2.4 6.8 5.7 5.1 2.9 1.9 1.4 1.4 1.2 KW0085 CW 25.5 27.6 1.9 2.7 7.6 6 5.1 3.3 2 1.6 1.6 1.4 KW0086 CW 21.2 30.3 1.5 2.2 5.7 4.7 4.8 2.7 1.6 1.6 1.5 1.3 KW0087 CW 21.5 29.6 1.4 2.1 5.8 4.7 5.1 2.8 1.5 1.3 1.3 1.2 KW0088 CW 20.5 26.1 1.1 1.9 5.7 5.3 4.6 2.7 1.6 1.3 1.3 1.3 KW0094 CW 26.3 n/a 2.4 2.7 8.4 6.2 5.9 3.3 1.8 1.6 1.6 1.5 KW0095 CW 32.8 40 2.4 2.8 7.3 7.2 5.8 3.5 2.1 1.7 1.7 1.4 KW0096 CW 33.8 24.3 2.4 3.3 8.3 6 6.9 3.5 2.7 1.8 1.8 1.6 KW0097 CW 26 42 2.2 2.1 6.8 6.2 5.4 3.2 2 1.4 1.5 1.4 KW0098 CW 26.5 37.4 2 2.4 6.8 5.6 6 3.1 1.7 1.7 1.6 1.3 KW0099 CW 24.4 26.4 2.2 2.7 6.3 5.4 4.6 2.9 1.6 1.6 1.6 1.2 KW0112 SS 20.8 13.1 1.4 2.1 5.8 5.2 4.3 2.6 1.6 1.2 1.2 1 KW0113 SS 21.4 28.3 1.2 2 6.2 5.4 4 2.7 1.6 1.6 1.3 1.1 KW0117 SS 32.7 33.8 2.4 2.4 8.7 7.9 6.1 3.8 1.8 2 1.6 1.5 KW0118 CW 27.9 n/a 2 2 7.4 6 5.3 3.4 1.9 1.6 1.6 1.3 KW0119 CW 24.6 n/a 2 2.2 7.1 6.1 4.9 3.2 1.9 1.4 1.4 1.3 KW0120 CW 28.3 42.2 1.9 2.1 6.8 5.5 5.6 3.5 1.8 1.6 1.7 1.5 KW0121 WP 23.6 26.8 1.7 1.9 6.6 5.8 4.7 2.6 1.7 1.2 1.3 1.2 KW0124 WP 30 55.14 2.1 2.6 7.1 6.4 6.1 3.6 2.1 1.7 1.6 1.3 KW0125 WP 27.4 40.6 1.7 2.1 7.9 6.3 5.7 3.2 1.9 1.6 1.7 1.4 KW0126 WP 31.7 33.5 1.7 2 9 7.7 6.3 3.4 2 2 1.6 1.4 KW0127 WP 28 46.8 1.7 2 7.3 6.4 5.5 3 1.7 1.8 1.8 1.3 KW0128 WP 22.6 18.3 1.4 1.7 7.2 6.3 4.9 3.1 1.8 1.6 1.6 1.3 KW0129 WP 24.3 n/a 1.8 2.1 7.3 6.3 5.6 3.3 1.8 1.8 1.8 1.3 KW0130 WP 26.11 43.5 2 2.1 7.2 5.9 4.8 3.2 2.1 1.2 1.2 1.2 KW0131 WP 26.9 37.4 1.7 2.2 7.4 6.1 5.5 3.4 1.6 1.7 1.5 1.1

120

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD KW0132 HS 28.8 24.6 2 2.3 8 6.7 5.3 3.2 2 1.6 1.5 1.5 KW0133 HS 26.2 36.8 1.6 1.9 8.1 7 5.2 3.2 1.8 1.8 1.7 1.3 KW0134 HS 28.6 37.9 2 2.8 7.6 5.9 5.9 3.3 2 1.6 1.8 1.3 KW0135 HS 31 36 2.3 2.5 8.4 6.1 5.8 3.6 1.9 1.6 1.6 1.3 KW0136 HS 28.3 43.2 2.3 2.6 8 6.8 5.5 3.3 1.8 1.6 1.6 1.4 KW0137 HS 29 41.7 2.2 2.5 7.5 5.8 5.9 3.2 1.7 1.6 1.7 1.3 KW0138 HS 25.9 32.8 1.9 2.1 6.8 5.6 5.2 2.8 1.7 1.6 1.5 1.5 KW0139 HS 25.9 32.6 1.7 2 6.9 5.9 4.2 3 1.7 1.4 1.4 1.4 KW0140 HS 26.2 35.2 1.7 2.4 6.7 6 5.2 3.2 1.7 1.5 1.5 1.3 KW0141 HS 25 24.8 1.8 2.2 6.2 5.5 5.1 3 1.7 1.6 1.6 1.3 KW0142 HS 30.7 42.1 2.5 2.5 8.4 5.7 5.8 3.8 2 2 1.8 1.3 KW0143 CW 30 44.9 2.4 2.7 7.6 6.3 6.2 3.4 1.9 1.7 1.7 1.6 KW0144 CW 21.3 15.6 1.4 1.9 5.8 4.3 4.3 2.6 1.4 1.2 1.2 1 KW0145 CW 23 35.5 2.2 2.5 6.4 4.3 4.9 2.9 1.7 1.4 1.4 1.1 KW0146 CW 21.3 27.9 1.7 2.3 5.5 5.2 4 2.6 1.5 1.2 1.3 1.3 KW0147 CW 32.3 n/a 1.3 1.6 8.1 6.5 5.8 3.1 1.8 1.8 1.8 1.6 KW0148 CW 29.2 n/a 2.1 2.8 7.1 5.7 5.7 3.2 2 1.7 1.5 1.5 KW0149 CW 23.5 n/a 1.4 1.5 6.4 5.3 4.7 2.8 1.6 1.5 1.5 1.3 UF128860 CW 26.5 34.5 2.2 2.7 6.4 4.5 5 3.2 2 1.6 1.3 1.3 UF128861 CW 29 25.8 2.5 2.7 6.6 5.9 4.6 3.2 2.2 1.6 1.3 1.4 UF128862 CW 26.5 39.5 2.2 2.8 6.1 5.1 4.7 3 2.1 1.6 1.3 1.5 UF128863 CW 29 45.7 2.4 2.9 6.5 5 5.6 3.6 2.1 1.8 1.6 1.5 UF128913 CW 33 46.5 3.2 3.6 7 6.8 6.2 4.1 2.6 1.9 1.7 1.9 UF128914 CW 31.8 52.7 2.9 3.6 7.3 7 6.1 3.9 2.3 2.2 1.9 1.7 UF128915 CW 32.7 46.5 3 3.2 7.3 7 6.2 4 2.8 1.9 2 1.9 UF128916 CW 32.4 45.7 2.7 2.8 8.2 7 6.3 3.9 2.3 2 2 1.6 UF128917 CW 37.1 48.6 3.2 3.4 9.5 7.3 6.4 4.3 2.5 2.3 2 1.9 UF128918 CW 34.5 52.5 3 3.3 8 6.8 6.2 4.2 2.8 2 2 1.8 UF128919 CW 37.2 50.2 3.4 3.8 8.1 6.9 6.6 4.6 3.1 1.9 2.2 1.9

