The Ecology and Evolutionary History of Two Musk Turtles in the Southeastern United States

Home , Diamondback terrapin, Eastern mud turtle, Flattened musk turtle, Kinosternidae, Kinosternon, Loggerhead musk turtle

The University of Southern Mississippi The Aquila Digital Community

Dissertations

Spring 2020

The Ecology and Evolutionary History of Two Musk Turtles in the Southeastern United States

Grover Brown

Follow this and additional works at: https://aquila.usm.edu/dissertations

Part of the Genetics Commons

Recommended Citation Brown, Grover, "The Ecology and Evolutionary History of Two Musk Turtles in the Southeastern United States" (2020). Dissertations. 1762. https://aquila.usm.edu/dissertations/1762

This Dissertation is brought to you for free and open access by The Aquila Digital Community. It has been accepted for inclusion in Dissertations by an authorized administrator of The Aquila Digital Community. For more information, please contact [email protected]. THE ECOLOGY AND EVOLUTIONARY HISTORY OF TWO MUSK TURTLES IN

THE SOUTHEASTERN UNITED STATES

Grover James Brown III

A Dissertation Submitted to the Graduate School, the College of Arts and Sciences and the School of Biological, Environmental, and Earth Sciences at The University of Southern Mississippi in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

Approved by:

Brian R. Kreiser, Committee Co-Chair Carl P. Qualls, Committee Co-Chair Jacob F. Schaefer Micheal A. Davis Willian W. Selman II

______Dr. Brian R. Kreiser Dr. Jacob Schaefer Dr. Karen S. Coats Committee Chair Director of School Dean of the Graduate School

May 2020

Grover James Brown III

2020

Published by the Graduate School

ABSTRACT

Turtles are among one of the most imperiled vertebrate groups on the planet with more than half of all species worldwide listed as threatened, endangered or extinct by the

International Union of the Conservation of Nature. The Southeastern United States is a global biodiversity hotspot for turtles, and it likely represents the last area on the planet of both high diversity and high abundance of turtle species. In the past century, there has been an exponential increase in the number of publications on turtles in the United States and Canada, however a recent review of the literature suggests that this research attention has not been spread evenly across taxa. The mud turtles (genus Kinosternon) and musk turtles (genus Sternotherus) of the family Kinosternidae remain one of the least studied turtle families. The lotic Sternotherus species have received very little research attention, particularly the stripe-necked musk turtle (Sternotherus peltifer) and the razorback musk turtle (Sternotherus carinatus) with the latter considered one of the top-ten least studied turtle species in the United States and Canada. No studies have examined the population genetics of these highly aquatic species across their geographic ranges, and this is particularly concerning as some states have had to issue moratoriums on commercial harvest of these species to prevent population declines. Harvested turtles are generally destined for overseas markets, as Southeast Asia has decimated its own turtle populations and must look to other areas of high diversity and abundance of turtles to satiate demand.

Considering that very little is still known about these species, it is important to have baseline data on the population genetics of these species, as well as fill in knowledge gaps on their ecology to help make more informed conservation and management decisions.

ACKNOWLEDGMENTS

I would like to thank my advisors, Brian Kreiser and Carl Qualls, for their insightful academic guidance throughout my graduate career, particularly for allowing me free reign to develop and pursue this research on these species. They have been pivotal to my development as a biologist. I would also like to thank Dr. Jen Lamb for other statistical, logistical and emotional support throughout my tenure at the University of Southern Mississippi. Dr. Will Selman for valuable insights not just into Mississippi turtle biology, but for valuable career advice along this journey.

This research would not have been possible without grants, funding and other support provided by (in alphabetical order): the American Museum of Natural History, the American Turtle Observatory, the Birmingham Audubon Society, the Chicago

Herpetological Society, the National Science Foundation, the Turtle Survival Alliance, and the University of Southern Mississippi.

iii

DEDICATION

I would like to partially dedicate this dissertation to the host of people who have inspired me to pursue a career in herpetology. I would first like to thank my family, but in particular my aunt Margaret Rose White, and my grandfather James Bryan White Jr., for introducing me to turtles at a young age and sparking a lifelong passion. I would like to thank Owen Kinney for taking me under his wing in high school and showing me what a career in herpetology looks like. I thank Dr. John Maerz for his mentoring in herpetology during my undergraduate career, and I also want to acknowledge Drs. Kurt Buhlmann and Whit Gibbons for showing me what a lifetime of passion for herpetology can accomplish.

I would be remiss not to mention the wonderful friends who have helped me, inspired me and taught me so much: Gabbie Berry, Cyndi Carter, Lucas Haralson, Cybil

Huntzinger, Jen Lamb, Luke Pearson, Chris Pellecchia, Todd Pierson, Theresa

Stratmann.

But mostly, I would like to dedicate this dissertation to my parents, Grover James

(Jimmy) Brown Jr. and Carolyn W. Brown, who fostered and supported my interest in wildlife, particularly turtles, from a young age, and who invested so heavily in my education to make sure I had every tool necessary to follow my passion and pursue my dreams. They exemplify selflessness. They are my personal heroes, and no words or document will ever be enough to express my gratitude.

TABLE OF CONTENTS

ABSTRACT ...... ii ACKNOWLEDGMENTS ...... iii DEDICATION ...... iv LIST OF TABLES ...... vii LIST OF ILLUSTRATIONS ...... ix CHAPTER I - INTRODUCTION TO TURTLES OF THE SOUTHEASTERN UNITED STATES ...... 1 1.1 Introduction ...... 1 1.2 Literature Cited ...... 11 CHAPTER II - CHARACTERIZATION OF MICROSATELLITE LOCI FOR THE RAZORBACK MUSK TURTLE (STERNOTHERUS CARINATUS; GRAY 1856) AND THEIR CROSS-AMPLIFICATION IN FIVE OTHER SPECIES IN THE FAMILY KINOSTERNIDAE ...... 19 2.1 Introduction ...... 19 2.2 Methods...... 21 2.3 Results ...... 23 2.4 Discussion ...... 23 2.5 Literature Cited ...... 26 CHAPTER III - RANGE-WIDE POPULATION GENETICS OF THE RAZORBACK MUSK TURTLE (STERNOTHERUS CARINATUS) ...... 35 3.1 Introduction ...... 35 3.2 Methods...... 39 3.2.1 Field Collections ...... 39 3.2.2 Molecular Methods ...... 40 3.3 Results ...... 43 3.4 Discussion ...... 46 3.5 Literature Cited ...... 63 CHAPTER IV – INTERACTIONS OF THE STRIPE-NECKED MUSK TURTLE AND RAZORBACK MUSK TURTLE IN SYMPATRY AND SYNTOPY IN THE PASCAGOULA RIVER SYSTEM ...... 71 4.1 Introduction ...... 72 4.2 Method ...... 76 4.2.1 Site Selection ...... 76 4.2.2 Trapping ...... 77 v

4.2.3 Morphometrics and Stream Characteristics ...... 78 4.2.4 Statistical Analysis ...... 79 4.2.5 Genetic Analysis ...... 81 4.3 Results ...... 83 4.3.1 Trapping ...... 83 4.3.2 Habitat Associations ...... 84 4.3.3 Genetics...... 85 4.3.4 Morphology...... 86 4.4 Discussion ...... 87 4.5 Literature Cited ...... 108 CHAPTER V – THE RIVERSCAPE GENETICS OF TWO LOTIC MUSK TURTLES IN THE PASCAGOULA RIVER DRAINAGE ...... 116 5.1 Introduction ...... 117 5.2 Methods...... 121 5.2.1 Study Area and Sampling Methods ...... 121 5.2.2 Molecular Methods ...... 121 5.2.3 Data Analysis ...... 122 5.3 Results ...... 125 5.3.1 Sites and Sampling ...... 125 5.3.2 Summary Statistics...... 125 5.3.3 Population Structure in Sternotherus carinatus ...... 126 5.3.4 Population Structure in Sternotherus peltifer ...... 127 5.4 Discussion ...... 128 5.5 Literature Cited ...... 164

LIST OF TABLES

Table 2.1 Characteristics of 40 microsatellite loci isolated for the razorback musk turtle (Sternotherus carinatus)...... 31 Table 2.2 All microsatellite loci developed for Sternotherus carinatus that cross-amplified in at least one kinosternid species...... 34 Table 3.1 Range wide sampling locations for Sternotherus carinatus...... 53 Table 3.2 Summary statistics for Sternotherus carinatus across river drainages...... 54 Table 3.3 Significance testing of allelic richness across populations of Sternotherus carinatus...... 55 Table 3.4 Significance testing of levels of observed heterozygosity across the range of the Sternotherus carinatus...... 56 Table 3.5 Significance testing of levels of expected heterozygosity across the range of the Sternotherus carinatus...... 57 Table 3.6 Significance testing of allelic richness including the Brazos Drainage...... 58 Table 3.7 Pairwise FST values across sites for Sternotherus carinatus...... 59 Table 4.1 Pascagoula River drainage sampling sites in 2018, and the number of turtles per species captured at each site...... 94 Table 4.2 List of all 64 sites where Sternotherus carinatus and S. peltifer were detected across all years in this study in the Pascagoula River drainage...... 99 Table 4.3 Average ancestry coefficients (q-scores) of 211 Sternotherus carinatus and 219 S. peltifer from the Pascagoula River drainage...... 103 Table 5.1 Grouped samples for analysis from the Pascagoula River Drainage for Sternotherus carinatus...... 136 Table 5.2 Grouped samples for analysis from the Pascagoula River Drainage for Sternotherus peltifer...... 139 Table 5.3 List of all sites where Sternotherus carinatus and S. peltifer were detected across all years from the Pascagoula River drainage...... 142 Table 5.4 Summary statistics for 8 microsatellite loci used to genotype Sternotherus carinatus in the Pascagoula River System...... 146 Table 5.5 Summary statistics for 10 microsatellite loci used to genotype Sternotherus peltifer in the Pascagoula River System...... 147 Table 5.6 Pairwise FST values (below diagonal) and river distances (above diagonal; in river kilometer) across 8 sites for Sternotherus carinatus within the Pascagoula River system...... 148 Table 5.7 Average group membership scores (q-scores) for Sternotherus carinatus at each site from the STRUCTURE analysis for K = 2 and K = 3, with the latter being a separate analysis without the Escatawpa River site...... 149 Table 5.8 Rates of recent migration between sites of Sternotherus carinatus as estimated with the program BAYESASS...... 150 Table 5.9 Pairwise FST values (below diagonal) and river distances (above diagonal; in river kilometer) across 7 sites for Sternotherus peltifer within the Pascagoula River system...... 151 Table 5.10 Average group membership scores (q-scores) for Sternotherus peltifer from each group from the STRUCTURE analysis for K = 2, K = 3 and K =4...... 152

vii

Table 5.11 Rates of recent migration between sites of Sternotherus peltifer as estimated with the program BAYESASS...... 153

viii

LIST OF ILLUSTRATIONS

Figure 3.1 The geographic range and sampling locations of Sternotherus carinatus ...... 60 Figure 3.2 Bar plots of membership coefficients from the STRUCTURE analysis of Sternotherus carinatus for values of K = 2 for the range wide analysis, a K = 4 for the GCP turtles, and a K = 2 for the Highlands turtles...... 61 Figure 3.3 Discriminant Analysis of Principle Coordinates for Sternotherus carinatus. .. 62 Figure 4.1 Range map of the lotic Sternotherus species in the Southeastern United States...... 104 Figure 4.2 Capture locations for Sternotherus within the Pascagoula River system ...... 105 Figure 4.3 A CCA plot of all turtle species encountered at >10% of all sites...... 106 Figure 4.4 STRUCTURE plot for 211 Sternotherus carinatus and 222 Sternotherus peltifer from the Pascagoula River drainage...... 107 Figure 5.1 The Pascagoula River drainage and the major subdrainages...... 154 Figure 5.2 Capture locations for Sternotherus within the Pascagoula River drainage. .. 155 Figure 5.3 Isolation by distance among all sites of Sternotherus carinatus as indicated by the significant positive correlation between pairwise genetic distances (FST) and river distance (river kilometers)...... 156 Figure 5.4 Isolation by distance among for Sternotherus carinatus (excluding the Escatawpa population) as indicated by a significant positive correlation between pairwise genetic distance (FST) and pairwise river distance (river kilometer)...... 157 Figure 5.5 ∆K analysis and log-likelihood plots for all Sternotherus carinatus across all sites in the Pascagoula River System...... 158 Figure 5.6 ∆K analysis and log-likelihood plots for Sternotherus carinatus across sites in the Pascagoula River System, excluding the Escatawpa drainage...... 159 Figure 5.7 Bar plots of membership coefficients from the STRUCTURE analysis of Sternotherus carinatus for values of K = 2 and K = 3 for analyses with and without the Escatawpa River turtles...... 160 Figure 5.8 Genetic distance by river km for Sternotherus peltifer. Sternotherus peltifer does not exhibit a significant correlation in pairwise genetic distance (FST) and river distance (river kilometer)...... 161 Figure 5.9 ∆K analysis and Log-likelihood plots for Sternotherus peltifer in the Pascagoula River System ...... 162 Figure 5.10 Bar plots of membership coefficients from the STRUCTURE analysis of Sternotherus peltifer for values of K = 2, K = 3, and K = 4...... 163

CHAPTER I - INTRODUCTION TO TURTLES OF THE SOUTHEASTERN UNITED

STATES

1.1 Introduction

Turtles are among the most imperiled vertebrates on the planet with an estimated

52.2-61% considered by the International Union for the Conservation of Nature (IUCN) to be extinct or in a threatened category - critically endangered, endangered, or vulnerable (TTWG 2017; Lovich et al. 2018). The major causes of these precipitous declines include habitat fragmentation, water quality degradation, and exploitation by

Asian markets; the latter was formerly a regional threat limited to Southeast Asia, but it is now a global threat (Moll and Moll, 2003; Cheung and Dudgeon 2006). The Southeastern

United States ranks number two globally in turtle diversity driven in a large part to the high degree of endemism (e.g., Graptemys spp.; Buhlmann et al. 2009; Lindeman 2013), which is still being characterized with new species described as recently as 2010 (e.g.,

Graptemys pearlensis - Ennen et al., 2010). The high degree of endemism is due in large part to the unique hydrology and geology of the Southeastern United States. The

Appalachian Mountains and the geologic features associated with them (including the

Central highlands) are very old, and they have for some 400 million years served as refugia during both sea level rise and fall (Soltis et al. 2006).

This unique and synergistic combination of time, geology, and hydrology has created a hotspot of turtle diversity in the Southeastern United States. This phenomenon is best illustrated from gamma diversity seen across the elevational gradient from the

Appalachians and associated geologic features down to the Atlantic and Gulf Coastal plains (e.g., in states like Georgia, Alabama, South Carolina, etc.). From the highland

endemics in the Blue Ridge mountains like the bog turtle (Glyptemys muhlenbergii) into the Piedmont where more common freshwater turtles can be found (e.g., Trachemys scripta, Pseudemys concinna and Apalone spinifera), to below the Fall Line where one begins to see coastal plains forms of the same genera (Apalone ferox, Pseudemys floridana) along with coastal plain endemics like the gopher tortoise (Gopherus polyphemus; Ernst and Lovich 2009). However, the bulk of Southeastern turtle diversity stems from highly aquatic turtle species endemic to a single Gulf Coast drainages like the map and sawback turtles (Genus Graptemys) where 11 of 14 recognized species are tied to one or two river systems, or like the alligator snapping turtles (Genus Macrochelys: 1 of 2 recognized species is endemic to a single drainage; Lindeman 2013; Thomas et al.

2014; Folt and Guyer 2015).

While there is a growing body of literature for some turtles in the Gulf coastal plain, there is still a substantial lack of fundamental research on the ecology and evolutionary history of other species, and this is especially pronounced in the small, bottom-walking turtles of the family Kinosternidae (Lovich and Ennen 2013). These small turtles seem to have been overlooked in the literature, perhaps due to their diminutive size and cryptic behavior (i.e., these turtles are not as visible as other turtle species that spend a great deal of time basking). For example, in a recent review of the literature, the razorback musk turtle (Sternotherus carinatus) is considered one of the top- ten least studied turtle species in the United States and Canada (Lovich and Ennen, 2013) despite being fairly common and broadly distributed across many drainages. Since the publication of Lovich and Ennen (2013), a new species of Sternotherus has been described from what was formerly believed to be a hybrid zone between two presumed

subspecies, Sternotherus minor peltifer and Sternotherus minor minor (Scott et al. 2018).

Genetic analysis determined that the “hybrid” zone actually reflected a unique and distinct evolutionary lineage, warranting the description of the intermediate musk turtle

(Sternotherus intermedius) and the elevation of the loggerhead (S. minor) and stripe- necked musk turtle (S. peltifer; Scott et al. 2018). If Lovich and Ennen (2013) were to reevaluate amount of knowledge (i.e., number of publications and text) on each species in the Sternotherus minor complex, the lotic Sternotherus (that is, all Sternotherus aside from S. odoratus) would likely constitute more of the top 10 least studied turtles than any other genus.

The foci of this dissertation are some of the small, pugnacious turtles of the family Kinosternidae from the Central and Southeastern United States. Kinosternidae is a

New World family of turtles that is comprised of twenty-five mud and musk turtle species (Spinks et al., 2014). The mud turtles (genus Kinosternon) are characterized by a double-hinged plastron, similar in function to that of North American box turtles (Genus

Terrapene). This allows some mud turtle species to close their carapace very tightly for protection, a feature that is tied to their ecology. Mud turtles are semi-aquatic, but many species spend significant portions of the year terrestrially, typically aestivating when wetlands or seasonal streams dry (Wygoda 1979; Christiansen et al. 1985; Ernst and

Lovich 2009; Legler and Vogt 2013). The musk turtles, on the other hand, are fully aquatic and rarely make overland journeys or aestivate on land (Tinkle 1958; Ernst and

Lovich 2009). Sternotherus come in two main varieties: lotic and lentic. There is one widespread musk turtle species, the eastern musk turtle (Sternotherus odoratus), a generalist that is primarily lentic, which ranges from Canada, across the Eastern United

States, and with historic records in Mexico (Ernst and Lovich 2009; Vogt and Legler

2013). The remaining species are primarily lotic and inhabit the streams and rivers of the

Central and Southeastern United States. From the Brazos River in Texas in the western portion of its range to the Pascagoula River drainage in the east, streams and rivers are inhabited by the razorback musk turtle (Sternotherus carinatus); the stripe-necked musk turtle (S. peltifer) inhabits lotic systems from the Pearl River to the Mobile River and into parts of the Tennessee River drainage; the critically-endangered (IUCN designation) and federally threatened flattened musk turtle (Sternotherus depressus) is endemic to the

Black Warrior River drainage above the fall line in Alabama; the intermediate musk turtle (S. intermedius) is found in the Perdido, Escambia and Choctawhatchee River drainages; and lastly, the loggerhead musk turtle (S. minor) ranges from the

Chattahoochee River drainages to the Altamaha River drainages in Georgia, and down into a number of springs and river systems into Central Florida. All musk turtles are primarily carnivorous, feeding on insects, insect larva, crustaceans, snails, and bivalves, but they also take some plant material (e.g. filamentous algae and seeds; Ernst and

Lovich 2009; Atkinson 2013).

Sternotherus carinatus is still considered a small-to-medium sized turtle (128 –

209 mm), despite being the largest in the genus (Lindeman 2008). It is associated with rivers, streams and oxbows where it probes the stream bottom for mollusks, seeds and carrion (Lindeman 2008; Ernst and Lovich 2009; Atkinson 2013). Males and females are sexually dimorphic with males attaining a larger size, larger heads and longer tail (the cloaca extends past the posterior marginal scutes) and rough pads of scales on the interior of the hind legs (Ernst and Lovich 2009; Kavanaugh and Kwiatkowski 2016).

Sternotherus carinatus differ from other members of the genus by having only 10 plastral scutes (lacking the gular shared by the other species). Sternotherus carinatus is highly aquatic and no published records of any overland movements exist. Because the species is so highly aquatic and is distributed across many Gulf Coast drainages, the potential for cryptic speciation cannot be ruled out. For example, there are other genera of aquatic turtles which are morphologically conserved like S. carinatus, yet they exhibit high levels of endemism to some of these same Gulf Coast drainages (e.g., Graptemys pulchra clade and Macrochelys – Lovich and McCoy 1992; Ennen et al. 2010, Thomas et al. 2012).

The stripe-necked musk turtle was originally described as a full species

(Sternotherus peltifer) by Smith and Glass (1947) from a specimen taken from an unknown stream in Bassfield, Mississippi, from the Pascagoula River watershed. The stripe-necked musk turtle was then moved and considered a subspecies to the razorback musk as S. carinatus peltifer until Tinkle and Webb (1955) re-evaluated the genus with the description of Sternotherus depressus. When the authors describe formally

Sternotherus depressus as a new species, they elevated Sternotherus carinatus minor to full species status, with two subspecies: the loggerhead musk turtle (Sternotherus minor minor) and the stripe-necked musk turtle (S. m. peltifer). They did this based on the distinct morphology of Sternotherus carinatus (e.g., the pronounced and acutely keeled carapace) as well as having only 10 plastral scutes (lacking the gular shared by other

Sternotherus). They also cite that the co-occurrence of S. carinatus and S. m. peltifer within the Pascagoula as further evidence of species status, even though they were unable to catch S. m. peltifer in the Pascagoula watershed after sampling the type locality twice, along with other streams, rivers and ponds in the drainage. They were only able to turn up

Sternotherus odoratus and S. carinatus, and they alluded that the type locality may be invalid. All Sternotherus were temporarily placed in the genus Kinosternon following

Seidel et al. (1986) and Iverson (1991) due to lack of synapomorphies for the species in

Sternotherus. However, Iverson (1998) later reversed this based on a larger phylogenetic analysis of the family. Taxonomy of the Sternotherus was stable until a cryptic species of

Sternotherus was described from what was formerly believed to be a zone of intergradation between S. minor minor and S. minor peltifer. Scott et al. (2018) used phylogenomic approaches that suggested the supposed intergrades from the Escambia,

Perdido and Choctawhatchee river drainages represented an independent evolutionary lineage. The description of Sternotherus intermedius elevated Sternotherus minor and S. peltifer to species status by default.

The species of the Sternotherus minor complex are smaller than Sternotherus carinatus. They range from 80-145 mm in carapace length (Ernst and Lovich 2009). As primarily molluscivorous turtles, many populations can develop a large, hypertrophied skulls with powerful jaws meant to consume snails, bivalves, and crayfish. Because of its recent description, there is not much literature on S. intermedius, but there is some literature on S. minor (Odum 1957; Iverson 1977; Iverson 1978; Gatten 1984; Cox et al.

1991; Pfaller et al. 2011). However, very little work has been done on S. peltifer, especially in the western, peripheral drainages such as the Pearl and Pascagoula where remarkably few records exist (Iverson 1977). In line with Tinkle and Webb (1955),

Iverson (1977) believed the type locality for S. peltifer may have been invalid, as there are some streams in Bassfield, MS, that are part of the Pearl River drainage (where other

records existed). The species was not officially (re)confirmed from the Pascagoula river drainage until 1978 (Vogt, McCoy and Bianculli 1978).

The Pearl River drainage is the westernmost extent of the stripe-necked musk turtle’s range, and the species terminates in Washington Parish, Louisiana. Given the very similar ecologies of Sternotherus peltifer and S. carinatus, it is surprising that the two can occur sympatrically in the Pearl and Pascagoula Rivers. Not only do the two occur sympatrically, but the two are known to occur in syntopy as well (McCoy,

Bianculli and Vogt 1978; Lindeman 1996). These two species occurring in syntopy is seemingly a violation of the competitive exclusion principle that suggests two species that compete for the same resources cannot coexist (Hardin 1960).

For two strikingly similar species to occur in syntopy creates a number of interesting questions, not only for the turtle species themselves, but in evolutionary biology and the forces that create and maintain biodiversity. There is also potential for the two species to hybridize, as hybridization in turtles has been reported widely, from sea turtles (Karl et al. 1995), map turtles (Freedberg and Myers, 2012; Godwin et al.

2014), Geoemydids (Stuart and Parham 2007) and even within Sternotherus itself (Scott and Rissler, 2015). What is unique to this congeneric interaction is the sympatry across two drainages as well as the allopatry across multiple drainages. There are many parallels to these two turtle species and their sympatric interactions to other taxa with interesting ecological and evolutionary histories. Duvernell et al. (2013) and Duvernell and Schaefer

(2014) examined a similar relationship between the two species of top-minnows,

Fundulus notatus and F. olivaceus. When in sympatry, F. olivaceus is typically more abundant in headwater stream habitats like S. peltifer, whereas F. notatus is more

abundant in larger rivers and backwater areas like S. carinatus (Braasch and Smith 1965;

Thomerson 1966; Thomerson and Wooldridge 1970; Duvernell et al. 2013). In allopatry, the two fish species can be found in all habitats (Thomerson 1966), just like these two species of Sternotherus (Kavanagh 2016; P. Scott, pers. comm.). The commonalities shared by these taxa provide a unique platform to compare how rivers shape the ecology and evolutionary history of highly aquatic organisms.

A commonality of each of these aforementioned Sternotherus species (with the exception of Sternotherus depressus – a federally protected species) is that currently all are classified as Least Concern by the International Union for Conservation of Nature

(IUCN). However, each species has become increasingly popular in the pet trade in

Southeast Asia with hundreds of thousands being exported annually (Mali et al. 2014; J.

Jensen and W. Selman pers. comm.). Unfortunately, Southeast Asia has already decimated native populations of turtles and tortoises (see Asian Turtle Crisis: Cheung and

Dudgeon, 2006), and in recent decades has turned to importing metric tons of turtles

(>1186 tons from 1992-1999) into markets to satiate appetites for this species in Chinese food and pet markets (Ades et al. 2000). Due to their diminutive size and amicable nature as pets, the popularity and demand for kinosternids has surged. As evidence, during my studies, I have been contacted multiple times via social media outlets about selling the turtles I catch during my research.

In order to ultimately conserve or manage these turtles, it is important to study the ecology and evolutionary histories of these species. To better understand the evolutionary histories of these Sternotherus species, it is important to pick informative genetic markers to elucidate the patterns of phylogeography and potential hybridization or introgression

of each species. Very few studies have looked to characterize the population structure or interactions amongst the focal species of this dissertation. For example, in Sternotherus, most phylogenetic studies are dated and characterized population genetic structure with restriction fragment length polymorphisms (RFLPs) or used the polyermase chain reaction (PCR) to amplify a short portion of the mitochondrial DNA (mtDNA) control region (Walker et al. 1995; Walker et al. 1998). These methods are useful for coarse characterization of genetic structure and are valuable for reconstructing phylogenies since the circular DNA of the mitochondria is not subject to recombination, and it tends to be uniparentally inherited (or in turtles at least; Rafinski and Babik 2000).

More recently, microsatellite loci (also known as variable number of tandem repeats to describe their short repeat motifs of 1-6 nucleotide repeats) are have been employed in many studies of population genetics. Rather than making inferences from a single locus, like in the previous studies of mtDNA, microsatellite loci are from multiple sites in the nuclear genome, they are selectively neutral, more prone to mutation, and therefore provide a more detailed look at evolutionary histories and trajectories of turtle populations (Jehle and Arntzen 2002). Microsatellites have been used to evaluate the conservation genetics of a variety of North American turtle species including the federally-listed Graptemys flavimaculata and G. oculifera (Ennen et al. 2010; Selman et al. 2013; Gaillard et al. 2015), as well as federally-listed populations of Gopherus polyphemus (Gaillard et al. 2011 ), and Clemmys guttata (Davy and Murphy 2012). They have also been used in studies to elucidate hybridization events in turtles (G. ernsti and

G. barbouri hybrids - Godwin et al. 2014). I employ the use of these versatile molecular markers herein to characterize levels of range wide population structure, evaluate levels

of introgression and hybridization, and compare levels of intra-drainage population structure in the stripe-necked and razorback musk turtles. The results of these studies have tangible applications to the fields of conservation biology, evolutionary biology, landscape (riverscape) ecology, and to the fields of population and community ecology.

1.2 Literature Cited

Ades, G., Banks, C., Buhlmann, K. A., Chan, B., Chang, H., Chen, T., Crow, P., Haupt,

H., Kan, R., Lai, J.-Y., Lau, M., Lin, H.-C., and Haitao, S. 2000. Turtle trade in

northeast Asia: regional summary. In Asian Turtle Trade-Proceedings of a

Workshop on Conservation and Trade of Freshwater Turtles and Tortoises in

Asia, Phnom Penh, Cambodia, 1-4 December 1999. Chelonian Research

Monographs 52–54.

Atkinson, C. 2013. Razor-backed musk turtle (Sternotherus carinatus) diet across a

gradient of invasion. Herpetological Conservation and Biology 8:561–570.

Barkhouser, T. D. 1996. Geographic Variation: A Test of the Conservative Effect of

Gene Flow in Sternotherus carinatus (Gray, 1856). Master’s Thesis. University of

Southwestern Louisiana, Lafayette, Louisiana.

Braasch, M. E., and Smith, P. W. 1965. Relationships of the topminnows Fundulus

notatus and Fundulus olivaceus in the Upper Mississippi River Valley. Copeia

18:46–53.

Buhlmann, K. A., Akre, T. S. B., Iverson, J. B., Karapatakis, D., Mittermeier, R. A.,

Georges, A., Rhodin, A. G. J., van Dijk, P. P., and Gibbons, J. W. 2009. A Global

Analysis of Tortoise and Freshwater Turtle Distributions with Identification of

Priority Conservation Areas. Chelonian Conservation and Biology. 8:116–149.

Cheung, S. M., and Dudgeon, D. 2006. Quantifying the Asian turtle crisis: Market

surveys in southern China, 2000–2003. Aquatic Conservation: Marine and

Freshwater Ecosystems. 16:751–770.

Christiansen, J. L., Cooper, J. A., Bickham, J. W., Gallaway, B. J., and Springer, M. D.

1985. Aspects of the Natural History of the yellow mud murtle (Kinosternon

flavescens; Kinosternidae) in Iowa: A proposed endangered species. The

Southwestern Naturalist 30: 413–425.

Cox, W. A., Hazelrig, J. B., Turner, M. E., Angus, R. A., and Marion, K. R. 1991. A

model for growth in the musk turtle, Sternotherus minor, in a north Florida spring.

