Citation
Phylogeography, Evolution, and Population Genetics of the Salamanders, Pseudobranchus, in the Southeastern United States

Material Information

Title:
Phylogeography, Evolution, and Population Genetics of the Salamanders, Pseudobranchus, in the Southeastern United States
Creator:
LIU, FU-GUO ROBERT
Copyright Date:
2008

Subjects

Subjects / Keywords:
Counties ( jstor )
Genetics ( jstor )
Haplotypes ( jstor )
Highlands ( jstor )
Mitochondrial DNA ( jstor )
Phylogenetics ( jstor )
Phylogeny ( jstor )
Population genetics ( jstor )
Sirens ( jstor )
Species ( jstor )
Greater Orlando ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Fu-Guo Robert Liu. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/31/2008
Resource Identifier:
436098813 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

PHYLOGEOGRAPHY, EVOLUTION, AND POPULATION GENETICS OF THE SALAMANDERS, Pseudobranchus, IN THE SOUTHEASTERN UNITED STATES By FU-GUO ROBERT LIU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Fu-Guo Robert Liu

PAGE 3

This document is dedicated to my deeply loved families, Der-Che and Jeo-Er, beloved wife Rachel, and daughters, Lucia, and Ramona.

PAGE 4

ACKNOWLEDGMENTS First I would like to thank my parents and my wife, R. Tseng. They have been totally dedicated in supporting me through the completion of my degree. I very much appreciate P. E. Moler, who traveled all around the southeastern United States to collect and provide me with most of the samples. Without his help, I had nowhere to start this project. Also, B. Mansell assisted with collection of Pseudobranchus; R. Ashton, D. B. Means, and D. Stevenson provided additional samples; A. M. Clark helped with data collection; and P. Kubilis contributed to Figure 2-1. Particularly, I must thank G. Clark who provided instructions and many useful suggestions to help me conduct molecular genetic work in the BEECS laboratory. I also thank P. Berlli, E. Braun, L. R. Franz, A. Larson, D. L. Swofford, and M. R. Tennant for their suggestions about this research. My appreciation goes to my committee members, M. M. Miyamoto, C. Mulligan, C. F. Baer, H. B. Lillywhite, and J. Reiskind for their support with this project. Their comments and suggestions have been invaluable. Also, the same goes to my ex-committee members, B. Bowen, C. Chapman, and the late L. R. McEdward. Certainly none of this would have been possible without M. M. Miyamoto, my advisor and mentor. He has supported me every step of the way to complete my PhD degree academically and personally. He has helped me formulate and implement the ideas for this project through countless demonstrations and discussions. His continued encouragement and patience have meant a great deal to me and words cannot express the appreciation I feel for all he has done. iv

PAGE 5

I also owe thanks to my fellow graduate students, T. Barko, N. P. Freire C. B. Gunnels IV, K. Hyndman, and T. Young, and post-doc, B. Knudsen, for their help and discussions throughout my work on this project. My sincerest appreciation also goes to the Department of Zoology, and the Biological Sciences Program, University of Florida, for financial assistance; National Institutes of Health and the National Science Foundation for attending the 2003 Summer Institute in Statistical Genetics at NC State University; and Game and Fish Division, Georgia Department of Natural Resources, for authorization to collect in Georgia. Florida collections were conducted under the auspices of the Florida Game and Fresh Water Fish Commission. Finally, I must thank my special friend, D. L. Southworth, for revising my grammar and much other help as an English teacher. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x LIST OF OBJECTS...........................................................................................................xi ABSTRACT......................................................................................................................xii CHAPTER 1 INTRODUCTION........................................................................................................1 Salamander–Pseudobranchus.......................................................................................2 Allozymes.....................................................................................................................2 Mitochondrial DNA Sequences....................................................................................3 Population Genetics......................................................................................................5 Population Structure.....................................................................................................6 Phylogeny.....................................................................................................................7 The Coalescent..............................................................................................................8 Biogeography................................................................................................................8 Sea Level Fluctuations..................................................................................................9 Current Reseach..........................................................................................................11 2 ALLOZYME VARIATION IN THE SALAMANDER GENUS Pseudobranchus: PHYLOGEOGRAPHIC AND TAXONOMIC SIGNIFICANCE.............................12 Materials and Methods...............................................................................................14 Results.........................................................................................................................17 Within-Sample Variation.....................................................................................17 Between-Sample Variation..................................................................................18 Discussion...................................................................................................................21 Within-Sample Variation and Paedomorphism...................................................21 Taxonomic and Phylogeographic Significance...................................................22 3 PHYLOGEOGRAPHY OF THE SALAMANDERS, Pseudobranchus, AND OTHER ENDEMIC GROUPS IN THE SOUTHEASTERN UNITED STATES.....26 vi

PAGE 7

Materials and Methods...............................................................................................28 Results and Discussion...............................................................................................32 Cytochrome b Sequences....................................................................................32 Phylogeographic Units........................................................................................33 Rooting the Pseudobranchus Phylogeny.............................................................38 Within-Locality Variation...................................................................................39 Regional Phylogeography...................................................................................40 Separate Species?................................................................................................42 4 GENETIC VARIATION WITHIN POPULATIONS AND SPECIES OF THE SALAMANDER Pseudobranchus: IMPACTS OF HISTORICAL OCEANIC FLOODING IN THE SOUTHEASTERN UNITED STATES..................................43 Methods......................................................................................................................45 Data Sets and Reference Phylogenies.................................................................45 Within-Population Diversities and F-statistics....................................................49 Dating the MRCA................................................................................................51 Results.........................................................................................................................53 Discussion...................................................................................................................59 Interglacial Oceanic Flooding.............................................................................59 Within-Population Diversity...............................................................................60 Species Ages and Phylogeographic Groups........................................................61 5 SUMMARY................................................................................................................64 APPENDIX A LOCALITY DATA FOR THE 245 SPECIMENS OF Pseudobranchus AND Siren (allozyme study)................................................................................................69 B GENOTYPE ARRAYS AT THE 13 VARIABLE LOCI FOR THE 27 SAMPLES OF Pseudobranchus AND Siren................................................................................72 C FULL LOCALITY DATA, SAMPLE SIZES, AND MUSEUM COLLECTION NUMBERS (UF) FOR THE 314 INDIVIDUALS OF Pseudobranchus AND Siren (mtdna study).....................................................................................................75 D ANNEALING SITES AND SEQUENCES FOR THE PCR AMPLIFICATION AND CYCLE SEQUENCING PRIMERS OF CYT B FOR Pseudobranchus AND Siren..................................................................................................................78 E MULTIPLE SEQUENCE ALIGNMENT FOR THE 173 DISTINCT HAPLOTYPES OF CYT B FOR Pseudobranchus AND Siren.................................79 F FREQUENCY DISTRIBUTION OF HAPLOTYPES AMONG THE 42 SAMPLES OF Pseudobranchus AND Siren.............................................................80 vii

PAGE 8

G COMPLETE SET OF D’S AND ’S FOR THE 42 SAMPLES OF Pseudobranchus AND Siren.......................................................................................82 LIST OF REFERENCES...................................................................................................83 BIOGRAPHICAL SKETCH.............................................................................................90 viii

PAGE 9

LIST OF TABLES Table page 4-1 Within-population diversity estimates for P. axanthus and P. striatus. .................54 4-2 F-statistics for the allozyme and mtDNA data. ......................................................56 4-3 Absolute ages for the MRCA of P. axanthus and P. striatus, as estimated from their allozyme and mtDNA data. ............................................................................59 B-1 Genotype arrays at the 13 variable loci for the 27 samples of Pseudobranchus and Siren. ................................................................................................................72 C-1 Full locality data, sample sizes, and museum collection numbers (UF) for the 314 individuals of Pseudobranchus and Siren (mtDNA study)...............................75 F-1 Frequency distribution of haplotypes among the 42 samples of Pseudobranchus and Siren...................................................................................................................80 ix

PAGE 10

LIST OF FIGURES Figure page 1-1 Chronostratigraphy and cycles of global sea level changes. ..................................10 2-1 Samples of Pseudobranchus and Siren used in this study. ....................................13 2-2 Continuous ML tree for the 26 samples of Pseudobranchus. ................................19 3-1 Maps of some endemic groups in the southeastern United States. ........................27 3-2 Map of the southeastern US for the 39 samples of Pseudobranchus and 3 samples of Siren. ....................................................................................................29 3-3 Unrooted fast ML phylogeny for the 39 Pseudobranchus samples. ......................34 3-4 Summary of the inter-sample divergences (d’s) for the major groups and subgroups of Pseudobranchus and Siren. ..............................................................37 4-1 Maps of the southeastern United States highlighting those terrestrial regions >30 m and >15 m above current sea levels and the distributions of the recognized phylogeographic groups within Pseudobranchus. .................................................44 4-2 Reference allozyme phylogeny for the 26 population samples of Pseudobranchus, as rooted against Siren. ..............................................................47 4-3 Reference mtDNA phylogeny for the 305 individual haplotypes of cyt b for Pseudobranchus. ....................................................................................................48 4-4 Summary of Nei genetic distances between all population pairs of the recognized phylogeographic groups. ......................................................................58 D-1 Annealing sites and sequences for the PCR amplification and cycle sequencing primers of cyt b for Pseudobranchus and Siren. ....................................................78 x

PAGE 11

LIST OF OBJECTS Object page E-1 Multiple sequence alignment for the 173 distinct haplotypes of cyt b for Pseudobranchus and Siren. .....................................................................................79 G-1 Complete set of d’s and ’s for the 42 samples of Pseudobranchus and Siren. ....82 xi

PAGE 12

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PHYLOGEOGRAPHY, EVOLUTION, AND POPULATION GENETICS OF THE SALAMANDERS, Pseudobranchus, IN THE SOUTHEASTERN UNITED STATES By Fu-Guo Robert Liu August 2005 Chair: Michael M. Miyamoto Major Department: Zoology This study examines allozyme and mitochondrial DNA variation for 238 and 305 individuals from 26 and 39 samples, respectively, to investigate the phylogeography, evolution, and population genetics of Pseudobranchus. Firstly, we used allele frequency and genotype array data from allozyme electrophoresis to investigate the phylogeographic and taxonomic affinities of groups within this salamander genus. Phylogenetic and population genetic analyses of the allozyme data confirm the validity of its two currently recognized species, P. axanthus and P. striatus. In concert with biogeographic data for other freshwater taxa, these analyses further support the recognition of western, southeastern, and northeastern phylogeographic groups within P. striatus. These species and phylogeographic groups offer new insights into the evolutionary history of this endemic genus in the southeastern United States and the taxonomic status of its nominal subspecies. xii

PAGE 13

Secondly, we generated an extensive set of mitochondrial cytochrome b sequences for Pseudobranchus, thereby providing new insights into the regional phylogeography of the southeastern United States. Results from multiple phylogenetic methods with these sequences supported a single set of major phylogeographic divisions for Pseudobranchus including those from the allozyme study. Correspondingly, these divisions highlight the Central Highland and Tifton/Vidalia uplands as a significant barrier to Atlantic versus Gulf coast groups, while reconfirming the phylogeographic significance of the Altamaha and Apalachicola river drainages. Thirdly, we examined allozyme and mitochondrial DNA variation within the populations and species of Pseudobranchus. Population genetic analyses of these data document that allozyme diversity is greater within P. striatus than within P. axanthus, but their mitochondrial DNA polymorphism is similar. This difference is tied to the greater population bottlenecking and extinction within P. axanthus due to interglacial marine flooding since the late Miocene, as well as to the faster sorting times for the mitochondrial DNA. Estimates of the ages and of the effective female population sizes (Nf) for both species show that P. striatus is at least 2.5 times older than P. axanthus despite their similar Nf. This age difference is linked to the more recent divergence of the phylogeographic groups within P. axanthus, again as a result of its greater lineage extinction due to interglacial marine flooding since the late Miocene. xiii

PAGE 14

CHAPTER 1 INTRODUCTION The secret of life always attracts human attention. This leads us to rapidly expand our knowledge of living systems as well as to constantly improve our technologies for research. Still, discovering the ultimate truth is a great challenge. Within my research, the major question is what are the major factors shaping the current biogeographic distribution patterns in the southeastern United States? My working hypothesis is that sea level fluctuations have shaped the geographic distributions evolution, and variation of entire communities in this region. Although we cannot go back in time to see what happened, these historical events may leave some tracks which we can find today. From many perspectives, species and their environments are connected to each other, in a scale as large as in an ecosystem or as small as in molecules. Thus, we can study one system in many different ways to find clues. With multiple evidence we can picture what happened very closely to the truth using different perspectives. Thus, in this study, we will start by clarifying the taxonomy of the target organism, Pseudobranchus, with allozyme and mtDNA data, then using different analysis tools (i.e., population genetic, phylogeny, coalescence and molecular clock) to see whether their patterns are consistent and how the information (i.e., migration, effective population size, geology, paeleontology, as well as results from other studies for different organisms) can enlighten us about the evolutionary history in the southeastern United States. The following sections introduce some major components used in this study. 1

PAGE 15

2 Salamander–Pseudobranchus Freshwater organisms possess ideal characteristics for historical biogeographic analysis (Hocutt and Wiley, 1986). One of the world’s richest freshwater faunas occurs in the southeastern United States. Since this area is relatively stable geophysically, the potential exists for testing many aspects of biogeographical and evolutionary hypotheses. Among these freshwater groups in the southeastern United States Pseudobranchus (dwarf siren) is one of the most ideal organisms for biogeographic studies. Unlike fishes swimming up and down stream frequently, dwarf sirens hide in the mud most of the time (Petranka, 1998). With this low migration potential each population is more restrictive in the local area and will likely accumulate unique variance and increase the variation between each population. This may also increase the sensitivity to detect their population structure and reflect their evolutionary history (Hedrick, 2005). Fossils of Pseudobranchus occur in the Pliocene and Pleistocene of Florida (Martof, 1972). Pseudobranchus are perennibranchs and exhibit larval traits such as absence of eyelids and the presence of one gill slit and external gills (Petranka, 1998). They have two front limbs that are greatly reduced and have three toes per limb. Females lack spermathecae and males lack cloacal glands. Males do not produce spermatophores and fertilization is presumed to be external. Many of the populations inhabit semipermanent habitats. During the dry season, i.e., summer, they would form a cocoon and aestivate. Siren, another paedomorphic group, is the only sister genus to Pseudobranchus. Allozymes Protein electrophoresis is among the most cost-efficient methods of investigating genetic phenomena at the molecular level (Hedrick, 2005). It can separate proteins by

PAGE 16

3 their different mobilities. When a protein has an amino acid substitution that leads to molecular size, electrical charge, or shape change, its mobility could be altered. Thus, protein electrophoresis has been widely used to detect genetic variation for studying natural populations, gene flow, hybridization, recognition of species boundaries, and phylogenetic relationships, among other questions. One of the most popular forms of protein data is allozyme data, allele frequencies, which are variants of polypeptides representing different allelic alternatives of the same gene locus (Hedrick, 2005). Extensive allozyme variation has been found in almost all natural populations studied by electrophoresis, including in Escherichia coli, plants, Drosophila, mice and humans. Then, by comparing different populations or groups we may see some alleles fixed at one genotype and other groups fixed for the alternative. More interestingly, we may find that these groups are separated by some geographic features as with other organisms. Two of the most common measures of genetic variation calculated from allele frequencies are heterozygosity and polymorphism. The average heterozygosity is estimated by totaling the number of heterozygous individuals of each gene, then dividing the result by the total number of individuals in the sample (Hedrick, 2005). Polymorphism is the estimated proportion of genes that are polymorphic. A polymorphic gene is a gene in which the most common allele has a frequency of less than 95%. On the other hand, protein electrophoresis is only able to detect the mobility difference between enzymes, but cannot indicate how different they are. It may underestimate genetic variation using either measure. Mitochondrial DNA Sequences Because of the rapid development of modern molecular techniques it has become possible to measure genetic variation at the level of DNA sequences (Avise, 2000a).