121

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD UF128920 CW 31.8 49 2.9 3.1 7.2 7.2 6 3.8 2.5 2.1 2.2 1.8 UF128921 CW 33.5 50.7 3.4 4 8.2 7.1 6.6 3.9 2.6 1.8 1.9 1.8 UF128980 WP 24 28.9 2.1 2.3 7.5 6.4 5.9 3 1.9 1.4 1.5 1.5 UF128981 WP 24.9 39.5 1.9 2.2 7.1 6.3 5.7 3.2 2.2 1.6 1.4 1.4 UF128982 WP 25.7 41.4 2.1 2.3 7.5 6.7 5.8 3 2.1 2 1.3 1.5 UF128983 WP 23.4 34.3 1.7 2.3 6.2 5.7 6.7 2.8 2 1.5 1.4 1.5 UF128984 CW 24.6 40.3 2.1 2.3 6.8 6 5.7 3.1 1.9 1.7 1.4 1.5 UF128985 CW 23.6 22.7 1.7 2.3 7.2 6.1 5.1 2.9 1.8 1.5 1.5 1.5 UF128986 CW 25.2 37.4 1.6 2.1 7.4 5.8 5.7 2.9 2 1.4 1.4 1.5 UF128987 CW 23.9 33 1.9 2.4 7.3 6.3 5.1 2.9 1.8 1.4 1.2 1.4 UF128988 CW 22 34.8 1.8 1.9 6.8 6 5.9 2.9 1.8 1.5 1.4 1.4 UF128989 CW 25.2 39.7 1.9 2.2 7.1 6.2 5.4 3 1.8 1.5 1.5 1.5 UF128990 CW 22.2 33.2 1.6 1.9 7.5 6.1 5.2 2.9 1.8 1.4 1.3 1.5 UF128991 CW 23.1 33.7 1.9 2.1 6.7 5.8 5.4 3 2.1 1.3 1.4 1.3 UF128992 CW 20.3 29.1 2 2.2 7.1 5.9 4.8 2.9 2 1.6 1.4 1.5 UF128994 CW 28.4 38.1 2.5 2.6 7.6 5.6 5.5 3.4 2.1 1.6 1.5 1.5 UF128995 CW 27.2 43.1 2.4 2.7 7.7 5.9 5.5 3.3 1.9 1.4 1.6 1.4 UF129029 CW 26.1 49.6 1.7 1.9 6.6 5.4 4.9 3.1 1.8 1.4 1.3 1.3 UF129032 CW 25.9 44.4 1.9 2.1 6.8 5.1 5 3 1.7 1.6 1.3 1.4 UF129034 CW 26.7 44.8 2 2.4 7.6 5.6 5.6 3.1 1.5 1.3 1.1 1.3 UF129044 CW 25.1 40.8 1.9 1.9 7.7 6.6 4.3 3.1 1.9 1.1 1.1 1.4 UF129045 CW 29.7 60.8 2.2 2.4 7.9 6.4 5.7 3.2 1.8 1.7 1.6 1.4 UF129046 CW 27.9 45.4 2.2 2.3 8 6.9 5.4 3.2 1.6 1.7 1.3 1.5 UF129048 CW 26.7 35.9 1.5 1.7 5.8 5.1 5.2 3.3 2 1.5 1.3 1.4 UF129056 CW 26 30.1 1.9 1.9 6.5 5.6 4.6 2.1 1.7 1.6 1.3 1.5 UF129062 CW 24.9 39.8 1.7 2.1 7.4 5.5 4.2 2.8 1.8 1.7 1.3 1.4 UF129064 CW 24.3 34.5 1.7 2.1 6.6 5.5 5 2.9 1.6 1.5 1.3 1.2 UF129066 CW 26.7 38.5 1.9 2.3 6.6 6.3 4.9 3.4 1.9 1.5 1.5 1.4 UF129069 CW 24.3 36 1.6 2.1 6.4 5.7 4.3 3 1.7 1.6 1.7 1.4

122

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD UF129070 CW 25.5 39.3 1.8 2.3 8.1 5.7 4.9 3.1 1.7 1.5 1.5 1.5 UF129071 CW 25.4 36.7 2 2.2 6.4 5.6 5.1 3.1 1.7 1.5 1.5 1.4 UF129072 CW 25.3 33.7 1.9 2.1 7.4 6.1 4.5 3.3 1.8 1.6 1.4 1.4 UF129100 CW 27.2 39.8 2.7 3.2 7.3 5.3 5.3 3.5 2 1.7 1.4 1.5 UF129101 CW 23.6 30.9 2.1 2.4 6.7 6 4.8 2.9 2 1.5 1.5 1.4 UF129102 CW 26.8 30.5 3 3.6 6.7 4.6 5.2 3.4 2.2 1.6 1.3 1.8 UF129103 CW 30.5 49.1 3.3 3.7 6.9 5.7 5 3.7 2.3 1.7 1.2 1.6 UF129104 CW 27.9 33.6 2.5 3.1 8.3 6.4 5.9 3.7 2.2 1.9 1.9 1.7 UF129105 CW 27.6 40 2.4 3.1 8.2 6.5 6 4.1 2 2.1 2.2 1.7 UF129106 CW 27.7 37.4 2.5 3.2 7.9 6.2 5.9 3.8 2.5 1.8 1.7 1.8 UF129107 CW 22.3 30 2 2.5 7 5.1 5.5 3.3 2.2 1.7 1.5 1.6 UF129108 CW 29 27.9 2.8 3.3 8.1 6.5 6 3.7 2.3 2 1.7 1.8 UF129109 CW 28.2 17.6 2.6 3.2 7.1 6.7 5.5 3.9 2.3 1.9 1.8 1.9 UF129114 CW 26.6 29 2.6 2.9 6.3 4.9 5.4 3.3 2.1 1.7 1.3 1.7 UF129115 CW 23.3 31.7 2.5 3 5.9 4.3 4.7 3 1.9 1.8 1 1.4 UF129117 CW 26.2 43.7 2.3 2.6 7.3 6.6 5 3.1 1.9 1.7 1.6 1.6 UF129118 CW 22.8 18.8 1.9 2.1 6.4 5.3 4.8 3 1.9 1.6 1.3 1.4 UF129119 CW 29.1 15 2.3 3 7.5 6.4 5.1 3.3 1.9 1.6 1.6 1.5 UF129146 SS 25.9 49.5 2.2 2.3 6.9 6.4 5.5 2.8 1.7 1.4 1.5 1.5 UF129147 SS 25.6 39.9 2.4 2.7 6.7 6 5.1 3 1.7 1.6 1.5 1.5 UF129148 CW 27.2 46 2.3 2.6 7 6.3 5.6 3.3 1.8 1.7 1.5 1.6 UF129149 SR 24.1 40.5 2.1 2.4 7.2 6.4 5.2 3 1.5 1.5 1.6 1.6 UF129150 SR 26 44.3 2.3 2.5 6.9 6.6 5.5 2.9 1.8 1.5 1.4 1.4 UF129151 SR 25 38.9 2.3 2.3 7.5 6.3 5.3 3.1 1.5 1.8 1.5 1.5 UF129197 CW 24.7 39.1 2 2.5 7.3 6.3 5.6 3.1 2 1.6 1.5 1.6 UF129199 CW 24.3 36.7 2.1 2.6 6.9 5.9 5.6 3.2 1.9 1.6 1.5 1.5 UF129200 CW 25.6 35.2 2.1 2.4 6.8 6.4 5 3.2 1.8 1.5 1.5 1.5 UF129201 SR 24 37.1 2 2.7 7.3 6.1 5.2 3.3 1.9 1.8 1.4 1.5 UF129202 SR 27.2 35.1 2.1 2.6 6.9 6.5 6.1 3.3 1.8 1.8 1.5 1.5

123

Table 3.3 Continued.

ID Population SVL TL TW TH HLL FLL HL HW HD CL IO OD UF129203 SR 25.3 32.8 2.1 2.5 7.1 6.6 5.4 3.2 1.8 1.5 1.4 1.5 UF129210 SR 27.8 32.6 2.5 3 7 6.6 5.2 3.4 2.3 1.9 1.5 1.6 UF129334 SR 23.5 29.4 2.2 2.3 6.2 5.6 5.3 3 1.9 1.7 1.7 1.4 UF129335 HS 24.2 36 2 2.4 7 5.8 5.4 2.9 2 1.6 1.5 1.5 UF129337 CW 26.8 42.6 2.2 2.6 6.7 5.7 5.7 3 1.9 1.9 1.8 1.5 UF129400 CW 26.7 43.9 2.5 3.1 6.8 5.8 4.9 2.9 2.1 1.5 1.5 1.3 UF24424 CW 26.6 41 2.3 2.6 7.3 6.3 5.2 3.1 1.8 1.5 1.2 1.3 UF25978 SS 27.2 45.6 2.1 2.6 6.7 5.5 4.4 3.3 1.9 1.6 1.3 1.5 UF26016 CW 29.4 50 2.6 3 7.7 6.3 5.5 5.5 2 1.7 1.6 1.6 UF26031 CW 25.3 40 2.3 3.1 7.3 6.4 4.8 3.2 1.9 1.6 1.5 1.6 UF26048 CW 29 40.4 2.4 2.8 6.9 5.4 5.2 3.3 1.9 1.7 1.3 1.5 UF68952 CW 26.9 48 1.9 2.5 6.6 6.4 5.3 3.2 1.8 1.8 1.4 1.4 UF68953 CW 23.8 37.3 1.8 2.3 6.4 5.5 4.9 3.1 1.6 1.5 1.3 1.3 UF68955 HS 25.4 37 1.8 2.3 6.5 5.4 4.7 3 1.6 1.6 1.3 1.3 UF77525 HS 24.8 38.1 1.7 2.2 6.3 5.9 5 3.1 1.9 1.6 1.5 1.4 UF77526 CW 23.7 27.7 2 2.5 6.2 5.3 4.8 3.1 2.1 1.5 1.3 1.3 UF77527 WP 26.1 36.2 2.9 3.1 6 5.7 5.1 3.4 2.6 1.6 1.4 1.2 UF77528 WP 24.9 36.3 1.9 2.2 6.2 5.9 5.2 3.2 2.1 1.6 1.4 1.4 UF77529 CW 25 25.8 1.6 1.6 7.4 5.9 5.2 3.1 1.8 1.6 1.4 1.5 UF77530 CW 26.1 37.1 2.2 2.1 8.2 5.8 5.3 3.2 1.8 1.7 1.6 1.6 UF77532 CW 23.5 27.7 1.5 1.8 6.9 5.4 5 2.8 1.7 1.4 1.2 1.5 UF77533 CW 23.3 34.3 1.9 2.4 6.1 5.4 4.7 3.3 2.2 1.2 1.2 1.4 UF77578 CW 30.9 57 2.1 2.5 7.7 5.8 5.9 3.5 2 1.9 1.4 1.4 UF97125 CW 29.1 32.6 2.4 2.5 7.5 6.4 5.2 3.2 1.9 1.6 1.5 1.4 UF97126 CW 28 29.8 2.5 2.9 6.5 6.1 5.1 3.3 2 1.7 1.6 1.4