Copeia 954–968.

Crain, J. L., and Cliburn, J. W. 1972. Sternothaerus minor peltifer in Louisiana. The

Southwestern Naturalist 16:444–445.

Davy, C. M., and Murphy, R. W. 2014. Conservation genetics of the endangered spotted

turtle (Clemmys guttata) illustrate the risks of “bottleneck tests.” Canadian

Journal of Zoology 92:149–162.

Duvernell, D. D., Meier, S. L., Schaefer, J. F., and Kreiser, B. R. 2013. Contrasting

phylogeographic histories between broadly sympatric topminnows in the

Fundulus notatus species complex. Molecular Phylogenetics and Evolution

69:653–663.

Duvernell, D. D., and Schaefer, J. F. 2014. Variation in contact zone dynamics between

two species of topminnows, Fundulus notatus and F. olivaceus, across isolated

drainage systems. Evolutionary Ecology 28:37–53.

Ennen, J. R., Kreiser, B. R., Qualls, C. P., and Lovich, J. E. 2010. Morphological and

molecular reassessment of Graptemys oculifera and Graptemys flavimaculata

(Testudines: Emydidae). Journal of Herpetology 44:544–554.

Ennen, J. R., Lovich, J. E., Kreiser, B. R., Selman, W., and Qualls, C. P. 2010. Genetic

and morphological variation between populations of the Pascagoula map turtle

(Graptemys gibbonsi) in the Pearl and Pascagoula Rivers with description of a

new species. Chelonian Conservation and Biology 9:98–113.

Ennen, J.R., R.D. Birkhead, B.R. Kreiser, D.L. Gaillard, C.P. Qualls and J.E. Lovich.

2011. The effects of isolation on the demography and genetic diversity of long-

lived species: implications for the conservation and management of the gopher

tortoise (Gopherus polyphemus). Herpetological Conservation and Biology 6:202-

214.

Ernst, C. H., and Lovich, J. E. 2009. Turtles of the United States and Canada. JHU Press.

Folt, B., and Guyer, C. 2015. Evaluating recent taxonomic changes for alligator snapping

turtles (Testudines: Chelydridae). Zootaxa 3947:447-450.

Freedberg, S., and Myers, E. M. 2012. Cytonuclear equilibrium following interspecific

introgression in a turtle lacking sex chromosomes. Biological Journal of the

Linnean Society 106:405–417.

Gaillard, D., Ennen, J. R., Kreiser, B. R., Qualls, C. P., Sweat, S. C., Birkhead, R.,

Tuberville, T. D., Aresco, M., McCoy, E. D., and Mushinsky, H. R. 2017. Range-

wide and regional patterns of population structure and genetic diversity in the

gopher tortoise. Journal of Fish and Wildlife Management 8:497–512.

Gaillard, D. L., Selman, W., Jones, R. L., Kreiser, B. R., Qualls, C. P., and Landry, K.

2015. High connectivity observed in populations of ringed sawbacks, Graptemys

oculifera, in the Pearl and Bogue Chitto Rivers using Six microsatellite loci.

Copeia 103:1075–1085.

Gatten Jr, R. E. 1984. Aerobic and anaerobic metabolism of freely-diving loggerhead

musk turtles (Sternotherus minor). Herpetologica 40:1–7.

Godwin, J. C., Lovich, J. E., Ennen, J. R., Kreiser, B. R., Folt, B., and Lechowicz, C.

2014. Hybridization of two megacephalic map turtles (Testudines: Emydidae:

Graptemys) in the Choctawhatchee River drainage of Alabama and Florida.

Copeia 2014:725–742.

Hardin, G. 1960. The Competitive Exclusion Principle. Science 131:1292–1297.

Iverson, J. B. 1998. Molecules, morphology, and mud turtle phylogenetics (family

Kinosternidae). Chelonian Conservation and Biology 3:113–117.

Iverson, John B. 1977. Geographic variation in the musk turtle, Sternotherus minor.

Copeia 1977:502–517.

Iverson, John B. 1978. Reproductive cycle of female loggerhead musk turtles

(Sternotherus minor minor) in Florida. Herpetologica 34:33–39.

Iverson, John B. 1991. Phylogenetic hypotheses for the evolution of modern Kinosternine

turtles. Herpetological Monographs 5:1–27.

Jehle, R., and Arntzen, J. W. 2002. Microsatellite markers in amphibian conservation

genetics. Herpetological Journal 12:1–9.

Karl, S. A., Bowen, B. W., and Avise, J. C. 1995. Hybridization among the ancient

mariners: characterization of marine turtle hybrids with molecular genetic assays.

Journal of Heredity 86:262–268.

Kavanagh, B. T., and Kwiatkowski, M. A. 2016. Sexual Dimorphism, Movement

Patterns, and Diets of Sternotherus carinatus (razorback musk turtle).

Southeastern Naturalist 15:117–133.

Legler, J. M., and Vogt, R. C. 2013. The Turtles of Mexico: Land and Freshwater Forms.

University of California Press.

Lindeman, P. V. 1996. Distribution, relative abundance, and basking ecology of the

razorback musk turtle, Kinosternon carinatum, in the Pearl and Pascagoula River

drainages. Herpetological Natural History 4:23–34.

Lindeman, P. V. 2008. Sternotherus carinatus. In Conservation Biology of Freshwater

Turtles and Tortoises: A Compilation Project of the IUCN/SSC Tortoise and

Freshwater Turtle Specialist Group pp. 012.1–012.6.

Lindeman, Peter V. 2013. The Map Turtle and Sawback Atlas: Ecology, Evolution,

Distribution, and Conservation. University of Oklahoma Press.

Lovich, J. E., and Ennen, J. R. 2013. A quantitative analysis of the state of knowledge of

turtles of the United States and Canada. Amphibia-Reptilia 34:11–23.

Lovich, J. E., Ennen, J. R., Agha, M., and Gibbons, J. W. 2018. Where have all the turtles

gone, and why does it matter? BioScience 68:771–781.

Mali, I., Vandewege, M. W., Davis, S. K., and Forstner, M. R. J. 2014. Magnitude of the

freshwater turtle exports from the US: long term trends and early effects of newly

implemented harvest management regimes. PLoS ONE 9:e86478.

Moll, D., and Moll, E. O. 2004. The Ecology, Exploitation and Conservation of River

Turtles. Oxford University Press, USA.

Odum, H. T. 1957. Trophic structure and productivity of Silver Springs, Florida.

Ecological Monographs 27:55–112.

Pfaller, J. B., Gignac, P. M., and Erickson, G. M. 2011. Ontogenetic changes in jaw-

muscle architecture facilitate durophagy in the turtle Sternotherus minor. Journal

of Experimental Biology 214:1655–1667.

Rafiński, J., and Babik, W. 2000. Genetic differentiation among northern and southern

populations of the moor frog Rana arvalis Nilsson in central Europe. Heredity

84:610–618.

Scott, P. A., Glenn, T. C., and Rissler, L. J. 2018. Resolving taxonomic turbulence and

uncovering cryptic diversity in the musk turtles (Sternotherus) using robust

demographic modeling. Molecular Phylogenetics and Evolution 120:1–15.

Scott, P. A., and Rissler, L. J. 2015. Integrating dynamic occupancy modeling and

genetics to infer the status of the imperiled flattened musk turtle. Biological

Conservation 192:294–303.

Seidel, M. E., and Lucchino, R. V. 1981. Allozymic and morphological variation among

the musk turtles Sternotherus carinatus, S. depressus and S. minor

(Kinosternidae). Copeia 1981:119–128.

Selman, W., Kreiser, B., and Qualls, C. 2013. Conservation genetics of the yellow-

blotched sawback Graptemys flavimaculata (Testudines: Emydidae).

Conservation Genetics 14:1193–1203.

Smith, H. M., and Glass, B. P. 1947. A new musk turtle from Southeastern United States.

Journal of the Washington Academy of Sciences 37:22–24.

Soltis, D. E., Morris, A. B., McLachlan, J. S., Manos, P. S., and Soltis, P. S. 2006.

Comparative phylogeography of unglaciated eastern North America:

phylogeography of unclaciated eastern North America. Molecular Ecology,

15:4261–4293.

Spinks, P. Q., Thomson, R. C., Gidiş, M., and Bradley Shaffer, H. 2014. Multilocus

phylogeny of the New-World mud turtles (Kinosternidae) supports the traditional

classification of the group. Molecular Phylogenetics and Evolution. 76:254–260.

Stuart, B. L., and Parham, J. F. 2007. Recent hybrid origin of three rare Chinese turtles.

Conservation Genetics 8:169–175.

Thomas, T. M., Granatosky, M. C., Bourque, J. R., Krysko, K. L., Moler, P. E., Gamble,

T., Suarez, E., Leone, E., and Roman, J. 2014. Taxonomic assessment of alligator

snapping turtles (Chelydridae: Macrochelys), with the description of two new

species from the southeastern United States. Zootaxa 3786:141–165.

Thomerson, J. E. 1966. A comparative biosystematic study of Fundulus notatus and

Fundulus olivaceus (Pisces: Cyprinodontidae). Tulane Studies of Zoology 13:29–

47.

Thomerson, J. E., and Wooldridge, D. P. 1970. Food habits of allotopic and syntopic

populations of the topminnows Fundulus olivaceus and Fundulus notatus.

American Midland Naturalist 84:573–576.

Tinkle, D. W. 1956. The Systematics and Ecology of the Sternotherus Carinatus Complex

(Testudinata: Chelydridae). Dissertation. Tulane University. New Orleans,

Louisiana.

Tinkle, D. W., and Webb, R. G. 1955. A new species of Sternotherus with a discussion of

the Sternotherus carinatus complex (Chelonia, Kinosternidae). Tulane Studies in

Zoology 3:53–67.

Turtle Taxonomy Working Group, Rhodin, A. G. J., Iverson, J. B., Bour, R., Fritz, U.,

Georges, A., Shaffer, H. B., and van Dijk, P. P. 2017. Turtles of the world:

annotated checklist and atlas of taxonomy, synonymy, distribution, and

conservation status (8th Ed.). Chelonian Research Foundation and Turtle

Conservancy.

Vogt, R. C., McCoy, C. J., and Bianculi, A. V. 1978. Sternotherus minor (Reptilia,

Kinosternidae) in the Pascagoula River system, Mississippi. Herpetological

Review 9:109.

Walker, D., Burke, V. J., Barák, I., and Avise, J. C. 1995. A comparison of mtDNA

restriction sites vs. control region sequences in phylogeographic assessment of the

musk turtle (Sternotherus minor). Molecular Ecology 4:365–374.

Walker, Deette, Ortí, G., and Avise, J. C. 1998. Phylogenetic distinctiveness of a

threatened aquatic turtle (Sternotherus depressus). Conservation Biology 12:639–

645.

Wygoda, M. L. (1979). Terrestrial activity of striped mud turtles, Kinosternon baurii

(Reptilia, Testudines, Kinosternidae) in West-Central Florida. Journal of

Herpetology, 13:469–480.

CHAPTER II - CHARACTERIZATION OF MICROSATELLITE LOCI FOR THE

RAZORBACK MUSK TURTLE (STERNOTHERUS CARINATUS; GRAY 1856) AND

THEIR CROSS-AMPLIFICATION IN FIVE OTHER SPECIES IN THE FAMILY

KINOSTERNIDAE

Abstract

Studies of mud and musk turtles (Family Kinosternidae) are underrepresented in the primary turtle literature. While many turtle species have benefitted from population genetics studies that used microsatellite loci, none have been isolated for any kinosternid species. We have developed and optimized loci for Sternotherus carinatus and tested their ability to cross-amplify in five other kinosternids. These loci should provide a useful set of tools for future population genetic studies of kinosternid species.

2.1 Introduction

Turtles are among the most imperiled vertebrates on the planet with an estimated

52.2-61% considered by the International Union for the Conservation of Nature (IUCN) to be extinct or in a threatened category - critically endangered, endangered, or vulnerable (TTWG 2017; Lovich et al. 2018). Genetic studies have become an important conservation tool in biology by evaluating the genetic health of populations, resolving taxonomic uncertainties and identifying evolutionarily significant units (Frankham 2003).

Microsatellite loci have long been used in biology to address questions of population genetics (Selkoe and Toonen 2006), and they have important applications in the field of conservation biology. While there is a rapidly expanding amount of literature on the

biology of turtles, some species are receiving more attention than others (Lovich and

Ennen 2013). For instance, the majority of microsatellite loci used in population genetic studies have been optimized for emydid species such as the European pond turtle (Emys orbicularis; Ciofi et al. 2009), the bog turtle (Glyptemys muhlenbergii; King and Julian

2004), the Blanding’s turtle (Emydoidea blandingii; Osentoski et al. 2002; Libants et al.

2004), and the diamondback terrapin (Malaclemmys terrapin; Hauswaldt and Glenn

2003). The microsatellite loci from these studies have also been tested for cross- amplification and optimized for other species such as the spotted turtle (Clemmys guttata;

Davy and Murphy 2014) and several Graptemys species (King and Julian 2004; Selman et al. 2009), but very few (only three out of forty) of these microsatellite loci have been successfully cross-amplified in a mud or musk turtle species (Hauswaldt and Glenn 2003;

Schwartz et al. 2003).

Even though the taxonomy of Kinosternidae (mud and musk turtles) is still a work in progress, as made evident by the recent recognition of cryptic diversity within the genera Sternotherus (Scott et al. 2018) and Kinosternon (López-Luna et al. 2018), there are currently thirty described species within Kinosternidae, making it one of the most diverse families of turtles globally (Turtle Taxonomy Working Group 2017). However, mud and musk turtles are among the least-studied turtles in the United States and Canada

(Lovich and Ennen 2013), and relatively little work has been conducted on many of the

Central and South American varieties. This highlights the need for versatile genetic markers for this group of turtles.

In this study, we identified microsatellite loci for the razorback musk turtle

(Sternotherus carinatus), a small to medium-sized (up to 18 cm) turtle with

predominately lotic tendencies. It occupies rivers, streams and oxbows from the Brazos

River drainage in Central Texas to the Escatawpa River drainage in Southwestern

Alabama and into some Mississippi River drainages in Southeastern Oklahoma and

Arkansas (Lindeman 2008). We also tested these loci for cross amplification in five other kinosternid species: the stripe-necked musk turtle (Sternotherus peltifer), the eastern musk turtle (Sternotherus odoratus), the eastern mud turtle (Kinosternon subrubrum), the striped mud turtle (Kinosternon baurii) as well as the federally-protected flattened musk turtle (Sternotherus depressus; US Fish and Wildlife Service 1987), a species endemic to Alabama where it is considered of “High Conservation Concern”, as well as critically endangered by the IUCN (2011). Identifying variable genetic markers for the razorback musk turtle will provide the tools necessary to help fill gaps in our understanding of the ecology and evolutionary history of this species and others in this overlooked family.

2.2 Methods

We trapped razorback musk turtles (Sternotherus carinatus) from the Leaf River in Hattiesburg, Mississippi, using 3-foot diameter hoop nets baited with sardines in soybean oil (Beach Cliff). We collected a 5 mm snippet of interdigital webbing from the hindfoot of each turtle to store as a genetic sample. The sample was preserved in 100% ethanol until total genomic DNA was extracted using a DNeasy Tissue Kit (QIAGEN

Inc.). Genomic DNA from one razorback musk turtle was sent to the Savannah River

Ecology Lab Molecular Ecology Lab (SREL MEL), where they prepared an Illumina pair-end shotgun library for sequencing (Lance et al. 2013) and subsequently used

PAL_FINDERv0.02.03 (Castoe et al. 2012) to identify di-, tri-, tetra-, penta- and

hexanucleotide microsatellite sequences (i.e., nucleotide repeat motifs) among the reads.

Primer3 (Rozen and Skaletsky 1999) was used to develop primers for 6,766 potential microsatellite loci. We screened a total of 60 penta- and tetranucleotide loci, focusing on loci whose sequences only appeared one to two times within the sequence reads, to reduce the likelihood of targeting repetitive loci.

We initially screened potential microsatellite loci for amplification success and variability across five individuals of Sternotherus carinatus. For those loci that amplified well, we subsequently genotyped 30 Sternotherus carinatus from the Leaf River in

Hattiesburg, Mississippi (31.337910°N, 89.280059°W; WGS84). We also tested all loci for cross-amplification across five other kinosternids (Kinosternon baurii [N = 5], K. subrubrum [N = 5], S. depressus [N = 5], S. peltifer [N = 5], and S. odoratus [N = 4]).

Amplifications were conducted in a total volume of 12.5 µL using 7.76 µL of dH2O, 1.25

µL 10× standard Taq (Mg-free) buffer (New England Biolabs), 0.75 µL 2 mM dNTPs,

0.75 µL 25 mM MgCl2, 0.25 units Taq polymerase (New England Biolabs), 0.4 µL of 10 mM M13 tailed forward (Boutin-Ganache et al. 2001) and reverse primers, 0.09 µL of 1

µM labeled M-13 primer (LICOR Co.), and 20-50 ng of DNA template. PCR cycling conditions were as follows: initial denaturation at 94°C for 2 minutes, 35 cycles of 30 seconds at 94°C, 30 seconds at a locus-specific annealing temperature of 54-56°C (see

Table 2.1) and 1 minute at 72°C, with a final elongation of 10 minutes at 72°C.

Microsatellites alleles were visualized on a polyacrylamide gel using a LICOR 4300

DNA Analyzer. Alleles were sized using GeneProfiler ver. 4.05 (LICOR Co.). Some loci for S. carinatus required further optimization in terms of PCR conditions with modifications made to the annealing temperature and the amount of Taq polymerase

(Table 2.1). For the additional kinosternid species in this study, we only tested for amplification under standard PCR conditions with an annealing temperature of 56°C. We calculated summary statistics for each locus using GenAlEx 6.5 (Peakall and Smouse

2012), and we tested for Hardy-Weinberg equilibrium within loci and linkage disequilibrium between loci using the probability tests of GENEPOP for the web

(Raymond and Rousset 1995; Rousset 2008). We adjusted our alpha values using a sequential Bonferonni correction to account for multiple comparisons (Rice 1989).

2.3 Results

Of the 60 loci we tested, 40 amplified in one or more species (Table 2.1). Of these, we optimized reaction conditions for 25 microsatellite loci that amplified in

Sternotherus carinatus samples (N=30). Numbers of alleles for these loci ranged from 2 to 15 (mean = 7.6, SE = 3.4) for S. carinatus, with observed heterozygosity values of

0.138 to 0.963 (mean = 0.657, SE = 0.202) and expected heterozygosity values of 0.128 to 0.910 (mean = 0.687, SE = 0.207). Only one locus deviated significantly from Hardy-

Weinberg equilibrium (Scar24), and no loci exhibited linkage disequilibrium. For the additional kinosternid species in this study, the total number of successfully amplified polymorphic loci was 26 for Sternotherus odoratus, 16 for S. peltifer, 13 for S. depressus,

14 for Kinosternon bauriii, and 6 for K. subrubrum (Table 2.2).

2.4 Discussion

Mud and musk turtles of the family Kinosternidae are among some of the least- studied turtles in the world (Lovich and Ennen 2013), and this study is the first to optimize microsatellite loci designed specifically for a kinosternid turtle. The loci were designed by using DNA from the razorback musk turtle, but they cross-amplified well in

five other species of mud and musk turtles. Microsatellite loci remain an affordable and effective method to study population genetics (Selkoe and Toonen 2006). They can be used at relatively little cost, which can be reduced further by multiplexing loci with alleles in different size ranges (e.g., Scar08 and Scar19; Scar24 and Scar36).

Furthermore, in contrast to next-gen approaches like RAD-Seq, microsatellite loci are amenable to adding additional individuals to the dataset as opportunity and funding allows.

Although populations of most kinosternids are considered stable across the

Southeastern United States (with the notable exception of the critically endangered

Sternotherus depressus), there is still much we do not know about the evolutionary history of these species, as made evident by the recent description of Sternotherus intermedius (Scott et al. 2018). The loci described herein should prove useful for studies evaluating the contemporary and historic levels of connectivity (i.e., gene flow) in kinosternid species as well as evaluating levels of genetic diversity within and among populations. Because lower genetic diversity has been correlated with lower fitness in turtles (Ennen et al. 2010), it is important to understand gene flow among isolated or fragmented populations that may be susceptible to genetic drift or inbreeding depression, particularly in threatened species (Ennen et al. 2010; Shaffer et al. 2015). A better understanding of the population genetics of these species will allow for more informed management decisions and will benefit the long-term survival of a species. We are currently using these loci to evaluate range-wide population structure in razorback musk turtles and to assess patterns of co-ocurrence and potential hybridization in S. carinatus and S. peltifer (unpublished data).

Acknowledgments

This work has been funded by the following grants (in alphabetical order): the

American Turtle Observatory, Birmingham Audubon Society, Chicago Herpetological

Society, NSF Graduate Research Fellowship (#1842492), and the Theodore Roosevelt

Memorial Grant through the American Museum of Natural History. We thank Rochelle

Beasley and Stacey Lance (Savannah River Ecology Lab) for Illumina Sequencing and processing of reads. We thank Cybil C. Huntzinger, Luke S. Pearson and Gabbie A.

Berry for assistance in the field and collection of tissue samples. All tissue samples were collected under scientific collecting permit MMNS #090915 in accordance with IACUC protocol #17101202 from the University of Southern Mississippi.

2.5 Literature Cited

Alacs, E. A., and Janzen, F. J. and Scribner, K. T. 2007. Genetic issues in freshwater

turtle and tortoise conservation. Chelonian Research Monographs 4:107–123

Allendorf, F.W. and G. Luikart. 2007. Conservation and the genetics of populations.

Blackwell Publishing, Malden, Massachusetts, USA, 664 pp.

Castoe, T. A., Poole, A. W., de Koning, A. P. J., Jones, K. L., Tomback, D. F., Oyler-

McCance, S. J., Lance, S. L., Streicher, J. W., Pollock, D. D. 2012. Rapid

microsatellite identification from Illumina paired-end genomic sequencing in two

birds and a snake. PLoS ONE 7:e30953.

Ciofi, C., Tzika, A. C., Natali, C., Chelazzi, G., Naziridis, T., and Milinkovitch, M. C.

2009. Characterization of microsatellite loci in the European pond turtle Emys

orbicularis. Molecular Ecology Resources 9:189–191.

Davy, C. M., and Murphy, R. W. 2014. Conservation genetics of the endangered Spotted

Turtle (Clemmys guttata) illustrate the risks of “bottleneck tests.” Canadian Journal

of Zoology 92:149–162.

Ennen, J. R., Kreiser, B. R., and Qualls, C. P. (2010). Low genetic diversity in several

gopher tortoise (Gopherus polyphemus) populations in the Desoto National Forest,

Mississippi. Herpetologica 66:31–38.

Frankham, R. 2003. Genetics and conservation biology. Comptes Rendus Biologies

326:22–29.

Hauswaldt, J. S., and Glenn, T. C. 2003. Microsatellite DNA loci from the Diamondback

terrapin (Malaclemys terrapin). Molecular Ecology Notes 3:174–176.

Iverson, J. B., Barthelmess, E. L., Smith, G. R., and deRiverea, C. E. 1991. Growth and

reproduction in the mud turtle (Kinosternon hirtipes) in Chihuahua, México. Journal

of Herpetology 25:64–72.

King, T. L., and Julian, S. E. 2004. Conservation of microsatellite DNA flanking

sequence across 13 Emydid genera assayed with novel bog turtle (Glyptemys

muhlenbergii) loci. Conservation Genetics 5:719–725.

Lance, S. L., Love, C. N., Nunziata, S. O., O’Bryhim, J. R., Scott, D. E., Flynn, R. W.,

and Jones, K. L. 2013. 32 species validation of a new Illumina paired-end approach

for the development of microsatellites. PLoS ONE 8:e81853.

Libants, S., Kamarainen, A. M., Scribner, K. T., and Congdon, J. D. 2004. Isolation and

cross-species amplification of seven microsatellite loci from Emydoidea blandingii.

Molecular Ecology Notes 4:300–302.

Lindeman, P.V. 2008. Sternotherus carinatus (Gray 1856) – Razorback musk turtle,

razor-backed musk turtle. In: Rhodin, A.G.J., Pritchard, P.C.H., van Dijk, P.P.,

Saumure, R.A., Buhlmann, K.A., and Iverson, J.B. (Eds.). Conservation Biology of

Freshwater Turtles and Tortoises: A Compilation Project of the IUCN/SSC Tortoise

and Freshwater Turtle Specialist Group. Chelonian Research Monographs 5:012.1-

012.6.

López-Luna, M. A., Cupul-Magaña, F. G., Escobedo-Galván, A. H., González-

Hernández, A. J., Centenero-Alcalá, E., Rangel-Mendoza, J. A., Ramírez-Ramírez

M. M., Cazares-Hernández, E. 2018. A distinctive new species of mud turtle from

Western México. Chelonian Conservation Biology 17:2–13.

Lovich, J. E., and Ennen, J. R. 2013. A quantitative analysis of the state of knowledge of

turtles of the United States and Canada. Amphibia-Reptilia 34:11–23.

Lovich, J. E., Ennen, J. R., Agha, M., and Gibbons, J. W. 2018. Where Have All the

Turtles Gone, and Why Does It Matter? BioScience, 68: 771–781.

Osentoski, M. F., Mockford, S., Wright, J. M., Snyder, M., Herman, T. B., and Hughes,

C. R. 2002. Isolation and characterization of microsatellite loci from the Blanding’s

turtle, Emydoidea blandingii. Molecular Ecology Notes 2:147–149.

Peakall, R., and Smouse, P. E. 2006. GENALEX 6: Genetic analysis in Excel. Population

genetic software for teaching and research. Molecular Ecology Notes. 6:288–295.

Raymond, M., and Rousset, F. 1995. GENEPOP (Version 1.2): Population genetics

software for exact tests and ecumenicism. Journal of Heredity 86:248–249.

Rice, W. R. 1989. Analyzing Tables of Statistical Tests. Evolution. 43:223–225.

Rhodin, A., Pritchard, P., van Dijk, P. P., and Saumure, R. (Eds.). 2008. Conservation

biology of freshwater turtles and tortoises: a compilation project of the IUCN/SSC

Tortoise and Freshwater Turtle Specialist Group. Chelonian Research Monographs

No. 5.

Turtle Conservation Coalition [Rhodin, A., Walde, A.D., Horne, B.D., Van Dijk, P.P.,

Blanck, T., and Hudson, R. (Eds.)]. 2011. Turtles in trouble: the world's 25+ most

endangered tortoises and freshwater turtles-2011. Lunenburg, MA: IUCN/SSC

Tortoise and Freshwater Turtle Specialist Group, Turtle Conservation Fund, Turtle

Survival Alliance, Turtle Conservancy, Chelonian Research Foundation,

Conservation International, Wildlife Conservation Society, and San Diego Zoo

Global, 54 pp.

Rozen, S., and Skaletsky, H. 1999. Primer3 on the www for general users and for

biologist programmers. In S. Misener and S. A. Krawetz, Bioinformatics Methods

and Protocols, 365–386. Humana Press. Clifton, New Jersey.

Rousset, F. 2008. Genepop’007: a complete re-implementation of the Genepop software

for Windows and Linux. Molecular Ecology Resources 8:103–106.

Scott, P. A., Glenn, T. C., and Rissler, L. J. 2018. Resolving taxonomic turbulence and

uncovering cryptic diversity in the musk turtles (Sternotherus) using robust

demographic modeling. Molecular Phylogenetics and Evolution 120:1–15.

Selman, W., Ennen, J., Kreiser, B. R., and Qualls, C. P. 2009. Cross-species

amplification of emydid microsatellite loci in three Graptemys species and their

utility as a conservation tool. Herpetological Review 40:32–37.

Shaffer, H. B., Gidiş, M., McCartney-Melstad, E., Neal, K. M., Oyamaguchi, H. M.,

Tellez, M., and Toffelmier, E. M. 2015. Conservation genetics and genomics of

amphibians and reptiles. Annual Review of Animal Biosciences 3:113–138.

Schwartz, T. S., Osentoski, M., Lamb, T., and Karl, S. A. 2003. Microsatellite loci for the

North American tortoises (genus Gopherus) and their applicability to other turtle

species. Molecular Ecology Notes 3:283–286.

Selkoe, K. A., and Toonen, R. J. 2006. Microsatellites for ecologists: a practical guide to

using and evaluating microsatellite markers. Ecology Letters 9:615–629.

Turtle Taxonomy Working Group [Rhodin, A.G.J., Iverson, J.B., Bour, R. Fritz, U.,

Georges, A., Shaffer, H.B., and van Dijk, P.P.]. 2017. Turtles of the world:

annotated checklist and atlas of taxonomy, synonymy, distribution, and conservation

status (8th Ed.). In: Rhodin, A.G.J., Iverson, J.B., van Dijk, P.P., Saumure, R.A.,

Buhlmann, K.A., Pritchard, P.C.H., and Mittermeier, R.A. (Eds.). Conservation biology of freshwater turtles and tortoises: A compilation project of the IUCN/SSC

Tortoise and Freshwater Turtle Specialist Group. Chelonian Research Monographs

7:1–292.