PAGE 17

4 DNA sequence data contain quantitative characters to compare variation between each pair of individuals, and are therefore much more usefully presented as gene genealogies, which describe the phylogenetic relationships of the sequence samples from a population. To understand the basis for the patterns of the DNA sequence variation within and between populations is the focus of molecular population genetics (Hedrick, 2005). One of the general measures to quantify the amount of variation is nucleotide diversity, (Nei and Li,1979; Nei and Tajima, 1981). There are two ways to approach . One is to average the proportion of nucleotide differences of all pair-wise comparisons between all sequences obtained from all individuals, i.e., to determine the proportion of nucleotide differences between pairs of distinct haplotypes and then weigh these differences by the frequencies of the distinct haplotypes. Additionally, another informative measure is haplotype diversity, h. It describes the number and frequency of different haplotypes in the sample, h = 1 f i 2 , where f i is the ith distinct haplotype frequency within the sample (Nei and Tajima, 1981). Animal mtDNA evolves rapidly, and is maternally inherited without recombination (but see Gantenbein et al., 2005; Sato et al., 2005), making it different from nuclear genes which are used in allozyme analyses and making it a valuable marker for the study of variation within and among populations (Avise, 2000a). Much of the variation in animal mtDNA is due to base substitution, with transitions greatly outnumbering transversions. Levels of differentiation among populations tend to be significantly higher than those within populations (Avise, 2000b). Correspondingly, mtDNA can be used to estimate phylogenies of populations, and thus to investigate patterns of historical

PAGE 18

5 phylogeography. This phylogenetic approach provides a qualitative assessment of genetic population structure within species. The unique characteristics of mtDNA also result in some disadvantages (Hedrick, 2005). The lack of recombination makes mtDNA comparable to a single allozyme locus with many alleles. From a functional perspective, mtDNA consists of about 37 genes, but from a phylogenetic perspective the entire mtDNA molecule represents one non-recombining genealogical unit with multiple alleles. Nevertheless, mtDNA has proved useful for resolving relationships of closely related species and it would be very interesting to examine the advantages and disadvantages between the maternally inherited system and the biparentally inherited system (i.e., mtDNA and nuclear genes, respectively). Population Genetics Population genetics is concerned with the genetic basis of evolution. It combines observation with theory for the purpose of understanding genetic changes that occur within and among populations. In general, we know the mutation rate is about one substitution in tens of thousands to millions of years (Avise, 2000a). To directly observe the changes in nature is not easy. However by constructing a mathematical model of evolution and comparing the difference between two or more individuals or populations, evolution can become “computable” in a practical manner. Some models or theories are more sophisticated than others. Given the genetic variation, population genetics can be extended to many interesting subjects, such as selection, genetic drift, effective population size, inbreeding, hybridization, gene flow, population structure, mutation, migration, phylogeny, molecular clock, coalescence, or neutral theory. For example, because selection leaves different signatures on the genetic variation, balancing selection

PAGE 19

6 would maintain the heterozygote in a population due to the fitness of heterozygotes being higher than homozygotes, and positive selection would likely fix an advantageous mutation according to environmental conditions (Hedrick, 2005). Genetic drift is the process of change in allele frequency resulting from the random sampling of gametes from generation to generation. If the population size is small, the allelic frequency can undergo large fluctuations in different generations and can result in chance fixation or loss of an allele. Genetic drift is a way of losing variation; in contrast, mutation generates new variation. Population Structure A population may have differences in genetic variation among its constituent parts for several different evolutionary reasons. When a population is subdivided, the amounts of genetic similarity among the parts of the population can be different. If the amount of gene flow between groups were high, gene flow would have the effect of homogenizing genetic variation over the groups (Weir, 1996). On the other hand, when the gene flow is low, genetic drift, selection, and mutation in the separate groups may lead to greater genetic differences. This would result in different population structure patterns. Several different approaches to the statistical description of population structure have been developed (Sokal and Oden, 1978; Weir, 1996). Wright (1951) developed one approach that detects the genetic variation in a subdivided population to provide an obvious description of differentiation and this has been employed most widely. His method consists of three F coefficients, F IS , F IT , and F ST, and their interrelationship can be written as 1F IT =(1F ST )(1F IS ). F IS is the correlation between homologous alleles within individuals with reference to the local population, and F IT is the corresponding allelic correlation with reference to the total population. In other words, F IS and F IT are

PAGE 20

7 measures of the deviation from Hardy-Weinberg proportions within subpopulations and in the total population, respectively, where positive values indicate a deficiency of heterozygotes, and negative values represent an excess of heterozygotes. F ST is a measure of the genetic differentiation among subpopulations and is always positive. Phylogeny Phylogeny is the study of historical relationships among organisms (Swofford et al., 1996). Any traits or characters that are independent of one another, heritable, and variable among the taxa in the study can help reconstruct the trees. Distance, parsimony, and maximum likelihood are three powerful and popular methods to find the optimal tree. Distance method converts the data into pairwise measures of evolutionary change that are used to find optimal trees, for example, with the minimum tree score. The parsimony method finds the tree with the minimum steps in the character state changes. The maximum likelihood method estimates and finds the tree with the greatest probability given an explicit evolutionary model. Relatively recent Bayesian approaches with sophisticated statistical estimates have also become available and popular (Ronquist and Huelsenbeck, 2003). Bayesian analysis estimates the posterior probability of finding trees for the given data and evolutionary model. Among the above methods, only Bayesian analysis provides the posterior probability of finding trees showing the reliability of the tree. Therefore, for other methods the bootstrap has become one of the most popular statistical tests to represent the confidence of one’s phylogenetic clades (Swofford et al., 1996). To do phylogenetic analyses is not only to generate a tree or find relationships, but it also helps us understand the histories and circumstances under which traits have evolved. In the same vein as knowledge about siblings is useful to understanding a

PAGE 21

8 person, knowledge about a group allows for predictions about its individual subgroups and species. The Coalescent Coalescent theory was developed by the mathematician Kingman (1982), and relates to the times at which lineages join up ("coalesce") as we follow a genealogy of alleles back into the past, to their most recent common ancestor. This is useful because it allows us, given a genealogy, to infer factors like theta (4Ne, four times the effective population size times the neutral mutation rate), T (t/2Ne, divergent time divided by two and the effective population size) and M (2Nem, two times the effective population size times the migration rate) (Edwards and Beerli, 2000; Beerli and Felsenstein, 2001). For a haploid gene the 2Ne is replaced by Ne. Because evolutionary processes leave different signatures in the structure of evolutionary trees, given coalescent theory, we would know what gene genealogies are expected to look like. For example, under neutral genetic drift a growing population would have deeper branch splits than a constant population (Donnelly and Tavare, 1995; Page and Holmes, 1998). Or, when we see a star-like tree or a clade with short branches, we can use multiple genes to tell whether it reflects positive selection for one gene or a bottleneck effect on the entire genome. Biogeography Simply stated, biogeography is the study of the distribution of organisms in space and time (Brown and Lomolino, 2003). However, merely describing these distribution patterns is not very interesting, but to discover which environmental and/or historical factors are the ones that determine or limit the distributions of the species makes this subject very remarkable. Numerous scientists have studied these factors, such as climate changes, sea level fluctuations, glaciations, plate tectonics, dispersals, speciation, and

PAGE 22

9 extinction (Nores, 2004; Liu et al., 2005; Weeks et al., 2005). For example, comparing the distributions of various groups for both their fossil and current species, we may find that a portion of the groups has very similar distributions but another has an extremely different pattern. For the former we may find evidence such as low dispersal mechanism or strong adaptations to local conditions. In the latter it may be because of dispersal, extinction, vicariance, or continental drift. These results apply not only to the studied organisms, but also allow for predications about others in the same area (Liu et al., 2005). Correspondingly, many studies have focused on the effects of climate change, glaciations, or sea level fluctuations on the environments which help us not only to understand the past, but also to predict our future. In these ways, biogeography is important to many fields. Sea Level Fluctuations Throughout the earth’s history the average sea level has never been constant. Changes in climate, ice sheets on both poles, ocean circulation, and plate tectonics are factors underlying changing sea levels. It has been well documented that the sea level at any particular place has varied over a wide range of time and space scales, periodically transgressing and regressing over millions of years (Geophysics Study Committee, 1990; Otvos, 1995). During the last five million years, on a global scale, there were several global significant high-stands and low-stands recognized (Figure 1-1). Although the magnitudes and the exact dates still need further study, recent studies support very consistent fluctuation patterns (Haq et al., 1987; Eberli, 2000). Also, these periodic sea level fluctuations have been recognized to strongly correspond to the glacial-interglacial cycles (Hallam, 1992; Brown and Lomolino, 2003). The effects of inundating and emerging landmass on the distributions and diversity of both terrestrial and marine biotas

PAGE 23

10 Figure 1-1. Chronostratigraphy and cycles of global sea level changes. The time scale (Ma: million years ago) is presented on the top from the beginning of the late Miocene to present. This chart is adapted from Haq et al. (1987) and Abreu and Erson (1998). have attracted the attention of many scientists involved in biogeography (Hallam, 1994; Nores, 2004). At late Miocene the sea level remained at a low stand for 1-2 million years at 80-100 m below present levels, then was followed by a rise to greater than 30 m above present levels for about 1 million years (~5.5 – ~4.5 Ma) (Haq et al., 1987; Abreu and Erson, 1998; Eberli, 2000). Another drop occurred after this high stand at late Pliocene, followed by a series of Pleistocene fluctuations, none of which rose more than 10-20 m

PAGE 24

11 above present levels and some of which may have fallen 80-100 m below current marks (Haq et al., 1987; Eberli, 2000; O’Neal and McGeary, 2002). Current Reseach In Chapter 2, we studied 14 allozyme loci of 238 Pseudobranchus from 26 samples estimated their heterozygosities, polymorphyisms, population structures, and phylogeny. Then, we investigated their taxonomic status and compared their phylogeography against other organisms and to geographic barriers and changing sea levels since Miocene to explain their evolutionary history. In chapter 3, 305 individuals were studied to generate mtDNA (cyt b) sequence from 39 samples, and genetic diversities and different phylogenetic analyses were then completed. As before, phylogeographic units were defined and compared with previous findings and with various geographic barriers and historical fluctuating sea levels. For chapter 4, we focused on the finer details of the genetic variation within the populations and species by re-examining their allozyme and mtDNA data. By applying the previous results and using coalescent theory and other methods, we estimated the diversities, divergence times, and effective population sizes for each species to test once again for the effects of changing sea levels since the late Miocene on the evolution of these salamanders in the southeastern United States.

PAGE 25

CHAPTER 2 ALLOZYME VARIATION IN THE SALAMANDER GENUS Pseudobranchus: PHYLOGEOGRAPHIC AND TAXONOMIC SIGNIFICANCE The salamander family Sirenidae includes the two genera Pseudobranchus (dwarf sirens) and Siren (sirens), both of which consist of paedomorphic, eel-like, freshwater salamanders with prominent external gills and no hind legs (Martof, 1972). The genus Pseudobranchus includes two recognized species, P. axanthus and P. striatus. Pseudobranchus axanthus is found throughout peninsular Florida, whereas P. striatus is largely restricted to the Atlantic and Gulf coastal terraces of southern South Carolina, Georgia, and central and northern Florida west to Choctawhatchee Bay (Figure 2-1). The distributions of the two species overlap by at least 130 km in north central Florida, where both have been collected syntopically at several locations (Moler and Kezer, 1993). Despite the current recognition of only P. axanthus and P. striatus, a significant degree of geographic variation exists within the genus, as reflected by the several nominal subspecies (Martof, 1972; Moler and Kezer, 1993). However, the taxonomic status and distributions of the described subspecies remain ambiguous, because no thorough, range-wide assessment of variation within the genus has been conducted. This is especially true for P. striatus along the Atlantic lowlands of Florida, a region traditionally thought to include only P. axanthus (Figure 2-1). In this study we examine allozyme variation, both within and between P. axanthus and P. striatus. The taxonomic validity of these two species relative to each other and to the recognized subspecies is evaluated along with their phylogeographic significance. 12

PAGE 26

13 Figure 2-1. Samples of Pseudobranchus and Siren used in this study. The numbering system for samples in this figure is followed throughout this chapter. Detailed locality and specimen data are listed on Appendix A. Grey areas indicate the Florida Central Highland and other upland areas that afford little or no suitable habitat for P. striatus. OB = Okefenokee Basin; TU = Tifton Upland; VU = Vidalia Upland.

PAGE 27

14 These phylogeographic and taxonomic evaluations offer further insights into the evolutionary history of this endemic genus from the southeastern United States. Materials and Methods A total of 238 specimens from 26 samples of P. axanthus and P. striatus were collected from 24 localities in Florida and Georgia (Figure 2-1). Specimens were assigned to species according to their color patterns and karyotypes (with the latter available for 21 of the 26 samples) (Martof, 1972; Moler and Kezer, 1993). In addition, seven individuals from a single sample of Siren intermedia were originally included as the outgroup (Figure 2-1). The allozyme data for S. intermedia were considered in the analyses of within-sample variation (see below). However, this outgroup was ultimately excluded from the phylogenetic analyses, because of its extensive allozyme divergence from Pseudobranchus (e.g., the two genera shared no alleles at high frequency at 10 of the 13 variable loci). Specimens were sacrificed by immersion in a solution of MS-222 (tricaine methane sulphonate) or by freezing at -76 C. Skeletal muscle and liver tissues were subsequently removed and stored at -76 C until used. Carcasses were fixed in 10% formalin, preserved in 70% ethanol, and placed in the herpetology collection at the Florida Museum of Natural History, University of Florida. Biochemical procedures consisted of horizontal starch gel electrophoresis with specific histochemical staining for the enzyme products of 14 presumptive loci (Murphy et al., 1996). Tissue homogenates were prepared from ~0.2 g of thawed tissue in 0.2 ml of chilled grinding buffer. Samples were run on 11.6% starch gels (Connaught Laboratories, lot #434-1) within 24 hours of homogenization. All electromorphs at a

PAGE 28

15 locus were run side by side on bridging gels to confirm their identities. These electromorphs were interpreted as genotypes and their represented alleles were assigned alphabetical designations in order of their associated anodal mobilities. Genotype arrays for the polyallelic loci of each sample were tested for their conformity to Hardy-Weinberg expectations (Guo and Thompson, 1992). These evaluations relied on the exact probability test for diallelic loci and on the chi-square statistic for goodness-of-fit (with pooled classes to compensate for small sample sizes) for all others. Estimates of polymorphism (i.e., the proportion of polymorphic loci) and mean heterozygosities (both direct count and Hardy-Weinberg expected) were also calculated from the allele frequencies and genotype counts (Swofford and Selander, 1981). In the calculation of the former, a polyallelic locus was scored as polymorphic when the allele frequency for its most common allele was < 0.95. Unrooted trees were estimated for the 26 samples of Pseudobranchus with the continuous maximum likelihood (contML), approximate frequency parsimony (aFREQPARS), and minimum evolution (ME, with arc chord distance) methods (Felsenstein, 1981; Swofford and Selander, 1981; Berlocher and Swofford, 1997; Wiens, 2000). These estimations each relied on heuristic searches with 1,000 random taxon additions and global branch swapping. Support for their final groups was evaluated by bootstrapping over loci with 200 pseudoreplicates apiece. Nei (1978) unbiased distances were calculated for all sample pairs. Genotype arrays for the variable loci of the syntopic samples were compared between species with the contingency chi-square test of heterogeneity. Genetic subdivision within P. axanthus was tested with the coancestry coefficient ( P or F ST ) using a standard, two-level,

PAGE 29

16 hierarchical analysis of the genotype counts and allele frequencies (Weir, 1996). Confidence limits (95%) for its F-statistics were established by bootstrapping over loci with 1,000 pseudoreplicates. Genetic subdivision within P. striatus was evaluated according to the three phylogeographic groups of this species, which were identified from the phylogenetic and geographic evidence (see Results). Genetic partitioning among these three was first assessed with a three-level hierarchical analysis of P using BIOSYS-2 (see below). However, as BIOSYS-2 does not generate confidence intervals for its F-statistics, the significance of P for these three groups was further evaluated in two ways. First, genetic partitioning between the southeastern and western groups was assessed with a three-level hierarchical analysis of P using GDA (see below). As for P. axanthus, confidence intervals for the F-statistics of this comparison were generated with 1,000 bootstrap pseudoreplicates. However, because GDA requires multiple samples per unit (unlike BIOSYS-2), genetic divergence between the northeastern group (i.e., its single representative <17>) versus the southeastern and western units was instead tested with the heterogeneity chi-square test. These chi-square tests compared <17> to <16> of the southeastern group and to both <5> and <6> of the western unit. Samples <5>, <6>, and <16> were chosen to represent their groups in these comparisons, because they included > 5 individuals and were closest to <17> in terms of their geographic and/or genetic distances (Figure 2-1). All population genetic analyses of withinand between-sample variation were conducted with BIOSYS-2 (Black, 1997; Swofford and Selander, 1981), except for the P calculations of P. axanthus and the southeastern versus western comparison of P.

PAGE 30

17 striatus, which were done with GDA, version 1.1 (Lewis and Zaykin, 2002). The contML analysis was accomplished with PHYLIP, version 3.6 (Felsenstein, 2002), whereas the aFREQPARS and ME estimations were completed with PAUP*, version 4.0b (Swofford, 2002). Results Within-Sample Variation Fourteen presumptive loci (13 variable) were scored for the 245 individuals of the study group and outgroup (Appendix B). None of the three coefficients for heterozygosity (both direct count and Hardy-Weinberg expected) and polymorphism was positively correlated with sample size (Spearman’s coefficient of rank correlation; P > 0.178 in every case). Thus, the heterozygosities and polymorphism differences among the samples were not simply due to their variable sample sizes of 1 to 21 individuals. Rather, the highest direct count (0.146 and 0.133) and expected (0.150 and 0.104) heterozygosities and polymorphisms (0.357 and 0.357) were associated with samples <6> (N = 5) and <16> (N = 7) of P. striatus. Similarly, the average direct count and expected heterozygosities and polymorphisms were higher for P. striatus (0.068 + 0.036 standard deviation, 0.078 + 0.035, and 0.206 + 0.101, respectively) than for P. axanthus (0.027 + 0.025, 0.047 + 0.052, and 0.119 + 0.101) and S. intermedia (0.000 + 0.000, 0.019 + 0.019, and 0.071). Seventy-one instances of a polyallelic locus were detected among the 27 samples of Pseudobranchus and Siren (Appendix B). The genotype counts of eight of these deviated significantly from Hardy-Weinberg expectations at P < 0.05. Significant deviations from Hardy-Weinberg equilibrium were expected by chance alone for approximately four of the 71 polyallelic cases (0.05 x 71). Furthermore, five of eight significant instances were

PAGE 31

18 scattered among different samples. In contrast, the remaining three cases (sAta-A, Pgm-A, and Pnp-A) involved sample <17>. These three polyallelic loci of <17> departed from Hardy-Weinberg expectations because of heterozygote deficiencies. Between-Sample Variation Because of the extensive allozyme divergence between the study group and outgroup, the contML, aFREQPARS, and ME trees for Pseudobranchus remained unrooted. Nevertheless, these unrooted trees still allowed for the recognition of potentially monophyletic groups within the genus. As illustrated by the contML solution, all three trees split the samples of P. axanthus and P. striatus into two distinct clusters that were separated by a long internode (Figure 2-2). This split between these two potentially monophyletic groups was defined in the contML analysis by a relatively high bootstrap score of 89%. Pseudobranchus axanthus and P. striatus were fixed for different diagnostic alleles at Ck-A and mSod-A and shared no alleles at their variable Ada-A and Gpi-A (Appendix B). Furthermore, syntopic samples of both from eastern Hernando County, Florida, (<20> and <3>) shared no alleles at Ada-A, Ck-A, Gpi-A, and mSod-A and almost none at sAcoh-A, sMdh-A, and G3pdh-A. Similarly, their syntopic samples from Volusia County, Florida, (<26> and <14>) shared no alleles at Ada-A, Ck-A, Gpi-A, mSod-A, and sAcoh-A. In both cases, the differences in genotype counts between the syntopic samples were highly significant (P < 0.0001). Finally, the mean Nei distance between the two species (0.512, range of 0.338-0.731) was 12.2 and 3.3 times greater than their separate intraspecific averages (0.042, 0.001-0.174 for P. axanthus; 0.153, 0.003-0.388 for P. striatus).