124

Table 3.4 Population sample means and standard deviations from 12 morphological traits of the Eurycea quadridigitata complex.

Pop. CL FLL HD HL HLL HW IO OD SVL TH TL TW 1.62 5.93 1.89 5.30 7.12 3.21 1.48 1.41 26.73 2.45 35.84 2.05 CW (0.18) (0.63) (0.23) (0.54) (0.74) (0.33) (0.16) (0.13) (2.44) (0.36) (9.12) (0.36) 1.51 5.59 1.88 5.04 6.64 3.04 1.48 1.34 23.87 2.14 28.19 1.64 SR (0.23) (0.87) (0.28) (0.48) (0.88) (0.32) (0.20) (0.13) (3.67) (0.30) (10.43) (0.42) 1.50 5.87 1.88 5.39 6.88 2.94 1.40 1.41 23.41 2.21 33.19 1.77 HS (0.17) (0.53) (0.19) (0.51) (0.63) (0.15) (0.10) (0.11) (1.75) (0.20) (5.30) (0.27) 1.97 6.47 2.37 6.00 7.82 3.81 1.83 1.72 30.45 3.10 40.66 2.62 WP (0.14) (0.67) (0.30) (0.33) (0.65) (0.36) (0.25) (0.16) (3.97) (0.55) (10.01) (0.52) 1.69 6.00 1.79 5.23 7.03 3.00 1.56 1.36 26.09 1.97 26.80 1.50 SS (0.11) (0.43) (0.13) (0.43) (0.59) (0.14) (0.19) (0.14) (3.75) (0.44) (14.81) (0.30)

Table 3.5 Principal component loadings for 11 morphological traits of the Eurycea quadridigitata complex.

Variable PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10 PC11 Canthus Length (CL) -0.30 0.10 0.56 0.30 -0.32 0.16 0.29 0.49 -0.06 -0.21 -0.03 Front Limb Length (FLL) -0.27 0.45 -0.40 -0.17 -0.16 -0.29 -0.11 0.33 0.50 -0.22 0.07 Head Depth (HD) -0.29 -0.33 -0.14 -0.35 0.50 -0.21 0.34 0.43 -0.24 0.02 -0.01 Head Length (HL) -0.30 0.20 0.28 -0.34 0.42 0.56 -0.38 -0.08 0.12 -0.17 -0.02 Hind Limb Length (HLL) -0.27 0.44 -0.42 0.08 -0.10 0.19 0.10 -0.15 -0.65 -0.09 -0.18 Head Width (HW) -0.33 -0.10 -0.04 0.26 0.20 -0.04 0.49 -0.54 0.35 -0.31 -0.07 Interocular Distance (IO) -0.29 0.22 0.41 0.07 0.15 -0.67 -0.31 -0.22 -0.20 0.14 -0.09 Ocular Diameter (OD) -0.31 -0.15 0.12 -0.62 -0.56 0.01 0.18 -0.29 -0.01 0.20 0.13 Snout-vent Length (SVL) -0.34 0.09 -0.12 0.32 0.12 0.20 0.04 0.06 0.16 0.76 0.31 Tail Height (TH) -0.30 -0.42 -0.16 0.23 -0.12 -0.02 -0.40 -0.02 -0.19 -0.35 0.57 Tail Width (TW) -0.30 -0.41 -0.17 0.14 -0.18 0.06 -0.32 0.08 0.13 0.12 -0.71 Component variance 6.58 1.33 0.63 0.43 0.43 0.39 0.34 0.27 0.26 0.19 0.15 Proportion of variance 0.60 0.12 0.06 0.04 0.04 0.04 0.03 0.02 0.02 0.02 0.01

125

Table 3.6 Results of one-way Type III SS ANOVAs and Kruskal-Wallis tests for 11 morphological traits of the Eurycea quadridigitata complex.

Response SS Residuals df Residuals SS (Type 3) df F value Pr(>F) Hind Limb Length (HLL) 104.40 194 14.90 4 6.94 3.06 x 10-5 Fore Limb Length (FLL) 81.07 194 7.83 4 4.69 1.24 x 10-3 Head Length (HL) 51.80 194 10.66 4 9.98 2.30 x 10-5 *ANOVA results

Response chi-squared df p-value Snout-vent Length (SVL) 49.28 4 5.09 x 10-10 Tail Width (TW) 49.48 4 4.63 x 10-10 Tail Height (TH) 46.47 4 1.97 x 10-9 Head Width (HW) 60.16 4 2.69 x 10-12 Head Depth (HD) 39.36 4 5.88 x 10-8 Canthus Length (CL) 55.72 4 2.30 x 10-11 Interocular Distance (IO) 36.95 4 1.85 x 10-7 Ocular Diameter (OD) 46.59 4 1.86 x 10-9 *Kruskal-Wallis results

126

Table 3.7 Results of pairwise t-tests and Wilcoxon Rank Sum tests for 11 morphological traits of the Eurycea quadridigitata complex. Only significant pairwise comparisons are shown for brevity.

Populations CW (n =136) SR HS WP SVL (0.033)° SR (n = 18) TH (0.01)° TW (0.005)°

CL (0.017)° HW (9.6 x 10-5)° HS (n = 18) SVL (3.0 x 10-6)° TH (0.027)° TW (0.015)° CL (1.3 x 10-9)° CL (6.8 x 10-6)° CL (9.0 x 10-6)° FLL (0.027)* FLL (0.016)* FLL (0.045)* HD (1.8 x 10-8)° HD (6.1 x 10-4)° HD (3.6 x 10-5)° HL (5.6 x 10-7)* HL (3.9 x 10-7)* HL (0.0039)* HLL (0.001)* HLL (1.7 x 10-5)* HLL (0.0012)* WP (n = 20) HW (2.1 x 10-8)° HW (4.1 x 10-5)° HW (1.8 x 10-6)° IO (6.1 x 10-7)° IO (0.0022)° IO (1.9 x 10-5)° OD (6.0 x 10-9)° OD (6.7 x 10-6)° OD (3.8 x 10-5)° SVL (4.1 x 10-4)° SVL (2.1 x 10-4)° SVL (9.7 x 10-6)° TH (3.8 x 10-5)° TH (1.3 x 10-4)° TH (2.4 x 10-4)° TW (1.2 x 10-4)° TW (1.2 x 10-4)° TW (1.6 x 10-4)° TW (8.4 x 10-3)° CL (0.003)° HD (0.002)° HL (0.0082)* SS (n =7) HW (0.001)° OD (0.006)° TH (0.005)° TW (0.003)° Note: Only significant pairwise comparisons are shown * Denotes pairwise t-test ° Denotes Wilcoxon Rank Sum test

127

Table 3.8 Group means of discriminant functions and discriminant function loadings for the Eurycea quadridigitata complex.

Variable CWmean SRmean HSmean WPmean SSmean LD1 LD2 LD3 LD4 Head Length (HL) 5.30 5.04 5.39 6.00 5.23 -0.44 -1.48 -1.37 1.80 Interocular Distance (IO) 1.48 1.48 1.40 1.83 1.56 3.20 -1.44 6.41 1.14 Ocular Diameter (OD) 1.41 1.34 1.41 1.72 1.36 4.99 -3.37 -2.72 -5.74 Snout-vent Length (SVL) 26.73 23.87 23.41 30.45 26.09 0.07 0.51 -0.09 0.02

128

REFERENCES

Adams, D. C., Berns, C. M., Kozak, K. H., & Wiens, J. J. (2009). Are rates of species diversification correlated with rates of morphological evolution? Proceedings of the Royal Society B: Biological Sciences, 276(1668), 2729–2738.