30 Table 2.1 Characteristics of 40 microsatellite loci isolated for the razorback musk turtle (Sternotherus carinatus).

Locus Primer Sequence 5'-3' Motif N Size (bp) NA HO HE T°C Scar02 CACGAGGCATGGTGTGTAGG ATCT ------CAGCTTGGAAACCACTGGC Scar06 TCAAGTGATGCCACTGAAACG AAAG 26 232-276 10 0.731 0.848 56 CCCTTTATGTGCAGCCAGC Scar07 CTGTTAGGTTGCTCACCGGC TCTG ------GCATTTCTCCACGATCGGC Scar08 AACACTCCAATCAGCATCAAGC AAAG 29 324-372 11 0.793 0.869 56 AACAGCGTTGGAAACACACC Scar10 TCAACAGTCGAACTCCAGCG AAAG 26 262-306 11 0.615 0.841 56 TCAGCTTGTCCATTATTTCCTTACC Scar11 GCATAATGATAGTTTCCAAGCACC AAAG 29 468-512 10 0.862 0.847 56 CAAAGTTTGTGCTCCTGTATTCC Scar13 GGGAGGTGCTGGATACTTGG ATCT 24 352-380 7 0.625 0.714 56 CCTTTGGTTAGAATAATTTACTCCGC Scar14 GGAGCCCTGGTCTTTAGTAATGG ATCT 28 196-284 15 0.857 0.897 56 TCTTTCAACTGCTCTGACACCC Scar15 GACGTGCTTTGGTCACCG AAAG 28 252-296 10 0.786 0.859 56 AACACTCCAATCAGCATCAAGC Scar16 CTCTGGGATAAATGTCCGGG AAAG ------GTCCAGGTATGATCTCGGTCG Scar17 ATGAGCGGCAATGAAAGTCC AAAG 27 286-314 8 0.889 0.823 56 ATCCACCGCAAACTCCTCG Scar19 TTTGAGATGGTTTCCTCAAAGC TTCC 28 148-180 6 0.571 0.656 56 CATTAATCAACCCAGTGCCG Scar20 GAGTGGGAATACTTGCATTTGG AAAG ------TGGAGGTATACCCATTTCCAGC Scar23 TAAATGTACAGCAGGCCTTAGATGG AAAG ------Table 2.1 (continued) CTATCAATGCCACTGAACCAGG Scar24 CAATGAACAGCTGCATTGGG AAAG 26 342-382 10 0.654 0.859 56 TCCCATACCGCTGAGAGAGG Scar26 AAGTGTCATAGCTTCAGAAACTCCC ATCT 28 268-316 10 0.786 0.831 56 CCCATCAGCGAGAATAGGC Scar27* GGTTAAAGGTTCAATCGGTCC AAAG 30 262-290 7 0.833 0.796 54 TTTATTGAAGGCGATGAGACG Scar28 TTTGGATTTGGGAAAGTCTGG AAAG ------CGTATTTGATTCCTTTAGCCAGC Scar29 TCAGTAGATTATGAACTCTGTAATGATACC ATCT 30 236-256 6 0.833 0.761 56 CAATTCCTGCACTTCGCC Scar31 CCAAATGGATTTCACTACAATGG ATCT 29 188-248 8 0.414 0.414 56 CAACAGTCTGCAATCGGAGC Scar32 CAGGATGTGTCCCAGTCAGC ATCT ------CGGATAGAGCAATTGATATTAGAATAGG Scar33 CATCTGTTGGGATGAATGGC ATCT 18 224-244 4 0.389 0.446 56 GCATCTTCAGTCCAATGAGCC Scar35 AGGGCATGCTGGTAGACTGG AAAG ------TCCCTTAAGGATCAGCTTTCTCC Scar36 AGGGAGCCTAACTCCCGAGG AAAG 28 144-164 6 0.643 0.594 56 ATAGGCTTTGCAATGTGCCC Scar37 TTCATGGAGCAGAGACCACC ATCT ------ATTTCCACAGAGTGCCAAGC Scar38* TTCAGGTCTTACCCACCTTGG AAAG 27 244-300 14 0.963 0.910 56 TCAACCATGTGTAATGGGAGC Scar39 AACATTTCATTCTCTCACACTTTCG AATG 29 196-252 8 0.759 0.718 56 CATCCTCTAGGAGCATCGGG

Table 2.1 (continued) Scar40 GCAAGGATCTAATCAGTTCTACAAGC AAAG 29 196-248 7 0.655 0.731 56 CCACTGATGGAGGACAGATCC Scar42 ACACTCAAACAGGGCACAGC AGAGC 23 198-233 7 0.696 0.619 56 TTCCCTTTCCCTTCCTTTCG Scar43 TTCTTTGAATGATCCTCTAACTTTGC ATATT 27 255-280 5 0.630 0.682 56 CCAGTGGGTACAAATGCAGG Scar46 AAAGTAGTTGGGTTGCGAGG ATCT ------GTGACGGAGCAAACATCTCC Scar47 GGATCCACCTTAAATGAACAACG AAAG ------TCTGCTCAATGGACAAATTGC Scar48 CAAACTTTGCCTCACATTGTCC AAAT 29 252-268 4 0.621 0.554 56 TTTCCTGGAGGAGTGTGTCC Scar49 GACACAGTCCGAGACCATGC AAAG ------TGGATAATTGTTTCCACCAACG Scar50 GTCATTCCCTGAGCACTGTCC AAAT 29 228 and 236 2 0.138 0.128 56 CATGCCGGTGAGGAATGG Scar51 TTGCAGTTTGTGGAATTGGG AAAT 30 220 and 228 2 0.267 0.231 56 AACTCTTCATTTGAGTACTCCTGCC Scar52 GAGATCCTGTATCTTCTCAAGCAGC AAAG 28 264-276 3 0.429 0.536 56 GGGTAAGCCTGTTCATGCC Scar54 TTTCTAGTCGTGGCATATCTCC AAAT ------ATGTGCCATATCCTGTCTGC Scar57 TCAAACTACCAGGTTCCCAGG ATAC ------AAATCCTGTTCTGTGAAATGAAGG Scar59 TTGACCAGTACACCCAAGTCG ATAC ------ATCTAGCTCAAACGGAGCGG Forward primers are reported without the M13 tail. Columns include repeat motif, number of individuals out of 30 (N) for which loci amplified, size range of alleles (Size bp), the number of alleles (NA), the observed and expected heterozygosities (HO and HE), and the annealing temperature (T°C) for razorback musk turtles. Those loci marked with an asterisk (*) were optimized with increased Taq (0.1 units) and forward/reverse primers (0.5 µL). Dashes represent loci that did not amplify in the razorback musk turtle but amplified in other kinosternid taxa at an annealing temperature of 56°C (Table 2.2).

Table 2.2 All microsatellite loci developed for Sternotherus carinatus that cross- amplified in at least one kinosternid species.

Locus S. odoratus S. peltifer S. depressus K. subrubrum K. bauri (4) (5) (5) (5) (5) Scar02 5 - - 7 - Scar06 5 5 - - 6 Scar07 - 2 - - - Scar08 7 6 - - 7 Scar10 - - - 3 3 Scar11 5 4 - 4 6 Scar13 5 - 3 - 2 Scar14 5 3 - - - Scar15 6 3 7 - - Scar16 5 6 - - - Scar17 - 6 - - - Scar19 5 - - - - Scar20 4 4 - - - Scar23 4 - - - 2 Scar26 - 2 6 - - Scar27 6 4 4 - - Scar28 - - 6 - - Scar29 6 - - 6 - Scar31 5 - - - - Scar32 5 5 - 3 - Scar35 5 5 - - - Scar37 3 - 3 - - Scar38 - - 8 - - Scar39 3 - - - 6 Scar40 7 - - - 6 Scar46 - - 5 - - Scar47 4 - 2 - 4 Scar48 3 - 2 - 5 Scar49 5 - 5 3 7 Scar50 - - - - 2 Scar52 2 2 - - 3 Scar54 3 3 2 - 2 Scar57 2 3 3 - - Scar59 3 - - - - Columns represent species, parentheses indicate number of each species tested, and numbers in column represent number of alleles amplified at a locus.

CHAPTER III - RANGE-WIDE POPULATION GENETICS OF THE RAZORBACK

MUSK TURTLE (STERNOTHERUS CARINATUS)

Abstract

The razorback musk turtle (Sternotherus carinatus) is a common inhabitant of river systems from the Brazos River drainage in Central Texas to the fringes of the

Pascagoula Drainage in southwestern Alabama. This small, highly aquatic turtle species is generally considered one of the least studied turtle species in the United States and

Canada, despite that it has been reported as an indicator species within the Gulf Coastal

Plain. While many other species of riverine turtles (e.g., map turtles, sawbacks and alligator snapping turtles) have benefitted from ecological and genetic studies across the

Gulf Coastal Plain, the razorback musk turtle has received little research attention. I used

10 microsatellite loci to evaluate the levels of population structure across the range of the razorback musk turtle. The species exhibits extensive levels of population structure, with the most marked differentiation between populations from the Ouachita Highlands in

Arkansas and southeastern Oklahoma and those in the remainder of the range. Given the popular nature of this species in overseas markets, and the alarming number of razorback musk turtles that have been exported in recent years, I suggest considering each of the

Ouachita populations as evolutionarily significant units for the species, and that these populations should be managed accordingly.

3.1 Introduction

Turtles and tortoises are plagued by a number of threats to their survival including habitat degradation, fragmentation, and alteration (Klemens 2000; Moll and Moll 2004).

This is why turtles are unfortunately considered the most endangered vertebrate taxa on

the planet with nearly 60% of all species classified as threatened or endangered by the

International Union for Conservation of Nature (Lovich et al. 2018). Though the

Southeastern United States represents a global diversity hotspot for chelonians, many species have been negatively affected by widespread anthropogenic alterations to the landscape or waterscape. Road mortality has caused precipitous declines in terrestrial and freshwater turtle populations (Gibbs and Shriver 2002; Cureton and Deaton 2012), whereas river channelization, impoundments, and pollution have imperiled freshwater species (Buhlmann and Gibbons 1997; Lindeman 1999; Bodie 2001; Shelby and

Mendonça 2001; Killebrew et al. 2002; Melancon et al. 2013).

In order to preserve and maintain the integrity of the biodiversity of southeastern turtle species, it is important to have a clear understanding of the ecology and evolutionary history of turtle species in the region. Effective management of declining, threatened, or endangered species requires an understanding of not only population health

(i.e., demographics), but population genetic health. Population genetics is important for identifying and defining appropriate management units (MU’s) and evolutionarily significant units (Ryder 1986; Moritz 1994; Vogler and DeSalle 1994). Population genetic and phylogeographic studies are essential to identify genetically disparate populations, which in some cases may represent cryptic diversity warranting taxonomic recognition. This phenomenon has been observed in several map turtle species (genus

Graptemys; Lovich and McCoy 1992; Ennen et al. 2010; Lindeman 2013) and the alligator snapping turtle (Thomas et al. 2014). These species have received a great amount of research attention, but a recent review of the literature highlights that other common species within the southeastern United States have received a paucity of

research attention, particularly smaller species of turtle (Lovich and Ennen 2010). For example, despite their abundances in streams, rivers and springs (Folkerts 1968; Cox and

Marion 1979; Lindeman 2008), Sternotherus are among the least studied turtles in North

America (Lovich and Ennen 2010). The recent description of the intermediate musk turtle

(Sternotherus intermedius; Scott et al. 2018) serves as an example of the impact of new studies on our understanding of the genus.. This species was formerly believed to be a hybrid from of the stripe-necked musk turtle (Sternotherus peltifer) and the loggerhead musk turtle (S. minor). However, Scott et al. (2018) determined that Sternotherus intermedius represented its own evolutionarily distinct lineage mostly restricted to the

Escambia River Drainage (a small Gulf Coast drainage), a pattern exemplified in other

Southeastern turtle species (viz. Graptemys – see Lindeman 2013).

The razorback musk turtle (Sternotherus carinatus) is another example of a wide- ranging, lotic turtle species that has received little research attention (ranking in the top- ten least studied in the United States and Canada; Lovich and Ennen 2010). The razorback musk turtle is a small- to medium-sized freshwater turtle found lotic habitats

(i.e., streams, rivers and associated oxbows) from the midwestern and southeastern

United States (Lindeman 2008). A recent study on the biogeography of North American turtle assemblages found that Sternotherus carinatus is one of two turtle species that is an indicator species for the Gulf Coastal Plain, the other species being the false map turtle

(Graptemys pseudogeographica). Sternotherus carinatus ranges from the Brazos River drainage in Central Texas, northward into the Red and Ouachita River drainages of southeastern Oklahoma and Arkansas, and terminates in the Escatawpa River of the

Pascagoula River drainage in extreme southwest Alabama (Ernst and Lovich 2009;

Lindeman 2008).

While the razorback musk turtle is currently considered of “Least Concern” by the

IUCN (IUCN 2016), the species has become increasing popular in the pet trade in

Southeast Asia, a region notorious for their demand for turtles in the food and pet trade

(Cheung and Dudgeon 2006, Rhodin et al. 2011). Reports of extremely large numbers of turtles (>100,000 Sternotherus spp.) have been reported as exported from some

Southeastern States (B. Glorioso [Louisiana] and J. Jenson [Georgia] pers. com.), forcing a moratorium on S. carinatus harvest in Louisiana (2016) and stringent turtle harvest laws in Georgia (2018). Given the life histories of most turtle species (i.e., delayed sexual maturity, low reproductive output; Congdon et al. 1994), collection of wild turtles is not sustainable, even at low levels of exploitation (Reed et al. 2002; Moll and Moll 2004).

Currently, there are no published studies on the population genetics of the razorback musk turtle. Recent studies on other southeastern turtle taxa have shown enough genetic divergence across Gulf coast drainages to warrant taxonomic recognition

(e.g. nearly ten species in the genus Graptemys, and the recent splitting of Macrochelys;

Lindeman 2013; Echelle et al., 2010, Thomas et al. 2014). Sternotherus carinatus shares a similar ecology to Macrochelys spp. and Graptemys spp. because all are highly aquatic and are infrequently encountered making overland movements (Tinkle 1958).

Furthermore, females nest close to the stream, though very few nests have ever been documented in the wild (Tinkle 1958; Ernst and Lovich 2009; Narum et al. in press).

Given the highly aquatic ecology of the species, gene flow between river drainages should be limited leading to population genetic structure.

The goal of this study was to characterize genetic structure across the range

Sternotherus carinatus from most of the major river systems across their geographic range using microsatellite loci. The primary objective was to identify range-wide geographic patterns of genetic structure and relate that to the biogeographic history of the species. I also wanted to compare levels of genetic diversity and assess the demographic history of populations across the range with the goal to inform conservation and management decisions. In this chapter, I investigate 1) the amount of genetic differentiation across populations of razorback musk turtles; and 2) and the landscape features responsible for the biogeography of the species.

3.2 Methods

3.2.1 Field Collections

I sampled sites from across the geographic range of the razorback musk turtle while other samples were acquired from colleagues (Table 3.1; Figure 3.1). These sites were generally small creeks or rivers near historic localities that could be easily waded or accessed by canoe. A combination of hoop traps, collapsible Promar© crab traps and minnow traps, and traditional crab traps were deployed at points of public access (e.g., bridge crossings and boat ramps) or on private property with permission. I set traps in shallow water near suitable locations (e.g. woody debris, log jams or rock ledges) and baited with sardines in soybean oil (Bumble Bee Foods ©) or fresh fish (e.g., nongame by-catch). Despite previous reports in the literature (Cagle and Chaney 1950; Tinkle

1958; and Trauth et al. 2004), Sternotherus carinatus was not bait averse. Traps were checked every 12-24 hours. During summer months, turtles were also opportunistically collected by hand while active in the early morning (0700-1000) and evening (2030-2230

by headlamp). I attempted to sample each site until I captured a representative number of turtles (20+ turtles), which generally took one to four days of sampling. However, some sites produced fewer turtles due to low abundance like the Brazos River drainage at the western extent of the species’ range or where inclement weather and river conditions precluded extensive sampling. I collected a small tissue sample from the interdigital webbing of the hind foot of each individual to minimize damage, stress, and injury to the turtle. Turtles were uniquely marked by filing the marginal scutes of the shell following a modification of Ernst’s (1974) filing scheme, so as not to be sampled twice, but also to retain the ability to individually identify animals for future studies. All turtles were ultimately released at the point of capture.

3.2.2 Molecular Methods

Total genomic DNA was extracted from the tissue samples with a DNeasy Tissue

Kit (QIAGEN Inc., Valencia, CA). The polymerase chain reaction (PCR) was used to amplify ten microsatellite loci designed for S. carinatus (Scar06, Scar08, Scar10, Scar19,

Scar26, Scar27, Scar29, Scar35, Scar38, Scar39, Scar40) using the reaction conditions and thermal cycler profiles outlined in Brown and Kreiser (in press). Microsatellite alleles were subsequently visualized using a LI-COR 4300 DNA sequencer. I tested for

Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium using Genepop on the

Web (Raymond and Rousset 1995; Rousset 2008). A Bonferroni correction was applied to the alpha value to account for multiple comparisons (Rice 1989).

I also calculated genetic diversity measures for each site including the number of alleles, allelic richness, and heterozygosity (observed and expected) using FSTAT v.

2.9.3 (Goudet 1995). Pairwise FST values were also calculated with FSTAT v. 2.9.3 to

determine levels of genetic differentiation between populations in each river drainage. I used ANOVA’s, or t-tests to test for any differences in allelic richness and levels of heterozygosity across populations. In addition to testing across all drainages, I also grouped turtle populations into one of three regions (west of the Mississippi River, east of the Mississippi River, or Highlands) and tested for differences in allelic richness among them. Before each analysis, I checked the genetic data for the assumptions of normality using a Shapiro-Wilks test and equal variances using a Bartlett’s test. When assumptions of normality or equal variances were violated, I used a nonparametric approach to compare groups by use of a Kruskal-Wallis test. I checked for significant differences among groups using a Tukey HSD post hoc test (parametric) or a Pairwise Mann-

Whitney U-test (nonparametric). All analyses were performed in the “stats” package built into R v. 3.6.1 with an alpha value of 0.05. The Jourdan and the Comite River sites were omitted from these statistical tests due to low sample sizes (n < 10 turtles). The Brazos also represented a small population, but I was still interested in its levels of allelic richness and heterozygosities, so I calculated allelic richness across all sites with greater than 10 turtles twice: once with the Brazos included, once with the Brazos excluded.

where K is the assumed number of populations – in this case 1/ [#of river drainages]).

For each value of K, I ran 20 replicates with 100,000 Markov chain Monte Carlo

(MCMC) iterations and a burn-in set at 50,000. I selected the best value of K by examining the likelihood values for each K and by using the ΔK method (Evanno et al.

2005) as calculated by the online program StructureSelector (Li and Liu 2018).

StructureSelector averaged the 20 runs for the best value of K using Clumpak (Kopelman et al. 2015), and I then visualized the average ancestry coefficient (q) for each individual with DISTRUCT v. 1.1 (Rosenberg 2004). I selected the best value of K as the smallest value that seemed to represent biologically meaningful genetic groups. Hereafter, I also use the term “region” to refer to the areas delineated by the sampling sites comprising these genetic groups. Since intraregional population structure was still possible, I performed another set of analyses on individuals within each of the initially recognized genetic groups (e.g., hierarchical analysis; Vähä et al. 2007). The same parameters were used for the STRUCTURE analyses with 20 iterations of each value of K from one to the number of drainages sampled plus two.

To corroborate the STRUCTURE results, I used a discriminant analysis of principal components (DAPC) in the R-package adegenet v2.1.1 to determine the number of genetic clusters (Jombart et al., 2010). This is a multivariate, model-free method also used to visualize genetic differentiation without using population information. A DAPC first transforms the genetic data using a PCA, and then clusters are identified using a discriminant analysis by maximizing among group variance and minimizing within group variability. By plotting the DAPC, it also allows the user to visualize genetic distance in ordination space. I used the find.clusters functions to determine the optimal K (from 1-

40) with the lowest Bayesian Inference Criterion (BIC). I used DAPC cross-validation

(30 replicates, 90% training dataset) to evaluate the optimal number of principal components and discriminant functions to retain, and I then plotted individuals against the top five DAPC axes.

3.3 Results

I genotyped a total of 227 Sternotherus carinatus from 12 river systems, and the number of turtles per drainage ranged from 7-30 individuals (Figure 3.1; Table 3.1). Of the ten microsatellite loci, none exhibited linkage disequilibrium nor differed significantly from Hardy-Weinberg Equilibrium after a Bonferroni correction. Observed and expected heterozygosity (HO/HE) ranged from 0.37-0.82 and 0.44-0.85 respectively

(Table 3.2). The Ouachita turtles had significantly lower observed heterozygosity values than all other drainages, including the Red River drainage (Table 3.4). The Red River drainage turtles had significantly lower expected heterozygosities than all river drainages west of the Mississippi River, except the Brazos which also had reduced heterozygosity, but it was not significantly so (Table 3.5).

Because allelic richness is calculated and rarefied to the smallest number of turtles in a population, the Comite and Jourdan turtles were excluded from this analysis. The

Brazos also represented a small population, but due to its peripheral nature, I was still interested in its allelic richness. When the Brazos was excluded from the dataset, the

Ouachita drainage had significantly lower allelic richness than all other sites (p < 0.05;

Table 3.3). When the Brazos was included in the dataset, the Ouachita had significantly lower allelic richness than all other populations except the Brazos and Pearl (p < 0.05), but this is likely because the dataset was rarefied to a smaller number of turtles and lost

some statistical power (Table 3.6). When the Brazos was included in the dataset, the

Brazos population had significantly lower allelic richness than all other Gulf Coastal

Plains drainages west of the Mississippi except the Sabine (e.g., significantly lower than the Trinity, Neches and Calcasieu; p < 0.05). Allelic richness was significantly different across the three regions (F2,87 = 13.5, p < 0.001). A Tukey HSD post hoc test determined that populations west of the Mississippi (when the Brazos was excluded) had significantly higher allelic richness than populations east of the Mississippi River (p =

0.05) and Highlands populations (p < 0.001), and that populations east of the Mississippi also had significantly higher allelic richness than the Highlands region (p = 0.01).

The ∆K method (Evanno et al. 2005) of the STRUCTURE results found the deepest level of population subdivision at a K of 2 corresponding to a Ouachita Highlands group

(Glover [Upper Red] and Saline [upper Ouachita] Rivers) and a Gulf Coastal Plains

(GCP) group with an average FST of 0.14 (Figure 3.2). The average individual ancestry coefficient (q) score of these two groups were 0.97 (s.d. = 0.05) for the Ouachita

Highlands group and 0.98 (s.d. 0.02) for the Gulf Coastal Plains group. The average FST values across all Sternotherus carinatus populations was 0.14, but the average FST between the Ouachita Highlands sites and the Gulf Coastal Plains sites was 0.25, or very high differentiation (Wright 1978; Hartl and Clark 1997).

After a range wide K of 2, I investigated the remaining population structure within those groups hierarchically. Within the Ouachita Highlands group, STRUCTURE results found a split of two between the Red and Ouachita River populations. Average q-scores were very high: for the Red River drainage turtles was 0.996 (s.d. = 0.003), and average q-scores of 0.991 (s.d. = 0.009) for the Ouachita River drainage turtles. The Red and

Ouachita River turtle populations were as differentiated from one another as they were from the rest of the Gulf Coastal Plain (FST = 0.30).

Within the Gulf Coastal Plain, the ∆K method of the STRUCTURE suggested that K

= 4 best explained the genetic structure in the data. These four groups were represented by the Brazos-Trinity (average q-score = 0.86), Neches-Sabine (average q-score = 0.87), a Pearl-Jourdan (average q-score = 0.86) and Pascagoula-Escatawpa (average q-score =

0.91). High levels of admixture were found in the Comite and Calcasieu (average q-score

= 0.53; Figure 3.2). The average FST across the Gulf Coastal Plains turtle populations was 0.09. The population of turtles from the Brazos River drainage was the most differentiated in the Gulf Coastal Plain with an average FST of 0.14, compared to an average of 0.08 among the remaining GCP populations (Table 3.7).

The DAPC corroborated the STRUCTURE results as it also yielded 6 groups: two highly differentiated Ouachita Highlands groups and four Gulf Coastal Plains groups

(Figure 3.3). The Ouachita was the most disparate genetic group identified in the DAPC, and it was highly differentiated along axis 1 (35.3% variance explained), whereas the Red

River turtles are separated along axis 2 (25.6% variance explained). The Gulf Coastal

Plains populations were more or less aligned from east to west along the second axis and like the STRUCTURE results, they correspond to a Brazos-Trinity group, Sabine-Neches group, Pearl-Jourdan group and a Pascagoula-Escatawpa group. There was overlap of the

Sabine-Neches and Pearl-Jourdan turtles with both the Brazos-Trinity and Pascagoula-

Escatawpa turtles, but the westernmost turtles of the Brazos-Trinity group did not overlap with the easternmost group of turtles in the Pascagoula-Escatawpa group.

3.4 Discussion

This is the first population genetics study of a wide-ranging Kinosternid species.

Because highly aquatic species are confined to their river drainages, it is not surprising that Sternotherus carinatus, like other highly aquatic, lotic species shows high levels of genetic differentiation across its range (Hackler et al. 2007; Scott et al. 2018). Both the

STRUCTURE and DPAC analyses suggest there are potentially 6 distinct genetic groups across the species’ geographic range. The Ouachita Highlands populations were as genetically differentiated from one another as they were from other Gulf Coastal Plains

(GCP) populations. The Gulf Coastal Plains populations had lower overall genetic differentiation from one another, with the highest levels of genetic differentiation between the easternmost (Pascagoula-Escatawpa) and westernmost (Brazos-Trinity) populations. For Sternotherus carinatus, it appears that it is predominantly the complex geology rather than hydrology across its geographic range that drives the genetic differentiation across populations. The Ouachita Highlands populations show similar genetic structuring to other species of fishes, amphibians and crayfish that also have distributions from the highlands to the coastal plain (Mayden 1985)

The Ouachita Mountains differ from the neighboring coastal plain/Mississippi

Delta in many regards (e.g., age, geology, topography; Soltis et al. 2006), and these acute differences in habitats (e.g., high gradient streams vs. low gradient meandering rivers) represent a unique geologic history and may present fundamentally different selection pressures. The Ouachita Mountain physiogeographic province extends from southwestern

Oklahoma to southeastern Arkansas and is drained predominantly by Red River tributaries in the west (e.g., Kiamichi River, Little River) and by the Ouachita and Saline

Rivers in the east (Thornbury 1965; Mayden 1985). The isolated nature Ouachita

Highland’s orogeny makes them a well-known biogeographic hotspot, with at least eight species of endemic fishes, five species of endemic crayfish, and four species of endemic salamanders (Mayden 1985; Hatcher et al. 1986). The curious distribution of many of these endemic fish species is thought to reflect historical paleodrainages. Mayden (1985) posits that the Blue, Little, Kiamichi and Ouachita once shared a common drainage, not shared by the Red River. Instead, there is evidence for a Plains River that drained parts of the upper Red and flowed into the Trinity River drainage directly into the Gulf of Mexico

(Metcalf 1966; Mayden 1985). Like the other Ouachita Highlands taxa, the razorback musk turtle also reflects this curious distribution from the Saline to the Blue Rivers.

Though both Ouachita Highlands populations were well differentiated from the Gulf

Coastal Plains population, the Red River turtles are less differentiated than the Ouachita

River turtles (see DAPC [Figure 3.3]). In fact, the Red River turtles show their lowest FST

(genetic differentiation) values with the turtles from the Trinity River drainage, though this should be interpreted cautiously, as the two populations still demonstrate very high genetic differentiation (FST = 0.153; Table 3.7). These lower levels of differentiation may reflect the historic connection of the Red River and the Upper Trinity and explain the slightly closer genetic proximity of the Red River turtles to those in the Gulf Coastal

Plain.

Sternotherus carinatus from the Ouachita Highlands were also morphologically dissimilar from Gulf Coastal Plains populations. Barkhouser (unpubl. thesis 1996) used a canonical variate analysis (CVA) on the morphology of 708 (472 adults, 236 juveniles) razorback musk turtles. Barkhouser found that there is broad overlap in the morphology

of the S. carinatus from the Gulf Coastal Plain, but those from the Ouachita Highlands separated from other population in ordination space by anal scute shape (Saline-Ouachita turtles) and relative head width (Kiamichi-Little turtles). The genetic data presented herein corroborate these morphological data. The DPAC plot of genetic distances (Fig.

3.3) closely mirrors the plots produced by the canonical variate analysis with the Saline

(Ouachita) turtles being most distinct from all others while the Red (upper Red, i.e.,

Glover) turtles as distinct, but more proximate genetically and morphologically to Gulf

Coastal Plains populations, which are also shallowly divided East from West both in the

DPAC and CVA. However, from personal observations within the Saline River population, relative head width appeared greater than elsewhere across the geographic range, but this was not tested statistically.

The S. carinatus from the Gulf Coastal Plains populations exhibited lower overall

FST values. Unlike the Ouachita Highlands rivers, the rivers of the GCP have been subject to eustatic sea level rise and fall since the Wisconsinan Glacial Stage (2.6 – 0.1 mya), and many of these rivers have shared deltas in recent geologic past (Morton and Price 1987;

Barkhouser 1996 unpub. Thesis; Near et al. 2003); further, the topography of the Gulf

Coastal Plain has been more dynamic than the Ouachita highlands, allowing for interdigitating headwater streams or shared bays to potentially provide opportunity for gene flow (Lohse 1955; Benke and Cushing 2005). Sea level rise and fall has played a key role in the dispersal and colonization of species across the GCP (Bermingham and

Avise 1986; Near et al. 2003). Near et al. (2003) found that there was a rapid diversification of black basses (genus Micropterus) during the Late Miocene to Pliocene due to rapidly fluctuating sea levels. Iverson (1977) also speculated that dispersal of

ancestral Sternotherus minor stock was due to sea level fluctuations during this time as well. Seismic reflection data of the late Wisconsinan depicts that many of the Texas rivers of the Gulf Coastal Plain (the Brazos, Trinity, Sabine and Calcasieu) may have shared paleo-fluvial channels or historic deltas, providing an opportunity for dispersal for this species (Winker 1979; Suter and Berryhill 1985; Morton and Price 1985). Eustatic sea level rise and fall in the Gulf Coastal Plain may have inundated entire river systems, including the rivers and streams of the Mississippi embayment, only to be recently (in geologic time) recolonized by neighboring populations when sea levels again receded, a pattern similar to the “progression rule” for arthropod phylogeography in Hawai’i

(Roderick and Gillespie 1998), and a proposed mechanism for Graptemys dispersal across the GCP (Thomson et al. 2018).

Sternotherus carinatus populations from river drainages that have shared deltas

(e.g., Neches-Sabine, Pascagoula-Escatawpa) or deltas in close proximity (Brazos-

Trinity, Sabine-Calcasieu) showed higher levels of shared ancestry in the STRUCTURE analysis. Younger, smaller river drainages, like the Jourdan River, show mixed ancestry between its most proximate drainages the Pearl (avg. q-score = 0.58) and Pascagoula

(avg. q-score = 0.32). Given that the most recent glacial maximum was approximately

20,000 years ago and that the corresponding sea levels were projected to be ~140 m lower than present day (Martinson et al. 1987; Hearty and Kaufman 2000), it is likely many of these Gulf Coastal Drainages in this study shared deltas or had deltas in closer geographic proximity, which could explain the patterns I found in this study.