PAGE 32

19 Figure 2-2. Continuous ML tree for the 26 samples of Pseudobranchus, with bootstrap scores shown for those groups with > 50% support. The contML method relies on an evolutionary model of allele frequency changes over time by a process of Brownian motion (Felsenstein, 1981, 2002). The positive ln likelihood score indicates that the density function in this case is > 1. Branch lengths are drawn proportional to their expected amounts of accumulated variance. The circled bootstrap score of 89% highlights the long internode that separates P. axanthus and P. striatus. The growing evidence for their full species status indicates that the root for this unrooted tree may lie somewhere along this long internode (see text).

PAGE 33

20 The contML, aFREQPARS, and ME trees also allowed for the recognition of three phylogeographic groups within P. striatus, representing the western, southeastern, and northeastern parts of its range in north central Florida and adjacent Georgia (Figures 2-1 and -2). The geographic partition of its western versus southeastern and northeastern groups was based on the Central Highland and Tifton/Vidalia Uplands, whereas the latter two were separated by the Altamaha River. The P for these three phylogeographic groups was 0.316 (with f = 0.135, F = 0.611, and S = 0.431; see below). However, as illustrated by the contML tree, these three groups were only weakly separated by bootstrap scores of 54% to 63%. Furthermore, the P for the southeastern versus western groups (0.381) was not significant given its overlapping 95% confidence interval with zero (-0.031-0.727). This lack of significance for the southeastern and western groups stood in contrast to their significant divergence among samples ( S = 0.656, 0.364-0.854) and deficit in individual heterozygosity versus total variation (F = 0.771, 0.432-0.904, with f [the inbreeding coefficient] = 0.160, -0.012-0.442). Nevertheless, these two groups were still defined by their fixed and nearly fixed allelic differences at mSod-A and sMdh-A (Appendix B). In contrast, the genotype counts for northeastern <17> differed significantly from those for the southeastern (<16>) and western (<5> and <6>) representatives according to their heterogeneity chi-square tests (P < 0.0001 in every case). In particular, the significant result for <17> versus <16> was critical, given that these two were obtained only ~40 km apart from north and south of the Altamaha River (Figure 2-1). Sample <17> shared the diagnostic allele of the western group at mSod-A, but that of the southeastern cluster at sMdh-A (Appendix B). The mean Nei distances for <17> versus

PAGE 34

21 the southeastern and western groups were 0.098 (0.079-0.122) and 0.151 (0.105-0.243), respectively. The contML, aFREQPARS, and ME trees revealed little hierarchical structure within P. axanthus, as its samples were clustered around a single branch point (Figure 2-2). This lack of resolution stood in contrast to its significant genetic divergence among samples ( P = 0.510, 0.206-0.651), as well as deficiencies in individual heterozygosity versus both samples (f = 0.313, 0.036-0.501) and total variation (F = 0.663, 0.332-0.749). These conspecific samples differed by a mean Nei distance of 0.042 (0.001-0.174). Discussion Within-Sample Variation and Paedomorphism The mean heterozygosities (both direct count and Hardy-Weinberg expected) and polymorphism for P. striatus are consistent with the general values that have been reported for amphibians and other vertebrates (Nei and Graur, 1984; Nevo and Beiles, 1991). For example, its average direct-count heterozygosity of 6.8% approximates the standard estimate of ~6% for a typical amphibian. In contrast, the average heterozygosities and polymorphisms for P. axanthus and S. intermedia are consistent with the reduced genetic variation reported for other species of paedomorphic salamanders (Shaffer and Breden, 1989). The underlying cause of this difference between P. striatus versus P. axanthus and S. intermedia remains unclear. Nevertheless, this dichotomy among sirenids suggests that further comparisons of P. axanthus and S. intermedia to P. striatus may provide insights into the demographic, ecological, and evolutionary factors that underlie the reduced genetic variation in most paedomorphic salamanders.

PAGE 35

22 Taxonomic and Phylogeographic Significance On both evolutionary and reproductive grounds, the allozyme data support the color pattern and karyological evidence for the current species status of P. axanthus and P. striatus (Martof, 1972; Moler and Kezer, 1993). This correspondence is consistent with their habitat differences in areas of sympatry, with P. axanthus usually found in open marshes in contrast to the cypress and gum swamps of P. striatus (Moler and Kezer, 1993). As illustrated by the contML solution, all three trees are consistent with the potential monophyly of both species (Figure 2-2). Their potential monophyly is bolstered by the biogeographic data for a vicariant origin of the two (see below). Furthermore, the two species are well differentiated according to their fixed allelic differences and genetic distances (Appendix B). In particular, the many fixed and nearly fixed allelic differences between their syntopic samples from Hernando County and Volusia County, Florida, reflect natural field experiments that indicate little to no gene flow between them (Bandoni et al., 2000). Despite their broadly overlapping ranges in north central Florida, P. axanthus and P. striatus are largely restricted to different geographic regions that continue south to the tip of peninsular Florida and north to South Carolina, respectively (Figure 2-1). This phylogeographic division is of general biogeographic significance, because it mirrors the geographic variation of other freshwater groups in the southeastern United States (e.g., Lepisosteus osseus [longnose gars] versus L. platyrhincus [Florida gars] [Hocutt and Wiley, 1986]; and two mud turtle subspecies, Kinosternon subrubrum, [Walker et al., 1998]). This biogeographic concordance has been attributed to marine transgressions in the late Miocene that isolated peninsular Florida (physically or ecologically) from the mainland (Gilbert et al., 1992). Correspondingly, Moler and Kezer (1993) argued that

PAGE 36

23 these marine transgressions were responsible for the speciation of P. axanthus and P. striatus. The allozyme data also allow for the recognition of western, southeastern, and northeastern phylogeographic groups within P. striatus (Figures. 2-1 and -2). These three groups are most strongly defined by their fixed or nearly fixed allelic differences at two loci and significant heterogeneity chi-square tests between the northeastern group and the other two (Appendix B). The western group is separated from the southeastern and northeastern ones by the Central Highland and Tifton/Vidalia Uplands, whereas the latter two are split by the Altamaha River (Figure 2-1). These geographic barriers are of known biogeographic importance to other freshwater groups in the southeastern United States (e.g., eastern mosquitofish, Gambusia holbrooki, [Wooten et al., 1988] and Fundulus auroguttatus versus F. rubrifrons, [Gilbert et al., 1992]). This biogeographic concordance with other freshwater taxa offers further support for the recognition of all three phylogeographic groups. A relatively large degree of genetic diversity exists among the samples of the western group, as evidenced by their long branches in the contML tree, their mean Nei distance of 0.094 (0.003-0.211), and their significant S and F with the southeastern unit (Figure 2-2). In particular, <6> and <7> are responsible for this greater diversity, because they differ from the other western samples by a mean Nei distance of 0.145 (0.059-0.211). For comparison, the mean Nei distance for samples of the southeastern group is 0.017 (0.004-0.034), whereas that between the two units is 0.237 (0.172-0.388). Increases in the within-group variation can be expected to obscure the differences that exist between groups, thereby making it more difficult to distinguish them (Weir and

PAGE 37

24 Hill, 2002). Thus, the inability to distinguish the southeastern and western groups with P can be attributed at least in part to the large genetic diversity within the latter. The relatively great divergence of <6> and <7> may be due to east-to-west introgression from the southeastern group via the Suwanee River, particularly during heavy flooding (Figure 2-1). This possibility is suggested by the increased heterozygosities and polymorphisms for the two western samples (see above) and by their unique alleles at Ada-A, Ldh-A, and mMdh-A (Appendix B). In hybrid populations, unique alleles are often found at fixed or high frequencies due to recombination, disrupted gene complexes, or local variation (Bradley et al., 1993). Thus, the Central Highland may not be a perfect barrier to gene flow between western and southeastern populations of P. striatus. However, such east-to-west introgression is probably rare, given that <6> and <7> are fixed for the same diagnostic alleles of mSod-A and sMdh-A that distinguish their western group from the southeastern one. These taxonomic and phylogeographic conclusions now allow for an evolutionary evaluation of the intraspecific taxonomy of P. striatus (Martof, 1972; Moler and Kezer, 1993). The western group corresponds well with P. s. spheniscus, with its holotype and 20 of 22 paratypes collected from within this phylogeographic region (Goin and Crenshaw, 1949). The western group also extends south along the Gulf Coast at least as far as Hernando County, Florida (Figure 2-1). Thus, this group also encompasses the reported range of P. s. lustricolus (Martof, 1972). However, all of our samples from this area conform to the description of P. s. spheniscus and differ markedly from that of P. s. lustricolus. Pseudobranchus s. lustricolus is based on a series of specimens that were

PAGE 38

25 collected in 1950 and 1951 from two localities near <5> of west central Florida. Since then, this subspecies has not been collected, and its taxonomic status remains uncertain. Traditionally, populations from the Atlantic coastal lowlands of north central Florida and adjacent Georgia have been assigned to either P. axanthus or P. s. striatus (Figure 2-1). Martof (1972) suggested that P. axanthus extends north to the Okefenokee Basin of southeastern Georgia and considered all Pseudobranchus from coastal Georgia further north to be P. s. striatus. However, he provided no records for P. striatus between the Okefenokee Basin and Altamaha River. In contrast, Moler and Kezer (1993) determined that both species occur within the Atlantic coastal lowlands of northern Florida, but that P. axanthus does not occur north beyond the Oklawaha River and Saint Johns River drainages. Our allozyme results agree with Moler and Kezer (1993) that some populations of Pseudobranchus from the Atlantic coastal lowlands of north central Florida and adjacent Georgia belong to P. striatus, with the southeastern group thereby representing a previously unrecognized phylogeographic unit of this species (Figure 2-2). In turn, our northeastern group from coastal Georgia north of the Altamaha River is consistent with P. s. striatus (Martof, 1972). The type locality for P. s. striatus is in this area and our representatives of northeastern sample <17> conform to its description. Nevertheless, this conclusion remains particularly tentative, because it is based on only a single sample (<17>).

PAGE 39

CHAPTER 3 PHYLOGEOGRAPHY OF THE SALAMANDERS, Pseudobranchus, AND OTHER ENDEMIC GROUPS IN THE SOUTHEASTERN UNITED STATES The regional, comparative, or landscape approach to phylogeography is concerned with the search for concordant patterns of geographic variation among the populations and closely related species of different sympatric groups (Avise, 2000b). Such concordant patterns of geographic variation are of general interest, because they reflect the shared historical factors that have shaped the distributions and divergence of entire assemblages. The power of this regional approach to phylogeography has been well illustrated with sympatric freshwater and terrestrial groups in the southeastern United States (US) (Hocutt and Wiley, 1986; Walker and Avise, 1998). Three major sub-regions have been recognized for aquatic groups in this region, with their geographic breaks corresponding to the Altamaha and Apalachicola river drainages along the Atlantic and Gulf coasts, respectively (Figure 3-1). Recurrent breaks at these river drainages have also been found for many terrestrial taxa in this region (e.g., within Geomys pinetis, southeastern pocket gopher, Avise et al., 1979). Thus, the Altamaha and Apalachicola river drainages are of major importance to the regional phylogeography of the southeastern US. During interglacial periods of the late Pliocene and Pleistocene, raised sea levels led to the periodic flooding of these two low-lying drainages and thereby to the observed phylogeographic splits between their surrounding communities (Colquhoun, 1995; Otvos, 1995). 26

PAGE 40

27 Figure 3-1. Maps of some endemic groups in the southeastern United States. (A) Major ecological sub-regions as recognized for aquatic groups in the southeastern US (Maxwell et al., 1995). These sub-regions are divided by the Altamaha River Drainage along the Atlantic Coast and by the Apalachicola River Drainage along the Gulf Coast. (B) Distributions for three sister species of Procambarus (crayfishes) illustrating the Central Highland and Tifton/Vidalia uplands, as well as the Altamaha River Drainage, as major phylogeographic breakpoints in this region (Hobbs, 1942). (C) Distributions for the Fundulus notti complex (starhead topminnows) and Rana virgatipes (carpenter frog) illustrating two lowland costal groups whose ranges terminate at the Central Highland and Tifton/Vidalia uplands (Lee et al., 1980; Conant and Collins, 1998). The grey shading in these three maps highlights the Central Highland and Tifton/Vidalia uplands that offer little to no suitable habitat for P. striatus.

PAGE 41

28 The family Sirenidae consists of Pseudobranchus (dwarf sirens) and Siren (sirens), two genera of eel-like, paedomorphic, freshwater salamanders (Martof, 1972). Pseudobranchus includes the recognized species P. axanthus of central and southern Florida and P. striatus of central and northern Florida, southern Georgia, and southernmost South Carolina (Figure 3-2). These two species are found in ponds, marshes, and swamps below 100 m elevation in the lower coastal plains of the southeastern US. Recognition of P. axanthus and P. striatus as valid species is based on karyological and allozyme differences, which support their separate phylogeographic histories, reproductive isolation, and genetic divergence (Moler and Kezer, 1993; Liu et al., 2004). Three phylogeographic groups of P. striatus have also been recognized from allozyme data, but their phylogenetic associations remain ambiguous. This study offers new insights into the phylogeography of the southeastern US by providing an extensive data set of mitochondrial DNA (mtDNA) sequences for the cytochrome b gene (cyt b) of Pseudobranchus and Siren. These new sequences support a well-defined phylogeny for Pseudobranchus that implicates the Central Highland and Tifton/Vidalia uplands as another significant phylogeographic barrier in addition to the Altamaha and Apalachicola river drainages. These results raise the question of whether P. striatus from west and east of these uplands represent distinct species, a possibility that is addressed with the concordance principles for species recognition (Avise and Ball, 1990). Materials and Methods A total of 305 specimens representing 39 samples of P. axanthus and P. striatus was collected from 36 localities from throughout their entire ranges (Figure 3-2). These specimens were assigned to species based on their color patterns, allozyme markers, and

PAGE 42

29 Figure 3-2. Map of the southeastern US for the 39 samples of Pseudobranchus and 3 samples of Siren. These 42 samples are referenced throughout by their numbers and are color coded according to their species and the phylogeographic units recognized in this study (see text). The gray shading highlights the Central Highland and Tifton/Vidalia uplands that provide little to no suitable habitat for P. striatus. “Other W samples” refers to those of the W Group for P. striatus without a well-defined subgroup assignment (Figure 3-3). Full locality data for these 42 samples are provided in Appendix C.