Akaike, H. (1974). A new look at statistical model identification. IEEE Transactions on Automatic Control, 19, 716–723.

Avise, J.C., & Wollenberg, K. (1997). Phylogenetics and the origin of species. Proceedings of the National Academy of Sciences of the United States of America, 94, 7748–7755.

Avise, John C. (1996). Toward a regional conservation genetics perspective: phylogeography of faunas of the southeastern United States. In John C. Avise & J. L. Hamrick (Eds.), Conservation Genetics: Case Histories from Nature (pp. 431–470). New York, NY: Chapman & Hall.

Avise, John C. (2000). Phylogeography: The History and Formation of Species (p. 464). Cambridge, MA: Harvard University Press.

Avise, John C., Arnold, J., Ball, R. M., Bermingham, E., Lamb, T., Neigel, J. E., Reeb, C. A., et al. (1987). Intraspecific phylogeography: The mitochondrial DNA bridge between population genetics and systematics. Annual Review of Ecology and Systematics, 18, 489– 522.

Avise, John C., & Ball, R. M. (1990). Principles of genealogical concordance in species concepts and biological taxonomy. Oxford Surveys in Evolutionary Biology, 7, 45–67.

Barraclough, T. G., & Nee, S. (2001). Phylogenetics and speciation. Trends in Ecology & Evolution, 16(7), 391–399.

Bauer, A. M., Parham, J. F., Brown, R. M., Stuart, B. L., Grismer, L., Papenfuss, T. J., Böhme, W., et al. (2011). Availability of new Bayesian-delimited gecko names and the importance of character-based species descriptions. Proceedings of the Royal Society B: Biological Sciences, 278(1705), 490–2; discussion 493–5.

Baum, D. A., & Shaw, K. L. (1995). Genealogical perspectives on the species problem. In P. C. Hoch & A. G. Stephenson (Eds.), Experimental and Molecular Approaches to Plant Biosystematics (pp. 289–303). St. Louis, MO: Missouri Botanical Garden.

Blankers, T., Adams, D. C., & Wiens, J. J. (2012). Ecological radiation with limited morphological diversification in salamanders. Journal of Evolutionary Biology, 25(4), 634– 646.

129

Bonett, R. M., & Chippindale, P. T. (2004). Speciation, phylogeography and evolution of life history and morphology in plethodontid salamanders of the Eurycea multiplicata complex. Molecular Ecology, 13(5), 1189–1203.

Brimley, C. S. (1923). The dwarf salamander at Raleigh, N.C. Copeia, 120, 81–83.

Brito, P. H., & Edwards, S. V. (2009). Multilocus phylogeography and phylogenetics using sequence-based markers. Genetica, 135(3), 439–455.

Bruce, R. C. (2011). Community assembly in the salamander genus Desmognathus. Herpetological Monographs, 25(1), 1–24.

Buckley, T. R. (2002). Model misspecification and probabilistic tests of topology: evidence from empirical data sets. Systematic Biology, 51, 509–523.

Burbrink, F. T., Lawson, R., & Slowinski, J. B. (2000). Mitochondrial DNA phylogeography of the polytypic North American rat snake (Elaphe obsoleta): a critique of the subspecies concept. Evolution, 54(6), 2107–2118.

Camargo, A., Morando, M., Avila, L. J., & Sites, J. W. (2012). Species delimitation with ABC and other coalescent-based methods: a test of accuracy with simulations and an empirical example with lizards of the Liolaemus darwinii complex (Squamata: Liolaemidae). Evolution 66(9): 2834-2849.

Camp, C. D., Peterman, W. E., Milanovich, J. R., Lamb, T., Maerz, J. C., & Wake, D. B. (2009). A new genus and species of lungless salamander (family Plethodontidae) from the Appalachian highlands of the southeastern United States. Journal of Zoology, 279(1), 86– 94.

Carr, A. F. (1940). A contribution to the herpetology of Florida. University of Florida Biological Sciences Series, 3(1), 1–118.

Chan, K. M. A., & Moore, B. R. (2005). SymmeTREE: whole-tree analysis of differential diversification rates. Bioinformatics, 21(8), 1709–10.

Chippindale, P. T., Bonett, R. M., Baldwin, A. S., & Wiens, J. J. (2004). Phylogenetic evidence for a major reversal of life-history evolution in plethodontid salamanders. Evolution, 58(12), 2809–2822.

Chippindale, P. T., Price, A. H., Wiens, J. J., & Hillis, D. M. (2000). Phylogenetic relationships and systematic revision of central Texas hemidactyliine plethodontid salamanders. Herpetological Monographs, 14, 1–80.

Clark, J. M. (1985). Fossil plethodontid salamanders from the latest Miocene of California. Journal of Herpetology, 19, 41–47.

130

Conant, R., & Collins, J. T. (1998). A Field Guide to Reptiles and Amphibians: Eastern and Central North America (3rd Ed., p. 616). Boston, MA: Hougton Mifflin Company.

Coyne, J. A., & Orr, H. A. (2004). Speciation (p. 545). Sutherland, MA: Sinauer Associates Inc.

Craig, T. P., Itami, J. K., Abrahamson, W. G., & Horner, J. D. (1993). Behavioral evidence for host-race formation in Eurosta solidaginis. Evolution, 47, 1696–1710.

Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jModelTest 2: more models, new heuristics and parallel computing. Nature Methods, 9(8), 772. de Queiroz, K. (1998). The general lineage concept of species, species criteria, and the process of speciation: A conceptual unification and terminological recommendations. In D. Howard & S. Berlocher (Eds.), Endless forms: Species and speciation (pp. 57–75). New York, NY: Oxford University Press.

Dieckmann, U., Doebeli, M., Metz, J. A. J., & Tautz, D. (Eds.). (2004). Adaptive Speciation (p. 460). Cambridge, UK: Cambridge University Press.

Dobzhansky, T. (1946). Complete reproductive isolation between two morphologically similar species of Drosophila. Ecology, 27, 205–211.

Dobzhansky, T. (1948). Darwin’s finches and evolution. Ecology, 29, 219–220.

Drummond, A. J., Ashton, B., Buxton, S., Cheung, M., Cooper, A., Duran, C., Field, M., et al. (2012). Geneious. Retrieved from http://www.geneious.com

Drummond, Alexei J, Ho, S. Y. W., Phillips, M. J., & Rambaut, A. (2006). Relaxed phylogenetics and dating with confidence. PLoS Biology, 4(5), e88.

Drummond, Alexei J, & Rambaut, A. (2007). BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology, 7, 214.

Dubois, A. (2005). Amphibia Mundi. 1.1. An ergotaxonomy of recent amphibians. Alytes, 23, 1– 24.

Duellman, W. E., & Trueb, L. (1986). Biology of Amphibians (p. 670). Baltimore, MD: John Hopkins University Press.

Edwards, A. W. F. (1992). Likelihood (p. 296). Baltimore, MD: John Hopkins University Press.

Elewa, A. M. T. (Ed.). (2004). Morphometrics: Applications in Biology and Paleontology (p. 263). Berlin, DE: Springer Berlin Heidelberg.

Feder, J. L., Opp, S. B., Wlazlo, B., Reynolds, K., Go, W., & Spsisak, S. (1994). Host fidelity is an effective premating barrier between sympatric races of the apple maggot fly.

131

Proceedings of the National Academy of Sciences of the United States of America, 91, 7990–7994. Fenneman, N. M. (1916). Physiographic subdivision of the United States. Proceedings of the National Academy of Sciences of the United States of America, 17–22.

François, O., & Durand, E. (2010). Spatially explicit Bayesian clustering models in population genetics. Molecular Ecology Resources, 10(5), 773–784.

Frost, D. R. (2013). Amphibian species of the world: an online reference. Version 5.6. New York, NY: American Museum of Natural History.

Fujita, M. K., & Leache, A. D. (2010). A coalescent perspective on delimiting and naming species: a reply to Bauer et al. Proceedings of the Royal Society B: Biological Sciences, 278(1705), 493–495.

Funk, D. J. (1998). Isolating a role for natural selection in speciation: host adaptation and sexual isolation in Neochlamisus bebbianae leaf beetles. Evolution, 52, 1744–1759.