The marked genetic differentiation across populations should be considered in making management recommendations for the species. Evolutionarily significant units

are a designation used within conservation genetics for populations of species that should be prioritized, if a species needs protection, based off that populations unique genetic

(i.e., divergent and/or locally adaptive alleles) and/or morphological features (Ryder

1986; Dizon 1992; Vogler 1993; Moritz 1994). While my sampling within the Ouachita

Highlands was not extensive, I did pick up on strong patterns of differentiation compared to all other populations. Considering these findings, I think that the two Ouachita

Highlands populations may likely represent ESU’s based on genetic data corroborated with morphological data from Barkhouser (1996; unpublished thesis). While the razorback musk turtle is currently considered of least concern by the IUCN, there are alarming numbers of turtles being exported from across the range. Louisiana implemented a moratorium for the collection of this species after a surge in exports in the

2010’s on the magnitude of tens of thousands of razorback musk turtles per year (in meeting of Louisiana Wildlife and Fisheries Commission, 07 January 2016).

I recommend investigating the population genetics of other Ouachita Highlands populations of Sternotherus carinatus. I collected samples for this study from the Saline and Glover rivers, however, the Saline River flows independently into the Ouachita River below the fall line in the state near the Arkansas-Louisiana border. Though where I sampled the Saline River in Benton, Arkansas, is in close geographic proximity (straight- line distance) to other Ouachita River drainages of the Ouachita Highlands (e.g., the

Ouachita River proper, the Caddo River and Little River), genetic studies of aquatic organisms are measured not “as the crow flies,” but “as the fish (or turtle) swims”

(Schick and Lindley 2007). Likewise, the Glover River represents only one of many montane rivers that empties into the Red River, and there could be other evolutionarily

significant units within this region. Additional sampling of these montane rivers will increase understanding of the role of the orogeny and paleodrainages on the population genetics of this species. Additional sampling from the Red and Ouachita Rivers in

Central Louisiana would also be informative to the delineation of any populations that might be considered ESU’s.

In conclusion, Sternotherus carinatus, like many other highly aquatic turtle species in the Gulf Coastal Plain, exhibits high levels of population structure and differentiation, particularly in the Ouachita Highlands. Additional sampling and research is necessary to determine the exact number of recommended management units, but certainly those populations in the Ouachita Highlands presented here are morphologically and genetically dissimilar from those in the Gulf Coastal Plain, and they should be managed accordingly.

Acknowledgements

While I was fortunate enough to travel and see many of these river drainages, this study would not have been possible without generous genetic samples donated by Luke

Pearson and Gabbie Berry (Pearl, Jourdan and Pascagoula), Cody Godwin (Comite) and

Brad “Bones” Glorioso (Calcasieu). All turtles were caught and handled under appropriate state permits (Alabama: #9029; Arkansas: #011720172; Louisiana: LNHP-

17-021; Mississippi: #0607172; Oklahoma: #6866; Texas: SPR-0616-157;) and IACUC numbers: 11092206 and 17101202 from the University of Southern Mississippi. This work has been funded by the following grants (in alphabetical order): the American

Turtle Observatory, Birmingham Audubon Society, Chicago Herpetological Society,

NSF Graduate Research Fellowship (#1842492), and the Theodore Roosevelt Memorial

Grant through the American Museum of Natural History.

Table 3.1 Range wide sampling locations for Sternotherus carinatus.

Drainage Number of S. carinatus General Location

Brazos 10 College Station, TX

Trinity 26 Livingston, TX

Neches 28 Nacodoches, TX

Sabine 14 Longview, TX

Calcasieu 21 Dry Creek, LA

Red 25 Wright City, OK

Ouachita 25 Benton, AR

Comite 7 Baton Rouge, LA

Pearl 23 Jackson, MS

Jourdan 8 Bay St. Louis, MS

Pascagoula 24 Hattiesburg, MS

Escatawpa 16 Hurley, MS

Table 3.2 Summary statistics for Sternotherus carinatus across river drainages.

Population N NA AR HO HE

Brazos 10 6.4 3.6 0.69 0.70

Trinity 25 11.3 4.3 0.80 0.83

Neches 26 10.7 4.2 0.82 0.83

Sabine 13 10.1 4.6 0.82 0.85

Calcasieu 18 10.8 4.6 0.81 0.86

Red 23 5.5 2.9 0.61 0.60

Ouachita 24 4.2 2.2 0.37 0.44

Comite 6 5.6 4.0 0.73 0.76

Pearl 21 8.8 4.0 0.80 0.80

Jourdan 6 3.9 3.4 0.59 0.60

Pascagoula 22 8.9 4.0 0.75 0.79

Escatawpa 13 7.6 4.0 0.75 0.77

Table organized from west to east. N = average number of turtles genotyped; N = average number of alleles per locus; A = average A R allelic richness per locus; H = average observed heterozygosity; H = average expected heterozygosity. O E

54 Table 3.3 Significance testing of allelic richness across populations of Sternotherus carinatus.

Drainage Trinity Neches Sabine Calcasieu Red Ouachita Pearl Pascagoula Escatawpa

Trinity -

Neches 1.00 -

Sabine 1.00 0.800 -

Calcasieu 0.970 1.00 0.570 -

Red 0.880 0.410 1.00 0.220 -

Ouachita >0.001 >0.001 0.004 >0.001 0.030 -

Pearl 0.500 0.120 0.940 0.050 1.00 0.150 -

Pascagoula 1.00 0.890 1.00 0.710 1.00 0.002 0.880 -

Escatawpa 1.00 0.890 1.000 0.710 1.00 0.002 0.870 1.00 -

Significant values are in bold and underlined. Sites with fewer than 10 turtles were excluded from this analysis.

Table 3.4 Significance testing of levels of observed heterozygosity across the range of the Sternotherus carinatus.

Drainage Brazos Calcasieu Escatawpa Neches Ouachita Pascagoula Pearl Red Sabine

Calcasieu 0.130 ------

Escatawpa 0.363 0.472 ------

Neches 0.104 0.791 0.427 ------

Ouachita 0.011 0.001 0.002 <0.001 - - - - -

Pascagoula 0.344 0.257 0.910 0.241 0.001 - - - -

Pearl 0.185 0.850 0.472 1.000 0.001 0.256 - - -

Red 0.570 0.031 0.173 0.038 0.031 0.226 0.031 - -

Sabine 0.081 0.940 0.447 0.850 0.001 0.570 0.970 0.075 -

Trinity 0.161 0.940 0.405 0.650 0.001 0.226 0.405 0.034 0.910

Significant values are in bold and underlined. Table 3.5 Significance testing of levels of expected heterozygosity across the range of the Sternotherus carinatus.

Drainage Brazos Calcasieu Escatawpa Neches Ouachita Pascagoula Pearl Red Sabine

Calcasieu 0.005 ------

Escatawpa 0.121 0.052 ------

Neches 0.011 0.082 0.427 ------

Ouachita 0.006 <0.001 <0.001 <0.001 - - - - -

Pascagoula 0.104 0.045 0.571 0.307 <0.001 - - - -

Pearl 0.076 0.035 0.821 0.820 <0.001 0.623 - - -

Red 0.162 <0.001 0.009 <0.001 0.151 0.006 0.005 - -

Sabine 0.003 0.481 0.096 0.850 <0.001 0.059 0.579 <0.001 -

Trinity 0.031 0.762 0.280 0.520 <0.001 241 0.315 0.001 1.00

Significant values are in bold and underlined.

Table 3.6 Significance testing of allelic richness including the Brazos Drainage.

Drainage Brazos Trinity Neches Sabine Calcasieu Red Ouachita Pearl Pascagoula Escatawpa

Brazos -

Trinity 0.004 -

Neches 0.003 1.00 -

Sabine 0.260 1.00 0.860 -

Calcasieu 0.001 0.980 1.00 0.660 -

Red 0.630 0.930 0.500 1.00 0.280 -

Ouachita 0.940 <0.001 <0.001 0.010 <0.001 0.040 -

Pearl 0.940 0.590 0.160 0.970 0.070 1.00 0.200 -

Pascagoula 0.170 1.00 0.930 1.00 0.780 1.00 0.004 0.920 -

Escatawpa 0.170 1.00 0.940 1.00 0.790 1.00 0.004 0.920 1.00 -

Significant values are in bold and underlined. Sites with fewer than 10 turtles were excluded from this analysis, with exception of the Brazos River turtles.

Table 3.7 Pairwise FST values across sites for Sternotherus carinatus.

Drainage Brazos Trinity Neches Sabine Calcasieu Red Ouachita Comite Pearl Jourdan Pascagoula Escatawpa

Brazos 0

Trinity 0.088 0

Neches 0.107 0.057 0

Sabine 0.107 0.052 0.041 0

Calcasieu 0.111 0.040 0.046 0.026 0

Red 0.172 0.153 0.183 0.185 0.179 0

Ouachita 0.376 0.253 0.259 0.262 0.252 0.304 0

Comite 0.149 0.069 0.064 0.035 0.048 0.206 0.300 0

Pearl 0.162 0.069 0.082 0.076 0.047 0.184 0.283 0.065 0

Jourdan 0.184 0.080 0.138 0.118 0.095 0.289 0.397 0.131 0.107 0

Pascagoula 0.175 0.081 0.096 0.057 0.070 0.208 0.244 0.097 0.096 0.112 0

Escatawpa 0.176 0.093 0.100 0.067 0.070 0.234 0.316 0.086 0.096 0.115 0.047 0

FST values significantly different from zero are in bold and underlined. The Comite and Jourdan sample sizes were too small for significance testing.

Figure 3.1 The geographic range and sampling locations of Sternotherus carinatus

Figure 3.2 Bar plots of membership coefficients from the STRUCTURE analysis of

Sternotherus carinatus for values of K = 2 for the range wide analysis, a K = 4 for the

GCP turtles, and a K = 2 for the Highlands turtles.

Legend 1. - Sabine/Neches 2. - Pascagoula-Escatawpa 3. - Brazos-Trinity 4. - Red River 5. - Ouachita Easternmost 6. - Pearl/Jourdan Ouachita

5 2

6 1

Westernmost

Red R. P C A e ig e n v a lu e s D A e ig e n v a lu e s 4

Figure 3.3 Discriminant Analysis of Principle Coordinates for Sternotherus carinatus.

3.5 Literature Cited

Avise, J. C. 2000. Phylogeography: the history and formation of species. Harvard

university press.

Benke, A. C., and Cushing, C. E. 2005. Rivers of North America. Academic Press.

Bermingham, E., and Avise, J. C. 1986. Molecular zoogeography of freshwater fishes in

the Southeastern United States. Genetics 113:939–965.

Bodie, J. R. 2001. Stream and riparian management for freshwater turtles. Journal of

Environmental Management 62:443–455.

Brown, G. J., and Kreiser, B. In Press. Chelonian Conservation and Biology.

Characterization of microsatellite loci for the Razorback Musk Turtle

(Sternotherus carinatus) and their cross-amplification in five other species in the

family Kinosternidae.

Buhlmann, K. A., and Gibbons, J. W. 1997. Imperiled aquatic reptiles of the southeastern

United States: Historical review and current conservation status. Aquatic Fauna in

Peril: The Southeastern Perspective 1:201–232.

Cagle, F. R., and Chaney, A. H. 1950. Turtle populations in Louisiana. The American

Midland Naturalist 43:383–388.

Cheung, S. M., and Dudgeon, D. 2006. Quantifying the Asian turtle crisis: Market

surveys in southern China, 2000–2003. Aquatic Conservation: Marine and

Freshwater Ecosystems 16:751–770.

Congdon, J. D., Dunham, A. E., and Sels, R. C. van L. 1994. Demographics of Common

Snapping Turtles (Chelydra serpentina): Implications for Conservation and

Management of Long-Lived Organisms. American Zoologist 34:397–408.

Cox, W. A., and Marion, K. R. 1979. Population structure and survivorship in the musk

turtle, Sternotherus minor, in a North Florida spring run (Reptilia: Chelonia).

Bulletin of the Association of Southeastern Biologists 28:84.

Cureton, J. C., and Deaton, R. 2012. Hot moments and hot spots: Identifying factors

explaining temporal and spatial variation in turtle road mortality. The Journal of

Wildlife Management 76:1047–1052.

Dizon, A. E., Lockyer, C., Perrin, W. F., Demaster, D. P., and Sisson, J. 1992. Rethinking

the Stock Concept: A Phylogeographic Approach. Conservation Biology 61:24–

36.

Echelle, A. A., Hackler, J. C., Lack, J. B., Ballard, S. R., Roman, J., Fox, S. F., Leslie, D.

M., and Van Den Bussche, R. A. 2010. Conservation genetics of the alligator

snapping turtle: Cytonuclear evidence of range-wide bottleneck effects and

unusually pronounced geographic structure. Conservation Genetics 11:1375–

1387.

Ennen, J. R., Lovich, J. E., Kreiser, B. R., Selman, W., and Qualls, C. P. 2010. Genetic

and morphological variation between populations of the Pascagoula map turtle

(Graptemys gibbonsi) in the Pearl and Pascagoula Rivers with description of a

new species. Chelonian Conservation and Biology 9:98–113.

Ennen, J. R., Matamoros, W. A., Agha, M., Lovich, J. E., Sweat, S. C., and Hoagstrom,

C. W. 2017. Hierarchical, quantitative biogeographic provinces for all North

American turtles and their contribution to the biogeography of turtles and the

continent. Herpetological Monographs 31:142–168.

Ernst, C. H., Hershey, M. F., and Barbour, R. W. 1974. A new coding system for

hardshelled turtles. Transactions of the Kentucky Academy of Science 35:27–28.

Ernst, Carl H., and Lovich, J. E. 2009. Turtles of the United States and Canada. JHU

Press.

Evanno, G., Regnaut, S., and Goudet, J. 2005. Detecting the number of clusters of

individuals using the software structure: A simulation study. Molecular Ecology

14:2611–2620.

Gibbs, J. P., and Shriver, W. G. 2002. Estimating the Effects of Road Mortality on Turtle

Populations. Conservation Biology 16:1647–1652.

Goudet, J. (1995). FSTAT (version 1.2): A computer program to calculate F-statistics.

Journal of Heredity 86:485–486.

Hackler, J. C., Van Den Bussche, R. A., and Leslie, D. M. 2007. Characterization of

microsatellite DNA markers for the alligator snapping turtle, Macrochelys

temminckii. Molecular Ecology Notes 7:474–476.

Hartl, D. L., and Clark, A. G. 2006. Principles of Population Genetics (4th edition).

Sinauer Associates is an imprint of Oxford University Press.

Hearty, P. J., and Kaufman, D. S. 2000. Whole-rock aminostratigraphy and quaternary

sea-level history of the Bahamas. Quaternary Research 54:163–173.

Hubisz, M. J., Falush, D., Stephens, M., and Pritchard, J. K. 2009. Inferring weak

population structure with the assistance of sample group information. Molecular

Ecology Resources 9:1322–1332.

Iverson, J. B. 1977. Geographic Variation in the Musk Turtle, Sternotherus minor. Copeia

1977:502–517.

Jombart, T., Devillard, S., and Balloux, F. 2010. Discriminant analysis of principal

components: A new method for the analysis of genetically structured populations.

BMC Genetics 11:94.

Jr, R. D. H., Thomas, W. A., and Viele, G. W. 1989. The Appalachian-Ouachita orogen

in the United States. Geological Society of America.

Killebrew, F. C., Rogers, W. J., Babitzke, J. B., and Authority, E. A. 2002. Assessment of

instream flow and habitat requirements for Cagle’s map turtle (Graptemys caglei).

Report to Edwards Aquifer Authority Contract #00-52-AS.

Klemens, M. W. 2000. Turtle conservation. Smithsonian Institution Press.

Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., and Mayrose, I. 2015.

Clumpak: A program for identifying clustering modes and packaging population

structure inferences across K. Molecular Ecology Resources 15:1179–1191.

Lindeman, P. V. 2008. Sternotherus carinatus. In Conservation Biology of Freshwater

Turtles and Tortoises: A Compilation Project of the IUCN/SSC Tortoise and

Freshwater Turtle Specialist Group pp. 012.1-012.6.

Lindeman, Peter V. 1999. Surveys of basking map turtles Graptemys spp. In three river

drainages and the importance of deadwood abundance. Biological Conservation

88:33–42.

Lindeman, Peter V. 2013. The Map Turtle and Sawback Atlas: Ecology, Evolution,

Distribution, and Conservation. University of Oklahoma Press.

Lohse, E. A. 1955. Dynamic geology of the modern coastal region, Northwest Gulf of

Mexico. http://archives.datapages.com/data/sepm_sp/SP3/Dynamic_Geology.htm

Lovich, J. E., and Ennen, J. R. 2013. A quantitative analysis of the state of knowledge of

turtles of the United States and Canada. Amphibia-Reptilia 34:11–23.

Lovich, J. E., Ennen, J. R., Agha, M., and Gibbons, J. W. 2018. Where Have All the

Turtles Gone, and Why Does It Matter? BioScience 68:771–781.

Lovich, J. E., and McCoy, C. J. 1992. Review of the Graptemys pulchra group (Reptilia:

Testudines: Emydidae), with descriptions of two new species. Annals of the

Carnegie Museum 61:293–315.

Mayden, R. L. 1985. Biogeography of Ouachita Highland fishes. The Southwestern

Naturalist 30:195–211.

Melancon, S. R., Angus, R. A., and Marion, K. R. 2013. Demographic comparisons

between reservoir-dwelling and stream-dwelling populations of a threatened turtle

(Sternotherus depressus). Southeastern Naturalist 12:684–691.

Metcalf, A. L. 1964. Fishes of the Kansas River system in relation to zoogeography of the

Great Plains. Doctoral dissertation, University of Kansas, Lawrence, Kansas.

Moll, D., and Moll, E. O. 2004. The Ecology, exploitation and conservation of river

turtles. Oxford University Press, USA.

Moritz, C. 1994. Defining ‘evolutionarily significant units’ for conservation. Trends in

Ecology and Evolution 9:373-375.

Morton, R. A., and Price, W. A. 1987. Late Quaternary sea-level fluctuations and

sedimentary phases of the Texas coastal plain and shelf. In Sea-Level Fluctuation

and Coastal Evolution pp. 181–198.

Near, T. J., Kassler, T. W., Koppelman, J. B., Dillman, C. B., and Philipp, D. P. 2003.

Speciation in North American Black Basses, Micropterus (actinopterygii:

Centrarchidae). Evolution 57:1610–1621.

Pritchard, J. K., Stephens, M., and Donnelly, P. 2000. Inference of population structure

using multilocus genotype data. Genetics 155:945–959.

Raymond, M., and Rousset, F. 1995. GENEPOP (Version 1.2): Population genetics

software for exact tests and ecumenicism. Journal of Heredity 86:248–249.

Reed, R. N., Congdon, J., and Gibbons, J. W. 2002. The alligator snapping turtle

[Macrochelys (Macroclemys) temminckii]: A review of ecology, life history, and

conservation, with demographic analyses of the sustainability of take from wild

populations. Report, Division of Scientific Authority, United States Fish and

Wildlife Service, Aiken, South Carolina 17 pp.

Rhodin, A. G. J., Walde, A. D., Horne, B. D., Van Dijk, P. P., Blanck, T., and Hudson, R.

2011. Turtles in Trouble: The World’s 25+ Most Endangered Tortoises and

Freshwater Turtles—2011. IUCN/SSC Tortoise and Freshwater Turtle Specialist

Group, Turtle Conservation Fund, Turtle Survival Alliance, Turtle Conservancy,

Chelonian Research Foundation, Conservation International. Wildlife

Conservation Society, and San Diego Zoo Global. Lunenburg, MA.

Rice, W. R. 1989. Analyzing Tables of Statistical Tests. Evolution 43:223–225.

Roderick, G. K., and Gillespie, R. G. 1998. Speciation and phylogeography of Hawaiian

terrestrial arthropods. Molecular Ecology 7:519–531

Rosenberg, N. A., Mahajan, S., Ramachandran, S., Zhao, C., Pritchard, J. K., and

Feldman, M. W. 2005. Clines, Clusters, and the Effect of Study Design on the

Inference of Human Population Structure. PLoS Genetics 1:e70.

Rousset, F. 2008. genepop’007: A complete re-implementation of the genepop software

for Windows and Linux. Molecular Ecology Resources 8:103–106.

Ryder, O. A. 1986. Species conservation and systematics: The dilemma of subspecies.

Trends in Ecology and Evolution 1:9–10.

Schick, R. S., and Lindley, S. T. 2007. Directed connectivity among fish populations in a

riverine network. Journal of Applied Ecology 44:1116–1126.

Scott, P. A., Glenn, T. C., and Rissler, L. J. 2018. Resolving taxonomic turbulence and

uncovering cryptic diversity in the musk turtles (Sternotherus) using robust

demographic modeling. Molecular Phylogenetics and Evolution 120:1–15.

Shelby, J. A., and Mendonça, M. T. 2001. Comparison of reproductive parameters in

male yellow-blotched map turtles (Graptemys flavimaculata) from a historically

contaminated site and a reference site. Comparative Biochemistry and Physiology

Part C: Toxicology and Pharmacology 129:233–242.

Suter, J. R., and Berryhill, H. L. 1985. Late quaternary shelf-margin deltas, Northwest

Gulf of Mexico. The American Association of Petroleum Geologies Bulletin

69:77–91.

Thomas, T. M., Granatosky, M. C., Bourque, J. R., Krysko, K. L., Moler, P. E., Gamble,

T., Suarez, E., Leone, E., and Roman, J. 2014. Taxonomic assessment of alligator

snapping turtles (Chelydridae: Macrochelys), with the description of two new

species from the southeastern United States. Zootaxa 3786:141–165.

Thomson, R. C., Spinks, P. Q., and Shaffer, H. B. 2018. Molecular phylogeny and

divergence of the map turtles (Emydidae: Graptemys). Molecular Phylogenetics

and Evolution 121:61–70.

Thornbury, W. D. 1965. Regional geomorpholoy of the United States. John Wiley and

Sons. New York.

Tinkle, D. W. 1958. The systematics and ecology of the Sternotherus carinatus complex

(Testudinata: Chelydridae). Tulane Studies of Zoology 6:1-56.

Trauth, S. E., Robison, H. W., and Plummer, M. V. 2004. The Amphibians and Reptiles

of Arkansas. The University of Arkansas Press.

Vähä, J.-P., Erkinaro, J., Niemelä, E., and Primmer, C. R. 2007. Life-history and habitat

features influence the within-river genetic structure of Atlantic salmon. Molecular

Ecology 16:2638–2654.

Vogler, A. P., and Desalle, R. 1994. Diagnosing Units of Conservation Management.

Conservation Biology 8:354–363.

Wang, Q., Yu, Y., Yuan, J., Zhang, X., Huang, H., Li, F., and Xiang, J. 2017. Effects of

marker density and population structure on the genomic prediction accuracy for

growth trait in Pacific white shrimp Litopenaeus vannamei. BMC Genetics 18:45.

Winker, C. D. 1979. Late Pleistocene fluvial-deltaic deposition, Texas coastal plain and

shelf. Gulf Coast Association of Geological Societies Transactions 29:215.

Wright, S. 1978. Evolution and the Genetics of Populations. Variability Within and

Among Natural Populations. University of Chicago Press, Chicago, IL.

CHAPTER IV – INTERACTIONS OF THE STRIPE-NECKED MUSK TURTLE AND

RAZORBACK MUSK TURTLE IN SYMPATRY AND SYNTOPY IN THE

PASCAGOULA RIVER SYSTEM

Abstract

Competition and hybridization between closely-related species is a topic of interest in many fields of biology: conservation biology, evolutionary biology and ecology. Gause’s Principle suggests that two species competing for the same resources in the same habitat cannot co-occur without sufficient differentiation in habitat use, quotidian activity or resource consumption. The ranges of the razorback musk turtle

(Sternotherus carinatus) and stripe-necked musk turtle (Sternotherus peltifer) are mostly non-overlapping, but these two species of lotic musk turtle are sympatric in South

Mississippi. In allopatry, both species fill the niche of a lotic, bottom-walking musk turtle species and can range in habitat from small, fast-flowing mountain stream of bedrock and boulder to the large, muddy bayous of the coastal plain. To determine how these species are able to coexist in sympatry despite their remarkably similar ecologies, I collected and analyzed environmental, morphological and genetic data to assess whether the two species exhibit different habitat associations, differ in size, and whether the two species are maintaining reproductive isolation. The two species differ significantly in habitat association, with the stripe-necked musk turtle occurring primarily in small streams or headwater stream habitats, and the razorback musk turtles in the larger river habitats. The two species occur in syntopy in intermediate habitats, and while the species can hybridize, hybridization is likely precluded by both pre- and postzygotic barriers like the

significant differences in body size between species and the reversed sexual size dimorphism exhibited by each species.

4.1 Introduction

Competition between closely related species has long been a focal point in ecology. Voltera (1926) and Gause (1934) were the first to posit that if two species’ requirements are sufficiently similar, then only one species should be able to persist.

Brown and Wilson (1956) coined the term “character displacement” as another condition that allowed two species to co-occur. Character displacement, as they defined it, is the divergent patterns ecology, behavior, physiology, and/or morphology of two species where they occur together. Shortly after, MacArthur (1958) was among one of the first to demonstrate this in sympatric warbler species. The five species of warbler he observed shared common prey resources, but they partitioned the tree canopy to reduce interspecific competition with one another. MacArthur’s work alluded to the concepts of realized and fundamental niches of species in sympatry and allopatry, where in allopatry, or in the absence of a competitor, a species can occupy its fundamental niche and access the full range of biotic and abiotic conditions under which is can grow and reproduce. In contrast, when two species exhibit interspecific competition, the niche that they occupy is narrower than their fundamental niche, and this is deemed their realized niche (Griesemer

1992).

Within chelonians, examples of resource partitioning are not uncommon. For example, niche overlap has been examined in marine turtles (Pike 2013), Australian turtles (Meathrel et al. 2002; Welsh et al. 2017), and South American turtles (Alcalde et al. 2010). Within the United States, map and sawback turtles (genus Graptemys) are the

most speciose group of turtles in the country (Vogt 1981; Fuselier and Edds 1994;

Lindeman 2000) and provide many examples of character displacement. When two species co-occur within a river drainage, one species specializes on a molluscivorous diet with females (the larger sex) developing large, well-developed alveolar plates and mega- or mesocephaly (Lindeman 2000; Lindeman 2013), while the congener has a small head used for a primarily spongivorous diet (Seigel and Brauman 1994; Lindeman 2013). This pattern is replicated across many Gulf Coast drainages (e.g., Graptemys pseudogeographica kohnii and G. sabinensis in the Sabine River drainage, G. pearlensis and G. oculifera in the Pearl River drainage, G. gibbonsi and G. flavimaculata in the

Pascagoula River drainage, and G. pulchra and G. nigronoda in the Mobile River drainage; Lindeman 2013).

Following a similar distribution to the Graptemys along the Gulf Coastal

Drainages, lotic versions of musk turtles (genus Sternotherus) will occupy anywhere from one to many drainages, but unlike the Graptemys, generally only one species occurs per drainage. There are currently five recognized forms of lotic Sternotherus (from west to east: S. carinatus, S. peltifer, S. depressus, S. intermedius and S. minor) that share very similar ecologies and have mostly non-overlapping distributions (Figure 4.1). The exceptions are S. peltifer and S. depressus which can co-occur at the fall line of the Black

Warrior River in Alabama (S. depressus occurs above the fall line, and S. peltifer below), and S. carinatus and S. peltifer in the Pearl and Pascagoula River drainages (Ernst and

Lovich 2009). Whereas S. depressus and S. peltifer have a singular contact point at the fall line, S. carinatus and S. peltifer co-occur throughout two river drainages (the Pearl and Pascagoula) at the eastern- and westernmost extent of each species’ ranges, while

also occurring allopatrically across other drainages (Fig. 4.1). This represents a somewhat unique phenomenon in the southeastern United States, where allopatrically the two species are fulfilling nearly identical niches (like the Graptemys across drainages).

However, little is known about what niches they fill and how they interact in sympatry.

Sternotherus carinatus and S. peltifer occupy very similar ecological roles in lotic habitats in allopatry (i.e., east of the Pascagoula [peltifer] and west of the Pearl

[carinatus]). For example, S. peltifer can be found throughout the entirety of the Mobile

River drainage (with the exception of the Black Warrior Drainage above the fall line which is inhabited by S. depressus) and large portions of the Tennessee River drainage from small streams dominated by boulders and cobble in the Blue Ridge Mountains of the Southern Appalachians all the way to the large, turbid and slow-moving Alabama

River near the Tensaw Delta (P. Scott, pers. comm.). Similarly, in the absence of S. peltifer, S. carinatus can be found in small creeks dominated by boulders, cobble and gravel in the Southern Ouachita Mountains and similarly into large rivers, oxbows and floodplains (Kavanagh and Kwiatkowski 2016; GJB pers. obvs.). To complicate matters,

Sternotherus peltifer was presumed to be absent in the Pascagoula (Iverson 1977). In fact, it was not until Vogt et al. (1978) trapped S. peltifer in the Chickasawhay River at

Waynesboro, Mississippi, that the species was formally documented within the

Pascagoula drainage. The authors noted the systematic and zoogeographic significance of

S. peltifer’s presence within the Pascagoula as it confirmed the type locality of the species as likely from the Pascagoula system near Bassfield, MS (Smith and Glass 1947), and it also provided an explanation for the dispersal of S. peltifer into the Pearl drainage

other than the previously proposed overland migration from the Tombigbee headwaters into the Pearl (Iverson 1977).

Still, no studies have attempted to fully document the distribution and abundance of S. peltifer within the Pearl and Pascagoula drainages. Since the species was confirmed in the Pascagoula River drainage in 1978, only seven other localities (seven individuals) for S. peltifer have been reported from the entirety of the Pascagoula river drainage, and currently there are still only three known localities within the Pearl river drainage

(Iverson 1977). Five of the seven Pascagoula River drainage localities were from the

Upper Chickasawhay drainage, an area that also has many S. carinatus records. Though no hybrids between these two species have ever been reported from the wild, hybridization events are not uncommon in turtles (Parham et al. 2013; Godwin et al.

2014), and hybridization has been documented in Sternotherus by Scott and Rissler

(2015) and Scott et al. (2019) who detected hybridization between S. peltifer and S. depressus at the fall line in Alabama.