PAGE 43

30 chromosome counts (electrophoretic and chromosomal data were available for 26 and 21 of the 39 samples, respectively). In addition, nine specimens of S. intermedia and S. lacertina (the outgroup) were collected from three Florida localities. Blood, liver, and/or skeletal muscle was obtained from each specimen. Representative vouchers were deposited in the herpetology collection at the Florida Museum of Natural History, University of Florida (Appendix C). Total genomic DNA was isolated from the tissues with either standard phenol/chloroform methodology (Hillis et al., 1996) or the DNeasy Tissue Kit (QIAGEN Company). An internal fragment of the cyt b coding sequence was then amplified from these DNA isolates with the polymerase chain reaction (PCR), with their amplified products purified with the QIAquick PCR Purification Kit (QIAGEN Company) or with Ultrafree-MC tubes (Millipore Corporation). The purified PCR products were then cycle sequenced at the University of Florida DNA Sequencing Core Laboratory with ABI Prism BigDye Terminator protocols (part number 4303237, Applied Biosystems/Perkin-Elmer Corporation). The fluorescent-labeled extension products were analyzed on a Model 373 Stretch DNA Sequencer (Applied Biosystems), 377 DNA Sequencer, or 3100/3100-Avant Genetic Analyzer (Perkin-Elmer Corporation). In brief, PCR amplifications and cycle sequencing were first conducted with universal cyt b primers (Moritz et al., 1992) and then with sirenid-specific ones derived from their preliminary data (Appendix D). The PCR thermal conditions were 1 cycle at 94C for 10 minutes; 1 cycle at 43C for 100 seconds; 46 cycles at 94C for 1.5 minutes, 43C for 1.5 minutes, and 72C for 2 minutes; and 1 cycle at 72C for 10 minutes, with final storage at 4C until purification. Otherwise, all DNA isolations, amplifications, and

PAGE 44

31 sequencing followed the manufacturers’ recommendations for their kits and products. These procedures included the use of both positive and negative controls in the PCR amplifications and the automated sequencing of each individual’s haplotype in both directions (i.e., on both strands). A multiple alignment was generated for the final cyt b sequences with Sequencer, version 4.2 (Gene Codes Corporation), and then checked against the inferred reading frame for its protein and known tertiary structure for the bovine CYT B subunit (Gao et al., 2003). The HKY + + I model for unequal transition versus transversion rates and rate heterogeneity among sites according to the gamma distribution with some proportion of invariable positions was chosen as most appropriate for the cyt b sequences by their series of likelihood ratio tests with MODELTEST, version 3.06 (Hasegawa et al., 1985; Posada and Crandall, 1998). This series of likelihood ratio tests followed the hierarchy of simple to complex models recommended by the MODELTEST authors. Correspondingly, inter-sample divergences (d) and intra-sample nucleotide diversities () were calculated for the 42 sirenid samples with the HKY + + I distances for their haplotype pairs (Nei, 1987). In these calculations, the gamma parameter and proportion of invariable sites were set at their MODELTEST values of = 1.298 and p = 0.506, respectively. Phylogenetic trees were then estimated with the multiple alignment by both equal and unequal weighted parsimony, minimum evolution with the HKY + + I distances, fast maximum likelihood (ML), and Bayesian phylogenetics (Swofford et al., 1996; Huelsenbeck et al., 2001; Guindon and Gascuel, 2003). These phylogenetic analyses were conducted with and without Siren to evaluate the effects of this divergent outgroup

PAGE 45

32 on the Pseudobranchus phylogenies. The parsimony and distance analyses were accomplished with PAUP*, version 4.0b (Swofford, 2002), the fast ML comparisons with PHYML, version 2.1 b1 (Guindon and Gascuel, 2003), and the Bayesian analyses with MRBAYES, version 3.0 b4 (Ronquist and Huelsenbeck, 2003). In the equally and unequally weighted parsimony analyses, first and second codon positions were weighted 1, 2, 3, 4, or 6 times more than third sites. The fast ML and Bayesian analyses relied on the HKY + + I model, but with the gamma parameter () and the proportion of invariable sites (p) estimated rather than fixed a priori. Optimal phylogenies were recovered in the parsimony, distance, and fast ML analyses by heuristic searches with 1,000 random starting trees and tree-bisection-and-reconnection branch swapping. The strength of support for each recovered group was evaluated with 1,000 bootstrap samples. The Bayesian analyses consisted of eight independent trials, with each relying on one cold and three incrementally heated chains that were started from different random trees and were swapped every 100 generations (T = 0.2). Each chain was run for 2,000,000 generations, with a tree sampled from the cold chain every 100 generations following a burn-in of 2,000 or 2,500 trees when analyzed without and with the outgroup, respectively. The final samples of trees were summarized by their maximum a posteriori (MAP) phylogenies along with the posterior probabilities for their recovered groups. Results and Discussion Cytochrome b Sequences A total of 166 distinct haplotypes was obtained for the 305 individuals of the 39 samples for Pseudobranchus and seven for the 9 specimens of the three samples for Siren (Appendices E and F). The final multiple alignment for these 173 haplotypes covered

PAGE 46

33 786 base pairs that corresponded to positions 16,267 to 17,052 of the complete mtDNA genome for the protein-coding region of cyt b for Xenopus laevis (African clawed frog, Roe et al., 1985). This alignment required only a single triplet of adjacent gaps for one missing codon of a single P. striatus individual (haplotype 99). Phylogeographic Units In terms of its major units, the same unrooted phylogeny for Pseudobranchus was recovered by the various parsimony, distance, ML, and Bayesian analyses. As illustrated by the optimal fast ML solution, this single unrooted phylogeny supported the division of P. axanthus and P. striatus and the recognition of the same three phylogeographic groups for P. striatus as identified in the earlier allozyme study (Figures 2-2 and 3-3) (Liu et al., 2004). As before, the Western (W) Group of P. striatus consisted of samples from west of the Central Highland and Tifton/Vidalia uplands (Figure 3-2). Correspondingly, the Northeastern (NE) and Southeastern (SE) groups of this species included samples from east of the Central Highland and Tifton/Vidalia uplands, but from within and north of versus south of the Altamaha River Drainage, respectively. These species and phylogeographic groups were all consistently defined by strong bootstrap scores and posterior probabilities of > 96%. As illustrated by the unrooted fast ML phylogeny, the different phylogenetic analyses were also consistent with the recognition of new phylogeographic units within both species (Figures 3-2 and -3). Two new phylogeographic units of P. axanthus were recognized from the southern end of its distribution in peninsular Florida versus further north into central Florida (its Southern [S] versus Northern [N] groups, respectively). Two new phylogeographic units were also identified within the W Group of P. striatus from west of versus within the Apalachicola River Drainage (its Westwestern [WW]

PAGE 47

Figure 3-3. Unrooted fast ML phylogeny for the 39 Pseudobranchus samples. Each triangle summarizes the associated haplotypes of a recognized phylogeographic unit as color coded to Figure 3-2. The height and basal width of each triangle are drawn proportional to the average total branch length of its haplotypes to their most recent common ancestor and to its number of included sequences, respectively. The included haplotypes and associated samples of each triangle are identified by their numbers before and after the slashes in the parentheses, which follow those of Appendix E and Figure 3-2. In numerical order, the 173 distinct haplotypes are available in GenBank upon accession numbers AY713120 to AY713292 (Benson et al., 2004), whereas the original optimal phylogeny upon which this fastML summary is based is provided in TreeBASE (Piel et al., 2001). The arrow points to the bold internal branch of the fast ML solution to which the Siren outgroup joins when included in this analysis. Bootstrap scores are given above and below the internal branches for the fast ML analyses without and with this outgroup, respectively. Although the Other W samples cluster here, this potential subgroup is not considered further, since it was never supported by bootstrap scores or posterior probabilities of > 58% and was not recovered by the MAP phylogenies of the Bayesian analyses. In contrast, as highlighted by the asterisk, haplotype 166 of the possible interspecific hybrid consistently joins types 35-38 of sample 13 for the W Group of P. striatus at bootstrap scores and posterior probabilities of 100% (see text).

PAGE 48

35

PAGE 49

36 versus Northwestern [NW] subgroups, respectively). As before, these new units were consistently defined by strong bootstrap scores and posterior probabilities of > 99%,except for the relatively strong to strong marks of 82% to 100% for the N Group of P. axanthus. Furthermore, in contrast to the ambiguous allozyme results, a connection between the NE and SE groups of P. striatus was also indicated by the different phylogenetic approaches (Figure 3-3). This association was consistently defined by strong bootstrap and posterior probability scores of > 92%, except for those of the unequally weighted parsimony analyses. For the latter, bootstrap scores for this association decreased from 86% to 56% as the weighting scheme was increased from 2:1 to 6:1. Inter-sample divergences (d’s) for P. axanthus versus P. striatus (mean, D = 0.271; range, R = 0.200-0.408) overlapped broadly with those for the W versus NE groups (D = 0.222; R = 0.200-0.267) and W versus SE groups (D = 0.258; R = 0.223-0.325) of the latter species (Figure 3-4). In contrast, the d’s for the NE versus SE groups of P. striatus (D = 0.138; R = 0.122-0.148) did not overlap with those from above for its W Group and the interspecific comparison. The mean d between the NW and WW subgroups of the W Group for P. striatus was 0.114 (R = 0.110-0.118), whereas that between the N and S groups of P axanthus was least for the major phylogeographic units of Pseudobranchus (D = 0.057; R = 0.040-0.069).

PAGE 50

37 Figure 3-4. Summary of the inter-sample divergences (d’s) for the major groups and subgroups of Pseudobranchus and Siren. This diagram highlights how sequence divergence for cyt b varies across the genera, species, and phylogeographic units of this study (Johns and Avise, 1998). The mean, standard deviation, and range for each comparison are indicated by the dot, double bars, and arrows, respectively, whereas its number of d’s is given in parentheses. The complete d matrix for all sample pairs is provided in Appendix G.

PAGE 51

38 Rooting the Pseudobranchus Phylogeny In contrast to their uniform strong support for specific phylogeographic units, the different parsimony, distance, ML, and Bayesian analyses failed to recover a single well-defined location for the root of the Pseudobranchus phylogeny. Instead, different positions within the same unrooted phylogeny for the major phylogeographic units were suggested for the Siren outgroup by these various phylogenetic approaches. The fast ML analyses assigned the root to the internal branch leading to the SE Group of P. striatus (Figure 3-3). In contrast, the other approaches placed the root at three alternative locations, with only the 4:1and 6:1-weighted parsimony analyses joining the outgroup to the internal branch between P. axanthus and P. striatus. Such lack of support for this location was surprising given the allozyme, karyological, phylogeographic, and ecological evidence for their full species status (Moler and Kezer, 1993; Liu et al., 2004). However, none of these alternative optimal placements for the root was ever supported by bootstrap scores or posterior probabilities of > 58% for both of its resultant basal lineages. For example, in the fast ML rooting, the resultant basal lineage of P. axanthus and the NE and W groups of P. striatus was defined by a bootstrap score of only 29% (Figure 3-3). Thus, as in the allozyme study (Liu et al., 2004), the assignment of the most basal node in the cyt b phylogeny of Pseudobranchus proved ambiguous and this tree was therefore left unrooted. Given their mean and maximum d’s of 0.377 and 0.457, respectively, the cyt b sequences for Pseudobranchus and Siren differed on average by more than one substitution for every three positions and sometimes by almost more than one for every two sites (Figure 3-4). Thus, this failure to root the cyt b phylogeny was very likely due to the great divergence between Pseudobranchus and

PAGE 52

39 Siren. In conclusion, what is now needed to root the Pseudobranchus phylogeny are new comparative data for more conserved characters of the study group and outgroup. Within-Locality Variation The cyt b haplotypes for all but one Pseudobranchus were consistent with their species assignments based on their allozyme, color pattern, and/or karyological characters. This consistency included 211 specimens for which allozyme genotypes were available (Liu et al., 2004). Indeed, 21 of these 211 were from two localities in central Florida where the N Group of P. axanthus occurs syntopically with either the SE (samples 35 versus 27) or W (33 versus 13) groups of P. striatus (Figure 3-2). The single inconsistent individual carried a cyt b haplotype for the W Group of P. striatus (type 166), which conflicted with its original allozyme and color pattern assignment to syntopic sample 33 of P. axanthus (Figure 3-3). Thus, this specimen may indicate a rare interspecific hybridization event between the two species. Haplotypes with separate associations to different samples and/or phylogeographic units also occurred in samples 36 of P. axanthus and 8 of P. striatus (Appendix F). Sample 36, from the southern edge of the N Group for P. axanthus, included two individuals with a haplotype (152) that was allied to types 153-165 of its S counterpart (Figures 3-2 and -3). Furthermore, sample 8 was a mixture of haplotypes with at least two separate histories relative to those of the other W samples for P. striatus (types 18 and 19 versus 20-23). However, none of these distinct haplotypes for sample 8 was closely related to any of those of the SE Group from across the Central Highland, in contrast to the possibility of east-to-west introgression as proposed in the allozyme study (Liu et al., 2004).

PAGE 53

40 Intra-sample nucleotide diversity () was <0.010 for all but three of the sirenid samples ( = 0.015, 0.017, and 0.026 for samples 23, 8, and 36, respectively) (Appendix G). As noted above, the large for sample 36 was due to its mixture of divergent haplotypes typical of the N and S groups for P. axanthus. Mean ’s were 0.006 (standard deviation, SD = 0.007; R = 0.000-0.026) for P. axanthus; 0.005 (SD = 0.004; R = 0.000-0.017) for P. striatus; 0.002 for S. intermedia; and 0.002 for S. lacertina. Thus, unlike the heterozygosity and polymorphism estimates for the allozyme data (Liu et al., 2004), these measures of within-sample variation are not greater for P. striatus relative to P. axanthus and S. intermedia. Regional Phylogeography Although unrooted, the well-supported cyt b phylogeny for Pseudobranchus still allows for the recognition of divisions that are of regional phylogeographic importance to the southeastern US (Avise, 2000b). Specifically, the divisions between the NE versus SE groups of P. striatus and among its WW Subgroup, NW Subgroup, and “other W samples” reconfirm the Altamaha and Apalachicola river drainages as major phylogeographic breakpoints along the Atlantic and Gulf coasts, respectively (Figures 3-1 ~ -3). Correspondingly, these divisions reinforce the importance of rising sea levels and periodic marine flooding of these river valleys during interglacial periods of the late Pliocene and Pleistocene to the regional phylogeography of this area (Hocutt and Wiley, 1986; Webb, 1990). Marine transgressions during an especially high-water stand in the early Pliocene have also been implicated in the south-to-north speciation between P. axanthus versus P. striatus (Moler and Kezer, 1993; Liu et al., 2004) and oceanic

PAGE 54

41 intrusions in the late Pleistocene may also underlie the split between the N and S groups of P. axanthus. The W versus NE/SE groups of P. striatus are divided by the Central Highland and Tifton/Vidalia uplands (Figures 3-2 and -3). This distinct west-to-east split within P. striatus is concordant with the phylogeographic divisions reported for other coastal lowland groups of the southeastern US (e.g., within the crayfish genus Procambarus, Figure 3-1). Furthermore, the Central Highland and Tifton/Vidalia uplands mark the western or eastern limits of the distributions for many lowland species found along either the Atlantic or Gulf coasts (e.g., the Fundulus notti complex, starhead topminnows; Rana virgatipes, carpenter frog). Such recurrent divisions and breaks among diverse lowland taxa identify these highlands as an additional phylogeographic barrier of importance to this region. Specifically, the Central Highland and Tifton/Vidalia uplands are implicated as a major phylogeographic barrier to groups that are restricted to the coastal lowlands of northern and central Florida and the adjacent mainland (Figures 3-1 and -2). For taxa that range into the uplands and/or further south into peninsular Florida, these highlands are of much less consequence as a phylogeographic barrier, since gene flow can occur in these groups across the ridge or around the southern edge. In support of this argument, a concordant west-to-east division is rarely seen for those sympatric groups with more wide-ranging distributions (Hocutt and Wiley, 1986; Walker and Avise, 1998). In conclusion, the Central Highland and Tifton/Vidalia uplands represent another important phylogeographic barrier of the southeastern US, but especially for its coastal lowland taxa north of southern Florida.

PAGE 55

42 Separate Species? Do the W versus NE/SE groups of P. striatus represent separate species? This possibility is raised by their distinct phylogeographic separation for cyt b, overlapping d’s with those for the confirmed P. axanthus and P. striatus, and nearly fixed allelic difference at the nuclear sMdh-A locus (Figures 3-2 ~ -4) (Liu et al., 2004). However, it is not corroborated by independent reproductive, karyological, and ecological evidence as for P. axanthus and P. striatus (Moler and Kezer, 1993). Thus, according to the concordance principles for species recognition (Avise and Ball, 1990), the W versus NE/SE groups do not warrant separate species status given the relative lack of independent data for their phylogenetic separation. This relative lack of independent evidence implies that any reproductive isolation between them is more likely due to extrinsic (geographic) factors rather than intrinsic ones. Following the concordance principles, we conclude that the W versus NE/SE groups of P. striatus do not merit separate species recognition at this time. Rather, as noted before for the rooting problem, we call for future studies of these phylogeographic groups with new nuclear genes and other independent internal characters.

PAGE 56

CHAPTER 4 GENETIC VARIATION WITHIN POPULATIONS AND SPECIES OF THE SALAMANDER Pseudobranchus: IMPACTS OF HISTORICAL OCEANIC FLOODING IN THE SOUTHEASTERN UNITED STATES The global cycling between glacial and interglacial periods over the last 5 million years has had a major impact on the terrestrial biota in temperate and subtropical North America (Cox and Moore, 2000). During the glacial periods, species and communities were often forced into refuges and/or the periphery of their distributions, thereby fragmenting them and limiting their numbers. In turn, during the alternate interglacial periods, rising sea levels often resulted in marine transgressions that flooded coastal and lowland species and communities, thereby leading to their fragmentation and population bottlenecking/extinction. These fragmentations and losses have led to the decreased levels of genetic variation that are commonly observed in the gene pools of contemporary species (Avise, 2000a, b). These decreases in genetic variation due to drift are evident even when current population sizes have rebounded and even for nuclear DNA (nDNA) as well as mitochondrial DNA (mtDNA) data. The latter emphasis is noteworthy, given the generally four-times greater effective population sizes (Ne) for the former. The target organisms, salamanders of the genus Pseudobranchus, are eel-like in appearance, paedomorphic, and endemic to freshwater habitats in the coastal lowlands and adjacent plains of the southeastern United States (Martof, 1972). This genus consists of two species: P. axanthus of central and southern Florida and P. striatus of northern and central Florida, Georgia, and South Carolina (Figure 4-1). Available karyological, 43

PAGE 57

44 Figure 4-1. Maps of the southeastern United States highlighting those terrestrial regions >30 m and > 15 m above current sea levels and the distributions of the recognized phylogeographic groups within Pseudobranchus. Population samples are identified by their numbers in brackets, with detailed locality and other information available for them in Liu et al. (2005). These numbers are followed throughout this chapter.