Gavrilets, S., & Losos, J. B. (2009). Adaptive radiation: contrasting theory with data. Science, 732–737.

Gernhard, T. (2008). The conditioned reconstructed process. Journal of Theoretical Biology, 253(4), 769–778.

Gittenberger, E. (1991). What about non-adaptive radiation? Biological Journal of the Linnean Society, 43(4), 263–272.

Givnish, T. J. (1997). Adaptive radiations and molecular systematics: issues and approaches. In T. J. Givnish & K. J. Sytsma (Eds.), Molecular Evolution and Adaptive Radiation (pp. 1– 54). Cambridge, UK: Cambridge University Press.

Glor, R. E. (2010). Phylogenetic insights on adaptive radiation. Annual Review of Ecology Evolution and Systematics, 41(1), 251–270.

Guindon, S., & Gascuel, O. (2003). A simple, fast and accurate method to estimate large phylogenies by maximum-likelihood. Systematic Biology, 52, 696–704.

Harrison, J. R., & Guttman, S. I. (2003). A new species of Eurycea (Caudata: Plethodontidae) from North and South Carolina. Southeastern Naturalist, 2(2), 159–178.

Heino, J., Muotka, T., Paavola, R., Hamalainen, H., & Koskenniemi, E. (2002). Correspondence between regional delineations and spatial patterns in macroinvertebrate assemblages of boreal headwater streams. Journal of the North American Benthological Society, 21, 397– 413.

132

Heled, J., & Drummond, A. J. (2010). Bayesian inference of species trees from multilocus data. Molecular Biology and Evolution, 27(3), 570–580.

Ho, S. Y. W. (2007). Calibrating molecular estimates of substitution rates and divergence times in birds. Journal of Avian Biology, 38(4), 409–414.

Hoelzer, G. A. (1997). Inferring phylogenies from mtDNA variation: Mitochondrial-gene trees versus nuclear-gene trees revisited. Evolution, 51(2), 622–626.

Holman, J. A. (1977). The Pleistocene (Kansan) herpetofauna of Cumberland Cave, Maryland. Annals of the Carnegie Museum, 46, 157–172.

Holman, J. A. (1995). Pleistocene Amphibians and Reptiles in North America (p. 243). New York, NY: Oxford University Press.

Holman, J. A. (2006). Fossil Salamanders of North America (p. 232). Indianapolis, IN: Indiana University Press.

Holman, J. A., & Grady, F. (1987). Herpetofauna of New Trout Cave. National Geographic Research, 3(3), 305–317.

Huelsenbeck, J. P., & Ronquist, F. (2001). MRBAYES: Bayesian inference of phylogeny. Bioinformatics, 17, 754–755.

Huelsenbeck, John P., Andolfatto, P., & Huelsenbeck, E. T. (2011). Stucturama: Bayesian inference of population structure. Evolutionary Bioinformatics, 7, 55–59.

Johnson, R. K. (2000). Spatial congruence between ecoregions and littoral macroinvertebrate assemblages. Journal of the North American Benthological Society, 19(3), 475–486.

Jolicoeur, P. (1963). The multivariate generalization of the allometry equation. Biometrics, 19(3), 497–499.

Katakura, H., Shioi, M., & Kira, Y. (1989). Reproductive isolation by host specificity in a pair of phytophagous ladybird beetles. Evolution, 43, 1045–1053.

Kingman, J. F. C. (1982a). On the genealogy of large populations. Journal of Applied Probability A, 19, 27–43.

Kingman, J. F. C. (1982b). The coalescent. Stochastic Processes and their Applications, 13(3), 235–248.

Knowles, L Lacey, & Carstens, B. C. (2007). Delimiting species without monophyletic gene trees. Systematic Biology, 56(6), 887–95.

133

Knowles, L. L. (2008). Why does a method that fails continue to be used? Evolution, 62(11), 2713–2717.

Knowles, Lacey L., & Madison, W. P. (2002). Statistical phylogeography. Molecular Ecology, 11, 2623–2635.

Kozak, K. H., Blaine, R. A., & Larson, A. (2006). Gene lineages and eastern North American palaeodrainage basins: phylogeography and speciation in salamanders of the Eurycea bislineata species complex. Molecular Ecology, 15(1), 191–207.

Kozak, K. H., Larson, A., Bonett, R. M., & Harmon, L. J. (2005). Phylogenetic analysis of ecomorphological divergence, community structure, and diversification rates in dusky salamanders (Plethodontidae: Desmognathus). Evolution, 59(9), 2000–2016.

Kozak, K. H., Mendyk, R. W., & Wiens, J. J. (2009). Can parallel diversification occur in sympatry? Repeated patterns of body-size evolution in coexisting clades of North American salamanders. Evolution, 63(7), 1769–1784.

Kozak, K. H., Weisrock, D. W., & Larson, A. (2006). Rapid lineage accumulation in a nonadaptive radiation: phylogenetic analysis of diversification rates in eastern North American woodland salamanders (Plethodontidae: Plethodon). Proceedings of the Royal Society B Biological Sciences, 273(1586), 539–546.

Kozak, K. H., & Wiens, J. J. (2006). Does niche conservatism promote speciation? A case study in North American salamanders. Evolution, 60(12), 2604–21.

Kozak, K. H., & Wiens, J. J. (2010). Niche conservatism drives elevational diversity patterns in Appalachian salamanders. The American Naturalist, 176(1), 40–54.

Lamb, T., & Beamer, D. A. (2012). Digits lost or gained? Evidence for pedal evolution in the Dwarf Salamander complex (Eurycea, Plethodontidae). PloS One, 7(5), e37544.

Leaché, A. D., & Fujita, M. K. (2010). Bayesian species delimitation in West African forest geckos (Hemidactylus fasciatus). Proceedings of the Royal Society B: Biological Sciences, 277(1697), 3071–3077.

Librado, P., & Rozas, J. (2009). DnaSP v5: A software for comprehensive analysis of DNA polymorphism data. Bioinformatics, 25(11), 1451–1452.

Linn, J. C., Feder, J. L., Nojima, S., Dambroski, H. R., Berlocher, S. H., & Roelofs, W. (2003). Fruit odor discrimination and sympatric host race formation in Rhagoletis. Proceedings of the National Academy of Sciences of the United States of America, 100, 11490–11493. Losos, J. B. (2010). Adaptive radiation, ecological opportunity, and evolutionary determinism. American Naturalist, 175, 623–639.

134

Losos, Jonathan B, & Mahler, D. L. (2010). Adaptive Radiation: The Interaction of Ecological Opportunity, Adaptation, and Speciation. In M. A. Bell, D. J. Futuyma, W. F. Eanes, & J. S. Levinton (Eds.), Evolution Since Darwin: The First 150 Years (pp. 381–420). Sunderland, Massachusetts: Sinauer Associates.

Lynch, J. D. (1978). The distribution of leopard frogs (Rana blairi and Rana pipiens) in Nebraska. Journal of Herpetology, 12, 157–162.

MacCallum, C. J., Nurnberger, B., Barton, N. H., & Szymura, J. M. (1998). Habitat preference in the Bombina hybrid zone in Croatia. Evolution, 52, 227–239.

Macey, J. R., Larson, A., Ananjeva, N. B., & Papenfuss, T. J. (1997). Evolutionary shifts in three major structural features of the mitochondrial genome among iguanian lizards. Journal of Molecular Evolution, 44(6), 660–674.

Maddison, W. P. (1997). Gene trees in species trees. Systematic Biology, 46(3), 523–536.

Marchant, R., Wells, F., & Newall, P. (2000). Assessment of an ecoregion approach for classifying macroinvertebrate assemblages from streams in Victoria, Australia. Journal of the North American Benthological Society, 19, 497–500.

Mayr, E. (1942). Systematics and the Origin of Species, from the Viewpoint of a Zoologist (p. 334). Cambridge, MA: Harvard University Press.

McDonald, R., McKnight, M., Weiss, D., Selig, E., O’Connor, M., Violin, C., & Moody, A. (2005). Species compositional similarity and ecoregions: Do ecoregion boundaries represent zones of high species turnover? Biological Conservation, 126(1), 24–40.

Means, D. B. (1974). The status of Desmognathus brimleyorum Stejneger and an analysis of the genus Desmognathus (Amphibia: Urodela) in Florida. Bulletin of the Florida State Museum, 18, 1–100.

Means, D. B. (1977). Aspects of the significance to terrestrial vertebrates of the Apalachicola River drainage basin. Florida Marine Research Publications, 26, 37–67.