While the Pearl and Pascagoula River systems have been surveyed extensively by turtle biologists for decades, the primary focus of much of this work were the federally protected sawback species: Graptemys flavimaculata and G. oculifera (Cagle 1953; Vogt

1980; McCoy and Vogt 1980, Jones and Hartfield 1995; Lindeman 1999; Selman and

Qualls 2011). Sampling for these turtles involved unbaited basking traps or fyke nets, so the number of Sternotherus in these surveys was small. The reason that Sternotherus peltifer eluded detection for so long was because in sympatry with Sternotherus carinatus, it appears that S. peltifer are more commonly encountered in smaller streams,

where fewer Graptemys are present, whereas S. carinatus occupies larger riverine habitats, where most of the aforementioned Graptemys studies were focused.

While there are some studies that have looked at sympatric interactions between species of Sternotherus (Berry, 1975; Brown 2008, unpublished thesis), these studies looked at the interactions between the lotic forms Sternotherus minor (Berry 1975) and

Sternotherus carinatus (Berry 2008, unpubl. thesis) with a lentic form (S. odoratus). No work has focused exclusively on the overlap of two lotic Sternotherus species. Studying two lotic turtle species in sympatry provides a unique opportunity to better understand the ecological (abiotic and/or biotic) drivers influencing these species distributions. In a broader context, this study can be used to address broader ecological questions like the forces that help create and maintain biodiversity, particularly as it pertains to coexistence between closely related taxa. In this study, I investigate the sympatric interactions between the razorback musk turtle (Sternotherus carinatus) and the stripe-necked musk turtle (Sternotherus peltifer). In this study I wanted to address: 1) do the two species differ significantly in their habitat association; 2) do the two species differ in morphology; and 3) are the two species are maintaining reproductive isolation, particularly at sites of syntopy.

4.2 Method

4.2.1 Site Selection

I collected Sternotherus tissue opportunistically from 2015 – 2019, as well as systematically from 28 April 2018 – 31 October 2019 as part of a study on habitat associations detailed below.

Preliminary sampling for Sternotherus peltifer and S. carinatus highlighted the importance of sampling a variety of lotic systems from small streams to larger rivers. I used stratified random sampling to select 60 potential trapping locations from points of public access (boat launches, bridge crossings, etc.). To identify potential sites, I used geographical information systems (GIS; ArcGIS 10.5; ERSI, Redland, CA, U.S.A.). First,

I downloaded the Elevation Derivatives for National Application (EDNA) watershed layer for the Pascagoula river drainage from the U.S. Geological Survey as well as GIS data for Mississippi roads from Mississippi Geospatial Clearinghouse

(http://www.census.gov/geo/www/tiger). I then intersected these two layers to create a pool of potential access points to streams and rivers, as well as downloaded the GPS coordinates for public boat launches in the Pascagoula river drainage from the

Mississippi Department of Wildlife, Parks and Fishers “Ramps and Piers” program. I classified sites into one of three groups based on discharge data from the EDNA watershed layer: small stream (0.5-10 m3/sec), intermediate sized streams (10-100 m3/sec) and larger rivers (100+ m3/sec). I used a random number generator to select twenty sites from each category. These sites were screened with GoogleMaps

(www.maps.google.com) to verify the stream crossing selected represented a natural, lotic environment and not irrigation ditches, which proved prevalent on the landscape.

4.2.2 Trapping

During systematic trapping efforts, stream or riverine sites were accessed by wading (small streams), canoe (intermediate streams and rivers without boat launches) or jon boat (rivers where boat launches were available). At each site, I set 15 collapsible minnow traps (36” x 12” Promar ©) for two nights, for a total of 30 trap-nights per site. I

set traps in shallow water (25cm -100 cm in depth) near suitable habitat (e.g. woody debris, log jams, or rock ledges) and baited with one can of sardines in soybean oil

(Bumble Bee Foods ©). All traps were tied to a stationary point (i.e., secure instream deadwood or the bank). An empty water bottle was place into each trap to account for any fluctuations in water level in order to reduce risk of drowning. Traps were checked every 24 hours for 48 hours. Turtles were also hand caught opportunistically while setting and checking traps.

4.2.3 Morphometrics and Stream Characteristics

For each Sternotherus species trapped and/or caught, I collected the following measurements: midline carapace length (CL), midline plastron length (PL), carapace width at the junction of the posterior bridge at the seam of the 6th and 7th marginal scute

(SW), carapace height measured behind the 2nd vertebral, head width measured immediately posterior to the orbital socket (HW), and head depth also measured directly posterior to the orbital socket. I also collected mass (g), dorsal and ventral photos (with ruler for scale), documented any injuries or abnormalities, and collected a small genetic sample by way of a 5-10 mm piece of interdigital tissue from the hind food, which was stored in 95% ethanol. Both Sternotherus carinatus and S. peltifer are sexually dimorphic. Males possess a longer tail and an oval patch of rough skin (rasping organs) on the interior of the hind legs (Smith and Glass 1947; Ernst and Lovich 2009). I also scored each adult Sternotherus based on the number of posterior marginals from left and right marginal scutes 9-11 that were injured or missing as part of a study on the potential aggressive interactions between the two sympatric Sternotherus species. A score of zero indicates no damage, a score of 6 was the highest score.

I collected stream habitat variables at each site to characterize stream size as well as the stream reach (e.g., a large silty river, a small stream dominated by soapstone, etc.).

I measured habitat characteristics on the first day of trap checking, 24 hours after traps were set. Stream habitat metrics were taken at 25, 50 and 75% of the length of transect.

Small stream transects were approximately 500 m, intermediate sites were approximately

1000 m, and large river sites were typically 1500 m. At these three points along the transect, I collected the following at 25, 50 and 75% of the stream width: stream substrate composition (using a modified Wentworth scale where 1 = silt/mud, 2 = sand, 3 = pea- sized gravel, 4 = large gravel, 5 = cobble and 6 = bedrock and/or soapstone; Cummins

1962), amount of deadwood (0 = absent, 1 = small amount woody debris, 2 = abundant deadwood), wetted width (m), stream depth (cm), canopy cover (approximated as 0%,

25%, 50%, 75% or 100%), turbidity (0 = very clear, 1 = clear, 2 = slightly turbid, 3 = very turbid). From GIS, I used the National Hydrology Plus database to extract the cumulative upstream drainage area (CDA) at each site as a descriptor of stream size

(USEPA, USGS, 2012).

4.2.4 Statistical Analysis

I used only adult turtles in statistical analyses of morphology (>80mm CL in

Sternotherus carinatus and >70 mm CL in S. peltifer; modified from Iverson 1977). For adults, I used a two-factor, factorial analysis of variance (ANOVA) with species and sex as factors to test three null hypotheses 1) the two species were equal in size (mass) 2) the two sexes were equal in size (mass) , and 3) there is no interaction between species and sex. I used mass to test these hypotheses as many turtles exhibited damage to the carapace or plastron, which affected carapace and plastron lengths. After a log

transformation of masses, data for each hypothesis met the necessary assumptions of normality and equal variances. A post hoc test (Tukey HSD), as needed, was applied to determine where significant differences occurred.

The posterior marginal damage data were meristic data, and not surprisingly, the distribution of these data was not normal. Since assumptions of normal distributions and equal variances could not be met, I used the Wilcoxon Rank-Sums test to determine which species exhibited more damage, and if there were statistical differences in the levels of damage between syntopic and allotopic (see Rivas 1964 for definitions of each term) populations of Sternotherus peltifer. All statistical analyses were performed in the

‘stats’ package in R 3.6.1. All alpha values were set at 0.05.

For the habitat data, I used a canonical correspondence analysis (CCA) to elucidate any habitat associations between turtles and different lotic habitats. A CCA is used to evaluate the associations between two sets of variables, and it allows one to visualize these associations in ordination space (Ter Braak 1986; Legendre and Legendre

2012). Because stream characteristics can be highly correlated, I checked for multicollinearity across habitat variables by examining the variance inflation factor scores (VIF scores; Legendre and De Cáceres 2012), and if two variables were too highly correlated (VIF scores > 4; Hair et al. 2010), one was removed from the analysis and VIF scores were re-evaluated before proceeding with the CCA. I used a PERMANOVA with either terms or CCA axes as fixed variables with 105 permutations to detect whether that axis or variable explained a significant amount of the variance in the dataset. Alpha levels were set at 0.05. In the first CCA, I included Sternotherus data to test for habitat associations with these two species, and in a second CCA, I plotted the habitat

associations of all turtle species with >5 detections during this study (i.e., observed >10% of the time). For this analysis, I used the “cca” function in the “vegan” package in R

3.6.1.

4.2.5 Genetic Analysis

The morphological identification (to species) of each individual was confirmed using through genetic analyses of microsatellite loci and a mitochondrial marker. Total genomic DNA was extracted from tissue samples with a DNeasy Tissue Kit (QIAGEN

Inc., Valencia, CA). Brown and Kreiser (in press) identified six diagnostic microsatellite loci (Scar06, Scar08, Scar26, Scar27, Scar29 and Scar39) for Sternotherus carinatus and

S. peltifer with species-specific allele sizes that do not overlap with those present in the other species. Polymerase chain reaction conditions and thermocycler profiles followed

Brown and Kreiser (in press). Microsatellite alleles were visualized on a polyacrylamide gel using a LICOR 4300 DNA Analyzer. Alleles were sized using GeneProfiler ver. 4.05

(LICOR Co.). I tested for linkage disequilibrium among loci within each species using

Genepop on the Web (Raymond and Rousset 1995; Rousset 2008). A Bonferroni correction was applied to the alpha value to account for multiple comparisons (Rice

1989).

I used STRUCTURE v. 2.3.4 to determine the proportion of each individual’s ancestry that was represented by Sternotherus carinatus and S. peltifer. I tested for a K of

2 for the two species, without using location as a prior, and assuming correlated allele frequencies with admixture between groups. I performed 10 independent runs of a K of 2 with a burn in of 50,000 and 100,000 MCMC replications. CLUMPP (Jakobsson and

Rosenberg 2007) as performed by StructureSelector (Li and Liu 2018) was used to

average all ten runs, which were then subsequently visualized using Distruct v. 1.1

(Rosenberg 2004).

In order to evaluate maternal ancestry of each turtle, I used a diagnostic restriction fragment length polymorphism (RFLP) of the mitochondrial gene, cytochrome b (cytb).

For this RFLP locus, sequences generated by other studies for all Kinosternids (Shaffer et al. 1997; Scott and Rissler 2015) were obtained from GenBank, and Sequencher 4.10.1

(Gene Codes; Ann Arbor, Michigan) was used to determine which restriction endonucleases produced species-specific fragment patterns. For this study, I used the restriction endonuclease Dpn II. I used the polymerase chain reaction to amplify this portion of the mitochondrial genome in 25 µL reactions consisting of 15.9 µL of dH2O,

2.5 µL 10× standard Taq (Mg-free) buffer (New England Biolabs), 2.0 µL 2 mM dNTPs,

2.0 µL 25 mM MgCl2, 0.5 units Taq polymerase (New England Biolabs), and 20-50 ng of

DNA template. PCR cycling conditions were as follows: initial denaturation at 94°C for 2 minutes, 30 cycles of 1 minute at 94°C, 1 minute at a 50°C, and 1 minute at 72°C, with a final elongation of 7 minutes at 72°C. Restriction digests using Dpn II (New England

Biolabs) took place in a 20 µL volumes of 7.9 µL of dH2O, 2 µL 10x Dpn II reaction buffer (New England Biolabs), 0.1 µL of Dpn II (New England Biolabs) and 10 µL of

PCR product. Samples incubated at 37°C for 4 hours. The restriction fragment digests were visualized on a 1.5% agarose gel stained with ethidium bromide. Individuals were scored as having a haplotype of either Sternotherus carinatus or Sternotherus peltifer.

4.3 Results

4.3.1 Trapping

I trapped 57 (20 small, 19 intermediate and 18 large) riverine and stream sites.

Fifty-two of these sites were chosen from the stratified random sampling of access points in GIS, and five were selected as known sites (Figure 4.1). This totaled 1,516 trap nights across the Pascagoula River watershed. A total of 174 trap nights were lost likely associated with alligator snapping turtle interference, whereby the turtle would rip open nets and steal the tin of bait. In total, 346 turtles were captured across 53 of the 57 sites totaling 9 species (Table 4.1). However, in addition to those species trapped, a few other species were observed during this study that were not caught in traps. I encountered several Chelydra serpentina near traps, as well as both map turtle species (Graptemys gibbonsi and G. flavimaculata) which are notoriously bait averse (Boyer 1965; Selman and Qualls 2009; Lindeman 2013; Lindeman 2014), as well as documented two county records for the southern painted turtle (Chrysemys dorsalis).

In 2018, I detected Sternotherus at 43 of the 57 sites (Table 4.1). Sternotherus carinatus were found at 22 river and stream sites. Sternotherus peltifer were caught at 30 stream and river sites, and I detected both species, in syntopy, at 9 sites (Figure 4.2). In total, 189 Sternotherus were trapped during these systematic trapping efforts – 88

Sternotherus carinatus (+12 hand captures) and 101 Sternotherus peltifer (+9 hand captures). In addition to these two Sternotherus species, I also caught 7 Sternotherus odoratus at six sites, as well as, very surprisingly, 3 Kinosternon subrubrum at one stream site. Kinosternids made up 57.5% of all turtles caught, illustrating the effectiveness of using small, baited traps for these species. Red-eared sliders (Trachemys

scripta elagans) were the third most abundant species caught in traps (N = 59), and the alligator snapping turtle (Macrochelys temminckii; N = 39 juveniles plus 100+ lost trap nights from adults), a candidate for federal listing, was the fourth most abundant species.

In addition to the trapping locations above, from 2015 to 2019, I opportunistically detected S. carinatus at an additional 15 sites, S. peltifer from an additional 11 sites, and

5 additional syntopic sites, for a total of 64 sites across the Pascagoula River drainage

(Table 4.2, Figure 4.2).

4.3.2 Habitat Associations

Upstream drainage area ranged from 12.0 km2 (Cooks Creek) to 21,195 km2

(Pascagoula River at Wade-Vancleave). Sternotherus were detected at sites ranging from

17.8 km2 (Garraway Creek) to 17,162 km2 (Pascagoula River in Benndale, MS; Table

4.1). The first CCA analysis of habitat associations between just Sternotherus carinatus and S. peltifer explained 65.2% of the variation (i.e., constrained inertia). Because there are only two species in this model, there is only 1 CCA axis which explained a significant amount of variation in the dataset (F = 78.803, p = 0.001), and this axis was driven by stream size. Four habitat variables were included in the model (all had VIF scores < 4;

Hair et al. 2010), and three of these were significant: stream depth (F = 48.3, p = 0.001), water temperature (F = 12.4, p = 0.005) and canopy cover (F = 7.94, p = 0.005).

Sternotherus carinatus and S. peltifer had a polar relationship along the first CCA axis

(1.17 and -0.86 respectively).

The analysis of the community dataset demonstrated similar patterns for the two musk turtle species, but the addition of other species to the model reduced the constrained variance explained to 23.1% (Fig. 4.3). The first axis (CCA1) remained significant (F =

12.0, p = 0.001). The three terms remained significant: stream depth (F = 6.60, p =

0.001), canopy cover (F = 3.93, p = 0.006), and water temperature (F = 2.49, p = 0.027).

CCA1 was again driven by stream size, and Sternotherus carinatus and S. peltifer are again polar in their position in ordination space (Figure 4.3). Trachemys scripta, a well- known generalist, not surprisingly occurred near the origin of the CCA (suggesting little habitat preference), while Macrochelys temminckii seems to be associated with larger, deeper stream reaches and Apalone spinifera with larger, wider stream reaches.

4.3.3 Genetics

In total, I collected tissue from 210 Sternotherus carinatus and S. peltifer from trapping in the spring, summer and fall of 2018. I collected an additional 185 tissue samples from Sternotherus carinatus and S. peltifer opportunistically for other studies of these species. In total 433 Sternotherus were analyzed in this study; 211 S. carinatus and

222 S. peltifer from the Pascagoula river drainage were genotyped across 6 diagnostic microsatellite loci. None of the six microsatellite loci demonstrated linkage disequilibrium after a Bonferroni correction. I did not explicitly test for Hardy-Weinberg

Equilibrium as I was not interested in population subdivision in these species but the ancestry of each, which does not hold the genetic data to the same assumptions.

In the nuclear (microsatellite) dataset, STRUCTURE determined that out of 433 turtles, only 3 demonstrated signs of mixed ancestry (q-score < 0.90; Figure 4.4, Table

4.3). Two were identified in the field as Sternotherus peltifer while the third appeared to have characteristics of each species, and it was noted as a putative hybrid. Excluding the turtles of mixed ancestry, the average ancestry coefficient (q-score) for S. carinatus was

0.991 (SD = 0.010), and the average q-score for S. peltifer 0.993 (SD = 0.007; Table 4.3).

Of the turtles with mixed ancestry, one was from an area of known syntopy (Sp136; Site

#58), and another was from a small stream immediately upstream from a syntopic stretch of river (Sp 85, Site #49). The third was from a small tributary of Red Creek, and though no Sternotherus were detected during trapping of Red Creek downstream of the tributary, the habitat looks suitable for co-occurrence of the two species (intermediate stream size, intermediate canopy cover, high substrate score; GJB pers. obs.; Sp93; Site #67). The individual identified as a hybrid had q-scores reflecting 50% ancestry from each species, whereas the other two had peltifer dominated q-scores (0.80 and 0.67; Table 4.3). The two S. peltifer of mixed ancestry had peltifer mtDNA haplotypes as did the putative hybrid, and all remaining RFLP haplotypes matched species designation upon capture.

4.3.4 Morphology

The two-factor, factorial ANOVA found a significant effect of species (F1,298 =

234.6, p <0.001) and a significant interaction of species and sex (F1,298 = 22.7, p < 0.001), but no significant effect of sex (F1,298 = 0.024, p = 0.878). S. carinatus was significantly larger (x̄ : 240.6 ± 75.1 g) than S. peltifer (x̄ : 140.1 ± 44.2 g) in the Pascagoula river drainage. A significant interaction between sex and species indicated that both species exhibited sexual dimorphism within the Pascagoula River drainage, though in polar directions. In Sternotherus carinatus males are significantly larger (x̄ : 261.2 ± 82.8 g) than females (x̄ : 219.7078 ± 59.9 g), while in Sternotherus peltifer, females are significantly larger (x̄ : 152.8 ± 44.9 g) than males (x̄ : 127.8 ± 40.1 g).

I scored posterior marginal scute damage for 332 individuals (179 S. carinatus,

153 S. peltifer) to estimate levels of aggression in Sternotherus species within the

Pascagoula River drainage. A Wilcoxon Rank Sum test indicated that Sternotherus

carinatus had significantly more injuries to the posterior marginals than Sternotherus peltifer in the Pascagoula river drainage (W = 19714, p < 0.001). For syntopic and allotopic populations of Sternotherus peltifer, the Wilcoxon Rank Rum test indicated that syntopic populations of Sternotherus peltifer had significantly more injuries to the posterior marginals than allotopic populations of Sternotherus peltifer in the Pascagoula river drainage (W = 2409.5, p < 0.01)

4.4 Discussion

Sternotherus carinatus and S. peltifer are both wide ranging lotic Sternotherus species that occupy very similar niches across their allopatric geographic distributions.

However, this study is the first to evaluate the ecology and the co-occurrence of

Sternotherus carinatus and S. peltifer in sympatry. These two species share a narrow range of overlap in the Pearl and Pascagoula River systems. While Sternotherus carinatus has been well documented throughout both the Pearl and Pascagoula drainages, prior to this study, remarkably few records of S. peltifer existed from either the Pearl or

Pascagoula River systems. Within the Pascagoula River system though, I found S. peltifer to be abundant and distributed throughout most of the drainage’s major tributaries (with exception of the Escatawpa River drainage). I was able to detect Sternotherus carinatus in all major tributaries of the Pascagoula River systems, in some counties where it was previously undocumented.

As previously mentioned, the two species fill very similar ecological roles, and they can both be found in a variety of lotic habitats from montane headwater streams to the large, meandering rivers of the Gulf Coastal Plain. Therefore, where these two species co-occur has provided an interesting framework for the study of the ecology of

competition in closely-related, sympatric species. It was originally Darwin (1859) who speculated that the most “closely-allied” forms (i.e., closely-related) experience the most intense competition with one another. This sentiment evolved into what is now known as the theory of competitive exclusion, or Gause’s principle (Gause 1934, Gilbert et al.

1952; Harper et al. 1961), and these concepts also laid the foundations for community assembly rules (Diamond 1975) which dictate that there should be less species co- occurrence than by chance (Gotelli and McCabe 2002). However, there are conditions under which closely-related species can coexist. Chunco et al. (2012) posited that species can co-occur at 1) intermediate environmental conditions, 2) under the most extreme conditions, or 3) independent of environmental conditions, but dependent on other biotic characteristics like dispersal ability.

In the case of sympatric Sternotherus carinatus and S. peltifer, a CCA illustrates that the two are significantly different in their habitat associations, with S. peltifer inhabiting smaller streams (higher canopy cover + high substrate score), and S. carinatus occupying larger water bodies (deeper + warmer; Figure 4.3). For the majority of sites

(~80%), the two species were detected in allotopy, that is, only one of the species was detected. However, there were nine (syntopic) sites where both species were detected during 2018 trapping efforts and an additional five syntopic sites were discovered through opportunistic and collaborative sampling forays. Most of these sites had commonalities that made them suitable for both species. Where the two species co- occurred was akin to a “Goldilocks” effect. These sites were typically in headwater reaches of rivers (11 out of 14) that had intermediate levels of canopy cover, high substrate scores, and provided an interface between small feeder streams and river

stretches. This sort of syntopy between closely related species in intermediate habitats has also been documented in salamanders (Wall 2009), birds (Swenson 2006), mammals

(Kamler et al. 1998), and fishes (Schaefer et al. 2011).

A potential consequence of this co-occurrence of congeneric species is hybridization. The study of hybridization between closely related species or genetically distinct populations has been of interest and has important implications across a number of fields in biology, particularly in conservation biology (Todesco et al. 2016), systematics (Berger 1973), and evolutionary biology (Hilbish et al 2012). In this study, I used both nuclear and mitochondrial markers to determine not just if there is gene flow between species, but also the directionality of the gene flow, as sometimes hybridization only happens unidirectionally (McGowan and Davidson 1992; Wirtz 1999; Beatty et al.

2010; Den Hartog et al. 2010; Scott et al. 2019). Because the two species in this study overlap at intermediate habitats, I expected that there could by hybrid zones along these stream reaches. However, this was not the case, I found no indication of a hybrid zone between these two species despite over a dozen documented sites of syntopy. While there was one documented hybrid, and two potentially back-crossed individuals, these were very rare in the dataset (3 out of 433, or a frequency of ~0.7%). In the Bouie River, both

Sternotherus carinatus and S. peltifer were captured at all 3 sites stretching from

Hattiesburg, Mississippi (Highway 49) to Seminary-Sumrall Road in Sumrall,

Mississippi, a contact zone stretching at minimum 24.1 km, and of the 27 Sternotherus carinatus and 20 Sternotherus peltifer, no turtles from this stretch of river came back with evidence of mixed ancestry. Likewise, in Black Creek, both species were detected from an upstream site (Highway 11, Purvis, Mississippi) down to Fairley Landing on

Black Creek, spanning at least 80 rkm of co-occurrence. However, there was still no discernable hybrid zone.

While I was not able to explicitly test whether pre- or postzygotic barriers prevented the formation of hybrid zones between these two species, some inferences can be drawn from phylogenetic, morphometric, and other data on these species. The phylogenetic position of razorback and stripe-necked musk turtles has fluctuated over time. Originally S. peltifer was described as a full species (Smith and Glass 1947), but it was later lumped in as a subspecies of S. minor and sister to S. carinatus (Tinkle 1958).

While early allozyme (Seidel and Lucchino 1981) and mitochondrial studies (Walker et al. 1998) found Sternotherus carinatus sister to the predominantly lentic eastern musk turtle, or stinkpot, Sternotherus odoratus, S. carinatus was again considered to be in the same clade as the other lotic sternotherines from 1998 to 2013. It was not until recent advances in sequencing and the inclusion of additional nuclear markers that the phylogeny was resolved, and Sternotherus carinatus was again placed sister to S. odoratus (Iverson et al. 2013). The most recent and well-supported phylogeny comes from Scott et al. (2018) after the description of Sternotherus intermedius. The minor clade includes Sternotherus minor, S. peltifer, S. depressus and S. intermedius, with the other two Sternotherus (S. odoratus and S. carinatus) as sister taxa. Phylogenetic distance between these two species may preclude some hybridization. The top-minnow species

Fundulus notatus and F. olivaceus are an ichthyological analog to these Sternotherus species (Schaefer et al. 2009; Duvernell et al. 2013). These fishes are sympatric with both

S. carinatus and S. peltifer in the Pascagoula River system, and they also demonstrate similar divergent habitat associations, with F. notatus occupying larger, slower moving

lotic habitats (like S. carinatus) and F. olivaceous occupying headwater or adventitious streams (like S. peltifer). However, these two species are known to hybridize at the interfaces of these lotic habitats. These two fishes are sister taxa, and their proximate position phylogenetically likely facilitates gene flow between the two species (Schaefer et al. 2009; Duvernell et al. 2013).

Morphologically, Sternotherus carinatus, at least within the Pascagoula river drainage, is a significantly larger species than the stripe-necked musk turtle. Furthermore, the two species have opposing sexual dimorphism with males being the larger sex in

Sternotherus carinatus and females being the larger sex in S. peltifer. Sexual dimorphism in razorback musk turtles has been well-documented (Ersnt and Lovich 2009; Kavanagh and Kwiatkowshi 2016), while most of the data on sexual dimorphism within the minor complex has focused on Sternotherus minor, with which Sternotheurs peltifer used to be considered conspecific. This may be the first study to report sexual dimorphism in

Sternotherus peltifer. The discrepancies between the opposing sexual dimorphism may explain the unidirectional gene flow seen in the mtDNA data of the three turtles of mixed ancestry that all had peltifer haplotypes. Gene flow was only observed from hybrid matings between male S. carinatus and female S. peltifer. Indeed, it seems far more likely that a small male razorback would attempt to copulate with a similarly-sized female stripe-necked musk turtle, than for a diminutive stripe-neck male attempt to mate with a much larger razorback female.

Intra- and interspecific aggression may also exacerbate this issue of unidirectional gene flow. In clear Floridian springs, the loggerhead musk turtles have sexual dimorphism similar to razorback musk turtles (males are significantly larger; Munscher

unpub. data). The male loggerhead musk turtles have been documented chasing and biting at (and occasionally breaking off) the posterior marginal scutes of the carapace of smaller male loggerheads. Munscher et al. (unpubl. data) used count data of damaged posterior marginal scutes in addition to field observations to evaluate the intraspecific, intrasexual aggression in this species. While I have not been able to personally observe this behavior in razorback musk turtles, predominantly due to lack of visibility in rivers in South Mississippi, I speculate the damage to the posterior marginals may also reflect male-on-male aggression. From the 332 scored turtles (179 S. carinatus, 153 S. peltifer), I found that Sternotherus carinatus had significantly more damage than S. peltifer.

Additionally, and interestingly, I found that S. peltifer caught in syntopy with S. carinatus had significantly more damage than S. peltifer in allotopy. This would suggest that

Sternotherus carinatus may also behave aggressively against conspecifics as well as congenerics. Thus, not only is it improbable that a diminutive S. peltifer male would attempt to mate with a much larger Sternotherus carinatus female, but it is possible that male S. carinatus are aggressive towards S. peltifer, reinforcing prezygotic barriers between the two species.

The results of this study closely parallel long-standing theories in ecology and evolutionary biology about the interactions and co-occurrence of closely related species

(Gause 1934; Diamond 1975). Lotic Sternotherus provide a unique example within turtles to study the effects of closely-related species competing for resources. Other species, like those in the genus Graptemys, have evolved to fill divergent niche space where congenerics co-occur (Lindeman 2008). When in sympatry, one Graptemys species will typically exhibit megacephaly for a molluscivorous diet, while the other

species is microcephalic for a diet of freshwater sponges or invertebrates (Lindeman

2008; Lindeman 2013; Lindeman 2016; Selman and Lindeman 2018). However, in this study, Sternotherus carinatus and S. peltifer differ from Graptemys by differentiating more in habitat. This pattern is not unusual in lotic taxa. There are analogous examples within fishes, most similarly in the Fundulus notatus species complex. Though

Sternotherus carinatus and S. peltifer are very similar in ecology in allopatry, in sympatry the two species differentiate from one another in habitat, size and aggression. It is likely that some combination of these factors as well as their phylogenetic distance preclude rampant hybridization between these species as has been seen in other syntopic

Sternotherus species (e.g., S. peltifer and S. depressus; Scott et al. 2019).

Acknowledgements

This study benefitted from genetic samples donated by Luke Pearson and Gabbie

Berry. All turtles were caught and handled under appropriate state permits (Alabama:

#9029; Mississippi: MMNS #’s: 0607172, 0530181, 0517191) and IACUC numbers:

11092206 and 17101202 from the University of Southern Mississippi. This work has been funded by the following grants (in alphabetical order): the American Turtle

Observatory, Birmingham Audubon Society, Chicago Herpetological Society, NSF

Graduate Research Fellowship (#1842492), and the Theodore Roosevelt Memorial Grant through the American Museum of Natural History.

93 Table 4.1 Pascagoula River drainage sampling sites in 2018, and the number of turtles per species captured at each site.