PAGE 58

45 allozyme, and mitochondrial cytochrome b (cyt b) sequence data strongly support the validity of the two species and the further recognition of several intra-specific phylogeographic groups (Moler and Kezer, 1993; Liu et al., 2004, 2005). Speciation and divergence within this genus have been linked to increasing sea levels and periodic oceanic flooding during the interglacial periods since the late Miocene. This study complements the existing ones on the higher-level systematics and evolution of Pseudobranchus by focusing instead on the genetic variation within its populations and species to explain how historical events resulted in these phenomena. It relies on estimates of the ages and effective female population sizes (Nf) for both P. axanthus and P. striatus, as well as on more standard coefficients of within-group variation, to document that the former is a less diverse and “younger” species than is the latter. These different patterns of genetic variation between these two species are linked to the greater population bottlenecking and extinction within P. axanthus due to interglacial marine flooding since the late Miocene, as well as to the longer sorting times, i.e., time to coalescence, for nDNA versus mtDNA. Thus, interglacial oceanic flooding is once again invoked as a primary evolutionary force of terrestrial groups in the southeastern United States (Avise, 2000b). Methods Data Sets and Reference Phylogenies The allozyme and mtDNA data sets for this study were obtained from Liu et al. (2004 and 2005, respectively). The former included the genotype frequencies for 14 presumptive nuclear loci as scored from 238 specimens for 26 population samples of P. axanthus and P. striatus. Of these 14 loci, only one was invariable (i.e., fixed for the same allele for all population samples). The latter consisted of the multiple sequence

PAGE 59

46 alignment for the individual cyt b haplotypes of 305 specimens from 39 population samples of the two species. All 26 samples of the allozyme data set were included among the 39 populations of the mtDNA matrix. The 305 individual cyt b sequences represented 166 distinct haplotypes. Their multiple sequence alignment covered 786 positions corresponding to sites 16,267 to 17,052 of the complete mtDNA genome for Xenopus laevis (clawed frog, Roe et al., 1985). Similarly, the reference allozyme and mtDNA phylogenies for this study were derived from those of Liu et al. (2004: fig. 2 and 2005: fig. 3, respectively) with one modification. An important difference between the original phylogenies for these two data sets was that the Northeastern (NE) Group of P. striatus was connected to its Western (W) Group in the allozyme phylogeny, rather than to its Southeastern (SE) Group as in the mtDNA tree. Otherwise, the phylogenetic arrangements for both were consistent with the split of P. axanthus versus P. striatus, the separation of the former into Northern (N) and Southern (S) Groups, and the subdivision of the latter into NE, Southeastern (SE), and W Groups (Figures 4-2 and 4-3). Given the much stronger bootstrap and posterior probability scores for the mtDNA results, the reference allozyme phylogeny was correspondingly changed to reflect the union of the NE and SE Groups for P. striatus. Liu et al. (2004, 2005) presented their phylogenies as unrooted trees. To root their phylogenies, comparable allozyme and mtDNA data were directly obtained from them for the other paedomorphic genus of Sirenidae (i.e., Siren intermedia and/or S. lacertina) and then incorporated into the two data sets for Pseudobranchus. Given the multiple lines of evidence for their reciprocal monophyly (i.e., from karyological, molecular,

PAGE 60

47 Figure 4-2. Reference allozyme phylogeny for the 26 population samples of Pseudobranchus, as rooted against Siren (not shown). Numbers identify these population samples (Figure 4-1), whereas “A” and “S” denote the most recent common ancestors of P. axanthus and P. striatus, respectively.

PAGE 61

48 reproductive, and biogeographic data), the unrooted allozyme and mtDNA phylogenies were then rooted by joining the Siren outgroup to the long internode interconnecting P. axanthus and P. striatus (Moler and Kezer, 1993; Liu et al., 2004, 2005). Branch lengths were then recalculated for these now rooted trees as described in Liu et al. (2004, 2005). Figure 4-3. Reference mtDNA phylogeny for the 305 individual haplotypes of cyt b for Pseudobranchus. Each terminal branch corresponds to a distinct haplotype, with numerals specifying the numbers of individuals for those with multiple representatives. All other details about this phylogeny follow Figure 4-2.

PAGE 62

49 Figure 4-3. Continued. Within-Population Diversities and F-statistics For the allozyme data, direct count and expected heterozygosities (Hd and He, with their standard deviations [SD], respectively) and polymorphisms (P) were directly obtained from Liu et al. (2004). For the mtDNA data, nucleotide and haplotype diversities ( and h, along with their SD for the former) were directly acquired from Liu et al. (2005) and were calculated by this study according to Nei (1978) for the 39 population samples, respectively. In the estimation of , sequence divergence between haplotype pairs was quantified with the HKY+I+ distance that accounted for different

PAGE 63

50 transition versus transversion rates and rate heterogeneity among sites according to the gamma distribution with some proportion of invariable positions (Hasegawa et al., 1985; Swofford et al., 1996). For the allozyme data, Hd, He, and P were each statistically compared between species in two different ways. First, each was directly compared between P. axanthus and P. striatus with the Wilcoxon two-sample test (Sokal and Rohlf, 1995). However, as this test does not account for phylogenetic covariation, a second, tree-based, permutation approach was also conducted with MacClade, version 4.07 (Maddison and Maddison 2000). In this second approach, these coefficients were each separately estimated for the most recent common ancestors (MRCA) of both species, first with the reference allozyme phylogeny (Figure 4-2) and then with 100 permutated phylogenies with their intraspecific branching orders randomly rearranged. The latter sets of paired ancestral estimates were then statistically compared across species with both the sign and Wilcoxon signed-ranks tests (Sokal and Rohlf, 1995). For the mtDNA data, h and were also evaluated between species with the Wilcoxon two-sample test. However, as neither was significant according to these evaluations, no follow-up, ancestral, permutation tests were conducted for them. For the allozyme data, F-statistics were directly obtained from Liu et al. (2004) as generated by them with GDA, version 1.1 (Lewis and Zaykin, 2002). As GDA requires multiple samples per unit at each hierarchical level to calculate confidence intervals, only the F-statistics for their two-tiered analyses (for populations and species) were considered in this study with one exception. This single exception involved their three-tiered comparison for P. striatus, where it was possible to include a third level for the multiple

PAGE 64

51 samples of its W and SE Groups. In each analysis, 95% confidence intervals were generated according to 1,000 bootstrap replicates over loci. For the mtDNA data, F-statistics were also calculated by this study with GDA. Given that mtDNA is haploid, these F-tests focused on two hierarchical levels and the estimation of coancestry coefficients ( P or F ST ). To evaluate the reliability of their P , jackknifed means and SD were also calculated for this coefficient by removing sequentially each population sample from its group. Dating the MRCA For both the allozyme and mtDNA data, absolute ages for the MRCA of P. axanthus and P. striatus were estimated with their calibration point set at 5.2 million years ago (Ma). This calibration point of 5.2 Ma corresponded to the accepted point estimate used to represent the time of speciation for the two species as established from both paleogeographic and biogeographic information (Haq et al., 1987; Moler and Kezer, 1993; Liu et al., 2004, 2005). These age estimates of P. axanthus and P. striatus were conducted in two different ways. In the first, ages were calculated according to the familiar approach of expressing the mean inter-population distance between the two descendant lineages of the MRCA in question as a proportion of the average rate for their calibration point (Figures 4-2 and -3). Here, these distance-based dates were calculated for the allozyme and mtDNA data from the standard genetic distances (Nei, 1978) and total inter-population / HKY+I+ divergences (Hasegawa et al., 1985; Nei, 1987), as obtained from Liu et al. (2004, 2005). In the second, ages were then estimated with the two reference phylogenies and their branch lengths by semiparametric smoothing with R8S, version 1.5 (Sanderson, 2002).

PAGE 65

52 In this penalized likelihood approach, evolutionary rates were allowed to vary across lineages, but their numbers were limited by the fact that each rate shift imposed a cost which decreased its probability. For the mtDNA data, absolute times for the MRCA were also estimated with the Markov chain Monte Carlo (MCMC) approach with Bayesian method of MDIV (Nielsen and Wakeley, 2001). In these calculations, the original cyt b sequences were analyzed two populations at a time under a Bayesian framework with a coalescent/finite-sites-mutation model. This model accounted for different transition/transversion rates with the HKY process and for ongoing migration between the two populations since their divergence. For each population pair, estimates of their mutation and migration rates were provided as compound parameters that were scaled by 2Nf (i.e., = 2Nf and M = Nf m, where = mutations per site per generation, and m = migrants per female per generation, respectively). In turn, an estimate of their divergence time was provided as the compound parameter T (where T = t / Nf and t = number of generations since their MRCA). All population pairs from across the two descendant lineages of each MRCA were compared to generate means and SD for the , T, and M of their species (Figure 4-3). To convert the average T into absolute time (Ma), the same calculations were also conducted for their calibration point of P. axanthus versus P. striatus. Assuming a constant mutation rate () for the two species, absolute ages for their MRCA were then derived from the following ratios of their means: T (either species) / T (calibration) = t (either species) / t (calibration) = Ma (either species) / 5.2 Ma (calibration).

PAGE 66

53 In this way, absolute ages for the MRCA were obtained without having to specify the number of absolute years per generation for Pseudobranchus. Furthermore, relative estimates of Nf for the two species were generated from the following ratio of their means: (P. axanthus) / (P. striatus) = Nf (P. axanthus) / Nf (P. striatus). Once again, a constant was assumed in these calculations. Because of its growing popularity, this study also initially considered the use of the Bayesian/MCMC dating method of Mulitdivtime, version 1.3 for the analysis of the mtDNA data (Thorne et al., 1998). However, as emphasized by its authors, this approach is not designed for intraspecific data and so this possibility was not pursued further (Kishino et al., 2001). Results For the allozyme data, mean Hd, He, and P were greater for P. striatus than for P. axanthus (Table 4-1). For example, mean Hd for P. striatus (0.068) was more than 2.5 times greater than the average for P. axanthus (0.027). This difference was significant according to its Wilcoxon two-sample test (Us = 128.5, P < 0.006). In the ancestral reconstructions, Hd was estimated as 0.024 and 0.045 for the MRCA of P. axanthus and P. striatus, respectively. In their ancestral permutation test, the Hd reconstructions for P. striatus were greater than those for P. axanthus in 98 of the 100 paired trials. Again, this difference was significant according to its sign test (P ~ 0.000), as well as its Wilcoxon signed-ranks test (Ts = 9.0, P ~ 0.000).

PAGE 67

Table 4-1. Within-population diversity estimates for P. axanthus and P. striatus. Standard deviations are given in parentheses, whereas “N” refers to the number of individuals. “-” refers to those cases where the sample size is too small for the estimation. 54 Sample Allozyme MtDNA number N Hd He P N h P. axanthus N Group 28 24 0.008 (0.006) 0.022 (0.018) 0.071 19 0.010 (0.007) 0.965 29 0 0.001(NA) 2 1.000 30 8 0.018 (0.012) 0.032 (0.024) 0.143 7 0.004 (0.003) 0.857 31 10 0.021 (0.015) 0.039 (0.032) 0.143 10 0.004 (0.005) 0.822 32 4 0.000 (0.000) 0.000 (0.000) 0.000 3 0.001 (0.001) 0.667 33 11 0.055 (0.022) 0.090 (0.040) 0.286 10 0.002 (0.001) 0.867 34 2 0.071 (0.049) 0.167 (0.074) 0.286 2 0.000 (NA) 0.000 35 3 0.000 (0.000) 0.000 (0.000) 0.000 14 0.006 (0.004) 0.923 36 8 0.036 (0.036) 0.038 (0.038) 0.071 7 0.026 (0.024) 0.905 S Group 37 20 0.037 (0.037) 0.036 (0.036) 0.071 21 0.003 (0.004) 0.686 38 0 0.006 (0.004) 4 0.833 39 0 0.007 (0.004) 10 0.889 Species means 0.027 (0.025) 0.047 (0.052) 0.119 (0.107) 0.006 (0.007) 0.784 (0.266) P. striatus NE Group 14 0 1 15 0 0.004 (0.004) 10 0.533 16 0 0.005 (0.002) 4 1.000 17 15 0.059 (0.046) 0.076 (0.038) 0.357 18 0.008 (0.004) 0.948 18 0 0.004 (0.003) 9 0.889 SE Group 19 7 0.133 (0.069) 0.104 (0.049) 0.357 13 0.003 (0.002) 0.808 20 0 0.007 (0.003) 4 0.833 21 0 0.008 (0.003) 10 0.978 22 12 0.071 (0.044) 0.072 (0.045) 0.214 9 0.008 (0.003) 1.000 23 0 0.015(NA) 2 1.000

PAGE 68

Table 4-1. Continued. 55 Sample Allozyme MtDNA number N Hd He P N h SE Group 24 1 0.071 (0.071) 0.071 (0.071) 0.071 1 25 20 0.093 (0.047) 0.119 (0.055) 0.286 16 0.004 (0.006) 0.825 26 8 0.039 (0.022) 0.079 (0.038) 0.286 6 0.000 (0.001) 0.333 27 8 0.009 (0.009) 0.009 (0.009) 0.071 5 0.001 (0.001) 0.600 W Group 1 0 5 0.000 (0.000) 0.000 2 4 0.054 (0.039) 0.059 (0.043) 0.143 4 0.005 (0.002) 1.000 3 0 6 0.002 (0.001) 0.733 4 0 6 0.005 (0.004) 0.733 5 6 0.012 (0.012) 0.034 (0.024) 0.143 4 0.001 (0.001) 0.500 6 16 0.040 (0.020) 0.059 (0.032) 0.214 13 0.000 (0.000) 0.154 7 5 0.086 (0.050) 0.127 (0.058) 0.286 3 0.003 (0.001) 1.000 8 5 0.146 (0.074) 0.150 (0.062) 0.357 10 0.017 (0.017) 0.844 9 1 0.071 (0.071) 0.071 (0.071) 0.071 1 10 12 0.054 (0.032) 0.075 (0.047) 0.143 12 0.006 (0.005) 0.879 11 17 0.063 (0.042) 0.069 (0.043) 0.143 15 0.001 (0.001) 0.552 12 3 0.095 (0.054) 0.105 (0.058) 0.214 3 0.000 (0.000) 0.000 13 5 0.057 (0.044) 0.048 (0.035) 0.143 5 0.002 (0.001) 0.700 Species means 0.068 (0.036) 0.078 (0.035) 0.206 (0.101) 0.005 (0.004) 0.702 (0.312) S. intermedia 40 7 0.000 (0.000) 0.019 (0.019) 0.071 3 0.002 (0.001) 1.000 S. lacertina 41 0 0.002 (0.001) 5 0.800 42 0 1

PAGE 69

56 This pattern of significantly greater diversity within P. striatus was corroborated by the results for He and P (Table 4-1). In contrast, it was not supported by the mtDNA data. For example, mean h and were not significantly different between the two species (Wilcoxon two-sample test, Us = 162.5, P > 0.52 and Us = 155.5, P > 0.68, respectively), with the absolute average for being greater for P. axanthus than for P. striatus (0.006 versus 0.005 substitutions per site, respectively). Instead, the two species were both characterized by similarly high averages for h (0.78 and 0.70, respectively), in addition to Table 4-2. F-statistics for the allozyme and mtDNA data. These results are all based on two-level hierarchical analyses, except for the additional three-tiered comparison of the allozyme data for the SE versus W Groups of P. striatus (see text). For this three-tiered comparison, s = 0.656 (0.364-0.854). “N” refers to the number of populations in each analysis. For the allozyme data, 95% confidence intervals are given in parentheses as obtained from 1,000 bootstrap replicates over loci. For the mtDNA data, these parenthetical values represent instead the jackknifed means and SD for P . N F IS (f) F IT (F) F ST ( P ) Allozyme P. axanthus 9 0.313 (0.036-0.501) 0.663 (0.332-0.749) 0.510 (0.206-0.651) N Group 8 0.433 (0.172-0.533) 0.598 (0.372-0.688) 0.291 (0.134-0.550) P. striatus 17 0.184 (0.019-0.508) 0.671 (0.427-0.898) 0.597 (0.337-0.822) SE versus W Groups 16 0.160 (-0.012-0.442) 0.711 (0.432-0.904) 0.381 (-0.031-0.727) SE Group 6 0.149 (-0.086-0.505) 0.498 (0.065-0.833) 0.410 (0.057-0.779) W Group 10 0.194 (-0.000-0.462) 0.592 (0.387-0.839) 0.493 (0.316-0.705) MtDNA P. axanthus 12 0.156 (0.156 + 0.045) N Group 9 0.119 (0.108 + 0.031) S Group 3 0.186 (0.259 + 0.076) P. striatus 27 0.265 (0.264 + 0.064) NE Group 5 0.150 (0.138 + 0.109) SE Group 9 0.187 (0.180 + 0.066) W Group 13 0.376 (0.383 + 0.118)

PAGE 70

57 their comparably low means for . Taken together, these similarly high versus low means were reflective of population samples for both species carrying multiple distinct, but similar (closely related) haplotypes. For the allozyme data, all F-statistics for the two species were significantly greater than zero, except for the inbreeding coefficient (f) and P in the three-level hierarchical analysis of P. striatus (Table 4-2). For the mtDNA data, the same was true as P was insignificant only for the NE Group of P. striatus. Thus, significant overall population differentiation existed within the phylogeographic groups as well as within their species according to both data sets. Indeed, Liu et al. (2004) noted that the insignificant P for the allozyme data was related to the extensive population differentiation within the W Group of P. striatus that thereby obscured its between-group differences. For the allozyme data, the distance (Figure 4-4) and semiparametric smoothing analyses resulted in similar dates of 0.93 and 0.88 Ma for the MRCA of P. axanthus and 2.28 and 2.64 Ma for the MRCA of P. striatus, respectively (Table 4-3). For the mtDNA data, these analyses converged onto comparable times of 0.99 and 1.21 MA for P. axanthus and 4.69 and 4.89 Ma for P. striatus, respectively. Thus, P. striatus was at least 2.5 to 4.7 times older than P. axanthus according to these analyses. For the mtDNA data, the coalescent/finite sites model provided mean , T, and M of 5.37, 39.30, and 0.03 for P. axanthus, 5.65, 107.77, and 0.02 for P. striatus, and 5.81, 115.55, and 0.02 for both species, respectively. These parameter estimates led to absolute ages of 1.77 and 4.85 Ma for P. axanthus versus P. striatus, respectively, and a relative Nf of 0.97 for the former versus the later. Thus, P. striatus was again at least 2.5

PAGE 71

58 Figure 4-4. Summary of Nei genetic distances between all population pairs of the recognized phylogeographic groups. Solid dots, cross-marks, arrows, and values in parentheses refer to the means, standard deviations, ranges, and numbers of pairwise distances for each inter-group comparison, respectively. A comparable summary for the mtDNA data is presented in Liu et al. (2005: fig. 4).