Means, D. B. (2000). Southeastern U.S. Coastal Plain habitats of the Plethodontidae: The importance of relief, ravines, and seepage. In R. C. Bruce, R. J. Jaeger, & L. D. Houck (Eds.), The Biology of Plethodontid Salamanders (pp. 287–302). New York, NY: Plenum Publishing Corporation.

Means, D. B., & Karlin, A. A. (1989). A new species of Desmognathus from the Eastern Gulf Coastal Plain. Herpetologica, 45(1), 37–46.

Miller, K. G., Kominz, M. A., Browning, J. V, Wright, J. D., Mountain, G. S., Katz, M. E., Sugarman, P. J., et al. (2005). The Phanerozoic record of global sea-level change. Science, 310, 1293–1298.

135

Miller, M. D. J. (1992). Analysis of fossil salamanders from the Cheek Bend Cave, Maury County, Tennessee. Appalachian State University, Boone, NC.

Mittleman, M. B. (1947). American Caudata . 1 . Geographic variation in Manculus quadridigitatus. Herpetologica, 3(6), 209–224.

Moore, W. S. (1995). Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nuclear-gene trees. Evolution, 49(4), 718–726.

Morell, V. (1999). Ecology returns to speciation studies. Science, 284(5423), 2106–2108.

Moritz, C., Schneider, C. J., & Wake, D. B. (1992). Evolutionary relationships within the Ensatina eschscholtzii complex confirm the ring species interpretation. Systematic Biology, 41, 273–291.

Mosimann, J. E., & James, F. C. (1979). New Statistical Methods for Allometry with Application to Florida Red-Winged Blackbirds, 33(1), 444–459.

Myers, R. L., & Ewel, J. J. (Eds.). (1990). Ecosystems of Florida (p. 765). Orlando, FL: University of Central Florida Press.

Niemiller, M. L., Fitzpatrick, B. M., & Miller, B. T. (2008). Recent divergence with gene flow in Tennessee cave salamanders (Plethodontidae: Gyrinophilus) inferred from gene genealogies. Molecular Ecology, 17(9), 2258–2275.

Noble, D. W. A., Qi, Y., & Fu, J. (2010). Species delineation using Bayesian model-based assignment tests: a case study using Chinese toad-headed agamas (genus Phrynocephalus). BMC Evolutionary Biology, 10, 197–211.

Noonan, B. P., & Yoder, A. D. (2009). Anonymous nuclear markers for Malagasy plated lizards (Zonosaurus). Molecular Ecology Resources, 9(1), 402–404.

Nylander, J. A. A., Wilgenbusch, J. C., Warren, D. L., & Swofford, D. L. (2008). AWTY (Are We There Yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics, 24(4), 581–583.

Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood, E. C., D’Amico, J. A., et al. (2001). Terrestrial ecoregions of the world: a new map of life on Earth. Bioscience, 51(11), 933–938.

Omernik, J. M. (1987). Ecoregions of the conterminous United States. Annals of the Association of American Geographers, 77, 118–125.

136

Omernik, J. M., & Griffith, G. E. (1991). Ecological regions versus hydrologic units: Frameworks for managing water quality. Journal of Soil and Water Conservation, 46, 334– 340.

Pamilo, P., & Nei, M. (1988). Relationships between gene trees and species trees. Molecular Biology and Evolution, 5, 568–583.

Paradis, E., Claude, J., & Strimmer, K. (2004). APE: analyses of phylogenetics and evolution in R language. Bioinformatics, 20, 289–290.

Pauly, G. B., Piskurek, O., & Shaffer, H. B. (2007). Phylogeographic concordance in the southeastern United States: the flatwoods salamander, Ambystoma cingulatum, as a test case. Molecular Ecology, 16(2), 415–429.

Petranka, J. W. (1998). Salamanders of the United States and Canada (p. 576). Washington, D.C.: Smithsonian Institution Press.

Pinzón, J. H., & LaJeunesse, T. C. (2011). Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Molecular Ecology, 20(2), 311–325.

Posada, D. (2004). Collapse ver. 1.2 A tool for collapsing sequences to haplotypes. Retrieved from http://darwin.uvigo.es

Posada, D., & Crandall, K. A. (1998). Modeltest: testing the model of DNA substitution. Bioinformatics, 14(9), 817–818.

Pybus, O. G., & Harvey, P. H. (2000). Testing macro-evolutionary models using incomplete molecular phylogenies. Proceedings of the Royal Society B: Biological Sciences, 267, 2267–2272.

Rambaut, A., & Drummond, A. J. (2007). Tracer v1.4. Retrieved from http://beast.bio.ed.ac.uk/Tracer

Rannala, B., & Yang, Z. (2003). Bayes estimation of species divergence times and ancestral population sizes using DNA sequences from multiple loci. Genetics, 164(4), 1645–1656.

Rannala, B., & Yang, Z. (2013). Improved reversible jump algorithms for Bayesian species delimitation. Genetics, 194(1), 245–253.

R Core Team. (2013). R: A language and environment for statistical computing. Vienna, AT: R Foundation for Statistical Computing. Retrieved from ISBN 3-900051-07-0, URL http://www.r-project.org/

137

Ree, R. H., Moore, B. R., Webb, C. O., & Donoghue, M. J. (2005). A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution, 59(11), 2299– 2311.

Ree, R. H., & Smith, S. A. (2008). Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Systematic Biology, 57(1), 4–14.

Ronquist, F., & Huelsenbeck, J. P. (2003). MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19(1572-1574).

Rundell, R. J., & Price, T. D. (2009). Adaptive radiation, nonadaptive radiation, ecological speciation and nonecological speciation. Trends in Ecology & Evolution, 24(7), 394–399.

Rundle, H. D., & Nosil, P. (2005). Ecological speciation. Ecology Letters, 8(3), 336–352.

Sambrook, J., & Russell, D. W. (2001). Molecular Cloning (3rd ed., p. 2344). Spring Harbour, NY: Cold Spring Harbour Laboratory Press.

Schemske, D. W. (2000). Understanding the origin of species. Evolution, 54(3), 1069–1073.

Schluter, D. (2000). The Ecology of Adaptive Radiation (p. 288). New York, NY: Oxford University Press.

Schluter, Dolph. (2001). Ecology and the origin of species. Trends in Ecology & Evolution, 16(7), 372–380.

Seaber, P. R., Kapinos, F. P., & Knapp, G. L. (1987). Hydrologic Unit Maps. U.S. Geological Survey Water Supply, Paper 2294, 1–63.

Shimodaira, H. (2002). An approximately unbiased test of phylogenetic tree selection. Systematic Biology, 51(3), 492–508.

Shimodaira, H., & Hasegawa, M. (1999). Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution, 16(8), 1114– 1116.

Simpson, G. G. (1953). The Major Features of Evolution (p. 434). New York, NY: Columbia University Press.

Simpson, George Gaylord. (1951). The species concept. Evolution, 5(4), 285–298.

Sites, J. W., & Marshall, J. C. (2004). Operational criteria for delimiting species. Annual Review of Ecology, Evolution, and Systematics, 35(1), 199–227.

Slaats, J. (1999). TNC Ecoregions and Divisions Map. Retrieved from http://gis.tnc.org/data/MapbookWebsite/map_page.php?map_id=9

138

Stamatakis, A. (2006). RAxML-VI-HPC: Maximum Likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics, 22(21), 2688–2690.

Stephens, M., & Scheet, P. (2005). Accounting for decay of linkage disequilibrium in haplotype inference and missing data imputation. American Journal of Human Genetics, 76, 449–462.

Stephens, M., Smith, N. J., & Donnelly, P. (2001). A new statistical method for haplotype reconstruction from population data. American Journal of Human Genetics, 68, 978–989.

Stephens, P. R., & Wiens, J. J. (2003). Explaining species richness from continents to communities: the time-for-speciation effect in emydid turtles. American Naturalist, 161, 112–128.

Swofford, D. L. (2003). PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sunderland, Massachusetts: Sinauer Associates.

Takahata, N. (1989). Gene genealogy in three related populations: consistency probability between gene and population trees. Genetics, 122, 957–966.

Tavormina, S. J. (1982). Sympatric genetic divergence in the leafmining insect Liriomyza brassicae (Diptera: Agromyzidae). Evolution, 36, 523–534.

Templeton, A. R. (2001). Using phylogeographic analyses of gene trees to test species status and boundaries. Molecular Ecology, 10, 779–791.

Tilley, S. G., Highton, R., & Wake, D. B. (2008). Caudata-Salamanders. IN B. I. Crother (ed.), Scientific and Standard English Names of Amphibians and Reptiles of North America North of Mexico. SSAR Herpetological Circular, 37, 13–23.