Site Date Sc Sp So Tse Mt Pc As Am Ks # TTN UDA

Myers Creek 4/28/18 0 4 0 0 0 0 0 0 0 30 33.4

Garraway Creek 4/28/18 0 3 0 0 0 0 0 0 0 30 17.9

Big Creek (Hattiesburg) 5/3/18 0 4 0 2 0 0 0 0 0 29 57.0

Bouie River (Private Access) 5/3/18 0 5 0 0 1 0 0 0 0 30 770.9

Red Creek (Wiggins) 5/9/18 0 0 0 0 0 0 0 0 0 30 256.3

Double Branch Creek 5/9/18 0 0 0 1 1 0 0 0 0 29 84.4

Okatoma Creek (Seminary) 5/12/18 1 2 0 0 0 0 0 0 0 29 527.9

Bouie River (Hwy 49) 5/12/18 11 1 0 0 2 0 0 0 0 29 788.1

Big Creek (Soso) 5/15/18 0 1 0 0 0 0 0 0 0 22 245.0

Oakey Woods Creek 5/15/18 0 3 0 0 0 0 0 0 0 28 144.4

Leaf River (Beaumont) 6/5/18 4 0 1 0 1 0 0 0 0 28 7815.6

Leaf River (Hattiesburg) 6/5/18 11 0 0 1 0 0 2 0 0 28 4698.2

Bouie River (Peps Point) 6/8/18 0 1 0 0 0 0 2 0 0 30 1563.5 Table 4.1 (continued)

Leaf River (Chain Park) 6/8/18 5 0 0 7 0 1 0 0 0 27 2795.6

Mill Creek 6/15/18 0 0 0 0 0 0 0 0 0 20 48.0

Flint Creek 6/15/18 0 0 0 0 0 0 0 0 0 16 82.9

Okatoma Creek (Lux) 6/19/18 2 1 0 0 0 0 0 0 0 29 697.3

Bouie River (Seminary) 6/19/18 2 1 0 0 0 0 0 0 0 26 565.6

Blacktom Creek 6/20/18 0 1 0 1 0 0 0 0 0 10 30.1

Pascagoula River (Wilkerson

Ferry) 6/22/18 0 0 0 1 1 0 1 0 0 28 17363.7

Pascagoula River (Benndale) 6/22/18 1 0 1 1 1 0 0 1 0 28 17162.3

Thompson Creek 6/26/18 1 1 0 4 2 0 0 0 0 29 550.0

Gaines Creek 6/26/18 2 0 0 0 0 0 0 0 0 26 312.2

Bogue Homa Creek (Richton) 6/29/18 3 0 0 4 2 0 1 0 0 27 922.4

Bogue Homa Creek (Ovett) 6/29/18 4 0 0 1 0 0 1 0 0 21 695.3

Black Creek (Hwy 11) 7/3/18 2 3 0 0 0 0 0 0 0 30 445.4 Table 4.1 (continued)

Mixon Creek 7/3/18 0 3 0 0 0 0 0 0 0 30 35.8

Terrible Creek 7/6/18 0 0 1 0 0 0 0 0 0 29 65.9

Cooks Branch 7/6/18 0 0 0 6 0 0 0 0 0 28 12.6

Chunky River (Hwy 80) 7/10/18 0 8 0 3 0 0 0 0 0 29 959.1

Allen Creek 7/10/18 0 0 0 0 0 0 0 0 3 29 20.2

Possum Creek 7/10/18 0 3 0 0 0 0 0 0 0 27 32.8

Black Creek (Fairley) 7/13/18 4 4 0 0 1 0 1 0 0 23 1740.8

Hickory Creek (Benndale) 7/13/18 0 0 0 2 1 0 0 0 0 30 60.6

Black Creek (Janice) 7/13/18 0 0 0 0 3 0 0 0 0 18 1257.4

Little Creek 7/24/18 0 7 0 3 0 0 0 0 0 30 35.4

Big Creek (Clara) 7/24/18 0 0 0 0 0 0 0 0 0 28 168.9

Buckatunna Creek (Waynesboro) 7/26/18 0 13 0 1 0 0 0 0 0 27 1319.0

Table 4.1 (continued)

Chickasawhay River

(Waynesboro) 7/26/18 0 17 0 2 0 0 0 0 0 28 4278.0

Mixons Creek 7/28/18 0 1 0 3 0 0 1 0 0 30 30.7

Tallahala Creek 7/29/18 3 0 0 0 0 0 0 0 0 20 1232.0

Chickasawhay River (Stonewall) 8/27/18 3 5 1 0 4 0 0 0 0 27 2392.7

Okatibbee Creek 8/27/18 0 3 0 0 1 0 0 0 0 30 977.6

Scotchenflipper Creek 8/27/18 0 1 0 0 0 0 0 0 0 28 40.8

Oakahay Creek 8/30/18 1 0 0 0 2 0 0 0 0 30 633.1

Clear Creek 8/30/18 0 1 0 0 0 0 0 0 0 28 34.3

Pascagoula River (Vancleave) 9/25/18 0 0 0 0 1 0 0 0 0 30 21195.1

Red Creek (Vestry) 9/25/18 9 0 0 0 1 0 0 0 0 23 1152.2

Black Creek (Hwy 57) 9/25/18 3 0 0 0 2 0 0 0 0 24 1952.2

Big Cedar Creek 9/26/18 0 1 0 0 0 0 0 0 0 20 68.6 Table 4.1 (continued)

Escatawpa River (Hurley) 9/27/18 13 0 1 0 1 0 0 0 0 30 1694.9

Chickasawhay River (Leakesville) 10/18/19 2 0 0 5 1 0 1 0 0 27 6986.4

Big Creek (Leakesville) 10/19/19 0 0 0 0 2 0 0 0 0 21 342.8

Brushy Creek 10/19/19 0 0 2 3 0 1 0 0 0 30 133.1

Buckatunna Creek (Frost Chapel) 10/30/18 1 1 0 1 0 1 0 0 0 24 1026.2

Cedar Creek (Quitman) 10/30/18 0 1 0 7 0 0 0 0 0 29 45.8

Shabuta Creek 10/30/18 0 1 0 0 0 0 0 0 0 20 88.4

Totals 88 101 7 59 31 3 10 1 3 1516

Species names are abbreviated: Sc = Sternotherus carinatus, Sp = S. peltifer, So = S. odoratus, Tse = Trachemys scripta elegans, Mt = Macrochelys temminckii, Pc = Pseudemys concinna, As = Apalone spinifera, Am = Apalone mutica, Ks = Kinosternon subrubrum. # TTN is the number of trap nights per site, and UDA is the upstream drainage area in km2 of each site

Table 4.2 List of all 64 sites where Sternotherus carinatus and S. peltifer were detected across all years in this study in the Pascagoula River drainage.

Site # Site Name Latitude Longitude

1 Oakahay Creek 31.74 -89.44

2 Leaf River @ Taylorsville, MS 31.97 -89.42

3 Leaf @ Eastabuchie, MS 31.44 -89.30

4 Black Creek @ Private Access 31.11 -89.29

5 Leaf @ Chain Park 31.35 -89.28

6 Leaf @ Sims Rd 31.26 -89.23

7 Tallahala Creek @ Ellisville 31.59 -89.17

8 Bogue Homa @ Ovett 31.48 -89.05

9 Bogue Homa @ Richton 31.36 -89.03

10 Leaf @ Beaumont, MS 31.18 -88.92

11 Red Creek @ Hwy 15 30.77 -88.91

12 Gaines Creek 31.25 -88.86

13 Red Creek @ Vestry 30.74 -88.78

14 Black Creek @ Hwy 57 30.78 -88.76

15 Pascagoula @ Benndale 30.90 -88.75

16 Pascagoula River @ Merrill 30.98 -88.73

17 Poticaw Bayou on Pascagoula River 30.51 -88.62

18 Chickasawhay River @ Leakesvillle, MS 31.15 -88.55

19 Escatawpa @ Presley Landing 30.48 -88.44

Table 4.2 (continued)

20 Escatawpa @ Hurley, MS 30.63 -88.43

21 Escatawpa @ Wilmer 30.86 -88.41

22 Escatawpa River @ Cintronelle, AL 31.09 -88.38

23 Clear Creek 31.87 -89.56

24 Mixon Creek @ Purvis 31.21 -89.41

25 Big Creek @ Hattiesburg 31.40 -89.41

26 Sandy Run Creek 31.21 -89.39

27 Bouie @ Peps Point 31.40 -89.37

28 Oakey Woods Creek 31.62 -89.36

29 Blacktom Creek @ Purvis 31.18 -89.36

30 Mixons Creek @ Hattiesburg 31.35 -89.35

31 Snows Creek 31.56 -89.33

32 Big Creek @ Soso, MS 31.72 -89.31

33 Double Branch Creek 30.93 -89.28

34 Myers Creek 31.25 -89.24

35 Garraway Creek 31.21 -89.13

36 Scotchenflipper Creek 32.22 -89.03

37 Thompson Creek 31.36 -88.93

38 Chunky River @ Hwy 80 32.33 -88.91

39 Possum Creek 32.27 -88.90

40 Sweetwater Creek 30.85 -88.85

100

Table 4.2 (continued)

41 Shabuta Creek 31.94 -88.84

42 Atkinson Creek 31.14 -88.80

43 Sowashee Creek 32.34 -88.73

44 Chickasawhay @ Waynesboro 31.68 -88.68

45 Little Creek @ Waynesboro 31.47 -88.65

46 Big Cedar Creek @ Agricola, MS 30.76 -88.57

47 Buckatunna @ Waynesboro 31.65 -88.52

48 Cedar Creek @ De Soto 31.96 -88.51

49 Red Creek @ Millry, AL 31.58 -88.45

50 Bouie @ Seminary-Sumrall 31.50 -89.56

51 Okatoma Creek @ Seminary 31.56 -89.50

52 Bouie River @ Private Access 31.45 -89.44

53 Bouie @ Hwy 49 31.43 -89.42

54 Okatoma Creek @ Lux 31.45 -89.41

55 Black Creek @ Hwy 11 31.19 -89.38

56 Black Creek @ Fairley 30.92 -88.97

57 Chunky @ Chunky, MS 32.32 -88.93

58 Thompson @ Hintonville 31.26 -88.90

59 Chunky @ Dunns Falls 32.23 -88.82

60 Chickasawhay River @ Stonewall 32.13 -88.81

61 Okatibbee Creek @ Enterprise, MS 32.20 -88.79

101

Table 4.2 (continued)

62 Pascagoula River @ Wilkerson Ferry 30.81 -88.74

63 Chickasawhay @ De Soto 31.98 -88.70

64 Buckatunna Creek @ Frost Church 31.80 -88.49

65 Terrible Creek 31.59 -89.63

66 Cooks Branch 31.51 -89.56

67 Double Branch Creek 31.25 -89.24

68 Red Creek @ Wiggins, MS 31.25 -89.24

69 Mill Creek @ Wiggins 30.77 -89.14

70 Flint Creek 30.81 -89.07

71 Black Creek @ Janice 30.99 -89.05

72 Hickory Creek @ Benndale 30.97 -88.97

73 Allen Creek 32.28 -88.85

74 Pascagoula River down from Benndale 30.83 -88.74

75 Big Creek @ Clara, MS 31.55 -88.68

76 Pascagoula River @ Wade Vancleave 30.61 -88.64

77 Big Creek @ Old 24 31.17 -88.63

78 Brushy Creek 30.94 -88.45

Sites 65-78 were also sampled but no Sternotherus were detected. All site numbers correspond to maps in Figure 4.2.

102 Table 4.3 Average ancestry coefficients (q-scores) of 211 Sternotherus carinatus and 219 S. peltifer from the Pascagoula River drainage.

Species Average carinatus ancestry Average peltifer ancestry SD

Sternotherus carinatus 0.992 0.008 0.010

Sternotherus peltifer 0.007 0.993 0.007

S. peltifer of Mixed Ancestry Location

Sp85 0.196 0.804 Millry, Alabama

Sp93 0.328 0.672 Wiggins, MS

Sp136 0.445 0.555 Beaumont, MS

The three S. peltifer of mixed ancestry were not included in the averages for the species, instead they are reported separately. SD = standard deviation.

Legend: — S. carinatus — S. peltifer — S. intermedius — S. depressus — S. minor

Figure 4.1 Range map of the lotic Sternotherus species in the Southeastern United States.

The area of sympatry between Sternotherus carinatus and Sternotherus peltifer is cross-hatched.

Figure 4.2 Capture locations for Sternotherus within the Pascagoula River system

a) Sternotherus carinatus , b) Sternotherus peltifer, c) syntopy, d) sites where no Sternotherus were detected. Site numbers correspond to those found in Table 4.2.

105

Figure 4.3 A CCA plot of all turtle species encountered at >10% of all sites.

Temp = Water temperature, Avg.Depth = Average stream depth, Canopy = average percent canopy cover and Avg.Substrate = average substrate score. Species are in red, Sternotherus carinatus, S. peltifer and Apalone spinifera are represented by their specific epithets. Trachemys scripta elegans and Macrochelys temminckii are represented by their genera.

106

Figure 4.4 STRUCTURE plot for 211 Sternotherus carinatus and 222 Sternotherus peltifer from the Pascagoula River drainage.

The bars in this bar plot represent the percent ancestry from each species for each turtle (1 bar = 1 turtle). Only three S. peltifer demonstrated signs of mixed ancestry.

4.5 Literature Cited

Alcalde, L., Derocco, N. N., and Rosset, S. D. 2010. Feeding in syntopy: diet of

Hydromedusa tectifera and Phrynops hilarii (Chelidae). Chelonian Conservation

and Biology 9:33–

Beatty, G. E., Philipp, M., and Provan, J. 2010. Unidirectional hybridization at a species’

range boundary: Implications for habitat tracking. Diversity and Distributions

16:1–9.

Berger, L. 1973. Systematics and hybridization in European green frogs of Rana

esculenta complex. Journal of Herpetology 7:1–10.

Berry, J. F. 1975. The Population effects of ecological sympatry on musk turtles in

Northern Florida. Copeia 1975:692–701.

Boyer, D. R. 1965. Ecology of the basking habit in turtles. Ecology 46:99–118.

Brown, G. J., and Kreiser, B. R. 2020. Characterization of microsatellite loci for the

razorback musk turtle (Sternotherus carinatus) and their cross-amplification in

five other species in the family Kinosternidae. Chelonian Conservation and

Biology. In Press.

Brown, W. L., and Wilson, E. O. 1956. Character Displacement. Systematic Zoology

5:49–64.

Chunco, A. J., Jobe, T., and Pfennig, K. S. 2012. Why do species co-occur? A test of

alternative hypotheses describing abiotic differences in sympatry versus allopatry

using spadefoot toads. PloS One 7:e32748.

108

Cummins, K. W. 1962. An evaluation of some techniques for the collection and analysis

of benthic samples with special emphasis on lotic waters. The American Midland

Naturalist 67:477–504.

Darwin, C., Duthie, J. F., and Hopkins, W. 1859. On the origin of species by means of

natural selection: Or, The preservation of favoured races in the struggle for life.

London: John Murray, Albemarle Street.

Diamond, J. M. 1975. Assembly of species communities. Pages 342–444 in Diamond J.

M., Cody M. L., eds. Ecology and Evolution of Communities.

Duvernell, D. D., Meier, S. L., Schaefer, J. F., and Kreiser, B. R. 2013. Contrasting

phylogeographic histories between broadly sympatric topminnows in the

Fundulus notatus species complex. Molecular Phylogenetics and Evolution

69:653–663.

Ernst, C. H., and Lovich, J. E. 2009. Turtles of the United States and Canada. JHU Press.

Fuselier, L., and Edds, D. 1994. Habitat partitioning among three sympatric species of

map turtles, genus Graptemys. Journal of Herpetology 28:154–158.

Gause, G. F. 1934. Experimental analysis of Vito Volterra’s mathematical theory of the

struggle for existence. Science 79:16–17.

Gilbert, O., Reynoldson, T. B., and Hobart, J. 1952. Gause’s hypothesis: An examination.

Journal of Animal Ecology 21:310–312.

Godwin, J. C., Lovich, J. E., Ennen, J. R., Kreiser, B. R., Folt, B., and Lechowicz, C.

2014. Hybridization of two megacephalic map turtles (Testudines: Emydidae:

Graptemys) in the Choctawhatchee River drainage of Alabama and Florida.

Copeia 2014:725–742.

109

Gotelli, N. J., and McCabe, D. J. 2002. Species co-occurrence: A meta-analysis of J. M.

Diamond’s Assembly Rules Model. Ecology 83:2091–2096.

Hair, J. F., Black, W. C., Babin, B. J., and Anderson, R. E. 2010. Multivariate data

analysis (8th edition edition). Cengage Learning EMEA.

Harper, J. L., Clatworthy, J. N., McNaughton, I. H., and Sagar, G. R. 1961. The evolution

and ecology of closely related species living in the same area. Evolution 15:209–

227.

Hilbish, T. J., Lima, F. P., Brannock, P. M., Fly, E. K., Rognstad, R. L., and Wethey, D.

S. 2012. Change and stasis in marine hybrid zones in response to climate

warming. Journal of Biogeography 39:676–687.

Iverson, J. B. 1977. Geographic Variation in the Musk Turtle, Sternotherus minor.

Copeia 1977:502–517.

Iverson, J. B., Le, M., and Ingram, C. 2013. Molecular phylogenetics of the mud and

musk turtle family Kinosternidae. Molecular Phylogenetics and Evolution

69:929–939.

Jakobsson, M., and Rosenberg, N. A. 2007. CLUMPP: A cluster matching and

permutation program for dealing with label switching and multimodality in

analysis of population structure. Bioinformatics 23:1801–1806.

Jones, R. L., and Hartfield, P. D. 1995. Population size and growth in the turtle

Graptemys oculifera. Journal of Herpetology 29:426–436.

Kamler, J. F., Pennock, D. S., Welch, C., and Pierotti, R. J. 1998. Variation in

morphological characteristics of the white-footed mouse (Peromyscus leucopus)

110

and the deer mouse (P. maniculatus) under allotopic and syntopic conditions. The

American Midland Naturalist 140:170–179.

Kavanagh, B. T., and Kwiatkowski, M. A. 2016. Sexual dimorphism, movement patterns,

and diets of Sternotherus carinatus (razorback musk turtle). Southeastern

Naturalist 15:117–133.

Legendre, P., and Legendre, L. F. J. 2012. Numerical Ecology. Elsevier.

Legendre, Pierre, and Cáceres, M. D. 2013. Beta diversity as the variance of community

data: Dissimilarity coefficients and partitioning. Ecology Letters 16:951–963.

Li, Y.-L., and Liu, J.-X. 2018. StructureSelector: A web-based software to select and

visualize the optimal number of clusters using multiple methods. Molecular

Ecology Resources 18:176–177.

Lindeman, P. V. 1999. Surveys of basking map turtles Graptemys spp. in three river

drainages and the importance of deadwood abundance. Biological Conservation

88:33–42.

Lindeman, P. V. 2000. Evolution of the relative width of the head and alveolar surfaces

in map turtles (Testudines: Emydidae: Graptemys). Biological Journal of the

Linnean Society 69:549–576.

Lindeman, P. V. 2008. Evolution of body size in the map turtles and sawbacks

(Emydidae: Deirochelyinae: Graptemys). Herpetologica 64:32–46.

Lindeman, P. V. 2014. New wine in old bottles: using modified hoopnets to catch bait-

averse basking turtles. Herpetological Review 45:597-600.

Lindeman, P. V. 2013. The Map Turtle and Sawback Atlas: Ecology, Evolution,

Distribution, and Conservation. University of Oklahoma Press.

111

Lindeman, P. V. 2015. Surveys of basking turtles in the rivers of northeastern Oklahoma,

with emphasis on Graptemys geographica (common map turtle). Proceedings of

the Oklahoma Academy of Science 94:1-9.

Lindeman, P. V. 2016. Diets of syntopic black-knobbed Sawbacks (Graptemys

nigrinoda) and Alabama map turtles (Graptemys pulchra) in the Alabama River.

The American Midland Naturalist 175:194–205.

MacArthur, R. H. 1958. Population ecology of some warblers of Northeastern coniferous

forests. Ecology 39:599–619.

McGowan, C., and Davidson, W. S. 1992. Unidirectional natural hybridization between

brown trout (Salmo trutta) and Atlantic salmon (S. salar) in Newfoundland.

Canadian Journal of Fisheries and Aquatic Sciences 49:1953–1958.

Meathrel, C. E., Suter, P. J., and Radford, N. M. 2002. Niche segregation between three

species of freshwater turtle in a large billabong during flood. Victorian

Naturalist 119:160-173.

Parham, J. F., Papenfuss, T. J., Dijk, P. P. van, Wilson, B. S., Marte, C., Schettino, L. R.,

and Brian Simison, W. 2013. Genetic introgression and hybridization in Antillean

freshwater turtles (Trachemys) revealed by coalescent analyses of mitochondrial

and cloned nuclear markers. Molecular Phylogenetics and Evolution 67:176–187.

Pike, D. A. 2013. Climate influences the global distribution of sea turtle nesting. Global

Ecology and Biogeography 22:555–566.

Raymond, M., and Rousset, F. 1995. GENEPOP (Version 1.2): Population genetics

software for exact tests and ecumenicism. Journal of Heredity 86:248–249.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223–225.

112

Rivas, L. R. 1964. A reinterpretation or the concepts “Sympatric” and “Allopatric” with

proposal or the additional terms “Syntopic” and “Allotopic”. Systematic

Zoology 13:42-43.

Rosenberg, N. A., Mahajan, S., Ramachandran, S., Zhao, C., Pritchard, J. K., and

Feldman, M. W. 2005. Clines, clusters, and the effect of study design on the

inference of human population structure. PLoS Genetics 1:0660-0671.

Rousset, F. 2008. genepop’007: A complete re-implementation of the genepop software

for Windows and Linux. Molecular Ecology Resources 8:103–106.

Schaefer, J., Kreiser, B. K., Champagne, C., Mickle, P. M., and Duvernell, D. D. 2009.

Patterns of co-existence and hybridization among two topminnows (Fundulus

euryzonus and F. olivaceus) in a riverine contact zone. Ecology Fresh Fish

18:360-368.

Schaefer, J. F., Duvernell, D. D., and Kreiser, B. R. 2011. Ecological and genetic

assessment of spatial structure among replicate contact zones between two

topminnow species. Evolutionary Ecology 25:1145–1161.

Scott, P. A., Glenn, T. C., and Rissler, L. J. 2019. Formation of a recent hybrid zone

offers insight to the geographic puzzle and maintenance of species boundaries in

musk turtles. Molecular Ecology 28:761-771.

Scott, P. A., Glenn, T. C., and Rissler, L. J. 2018. Resolving taxonomic turbulence and

uncovering cryptic diversity in the musk turtles (Sternotherus) using robust

demographic modeling. Molecular Phylogenetics and Evolution 120:1–15.

113

Scott, P. A., and Rissler, L. J. 2015. Integrating dynamic occupancy modeling and

genetics to infer the status of the imperiled flattened musk turtle. Biological

Conservation 192:294–303.

Seidel, M. E., and Lucchino, R. V. 1981. Allozymic and morphological variation among

the musk turtles Sternotherus carinatus, S. depressus and S. minor

(Kinosternidae). Copeia 1981:119–128.

Selman, W., and Lindeman, P. V. 2018. Spatial, seasonal, and sexual variation in the diet

of Graptemys flavimaculata, a threatened turtle of the Pascagoula River system,

Mississippi, USA. Copeia 106:247–254.

Selman, W., and Qualls, C. P. 2011. Basking ecology of the yellow-blotched sawback

(Graptemys flavimaculata), an imperiled turtle species of the Pascagoula River

system, Mississippi, United States. Chelonian Conservation and Biology 10:188–

197.

Shaffer, H. B., Meylan, P., and McKnight, M. L. 1997. Tests of turtle phylogeny:

molecular, morphological, and paleontological approaches. Systematic

Biology 46:235-268.

Smith, H. M., and Glass, B. P. 1947. A new musk turtle from Southeastern United States.

Journal of the Washington Academy of Sciences 37:22–24.

Ter Braak, C. J. F. 1987. The analysis of vegetation-environment relationships by

canonical correspondence analysis. Vegetation 69:69–77.

Tinkle, D. W. (1958). The systematics and ecology of the Sternotherus carinatus

complex (Testudinata: Chelydridae). Tulane Studies of Zoology 6:1-56.

114

Todesco, M., Pascual, M. A., Owens, G. L., Ostevik, K. L., Moyers, B. T., Hübner, S., ...

and Rieseberg, L. H. 2016. Hybridization and extinction. Evolutionary

Applications 9:892-908.

Vogt, Richard C. 1980. New methods for trapping aquatic turtles. Copeia 1980:368–371.

Vogt, Richard C. 1981. Food partitioning in three sympatric species of map turtle, Genus

Graptemys (Testudinata, Emydidae). The American Midland Naturalist 105:102–

111.

Vogt, R. C., McCoy, C. J., and Bianculi, A. V. 1978. Sternotherus minor (Reptilia,

Kinosternidae) in the Pascagoula River system, Mississippi. Herpetological

Review 9:109.

Walker, D., Ortí, G., and Avise, J. C. 1998. Phylogenetic distinctiveness of a threatened

aquatic turtle (Sternotherus depressus). Conservation Biology 12:639–645.

Walls, S. C. 2009. The role of climate in the dynamics of a hybrid zone in Appalachian

salamanders. Global Change Biology 15:1903–1910.

Welsh, M. A., Doody, J. S., and Georges, A. 2017. Resource partitioning among five

sympatric species of freshwater turtles from the wet–dry tropics of northern

Australia. Wildlife Research 443:219–229.

Wirtz, P. 1999. Mother species–father species: Unidirectional hybridization in animals

with female choice. Animal Behaviour 58:1–12.

115

CHAPTER V – THE RIVERSCAPE GENETICS OF TWO LOTIC MUSK TURTLES

IN THE PASCAGOULA RIVER DRAINAGE

Abstract

Landscape genetics is a young but growing field at the intersection of landscape ecology and population genetics. Early on, studies focused on terrestrial systems and resistance points for gene flow, but the field quickly expanded to include the study of riverscape genetics I compared and contrasted the intra-drainage population genetics of the razorback musk turtle (Sternotherus carinatus) and the stripe-necked musk turtle

(Sternotherus peltifer) where the two species overlap in geographic distribution in the

Pascagoula River system. In allopatry, both species occur across the river continuum, but in sympatry, the two species have different habitat associations, where S. carinatus occupies large streams and rivers, and S. peltifer occurs predominantly in small or medium sized streams. I used microsatellite loci to evaluate the population genetics of each species within the Pascagoula River system. There is population structure in both species with higher intra-drainage population structure detected in Sternotherus peltifer, and a pattern of isolation by distance in Sternotherus carinatus. These patterns of population structure correspond with the ecology of each species, with Sternotherus peltifer exhibiting more isolation due to its affinity for smaller stream habitats that are isolated by stretches of larger river, whereas Sternotherus carinatus has a more contiguous distribution throughout the drainage and shows clinal variation in population genetics. These data provide important baseline data on the population genetics of two freshwater turtle species in an unimpounded river system.

116

5.1 Introduction

Landscape genetics is a young but growing field at the intersection of landscape ecology and population genetics that investigates how the geographical and environmental features of a landscape influence or structure the genetic variation at the population and individual levels (Manel et al. 2003). Landscape genetics differs from other fields of genetics, like phylogeography, by focusing on a finer geospatial scale.

Landscape genetics was conceived for the analysis of terrestrial landscapes but the concept has been expanded to include unique landscapes such as the sea and sea floor

(seascape genetics; Selkoe et al. 2008; Storfer et al. 2010) and the linear, dendritic networks of rivers (riverscape genetics; Kanno et al. 2011; Davis et al. 2018). Landscape genetics is a versatile field, and while many studies have examined how the natural features of the landscape can shape population structure, the same tools can also be applied to understanding the how anthropogenic barriers and threats can impact genetic structure (e.g., climate change – Hoffmann and Willi 2008; Norberg et al. 2012; Pauls et al. 2013; Manel and Holderegger 2013). The study of landscape genetics, therefore, has important applications and implications in the fields of evolution, ecology and conservation biology.

Most landscape genetics studies seek to test whether landscape/seascape/riverscape features affect gene flow. Examples of landscape genetic studies include some that have looked at the effect of roads (Riley et al. 2006), rivers

(Antolin et al. 2006) and mountain chains (Funk et al. 2005) on gene flow, as well as more gradual effects on gene flow like isolation by distance (Wright 1943; Rousset 1997;

Geffen et al. 2004; Kanno 2011). Impediments to gene flow are not only limited to

117

physical barriers, but they can also manifest through microhabitats associations of species that are not contiguous throughout the landscape or aquatic system like moisture levels, temperature, and water quality (Geffen et al. 2004; Storfer et al. 2007; Earnest 2014).

As previously mentioned, riverscape genetics is a subfield of landscape genetics.

Davis et al (2018) define a “riverscape” as the hierarchically structured mosaic of differentially distributed habitat within freshwater environments. Riverscapes are contiguous and have permeable borders, but they have discontinuities that form along the river continuum that are functionally important for many aquatic organisms (Vannote et al. 1980; Benda et al. 2004; Carbonneau et al. 2014; Davis et al. 2018). Processes that affect one river patch can affect a neighboring river patch, creating complex interactions within and among populations of aquatic organisms across a range of spatial scales.

Because highly or fully-aquatic species can only move in linear directions within the riverscape, estimating rates of gene flow and isolation by distance are not calculated in a

Euclidean manner (e.g., “as the crow flies”), but they must be measured along river and stream corridors “as the fish [or turtle] swims” according to dendritic ecological networks

(Fagan 2002; Elliot et al. 2003; Kanno 2011; Altermatt et al. 2013). For examples, despite how headwater stream may interdigitate and otherwise be geographically very proximate, the degree of isolation is likely magnitudes higher than the Euclidean distance

(straight-line distance) between those sites. Ultimately, though, it will be the vagility of a species that affects gene flow within a lotic system.

Aquatic biologists have used riverscape genetics to investigate the genetic structure of fishes (Kanno et al. 2011; Earnest et al. 2014; Bowlby et al. 2016; Brauer et al. 2016), mollusks (Galbraith et al. 2015; Inoue et al. 2014, Whelan et al. 2019), and

118

parasites (Sprehn et al. 2015). There have been riverscape genetics studies of turtles, though the authors may not have used the term “riverscape genetics” explicitly. Bennett et al. (2010) studied the effects of dams on gene flow in Graptemys geographica; Selman et al. (2013) looked at the population genetics of the riverine Graptemys flavimaculata in the Pascagoula River drainage and how paleodrainages may shape genetic structure; and

Dos Santos et al. (2016) investigated the population genetics of Podocnemis erythrocephala in South America and determined waterfalls, rapids, and the Amazon

River reduced gene flow across the river continuum. Reid et al. (2016) studied the landscape genetics of three sympatric turtle species (Chelydra serpentina, Chrysemys picta, and Emydoidea blandingii) in Wisconsin, USA. These three turtle species typically inhabit lentic environments, and the authors found that the more highly aquatic Chelydra and Chrysemys species exhibited a trend of isolation by distance, versus the more semi- terrestrial Emydoidea which did not exhibit a trend of isolation by distance. Reid et al.