PAGE 72

59 Table 4-3. Absolute ages for the MRCA of P. axanthus and P. striatus, as estimated from their allozyme and mtDNA data. Standard deviations are given in parentheses for those estimates representing means. MRCA Age of P. striatus P. axanthus P. striatus to P. axanthus MtDNA Mean distances 0.99 (0.25) 4.69 (0.56) 4.74 Semi-parametric smoothing 1.21 (0.86) 4.98 (0.28) 4.11 Coalescent 1.77 (0.12) 4.85 (0.61) 2.74 Allozyme Mean distances 0.93 (0.41) 2.28 (0.61) 2.45 Semi-parametric smoothing 0.88 (1.00) 2.64 (0.76) 3.00 times older than P. axanthus, in contrast to their nearly identical Nf and minimum migration between the population samples of their basal descendant lineages. Discussion Interglacial Oceanic Flooding Over the last 5 million years, interglacial marine transgressions have flooded the low-lying Altamaha and Appalachicola river valleys, thereby acting as geographic barriers for the fragmentation of their surrounding coastal and lowland communities (Figure 4-1). These historical barriers have been recognized as of primary importance to the divergence of the three major ecological sub-regions for aquatic groups within the southeastern United States, as well as the phylogeographic units within Pseudobranchus (Moler and Kezer, 1993; Maxwell et al., 1995; Liu et al., 2004, 2005). These acknowledged “barrier” effects are now extended to include the impact of general habitat reduction due to widespread oceanic flooding over the last 5 million years. Since the late Miocene, peninsular Florida has been periodically inundated during the interglacial periods, in contrast to the less extensive oceanic transgressions to the north. Such differences in marine flooding have resulted in more extensive population bottlenecking

PAGE 73

60 and extinction within P. axanthus than within P. striatus, as discussed below in light of their current allozyme and mtDNA data. Within-Population Diversity The allozyme diversity within P. striatus is significantly greater than that within P. axanthus (Table 4-1). This conclusion is supported by their statistical comparisons and ancestral permutation tests of Hd, He, and P. In particular, the ancestral permutation tests account for the phylogenetic covariation within each species, while documenting the greater predicted diversity of the MRCA for P. striatus. Furthermore, this conclusion is corroborated by their F-statistics, which show a significant heterozygote deficiency within P. axanthus but not within P. striatus (Table 4-2). The question is now whether the greater allozyme variation within P. striatus is unexpectedly high and/or whether the lower diversity within P. axanthus is surprisingly low for two paedomorphic species. This question is addressed with two different lines of evidence. The first involves the available allozyme data for their paedomorphic Siren outgroup (Liu et al. 2004). Although limited in terms of its population samples, mean Hd, He, and P for S.intermedia are even lower than the averages for P. axanthus (0.000, 0.019, and 0.071 versus 0.027, 0.047, and 0.119, respectively) (Table 4-1). The second line of evidence includes the general reviews of allozyme H and P for vertebrates (Nei and Graur, 1984; Shaffer and Breden, 1989; Nevo and Beiles, 1991). According to these reviews, H and P average ~0.06 and ~0.20 for vertebrates in general, ~0.073 and ~0.255 for typical amphibians, and ~0.050 and ~0.132 for non-transforming salamanders, respectively. Given these different baselines, the allozyme diversity for P. axanthus is seen as not particularly low, whereas that for P. striatus is viewed as unusually high for paedomorphic salamanders.

PAGE 74

61 Indeed, the mean Hd, He, and P for P. striatus overlap with the averages for typical vertebrates including transforming salamanders (0.068, 0.078, and 0.206, respectively). In contrast to this pattern of unequal allozyme diversity, the mtDNA data support one of comparable variation for the two species, whereby their populations consist of multiple distinct but recently derived haplotypes (Table 4-1). Although different, these two patterns are not necessarily inconsistent, since they are most likely the result of the unequal Ne for nDNA versus mtDNA. Assuming a 1:1 sex ratio, Ne is four times greater for nDNA than for mtDNA (Hedrick, 2005). Thus, during interglacial periods of population bottlenecking due to widespread marine flooding, more mtDNA than nDNA diversity was lost by drift from both P. axanthus and P. striatus, as indicated by their similar mean h and but different average Hd, He, and P. In this way, a more detailed (older) record of diversity was preserved for the two species in their allozyme data, but not in their mtDNA counterpart. Species Ages and Phylogeographic Groups The five alternative dates for the MRCA of P. axanthus range from 0.88 to 1.60 Ma, whereas those for the MRCA of P. striatus vary from 2.28 to 4.89 Ma (Table 4-3). Thus, the age estimates for both species vary by about a factor of two among the five different pairs of data sets and approaches. The allozyme dates are consistently less than those for the mtDNA data, particularly for the older estimates of P. striatus. Given that they are based on a fully probabilistic model for intraspecific sequences, the coalescent/finite sites estimates for the mtDNA data can be regarded as rigorous (Nielsen and Wakeley, 2001). Its estimate of 1.77 Ma for P. axanthus is the largest of the five for

PAGE 75

62 this species, whereas its date of 4.85 Ma for P. striatus falls in between those for the allozyme versus other mtDNA ages. Despite these discrepancies about exact dates, the genetic variation within P. axanthus is clearly younger or “shallower” than within P. striatus (Table 4-3). This conclusion is apparent from the direct visual inspection of their reference phylogenies (Figures 4-2 and -3) and is now confirmed by the different age estimates that all agree that the MRCA of P. axanthus is at least 2.5 times younger than the MRCA of P. striatus. This shallower variation is a direct reflection of the younger N and S Groups within P. axanthus versus the older W and NE/SE lineages within P. striatus. Thus, as indicated by its relative Nf of 0.97 with P. striatus, this younger variation is related to the direct loss of phylogeographic groups within P. axanthus, rather than to any reductions in Ne as implicated for their allozyme and mtDNA diversity. Specifically, it is tied to the greater extinction of lineages within P. axanthus during the widespread marine inundations of peninsular Florida since the late Miocene (Figure 4-1). Despite their uncertainties about exact dates, the different age estimates for the MRCA of P. axanthus and P. striatus are sufficiently narrow to implicate two separate high-water stands at ~3.5 and ~1.5 Ma as potentially crucial to the establishment of their standing genetic diversities (Table 4-3). These two dates mark the ends of interglacial high-water stands that exceeded ~30 and ~15 m above current sea levels, respectively (Stanford et al., 2001; O’Neal and McGeary, 2002). After the first stand, interglacial marine transgressions remained sufficiently low for most of north/central Florida and the adjacent mainland to remain emergent to the present (Figure 4-1). In turn, after the second, interglacial sea levels remained sufficiently low for most of peninsular Florida to

PAGE 76

63 remain permanent. Thus, these two high-water stands may mark the beginnings of significant permanent land for the establishment of the older versus younger genetic variation within P. striatus and P. axanthus, respectively.

PAGE 77

CHAPTER 5 SUMMARY This study examines allozyme and mitochondrial DNA variation for 238 and 305 individuals from 26 and 39 samples, respectively, to investigate the population genetics, evolution, and phylogeography of Pseudobranchus. The following are major conclusions from this study. 1. The heterozygosities and polymorphisms estimated from 14 allozyme loci were not positively correlated to sample size according to Spearman’s coefficient of rank test, so the differences among these estimates are not due to uneven sample sizes (Table 4-1 and Appendix B). 2. Because of the extensive allozyme divergence between the study group and outgroup, Siren is not an ideal outgroup for the allozyme data, although it is the only sister taxa to Pseudobranchus (Figure 2-2). 3. The allozyme data supported the recognized species, P. axanthus and P. striatus, by a long phylogenetic branch with 89% bootstrap support between them. Also, these two species were fixed for different alleles, (Ck-A and mSod-A), as well as sharing no alleles at their more polymorphic loci, Ada-A and Gpi-A. Furthermore, the syntopic samples from the two localities shared no alleles at four or five loci (Figure 2-2 and Appendix B). 4. The mean heterozygosities and polymorphisms for P. striatus are consistent with those vertebrates in general (Table 4-1). In contrast, those for P. axanthus are similar to the reduced values for other paedomorphic salamanders. Thus, 64

PAGE 78

65 Pseudobranchus provides a good opportunity to study the demographic, ecological, and evolutionary factors that underlie the reduced genetic variation in non-transforming salamanders. 5. Based on different phylogenetic analyses of the allozyme data, three phylogeographic groups within P. striatus were consistently recognized. Although they were only weakly separated by bootstrap scores of 54% to 63% and by an insignificant P (Figure 2-2; Table 4-2), they were recognizable from their great divergences and fixed and nearly fixed allelic differences at mSod-A and sMdh-A (Figure 4-4; Table 4-2.; Appendix B) 6. These phylogeographic groups of P. striatus not only correspond to their own recognized subspecies (Goin and Crenshaw, 1949; Martof, 1972; Moler and Kezer, 1993; Liu et al., 2004, 2005), but also to known geographic barriers that separate other aquatic groups in the southeastern United States (i.e., the Central Highland and Tifton/Vidalia Uplands, and the Altamaha River) (Figures 2-1, 3-1, 3-2 and 4-1). 7. The cyt b sequences of Pseudobranchus were represented by 166 distinct haplotypes that corresponded to an alignment of 786 base pairs (Appendix E). 8. Given various phylogenetic analyses of the cyt b sequence data, Siren was found not to be an ideal outgroup for Pseudobranchus, because of its great divergence to the latter (Figures 3-3 and 4-4; Appendix G). Because there is no other “close” living relative to Pseudobranchus, the use of more conserved genes is now needed to root the Pseudobranchus phylogeny.

PAGE 79

66 9. Major phylogeographic groups were recognized in the unrooted cyt b phylogeny by the various parsimony, distances, maximum likelihood, and Bayesian analyses, including the two species, P. axanthus and P. striatus, and the three phylogeographic groups within P. striatus (NE, SE, and W) from the allozyme study (Figures 2-2 and 3-3). These groups re-emphasized the effects of the corresponding geographic barriers in this area, i.e., the Central Highland and Tifton/Vidalia Uplands, and the Altamaha River, as well as status of their nominal taxa (Figures 3-1, -2 and 4). 10. Also, new subgroups were consistently recovered within P. axanthus and the W group of P. striatus given their bootstrap scores of 82% to 100% (Figure 3-3). The N and S groups of P. axanthus represent two recognized subspecies, P. a. axanthus and P. a. belli, respectively. The subdivision of the NW, WW, and “other W” subgroups within the W group of P. striatus conforms to of the Apalachicola River Drainage (Figures 3-2 and 4-1). Again, these phylogeographic units and biogeographic boundaries are consistent with the recognized ecological sub-regions for aquatic groups in the southeastern United States (Hocutt and Wiley, 1986; Moler and Kezer, 1993; Maxwell et al. 1995; Avise, 2000b; Liu et al., 2004, 2005). 11. Among the phylogeographic groups of P. striatus, cyt b data support the NE group as closely related to the SE group in contrast to the ambiguous allozyme results (Figure 3-3). This union highlights the importance of the Central Highland and Tifton/Vidalia Uplands, although NE and SE groups may not be recognized

PAGE 80

67 as separate species (Figures 3-2 and 4-1). These highlands represent a limit to many lowland organisms in the southeastern United States (Figures 3-1 and 4-1). 12. The Altamaha and Apalachicola river valleys correspond to recurrent biogeographic barriers and thereby reinforce the importance of rising sea levels and flooding of these low-lying valleys during interglacial periods of the late Pliocene and Pleistocene (Figures 1-1, 3-1, and 4-1). During these marine transgressions, those river valleys not only became wider and deeper, but also increased in salinity thereby separating these regions (Hocutt and Wiley 1986; Webb, 1990). 13. Within-population allozyme variation is significantly lower within P. axanthus than P. striatus (Table 4-1). This is consistent with their F-statistics which show a significant heterozygote deficiency within P. axanthus but not within P. striatus (Table 4-2) (Liu et al., 2004). 14. In contrast to this low allozyme variation in P. axanthus (or conversely, high variation in P. striatus), their genetic diversities from cyt b data are very similar (Table 4-1). This difference between the two genetic systems may be explained by the faster sorting times for mtDNA versus nDNA in addition to interglacial marine transgressions since the late Miocene that caused greater population bottlenecking within P. axanthus than P. striatus (Li, 1997; Moler and Kezer, 1993; Liu et al., 2004, 2005). 15. The coalescent theory allowed us to estimate the effective population size, divergence time, and migration rate at the population level (Edwards and Beerli, 2000; Beerli and Felsenstein, 2001). Thus these parameters provided us with

PAGE 81

68 further insights into the population demographics of P. axanthus than P. striatus (Nielsen and Wakeley, 2001). 16. Based on five dating analyses P. striatus is at least 2.5 times older than P. axanthus, even though their Nf are very similar (Table 4-3). The younger age of P. axanthus represents the more recent divergence of its phylogeographic groups and is due to its greater lineage extinction by widespread marine flooding during the last 5 million years. 17. About 3.5 and 1.5 Ma, interglacial high-water stands in the southeastern United States reached ~30 and ~15 m above current sea levels and thereafter stayed below these marks to the present, respectively. These two points in time are highlighted as most likely when the greater standing genetic diversity within P. striatus than within P. axanthus trace back to, respectively. (Figure 1-1 and 4-1) (Stanford et al. 2001; O’Neal and McGeary, 2002).

PAGE 82

APPENDIX A LOCALITY DATA FOR THE 245 SPECIMENS OF Pseudobranchus AND Siren (ALLOZYME STUDY) Locality data for the 245 specimens of Pseudobranchus and Siren (allozyme study). Sample sizes for these samples are given in parentheses. Voucher specimens were deposited in the collections of the Florida Museum of Natural History, University of Florida (UF numbers). ______________________________________________________________________________ Pseudobranchus striatus <1> Dixie County, Florida; State Route 349, pond 12.2 km south of Old Town. UF 73011 (1). <2> Gilchrist County, Florida; State Route 26, approximately 8.0 km west of Alachua County line. UF 73121-73123, 73130-73138 (12). <3> Eastern Hernando County, Florida; State Route 50, 2.9 km east of US Route 301. UF 73012-73016 (5). <4> Western Hernando County, Florida; US Route 19, 1.3 km north of junction with State Route 50. UF 73074, 73085-73086 (3). <5> Levy County, Florida; State Route 24, 2.9 to 11.2 km west of Bronson. UF 72966-72979, 73073, 73087, 73088 (17). <6> Eastern Madison County, Florida; State Route 53, 9.9 km south of Interstate 10. UF 73103-73107 (5). <7> Western Madison County, Florida; Interstate 10, 9.6 km east of US Route 221. UF 69

PAGE 83

70 73108-73112 (5). <8> Wakulla County, Florida; approximately 6.4 km west of Crawfordville. UF 73124-73129; 11 specimens not preserved (17). <9> Washington County, Florida; US Route 90, 7.7 km east of Choctawhatchee River. UF 73099-73102 (4). <10> Lee County, Georgia; Eagle Pond, approximately 8.0 km south of Smithville on State Route 19. Six specimens not preserved (6). <11> Northern Alachua County, Florida; Santa Fe River swamp at US Route 301, north of Waldo. UF 73033-73051; 1 specimen not preserved (20). <12> Southern Alachua County, Florida; State Route 346, approximately 0.2 km east of River Styx. UF 72961, 72962, 73077-73082 (8). <13> Baker County, Florida; State Route 2, approximately 4.8 km east of Columbia County line. UF 73021-73032 (12). <14> Volusia County, Florida; Maytown Road, 1.3 km north of Interstate 95 overpass. UF 72959, 72960; 6 specimens not preserved (8). <15> Charlton County, Georgia; State Route 121, 5.3 km south of St. George. UF 73083 (1). <16> Wayne County, Georgia; US Route 301, 0.8 km north of Brantley County line. UF 73066-73072 (7). <17> Long County, Georgia; State Route 99, 1.4 km west of McIntosh County line. UF 72993-73010; 2 specimens not preserved (20). Pseudobranchus axanthus <18> Citrus County, Florida; State Route 48 at Withlacoochee River. UF 73089-73098

PAGE 84

71 (10). <19> Central Hernando County, Florida; Mountain Lake. UF 73017-73020 (4). <20> Eastern Hernando County, Florida; State Route 50, 2.9 km east of US Route 301. Eleven specimens not preserved (11). <21> Marion County, Florida; creek between Moore’s Pond and Ledwith Lake. UF 72939-72958 (20). <22> Osceola County, Florida; US Route 441, 4.8 km south of Kenansville. UF 73113-73120 (8). <23> Pasco County, Florida; State Route 54, 1.3 km west of State Route 35A. UF 73075, 73084 (2). <24> Putnam County, Florida; Rodman Reservoir, State Route 310 at Deep Creek. Eight specimens not preserved (8). <25> Sarasota County, Florida; Interstate 75, 0.8 km north of Laurel Road. UF 72963-72965, 73052-73058; 11 specimens not preserved (21). <26> Volusia County, Florida. Maytown Road, 1.3 km north of Interstate 95 overpass. UF 73076; 2 specimens not preserved (3). Siren intermedia (outgroup) <27> Northern Alachua County, Florida; Santa Fe River swamp at US Route 301, north of Waldo. UF 73059-73065 (7). ______________________________________________________________________________