Timpe EK, Graham SP, Bonett RM (2009) Phylogeography of the Brownback Salamander reveals patterns of local endemism in Southern Appalachian springs. Molecular Phylogenetics and Evolution, 52, 368–376.

Via, S. (1999). Reproductive isolation between sympatric races of pea aphids. I. Gene flow restriction and habitat choice. Evolution, 53(53), 1446–1457.

Vieites, D. R., Min, M.-S., & Wake, D. B. (2007). Rapid diversification and dispersal during periods of global warming by plethodontid salamanders. Proceedings of the National Academy of Sciences of the United States of America, 104(50), 19903–19907.

Vigil, J. F., Pike, R. J., & Howell, D. G. (2000). A tapestry of time and terrain: U.S. Geological Survey Pamphlet to accompany Geologic Investigations Series I–2720.

Wake, D. B. (1987). Adaptive radiation of salamanders in Middle American cloud forests. Missouri Botanical Garden, 74(2), 242–264.

139

Wake, D. B. (2006). Problems with species: Patterns and processes of species formation in salamanders. Annals of the Missouri Botanical Garden, 93(1), 8–23.

Wakeley, J. (2008). Coalescent Theory: An Introduction (1st ed., p. 432). Greenwood Village, CO: Roberts & Company Publishers.

Walker, D., Moler, P. E., Buhlmann, K. A., & Avise, J. C. (1998). Phylogeographic patterns in Kinosternon subrubrum and K . baurii based on mitochondrial DNA restriction analyses. Herpetologica, 54(2), 174–184. Wallace, A. R. (1852). No Title. Proceedings of the Zoological Society, London, 20, 107–110.

Wang, H., McArthur, E. D., Sanderson, S. C., Graham, J. H., & Freeman, D. C. (1997). Narrow hybrid zone between two subspecies of big sagebrush (Artemisia tridentata). IV. Reciprocal transplant experiments. Evolution, 51, 95–102.

Weisrock, D. W., Macey, J. R., Ugurtas, I. H., Larson, A., & Papenfuss, T. J. (2001). Molecular phylogenetics and historical biogeography among salamandrids of the “true” salamander clade: rapid branching of numerous highly divergent lineages in Mertensiella luschani associated with the rise of Anatolia. Molecular Phylogenetics and Evolution, 18(3), 434– 448.

Whitney, E., Means, D. B., & Rudloe, A. (2011). Priceless Florida: Natural Ecosystems and Native Species (p. 423). Sarasota, FL: Pineapple Press Inc.

Wiens, J. J., Ackerly, D. D., Allen, A. P., Anacker, B. L., Buckley, L. B., Cornell, H. V, Damschen, E. I., et al. (2010). Niche conservatism as an emerging principle in ecology and conservation biology. Ecology Letters, 13(10), 1310–1324.

Wiens, J. J., Chippindale, P. T., & Hillis, D. M. (2003). When are phylogenetic analyses misled by convergence? A case study in Texas cave salamanders. Systematic Biology, 52(4), 501– 514.

Wiens, J. J., Engstrom, T. N., & Chippindale, P. T. (2006). Rapid diversification, incomplete isolation, and the “speciation clock” in North American salamanders (Genus Plethodon): Testing the hybrid swarm hypothesis of rapid radiation. Evolution, 60(12), 2585–2603.

Wiens, J. J., & Graham, C. H. (2005). Niche conservatism: Integrating evolution, ecology, and conservation biology. Annual Review of Ecology, Evolution, and Systematics, 36(1), 519– 539.

Wiens, J. J., & Penkrot, T. A. (2002). Delimiting species using DNA and morphological variation and discordant species limits in spiny lizards (Sceloporus). Systematic Biology, 51(1), 69–91.

140

Wiley, E. O. (1978). The evolutionary species concept reconsidered. Systematic Zoology, 27(1), 17–26.

Wright, R. G., Murray, M. P., & Merrill, T. (1998). Ecoregions as a level of ecological analysis. Biological Conservation, 86(2), 207–213.

Yang, Z., & Rannala, B. (2010). Bayesian species delimitation using multilocus sequence data. Proceedings of the National Academy of Sciences of the United States of America, 107(20), 9264–9269.

Yoder, J. B., Clancey, E., Des Roches, S., Eastman, J. M., Gentry, L., Godsoe, W., Hagey, T. J., et al. (2010). Ecological opportunity and the origin of adaptive radiations. Journal of Evolutionary Biology, 23(8), 1581–1596.

Zelditch, M. L., Swiderski, D. L., & Sheets, H. D. (2012). Geometric Morphometrics for Biologists: A Primer (2nd ed., p. 488). New York, NY: Academic Press.

Zellmer, A. J., Hanes, M. M., Hird, S. M., & Carstens, B. C. (2012). Deep phylogeographic structure and environmental differentiation in the carnivorous plant Sarracenia alata. Systematic Biology, 61(5), 763–777.

Zhang, C., Zhang, D.-X., Zhu, T., & Yang, Z. (2011). Evaluation of a bayesian coalescent method of species delimitation. Systematic Biology, 60(6), 747–761.

Zhang, D. X., & Hewitt, G. M. (1996). Nuclear integrations: Challenges for mitochondrial DNA markers. Trends in Ecology & Evolution, 11, 247–251.

141

BIOGRAPHICAL SKETCH

Education: • 2005-2013. Doctor of Philosopy in Biology. Florida State University. Major Advisor: Scott Steppan. • 1997-2000. Master of Science in Biology. University of Texas at Arlington. Masters Thesis: Nuptial Excrescences of Central American Stream Breeding Frogs: Microstructure and Phylogenetic Significance. Major Advisor: Jonathan A. Campbell. • 1993-1997. Bachelors of Science in Zoology. University of Florida. Peer-reviewed Publications: • Rokyta, D.R., K.P. Wray, M.J. Margres. 2013. The genesis of an exceptionally lethal venom in the timber rattlesnake (Crotalus horridus) revealed through comparative venom-gland transcriptomics. BMC Genomics 14:394 • Rokyta, D.R., K.P. Wray, A.R. Lemmon, E.C. Moriarty Lemmon, and S.B. Caudle. 2011. A high-throughput venom-gland transcriptome for the Eastern Diamondback Rattlesnake (Crotalus adamanteus) and evidence for pervasive positive selection across toxin classes. Toxicon 57:657-671. • Noonan, B.P. and K.P. Wray. 2006. Neotropical Diversification: The effects of a complex history on diversity within the poison frog genus Dendrobates. Journal of Biogeography, 33(6): 1007-1022. Nonpeer-reviewed Publications: • Castoe, T.A., E.L. Braun, A.M. Bronikowski, C.L. Cox, A.R.D. Rabosky, A.P. Jason de Koning, J. Dobry, M. K. Fujita, M.W. Giorgianni, A. Hargreaves, C.V. Henkel, S.P. Mackessy, D. O’Meally, D.R. Rokyta, S.M. Secor, J.W. Streicher, K.P. Wray, K.D. Yokoyama, D.D. Pollock. 2012. Report from the First Snake Genomics and Integrative Biology Meeting. Standards in Genomic Sciences 7:150-152. • Bartlett, R.D. and K. Wray. 2005. Vipers, A Guide for the Advanced Hobbyist. Barron’s Educational Series Inc. Hauppauge, NY 11788-3917. • Wray, K.P. and F.M. Morrissiey. 1999. Florida Green Watersnake (Nerodia floridana): Reproduction. Herpetological Review 30:47. • Wray, K.P. and R. Owen. 1999. New County Records of Amphibians and Reptiles from Nassau County, Florida. Herpetological Review 30: 237-238. • King, F. Wayne and Kenneth Wray. 1996. Guide to Florida’s Venomous Snakes. http://www.flmnh.ufl.edu/herpetology/FL-GUIDE/onlineguide.htm

Research Experience: • 2011-2012. Research Assistant, Steppan lab at Florida State University. Responsible for training undergraduate and graduate students on laboratory procedures and for carrying