(2016) stated the utility of expanding the scope of landscape genetics beyond a single species to help infer how species responses to environmental variables are expressed differentially. Along similar lines, expanding the focus of riverscape genetics beyond a single species to two ecologically similar, congeneric species will also provide unique insight on the effects of habitat associations and species interactions on genetic structure.

The lotic musk turtles (genus Sternotherus) are common inhabitants of rivers and streams across the southeast. Currently, there are 5 recognized species of lotic musk turtle that replace each other more or less in geographic sequence from the Brazos river drainage in Central Texas (Sternotherus carinatus) to the Altamaha river drainage in

Georgia (S. minor; Ernst and Lovich 2009; Scott et al. 2018). These musk turtles share

119

very similar ecologies and fill the niche of a small bottom walking, durophagous turtle species that can be found in a variety of habitats from small montane streams to larger coastal plains rivers and associated floodplains (Ernst and Lovich 2009). While most of these species occur in allopatry of one another, this artenkreis of musk turtles is disrupted in south Mississippi where there is overlap in the ranges of Sternotherus carinatus and S. peltifer. In sympatry, these two species tend to be found in different habitats with

Sternotherus carinatus occupying primarily the larger streams and rivers, and S. peltifer typifying small stream habitat (headwater and adventitious streams) or small headwater rivers (Brown et al. in prep).

Given the unique circumstance of the overlap of these two species, I wanted to test whether interspecific habitat associations affected the genetic structure of these species within the Pascagoula River drainage by comparing and contrasting the intra- drainage population structure of each species. The Pascagoula River system is the largest unimpounded system left in the contiguous United States (Dynesius and Nilsson 1994), and it affords a unique opportunity to look at gene flow from an intrinsic, natural perspective as there are no impoundments to impede gene flow. Since river systems are dendritic, smaller headwater streams and adventitious streams are isolated by stretches of larger river habitat. Because Sternotherus peltifer is associated with these habitats in the

Pascagoula watershed, I predicted this species would demonstrate higher levels of population structure and more signatures of genetic isolation across the Pascagoula River drainage. I predicted that the larger river species, Sternotherus carinatus, would have higher levels of gene flow due to contiguity of habitat and therefore demonstrate less genetic structure across the drainage.

120

5.2 Methods

5.2.1 Study Area and Sampling Methods

I collected lotic Sternotherus carinatus and S. peltifer tissue from 2015-2019 from the major tributaries of the Pascagoula river system: Red Creek, Black Creek, Leaf River,

Chickasawhay River and the mainstem of the Pascagoula River (Fig. 5.1). During trapping, turtles were caught by setting hoop nets, minnow traps (ProMar ©) and crab traps baited with either fresh fish or sardines in soybean oil (Bumble Bee Foods ©).

Turtles were also caught opportunistically via wading stream and rivers, and this was typically done at dawn or dusk when these species are most active (Ernst and Lovich

2009). Upon capture, turtles were measured, marked uniquely (to prevent resampling), and a 10mm sample of interdigital tissue was collected from the back foot. Tissue was stored in 100% ethanol and returned to the lab to be stored at -20°C until extracted.

I was able to catch and collect tissue from turtles in every subdrainage within the

Pascagoula River system. In order to draw more meaningful inferences with more power,

I grouped these samples into seven subdrainages and river stretches for Sternotherus peltifer and eight subdrainages and river stretches for S. carinatus (Tables 5.1 and 5.2). I was able to collect Sternotherus peltifer from most of the major subdrainages of the

Pascagoula River drainage, but no S. peltifer were found in the Escatawpa drainage. Sites from where Sternotherus tissue was collected can be found in Table 5.3 and seen on maps in Figure 5.2.

5.2.2 Molecular Methods

Total genomic DNA was extracted from tissue samples with a DNeasy Tissue Kit

(QIAGEN Inc., Valencia, CA). I genotyped Sternotherus carinatus for 8 microsatellite

121

loci (Scar06, Scar08, Scar26, Scar27, Scar29, Scar35, Scar39, Scar40) and stripe-necked musk turtles at 10 microsatellite loci (Scar06, Scar08, Scar14, Scar26, Scar27, Scar29,

Scar35, Scar38, Scar39, Scar54). Polymerase chain reaction conditions and thermocycler profiles followed Brown and Kreiser (in press). Microsatellite alleles were visualized on a polyacrylamide gel using a LICOR 4300 DNA Analyzer and scored using GeneProfiler ver. 4.05 (LICOR Co.). I tested for Hardy-Weinberg equilibrium and linkage disequilibrium using Genepop on the Web (Raymond and Rousset 1995; Rousset 2008).

A Bonferroni correction was applied to the alpha value to account for multiple comparisons (Rice 1989).

5.2.3 Data Analysis

I calculated genetic diversity measures for each site including the number of alleles, allelic richness (Ar), and heterozygosity (observed and expected) using FSTAT v.

2.9.3 and GenAlEx 6.502 (Goudet 1995; Peakall and Smouse 2006; Peakall and Smouse

2012). When calculating allelic richness, the number of individuals was rarefied to the smallest population size. To limit the impact of sites with low sample sizes on my estimates of Ar, I excluded sites with fewer than 10 individuals (this the Pascagoula population in S. peltifer). I used ANOVA’s to test for any differences in allelic richness and levels of heterozygosity (observed and expected) across populations from different subdrainages. Before each analysis, I checked the data for the assumptions of normality using a Shapiro-Wilks test and equal variances using a Bartlett’s test. When assumptions of normality or equal variances were violated, I used a nonparametric approach to compare groups by use of a Kruskal-Wallis test. All analyses were performed in the

“stats” package built into R v. 3.6.1. All alpha values were set at 0.05.

122

I calculated pairwise FST values (using θ - Weir and Cockerham’s (1984) unbiased estimator of FST) and significance tests in FSTAT v. 2.9.3 (Goudet 1995) to determine levels of population differentiation across turtles between each river subdrainage. I tested for isolation by distance by using the ‘mantel’ function in the “vegan” package in R. This function tests for correlations between two matrices, in this case, pairwise genetic distances (from FSTAT v. 2.9.3) and geographic distances. Pairwise geographic distances between sites were calculated using the “riverdistancemat” function in R-package

“riverdist.” This package uses a river network shapefile and a shapefile of site coordinates to calculate pairwise river distance between sites (Tyers 2017).

I used STRUCTURE v 2.3.4 (Pritchard et al. 2000) to determine the number of discrete populations (K). STRUCTURE uses a Bayesian approach to assign individuals to a predetermined number of groups given the populations are in Hardy-Weinberg and linkage equilibrium. I tested values of K from 1-10 using a model of admixed ancestry and assuming correlated allele frequencies between groups using population location as a prior (Hubisz et al., 2009). The program STRUCTURE is sensitive to uneven sampling across drainages (Wang 2017), therefore I used a smaller alpha value (1/~K, where K is the assumed number of populations – in this case 1/ [# of subdrainages]). For each value of K, I ran 20 replicates with 100,000 Markov chain Monte Carlo (MCMC) iterations and a burn-in set at 50,000. I selected the best value of K by examining the likelihood values for each K and by using the ΔK method (Evanno et al. 2005) as calculated by the online program StructureSelector (Li and Liu 2018). StructureSelector averaged the 20 runs for the best value of K using Clumpak (Kopelman et al. 2015), and I then visualized the average ancestry coefficient (q) for each individual with DISTRUCT v. 1.1 (Rosenberg

123

2004). I selected the best value of K as the smallest value that seemed to represent biologically meaningful genetic groups. For S. carinatus, I performed an additional

STRUCTURE analysis where I excluded the Escatawpa population (e.g., hierarchical analysis; Vähä et al. 2007) to make more direct comparisons between the two species, since S. peltifer was not detected in the Escatawpa.

Gene flow among subdrainages was estimated using BAYESASS, version 1.3

(Wilson and Rannala 2003). BAYESASS uses a Markov chain Monte Carlo method that does not require populations to be in Hardy-Weinberg Equilibrium or migration-drift equilibrium. Initial runs were run at default settings to determine the delta values for migration rate, inbreeding, and allele frequencies. I made modifications to the mixing parameter for allele frequencies (a) and missing parameters for inbreeding coefficients (f) to meet the recommended acceptance rates between 0.2 and 0.6 for those parameters

(Wilson and Rannala 2003). Using random seeds to check for local optima, I ran three independent runs for a total of 108 iterations with a burn in of 107. After discarding the initial burn-in, samples were collected after every 100 generations to infer the posterior probability distributions of migration rates among the genetic groups. I used Trace v. 1.5

(Rambaut and Drummond 2007) to ensure the analysis was run long enough for convergence, and that inferences may be drawn from migration rates. I considered migration rates significantly different if the 95% credible sets did not overlap across groups.

124

5.3 Results

5.3.1 Sites and Sampling

From 2015-2019, I collected 217 Sternotherus carinatus samples and 228 S. peltifer samples from 30 and 36 sites, respectively, within the Pascagoula River drainage

(Figure 5.2, Table 5.3). Sample sizes per location ranged from 6 to 44 for Sternotherus carinatus with a median of 28 turtles, and 4 to 61 for Sternotherus peltifer with a median of 14.5 turtles. Because sample sizes for many locations were low (e.g., the Pascagoula

River for S. carinatus, and Red Creek for S. peltifer), Sternotherus carinatus were grouped into 8 subdrainages or river stretches, and S. peltifer

5.3.2 Summary Statistics

None of the populations for either Sternotherus carinatus or S. peltifer deviated significantly from Hardy-Weinberg equilibrium, nor did either species exhibit linkage disequilibrium across loci after a Bonferroni correction (Rice 1989). Summary statistics for each species were similar. For Sternotherus carinatus allelic richness did not differ significantly across the grouped sites (F7,56 = 0.68, p = 0.69), nor did observed heterozygosity (F7,56 = 0.77, p = 0.61) or expected heterozygosity (F7,56 = 0.94, p = 0.48;

Table 5.4). Similarly in S. peltifer, allelic richness (for those populations with >10 turtles) did not differ significantly across grouped sites (F5,54 = 0.23, p = 0.95), and observed and expected levels of heterozygosity, likewise, did not differ significantly across sites (F6,63

= 0.271, p = 0.95 and χ2 = 1.95, p = 0.9244, d.f. = 6, respectively; Table 5.5). However, across species, Sternotherus carinatus had significantly higher allelic richness (W =

2705.5, p < 0.001), observed heterozygosity (W = 3133.5, p < 0.001), and expected heterozygosity (W = 3325.5, p < 0.001) than S. peltifer.

125

5.3.3 Population Structure in Sternotherus carinatus

Most pairwise FST values between Sternotherus carinatus subdrainages were small, but some comparisons differed significantly from zero (Table 5.6). Several values were less than 0.01 across sites, often in neighboring drainages (e.g., Upper Leaf-Middle

Leaf and Red Creek-Black Creek), and most of the highest FST values were associated with pairwise comparisons with the Escatawpa River drainage. Indeed, the highest average FST values came from the Escatawpa River, but those values were followed closely by headwater populations in the Upper Chickasawhay River and the Bouie River.

The Mantel test detected a significant correlation between river distance and genetic distance both when the Escatawpa River turtles were included in the analysis (p = 0.002, r = 0.700; Fig. 5.3), and when they were excluded in the analysis (p = 0.002, r = 0.710;

Fig. 5.4)

A ∆K analysis suggested that initially there were two groups that best represented the structure in the data. This corresponded to an Escatawpa group and the rest of the subdrainages (Fig. 5.5). The log-likelihood plot plateaued at K = 3 (Fig. 5.5). However, this structure was within the remaining subdrainages, which I explored hierarchically.

The ∆K analysis of the STRUCTURE results after omitting the Escatawpa suggested two groups best represented the genetic structure in the data (Fig. 5.6 and 5.7). At this K=2, the Middle Leaf, Upper Leaf and Bouie River turtles formed a cohesive group marked by high membership coefficients (q-scores), and a Red Creek, Black Creek, and Upper

Chickasawhay River group also marked by high q-scores, with the Lower Chickasawhay-

Pascagoula River exhibiting a mixture of these two groups (Table 5.7). Though it was not the most well-supported in the ∆K analysis or log-likelihood (lnK) values (Fig. 5.6), K =

126

3 also seems to reflect a biologically relevant group: a Red Creek-Black Creek group, a

Leaf-Bouie group and an Upper Chickasawhay group, which was also reflected in the FST values.

Estimated rates of contemporary migration from the BAYESASS analysis suggested that migration rates between sites were generally low with mostly overlapping confidence intervals, and that most turtles in each population were derived from the source site (Table 5.8). The Escatawpa and Bouie Rivers turtles had the highest estimated proportion of turtles derived from the source location (m = 0.92 and 0.856 respectively).

While most migration rates were very low, the highest migration rates came from the

Upper and Middle Leaf sites into the Bouie River system (m = 0.262 and 0.226 respectively).

5.3.4 Population Structure in Sternotherus peltifer

Most pairwise FST between subdrainages suggested moderate levels of population structure. All pairwise FST values (except those with the Pascagoula, which suffered from a small sample size [n = 6]) were significantly different from zero except for the Leaf

River and Bouie River populations (Table 5.9). The Mantel test did not find a significant correlation between river distance and genetic distance between sites (r = 0.2143, p =

0.149; Fig. 5.8).

The results of the STRUCTURE analysis demonstrated marked population structure for Sternotherus peltifer within the Pascagoula River drainage. An initial ∆K analysis suggested a K=2 explained the most genetic structure within the dataset (Fig.

5.9). This corresponded to a Bouie-Leaf River group well differentiated from all others

(average q-scores = 0.88 and 0.93; Table 5.10). However, the log-likelihood (lnP[K]) plot

127

plateaued at a K of 4 (Fig. 5.9). This corresponds to a well-supported Red Creek-Black

Creek group (average q-score = 0.83), a Bouie-Leaf group (average q-score = 0.87 and

0.92), a Chickasawhay group (average q-score = 0.88) and a Buckatunna group (average q-score = 0.68; Table 5.10). The intermediate value of K = 3 was also informative, and is therefore included in Figure 5.9. A K of 3 corresponds with a Red Creek-Black Creek grouping, a Bouie-Leaf grouping and a Chickasawhay-Pascagoula grouping (Fig. 5.10).

Estimated rates of contemporary migration from the BAYESASS analysis suggested that migration rates between sites were low with mostly overlapping confidence intervals, and that most turtles in each population were derived from the source site (Table 5.11). The Bouie and Red Creek-Black Creek populations both had very high proportions of turtles derived from the source site (>0.90). Migration rates were generally very low (i.e., had overlapping confidence intervals), except for migration from

1) Leaf turtles into the Bouie River, 2) Buckatunna Creek into the Upper Chickasawhay.

However, these and other values should be interpreted with caution as shared alleles between populations could give the appearance of increased migration rates (e.g.,

Chickasawhay into Red-Black and Upper Chickasawhay into the Bouie).

5.4 Discussion

Despite the relatively small spatial scale of this study, both S. carinatus and S. peltifer exhibited population structure within the Pascagoula River drainage. There were some parallels between the two species, particularly where population structure was most prominent, as well as where rates of contemporary migration were highest. Sternotherus carinatus was found throughout the large streams and rivers of the Pascagoula River drainage, though it was seemingly not abundant in the lower reaches of the Pascagoula

128

River despite multiple sampling efforts using a variety of traps, bait and sampling locations to bolster sample size. Sternotherus carinatus was surprisingly abundant in the

Escatawpa River in southwestern Alabama and southeastern Mississippi, whereas I did not detect any populations of S. peltifer in the Escatawpa River drainage. However, S. peltifer were otherwise commonly encountered in small streams and small headwater rivers, despite very few individuals of this species being documented prior to this study

(Iverson 1977; Vogt et al. 1978). This is likely because the two species show diametrically opposed relationship when it comes to the stream size (Brown et al. in prep), and most of the work on chelonians in the Pascagoula River system has focused on the genus Graptemys which have more similar habitat associations as Sternotherus carinatus (McCoy and Vogt 1980; Lindeman 1999; Shelby and Mendonça 2001; Selman and Qualls 2011; Selman et al. 2013, Selman and Lindeman 2018). The two different ecologies of these musk turtles in syntopy has also shaped their intra-drainage population structure.

An initial STRUCTURE analysis for a K = 2 in the razorback musk turtle corresponded to a split between turtles in the Escatawpa River and those in the rest of the drainage (Fig. 5.7). This was reflected in the pairwise FST values as well (Table 5.6). The

Escatawpa sites had consistently higher pairwise FST values within the drainage. This result is very similar to the population structure of a sympatric map turtle, the yellow- blotched sawback (Graptemys flavimaculata; Selman et al. 2013). Selman et al. (2013) speculated that the significant population structure between the Escatawpa River and all other drainages had to do with the geologic history of the Escatawpa River. The

Escatawpa previously flowed independently into the Gulf of Mexico at Grand Bay (Otvos

129

2007). The Escatawpa is believed to have been captured only recently by the Pascagoula

River about 11,500 years ago based off the cessation of fluvial deposits into Grand Bay around this time (Otvos 2007). Currently 34.3 river km of brackish marsh separate the

Pascagoula River from the Escatawpa, and since very little is known about the salt tolerance or estuarine habitat use by S. carinatus, this may facilitate the persistence of genetic differentiation between these populations (Selman et al. 2013).

When the Escatawpa was excluded from the subsequent STRUCTURE analysis

(Fig. 5.7), the results were again similar to the population structure found in Graptemys flavimaculata. Selman et al. (2013) found strong support (i.e., high q-scores) for populations in the Upper Chickasawhay and Upper Leaf, which was also reflected in S. carinatus (Figure 5.7; Table 5.7). Selman et al. (2013) detected a third group of turtles with similar ancestry in the Pascagoula River proper, from where I lacked many samples.

Instead, I detected evidence of a third group in the Red and Black Creek drainages.

However, this K of 3 was not well supported in the ∆K analysis or log-likelihood plot, but it is reflected in the q-scores (Table 5.7). However, because Sternotherus carinatus exhibited a strong isolation by distance trend, STRUCTURE results should be interpreted with caution. STRUCTURE has difficulty inferring number of clusters when there is isolation by distance (Frantz et al. 2009; Guillot et al. 2009; Perez et al. 2018). More specifically, STRUCTURE may overestimate genetic clustering when there is very strong isolation by distance (Frantz et al. 2009).

As previously mentioned, no S. peltifer were caught in the Escatawpa River drainage, so I cannot address whether the species would also exhibit marked differentiation in the drainage, though it would have made for an interesting comparison.

130

However, S. peltifer were found in all remaining subdrainages within the Pascagoula

River drainage. Across these subdrainages, a STRUCTURE analysis found that a K = 2 best represented the genetic data (Figures 5.9 and 5.10). Just like in the razorback musk turtle, the split was again between the Bouie-Leaf sites and the rest of the drainage.

However, the STRUCTURE analysis (log-likelihood plot) also showed support for K = 4, which found the Red Creek-Black Creek sites as distinct along with the Chickasawhay, similar to the K = 3 in S. carinatus (Figures 5.7 and 5.10). The last split in the

STRUCTURE analysis that I found biologically meaningful was K = 3, where there were high ancestry coefficients for Red Creek-Black Creek, Bouie and Leaf Rivers, and the

Chickasawhay-Buckatunna-Pascagoula drainages.

Within each species, and across all sites, neither species differed with conspecifics in allelic richness, observed heterozygosity, or expected heterozygosity.

However, when comparing one species to the other, S. carinatus did have significantly higher allelic richness, observed heterozygosity, and expected heterozygosity than S. peltifer. However, this is not unusual when cross-amplifying loci designed for one species for another closely related species (Primmer et al. 1996).

The higher levels of population substructure seen in S. peltifer likely reflects the ecology of this species within the Pascagoula River system. Because the species is associated with small or adventitious streams and headwater rivers, populations are increasingly isolated from one another in each subdrainage because populations are separated by long distances of unsuitable habitat, in this case, large rivers. This geographic isolation of S. peltifer populations could also explain the reduced levels of allelic richness and heterozygosities found in this study (Kekkonen et al. 2012; Eterovick

131

et al. 2016). All of these metrics were significantly lower in S. peltifer than in S. carinatus, but this could be due to cross-amplifying S. peltifer with primers designed for

S. carinatus as cross-amplication across species can produce lower values of heterozygosity and allelic richness (Primmer et al. 1996). Conversely, because S. carinatus is found throughout the larger lotic systems of the drainage, this contiguous network of streams and rivers has allowed for more gene flow across populations. This opportunity for gene flow has resulted in lower levels of population structure, genetic differentiation, and potentially higher levels of allelic richness and heterozygosity values for the Sternotherus carinatus in the Pascagoula River system.

It was not altogether surprising to detect population structure in either species of

Sternotherus. While movement studies have not been conducted on these species in sympatry, movement studies in allopatry have both demonstrated these turtles both have small home ranges. Ennen and Scott (2013) found that a population of S. peltifer in

Tennessee to have a home range size of 346 ± 224.7 m for females and 335.2 ± 475.7 m, and Kavanagh and Kwiatkowshi (2016) found comparable home ranges sizes in S. carinatus in which ranged from 110-334 m in male and 333-369 m in females. Given these relatively small home range sizes, gene flow between and among populations would likely demonstrate isolation by distance, or the isolation of Sternotherus peltifer across subdrainages.

Small home range sizes may also help explain the low levels of contemporary migration rates. There were interesting similarities between the two species, particularly that both species from the Bouie River turtles had the highest proportion of individuals being derived from that site. While most confidence intervals for migration rates

132

overlapped, in both species of Sternotherus there seemed to be high rates of migration from the Leaf River into the Bouie system (Tables 5.8 and 5.11); this could be due to the close proximity and shared confluence of these two systems. Other notable rates of contemporary migration were S. carinatus from Black Creek moving into Red Creek.

This is not surprising as the two creeks share a confluence to form the Big Black River, which travels a short distance before joining the Pascagoula River. The two sites on

Black Creek and Red Creek were only separated by 41.5 river km, one of the smallest pairwise distance between any S. carinatus sites. However, these migration values should be interpreted with caution because shared alleles and insufficient differentiation could give the appearance of migration. This would be more pronounced on a study of intra- drainage population structure than it would for inter-drainage population structures.

Future directions could incorporate more loci or SNP data to better disentangle contemporary rates of migration between populations.

This study was the first to compare and contrast the riverscape genetics of two sympatric species of turtle. While this may have been the first study on turtles, similar research has been conducted on fish species. Earnest et al. (2014) addressed similar research questions in two species of top-minnow with similar habitat associations as S. carinatus and S. peltifer. Fundulus olivaceus and F. notatus are sister species of topminnow that also co-occur. Fundulus olivaceus is a headwater specialist while F. notatus is a large river specialist. Earnest et al. (2014) studied these two species in sympatry in the Saline River in southern Illinois. Similarly, the headwater stream specialist, F. olivaceus, had more pronounced population structure when compared to the large river specialist. Our two studies differed in that Fundulus olivaceus had a

133

significant pattern of isolation by distance while F. notatus did not, whereas Sternotherus showed the opposite trend of the large river inhabitant demonstrating isolation by distance and the small stream inhabitant exhibiting no isolation by distance. However,

Ernest et al. (2014) found F. olivaceus at all but two of the sites sampled, but the species was generally found at low densities in larger streams. This presumably allowed for limited gene flow between populations, and this could have produced the pattern of isolation by distance. Schmidt and Schaefer (2018) also found that increased headwater specialization in other fishes led to a loss of genetic connectivity across populations. The results of this study corroborate others in the fish literature that suggest that large rivers can fragment headwater specialists. The results of the riverscape genetics of lotic

Sternotherus in the Pascagoula River system are concurrent with patterns seen in fishes, where the biology of the organism (e.g., large river specialist vs. headwater stream specialist) affects the genetic structure of the organism. The two musk turtles in this study demonstrate how biotic (e.g., habitat associations in sympatry) and abiotic factors (fluvial geomorphology) in the same river system play an important role in the genetic structuring of lotic species. Riverscape genetics is still a young branch of ecology, but it has utility in the study of the evolutionary biology, ecology and conservation management of riverine organisms. Future studies of the riverscape genetics of other turtle species within the Pascagoula River drainage could provide a larger, comparative approach to understanding how the ecology of turtle species within the Pascagoula River system is reflected in their riverscape genetics. The Pascagoula River and its backwaters are home to more species of turtle than almost any other river system in the continental United

States (N = 14-15; Ernst and Lovich 2009), and it is also the largest unimpounded river

134

system left in the continental United States. These criteria lend themselves well for a meta-analysis of the riverscape genetics of freshwater turtles, a better understanding of turtle ecology, and potentially a null model for other studies of riverscape genetics in modified lotic environments.

Acknowledgements

This study benefitted from genetic samples donated by Luke Pearson and Gabbie

Berry. All turtles were caught and handled under appropriate state permits (Alabama:

#9029; Mississippi: MMNS #’s: 0607172, 0530181, 0517191) and IACUC numbers:

11092206 and 17101202 from the University of Southern Mississippi. This work has been funded by the following grants (in alphabetical order): the American Turtle

Observatory, Birmingham Audubon Society, Chicago Herpetological Society, NSF

Graduate Research Fellowship (#1842492), and the Theodore Roosevelt Memorial Grant through the American Museum of Natural History.

135

Table 5.1 Grouped samples for analysis from the Pascagoula River Drainage for

Sternotherus carinatus.

Drainage Site Lat. Long.

Red Creek

Red Creek @ Hwy 15 30.774 -88.913

Red Creek @ Vestry 30.736 -88.781

Black Creek

Black Creek @ Private Access 31.106 -89.286

Black Creek @ Hwy 57 30.780 -88.763

Black Creek @ Hwy 11 31.191 -89.377

Black Creek @ Fairley 30.919 -88.966

Bouie River

Bouie @ Seminary-Sumrall 31.497 -89.558

Okatoma Creek @ Seminary 31.563 -89.504

Bouie River @ Private Access 31.446 -89.441

Bouie @ Hwy 49 31.426 -89.415

Okatoma Creek @ Lux 31.446 -89.409

Upper Leaf

Oakahay Creek 31.743 -89.443

Leaf River @ Taylorsville, MS 31.974 -89.419

Leaf @ Eastabuchie, MS 31.438 -89.300

136

Table 5.1 (continued)

Leaf @ Chain Park 31.345 -89.280

Leaf @ Sims Rd 31.261 -89.226

Middle Leaf

Tallahala Creek @ Ellisville 31.591 -89.167

Bogue Homa @ Ovett 31.480 -89.046

Bogue Homa @ Richton 31.358 -89.026

Leaf @ Beaumont, MS 31.183 -88.919

Gaines Creek 31.253 -88.865

Thompson @ Hintonville 31.261 -88.902

Upper

Chickasawhay

Chunky @ Chunky, MS 32.321 -88.932

Chunky @ Dunns Falls 32.229 -88.822

Chickasawhay River @ Stonewall 32.130 -88.812

Okatibee Creek @ Enterprise, MS 32.199 -88.786

Chickasawhay @ De Soto 31.975 -88.705

Buckatunna Creek @ Frost Church 31.799 -88.491

Lower

Chickasawhay

Chickasawhay River @ Leakesville,

MS 31.148 -88.548

137

Table 5.1 (continued)

Pascagoula River

Pascagoula @ Benndale 30.904 -88.748

Pascagoula River @ Merrill 30.977 -88.726

Poticaw Bayou on Pascagoula River 30.513 -88.619

Pascagoula River @ Wilkerson Ferry 30.811 -88.743

Escatawpa River

Escatawpa @ Presley Landing 30.485 -88.436

Escatawpa @ Hurley, MS 30.630 -88.433

Escatawpa @ Wilmer 30.864 -88.414

Escatawpa River @ Cintronelle 31.086 -88.376

138

Table 5.2 Grouped samples for analysis from the Pascagoula River Drainage for

Sternotherus peltifer.

Drainage Site Lat. Long.

Red Creek

Double Branch Creek 30.933 -89.283

Black Creek

Mixon Creek @ Purvis 31.209 -89.414

Sandy Run Creek 31.207 -89.388

Blacktom Creek @ Purvis 31.183 -89.358

Sweetwater Creek 30.854 -88.847

Black Creek @ Hwy 11 31.191 -89.377

Black Creek @ Fairley 30.919 -88.966

Bouie River

Big Creek @ Hattiesburg 31.398 -89.409

Bouie @ Peps Point 31.396 -89.367

Mixons Creek @ Hattiesburg 31.348 -89.351

Bouie @ Seminary-Sumrall 31.497 -89.558

Okatoma Creek @ Seminary 31.563 -89.504

Bouie River @ Private Access 31.446 -89.441

Bouie @ Hwy 49 31.426 -89.415

Okatoma Creek @ Lux 31.446 -89.409

139

Table 5.2 (continued)

Upper Leaf

Clear Creek 31.867 -89.562

Oakey Woods Creek 31.625 -89.358

Snows Creek 31.557 -89.329

Big Creek @ Soso, MS 31.717 -89.311

Middle Leaf

Myers Creek 31.245 -89.240

Garraway Creek 31.211 -89.127

Thompson Creek 31.357 -88.925

Atkinson Creek 31.140 -88.800

Thompson @ Hintonville 31.261 -88.902

Upper

Chickasawhay

Chunky River

Chunky River @ Hwy 80 32.326 -88.909

Possum Creek 32.273 -88.898

Chunky @ Chunky, MS 32.321 -88.932

Chunky @ Dunns Falls 32.229 -88.822

Okatibbee Creek

Sowashee Creek 32.344 -88.727

Okatibee Creek @ Enterprise, MS 32.199 -88.786

140

Table 5.2 (continued)

Chickasawhay River

Scotchenflipper Creek 32.221 -89.029

Shabuta Creek 31.944 -88.838

Chickasawhay @ Waynesboro 31.680 -88.684

Little Creek @ Waynesboro 31.470 -88.650

Chickasawhay River @ Stonewall 32.130 -88.812

Chickasawhay @ De Soto 31.975 -88.705

Buckatunna Creek

Buckatunna @ Waynesboro 31.654 -88.522

Cedar Creek @ De Soto 31.961 -88.509

Red Creek @ Millry, AL 31.581 -88.451

Buckatunna Creek @ Frost Church 31.799 -88.491

Pascagoula River

Big Cedar Creek @ Agricola, MS 30.763 -88.575

Pascagoula River @ Wilkerson Ferry 30.811 -88.743

141

Table 5.3 List of all sites where Sternotherus carinatus and S. peltifer were detected across all years from the Pascagoula River drainage.