PAGE 85

APPENDIX B GENOTYPE ARRAYS AT THE 13 VARIABLE LOCI FOR THE 27 SAMPLES OF Pseudobranchus AND Siren Table B-1. Genotype arrays at the 13 variable loci for the 27 samples of Pseudobranchus and Siren. Genotypes and their counts are given before and after the colons, respectively. Those in bold highlight the eight instances where the variable locus of a sample is not in Hardy-Weinberg equilibrium. Samples Locus 1 <1> <2> <3> <4> <5> <6> <7> <8> <9> sAcoh-A bb:1 bb:12 bb:5 bb:3 bb17 aa:1 cc:5 bb:7 bb:4 ac:1 cc:3 Ada-A bc:1 bb:6 ab:3 ab:2 bb:7 ac:1 ac:1 aa:11 bb:4 bc:4 bb2 bb:1 bc:9 ce:4 bb:2 ab:2 cc:2 cc:1 bc:2 bb:1 sAta-A cc:1 cc:12 cc:5 cc:3 cc:17 cc:5 cc:5 cc:17 cc:4 Ck-A aa:1 aa:12 aa:5 aa:3 aa:17 aa:5 aa:5 aa:17 aa:4 Gpi-A cc:1 aa:5 cc:5 cc:1 aa:3 ac:1 cc:5 cc:15 cc:3 ac:4 cd:1 ac:5 cc:3 cd:2 cd:1 cc:3 dd:1 cc:9 G3pdh-A cc:1 cc:11 ce:1 aa:2 cc17 aa:1 aa:3 aa:1 cc:4 cd:1 ee:4 ac:1 ac:2 ac:1 ac:4 cc:2 cc:1 cc:11 dd:1 Ldh-A bb:1 bb:12 bb:5 bb:3 bb:17 bb:4 bb:1 bb:17 bb:4 bc:1 cc:4 mMdh-A aa:1 aa:12 aa:5 aa:3 aa:17 aa:5 aa:3 aa:16 aa:4 ac:2 sMdh-A bb:1 bb:12 bb:5 bb:3 ab:1 bb:5 bb:5 ab:1 bb:4 bb:16 bb:16 Mpi-A cc:1 cc:12 cc:5 cc:3 cc:15 cc:5 cc:5 cc:17 cc:4 Pgm-A bb:1 bb:12 bb:5 bb:3 bb:17 bb:5 bb:5 bb:17 bb:1 be2 ee:1 Pnp-A cc:1 cc:11 cc:5 cc:3 cc:17 cc:5 cc:5 cc:17 cc:4 mSod-A aa:1 aa:12 aa:5 aa:3 aa:17 aa:5 aa:5 aa:17 aa:4 72

PAGE 86

73 Table B-1. Continued. Samples Locus 1 <10> <11> <12> <13> <14> <15> <16> <17> <18> sAcoh-A bb:1 ab:1 aa:2 bb:12 bb:8 bb:1 bb:7 bb:20 cc:10 cc:5 bb:10 ab:1 bc:2 bb:5 cc:7 Ada-A ab:1 ab:2 ab:1 ab:1 bb:8 bb:1 ab:4 ab:4 dd:10 bb:5 bb:17 bb:6 bb:10 bb:3 bb:16 bc:1 bc:1 bc:1 sAta-A cc:6 cc:20 ac:1 cc:7 cc:8 cc:1 ac:1 cc:19 cc:9 cc:5 cc:6 dd:1 cd:1 Ck-A aa:6 aa:20 aa:8 aa:12 aa:8 aa:1 aa:7 aa:20 cc:10 Gpi-A cc:6 cc:19 cc:8 ac:1 cc:8 cc:1 cc:6 cc:20 bb:10 cd:1 cc:9 cd:1 cd:2 G3pdh-A cc:6 cc:5 aa:1 ac:1 cc:8 ac:1 ac:5 aa:5 cc:10 cf:12 cc:7 ae:2 cc:1 ac:13 ff:3 cc:4 ce:1 cc:2 ce:4 ee:1 Ldh-A bb:6 bb:20 bb:8 bb:12 bb:8 bb:1 bb:7 bb:20 bb:10 mMdh-A aa:6 aa:20 aa:8 aa:12 aa:8 aa:1 aa:7 aa:15 aa:10 sMdh-A bb:6 aa:20 aa:8 aa:12 aa:8 aa:1 aa:7 aa:13 aa:10 Mpi-A cc:6 cc:18 cc:8 cc:12 cc:8 cc:1 cc:7 cc:16 cc:6 cd:2 dd:2 Pgm-A bb:6 bb:20 bb:8 bb:12 bb:8 bb:1 ab:1 aa:1 bb:10 bb:6 bb:18 dd:1 Pnp-A cc:6 aa:2 cc:8 cc:12 cc:7 cc:1 cc:7 bb:1 cc:10 ac:1 cd:1 cc:15 bc:6 cd:1 cc:11 mSod-A aa:6 bb:20 bb:8 bb:12 bb:8 bb:1 bb:7 aa:20 cc:10

PAGE 87

74 Table B-1. Continued. Samples Locus 1 <19> <20> <21> <22> <23> <24> <25> <26> <27> sAcoh-A bb:4 bc:2 cc:20 cc:8 bc:1 cc:6 cc:21 cc:3 dd:1 cc:9 cc:1 cd:1 ee:6 dd:1 Ada-A dd:4 dd:11 dd:20 dd:8 dd:2 dd:8 dd:21 dd:3 bb:7 sAta-A cc:4 ac:1 cc:20 cc:8 aa:1 cc:8 cc:21 cc:3 bb:7 cc:10 cc:1 Ck-A cc:4 cc:11 cc:20 cc:8 cc:2 cc:8 cc:21 cc:3 bb:7 Gpi-A bb:4 bb:11 bb:20 bb:7 bb:2 bb:8 bb:21 bb:3 cc:7 G3pdh-A cc:4 cc:8 aa:2 aa:2 bc:1 ac:1 aa:21 cc:3 cc:7 cd:1 ac:1 ac:4 cc:1 cc:7 dd:2 cc:15 cc:2 Ldh-A bb:4 bb:11 bb:20 bb:8 bb:2 bb:8 bb:21 bb:3 aa:7 mMdh-A aa:4 aa:11 aa:20 aa:8 aa:2 aa:8 aa:20 aa:3 bb:7 sMdh-A aa:4 aa:9 aa:20 aa:8 aa:2 aa:8 aa:6 aa:3 cc:7 ab:2 ab:11 bb:4 Mpi-A cc:3 ad:1 cc:20 cc:8 aa:1 cc:8 cc:21 cc:3 bb:7 cc:6 cc:1 cd:1 dd:1 Pgm-A bb:4 bb:11 bb:20 bb:8 bb:2 bb:8 bb:21 bb:3 cc:7 Pnp-A cc:4 cc:11 bc:1 cc:8 cc:2 cc:8 cc:21 cc:3 dd:7 cc:15 mSod-A cc:4 cc:11 cc:20 cc:8 cc:2 cc:8 cc:21 cc:3 dd:7 1 sAcoh-A (aconitate hydratase, 4.2.1.3, liver, buffer D); Ada-A (adenosine deaminase, 3.5.4.4, liver, buffer D); sAta-A (aspartate transaminase, 2.6.1.1, muscle, buffer B); Ck-A (creatine kinase, 2.7.3.2, muscle, buffer E); Gpi-A (glucose-6-phosphate isomerase, 5.3.1.9, muscle, buffer A); G3pdh-A (glycerol-3-phosphate dehydrogenase, 1.1.1.8, muscle, buffer E); Ldh-A (L-lactate dehydrogenase, 1.1.1.27, muscle, buffer C); mMdh-A (malate dehydrogenase, 1.1.1.37, muscle, buffer D); sMdh-A (malate dehydrogenase, 1.1.1.37, muscle, buffer D); Mpi-A (mannose-6-phosphate isomerase, 5.3.1.8, muscle, buffer B); Pgm-A (phosphoglucomutase, 5.4.2.2, muscle, buffer A); Pnp-A (purine nucleoside phosphorylase, 2.4.2.1, muscle, buffer D); mSod-A (superoxide dismutase, 1.15.1.1, muscle, buffer D). Enzyme, Enzyme Commission number, tissue source, and buffer system are presented in that order for each locus (International Union of Biochemistry and Molecular Biology, 1992; Murphy et al., 1996). sSod-A (superoxide dismutase, 1.15.1.1, muscle, buffer D) is monoallelic. The buffer systems include: (A) amine-citrate (morpholine), pH 6.1; (B) Tris-borate-EDTA II, pH 8.0; (C) Tris-citrate II, pH 8.0; (D) Tris-citrate-EDTA, pH 7.0; and (E) Tris-HCL, pH 8.5/8.2.

PAGE 88

APPENDIX C FULL LOCALITY DATA, SAMPLE SIZES, AND MUSEUM COLLECTION NUMBERS (UF) FOR THE 314 INDIVIDUALS OF Pseudobranchus AND Siren (MTDNA STUDY) Table C-1. Full locality data, sample sizes, and museum collection numbers (UF) for the 314 individuals of Pseudobranchus and Siren (mtDNA study). Sample number County, state Location Sample size Museum number Pseudobranchus striatus WW Subgroup 1 Santa Rosa Co., FL Vic. Chumuckla Spring, near confluence McDavid Creek with Escambia River 5 No specimens preserved 2 Washington Co., FL US Rte. 90, 7.7 km E Choctawhatchee River 4 UF 73099-73102 3 Holmes Co., FL Interstate Hwy 10, 3.5 km W FL Rte. 79 6 UF 141754-141759 NW Subgroup 4 Baker Co., GA Joseph Jones Ecological Research Station 6 No specimens preserved 5 Lee Co., GA Eagle Pond, ca. 8 km S Smithville on GA Rte. 19 4 No specimens preserved Other W samples 6 Wakulla Co., FL Approximately 6.4 km W Crawfordville 13 UF 73124-73126, 73128; 9 specimens not preserved 7 Madison Co., FL FL Rte. 53, 9.9 km S Interstate Hwy 10 3 UF 73104-73106 8 Madison Co., FL Interstate Hwy 10, 9.6 km E US Rte. 221 10 UF 73108, 73110-73112; 141760-141765 9 Dixie Co., FL FL Rte. 349, 12.2 km S Old Town 1 UF 73011 10 Gilchrist Co., FL FL Rte. 26, ca. 8 km W Alachua County line 12 UF 73121-73123, 73130-73138 11 Levy Co., FL FL Rte. 24, 2.9, 6.7, and 11.2 km W Bronson 15 UF 72966, 72971, 73073-72975, 72977-72979; 73087-73088; 4 specimens not preserved 75

PAGE 89

76 Table C-1. Continued. Sample number County, state Location Sample size Museum number 12 Hernando Co., FL Weekiwachee. US Rte. 19, 1.3 km N FL Rte. 50 3 UF 73074, 73085, 73086 13 Hernando Co., FL FL Rte. 50, 2.9 km E US Rte. 301 5 UF 73012-73016 NE Group 14 Orangeburg Co., SC SC Rte. 21, 6.2 km N Branchville 1 No specimens preserved 15 Bryan Co., GA Fort Stewart 10 UF141733-141742 16 Long Co., GA 15.5 km E Glennville 4 No specimens preserved 17 Long Co., GA GA Rte. 99, 1.4 km W McIntosh County line 18 UF 72993-72997, 72999, 73001, 73002, 73004-73006, 73008-73010; 4 specimens not preserved 18 Wayne Co., GA US 341, 23.2 km SE Jesup 9 UF141704; 8 specimens not preserved SE Group 19 Wayne Co., GA US Rte. 301, 0.8 km N Brantley County line 13 UF 73066-73072, 141732; 5 specimens not preserved 20 Hamilton Co., FL 4 km NNE White Springs 4 No specimens preserved 21 Clinch Co., GA Fargo, US 441 S Suwannee River 11 UF141743-141753 22 Baker Co., FL FL Rte. 2, ca. 4.8 km E Columbia County line 9 UF 73021, 73023, 73025, 73027-73032 23 Baker Co., FL Co. Rd 122 E Taylor, at Middle Prong St. Marys River 2 No specimens preserved 24 Charlton Co., GA GA Rte. 121, 5.3 km S St. George 1 UF 73083 25 Alachua Co., FL Santa Fe River swamp at US Rte. 301, N Waldo 16 UF 73033, 73035, 73037-73041, 73043, 73044, 73046-73051; 1 specimen not preserved 26 Alachua Co., FL FL Rte. 346 ca. 200 m E River Styx 6 UF 72961, 72962, 73077, 73080-73082 27 Volusia Co., FL Maytown Road, 1.3 km N Interstate Hwy 95 overpass 5 UF 72959, 123827; 3 specimens not preserved

PAGE 90

77 Table C-1. Continued. Sample number County, state Location Sample size Museum number Pseudobranchus axanthus N Group 28 Marion Co., FL Creek between Moores Pond and Ledwith Lake 19 UF 72940-72958 29 Alachua Co., FL FL Rte. 346 at River Styx, E Micanopy 2 UF141705, 141706 30 Putnam Co., FL Rodman Reservoir, FL Rte. 310 at Deep Creek 7 UF124724-124730 31 Citrus Co., FL FL Rte. 48 at Withlacoochee River 10 UF 73089-73098 32 Hernando Co., FL Mountain Lake 3 UF 73017-73019 33 Hernando Co., FL FL Rte. 50, 2.9 km E US Rte. 301 10 UF124716-124721, UF1214723, UF124742, UF124743; 1 specimen not preserved 34 Pasco Co., FL FL Rte. 54, 1.3 km W FL Rte. 35A 2 UF 73075, 73084 35 Volusia Co., FL Maytown Road, 1.3 km N Interstate Hwy 95 overpass 14 UF 73076, 123830, 123831, 141712-141717; 5 specimens not preserved 36 Osceola Co., FL US Rte. 441, 4.8 km S Kenansville 7 UF 73113-73115, 73117-73120 S Group 37 Sarasota Co., FL Interstate Hwy 75, 0.8 km N Laurel Road 21 UF 72963-72965, 73052-73058; 11 specimens not preserved 38 Sarasota Co., FL FL Rte. 72, ca. 35.2 km W Interstate Hwy 75 4 UF 141718-141721 39 Glades Co., FL FL Rte. 720 at Pollywog Creek, N Labelle 10 UF 141722-141731 Siren intermedia 40 Alachua Co., FL Santa Fe River swamp at US Rte. 301, N Waldo 3 No specimens preserved Siren lacertina 41 Volusia Co., FL Maytown Road, 1.3 km N Interstate Hwy 95 overpass 5 UF 141707-141711 42 Sarasota Co., FL FL Rte. 72 ca. 20.8 km E Interstate Hwy 75 1 No specimens preserved

PAGE 91

APPENDIX D ANNEALING SITES AND SEQUENCES FOR THE PCR AMPLIFICATION AND CYCLE SEQUENCING PRIMERS OF CYT B FOR Pseudobranchus AND Siren Figure D-1. Annealing sites and sequences for the PCR amplification and cycle sequencing primers of cyt b for Pseudobranchus and Siren. Universal versus sirenid-specific primers are represented by the solid versus broken arrows, whereas those for the heavy versus light strands of this mtDNA gene are presented above and below the diagram, respectively. The universal primers are from Moritz et al. (1992), whereas the sirenid-specific ones are derived from the preliminary data of this study. 78

PAGE 92

APPENDIX E MULTIPLE SEQUENCE ALIGNMENT FOR THE 173 DISTINCT HAPLOTYPES OF CYT B FOR Pseudobranchus AND Siren Object E-1: Multiple sequence alignment for the 173 distinct haplotypes of cyt b for Pseudobranchus and Siren. Haplotype 1 is shown in its entirety, with periods denoting the same nucleotide as that of this reference. Dashes represent gaps, whereas question marks refer to missing data. 79

PAGE 93

APPENDIX F FREQUENCY DISTRIBUTION OF HAPLOTYPES AMONG THE 42 SAMPLES OF Pseudobranchus AND Siren Table F-1. Frequency distribution of haplotypes among the 42 samples of Pseudobranchus and Siren. Sample 1 2 3 4 5 6 7 8 9 10 Distinct haplotypes (number of sequences) 1 (5) 2 (1) 3 (1) 4 (1) 5 (1) 5 (3) 6 (2) 7 (1) 8 (3) 9 (1) 1 0 (2) 11 (1) 12 (3) 13 (12) 14 (1) 15 (1) 16 (1) 17 (1) 18 (1) 19 (2) 20 (4) 21 (1) 22 (1) 23 (1) 24 (1) 25 (2) 26 (1) 27 (1) 28 (2) 29 (3) 30 (3) Sample 11 12 13 14 15 16 17 18 19 Distinct haplotypes (number of sequences) 30 (10) 31 (1) 32 (2) 33 (2) 34 (3) 35 (1) 36 (2) 37 (1) 38 (1) 39 (1) 40 (6) 41 (4) 41 (1) 42 (1) 43 (1) 44 (1) 41 (2) 45 (1) 46 (1) 47 (2) 48 (3) 49 (3) 50 (1) 51 (1) 52 (1) 53 (1) 54 (1) 55 (1) 56 (3) 57 (2) 58 (1) 59 (1) 60 (1) 61 (1) 62 (1) 63 (1) 64 (1) 65 (1) 66 (1) 67 (6) 68 (1) 69 (1) Sample 20 21 22 23 24 25 26 27 Distinct haplotypes (number of sequences) 70 (1) 71 (1) 72 (2) 73 (1) 74 (1) 75 (1) 76 (1) 77 (2) 78 (1) 79 (1) 80 (1) 81 (1) 81 (1) 82 (1) 83 (1) 84 (1) 85 (1) 86 (1) 87 (1) 88 (1) 89 (1) 90 (1) 91 (1) 92 (1) 93 (1) 94 (7) 95 (1) 96 (1) 97 (1) 98 (1) 99 (1) 100 (1) 101 (1) 102 (1) 103 (5) 104 (1) 105 (2) 106 (3) 80