142

out extractions, PCR, sequencing, troubleshooting, and analysis of DNA samples for NSF funded research project involving Muroid rodent phylogeny. • 1997-2000. Assistant Collections Manager, University of Texas at Arlington Vertebrate Collection, University of Texas at Arlington, Arlington, TX. Duties included collection, preservation, and cataloguing in the 180,000+ collection. • 1997. Biology Internship, Howard Gillman White Oak Plantation, Yulee, FL. Duties consisted of an eight week field collection, identification, and preservation of the lower vertebrates (reptiles, amphibians, and fish) at the 10,000 acre property in extreme northeastern, FL and to develop a conservation plan between the native fauna and the captive, zoo-based, exotic breeding program of rare and endangered fauna of the world. • 1994-1997. Assistant Collections Manager, Florida Museum of Natural History, University of Florida, Gainesville, FL. Duties included public education, cataloguing and maintenance for the 175,000+ wet, dry, and skeletal specimens of amphibians and reptiles, and the cataloguing and maintenance of the 1000+ books and 50,000 + reprints in the herpetology library. • 1993-1994. Laboratory Assistant, Wakeland Immunopathology Lab, Shands Hospital, University of Florida, Gainesville, FL. Duties included sole maintenance of 2000+ rodent, research colony, general laboratory maintenance, preparation of electrophoretic gels, cataloguing and maintenance of 3000+ journal library. Research Grants and Awards: • 2013. Theodore Roosevelt Memorial Grant, American Museum of Natural History. Phylogenetics and Morphological Relationships of the Dwarf Salamander (Eurycea quadridigitata) complex. $2000. • 2013. Sigma Xi Biological Society. Phylogenetics Relationships of the Dwarf Salamander (Eurycea quadridigitata). $500. • 2011. Texas Herpetological Society. Phylogenetics Relationships of the Dwarf Salamander (Eurycea quadridigitata) Complex in East Texas. $500. • 2011. East Texas Herpetological Society. Phylogenetics Relationships of the Dwarf Salamander (Eurycea quadridigitata) Complex in East Texas. $100. • 2010. Brenda W. Bennison Memorial Scholarship, Florida State University. Phylogenetic Relationships in the Ratsnake (Pantherophis obsoletus): Is the Apalachicola River a historic barrier to gene flow? $1000 • 2007. Robert B. Short Grant in Zoology, Florida State University. Development of Anonymous Loci and the Phylogenetic Relationships of the Dwarf Salamander (Eurycea quadridigitata) Complex. $2000

Professional Experience:

143

• 2001-2005. Composite Science Teacher in Arlington Independent School District. Arlington, TX. Teaching experience included Integrated Physics and Chemistry, Chemistry, Environmental Science, Biology, Algebra I, Algebra II, and Geometry. • 1997-2001. Head Veterinary Technician, Parkway Animal Hospital, Grand Prairie, Texas. Duties included laboratory analysis, culturing, histology, radiology, surgery, diagnosis (team lead on all small animal exotic cases), treatment, client education, and animal health and maintenance. Teaching Experience: • Animal Diversity: Lab Assistant Coordinator, Florida State University; Summer 2007, 2010, 2011. • Animal Diversity: Teaching Assistant, Florida State University; Spring 2007, Fall 2007. • Comparative Vertebrate Anatomy: Teaching Assistant, Florida State University; Spring 2006, 2008, 2009, 2010, 2011, 2013. • Evolution: Teaching Assistant, Florida State University; Fall 2005. • Herpetology: Teaching Assistant, Florida State University; Fall 2006, 2008, 2009, 2010, 2012. • Natural History of Amphibians and Reptiles: Teaching Assistant, University of Florida; Spring 1997 • Composite Science Teacher in Arlington Independent School District, Arlington, TX; 2001-2005. Teaching experience included Integrated Physics and Chemistry, Chemistry, Environmental Science, Biology, Algebra I, Algebra II, and Geometry. Professional Meeting Presentations: • February 2012. Southeastern Partners in Amphibian and Reptile Conservation Annual Meeting, Fall Creek Falls State Park, TN. Exploring Cryptic Diversity in the Widespread Dwarf Salamander (Eurycea quadridigitata) and its Implications for Conservation. • October 2011. Snake Genomics and Integrative Biology Inaugural Meeting, Vail, CO. Venom Transcriptomics and Proteomics at Florida State University. • June 2011. Evolution Meetings, Norman, OK. Testing hypotheses of diversification in the Dwarf Salamander (Eurycea quadridigitata) complex. • March 2010. The 33rd Annual Herpetology Conference, Gainesville, FL. Divergence in the Dwarf Salamander species complex (Eurycea quadridigitata): A case of ecological speciation? • June 2009. Evolution Meetings, Moscow, ID. Phylogenetics of the Dwarf Salamander (Eurycea quadridigitata) complex. • March 2006. I International Congress on the Biodiversity of the Guiana Shield, Santa Elena, Venezuela. The Herpetofauna of Imbaimadai and Surrounding Areas. Guest and Invited Lectures: • September 2012. Dr. Greg Erickson’s Herpetology undergraduate course at Florida State University, Tallahassee, FL. Rise of the Tetrapods.

144

• September 2012. Dr. Greg Erickson’s Herpetology undergraduate course at Florida State University, Tallahassee, FL. Introduction to Herpetology. • November 2010. Dr. Greg Erickson’s Herpetology undergraduate course at Florida State University, Tallahassee, FL. Research in Herpetology. • October 2009. Dr. Austin Mast’s Field Botany undergraduate course at Florida State University, Tallahassee, FL. Wetland Diversity in the Southeastern United States. • June 2009. Florida State University Department of Biology Seminar, Tallahassee, FL. Anonymous Loci Development and Their Use in Phylogenetic and Phylogeographic Studies. • March 2009. Dr. Peter Beerli’s Conservation Biology undergraduate course at Florida State University, Tallahassee, FL. Habitat Restoration of Breeding Wetlands and Conservation of the Endangered Flatwoods Salamander (Ambystoma cingulatum). • September 2005. Dr. Scott Steppan’s Evolution undergraduate course at Florida State University, Tallahassee, FL. Phylogenetics, Trees, and Terminology. • June 2005. Kentuckiana Herpetological Society, Louisville, KY. Herpetological Survey of Imbaimadai, Guyana: Tepuis, Endemism, and a Lost World. • September 2002. Arlington Independent School District, Arlington, TX. The Preservation of Vertebrates for Museum and Teaching Collections. • May 2001. Dallas-Ft. Worth Herpetological Society, Grand Prairie, TX. The Herpetofauna of Baja California and the Northwest Coast of Mexico. • March 2001. West Texas Herpetological Society, San Angelo, TX. The Reptiles and Amphibians of Florida. • August 2000. University of Texas at Arlington Biology Department Seminar, Arlington, TX. Nuptial Excrescences in Hylid Frogs: SEM Morphology. • June 2000. City of Grand Prairie, Grand Prairie, TX. Venomous Snakes of the Dallas-Ft. Worth Metroplex. • August 1999. Society of Midwestern Animal Cruelty Investigators Annual Conference, Ft. Worth, TX. Care of Reptiles and Amphibians in Captivity. • June 1999. North Texas Animal Control Society Annual Meeting, Grapevine, TX. Venomous Reptiles: Identification and Care in Captivity. • October 1998. North Texas Herpetological Society, Arlington, TX. Herpetological Diversity of Florida. • August 1997. Howard Gillman Corporation and White Oak Plantation, Yulee, FL. A Checklist and Conservation Plan for the Lower Vertebrates of the White Oak Plantation, Yulee, Florida. • April 1997. Jacksonville Herpetological Society, Jacksonville, FL. Herpetology of the Desert Southwest. • February 1997. Gainesville Herpetological Society, Gainesville, FL. Herpetology of the Sonora and Colorado Deserts.

145

Field Experience: • 2009-2013. Collection of venomous snakes of the southeastern U.S. for venom and tissue analyses. • June-September 2008. Collection of reptiles and amphibians in western Guyana. • April 2007. Three-week collecting trip of salamanders in the Appalachian Mountain states. • December 2006. Three-week collecting trip of salamanders in western U.S. • April 2006. Two-week collecting trip of salamanders in western U.S. • March 2006. Two-week expedition to Tepui Tablelands in Venezuela. • May-June 2004. Four-week photographic and collecting expedition of fish, amphibians, and reptiles in the Tepui Tablelands of western Guyana. • June-July 2002. Five-week collection of amphibians and reptiles in the western United States. • May 1998. Four-week photographic expedition in Baja California, Baja Sur, Sinaloa, and Sonora Mexico. • 1997-2002. Six years experience collecting amphibians and reptiles in the southwestern United States for the University of Texas at Arlington Vertebrate Collection. • July-August 1997. Eight-week collection of fish, amphibians, and reptiles for the Howard Gillman Corporation at the 10,000-acre White Oak Plantation. • 1994-1997. Three years experience collecting fossils, invertebrates, fish, amphibians, and reptiles in the southeastern United States for the Florida Museum of Natural History.

146