Site # Site Name Latitude Longitude

1 Oakahay Creek 31.74 -89.44

2 Leaf River @ Taylorsville, MS 31.97 -89.42

3 Leaf @ Eastabuchie, MS 31.44 -89.30

4 Black Creek @ Private Access 31.11 -89.29

5 Leaf @ Chain Park 31.35 -89.28

6 Leaf @ Sims Rd 31.26 -89.23

7 Tallahala Creek @ Ellisville 31.59 -89.17

8 Bogue Homa @ Ovett 31.48 -89.05

9 Bogue Homa @ Richton 31.36 -89.03

10 Leaf @ Beaumont, MS 31.18 -88.92

11 Red Creek @ Hwy 15 30.77 -88.91

12 Gaines Creek 31.25 -88.86

13 Red Creek @ Vestry 30.74 -88.78

14 Black Creek @ Hwy 57 30.78 -88.76

15 Pascagoula @ Benndale 30.90 -88.75

16 Pascagoula River @ Merrill 30.98 -88.73

17 Poticaw Bayou on Pascagoula River 30.51 -88.62

18 Chickasawhay River @ Leakesvillle, MS 31.15 -88.55

142

Table 5.3 (continued)

19 Escatawpa @ Presley Landing 30.48 -88.44

20 Escatawpa @ Hurley, MS 30.63 -88.43

21 Escatawpa @ Wilmer 30.86 -88.41

22 Escatawpa River @ Cintronelle 31.09 -88.38

23 Clear Creek 31.87 -89.56

24 Mixon Creek @ Purvis 31.21 -89.41

25 Big Creek @ Hattiesburg 31.40 -89.41

26 Sandy Run Creek 31.21 -89.39

27 Bouie @ Peps Point 31.40 -89.37

28 Oakey Woods Creek 31.62 -89.36

29 Blacktom Creek @ Purvis 31.18 -89.36

30 Mixons Creek @ Hattiesburg 31.35 -89.35

31 Snows Creek 31.56 -89.33

32 Big Creek @ Soso, MS 31.72 -89.31

33 Double Branch Creek 30.93 -89.28

34 Myers Creek 31.25 -89.24

35 Garraway Creek 31.21 -89.13

36 Scotchenflipper Creek 32.22 -89.03

37 Thompson Creek 31.36 -88.93

38 Chunky River @ Hwy 80 32.33 -88.91

39 Possum Creek 32.27 -88.90

143

Table 5.3 (continued)

40 Sweetwater Creek 30.85 -88.85

41 Shabuta Creek 31.94 -88.84

42 Atkinson Creek 31.14 -88.80

43 Sowashee Creek 32.34 -88.73

44 Chickasawhay @ Waynesboro 31.68 -88.68

45 Little Creek @ Waynesboro 31.47 -88.65

46 Big Cedar Creek @ Agricola, MS 30.76 -88.57

47 Buckatunna @ Waynesboro 31.65 -88.52

48 Cedar Creek @ De Soto 31.96 -88.51

49 Red Creek @ Millry, AL 31.58 -88.45

50 Bouie @ Seminary-Sumrall 31.50 -89.56

51 Okatoma Creek @ Seminary 31.56 -89.50

52 Bouie River @ Private Access 31.45 -89.44

53 Bouie @ Hwy 49 31.43 -89.42

54 Okatoma Creek @ Lux 31.45 -89.41

55 Black Creek @ Hwy 11 31.19 -89.38

56 Black Creek @ Fairley 30.92 -88.97

57 Chunky @ Chunky, MS 32.32 -88.93

58 Thompson @ Hintonville 31.26 -88.90

59 Chunky @ Dunns Falls 32.23 -88.82

60 Chickasawhay River @ Stonewall 32.13 -88.81

144

Table 5.3 (continued)

61 Okatibbee Creek @ Enterprise, MS 32.20 -88.79

62 Pascagoula River @ Wilkerson Ferry 30.81 -88.74

63 Chickasawhay @ De Soto 31.98 -88.70

64 Buckatunna Creek @ Frost Church 31.80 -88.49

65 Terrible Creek 31.59 -89.63

66 Cooks Branch 31.51 -89.56

67 Double Branch Creek 31.25 -89.24

68 Red Creek @ Wiggins, MS 31.25 -89.24

69 Mill Creek @ Wiggins 30.77 -89.14

70 Flint Creek 30.81 -89.07

71 Black Creek @ Janice 30.99 -89.05

72 Hickory Creek @ Benndale 30.97 -88.97

73 Allen Creek 32.28 -88.85

74 Pascagoula River down from Benndale 30.83 -88.74

75 Big Creek @ Clara, MS 31.55 -88.68

76 Pascagoua River @ Wade Vancleace 30.61 -88.64

77 Big Creek @ Old 24 31.17 -88.63

78 Brushy Creek 30.94 -88.45

Sites 65-78 were sampled but no Sternotherus were detected.

145

Table 5.4 Summary statistics for 8 microsatellite loci used to genotype Sternotherus carinatus in the Pascagoula River System.

Subdrainage N A Ar Ho He

Red 28 9.9 7.00 0.825 0.817

Black 13 7.6 6.77 0.756 0.778

Bouie 34 8.6 5.71 0.721 0.726

Upper Leaf 44 9.9 6.00 0.783 0.786

Middle Leaf 30 10.0 6.21 0.775 0.767

Upper

Chick. 13 7.0 6.39 0.831 0.798

L. Chick-

Pasc 13 7.5 6.38 0.750 0.777

Escatawpa 38 9.0 6.69 0.758 0.779 N = number of turtles genotyped, A = number of alleles, Ar = allelic richness, Ho = observed heterozygosity, and He = expected heterozygosity. Upper Chick. = Upper Chickasawhay, L. Chick-Pasc. = Lower Chickasawhay-Pascagoula.

146

Table 5.5 Summary statistics for 10 microsatellite loci used to genotype Sternotherus peltifer in the Pascagoula River System.

Subdrainage N A Ar Ho He

Red-Black 42 6.800 4.7703 0.596 0.649

Bouie 61 6.800 4.4293 0.580 0.602

Leaf 47 6.400 4.2745 0.545 0.588

Upper Chick. 19 6.300 4.8828 0.555 0.603

Chickasawhay 32 7.800 5.3559 0.605 0.635

Buckatunna 14 5.500 4.7155 0.492 0.565

Pascagoula 6 4.300 - 0.483 0.574

N = number of turtles genotyped, A = number of alleles, Ar = allelic richness, Ho = observed heterozygosity, and He = expected heterozygosity. Upper Chick. = Upper Chickasawhay

147 Table 5.6 Pairwise FST values (below diagonal) and river distances (above diagonal; in river kilometer) across 8 sites for

Sternotherus carinatus within the Pascagoula River system. Upper L. Chick - Site Red Black Bouie Upper Leaf Middle Leaf Chick. Pasc Escatawpa Red 0 41.5 210.9 253.8 156.6 318.5 72.9 166.4

Black 0.007 0 213.4 256.4 159.2 321.1 75.5 169.3

Bouie 0.040 0.049 0 92.7 54.2 383.6 138.0 329.9

Upper Leaf 0.017 0.025 0.020 0 97.2 426.6 180.9 372.9

Middle Leaf 0.022 0.041 0.016 0.007 0 329.4 83.7 275.7

Upper Chick. 0.018 0.053 0.078 0.042 0.051 0 245.6 437.6

L. Chick-Pasc 0.011 0.022 0.021 0.009 0.005 0.026 0 192.0

Escatawpa 0.028 0.036 0.083 0.063 0.072 0.047 0.051 0

Upper Chick. = Upper Chickasawhay, L. Chick.-Pasc. = Lower Chickasawhay-Pascagoula. Values significantly different from zero are bolded and underlined. Table 5.7 Average group membership scores (q-scores) for Sternotherus carinatus at each site from the STRUCTURE analysis for K

= 2 and K = 3, with the latter being a separate analysis without the Escatawpa River site.

K = 2 K = 3 K = 2 K = 3 1 2 1 2 3 1 2 1 2 3 Red 0.367 0.633 0.125 0.803 0.072 0.910 0.090 0.105 0.810 0.086 Black 0.394 0.606 0.176 0.706 0.118 0.873 0.127 0.031 0.872 0.097 Bouie 0.974 0.026 0.936 0.032 0.032 0.051 0.949 0.015 0.047 0.938 Upper Leaf 0.941 0.059 0.795 0.193 0.012 0.217 0.783 0.109 0.126 0.765 Middle Leaf 0.887 0.113 0.758 0.211 0.031 0.266 0.734 0.115 0.152 0.733 Upper Chickasawhay 0.239 0.761 0.063 0.882 0.056 0.940 0.060 0.857 0.063 0.080 L. Chick-Pascagoula 0.677 0.323 0.513 0.433 0.054 0.536 0.464 0.252 0.295 0.453 Escatawpa 0.020 0.980 0.018 0.044 0.938 - - - - -

Table 5.8 Rates of recent migration between sites of Sternotherus carinatus as estimated with the program BAYESASS.

Migration from Migration into Red Black Bouie Upper Leaf Middle Leaf Upper Chick L. Chick-Pasc Escatawpa Red 0.689 (0.67-0.709) 0.015 (0-0.032) 0.047 (0.008-0.086) 0.109 (0.049-0.169) 0.061 (0.015-0.108) 0.018 (0-0.038) 0.011 (0-0.021) 0.05 (0.013-0.086) Black 0.112 (0.061-0.164) 0.687 (0.668-0.705) 0.074 (0.032-0.115) 0.044 (0.004-0.085) 0.024 (0-0.049) 0.02 (0.001-0.038) 0.018 (0.001-0.034) 0.022 (0.001-0.043) Bouie 0.011 (0-0.021) 0.009 (0-0.017) 0.92 (0.893-0.947) 0.017 (0.001-0.033) 0.016 (0.001-0.031) 0.009 (0-0.017) 0.008 (0-0.016) 0.011 (0-0.021) Upper Leaf 0.01 (0-0.019) 0.007 (0-0.014) 0.262 (0.237-0.287) 0.684 (0.669-0.7) 0.015 (0-0.029) 0.007 (0-0.015) 0.007 (0-0.013) 0.008 (0-0.016) Middle Leaf 0.011 (0-0.022) 0.009 (0-0.018) 0.226 (0.19-0.261) 0.044 (0.014-0.074) 0.679 (0.667-0.69) 0.01 (0-0.019) 0.009 (0-0.018) 0.013 (0.001-0.025) Upper Chick 0.048 (0-0.095) 0.019 (0.001-0.037) 0.042 (0.007-0.077) 0.083 (0.018-0.148) 0.057 (0.01-0.104) 0.699 (0.668-0.73) 0.018 (0.001-0.035) 0.035 (0.005-0.066) L. Chick-Pasc 0.028 (0.004-0.052) 0.018 (0.001-0.036) 0.127 (0.081-0.173) 0.054 (0.012-0.095) 0.048 (0.013-0.082) 0.024 (0.003-0.045) 0.683 (0.667-0.698) 0.019 (0.001-0.037) Escatawpa 0.032 (0-0.066) 0.011 (0-0.022) 0.026 (0.002-0.049) 0.033 (0-0.066) 0.02 (0-0.04) 0.013 (0-0.027) 0.009 (0-0.017) 0.856 (0.807-0.906)

The values represent the migration rate into a given site along with the 95% confidence intervals in parentheses. Values along the diagonal (bolded) are the proportion of individuals derived from the source site. L. Chick.-Pasc = Lower Chickasawhay-Pascagoula. Table 5.9 Pairwise FST values (below diagonal) and river distances (above diagonal; in river kilometer) across 7 sites for

Sternotherus peltifer within the Pascagoula River system.

Site Red-Black Bouie Leaf Upper Chick. Chickasawhay Buckatunna Pascagoula

Red-Black 0 191.0 163.7 334.6 305.3 197.1 4.5

Bouie 0.061 0 27.2 419.6 390.3 282.1 187.6

Leaf 0.079 0.012 0 392.4 363.0 254.9 160.4

Upper Chick 0.050 0.067 0.097 0 29.4 215.5 331.3

Chickasawhay 0.032 0.074 0.101 0.048 0 186.1 302.0

Buckatunna 0.112 0.123 0.152 0.080 0.077 0 193.8

Pascagoula 0.041 0.077 0.114 0.036 0.019 0.064 0

Upper Chick. = Upper Chickasawhay

Table 5.10 Average group membership scores (q-scores) for Sternotherus peltifer from each group from the STRUCTURE analysis for

K = 2, K = 3 and K =4.

K = 2 K = 3 K = 4 1 2 1 2 3 1 2 3 4 Red-Black 0.148 0.852 0.117 0.020 0.863 0.825 0.052 0.006 0.117 Bouie 0.877 0.123 0.891 0.048 0.061 0.041 0.078 0.013 0.868 Leaf 0.931 0.069 0.935 0.041 0.024 0.019 0.054 0.007 0.920 Upper Chickasawhay 0.288 0.712 0.299 0.642 0.058 0.049 0.322 0.323 0.306 Chickasawhay 0.068 0.933 0.103 0.773 0.124 0.030 0.875 0.036 0.059 Buckatunna 0.108 0.892 0.073 0.915 0.012 0.023 0.188 0.678 0.111 Pascagoula 0.194 0.866 0.179 0.691 0.131 0.060 0.723 0.043 0.174

Table 5.11 Rates of recent migration between sites of Sternotherus peltifer as estimated with the program BAYESASS.

Migration from Migration into Red-Black Bouie Leaf Upper Chickasawhay Chickasawhay Buckatunna Pascagoula Red-Black 0.923 (0.892-0.953) 0.024 (0-0.043) 0.014 (0-0.032) 0.011 (0.001-0.021) 0.015 (0.001-0.029) 0.007 (0-0.014) 0.007 (0-0.014) Bouie 0.025 (0.008-0.042) 0.937 (0.916-0.959) 0.006 (-0.001-0.013) 0.011 (0.002-0.02) 0.011 (0.002-0.019) 0.005 (0-0.01) 0.005 (0-0.01) Leaf 0.007 (0-0.015) 0.292 (0.276-0.308) 0.674 (0.667-0.681) 0.007 (0-0.013) 0.007 (0-0.013) 0.007 (0-0.013) 0.007 (0-0.013) Upper Chick 0.028 (0.003-0.053) 0.155 (0.115-0.194) 0.019 (0-0.039) 0.738 (0.705-0.771) 0.034 (0.006-0.062) 0.013 (0.001-0.026) 0.013 (0.001-0.026) Chickasawhay 0.129 (0.083-0.174) 0.039 (0.013-0.065) 0.03 (0-0.068) 0.065 (0.03-0.1) 0.721 (0.689-0.752) 0.009 (0-0.018) 0.009 (0-0.017) Buckatunna 0.026 (0.004-0.049) 0.036 (0.012-0.06) 0.022 (0.002-0.041) 0.179 (0.138-0.22) 0.039 (0.011-0.067) 0.683 (0.667-0.698) 0.016 (0.001-0.031) Pascagoula 0.068 (0.029-0.107) 0.053 (0.02-0.087) 0.039 (0.002-0.077) 0.031 (0.002-0.06) 0.092 (0.041-0.142) 0.026 (0.002-0.049) 0.692 (0.669-0.716)

Up per C hick asaw hay

Bu ck at un na C r.

B o u C i h e ic k C a re s e a k w h a y R iv e r

Le af Riv er Bl ack Cr eek

E sc at aw pa Red R Cre iv ek er

P as ca go ul a R iv er

Figure 5.1 The Pascagoula River drainage and the major subdrainages.

154

Figure 5.2 Capture locations for Sternotherus within the Pascagoula River drainage.

a) Sternotherus carinatus b) Sternotherus peltifer, c) syntopy, d) sites where no Sternotherus were detected. Numbers correspond to those found in Table 3.

155

Figure 5.3 Isolation by distance among all sites of Sternotherus carinatus as indicated by the significant positive correlation between pairwise genetic distances (FST) and river distance (river kilometers).

156

Figure 5.4 Isolation by distance among for Sternotherus carinatus (excluding the

Escatawpa population) as indicated by a significant positive correlation between pairwise genetic distance (FST) and pairwise river distance (river kilometer).

157

Delta K Mean LnP(K) ± Stdev

0 9

● 5 −

0 ●

3 ● 0

5 ●

5 2

− ●

)

(

−

5 1

● 6 −

● ● 0 1 2 3 4 5 1 2 3 4 5

K K

Figure 5.5 ∆K analysis and log-likelihood plots for all Sternotherus carinatus across all sites in the Pascagoula River System. Delta K Mean LnP(K) ± Stdev

● ●

0 ●

−

5 2

● 0

5 ●

0 5

0 ●

−

)

(

0 n

a ●

−

e M

●

5 −

●

5 − ● ●

● 0

1 2 3 4 5 6 7 1 2 3 4 5 6 7

K K

Figure 5.6 ∆K analysis and log-likelihood plots for Sternotherus carinatus across sites in the Pascagoula River System, excluding the

Escatawpa drainage.

Figure 5.7 Bar plots of membership coefficients from the STRUCTURE analysis of

Sternotherus carinatus for values of K = 2 and K = 3 for analyses with and without the

Escatawpa River turtles.

L. Chick-Pascagoula = Lower Chickasawhay-Pascagoula.

160

Figure 5.8 Genetic distance by river km for Sternotherus peltifer. Sternotherus peltifer does not exhibit a significant correlation in pairwise genetic distance (FST) and river distance (river kilometer).

161

Delta K Mean LnP(K) ± Stdev

0 8

● 5 −

0 ● 2 ●

●

0 ●

−

5 1

) ●

(

−

5 1

● 6 −

●

● ● 0

1 2 3 4 5 6 1 2 3 4 5 6

K K

Figure 5.9 ∆K analysis and Log-likelihood plots for Sternotherus peltifer in the Pascagoula River System

Figure 5.10 Bar plots of membership coefficients from the STRUCTURE analysis of

Sternotherus peltifer for values of K = 2, K = 3, and K = 4.

163

5.5 Literature Cited

Altermatt, F. 2013. Diversity in riverine metacommunities: A network perspective.

Aquatic Ecology 47:365–377.

Antolin, M. F., Savage, L. T., and Eisen, R. J. 2006. Landscape features influence genetic

structure of black-tailed prairie dogs (Cynomys ludovicianus). Landscape

Ecology, 21:867–875.

Bennett, A. M., Keevil, M., and Litzgus, J. D. 2010. Spatial ecology and population

genetics of northern map turtles (Graptemys geographica) in fragmented and

continuous habitats in Canada. Chelonian Conservation and Biology 9:185–195.

Bowlby, H. D., Fleming, I. A., and Gibson, A. J. F. 2016. Applying landscape genetics to

evaluate threats affecting endangered Atlantic salmon populations. Conservation

Genetics 17:823–838.

Brown, G. J., and Kreiser, B. R. 2020. Characterization of microsatellite loci for the

razorback musk turtle (Sternotherus carinatus) and their cross-amplification in

five other species in the family Kinosternidae. Chelonian Conservation and

Biology. In Press.

Carbonneau, P., Fonstad, M. A., Marcus, W. A., and Dugdale, S. J. 2012. Making

riverscapes real. Geomorphology 137:74–86.

Davis, C. D., Epps, C. W., Flitcroft, R. L., and Banks, M. A. 2018. Refining and defining

riverscape genetics: How rivers influence population genetic structure. WIREs

Water 5:e1269. dos Santos, R. C., das Neves Silva Viana, M., dos Santos Monjeló, L. A., Andrade, P. C.

M., Pantoja-Lima, J., Oliveira, P. H. G., Vogt, R. C., Pezzuti, J. C. B., Sites, J.

164

W., Hrbek, T., and Farias, I. P. 2016. Testing the effects of barriers on the genetic

connectivity in Podocnemis erythrocephala (red-headed amazon river turtle):

Implications for management and conservation. Chelonian Conservation and

Biology 15:12–22.

Earnest, K., Scott, J., Schaefer, J., and Duvernell, D. 2014. The landscape genetics of

syntopic topminnows (Fundulus notatus and F. olivaceus) in a riverine contact

zone. Ecology of Freshwater Fish 23:572–580.

Elliott, J. M. 2003. A comparative study of the dispersal of 10 species of stream

invertebrates. Freshwater Biology, 48:1652–1668.

Ennen, J. R., and Scott, A. F. 2013. Home range characteristics and overwintering

ecology of the stripe-necked musk turtle (Sternotherus minor peltifer) in Middle

Tennessee. Chelonian Conservation and Biology 12:199–203.

Ernst, C. H., and Lovich, J. E. 2009. Turtles of the United States and Canada (2nd ed).

Johns Hopkins University Press.

Eterovick, P. C., Sloss, B. L., Scalzo, J. A. M., and Alford, R. A. 2016. Isolated frogs in a

crowded world: Effects of human-caused habitat loss on frog heterozygosity and

fluctuating asymmetry. Biological Conservation 195:52–59.

Evanno, G., Regnaut, S., and Goudet, J. 2005. Detecting the number of clusters of

individuals using the software structure: A simulation study. Molecular Ecology

14:2611–2620.

Fagan, W. F. 2002. Connectivity, Fragmentation, and Extinction Risk in Dendritic

Metapopulations. Ecology, 83:3243–3249.

165

Frantz, A. C., Cellina, S., Krier, A., Schley, L., and Burke, T. 2009. Using spatial

Bayesian methods to determine the genetic structure of a continuously distributed

population: Clusters or isolation by distance? Journal of Applied Ecology 46:493–

505.

Galbraith, H. S., Zanatta, D. T., and Wilson, C. C. 2015. Comparative analysis of

riverscape genetic structure in rare, threatened and common freshwater mussels.

Conservation Genetics, 16:845–857.

Geffen, E., Anderson, M. J., and Wayne, R. K. 2004. Climate and habitat barriers to

dispersal in the highly mobile grey wolf. Molecular Ecology, 13:2481–2490.

Hubisz, M. J., Falush, D., Stephens, M., and Pritchard, J. K. 2009. Inferring weak

population structure with the assistance of sample group information. Molecular

Ecology Resources 9:1322–1332.

Inoue, K., Monroe, E. M., Elderkin, C. L., and Berg, D. J. 2014. Phylogeographic and

population genetic analyses reveal Pleistocene isolation followed by high gene

flow in a wide ranging, but endangered, freshwater mussel. Heredity 112:282–

290.

Iverson, J. B. 1977. Geographic Variation in the Musk Turtle, Sternotherus minor. Copeia

1977:502–517.

Kanno, Y., Vokoun, J. C., and Letcher, B. H. 2011. Fine-scale population structure and

riverscape genetics of brook trout (Salvelinus fontinalis) distributed continuously

along headwater channel networks. Molecular Ecology 20:3711–3729.

166

Kavanagh, B. T., and Kwiatkowski, M. A. 2016. Sexual dimorphism, movement patterns,

and diets of Sternotherus carinatus (razorback musk turtle). Southeastern

Naturalist 15:117–133.

Kekkonen, J., Wikström, M., and Brommer, J. E. 2012. Heterozygosity in an isolated

population of a large mammal founded by four individuals is predicted by an

individual-based genetic model. PLoS ONE 7: e43482.

Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., and Mayrose, I. 2015.

Clumpak: A program for identifying clustering modes and packaging population

structure inferences across K. Molecular Ecology Resources 15:1179–1191.

Li, Y.-L., and Liu, J.-X. 2018. StructureSelector: A web-based software to select and

visualize the optimal number of clusters using multiple methods. Molecular

Ecology Resources, 18:176–177.

Lindeman, P. V. 1999. Surveys of basking map turtles Graptemys spp. in three river

drainages and the importance of deadwood abundance. Biological Conservation

88:33–42.

Manel, S., and Holderegger, R. 2013. Ten years of landscape genetics. Trends in Ecology

and Evolution 28:614–621.

Manel, S., Schwartz, M. K., Luikart, G., and Taberlet, P. 2003. Landscape genetics:

Combining landscape ecology and population genetics. Trends in Ecology and

Evolution 18:189–197.

Otvos EG (2007) Geological framework and evolution history. In: Peterson MS, Waggy

GL, Woodrey MS (eds) Grand Bay National Estuarine Research Reserve: an

167

ecological characterization. Grand Bay National Estuarine Research Reserve,

Moss Point, pp 22–46

McCoy, C. J., and R. C. Vogt. 1980. Distribution and population status of the ringed

sawback Graptemys oculifera (Baur) in Mississippi and Louisiana. Final Report,

U.S. Fish and Wildlife Service contract number 14-16-0004-79-038.

Pauls, S. U., Nowak, C., Bálint, M., and Pfenninger, M. 2013. The impact of global

climate change on genetic diversity within populations and species. Molecular

Ecology 22:925–946.

Peakall, R., and Smouse, P. E. 2006. genalex 6: Genetic analysis in Excel. Population

genetic software for teaching and research. Molecular Ecology Notes 6:288–295.

Peakall, P., and R. Smouse. 2012. GenAlEx 6.5: genetic analysis in Excel. Population

genetic software for teaching and research–an update. Bioinformatics 28:2537.

Perez, M. F., Franco, F. F., Bombonato, J. R., Bonatelli, I. A. S., Khan, G., Romeiro‐

Brito, M., Fegies, A. C., Ribeiro, P. M., Silva, G. A. R., and Moraes, E. M. 2018.

Assessing population structure in the face of isolation by distance: Are we

neglecting the problem? Diversity and Distributions 24:1883–1889.

Primmer, C. R., Møller, A. P., and Ellegren, H. 1996. A wide-range survey of cross-

species microsatellite amplification in birds. Molecular Ecology 5:365–378.

Pritchard, J. K., Stephens, M., and Donnelly, P. 2000. Inference of population structure

using multilocus genotype data. Genetics 155:945–959.

Rambaut, A., and Drummond, A. J. 2003. Tracer: MCMC trace analysis tool. University

of Oxford.

168

Raymond, M., and Rousset, F. 1995. GENEPOP (Version 1.2): Population genetics

software for exact tests and ecumenicism. Journal of Heredity 86:248–249.

Reid, B. N., Mladenoff, D. J., and Peery, M. Z. 2017. Genetic effects of landscape,

habitat preference and demography on three co-occurring turtle species.

Molecular Ecology 26:781–798.

Rice, W. R. 1989. Analyzing Tables of Statistical Tests. Evolution 43:223:–225. JSTOR.

Riley, S. P. D., Pollinger, J. P., Sauvajot, R. M., York, E. C., Bromley, C., Fuller, T. K.,

and Wayne, R. K. 2006. FAST-TRACK: A southern California freeway is a

physical and social barrier to gene flow in carnivores. Molecular Ecology

15:1733–1741.

Rosenberg, N. A. 2004. distruct: A program for the graphical display of population

structure. Molecular Ecology Notes 4:137–138.

Rousset, F. 2008. genepop’007: A complete re-implementation of the genepop software

for Windows and Linux. Molecular Ecology Resources 8:103–106.

Schmidt, B. V., and Schaefer, J. 2018. Ecological and landscape effects on genetic

distance in an assemblage of headwater fishes. Ecology of Freshwater Fish

27:617–631.

Selman, W., Kreiser, B., and Qualls, C. 2013. Conservation genetics of the yellow-

blotched sawback Graptemys flavimaculata (Testudines: Emydidae).

Conservation Genetics, 14:1193–1203.

Selman, Will, and Lindeman, P. V. 2018. Spatial, seasonal, and sexual variation in the

diet of Graptemys flavimaculata, a threatened turtle of the Pascagoula River

system, Mississippi, USA. Copeia, 106:247–254.

169

Selman, Will, and Qualls, C. P. 2011. Basking ecology of the yellow-blotched sawback

(Graptemys flavimaculata), an imperiled turtle species of the Pascagoula River

system, Mississippi, United States. Chelonian Conservation and Biology, 10:188–

197.

Shelby, J. A., and Mendonça, M. T. 2001. Comparison of reproductive parameters in

male yellow-blotched map turtles (Graptemys flavimaculata) from a historically

contaminated site and a reference site. Comparative Biochemistry and Physiology

Part C: Toxicology and Pharmacology 129:233–242.

Sprehn, C. G., Blum, M. J., Quinn, T. P., and Heins, D. C. 2015. Landscape genetics of

Schistocephalus solidus parasites in threespine stickleback (Gasterosteus

aculeatus) from Alaska. PLoS ONE,10:1–17

Storfer, A., Murphy, M. A., Evans, J. S., Goldberg, C. S., Robinson, S., Spear, S. F.,

Dezzani, R., Delmelle, E., Vierling, L., and Waits, L. P. 2007. Putting the

‘landscape’ in landscape genetics. Heredity 98:128–142.

Tyers, M. 2016. riverdist: River network distance computation and applications. Package

version 0.13. 0.

Vähä, J.-P., Erkinaro, J., Niemelä, E., and Primmer, C. R. 2007. Life-history and habitat

features influence the within-river genetic structure of Atlantic salmon. Molecular

Ecology 16:2638–2654.

Vogt, R. C., McCoy, C. J., and Bianculi, A. V. 1978. Sternotherus minor (Reptilia,

Kinosternidae) in the Pascagoula River system, Mississippi. Herpetological

Review 9:109.

170

Wang, Q., Yu, Y., Yuan, J., Zhang, X., Huang, H., Li, F., and Xiang, J. 2017. Effects of

marker density and population structure on the genomic prediction accuracy for

growth trait in Pacific white shrimp Litopenaeus vannamei. BMC Genetics 18:45.

Weir, B. S., and Cockerham, C. C. (1984). Estimating F-statistics for the analysis of

population structure. Evolution, 38:1358–1370.

Whelan, N. V., Galaska, M. P., Sipley, B. N., Weber, J. M., Johnson, P. D., Halanych, K.

M., and Helms, B. S. 2019. Riverscape genetic variation, migration patterns, and

morphological variation of the threatened round rocksnail, Leptoxis ampla.

Molecular Ecology 28:1593–1610.

Wilson, G. A., and Rannala, B. 2005. Documentation for BayesAss 1.3. University of

California Berkeley, Berkeley, CA.

Wright, S. 1943. Isolation by distance. Genetics 28:114–138.

171