PAGE 94

81 Table F-1. Continued. Sample 28 29 30 31 32 33 34 35 Distinct haplotypes (number of sequences) 107 (1) 108 (1) 109 (2) 110 (3) 111 (2) 112 (1) 113 (1) 114 (1) 115 (1) 116 (1) 117 (1) 118 (1) 119 (2) 120 (1) 121 (1) 122 (1) 123 (3) 124 (1) 125 (1) 126 (1) 127 (1) 128 (4) 129 (1) 130 (2) 131 (1) 132 (2) 133 (2) 134 (1) 128 (1) 135 (3) 136 (1) 137 (1) 138 (3) 166 (1) 128 (2) 139 (3) 140 (1) 141 (1) 142 (1) 143 (2) 144 (3) 145 (1) 146 (1) 147 (1) Sample 36 37 38 39 40 41 42 Distinct haplotypes (number of sequences) 148 (1) 149 (1) 150 (1) 151 (2) 152 (2) 153 (7) 154 (10) 155 (1) 156 (1) 157 (1) 158 (1) 153 (2) 159 (1) 160 (1) 160 (1) 161 (2) 162 (1) 163 (1) 164 (2) 165 (3) 167 (1) 168 (1) 169 (1) 170 (2) 171 (2) 172 (1) 173 (1)

PAGE 95

APPENDIX G COMPLETE SET OF D’S AND ’S FOR THE 42 SAMPLES OF Pseudobranchus AND Siren Object G-1. Complete set of d’s and ’s for the 42 samples of Pseudobranchus and Siren (below and along the diagonal, respectively). Both are based on the HKY + + I distances for their haplotype pairs (with = 1.298 and p = 0.506). “NA” refers to those samples where is not applicable because of their sample sizes of one. 82

PAGE 96

LIST OF REFERENCES ABREU, V. S., AND J. B. ERSON. 1998. Glacial eustasy during the Cenozoic: Sequence stratigraphic implications. AAPG Bull. 82: 1385-1400. AVISE, J. C. 2000a. Molecular markers, natural history and evolution, 2nd edition. Chapman and Hall, New York. AVISE, J. C. 2000b. Phylogeography: The history and formation of species. Harvard University Press, Cambridge, Massachusetts. AVISE, J. C., AND BALL, R. M. JR. 1990. Principles of genealogical concordance in species concepts and biological taxonomy. Oxford Surv. Evol. Biol. 7: 45-67. AVISE, J. C., C. GIBLIN-DAVIDSON, J. LAERM, J. C. PATTON, AND R. A. LANSMAN. 1979. Mitochondrial DNA clones and matriarchal phylogeny within and among geographic populations of the pocket gopher, Geomys pinetis. Proc. Natl. Acad. Sci. USA 76: 6694-6698. BANDONI, S. M., M. MULVEY, AND E. S. LOKER. 2000. Population structure and taxonomic discrimination among three species of Biomphalaria Preston, 1910 (Gastropoda: Planorbidae) from Kenya. Zool. J. Linn. Soc. 129: 387-401. BEERLI, P., AND J. FELSENSTEIN. 2001. Maximun likelihood estimation of a migration matrix and effective population size in n subpopulations by using a coalescent approach. Proc. Natl. Acad. Sci. USA 98: 4563-4568. BENSON, D. A., I. KARSCH-MIZRACHI, D. J. LIPMAN, J. OSTELL, AND D. L. WHEELER. 2004. GenBank: update. Nuc. Acids Res. 32: D23-D26. BERLOCHER, S. H., AND D. L. SWOFFORD. 1997. Searching for phylogenetic trees under the frequency parsimony criterion: an approximation using generalized parsimony. Syst. Biol. 46: 211-215. BLACK, W. C. 1997. BIOSYS-2. Dept. Microbiology, Colorado State Univ., Fort Collins, Colorado. BRADLEY, R. D., J. J. BULL, A. D. JOHNSON, AND D. M. HILLIS. 1993. Origin of a novel allele in a mammalian hybrid zone. Proc. Natl. Acad. Sci. USA 90: 8939-8941. BROWN, J. H., AND M. V. LOMOLINO. 2003. Biogeography, 2nd edition. Sinauer Assc., Sunderland, Massachusetts. 83

PAGE 97

84 COLQUHOUN, D. J. 1995. A review of Cenozoic evolution of the southeastern United States Atlantic coast north of the Georgia trough. Quaternary Inter. 26: 35-41. CONANT, R., AND J. T. COLLINS. 1998. A field guide to reptiles and amphibians of eastern and central North America. Houghton Mifflin, New York. COX, C. B., AND P. D. MOORE. 2000. Biogeography: An ecological and evolutionary approach, 6th edition. Blackwell Science Ltd, Malden, Massachusetts. DONNELLY, P., AND S. TAVARE. 1995. Coalescents and genealogical structure under neutrality. Annu. Rev. Genet. 29: 401-421. EBERLI, G. P. 2000. The record of Neogene sea-level changes in the prograding carbonates along the Bahamas transsect-LEG 166 synthesis. Proc. Ocean Drilling Program, Scientific Results 166: 167-177. EDWARDS, S. V., AND BEERLI, P. 2000. Gene divergence, population divergence, and the variance in coalescence time in phylogrographic studies. Evolution 54: 1839-1854. FELSENSTEIN, J. 1981. Evolutionary trees from gene frequencies and quantitative characters: Finding maximum likelihood estimates. Evolution 35: 1229-1242. FELSENSTEIN, J. 2002. PHYLIP, version 3.6. Dept. Genetics, Univ. Washington, Seattle, Washington. GANTENBEIN, B., V. FET, I. A. GANTENBEIN-RITTER, AND F. BALLOUX. 2005. Evidence for recombination in scorpion mitochondrial DNA (Scorpiones: Buthidae). Proc. R. Soc. London B 272: 697-704. GAO, X., X. WEN, L. ESSER, B. QUINN, L. YU, C.-A. YU, AND D. XIA. 2003. Structural basis for the quinone reduction in the bc 1 complex: A comparative analysis of crystal structures of mitochondrial cytochrome bc 1 with bound substrate and inhibitors at the Q i site. Biochemistry 42: 9067-9080. GEOPHYSICS STUDY COMMITTEE. 1990. Studies in Geophysics: Sea-Level Change. National Academy Press, Washington, DC. GILBERT, C. R., R. C. CASHNER, AND E. O. WILEY. 1992. Taxonomic and nomenclatural status of the banded topminnow, Fundulus cingulatus (Cyprinodontiformes: Cyprinodontidae). Copeia 1992: 747-759. GOIN, C. J., AND J. W. CRENSHAW, JR. 1949. Description of a new race of the salamander Pseudobranchus striatus (Le Conte). Annals Carnegie Mus. 31: 277-280. GUINDON, S., AND O. GASCUEL. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst. Biol. 52: 696-704.

PAGE 98

85 GUO, S. W., AND E. A. THOMPSON. 1992. Performing the exact test of Hardy-Weinberg proportions for multiple alleles. Biometrics 48: 361-372. HALLAM, A. A. 1992. Phanerozoic sea-level changes. Columbia Univ. Press, New York. HALLAM, A. A. 1994. An outline of Phanerozoic biogeography. Oxford Univ. Press, New York. HAQ, B. U., J. HARDENBOL, AND P. R. VAIL. 1987. Chronology of fluctuating sea levels since the Triassic. Science 235: 1156-1166. HASEGAWA, M., H. KISHINO, AND T. YANO. 1985. Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22: 160-174. HEDRICK, P. W. 2005. Genetics of populations, 3rd ed. Jones and Bartlett publishers, Sudbury, Massachusetts. HILLIS, D. M., B. K. MABLE, A. LARSON, S. K. DAVIS, AND E. A. ZIMMER. 1996. Nucleic acids IV: Sequencing and cloning. Pages 321-381 in Molecular systematics, 2nd edition. (D. M. Hillis, C. Moritz, and B. K. Mable, eds.). Sinauer Assc., Sunderland, Massachusetts. HOBBS, H. H., JR. 1942. The crayfishes of Florida. Biol. Sci. Ser. 3: 1-179. HOCUTT, C. H., AND E. O. WILEY. 1986. The zoogeography of North American freshwater fishes. John Wiley, New York. HUELSENBECK, J. P., F. RONQUIST, R. NIELSEN, AND J. P. BOLLBACK. 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294: 2310-2314. INTERNATIONAL UNION OF BIOCHEMISTRY AND MOLECULAR BIOLOGY. 1992. Enzyme nomenclature 1992. Academic Press, San Diego, California. JOHNS, G. C., AND J. C. AVISE. 1998. A comparative summary of genetic distances in the vertebrates from the mitochondrial cytochrome b gene. Mol. Biol. Evol. 15: 1481-1490. KINGMAN, J. F. C. 1982. The coalescent. Stoch. Proc. Theo. Appl. 13: 235-248. KISHINO, H., J. L. THORNE, AND W. J. BRUNO. 2001. Performance of a divergence time estimation method under a probabilistic model of rate evolution. Mol. Biol. Evol. 18: 352-361. LEE, D. S., GILBERT, C. R., HOCUTT, C. H., JENKINS, R. E., MCALLISTER, D. E., AND J. R. STAUFFER JR. 1980. Atlas of North American freshwater fishes. North Carolina State Mus. Nat. Hist., Raleigh, North Carolina.

PAGE 99

86 LEWIS, P. O., AND D. ZAYKIN. 2002. Dept. Ecol. Evol. Biol., Univ. Connecticut and Stat. Genet., Glaxo Wellcome, North Carolina. LI, W.–H. 1997. Molecular evolution. Sinauer Assc., Sunderland, Massachusetts. LIU, F.-G. R., P. E. MOLER, AND M. M. MIYAMOTO. 2005. Phylogeography of the salamanders, Pseudobranchus, and other endemic groups in the southeastern United States. Mol. Phyl. Evol. (in review). LIU, F.-G. R., P. E. MOLER, H. P. WHIDDEN, AND M. M. MIYAMOTO. 2004. Allozyme variation in the salamander genus Pseudobranchus: Phylogeographic and taxonomic significance. Copeia 2004: 136-144. MADDISON, W. P., AND D. R. MADDISON. 2000. MacClade 4: Analysis of phylogeny and character evolution, version 4.07. Sinauer Assc, Sunderland, Massachusetts. MARTOF, B. S. 1972. Pseudobranchus, P. striatus. Cat. Am. Amphib. Rept. 118: 1-4. MAXWELL, J. R., C. J. EDWARDS, M. E. JENSEN, S. J. PAUSTIAN, H. PARROT, AND D. M. HILL. 1995. A hierarchical framework of aquatic ecological units in North America (Nearctic Zone). USDA Forest Serv. General Tech. Rep. NC-176, St. Paul, Minnesota. MOLER, P. E., AND J. KEZER. 1993. Karyology and systematics of the salamander genus Pseudobranchus (Sirenidae). Copeia 1993: 39-47. MORITZ, C., C. J. SCHNEIDER, AND D. B. WAKE. 1992. Evolutionary relationships within the Ensatina eschscholtzii complex confirm the ring species interpretation. Syst. Biol. 41: 273-291. MURPHY, R. W., J. W. SITES JR., D. G. BUTH, AND C. H. HAUFLER. 1996. Proteins: Isozyme electrophoresis, Pages 51-120 in: Molecular systematics, 2nd edition. (D. M. Hillis, C. Moritz, and B. K. Mable eds.). Sinauer Assc., Sunderland, Massachusetts. NEI, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583-590. NEI, M. 1987. Molecular evolutionary genetics. Columbia Univ. Press, New York. NEI, M., AND D. GRAUR. 1984. Extent of protein polymorphism and the neutral mutation theory. Evol. Biol. 17: 73-118. NEI, M., AND W.–H. LI. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76: 5269-5273. NEI, M., AND F. TAJIMA. 1981. DNA polymorphism detectable by restriction endonucleases. Genetics 97: 154-163.

PAGE 100

87 NEVO, E., AND A. BEILES. 1991. Genetic diversity and ecological heterogeneity in amphibian evolution. Copeia 1991: 565-592. NIELSEN, R., AND J. WAKELEY. 2001. Distinguishing migration from isolation: a Markov Chain Monte Carlo approach. Genetics 158: 885-896. NORES, M. 2004. The implications of Tertiary and Quaternary sea level rise events for avian distribution patterns in the lowlands of northern South America. Glob. Ecol. Biogeo. 13: 149-161. O’NEAL, M. L., AND S. MCGEARY. 2002. Late Quaternary stratigraphy and sea-level history of the northern Delaware Bay margin, southern New Jersey, USA: a ground penetrating radar analysis of composite Quaternary coastal terraces. Quat. Sci. Rev. 21: 929-946. OTVOS, E. G. 1995. Multiple Pliocene-Quaternary marine highstands, northeast Gulf coastal plain fallacies and facts. J. Coast. Res. 11: 984-1002. PAGE, R. D. M., AND E. C. HOLMES. 1998. Molecular evolution: A phylogenetic approach. Blackwell Science, Malden, Massachusetts. PETRANKA, J. W. 1998. Salamanders of the United States and Canada. Smithsonian Institution Press, Washington DC. PIEL, W. H., M. J. DONOGHUE, AND M. J. SANDERSON. 2001. TreeBASE: a database of phylogenetic information. 2nd International Workshop of Species 2000, Natl. Inst. Environ. Stud. Tsukuba, Japan. POSADA, D., AND K. A. CRANDALL. 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14: 817-818. ROE, B. A., D. P. MA, R. K. WILSON, AND J. F. WONG. 1985. The complete nucleotide sequence of the Xenopus laevis mitochondrial genome. J. Biol. Chem. 260: 9759-9774. RONQUIST, F., AND J. P. HUELSENBECK. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572-1574. SANDERSON, M. J. 2002. Estimating absolute rates of molecular evolution and divergence time: a penalized likelihood approach. Mol. Biol. Evol. 19: 101-109. SATO, A., K. NAKADA, M. AKIMOTO, K. ISHIKAWA, T. ONO, H. SHITARA, H. YONEKAWA, AND J. L. HAYASHI. 2005. Rare creation of recombinant mtDNA haplotypes in mammalian tissues. Proc. Natl. Acad. Sci. USA 102: 6057-6062. SHAFFER, H. B., AND F. BREDEN. 1989. The relationship between allozyme variation and life history: non-transforming salamanders are less variable. Copeia 1989:1016-1023.

PAGE 101

88 SOKAL, R. R., AND N. L. ODEN. 1978. Spatial autocorrelation in biology. 1. Methodology. Biol. J. Linn. Soc. 10: 199-228. SOKAL, R. R., AND F. J. ROHLF. 1995. Biometry. 3rd edition, W. H. Freeman, New York. STANFORD, S. D., G. M. ASHLEY, AND G. J. BRENNER. 2001. Late Cenozoic fluvial stratigraphy of the New Jersey piedmont: low-relief passive margin. J. Geol. 109: 265-276. SWOFFORD, D. L. 2002. PAUP*: Phylogenetic analysis using parsimony (* and other methods), version 4.0b. Sinauer Assc., Sunderland, Massachusetts. SWOFFORD, D. L., G. J. OLSEN, P. J. WADDELL, AND D. M. HILLIS. 1996. Phylogenetic inference. Pages 407-514 in Molecular systematics, 2nd edition. (D. M. Hillis, C. Moritz, and B. K. Mable, eds.). Sinauer Assc., Sunderland, Massachusetts. SWOFFORD, D. L., AND R. B. SELANDER. 1981. BIOSYS-1: a Fortran program for the comprehensive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72: 281-283. THORNE, J. L., H. KISHINO, AND I. S. PAINTER. 1998. Estimating the rate of evolution of the rate of molecular evolution. Mol. Biol. Evol. 15: 1647-1657. WALKER, D., AND J. C. AVISE. 1998. Principles of phylogeography as illustrated by freshwater and terrestrial turtles in the southeastern United States. Annu. Rev. Ecol. Syst. 29: 23-58. WALKER, D., P. E. MOLER, K. A. BUHLMANN, AND J. C. AVISE. 1998. Phylogeographic patterns in Kinosternon subrubrum and K. baurii based on mitochondrial DNA restriction analyses. Herpetologica 54: 174-184. WEBB, S. D. 1990. Historical biogeography. Pages 70-102 in Ecosystems of Florida (R. L. Myers, and J. J. Ewel, eds.). Univ. Presses Florida, Gainesville, Florida. WEEKS, A., S. C. DALY, AND B. B. SIMPSON. 2005. The phylogenetic history and biogeography of the frankincense and myrrh family (Burseraceae) based on nuclear and chloroplast sequence data. Mol. Phyl. Evol. 35: 85-101. WEIR, B. S. 1996. Genetic data analysis II. Sinauer Assc., Sunderland, Massachusetts. WEIR, B. S., AND W. G. HILL. 2002. Estimating F-statistics. Annu. Rev. Genet. 36: 721-750. WIENS, J. J. 2000. Reconstructing phylogenies from allozyme data: comparing method performance with congruence. Biol. J. Linn. Soc. 70: 613-632. WOOTEN, M. C., K. T. SCRIBNER, AND M. H. SMITH. 1988. Genetic variability and systematics of Gambusia in the southeastern United States. Copeia 1988: 283-289.

PAGE 102

89 WRIGHT, S. 1951. The genetical structure of populations. Ann. Eugen. 15: 323-354.

PAGE 103

BIOGRAPHICAL SKETCH Fu-Guo Robert Liu was born on October 19, 1966, in Taipei, Taiwan. He received his Bachelor of Science degree in biology from the Chinese Culture University in August 1992. The following two years he served in the Marine military service. After that, he became a research assistant at the Academic Sinica in Taiwan. Thereafter, he decided to go abroad to pursue his PhD degree at the University of Florida. During this time, he was involved in various projects that ultimately led to five publications and submitted manuscripts, most notably his article in 2001 in the journal Science. Also, the magazine National Geographic interviewed him for its article on mammalian phylogeny in its April 2003 issue. After completing his Ph.D. degree on the biogeography of the southeastern United States, Robert will begin a post-doctorate with Professor W.-H. Li in genomic research on yeast at the University of Chicago. After his post-doctorate training, he plans to pursue a research and teaching career back in his home country at Academic Sinica, Taiwan. 90